{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 200.0, "eval_steps": 500, "global_step": 32200, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.006230529595015576, "grad_norm": 21.625, "learning_rate": 0.0, "loss": 4.4544, "step": 1 }, { "epoch": 0.012461059190031152, "grad_norm": 21.75, "learning_rate": 1.5527950310559006e-07, "loss": 4.4809, "step": 2 }, { "epoch": 0.018691588785046728, "grad_norm": 20.5, "learning_rate": 3.1055900621118013e-07, "loss": 4.3941, "step": 3 }, { "epoch": 0.024922118380062305, "grad_norm": 20.5, "learning_rate": 4.6583850931677024e-07, "loss": 4.4228, "step": 4 }, { "epoch": 0.03115264797507788, "grad_norm": 20.875, "learning_rate": 6.211180124223603e-07, "loss": 4.3986, "step": 5 }, { "epoch": 0.037383177570093455, "grad_norm": 21.375, "learning_rate": 7.763975155279503e-07, "loss": 4.5342, "step": 6 }, { "epoch": 0.04361370716510903, "grad_norm": 21.375, "learning_rate": 9.316770186335405e-07, "loss": 4.4011, "step": 7 }, { "epoch": 0.04984423676012461, "grad_norm": 21.0, "learning_rate": 1.0869565217391306e-06, "loss": 4.4421, "step": 8 }, { "epoch": 0.056074766355140186, "grad_norm": 20.875, "learning_rate": 1.2422360248447205e-06, "loss": 4.4386, "step": 9 }, { "epoch": 0.06230529595015576, "grad_norm": 21.0, "learning_rate": 1.3975155279503107e-06, "loss": 4.4903, "step": 10 }, { "epoch": 0.06853582554517133, "grad_norm": 20.875, "learning_rate": 1.5527950310559006e-06, "loss": 4.4575, "step": 11 }, { "epoch": 0.07476635514018691, "grad_norm": 21.125, "learning_rate": 1.7080745341614908e-06, "loss": 4.5027, "step": 12 }, { "epoch": 0.08099688473520249, "grad_norm": 21.0, "learning_rate": 1.863354037267081e-06, "loss": 4.4064, "step": 13 }, { "epoch": 0.08722741433021806, "grad_norm": 20.125, "learning_rate": 2.018633540372671e-06, "loss": 4.4408, "step": 14 }, { "epoch": 0.09345794392523364, "grad_norm": 21.375, "learning_rate": 2.173913043478261e-06, "loss": 4.4542, "step": 15 }, { "epoch": 0.09968847352024922, "grad_norm": 21.5, "learning_rate": 2.329192546583851e-06, "loss": 4.4183, "step": 16 }, { "epoch": 0.1059190031152648, "grad_norm": 20.625, "learning_rate": 2.484472049689441e-06, "loss": 4.2817, "step": 17 }, { "epoch": 0.11214953271028037, "grad_norm": 20.375, "learning_rate": 2.639751552795031e-06, "loss": 4.4335, "step": 18 }, { "epoch": 0.11838006230529595, "grad_norm": 20.875, "learning_rate": 2.7950310559006214e-06, "loss": 4.3935, "step": 19 }, { "epoch": 0.12461059190031153, "grad_norm": 20.5, "learning_rate": 2.9503105590062115e-06, "loss": 4.3969, "step": 20 }, { "epoch": 0.1308411214953271, "grad_norm": 20.25, "learning_rate": 3.1055900621118013e-06, "loss": 4.3285, "step": 21 }, { "epoch": 0.13707165109034267, "grad_norm": 20.375, "learning_rate": 3.2608695652173914e-06, "loss": 4.2989, "step": 22 }, { "epoch": 0.14330218068535824, "grad_norm": 20.875, "learning_rate": 3.4161490683229816e-06, "loss": 4.3591, "step": 23 }, { "epoch": 0.14953271028037382, "grad_norm": 20.625, "learning_rate": 3.5714285714285714e-06, "loss": 4.3529, "step": 24 }, { "epoch": 0.1557632398753894, "grad_norm": 20.375, "learning_rate": 3.726708074534162e-06, "loss": 4.3015, "step": 25 }, { "epoch": 0.16199376947040497, "grad_norm": 20.5, "learning_rate": 3.881987577639752e-06, "loss": 4.2567, "step": 26 }, { "epoch": 0.16822429906542055, "grad_norm": 20.625, "learning_rate": 4.037267080745342e-06, "loss": 4.3496, "step": 27 }, { "epoch": 0.17445482866043613, "grad_norm": 20.125, "learning_rate": 4.192546583850932e-06, "loss": 4.29, "step": 28 }, { "epoch": 0.1806853582554517, "grad_norm": 20.75, "learning_rate": 4.347826086956522e-06, "loss": 4.2808, "step": 29 }, { "epoch": 0.18691588785046728, "grad_norm": 19.125, "learning_rate": 4.503105590062112e-06, "loss": 4.1188, "step": 30 }, { "epoch": 0.19314641744548286, "grad_norm": 19.375, "learning_rate": 4.658385093167702e-06, "loss": 4.1089, "step": 31 }, { "epoch": 0.19937694704049844, "grad_norm": 19.125, "learning_rate": 4.813664596273292e-06, "loss": 4.0274, "step": 32 }, { "epoch": 0.205607476635514, "grad_norm": 19.375, "learning_rate": 4.968944099378882e-06, "loss": 4.0535, "step": 33 }, { "epoch": 0.2118380062305296, "grad_norm": 18.25, "learning_rate": 5.124223602484472e-06, "loss": 4.0217, "step": 34 }, { "epoch": 0.21806853582554517, "grad_norm": 19.125, "learning_rate": 5.279503105590062e-06, "loss": 4.124, "step": 35 }, { "epoch": 0.22429906542056074, "grad_norm": 19.25, "learning_rate": 5.4347826086956525e-06, "loss": 4.1087, "step": 36 }, { "epoch": 0.23052959501557632, "grad_norm": 18.5, "learning_rate": 5.590062111801243e-06, "loss": 3.9983, "step": 37 }, { "epoch": 0.2367601246105919, "grad_norm": 17.625, "learning_rate": 5.745341614906832e-06, "loss": 3.8408, "step": 38 }, { "epoch": 0.24299065420560748, "grad_norm": 17.625, "learning_rate": 5.900621118012423e-06, "loss": 3.8556, "step": 39 }, { "epoch": 0.24922118380062305, "grad_norm": 18.0, "learning_rate": 6.055900621118013e-06, "loss": 3.8305, "step": 40 }, { "epoch": 0.2554517133956386, "grad_norm": 17.125, "learning_rate": 6.2111801242236025e-06, "loss": 3.7877, "step": 41 }, { "epoch": 0.2616822429906542, "grad_norm": 17.125, "learning_rate": 6.366459627329193e-06, "loss": 3.8079, "step": 42 }, { "epoch": 0.26791277258566976, "grad_norm": 16.5, "learning_rate": 6.521739130434783e-06, "loss": 3.7283, "step": 43 }, { "epoch": 0.27414330218068533, "grad_norm": 15.5625, "learning_rate": 6.677018633540373e-06, "loss": 3.6548, "step": 44 }, { "epoch": 0.2803738317757009, "grad_norm": 16.5, "learning_rate": 6.832298136645963e-06, "loss": 3.7493, "step": 45 }, { "epoch": 0.2866043613707165, "grad_norm": 15.1875, "learning_rate": 6.9875776397515525e-06, "loss": 3.621, "step": 46 }, { "epoch": 0.29283489096573206, "grad_norm": 15.75, "learning_rate": 7.142857142857143e-06, "loss": 3.6482, "step": 47 }, { "epoch": 0.29906542056074764, "grad_norm": 16.125, "learning_rate": 7.298136645962733e-06, "loss": 3.534, "step": 48 }, { "epoch": 0.3052959501557632, "grad_norm": 15.4375, "learning_rate": 7.453416149068324e-06, "loss": 3.5081, "step": 49 }, { "epoch": 0.3115264797507788, "grad_norm": 14.6875, "learning_rate": 7.608695652173914e-06, "loss": 3.4946, "step": 50 }, { "epoch": 0.3177570093457944, "grad_norm": 13.875, "learning_rate": 7.763975155279503e-06, "loss": 3.328, "step": 51 }, { "epoch": 0.32398753894080995, "grad_norm": 13.9375, "learning_rate": 7.919254658385093e-06, "loss": 3.3846, "step": 52 }, { "epoch": 0.3302180685358255, "grad_norm": 12.875, "learning_rate": 8.074534161490684e-06, "loss": 3.2337, "step": 53 }, { "epoch": 0.3364485981308411, "grad_norm": 12.8125, "learning_rate": 8.229813664596275e-06, "loss": 3.1147, "step": 54 }, { "epoch": 0.3426791277258567, "grad_norm": 12.0625, "learning_rate": 8.385093167701864e-06, "loss": 3.1117, "step": 55 }, { "epoch": 0.34890965732087226, "grad_norm": 11.4375, "learning_rate": 8.540372670807453e-06, "loss": 3.0921, "step": 56 }, { "epoch": 0.35514018691588783, "grad_norm": 11.625, "learning_rate": 8.695652173913044e-06, "loss": 3.038, "step": 57 }, { "epoch": 0.3613707165109034, "grad_norm": 10.5625, "learning_rate": 8.850931677018634e-06, "loss": 2.9558, "step": 58 }, { "epoch": 0.367601246105919, "grad_norm": 8.9375, "learning_rate": 9.006211180124225e-06, "loss": 2.9024, "step": 59 }, { "epoch": 0.37383177570093457, "grad_norm": 9.0, "learning_rate": 9.161490683229814e-06, "loss": 2.8925, "step": 60 }, { "epoch": 0.38006230529595014, "grad_norm": 8.125, "learning_rate": 9.316770186335403e-06, "loss": 2.8254, "step": 61 }, { "epoch": 0.3862928348909657, "grad_norm": 7.65625, "learning_rate": 9.472049689440994e-06, "loss": 2.7155, "step": 62 }, { "epoch": 0.3925233644859813, "grad_norm": 6.75, "learning_rate": 9.627329192546584e-06, "loss": 2.71, "step": 63 }, { "epoch": 0.3987538940809969, "grad_norm": 6.125, "learning_rate": 9.782608695652175e-06, "loss": 2.6395, "step": 64 }, { "epoch": 0.40498442367601245, "grad_norm": 6.59375, "learning_rate": 9.937888198757764e-06, "loss": 2.4913, "step": 65 }, { "epoch": 0.411214953271028, "grad_norm": 5.34375, "learning_rate": 1.0093167701863353e-05, "loss": 2.6048, "step": 66 }, { "epoch": 0.4174454828660436, "grad_norm": 5.1875, "learning_rate": 1.0248447204968944e-05, "loss": 2.6173, "step": 67 }, { "epoch": 0.4236760124610592, "grad_norm": 5.25, "learning_rate": 1.0403726708074535e-05, "loss": 2.599, "step": 68 }, { "epoch": 0.42990654205607476, "grad_norm": 4.25, "learning_rate": 1.0559006211180125e-05, "loss": 2.4109, "step": 69 }, { "epoch": 0.43613707165109034, "grad_norm": 4.0625, "learning_rate": 1.0714285714285714e-05, "loss": 2.4219, "step": 70 }, { "epoch": 0.4423676012461059, "grad_norm": 4.09375, "learning_rate": 1.0869565217391305e-05, "loss": 2.5139, "step": 71 }, { "epoch": 0.4485981308411215, "grad_norm": 3.65625, "learning_rate": 1.1024844720496894e-05, "loss": 2.3694, "step": 72 }, { "epoch": 0.45482866043613707, "grad_norm": 3.6875, "learning_rate": 1.1180124223602485e-05, "loss": 2.4366, "step": 73 }, { "epoch": 0.46105919003115264, "grad_norm": 3.421875, "learning_rate": 1.1335403726708076e-05, "loss": 2.3894, "step": 74 }, { "epoch": 0.4672897196261682, "grad_norm": 3.484375, "learning_rate": 1.1490683229813664e-05, "loss": 2.3738, "step": 75 }, { "epoch": 0.4735202492211838, "grad_norm": 3.21875, "learning_rate": 1.1645962732919255e-05, "loss": 2.3221, "step": 76 }, { "epoch": 0.4797507788161994, "grad_norm": 3.328125, "learning_rate": 1.1801242236024846e-05, "loss": 2.2973, "step": 77 }, { "epoch": 0.48598130841121495, "grad_norm": 3.5, "learning_rate": 1.1956521739130435e-05, "loss": 2.2083, "step": 78 }, { "epoch": 0.49221183800623053, "grad_norm": 3.265625, "learning_rate": 1.2111801242236026e-05, "loss": 2.3984, "step": 79 }, { "epoch": 0.4984423676012461, "grad_norm": 4.0625, "learning_rate": 1.2267080745341616e-05, "loss": 2.4198, "step": 80 }, { "epoch": 0.5046728971962616, "grad_norm": 4.1875, "learning_rate": 1.2422360248447205e-05, "loss": 2.4435, "step": 81 }, { "epoch": 0.5109034267912772, "grad_norm": 3.421875, "learning_rate": 1.2577639751552794e-05, "loss": 2.3175, "step": 82 }, { "epoch": 0.5171339563862928, "grad_norm": 3.65625, "learning_rate": 1.2732919254658385e-05, "loss": 2.415, "step": 83 }, { "epoch": 0.5233644859813084, "grad_norm": 3.53125, "learning_rate": 1.2888198757763975e-05, "loss": 2.4788, "step": 84 }, { "epoch": 0.5295950155763239, "grad_norm": 3.546875, "learning_rate": 1.3043478260869566e-05, "loss": 2.4395, "step": 85 }, { "epoch": 0.5358255451713395, "grad_norm": 3.171875, "learning_rate": 1.3198757763975155e-05, "loss": 2.4031, "step": 86 }, { "epoch": 0.5420560747663551, "grad_norm": 3.078125, "learning_rate": 1.3354037267080746e-05, "loss": 2.3965, "step": 87 }, { "epoch": 0.5482866043613707, "grad_norm": 4.1875, "learning_rate": 1.3509316770186337e-05, "loss": 2.2563, "step": 88 }, { "epoch": 0.5545171339563862, "grad_norm": 3.453125, "learning_rate": 1.3664596273291926e-05, "loss": 2.3873, "step": 89 }, { "epoch": 0.5607476635514018, "grad_norm": 3.484375, "learning_rate": 1.3819875776397517e-05, "loss": 2.4077, "step": 90 }, { "epoch": 0.5669781931464174, "grad_norm": 3.53125, "learning_rate": 1.3975155279503105e-05, "loss": 2.3796, "step": 91 }, { "epoch": 0.573208722741433, "grad_norm": 3.34375, "learning_rate": 1.4130434782608694e-05, "loss": 2.3295, "step": 92 }, { "epoch": 0.5794392523364486, "grad_norm": 2.984375, "learning_rate": 1.4285714285714285e-05, "loss": 2.3338, "step": 93 }, { "epoch": 0.5856697819314641, "grad_norm": 3.15625, "learning_rate": 1.4440993788819876e-05, "loss": 2.3082, "step": 94 }, { "epoch": 0.5919003115264797, "grad_norm": 3.328125, "learning_rate": 1.4596273291925466e-05, "loss": 2.3224, "step": 95 }, { "epoch": 0.5981308411214953, "grad_norm": 3.375, "learning_rate": 1.4751552795031057e-05, "loss": 2.3802, "step": 96 }, { "epoch": 0.6043613707165109, "grad_norm": 3.203125, "learning_rate": 1.4906832298136648e-05, "loss": 2.4522, "step": 97 }, { "epoch": 0.6105919003115264, "grad_norm": 3.453125, "learning_rate": 1.5062111801242237e-05, "loss": 2.3779, "step": 98 }, { "epoch": 0.616822429906542, "grad_norm": 2.84375, "learning_rate": 1.5217391304347828e-05, "loss": 2.3352, "step": 99 }, { "epoch": 0.6230529595015576, "grad_norm": 2.859375, "learning_rate": 1.537267080745342e-05, "loss": 2.4483, "step": 100 }, { "epoch": 0.6292834890965732, "grad_norm": 2.96875, "learning_rate": 1.5527950310559007e-05, "loss": 2.3519, "step": 101 }, { "epoch": 0.6355140186915887, "grad_norm": 4.0625, "learning_rate": 1.5683229813664594e-05, "loss": 2.2655, "step": 102 }, { "epoch": 0.6417445482866043, "grad_norm": 3.109375, "learning_rate": 1.5838509316770185e-05, "loss": 2.3304, "step": 103 }, { "epoch": 0.6479750778816199, "grad_norm": 3.140625, "learning_rate": 1.5993788819875776e-05, "loss": 2.3793, "step": 104 }, { "epoch": 0.6542056074766355, "grad_norm": 3.28125, "learning_rate": 1.6149068322981367e-05, "loss": 2.241, "step": 105 }, { "epoch": 0.660436137071651, "grad_norm": 2.96875, "learning_rate": 1.630434782608696e-05, "loss": 2.3329, "step": 106 }, { "epoch": 0.6666666666666666, "grad_norm": 3.09375, "learning_rate": 1.645962732919255e-05, "loss": 2.3001, "step": 107 }, { "epoch": 0.6728971962616822, "grad_norm": 2.875, "learning_rate": 1.6614906832298137e-05, "loss": 2.3532, "step": 108 }, { "epoch": 0.6791277258566978, "grad_norm": 3.21875, "learning_rate": 1.6770186335403728e-05, "loss": 2.3358, "step": 109 }, { "epoch": 0.6853582554517134, "grad_norm": 3.0625, "learning_rate": 1.6925465838509316e-05, "loss": 2.4115, "step": 110 }, { "epoch": 0.6915887850467289, "grad_norm": 2.90625, "learning_rate": 1.7080745341614907e-05, "loss": 2.2682, "step": 111 }, { "epoch": 0.6978193146417445, "grad_norm": 3.046875, "learning_rate": 1.7236024844720498e-05, "loss": 2.3385, "step": 112 }, { "epoch": 0.7040498442367601, "grad_norm": 3.890625, "learning_rate": 1.739130434782609e-05, "loss": 2.4165, "step": 113 }, { "epoch": 0.7102803738317757, "grad_norm": 3.203125, "learning_rate": 1.7546583850931676e-05, "loss": 2.3243, "step": 114 }, { "epoch": 0.7165109034267912, "grad_norm": 3.015625, "learning_rate": 1.7701863354037267e-05, "loss": 2.375, "step": 115 }, { "epoch": 0.7227414330218068, "grad_norm": 3.0, "learning_rate": 1.785714285714286e-05, "loss": 2.3254, "step": 116 }, { "epoch": 0.7289719626168224, "grad_norm": 3.328125, "learning_rate": 1.801242236024845e-05, "loss": 2.37, "step": 117 }, { "epoch": 0.735202492211838, "grad_norm": 3.46875, "learning_rate": 1.816770186335404e-05, "loss": 2.3857, "step": 118 }, { "epoch": 0.7414330218068536, "grad_norm": 3.34375, "learning_rate": 1.8322981366459628e-05, "loss": 2.3181, "step": 119 }, { "epoch": 0.7476635514018691, "grad_norm": 3.296875, "learning_rate": 1.8478260869565216e-05, "loss": 2.2694, "step": 120 }, { "epoch": 0.7538940809968847, "grad_norm": 3.046875, "learning_rate": 1.8633540372670807e-05, "loss": 2.3064, "step": 121 }, { "epoch": 0.7601246105919003, "grad_norm": 3.46875, "learning_rate": 1.8788819875776398e-05, "loss": 2.3837, "step": 122 }, { "epoch": 0.7663551401869159, "grad_norm": 3.203125, "learning_rate": 1.894409937888199e-05, "loss": 2.3447, "step": 123 }, { "epoch": 0.7725856697819314, "grad_norm": 2.828125, "learning_rate": 1.909937888198758e-05, "loss": 2.2986, "step": 124 }, { "epoch": 0.778816199376947, "grad_norm": 3.390625, "learning_rate": 1.9254658385093167e-05, "loss": 2.3389, "step": 125 }, { "epoch": 0.7850467289719626, "grad_norm": 2.875, "learning_rate": 1.940993788819876e-05, "loss": 2.3204, "step": 126 }, { "epoch": 0.7912772585669782, "grad_norm": 3.3125, "learning_rate": 1.956521739130435e-05, "loss": 2.279, "step": 127 }, { "epoch": 0.7975077881619937, "grad_norm": 2.84375, "learning_rate": 1.972049689440994e-05, "loss": 2.3519, "step": 128 }, { "epoch": 0.8037383177570093, "grad_norm": 3.375, "learning_rate": 1.9875776397515528e-05, "loss": 2.3865, "step": 129 }, { "epoch": 0.8099688473520249, "grad_norm": 3.140625, "learning_rate": 2.003105590062112e-05, "loss": 2.3323, "step": 130 }, { "epoch": 0.8161993769470405, "grad_norm": 3.453125, "learning_rate": 2.0186335403726707e-05, "loss": 2.2402, "step": 131 }, { "epoch": 0.822429906542056, "grad_norm": 3.140625, "learning_rate": 2.0341614906832298e-05, "loss": 2.3353, "step": 132 }, { "epoch": 0.8286604361370716, "grad_norm": 3.078125, "learning_rate": 2.049689440993789e-05, "loss": 2.2439, "step": 133 }, { "epoch": 0.8348909657320872, "grad_norm": 3.078125, "learning_rate": 2.065217391304348e-05, "loss": 2.3234, "step": 134 }, { "epoch": 0.8411214953271028, "grad_norm": 3.640625, "learning_rate": 2.080745341614907e-05, "loss": 2.4313, "step": 135 }, { "epoch": 0.8473520249221184, "grad_norm": 3.5, "learning_rate": 2.096273291925466e-05, "loss": 2.3773, "step": 136 }, { "epoch": 0.8535825545171339, "grad_norm": 2.96875, "learning_rate": 2.111801242236025e-05, "loss": 2.3734, "step": 137 }, { "epoch": 0.8598130841121495, "grad_norm": 3.3125, "learning_rate": 2.127329192546584e-05, "loss": 2.2644, "step": 138 }, { "epoch": 0.8660436137071651, "grad_norm": 2.75, "learning_rate": 2.1428571428571428e-05, "loss": 2.2843, "step": 139 }, { "epoch": 0.8722741433021807, "grad_norm": 2.578125, "learning_rate": 2.158385093167702e-05, "loss": 2.3232, "step": 140 }, { "epoch": 0.8785046728971962, "grad_norm": 3.203125, "learning_rate": 2.173913043478261e-05, "loss": 2.2752, "step": 141 }, { "epoch": 0.8847352024922118, "grad_norm": 2.765625, "learning_rate": 2.1894409937888198e-05, "loss": 2.3752, "step": 142 }, { "epoch": 0.8909657320872274, "grad_norm": 2.9375, "learning_rate": 2.204968944099379e-05, "loss": 2.3247, "step": 143 }, { "epoch": 0.897196261682243, "grad_norm": 2.765625, "learning_rate": 2.220496894409938e-05, "loss": 2.3056, "step": 144 }, { "epoch": 0.9034267912772586, "grad_norm": 2.953125, "learning_rate": 2.236024844720497e-05, "loss": 2.3318, "step": 145 }, { "epoch": 0.9096573208722741, "grad_norm": 3.265625, "learning_rate": 2.2515527950310562e-05, "loss": 2.2385, "step": 146 }, { "epoch": 0.9158878504672897, "grad_norm": 2.859375, "learning_rate": 2.2670807453416153e-05, "loss": 2.2781, "step": 147 }, { "epoch": 0.9221183800623053, "grad_norm": 3.203125, "learning_rate": 2.282608695652174e-05, "loss": 2.267, "step": 148 }, { "epoch": 0.9283489096573209, "grad_norm": 3.109375, "learning_rate": 2.2981366459627328e-05, "loss": 2.2643, "step": 149 }, { "epoch": 0.9345794392523364, "grad_norm": 2.984375, "learning_rate": 2.313664596273292e-05, "loss": 2.3215, "step": 150 }, { "epoch": 0.940809968847352, "grad_norm": 3.03125, "learning_rate": 2.329192546583851e-05, "loss": 2.3077, "step": 151 }, { "epoch": 0.9470404984423676, "grad_norm": 3.171875, "learning_rate": 2.34472049689441e-05, "loss": 2.3486, "step": 152 }, { "epoch": 0.9532710280373832, "grad_norm": 3.078125, "learning_rate": 2.3602484472049692e-05, "loss": 2.2695, "step": 153 }, { "epoch": 0.9595015576323987, "grad_norm": 2.734375, "learning_rate": 2.375776397515528e-05, "loss": 2.383, "step": 154 }, { "epoch": 0.9657320872274143, "grad_norm": 3.46875, "learning_rate": 2.391304347826087e-05, "loss": 2.155, "step": 155 }, { "epoch": 0.9719626168224299, "grad_norm": 3.125, "learning_rate": 2.4068322981366462e-05, "loss": 2.1467, "step": 156 }, { "epoch": 0.9781931464174455, "grad_norm": 3.1875, "learning_rate": 2.4223602484472053e-05, "loss": 2.2839, "step": 157 }, { "epoch": 0.9844236760124611, "grad_norm": 3.109375, "learning_rate": 2.437888198757764e-05, "loss": 2.3656, "step": 158 }, { "epoch": 0.9906542056074766, "grad_norm": 3.0625, "learning_rate": 2.453416149068323e-05, "loss": 2.1615, "step": 159 }, { "epoch": 0.9968847352024922, "grad_norm": 3.046875, "learning_rate": 2.468944099378882e-05, "loss": 2.1974, "step": 160 }, { "epoch": 1.0, "grad_norm": 2.5, "learning_rate": 2.484472049689441e-05, "loss": 1.1735, "step": 161 }, { "epoch": 1.0062305295950156, "grad_norm": 3.859375, "learning_rate": 2.5e-05, "loss": 2.379, "step": 162 }, { "epoch": 1.0124610591900312, "grad_norm": 2.953125, "learning_rate": 2.515527950310559e-05, "loss": 2.2372, "step": 163 }, { "epoch": 1.0186915887850467, "grad_norm": 2.765625, "learning_rate": 2.5310559006211183e-05, "loss": 2.3076, "step": 164 }, { "epoch": 1.0249221183800623, "grad_norm": 3.3125, "learning_rate": 2.546583850931677e-05, "loss": 2.2718, "step": 165 }, { "epoch": 1.0311526479750779, "grad_norm": 2.8125, "learning_rate": 2.5621118012422362e-05, "loss": 2.2971, "step": 166 }, { "epoch": 1.0373831775700935, "grad_norm": 3.515625, "learning_rate": 2.577639751552795e-05, "loss": 2.2479, "step": 167 }, { "epoch": 1.043613707165109, "grad_norm": 3.703125, "learning_rate": 2.5931677018633544e-05, "loss": 2.2062, "step": 168 }, { "epoch": 1.0498442367601246, "grad_norm": 2.953125, "learning_rate": 2.608695652173913e-05, "loss": 2.2874, "step": 169 }, { "epoch": 1.0560747663551402, "grad_norm": 2.953125, "learning_rate": 2.6242236024844723e-05, "loss": 2.2644, "step": 170 }, { "epoch": 1.0623052959501558, "grad_norm": 2.765625, "learning_rate": 2.639751552795031e-05, "loss": 2.2658, "step": 171 }, { "epoch": 1.0685358255451713, "grad_norm": 4.0, "learning_rate": 2.6552795031055898e-05, "loss": 2.1962, "step": 172 }, { "epoch": 1.074766355140187, "grad_norm": 3.09375, "learning_rate": 2.6708074534161492e-05, "loss": 2.1297, "step": 173 }, { "epoch": 1.0809968847352025, "grad_norm": 2.796875, "learning_rate": 2.686335403726708e-05, "loss": 2.2106, "step": 174 }, { "epoch": 1.087227414330218, "grad_norm": 5.21875, "learning_rate": 2.7018633540372674e-05, "loss": 2.2945, "step": 175 }, { "epoch": 1.0934579439252337, "grad_norm": 4.375, "learning_rate": 2.7173913043478262e-05, "loss": 2.2794, "step": 176 }, { "epoch": 1.0996884735202492, "grad_norm": 3.6875, "learning_rate": 2.7329192546583853e-05, "loss": 2.3223, "step": 177 }, { "epoch": 1.1059190031152648, "grad_norm": 3.234375, "learning_rate": 2.748447204968944e-05, "loss": 2.1913, "step": 178 }, { "epoch": 1.1121495327102804, "grad_norm": 3.359375, "learning_rate": 2.7639751552795035e-05, "loss": 2.2562, "step": 179 }, { "epoch": 1.118380062305296, "grad_norm": 3.34375, "learning_rate": 2.7795031055900623e-05, "loss": 2.2384, "step": 180 }, { "epoch": 1.1246105919003115, "grad_norm": 3.40625, "learning_rate": 2.795031055900621e-05, "loss": 2.1845, "step": 181 }, { "epoch": 1.1308411214953271, "grad_norm": 4.1875, "learning_rate": 2.8105590062111805e-05, "loss": 2.1903, "step": 182 }, { "epoch": 1.1370716510903427, "grad_norm": 3.046875, "learning_rate": 2.826086956521739e-05, "loss": 2.2417, "step": 183 }, { "epoch": 1.1433021806853583, "grad_norm": 3.484375, "learning_rate": 2.8416149068322983e-05, "loss": 2.2426, "step": 184 }, { "epoch": 1.1495327102803738, "grad_norm": 3.15625, "learning_rate": 2.857142857142857e-05, "loss": 2.1961, "step": 185 }, { "epoch": 1.1557632398753894, "grad_norm": 3.59375, "learning_rate": 2.8726708074534165e-05, "loss": 2.2203, "step": 186 }, { "epoch": 1.161993769470405, "grad_norm": 4.46875, "learning_rate": 2.8881987577639753e-05, "loss": 2.2275, "step": 187 }, { "epoch": 1.1682242990654206, "grad_norm": 3.453125, "learning_rate": 2.9037267080745344e-05, "loss": 2.201, "step": 188 }, { "epoch": 1.1744548286604362, "grad_norm": 3.328125, "learning_rate": 2.919254658385093e-05, "loss": 2.2425, "step": 189 }, { "epoch": 1.1806853582554517, "grad_norm": 3.53125, "learning_rate": 2.9347826086956526e-05, "loss": 2.1958, "step": 190 }, { "epoch": 1.1869158878504673, "grad_norm": 4.03125, "learning_rate": 2.9503105590062114e-05, "loss": 2.1724, "step": 191 }, { "epoch": 1.1931464174454829, "grad_norm": 3.1875, "learning_rate": 2.96583850931677e-05, "loss": 2.1394, "step": 192 }, { "epoch": 1.1993769470404985, "grad_norm": 2.859375, "learning_rate": 2.9813664596273296e-05, "loss": 2.2307, "step": 193 }, { "epoch": 1.205607476635514, "grad_norm": 2.8125, "learning_rate": 2.9968944099378883e-05, "loss": 2.2231, "step": 194 }, { "epoch": 1.2118380062305296, "grad_norm": 3.515625, "learning_rate": 3.0124223602484474e-05, "loss": 2.1515, "step": 195 }, { "epoch": 1.2180685358255452, "grad_norm": 3.078125, "learning_rate": 3.0279503105590062e-05, "loss": 2.1269, "step": 196 }, { "epoch": 1.2242990654205608, "grad_norm": 3.6875, "learning_rate": 3.0434782608695656e-05, "loss": 2.1719, "step": 197 }, { "epoch": 1.2305295950155763, "grad_norm": 3.046875, "learning_rate": 3.059006211180124e-05, "loss": 2.1608, "step": 198 }, { "epoch": 1.236760124610592, "grad_norm": 3.0625, "learning_rate": 3.074534161490684e-05, "loss": 2.162, "step": 199 }, { "epoch": 1.2429906542056075, "grad_norm": 3.3125, "learning_rate": 3.090062111801242e-05, "loss": 2.2085, "step": 200 }, { "epoch": 1.249221183800623, "grad_norm": 2.890625, "learning_rate": 3.1055900621118014e-05, "loss": 2.1744, "step": 201 }, { "epoch": 1.2554517133956387, "grad_norm": 3.359375, "learning_rate": 3.1211180124223605e-05, "loss": 2.1267, "step": 202 }, { "epoch": 1.2616822429906542, "grad_norm": 3.328125, "learning_rate": 3.136645962732919e-05, "loss": 2.1804, "step": 203 }, { "epoch": 1.2679127725856698, "grad_norm": 3.859375, "learning_rate": 3.152173913043479e-05, "loss": 2.1822, "step": 204 }, { "epoch": 1.2741433021806854, "grad_norm": 3.046875, "learning_rate": 3.167701863354037e-05, "loss": 2.1019, "step": 205 }, { "epoch": 1.280373831775701, "grad_norm": 3.125, "learning_rate": 3.183229813664597e-05, "loss": 2.1302, "step": 206 }, { "epoch": 1.2866043613707165, "grad_norm": 3.03125, "learning_rate": 3.198757763975155e-05, "loss": 2.1363, "step": 207 }, { "epoch": 1.2928348909657321, "grad_norm": 2.84375, "learning_rate": 3.2142857142857144e-05, "loss": 2.0888, "step": 208 }, { "epoch": 1.2990654205607477, "grad_norm": 2.84375, "learning_rate": 3.2298136645962735e-05, "loss": 2.1384, "step": 209 }, { "epoch": 1.3052959501557633, "grad_norm": 3.15625, "learning_rate": 3.245341614906832e-05, "loss": 2.1225, "step": 210 }, { "epoch": 1.3115264797507789, "grad_norm": 3.21875, "learning_rate": 3.260869565217392e-05, "loss": 2.1774, "step": 211 }, { "epoch": 1.3177570093457944, "grad_norm": 3.109375, "learning_rate": 3.27639751552795e-05, "loss": 2.1079, "step": 212 }, { "epoch": 1.32398753894081, "grad_norm": 3.046875, "learning_rate": 3.29192546583851e-05, "loss": 2.1944, "step": 213 }, { "epoch": 1.3302180685358256, "grad_norm": 3.09375, "learning_rate": 3.307453416149068e-05, "loss": 2.2024, "step": 214 }, { "epoch": 1.3364485981308412, "grad_norm": 2.84375, "learning_rate": 3.3229813664596274e-05, "loss": 2.1546, "step": 215 }, { "epoch": 1.3426791277258567, "grad_norm": 3.046875, "learning_rate": 3.3385093167701865e-05, "loss": 2.1006, "step": 216 }, { "epoch": 1.3489096573208723, "grad_norm": 2.859375, "learning_rate": 3.3540372670807456e-05, "loss": 2.1005, "step": 217 }, { "epoch": 1.355140186915888, "grad_norm": 3.4375, "learning_rate": 3.369565217391305e-05, "loss": 2.1224, "step": 218 }, { "epoch": 1.3613707165109035, "grad_norm": 2.984375, "learning_rate": 3.385093167701863e-05, "loss": 2.2004, "step": 219 }, { "epoch": 1.367601246105919, "grad_norm": 2.765625, "learning_rate": 3.400621118012422e-05, "loss": 2.1711, "step": 220 }, { "epoch": 1.3738317757009346, "grad_norm": 3.390625, "learning_rate": 3.4161490683229814e-05, "loss": 2.1242, "step": 221 }, { "epoch": 1.3800623052959502, "grad_norm": 3.546875, "learning_rate": 3.4316770186335405e-05, "loss": 2.1552, "step": 222 }, { "epoch": 1.3862928348909658, "grad_norm": 2.65625, "learning_rate": 3.4472049689440996e-05, "loss": 2.1365, "step": 223 }, { "epoch": 1.3925233644859814, "grad_norm": 3.96875, "learning_rate": 3.462732919254659e-05, "loss": 2.085, "step": 224 }, { "epoch": 1.398753894080997, "grad_norm": 3.203125, "learning_rate": 3.478260869565218e-05, "loss": 2.1696, "step": 225 }, { "epoch": 1.4049844236760125, "grad_norm": 3.375, "learning_rate": 3.493788819875777e-05, "loss": 2.1544, "step": 226 }, { "epoch": 1.411214953271028, "grad_norm": 3.25, "learning_rate": 3.509316770186335e-05, "loss": 2.161, "step": 227 }, { "epoch": 1.4174454828660437, "grad_norm": 3.25, "learning_rate": 3.524844720496895e-05, "loss": 2.2387, "step": 228 }, { "epoch": 1.4236760124610592, "grad_norm": 3.046875, "learning_rate": 3.5403726708074535e-05, "loss": 2.0604, "step": 229 }, { "epoch": 1.4299065420560748, "grad_norm": 3.09375, "learning_rate": 3.5559006211180126e-05, "loss": 2.0908, "step": 230 }, { "epoch": 1.4361370716510904, "grad_norm": 2.71875, "learning_rate": 3.571428571428572e-05, "loss": 2.0388, "step": 231 }, { "epoch": 1.442367601246106, "grad_norm": 3.078125, "learning_rate": 3.58695652173913e-05, "loss": 2.1781, "step": 232 }, { "epoch": 1.4485981308411215, "grad_norm": 3.484375, "learning_rate": 3.60248447204969e-05, "loss": 2.1808, "step": 233 }, { "epoch": 1.4548286604361371, "grad_norm": 3.34375, "learning_rate": 3.618012422360248e-05, "loss": 2.1055, "step": 234 }, { "epoch": 1.4610591900311527, "grad_norm": 3.5625, "learning_rate": 3.633540372670808e-05, "loss": 2.1293, "step": 235 }, { "epoch": 1.4672897196261683, "grad_norm": 2.84375, "learning_rate": 3.6490683229813665e-05, "loss": 2.1053, "step": 236 }, { "epoch": 1.4735202492211839, "grad_norm": 3.65625, "learning_rate": 3.6645962732919256e-05, "loss": 2.1123, "step": 237 }, { "epoch": 1.4797507788161994, "grad_norm": 4.28125, "learning_rate": 3.680124223602485e-05, "loss": 2.0667, "step": 238 }, { "epoch": 1.485981308411215, "grad_norm": 3.984375, "learning_rate": 3.695652173913043e-05, "loss": 2.0825, "step": 239 }, { "epoch": 1.4922118380062306, "grad_norm": 2.78125, "learning_rate": 3.711180124223603e-05, "loss": 2.0364, "step": 240 }, { "epoch": 1.4984423676012462, "grad_norm": 6.5, "learning_rate": 3.7267080745341614e-05, "loss": 2.1739, "step": 241 }, { "epoch": 1.5046728971962615, "grad_norm": 3.90625, "learning_rate": 3.742236024844721e-05, "loss": 2.0776, "step": 242 }, { "epoch": 1.5109034267912773, "grad_norm": 4.71875, "learning_rate": 3.7577639751552796e-05, "loss": 2.0737, "step": 243 }, { "epoch": 1.5171339563862927, "grad_norm": 4.15625, "learning_rate": 3.773291925465839e-05, "loss": 2.1228, "step": 244 }, { "epoch": 1.5233644859813085, "grad_norm": 4.78125, "learning_rate": 3.788819875776398e-05, "loss": 2.0697, "step": 245 }, { "epoch": 1.5295950155763238, "grad_norm": 4.84375, "learning_rate": 3.804347826086957e-05, "loss": 2.0748, "step": 246 }, { "epoch": 1.5358255451713396, "grad_norm": 3.125, "learning_rate": 3.819875776397516e-05, "loss": 2.046, "step": 247 }, { "epoch": 1.542056074766355, "grad_norm": 3.53125, "learning_rate": 3.8354037267080744e-05, "loss": 2.0953, "step": 248 }, { "epoch": 1.5482866043613708, "grad_norm": 3.9375, "learning_rate": 3.8509316770186335e-05, "loss": 2.074, "step": 249 }, { "epoch": 1.5545171339563861, "grad_norm": 2.609375, "learning_rate": 3.8664596273291926e-05, "loss": 2.1043, "step": 250 }, { "epoch": 1.560747663551402, "grad_norm": 3.09375, "learning_rate": 3.881987577639752e-05, "loss": 2.0944, "step": 251 }, { "epoch": 1.5669781931464173, "grad_norm": 3.046875, "learning_rate": 3.897515527950311e-05, "loss": 2.0783, "step": 252 }, { "epoch": 1.573208722741433, "grad_norm": 4.5, "learning_rate": 3.91304347826087e-05, "loss": 2.103, "step": 253 }, { "epoch": 1.5794392523364484, "grad_norm": 3.46875, "learning_rate": 3.928571428571429e-05, "loss": 2.0657, "step": 254 }, { "epoch": 1.5856697819314642, "grad_norm": 2.9375, "learning_rate": 3.944099378881988e-05, "loss": 2.0382, "step": 255 }, { "epoch": 1.5919003115264796, "grad_norm": 3.90625, "learning_rate": 3.9596273291925465e-05, "loss": 2.1199, "step": 256 }, { "epoch": 1.5981308411214954, "grad_norm": 3.5625, "learning_rate": 3.9751552795031056e-05, "loss": 2.0703, "step": 257 }, { "epoch": 1.6043613707165107, "grad_norm": 3.59375, "learning_rate": 3.990683229813665e-05, "loss": 2.0715, "step": 258 }, { "epoch": 1.6105919003115265, "grad_norm": 3.34375, "learning_rate": 4.006211180124224e-05, "loss": 2.0849, "step": 259 }, { "epoch": 1.616822429906542, "grad_norm": 2.765625, "learning_rate": 4.021739130434783e-05, "loss": 2.0223, "step": 260 }, { "epoch": 1.6230529595015577, "grad_norm": 4.9375, "learning_rate": 4.0372670807453414e-05, "loss": 2.0735, "step": 261 }, { "epoch": 1.629283489096573, "grad_norm": 3.796875, "learning_rate": 4.052795031055901e-05, "loss": 2.1129, "step": 262 }, { "epoch": 1.6355140186915889, "grad_norm": 2.890625, "learning_rate": 4.0683229813664596e-05, "loss": 2.0814, "step": 263 }, { "epoch": 1.6417445482866042, "grad_norm": 3.265625, "learning_rate": 4.0838509316770193e-05, "loss": 2.0845, "step": 264 }, { "epoch": 1.64797507788162, "grad_norm": 4.90625, "learning_rate": 4.099378881987578e-05, "loss": 2.071, "step": 265 }, { "epoch": 1.6542056074766354, "grad_norm": 5.09375, "learning_rate": 4.114906832298137e-05, "loss": 2.066, "step": 266 }, { "epoch": 1.6604361370716512, "grad_norm": 3.390625, "learning_rate": 4.130434782608696e-05, "loss": 2.1095, "step": 267 }, { "epoch": 1.6666666666666665, "grad_norm": 3.375, "learning_rate": 4.1459627329192544e-05, "loss": 2.1222, "step": 268 }, { "epoch": 1.6728971962616823, "grad_norm": 3.375, "learning_rate": 4.161490683229814e-05, "loss": 1.9753, "step": 269 }, { "epoch": 1.6791277258566977, "grad_norm": 2.859375, "learning_rate": 4.1770186335403726e-05, "loss": 1.9943, "step": 270 }, { "epoch": 1.6853582554517135, "grad_norm": 3.078125, "learning_rate": 4.192546583850932e-05, "loss": 2.0052, "step": 271 }, { "epoch": 1.6915887850467288, "grad_norm": 2.90625, "learning_rate": 4.208074534161491e-05, "loss": 2.0468, "step": 272 }, { "epoch": 1.6978193146417446, "grad_norm": 3.6875, "learning_rate": 4.22360248447205e-05, "loss": 2.0488, "step": 273 }, { "epoch": 1.70404984423676, "grad_norm": 3.390625, "learning_rate": 4.239130434782609e-05, "loss": 1.9979, "step": 274 }, { "epoch": 1.7102803738317758, "grad_norm": 5.125, "learning_rate": 4.254658385093168e-05, "loss": 2.082, "step": 275 }, { "epoch": 1.7165109034267911, "grad_norm": 4.6875, "learning_rate": 4.270186335403727e-05, "loss": 2.0598, "step": 276 }, { "epoch": 1.722741433021807, "grad_norm": 3.90625, "learning_rate": 4.2857142857142856e-05, "loss": 2.0573, "step": 277 }, { "epoch": 1.7289719626168223, "grad_norm": 3.109375, "learning_rate": 4.301242236024845e-05, "loss": 2.0267, "step": 278 }, { "epoch": 1.735202492211838, "grad_norm": 2.984375, "learning_rate": 4.316770186335404e-05, "loss": 2.1038, "step": 279 }, { "epoch": 1.7414330218068534, "grad_norm": 5.3125, "learning_rate": 4.332298136645963e-05, "loss": 2.0721, "step": 280 }, { "epoch": 1.7476635514018692, "grad_norm": 5.40625, "learning_rate": 4.347826086956522e-05, "loss": 2.0936, "step": 281 }, { "epoch": 1.7538940809968846, "grad_norm": 2.875, "learning_rate": 4.363354037267081e-05, "loss": 2.0286, "step": 282 }, { "epoch": 1.7601246105919004, "grad_norm": 4.125, "learning_rate": 4.3788819875776396e-05, "loss": 1.9836, "step": 283 }, { "epoch": 1.7663551401869158, "grad_norm": 5.96875, "learning_rate": 4.3944099378881993e-05, "loss": 2.0922, "step": 284 }, { "epoch": 1.7725856697819315, "grad_norm": 5.25, "learning_rate": 4.409937888198758e-05, "loss": 2.0774, "step": 285 }, { "epoch": 1.778816199376947, "grad_norm": 3.421875, "learning_rate": 4.425465838509317e-05, "loss": 1.9435, "step": 286 }, { "epoch": 1.7850467289719627, "grad_norm": 8.1875, "learning_rate": 4.440993788819876e-05, "loss": 2.0752, "step": 287 }, { "epoch": 1.791277258566978, "grad_norm": 7.75, "learning_rate": 4.456521739130435e-05, "loss": 2.0397, "step": 288 }, { "epoch": 1.7975077881619939, "grad_norm": 6.46875, "learning_rate": 4.472049689440994e-05, "loss": 1.9849, "step": 289 }, { "epoch": 1.8037383177570092, "grad_norm": 4.71875, "learning_rate": 4.4875776397515526e-05, "loss": 2.0359, "step": 290 }, { "epoch": 1.809968847352025, "grad_norm": 7.3125, "learning_rate": 4.5031055900621124e-05, "loss": 1.9925, "step": 291 }, { "epoch": 1.8161993769470404, "grad_norm": 7.59375, "learning_rate": 4.518633540372671e-05, "loss": 2.0804, "step": 292 }, { "epoch": 1.8224299065420562, "grad_norm": 7.59375, "learning_rate": 4.5341614906832306e-05, "loss": 2.0163, "step": 293 }, { "epoch": 1.8286604361370715, "grad_norm": 6.03125, "learning_rate": 4.549689440993789e-05, "loss": 2.0441, "step": 294 }, { "epoch": 1.8348909657320873, "grad_norm": 3.828125, "learning_rate": 4.565217391304348e-05, "loss": 2.0724, "step": 295 }, { "epoch": 1.8411214953271027, "grad_norm": 9.0, "learning_rate": 4.580745341614907e-05, "loss": 2.1115, "step": 296 }, { "epoch": 1.8473520249221185, "grad_norm": 11.625, "learning_rate": 4.5962732919254656e-05, "loss": 2.144, "step": 297 }, { "epoch": 1.8535825545171338, "grad_norm": 13.0, "learning_rate": 4.6118012422360254e-05, "loss": 2.0399, "step": 298 }, { "epoch": 1.8598130841121496, "grad_norm": 11.1875, "learning_rate": 4.627329192546584e-05, "loss": 2.0735, "step": 299 }, { "epoch": 1.866043613707165, "grad_norm": 3.25, "learning_rate": 4.642857142857143e-05, "loss": 2.0003, "step": 300 }, { "epoch": 1.8722741433021808, "grad_norm": 4.8125, "learning_rate": 4.658385093167702e-05, "loss": 1.9765, "step": 301 }, { "epoch": 1.8785046728971961, "grad_norm": 4.75, "learning_rate": 4.673913043478261e-05, "loss": 1.9863, "step": 302 }, { "epoch": 1.884735202492212, "grad_norm": 4.46875, "learning_rate": 4.68944099378882e-05, "loss": 1.9618, "step": 303 }, { "epoch": 1.8909657320872273, "grad_norm": 3.015625, "learning_rate": 4.7049689440993793e-05, "loss": 2.0481, "step": 304 }, { "epoch": 1.897196261682243, "grad_norm": 3.71875, "learning_rate": 4.7204968944099384e-05, "loss": 2.0122, "step": 305 }, { "epoch": 1.9034267912772584, "grad_norm": 7.375, "learning_rate": 4.736024844720497e-05, "loss": 2.0421, "step": 306 }, { "epoch": 1.9096573208722742, "grad_norm": 7.71875, "learning_rate": 4.751552795031056e-05, "loss": 2.0222, "step": 307 }, { "epoch": 1.9158878504672896, "grad_norm": 7.3125, "learning_rate": 4.767080745341615e-05, "loss": 1.9617, "step": 308 }, { "epoch": 1.9221183800623054, "grad_norm": 4.09375, "learning_rate": 4.782608695652174e-05, "loss": 2.0196, "step": 309 }, { "epoch": 1.9283489096573208, "grad_norm": 4.71875, "learning_rate": 4.798136645962733e-05, "loss": 2.0554, "step": 310 }, { "epoch": 1.9345794392523366, "grad_norm": 7.25, "learning_rate": 4.8136645962732924e-05, "loss": 1.9988, "step": 311 }, { "epoch": 1.940809968847352, "grad_norm": 6.34375, "learning_rate": 4.829192546583851e-05, "loss": 1.9776, "step": 312 }, { "epoch": 1.9470404984423677, "grad_norm": 4.96875, "learning_rate": 4.8447204968944106e-05, "loss": 1.9908, "step": 313 }, { "epoch": 1.953271028037383, "grad_norm": 3.234375, "learning_rate": 4.860248447204969e-05, "loss": 2.0315, "step": 314 }, { "epoch": 1.9595015576323989, "grad_norm": 4.84375, "learning_rate": 4.875776397515528e-05, "loss": 2.0451, "step": 315 }, { "epoch": 1.9657320872274142, "grad_norm": 7.09375, "learning_rate": 4.891304347826087e-05, "loss": 1.9959, "step": 316 }, { "epoch": 1.97196261682243, "grad_norm": 8.0, "learning_rate": 4.906832298136646e-05, "loss": 2.0517, "step": 317 }, { "epoch": 1.9781931464174454, "grad_norm": 3.78125, "learning_rate": 4.9223602484472054e-05, "loss": 1.9954, "step": 318 }, { "epoch": 1.9844236760124612, "grad_norm": 3.578125, "learning_rate": 4.937888198757764e-05, "loss": 1.9955, "step": 319 }, { "epoch": 1.9906542056074765, "grad_norm": 5.65625, "learning_rate": 4.9534161490683236e-05, "loss": 1.9768, "step": 320 }, { "epoch": 1.9968847352024923, "grad_norm": 5.0625, "learning_rate": 4.968944099378882e-05, "loss": 1.9917, "step": 321 }, { "epoch": 2.0, "grad_norm": 3.1875, "learning_rate": 4.984472049689442e-05, "loss": 0.977, "step": 322 }, { "epoch": 2.0062305295950154, "grad_norm": 3.3125, "learning_rate": 5e-05, "loss": 1.9575, "step": 323 }, { "epoch": 2.012461059190031, "grad_norm": 3.796875, "learning_rate": 4.99999998785975e-05, "loss": 2.0026, "step": 324 }, { "epoch": 2.0186915887850465, "grad_norm": 6.5625, "learning_rate": 4.999999951439e-05, "loss": 2.0183, "step": 325 }, { "epoch": 2.0249221183800623, "grad_norm": 4.65625, "learning_rate": 4.999999890737752e-05, "loss": 2.018, "step": 326 }, { "epoch": 2.0311526479750777, "grad_norm": 3.53125, "learning_rate": 4.9999998057560046e-05, "loss": 1.9771, "step": 327 }, { "epoch": 2.0373831775700935, "grad_norm": 5.4375, "learning_rate": 4.999999696493759e-05, "loss": 2.0226, "step": 328 }, { "epoch": 2.043613707165109, "grad_norm": 5.9375, "learning_rate": 4.999999562951018e-05, "loss": 2.0068, "step": 329 }, { "epoch": 2.0498442367601246, "grad_norm": 5.40625, "learning_rate": 4.99999940512778e-05, "loss": 1.9871, "step": 330 }, { "epoch": 2.05607476635514, "grad_norm": 3.640625, "learning_rate": 4.9999992230240476e-05, "loss": 1.9577, "step": 331 }, { "epoch": 2.0623052959501558, "grad_norm": 5.78125, "learning_rate": 4.999999016639825e-05, "loss": 1.9738, "step": 332 }, { "epoch": 2.068535825545171, "grad_norm": 6.15625, "learning_rate": 4.9999987859751106e-05, "loss": 1.9128, "step": 333 }, { "epoch": 2.074766355140187, "grad_norm": 3.625, "learning_rate": 4.999998531029908e-05, "loss": 1.9824, "step": 334 }, { "epoch": 2.0809968847352023, "grad_norm": 3.71875, "learning_rate": 4.999998251804221e-05, "loss": 1.9932, "step": 335 }, { "epoch": 2.087227414330218, "grad_norm": 4.09375, "learning_rate": 4.999997948298051e-05, "loss": 1.9354, "step": 336 }, { "epoch": 2.0934579439252334, "grad_norm": 5.1875, "learning_rate": 4.999997620511401e-05, "loss": 1.9062, "step": 337 }, { "epoch": 2.0996884735202492, "grad_norm": 3.453125, "learning_rate": 4.999997268444275e-05, "loss": 1.9358, "step": 338 }, { "epoch": 2.1059190031152646, "grad_norm": 4.65625, "learning_rate": 4.999996892096675e-05, "loss": 2.0227, "step": 339 }, { "epoch": 2.1121495327102804, "grad_norm": 4.125, "learning_rate": 4.999996491468606e-05, "loss": 1.884, "step": 340 }, { "epoch": 2.1183800623052957, "grad_norm": 3.859375, "learning_rate": 4.999996066560071e-05, "loss": 1.9747, "step": 341 }, { "epoch": 2.1246105919003115, "grad_norm": 4.375, "learning_rate": 4.999995617371074e-05, "loss": 1.9467, "step": 342 }, { "epoch": 2.130841121495327, "grad_norm": 3.53125, "learning_rate": 4.99999514390162e-05, "loss": 1.9679, "step": 343 }, { "epoch": 2.1370716510903427, "grad_norm": 3.71875, "learning_rate": 4.9999946461517144e-05, "loss": 1.924, "step": 344 }, { "epoch": 2.143302180685358, "grad_norm": 5.5625, "learning_rate": 4.99999412412136e-05, "loss": 1.9707, "step": 345 }, { "epoch": 2.149532710280374, "grad_norm": 4.28125, "learning_rate": 4.999993577810563e-05, "loss": 1.9218, "step": 346 }, { "epoch": 2.155763239875389, "grad_norm": 4.71875, "learning_rate": 4.999993007219329e-05, "loss": 1.9323, "step": 347 }, { "epoch": 2.161993769470405, "grad_norm": 4.03125, "learning_rate": 4.9999924123476636e-05, "loss": 1.9096, "step": 348 }, { "epoch": 2.1682242990654204, "grad_norm": 3.390625, "learning_rate": 4.999991793195572e-05, "loss": 1.8822, "step": 349 }, { "epoch": 2.174454828660436, "grad_norm": 3.421875, "learning_rate": 4.9999911497630595e-05, "loss": 1.9357, "step": 350 }, { "epoch": 2.1806853582554515, "grad_norm": 3.359375, "learning_rate": 4.999990482050134e-05, "loss": 1.9405, "step": 351 }, { "epoch": 2.1869158878504673, "grad_norm": 4.65625, "learning_rate": 4.9999897900568005e-05, "loss": 1.9718, "step": 352 }, { "epoch": 2.1931464174454827, "grad_norm": 3.484375, "learning_rate": 4.999989073783067e-05, "loss": 1.8096, "step": 353 }, { "epoch": 2.1993769470404985, "grad_norm": 4.6875, "learning_rate": 4.99998833322894e-05, "loss": 1.81, "step": 354 }, { "epoch": 2.205607476635514, "grad_norm": 3.984375, "learning_rate": 4.999987568394426e-05, "loss": 1.8839, "step": 355 }, { "epoch": 2.2118380062305296, "grad_norm": 4.03125, "learning_rate": 4.999986779279533e-05, "loss": 1.8987, "step": 356 }, { "epoch": 2.218068535825545, "grad_norm": 3.875, "learning_rate": 4.9999859658842694e-05, "loss": 1.9525, "step": 357 }, { "epoch": 2.2242990654205608, "grad_norm": 3.484375, "learning_rate": 4.999985128208642e-05, "loss": 1.8755, "step": 358 }, { "epoch": 2.230529595015576, "grad_norm": 4.03125, "learning_rate": 4.9999842662526596e-05, "loss": 1.8876, "step": 359 }, { "epoch": 2.236760124610592, "grad_norm": 4.375, "learning_rate": 4.99998338001633e-05, "loss": 1.8129, "step": 360 }, { "epoch": 2.2429906542056073, "grad_norm": 5.65625, "learning_rate": 4.999982469499662e-05, "loss": 1.8265, "step": 361 }, { "epoch": 2.249221183800623, "grad_norm": 3.953125, "learning_rate": 4.9999815347026644e-05, "loss": 1.8478, "step": 362 }, { "epoch": 2.2554517133956384, "grad_norm": 4.8125, "learning_rate": 4.999980575625346e-05, "loss": 1.8146, "step": 363 }, { "epoch": 2.2616822429906542, "grad_norm": 5.5, "learning_rate": 4.999979592267717e-05, "loss": 1.8744, "step": 364 }, { "epoch": 2.2679127725856696, "grad_norm": 4.34375, "learning_rate": 4.999978584629787e-05, "loss": 1.7636, "step": 365 }, { "epoch": 2.2741433021806854, "grad_norm": 4.375, "learning_rate": 4.9999775527115644e-05, "loss": 1.7741, "step": 366 }, { "epoch": 2.2803738317757007, "grad_norm": 5.5, "learning_rate": 4.99997649651306e-05, "loss": 1.6826, "step": 367 }, { "epoch": 2.2866043613707165, "grad_norm": 5.15625, "learning_rate": 4.999975416034285e-05, "loss": 1.8612, "step": 368 }, { "epoch": 2.292834890965732, "grad_norm": 4.3125, "learning_rate": 4.9999743112752485e-05, "loss": 1.7158, "step": 369 }, { "epoch": 2.2990654205607477, "grad_norm": 6.25, "learning_rate": 4.9999731822359616e-05, "loss": 1.7513, "step": 370 }, { "epoch": 2.305295950155763, "grad_norm": 4.65625, "learning_rate": 4.999972028916436e-05, "loss": 1.7027, "step": 371 }, { "epoch": 2.311526479750779, "grad_norm": 4.375, "learning_rate": 4.9999708513166824e-05, "loss": 1.7601, "step": 372 }, { "epoch": 2.317757009345794, "grad_norm": 4.8125, "learning_rate": 4.999969649436711e-05, "loss": 1.8029, "step": 373 }, { "epoch": 2.32398753894081, "grad_norm": 5.40625, "learning_rate": 4.9999684232765357e-05, "loss": 1.7567, "step": 374 }, { "epoch": 2.3302180685358254, "grad_norm": 5.4375, "learning_rate": 4.999967172836167e-05, "loss": 1.7159, "step": 375 }, { "epoch": 2.336448598130841, "grad_norm": 4.8125, "learning_rate": 4.999965898115617e-05, "loss": 1.7067, "step": 376 }, { "epoch": 2.3426791277258565, "grad_norm": 4.625, "learning_rate": 4.9999645991148994e-05, "loss": 1.7078, "step": 377 }, { "epoch": 2.3489096573208723, "grad_norm": 5.125, "learning_rate": 4.999963275834025e-05, "loss": 1.7997, "step": 378 }, { "epoch": 2.3551401869158877, "grad_norm": 4.84375, "learning_rate": 4.9999619282730084e-05, "loss": 1.6324, "step": 379 }, { "epoch": 2.3613707165109035, "grad_norm": 4.875, "learning_rate": 4.9999605564318606e-05, "loss": 1.746, "step": 380 }, { "epoch": 2.367601246105919, "grad_norm": 4.625, "learning_rate": 4.999959160310597e-05, "loss": 1.5321, "step": 381 }, { "epoch": 2.3738317757009346, "grad_norm": 5.15625, "learning_rate": 4.9999577399092304e-05, "loss": 1.5984, "step": 382 }, { "epoch": 2.38006230529595, "grad_norm": 5.90625, "learning_rate": 4.999956295227774e-05, "loss": 1.5551, "step": 383 }, { "epoch": 2.3862928348909658, "grad_norm": 5.34375, "learning_rate": 4.999954826266242e-05, "loss": 1.6489, "step": 384 }, { "epoch": 2.392523364485981, "grad_norm": 5.0, "learning_rate": 4.99995333302465e-05, "loss": 1.6522, "step": 385 }, { "epoch": 2.398753894080997, "grad_norm": 5.5, "learning_rate": 4.999951815503011e-05, "loss": 1.7643, "step": 386 }, { "epoch": 2.4049844236760123, "grad_norm": 5.875, "learning_rate": 4.99995027370134e-05, "loss": 1.6512, "step": 387 }, { "epoch": 2.411214953271028, "grad_norm": 5.5, "learning_rate": 4.999948707619653e-05, "loss": 1.6847, "step": 388 }, { "epoch": 2.4174454828660434, "grad_norm": 5.4375, "learning_rate": 4.999947117257964e-05, "loss": 1.5806, "step": 389 }, { "epoch": 2.4236760124610592, "grad_norm": 7.0, "learning_rate": 4.999945502616289e-05, "loss": 1.7021, "step": 390 }, { "epoch": 2.4299065420560746, "grad_norm": 7.90625, "learning_rate": 4.9999438636946427e-05, "loss": 1.6021, "step": 391 }, { "epoch": 2.4361370716510904, "grad_norm": 6.03125, "learning_rate": 4.999942200493043e-05, "loss": 1.5013, "step": 392 }, { "epoch": 2.4423676012461057, "grad_norm": 6.34375, "learning_rate": 4.999940513011504e-05, "loss": 1.6075, "step": 393 }, { "epoch": 2.4485981308411215, "grad_norm": 5.4375, "learning_rate": 4.999938801250044e-05, "loss": 1.4568, "step": 394 }, { "epoch": 2.454828660436137, "grad_norm": 5.15625, "learning_rate": 4.9999370652086775e-05, "loss": 1.5305, "step": 395 }, { "epoch": 2.4610591900311527, "grad_norm": 6.78125, "learning_rate": 4.9999353048874234e-05, "loss": 1.538, "step": 396 }, { "epoch": 2.467289719626168, "grad_norm": 5.96875, "learning_rate": 4.999933520286297e-05, "loss": 1.5722, "step": 397 }, { "epoch": 2.473520249221184, "grad_norm": 5.65625, "learning_rate": 4.999931711405318e-05, "loss": 1.5706, "step": 398 }, { "epoch": 2.479750778816199, "grad_norm": 6.03125, "learning_rate": 4.9999298782445007e-05, "loss": 1.4663, "step": 399 }, { "epoch": 2.485981308411215, "grad_norm": 6.15625, "learning_rate": 4.999928020803866e-05, "loss": 1.4853, "step": 400 }, { "epoch": 2.4922118380062304, "grad_norm": 5.9375, "learning_rate": 4.99992613908343e-05, "loss": 1.4933, "step": 401 }, { "epoch": 2.498442367601246, "grad_norm": 7.375, "learning_rate": 4.999924233083212e-05, "loss": 1.4772, "step": 402 }, { "epoch": 2.5046728971962615, "grad_norm": 5.46875, "learning_rate": 4.9999223028032295e-05, "loss": 1.4427, "step": 403 }, { "epoch": 2.5109034267912773, "grad_norm": 6.53125, "learning_rate": 4.9999203482435024e-05, "loss": 1.402, "step": 404 }, { "epoch": 2.5171339563862927, "grad_norm": 8.125, "learning_rate": 4.999918369404049e-05, "loss": 1.527, "step": 405 }, { "epoch": 2.5233644859813085, "grad_norm": 6.21875, "learning_rate": 4.999916366284889e-05, "loss": 1.5608, "step": 406 }, { "epoch": 2.529595015576324, "grad_norm": 7.3125, "learning_rate": 4.999914338886042e-05, "loss": 1.5399, "step": 407 }, { "epoch": 2.5358255451713396, "grad_norm": 5.59375, "learning_rate": 4.999912287207525e-05, "loss": 1.4036, "step": 408 }, { "epoch": 2.542056074766355, "grad_norm": 5.4375, "learning_rate": 4.9999102112493625e-05, "loss": 1.4973, "step": 409 }, { "epoch": 2.5482866043613708, "grad_norm": 5.875, "learning_rate": 4.999908111011571e-05, "loss": 1.402, "step": 410 }, { "epoch": 2.554517133956386, "grad_norm": 5.84375, "learning_rate": 4.9999059864941726e-05, "loss": 1.4288, "step": 411 }, { "epoch": 2.560747663551402, "grad_norm": 6.90625, "learning_rate": 4.9999038376971875e-05, "loss": 1.5201, "step": 412 }, { "epoch": 2.5669781931464173, "grad_norm": 6.3125, "learning_rate": 4.9999016646206366e-05, "loss": 1.4728, "step": 413 }, { "epoch": 2.573208722741433, "grad_norm": 5.75, "learning_rate": 4.9998994672645415e-05, "loss": 1.3927, "step": 414 }, { "epoch": 2.5794392523364484, "grad_norm": 5.96875, "learning_rate": 4.9998972456289226e-05, "loss": 1.4501, "step": 415 }, { "epoch": 2.5856697819314642, "grad_norm": 7.1875, "learning_rate": 4.9998949997138015e-05, "loss": 1.4107, "step": 416 }, { "epoch": 2.5919003115264796, "grad_norm": 5.875, "learning_rate": 4.999892729519201e-05, "loss": 1.4302, "step": 417 }, { "epoch": 2.5981308411214954, "grad_norm": 7.625, "learning_rate": 4.999890435045142e-05, "loss": 1.2293, "step": 418 }, { "epoch": 2.6043613707165107, "grad_norm": 7.28125, "learning_rate": 4.999888116291649e-05, "loss": 1.4372, "step": 419 }, { "epoch": 2.6105919003115265, "grad_norm": 5.625, "learning_rate": 4.9998857732587414e-05, "loss": 1.2784, "step": 420 }, { "epoch": 2.616822429906542, "grad_norm": 5.71875, "learning_rate": 4.9998834059464436e-05, "loss": 1.3441, "step": 421 }, { "epoch": 2.6230529595015577, "grad_norm": 5.625, "learning_rate": 4.999881014354779e-05, "loss": 1.4206, "step": 422 }, { "epoch": 2.629283489096573, "grad_norm": 6.84375, "learning_rate": 4.99987859848377e-05, "loss": 1.4206, "step": 423 }, { "epoch": 2.635514018691589, "grad_norm": 5.84375, "learning_rate": 4.99987615833344e-05, "loss": 1.2258, "step": 424 }, { "epoch": 2.641744548286604, "grad_norm": 5.46875, "learning_rate": 4.999873693903814e-05, "loss": 1.291, "step": 425 }, { "epoch": 2.64797507788162, "grad_norm": 6.21875, "learning_rate": 4.999871205194914e-05, "loss": 1.4762, "step": 426 }, { "epoch": 2.6542056074766354, "grad_norm": 7.375, "learning_rate": 4.999868692206765e-05, "loss": 1.3724, "step": 427 }, { "epoch": 2.660436137071651, "grad_norm": 6.25, "learning_rate": 4.9998661549393924e-05, "loss": 1.2008, "step": 428 }, { "epoch": 2.6666666666666665, "grad_norm": 6.0, "learning_rate": 4.9998635933928196e-05, "loss": 1.2335, "step": 429 }, { "epoch": 2.6728971962616823, "grad_norm": 6.21875, "learning_rate": 4.999861007567072e-05, "loss": 1.5353, "step": 430 }, { "epoch": 2.6791277258566977, "grad_norm": 6.53125, "learning_rate": 4.9998583974621736e-05, "loss": 1.2602, "step": 431 }, { "epoch": 2.6853582554517135, "grad_norm": 5.90625, "learning_rate": 4.9998557630781514e-05, "loss": 1.3727, "step": 432 }, { "epoch": 2.691588785046729, "grad_norm": 5.15625, "learning_rate": 4.999853104415031e-05, "loss": 1.1685, "step": 433 }, { "epoch": 2.6978193146417446, "grad_norm": 5.78125, "learning_rate": 4.999850421472837e-05, "loss": 1.3324, "step": 434 }, { "epoch": 2.70404984423676, "grad_norm": 5.40625, "learning_rate": 4.999847714251596e-05, "loss": 1.2054, "step": 435 }, { "epoch": 2.710280373831776, "grad_norm": 5.65625, "learning_rate": 4.9998449827513344e-05, "loss": 1.3234, "step": 436 }, { "epoch": 2.716510903426791, "grad_norm": 7.125, "learning_rate": 4.999842226972078e-05, "loss": 1.3857, "step": 437 }, { "epoch": 2.722741433021807, "grad_norm": 5.34375, "learning_rate": 4.999839446913855e-05, "loss": 1.1641, "step": 438 }, { "epoch": 2.7289719626168223, "grad_norm": 6.25, "learning_rate": 4.999836642576692e-05, "loss": 1.3435, "step": 439 }, { "epoch": 2.735202492211838, "grad_norm": 6.3125, "learning_rate": 4.999833813960615e-05, "loss": 1.2676, "step": 440 }, { "epoch": 2.7414330218068534, "grad_norm": 6.5625, "learning_rate": 4.999830961065652e-05, "loss": 1.3203, "step": 441 }, { "epoch": 2.7476635514018692, "grad_norm": 6.28125, "learning_rate": 4.999828083891831e-05, "loss": 1.2945, "step": 442 }, { "epoch": 2.7538940809968846, "grad_norm": 5.71875, "learning_rate": 4.999825182439181e-05, "loss": 1.2872, "step": 443 }, { "epoch": 2.7601246105919004, "grad_norm": 6.78125, "learning_rate": 4.999822256707728e-05, "loss": 1.1761, "step": 444 }, { "epoch": 2.7663551401869158, "grad_norm": 5.28125, "learning_rate": 4.9998193066975016e-05, "loss": 1.174, "step": 445 }, { "epoch": 2.7725856697819315, "grad_norm": 5.9375, "learning_rate": 4.999816332408531e-05, "loss": 1.207, "step": 446 }, { "epoch": 2.778816199376947, "grad_norm": 5.5625, "learning_rate": 4.999813333840845e-05, "loss": 1.1859, "step": 447 }, { "epoch": 2.7850467289719627, "grad_norm": 6.21875, "learning_rate": 4.9998103109944714e-05, "loss": 1.3363, "step": 448 }, { "epoch": 2.791277258566978, "grad_norm": 5.96875, "learning_rate": 4.99980726386944e-05, "loss": 1.125, "step": 449 }, { "epoch": 2.797507788161994, "grad_norm": 5.53125, "learning_rate": 4.9998041924657815e-05, "loss": 1.2218, "step": 450 }, { "epoch": 2.803738317757009, "grad_norm": 6.3125, "learning_rate": 4.999801096783524e-05, "loss": 1.3417, "step": 451 }, { "epoch": 2.809968847352025, "grad_norm": 5.0, "learning_rate": 4.9997979768227e-05, "loss": 1.0431, "step": 452 }, { "epoch": 2.8161993769470404, "grad_norm": 6.1875, "learning_rate": 4.999794832583337e-05, "loss": 1.0923, "step": 453 }, { "epoch": 2.822429906542056, "grad_norm": 5.34375, "learning_rate": 4.999791664065467e-05, "loss": 1.3176, "step": 454 }, { "epoch": 2.8286604361370715, "grad_norm": 5.5, "learning_rate": 4.999788471269122e-05, "loss": 1.2173, "step": 455 }, { "epoch": 2.8348909657320873, "grad_norm": 5.3125, "learning_rate": 4.99978525419433e-05, "loss": 1.1774, "step": 456 }, { "epoch": 2.8411214953271027, "grad_norm": 6.125, "learning_rate": 4.999782012841126e-05, "loss": 1.062, "step": 457 }, { "epoch": 2.8473520249221185, "grad_norm": 5.875, "learning_rate": 4.999778747209537e-05, "loss": 1.0994, "step": 458 }, { "epoch": 2.853582554517134, "grad_norm": 6.125, "learning_rate": 4.999775457299598e-05, "loss": 1.2723, "step": 459 }, { "epoch": 2.8598130841121496, "grad_norm": 5.9375, "learning_rate": 4.999772143111341e-05, "loss": 1.2173, "step": 460 }, { "epoch": 2.866043613707165, "grad_norm": 5.96875, "learning_rate": 4.999768804644796e-05, "loss": 1.1241, "step": 461 }, { "epoch": 2.872274143302181, "grad_norm": 5.34375, "learning_rate": 4.999765441899997e-05, "loss": 1.1008, "step": 462 }, { "epoch": 2.878504672897196, "grad_norm": 5.46875, "learning_rate": 4.999762054876977e-05, "loss": 0.985, "step": 463 }, { "epoch": 2.884735202492212, "grad_norm": 6.0625, "learning_rate": 4.999758643575768e-05, "loss": 1.1343, "step": 464 }, { "epoch": 2.8909657320872273, "grad_norm": 6.375, "learning_rate": 4.999755207996403e-05, "loss": 1.2552, "step": 465 }, { "epoch": 2.897196261682243, "grad_norm": 6.0625, "learning_rate": 4.9997517481389156e-05, "loss": 1.1362, "step": 466 }, { "epoch": 2.9034267912772584, "grad_norm": 5.34375, "learning_rate": 4.99974826400334e-05, "loss": 0.9222, "step": 467 }, { "epoch": 2.9096573208722742, "grad_norm": 5.25, "learning_rate": 4.999744755589709e-05, "loss": 1.3016, "step": 468 }, { "epoch": 2.9158878504672896, "grad_norm": 5.25, "learning_rate": 4.999741222898058e-05, "loss": 1.0441, "step": 469 }, { "epoch": 2.9221183800623054, "grad_norm": 5.8125, "learning_rate": 4.99973766592842e-05, "loss": 1.0088, "step": 470 }, { "epoch": 2.9283489096573208, "grad_norm": 6.0625, "learning_rate": 4.99973408468083e-05, "loss": 1.1416, "step": 471 }, { "epoch": 2.9345794392523366, "grad_norm": 6.6875, "learning_rate": 4.999730479155323e-05, "loss": 0.9618, "step": 472 }, { "epoch": 2.940809968847352, "grad_norm": 6.28125, "learning_rate": 4.9997268493519345e-05, "loss": 1.0906, "step": 473 }, { "epoch": 2.9470404984423677, "grad_norm": 7.15625, "learning_rate": 4.999723195270698e-05, "loss": 1.0354, "step": 474 }, { "epoch": 2.953271028037383, "grad_norm": 6.40625, "learning_rate": 4.9997195169116514e-05, "loss": 1.047, "step": 475 }, { "epoch": 2.959501557632399, "grad_norm": 6.78125, "learning_rate": 4.999715814274828e-05, "loss": 1.0395, "step": 476 }, { "epoch": 2.965732087227414, "grad_norm": 5.25, "learning_rate": 4.9997120873602654e-05, "loss": 1.193, "step": 477 }, { "epoch": 2.97196261682243, "grad_norm": 5.28125, "learning_rate": 4.9997083361679995e-05, "loss": 1.1083, "step": 478 }, { "epoch": 2.9781931464174454, "grad_norm": 6.03125, "learning_rate": 4.999704560698067e-05, "loss": 0.9952, "step": 479 }, { "epoch": 2.984423676012461, "grad_norm": 5.125, "learning_rate": 4.999700760950503e-05, "loss": 0.9787, "step": 480 }, { "epoch": 2.9906542056074765, "grad_norm": 5.71875, "learning_rate": 4.999696936925347e-05, "loss": 1.0184, "step": 481 }, { "epoch": 2.9968847352024923, "grad_norm": 6.9375, "learning_rate": 4.999693088622633e-05, "loss": 1.0158, "step": 482 }, { "epoch": 3.0, "grad_norm": 3.53125, "learning_rate": 4.9996892160424016e-05, "loss": 0.5845, "step": 483 }, { "epoch": 3.0062305295950154, "grad_norm": 7.0, "learning_rate": 4.9996853191846885e-05, "loss": 1.3792, "step": 484 }, { "epoch": 3.012461059190031, "grad_norm": 6.65625, "learning_rate": 4.9996813980495305e-05, "loss": 1.2228, "step": 485 }, { "epoch": 3.0186915887850465, "grad_norm": 5.375, "learning_rate": 4.999677452636968e-05, "loss": 1.0208, "step": 486 }, { "epoch": 3.0249221183800623, "grad_norm": 5.53125, "learning_rate": 4.999673482947039e-05, "loss": 0.9694, "step": 487 }, { "epoch": 3.0311526479750777, "grad_norm": 6.84375, "learning_rate": 4.999669488979781e-05, "loss": 0.9819, "step": 488 }, { "epoch": 3.0373831775700935, "grad_norm": 6.46875, "learning_rate": 4.999665470735233e-05, "loss": 0.9272, "step": 489 }, { "epoch": 3.043613707165109, "grad_norm": 5.5625, "learning_rate": 4.999661428213435e-05, "loss": 0.8949, "step": 490 }, { "epoch": 3.0498442367601246, "grad_norm": 7.25, "learning_rate": 4.999657361414425e-05, "loss": 1.0846, "step": 491 }, { "epoch": 3.05607476635514, "grad_norm": 7.59375, "learning_rate": 4.999653270338242e-05, "loss": 0.9503, "step": 492 }, { "epoch": 3.0623052959501558, "grad_norm": 5.625, "learning_rate": 4.9996491549849286e-05, "loss": 0.9352, "step": 493 }, { "epoch": 3.068535825545171, "grad_norm": 8.25, "learning_rate": 4.999645015354522e-05, "loss": 0.9721, "step": 494 }, { "epoch": 3.074766355140187, "grad_norm": 9.875, "learning_rate": 4.999640851447064e-05, "loss": 0.7818, "step": 495 }, { "epoch": 3.0809968847352023, "grad_norm": 5.46875, "learning_rate": 4.999636663262594e-05, "loss": 0.8147, "step": 496 }, { "epoch": 3.087227414330218, "grad_norm": 6.03125, "learning_rate": 4.999632450801153e-05, "loss": 0.9981, "step": 497 }, { "epoch": 3.0934579439252334, "grad_norm": 7.03125, "learning_rate": 4.999628214062782e-05, "loss": 0.9878, "step": 498 }, { "epoch": 3.0996884735202492, "grad_norm": 6.5, "learning_rate": 4.999623953047522e-05, "loss": 0.9994, "step": 499 }, { "epoch": 3.1059190031152646, "grad_norm": 5.03125, "learning_rate": 4.999619667755414e-05, "loss": 0.7525, "step": 500 }, { "epoch": 3.1121495327102804, "grad_norm": 5.53125, "learning_rate": 4.9996153581865006e-05, "loss": 0.9349, "step": 501 }, { "epoch": 3.1183800623052957, "grad_norm": 7.125, "learning_rate": 4.9996110243408234e-05, "loss": 0.9448, "step": 502 }, { "epoch": 3.1246105919003115, "grad_norm": 6.0, "learning_rate": 4.999606666218424e-05, "loss": 0.7907, "step": 503 }, { "epoch": 3.130841121495327, "grad_norm": 8.8125, "learning_rate": 4.9996022838193454e-05, "loss": 0.886, "step": 504 }, { "epoch": 3.1370716510903427, "grad_norm": 8.5625, "learning_rate": 4.9995978771436294e-05, "loss": 0.9978, "step": 505 }, { "epoch": 3.143302180685358, "grad_norm": 5.8125, "learning_rate": 4.999593446191319e-05, "loss": 0.9223, "step": 506 }, { "epoch": 3.149532710280374, "grad_norm": 4.8125, "learning_rate": 4.999588990962458e-05, "loss": 0.78, "step": 507 }, { "epoch": 3.155763239875389, "grad_norm": 11.0625, "learning_rate": 4.9995845114570886e-05, "loss": 1.0236, "step": 508 }, { "epoch": 3.161993769470405, "grad_norm": 11.5625, "learning_rate": 4.999580007675255e-05, "loss": 1.0602, "step": 509 }, { "epoch": 3.1682242990654204, "grad_norm": 6.78125, "learning_rate": 4.999575479617e-05, "loss": 0.8545, "step": 510 }, { "epoch": 3.174454828660436, "grad_norm": 7.28125, "learning_rate": 4.9995709272823686e-05, "loss": 1.1637, "step": 511 }, { "epoch": 3.1806853582554515, "grad_norm": 8.3125, "learning_rate": 4.9995663506714054e-05, "loss": 0.8286, "step": 512 }, { "epoch": 3.1869158878504673, "grad_norm": 8.25, "learning_rate": 4.999561749784154e-05, "loss": 0.9416, "step": 513 }, { "epoch": 3.1931464174454827, "grad_norm": 5.34375, "learning_rate": 4.9995571246206586e-05, "loss": 0.6763, "step": 514 }, { "epoch": 3.1993769470404985, "grad_norm": 8.3125, "learning_rate": 4.999552475180965e-05, "loss": 0.8274, "step": 515 }, { "epoch": 3.205607476635514, "grad_norm": 12.0625, "learning_rate": 4.999547801465119e-05, "loss": 0.9876, "step": 516 }, { "epoch": 3.2118380062305296, "grad_norm": 8.875, "learning_rate": 4.999543103473164e-05, "loss": 0.8322, "step": 517 }, { "epoch": 3.218068535825545, "grad_norm": 4.65625, "learning_rate": 4.999538381205147e-05, "loss": 0.8505, "step": 518 }, { "epoch": 3.2242990654205608, "grad_norm": 6.34375, "learning_rate": 4.999533634661113e-05, "loss": 0.9754, "step": 519 }, { "epoch": 3.230529595015576, "grad_norm": 7.1875, "learning_rate": 4.99952886384111e-05, "loss": 0.989, "step": 520 }, { "epoch": 3.236760124610592, "grad_norm": 5.3125, "learning_rate": 4.999524068745182e-05, "loss": 0.66, "step": 521 }, { "epoch": 3.2429906542056073, "grad_norm": 5.34375, "learning_rate": 4.999519249373377e-05, "loss": 0.91, "step": 522 }, { "epoch": 3.249221183800623, "grad_norm": 8.25, "learning_rate": 4.9995144057257413e-05, "loss": 0.9608, "step": 523 }, { "epoch": 3.2554517133956384, "grad_norm": 7.0, "learning_rate": 4.999509537802322e-05, "loss": 0.7531, "step": 524 }, { "epoch": 3.2616822429906542, "grad_norm": 4.65625, "learning_rate": 4.999504645603167e-05, "loss": 0.7666, "step": 525 }, { "epoch": 3.2679127725856696, "grad_norm": 6.09375, "learning_rate": 4.999499729128322e-05, "loss": 0.7864, "step": 526 }, { "epoch": 3.2741433021806854, "grad_norm": 6.28125, "learning_rate": 4.9994947883778375e-05, "loss": 0.8694, "step": 527 }, { "epoch": 3.2803738317757007, "grad_norm": 6.96875, "learning_rate": 4.999489823351759e-05, "loss": 1.0523, "step": 528 }, { "epoch": 3.2866043613707165, "grad_norm": 5.875, "learning_rate": 4.999484834050136e-05, "loss": 0.9617, "step": 529 }, { "epoch": 3.292834890965732, "grad_norm": 9.75, "learning_rate": 4.9994798204730166e-05, "loss": 0.9407, "step": 530 }, { "epoch": 3.2990654205607477, "grad_norm": 10.75, "learning_rate": 4.999474782620449e-05, "loss": 1.0201, "step": 531 }, { "epoch": 3.305295950155763, "grad_norm": 7.21875, "learning_rate": 4.999469720492483e-05, "loss": 0.8005, "step": 532 }, { "epoch": 3.311526479750779, "grad_norm": 5.9375, "learning_rate": 4.999464634089168e-05, "loss": 1.0173, "step": 533 }, { "epoch": 3.317757009345794, "grad_norm": 7.3125, "learning_rate": 4.999459523410553e-05, "loss": 1.0147, "step": 534 }, { "epoch": 3.32398753894081, "grad_norm": 9.125, "learning_rate": 4.999454388456687e-05, "loss": 0.9798, "step": 535 }, { "epoch": 3.3302180685358254, "grad_norm": 6.40625, "learning_rate": 4.99944922922762e-05, "loss": 0.7245, "step": 536 }, { "epoch": 3.336448598130841, "grad_norm": 5.96875, "learning_rate": 4.999444045723403e-05, "loss": 1.0686, "step": 537 }, { "epoch": 3.3426791277258565, "grad_norm": 4.625, "learning_rate": 4.9994388379440854e-05, "loss": 0.7834, "step": 538 }, { "epoch": 3.3489096573208723, "grad_norm": 5.5, "learning_rate": 4.9994336058897184e-05, "loss": 0.9634, "step": 539 }, { "epoch": 3.3551401869158877, "grad_norm": 6.0625, "learning_rate": 4.999428349560352e-05, "loss": 0.8822, "step": 540 }, { "epoch": 3.3613707165109035, "grad_norm": 5.46875, "learning_rate": 4.9994230689560387e-05, "loss": 0.9426, "step": 541 }, { "epoch": 3.367601246105919, "grad_norm": 6.375, "learning_rate": 4.999417764076829e-05, "loss": 1.0417, "step": 542 }, { "epoch": 3.3738317757009346, "grad_norm": 5.25, "learning_rate": 4.9994124349227736e-05, "loss": 0.6999, "step": 543 }, { "epoch": 3.38006230529595, "grad_norm": 5.9375, "learning_rate": 4.9994070814939256e-05, "loss": 0.9116, "step": 544 }, { "epoch": 3.3862928348909658, "grad_norm": 6.3125, "learning_rate": 4.9994017037903365e-05, "loss": 1.1158, "step": 545 }, { "epoch": 3.392523364485981, "grad_norm": 4.5625, "learning_rate": 4.999396301812058e-05, "loss": 0.7059, "step": 546 }, { "epoch": 3.398753894080997, "grad_norm": 5.84375, "learning_rate": 4.999390875559143e-05, "loss": 0.747, "step": 547 }, { "epoch": 3.4049844236760123, "grad_norm": 5.53125, "learning_rate": 4.999385425031644e-05, "loss": 0.8145, "step": 548 }, { "epoch": 3.411214953271028, "grad_norm": 6.09375, "learning_rate": 4.999379950229615e-05, "loss": 0.7708, "step": 549 }, { "epoch": 3.4174454828660434, "grad_norm": 6.21875, "learning_rate": 4.999374451153107e-05, "loss": 0.8079, "step": 550 }, { "epoch": 3.4236760124610592, "grad_norm": 5.3125, "learning_rate": 4.999368927802176e-05, "loss": 0.7345, "step": 551 }, { "epoch": 3.4299065420560746, "grad_norm": 6.28125, "learning_rate": 4.999363380176874e-05, "loss": 0.9347, "step": 552 }, { "epoch": 3.4361370716510904, "grad_norm": 5.34375, "learning_rate": 4.999357808277255e-05, "loss": 0.9109, "step": 553 }, { "epoch": 3.4423676012461057, "grad_norm": 5.25, "learning_rate": 4.999352212103373e-05, "loss": 0.781, "step": 554 }, { "epoch": 3.4485981308411215, "grad_norm": 6.34375, "learning_rate": 4.999346591655284e-05, "loss": 1.1715, "step": 555 }, { "epoch": 3.454828660436137, "grad_norm": 5.96875, "learning_rate": 4.99934094693304e-05, "loss": 0.9869, "step": 556 }, { "epoch": 3.4610591900311527, "grad_norm": 4.96875, "learning_rate": 4.999335277936698e-05, "loss": 0.8633, "step": 557 }, { "epoch": 3.467289719626168, "grad_norm": 5.03125, "learning_rate": 4.999329584666311e-05, "loss": 0.726, "step": 558 }, { "epoch": 3.473520249221184, "grad_norm": 6.71875, "learning_rate": 4.999323867121936e-05, "loss": 0.8241, "step": 559 }, { "epoch": 3.479750778816199, "grad_norm": 5.1875, "learning_rate": 4.999318125303628e-05, "loss": 0.7236, "step": 560 }, { "epoch": 3.485981308411215, "grad_norm": 4.15625, "learning_rate": 4.9993123592114435e-05, "loss": 0.6526, "step": 561 }, { "epoch": 3.4922118380062304, "grad_norm": 4.65625, "learning_rate": 4.9993065688454365e-05, "loss": 0.8017, "step": 562 }, { "epoch": 3.498442367601246, "grad_norm": 4.21875, "learning_rate": 4.999300754205666e-05, "loss": 0.6326, "step": 563 }, { "epoch": 3.5046728971962615, "grad_norm": 5.59375, "learning_rate": 4.999294915292186e-05, "loss": 0.9381, "step": 564 }, { "epoch": 3.5109034267912773, "grad_norm": 5.96875, "learning_rate": 4.999289052105054e-05, "loss": 0.8984, "step": 565 }, { "epoch": 3.5171339563862927, "grad_norm": 5.90625, "learning_rate": 4.999283164644328e-05, "loss": 0.9072, "step": 566 }, { "epoch": 3.5233644859813085, "grad_norm": 5.9375, "learning_rate": 4.999277252910063e-05, "loss": 0.7769, "step": 567 }, { "epoch": 3.529595015576324, "grad_norm": 5.9375, "learning_rate": 4.9992713169023184e-05, "loss": 0.8851, "step": 568 }, { "epoch": 3.5358255451713396, "grad_norm": 5.84375, "learning_rate": 4.9992653566211525e-05, "loss": 0.8517, "step": 569 }, { "epoch": 3.542056074766355, "grad_norm": 6.0, "learning_rate": 4.99925937206662e-05, "loss": 0.6443, "step": 570 }, { "epoch": 3.5482866043613708, "grad_norm": 7.875, "learning_rate": 4.9992533632387815e-05, "loss": 0.9067, "step": 571 }, { "epoch": 3.554517133956386, "grad_norm": 6.71875, "learning_rate": 4.999247330137695e-05, "loss": 1.0451, "step": 572 }, { "epoch": 3.560747663551402, "grad_norm": 5.34375, "learning_rate": 4.9992412727634186e-05, "loss": 0.8341, "step": 573 }, { "epoch": 3.5669781931464173, "grad_norm": 4.96875, "learning_rate": 4.999235191116011e-05, "loss": 0.8516, "step": 574 }, { "epoch": 3.573208722741433, "grad_norm": 6.21875, "learning_rate": 4.9992290851955325e-05, "loss": 0.8938, "step": 575 }, { "epoch": 3.5794392523364484, "grad_norm": 4.75, "learning_rate": 4.999222955002041e-05, "loss": 0.8985, "step": 576 }, { "epoch": 3.5856697819314642, "grad_norm": 6.125, "learning_rate": 4.9992168005355963e-05, "loss": 0.7256, "step": 577 }, { "epoch": 3.5919003115264796, "grad_norm": 6.71875, "learning_rate": 4.999210621796259e-05, "loss": 0.7756, "step": 578 }, { "epoch": 3.5981308411214954, "grad_norm": 5.09375, "learning_rate": 4.999204418784088e-05, "loss": 0.6746, "step": 579 }, { "epoch": 3.6043613707165107, "grad_norm": 6.1875, "learning_rate": 4.999198191499145e-05, "loss": 0.9187, "step": 580 }, { "epoch": 3.6105919003115265, "grad_norm": 6.09375, "learning_rate": 4.9991919399414887e-05, "loss": 0.7648, "step": 581 }, { "epoch": 3.616822429906542, "grad_norm": 4.75, "learning_rate": 4.999185664111181e-05, "loss": 0.6462, "step": 582 }, { "epoch": 3.6230529595015577, "grad_norm": 5.84375, "learning_rate": 4.999179364008283e-05, "loss": 0.6839, "step": 583 }, { "epoch": 3.629283489096573, "grad_norm": 5.0625, "learning_rate": 4.999173039632855e-05, "loss": 0.6885, "step": 584 }, { "epoch": 3.635514018691589, "grad_norm": 9.6875, "learning_rate": 4.9991666909849586e-05, "loss": 0.8094, "step": 585 }, { "epoch": 3.641744548286604, "grad_norm": 7.09375, "learning_rate": 4.999160318064656e-05, "loss": 0.8606, "step": 586 }, { "epoch": 3.64797507788162, "grad_norm": 6.46875, "learning_rate": 4.9991539208720096e-05, "loss": 0.8589, "step": 587 }, { "epoch": 3.6542056074766354, "grad_norm": 6.53125, "learning_rate": 4.99914749940708e-05, "loss": 0.6127, "step": 588 }, { "epoch": 3.660436137071651, "grad_norm": 7.59375, "learning_rate": 4.9991410536699304e-05, "loss": 0.5971, "step": 589 }, { "epoch": 3.6666666666666665, "grad_norm": 7.21875, "learning_rate": 4.999134583660623e-05, "loss": 0.7092, "step": 590 }, { "epoch": 3.6728971962616823, "grad_norm": 4.5625, "learning_rate": 4.999128089379221e-05, "loss": 0.6235, "step": 591 }, { "epoch": 3.6791277258566977, "grad_norm": 8.5625, "learning_rate": 4.999121570825788e-05, "loss": 0.7013, "step": 592 }, { "epoch": 3.6853582554517135, "grad_norm": 10.75, "learning_rate": 4.9991150280003864e-05, "loss": 0.9585, "step": 593 }, { "epoch": 3.691588785046729, "grad_norm": 6.96875, "learning_rate": 4.99910846090308e-05, "loss": 0.7, "step": 594 }, { "epoch": 3.6978193146417446, "grad_norm": 4.375, "learning_rate": 4.999101869533932e-05, "loss": 0.6818, "step": 595 }, { "epoch": 3.70404984423676, "grad_norm": 6.53125, "learning_rate": 4.999095253893008e-05, "loss": 0.6139, "step": 596 }, { "epoch": 3.710280373831776, "grad_norm": 6.0, "learning_rate": 4.9990886139803714e-05, "loss": 0.7346, "step": 597 }, { "epoch": 3.716510903426791, "grad_norm": 7.28125, "learning_rate": 4.999081949796086e-05, "loss": 0.9631, "step": 598 }, { "epoch": 3.722741433021807, "grad_norm": 5.375, "learning_rate": 4.999075261340218e-05, "loss": 0.5707, "step": 599 }, { "epoch": 3.7289719626168223, "grad_norm": 8.1875, "learning_rate": 4.999068548612831e-05, "loss": 0.8877, "step": 600 }, { "epoch": 3.735202492211838, "grad_norm": 8.125, "learning_rate": 4.9990618116139904e-05, "loss": 0.7875, "step": 601 }, { "epoch": 3.7414330218068534, "grad_norm": 5.875, "learning_rate": 4.999055050343762e-05, "loss": 0.9689, "step": 602 }, { "epoch": 3.7476635514018692, "grad_norm": 4.3125, "learning_rate": 4.999048264802212e-05, "loss": 0.8712, "step": 603 }, { "epoch": 3.7538940809968846, "grad_norm": 6.71875, "learning_rate": 4.999041454989405e-05, "loss": 0.8917, "step": 604 }, { "epoch": 3.7601246105919004, "grad_norm": 6.34375, "learning_rate": 4.999034620905408e-05, "loss": 0.7225, "step": 605 }, { "epoch": 3.7663551401869158, "grad_norm": 5.375, "learning_rate": 4.9990277625502876e-05, "loss": 0.5459, "step": 606 }, { "epoch": 3.7725856697819315, "grad_norm": 4.90625, "learning_rate": 4.99902087992411e-05, "loss": 0.8627, "step": 607 }, { "epoch": 3.778816199376947, "grad_norm": 4.40625, "learning_rate": 4.999013973026942e-05, "loss": 0.7303, "step": 608 }, { "epoch": 3.7850467289719627, "grad_norm": 4.8125, "learning_rate": 4.999007041858851e-05, "loss": 0.7531, "step": 609 }, { "epoch": 3.791277258566978, "grad_norm": 5.4375, "learning_rate": 4.999000086419903e-05, "loss": 0.8009, "step": 610 }, { "epoch": 3.797507788161994, "grad_norm": 5.375, "learning_rate": 4.998993106710168e-05, "loss": 0.7496, "step": 611 }, { "epoch": 3.803738317757009, "grad_norm": 4.59375, "learning_rate": 4.9989861027297115e-05, "loss": 0.859, "step": 612 }, { "epoch": 3.809968847352025, "grad_norm": 5.3125, "learning_rate": 4.998979074478603e-05, "loss": 0.9045, "step": 613 }, { "epoch": 3.8161993769470404, "grad_norm": 5.1875, "learning_rate": 4.998972021956909e-05, "loss": 0.756, "step": 614 }, { "epoch": 3.822429906542056, "grad_norm": 4.125, "learning_rate": 4.9989649451647e-05, "loss": 0.5173, "step": 615 }, { "epoch": 3.8286604361370715, "grad_norm": 4.5, "learning_rate": 4.998957844102044e-05, "loss": 0.5987, "step": 616 }, { "epoch": 3.8348909657320873, "grad_norm": 5.25, "learning_rate": 4.9989507187690106e-05, "loss": 0.7363, "step": 617 }, { "epoch": 3.8411214953271027, "grad_norm": 5.25, "learning_rate": 4.9989435691656674e-05, "loss": 0.8524, "step": 618 }, { "epoch": 3.8473520249221185, "grad_norm": 5.6875, "learning_rate": 4.9989363952920846e-05, "loss": 0.8353, "step": 619 }, { "epoch": 3.853582554517134, "grad_norm": 4.71875, "learning_rate": 4.9989291971483325e-05, "loss": 0.4733, "step": 620 }, { "epoch": 3.8598130841121496, "grad_norm": 5.0625, "learning_rate": 4.9989219747344804e-05, "loss": 0.855, "step": 621 }, { "epoch": 3.866043613707165, "grad_norm": 4.9375, "learning_rate": 4.998914728050598e-05, "loss": 0.6563, "step": 622 }, { "epoch": 3.872274143302181, "grad_norm": 5.875, "learning_rate": 4.998907457096757e-05, "loss": 0.7815, "step": 623 }, { "epoch": 3.878504672897196, "grad_norm": 4.53125, "learning_rate": 4.998900161873027e-05, "loss": 0.593, "step": 624 }, { "epoch": 3.884735202492212, "grad_norm": 4.78125, "learning_rate": 4.998892842379479e-05, "loss": 0.5736, "step": 625 }, { "epoch": 3.8909657320872273, "grad_norm": 5.03125, "learning_rate": 4.998885498616184e-05, "loss": 0.7603, "step": 626 }, { "epoch": 3.897196261682243, "grad_norm": 5.09375, "learning_rate": 4.998878130583214e-05, "loss": 0.8182, "step": 627 }, { "epoch": 3.9034267912772584, "grad_norm": 5.84375, "learning_rate": 4.99887073828064e-05, "loss": 0.782, "step": 628 }, { "epoch": 3.9096573208722742, "grad_norm": 5.65625, "learning_rate": 4.998863321708534e-05, "loss": 0.7215, "step": 629 }, { "epoch": 3.9158878504672896, "grad_norm": 5.78125, "learning_rate": 4.998855880866967e-05, "loss": 0.7405, "step": 630 }, { "epoch": 3.9221183800623054, "grad_norm": 5.375, "learning_rate": 4.9988484157560136e-05, "loss": 0.6579, "step": 631 }, { "epoch": 3.9283489096573208, "grad_norm": 3.953125, "learning_rate": 4.9988409263757436e-05, "loss": 0.5864, "step": 632 }, { "epoch": 3.9345794392523366, "grad_norm": 6.5, "learning_rate": 4.998833412726231e-05, "loss": 0.7875, "step": 633 }, { "epoch": 3.940809968847352, "grad_norm": 5.09375, "learning_rate": 4.99882587480755e-05, "loss": 0.5504, "step": 634 }, { "epoch": 3.9470404984423677, "grad_norm": 4.6875, "learning_rate": 4.998818312619772e-05, "loss": 0.5876, "step": 635 }, { "epoch": 3.953271028037383, "grad_norm": 4.5625, "learning_rate": 4.99881072616297e-05, "loss": 0.5262, "step": 636 }, { "epoch": 3.959501557632399, "grad_norm": 5.0625, "learning_rate": 4.99880311543722e-05, "loss": 0.7, "step": 637 }, { "epoch": 3.965732087227414, "grad_norm": 4.90625, "learning_rate": 4.998795480442595e-05, "loss": 0.7841, "step": 638 }, { "epoch": 3.97196261682243, "grad_norm": 5.96875, "learning_rate": 4.9987878211791674e-05, "loss": 0.7631, "step": 639 }, { "epoch": 3.9781931464174454, "grad_norm": 4.40625, "learning_rate": 4.998780137647014e-05, "loss": 0.7605, "step": 640 }, { "epoch": 3.984423676012461, "grad_norm": 5.09375, "learning_rate": 4.998772429846208e-05, "loss": 0.6811, "step": 641 }, { "epoch": 3.9906542056074765, "grad_norm": 6.96875, "learning_rate": 4.998764697776824e-05, "loss": 0.7619, "step": 642 }, { "epoch": 3.9968847352024923, "grad_norm": 4.65625, "learning_rate": 4.998756941438939e-05, "loss": 0.5984, "step": 643 }, { "epoch": 4.0, "grad_norm": 2.828125, "learning_rate": 4.9987491608326267e-05, "loss": 0.3537, "step": 644 }, { "epoch": 4.006230529595015, "grad_norm": 6.46875, "learning_rate": 4.9987413559579636e-05, "loss": 0.8512, "step": 645 }, { "epoch": 4.012461059190031, "grad_norm": 5.21875, "learning_rate": 4.998733526815024e-05, "loss": 0.5929, "step": 646 }, { "epoch": 4.018691588785047, "grad_norm": 6.65625, "learning_rate": 4.998725673403885e-05, "loss": 0.5444, "step": 647 }, { "epoch": 4.024922118380062, "grad_norm": 6.71875, "learning_rate": 4.998717795724624e-05, "loss": 0.7869, "step": 648 }, { "epoch": 4.031152647975078, "grad_norm": 6.5625, "learning_rate": 4.998709893777315e-05, "loss": 0.6958, "step": 649 }, { "epoch": 4.037383177570093, "grad_norm": 5.59375, "learning_rate": 4.998701967562035e-05, "loss": 0.6092, "step": 650 }, { "epoch": 4.043613707165109, "grad_norm": 5.125, "learning_rate": 4.9986940170788646e-05, "loss": 0.7544, "step": 651 }, { "epoch": 4.049844236760125, "grad_norm": 6.0625, "learning_rate": 4.998686042327877e-05, "loss": 0.5201, "step": 652 }, { "epoch": 4.05607476635514, "grad_norm": 4.5625, "learning_rate": 4.998678043309151e-05, "loss": 0.4694, "step": 653 }, { "epoch": 4.062305295950155, "grad_norm": 4.8125, "learning_rate": 4.9986700200227645e-05, "loss": 0.6531, "step": 654 }, { "epoch": 4.068535825545172, "grad_norm": 6.40625, "learning_rate": 4.9986619724687954e-05, "loss": 0.8526, "step": 655 }, { "epoch": 4.074766355140187, "grad_norm": 5.3125, "learning_rate": 4.9986539006473215e-05, "loss": 0.6849, "step": 656 }, { "epoch": 4.080996884735202, "grad_norm": 4.875, "learning_rate": 4.9986458045584215e-05, "loss": 0.5885, "step": 657 }, { "epoch": 4.087227414330218, "grad_norm": 5.28125, "learning_rate": 4.9986376842021745e-05, "loss": 0.6471, "step": 658 }, { "epoch": 4.093457943925234, "grad_norm": 5.78125, "learning_rate": 4.9986295395786586e-05, "loss": 0.6092, "step": 659 }, { "epoch": 4.099688473520249, "grad_norm": 4.34375, "learning_rate": 4.998621370687952e-05, "loss": 0.6097, "step": 660 }, { "epoch": 4.105919003115265, "grad_norm": 5.34375, "learning_rate": 4.998613177530137e-05, "loss": 0.7251, "step": 661 }, { "epoch": 4.11214953271028, "grad_norm": 5.6875, "learning_rate": 4.998604960105291e-05, "loss": 0.7224, "step": 662 }, { "epoch": 4.118380062305296, "grad_norm": 4.46875, "learning_rate": 4.9985967184134936e-05, "loss": 0.6455, "step": 663 }, { "epoch": 4.1246105919003115, "grad_norm": 5.8125, "learning_rate": 4.9985884524548254e-05, "loss": 0.4878, "step": 664 }, { "epoch": 4.130841121495327, "grad_norm": 4.96875, "learning_rate": 4.9985801622293674e-05, "loss": 0.5062, "step": 665 }, { "epoch": 4.137071651090342, "grad_norm": 3.90625, "learning_rate": 4.998571847737198e-05, "loss": 0.6016, "step": 666 }, { "epoch": 4.1433021806853585, "grad_norm": 5.46875, "learning_rate": 4.9985635089784e-05, "loss": 0.524, "step": 667 }, { "epoch": 4.149532710280374, "grad_norm": 4.0625, "learning_rate": 4.998555145953054e-05, "loss": 0.458, "step": 668 }, { "epoch": 4.155763239875389, "grad_norm": 4.59375, "learning_rate": 4.998546758661241e-05, "loss": 0.4851, "step": 669 }, { "epoch": 4.161993769470405, "grad_norm": 4.40625, "learning_rate": 4.9985383471030425e-05, "loss": 0.5916, "step": 670 }, { "epoch": 4.168224299065421, "grad_norm": 6.09375, "learning_rate": 4.9985299112785406e-05, "loss": 0.5923, "step": 671 }, { "epoch": 4.174454828660436, "grad_norm": 5.4375, "learning_rate": 4.998521451187816e-05, "loss": 0.6733, "step": 672 }, { "epoch": 4.1806853582554515, "grad_norm": 5.625, "learning_rate": 4.998512966830951e-05, "loss": 0.8544, "step": 673 }, { "epoch": 4.186915887850467, "grad_norm": 5.5625, "learning_rate": 4.99850445820803e-05, "loss": 0.6227, "step": 674 }, { "epoch": 4.193146417445483, "grad_norm": 4.65625, "learning_rate": 4.998495925319134e-05, "loss": 0.4555, "step": 675 }, { "epoch": 4.1993769470404985, "grad_norm": 5.5625, "learning_rate": 4.998487368164346e-05, "loss": 0.6757, "step": 676 }, { "epoch": 4.205607476635514, "grad_norm": 4.90625, "learning_rate": 4.998478786743749e-05, "loss": 0.6447, "step": 677 }, { "epoch": 4.211838006230529, "grad_norm": 4.75, "learning_rate": 4.998470181057426e-05, "loss": 0.613, "step": 678 }, { "epoch": 4.218068535825545, "grad_norm": 5.0625, "learning_rate": 4.998461551105462e-05, "loss": 0.6352, "step": 679 }, { "epoch": 4.224299065420561, "grad_norm": 5.46875, "learning_rate": 4.9984528968879394e-05, "loss": 0.7015, "step": 680 }, { "epoch": 4.230529595015576, "grad_norm": 6.53125, "learning_rate": 4.998444218404944e-05, "loss": 0.7203, "step": 681 }, { "epoch": 4.2367601246105915, "grad_norm": 4.8125, "learning_rate": 4.9984355156565576e-05, "loss": 0.6342, "step": 682 }, { "epoch": 4.242990654205608, "grad_norm": 5.0625, "learning_rate": 4.998426788642867e-05, "loss": 0.6383, "step": 683 }, { "epoch": 4.249221183800623, "grad_norm": 5.4375, "learning_rate": 4.9984180373639555e-05, "loss": 0.5606, "step": 684 }, { "epoch": 4.255451713395638, "grad_norm": 5.0, "learning_rate": 4.998409261819909e-05, "loss": 0.5101, "step": 685 }, { "epoch": 4.261682242990654, "grad_norm": 4.21875, "learning_rate": 4.998400462010812e-05, "loss": 0.5743, "step": 686 }, { "epoch": 4.26791277258567, "grad_norm": 5.28125, "learning_rate": 4.99839163793675e-05, "loss": 0.6679, "step": 687 }, { "epoch": 4.274143302180685, "grad_norm": 5.46875, "learning_rate": 4.998382789597809e-05, "loss": 0.5584, "step": 688 }, { "epoch": 4.280373831775701, "grad_norm": 5.125, "learning_rate": 4.9983739169940756e-05, "loss": 0.5341, "step": 689 }, { "epoch": 4.286604361370716, "grad_norm": 4.90625, "learning_rate": 4.998365020125635e-05, "loss": 0.5683, "step": 690 }, { "epoch": 4.292834890965732, "grad_norm": 5.21875, "learning_rate": 4.9983560989925736e-05, "loss": 0.4846, "step": 691 }, { "epoch": 4.299065420560748, "grad_norm": 5.125, "learning_rate": 4.998347153594979e-05, "loss": 0.6068, "step": 692 }, { "epoch": 4.305295950155763, "grad_norm": 4.4375, "learning_rate": 4.998338183932937e-05, "loss": 0.4914, "step": 693 }, { "epoch": 4.311526479750778, "grad_norm": 5.5, "learning_rate": 4.998329190006535e-05, "loss": 0.7465, "step": 694 }, { "epoch": 4.317757009345795, "grad_norm": 5.1875, "learning_rate": 4.99832017181586e-05, "loss": 0.7215, "step": 695 }, { "epoch": 4.32398753894081, "grad_norm": 4.28125, "learning_rate": 4.9983111293610015e-05, "loss": 0.6023, "step": 696 }, { "epoch": 4.330218068535825, "grad_norm": 5.375, "learning_rate": 4.998302062642045e-05, "loss": 0.7708, "step": 697 }, { "epoch": 4.336448598130841, "grad_norm": 5.375, "learning_rate": 4.99829297165908e-05, "loss": 0.5709, "step": 698 }, { "epoch": 4.342679127725857, "grad_norm": 4.0625, "learning_rate": 4.998283856412194e-05, "loss": 0.4831, "step": 699 }, { "epoch": 4.348909657320872, "grad_norm": 4.28125, "learning_rate": 4.998274716901475e-05, "loss": 0.5089, "step": 700 }, { "epoch": 4.355140186915888, "grad_norm": 3.90625, "learning_rate": 4.998265553127013e-05, "loss": 0.5293, "step": 701 }, { "epoch": 4.361370716510903, "grad_norm": 8.4375, "learning_rate": 4.998256365088897e-05, "loss": 0.6808, "step": 702 }, { "epoch": 4.367601246105919, "grad_norm": 7.125, "learning_rate": 4.998247152787216e-05, "loss": 0.4541, "step": 703 }, { "epoch": 4.373831775700935, "grad_norm": 6.1875, "learning_rate": 4.9982379162220586e-05, "loss": 0.6551, "step": 704 }, { "epoch": 4.38006230529595, "grad_norm": 7.125, "learning_rate": 4.998228655393515e-05, "loss": 0.8244, "step": 705 }, { "epoch": 4.386292834890965, "grad_norm": 7.25, "learning_rate": 4.9982193703016765e-05, "loss": 0.5893, "step": 706 }, { "epoch": 4.392523364485982, "grad_norm": 7.875, "learning_rate": 4.998210060946631e-05, "loss": 0.4978, "step": 707 }, { "epoch": 4.398753894080997, "grad_norm": 4.46875, "learning_rate": 4.99820072732847e-05, "loss": 0.6438, "step": 708 }, { "epoch": 4.404984423676012, "grad_norm": 9.125, "learning_rate": 4.9981913694472846e-05, "loss": 0.5922, "step": 709 }, { "epoch": 4.411214953271028, "grad_norm": 6.96875, "learning_rate": 4.9981819873031656e-05, "loss": 0.6971, "step": 710 }, { "epoch": 4.417445482866044, "grad_norm": 6.65625, "learning_rate": 4.998172580896203e-05, "loss": 0.5714, "step": 711 }, { "epoch": 4.423676012461059, "grad_norm": 7.3125, "learning_rate": 4.998163150226489e-05, "loss": 0.6378, "step": 712 }, { "epoch": 4.429906542056075, "grad_norm": 4.78125, "learning_rate": 4.998153695294116e-05, "loss": 0.4493, "step": 713 }, { "epoch": 4.43613707165109, "grad_norm": 4.1875, "learning_rate": 4.998144216099174e-05, "loss": 0.3921, "step": 714 }, { "epoch": 4.442367601246106, "grad_norm": 4.625, "learning_rate": 4.9981347126417564e-05, "loss": 0.5869, "step": 715 }, { "epoch": 4.4485981308411215, "grad_norm": 4.96875, "learning_rate": 4.998125184921955e-05, "loss": 0.4429, "step": 716 }, { "epoch": 4.454828660436137, "grad_norm": 5.03125, "learning_rate": 4.9981156329398626e-05, "loss": 0.5401, "step": 717 }, { "epoch": 4.461059190031152, "grad_norm": 5.46875, "learning_rate": 4.9981060566955715e-05, "loss": 0.6392, "step": 718 }, { "epoch": 4.4672897196261685, "grad_norm": 6.15625, "learning_rate": 4.998096456189175e-05, "loss": 0.4644, "step": 719 }, { "epoch": 4.473520249221184, "grad_norm": 4.375, "learning_rate": 4.9980868314207665e-05, "loss": 0.5941, "step": 720 }, { "epoch": 4.479750778816199, "grad_norm": 4.90625, "learning_rate": 4.9980771823904396e-05, "loss": 0.637, "step": 721 }, { "epoch": 4.485981308411215, "grad_norm": 4.375, "learning_rate": 4.998067509098288e-05, "loss": 0.597, "step": 722 }, { "epoch": 4.492211838006231, "grad_norm": 5.15625, "learning_rate": 4.9980578115444045e-05, "loss": 0.8217, "step": 723 }, { "epoch": 4.498442367601246, "grad_norm": 5.1875, "learning_rate": 4.998048089728884e-05, "loss": 0.8387, "step": 724 }, { "epoch": 4.5046728971962615, "grad_norm": 6.625, "learning_rate": 4.998038343651822e-05, "loss": 0.6024, "step": 725 }, { "epoch": 4.510903426791277, "grad_norm": 4.0, "learning_rate": 4.998028573313312e-05, "loss": 0.5134, "step": 726 }, { "epoch": 4.517133956386293, "grad_norm": 5.34375, "learning_rate": 4.9980187787134495e-05, "loss": 0.6402, "step": 727 }, { "epoch": 4.5233644859813085, "grad_norm": 6.5625, "learning_rate": 4.998008959852328e-05, "loss": 0.8354, "step": 728 }, { "epoch": 4.529595015576324, "grad_norm": 5.75, "learning_rate": 4.9979991167300453e-05, "loss": 0.7811, "step": 729 }, { "epoch": 4.535825545171339, "grad_norm": 4.46875, "learning_rate": 4.997989249346695e-05, "loss": 0.4961, "step": 730 }, { "epoch": 4.542056074766355, "grad_norm": 4.1875, "learning_rate": 4.9979793577023746e-05, "loss": 0.3693, "step": 731 }, { "epoch": 4.548286604361371, "grad_norm": 4.09375, "learning_rate": 4.997969441797179e-05, "loss": 0.4789, "step": 732 }, { "epoch": 4.554517133956386, "grad_norm": 4.03125, "learning_rate": 4.997959501631205e-05, "loss": 0.5738, "step": 733 }, { "epoch": 4.5607476635514015, "grad_norm": 3.796875, "learning_rate": 4.997949537204549e-05, "loss": 0.5126, "step": 734 }, { "epoch": 4.566978193146418, "grad_norm": 4.0, "learning_rate": 4.997939548517308e-05, "loss": 0.4648, "step": 735 }, { "epoch": 4.573208722741433, "grad_norm": 3.453125, "learning_rate": 4.9979295355695776e-05, "loss": 0.5322, "step": 736 }, { "epoch": 4.579439252336448, "grad_norm": 4.96875, "learning_rate": 4.997919498361457e-05, "loss": 0.554, "step": 737 }, { "epoch": 4.585669781931464, "grad_norm": 3.59375, "learning_rate": 4.997909436893043e-05, "loss": 0.6386, "step": 738 }, { "epoch": 4.59190031152648, "grad_norm": 4.28125, "learning_rate": 4.9978993511644334e-05, "loss": 0.5965, "step": 739 }, { "epoch": 4.598130841121495, "grad_norm": 3.890625, "learning_rate": 4.997889241175726e-05, "loss": 0.416, "step": 740 }, { "epoch": 4.604361370716511, "grad_norm": 4.6875, "learning_rate": 4.9978791069270184e-05, "loss": 0.3923, "step": 741 }, { "epoch": 4.610591900311526, "grad_norm": 4.96875, "learning_rate": 4.9978689484184104e-05, "loss": 0.5956, "step": 742 }, { "epoch": 4.616822429906542, "grad_norm": 4.09375, "learning_rate": 4.9978587656499995e-05, "loss": 0.4647, "step": 743 }, { "epoch": 4.623052959501558, "grad_norm": 3.40625, "learning_rate": 4.9978485586218846e-05, "loss": 0.4451, "step": 744 }, { "epoch": 4.629283489096573, "grad_norm": 4.21875, "learning_rate": 4.9978383273341655e-05, "loss": 0.5009, "step": 745 }, { "epoch": 4.635514018691588, "grad_norm": 4.625, "learning_rate": 4.997828071786941e-05, "loss": 0.5868, "step": 746 }, { "epoch": 4.641744548286605, "grad_norm": 5.125, "learning_rate": 4.997817791980311e-05, "loss": 0.7593, "step": 747 }, { "epoch": 4.64797507788162, "grad_norm": 6.625, "learning_rate": 4.9978074879143754e-05, "loss": 0.5129, "step": 748 }, { "epoch": 4.654205607476635, "grad_norm": 4.53125, "learning_rate": 4.997797159589235e-05, "loss": 0.4364, "step": 749 }, { "epoch": 4.660436137071651, "grad_norm": 5.53125, "learning_rate": 4.997786807004988e-05, "loss": 0.5048, "step": 750 }, { "epoch": 4.666666666666667, "grad_norm": 7.40625, "learning_rate": 4.997776430161737e-05, "loss": 0.7256, "step": 751 }, { "epoch": 4.672897196261682, "grad_norm": 3.78125, "learning_rate": 4.997766029059582e-05, "loss": 0.4477, "step": 752 }, { "epoch": 4.679127725856698, "grad_norm": 4.3125, "learning_rate": 4.9977556036986236e-05, "loss": 0.4604, "step": 753 }, { "epoch": 4.685358255451713, "grad_norm": 5.96875, "learning_rate": 4.9977451540789635e-05, "loss": 0.6264, "step": 754 }, { "epoch": 4.691588785046729, "grad_norm": 3.421875, "learning_rate": 4.997734680200703e-05, "loss": 0.5397, "step": 755 }, { "epoch": 4.697819314641745, "grad_norm": 5.0, "learning_rate": 4.997724182063945e-05, "loss": 0.4754, "step": 756 }, { "epoch": 4.70404984423676, "grad_norm": 4.1875, "learning_rate": 4.99771365966879e-05, "loss": 0.542, "step": 757 }, { "epoch": 4.710280373831775, "grad_norm": 6.4375, "learning_rate": 4.997703113015339e-05, "loss": 0.4681, "step": 758 }, { "epoch": 4.716510903426792, "grad_norm": 3.703125, "learning_rate": 4.9976925421036986e-05, "loss": 0.5615, "step": 759 }, { "epoch": 4.722741433021807, "grad_norm": 6.53125, "learning_rate": 4.997681946933967e-05, "loss": 0.687, "step": 760 }, { "epoch": 4.728971962616822, "grad_norm": 5.0625, "learning_rate": 4.9976713275062506e-05, "loss": 0.5605, "step": 761 }, { "epoch": 4.735202492211838, "grad_norm": 4.65625, "learning_rate": 4.9976606838206506e-05, "loss": 0.6131, "step": 762 }, { "epoch": 4.741433021806854, "grad_norm": 4.78125, "learning_rate": 4.9976500158772706e-05, "loss": 0.5277, "step": 763 }, { "epoch": 4.747663551401869, "grad_norm": 5.03125, "learning_rate": 4.997639323676214e-05, "loss": 0.4088, "step": 764 }, { "epoch": 4.753894080996885, "grad_norm": 4.59375, "learning_rate": 4.997628607217585e-05, "loss": 0.6742, "step": 765 }, { "epoch": 4.7601246105919, "grad_norm": 5.90625, "learning_rate": 4.997617866501488e-05, "loss": 0.6446, "step": 766 }, { "epoch": 4.766355140186916, "grad_norm": 4.4375, "learning_rate": 4.9976071015280276e-05, "loss": 0.392, "step": 767 }, { "epoch": 4.7725856697819315, "grad_norm": 4.4375, "learning_rate": 4.9975963122973066e-05, "loss": 0.6358, "step": 768 }, { "epoch": 4.778816199376947, "grad_norm": 3.828125, "learning_rate": 4.9975854988094316e-05, "loss": 0.5524, "step": 769 }, { "epoch": 4.785046728971962, "grad_norm": 5.125, "learning_rate": 4.9975746610645075e-05, "loss": 0.423, "step": 770 }, { "epoch": 4.7912772585669785, "grad_norm": 4.375, "learning_rate": 4.997563799062638e-05, "loss": 0.4124, "step": 771 }, { "epoch": 4.797507788161994, "grad_norm": 4.9375, "learning_rate": 4.9975529128039305e-05, "loss": 0.5648, "step": 772 }, { "epoch": 4.803738317757009, "grad_norm": 6.59375, "learning_rate": 4.9975420022884896e-05, "loss": 0.4473, "step": 773 }, { "epoch": 4.809968847352025, "grad_norm": 4.375, "learning_rate": 4.997531067516421e-05, "loss": 0.548, "step": 774 }, { "epoch": 4.816199376947041, "grad_norm": 4.125, "learning_rate": 4.997520108487832e-05, "loss": 0.3893, "step": 775 }, { "epoch": 4.822429906542056, "grad_norm": 3.78125, "learning_rate": 4.997509125202828e-05, "loss": 0.5388, "step": 776 }, { "epoch": 4.8286604361370715, "grad_norm": 4.75, "learning_rate": 4.997498117661517e-05, "loss": 0.5643, "step": 777 }, { "epoch": 4.834890965732087, "grad_norm": 6.15625, "learning_rate": 4.997487085864004e-05, "loss": 0.6649, "step": 778 }, { "epoch": 4.841121495327103, "grad_norm": 4.46875, "learning_rate": 4.997476029810398e-05, "loss": 0.5124, "step": 779 }, { "epoch": 4.8473520249221185, "grad_norm": 4.46875, "learning_rate": 4.9974649495008055e-05, "loss": 0.6564, "step": 780 }, { "epoch": 4.853582554517134, "grad_norm": 3.21875, "learning_rate": 4.9974538449353335e-05, "loss": 0.4781, "step": 781 }, { "epoch": 4.859813084112149, "grad_norm": 7.4375, "learning_rate": 4.997442716114091e-05, "loss": 0.447, "step": 782 }, { "epoch": 4.866043613707165, "grad_norm": 6.65625, "learning_rate": 4.9974315630371855e-05, "loss": 0.3421, "step": 783 }, { "epoch": 4.872274143302181, "grad_norm": 4.84375, "learning_rate": 4.997420385704725e-05, "loss": 0.4854, "step": 784 }, { "epoch": 4.878504672897196, "grad_norm": 4.6875, "learning_rate": 4.9974091841168195e-05, "loss": 0.4635, "step": 785 }, { "epoch": 4.8847352024922115, "grad_norm": 6.5, "learning_rate": 4.997397958273576e-05, "loss": 0.5343, "step": 786 }, { "epoch": 4.890965732087228, "grad_norm": 5.125, "learning_rate": 4.997386708175105e-05, "loss": 0.4888, "step": 787 }, { "epoch": 4.897196261682243, "grad_norm": 3.484375, "learning_rate": 4.997375433821514e-05, "loss": 0.3721, "step": 788 }, { "epoch": 4.9034267912772584, "grad_norm": 4.3125, "learning_rate": 4.9973641352129144e-05, "loss": 0.4948, "step": 789 }, { "epoch": 4.909657320872274, "grad_norm": 9.4375, "learning_rate": 4.9973528123494146e-05, "loss": 0.485, "step": 790 }, { "epoch": 4.91588785046729, "grad_norm": 8.0625, "learning_rate": 4.997341465231126e-05, "loss": 0.5562, "step": 791 }, { "epoch": 4.922118380062305, "grad_norm": 5.15625, "learning_rate": 4.997330093858157e-05, "loss": 0.5911, "step": 792 }, { "epoch": 4.928348909657321, "grad_norm": 5.28125, "learning_rate": 4.997318698230619e-05, "loss": 0.5905, "step": 793 }, { "epoch": 4.934579439252336, "grad_norm": 6.5625, "learning_rate": 4.9973072783486226e-05, "loss": 0.5666, "step": 794 }, { "epoch": 4.940809968847352, "grad_norm": 5.34375, "learning_rate": 4.997295834212279e-05, "loss": 0.5233, "step": 795 }, { "epoch": 4.947040498442368, "grad_norm": 3.28125, "learning_rate": 4.997284365821699e-05, "loss": 0.5307, "step": 796 }, { "epoch": 4.953271028037383, "grad_norm": 4.78125, "learning_rate": 4.997272873176994e-05, "loss": 0.6502, "step": 797 }, { "epoch": 4.959501557632398, "grad_norm": 4.34375, "learning_rate": 4.9972613562782756e-05, "loss": 0.4848, "step": 798 }, { "epoch": 4.965732087227415, "grad_norm": 3.796875, "learning_rate": 4.9972498151256555e-05, "loss": 0.4976, "step": 799 }, { "epoch": 4.97196261682243, "grad_norm": 4.8125, "learning_rate": 4.9972382497192464e-05, "loss": 0.7667, "step": 800 }, { "epoch": 4.978193146417445, "grad_norm": 5.3125, "learning_rate": 4.99722666005916e-05, "loss": 0.4827, "step": 801 }, { "epoch": 4.984423676012461, "grad_norm": 5.15625, "learning_rate": 4.997215046145509e-05, "loss": 0.4081, "step": 802 }, { "epoch": 4.990654205607477, "grad_norm": 4.71875, "learning_rate": 4.997203407978407e-05, "loss": 0.4031, "step": 803 }, { "epoch": 4.996884735202492, "grad_norm": 6.75, "learning_rate": 4.9971917455579654e-05, "loss": 0.3994, "step": 804 }, { "epoch": 5.0, "grad_norm": 2.9375, "learning_rate": 4.997180058884299e-05, "loss": 0.29, "step": 805 }, { "epoch": 5.006230529595015, "grad_norm": 5.625, "learning_rate": 4.99716834795752e-05, "loss": 0.5152, "step": 806 }, { "epoch": 5.012461059190031, "grad_norm": 5.53125, "learning_rate": 4.9971566127777436e-05, "loss": 0.5324, "step": 807 }, { "epoch": 5.018691588785047, "grad_norm": 3.15625, "learning_rate": 4.9971448533450825e-05, "loss": 0.4167, "step": 808 }, { "epoch": 5.024922118380062, "grad_norm": 4.21875, "learning_rate": 4.997133069659652e-05, "loss": 0.3634, "step": 809 }, { "epoch": 5.031152647975078, "grad_norm": 5.09375, "learning_rate": 4.997121261721565e-05, "loss": 0.5812, "step": 810 }, { "epoch": 5.037383177570093, "grad_norm": 4.21875, "learning_rate": 4.997109429530938e-05, "loss": 0.5952, "step": 811 }, { "epoch": 5.043613707165109, "grad_norm": 4.34375, "learning_rate": 4.997097573087884e-05, "loss": 0.5939, "step": 812 }, { "epoch": 5.049844236760125, "grad_norm": 4.8125, "learning_rate": 4.997085692392521e-05, "loss": 0.6911, "step": 813 }, { "epoch": 5.05607476635514, "grad_norm": 4.96875, "learning_rate": 4.9970737874449615e-05, "loss": 0.5382, "step": 814 }, { "epoch": 5.062305295950155, "grad_norm": 4.90625, "learning_rate": 4.997061858245322e-05, "loss": 0.4517, "step": 815 }, { "epoch": 5.068535825545172, "grad_norm": 5.1875, "learning_rate": 4.997049904793719e-05, "loss": 0.7311, "step": 816 }, { "epoch": 5.074766355140187, "grad_norm": 6.59375, "learning_rate": 4.997037927090268e-05, "loss": 0.5173, "step": 817 }, { "epoch": 5.080996884735202, "grad_norm": 5.40625, "learning_rate": 4.997025925135086e-05, "loss": 0.4499, "step": 818 }, { "epoch": 5.087227414330218, "grad_norm": 5.375, "learning_rate": 4.997013898928288e-05, "loss": 0.5604, "step": 819 }, { "epoch": 5.093457943925234, "grad_norm": 4.71875, "learning_rate": 4.997001848469993e-05, "loss": 0.6147, "step": 820 }, { "epoch": 5.099688473520249, "grad_norm": 4.34375, "learning_rate": 4.996989773760316e-05, "loss": 0.4573, "step": 821 }, { "epoch": 5.105919003115265, "grad_norm": 4.6875, "learning_rate": 4.996977674799376e-05, "loss": 0.538, "step": 822 }, { "epoch": 5.11214953271028, "grad_norm": 3.359375, "learning_rate": 4.9969655515872895e-05, "loss": 0.4125, "step": 823 }, { "epoch": 5.118380062305296, "grad_norm": 4.25, "learning_rate": 4.996953404124174e-05, "loss": 0.5349, "step": 824 }, { "epoch": 5.1246105919003115, "grad_norm": 4.46875, "learning_rate": 4.996941232410148e-05, "loss": 0.5448, "step": 825 }, { "epoch": 5.130841121495327, "grad_norm": 5.21875, "learning_rate": 4.996929036445329e-05, "loss": 0.6183, "step": 826 }, { "epoch": 5.137071651090342, "grad_norm": 4.03125, "learning_rate": 4.996916816229837e-05, "loss": 0.5212, "step": 827 }, { "epoch": 5.1433021806853585, "grad_norm": 6.0, "learning_rate": 4.9969045717637896e-05, "loss": 0.5104, "step": 828 }, { "epoch": 5.149532710280374, "grad_norm": 4.59375, "learning_rate": 4.996892303047306e-05, "loss": 0.455, "step": 829 }, { "epoch": 5.155763239875389, "grad_norm": 4.3125, "learning_rate": 4.9968800100805054e-05, "loss": 0.5589, "step": 830 }, { "epoch": 5.161993769470405, "grad_norm": 3.859375, "learning_rate": 4.996867692863507e-05, "loss": 0.4931, "step": 831 }, { "epoch": 5.168224299065421, "grad_norm": 5.34375, "learning_rate": 4.99685535139643e-05, "loss": 0.8132, "step": 832 }, { "epoch": 5.174454828660436, "grad_norm": 4.40625, "learning_rate": 4.9968429856793945e-05, "loss": 0.3563, "step": 833 }, { "epoch": 5.1806853582554515, "grad_norm": 4.25, "learning_rate": 4.9968305957125215e-05, "loss": 0.3753, "step": 834 }, { "epoch": 5.186915887850467, "grad_norm": 3.515625, "learning_rate": 4.99681818149593e-05, "loss": 0.5692, "step": 835 }, { "epoch": 5.193146417445483, "grad_norm": 6.75, "learning_rate": 4.9968057430297415e-05, "loss": 0.4894, "step": 836 }, { "epoch": 5.1993769470404985, "grad_norm": 6.53125, "learning_rate": 4.9967932803140764e-05, "loss": 0.4608, "step": 837 }, { "epoch": 5.205607476635514, "grad_norm": 4.75, "learning_rate": 4.996780793349056e-05, "loss": 0.412, "step": 838 }, { "epoch": 5.211838006230529, "grad_norm": 4.1875, "learning_rate": 4.996768282134802e-05, "loss": 0.6614, "step": 839 }, { "epoch": 5.218068535825545, "grad_norm": 4.9375, "learning_rate": 4.996755746671434e-05, "loss": 0.3526, "step": 840 }, { "epoch": 5.224299065420561, "grad_norm": 5.6875, "learning_rate": 4.9967431869590766e-05, "loss": 0.5028, "step": 841 }, { "epoch": 5.230529595015576, "grad_norm": 6.03125, "learning_rate": 4.9967306029978497e-05, "loss": 0.4771, "step": 842 }, { "epoch": 5.2367601246105915, "grad_norm": 5.34375, "learning_rate": 4.996717994787876e-05, "loss": 0.5306, "step": 843 }, { "epoch": 5.242990654205608, "grad_norm": 5.75, "learning_rate": 4.996705362329278e-05, "loss": 0.6897, "step": 844 }, { "epoch": 5.249221183800623, "grad_norm": 4.875, "learning_rate": 4.996692705622178e-05, "loss": 0.4081, "step": 845 }, { "epoch": 5.255451713395638, "grad_norm": 5.6875, "learning_rate": 4.9966800246667e-05, "loss": 0.5947, "step": 846 }, { "epoch": 5.261682242990654, "grad_norm": 4.0, "learning_rate": 4.996667319462966e-05, "loss": 0.4354, "step": 847 }, { "epoch": 5.26791277258567, "grad_norm": 4.9375, "learning_rate": 4.9966545900111006e-05, "loss": 0.5428, "step": 848 }, { "epoch": 5.274143302180685, "grad_norm": 5.40625, "learning_rate": 4.996641836311226e-05, "loss": 0.5329, "step": 849 }, { "epoch": 5.280373831775701, "grad_norm": 3.953125, "learning_rate": 4.996629058363468e-05, "loss": 0.4647, "step": 850 }, { "epoch": 5.286604361370716, "grad_norm": 3.046875, "learning_rate": 4.9966162561679486e-05, "loss": 0.4125, "step": 851 }, { "epoch": 5.292834890965732, "grad_norm": 4.5625, "learning_rate": 4.996603429724793e-05, "loss": 0.5113, "step": 852 }, { "epoch": 5.299065420560748, "grad_norm": 4.90625, "learning_rate": 4.996590579034126e-05, "loss": 0.3235, "step": 853 }, { "epoch": 5.305295950155763, "grad_norm": 3.609375, "learning_rate": 4.996577704096072e-05, "loss": 0.3765, "step": 854 }, { "epoch": 5.311526479750778, "grad_norm": 4.90625, "learning_rate": 4.9965648049107575e-05, "loss": 0.569, "step": 855 }, { "epoch": 5.317757009345795, "grad_norm": 4.3125, "learning_rate": 4.996551881478305e-05, "loss": 0.5853, "step": 856 }, { "epoch": 5.32398753894081, "grad_norm": 4.9375, "learning_rate": 4.996538933798843e-05, "loss": 0.5972, "step": 857 }, { "epoch": 5.330218068535825, "grad_norm": 6.34375, "learning_rate": 4.996525961872495e-05, "loss": 0.8639, "step": 858 }, { "epoch": 5.336448598130841, "grad_norm": 6.0625, "learning_rate": 4.9965129656993875e-05, "loss": 0.5853, "step": 859 }, { "epoch": 5.342679127725857, "grad_norm": 5.375, "learning_rate": 4.996499945279648e-05, "loss": 0.3496, "step": 860 }, { "epoch": 5.348909657320872, "grad_norm": 6.28125, "learning_rate": 4.9964869006134004e-05, "loss": 0.6613, "step": 861 }, { "epoch": 5.355140186915888, "grad_norm": 4.53125, "learning_rate": 4.996473831700774e-05, "loss": 0.4269, "step": 862 }, { "epoch": 5.361370716510903, "grad_norm": 7.28125, "learning_rate": 4.996460738541895e-05, "loss": 0.789, "step": 863 }, { "epoch": 5.367601246105919, "grad_norm": 3.015625, "learning_rate": 4.996447621136891e-05, "loss": 0.4311, "step": 864 }, { "epoch": 5.373831775700935, "grad_norm": 4.3125, "learning_rate": 4.996434479485886e-05, "loss": 0.4077, "step": 865 }, { "epoch": 5.38006230529595, "grad_norm": 6.21875, "learning_rate": 4.996421313589013e-05, "loss": 0.4448, "step": 866 }, { "epoch": 5.386292834890965, "grad_norm": 8.4375, "learning_rate": 4.996408123446396e-05, "loss": 0.5301, "step": 867 }, { "epoch": 5.392523364485982, "grad_norm": 4.0625, "learning_rate": 4.996394909058164e-05, "loss": 0.4028, "step": 868 }, { "epoch": 5.398753894080997, "grad_norm": 6.5, "learning_rate": 4.996381670424446e-05, "loss": 0.6803, "step": 869 }, { "epoch": 5.404984423676012, "grad_norm": 10.8125, "learning_rate": 4.99636840754537e-05, "loss": 0.4817, "step": 870 }, { "epoch": 5.411214953271028, "grad_norm": 4.21875, "learning_rate": 4.9963551204210655e-05, "loss": 0.3013, "step": 871 }, { "epoch": 5.417445482866044, "grad_norm": 3.515625, "learning_rate": 4.996341809051661e-05, "loss": 0.3417, "step": 872 }, { "epoch": 5.423676012461059, "grad_norm": 10.0, "learning_rate": 4.9963284734372855e-05, "loss": 0.6224, "step": 873 }, { "epoch": 5.429906542056075, "grad_norm": 10.75, "learning_rate": 4.996315113578069e-05, "loss": 0.4565, "step": 874 }, { "epoch": 5.43613707165109, "grad_norm": 9.5625, "learning_rate": 4.9963017294741407e-05, "loss": 0.5562, "step": 875 }, { "epoch": 5.442367601246106, "grad_norm": 4.53125, "learning_rate": 4.9962883211256315e-05, "loss": 0.398, "step": 876 }, { "epoch": 5.4485981308411215, "grad_norm": 6.625, "learning_rate": 4.996274888532672e-05, "loss": 0.5051, "step": 877 }, { "epoch": 5.454828660436137, "grad_norm": 9.875, "learning_rate": 4.99626143169539e-05, "loss": 0.5218, "step": 878 }, { "epoch": 5.461059190031152, "grad_norm": 7.96875, "learning_rate": 4.9962479506139195e-05, "loss": 0.4092, "step": 879 }, { "epoch": 5.4672897196261685, "grad_norm": 3.578125, "learning_rate": 4.996234445288389e-05, "loss": 0.5676, "step": 880 }, { "epoch": 5.473520249221184, "grad_norm": 5.125, "learning_rate": 4.996220915718931e-05, "loss": 0.3629, "step": 881 }, { "epoch": 5.479750778816199, "grad_norm": 5.40625, "learning_rate": 4.996207361905676e-05, "loss": 0.5503, "step": 882 }, { "epoch": 5.485981308411215, "grad_norm": 7.875, "learning_rate": 4.996193783848756e-05, "loss": 0.5268, "step": 883 }, { "epoch": 5.492211838006231, "grad_norm": 3.921875, "learning_rate": 4.996180181548303e-05, "loss": 0.4549, "step": 884 }, { "epoch": 5.498442367601246, "grad_norm": 6.625, "learning_rate": 4.99616655500445e-05, "loss": 0.3825, "step": 885 }, { "epoch": 5.5046728971962615, "grad_norm": 6.71875, "learning_rate": 4.996152904217328e-05, "loss": 0.4499, "step": 886 }, { "epoch": 5.510903426791277, "grad_norm": 4.3125, "learning_rate": 4.99613922918707e-05, "loss": 0.4498, "step": 887 }, { "epoch": 5.517133956386293, "grad_norm": 3.953125, "learning_rate": 4.996125529913809e-05, "loss": 0.4688, "step": 888 }, { "epoch": 5.5233644859813085, "grad_norm": 3.203125, "learning_rate": 4.996111806397677e-05, "loss": 0.4939, "step": 889 }, { "epoch": 5.529595015576324, "grad_norm": 5.65625, "learning_rate": 4.996098058638809e-05, "loss": 0.6239, "step": 890 }, { "epoch": 5.535825545171339, "grad_norm": 3.125, "learning_rate": 4.996084286637338e-05, "loss": 0.5127, "step": 891 }, { "epoch": 5.542056074766355, "grad_norm": 4.375, "learning_rate": 4.996070490393397e-05, "loss": 0.601, "step": 892 }, { "epoch": 5.548286604361371, "grad_norm": 2.796875, "learning_rate": 4.9960566699071204e-05, "loss": 0.3069, "step": 893 }, { "epoch": 5.554517133956386, "grad_norm": 3.484375, "learning_rate": 4.9960428251786426e-05, "loss": 0.4631, "step": 894 }, { "epoch": 5.5607476635514015, "grad_norm": 4.53125, "learning_rate": 4.9960289562080974e-05, "loss": 0.4522, "step": 895 }, { "epoch": 5.566978193146418, "grad_norm": 4.40625, "learning_rate": 4.996015062995621e-05, "loss": 0.3931, "step": 896 }, { "epoch": 5.573208722741433, "grad_norm": 3.703125, "learning_rate": 4.996001145541347e-05, "loss": 0.5091, "step": 897 }, { "epoch": 5.579439252336448, "grad_norm": 3.34375, "learning_rate": 4.99598720384541e-05, "loss": 0.4879, "step": 898 }, { "epoch": 5.585669781931464, "grad_norm": 6.21875, "learning_rate": 4.9959732379079475e-05, "loss": 0.6151, "step": 899 }, { "epoch": 5.59190031152648, "grad_norm": 7.34375, "learning_rate": 4.9959592477290936e-05, "loss": 0.4314, "step": 900 }, { "epoch": 5.598130841121495, "grad_norm": 4.65625, "learning_rate": 4.995945233308985e-05, "loss": 0.5442, "step": 901 }, { "epoch": 5.604361370716511, "grad_norm": 4.25, "learning_rate": 4.995931194647757e-05, "loss": 0.4462, "step": 902 }, { "epoch": 5.610591900311526, "grad_norm": 6.25, "learning_rate": 4.9959171317455455e-05, "loss": 0.6712, "step": 903 }, { "epoch": 5.616822429906542, "grad_norm": 5.03125, "learning_rate": 4.995903044602489e-05, "loss": 0.3918, "step": 904 }, { "epoch": 5.623052959501558, "grad_norm": 3.96875, "learning_rate": 4.995888933218723e-05, "loss": 0.4826, "step": 905 }, { "epoch": 5.629283489096573, "grad_norm": 4.96875, "learning_rate": 4.995874797594384e-05, "loss": 0.4262, "step": 906 }, { "epoch": 5.635514018691588, "grad_norm": 5.03125, "learning_rate": 4.995860637729611e-05, "loss": 0.5213, "step": 907 }, { "epoch": 5.641744548286605, "grad_norm": 3.453125, "learning_rate": 4.9958464536245406e-05, "loss": 0.3488, "step": 908 }, { "epoch": 5.64797507788162, "grad_norm": 4.40625, "learning_rate": 4.9958322452793096e-05, "loss": 0.3977, "step": 909 }, { "epoch": 5.654205607476635, "grad_norm": 3.828125, "learning_rate": 4.9958180126940577e-05, "loss": 0.4696, "step": 910 }, { "epoch": 5.660436137071651, "grad_norm": 5.71875, "learning_rate": 4.9958037558689216e-05, "loss": 0.6562, "step": 911 }, { "epoch": 5.666666666666667, "grad_norm": 4.1875, "learning_rate": 4.995789474804042e-05, "loss": 0.4133, "step": 912 }, { "epoch": 5.672897196261682, "grad_norm": 4.34375, "learning_rate": 4.9957751694995545e-05, "loss": 0.5714, "step": 913 }, { "epoch": 5.679127725856698, "grad_norm": 3.625, "learning_rate": 4.995760839955601e-05, "loss": 0.4018, "step": 914 }, { "epoch": 5.685358255451713, "grad_norm": 4.125, "learning_rate": 4.995746486172318e-05, "loss": 0.4565, "step": 915 }, { "epoch": 5.691588785046729, "grad_norm": 5.375, "learning_rate": 4.9957321081498475e-05, "loss": 0.3611, "step": 916 }, { "epoch": 5.697819314641745, "grad_norm": 4.875, "learning_rate": 4.995717705888327e-05, "loss": 0.3855, "step": 917 }, { "epoch": 5.70404984423676, "grad_norm": 3.828125, "learning_rate": 4.995703279387898e-05, "loss": 0.68, "step": 918 }, { "epoch": 5.710280373831775, "grad_norm": 4.15625, "learning_rate": 4.995688828648699e-05, "loss": 0.7096, "step": 919 }, { "epoch": 5.716510903426792, "grad_norm": 4.09375, "learning_rate": 4.995674353670872e-05, "loss": 0.4275, "step": 920 }, { "epoch": 5.722741433021807, "grad_norm": 4.625, "learning_rate": 4.9956598544545566e-05, "loss": 0.5967, "step": 921 }, { "epoch": 5.728971962616822, "grad_norm": 4.09375, "learning_rate": 4.995645330999893e-05, "loss": 0.3085, "step": 922 }, { "epoch": 5.735202492211838, "grad_norm": 3.703125, "learning_rate": 4.995630783307024e-05, "loss": 0.5602, "step": 923 }, { "epoch": 5.741433021806854, "grad_norm": 3.953125, "learning_rate": 4.995616211376089e-05, "loss": 0.5846, "step": 924 }, { "epoch": 5.747663551401869, "grad_norm": 5.4375, "learning_rate": 4.9956016152072316e-05, "loss": 0.8082, "step": 925 }, { "epoch": 5.753894080996885, "grad_norm": 3.859375, "learning_rate": 4.995586994800592e-05, "loss": 0.3642, "step": 926 }, { "epoch": 5.7601246105919, "grad_norm": 4.03125, "learning_rate": 4.995572350156313e-05, "loss": 0.6573, "step": 927 }, { "epoch": 5.766355140186916, "grad_norm": 3.65625, "learning_rate": 4.9955576812745366e-05, "loss": 0.4611, "step": 928 }, { "epoch": 5.7725856697819315, "grad_norm": 4.28125, "learning_rate": 4.995542988155404e-05, "loss": 0.5372, "step": 929 }, { "epoch": 5.778816199376947, "grad_norm": 3.765625, "learning_rate": 4.99552827079906e-05, "loss": 0.4261, "step": 930 }, { "epoch": 5.785046728971962, "grad_norm": 4.6875, "learning_rate": 4.995513529205647e-05, "loss": 0.7184, "step": 931 }, { "epoch": 5.7912772585669785, "grad_norm": 4.28125, "learning_rate": 4.995498763375306e-05, "loss": 0.5393, "step": 932 }, { "epoch": 5.797507788161994, "grad_norm": 3.078125, "learning_rate": 4.995483973308184e-05, "loss": 0.4515, "step": 933 }, { "epoch": 5.803738317757009, "grad_norm": 3.234375, "learning_rate": 4.9954691590044215e-05, "loss": 0.3811, "step": 934 }, { "epoch": 5.809968847352025, "grad_norm": 4.53125, "learning_rate": 4.9954543204641644e-05, "loss": 0.5241, "step": 935 }, { "epoch": 5.816199376947041, "grad_norm": 4.0, "learning_rate": 4.9954394576875564e-05, "loss": 0.5225, "step": 936 }, { "epoch": 5.822429906542056, "grad_norm": 4.15625, "learning_rate": 4.9954245706747405e-05, "loss": 0.4893, "step": 937 }, { "epoch": 5.8286604361370715, "grad_norm": 3.734375, "learning_rate": 4.9954096594258637e-05, "loss": 0.4085, "step": 938 }, { "epoch": 5.834890965732087, "grad_norm": 3.46875, "learning_rate": 4.995394723941068e-05, "loss": 0.3327, "step": 939 }, { "epoch": 5.841121495327103, "grad_norm": 3.453125, "learning_rate": 4.995379764220501e-05, "loss": 0.3659, "step": 940 }, { "epoch": 5.8473520249221185, "grad_norm": 3.109375, "learning_rate": 4.9953647802643065e-05, "loss": 0.3575, "step": 941 }, { "epoch": 5.853582554517134, "grad_norm": 4.0625, "learning_rate": 4.995349772072631e-05, "loss": 0.479, "step": 942 }, { "epoch": 5.859813084112149, "grad_norm": 4.25, "learning_rate": 4.9953347396456194e-05, "loss": 0.5983, "step": 943 }, { "epoch": 5.866043613707165, "grad_norm": 4.34375, "learning_rate": 4.995319682983418e-05, "loss": 0.4943, "step": 944 }, { "epoch": 5.872274143302181, "grad_norm": 4.1875, "learning_rate": 4.9953046020861724e-05, "loss": 0.3514, "step": 945 }, { "epoch": 5.878504672897196, "grad_norm": 3.375, "learning_rate": 4.9952894969540296e-05, "loss": 0.4107, "step": 946 }, { "epoch": 5.8847352024922115, "grad_norm": 4.96875, "learning_rate": 4.995274367587137e-05, "loss": 0.3171, "step": 947 }, { "epoch": 5.890965732087228, "grad_norm": 7.21875, "learning_rate": 4.9952592139856413e-05, "loss": 0.605, "step": 948 }, { "epoch": 5.897196261682243, "grad_norm": 3.4375, "learning_rate": 4.9952440361496886e-05, "loss": 0.4364, "step": 949 }, { "epoch": 5.9034267912772584, "grad_norm": 3.59375, "learning_rate": 4.995228834079428e-05, "loss": 0.5196, "step": 950 }, { "epoch": 5.909657320872274, "grad_norm": 9.5625, "learning_rate": 4.995213607775005e-05, "loss": 0.4771, "step": 951 }, { "epoch": 5.91588785046729, "grad_norm": 4.96875, "learning_rate": 4.9951983572365694e-05, "loss": 0.4298, "step": 952 }, { "epoch": 5.922118380062305, "grad_norm": 5.84375, "learning_rate": 4.995183082464269e-05, "loss": 0.5363, "step": 953 }, { "epoch": 5.928348909657321, "grad_norm": 3.421875, "learning_rate": 4.9951677834582514e-05, "loss": 0.42, "step": 954 }, { "epoch": 5.934579439252336, "grad_norm": 6.15625, "learning_rate": 4.995152460218665e-05, "loss": 0.3586, "step": 955 }, { "epoch": 5.940809968847352, "grad_norm": 4.90625, "learning_rate": 4.995137112745659e-05, "loss": 0.5767, "step": 956 }, { "epoch": 5.947040498442368, "grad_norm": 4.09375, "learning_rate": 4.995121741039383e-05, "loss": 0.5265, "step": 957 }, { "epoch": 5.953271028037383, "grad_norm": 4.03125, "learning_rate": 4.9951063450999856e-05, "loss": 0.5286, "step": 958 }, { "epoch": 5.959501557632398, "grad_norm": 4.125, "learning_rate": 4.9950909249276166e-05, "loss": 0.508, "step": 959 }, { "epoch": 5.965732087227415, "grad_norm": 4.1875, "learning_rate": 4.9950754805224265e-05, "loss": 0.5642, "step": 960 }, { "epoch": 5.97196261682243, "grad_norm": 3.84375, "learning_rate": 4.9950600118845644e-05, "loss": 0.5373, "step": 961 }, { "epoch": 5.978193146417445, "grad_norm": 3.40625, "learning_rate": 4.9950445190141805e-05, "loss": 0.3396, "step": 962 }, { "epoch": 5.984423676012461, "grad_norm": 4.4375, "learning_rate": 4.9950290019114255e-05, "loss": 0.4789, "step": 963 }, { "epoch": 5.990654205607477, "grad_norm": 8.25, "learning_rate": 4.99501346057645e-05, "loss": 0.4476, "step": 964 }, { "epoch": 5.996884735202492, "grad_norm": 3.765625, "learning_rate": 4.994997895009404e-05, "loss": 0.4666, "step": 965 }, { "epoch": 6.0, "grad_norm": 2.546875, "learning_rate": 4.9949823052104416e-05, "loss": 0.2556, "step": 966 }, { "epoch": 6.006230529595015, "grad_norm": 2.5625, "learning_rate": 4.994966691179711e-05, "loss": 0.2487, "step": 967 }, { "epoch": 6.012461059190031, "grad_norm": 5.0625, "learning_rate": 4.994951052917366e-05, "loss": 0.4049, "step": 968 }, { "epoch": 6.018691588785047, "grad_norm": 3.90625, "learning_rate": 4.9949353904235574e-05, "loss": 0.3957, "step": 969 }, { "epoch": 6.024922118380062, "grad_norm": 4.75, "learning_rate": 4.994919703698438e-05, "loss": 0.4916, "step": 970 }, { "epoch": 6.031152647975078, "grad_norm": 5.3125, "learning_rate": 4.9949039927421584e-05, "loss": 0.3957, "step": 971 }, { "epoch": 6.037383177570093, "grad_norm": 5.46875, "learning_rate": 4.9948882575548726e-05, "loss": 0.4501, "step": 972 }, { "epoch": 6.043613707165109, "grad_norm": 4.21875, "learning_rate": 4.994872498136734e-05, "loss": 0.5637, "step": 973 }, { "epoch": 6.049844236760125, "grad_norm": 3.015625, "learning_rate": 4.9948567144878956e-05, "loss": 0.3254, "step": 974 }, { "epoch": 6.05607476635514, "grad_norm": 6.03125, "learning_rate": 4.9948409066085086e-05, "loss": 0.6274, "step": 975 }, { "epoch": 6.062305295950155, "grad_norm": 4.1875, "learning_rate": 4.994825074498729e-05, "loss": 0.3241, "step": 976 }, { "epoch": 6.068535825545172, "grad_norm": 4.40625, "learning_rate": 4.9948092181587095e-05, "loss": 0.3067, "step": 977 }, { "epoch": 6.074766355140187, "grad_norm": 3.796875, "learning_rate": 4.994793337588605e-05, "loss": 0.4279, "step": 978 }, { "epoch": 6.080996884735202, "grad_norm": 6.125, "learning_rate": 4.994777432788568e-05, "loss": 0.7804, "step": 979 }, { "epoch": 6.087227414330218, "grad_norm": 4.46875, "learning_rate": 4.994761503758753e-05, "loss": 0.4237, "step": 980 }, { "epoch": 6.093457943925234, "grad_norm": 4.78125, "learning_rate": 4.994745550499317e-05, "loss": 0.4527, "step": 981 }, { "epoch": 6.099688473520249, "grad_norm": 6.71875, "learning_rate": 4.994729573010413e-05, "loss": 0.5885, "step": 982 }, { "epoch": 6.105919003115265, "grad_norm": 3.015625, "learning_rate": 4.994713571292196e-05, "loss": 0.4919, "step": 983 }, { "epoch": 6.11214953271028, "grad_norm": 4.875, "learning_rate": 4.994697545344823e-05, "loss": 0.3116, "step": 984 }, { "epoch": 6.118380062305296, "grad_norm": 3.3125, "learning_rate": 4.9946814951684484e-05, "loss": 0.3766, "step": 985 }, { "epoch": 6.1246105919003115, "grad_norm": 7.65625, "learning_rate": 4.994665420763229e-05, "loss": 0.4124, "step": 986 }, { "epoch": 6.130841121495327, "grad_norm": 6.90625, "learning_rate": 4.994649322129319e-05, "loss": 0.4193, "step": 987 }, { "epoch": 6.137071651090342, "grad_norm": 4.65625, "learning_rate": 4.9946331992668775e-05, "loss": 0.4742, "step": 988 }, { "epoch": 6.1433021806853585, "grad_norm": 6.3125, "learning_rate": 4.994617052176059e-05, "loss": 0.6975, "step": 989 }, { "epoch": 6.149532710280374, "grad_norm": 5.9375, "learning_rate": 4.994600880857022e-05, "loss": 0.2893, "step": 990 }, { "epoch": 6.155763239875389, "grad_norm": 5.25, "learning_rate": 4.994584685309921e-05, "loss": 0.2645, "step": 991 }, { "epoch": 6.161993769470405, "grad_norm": 3.375, "learning_rate": 4.994568465534916e-05, "loss": 0.5542, "step": 992 }, { "epoch": 6.168224299065421, "grad_norm": 5.0625, "learning_rate": 4.9945522215321635e-05, "loss": 0.4482, "step": 993 }, { "epoch": 6.174454828660436, "grad_norm": 8.25, "learning_rate": 4.994535953301821e-05, "loss": 0.3642, "step": 994 }, { "epoch": 6.1806853582554515, "grad_norm": 7.875, "learning_rate": 4.9945196608440456e-05, "loss": 0.4384, "step": 995 }, { "epoch": 6.186915887850467, "grad_norm": 3.46875, "learning_rate": 4.994503344158998e-05, "loss": 0.5048, "step": 996 }, { "epoch": 6.193146417445483, "grad_norm": 2.734375, "learning_rate": 4.9944870032468346e-05, "loss": 0.3639, "step": 997 }, { "epoch": 6.1993769470404985, "grad_norm": 6.15625, "learning_rate": 4.9944706381077154e-05, "loss": 0.3666, "step": 998 }, { "epoch": 6.205607476635514, "grad_norm": 7.4375, "learning_rate": 4.994454248741798e-05, "loss": 0.4868, "step": 999 }, { "epoch": 6.211838006230529, "grad_norm": 6.6875, "learning_rate": 4.9944378351492424e-05, "loss": 0.6463, "step": 1000 }, { "epoch": 6.218068535825545, "grad_norm": 5.96875, "learning_rate": 4.9944213973302084e-05, "loss": 0.5097, "step": 1001 }, { "epoch": 6.224299065420561, "grad_norm": 5.03125, "learning_rate": 4.994404935284855e-05, "loss": 0.3176, "step": 1002 }, { "epoch": 6.230529595015576, "grad_norm": 6.3125, "learning_rate": 4.994388449013342e-05, "loss": 0.4336, "step": 1003 }, { "epoch": 6.2367601246105915, "grad_norm": 3.15625, "learning_rate": 4.99437193851583e-05, "loss": 0.4893, "step": 1004 }, { "epoch": 6.242990654205608, "grad_norm": 7.25, "learning_rate": 4.994355403792479e-05, "loss": 0.6182, "step": 1005 }, { "epoch": 6.249221183800623, "grad_norm": 7.65625, "learning_rate": 4.994338844843449e-05, "loss": 0.4564, "step": 1006 }, { "epoch": 6.255451713395638, "grad_norm": 5.84375, "learning_rate": 4.9943222616689025e-05, "loss": 0.6317, "step": 1007 }, { "epoch": 6.261682242990654, "grad_norm": 4.0, "learning_rate": 4.994305654268999e-05, "loss": 0.6705, "step": 1008 }, { "epoch": 6.26791277258567, "grad_norm": 4.625, "learning_rate": 4.994289022643901e-05, "loss": 0.4719, "step": 1009 }, { "epoch": 6.274143302180685, "grad_norm": 4.125, "learning_rate": 4.99427236679377e-05, "loss": 0.3821, "step": 1010 }, { "epoch": 6.280373831775701, "grad_norm": 3.03125, "learning_rate": 4.9942556867187654e-05, "loss": 0.3258, "step": 1011 }, { "epoch": 6.286604361370716, "grad_norm": 5.25, "learning_rate": 4.994238982419052e-05, "loss": 0.5844, "step": 1012 }, { "epoch": 6.292834890965732, "grad_norm": 4.3125, "learning_rate": 4.994222253894791e-05, "loss": 0.4033, "step": 1013 }, { "epoch": 6.299065420560748, "grad_norm": 2.890625, "learning_rate": 4.9942055011461455e-05, "loss": 0.2575, "step": 1014 }, { "epoch": 6.305295950155763, "grad_norm": 2.609375, "learning_rate": 4.994188724173276e-05, "loss": 0.3217, "step": 1015 }, { "epoch": 6.311526479750778, "grad_norm": 4.96875, "learning_rate": 4.994171922976348e-05, "loss": 0.3689, "step": 1016 }, { "epoch": 6.317757009345795, "grad_norm": 6.9375, "learning_rate": 4.994155097555524e-05, "loss": 0.502, "step": 1017 }, { "epoch": 6.32398753894081, "grad_norm": 4.25, "learning_rate": 4.9941382479109665e-05, "loss": 0.3992, "step": 1018 }, { "epoch": 6.330218068535825, "grad_norm": 3.65625, "learning_rate": 4.99412137404284e-05, "loss": 0.3359, "step": 1019 }, { "epoch": 6.336448598130841, "grad_norm": 2.96875, "learning_rate": 4.994104475951308e-05, "loss": 0.3605, "step": 1020 }, { "epoch": 6.342679127725857, "grad_norm": 3.5625, "learning_rate": 4.994087553636535e-05, "loss": 0.3842, "step": 1021 }, { "epoch": 6.348909657320872, "grad_norm": 4.65625, "learning_rate": 4.9940706070986854e-05, "loss": 0.6774, "step": 1022 }, { "epoch": 6.355140186915888, "grad_norm": 3.46875, "learning_rate": 4.9940536363379234e-05, "loss": 0.2911, "step": 1023 }, { "epoch": 6.361370716510903, "grad_norm": 3.65625, "learning_rate": 4.994036641354413e-05, "loss": 0.4656, "step": 1024 }, { "epoch": 6.367601246105919, "grad_norm": 3.703125, "learning_rate": 4.9940196221483214e-05, "loss": 0.3486, "step": 1025 }, { "epoch": 6.373831775700935, "grad_norm": 3.6875, "learning_rate": 4.994002578719812e-05, "loss": 0.5038, "step": 1026 }, { "epoch": 6.38006230529595, "grad_norm": 4.46875, "learning_rate": 4.993985511069051e-05, "loss": 0.5796, "step": 1027 }, { "epoch": 6.386292834890965, "grad_norm": 4.5, "learning_rate": 4.993968419196204e-05, "loss": 0.4349, "step": 1028 }, { "epoch": 6.392523364485982, "grad_norm": 4.84375, "learning_rate": 4.9939513031014376e-05, "loss": 0.3734, "step": 1029 }, { "epoch": 6.398753894080997, "grad_norm": 3.5, "learning_rate": 4.9939341627849176e-05, "loss": 0.4435, "step": 1030 }, { "epoch": 6.404984423676012, "grad_norm": 2.84375, "learning_rate": 4.993916998246811e-05, "loss": 0.3772, "step": 1031 }, { "epoch": 6.411214953271028, "grad_norm": 4.25, "learning_rate": 4.993899809487284e-05, "loss": 0.264, "step": 1032 }, { "epoch": 6.417445482866044, "grad_norm": 3.09375, "learning_rate": 4.993882596506502e-05, "loss": 0.484, "step": 1033 }, { "epoch": 6.423676012461059, "grad_norm": 3.3125, "learning_rate": 4.9938653593046345e-05, "loss": 0.3647, "step": 1034 }, { "epoch": 6.429906542056075, "grad_norm": 3.6875, "learning_rate": 4.993848097881849e-05, "loss": 0.3569, "step": 1035 }, { "epoch": 6.43613707165109, "grad_norm": 3.890625, "learning_rate": 4.993830812238311e-05, "loss": 0.5153, "step": 1036 }, { "epoch": 6.442367601246106, "grad_norm": 3.9375, "learning_rate": 4.9938135023741904e-05, "loss": 0.4638, "step": 1037 }, { "epoch": 6.4485981308411215, "grad_norm": 3.703125, "learning_rate": 4.993796168289654e-05, "loss": 0.3317, "step": 1038 }, { "epoch": 6.454828660436137, "grad_norm": 2.96875, "learning_rate": 4.9937788099848713e-05, "loss": 0.3571, "step": 1039 }, { "epoch": 6.461059190031152, "grad_norm": 4.125, "learning_rate": 4.99376142746001e-05, "loss": 0.4932, "step": 1040 }, { "epoch": 6.4672897196261685, "grad_norm": 3.578125, "learning_rate": 4.99374402071524e-05, "loss": 0.4275, "step": 1041 }, { "epoch": 6.473520249221184, "grad_norm": 2.28125, "learning_rate": 4.993726589750728e-05, "loss": 0.2699, "step": 1042 }, { "epoch": 6.479750778816199, "grad_norm": 3.359375, "learning_rate": 4.993709134566646e-05, "loss": 0.5922, "step": 1043 }, { "epoch": 6.485981308411215, "grad_norm": 7.1875, "learning_rate": 4.993691655163162e-05, "loss": 0.4393, "step": 1044 }, { "epoch": 6.492211838006231, "grad_norm": 4.34375, "learning_rate": 4.993674151540446e-05, "loss": 0.2813, "step": 1045 }, { "epoch": 6.498442367601246, "grad_norm": 3.4375, "learning_rate": 4.993656623698669e-05, "loss": 0.4869, "step": 1046 }, { "epoch": 6.5046728971962615, "grad_norm": 4.59375, "learning_rate": 4.993639071638e-05, "loss": 0.6756, "step": 1047 }, { "epoch": 6.510903426791277, "grad_norm": 4.0625, "learning_rate": 4.993621495358609e-05, "loss": 0.3896, "step": 1048 }, { "epoch": 6.517133956386293, "grad_norm": 3.828125, "learning_rate": 4.993603894860669e-05, "loss": 0.3984, "step": 1049 }, { "epoch": 6.5233644859813085, "grad_norm": 3.84375, "learning_rate": 4.993586270144349e-05, "loss": 0.3634, "step": 1050 }, { "epoch": 6.529595015576324, "grad_norm": 4.03125, "learning_rate": 4.99356862120982e-05, "loss": 0.5371, "step": 1051 }, { "epoch": 6.535825545171339, "grad_norm": 3.65625, "learning_rate": 4.993550948057255e-05, "loss": 0.5111, "step": 1052 }, { "epoch": 6.542056074766355, "grad_norm": 3.109375, "learning_rate": 4.993533250686824e-05, "loss": 0.3913, "step": 1053 }, { "epoch": 6.548286604361371, "grad_norm": 3.75, "learning_rate": 4.9935155290987e-05, "loss": 0.446, "step": 1054 }, { "epoch": 6.554517133956386, "grad_norm": 4.28125, "learning_rate": 4.9934977832930546e-05, "loss": 0.4998, "step": 1055 }, { "epoch": 6.5607476635514015, "grad_norm": 5.21875, "learning_rate": 4.99348001327006e-05, "loss": 0.5589, "step": 1056 }, { "epoch": 6.566978193146418, "grad_norm": 3.53125, "learning_rate": 4.9934622190298896e-05, "loss": 0.5125, "step": 1057 }, { "epoch": 6.573208722741433, "grad_norm": 4.375, "learning_rate": 4.993444400572716e-05, "loss": 0.4332, "step": 1058 }, { "epoch": 6.579439252336448, "grad_norm": 3.40625, "learning_rate": 4.993426557898711e-05, "loss": 0.4533, "step": 1059 }, { "epoch": 6.585669781931464, "grad_norm": 2.71875, "learning_rate": 4.9934086910080505e-05, "loss": 0.3055, "step": 1060 }, { "epoch": 6.59190031152648, "grad_norm": 4.5625, "learning_rate": 4.993390799900904e-05, "loss": 0.4728, "step": 1061 }, { "epoch": 6.598130841121495, "grad_norm": 6.25, "learning_rate": 4.9933728845774496e-05, "loss": 0.6341, "step": 1062 }, { "epoch": 6.604361370716511, "grad_norm": 4.03125, "learning_rate": 4.993354945037858e-05, "loss": 0.4096, "step": 1063 }, { "epoch": 6.610591900311526, "grad_norm": 3.265625, "learning_rate": 4.993336981282306e-05, "loss": 0.3311, "step": 1064 }, { "epoch": 6.616822429906542, "grad_norm": 4.875, "learning_rate": 4.9933189933109657e-05, "loss": 0.5484, "step": 1065 }, { "epoch": 6.623052959501558, "grad_norm": 3.53125, "learning_rate": 4.9933009811240136e-05, "loss": 0.4002, "step": 1066 }, { "epoch": 6.629283489096573, "grad_norm": 4.5625, "learning_rate": 4.993282944721624e-05, "loss": 0.4628, "step": 1067 }, { "epoch": 6.635514018691588, "grad_norm": 3.5, "learning_rate": 4.9932648841039716e-05, "loss": 0.3231, "step": 1068 }, { "epoch": 6.641744548286605, "grad_norm": 4.28125, "learning_rate": 4.993246799271233e-05, "loss": 0.3077, "step": 1069 }, { "epoch": 6.64797507788162, "grad_norm": 3.84375, "learning_rate": 4.9932286902235824e-05, "loss": 0.3891, "step": 1070 }, { "epoch": 6.654205607476635, "grad_norm": 2.828125, "learning_rate": 4.993210556961197e-05, "loss": 0.3203, "step": 1071 }, { "epoch": 6.660436137071651, "grad_norm": 3.734375, "learning_rate": 4.993192399484251e-05, "loss": 0.3305, "step": 1072 }, { "epoch": 6.666666666666667, "grad_norm": 3.484375, "learning_rate": 4.993174217792923e-05, "loss": 0.281, "step": 1073 }, { "epoch": 6.672897196261682, "grad_norm": 4.71875, "learning_rate": 4.993156011887389e-05, "loss": 0.6112, "step": 1074 }, { "epoch": 6.679127725856698, "grad_norm": 5.0625, "learning_rate": 4.9931377817678246e-05, "loss": 0.5132, "step": 1075 }, { "epoch": 6.685358255451713, "grad_norm": 5.1875, "learning_rate": 4.993119527434408e-05, "loss": 0.5102, "step": 1076 }, { "epoch": 6.691588785046729, "grad_norm": 5.0, "learning_rate": 4.993101248887316e-05, "loss": 0.5021, "step": 1077 }, { "epoch": 6.697819314641745, "grad_norm": 7.53125, "learning_rate": 4.993082946126727e-05, "loss": 0.5001, "step": 1078 }, { "epoch": 6.70404984423676, "grad_norm": 3.84375, "learning_rate": 4.9930646191528175e-05, "loss": 0.4475, "step": 1079 }, { "epoch": 6.710280373831775, "grad_norm": 3.0, "learning_rate": 4.993046267965766e-05, "loss": 0.3082, "step": 1080 }, { "epoch": 6.716510903426792, "grad_norm": 7.03125, "learning_rate": 4.993027892565751e-05, "loss": 0.6205, "step": 1081 }, { "epoch": 6.722741433021807, "grad_norm": 6.9375, "learning_rate": 4.9930094929529506e-05, "loss": 0.3602, "step": 1082 }, { "epoch": 6.728971962616822, "grad_norm": 4.84375, "learning_rate": 4.992991069127544e-05, "loss": 0.3406, "step": 1083 }, { "epoch": 6.735202492211838, "grad_norm": 6.71875, "learning_rate": 4.992972621089709e-05, "loss": 0.4045, "step": 1084 }, { "epoch": 6.741433021806854, "grad_norm": 8.25, "learning_rate": 4.9929541488396266e-05, "loss": 0.4651, "step": 1085 }, { "epoch": 6.747663551401869, "grad_norm": 4.46875, "learning_rate": 4.9929356523774745e-05, "loss": 0.3925, "step": 1086 }, { "epoch": 6.753894080996885, "grad_norm": 3.53125, "learning_rate": 4.992917131703433e-05, "loss": 0.4, "step": 1087 }, { "epoch": 6.7601246105919, "grad_norm": 3.828125, "learning_rate": 4.992898586817682e-05, "loss": 0.2424, "step": 1088 }, { "epoch": 6.766355140186916, "grad_norm": 6.84375, "learning_rate": 4.9928800177204025e-05, "loss": 0.4258, "step": 1089 }, { "epoch": 6.7725856697819315, "grad_norm": 4.09375, "learning_rate": 4.9928614244117735e-05, "loss": 0.428, "step": 1090 }, { "epoch": 6.778816199376947, "grad_norm": 4.78125, "learning_rate": 4.9928428068919755e-05, "loss": 0.5986, "step": 1091 }, { "epoch": 6.785046728971962, "grad_norm": 8.375, "learning_rate": 4.99282416516119e-05, "loss": 0.5953, "step": 1092 }, { "epoch": 6.7912772585669785, "grad_norm": 5.34375, "learning_rate": 4.9928054992195985e-05, "loss": 0.4983, "step": 1093 }, { "epoch": 6.797507788161994, "grad_norm": 4.78125, "learning_rate": 4.992786809067381e-05, "loss": 0.5495, "step": 1094 }, { "epoch": 6.803738317757009, "grad_norm": 3.0, "learning_rate": 4.99276809470472e-05, "loss": 0.3395, "step": 1095 }, { "epoch": 6.809968847352025, "grad_norm": 3.578125, "learning_rate": 4.9927493561317974e-05, "loss": 0.2733, "step": 1096 }, { "epoch": 6.816199376947041, "grad_norm": 6.5625, "learning_rate": 4.992730593348794e-05, "loss": 0.5006, "step": 1097 }, { "epoch": 6.822429906542056, "grad_norm": 4.6875, "learning_rate": 4.992711806355893e-05, "loss": 0.4895, "step": 1098 }, { "epoch": 6.8286604361370715, "grad_norm": 4.78125, "learning_rate": 4.992692995153276e-05, "loss": 0.5019, "step": 1099 }, { "epoch": 6.834890965732087, "grad_norm": 7.6875, "learning_rate": 4.992674159741128e-05, "loss": 0.4564, "step": 1100 }, { "epoch": 6.841121495327103, "grad_norm": 4.59375, "learning_rate": 4.992655300119629e-05, "loss": 0.2719, "step": 1101 }, { "epoch": 6.8473520249221185, "grad_norm": 3.34375, "learning_rate": 4.992636416288964e-05, "loss": 0.4087, "step": 1102 }, { "epoch": 6.853582554517134, "grad_norm": 3.140625, "learning_rate": 4.992617508249316e-05, "loss": 0.2165, "step": 1103 }, { "epoch": 6.859813084112149, "grad_norm": 6.78125, "learning_rate": 4.992598576000868e-05, "loss": 0.5951, "step": 1104 }, { "epoch": 6.866043613707165, "grad_norm": 3.375, "learning_rate": 4.9925796195438044e-05, "loss": 0.3582, "step": 1105 }, { "epoch": 6.872274143302181, "grad_norm": 2.59375, "learning_rate": 4.99256063887831e-05, "loss": 0.3054, "step": 1106 }, { "epoch": 6.878504672897196, "grad_norm": 6.5625, "learning_rate": 4.992541634004567e-05, "loss": 0.4324, "step": 1107 }, { "epoch": 6.8847352024922115, "grad_norm": 5.21875, "learning_rate": 4.992522604922762e-05, "loss": 0.2989, "step": 1108 }, { "epoch": 6.890965732087228, "grad_norm": 3.21875, "learning_rate": 4.9925035516330796e-05, "loss": 0.4028, "step": 1109 }, { "epoch": 6.897196261682243, "grad_norm": 5.34375, "learning_rate": 4.992484474135705e-05, "loss": 0.4615, "step": 1110 }, { "epoch": 6.9034267912772584, "grad_norm": 7.84375, "learning_rate": 4.9924653724308225e-05, "loss": 0.4422, "step": 1111 }, { "epoch": 6.909657320872274, "grad_norm": 5.84375, "learning_rate": 4.9924462465186174e-05, "loss": 0.3251, "step": 1112 }, { "epoch": 6.91588785046729, "grad_norm": 3.875, "learning_rate": 4.992427096399277e-05, "loss": 0.3764, "step": 1113 }, { "epoch": 6.922118380062305, "grad_norm": 6.8125, "learning_rate": 4.9924079220729855e-05, "loss": 0.3444, "step": 1114 }, { "epoch": 6.928348909657321, "grad_norm": 6.96875, "learning_rate": 4.992388723539931e-05, "loss": 0.2993, "step": 1115 }, { "epoch": 6.934579439252336, "grad_norm": 7.5625, "learning_rate": 4.992369500800299e-05, "loss": 0.5292, "step": 1116 }, { "epoch": 6.940809968847352, "grad_norm": 3.46875, "learning_rate": 4.992350253854276e-05, "loss": 0.3734, "step": 1117 }, { "epoch": 6.947040498442368, "grad_norm": 6.15625, "learning_rate": 4.992330982702048e-05, "loss": 0.4628, "step": 1118 }, { "epoch": 6.953271028037383, "grad_norm": 5.90625, "learning_rate": 4.992311687343804e-05, "loss": 0.3497, "step": 1119 }, { "epoch": 6.959501557632398, "grad_norm": 8.25, "learning_rate": 4.992292367779731e-05, "loss": 0.3545, "step": 1120 }, { "epoch": 6.965732087227415, "grad_norm": 4.28125, "learning_rate": 4.992273024010016e-05, "loss": 0.3712, "step": 1121 }, { "epoch": 6.97196261682243, "grad_norm": 6.3125, "learning_rate": 4.9922536560348473e-05, "loss": 0.549, "step": 1122 }, { "epoch": 6.978193146417445, "grad_norm": 6.53125, "learning_rate": 4.9922342638544125e-05, "loss": 0.294, "step": 1123 }, { "epoch": 6.984423676012461, "grad_norm": 6.46875, "learning_rate": 4.992214847468901e-05, "loss": 0.5027, "step": 1124 }, { "epoch": 6.990654205607477, "grad_norm": 4.0625, "learning_rate": 4.992195406878501e-05, "loss": 0.4068, "step": 1125 }, { "epoch": 6.996884735202492, "grad_norm": 5.84375, "learning_rate": 4.9921759420834e-05, "loss": 0.299, "step": 1126 }, { "epoch": 7.0, "grad_norm": 4.9375, "learning_rate": 4.9921564530837894e-05, "loss": 0.3624, "step": 1127 }, { "epoch": 7.006230529595015, "grad_norm": 7.46875, "learning_rate": 4.992136939879856e-05, "loss": 0.5005, "step": 1128 }, { "epoch": 7.012461059190031, "grad_norm": 3.234375, "learning_rate": 4.992117402471791e-05, "loss": 0.4705, "step": 1129 }, { "epoch": 7.018691588785047, "grad_norm": 3.921875, "learning_rate": 4.9920978408597837e-05, "loss": 0.3865, "step": 1130 }, { "epoch": 7.024922118380062, "grad_norm": 6.46875, "learning_rate": 4.9920782550440236e-05, "loss": 0.5202, "step": 1131 }, { "epoch": 7.031152647975078, "grad_norm": 4.5, "learning_rate": 4.992058645024702e-05, "loss": 0.2611, "step": 1132 }, { "epoch": 7.037383177570093, "grad_norm": 4.125, "learning_rate": 4.9920390108020075e-05, "loss": 0.3743, "step": 1133 }, { "epoch": 7.043613707165109, "grad_norm": 2.4375, "learning_rate": 4.992019352376133e-05, "loss": 0.2646, "step": 1134 }, { "epoch": 7.049844236760125, "grad_norm": 5.28125, "learning_rate": 4.9919996697472684e-05, "loss": 0.3622, "step": 1135 }, { "epoch": 7.05607476635514, "grad_norm": 4.5625, "learning_rate": 4.9919799629156036e-05, "loss": 0.4537, "step": 1136 }, { "epoch": 7.062305295950155, "grad_norm": 4.46875, "learning_rate": 4.991960231881333e-05, "loss": 0.4496, "step": 1137 }, { "epoch": 7.068535825545172, "grad_norm": 5.71875, "learning_rate": 4.9919404766446456e-05, "loss": 0.3433, "step": 1138 }, { "epoch": 7.074766355140187, "grad_norm": 3.96875, "learning_rate": 4.991920697205734e-05, "loss": 0.3525, "step": 1139 }, { "epoch": 7.080996884735202, "grad_norm": 3.578125, "learning_rate": 4.9919008935647904e-05, "loss": 0.512, "step": 1140 }, { "epoch": 7.087227414330218, "grad_norm": 3.421875, "learning_rate": 4.9918810657220075e-05, "loss": 0.366, "step": 1141 }, { "epoch": 7.093457943925234, "grad_norm": 3.859375, "learning_rate": 4.9918612136775776e-05, "loss": 0.2886, "step": 1142 }, { "epoch": 7.099688473520249, "grad_norm": 3.71875, "learning_rate": 4.991841337431694e-05, "loss": 0.4487, "step": 1143 }, { "epoch": 7.105919003115265, "grad_norm": 3.53125, "learning_rate": 4.991821436984548e-05, "loss": 0.4561, "step": 1144 }, { "epoch": 7.11214953271028, "grad_norm": 4.71875, "learning_rate": 4.9918015123363355e-05, "loss": 0.5704, "step": 1145 }, { "epoch": 7.118380062305296, "grad_norm": 4.59375, "learning_rate": 4.991781563487248e-05, "loss": 0.5129, "step": 1146 }, { "epoch": 7.1246105919003115, "grad_norm": 3.359375, "learning_rate": 4.991761590437479e-05, "loss": 0.2844, "step": 1147 }, { "epoch": 7.130841121495327, "grad_norm": 3.4375, "learning_rate": 4.991741593187225e-05, "loss": 0.5552, "step": 1148 }, { "epoch": 7.137071651090342, "grad_norm": 3.765625, "learning_rate": 4.991721571736677e-05, "loss": 0.5532, "step": 1149 }, { "epoch": 7.1433021806853585, "grad_norm": 5.65625, "learning_rate": 4.991701526086032e-05, "loss": 0.2973, "step": 1150 }, { "epoch": 7.149532710280374, "grad_norm": 3.90625, "learning_rate": 4.991681456235483e-05, "loss": 0.3134, "step": 1151 }, { "epoch": 7.155763239875389, "grad_norm": 3.609375, "learning_rate": 4.9916613621852265e-05, "loss": 0.3036, "step": 1152 }, { "epoch": 7.161993769470405, "grad_norm": 6.8125, "learning_rate": 4.991641243935457e-05, "loss": 0.3793, "step": 1153 }, { "epoch": 7.168224299065421, "grad_norm": 6.15625, "learning_rate": 4.991621101486368e-05, "loss": 0.439, "step": 1154 }, { "epoch": 7.174454828660436, "grad_norm": 6.03125, "learning_rate": 4.9916009348381574e-05, "loss": 0.4148, "step": 1155 }, { "epoch": 7.1806853582554515, "grad_norm": 4.9375, "learning_rate": 4.9915807439910213e-05, "loss": 0.481, "step": 1156 }, { "epoch": 7.186915887850467, "grad_norm": 8.3125, "learning_rate": 4.991560528945154e-05, "loss": 0.4532, "step": 1157 }, { "epoch": 7.193146417445483, "grad_norm": 7.375, "learning_rate": 4.991540289700754e-05, "loss": 0.4857, "step": 1158 }, { "epoch": 7.1993769470404985, "grad_norm": 4.90625, "learning_rate": 4.991520026258015e-05, "loss": 0.387, "step": 1159 }, { "epoch": 7.205607476635514, "grad_norm": 3.0, "learning_rate": 4.991499738617137e-05, "loss": 0.2769, "step": 1160 }, { "epoch": 7.211838006230529, "grad_norm": 4.78125, "learning_rate": 4.9914794267783146e-05, "loss": 0.3248, "step": 1161 }, { "epoch": 7.218068535825545, "grad_norm": 5.1875, "learning_rate": 4.991459090741746e-05, "loss": 0.436, "step": 1162 }, { "epoch": 7.224299065420561, "grad_norm": 1.8671875, "learning_rate": 4.991438730507629e-05, "loss": 0.2429, "step": 1163 }, { "epoch": 7.230529595015576, "grad_norm": 3.328125, "learning_rate": 4.991418346076161e-05, "loss": 0.3676, "step": 1164 }, { "epoch": 7.2367601246105915, "grad_norm": 6.625, "learning_rate": 4.991397937447541e-05, "loss": 0.4761, "step": 1165 }, { "epoch": 7.242990654205608, "grad_norm": 5.59375, "learning_rate": 4.991377504621965e-05, "loss": 0.6998, "step": 1166 }, { "epoch": 7.249221183800623, "grad_norm": 4.21875, "learning_rate": 4.991357047599632e-05, "loss": 0.4308, "step": 1167 }, { "epoch": 7.255451713395638, "grad_norm": 4.84375, "learning_rate": 4.991336566380742e-05, "loss": 0.4071, "step": 1168 }, { "epoch": 7.261682242990654, "grad_norm": 5.9375, "learning_rate": 4.991316060965494e-05, "loss": 0.3881, "step": 1169 }, { "epoch": 7.26791277258567, "grad_norm": 4.90625, "learning_rate": 4.991295531354085e-05, "loss": 0.4372, "step": 1170 }, { "epoch": 7.274143302180685, "grad_norm": 3.328125, "learning_rate": 4.991274977546717e-05, "loss": 0.424, "step": 1171 }, { "epoch": 7.280373831775701, "grad_norm": 8.3125, "learning_rate": 4.9912543995435875e-05, "loss": 0.4042, "step": 1172 }, { "epoch": 7.286604361370716, "grad_norm": 6.8125, "learning_rate": 4.9912337973448984e-05, "loss": 0.3438, "step": 1173 }, { "epoch": 7.292834890965732, "grad_norm": 9.25, "learning_rate": 4.991213170950848e-05, "loss": 0.4577, "step": 1174 }, { "epoch": 7.299065420560748, "grad_norm": 2.90625, "learning_rate": 4.9911925203616374e-05, "loss": 0.3981, "step": 1175 }, { "epoch": 7.305295950155763, "grad_norm": 3.5625, "learning_rate": 4.991171845577467e-05, "loss": 0.4004, "step": 1176 }, { "epoch": 7.311526479750778, "grad_norm": 7.9375, "learning_rate": 4.9911511465985386e-05, "loss": 0.6137, "step": 1177 }, { "epoch": 7.317757009345795, "grad_norm": 3.96875, "learning_rate": 4.9911304234250514e-05, "loss": 0.3317, "step": 1178 }, { "epoch": 7.32398753894081, "grad_norm": 4.21875, "learning_rate": 4.9911096760572066e-05, "loss": 0.4097, "step": 1179 }, { "epoch": 7.330218068535825, "grad_norm": 4.84375, "learning_rate": 4.991088904495208e-05, "loss": 0.424, "step": 1180 }, { "epoch": 7.336448598130841, "grad_norm": 5.90625, "learning_rate": 4.9910681087392554e-05, "loss": 0.3834, "step": 1181 }, { "epoch": 7.342679127725857, "grad_norm": 2.734375, "learning_rate": 4.9910472887895524e-05, "loss": 0.3285, "step": 1182 }, { "epoch": 7.348909657320872, "grad_norm": 3.578125, "learning_rate": 4.9910264446462986e-05, "loss": 0.6835, "step": 1183 }, { "epoch": 7.355140186915888, "grad_norm": 3.234375, "learning_rate": 4.991005576309699e-05, "loss": 0.3767, "step": 1184 }, { "epoch": 7.361370716510903, "grad_norm": 3.90625, "learning_rate": 4.990984683779955e-05, "loss": 0.4639, "step": 1185 }, { "epoch": 7.367601246105919, "grad_norm": 3.859375, "learning_rate": 4.99096376705727e-05, "loss": 0.595, "step": 1186 }, { "epoch": 7.373831775700935, "grad_norm": 4.78125, "learning_rate": 4.990942826141847e-05, "loss": 0.3647, "step": 1187 }, { "epoch": 7.38006230529595, "grad_norm": 4.90625, "learning_rate": 4.990921861033889e-05, "loss": 0.31, "step": 1188 }, { "epoch": 7.386292834890965, "grad_norm": 3.328125, "learning_rate": 4.9909008717336e-05, "loss": 0.4912, "step": 1189 }, { "epoch": 7.392523364485982, "grad_norm": 3.359375, "learning_rate": 4.9908798582411845e-05, "loss": 0.3261, "step": 1190 }, { "epoch": 7.398753894080997, "grad_norm": 4.1875, "learning_rate": 4.990858820556845e-05, "loss": 0.2408, "step": 1191 }, { "epoch": 7.404984423676012, "grad_norm": 5.125, "learning_rate": 4.990837758680787e-05, "loss": 0.576, "step": 1192 }, { "epoch": 7.411214953271028, "grad_norm": 4.3125, "learning_rate": 4.990816672613216e-05, "loss": 0.4293, "step": 1193 }, { "epoch": 7.417445482866044, "grad_norm": 4.15625, "learning_rate": 4.990795562354333e-05, "loss": 0.4167, "step": 1194 }, { "epoch": 7.423676012461059, "grad_norm": 2.734375, "learning_rate": 4.9907744279043475e-05, "loss": 0.317, "step": 1195 }, { "epoch": 7.429906542056075, "grad_norm": 6.03125, "learning_rate": 4.990753269263463e-05, "loss": 0.4487, "step": 1196 }, { "epoch": 7.43613707165109, "grad_norm": 5.375, "learning_rate": 4.990732086431884e-05, "loss": 0.4674, "step": 1197 }, { "epoch": 7.442367601246106, "grad_norm": 4.21875, "learning_rate": 4.9907108794098176e-05, "loss": 0.5019, "step": 1198 }, { "epoch": 7.4485981308411215, "grad_norm": 3.5625, "learning_rate": 4.990689648197469e-05, "loss": 0.39, "step": 1199 }, { "epoch": 7.454828660436137, "grad_norm": 5.96875, "learning_rate": 4.9906683927950446e-05, "loss": 0.3816, "step": 1200 }, { "epoch": 7.461059190031152, "grad_norm": 4.625, "learning_rate": 4.990647113202751e-05, "loss": 0.4683, "step": 1201 }, { "epoch": 7.4672897196261685, "grad_norm": 4.96875, "learning_rate": 4.9906258094207945e-05, "loss": 0.5677, "step": 1202 }, { "epoch": 7.473520249221184, "grad_norm": 2.96875, "learning_rate": 4.9906044814493824e-05, "loss": 0.243, "step": 1203 }, { "epoch": 7.479750778816199, "grad_norm": 4.90625, "learning_rate": 4.990583129288722e-05, "loss": 0.3101, "step": 1204 }, { "epoch": 7.485981308411215, "grad_norm": 5.375, "learning_rate": 4.9905617529390203e-05, "loss": 0.433, "step": 1205 }, { "epoch": 7.492211838006231, "grad_norm": 5.8125, "learning_rate": 4.9905403524004844e-05, "loss": 0.4636, "step": 1206 }, { "epoch": 7.498442367601246, "grad_norm": 5.96875, "learning_rate": 4.990518927673324e-05, "loss": 0.4162, "step": 1207 }, { "epoch": 7.5046728971962615, "grad_norm": 7.25, "learning_rate": 4.9904974787577455e-05, "loss": 0.4913, "step": 1208 }, { "epoch": 7.510903426791277, "grad_norm": 6.1875, "learning_rate": 4.9904760056539565e-05, "loss": 0.3615, "step": 1209 }, { "epoch": 7.517133956386293, "grad_norm": 5.21875, "learning_rate": 4.990454508362168e-05, "loss": 0.7119, "step": 1210 }, { "epoch": 7.5233644859813085, "grad_norm": 3.703125, "learning_rate": 4.9904329868825874e-05, "loss": 0.4234, "step": 1211 }, { "epoch": 7.529595015576324, "grad_norm": 3.0, "learning_rate": 4.990411441215423e-05, "loss": 0.3564, "step": 1212 }, { "epoch": 7.535825545171339, "grad_norm": 5.9375, "learning_rate": 4.9903898713608856e-05, "loss": 0.6968, "step": 1213 }, { "epoch": 7.542056074766355, "grad_norm": 6.0625, "learning_rate": 4.990368277319183e-05, "loss": 0.2672, "step": 1214 }, { "epoch": 7.548286604361371, "grad_norm": 6.75, "learning_rate": 4.990346659090527e-05, "loss": 0.4244, "step": 1215 }, { "epoch": 7.554517133956386, "grad_norm": 6.03125, "learning_rate": 4.990325016675126e-05, "loss": 0.5732, "step": 1216 }, { "epoch": 7.5607476635514015, "grad_norm": 6.28125, "learning_rate": 4.9903033500731906e-05, "loss": 0.3504, "step": 1217 }, { "epoch": 7.566978193146418, "grad_norm": 6.46875, "learning_rate": 4.990281659284931e-05, "loss": 0.4776, "step": 1218 }, { "epoch": 7.573208722741433, "grad_norm": 4.84375, "learning_rate": 4.9902599443105584e-05, "loss": 0.5786, "step": 1219 }, { "epoch": 7.579439252336448, "grad_norm": 3.5625, "learning_rate": 4.990238205150284e-05, "loss": 0.2676, "step": 1220 }, { "epoch": 7.585669781931464, "grad_norm": 5.46875, "learning_rate": 4.990216441804318e-05, "loss": 0.7115, "step": 1221 }, { "epoch": 7.59190031152648, "grad_norm": 5.65625, "learning_rate": 4.9901946542728714e-05, "loss": 0.2844, "step": 1222 }, { "epoch": 7.598130841121495, "grad_norm": 8.8125, "learning_rate": 4.9901728425561564e-05, "loss": 0.5509, "step": 1223 }, { "epoch": 7.604361370716511, "grad_norm": 4.90625, "learning_rate": 4.990151006654386e-05, "loss": 0.5199, "step": 1224 }, { "epoch": 7.610591900311526, "grad_norm": 4.65625, "learning_rate": 4.9901291465677706e-05, "loss": 0.5587, "step": 1225 }, { "epoch": 7.616822429906542, "grad_norm": 4.25, "learning_rate": 4.990107262296524e-05, "loss": 0.3305, "step": 1226 }, { "epoch": 7.623052959501558, "grad_norm": 3.46875, "learning_rate": 4.990085353840857e-05, "loss": 0.307, "step": 1227 }, { "epoch": 7.629283489096573, "grad_norm": 4.125, "learning_rate": 4.9900634212009834e-05, "loss": 0.3406, "step": 1228 }, { "epoch": 7.635514018691588, "grad_norm": 5.71875, "learning_rate": 4.990041464377117e-05, "loss": 0.7568, "step": 1229 }, { "epoch": 7.641744548286605, "grad_norm": 4.75, "learning_rate": 4.990019483369469e-05, "loss": 0.3004, "step": 1230 }, { "epoch": 7.64797507788162, "grad_norm": 4.78125, "learning_rate": 4.9899974781782546e-05, "loss": 0.3819, "step": 1231 }, { "epoch": 7.654205607476635, "grad_norm": 3.21875, "learning_rate": 4.9899754488036864e-05, "loss": 0.3096, "step": 1232 }, { "epoch": 7.660436137071651, "grad_norm": 3.71875, "learning_rate": 4.98995339524598e-05, "loss": 0.3275, "step": 1233 }, { "epoch": 7.666666666666667, "grad_norm": 4.03125, "learning_rate": 4.9899313175053474e-05, "loss": 0.3751, "step": 1234 }, { "epoch": 7.672897196261682, "grad_norm": 2.84375, "learning_rate": 4.989909215582005e-05, "loss": 0.2836, "step": 1235 }, { "epoch": 7.679127725856698, "grad_norm": 3.609375, "learning_rate": 4.9898870894761664e-05, "loss": 0.5198, "step": 1236 }, { "epoch": 7.685358255451713, "grad_norm": 2.953125, "learning_rate": 4.989864939188046e-05, "loss": 0.3102, "step": 1237 }, { "epoch": 7.691588785046729, "grad_norm": 3.21875, "learning_rate": 4.9898427647178604e-05, "loss": 0.2529, "step": 1238 }, { "epoch": 7.697819314641745, "grad_norm": 4.03125, "learning_rate": 4.989820566065824e-05, "loss": 0.5137, "step": 1239 }, { "epoch": 7.70404984423676, "grad_norm": 3.234375, "learning_rate": 4.989798343232151e-05, "loss": 0.3336, "step": 1240 }, { "epoch": 7.710280373831775, "grad_norm": 2.984375, "learning_rate": 4.989776096217061e-05, "loss": 0.3, "step": 1241 }, { "epoch": 7.716510903426792, "grad_norm": 3.078125, "learning_rate": 4.989753825020766e-05, "loss": 0.3739, "step": 1242 }, { "epoch": 7.722741433021807, "grad_norm": 4.875, "learning_rate": 4.989731529643486e-05, "loss": 0.6294, "step": 1243 }, { "epoch": 7.728971962616822, "grad_norm": 4.5, "learning_rate": 4.989709210085435e-05, "loss": 0.4821, "step": 1244 }, { "epoch": 7.735202492211838, "grad_norm": 3.3125, "learning_rate": 4.9896868663468296e-05, "loss": 0.36, "step": 1245 }, { "epoch": 7.741433021806854, "grad_norm": 4.40625, "learning_rate": 4.989664498427889e-05, "loss": 0.4021, "step": 1246 }, { "epoch": 7.747663551401869, "grad_norm": 3.765625, "learning_rate": 4.9896421063288286e-05, "loss": 0.281, "step": 1247 }, { "epoch": 7.753894080996885, "grad_norm": 3.421875, "learning_rate": 4.989619690049866e-05, "loss": 0.4404, "step": 1248 }, { "epoch": 7.7601246105919, "grad_norm": 3.890625, "learning_rate": 4.98959724959122e-05, "loss": 0.3461, "step": 1249 }, { "epoch": 7.766355140186916, "grad_norm": 4.0, "learning_rate": 4.989574784953107e-05, "loss": 0.3134, "step": 1250 }, { "epoch": 7.7725856697819315, "grad_norm": 4.3125, "learning_rate": 4.989552296135747e-05, "loss": 0.4651, "step": 1251 }, { "epoch": 7.778816199376947, "grad_norm": 3.390625, "learning_rate": 4.989529783139356e-05, "loss": 0.3497, "step": 1252 }, { "epoch": 7.785046728971962, "grad_norm": 5.375, "learning_rate": 4.9895072459641554e-05, "loss": 0.2959, "step": 1253 }, { "epoch": 7.7912772585669785, "grad_norm": 4.09375, "learning_rate": 4.9894846846103625e-05, "loss": 0.353, "step": 1254 }, { "epoch": 7.797507788161994, "grad_norm": 2.546875, "learning_rate": 4.989462099078197e-05, "loss": 0.2935, "step": 1255 }, { "epoch": 7.803738317757009, "grad_norm": 3.5625, "learning_rate": 4.989439489367877e-05, "loss": 0.317, "step": 1256 }, { "epoch": 7.809968847352025, "grad_norm": 3.921875, "learning_rate": 4.989416855479624e-05, "loss": 0.3485, "step": 1257 }, { "epoch": 7.816199376947041, "grad_norm": 5.625, "learning_rate": 4.989394197413656e-05, "loss": 0.3505, "step": 1258 }, { "epoch": 7.822429906542056, "grad_norm": 4.09375, "learning_rate": 4.989371515170195e-05, "loss": 0.3131, "step": 1259 }, { "epoch": 7.8286604361370715, "grad_norm": 2.796875, "learning_rate": 4.989348808749459e-05, "loss": 0.2893, "step": 1260 }, { "epoch": 7.834890965732087, "grad_norm": 5.0625, "learning_rate": 4.989326078151671e-05, "loss": 0.5849, "step": 1261 }, { "epoch": 7.841121495327103, "grad_norm": 4.78125, "learning_rate": 4.98930332337705e-05, "loss": 0.2642, "step": 1262 }, { "epoch": 7.8473520249221185, "grad_norm": 6.875, "learning_rate": 4.9892805444258174e-05, "loss": 0.4094, "step": 1263 }, { "epoch": 7.853582554517134, "grad_norm": 3.859375, "learning_rate": 4.9892577412981946e-05, "loss": 0.271, "step": 1264 }, { "epoch": 7.859813084112149, "grad_norm": 4.28125, "learning_rate": 4.989234913994403e-05, "loss": 0.4152, "step": 1265 }, { "epoch": 7.866043613707165, "grad_norm": 6.0625, "learning_rate": 4.989212062514664e-05, "loss": 0.3011, "step": 1266 }, { "epoch": 7.872274143302181, "grad_norm": 3.921875, "learning_rate": 4.989189186859201e-05, "loss": 0.3179, "step": 1267 }, { "epoch": 7.878504672897196, "grad_norm": 4.46875, "learning_rate": 4.989166287028234e-05, "loss": 0.3416, "step": 1268 }, { "epoch": 7.8847352024922115, "grad_norm": 2.515625, "learning_rate": 4.989143363021987e-05, "loss": 0.2686, "step": 1269 }, { "epoch": 7.890965732087228, "grad_norm": 4.375, "learning_rate": 4.989120414840681e-05, "loss": 0.3523, "step": 1270 }, { "epoch": 7.897196261682243, "grad_norm": 3.53125, "learning_rate": 4.989097442484542e-05, "loss": 0.4397, "step": 1271 }, { "epoch": 7.9034267912772584, "grad_norm": 5.5, "learning_rate": 4.989074445953789e-05, "loss": 0.3698, "step": 1272 }, { "epoch": 7.909657320872274, "grad_norm": 3.328125, "learning_rate": 4.9890514252486486e-05, "loss": 0.2742, "step": 1273 }, { "epoch": 7.91588785046729, "grad_norm": 5.21875, "learning_rate": 4.989028380369343e-05, "loss": 0.4423, "step": 1274 }, { "epoch": 7.922118380062305, "grad_norm": 5.46875, "learning_rate": 4.989005311316096e-05, "loss": 0.5309, "step": 1275 }, { "epoch": 7.928348909657321, "grad_norm": 3.75, "learning_rate": 4.9889822180891324e-05, "loss": 0.4353, "step": 1276 }, { "epoch": 7.934579439252336, "grad_norm": 3.609375, "learning_rate": 4.988959100688675e-05, "loss": 0.2706, "step": 1277 }, { "epoch": 7.940809968847352, "grad_norm": 5.28125, "learning_rate": 4.9889359591149496e-05, "loss": 0.3171, "step": 1278 }, { "epoch": 7.947040498442368, "grad_norm": 3.296875, "learning_rate": 4.988912793368181e-05, "loss": 0.3996, "step": 1279 }, { "epoch": 7.953271028037383, "grad_norm": 3.171875, "learning_rate": 4.9888896034485935e-05, "loss": 0.4366, "step": 1280 }, { "epoch": 7.959501557632398, "grad_norm": 5.6875, "learning_rate": 4.988866389356413e-05, "loss": 0.5389, "step": 1281 }, { "epoch": 7.965732087227415, "grad_norm": 4.59375, "learning_rate": 4.9888431510918645e-05, "loss": 0.3711, "step": 1282 }, { "epoch": 7.97196261682243, "grad_norm": 3.28125, "learning_rate": 4.988819888655174e-05, "loss": 0.4037, "step": 1283 }, { "epoch": 7.978193146417445, "grad_norm": 5.25, "learning_rate": 4.988796602046567e-05, "loss": 0.3255, "step": 1284 }, { "epoch": 7.984423676012461, "grad_norm": 3.6875, "learning_rate": 4.98877329126627e-05, "loss": 0.3119, "step": 1285 }, { "epoch": 7.990654205607477, "grad_norm": 4.21875, "learning_rate": 4.988749956314509e-05, "loss": 0.4487, "step": 1286 }, { "epoch": 7.996884735202492, "grad_norm": 3.109375, "learning_rate": 4.988726597191511e-05, "loss": 0.3174, "step": 1287 }, { "epoch": 8.0, "grad_norm": 3.28125, "learning_rate": 4.988703213897503e-05, "loss": 0.366, "step": 1288 }, { "epoch": 8.006230529595015, "grad_norm": 3.453125, "learning_rate": 4.988679806432712e-05, "loss": 0.3347, "step": 1289 }, { "epoch": 8.01246105919003, "grad_norm": 4.65625, "learning_rate": 4.988656374797365e-05, "loss": 0.4169, "step": 1290 }, { "epoch": 8.018691588785046, "grad_norm": 3.640625, "learning_rate": 4.98863291899169e-05, "loss": 0.4884, "step": 1291 }, { "epoch": 8.024922118380061, "grad_norm": 3.203125, "learning_rate": 4.9886094390159144e-05, "loss": 0.3929, "step": 1292 }, { "epoch": 8.031152647975079, "grad_norm": 3.53125, "learning_rate": 4.9885859348702665e-05, "loss": 0.4501, "step": 1293 }, { "epoch": 8.037383177570094, "grad_norm": 3.015625, "learning_rate": 4.9885624065549747e-05, "loss": 0.3085, "step": 1294 }, { "epoch": 8.04361370716511, "grad_norm": 4.34375, "learning_rate": 4.988538854070267e-05, "loss": 0.486, "step": 1295 }, { "epoch": 8.049844236760125, "grad_norm": 3.78125, "learning_rate": 4.988515277416373e-05, "loss": 0.4864, "step": 1296 }, { "epoch": 8.05607476635514, "grad_norm": 4.40625, "learning_rate": 4.9884916765935205e-05, "loss": 0.39, "step": 1297 }, { "epoch": 8.062305295950155, "grad_norm": 3.96875, "learning_rate": 4.98846805160194e-05, "loss": 0.2917, "step": 1298 }, { "epoch": 8.06853582554517, "grad_norm": 3.125, "learning_rate": 4.988444402441861e-05, "loss": 0.2848, "step": 1299 }, { "epoch": 8.074766355140186, "grad_norm": 3.890625, "learning_rate": 4.988420729113511e-05, "loss": 0.3825, "step": 1300 }, { "epoch": 8.080996884735203, "grad_norm": 4.03125, "learning_rate": 4.9883970316171226e-05, "loss": 0.5276, "step": 1301 }, { "epoch": 8.087227414330219, "grad_norm": 3.59375, "learning_rate": 4.988373309952924e-05, "loss": 0.4851, "step": 1302 }, { "epoch": 8.093457943925234, "grad_norm": 4.84375, "learning_rate": 4.988349564121147e-05, "loss": 0.4123, "step": 1303 }, { "epoch": 8.09968847352025, "grad_norm": 4.375, "learning_rate": 4.988325794122022e-05, "loss": 0.3266, "step": 1304 }, { "epoch": 8.105919003115265, "grad_norm": 2.140625, "learning_rate": 4.9883019999557786e-05, "loss": 0.2324, "step": 1305 }, { "epoch": 8.11214953271028, "grad_norm": 3.765625, "learning_rate": 4.9882781816226485e-05, "loss": 0.4671, "step": 1306 }, { "epoch": 8.118380062305295, "grad_norm": 4.09375, "learning_rate": 4.988254339122864e-05, "loss": 0.3285, "step": 1307 }, { "epoch": 8.12461059190031, "grad_norm": 2.90625, "learning_rate": 4.9882304724566565e-05, "loss": 0.465, "step": 1308 }, { "epoch": 8.130841121495328, "grad_norm": 4.3125, "learning_rate": 4.988206581624256e-05, "loss": 0.5063, "step": 1309 }, { "epoch": 8.137071651090343, "grad_norm": 4.3125, "learning_rate": 4.988182666625897e-05, "loss": 0.4262, "step": 1310 }, { "epoch": 8.143302180685358, "grad_norm": 3.90625, "learning_rate": 4.988158727461809e-05, "loss": 0.2515, "step": 1311 }, { "epoch": 8.149532710280374, "grad_norm": 3.578125, "learning_rate": 4.9881347641322277e-05, "loss": 0.4284, "step": 1312 }, { "epoch": 8.15576323987539, "grad_norm": 3.25, "learning_rate": 4.988110776637383e-05, "loss": 0.2837, "step": 1313 }, { "epoch": 8.161993769470405, "grad_norm": 5.21875, "learning_rate": 4.9880867649775096e-05, "loss": 0.2663, "step": 1314 }, { "epoch": 8.16822429906542, "grad_norm": 4.8125, "learning_rate": 4.9880627291528403e-05, "loss": 0.3982, "step": 1315 }, { "epoch": 8.174454828660435, "grad_norm": 3.109375, "learning_rate": 4.9880386691636086e-05, "loss": 0.2316, "step": 1316 }, { "epoch": 8.180685358255452, "grad_norm": 2.40625, "learning_rate": 4.988014585010048e-05, "loss": 0.3048, "step": 1317 }, { "epoch": 8.186915887850468, "grad_norm": 4.6875, "learning_rate": 4.987990476692392e-05, "loss": 0.3945, "step": 1318 }, { "epoch": 8.193146417445483, "grad_norm": 5.6875, "learning_rate": 4.987966344210875e-05, "loss": 0.558, "step": 1319 }, { "epoch": 8.199376947040498, "grad_norm": 5.96875, "learning_rate": 4.9879421875657314e-05, "loss": 0.4607, "step": 1320 }, { "epoch": 8.205607476635514, "grad_norm": 4.4375, "learning_rate": 4.9879180067571966e-05, "loss": 0.4133, "step": 1321 }, { "epoch": 8.21183800623053, "grad_norm": 4.40625, "learning_rate": 4.9878938017855046e-05, "loss": 0.3801, "step": 1322 }, { "epoch": 8.218068535825545, "grad_norm": 5.3125, "learning_rate": 4.987869572650891e-05, "loss": 0.613, "step": 1323 }, { "epoch": 8.22429906542056, "grad_norm": 4.59375, "learning_rate": 4.98784531935359e-05, "loss": 0.4078, "step": 1324 }, { "epoch": 8.230529595015577, "grad_norm": 3.40625, "learning_rate": 4.987821041893839e-05, "loss": 0.4506, "step": 1325 }, { "epoch": 8.236760124610592, "grad_norm": 3.1875, "learning_rate": 4.987796740271872e-05, "loss": 0.3616, "step": 1326 }, { "epoch": 8.242990654205608, "grad_norm": 4.875, "learning_rate": 4.987772414487926e-05, "loss": 0.6186, "step": 1327 }, { "epoch": 8.249221183800623, "grad_norm": 4.875, "learning_rate": 4.9877480645422365e-05, "loss": 0.3585, "step": 1328 }, { "epoch": 8.255451713395638, "grad_norm": 3.390625, "learning_rate": 4.987723690435041e-05, "loss": 0.2844, "step": 1329 }, { "epoch": 8.261682242990654, "grad_norm": 3.390625, "learning_rate": 4.9876992921665754e-05, "loss": 0.404, "step": 1330 }, { "epoch": 8.26791277258567, "grad_norm": 3.6875, "learning_rate": 4.987674869737077e-05, "loss": 0.3388, "step": 1331 }, { "epoch": 8.274143302180685, "grad_norm": 3.484375, "learning_rate": 4.987650423146783e-05, "loss": 0.3457, "step": 1332 }, { "epoch": 8.280373831775702, "grad_norm": 3.3125, "learning_rate": 4.987625952395931e-05, "loss": 0.2975, "step": 1333 }, { "epoch": 8.286604361370717, "grad_norm": 3.265625, "learning_rate": 4.987601457484758e-05, "loss": 0.519, "step": 1334 }, { "epoch": 8.292834890965732, "grad_norm": 3.203125, "learning_rate": 4.987576938413504e-05, "loss": 0.2646, "step": 1335 }, { "epoch": 8.299065420560748, "grad_norm": 3.765625, "learning_rate": 4.9875523951824045e-05, "loss": 0.5175, "step": 1336 }, { "epoch": 8.305295950155763, "grad_norm": 5.0, "learning_rate": 4.987527827791698e-05, "loss": 0.5623, "step": 1337 }, { "epoch": 8.311526479750778, "grad_norm": 3.296875, "learning_rate": 4.9875032362416254e-05, "loss": 0.5291, "step": 1338 }, { "epoch": 8.317757009345794, "grad_norm": 3.515625, "learning_rate": 4.987478620532423e-05, "loss": 0.4635, "step": 1339 }, { "epoch": 8.32398753894081, "grad_norm": 3.40625, "learning_rate": 4.987453980664332e-05, "loss": 0.4083, "step": 1340 }, { "epoch": 8.330218068535826, "grad_norm": 3.59375, "learning_rate": 4.98742931663759e-05, "loss": 0.5085, "step": 1341 }, { "epoch": 8.336448598130842, "grad_norm": 4.90625, "learning_rate": 4.987404628452438e-05, "loss": 0.4653, "step": 1342 }, { "epoch": 8.342679127725857, "grad_norm": 4.125, "learning_rate": 4.987379916109114e-05, "loss": 0.489, "step": 1343 }, { "epoch": 8.348909657320872, "grad_norm": 4.0, "learning_rate": 4.987355179607859e-05, "loss": 0.5742, "step": 1344 }, { "epoch": 8.355140186915888, "grad_norm": 4.875, "learning_rate": 4.9873304189489144e-05, "loss": 0.5105, "step": 1345 }, { "epoch": 8.361370716510903, "grad_norm": 4.40625, "learning_rate": 4.987305634132519e-05, "loss": 0.3681, "step": 1346 }, { "epoch": 8.367601246105918, "grad_norm": 5.25, "learning_rate": 4.9872808251589144e-05, "loss": 0.375, "step": 1347 }, { "epoch": 8.373831775700934, "grad_norm": 3.40625, "learning_rate": 4.98725599202834e-05, "loss": 0.4207, "step": 1348 }, { "epoch": 8.38006230529595, "grad_norm": 4.65625, "learning_rate": 4.98723113474104e-05, "loss": 0.308, "step": 1349 }, { "epoch": 8.386292834890966, "grad_norm": 5.21875, "learning_rate": 4.987206253297253e-05, "loss": 0.405, "step": 1350 }, { "epoch": 8.392523364485982, "grad_norm": 2.609375, "learning_rate": 4.987181347697221e-05, "loss": 0.2778, "step": 1351 }, { "epoch": 8.398753894080997, "grad_norm": 4.4375, "learning_rate": 4.987156417941188e-05, "loss": 0.363, "step": 1352 }, { "epoch": 8.404984423676012, "grad_norm": 4.34375, "learning_rate": 4.987131464029394e-05, "loss": 0.2555, "step": 1353 }, { "epoch": 8.411214953271028, "grad_norm": 3.78125, "learning_rate": 4.987106485962082e-05, "loss": 0.4244, "step": 1354 }, { "epoch": 8.417445482866043, "grad_norm": 2.640625, "learning_rate": 4.987081483739495e-05, "loss": 0.3264, "step": 1355 }, { "epoch": 8.423676012461058, "grad_norm": 3.265625, "learning_rate": 4.9870564573618747e-05, "loss": 0.3783, "step": 1356 }, { "epoch": 8.429906542056075, "grad_norm": 8.125, "learning_rate": 4.9870314068294655e-05, "loss": 0.5975, "step": 1357 }, { "epoch": 8.43613707165109, "grad_norm": 5.375, "learning_rate": 4.9870063321425105e-05, "loss": 0.4025, "step": 1358 }, { "epoch": 8.442367601246106, "grad_norm": 2.921875, "learning_rate": 4.986981233301252e-05, "loss": 0.2771, "step": 1359 }, { "epoch": 8.448598130841122, "grad_norm": 3.765625, "learning_rate": 4.9869561103059354e-05, "loss": 0.3953, "step": 1360 }, { "epoch": 8.454828660436137, "grad_norm": 5.875, "learning_rate": 4.986930963156804e-05, "loss": 0.4403, "step": 1361 }, { "epoch": 8.461059190031152, "grad_norm": 4.5625, "learning_rate": 4.986905791854101e-05, "loss": 0.4021, "step": 1362 }, { "epoch": 8.467289719626168, "grad_norm": 3.75, "learning_rate": 4.9868805963980726e-05, "loss": 0.601, "step": 1363 }, { "epoch": 8.473520249221183, "grad_norm": 3.40625, "learning_rate": 4.986855376788963e-05, "loss": 0.515, "step": 1364 }, { "epoch": 8.4797507788162, "grad_norm": 3.625, "learning_rate": 4.986830133027016e-05, "loss": 0.3628, "step": 1365 }, { "epoch": 8.485981308411215, "grad_norm": 4.6875, "learning_rate": 4.986804865112478e-05, "loss": 0.5495, "step": 1366 }, { "epoch": 8.49221183800623, "grad_norm": 3.890625, "learning_rate": 4.986779573045594e-05, "loss": 0.2991, "step": 1367 }, { "epoch": 8.498442367601246, "grad_norm": 4.75, "learning_rate": 4.98675425682661e-05, "loss": 0.3734, "step": 1368 }, { "epoch": 8.504672897196262, "grad_norm": 4.03125, "learning_rate": 4.986728916455771e-05, "loss": 0.48, "step": 1369 }, { "epoch": 8.510903426791277, "grad_norm": 2.71875, "learning_rate": 4.986703551933324e-05, "loss": 0.2212, "step": 1370 }, { "epoch": 8.517133956386292, "grad_norm": 3.140625, "learning_rate": 4.986678163259515e-05, "loss": 0.2972, "step": 1371 }, { "epoch": 8.523364485981308, "grad_norm": 3.875, "learning_rate": 4.986652750434591e-05, "loss": 0.5386, "step": 1372 }, { "epoch": 8.529595015576325, "grad_norm": 3.859375, "learning_rate": 4.986627313458798e-05, "loss": 0.2935, "step": 1373 }, { "epoch": 8.53582554517134, "grad_norm": 3.921875, "learning_rate": 4.986601852332383e-05, "loss": 0.4043, "step": 1374 }, { "epoch": 8.542056074766355, "grad_norm": 2.890625, "learning_rate": 4.986576367055594e-05, "loss": 0.2797, "step": 1375 }, { "epoch": 8.54828660436137, "grad_norm": 4.59375, "learning_rate": 4.986550857628679e-05, "loss": 0.4316, "step": 1376 }, { "epoch": 8.554517133956386, "grad_norm": 3.65625, "learning_rate": 4.986525324051885e-05, "loss": 0.3692, "step": 1377 }, { "epoch": 8.560747663551401, "grad_norm": 3.796875, "learning_rate": 4.986499766325459e-05, "loss": 0.408, "step": 1378 }, { "epoch": 8.566978193146417, "grad_norm": 4.65625, "learning_rate": 4.9864741844496506e-05, "loss": 0.4166, "step": 1379 }, { "epoch": 8.573208722741432, "grad_norm": 4.65625, "learning_rate": 4.9864485784247086e-05, "loss": 0.381, "step": 1380 }, { "epoch": 8.57943925233645, "grad_norm": 3.0625, "learning_rate": 4.9864229482508804e-05, "loss": 0.2736, "step": 1381 }, { "epoch": 8.585669781931465, "grad_norm": 4.3125, "learning_rate": 4.986397293928415e-05, "loss": 0.2526, "step": 1382 }, { "epoch": 8.59190031152648, "grad_norm": 7.5, "learning_rate": 4.986371615457564e-05, "loss": 0.3489, "step": 1383 }, { "epoch": 8.598130841121495, "grad_norm": 5.28125, "learning_rate": 4.986345912838573e-05, "loss": 0.6103, "step": 1384 }, { "epoch": 8.60436137071651, "grad_norm": 4.6875, "learning_rate": 4.986320186071694e-05, "loss": 0.2931, "step": 1385 }, { "epoch": 8.610591900311526, "grad_norm": 2.484375, "learning_rate": 4.986294435157176e-05, "loss": 0.2484, "step": 1386 }, { "epoch": 8.616822429906541, "grad_norm": 6.5625, "learning_rate": 4.986268660095269e-05, "loss": 0.334, "step": 1387 }, { "epoch": 8.623052959501557, "grad_norm": 5.125, "learning_rate": 4.9862428608862253e-05, "loss": 0.5123, "step": 1388 }, { "epoch": 8.629283489096574, "grad_norm": 7.34375, "learning_rate": 4.986217037530293e-05, "loss": 0.4144, "step": 1389 }, { "epoch": 8.63551401869159, "grad_norm": 4.21875, "learning_rate": 4.9861911900277236e-05, "loss": 0.4631, "step": 1390 }, { "epoch": 8.641744548286605, "grad_norm": 4.71875, "learning_rate": 4.986165318378769e-05, "loss": 0.2088, "step": 1391 }, { "epoch": 8.64797507788162, "grad_norm": 7.0, "learning_rate": 4.98613942258368e-05, "loss": 0.5726, "step": 1392 }, { "epoch": 8.654205607476635, "grad_norm": 4.84375, "learning_rate": 4.9861135026427074e-05, "loss": 0.4645, "step": 1393 }, { "epoch": 8.66043613707165, "grad_norm": 3.78125, "learning_rate": 4.986087558556104e-05, "loss": 0.4516, "step": 1394 }, { "epoch": 8.666666666666666, "grad_norm": 4.34375, "learning_rate": 4.98606159032412e-05, "loss": 0.342, "step": 1395 }, { "epoch": 8.672897196261681, "grad_norm": 5.21875, "learning_rate": 4.986035597947011e-05, "loss": 0.6123, "step": 1396 }, { "epoch": 8.679127725856699, "grad_norm": 3.796875, "learning_rate": 4.986009581425026e-05, "loss": 0.3505, "step": 1397 }, { "epoch": 8.685358255451714, "grad_norm": 4.6875, "learning_rate": 4.9859835407584186e-05, "loss": 0.3487, "step": 1398 }, { "epoch": 8.69158878504673, "grad_norm": 3.890625, "learning_rate": 4.985957475947443e-05, "loss": 0.4335, "step": 1399 }, { "epoch": 8.697819314641745, "grad_norm": 4.40625, "learning_rate": 4.985931386992351e-05, "loss": 0.3712, "step": 1400 }, { "epoch": 8.70404984423676, "grad_norm": 4.0, "learning_rate": 4.9859052738933966e-05, "loss": 0.3335, "step": 1401 }, { "epoch": 8.710280373831775, "grad_norm": 2.796875, "learning_rate": 4.985879136650834e-05, "loss": 0.373, "step": 1402 }, { "epoch": 8.71651090342679, "grad_norm": 2.984375, "learning_rate": 4.985852975264915e-05, "loss": 0.3417, "step": 1403 }, { "epoch": 8.722741433021806, "grad_norm": 2.6875, "learning_rate": 4.9858267897358956e-05, "loss": 0.341, "step": 1404 }, { "epoch": 8.728971962616823, "grad_norm": 3.359375, "learning_rate": 4.985800580064029e-05, "loss": 0.3096, "step": 1405 }, { "epoch": 8.735202492211839, "grad_norm": 2.90625, "learning_rate": 4.985774346249571e-05, "loss": 0.3406, "step": 1406 }, { "epoch": 8.741433021806854, "grad_norm": 4.0625, "learning_rate": 4.985748088292775e-05, "loss": 0.6418, "step": 1407 }, { "epoch": 8.74766355140187, "grad_norm": 4.03125, "learning_rate": 4.9857218061938976e-05, "loss": 0.4699, "step": 1408 }, { "epoch": 8.753894080996885, "grad_norm": 3.21875, "learning_rate": 4.985695499953193e-05, "loss": 0.5428, "step": 1409 }, { "epoch": 8.7601246105919, "grad_norm": 3.8125, "learning_rate": 4.985669169570916e-05, "loss": 0.3722, "step": 1410 }, { "epoch": 8.766355140186915, "grad_norm": 3.375, "learning_rate": 4.985642815047323e-05, "loss": 0.4381, "step": 1411 }, { "epoch": 8.77258566978193, "grad_norm": 3.578125, "learning_rate": 4.985616436382671e-05, "loss": 0.4347, "step": 1412 }, { "epoch": 8.778816199376948, "grad_norm": 3.171875, "learning_rate": 4.985590033577215e-05, "loss": 0.2959, "step": 1413 }, { "epoch": 8.785046728971963, "grad_norm": 4.4375, "learning_rate": 4.985563606631212e-05, "loss": 0.433, "step": 1414 }, { "epoch": 8.791277258566979, "grad_norm": 3.53125, "learning_rate": 4.9855371555449185e-05, "loss": 0.3808, "step": 1415 }, { "epoch": 8.797507788161994, "grad_norm": 3.71875, "learning_rate": 4.98551068031859e-05, "loss": 0.3154, "step": 1416 }, { "epoch": 8.80373831775701, "grad_norm": 4.625, "learning_rate": 4.9854841809524866e-05, "loss": 0.4619, "step": 1417 }, { "epoch": 8.809968847352025, "grad_norm": 2.734375, "learning_rate": 4.985457657446863e-05, "loss": 0.3041, "step": 1418 }, { "epoch": 8.81619937694704, "grad_norm": 3.546875, "learning_rate": 4.985431109801978e-05, "loss": 0.3683, "step": 1419 }, { "epoch": 8.822429906542055, "grad_norm": 6.28125, "learning_rate": 4.985404538018089e-05, "loss": 0.3789, "step": 1420 }, { "epoch": 8.828660436137072, "grad_norm": 6.46875, "learning_rate": 4.985377942095455e-05, "loss": 0.4158, "step": 1421 }, { "epoch": 8.834890965732088, "grad_norm": 5.21875, "learning_rate": 4.985351322034333e-05, "loss": 0.4277, "step": 1422 }, { "epoch": 8.841121495327103, "grad_norm": 3.796875, "learning_rate": 4.985324677834983e-05, "loss": 0.4379, "step": 1423 }, { "epoch": 8.847352024922118, "grad_norm": 5.9375, "learning_rate": 4.985298009497662e-05, "loss": 0.4923, "step": 1424 }, { "epoch": 8.853582554517134, "grad_norm": 9.125, "learning_rate": 4.9852713170226294e-05, "loss": 0.6862, "step": 1425 }, { "epoch": 8.85981308411215, "grad_norm": 6.53125, "learning_rate": 4.985244600410146e-05, "loss": 0.3098, "step": 1426 }, { "epoch": 8.866043613707165, "grad_norm": 5.59375, "learning_rate": 4.9852178596604705e-05, "loss": 0.3421, "step": 1427 }, { "epoch": 8.87227414330218, "grad_norm": 4.25, "learning_rate": 4.985191094773861e-05, "loss": 0.352, "step": 1428 }, { "epoch": 8.878504672897197, "grad_norm": 7.125, "learning_rate": 4.9851643057505805e-05, "loss": 0.3544, "step": 1429 }, { "epoch": 8.884735202492212, "grad_norm": 6.25, "learning_rate": 4.985137492590887e-05, "loss": 0.2265, "step": 1430 }, { "epoch": 8.890965732087228, "grad_norm": 4.1875, "learning_rate": 4.9851106552950404e-05, "loss": 0.4324, "step": 1431 }, { "epoch": 8.897196261682243, "grad_norm": 4.375, "learning_rate": 4.985083793863303e-05, "loss": 0.4722, "step": 1432 }, { "epoch": 8.903426791277258, "grad_norm": 3.375, "learning_rate": 4.985056908295935e-05, "loss": 0.4213, "step": 1433 }, { "epoch": 8.909657320872274, "grad_norm": 4.25, "learning_rate": 4.985029998593198e-05, "loss": 0.428, "step": 1434 }, { "epoch": 8.91588785046729, "grad_norm": 4.03125, "learning_rate": 4.9850030647553525e-05, "loss": 0.4185, "step": 1435 }, { "epoch": 8.922118380062305, "grad_norm": 4.59375, "learning_rate": 4.984976106782661e-05, "loss": 0.4633, "step": 1436 }, { "epoch": 8.928348909657322, "grad_norm": 3.03125, "learning_rate": 4.984949124675383e-05, "loss": 0.2845, "step": 1437 }, { "epoch": 8.934579439252337, "grad_norm": 2.84375, "learning_rate": 4.9849221184337843e-05, "loss": 0.2843, "step": 1438 }, { "epoch": 8.940809968847352, "grad_norm": 3.84375, "learning_rate": 4.9848950880581246e-05, "loss": 0.2972, "step": 1439 }, { "epoch": 8.947040498442368, "grad_norm": 3.1875, "learning_rate": 4.984868033548667e-05, "loss": 0.2857, "step": 1440 }, { "epoch": 8.953271028037383, "grad_norm": 3.84375, "learning_rate": 4.9848409549056745e-05, "loss": 0.3891, "step": 1441 }, { "epoch": 8.959501557632398, "grad_norm": 3.890625, "learning_rate": 4.984813852129409e-05, "loss": 0.4106, "step": 1442 }, { "epoch": 8.965732087227414, "grad_norm": 2.5, "learning_rate": 4.984786725220136e-05, "loss": 0.2219, "step": 1443 }, { "epoch": 8.97196261682243, "grad_norm": 4.03125, "learning_rate": 4.9847595741781175e-05, "loss": 0.3955, "step": 1444 }, { "epoch": 8.978193146417446, "grad_norm": 3.484375, "learning_rate": 4.9847323990036174e-05, "loss": 0.4804, "step": 1445 }, { "epoch": 8.984423676012462, "grad_norm": 3.9375, "learning_rate": 4.984705199696899e-05, "loss": 0.3511, "step": 1446 }, { "epoch": 8.990654205607477, "grad_norm": 3.109375, "learning_rate": 4.984677976258227e-05, "loss": 0.2896, "step": 1447 }, { "epoch": 8.996884735202492, "grad_norm": 3.75, "learning_rate": 4.9846507286878665e-05, "loss": 0.4947, "step": 1448 }, { "epoch": 9.0, "grad_norm": 2.453125, "learning_rate": 4.9846234569860806e-05, "loss": 0.24, "step": 1449 }, { "epoch": 9.006230529595015, "grad_norm": 2.515625, "learning_rate": 4.984596161153136e-05, "loss": 0.2993, "step": 1450 }, { "epoch": 9.01246105919003, "grad_norm": 4.1875, "learning_rate": 4.9845688411892956e-05, "loss": 0.4036, "step": 1451 }, { "epoch": 9.018691588785046, "grad_norm": 3.421875, "learning_rate": 4.984541497094827e-05, "loss": 0.3393, "step": 1452 }, { "epoch": 9.024922118380061, "grad_norm": 3.8125, "learning_rate": 4.984514128869994e-05, "loss": 0.411, "step": 1453 }, { "epoch": 9.031152647975079, "grad_norm": 4.34375, "learning_rate": 4.9844867365150636e-05, "loss": 0.4003, "step": 1454 }, { "epoch": 9.037383177570094, "grad_norm": 4.4375, "learning_rate": 4.984459320030301e-05, "loss": 0.3377, "step": 1455 }, { "epoch": 9.04361370716511, "grad_norm": 3.625, "learning_rate": 4.984431879415973e-05, "loss": 0.2611, "step": 1456 }, { "epoch": 9.049844236760125, "grad_norm": 3.984375, "learning_rate": 4.984404414672346e-05, "loss": 0.4152, "step": 1457 }, { "epoch": 9.05607476635514, "grad_norm": 7.40625, "learning_rate": 4.984376925799687e-05, "loss": 0.344, "step": 1458 }, { "epoch": 9.062305295950155, "grad_norm": 8.0625, "learning_rate": 4.984349412798263e-05, "loss": 0.2804, "step": 1459 }, { "epoch": 9.06853582554517, "grad_norm": 9.25, "learning_rate": 4.98432187566834e-05, "loss": 0.4757, "step": 1460 }, { "epoch": 9.074766355140186, "grad_norm": 4.84375, "learning_rate": 4.984294314410186e-05, "loss": 0.4826, "step": 1461 }, { "epoch": 9.080996884735203, "grad_norm": 5.0, "learning_rate": 4.98426672902407e-05, "loss": 0.4146, "step": 1462 }, { "epoch": 9.087227414330219, "grad_norm": 4.0625, "learning_rate": 4.9842391195102585e-05, "loss": 0.4338, "step": 1463 }, { "epoch": 9.093457943925234, "grad_norm": 4.40625, "learning_rate": 4.98421148586902e-05, "loss": 0.4155, "step": 1464 }, { "epoch": 9.09968847352025, "grad_norm": 3.0625, "learning_rate": 4.9841838281006235e-05, "loss": 0.2522, "step": 1465 }, { "epoch": 9.105919003115265, "grad_norm": 3.3125, "learning_rate": 4.984156146205336e-05, "loss": 0.2294, "step": 1466 }, { "epoch": 9.11214953271028, "grad_norm": 4.09375, "learning_rate": 4.984128440183429e-05, "loss": 0.2848, "step": 1467 }, { "epoch": 9.118380062305295, "grad_norm": 3.328125, "learning_rate": 4.984100710035169e-05, "loss": 0.2469, "step": 1468 }, { "epoch": 9.12461059190031, "grad_norm": 2.90625, "learning_rate": 4.9840729557608265e-05, "loss": 0.3061, "step": 1469 }, { "epoch": 9.130841121495328, "grad_norm": 5.0625, "learning_rate": 4.984045177360671e-05, "loss": 0.423, "step": 1470 }, { "epoch": 9.137071651090343, "grad_norm": 3.703125, "learning_rate": 4.9840173748349716e-05, "loss": 0.4468, "step": 1471 }, { "epoch": 9.143302180685358, "grad_norm": 3.078125, "learning_rate": 4.9839895481839994e-05, "loss": 0.2443, "step": 1472 }, { "epoch": 9.149532710280374, "grad_norm": 2.171875, "learning_rate": 4.9839616974080246e-05, "loss": 0.2, "step": 1473 }, { "epoch": 9.15576323987539, "grad_norm": 3.265625, "learning_rate": 4.983933822507316e-05, "loss": 0.3402, "step": 1474 }, { "epoch": 9.161993769470405, "grad_norm": 3.203125, "learning_rate": 4.983905923482147e-05, "loss": 0.3279, "step": 1475 }, { "epoch": 9.16822429906542, "grad_norm": 2.96875, "learning_rate": 4.983878000332787e-05, "loss": 0.2908, "step": 1476 }, { "epoch": 9.174454828660435, "grad_norm": 3.828125, "learning_rate": 4.9838500530595066e-05, "loss": 0.3899, "step": 1477 }, { "epoch": 9.180685358255452, "grad_norm": 1.8203125, "learning_rate": 4.983822081662578e-05, "loss": 0.2113, "step": 1478 }, { "epoch": 9.186915887850468, "grad_norm": 3.453125, "learning_rate": 4.983794086142273e-05, "loss": 0.3908, "step": 1479 }, { "epoch": 9.193146417445483, "grad_norm": 4.0, "learning_rate": 4.9837660664988636e-05, "loss": 0.2624, "step": 1480 }, { "epoch": 9.199376947040498, "grad_norm": 3.0625, "learning_rate": 4.983738022732621e-05, "loss": 0.4495, "step": 1481 }, { "epoch": 9.205607476635514, "grad_norm": 2.484375, "learning_rate": 4.9837099548438195e-05, "loss": 0.2885, "step": 1482 }, { "epoch": 9.21183800623053, "grad_norm": 3.203125, "learning_rate": 4.98368186283273e-05, "loss": 0.3212, "step": 1483 }, { "epoch": 9.218068535825545, "grad_norm": 2.1875, "learning_rate": 4.9836537466996244e-05, "loss": 0.2248, "step": 1484 }, { "epoch": 9.22429906542056, "grad_norm": 2.265625, "learning_rate": 4.983625606444779e-05, "loss": 0.257, "step": 1485 }, { "epoch": 9.230529595015577, "grad_norm": 2.390625, "learning_rate": 4.983597442068464e-05, "loss": 0.257, "step": 1486 }, { "epoch": 9.236760124610592, "grad_norm": 3.21875, "learning_rate": 4.983569253570955e-05, "loss": 0.2767, "step": 1487 }, { "epoch": 9.242990654205608, "grad_norm": 3.671875, "learning_rate": 4.9835410409525246e-05, "loss": 0.3661, "step": 1488 }, { "epoch": 9.249221183800623, "grad_norm": 4.1875, "learning_rate": 4.983512804213447e-05, "loss": 0.4692, "step": 1489 }, { "epoch": 9.255451713395638, "grad_norm": 3.25, "learning_rate": 4.9834845433539966e-05, "loss": 0.3731, "step": 1490 }, { "epoch": 9.261682242990654, "grad_norm": 3.4375, "learning_rate": 4.983456258374448e-05, "loss": 0.2799, "step": 1491 }, { "epoch": 9.26791277258567, "grad_norm": 4.375, "learning_rate": 4.983427949275076e-05, "loss": 0.3337, "step": 1492 }, { "epoch": 9.274143302180685, "grad_norm": 3.78125, "learning_rate": 4.983399616056155e-05, "loss": 0.4058, "step": 1493 }, { "epoch": 9.280373831775702, "grad_norm": 4.75, "learning_rate": 4.983371258717961e-05, "loss": 0.3449, "step": 1494 }, { "epoch": 9.286604361370717, "grad_norm": 6.625, "learning_rate": 4.9833428772607685e-05, "loss": 0.5227, "step": 1495 }, { "epoch": 9.292834890965732, "grad_norm": 5.96875, "learning_rate": 4.983314471684853e-05, "loss": 0.2311, "step": 1496 }, { "epoch": 9.299065420560748, "grad_norm": 2.1875, "learning_rate": 4.983286041990492e-05, "loss": 0.3054, "step": 1497 }, { "epoch": 9.305295950155763, "grad_norm": 5.125, "learning_rate": 4.98325758817796e-05, "loss": 0.379, "step": 1498 }, { "epoch": 9.311526479750778, "grad_norm": 5.46875, "learning_rate": 4.983229110247534e-05, "loss": 0.6068, "step": 1499 }, { "epoch": 9.317757009345794, "grad_norm": 3.84375, "learning_rate": 4.9832006081994907e-05, "loss": 0.4105, "step": 1500 }, { "epoch": 9.32398753894081, "grad_norm": 3.71875, "learning_rate": 4.983172082034106e-05, "loss": 0.5494, "step": 1501 }, { "epoch": 9.330218068535826, "grad_norm": 2.75, "learning_rate": 4.983143531751659e-05, "loss": 0.4175, "step": 1502 }, { "epoch": 9.336448598130842, "grad_norm": 3.765625, "learning_rate": 4.983114957352425e-05, "loss": 0.3544, "step": 1503 }, { "epoch": 9.342679127725857, "grad_norm": 4.8125, "learning_rate": 4.9830863588366815e-05, "loss": 0.2478, "step": 1504 }, { "epoch": 9.348909657320872, "grad_norm": 2.734375, "learning_rate": 4.9830577362047076e-05, "loss": 0.2417, "step": 1505 }, { "epoch": 9.355140186915888, "grad_norm": 3.828125, "learning_rate": 4.98302908945678e-05, "loss": 0.3977, "step": 1506 }, { "epoch": 9.361370716510903, "grad_norm": 3.21875, "learning_rate": 4.983000418593179e-05, "loss": 0.2396, "step": 1507 }, { "epoch": 9.367601246105918, "grad_norm": 2.375, "learning_rate": 4.98297172361418e-05, "loss": 0.2267, "step": 1508 }, { "epoch": 9.373831775700934, "grad_norm": 3.015625, "learning_rate": 4.982943004520064e-05, "loss": 0.2676, "step": 1509 }, { "epoch": 9.38006230529595, "grad_norm": 3.90625, "learning_rate": 4.982914261311109e-05, "loss": 0.37, "step": 1510 }, { "epoch": 9.386292834890966, "grad_norm": 4.3125, "learning_rate": 4.9828854939875945e-05, "loss": 0.3094, "step": 1511 }, { "epoch": 9.392523364485982, "grad_norm": 3.953125, "learning_rate": 4.9828567025498e-05, "loss": 0.3521, "step": 1512 }, { "epoch": 9.398753894080997, "grad_norm": 3.625, "learning_rate": 4.9828278869980044e-05, "loss": 0.3441, "step": 1513 }, { "epoch": 9.404984423676012, "grad_norm": 2.84375, "learning_rate": 4.982799047332488e-05, "loss": 0.3019, "step": 1514 }, { "epoch": 9.411214953271028, "grad_norm": 4.5625, "learning_rate": 4.9827701835535315e-05, "loss": 0.307, "step": 1515 }, { "epoch": 9.417445482866043, "grad_norm": 3.484375, "learning_rate": 4.9827412956614145e-05, "loss": 0.4376, "step": 1516 }, { "epoch": 9.423676012461058, "grad_norm": 4.25, "learning_rate": 4.9827123836564175e-05, "loss": 0.2868, "step": 1517 }, { "epoch": 9.429906542056075, "grad_norm": 2.484375, "learning_rate": 4.982683447538822e-05, "loss": 0.2598, "step": 1518 }, { "epoch": 9.43613707165109, "grad_norm": 4.40625, "learning_rate": 4.982654487308908e-05, "loss": 0.4036, "step": 1519 }, { "epoch": 9.442367601246106, "grad_norm": 2.484375, "learning_rate": 4.9826255029669577e-05, "loss": 0.2996, "step": 1520 }, { "epoch": 9.448598130841122, "grad_norm": 6.375, "learning_rate": 4.982596494513252e-05, "loss": 0.4265, "step": 1521 }, { "epoch": 9.454828660436137, "grad_norm": 4.90625, "learning_rate": 4.982567461948073e-05, "loss": 0.2577, "step": 1522 }, { "epoch": 9.461059190031152, "grad_norm": 5.84375, "learning_rate": 4.982538405271703e-05, "loss": 0.5314, "step": 1523 }, { "epoch": 9.467289719626168, "grad_norm": 3.15625, "learning_rate": 4.982509324484424e-05, "loss": 0.2529, "step": 1524 }, { "epoch": 9.473520249221183, "grad_norm": 6.46875, "learning_rate": 4.9824802195865164e-05, "loss": 0.3395, "step": 1525 }, { "epoch": 9.4797507788162, "grad_norm": 2.921875, "learning_rate": 4.982451090578266e-05, "loss": 0.3474, "step": 1526 }, { "epoch": 9.485981308411215, "grad_norm": 4.78125, "learning_rate": 4.9824219374599543e-05, "loss": 0.335, "step": 1527 }, { "epoch": 9.49221183800623, "grad_norm": 4.125, "learning_rate": 4.982392760231864e-05, "loss": 0.4863, "step": 1528 }, { "epoch": 9.498442367601246, "grad_norm": 2.90625, "learning_rate": 4.98236355889428e-05, "loss": 0.3372, "step": 1529 }, { "epoch": 9.504672897196262, "grad_norm": 3.8125, "learning_rate": 4.9823343334474844e-05, "loss": 0.4272, "step": 1530 }, { "epoch": 9.510903426791277, "grad_norm": 4.0, "learning_rate": 4.982305083891761e-05, "loss": 0.4918, "step": 1531 }, { "epoch": 9.517133956386292, "grad_norm": 2.296875, "learning_rate": 4.982275810227395e-05, "loss": 0.2727, "step": 1532 }, { "epoch": 9.523364485981308, "grad_norm": 3.0, "learning_rate": 4.9822465124546705e-05, "loss": 0.2535, "step": 1533 }, { "epoch": 9.529595015576325, "grad_norm": 3.96875, "learning_rate": 4.982217190573871e-05, "loss": 0.4352, "step": 1534 }, { "epoch": 9.53582554517134, "grad_norm": 2.828125, "learning_rate": 4.982187844585282e-05, "loss": 0.2624, "step": 1535 }, { "epoch": 9.542056074766355, "grad_norm": 2.59375, "learning_rate": 4.982158474489189e-05, "loss": 0.2743, "step": 1536 }, { "epoch": 9.54828660436137, "grad_norm": 1.9453125, "learning_rate": 4.982129080285876e-05, "loss": 0.2233, "step": 1537 }, { "epoch": 9.554517133956386, "grad_norm": 3.828125, "learning_rate": 4.98209966197563e-05, "loss": 0.3479, "step": 1538 }, { "epoch": 9.560747663551401, "grad_norm": 2.53125, "learning_rate": 4.9820702195587356e-05, "loss": 0.2525, "step": 1539 }, { "epoch": 9.566978193146417, "grad_norm": 3.34375, "learning_rate": 4.9820407530354786e-05, "loss": 0.3433, "step": 1540 }, { "epoch": 9.573208722741432, "grad_norm": 4.0625, "learning_rate": 4.982011262406145e-05, "loss": 0.3334, "step": 1541 }, { "epoch": 9.57943925233645, "grad_norm": 3.078125, "learning_rate": 4.981981747671024e-05, "loss": 0.3732, "step": 1542 }, { "epoch": 9.585669781931465, "grad_norm": 3.828125, "learning_rate": 4.981952208830399e-05, "loss": 0.4863, "step": 1543 }, { "epoch": 9.59190031152648, "grad_norm": 4.59375, "learning_rate": 4.981922645884558e-05, "loss": 0.3562, "step": 1544 }, { "epoch": 9.598130841121495, "grad_norm": 4.875, "learning_rate": 4.981893058833788e-05, "loss": 0.3072, "step": 1545 }, { "epoch": 9.60436137071651, "grad_norm": 3.1875, "learning_rate": 4.9818634476783764e-05, "loss": 0.3958, "step": 1546 }, { "epoch": 9.610591900311526, "grad_norm": 3.9375, "learning_rate": 4.981833812418612e-05, "loss": 0.4716, "step": 1547 }, { "epoch": 9.616822429906541, "grad_norm": 3.6875, "learning_rate": 4.9818041530547796e-05, "loss": 0.3006, "step": 1548 }, { "epoch": 9.623052959501557, "grad_norm": 4.40625, "learning_rate": 4.98177446958717e-05, "loss": 0.5452, "step": 1549 }, { "epoch": 9.629283489096574, "grad_norm": 3.609375, "learning_rate": 4.981744762016072e-05, "loss": 0.5232, "step": 1550 }, { "epoch": 9.63551401869159, "grad_norm": 3.734375, "learning_rate": 4.9817150303417716e-05, "loss": 0.3052, "step": 1551 }, { "epoch": 9.641744548286605, "grad_norm": 4.5, "learning_rate": 4.981685274564558e-05, "loss": 0.4713, "step": 1552 }, { "epoch": 9.64797507788162, "grad_norm": 4.4375, "learning_rate": 4.9816554946847216e-05, "loss": 0.3603, "step": 1553 }, { "epoch": 9.654205607476635, "grad_norm": 2.78125, "learning_rate": 4.981625690702552e-05, "loss": 0.2546, "step": 1554 }, { "epoch": 9.66043613707165, "grad_norm": 3.890625, "learning_rate": 4.981595862618336e-05, "loss": 0.4446, "step": 1555 }, { "epoch": 9.666666666666666, "grad_norm": 3.1875, "learning_rate": 4.9815660104323656e-05, "loss": 0.3243, "step": 1556 }, { "epoch": 9.672897196261681, "grad_norm": 5.25, "learning_rate": 4.98153613414493e-05, "loss": 0.3055, "step": 1557 }, { "epoch": 9.679127725856699, "grad_norm": 4.25, "learning_rate": 4.98150623375632e-05, "loss": 0.3349, "step": 1558 }, { "epoch": 9.685358255451714, "grad_norm": 2.5, "learning_rate": 4.9814763092668245e-05, "loss": 0.2742, "step": 1559 }, { "epoch": 9.69158878504673, "grad_norm": 2.921875, "learning_rate": 4.981446360676736e-05, "loss": 0.2665, "step": 1560 }, { "epoch": 9.697819314641745, "grad_norm": 3.734375, "learning_rate": 4.9814163879863436e-05, "loss": 0.2946, "step": 1561 }, { "epoch": 9.70404984423676, "grad_norm": 4.59375, "learning_rate": 4.981386391195939e-05, "loss": 0.4124, "step": 1562 }, { "epoch": 9.710280373831775, "grad_norm": 4.6875, "learning_rate": 4.981356370305815e-05, "loss": 0.2222, "step": 1563 }, { "epoch": 9.71651090342679, "grad_norm": 2.984375, "learning_rate": 4.981326325316261e-05, "loss": 0.2866, "step": 1564 }, { "epoch": 9.722741433021806, "grad_norm": 2.59375, "learning_rate": 4.981296256227569e-05, "loss": 0.2444, "step": 1565 }, { "epoch": 9.728971962616823, "grad_norm": 3.421875, "learning_rate": 4.981266163040033e-05, "loss": 0.2655, "step": 1566 }, { "epoch": 9.735202492211839, "grad_norm": 4.0625, "learning_rate": 4.9812360457539444e-05, "loss": 0.2508, "step": 1567 }, { "epoch": 9.741433021806854, "grad_norm": 5.21875, "learning_rate": 4.981205904369594e-05, "loss": 0.4171, "step": 1568 }, { "epoch": 9.74766355140187, "grad_norm": 5.09375, "learning_rate": 4.9811757388872764e-05, "loss": 0.3575, "step": 1569 }, { "epoch": 9.753894080996885, "grad_norm": 3.390625, "learning_rate": 4.981145549307285e-05, "loss": 0.2845, "step": 1570 }, { "epoch": 9.7601246105919, "grad_norm": 2.953125, "learning_rate": 4.98111533562991e-05, "loss": 0.4146, "step": 1571 }, { "epoch": 9.766355140186915, "grad_norm": 3.25, "learning_rate": 4.981085097855449e-05, "loss": 0.2509, "step": 1572 }, { "epoch": 9.77258566978193, "grad_norm": 3.21875, "learning_rate": 4.981054835984192e-05, "loss": 0.3957, "step": 1573 }, { "epoch": 9.778816199376948, "grad_norm": 2.6875, "learning_rate": 4.981024550016435e-05, "loss": 0.2785, "step": 1574 }, { "epoch": 9.785046728971963, "grad_norm": 2.875, "learning_rate": 4.980994239952472e-05, "loss": 0.2763, "step": 1575 }, { "epoch": 9.791277258566979, "grad_norm": 4.0625, "learning_rate": 4.9809639057925966e-05, "loss": 0.3019, "step": 1576 }, { "epoch": 9.797507788161994, "grad_norm": 4.0625, "learning_rate": 4.9809335475371045e-05, "loss": 0.3584, "step": 1577 }, { "epoch": 9.80373831775701, "grad_norm": 3.609375, "learning_rate": 4.980903165186289e-05, "loss": 0.3667, "step": 1578 }, { "epoch": 9.809968847352025, "grad_norm": 3.296875, "learning_rate": 4.9808727587404464e-05, "loss": 0.3258, "step": 1579 }, { "epoch": 9.81619937694704, "grad_norm": 5.53125, "learning_rate": 4.980842328199872e-05, "loss": 0.3599, "step": 1580 }, { "epoch": 9.822429906542055, "grad_norm": 2.609375, "learning_rate": 4.98081187356486e-05, "loss": 0.3065, "step": 1581 }, { "epoch": 9.828660436137072, "grad_norm": 3.359375, "learning_rate": 4.980781394835707e-05, "loss": 0.3058, "step": 1582 }, { "epoch": 9.834890965732088, "grad_norm": 3.265625, "learning_rate": 4.980750892012711e-05, "loss": 0.4658, "step": 1583 }, { "epoch": 9.841121495327103, "grad_norm": 3.59375, "learning_rate": 4.9807203650961654e-05, "loss": 0.3304, "step": 1584 }, { "epoch": 9.847352024922118, "grad_norm": 2.796875, "learning_rate": 4.980689814086368e-05, "loss": 0.3275, "step": 1585 }, { "epoch": 9.853582554517134, "grad_norm": 2.578125, "learning_rate": 4.980659238983615e-05, "loss": 0.3111, "step": 1586 }, { "epoch": 9.85981308411215, "grad_norm": 2.859375, "learning_rate": 4.980628639788203e-05, "loss": 0.2771, "step": 1587 }, { "epoch": 9.866043613707165, "grad_norm": 2.546875, "learning_rate": 4.9805980165004304e-05, "loss": 0.2486, "step": 1588 }, { "epoch": 9.87227414330218, "grad_norm": 4.96875, "learning_rate": 4.980567369120594e-05, "loss": 0.4979, "step": 1589 }, { "epoch": 9.878504672897197, "grad_norm": 2.296875, "learning_rate": 4.9805366976489915e-05, "loss": 0.2439, "step": 1590 }, { "epoch": 9.884735202492212, "grad_norm": 3.875, "learning_rate": 4.980506002085921e-05, "loss": 0.4273, "step": 1591 }, { "epoch": 9.890965732087228, "grad_norm": 3.234375, "learning_rate": 4.9804752824316794e-05, "loss": 0.3615, "step": 1592 }, { "epoch": 9.897196261682243, "grad_norm": 5.0, "learning_rate": 4.980444538686567e-05, "loss": 0.4937, "step": 1593 }, { "epoch": 9.903426791277258, "grad_norm": 3.203125, "learning_rate": 4.9804137708508814e-05, "loss": 0.3817, "step": 1594 }, { "epoch": 9.909657320872274, "grad_norm": 4.125, "learning_rate": 4.9803829789249205e-05, "loss": 0.3778, "step": 1595 }, { "epoch": 9.91588785046729, "grad_norm": 3.265625, "learning_rate": 4.980352162908985e-05, "loss": 0.4179, "step": 1596 }, { "epoch": 9.922118380062305, "grad_norm": 2.671875, "learning_rate": 4.980321322803374e-05, "loss": 0.2771, "step": 1597 }, { "epoch": 9.928348909657322, "grad_norm": 2.265625, "learning_rate": 4.9802904586083856e-05, "loss": 0.3559, "step": 1598 }, { "epoch": 9.934579439252337, "grad_norm": 2.875, "learning_rate": 4.9802595703243205e-05, "loss": 0.2838, "step": 1599 }, { "epoch": 9.940809968847352, "grad_norm": 2.671875, "learning_rate": 4.9802286579514787e-05, "loss": 0.3266, "step": 1600 }, { "epoch": 9.947040498442368, "grad_norm": 2.171875, "learning_rate": 4.980197721490161e-05, "loss": 0.2071, "step": 1601 }, { "epoch": 9.953271028037383, "grad_norm": 3.875, "learning_rate": 4.980166760940667e-05, "loss": 0.3106, "step": 1602 }, { "epoch": 9.959501557632398, "grad_norm": 2.90625, "learning_rate": 4.980135776303297e-05, "loss": 0.3547, "step": 1603 }, { "epoch": 9.965732087227414, "grad_norm": 3.28125, "learning_rate": 4.980104767578353e-05, "loss": 0.4603, "step": 1604 }, { "epoch": 9.97196261682243, "grad_norm": 3.578125, "learning_rate": 4.980073734766136e-05, "loss": 0.3774, "step": 1605 }, { "epoch": 9.978193146417446, "grad_norm": 2.578125, "learning_rate": 4.980042677866947e-05, "loss": 0.2673, "step": 1606 }, { "epoch": 9.984423676012462, "grad_norm": 3.4375, "learning_rate": 4.980011596881087e-05, "loss": 0.3239, "step": 1607 }, { "epoch": 9.990654205607477, "grad_norm": 3.53125, "learning_rate": 4.97998049180886e-05, "loss": 0.3667, "step": 1608 }, { "epoch": 9.996884735202492, "grad_norm": 2.515625, "learning_rate": 4.9799493626505653e-05, "loss": 0.2338, "step": 1609 }, { "epoch": 10.0, "grad_norm": 2.21875, "learning_rate": 4.9799182094065075e-05, "loss": 0.1516, "step": 1610 }, { "epoch": 10.006230529595015, "grad_norm": 3.578125, "learning_rate": 4.9798870320769886e-05, "loss": 0.4352, "step": 1611 }, { "epoch": 10.01246105919003, "grad_norm": 3.34375, "learning_rate": 4.979855830662311e-05, "loss": 0.2902, "step": 1612 }, { "epoch": 10.018691588785046, "grad_norm": 3.03125, "learning_rate": 4.9798246051627776e-05, "loss": 0.4102, "step": 1613 }, { "epoch": 10.024922118380061, "grad_norm": 4.09375, "learning_rate": 4.979793355578691e-05, "loss": 0.6046, "step": 1614 }, { "epoch": 10.031152647975079, "grad_norm": 4.125, "learning_rate": 4.9797620819103574e-05, "loss": 0.4895, "step": 1615 }, { "epoch": 10.037383177570094, "grad_norm": 3.734375, "learning_rate": 4.979730784158078e-05, "loss": 0.4159, "step": 1616 }, { "epoch": 10.04361370716511, "grad_norm": 3.140625, "learning_rate": 4.979699462322157e-05, "loss": 0.4023, "step": 1617 }, { "epoch": 10.049844236760125, "grad_norm": 6.84375, "learning_rate": 4.9796681164029e-05, "loss": 0.5295, "step": 1618 }, { "epoch": 10.05607476635514, "grad_norm": 4.125, "learning_rate": 4.97963674640061e-05, "loss": 0.3447, "step": 1619 }, { "epoch": 10.062305295950155, "grad_norm": 3.078125, "learning_rate": 4.979605352315592e-05, "loss": 0.269, "step": 1620 }, { "epoch": 10.06853582554517, "grad_norm": 6.0, "learning_rate": 4.979573934148152e-05, "loss": 0.369, "step": 1621 }, { "epoch": 10.074766355140186, "grad_norm": 6.21875, "learning_rate": 4.9795424918985936e-05, "loss": 0.4843, "step": 1622 }, { "epoch": 10.080996884735203, "grad_norm": 4.9375, "learning_rate": 4.979511025567223e-05, "loss": 0.3495, "step": 1623 }, { "epoch": 10.087227414330219, "grad_norm": 3.015625, "learning_rate": 4.979479535154346e-05, "loss": 0.2173, "step": 1624 }, { "epoch": 10.093457943925234, "grad_norm": 2.890625, "learning_rate": 4.979448020660268e-05, "loss": 0.236, "step": 1625 }, { "epoch": 10.09968847352025, "grad_norm": 4.1875, "learning_rate": 4.979416482085295e-05, "loss": 0.3235, "step": 1626 }, { "epoch": 10.105919003115265, "grad_norm": 3.109375, "learning_rate": 4.9793849194297335e-05, "loss": 0.244, "step": 1627 }, { "epoch": 10.11214953271028, "grad_norm": 4.40625, "learning_rate": 4.97935333269389e-05, "loss": 0.3962, "step": 1628 }, { "epoch": 10.118380062305295, "grad_norm": 3.5, "learning_rate": 4.979321721878072e-05, "loss": 0.386, "step": 1629 }, { "epoch": 10.12461059190031, "grad_norm": 3.609375, "learning_rate": 4.9792900869825854e-05, "loss": 0.4172, "step": 1630 }, { "epoch": 10.130841121495328, "grad_norm": 3.828125, "learning_rate": 4.979258428007738e-05, "loss": 0.4187, "step": 1631 }, { "epoch": 10.137071651090343, "grad_norm": 4.21875, "learning_rate": 4.979226744953837e-05, "loss": 0.3376, "step": 1632 }, { "epoch": 10.143302180685358, "grad_norm": 4.46875, "learning_rate": 4.97919503782119e-05, "loss": 0.283, "step": 1633 }, { "epoch": 10.149532710280374, "grad_norm": 3.171875, "learning_rate": 4.979163306610105e-05, "loss": 0.3432, "step": 1634 }, { "epoch": 10.15576323987539, "grad_norm": 4.34375, "learning_rate": 4.979131551320891e-05, "loss": 0.4562, "step": 1635 }, { "epoch": 10.161993769470405, "grad_norm": 3.9375, "learning_rate": 4.9790997719538554e-05, "loss": 0.224, "step": 1636 }, { "epoch": 10.16822429906542, "grad_norm": 3.53125, "learning_rate": 4.9790679685093075e-05, "loss": 0.4093, "step": 1637 }, { "epoch": 10.174454828660435, "grad_norm": 2.78125, "learning_rate": 4.979036140987556e-05, "loss": 0.3773, "step": 1638 }, { "epoch": 10.180685358255452, "grad_norm": 4.65625, "learning_rate": 4.97900428938891e-05, "loss": 0.5101, "step": 1639 }, { "epoch": 10.186915887850468, "grad_norm": 2.5625, "learning_rate": 4.978972413713678e-05, "loss": 0.3439, "step": 1640 }, { "epoch": 10.193146417445483, "grad_norm": 5.40625, "learning_rate": 4.9789405139621714e-05, "loss": 0.2689, "step": 1641 }, { "epoch": 10.199376947040498, "grad_norm": 4.125, "learning_rate": 4.978908590134699e-05, "loss": 0.3486, "step": 1642 }, { "epoch": 10.205607476635514, "grad_norm": 3.875, "learning_rate": 4.97887664223157e-05, "loss": 0.3506, "step": 1643 }, { "epoch": 10.21183800623053, "grad_norm": 4.40625, "learning_rate": 4.978844670253096e-05, "loss": 0.2343, "step": 1644 }, { "epoch": 10.218068535825545, "grad_norm": 4.65625, "learning_rate": 4.978812674199587e-05, "loss": 0.3374, "step": 1645 }, { "epoch": 10.22429906542056, "grad_norm": 4.625, "learning_rate": 4.9787806540713546e-05, "loss": 0.3573, "step": 1646 }, { "epoch": 10.230529595015577, "grad_norm": 2.765625, "learning_rate": 4.9787486098687076e-05, "loss": 0.249, "step": 1647 }, { "epoch": 10.236760124610592, "grad_norm": 2.515625, "learning_rate": 4.97871654159196e-05, "loss": 0.3293, "step": 1648 }, { "epoch": 10.242990654205608, "grad_norm": 5.53125, "learning_rate": 4.978684449241421e-05, "loss": 0.3728, "step": 1649 }, { "epoch": 10.249221183800623, "grad_norm": 4.0, "learning_rate": 4.978652332817403e-05, "loss": 0.2792, "step": 1650 }, { "epoch": 10.255451713395638, "grad_norm": 3.359375, "learning_rate": 4.978620192320218e-05, "loss": 0.3799, "step": 1651 }, { "epoch": 10.261682242990654, "grad_norm": 5.9375, "learning_rate": 4.978588027750178e-05, "loss": 0.328, "step": 1652 }, { "epoch": 10.26791277258567, "grad_norm": 3.984375, "learning_rate": 4.978555839107597e-05, "loss": 0.2989, "step": 1653 }, { "epoch": 10.274143302180685, "grad_norm": 5.3125, "learning_rate": 4.9785236263927854e-05, "loss": 0.2172, "step": 1654 }, { "epoch": 10.280373831775702, "grad_norm": 3.4375, "learning_rate": 4.978491389606057e-05, "loss": 0.3088, "step": 1655 }, { "epoch": 10.286604361370717, "grad_norm": 3.6875, "learning_rate": 4.978459128747724e-05, "loss": 0.2624, "step": 1656 }, { "epoch": 10.292834890965732, "grad_norm": 6.125, "learning_rate": 4.9784268438181016e-05, "loss": 0.4416, "step": 1657 }, { "epoch": 10.299065420560748, "grad_norm": 5.1875, "learning_rate": 4.978394534817502e-05, "loss": 0.2592, "step": 1658 }, { "epoch": 10.305295950155763, "grad_norm": 5.21875, "learning_rate": 4.978362201746239e-05, "loss": 0.4515, "step": 1659 }, { "epoch": 10.311526479750778, "grad_norm": 3.21875, "learning_rate": 4.978329844604627e-05, "loss": 0.2882, "step": 1660 }, { "epoch": 10.317757009345794, "grad_norm": 5.84375, "learning_rate": 4.97829746339298e-05, "loss": 0.4842, "step": 1661 }, { "epoch": 10.32398753894081, "grad_norm": 6.25, "learning_rate": 4.978265058111613e-05, "loss": 0.5348, "step": 1662 }, { "epoch": 10.330218068535826, "grad_norm": 7.15625, "learning_rate": 4.978232628760841e-05, "loss": 0.5213, "step": 1663 }, { "epoch": 10.336448598130842, "grad_norm": 2.265625, "learning_rate": 4.978200175340977e-05, "loss": 0.3275, "step": 1664 }, { "epoch": 10.342679127725857, "grad_norm": 2.53125, "learning_rate": 4.978167697852338e-05, "loss": 0.2535, "step": 1665 }, { "epoch": 10.348909657320872, "grad_norm": 2.328125, "learning_rate": 4.9781351962952384e-05, "loss": 0.3194, "step": 1666 }, { "epoch": 10.355140186915888, "grad_norm": 3.890625, "learning_rate": 4.978102670669995e-05, "loss": 0.2694, "step": 1667 }, { "epoch": 10.361370716510903, "grad_norm": 4.625, "learning_rate": 4.978070120976923e-05, "loss": 0.3476, "step": 1668 }, { "epoch": 10.367601246105918, "grad_norm": 2.578125, "learning_rate": 4.9780375472163385e-05, "loss": 0.3262, "step": 1669 }, { "epoch": 10.373831775700934, "grad_norm": 3.828125, "learning_rate": 4.978004949388558e-05, "loss": 0.4028, "step": 1670 }, { "epoch": 10.38006230529595, "grad_norm": 6.28125, "learning_rate": 4.977972327493899e-05, "loss": 0.3731, "step": 1671 }, { "epoch": 10.386292834890966, "grad_norm": 4.625, "learning_rate": 4.9779396815326765e-05, "loss": 0.4095, "step": 1672 }, { "epoch": 10.392523364485982, "grad_norm": 3.171875, "learning_rate": 4.977907011505208e-05, "loss": 0.2811, "step": 1673 }, { "epoch": 10.398753894080997, "grad_norm": 4.0625, "learning_rate": 4.977874317411813e-05, "loss": 0.2559, "step": 1674 }, { "epoch": 10.404984423676012, "grad_norm": 4.1875, "learning_rate": 4.977841599252806e-05, "loss": 0.3463, "step": 1675 }, { "epoch": 10.411214953271028, "grad_norm": 3.78125, "learning_rate": 4.977808857028506e-05, "loss": 0.3652, "step": 1676 }, { "epoch": 10.417445482866043, "grad_norm": 4.78125, "learning_rate": 4.977776090739231e-05, "loss": 0.4156, "step": 1677 }, { "epoch": 10.423676012461058, "grad_norm": 3.796875, "learning_rate": 4.9777433003853e-05, "loss": 0.2205, "step": 1678 }, { "epoch": 10.429906542056075, "grad_norm": 4.21875, "learning_rate": 4.977710485967031e-05, "loss": 0.2415, "step": 1679 }, { "epoch": 10.43613707165109, "grad_norm": 4.15625, "learning_rate": 4.977677647484741e-05, "loss": 0.449, "step": 1680 }, { "epoch": 10.442367601246106, "grad_norm": 3.421875, "learning_rate": 4.977644784938752e-05, "loss": 0.3045, "step": 1681 }, { "epoch": 10.448598130841122, "grad_norm": 3.359375, "learning_rate": 4.9776118983293805e-05, "loss": 0.3476, "step": 1682 }, { "epoch": 10.454828660436137, "grad_norm": 5.96875, "learning_rate": 4.9775789876569476e-05, "loss": 0.2566, "step": 1683 }, { "epoch": 10.461059190031152, "grad_norm": 4.21875, "learning_rate": 4.9775460529217724e-05, "loss": 0.2245, "step": 1684 }, { "epoch": 10.467289719626168, "grad_norm": 2.765625, "learning_rate": 4.9775130941241747e-05, "loss": 0.3771, "step": 1685 }, { "epoch": 10.473520249221183, "grad_norm": 3.34375, "learning_rate": 4.977480111264474e-05, "loss": 0.2818, "step": 1686 }, { "epoch": 10.4797507788162, "grad_norm": 2.8125, "learning_rate": 4.977447104342992e-05, "loss": 0.3472, "step": 1687 }, { "epoch": 10.485981308411215, "grad_norm": 2.515625, "learning_rate": 4.9774140733600475e-05, "loss": 0.3812, "step": 1688 }, { "epoch": 10.49221183800623, "grad_norm": 3.28125, "learning_rate": 4.977381018315963e-05, "loss": 0.3612, "step": 1689 }, { "epoch": 10.498442367601246, "grad_norm": 2.625, "learning_rate": 4.977347939211058e-05, "loss": 0.193, "step": 1690 }, { "epoch": 10.504672897196262, "grad_norm": 3.234375, "learning_rate": 4.9773148360456555e-05, "loss": 0.4167, "step": 1691 }, { "epoch": 10.510903426791277, "grad_norm": 5.0, "learning_rate": 4.977281708820076e-05, "loss": 0.4475, "step": 1692 }, { "epoch": 10.517133956386292, "grad_norm": 4.0625, "learning_rate": 4.977248557534641e-05, "loss": 0.5483, "step": 1693 }, { "epoch": 10.523364485981308, "grad_norm": 3.28125, "learning_rate": 4.9772153821896725e-05, "loss": 0.5637, "step": 1694 }, { "epoch": 10.529595015576325, "grad_norm": 3.609375, "learning_rate": 4.977182182785494e-05, "loss": 0.406, "step": 1695 }, { "epoch": 10.53582554517134, "grad_norm": 3.203125, "learning_rate": 4.977148959322426e-05, "loss": 0.2589, "step": 1696 }, { "epoch": 10.542056074766355, "grad_norm": 3.09375, "learning_rate": 4.977115711800793e-05, "loss": 0.3682, "step": 1697 }, { "epoch": 10.54828660436137, "grad_norm": 4.125, "learning_rate": 4.977082440220917e-05, "loss": 0.3624, "step": 1698 }, { "epoch": 10.554517133956386, "grad_norm": 3.953125, "learning_rate": 4.977049144583121e-05, "loss": 0.3664, "step": 1699 }, { "epoch": 10.560747663551401, "grad_norm": 3.796875, "learning_rate": 4.9770158248877286e-05, "loss": 0.2837, "step": 1700 }, { "epoch": 10.566978193146417, "grad_norm": 4.90625, "learning_rate": 4.976982481135063e-05, "loss": 0.6075, "step": 1701 }, { "epoch": 10.573208722741432, "grad_norm": 4.3125, "learning_rate": 4.9769491133254485e-05, "loss": 0.36, "step": 1702 }, { "epoch": 10.57943925233645, "grad_norm": 3.140625, "learning_rate": 4.976915721459209e-05, "loss": 0.2108, "step": 1703 }, { "epoch": 10.585669781931465, "grad_norm": 3.453125, "learning_rate": 4.976882305536669e-05, "loss": 0.5151, "step": 1704 }, { "epoch": 10.59190031152648, "grad_norm": 2.796875, "learning_rate": 4.976848865558153e-05, "loss": 0.2693, "step": 1705 }, { "epoch": 10.598130841121495, "grad_norm": 5.28125, "learning_rate": 4.9768154015239854e-05, "loss": 0.3783, "step": 1706 }, { "epoch": 10.60436137071651, "grad_norm": 3.8125, "learning_rate": 4.9767819134344914e-05, "loss": 0.403, "step": 1707 }, { "epoch": 10.610591900311526, "grad_norm": 2.703125, "learning_rate": 4.976748401289997e-05, "loss": 0.3663, "step": 1708 }, { "epoch": 10.616822429906541, "grad_norm": 2.21875, "learning_rate": 4.976714865090827e-05, "loss": 0.2542, "step": 1709 }, { "epoch": 10.623052959501557, "grad_norm": 3.453125, "learning_rate": 4.976681304837307e-05, "loss": 0.3907, "step": 1710 }, { "epoch": 10.629283489096574, "grad_norm": 4.21875, "learning_rate": 4.9766477205297634e-05, "loss": 0.411, "step": 1711 }, { "epoch": 10.63551401869159, "grad_norm": 4.625, "learning_rate": 4.976614112168522e-05, "loss": 0.2636, "step": 1712 }, { "epoch": 10.641744548286605, "grad_norm": 2.421875, "learning_rate": 4.9765804797539086e-05, "loss": 0.2373, "step": 1713 }, { "epoch": 10.64797507788162, "grad_norm": 2.5625, "learning_rate": 4.9765468232862505e-05, "loss": 0.254, "step": 1714 }, { "epoch": 10.654205607476635, "grad_norm": 3.046875, "learning_rate": 4.976513142765875e-05, "loss": 0.4048, "step": 1715 }, { "epoch": 10.66043613707165, "grad_norm": 3.203125, "learning_rate": 4.976479438193109e-05, "loss": 0.3122, "step": 1716 }, { "epoch": 10.666666666666666, "grad_norm": 3.34375, "learning_rate": 4.976445709568279e-05, "loss": 0.3337, "step": 1717 }, { "epoch": 10.672897196261681, "grad_norm": 3.78125, "learning_rate": 4.976411956891714e-05, "loss": 0.4724, "step": 1718 }, { "epoch": 10.679127725856699, "grad_norm": 2.734375, "learning_rate": 4.976378180163741e-05, "loss": 0.3219, "step": 1719 }, { "epoch": 10.685358255451714, "grad_norm": 3.109375, "learning_rate": 4.976344379384688e-05, "loss": 0.413, "step": 1720 }, { "epoch": 10.69158878504673, "grad_norm": 2.421875, "learning_rate": 4.976310554554883e-05, "loss": 0.2792, "step": 1721 }, { "epoch": 10.697819314641745, "grad_norm": 3.171875, "learning_rate": 4.976276705674655e-05, "loss": 0.3504, "step": 1722 }, { "epoch": 10.70404984423676, "grad_norm": 4.125, "learning_rate": 4.976242832744332e-05, "loss": 0.3641, "step": 1723 }, { "epoch": 10.710280373831775, "grad_norm": 3.53125, "learning_rate": 4.976208935764245e-05, "loss": 0.455, "step": 1724 }, { "epoch": 10.71651090342679, "grad_norm": 2.890625, "learning_rate": 4.976175014734721e-05, "loss": 0.4284, "step": 1725 }, { "epoch": 10.722741433021806, "grad_norm": 3.125, "learning_rate": 4.976141069656091e-05, "loss": 0.316, "step": 1726 }, { "epoch": 10.728971962616823, "grad_norm": 2.234375, "learning_rate": 4.976107100528683e-05, "loss": 0.251, "step": 1727 }, { "epoch": 10.735202492211839, "grad_norm": 4.71875, "learning_rate": 4.9760731073528285e-05, "loss": 0.5486, "step": 1728 }, { "epoch": 10.741433021806854, "grad_norm": 2.640625, "learning_rate": 4.976039090128857e-05, "loss": 0.1993, "step": 1729 }, { "epoch": 10.74766355140187, "grad_norm": 4.4375, "learning_rate": 4.9760050488570984e-05, "loss": 0.4187, "step": 1730 }, { "epoch": 10.753894080996885, "grad_norm": 2.65625, "learning_rate": 4.975970983537884e-05, "loss": 0.3085, "step": 1731 }, { "epoch": 10.7601246105919, "grad_norm": 3.8125, "learning_rate": 4.9759368941715445e-05, "loss": 0.3054, "step": 1732 }, { "epoch": 10.766355140186915, "grad_norm": 2.71875, "learning_rate": 4.975902780758411e-05, "loss": 0.2451, "step": 1733 }, { "epoch": 10.77258566978193, "grad_norm": 2.859375, "learning_rate": 4.9758686432988155e-05, "loss": 0.2737, "step": 1734 }, { "epoch": 10.778816199376948, "grad_norm": 2.859375, "learning_rate": 4.975834481793088e-05, "loss": 0.2639, "step": 1735 }, { "epoch": 10.785046728971963, "grad_norm": 4.59375, "learning_rate": 4.975800296241561e-05, "loss": 0.6028, "step": 1736 }, { "epoch": 10.791277258566979, "grad_norm": 3.375, "learning_rate": 4.975766086644567e-05, "loss": 0.3264, "step": 1737 }, { "epoch": 10.797507788161994, "grad_norm": 3.65625, "learning_rate": 4.9757318530024376e-05, "loss": 0.4381, "step": 1738 }, { "epoch": 10.80373831775701, "grad_norm": 3.3125, "learning_rate": 4.975697595315506e-05, "loss": 0.3753, "step": 1739 }, { "epoch": 10.809968847352025, "grad_norm": 3.8125, "learning_rate": 4.975663313584104e-05, "loss": 0.2354, "step": 1740 }, { "epoch": 10.81619937694704, "grad_norm": 3.546875, "learning_rate": 4.975629007808565e-05, "loss": 0.2258, "step": 1741 }, { "epoch": 10.822429906542055, "grad_norm": 2.171875, "learning_rate": 4.9755946779892225e-05, "loss": 0.201, "step": 1742 }, { "epoch": 10.828660436137072, "grad_norm": 6.34375, "learning_rate": 4.97556032412641e-05, "loss": 0.5661, "step": 1743 }, { "epoch": 10.834890965732088, "grad_norm": 7.375, "learning_rate": 4.9755259462204604e-05, "loss": 0.3587, "step": 1744 }, { "epoch": 10.841121495327103, "grad_norm": 5.96875, "learning_rate": 4.975491544271708e-05, "loss": 0.4029, "step": 1745 }, { "epoch": 10.847352024922118, "grad_norm": 4.0, "learning_rate": 4.975457118280487e-05, "loss": 0.2886, "step": 1746 }, { "epoch": 10.853582554517134, "grad_norm": 5.3125, "learning_rate": 4.975422668247131e-05, "loss": 0.3636, "step": 1747 }, { "epoch": 10.85981308411215, "grad_norm": 6.0, "learning_rate": 4.975388194171975e-05, "loss": 0.3863, "step": 1748 }, { "epoch": 10.866043613707165, "grad_norm": 5.8125, "learning_rate": 4.9753536960553545e-05, "loss": 0.2802, "step": 1749 }, { "epoch": 10.87227414330218, "grad_norm": 5.03125, "learning_rate": 4.9753191738976045e-05, "loss": 0.388, "step": 1750 }, { "epoch": 10.878504672897197, "grad_norm": 3.1875, "learning_rate": 4.97528462769906e-05, "loss": 0.2512, "step": 1751 }, { "epoch": 10.884735202492212, "grad_norm": 4.75, "learning_rate": 4.9752500574600556e-05, "loss": 0.3566, "step": 1752 }, { "epoch": 10.890965732087228, "grad_norm": 2.546875, "learning_rate": 4.975215463180929e-05, "loss": 0.1771, "step": 1753 }, { "epoch": 10.897196261682243, "grad_norm": 2.75, "learning_rate": 4.975180844862014e-05, "loss": 0.2113, "step": 1754 }, { "epoch": 10.903426791277258, "grad_norm": 3.375, "learning_rate": 4.975146202503648e-05, "loss": 0.2725, "step": 1755 }, { "epoch": 10.909657320872274, "grad_norm": 4.84375, "learning_rate": 4.975111536106167e-05, "loss": 0.387, "step": 1756 }, { "epoch": 10.91588785046729, "grad_norm": 5.3125, "learning_rate": 4.9750768456699095e-05, "loss": 0.2607, "step": 1757 }, { "epoch": 10.922118380062305, "grad_norm": 6.0, "learning_rate": 4.975042131195209e-05, "loss": 0.2711, "step": 1758 }, { "epoch": 10.928348909657322, "grad_norm": 4.90625, "learning_rate": 4.975007392682406e-05, "loss": 0.386, "step": 1759 }, { "epoch": 10.934579439252337, "grad_norm": 3.578125, "learning_rate": 4.974972630131836e-05, "loss": 0.3828, "step": 1760 }, { "epoch": 10.940809968847352, "grad_norm": 4.1875, "learning_rate": 4.974937843543836e-05, "loss": 0.291, "step": 1761 }, { "epoch": 10.947040498442368, "grad_norm": 7.3125, "learning_rate": 4.9749030329187466e-05, "loss": 0.3035, "step": 1762 }, { "epoch": 10.953271028037383, "grad_norm": 7.09375, "learning_rate": 4.9748681982569035e-05, "loss": 0.3721, "step": 1763 }, { "epoch": 10.959501557632398, "grad_norm": 4.25, "learning_rate": 4.974833339558645e-05, "loss": 0.4957, "step": 1764 }, { "epoch": 10.965732087227414, "grad_norm": 3.359375, "learning_rate": 4.974798456824312e-05, "loss": 0.3321, "step": 1765 }, { "epoch": 10.97196261682243, "grad_norm": 7.15625, "learning_rate": 4.974763550054241e-05, "loss": 0.3322, "step": 1766 }, { "epoch": 10.978193146417446, "grad_norm": 5.4375, "learning_rate": 4.974728619248772e-05, "loss": 0.283, "step": 1767 }, { "epoch": 10.984423676012462, "grad_norm": 3.3125, "learning_rate": 4.974693664408243e-05, "loss": 0.2918, "step": 1768 }, { "epoch": 10.990654205607477, "grad_norm": 4.0, "learning_rate": 4.9746586855329944e-05, "loss": 0.4142, "step": 1769 }, { "epoch": 10.996884735202492, "grad_norm": 4.625, "learning_rate": 4.9746236826233674e-05, "loss": 0.339, "step": 1770 }, { "epoch": 11.0, "grad_norm": 3.5625, "learning_rate": 4.9745886556796995e-05, "loss": 0.1028, "step": 1771 }, { "epoch": 11.006230529595015, "grad_norm": 6.5, "learning_rate": 4.9745536047023324e-05, "loss": 0.4129, "step": 1772 }, { "epoch": 11.01246105919003, "grad_norm": 5.34375, "learning_rate": 4.9745185296916054e-05, "loss": 0.32, "step": 1773 }, { "epoch": 11.018691588785046, "grad_norm": 3.25, "learning_rate": 4.97448343064786e-05, "loss": 0.2749, "step": 1774 }, { "epoch": 11.024922118380061, "grad_norm": 3.515625, "learning_rate": 4.974448307571437e-05, "loss": 0.4339, "step": 1775 }, { "epoch": 11.031152647975079, "grad_norm": 4.375, "learning_rate": 4.974413160462678e-05, "loss": 0.2843, "step": 1776 }, { "epoch": 11.037383177570094, "grad_norm": 4.34375, "learning_rate": 4.974377989321923e-05, "loss": 0.3586, "step": 1777 }, { "epoch": 11.04361370716511, "grad_norm": 2.84375, "learning_rate": 4.974342794149515e-05, "loss": 0.2819, "step": 1778 }, { "epoch": 11.049844236760125, "grad_norm": 2.125, "learning_rate": 4.974307574945794e-05, "loss": 0.2186, "step": 1779 }, { "epoch": 11.05607476635514, "grad_norm": 3.90625, "learning_rate": 4.974272331711104e-05, "loss": 0.493, "step": 1780 }, { "epoch": 11.062305295950155, "grad_norm": 2.234375, "learning_rate": 4.974237064445787e-05, "loss": 0.2194, "step": 1781 }, { "epoch": 11.06853582554517, "grad_norm": 2.484375, "learning_rate": 4.974201773150184e-05, "loss": 0.2184, "step": 1782 }, { "epoch": 11.074766355140186, "grad_norm": 3.578125, "learning_rate": 4.9741664578246394e-05, "loss": 0.2962, "step": 1783 }, { "epoch": 11.080996884735203, "grad_norm": 3.796875, "learning_rate": 4.974131118469495e-05, "loss": 0.286, "step": 1784 }, { "epoch": 11.087227414330219, "grad_norm": 4.21875, "learning_rate": 4.9740957550850956e-05, "loss": 0.2256, "step": 1785 }, { "epoch": 11.093457943925234, "grad_norm": 4.0, "learning_rate": 4.974060367671783e-05, "loss": 0.2587, "step": 1786 }, { "epoch": 11.09968847352025, "grad_norm": 2.6875, "learning_rate": 4.974024956229901e-05, "loss": 0.3186, "step": 1787 }, { "epoch": 11.105919003115265, "grad_norm": 4.09375, "learning_rate": 4.973989520759795e-05, "loss": 0.3566, "step": 1788 }, { "epoch": 11.11214953271028, "grad_norm": 4.03125, "learning_rate": 4.973954061261809e-05, "loss": 0.4185, "step": 1789 }, { "epoch": 11.118380062305295, "grad_norm": 5.4375, "learning_rate": 4.973918577736285e-05, "loss": 0.439, "step": 1790 }, { "epoch": 11.12461059190031, "grad_norm": 2.765625, "learning_rate": 4.97388307018357e-05, "loss": 0.3391, "step": 1791 }, { "epoch": 11.130841121495328, "grad_norm": 4.0625, "learning_rate": 4.973847538604008e-05, "loss": 0.4999, "step": 1792 }, { "epoch": 11.137071651090343, "grad_norm": 3.546875, "learning_rate": 4.973811982997944e-05, "loss": 0.4053, "step": 1793 }, { "epoch": 11.143302180685358, "grad_norm": 3.046875, "learning_rate": 4.973776403365724e-05, "loss": 0.3265, "step": 1794 }, { "epoch": 11.149532710280374, "grad_norm": 4.28125, "learning_rate": 4.973740799707692e-05, "loss": 0.4779, "step": 1795 }, { "epoch": 11.15576323987539, "grad_norm": 2.203125, "learning_rate": 4.973705172024196e-05, "loss": 0.2573, "step": 1796 }, { "epoch": 11.161993769470405, "grad_norm": 4.53125, "learning_rate": 4.9736695203155805e-05, "loss": 0.2318, "step": 1797 }, { "epoch": 11.16822429906542, "grad_norm": 4.78125, "learning_rate": 4.9736338445821925e-05, "loss": 0.5799, "step": 1798 }, { "epoch": 11.174454828660435, "grad_norm": 2.640625, "learning_rate": 4.973598144824377e-05, "loss": 0.2384, "step": 1799 }, { "epoch": 11.180685358255452, "grad_norm": 4.6875, "learning_rate": 4.973562421042483e-05, "loss": 0.4389, "step": 1800 }, { "epoch": 11.186915887850468, "grad_norm": 4.90625, "learning_rate": 4.973526673236856e-05, "loss": 0.3759, "step": 1801 }, { "epoch": 11.193146417445483, "grad_norm": 2.15625, "learning_rate": 4.9734909014078435e-05, "loss": 0.3147, "step": 1802 }, { "epoch": 11.199376947040498, "grad_norm": 1.8828125, "learning_rate": 4.973455105555793e-05, "loss": 0.2257, "step": 1803 }, { "epoch": 11.205607476635514, "grad_norm": 3.734375, "learning_rate": 4.973419285681052e-05, "loss": 0.4167, "step": 1804 }, { "epoch": 11.21183800623053, "grad_norm": 2.515625, "learning_rate": 4.973383441783969e-05, "loss": 0.2767, "step": 1805 }, { "epoch": 11.218068535825545, "grad_norm": 3.5625, "learning_rate": 4.973347573864891e-05, "loss": 0.3753, "step": 1806 }, { "epoch": 11.22429906542056, "grad_norm": 2.15625, "learning_rate": 4.9733116819241664e-05, "loss": 0.2748, "step": 1807 }, { "epoch": 11.230529595015577, "grad_norm": 2.78125, "learning_rate": 4.973275765962145e-05, "loss": 0.3352, "step": 1808 }, { "epoch": 11.236760124610592, "grad_norm": 2.53125, "learning_rate": 4.973239825979175e-05, "loss": 0.3357, "step": 1809 }, { "epoch": 11.242990654205608, "grad_norm": 3.6875, "learning_rate": 4.973203861975605e-05, "loss": 0.2944, "step": 1810 }, { "epoch": 11.249221183800623, "grad_norm": 2.90625, "learning_rate": 4.973167873951785e-05, "loss": 0.2527, "step": 1811 }, { "epoch": 11.255451713395638, "grad_norm": 3.0625, "learning_rate": 4.973131861908064e-05, "loss": 0.296, "step": 1812 }, { "epoch": 11.261682242990654, "grad_norm": 3.265625, "learning_rate": 4.973095825844793e-05, "loss": 0.3883, "step": 1813 }, { "epoch": 11.26791277258567, "grad_norm": 4.75, "learning_rate": 4.973059765762319e-05, "loss": 0.2579, "step": 1814 }, { "epoch": 11.274143302180685, "grad_norm": 5.6875, "learning_rate": 4.973023681660996e-05, "loss": 0.4065, "step": 1815 }, { "epoch": 11.280373831775702, "grad_norm": 3.375, "learning_rate": 4.972987573541172e-05, "loss": 0.2592, "step": 1816 }, { "epoch": 11.286604361370717, "grad_norm": 4.03125, "learning_rate": 4.9729514414031985e-05, "loss": 0.2207, "step": 1817 }, { "epoch": 11.292834890965732, "grad_norm": 5.59375, "learning_rate": 4.972915285247426e-05, "loss": 0.3439, "step": 1818 }, { "epoch": 11.299065420560748, "grad_norm": 6.375, "learning_rate": 4.9728791050742064e-05, "loss": 0.2102, "step": 1819 }, { "epoch": 11.305295950155763, "grad_norm": 3.4375, "learning_rate": 4.9728429008838906e-05, "loss": 0.2149, "step": 1820 }, { "epoch": 11.311526479750778, "grad_norm": 3.453125, "learning_rate": 4.97280667267683e-05, "loss": 0.2055, "step": 1821 }, { "epoch": 11.317757009345794, "grad_norm": 5.6875, "learning_rate": 4.972770420453376e-05, "loss": 0.3262, "step": 1822 }, { "epoch": 11.32398753894081, "grad_norm": 6.6875, "learning_rate": 4.972734144213882e-05, "loss": 0.3478, "step": 1823 }, { "epoch": 11.330218068535826, "grad_norm": 2.828125, "learning_rate": 4.9726978439587e-05, "loss": 0.2307, "step": 1824 }, { "epoch": 11.336448598130842, "grad_norm": 3.078125, "learning_rate": 4.972661519688182e-05, "loss": 0.3896, "step": 1825 }, { "epoch": 11.342679127725857, "grad_norm": 5.71875, "learning_rate": 4.9726251714026804e-05, "loss": 0.2397, "step": 1826 }, { "epoch": 11.348909657320872, "grad_norm": 5.46875, "learning_rate": 4.97258879910255e-05, "loss": 0.3574, "step": 1827 }, { "epoch": 11.355140186915888, "grad_norm": 6.96875, "learning_rate": 4.972552402788142e-05, "loss": 0.2377, "step": 1828 }, { "epoch": 11.361370716510903, "grad_norm": 3.25, "learning_rate": 4.972515982459812e-05, "loss": 0.2967, "step": 1829 }, { "epoch": 11.367601246105918, "grad_norm": 3.6875, "learning_rate": 4.972479538117911e-05, "loss": 0.5257, "step": 1830 }, { "epoch": 11.373831775700934, "grad_norm": 3.8125, "learning_rate": 4.972443069762795e-05, "loss": 0.3176, "step": 1831 }, { "epoch": 11.38006230529595, "grad_norm": 6.53125, "learning_rate": 4.9724065773948185e-05, "loss": 0.3765, "step": 1832 }, { "epoch": 11.386292834890966, "grad_norm": 3.21875, "learning_rate": 4.9723700610143344e-05, "loss": 0.4771, "step": 1833 }, { "epoch": 11.392523364485982, "grad_norm": 3.359375, "learning_rate": 4.972333520621698e-05, "loss": 0.2672, "step": 1834 }, { "epoch": 11.398753894080997, "grad_norm": 2.5, "learning_rate": 4.972296956217265e-05, "loss": 0.2363, "step": 1835 }, { "epoch": 11.404984423676012, "grad_norm": 4.71875, "learning_rate": 4.972260367801389e-05, "loss": 0.3016, "step": 1836 }, { "epoch": 11.411214953271028, "grad_norm": 6.5625, "learning_rate": 4.972223755374426e-05, "loss": 0.4322, "step": 1837 }, { "epoch": 11.417445482866043, "grad_norm": 3.609375, "learning_rate": 4.972187118936732e-05, "loss": 0.4128, "step": 1838 }, { "epoch": 11.423676012461058, "grad_norm": 3.859375, "learning_rate": 4.9721504584886624e-05, "loss": 0.3934, "step": 1839 }, { "epoch": 11.429906542056075, "grad_norm": 2.84375, "learning_rate": 4.9721137740305734e-05, "loss": 0.2586, "step": 1840 }, { "epoch": 11.43613707165109, "grad_norm": 4.28125, "learning_rate": 4.972077065562821e-05, "loss": 0.4579, "step": 1841 }, { "epoch": 11.442367601246106, "grad_norm": 3.703125, "learning_rate": 4.972040333085763e-05, "loss": 0.2475, "step": 1842 }, { "epoch": 11.448598130841122, "grad_norm": 3.921875, "learning_rate": 4.972003576599754e-05, "loss": 0.3643, "step": 1843 }, { "epoch": 11.454828660436137, "grad_norm": 3.96875, "learning_rate": 4.971966796105153e-05, "loss": 0.3893, "step": 1844 }, { "epoch": 11.461059190031152, "grad_norm": 6.53125, "learning_rate": 4.971929991602315e-05, "loss": 0.3468, "step": 1845 }, { "epoch": 11.467289719626168, "grad_norm": 6.53125, "learning_rate": 4.9718931630916e-05, "loss": 0.3757, "step": 1846 }, { "epoch": 11.473520249221183, "grad_norm": 3.765625, "learning_rate": 4.971856310573365e-05, "loss": 0.3579, "step": 1847 }, { "epoch": 11.4797507788162, "grad_norm": 3.5625, "learning_rate": 4.9718194340479663e-05, "loss": 0.2603, "step": 1848 }, { "epoch": 11.485981308411215, "grad_norm": 5.0625, "learning_rate": 4.971782533515763e-05, "loss": 0.2998, "step": 1849 }, { "epoch": 11.49221183800623, "grad_norm": 4.53125, "learning_rate": 4.971745608977114e-05, "loss": 0.2668, "step": 1850 }, { "epoch": 11.498442367601246, "grad_norm": 3.84375, "learning_rate": 4.971708660432378e-05, "loss": 0.35, "step": 1851 }, { "epoch": 11.504672897196262, "grad_norm": 2.53125, "learning_rate": 4.971671687881913e-05, "loss": 0.3376, "step": 1852 }, { "epoch": 11.510903426791277, "grad_norm": 5.0625, "learning_rate": 4.971634691326079e-05, "loss": 0.2987, "step": 1853 }, { "epoch": 11.517133956386292, "grad_norm": 3.421875, "learning_rate": 4.971597670765234e-05, "loss": 0.2326, "step": 1854 }, { "epoch": 11.523364485981308, "grad_norm": 4.8125, "learning_rate": 4.9715606261997386e-05, "loss": 0.2439, "step": 1855 }, { "epoch": 11.529595015576325, "grad_norm": 3.390625, "learning_rate": 4.9715235576299523e-05, "loss": 0.3162, "step": 1856 }, { "epoch": 11.53582554517134, "grad_norm": 5.0, "learning_rate": 4.9714864650562356e-05, "loss": 0.4207, "step": 1857 }, { "epoch": 11.542056074766355, "grad_norm": 2.78125, "learning_rate": 4.9714493484789476e-05, "loss": 0.2445, "step": 1858 }, { "epoch": 11.54828660436137, "grad_norm": 3.34375, "learning_rate": 4.971412207898451e-05, "loss": 0.3018, "step": 1859 }, { "epoch": 11.554517133956386, "grad_norm": 2.4375, "learning_rate": 4.971375043315103e-05, "loss": 0.2303, "step": 1860 }, { "epoch": 11.560747663551401, "grad_norm": 4.09375, "learning_rate": 4.971337854729268e-05, "loss": 0.3378, "step": 1861 }, { "epoch": 11.566978193146417, "grad_norm": 4.53125, "learning_rate": 4.9713006421413047e-05, "loss": 0.4347, "step": 1862 }, { "epoch": 11.573208722741432, "grad_norm": 4.3125, "learning_rate": 4.971263405551576e-05, "loss": 0.3612, "step": 1863 }, { "epoch": 11.57943925233645, "grad_norm": 3.4375, "learning_rate": 4.971226144960443e-05, "loss": 0.375, "step": 1864 }, { "epoch": 11.585669781931465, "grad_norm": 3.0, "learning_rate": 4.971188860368268e-05, "loss": 0.2507, "step": 1865 }, { "epoch": 11.59190031152648, "grad_norm": 4.9375, "learning_rate": 4.971151551775413e-05, "loss": 0.2766, "step": 1866 }, { "epoch": 11.598130841121495, "grad_norm": 2.0625, "learning_rate": 4.971114219182239e-05, "loss": 0.2393, "step": 1867 }, { "epoch": 11.60436137071651, "grad_norm": 2.5, "learning_rate": 4.9710768625891114e-05, "loss": 0.3786, "step": 1868 }, { "epoch": 11.610591900311526, "grad_norm": 3.53125, "learning_rate": 4.97103948199639e-05, "loss": 0.2394, "step": 1869 }, { "epoch": 11.616822429906541, "grad_norm": 3.125, "learning_rate": 4.9710020774044394e-05, "loss": 0.2407, "step": 1870 }, { "epoch": 11.623052959501557, "grad_norm": 3.71875, "learning_rate": 4.970964648813623e-05, "loss": 0.2156, "step": 1871 }, { "epoch": 11.629283489096574, "grad_norm": 4.1875, "learning_rate": 4.9709271962243044e-05, "loss": 0.3862, "step": 1872 }, { "epoch": 11.63551401869159, "grad_norm": 3.1875, "learning_rate": 4.970889719636847e-05, "loss": 0.2339, "step": 1873 }, { "epoch": 11.641744548286605, "grad_norm": 2.171875, "learning_rate": 4.9708522190516135e-05, "loss": 0.2157, "step": 1874 }, { "epoch": 11.64797507788162, "grad_norm": 2.796875, "learning_rate": 4.9708146944689705e-05, "loss": 0.2994, "step": 1875 }, { "epoch": 11.654205607476635, "grad_norm": 2.828125, "learning_rate": 4.9707771458892804e-05, "loss": 0.335, "step": 1876 }, { "epoch": 11.66043613707165, "grad_norm": 3.25, "learning_rate": 4.970739573312909e-05, "loss": 0.3362, "step": 1877 }, { "epoch": 11.666666666666666, "grad_norm": 3.328125, "learning_rate": 4.970701976740222e-05, "loss": 0.313, "step": 1878 }, { "epoch": 11.672897196261681, "grad_norm": 3.390625, "learning_rate": 4.970664356171583e-05, "loss": 0.3884, "step": 1879 }, { "epoch": 11.679127725856699, "grad_norm": 3.546875, "learning_rate": 4.9706267116073565e-05, "loss": 0.2666, "step": 1880 }, { "epoch": 11.685358255451714, "grad_norm": 2.9375, "learning_rate": 4.970589043047911e-05, "loss": 0.3425, "step": 1881 }, { "epoch": 11.69158878504673, "grad_norm": 3.859375, "learning_rate": 4.970551350493611e-05, "loss": 0.2864, "step": 1882 }, { "epoch": 11.697819314641745, "grad_norm": 2.59375, "learning_rate": 4.9705136339448214e-05, "loss": 0.2726, "step": 1883 }, { "epoch": 11.70404984423676, "grad_norm": 3.0625, "learning_rate": 4.97047589340191e-05, "loss": 0.3397, "step": 1884 }, { "epoch": 11.710280373831775, "grad_norm": 3.296875, "learning_rate": 4.9704381288652436e-05, "loss": 0.2343, "step": 1885 }, { "epoch": 11.71651090342679, "grad_norm": 1.78125, "learning_rate": 4.970400340335187e-05, "loss": 0.2233, "step": 1886 }, { "epoch": 11.722741433021806, "grad_norm": 3.921875, "learning_rate": 4.970362527812109e-05, "loss": 0.4212, "step": 1887 }, { "epoch": 11.728971962616823, "grad_norm": 3.28125, "learning_rate": 4.9703246912963764e-05, "loss": 0.2776, "step": 1888 }, { "epoch": 11.735202492211839, "grad_norm": 3.0625, "learning_rate": 4.970286830788357e-05, "loss": 0.4026, "step": 1889 }, { "epoch": 11.741433021806854, "grad_norm": 3.765625, "learning_rate": 4.970248946288417e-05, "loss": 0.5104, "step": 1890 }, { "epoch": 11.74766355140187, "grad_norm": 3.953125, "learning_rate": 4.970211037796927e-05, "loss": 0.632, "step": 1891 }, { "epoch": 11.753894080996885, "grad_norm": 2.90625, "learning_rate": 4.970173105314252e-05, "loss": 0.3062, "step": 1892 }, { "epoch": 11.7601246105919, "grad_norm": 2.640625, "learning_rate": 4.970135148840763e-05, "loss": 0.2463, "step": 1893 }, { "epoch": 11.766355140186915, "grad_norm": 3.34375, "learning_rate": 4.970097168376827e-05, "loss": 0.3414, "step": 1894 }, { "epoch": 11.77258566978193, "grad_norm": 2.90625, "learning_rate": 4.9700591639228144e-05, "loss": 0.266, "step": 1895 }, { "epoch": 11.778816199376948, "grad_norm": 3.984375, "learning_rate": 4.970021135479093e-05, "loss": 0.3885, "step": 1896 }, { "epoch": 11.785046728971963, "grad_norm": 4.125, "learning_rate": 4.9699830830460324e-05, "loss": 0.3919, "step": 1897 }, { "epoch": 11.791277258566979, "grad_norm": 2.78125, "learning_rate": 4.969945006624003e-05, "loss": 0.2688, "step": 1898 }, { "epoch": 11.797507788161994, "grad_norm": 3.53125, "learning_rate": 4.969906906213373e-05, "loss": 0.323, "step": 1899 }, { "epoch": 11.80373831775701, "grad_norm": 1.5546875, "learning_rate": 4.969868781814514e-05, "loss": 0.2089, "step": 1900 }, { "epoch": 11.809968847352025, "grad_norm": 3.34375, "learning_rate": 4.9698306334277956e-05, "loss": 0.4002, "step": 1901 }, { "epoch": 11.81619937694704, "grad_norm": 3.34375, "learning_rate": 4.969792461053588e-05, "loss": 0.3422, "step": 1902 }, { "epoch": 11.822429906542055, "grad_norm": 4.3125, "learning_rate": 4.969754264692264e-05, "loss": 0.2768, "step": 1903 }, { "epoch": 11.828660436137072, "grad_norm": 3.546875, "learning_rate": 4.969716044344191e-05, "loss": 0.3393, "step": 1904 }, { "epoch": 11.834890965732088, "grad_norm": 4.53125, "learning_rate": 4.9696778000097424e-05, "loss": 0.4317, "step": 1905 }, { "epoch": 11.841121495327103, "grad_norm": 4.28125, "learning_rate": 4.9696395316892895e-05, "loss": 0.3583, "step": 1906 }, { "epoch": 11.847352024922118, "grad_norm": 4.03125, "learning_rate": 4.9696012393832034e-05, "loss": 0.2594, "step": 1907 }, { "epoch": 11.853582554517134, "grad_norm": 3.375, "learning_rate": 4.9695629230918574e-05, "loss": 0.3005, "step": 1908 }, { "epoch": 11.85981308411215, "grad_norm": 3.875, "learning_rate": 4.969524582815622e-05, "loss": 0.2791, "step": 1909 }, { "epoch": 11.866043613707165, "grad_norm": 1.9609375, "learning_rate": 4.969486218554871e-05, "loss": 0.1948, "step": 1910 }, { "epoch": 11.87227414330218, "grad_norm": 3.75, "learning_rate": 4.969447830309975e-05, "loss": 0.3182, "step": 1911 }, { "epoch": 11.878504672897197, "grad_norm": 3.9375, "learning_rate": 4.969409418081309e-05, "loss": 0.4386, "step": 1912 }, { "epoch": 11.884735202492212, "grad_norm": 1.703125, "learning_rate": 4.9693709818692444e-05, "loss": 0.2015, "step": 1913 }, { "epoch": 11.890965732087228, "grad_norm": 3.953125, "learning_rate": 4.969332521674155e-05, "loss": 0.3825, "step": 1914 }, { "epoch": 11.897196261682243, "grad_norm": 3.96875, "learning_rate": 4.9692940374964155e-05, "loss": 0.3792, "step": 1915 }, { "epoch": 11.903426791277258, "grad_norm": 3.0, "learning_rate": 4.969255529336398e-05, "loss": 0.4019, "step": 1916 }, { "epoch": 11.909657320872274, "grad_norm": 2.234375, "learning_rate": 4.9692169971944774e-05, "loss": 0.219, "step": 1917 }, { "epoch": 11.91588785046729, "grad_norm": 4.1875, "learning_rate": 4.969178441071028e-05, "loss": 0.4068, "step": 1918 }, { "epoch": 11.922118380062305, "grad_norm": 2.9375, "learning_rate": 4.9691398609664233e-05, "loss": 0.2436, "step": 1919 }, { "epoch": 11.928348909657322, "grad_norm": 2.765625, "learning_rate": 4.969101256881039e-05, "loss": 0.2341, "step": 1920 }, { "epoch": 11.934579439252337, "grad_norm": 2.71875, "learning_rate": 4.9690626288152504e-05, "loss": 0.262, "step": 1921 }, { "epoch": 11.940809968847352, "grad_norm": 1.9921875, "learning_rate": 4.969023976769431e-05, "loss": 0.1892, "step": 1922 }, { "epoch": 11.947040498442368, "grad_norm": 3.578125, "learning_rate": 4.9689853007439577e-05, "loss": 0.2112, "step": 1923 }, { "epoch": 11.953271028037383, "grad_norm": 3.515625, "learning_rate": 4.968946600739206e-05, "loss": 0.3895, "step": 1924 }, { "epoch": 11.959501557632398, "grad_norm": 3.15625, "learning_rate": 4.9689078767555505e-05, "loss": 0.2395, "step": 1925 }, { "epoch": 11.965732087227414, "grad_norm": 3.84375, "learning_rate": 4.968869128793369e-05, "loss": 0.2977, "step": 1926 }, { "epoch": 11.97196261682243, "grad_norm": 3.21875, "learning_rate": 4.968830356853037e-05, "loss": 0.5132, "step": 1927 }, { "epoch": 11.978193146417446, "grad_norm": 3.078125, "learning_rate": 4.9687915609349305e-05, "loss": 0.2922, "step": 1928 }, { "epoch": 11.984423676012462, "grad_norm": 4.03125, "learning_rate": 4.9687527410394275e-05, "loss": 0.3033, "step": 1929 }, { "epoch": 11.990654205607477, "grad_norm": 3.5, "learning_rate": 4.968713897166903e-05, "loss": 0.3741, "step": 1930 }, { "epoch": 11.996884735202492, "grad_norm": 3.59375, "learning_rate": 4.968675029317738e-05, "loss": 0.3112, "step": 1931 }, { "epoch": 12.0, "grad_norm": 3.03125, "learning_rate": 4.968636137492305e-05, "loss": 0.1329, "step": 1932 }, { "epoch": 12.006230529595015, "grad_norm": 2.546875, "learning_rate": 4.968597221690986e-05, "loss": 0.2455, "step": 1933 }, { "epoch": 12.01246105919003, "grad_norm": 3.140625, "learning_rate": 4.968558281914158e-05, "loss": 0.317, "step": 1934 }, { "epoch": 12.018691588785046, "grad_norm": 1.96875, "learning_rate": 4.968519318162196e-05, "loss": 0.203, "step": 1935 }, { "epoch": 12.024922118380061, "grad_norm": 3.90625, "learning_rate": 4.968480330435483e-05, "loss": 0.427, "step": 1936 }, { "epoch": 12.031152647975079, "grad_norm": 4.46875, "learning_rate": 4.9684413187343946e-05, "loss": 0.3895, "step": 1937 }, { "epoch": 12.037383177570094, "grad_norm": 4.34375, "learning_rate": 4.9684022830593116e-05, "loss": 0.2898, "step": 1938 }, { "epoch": 12.04361370716511, "grad_norm": 1.4921875, "learning_rate": 4.968363223410612e-05, "loss": 0.1643, "step": 1939 }, { "epoch": 12.049844236760125, "grad_norm": 4.21875, "learning_rate": 4.9683241397886745e-05, "loss": 0.2907, "step": 1940 }, { "epoch": 12.05607476635514, "grad_norm": 2.953125, "learning_rate": 4.96828503219388e-05, "loss": 0.3146, "step": 1941 }, { "epoch": 12.062305295950155, "grad_norm": 2.65625, "learning_rate": 4.968245900626608e-05, "loss": 0.2888, "step": 1942 }, { "epoch": 12.06853582554517, "grad_norm": 2.640625, "learning_rate": 4.9682067450872385e-05, "loss": 0.2736, "step": 1943 }, { "epoch": 12.074766355140186, "grad_norm": 1.953125, "learning_rate": 4.968167565576151e-05, "loss": 0.1886, "step": 1944 }, { "epoch": 12.080996884735203, "grad_norm": 4.03125, "learning_rate": 4.968128362093728e-05, "loss": 0.2558, "step": 1945 }, { "epoch": 12.087227414330219, "grad_norm": 2.703125, "learning_rate": 4.968089134640348e-05, "loss": 0.2106, "step": 1946 }, { "epoch": 12.093457943925234, "grad_norm": 3.796875, "learning_rate": 4.968049883216393e-05, "loss": 0.4318, "step": 1947 }, { "epoch": 12.09968847352025, "grad_norm": 2.640625, "learning_rate": 4.9680106078222444e-05, "loss": 0.3833, "step": 1948 }, { "epoch": 12.105919003115265, "grad_norm": 4.3125, "learning_rate": 4.967971308458283e-05, "loss": 0.3959, "step": 1949 }, { "epoch": 12.11214953271028, "grad_norm": 3.71875, "learning_rate": 4.967931985124892e-05, "loss": 0.2953, "step": 1950 }, { "epoch": 12.118380062305295, "grad_norm": 3.28125, "learning_rate": 4.967892637822451e-05, "loss": 0.2887, "step": 1951 }, { "epoch": 12.12461059190031, "grad_norm": 3.5625, "learning_rate": 4.9678532665513444e-05, "loss": 0.4137, "step": 1952 }, { "epoch": 12.130841121495328, "grad_norm": 2.875, "learning_rate": 4.9678138713119524e-05, "loss": 0.2699, "step": 1953 }, { "epoch": 12.137071651090343, "grad_norm": 2.984375, "learning_rate": 4.96777445210466e-05, "loss": 0.3674, "step": 1954 }, { "epoch": 12.143302180685358, "grad_norm": 4.0, "learning_rate": 4.967735008929848e-05, "loss": 0.4276, "step": 1955 }, { "epoch": 12.149532710280374, "grad_norm": 2.53125, "learning_rate": 4.967695541787901e-05, "loss": 0.2574, "step": 1956 }, { "epoch": 12.15576323987539, "grad_norm": 4.21875, "learning_rate": 4.967656050679201e-05, "loss": 0.4382, "step": 1957 }, { "epoch": 12.161993769470405, "grad_norm": 4.03125, "learning_rate": 4.967616535604133e-05, "loss": 0.3072, "step": 1958 }, { "epoch": 12.16822429906542, "grad_norm": 3.1875, "learning_rate": 4.967576996563079e-05, "loss": 0.2973, "step": 1959 }, { "epoch": 12.174454828660435, "grad_norm": 2.59375, "learning_rate": 4.967537433556424e-05, "loss": 0.3327, "step": 1960 }, { "epoch": 12.180685358255452, "grad_norm": 3.03125, "learning_rate": 4.967497846584552e-05, "loss": 0.2096, "step": 1961 }, { "epoch": 12.186915887850468, "grad_norm": 4.75, "learning_rate": 4.967458235647849e-05, "loss": 0.2097, "step": 1962 }, { "epoch": 12.193146417445483, "grad_norm": 5.375, "learning_rate": 4.967418600746697e-05, "loss": 0.3963, "step": 1963 }, { "epoch": 12.199376947040498, "grad_norm": 3.875, "learning_rate": 4.967378941881483e-05, "loss": 0.4216, "step": 1964 }, { "epoch": 12.205607476635514, "grad_norm": 3.6875, "learning_rate": 4.9673392590525915e-05, "loss": 0.2706, "step": 1965 }, { "epoch": 12.21183800623053, "grad_norm": 4.25, "learning_rate": 4.9672995522604085e-05, "loss": 0.4475, "step": 1966 }, { "epoch": 12.218068535825545, "grad_norm": 2.828125, "learning_rate": 4.967259821505318e-05, "loss": 0.2232, "step": 1967 }, { "epoch": 12.22429906542056, "grad_norm": 4.6875, "learning_rate": 4.9672200667877065e-05, "loss": 0.2916, "step": 1968 }, { "epoch": 12.230529595015577, "grad_norm": 3.828125, "learning_rate": 4.967180288107961e-05, "loss": 0.3716, "step": 1969 }, { "epoch": 12.236760124610592, "grad_norm": 4.4375, "learning_rate": 4.9671404854664685e-05, "loss": 0.3205, "step": 1970 }, { "epoch": 12.242990654205608, "grad_norm": 4.375, "learning_rate": 4.967100658863613e-05, "loss": 0.2405, "step": 1971 }, { "epoch": 12.249221183800623, "grad_norm": 4.46875, "learning_rate": 4.967060808299784e-05, "loss": 0.4361, "step": 1972 }, { "epoch": 12.255451713395638, "grad_norm": 2.53125, "learning_rate": 4.967020933775366e-05, "loss": 0.218, "step": 1973 }, { "epoch": 12.261682242990654, "grad_norm": 4.53125, "learning_rate": 4.966981035290748e-05, "loss": 0.5257, "step": 1974 }, { "epoch": 12.26791277258567, "grad_norm": 3.203125, "learning_rate": 4.966941112846317e-05, "loss": 0.3195, "step": 1975 }, { "epoch": 12.274143302180685, "grad_norm": 3.0625, "learning_rate": 4.9669011664424616e-05, "loss": 0.2527, "step": 1976 }, { "epoch": 12.280373831775702, "grad_norm": 3.640625, "learning_rate": 4.9668611960795683e-05, "loss": 0.436, "step": 1977 }, { "epoch": 12.286604361370717, "grad_norm": 3.34375, "learning_rate": 4.966821201758025e-05, "loss": 0.2907, "step": 1978 }, { "epoch": 12.292834890965732, "grad_norm": 2.640625, "learning_rate": 4.9667811834782224e-05, "loss": 0.1936, "step": 1979 }, { "epoch": 12.299065420560748, "grad_norm": 2.890625, "learning_rate": 4.9667411412405477e-05, "loss": 0.282, "step": 1980 }, { "epoch": 12.305295950155763, "grad_norm": 2.484375, "learning_rate": 4.9667010750453896e-05, "loss": 0.274, "step": 1981 }, { "epoch": 12.311526479750778, "grad_norm": 3.390625, "learning_rate": 4.966660984893138e-05, "loss": 0.5162, "step": 1982 }, { "epoch": 12.317757009345794, "grad_norm": 3.578125, "learning_rate": 4.966620870784181e-05, "loss": 0.4643, "step": 1983 }, { "epoch": 12.32398753894081, "grad_norm": 2.78125, "learning_rate": 4.9665807327189096e-05, "loss": 0.2721, "step": 1984 }, { "epoch": 12.330218068535826, "grad_norm": 2.21875, "learning_rate": 4.966540570697713e-05, "loss": 0.1726, "step": 1985 }, { "epoch": 12.336448598130842, "grad_norm": 2.78125, "learning_rate": 4.966500384720981e-05, "loss": 0.3056, "step": 1986 }, { "epoch": 12.342679127725857, "grad_norm": 2.578125, "learning_rate": 4.966460174789105e-05, "loss": 0.2082, "step": 1987 }, { "epoch": 12.348909657320872, "grad_norm": 4.4375, "learning_rate": 4.9664199409024745e-05, "loss": 0.3279, "step": 1988 }, { "epoch": 12.355140186915888, "grad_norm": 3.640625, "learning_rate": 4.96637968306148e-05, "loss": 0.2231, "step": 1989 }, { "epoch": 12.361370716510903, "grad_norm": 4.21875, "learning_rate": 4.9663394012665135e-05, "loss": 0.3131, "step": 1990 }, { "epoch": 12.367601246105918, "grad_norm": 3.9375, "learning_rate": 4.9662990955179656e-05, "loss": 0.3297, "step": 1991 }, { "epoch": 12.373831775700934, "grad_norm": 3.53125, "learning_rate": 4.966258765816227e-05, "loss": 0.3684, "step": 1992 }, { "epoch": 12.38006230529595, "grad_norm": 4.5, "learning_rate": 4.966218412161692e-05, "loss": 0.3489, "step": 1993 }, { "epoch": 12.386292834890966, "grad_norm": 2.140625, "learning_rate": 4.9661780345547495e-05, "loss": 0.19, "step": 1994 }, { "epoch": 12.392523364485982, "grad_norm": 4.34375, "learning_rate": 4.966137632995793e-05, "loss": 0.3512, "step": 1995 }, { "epoch": 12.398753894080997, "grad_norm": 4.03125, "learning_rate": 4.9660972074852154e-05, "loss": 0.2497, "step": 1996 }, { "epoch": 12.404984423676012, "grad_norm": 1.5703125, "learning_rate": 4.966056758023408e-05, "loss": 0.2018, "step": 1997 }, { "epoch": 12.411214953271028, "grad_norm": 3.09375, "learning_rate": 4.9660162846107654e-05, "loss": 0.4691, "step": 1998 }, { "epoch": 12.417445482866043, "grad_norm": 3.703125, "learning_rate": 4.96597578724768e-05, "loss": 0.3415, "step": 1999 }, { "epoch": 12.423676012461058, "grad_norm": 2.640625, "learning_rate": 4.965935265934544e-05, "loss": 0.2843, "step": 2000 }, { "epoch": 12.429906542056075, "grad_norm": 3.5, "learning_rate": 4.965894720671751e-05, "loss": 0.4128, "step": 2001 }, { "epoch": 12.43613707165109, "grad_norm": 2.65625, "learning_rate": 4.965854151459697e-05, "loss": 0.2927, "step": 2002 }, { "epoch": 12.442367601246106, "grad_norm": 3.15625, "learning_rate": 4.965813558298773e-05, "loss": 0.4498, "step": 2003 }, { "epoch": 12.448598130841122, "grad_norm": 3.3125, "learning_rate": 4.965772941189376e-05, "loss": 0.3049, "step": 2004 }, { "epoch": 12.454828660436137, "grad_norm": 2.96875, "learning_rate": 4.965732300131899e-05, "loss": 0.2197, "step": 2005 }, { "epoch": 12.461059190031152, "grad_norm": 4.03125, "learning_rate": 4.9656916351267375e-05, "loss": 0.3732, "step": 2006 }, { "epoch": 12.467289719626168, "grad_norm": 2.5625, "learning_rate": 4.965650946174285e-05, "loss": 0.2442, "step": 2007 }, { "epoch": 12.473520249221183, "grad_norm": 1.71875, "learning_rate": 4.965610233274939e-05, "loss": 0.19, "step": 2008 }, { "epoch": 12.4797507788162, "grad_norm": 2.6875, "learning_rate": 4.965569496429092e-05, "loss": 0.2317, "step": 2009 }, { "epoch": 12.485981308411215, "grad_norm": 3.328125, "learning_rate": 4.9655287356371416e-05, "loss": 0.4297, "step": 2010 }, { "epoch": 12.49221183800623, "grad_norm": 3.65625, "learning_rate": 4.965487950899484e-05, "loss": 0.2787, "step": 2011 }, { "epoch": 12.498442367601246, "grad_norm": 3.03125, "learning_rate": 4.9654471422165135e-05, "loss": 0.2729, "step": 2012 }, { "epoch": 12.504672897196262, "grad_norm": 3.265625, "learning_rate": 4.9654063095886286e-05, "loss": 0.2213, "step": 2013 }, { "epoch": 12.510903426791277, "grad_norm": 3.578125, "learning_rate": 4.965365453016224e-05, "loss": 0.4686, "step": 2014 }, { "epoch": 12.517133956386292, "grad_norm": 3.640625, "learning_rate": 4.965324572499698e-05, "loss": 0.3015, "step": 2015 }, { "epoch": 12.523364485981308, "grad_norm": 2.90625, "learning_rate": 4.965283668039447e-05, "loss": 0.2362, "step": 2016 }, { "epoch": 12.529595015576325, "grad_norm": 6.4375, "learning_rate": 4.965242739635867e-05, "loss": 0.3562, "step": 2017 }, { "epoch": 12.53582554517134, "grad_norm": 6.46875, "learning_rate": 4.965201787289358e-05, "loss": 0.248, "step": 2018 }, { "epoch": 12.542056074766355, "grad_norm": 4.59375, "learning_rate": 4.965160811000316e-05, "loss": 0.3131, "step": 2019 }, { "epoch": 12.54828660436137, "grad_norm": 3.828125, "learning_rate": 4.96511981076914e-05, "loss": 0.4679, "step": 2020 }, { "epoch": 12.554517133956386, "grad_norm": 5.1875, "learning_rate": 4.965078786596227e-05, "loss": 0.3069, "step": 2021 }, { "epoch": 12.560747663551401, "grad_norm": 4.5625, "learning_rate": 4.965037738481976e-05, "loss": 0.1957, "step": 2022 }, { "epoch": 12.566978193146417, "grad_norm": 2.921875, "learning_rate": 4.9649966664267856e-05, "loss": 0.1921, "step": 2023 }, { "epoch": 12.573208722741432, "grad_norm": 4.03125, "learning_rate": 4.964955570431055e-05, "loss": 0.2409, "step": 2024 }, { "epoch": 12.57943925233645, "grad_norm": 3.65625, "learning_rate": 4.964914450495183e-05, "loss": 0.2861, "step": 2025 }, { "epoch": 12.585669781931465, "grad_norm": 6.46875, "learning_rate": 4.9648733066195694e-05, "loss": 0.2647, "step": 2026 }, { "epoch": 12.59190031152648, "grad_norm": 4.71875, "learning_rate": 4.9648321388046137e-05, "loss": 0.2406, "step": 2027 }, { "epoch": 12.598130841121495, "grad_norm": 4.3125, "learning_rate": 4.964790947050715e-05, "loss": 0.4196, "step": 2028 }, { "epoch": 12.60436137071651, "grad_norm": 2.96875, "learning_rate": 4.9647497313582744e-05, "loss": 0.314, "step": 2029 }, { "epoch": 12.610591900311526, "grad_norm": 3.6875, "learning_rate": 4.964708491727692e-05, "loss": 0.2913, "step": 2030 }, { "epoch": 12.616822429906541, "grad_norm": 6.3125, "learning_rate": 4.9646672281593674e-05, "loss": 0.3647, "step": 2031 }, { "epoch": 12.623052959501557, "grad_norm": 4.9375, "learning_rate": 4.964625940653702e-05, "loss": 0.379, "step": 2032 }, { "epoch": 12.629283489096574, "grad_norm": 3.5625, "learning_rate": 4.964584629211098e-05, "loss": 0.2402, "step": 2033 }, { "epoch": 12.63551401869159, "grad_norm": 2.859375, "learning_rate": 4.9645432938319536e-05, "loss": 0.2692, "step": 2034 }, { "epoch": 12.641744548286605, "grad_norm": 5.6875, "learning_rate": 4.9645019345166734e-05, "loss": 0.3807, "step": 2035 }, { "epoch": 12.64797507788162, "grad_norm": 4.03125, "learning_rate": 4.964460551265657e-05, "loss": 0.3105, "step": 2036 }, { "epoch": 12.654205607476635, "grad_norm": 4.125, "learning_rate": 4.9644191440793076e-05, "loss": 0.2636, "step": 2037 }, { "epoch": 12.66043613707165, "grad_norm": 2.578125, "learning_rate": 4.9643777129580263e-05, "loss": 0.2434, "step": 2038 }, { "epoch": 12.666666666666666, "grad_norm": 2.765625, "learning_rate": 4.964336257902217e-05, "loss": 0.3072, "step": 2039 }, { "epoch": 12.672897196261681, "grad_norm": 3.125, "learning_rate": 4.964294778912281e-05, "loss": 0.2468, "step": 2040 }, { "epoch": 12.679127725856699, "grad_norm": 2.9375, "learning_rate": 4.9642532759886205e-05, "loss": 0.2867, "step": 2041 }, { "epoch": 12.685358255451714, "grad_norm": 4.375, "learning_rate": 4.96421174913164e-05, "loss": 0.4539, "step": 2042 }, { "epoch": 12.69158878504673, "grad_norm": 2.71875, "learning_rate": 4.9641701983417434e-05, "loss": 0.2407, "step": 2043 }, { "epoch": 12.697819314641745, "grad_norm": 2.90625, "learning_rate": 4.964128623619332e-05, "loss": 0.3282, "step": 2044 }, { "epoch": 12.70404984423676, "grad_norm": 5.375, "learning_rate": 4.964087024964812e-05, "loss": 0.3078, "step": 2045 }, { "epoch": 12.710280373831775, "grad_norm": 4.375, "learning_rate": 4.964045402378585e-05, "loss": 0.2152, "step": 2046 }, { "epoch": 12.71651090342679, "grad_norm": 4.5625, "learning_rate": 4.964003755861057e-05, "loss": 0.3074, "step": 2047 }, { "epoch": 12.722741433021806, "grad_norm": 3.59375, "learning_rate": 4.9639620854126326e-05, "loss": 0.3332, "step": 2048 }, { "epoch": 12.728971962616823, "grad_norm": 3.953125, "learning_rate": 4.963920391033715e-05, "loss": 0.3088, "step": 2049 }, { "epoch": 12.735202492211839, "grad_norm": 4.34375, "learning_rate": 4.963878672724711e-05, "loss": 0.3381, "step": 2050 }, { "epoch": 12.741433021806854, "grad_norm": 4.21875, "learning_rate": 4.963836930486025e-05, "loss": 0.3092, "step": 2051 }, { "epoch": 12.74766355140187, "grad_norm": 2.90625, "learning_rate": 4.9637951643180605e-05, "loss": 0.2462, "step": 2052 }, { "epoch": 12.753894080996885, "grad_norm": 6.46875, "learning_rate": 4.963753374221226e-05, "loss": 0.2742, "step": 2053 }, { "epoch": 12.7601246105919, "grad_norm": 5.28125, "learning_rate": 4.963711560195926e-05, "loss": 0.3235, "step": 2054 }, { "epoch": 12.766355140186915, "grad_norm": 5.6875, "learning_rate": 4.963669722242567e-05, "loss": 0.2465, "step": 2055 }, { "epoch": 12.77258566978193, "grad_norm": 4.9375, "learning_rate": 4.963627860361555e-05, "loss": 0.33, "step": 2056 }, { "epoch": 12.778816199376948, "grad_norm": 2.125, "learning_rate": 4.9635859745532974e-05, "loss": 0.2064, "step": 2057 }, { "epoch": 12.785046728971963, "grad_norm": 2.703125, "learning_rate": 4.9635440648181994e-05, "loss": 0.2536, "step": 2058 }, { "epoch": 12.791277258566979, "grad_norm": 4.96875, "learning_rate": 4.9635021311566696e-05, "loss": 0.3027, "step": 2059 }, { "epoch": 12.797507788161994, "grad_norm": 3.78125, "learning_rate": 4.963460173569114e-05, "loss": 0.4137, "step": 2060 }, { "epoch": 12.80373831775701, "grad_norm": 2.421875, "learning_rate": 4.963418192055942e-05, "loss": 0.2092, "step": 2061 }, { "epoch": 12.809968847352025, "grad_norm": 4.4375, "learning_rate": 4.9633761866175586e-05, "loss": 0.3685, "step": 2062 }, { "epoch": 12.81619937694704, "grad_norm": 5.78125, "learning_rate": 4.9633341572543745e-05, "loss": 0.4078, "step": 2063 }, { "epoch": 12.822429906542055, "grad_norm": 6.09375, "learning_rate": 4.963292103966796e-05, "loss": 0.3216, "step": 2064 }, { "epoch": 12.828660436137072, "grad_norm": 3.90625, "learning_rate": 4.963250026755232e-05, "loss": 0.2011, "step": 2065 }, { "epoch": 12.834890965732088, "grad_norm": 3.25, "learning_rate": 4.963207925620092e-05, "loss": 0.3246, "step": 2066 }, { "epoch": 12.841121495327103, "grad_norm": 5.03125, "learning_rate": 4.963165800561784e-05, "loss": 0.2769, "step": 2067 }, { "epoch": 12.847352024922118, "grad_norm": 5.0, "learning_rate": 4.963123651580717e-05, "loss": 0.4097, "step": 2068 }, { "epoch": 12.853582554517134, "grad_norm": 3.25, "learning_rate": 4.963081478677301e-05, "loss": 0.253, "step": 2069 }, { "epoch": 12.85981308411215, "grad_norm": 3.265625, "learning_rate": 4.9630392818519457e-05, "loss": 0.4105, "step": 2070 }, { "epoch": 12.866043613707165, "grad_norm": 3.8125, "learning_rate": 4.96299706110506e-05, "loss": 0.3166, "step": 2071 }, { "epoch": 12.87227414330218, "grad_norm": 3.78125, "learning_rate": 4.9629548164370546e-05, "loss": 0.3095, "step": 2072 }, { "epoch": 12.878504672897197, "grad_norm": 4.625, "learning_rate": 4.962912547848339e-05, "loss": 0.3666, "step": 2073 }, { "epoch": 12.884735202492212, "grad_norm": 4.75, "learning_rate": 4.9628702553393255e-05, "loss": 0.4749, "step": 2074 }, { "epoch": 12.890965732087228, "grad_norm": 3.109375, "learning_rate": 4.962827938910424e-05, "loss": 0.2609, "step": 2075 }, { "epoch": 12.897196261682243, "grad_norm": 3.609375, "learning_rate": 4.962785598562044e-05, "loss": 0.241, "step": 2076 }, { "epoch": 12.903426791277258, "grad_norm": 3.265625, "learning_rate": 4.962743234294599e-05, "loss": 0.2721, "step": 2077 }, { "epoch": 12.909657320872274, "grad_norm": 3.296875, "learning_rate": 4.9627008461084987e-05, "loss": 0.3123, "step": 2078 }, { "epoch": 12.91588785046729, "grad_norm": 4.125, "learning_rate": 4.962658434004156e-05, "loss": 0.4008, "step": 2079 }, { "epoch": 12.922118380062305, "grad_norm": 5.03125, "learning_rate": 4.9626159979819826e-05, "loss": 0.3552, "step": 2080 }, { "epoch": 12.928348909657322, "grad_norm": 4.46875, "learning_rate": 4.96257353804239e-05, "loss": 0.4172, "step": 2081 }, { "epoch": 12.934579439252337, "grad_norm": 2.78125, "learning_rate": 4.962531054185791e-05, "loss": 0.3121, "step": 2082 }, { "epoch": 12.940809968847352, "grad_norm": 2.703125, "learning_rate": 4.962488546412598e-05, "loss": 0.2668, "step": 2083 }, { "epoch": 12.947040498442368, "grad_norm": 4.75, "learning_rate": 4.962446014723224e-05, "loss": 0.3662, "step": 2084 }, { "epoch": 12.953271028037383, "grad_norm": 2.859375, "learning_rate": 4.9624034591180815e-05, "loss": 0.2082, "step": 2085 }, { "epoch": 12.959501557632398, "grad_norm": 3.78125, "learning_rate": 4.962360879597585e-05, "loss": 0.3149, "step": 2086 }, { "epoch": 12.965732087227414, "grad_norm": 3.390625, "learning_rate": 4.962318276162148e-05, "loss": 0.3184, "step": 2087 }, { "epoch": 12.97196261682243, "grad_norm": 3.375, "learning_rate": 4.9622756488121826e-05, "loss": 0.2869, "step": 2088 }, { "epoch": 12.978193146417446, "grad_norm": 2.453125, "learning_rate": 4.962232997548105e-05, "loss": 0.1648, "step": 2089 }, { "epoch": 12.984423676012462, "grad_norm": 4.53125, "learning_rate": 4.962190322370327e-05, "loss": 0.3891, "step": 2090 }, { "epoch": 12.990654205607477, "grad_norm": 2.375, "learning_rate": 4.9621476232792654e-05, "loss": 0.2075, "step": 2091 }, { "epoch": 12.996884735202492, "grad_norm": 2.578125, "learning_rate": 4.962104900275334e-05, "loss": 0.2254, "step": 2092 }, { "epoch": 13.0, "grad_norm": 2.28125, "learning_rate": 4.962062153358947e-05, "loss": 0.195, "step": 2093 }, { "epoch": 13.006230529595015, "grad_norm": 5.6875, "learning_rate": 4.962019382530521e-05, "loss": 0.301, "step": 2094 }, { "epoch": 13.01246105919003, "grad_norm": 3.078125, "learning_rate": 4.9619765877904696e-05, "loss": 0.248, "step": 2095 }, { "epoch": 13.018691588785046, "grad_norm": 3.4375, "learning_rate": 4.9619337691392096e-05, "loss": 0.2698, "step": 2096 }, { "epoch": 13.024922118380061, "grad_norm": 2.328125, "learning_rate": 4.9618909265771574e-05, "loss": 0.2458, "step": 2097 }, { "epoch": 13.031152647975079, "grad_norm": 5.65625, "learning_rate": 4.961848060104728e-05, "loss": 0.3919, "step": 2098 }, { "epoch": 13.037383177570094, "grad_norm": 3.296875, "learning_rate": 4.961805169722338e-05, "loss": 0.4126, "step": 2099 }, { "epoch": 13.04361370716511, "grad_norm": 2.625, "learning_rate": 4.9617622554304044e-05, "loss": 0.2848, "step": 2100 }, { "epoch": 13.049844236760125, "grad_norm": 3.84375, "learning_rate": 4.961719317229344e-05, "loss": 0.2336, "step": 2101 }, { "epoch": 13.05607476635514, "grad_norm": 2.234375, "learning_rate": 4.961676355119572e-05, "loss": 0.2096, "step": 2102 }, { "epoch": 13.062305295950155, "grad_norm": 2.4375, "learning_rate": 4.961633369101509e-05, "loss": 0.2514, "step": 2103 }, { "epoch": 13.06853582554517, "grad_norm": 3.15625, "learning_rate": 4.96159035917557e-05, "loss": 0.27, "step": 2104 }, { "epoch": 13.074766355140186, "grad_norm": 2.390625, "learning_rate": 4.9615473253421727e-05, "loss": 0.3274, "step": 2105 }, { "epoch": 13.080996884735203, "grad_norm": 2.890625, "learning_rate": 4.961504267601737e-05, "loss": 0.3066, "step": 2106 }, { "epoch": 13.087227414330219, "grad_norm": 1.984375, "learning_rate": 4.961461185954678e-05, "loss": 0.1936, "step": 2107 }, { "epoch": 13.093457943925234, "grad_norm": 2.28125, "learning_rate": 4.961418080401418e-05, "loss": 0.2279, "step": 2108 }, { "epoch": 13.09968847352025, "grad_norm": 3.3125, "learning_rate": 4.961374950942372e-05, "loss": 0.346, "step": 2109 }, { "epoch": 13.105919003115265, "grad_norm": 3.296875, "learning_rate": 4.9613317975779605e-05, "loss": 0.3768, "step": 2110 }, { "epoch": 13.11214953271028, "grad_norm": 3.171875, "learning_rate": 4.961288620308603e-05, "loss": 0.3256, "step": 2111 }, { "epoch": 13.118380062305295, "grad_norm": 2.671875, "learning_rate": 4.961245419134719e-05, "loss": 0.2556, "step": 2112 }, { "epoch": 13.12461059190031, "grad_norm": 3.640625, "learning_rate": 4.961202194056727e-05, "loss": 0.2642, "step": 2113 }, { "epoch": 13.130841121495328, "grad_norm": 4.625, "learning_rate": 4.9611589450750475e-05, "loss": 0.5115, "step": 2114 }, { "epoch": 13.137071651090343, "grad_norm": 3.5625, "learning_rate": 4.9611156721901e-05, "loss": 0.2911, "step": 2115 }, { "epoch": 13.143302180685358, "grad_norm": 3.546875, "learning_rate": 4.961072375402305e-05, "loss": 0.4232, "step": 2116 }, { "epoch": 13.149532710280374, "grad_norm": 2.3125, "learning_rate": 4.9610290547120835e-05, "loss": 0.2185, "step": 2117 }, { "epoch": 13.15576323987539, "grad_norm": 3.65625, "learning_rate": 4.960985710119855e-05, "loss": 0.4968, "step": 2118 }, { "epoch": 13.161993769470405, "grad_norm": 3.75, "learning_rate": 4.9609423416260426e-05, "loss": 0.217, "step": 2119 }, { "epoch": 13.16822429906542, "grad_norm": 2.046875, "learning_rate": 4.960898949231066e-05, "loss": 0.2573, "step": 2120 }, { "epoch": 13.174454828660435, "grad_norm": 3.296875, "learning_rate": 4.960855532935347e-05, "loss": 0.4622, "step": 2121 }, { "epoch": 13.180685358255452, "grad_norm": 3.6875, "learning_rate": 4.960812092739306e-05, "loss": 0.5526, "step": 2122 }, { "epoch": 13.186915887850468, "grad_norm": 2.78125, "learning_rate": 4.9607686286433676e-05, "loss": 0.2338, "step": 2123 }, { "epoch": 13.193146417445483, "grad_norm": 2.390625, "learning_rate": 4.960725140647952e-05, "loss": 0.293, "step": 2124 }, { "epoch": 13.199376947040498, "grad_norm": 6.4375, "learning_rate": 4.9606816287534805e-05, "loss": 0.5162, "step": 2125 }, { "epoch": 13.205607476635514, "grad_norm": 3.703125, "learning_rate": 4.9606380929603784e-05, "loss": 0.3307, "step": 2126 }, { "epoch": 13.21183800623053, "grad_norm": 2.5, "learning_rate": 4.960594533269067e-05, "loss": 0.2355, "step": 2127 }, { "epoch": 13.218068535825545, "grad_norm": 2.53125, "learning_rate": 4.960550949679969e-05, "loss": 0.2363, "step": 2128 }, { "epoch": 13.22429906542056, "grad_norm": 3.921875, "learning_rate": 4.96050734219351e-05, "loss": 0.4406, "step": 2129 }, { "epoch": 13.230529595015577, "grad_norm": 4.21875, "learning_rate": 4.960463710810111e-05, "loss": 0.444, "step": 2130 }, { "epoch": 13.236760124610592, "grad_norm": 3.359375, "learning_rate": 4.960420055530196e-05, "loss": 0.2749, "step": 2131 }, { "epoch": 13.242990654205608, "grad_norm": 3.40625, "learning_rate": 4.96037637635419e-05, "loss": 0.3246, "step": 2132 }, { "epoch": 13.249221183800623, "grad_norm": 2.328125, "learning_rate": 4.9603326732825175e-05, "loss": 0.1732, "step": 2133 }, { "epoch": 13.255451713395638, "grad_norm": 4.4375, "learning_rate": 4.960288946315601e-05, "loss": 0.5064, "step": 2134 }, { "epoch": 13.261682242990654, "grad_norm": 4.375, "learning_rate": 4.960245195453868e-05, "loss": 0.501, "step": 2135 }, { "epoch": 13.26791277258567, "grad_norm": 3.03125, "learning_rate": 4.96020142069774e-05, "loss": 0.2776, "step": 2136 }, { "epoch": 13.274143302180685, "grad_norm": 3.1875, "learning_rate": 4.960157622047645e-05, "loss": 0.3631, "step": 2137 }, { "epoch": 13.280373831775702, "grad_norm": 3.09375, "learning_rate": 4.960113799504008e-05, "loss": 0.3345, "step": 2138 }, { "epoch": 13.286604361370717, "grad_norm": 3.0, "learning_rate": 4.9600699530672536e-05, "loss": 0.3393, "step": 2139 }, { "epoch": 13.292834890965732, "grad_norm": 3.0, "learning_rate": 4.9600260827378074e-05, "loss": 0.511, "step": 2140 }, { "epoch": 13.299065420560748, "grad_norm": 1.8984375, "learning_rate": 4.9599821885160966e-05, "loss": 0.2118, "step": 2141 }, { "epoch": 13.305295950155763, "grad_norm": 2.265625, "learning_rate": 4.959938270402547e-05, "loss": 0.2322, "step": 2142 }, { "epoch": 13.311526479750778, "grad_norm": 2.953125, "learning_rate": 4.9598943283975854e-05, "loss": 0.2519, "step": 2143 }, { "epoch": 13.317757009345794, "grad_norm": 2.65625, "learning_rate": 4.959850362501638e-05, "loss": 0.2695, "step": 2144 }, { "epoch": 13.32398753894081, "grad_norm": 3.921875, "learning_rate": 4.9598063727151325e-05, "loss": 0.2793, "step": 2145 }, { "epoch": 13.330218068535826, "grad_norm": 2.5625, "learning_rate": 4.9597623590384956e-05, "loss": 0.1844, "step": 2146 }, { "epoch": 13.336448598130842, "grad_norm": 4.625, "learning_rate": 4.959718321472155e-05, "loss": 0.4477, "step": 2147 }, { "epoch": 13.342679127725857, "grad_norm": 2.90625, "learning_rate": 4.959674260016538e-05, "loss": 0.3134, "step": 2148 }, { "epoch": 13.348909657320872, "grad_norm": 3.28125, "learning_rate": 4.9596301746720734e-05, "loss": 0.3251, "step": 2149 }, { "epoch": 13.355140186915888, "grad_norm": 3.015625, "learning_rate": 4.959586065439189e-05, "loss": 0.3112, "step": 2150 }, { "epoch": 13.361370716510903, "grad_norm": 3.28125, "learning_rate": 4.9595419323183125e-05, "loss": 0.2154, "step": 2151 }, { "epoch": 13.367601246105918, "grad_norm": 3.46875, "learning_rate": 4.9594977753098734e-05, "loss": 0.3004, "step": 2152 }, { "epoch": 13.373831775700934, "grad_norm": 3.640625, "learning_rate": 4.959453594414301e-05, "loss": 0.256, "step": 2153 }, { "epoch": 13.38006230529595, "grad_norm": 2.59375, "learning_rate": 4.959409389632022e-05, "loss": 0.324, "step": 2154 }, { "epoch": 13.386292834890966, "grad_norm": 3.578125, "learning_rate": 4.9593651609634696e-05, "loss": 0.2222, "step": 2155 }, { "epoch": 13.392523364485982, "grad_norm": 2.53125, "learning_rate": 4.9593209084090696e-05, "loss": 0.2295, "step": 2156 }, { "epoch": 13.398753894080997, "grad_norm": 2.5, "learning_rate": 4.959276631969253e-05, "loss": 0.2218, "step": 2157 }, { "epoch": 13.404984423676012, "grad_norm": 4.625, "learning_rate": 4.959232331644451e-05, "loss": 0.3177, "step": 2158 }, { "epoch": 13.411214953271028, "grad_norm": 2.421875, "learning_rate": 4.9591880074350926e-05, "loss": 0.2643, "step": 2159 }, { "epoch": 13.417445482866043, "grad_norm": 3.5625, "learning_rate": 4.959143659341609e-05, "loss": 0.2342, "step": 2160 }, { "epoch": 13.423676012461058, "grad_norm": 3.6875, "learning_rate": 4.95909928736443e-05, "loss": 0.4089, "step": 2161 }, { "epoch": 13.429906542056075, "grad_norm": 2.953125, "learning_rate": 4.959054891503988e-05, "loss": 0.2234, "step": 2162 }, { "epoch": 13.43613707165109, "grad_norm": 4.0, "learning_rate": 4.9590104717607135e-05, "loss": 0.6074, "step": 2163 }, { "epoch": 13.442367601246106, "grad_norm": 2.921875, "learning_rate": 4.958966028135037e-05, "loss": 0.3639, "step": 2164 }, { "epoch": 13.448598130841122, "grad_norm": 2.734375, "learning_rate": 4.9589215606273905e-05, "loss": 0.2445, "step": 2165 }, { "epoch": 13.454828660436137, "grad_norm": 3.140625, "learning_rate": 4.958877069238208e-05, "loss": 0.2173, "step": 2166 }, { "epoch": 13.461059190031152, "grad_norm": 3.78125, "learning_rate": 4.9588325539679185e-05, "loss": 0.3397, "step": 2167 }, { "epoch": 13.467289719626168, "grad_norm": 3.265625, "learning_rate": 4.9587880148169554e-05, "loss": 0.2293, "step": 2168 }, { "epoch": 13.473520249221183, "grad_norm": 4.9375, "learning_rate": 4.9587434517857525e-05, "loss": 0.2203, "step": 2169 }, { "epoch": 13.4797507788162, "grad_norm": 3.390625, "learning_rate": 4.958698864874741e-05, "loss": 0.3538, "step": 2170 }, { "epoch": 13.485981308411215, "grad_norm": 3.625, "learning_rate": 4.958654254084355e-05, "loss": 0.3047, "step": 2171 }, { "epoch": 13.49221183800623, "grad_norm": 2.953125, "learning_rate": 4.958609619415028e-05, "loss": 0.1773, "step": 2172 }, { "epoch": 13.498442367601246, "grad_norm": 2.8125, "learning_rate": 4.958564960867192e-05, "loss": 0.2451, "step": 2173 }, { "epoch": 13.504672897196262, "grad_norm": 2.953125, "learning_rate": 4.958520278441282e-05, "loss": 0.2097, "step": 2174 }, { "epoch": 13.510903426791277, "grad_norm": 4.25, "learning_rate": 4.958475572137732e-05, "loss": 0.29, "step": 2175 }, { "epoch": 13.517133956386292, "grad_norm": 2.328125, "learning_rate": 4.9584308419569755e-05, "loss": 0.2186, "step": 2176 }, { "epoch": 13.523364485981308, "grad_norm": 2.984375, "learning_rate": 4.958386087899447e-05, "loss": 0.2766, "step": 2177 }, { "epoch": 13.529595015576325, "grad_norm": 2.53125, "learning_rate": 4.958341309965582e-05, "loss": 0.2522, "step": 2178 }, { "epoch": 13.53582554517134, "grad_norm": 1.34375, "learning_rate": 4.958296508155814e-05, "loss": 0.1498, "step": 2179 }, { "epoch": 13.542056074766355, "grad_norm": 1.8671875, "learning_rate": 4.9582516824705795e-05, "loss": 0.1581, "step": 2180 }, { "epoch": 13.54828660436137, "grad_norm": 2.0625, "learning_rate": 4.9582068329103125e-05, "loss": 0.1992, "step": 2181 }, { "epoch": 13.554517133956386, "grad_norm": 3.0, "learning_rate": 4.95816195947545e-05, "loss": 0.3419, "step": 2182 }, { "epoch": 13.560747663551401, "grad_norm": 2.203125, "learning_rate": 4.958117062166427e-05, "loss": 0.2124, "step": 2183 }, { "epoch": 13.566978193146417, "grad_norm": 4.875, "learning_rate": 4.958072140983679e-05, "loss": 0.2445, "step": 2184 }, { "epoch": 13.573208722741432, "grad_norm": 4.21875, "learning_rate": 4.9580271959276445e-05, "loss": 0.2612, "step": 2185 }, { "epoch": 13.57943925233645, "grad_norm": 2.828125, "learning_rate": 4.9579822269987574e-05, "loss": 0.2111, "step": 2186 }, { "epoch": 13.585669781931465, "grad_norm": 3.421875, "learning_rate": 4.957937234197456e-05, "loss": 0.4228, "step": 2187 }, { "epoch": 13.59190031152648, "grad_norm": 2.703125, "learning_rate": 4.957892217524177e-05, "loss": 0.2392, "step": 2188 }, { "epoch": 13.598130841121495, "grad_norm": 1.6953125, "learning_rate": 4.957847176979357e-05, "loss": 0.1894, "step": 2189 }, { "epoch": 13.60436137071651, "grad_norm": 2.25, "learning_rate": 4.957802112563434e-05, "loss": 0.3218, "step": 2190 }, { "epoch": 13.610591900311526, "grad_norm": 3.4375, "learning_rate": 4.957757024276846e-05, "loss": 0.2906, "step": 2191 }, { "epoch": 13.616822429906541, "grad_norm": 4.75, "learning_rate": 4.957711912120031e-05, "loss": 0.3414, "step": 2192 }, { "epoch": 13.623052959501557, "grad_norm": 2.3125, "learning_rate": 4.9576667760934256e-05, "loss": 0.1836, "step": 2193 }, { "epoch": 13.629283489096574, "grad_norm": 4.15625, "learning_rate": 4.957621616197469e-05, "loss": 0.3271, "step": 2194 }, { "epoch": 13.63551401869159, "grad_norm": 2.84375, "learning_rate": 4.9575764324326004e-05, "loss": 0.2161, "step": 2195 }, { "epoch": 13.641744548286605, "grad_norm": 4.78125, "learning_rate": 4.9575312247992584e-05, "loss": 0.4427, "step": 2196 }, { "epoch": 13.64797507788162, "grad_norm": 2.5625, "learning_rate": 4.957485993297882e-05, "loss": 0.2373, "step": 2197 }, { "epoch": 13.654205607476635, "grad_norm": 4.15625, "learning_rate": 4.95744073792891e-05, "loss": 0.3088, "step": 2198 }, { "epoch": 13.66043613707165, "grad_norm": 3.078125, "learning_rate": 4.957395458692782e-05, "loss": 0.215, "step": 2199 }, { "epoch": 13.666666666666666, "grad_norm": 3.4375, "learning_rate": 4.9573501555899394e-05, "loss": 0.1828, "step": 2200 }, { "epoch": 13.672897196261681, "grad_norm": 3.03125, "learning_rate": 4.95730482862082e-05, "loss": 0.2356, "step": 2201 }, { "epoch": 13.679127725856699, "grad_norm": 2.609375, "learning_rate": 4.9572594777858644e-05, "loss": 0.4203, "step": 2202 }, { "epoch": 13.685358255451714, "grad_norm": 4.78125, "learning_rate": 4.957214103085514e-05, "loss": 0.2602, "step": 2203 }, { "epoch": 13.69158878504673, "grad_norm": 4.40625, "learning_rate": 4.957168704520209e-05, "loss": 0.3767, "step": 2204 }, { "epoch": 13.697819314641745, "grad_norm": 3.265625, "learning_rate": 4.9571232820903903e-05, "loss": 0.3451, "step": 2205 }, { "epoch": 13.70404984423676, "grad_norm": 2.25, "learning_rate": 4.9570778357965e-05, "loss": 0.2051, "step": 2206 }, { "epoch": 13.710280373831775, "grad_norm": 3.0, "learning_rate": 4.9570323656389776e-05, "loss": 0.3567, "step": 2207 }, { "epoch": 13.71651090342679, "grad_norm": 3.375, "learning_rate": 4.956986871618266e-05, "loss": 0.3363, "step": 2208 }, { "epoch": 13.722741433021806, "grad_norm": 2.890625, "learning_rate": 4.956941353734807e-05, "loss": 0.234, "step": 2209 }, { "epoch": 13.728971962616823, "grad_norm": 2.703125, "learning_rate": 4.9568958119890417e-05, "loss": 0.3315, "step": 2210 }, { "epoch": 13.735202492211839, "grad_norm": 2.84375, "learning_rate": 4.956850246381414e-05, "loss": 0.2427, "step": 2211 }, { "epoch": 13.741433021806854, "grad_norm": 3.296875, "learning_rate": 4.956804656912365e-05, "loss": 0.3098, "step": 2212 }, { "epoch": 13.74766355140187, "grad_norm": 2.859375, "learning_rate": 4.9567590435823383e-05, "loss": 0.2023, "step": 2213 }, { "epoch": 13.753894080996885, "grad_norm": 4.03125, "learning_rate": 4.956713406391777e-05, "loss": 0.6036, "step": 2214 }, { "epoch": 13.7601246105919, "grad_norm": 2.703125, "learning_rate": 4.956667745341124e-05, "loss": 0.3741, "step": 2215 }, { "epoch": 13.766355140186915, "grad_norm": 2.671875, "learning_rate": 4.9566220604308224e-05, "loss": 0.2995, "step": 2216 }, { "epoch": 13.77258566978193, "grad_norm": 2.734375, "learning_rate": 4.9565763516613165e-05, "loss": 0.2605, "step": 2217 }, { "epoch": 13.778816199376948, "grad_norm": 3.28125, "learning_rate": 4.956530619033049e-05, "loss": 0.3205, "step": 2218 }, { "epoch": 13.785046728971963, "grad_norm": 2.84375, "learning_rate": 4.956484862546467e-05, "loss": 0.3675, "step": 2219 }, { "epoch": 13.791277258566979, "grad_norm": 4.15625, "learning_rate": 4.9564390822020114e-05, "loss": 0.258, "step": 2220 }, { "epoch": 13.797507788161994, "grad_norm": 3.234375, "learning_rate": 4.956393278000129e-05, "loss": 0.2464, "step": 2221 }, { "epoch": 13.80373831775701, "grad_norm": 3.03125, "learning_rate": 4.956347449941264e-05, "loss": 0.2326, "step": 2222 }, { "epoch": 13.809968847352025, "grad_norm": 3.546875, "learning_rate": 4.956301598025861e-05, "loss": 0.3667, "step": 2223 }, { "epoch": 13.81619937694704, "grad_norm": 2.859375, "learning_rate": 4.956255722254367e-05, "loss": 0.3655, "step": 2224 }, { "epoch": 13.822429906542055, "grad_norm": 5.21875, "learning_rate": 4.956209822627226e-05, "loss": 0.3301, "step": 2225 }, { "epoch": 13.828660436137072, "grad_norm": 3.09375, "learning_rate": 4.956163899144884e-05, "loss": 0.2702, "step": 2226 }, { "epoch": 13.834890965732088, "grad_norm": 2.828125, "learning_rate": 4.9561179518077875e-05, "loss": 0.2701, "step": 2227 }, { "epoch": 13.841121495327103, "grad_norm": 5.15625, "learning_rate": 4.956071980616382e-05, "loss": 0.3026, "step": 2228 }, { "epoch": 13.847352024922118, "grad_norm": 3.9375, "learning_rate": 4.956025985571115e-05, "loss": 0.2596, "step": 2229 }, { "epoch": 13.853582554517134, "grad_norm": 2.5, "learning_rate": 4.955979966672432e-05, "loss": 0.3511, "step": 2230 }, { "epoch": 13.85981308411215, "grad_norm": 3.15625, "learning_rate": 4.9559339239207815e-05, "loss": 0.3386, "step": 2231 }, { "epoch": 13.866043613707165, "grad_norm": 5.3125, "learning_rate": 4.955887857316609e-05, "loss": 0.4464, "step": 2232 }, { "epoch": 13.87227414330218, "grad_norm": 3.71875, "learning_rate": 4.955841766860363e-05, "loss": 0.2741, "step": 2233 }, { "epoch": 13.878504672897197, "grad_norm": 4.125, "learning_rate": 4.9557956525524904e-05, "loss": 0.2682, "step": 2234 }, { "epoch": 13.884735202492212, "grad_norm": 4.09375, "learning_rate": 4.9557495143934405e-05, "loss": 0.3089, "step": 2235 }, { "epoch": 13.890965732087228, "grad_norm": 2.515625, "learning_rate": 4.95570335238366e-05, "loss": 0.2121, "step": 2236 }, { "epoch": 13.897196261682243, "grad_norm": 3.015625, "learning_rate": 4.955657166523597e-05, "loss": 0.2475, "step": 2237 }, { "epoch": 13.903426791277258, "grad_norm": 4.0, "learning_rate": 4.955610956813702e-05, "loss": 0.3998, "step": 2238 }, { "epoch": 13.909657320872274, "grad_norm": 3.1875, "learning_rate": 4.9555647232544214e-05, "loss": 0.2644, "step": 2239 }, { "epoch": 13.91588785046729, "grad_norm": 2.03125, "learning_rate": 4.955518465846206e-05, "loss": 0.2186, "step": 2240 }, { "epoch": 13.922118380062305, "grad_norm": 2.109375, "learning_rate": 4.955472184589505e-05, "loss": 0.2432, "step": 2241 }, { "epoch": 13.928348909657322, "grad_norm": 2.234375, "learning_rate": 4.955425879484766e-05, "loss": 0.2431, "step": 2242 }, { "epoch": 13.934579439252337, "grad_norm": 3.5625, "learning_rate": 4.9553795505324406e-05, "loss": 0.244, "step": 2243 }, { "epoch": 13.940809968847352, "grad_norm": 5.875, "learning_rate": 4.955333197732978e-05, "loss": 0.2789, "step": 2244 }, { "epoch": 13.947040498442368, "grad_norm": 4.21875, "learning_rate": 4.955286821086829e-05, "loss": 0.4741, "step": 2245 }, { "epoch": 13.953271028037383, "grad_norm": 2.96875, "learning_rate": 4.955240420594443e-05, "loss": 0.252, "step": 2246 }, { "epoch": 13.959501557632398, "grad_norm": 3.46875, "learning_rate": 4.9551939962562714e-05, "loss": 0.2861, "step": 2247 }, { "epoch": 13.965732087227414, "grad_norm": 4.71875, "learning_rate": 4.955147548072765e-05, "loss": 0.3961, "step": 2248 }, { "epoch": 13.97196261682243, "grad_norm": 4.09375, "learning_rate": 4.955101076044375e-05, "loss": 0.4026, "step": 2249 }, { "epoch": 13.978193146417446, "grad_norm": 2.375, "learning_rate": 4.955054580171553e-05, "loss": 0.1723, "step": 2250 }, { "epoch": 13.984423676012462, "grad_norm": 2.390625, "learning_rate": 4.955008060454749e-05, "loss": 0.3364, "step": 2251 }, { "epoch": 13.990654205607477, "grad_norm": 5.4375, "learning_rate": 4.954961516894417e-05, "loss": 0.3447, "step": 2252 }, { "epoch": 13.996884735202492, "grad_norm": 5.15625, "learning_rate": 4.954914949491007e-05, "loss": 0.2855, "step": 2253 }, { "epoch": 14.0, "grad_norm": 2.015625, "learning_rate": 4.9548683582449733e-05, "loss": 0.128, "step": 2254 }, { "epoch": 14.006230529595015, "grad_norm": 3.453125, "learning_rate": 4.9548217431567665e-05, "loss": 0.2803, "step": 2255 }, { "epoch": 14.01246105919003, "grad_norm": 3.21875, "learning_rate": 4.954775104226841e-05, "loss": 0.3978, "step": 2256 }, { "epoch": 14.018691588785046, "grad_norm": 4.1875, "learning_rate": 4.9547284414556487e-05, "loss": 0.2609, "step": 2257 }, { "epoch": 14.024922118380061, "grad_norm": 4.6875, "learning_rate": 4.954681754843643e-05, "loss": 0.2188, "step": 2258 }, { "epoch": 14.031152647975079, "grad_norm": 3.890625, "learning_rate": 4.9546350443912783e-05, "loss": 0.5006, "step": 2259 }, { "epoch": 14.037383177570094, "grad_norm": 3.203125, "learning_rate": 4.954588310099007e-05, "loss": 0.546, "step": 2260 }, { "epoch": 14.04361370716511, "grad_norm": 3.359375, "learning_rate": 4.954541551967282e-05, "loss": 0.3129, "step": 2261 }, { "epoch": 14.049844236760125, "grad_norm": 5.0625, "learning_rate": 4.9544947699965594e-05, "loss": 0.2844, "step": 2262 }, { "epoch": 14.05607476635514, "grad_norm": 3.828125, "learning_rate": 4.954447964187294e-05, "loss": 0.4454, "step": 2263 }, { "epoch": 14.062305295950155, "grad_norm": 2.078125, "learning_rate": 4.9544011345399384e-05, "loss": 0.2032, "step": 2264 }, { "epoch": 14.06853582554517, "grad_norm": 3.4375, "learning_rate": 4.954354281054948e-05, "loss": 0.3154, "step": 2265 }, { "epoch": 14.074766355140186, "grad_norm": 3.078125, "learning_rate": 4.95430740373278e-05, "loss": 0.3783, "step": 2266 }, { "epoch": 14.080996884735203, "grad_norm": 2.375, "learning_rate": 4.954260502573886e-05, "loss": 0.2467, "step": 2267 }, { "epoch": 14.087227414330219, "grad_norm": 2.1875, "learning_rate": 4.954213577578724e-05, "loss": 0.2121, "step": 2268 }, { "epoch": 14.093457943925234, "grad_norm": 2.765625, "learning_rate": 4.9541666287477494e-05, "loss": 0.2579, "step": 2269 }, { "epoch": 14.09968847352025, "grad_norm": 3.046875, "learning_rate": 4.9541196560814173e-05, "loss": 0.3842, "step": 2270 }, { "epoch": 14.105919003115265, "grad_norm": 2.375, "learning_rate": 4.954072659580185e-05, "loss": 0.2964, "step": 2271 }, { "epoch": 14.11214953271028, "grad_norm": 2.171875, "learning_rate": 4.9540256392445086e-05, "loss": 0.188, "step": 2272 }, { "epoch": 14.118380062305295, "grad_norm": 2.59375, "learning_rate": 4.953978595074845e-05, "loss": 0.2089, "step": 2273 }, { "epoch": 14.12461059190031, "grad_norm": 2.75, "learning_rate": 4.95393152707165e-05, "loss": 0.3641, "step": 2274 }, { "epoch": 14.130841121495328, "grad_norm": 3.328125, "learning_rate": 4.953884435235382e-05, "loss": 0.4728, "step": 2275 }, { "epoch": 14.137071651090343, "grad_norm": 4.46875, "learning_rate": 4.953837319566497e-05, "loss": 0.3327, "step": 2276 }, { "epoch": 14.143302180685358, "grad_norm": 4.125, "learning_rate": 4.953790180065454e-05, "loss": 0.3896, "step": 2277 }, { "epoch": 14.149532710280374, "grad_norm": 3.515625, "learning_rate": 4.95374301673271e-05, "loss": 0.3846, "step": 2278 }, { "epoch": 14.15576323987539, "grad_norm": 5.5, "learning_rate": 4.9536958295687244e-05, "loss": 0.5431, "step": 2279 }, { "epoch": 14.161993769470405, "grad_norm": 5.90625, "learning_rate": 4.953648618573953e-05, "loss": 0.241, "step": 2280 }, { "epoch": 14.16822429906542, "grad_norm": 4.78125, "learning_rate": 4.953601383748857e-05, "loss": 0.3138, "step": 2281 }, { "epoch": 14.174454828660435, "grad_norm": 2.609375, "learning_rate": 4.953554125093893e-05, "loss": 0.2451, "step": 2282 }, { "epoch": 14.180685358255452, "grad_norm": 1.84375, "learning_rate": 4.953506842609521e-05, "loss": 0.1764, "step": 2283 }, { "epoch": 14.186915887850468, "grad_norm": 2.859375, "learning_rate": 4.9534595362962e-05, "loss": 0.2265, "step": 2284 }, { "epoch": 14.193146417445483, "grad_norm": 3.9375, "learning_rate": 4.95341220615439e-05, "loss": 0.292, "step": 2285 }, { "epoch": 14.199376947040498, "grad_norm": 3.671875, "learning_rate": 4.95336485218455e-05, "loss": 0.3642, "step": 2286 }, { "epoch": 14.205607476635514, "grad_norm": 3.109375, "learning_rate": 4.95331747438714e-05, "loss": 0.2656, "step": 2287 }, { "epoch": 14.21183800623053, "grad_norm": 3.78125, "learning_rate": 4.9532700727626205e-05, "loss": 0.3824, "step": 2288 }, { "epoch": 14.218068535825545, "grad_norm": 2.65625, "learning_rate": 4.953222647311452e-05, "loss": 0.3081, "step": 2289 }, { "epoch": 14.22429906542056, "grad_norm": 2.34375, "learning_rate": 4.953175198034094e-05, "loss": 0.2351, "step": 2290 }, { "epoch": 14.230529595015577, "grad_norm": 2.890625, "learning_rate": 4.953127724931008e-05, "loss": 0.2996, "step": 2291 }, { "epoch": 14.236760124610592, "grad_norm": 4.25, "learning_rate": 4.953080228002656e-05, "loss": 0.4607, "step": 2292 }, { "epoch": 14.242990654205608, "grad_norm": 3.40625, "learning_rate": 4.9530327072494985e-05, "loss": 0.2112, "step": 2293 }, { "epoch": 14.249221183800623, "grad_norm": 3.046875, "learning_rate": 4.952985162671996e-05, "loss": 0.3388, "step": 2294 }, { "epoch": 14.255451713395638, "grad_norm": 2.984375, "learning_rate": 4.9529375942706124e-05, "loss": 0.274, "step": 2295 }, { "epoch": 14.261682242990654, "grad_norm": 3.5, "learning_rate": 4.952890002045808e-05, "loss": 0.2085, "step": 2296 }, { "epoch": 14.26791277258567, "grad_norm": 4.25, "learning_rate": 4.952842385998046e-05, "loss": 0.412, "step": 2297 }, { "epoch": 14.274143302180685, "grad_norm": 2.609375, "learning_rate": 4.952794746127787e-05, "loss": 0.2311, "step": 2298 }, { "epoch": 14.280373831775702, "grad_norm": 3.40625, "learning_rate": 4.952747082435496e-05, "loss": 0.2808, "step": 2299 }, { "epoch": 14.286604361370717, "grad_norm": 2.671875, "learning_rate": 4.9526993949216355e-05, "loss": 0.3129, "step": 2300 }, { "epoch": 14.292834890965732, "grad_norm": 3.65625, "learning_rate": 4.952651683586668e-05, "loss": 0.3815, "step": 2301 }, { "epoch": 14.299065420560748, "grad_norm": 5.75, "learning_rate": 4.952603948431057e-05, "loss": 0.345, "step": 2302 }, { "epoch": 14.305295950155763, "grad_norm": 4.125, "learning_rate": 4.952556189455266e-05, "loss": 0.3423, "step": 2303 }, { "epoch": 14.311526479750778, "grad_norm": 3.546875, "learning_rate": 4.9525084066597594e-05, "loss": 0.4491, "step": 2304 }, { "epoch": 14.317757009345794, "grad_norm": 3.5, "learning_rate": 4.9524606000450005e-05, "loss": 0.2789, "step": 2305 }, { "epoch": 14.32398753894081, "grad_norm": 3.421875, "learning_rate": 4.952412769611454e-05, "loss": 0.2266, "step": 2306 }, { "epoch": 14.330218068535826, "grad_norm": 2.9375, "learning_rate": 4.952364915359584e-05, "loss": 0.2462, "step": 2307 }, { "epoch": 14.336448598130842, "grad_norm": 2.390625, "learning_rate": 4.952317037289857e-05, "loss": 0.2211, "step": 2308 }, { "epoch": 14.342679127725857, "grad_norm": 3.390625, "learning_rate": 4.952269135402736e-05, "loss": 0.276, "step": 2309 }, { "epoch": 14.348909657320872, "grad_norm": 3.5625, "learning_rate": 4.952221209698687e-05, "loss": 0.345, "step": 2310 }, { "epoch": 14.355140186915888, "grad_norm": 2.078125, "learning_rate": 4.952173260178176e-05, "loss": 0.3144, "step": 2311 }, { "epoch": 14.361370716510903, "grad_norm": 4.40625, "learning_rate": 4.9521252868416665e-05, "loss": 0.3777, "step": 2312 }, { "epoch": 14.367601246105918, "grad_norm": 2.53125, "learning_rate": 4.952077289689627e-05, "loss": 0.171, "step": 2313 }, { "epoch": 14.373831775700934, "grad_norm": 2.40625, "learning_rate": 4.952029268722523e-05, "loss": 0.2124, "step": 2314 }, { "epoch": 14.38006230529595, "grad_norm": 2.734375, "learning_rate": 4.951981223940821e-05, "loss": 0.1809, "step": 2315 }, { "epoch": 14.386292834890966, "grad_norm": 2.890625, "learning_rate": 4.951933155344987e-05, "loss": 0.3076, "step": 2316 }, { "epoch": 14.392523364485982, "grad_norm": 4.03125, "learning_rate": 4.951885062935487e-05, "loss": 0.2891, "step": 2317 }, { "epoch": 14.398753894080997, "grad_norm": 3.09375, "learning_rate": 4.95183694671279e-05, "loss": 0.2137, "step": 2318 }, { "epoch": 14.404984423676012, "grad_norm": 2.1875, "learning_rate": 4.951788806677362e-05, "loss": 0.2019, "step": 2319 }, { "epoch": 14.411214953271028, "grad_norm": 3.21875, "learning_rate": 4.951740642829671e-05, "loss": 0.2798, "step": 2320 }, { "epoch": 14.417445482866043, "grad_norm": 3.109375, "learning_rate": 4.951692455170185e-05, "loss": 0.2404, "step": 2321 }, { "epoch": 14.423676012461058, "grad_norm": 5.15625, "learning_rate": 4.9516442436993724e-05, "loss": 0.5644, "step": 2322 }, { "epoch": 14.429906542056075, "grad_norm": 5.03125, "learning_rate": 4.9515960084177005e-05, "loss": 0.3819, "step": 2323 }, { "epoch": 14.43613707165109, "grad_norm": 5.625, "learning_rate": 4.951547749325638e-05, "loss": 0.2824, "step": 2324 }, { "epoch": 14.442367601246106, "grad_norm": 3.328125, "learning_rate": 4.9514994664236535e-05, "loss": 0.3405, "step": 2325 }, { "epoch": 14.448598130841122, "grad_norm": 2.875, "learning_rate": 4.951451159712216e-05, "loss": 0.2931, "step": 2326 }, { "epoch": 14.454828660436137, "grad_norm": 3.984375, "learning_rate": 4.951402829191795e-05, "loss": 0.3152, "step": 2327 }, { "epoch": 14.461059190031152, "grad_norm": 4.71875, "learning_rate": 4.9513544748628596e-05, "loss": 0.2765, "step": 2328 }, { "epoch": 14.467289719626168, "grad_norm": 6.0625, "learning_rate": 4.9513060967258794e-05, "loss": 0.243, "step": 2329 }, { "epoch": 14.473520249221183, "grad_norm": 5.71875, "learning_rate": 4.951257694781325e-05, "loss": 0.3524, "step": 2330 }, { "epoch": 14.4797507788162, "grad_norm": 3.140625, "learning_rate": 4.951209269029665e-05, "loss": 0.2247, "step": 2331 }, { "epoch": 14.485981308411215, "grad_norm": 3.765625, "learning_rate": 4.951160819471372e-05, "loss": 0.2427, "step": 2332 }, { "epoch": 14.49221183800623, "grad_norm": 5.3125, "learning_rate": 4.951112346106914e-05, "loss": 0.3791, "step": 2333 }, { "epoch": 14.498442367601246, "grad_norm": 2.953125, "learning_rate": 4.951063848936763e-05, "loss": 0.3236, "step": 2334 }, { "epoch": 14.504672897196262, "grad_norm": 3.296875, "learning_rate": 4.95101532796139e-05, "loss": 0.401, "step": 2335 }, { "epoch": 14.510903426791277, "grad_norm": 4.0625, "learning_rate": 4.950966783181267e-05, "loss": 0.2776, "step": 2336 }, { "epoch": 14.517133956386292, "grad_norm": 4.0625, "learning_rate": 4.950918214596863e-05, "loss": 0.2809, "step": 2337 }, { "epoch": 14.523364485981308, "grad_norm": 3.4375, "learning_rate": 4.950869622208653e-05, "loss": 0.2445, "step": 2338 }, { "epoch": 14.529595015576325, "grad_norm": 2.25, "learning_rate": 4.950821006017107e-05, "loss": 0.2589, "step": 2339 }, { "epoch": 14.53582554517134, "grad_norm": 3.125, "learning_rate": 4.950772366022697e-05, "loss": 0.2652, "step": 2340 }, { "epoch": 14.542056074766355, "grad_norm": 1.34375, "learning_rate": 4.9507237022258956e-05, "loss": 0.1625, "step": 2341 }, { "epoch": 14.54828660436137, "grad_norm": 2.71875, "learning_rate": 4.950675014627176e-05, "loss": 0.4458, "step": 2342 }, { "epoch": 14.554517133956386, "grad_norm": 1.9453125, "learning_rate": 4.950626303227011e-05, "loss": 0.2235, "step": 2343 }, { "epoch": 14.560747663551401, "grad_norm": 2.671875, "learning_rate": 4.9505775680258734e-05, "loss": 0.3427, "step": 2344 }, { "epoch": 14.566978193146417, "grad_norm": 2.875, "learning_rate": 4.9505288090242364e-05, "loss": 0.2948, "step": 2345 }, { "epoch": 14.573208722741432, "grad_norm": 3.03125, "learning_rate": 4.950480026222574e-05, "loss": 0.3456, "step": 2346 }, { "epoch": 14.57943925233645, "grad_norm": 3.046875, "learning_rate": 4.9504312196213596e-05, "loss": 0.2488, "step": 2347 }, { "epoch": 14.585669781931465, "grad_norm": 3.296875, "learning_rate": 4.950382389221068e-05, "loss": 0.2333, "step": 2348 }, { "epoch": 14.59190031152648, "grad_norm": 3.125, "learning_rate": 4.9503335350221717e-05, "loss": 0.3337, "step": 2349 }, { "epoch": 14.598130841121495, "grad_norm": 2.84375, "learning_rate": 4.950284657025147e-05, "loss": 0.3991, "step": 2350 }, { "epoch": 14.60436137071651, "grad_norm": 2.484375, "learning_rate": 4.950235755230468e-05, "loss": 0.1868, "step": 2351 }, { "epoch": 14.610591900311526, "grad_norm": 2.65625, "learning_rate": 4.950186829638609e-05, "loss": 0.3405, "step": 2352 }, { "epoch": 14.616822429906541, "grad_norm": 2.28125, "learning_rate": 4.950137880250046e-05, "loss": 0.1922, "step": 2353 }, { "epoch": 14.623052959501557, "grad_norm": 2.828125, "learning_rate": 4.9500889070652545e-05, "loss": 0.3919, "step": 2354 }, { "epoch": 14.629283489096574, "grad_norm": 1.8203125, "learning_rate": 4.9500399100847094e-05, "loss": 0.1998, "step": 2355 }, { "epoch": 14.63551401869159, "grad_norm": 2.734375, "learning_rate": 4.949990889308887e-05, "loss": 0.2343, "step": 2356 }, { "epoch": 14.641744548286605, "grad_norm": 3.28125, "learning_rate": 4.949941844738263e-05, "loss": 0.3984, "step": 2357 }, { "epoch": 14.64797507788162, "grad_norm": 2.890625, "learning_rate": 4.949892776373315e-05, "loss": 0.2155, "step": 2358 }, { "epoch": 14.654205607476635, "grad_norm": 3.15625, "learning_rate": 4.949843684214518e-05, "loss": 0.3008, "step": 2359 }, { "epoch": 14.66043613707165, "grad_norm": 2.546875, "learning_rate": 4.9497945682623495e-05, "loss": 0.3047, "step": 2360 }, { "epoch": 14.666666666666666, "grad_norm": 3.875, "learning_rate": 4.9497454285172865e-05, "loss": 0.2358, "step": 2361 }, { "epoch": 14.672897196261681, "grad_norm": 5.3125, "learning_rate": 4.949696264979806e-05, "loss": 0.4388, "step": 2362 }, { "epoch": 14.679127725856699, "grad_norm": 3.078125, "learning_rate": 4.9496470776503855e-05, "loss": 0.2036, "step": 2363 }, { "epoch": 14.685358255451714, "grad_norm": 3.203125, "learning_rate": 4.949597866529503e-05, "loss": 0.2752, "step": 2364 }, { "epoch": 14.69158878504673, "grad_norm": 3.625, "learning_rate": 4.949548631617637e-05, "loss": 0.2998, "step": 2365 }, { "epoch": 14.697819314641745, "grad_norm": 3.78125, "learning_rate": 4.949499372915264e-05, "loss": 0.2853, "step": 2366 }, { "epoch": 14.70404984423676, "grad_norm": 3.46875, "learning_rate": 4.949450090422865e-05, "loss": 0.2515, "step": 2367 }, { "epoch": 14.710280373831775, "grad_norm": 2.9375, "learning_rate": 4.949400784140915e-05, "loss": 0.306, "step": 2368 }, { "epoch": 14.71651090342679, "grad_norm": 1.859375, "learning_rate": 4.9493514540698963e-05, "loss": 0.1833, "step": 2369 }, { "epoch": 14.722741433021806, "grad_norm": 4.21875, "learning_rate": 4.949302100210287e-05, "loss": 0.6816, "step": 2370 }, { "epoch": 14.728971962616823, "grad_norm": 2.828125, "learning_rate": 4.949252722562565e-05, "loss": 0.3138, "step": 2371 }, { "epoch": 14.735202492211839, "grad_norm": 3.5625, "learning_rate": 4.949203321127212e-05, "loss": 0.2012, "step": 2372 }, { "epoch": 14.741433021806854, "grad_norm": 3.890625, "learning_rate": 4.9491538959047053e-05, "loss": 0.3492, "step": 2373 }, { "epoch": 14.74766355140187, "grad_norm": 2.484375, "learning_rate": 4.949104446895528e-05, "loss": 0.4013, "step": 2374 }, { "epoch": 14.753894080996885, "grad_norm": 2.046875, "learning_rate": 4.949054974100158e-05, "loss": 0.2064, "step": 2375 }, { "epoch": 14.7601246105919, "grad_norm": 2.546875, "learning_rate": 4.949005477519077e-05, "loss": 0.2433, "step": 2376 }, { "epoch": 14.766355140186915, "grad_norm": 2.421875, "learning_rate": 4.9489559571527644e-05, "loss": 0.2379, "step": 2377 }, { "epoch": 14.77258566978193, "grad_norm": 2.140625, "learning_rate": 4.9489064130017025e-05, "loss": 0.2225, "step": 2378 }, { "epoch": 14.778816199376948, "grad_norm": 3.25, "learning_rate": 4.948856845066372e-05, "loss": 0.3306, "step": 2379 }, { "epoch": 14.785046728971963, "grad_norm": 3.015625, "learning_rate": 4.9488072533472546e-05, "loss": 0.3334, "step": 2380 }, { "epoch": 14.791277258566979, "grad_norm": 3.90625, "learning_rate": 4.948757637844831e-05, "loss": 0.2455, "step": 2381 }, { "epoch": 14.797507788161994, "grad_norm": 2.328125, "learning_rate": 4.948707998559584e-05, "loss": 0.1943, "step": 2382 }, { "epoch": 14.80373831775701, "grad_norm": 3.203125, "learning_rate": 4.948658335491996e-05, "loss": 0.316, "step": 2383 }, { "epoch": 14.809968847352025, "grad_norm": 2.515625, "learning_rate": 4.9486086486425484e-05, "loss": 0.2979, "step": 2384 }, { "epoch": 14.81619937694704, "grad_norm": 2.78125, "learning_rate": 4.948558938011724e-05, "loss": 0.3352, "step": 2385 }, { "epoch": 14.822429906542055, "grad_norm": 3.953125, "learning_rate": 4.9485092036000055e-05, "loss": 0.3413, "step": 2386 }, { "epoch": 14.828660436137072, "grad_norm": 4.125, "learning_rate": 4.9484594454078765e-05, "loss": 0.317, "step": 2387 }, { "epoch": 14.834890965732088, "grad_norm": 3.46875, "learning_rate": 4.94840966343582e-05, "loss": 0.3701, "step": 2388 }, { "epoch": 14.841121495327103, "grad_norm": 3.953125, "learning_rate": 4.9483598576843196e-05, "loss": 0.2863, "step": 2389 }, { "epoch": 14.847352024922118, "grad_norm": 4.1875, "learning_rate": 4.9483100281538573e-05, "loss": 0.2802, "step": 2390 }, { "epoch": 14.853582554517134, "grad_norm": 4.78125, "learning_rate": 4.9482601748449204e-05, "loss": 0.2723, "step": 2391 }, { "epoch": 14.85981308411215, "grad_norm": 2.8125, "learning_rate": 4.9482102977579905e-05, "loss": 0.4152, "step": 2392 }, { "epoch": 14.866043613707165, "grad_norm": 3.4375, "learning_rate": 4.948160396893553e-05, "loss": 0.6227, "step": 2393 }, { "epoch": 14.87227414330218, "grad_norm": 4.1875, "learning_rate": 4.948110472252092e-05, "loss": 0.2582, "step": 2394 }, { "epoch": 14.878504672897197, "grad_norm": 2.375, "learning_rate": 4.948060523834094e-05, "loss": 0.234, "step": 2395 }, { "epoch": 14.884735202492212, "grad_norm": 4.75, "learning_rate": 4.948010551640041e-05, "loss": 0.3117, "step": 2396 }, { "epoch": 14.890965732087228, "grad_norm": 3.46875, "learning_rate": 4.947960555670421e-05, "loss": 0.3776, "step": 2397 }, { "epoch": 14.897196261682243, "grad_norm": 2.578125, "learning_rate": 4.947910535925719e-05, "loss": 0.2565, "step": 2398 }, { "epoch": 14.903426791277258, "grad_norm": 3.921875, "learning_rate": 4.947860492406421e-05, "loss": 0.2808, "step": 2399 }, { "epoch": 14.909657320872274, "grad_norm": 2.953125, "learning_rate": 4.947810425113011e-05, "loss": 0.244, "step": 2400 }, { "epoch": 14.91588785046729, "grad_norm": 2.1875, "learning_rate": 4.947760334045978e-05, "loss": 0.2634, "step": 2401 }, { "epoch": 14.922118380062305, "grad_norm": 2.96875, "learning_rate": 4.947710219205808e-05, "loss": 0.2996, "step": 2402 }, { "epoch": 14.928348909657322, "grad_norm": 5.03125, "learning_rate": 4.947660080592986e-05, "loss": 0.3631, "step": 2403 }, { "epoch": 14.934579439252337, "grad_norm": 5.40625, "learning_rate": 4.947609918208e-05, "loss": 0.2958, "step": 2404 }, { "epoch": 14.940809968847352, "grad_norm": 3.625, "learning_rate": 4.9475597320513374e-05, "loss": 0.3368, "step": 2405 }, { "epoch": 14.947040498442368, "grad_norm": 2.9375, "learning_rate": 4.947509522123485e-05, "loss": 0.2063, "step": 2406 }, { "epoch": 14.953271028037383, "grad_norm": 4.0, "learning_rate": 4.947459288424932e-05, "loss": 0.2057, "step": 2407 }, { "epoch": 14.959501557632398, "grad_norm": 2.96875, "learning_rate": 4.9474090309561635e-05, "loss": 0.2013, "step": 2408 }, { "epoch": 14.965732087227414, "grad_norm": 3.078125, "learning_rate": 4.947358749717671e-05, "loss": 0.3805, "step": 2409 }, { "epoch": 14.97196261682243, "grad_norm": 3.375, "learning_rate": 4.9473084447099404e-05, "loss": 0.3545, "step": 2410 }, { "epoch": 14.978193146417446, "grad_norm": 3.609375, "learning_rate": 4.947258115933461e-05, "loss": 0.2234, "step": 2411 }, { "epoch": 14.984423676012462, "grad_norm": 3.59375, "learning_rate": 4.947207763388722e-05, "loss": 0.363, "step": 2412 }, { "epoch": 14.990654205607477, "grad_norm": 3.28125, "learning_rate": 4.94715738707621e-05, "loss": 0.2674, "step": 2413 }, { "epoch": 14.996884735202492, "grad_norm": 2.609375, "learning_rate": 4.947106986996418e-05, "loss": 0.2425, "step": 2414 }, { "epoch": 15.0, "grad_norm": 2.234375, "learning_rate": 4.947056563149834e-05, "loss": 0.1624, "step": 2415 }, { "epoch": 15.006230529595015, "grad_norm": 3.578125, "learning_rate": 4.947006115536947e-05, "loss": 0.4272, "step": 2416 }, { "epoch": 15.01246105919003, "grad_norm": 3.9375, "learning_rate": 4.946955644158248e-05, "loss": 0.3557, "step": 2417 }, { "epoch": 15.018691588785046, "grad_norm": 2.734375, "learning_rate": 4.946905149014226e-05, "loss": 0.2737, "step": 2418 }, { "epoch": 15.024922118380061, "grad_norm": 3.09375, "learning_rate": 4.946854630105372e-05, "loss": 0.2399, "step": 2419 }, { "epoch": 15.031152647975079, "grad_norm": 5.4375, "learning_rate": 4.946804087432177e-05, "loss": 0.4785, "step": 2420 }, { "epoch": 15.037383177570094, "grad_norm": 4.0, "learning_rate": 4.9467535209951315e-05, "loss": 0.2515, "step": 2421 }, { "epoch": 15.04361370716511, "grad_norm": 4.125, "learning_rate": 4.946702930794726e-05, "loss": 0.3757, "step": 2422 }, { "epoch": 15.049844236760125, "grad_norm": 5.0, "learning_rate": 4.9466523168314535e-05, "loss": 0.2232, "step": 2423 }, { "epoch": 15.05607476635514, "grad_norm": 3.5625, "learning_rate": 4.946601679105805e-05, "loss": 0.3595, "step": 2424 }, { "epoch": 15.062305295950155, "grad_norm": 3.546875, "learning_rate": 4.9465510176182714e-05, "loss": 0.311, "step": 2425 }, { "epoch": 15.06853582554517, "grad_norm": 2.953125, "learning_rate": 4.946500332369345e-05, "loss": 0.3118, "step": 2426 }, { "epoch": 15.074766355140186, "grad_norm": 3.46875, "learning_rate": 4.946449623359519e-05, "loss": 0.2033, "step": 2427 }, { "epoch": 15.080996884735203, "grad_norm": 4.90625, "learning_rate": 4.9463988905892847e-05, "loss": 0.3474, "step": 2428 }, { "epoch": 15.087227414330219, "grad_norm": 5.09375, "learning_rate": 4.946348134059136e-05, "loss": 0.3397, "step": 2429 }, { "epoch": 15.093457943925234, "grad_norm": 4.90625, "learning_rate": 4.946297353769564e-05, "loss": 0.1987, "step": 2430 }, { "epoch": 15.09968847352025, "grad_norm": 2.71875, "learning_rate": 4.946246549721064e-05, "loss": 0.2872, "step": 2431 }, { "epoch": 15.105919003115265, "grad_norm": 3.46875, "learning_rate": 4.946195721914128e-05, "loss": 0.2256, "step": 2432 }, { "epoch": 15.11214953271028, "grad_norm": 4.46875, "learning_rate": 4.94614487034925e-05, "loss": 0.2991, "step": 2433 }, { "epoch": 15.118380062305295, "grad_norm": 2.9375, "learning_rate": 4.946093995026925e-05, "loss": 0.3223, "step": 2434 }, { "epoch": 15.12461059190031, "grad_norm": 2.15625, "learning_rate": 4.9460430959476456e-05, "loss": 0.2035, "step": 2435 }, { "epoch": 15.130841121495328, "grad_norm": 2.046875, "learning_rate": 4.9459921731119065e-05, "loss": 0.2061, "step": 2436 }, { "epoch": 15.137071651090343, "grad_norm": 2.421875, "learning_rate": 4.945941226520203e-05, "loss": 0.1748, "step": 2437 }, { "epoch": 15.143302180685358, "grad_norm": 3.9375, "learning_rate": 4.945890256173029e-05, "loss": 0.3059, "step": 2438 }, { "epoch": 15.149532710280374, "grad_norm": 2.640625, "learning_rate": 4.945839262070879e-05, "loss": 0.2505, "step": 2439 }, { "epoch": 15.15576323987539, "grad_norm": 2.734375, "learning_rate": 4.945788244214251e-05, "loss": 0.3505, "step": 2440 }, { "epoch": 15.161993769470405, "grad_norm": 2.921875, "learning_rate": 4.945737202603637e-05, "loss": 0.1788, "step": 2441 }, { "epoch": 15.16822429906542, "grad_norm": 2.734375, "learning_rate": 4.9456861372395356e-05, "loss": 0.2579, "step": 2442 }, { "epoch": 15.174454828660435, "grad_norm": 2.75, "learning_rate": 4.9456350481224415e-05, "loss": 0.2005, "step": 2443 }, { "epoch": 15.180685358255452, "grad_norm": 3.953125, "learning_rate": 4.94558393525285e-05, "loss": 0.357, "step": 2444 }, { "epoch": 15.186915887850468, "grad_norm": 2.796875, "learning_rate": 4.9455327986312595e-05, "loss": 0.2233, "step": 2445 }, { "epoch": 15.193146417445483, "grad_norm": 2.875, "learning_rate": 4.9454816382581645e-05, "loss": 0.2841, "step": 2446 }, { "epoch": 15.199376947040498, "grad_norm": 4.625, "learning_rate": 4.945430454134064e-05, "loss": 0.3903, "step": 2447 }, { "epoch": 15.205607476635514, "grad_norm": 3.046875, "learning_rate": 4.945379246259454e-05, "loss": 0.1979, "step": 2448 }, { "epoch": 15.21183800623053, "grad_norm": 4.5625, "learning_rate": 4.9453280146348315e-05, "loss": 0.4923, "step": 2449 }, { "epoch": 15.218068535825545, "grad_norm": 2.171875, "learning_rate": 4.945276759260695e-05, "loss": 0.1888, "step": 2450 }, { "epoch": 15.22429906542056, "grad_norm": 4.03125, "learning_rate": 4.9452254801375407e-05, "loss": 0.2467, "step": 2451 }, { "epoch": 15.230529595015577, "grad_norm": 4.40625, "learning_rate": 4.945174177265869e-05, "loss": 0.3817, "step": 2452 }, { "epoch": 15.236760124610592, "grad_norm": 3.4375, "learning_rate": 4.945122850646176e-05, "loss": 0.4986, "step": 2453 }, { "epoch": 15.242990654205608, "grad_norm": 3.03125, "learning_rate": 4.945071500278962e-05, "loss": 0.2595, "step": 2454 }, { "epoch": 15.249221183800623, "grad_norm": 2.734375, "learning_rate": 4.945020126164724e-05, "loss": 0.2244, "step": 2455 }, { "epoch": 15.255451713395638, "grad_norm": 2.75, "learning_rate": 4.944968728303962e-05, "loss": 0.1899, "step": 2456 }, { "epoch": 15.261682242990654, "grad_norm": 2.1875, "learning_rate": 4.944917306697175e-05, "loss": 0.1796, "step": 2457 }, { "epoch": 15.26791277258567, "grad_norm": 3.28125, "learning_rate": 4.944865861344863e-05, "loss": 0.2832, "step": 2458 }, { "epoch": 15.274143302180685, "grad_norm": 1.7265625, "learning_rate": 4.944814392247525e-05, "loss": 0.163, "step": 2459 }, { "epoch": 15.280373831775702, "grad_norm": 3.421875, "learning_rate": 4.9447628994056604e-05, "loss": 0.3247, "step": 2460 }, { "epoch": 15.286604361370717, "grad_norm": 3.5, "learning_rate": 4.94471138281977e-05, "loss": 0.3648, "step": 2461 }, { "epoch": 15.292834890965732, "grad_norm": 2.625, "learning_rate": 4.944659842490354e-05, "loss": 0.1958, "step": 2462 }, { "epoch": 15.299065420560748, "grad_norm": 3.3125, "learning_rate": 4.944608278417912e-05, "loss": 0.3274, "step": 2463 }, { "epoch": 15.305295950155763, "grad_norm": 2.484375, "learning_rate": 4.944556690602947e-05, "loss": 0.3075, "step": 2464 }, { "epoch": 15.311526479750778, "grad_norm": 4.0, "learning_rate": 4.944505079045958e-05, "loss": 0.4914, "step": 2465 }, { "epoch": 15.317757009345794, "grad_norm": 3.53125, "learning_rate": 4.9444534437474476e-05, "loss": 0.2992, "step": 2466 }, { "epoch": 15.32398753894081, "grad_norm": 2.84375, "learning_rate": 4.944401784707916e-05, "loss": 0.2899, "step": 2467 }, { "epoch": 15.330218068535826, "grad_norm": 2.390625, "learning_rate": 4.9443501019278664e-05, "loss": 0.3159, "step": 2468 }, { "epoch": 15.336448598130842, "grad_norm": 2.15625, "learning_rate": 4.944298395407799e-05, "loss": 0.2631, "step": 2469 }, { "epoch": 15.342679127725857, "grad_norm": 2.984375, "learning_rate": 4.9442466651482184e-05, "loss": 0.2585, "step": 2470 }, { "epoch": 15.348909657320872, "grad_norm": 2.171875, "learning_rate": 4.944194911149625e-05, "loss": 0.1952, "step": 2471 }, { "epoch": 15.355140186915888, "grad_norm": 3.03125, "learning_rate": 4.944143133412521e-05, "loss": 0.3609, "step": 2472 }, { "epoch": 15.361370716510903, "grad_norm": 1.8359375, "learning_rate": 4.944091331937412e-05, "loss": 0.1489, "step": 2473 }, { "epoch": 15.367601246105918, "grad_norm": 2.5, "learning_rate": 4.944039506724798e-05, "loss": 0.2042, "step": 2474 }, { "epoch": 15.373831775700934, "grad_norm": 2.515625, "learning_rate": 4.943987657775185e-05, "loss": 0.3098, "step": 2475 }, { "epoch": 15.38006230529595, "grad_norm": 2.90625, "learning_rate": 4.9439357850890744e-05, "loss": 0.1791, "step": 2476 }, { "epoch": 15.386292834890966, "grad_norm": 2.9375, "learning_rate": 4.9438838886669714e-05, "loss": 0.2815, "step": 2477 }, { "epoch": 15.392523364485982, "grad_norm": 2.203125, "learning_rate": 4.943831968509379e-05, "loss": 0.1873, "step": 2478 }, { "epoch": 15.398753894080997, "grad_norm": 1.78125, "learning_rate": 4.9437800246168025e-05, "loss": 0.2024, "step": 2479 }, { "epoch": 15.404984423676012, "grad_norm": 3.890625, "learning_rate": 4.943728056989746e-05, "loss": 0.6106, "step": 2480 }, { "epoch": 15.411214953271028, "grad_norm": 2.625, "learning_rate": 4.943676065628714e-05, "loss": 0.2413, "step": 2481 }, { "epoch": 15.417445482866043, "grad_norm": 2.8125, "learning_rate": 4.9436240505342115e-05, "loss": 0.2337, "step": 2482 }, { "epoch": 15.423676012461058, "grad_norm": 2.78125, "learning_rate": 4.9435720117067434e-05, "loss": 0.3213, "step": 2483 }, { "epoch": 15.429906542056075, "grad_norm": 3.59375, "learning_rate": 4.943519949146816e-05, "loss": 0.2852, "step": 2484 }, { "epoch": 15.43613707165109, "grad_norm": 2.671875, "learning_rate": 4.943467862854934e-05, "loss": 0.2017, "step": 2485 }, { "epoch": 15.442367601246106, "grad_norm": 3.140625, "learning_rate": 4.9434157528316036e-05, "loss": 0.375, "step": 2486 }, { "epoch": 15.448598130841122, "grad_norm": 2.953125, "learning_rate": 4.943363619077331e-05, "loss": 0.3216, "step": 2487 }, { "epoch": 15.454828660436137, "grad_norm": 2.203125, "learning_rate": 4.943311461592623e-05, "loss": 0.3025, "step": 2488 }, { "epoch": 15.461059190031152, "grad_norm": 2.296875, "learning_rate": 4.943259280377986e-05, "loss": 0.2145, "step": 2489 }, { "epoch": 15.467289719626168, "grad_norm": 3.09375, "learning_rate": 4.9432070754339254e-05, "loss": 0.3284, "step": 2490 }, { "epoch": 15.473520249221183, "grad_norm": 4.40625, "learning_rate": 4.94315484676095e-05, "loss": 0.4496, "step": 2491 }, { "epoch": 15.4797507788162, "grad_norm": 2.5, "learning_rate": 4.943102594359566e-05, "loss": 0.3026, "step": 2492 }, { "epoch": 15.485981308411215, "grad_norm": 1.953125, "learning_rate": 4.943050318230281e-05, "loss": 0.1781, "step": 2493 }, { "epoch": 15.49221183800623, "grad_norm": 4.125, "learning_rate": 4.9429980183736034e-05, "loss": 0.5655, "step": 2494 }, { "epoch": 15.498442367601246, "grad_norm": 1.96875, "learning_rate": 4.942945694790041e-05, "loss": 0.1691, "step": 2495 }, { "epoch": 15.504672897196262, "grad_norm": 2.890625, "learning_rate": 4.942893347480101e-05, "loss": 0.2291, "step": 2496 }, { "epoch": 15.510903426791277, "grad_norm": 3.125, "learning_rate": 4.9428409764442926e-05, "loss": 0.3743, "step": 2497 }, { "epoch": 15.517133956386292, "grad_norm": 1.75, "learning_rate": 4.9427885816831256e-05, "loss": 0.2068, "step": 2498 }, { "epoch": 15.523364485981308, "grad_norm": 4.8125, "learning_rate": 4.942736163197106e-05, "loss": 0.2725, "step": 2499 }, { "epoch": 15.529595015576325, "grad_norm": 3.109375, "learning_rate": 4.942683720986745e-05, "loss": 0.2625, "step": 2500 }, { "epoch": 15.53582554517134, "grad_norm": 3.171875, "learning_rate": 4.942631255052551e-05, "loss": 0.2338, "step": 2501 }, { "epoch": 15.542056074766355, "grad_norm": 3.59375, "learning_rate": 4.942578765395035e-05, "loss": 0.3985, "step": 2502 }, { "epoch": 15.54828660436137, "grad_norm": 4.375, "learning_rate": 4.9425262520147055e-05, "loss": 0.2615, "step": 2503 }, { "epoch": 15.554517133956386, "grad_norm": 4.0625, "learning_rate": 4.942473714912072e-05, "loss": 0.3357, "step": 2504 }, { "epoch": 15.560747663551401, "grad_norm": 2.421875, "learning_rate": 4.9424211540876456e-05, "loss": 0.1877, "step": 2505 }, { "epoch": 15.566978193146417, "grad_norm": 4.125, "learning_rate": 4.9423685695419374e-05, "loss": 0.4378, "step": 2506 }, { "epoch": 15.573208722741432, "grad_norm": 2.59375, "learning_rate": 4.942315961275457e-05, "loss": 0.2436, "step": 2507 }, { "epoch": 15.57943925233645, "grad_norm": 3.875, "learning_rate": 4.942263329288716e-05, "loss": 0.356, "step": 2508 }, { "epoch": 15.585669781931465, "grad_norm": 4.53125, "learning_rate": 4.942210673582226e-05, "loss": 0.4407, "step": 2509 }, { "epoch": 15.59190031152648, "grad_norm": 3.1875, "learning_rate": 4.9421579941564965e-05, "loss": 0.2685, "step": 2510 }, { "epoch": 15.598130841121495, "grad_norm": 1.9296875, "learning_rate": 4.942105291012041e-05, "loss": 0.1673, "step": 2511 }, { "epoch": 15.60436137071651, "grad_norm": 3.53125, "learning_rate": 4.94205256414937e-05, "loss": 0.4421, "step": 2512 }, { "epoch": 15.610591900311526, "grad_norm": 3.546875, "learning_rate": 4.9419998135689974e-05, "loss": 0.2336, "step": 2513 }, { "epoch": 15.616822429906541, "grad_norm": 3.1875, "learning_rate": 4.941947039271434e-05, "loss": 0.2929, "step": 2514 }, { "epoch": 15.623052959501557, "grad_norm": 2.578125, "learning_rate": 4.941894241257193e-05, "loss": 0.2444, "step": 2515 }, { "epoch": 15.629283489096574, "grad_norm": 2.671875, "learning_rate": 4.9418414195267873e-05, "loss": 0.3333, "step": 2516 }, { "epoch": 15.63551401869159, "grad_norm": 3.96875, "learning_rate": 4.941788574080729e-05, "loss": 0.2825, "step": 2517 }, { "epoch": 15.641744548286605, "grad_norm": 3.703125, "learning_rate": 4.941735704919532e-05, "loss": 0.3375, "step": 2518 }, { "epoch": 15.64797507788162, "grad_norm": 3.703125, "learning_rate": 4.9416828120437095e-05, "loss": 0.2876, "step": 2519 }, { "epoch": 15.654205607476635, "grad_norm": 2.828125, "learning_rate": 4.9416298954537763e-05, "loss": 0.338, "step": 2520 }, { "epoch": 15.66043613707165, "grad_norm": 2.25, "learning_rate": 4.941576955150245e-05, "loss": 0.2334, "step": 2521 }, { "epoch": 15.666666666666666, "grad_norm": 3.4375, "learning_rate": 4.94152399113363e-05, "loss": 0.1925, "step": 2522 }, { "epoch": 15.672897196261681, "grad_norm": 3.171875, "learning_rate": 4.9414710034044464e-05, "loss": 0.2885, "step": 2523 }, { "epoch": 15.679127725856699, "grad_norm": 2.3125, "learning_rate": 4.941417991963208e-05, "loss": 0.2117, "step": 2524 }, { "epoch": 15.685358255451714, "grad_norm": 1.859375, "learning_rate": 4.9413649568104305e-05, "loss": 0.1759, "step": 2525 }, { "epoch": 15.69158878504673, "grad_norm": 4.1875, "learning_rate": 4.941311897946628e-05, "loss": 0.2033, "step": 2526 }, { "epoch": 15.697819314641745, "grad_norm": 2.1875, "learning_rate": 4.9412588153723174e-05, "loss": 0.2638, "step": 2527 }, { "epoch": 15.70404984423676, "grad_norm": 3.78125, "learning_rate": 4.941205709088011e-05, "loss": 0.3767, "step": 2528 }, { "epoch": 15.710280373831775, "grad_norm": 2.796875, "learning_rate": 4.941152579094229e-05, "loss": 0.2264, "step": 2529 }, { "epoch": 15.71651090342679, "grad_norm": 4.0, "learning_rate": 4.9410994253914845e-05, "loss": 0.4684, "step": 2530 }, { "epoch": 15.722741433021806, "grad_norm": 2.546875, "learning_rate": 4.9410462479802945e-05, "loss": 0.2088, "step": 2531 }, { "epoch": 15.728971962616823, "grad_norm": 2.4375, "learning_rate": 4.940993046861174e-05, "loss": 0.2815, "step": 2532 }, { "epoch": 15.735202492211839, "grad_norm": 3.15625, "learning_rate": 4.940939822034643e-05, "loss": 0.2404, "step": 2533 }, { "epoch": 15.741433021806854, "grad_norm": 2.3125, "learning_rate": 4.9408865735012154e-05, "loss": 0.2008, "step": 2534 }, { "epoch": 15.74766355140187, "grad_norm": 2.1875, "learning_rate": 4.94083330126141e-05, "loss": 0.2032, "step": 2535 }, { "epoch": 15.753894080996885, "grad_norm": 2.71875, "learning_rate": 4.940780005315744e-05, "loss": 0.2162, "step": 2536 }, { "epoch": 15.7601246105919, "grad_norm": 2.5, "learning_rate": 4.940726685664734e-05, "loss": 0.1769, "step": 2537 }, { "epoch": 15.766355140186915, "grad_norm": 3.421875, "learning_rate": 4.940673342308898e-05, "loss": 0.3786, "step": 2538 }, { "epoch": 15.77258566978193, "grad_norm": 3.453125, "learning_rate": 4.940619975248756e-05, "loss": 0.3251, "step": 2539 }, { "epoch": 15.778816199376948, "grad_norm": 3.453125, "learning_rate": 4.940566584484824e-05, "loss": 0.3536, "step": 2540 }, { "epoch": 15.785046728971963, "grad_norm": 2.984375, "learning_rate": 4.9405131700176215e-05, "loss": 0.3622, "step": 2541 }, { "epoch": 15.791277258566979, "grad_norm": 3.953125, "learning_rate": 4.940459731847667e-05, "loss": 0.4363, "step": 2542 }, { "epoch": 15.797507788161994, "grad_norm": 3.046875, "learning_rate": 4.94040626997548e-05, "loss": 0.3148, "step": 2543 }, { "epoch": 15.80373831775701, "grad_norm": 4.65625, "learning_rate": 4.9403527844015805e-05, "loss": 0.5927, "step": 2544 }, { "epoch": 15.809968847352025, "grad_norm": 3.0, "learning_rate": 4.940299275126486e-05, "loss": 0.2375, "step": 2545 }, { "epoch": 15.81619937694704, "grad_norm": 2.28125, "learning_rate": 4.940245742150718e-05, "loss": 0.2705, "step": 2546 }, { "epoch": 15.822429906542055, "grad_norm": 3.21875, "learning_rate": 4.940192185474794e-05, "loss": 0.3224, "step": 2547 }, { "epoch": 15.828660436137072, "grad_norm": 2.265625, "learning_rate": 4.9401386050992365e-05, "loss": 0.2143, "step": 2548 }, { "epoch": 15.834890965732088, "grad_norm": 2.5625, "learning_rate": 4.9400850010245656e-05, "loss": 0.2607, "step": 2549 }, { "epoch": 15.841121495327103, "grad_norm": 1.4921875, "learning_rate": 4.9400313732513006e-05, "loss": 0.1591, "step": 2550 }, { "epoch": 15.847352024922118, "grad_norm": 2.375, "learning_rate": 4.939977721779964e-05, "loss": 0.3345, "step": 2551 }, { "epoch": 15.853582554517134, "grad_norm": 2.734375, "learning_rate": 4.9399240466110754e-05, "loss": 0.2043, "step": 2552 }, { "epoch": 15.85981308411215, "grad_norm": 3.25, "learning_rate": 4.939870347745156e-05, "loss": 0.2457, "step": 2553 }, { "epoch": 15.866043613707165, "grad_norm": 1.9765625, "learning_rate": 4.939816625182729e-05, "loss": 0.2094, "step": 2554 }, { "epoch": 15.87227414330218, "grad_norm": 2.453125, "learning_rate": 4.9397628789243155e-05, "loss": 0.2692, "step": 2555 }, { "epoch": 15.878504672897197, "grad_norm": 3.25, "learning_rate": 4.9397091089704364e-05, "loss": 0.3039, "step": 2556 }, { "epoch": 15.884735202492212, "grad_norm": 3.53125, "learning_rate": 4.939655315321615e-05, "loss": 0.3357, "step": 2557 }, { "epoch": 15.890965732087228, "grad_norm": 2.359375, "learning_rate": 4.9396014979783736e-05, "loss": 0.2226, "step": 2558 }, { "epoch": 15.897196261682243, "grad_norm": 3.046875, "learning_rate": 4.9395476569412355e-05, "loss": 0.1659, "step": 2559 }, { "epoch": 15.903426791277258, "grad_norm": 4.90625, "learning_rate": 4.9394937922107224e-05, "loss": 0.3771, "step": 2560 }, { "epoch": 15.909657320872274, "grad_norm": 3.296875, "learning_rate": 4.9394399037873574e-05, "loss": 0.4929, "step": 2561 }, { "epoch": 15.91588785046729, "grad_norm": 2.90625, "learning_rate": 4.939385991671664e-05, "loss": 0.2242, "step": 2562 }, { "epoch": 15.922118380062305, "grad_norm": 3.765625, "learning_rate": 4.9393320558641676e-05, "loss": 0.3404, "step": 2563 }, { "epoch": 15.928348909657322, "grad_norm": 5.03125, "learning_rate": 4.9392780963653905e-05, "loss": 0.3927, "step": 2564 }, { "epoch": 15.934579439252337, "grad_norm": 5.40625, "learning_rate": 4.939224113175857e-05, "loss": 0.2312, "step": 2565 }, { "epoch": 15.940809968847352, "grad_norm": 3.65625, "learning_rate": 4.939170106296091e-05, "loss": 0.3743, "step": 2566 }, { "epoch": 15.947040498442368, "grad_norm": 2.578125, "learning_rate": 4.9391160757266164e-05, "loss": 0.3677, "step": 2567 }, { "epoch": 15.953271028037383, "grad_norm": 4.03125, "learning_rate": 4.9390620214679595e-05, "loss": 0.3727, "step": 2568 }, { "epoch": 15.959501557632398, "grad_norm": 4.9375, "learning_rate": 4.9390079435206446e-05, "loss": 0.3201, "step": 2569 }, { "epoch": 15.965732087227414, "grad_norm": 5.0, "learning_rate": 4.9389538418851974e-05, "loss": 0.2881, "step": 2570 }, { "epoch": 15.97196261682243, "grad_norm": 3.453125, "learning_rate": 4.938899716562143e-05, "loss": 0.3272, "step": 2571 }, { "epoch": 15.978193146417446, "grad_norm": 3.4375, "learning_rate": 4.938845567552007e-05, "loss": 0.3433, "step": 2572 }, { "epoch": 15.984423676012462, "grad_norm": 4.75, "learning_rate": 4.938791394855316e-05, "loss": 0.2934, "step": 2573 }, { "epoch": 15.990654205607477, "grad_norm": 4.03125, "learning_rate": 4.938737198472594e-05, "loss": 0.3508, "step": 2574 }, { "epoch": 15.996884735202492, "grad_norm": 2.265625, "learning_rate": 4.93868297840437e-05, "loss": 0.263, "step": 2575 }, { "epoch": 16.0, "grad_norm": 1.96875, "learning_rate": 4.938628734651168e-05, "loss": 0.1443, "step": 2576 }, { "epoch": 16.006230529595015, "grad_norm": 3.390625, "learning_rate": 4.938574467213518e-05, "loss": 0.2562, "step": 2577 }, { "epoch": 16.01246105919003, "grad_norm": 4.34375, "learning_rate": 4.938520176091944e-05, "loss": 0.3296, "step": 2578 }, { "epoch": 16.018691588785046, "grad_norm": 2.359375, "learning_rate": 4.9384658612869754e-05, "loss": 0.2099, "step": 2579 }, { "epoch": 16.02492211838006, "grad_norm": 5.65625, "learning_rate": 4.938411522799139e-05, "loss": 0.4087, "step": 2580 }, { "epoch": 16.031152647975077, "grad_norm": 2.65625, "learning_rate": 4.938357160628962e-05, "loss": 0.2312, "step": 2581 }, { "epoch": 16.037383177570092, "grad_norm": 2.78125, "learning_rate": 4.938302774776973e-05, "loss": 0.2505, "step": 2582 }, { "epoch": 16.043613707165107, "grad_norm": 2.1875, "learning_rate": 4.9382483652437e-05, "loss": 0.1954, "step": 2583 }, { "epoch": 16.049844236760123, "grad_norm": 4.0625, "learning_rate": 4.938193932029671e-05, "loss": 0.3704, "step": 2584 }, { "epoch": 16.05607476635514, "grad_norm": 3.296875, "learning_rate": 4.938139475135416e-05, "loss": 0.5484, "step": 2585 }, { "epoch": 16.062305295950157, "grad_norm": 3.640625, "learning_rate": 4.938084994561463e-05, "loss": 0.4022, "step": 2586 }, { "epoch": 16.068535825545172, "grad_norm": 4.03125, "learning_rate": 4.9380304903083404e-05, "loss": 0.3085, "step": 2587 }, { "epoch": 16.074766355140188, "grad_norm": 2.578125, "learning_rate": 4.937975962376579e-05, "loss": 0.1838, "step": 2588 }, { "epoch": 16.080996884735203, "grad_norm": 2.921875, "learning_rate": 4.937921410766707e-05, "loss": 0.3536, "step": 2589 }, { "epoch": 16.08722741433022, "grad_norm": 3.140625, "learning_rate": 4.937866835479256e-05, "loss": 0.1995, "step": 2590 }, { "epoch": 16.093457943925234, "grad_norm": 3.890625, "learning_rate": 4.9378122365147536e-05, "loss": 0.256, "step": 2591 }, { "epoch": 16.09968847352025, "grad_norm": 3.515625, "learning_rate": 4.937757613873732e-05, "loss": 0.2492, "step": 2592 }, { "epoch": 16.105919003115265, "grad_norm": 3.25, "learning_rate": 4.937702967556722e-05, "loss": 0.2563, "step": 2593 }, { "epoch": 16.11214953271028, "grad_norm": 2.5, "learning_rate": 4.937648297564252e-05, "loss": 0.1782, "step": 2594 }, { "epoch": 16.118380062305295, "grad_norm": 2.90625, "learning_rate": 4.9375936038968565e-05, "loss": 0.2522, "step": 2595 }, { "epoch": 16.12461059190031, "grad_norm": 3.171875, "learning_rate": 4.9375388865550635e-05, "loss": 0.2217, "step": 2596 }, { "epoch": 16.130841121495326, "grad_norm": 2.734375, "learning_rate": 4.937484145539405e-05, "loss": 0.2828, "step": 2597 }, { "epoch": 16.13707165109034, "grad_norm": 2.1875, "learning_rate": 4.9374293808504136e-05, "loss": 0.1846, "step": 2598 }, { "epoch": 16.143302180685357, "grad_norm": 2.671875, "learning_rate": 4.937374592488621e-05, "loss": 0.2246, "step": 2599 }, { "epoch": 16.149532710280372, "grad_norm": 3.03125, "learning_rate": 4.937319780454559e-05, "loss": 0.3297, "step": 2600 }, { "epoch": 16.15576323987539, "grad_norm": 3.015625, "learning_rate": 4.9372649447487604e-05, "loss": 0.3195, "step": 2601 }, { "epoch": 16.161993769470406, "grad_norm": 3.921875, "learning_rate": 4.9372100853717577e-05, "loss": 0.437, "step": 2602 }, { "epoch": 16.16822429906542, "grad_norm": 3.46875, "learning_rate": 4.937155202324083e-05, "loss": 0.4745, "step": 2603 }, { "epoch": 16.174454828660437, "grad_norm": 2.296875, "learning_rate": 4.9371002956062694e-05, "loss": 0.2687, "step": 2604 }, { "epoch": 16.180685358255452, "grad_norm": 3.265625, "learning_rate": 4.937045365218851e-05, "loss": 0.2725, "step": 2605 }, { "epoch": 16.186915887850468, "grad_norm": 2.5, "learning_rate": 4.936990411162361e-05, "loss": 0.2215, "step": 2606 }, { "epoch": 16.193146417445483, "grad_norm": 4.15625, "learning_rate": 4.936935433437333e-05, "loss": 0.2344, "step": 2607 }, { "epoch": 16.1993769470405, "grad_norm": 2.890625, "learning_rate": 4.9368804320443004e-05, "loss": 0.1956, "step": 2608 }, { "epoch": 16.205607476635514, "grad_norm": 2.734375, "learning_rate": 4.936825406983798e-05, "loss": 0.2581, "step": 2609 }, { "epoch": 16.21183800623053, "grad_norm": 4.28125, "learning_rate": 4.9367703582563606e-05, "loss": 0.3117, "step": 2610 }, { "epoch": 16.218068535825545, "grad_norm": 4.46875, "learning_rate": 4.936715285862523e-05, "loss": 0.3035, "step": 2611 }, { "epoch": 16.22429906542056, "grad_norm": 4.09375, "learning_rate": 4.936660189802818e-05, "loss": 0.4918, "step": 2612 }, { "epoch": 16.230529595015575, "grad_norm": 2.5625, "learning_rate": 4.936605070077783e-05, "loss": 0.206, "step": 2613 }, { "epoch": 16.23676012461059, "grad_norm": 3.078125, "learning_rate": 4.936549926687952e-05, "loss": 0.2708, "step": 2614 }, { "epoch": 16.242990654205606, "grad_norm": 2.65625, "learning_rate": 4.936494759633862e-05, "loss": 0.2371, "step": 2615 }, { "epoch": 16.24922118380062, "grad_norm": 2.640625, "learning_rate": 4.936439568916047e-05, "loss": 0.2574, "step": 2616 }, { "epoch": 16.25545171339564, "grad_norm": 3.28125, "learning_rate": 4.936384354535044e-05, "loss": 0.3581, "step": 2617 }, { "epoch": 16.261682242990656, "grad_norm": 3.203125, "learning_rate": 4.93632911649139e-05, "loss": 0.2932, "step": 2618 }, { "epoch": 16.26791277258567, "grad_norm": 5.21875, "learning_rate": 4.936273854785619e-05, "loss": 0.2479, "step": 2619 }, { "epoch": 16.274143302180686, "grad_norm": 3.421875, "learning_rate": 4.9362185694182706e-05, "loss": 0.2978, "step": 2620 }, { "epoch": 16.2803738317757, "grad_norm": 3.265625, "learning_rate": 4.93616326038988e-05, "loss": 0.2741, "step": 2621 }, { "epoch": 16.286604361370717, "grad_norm": 2.640625, "learning_rate": 4.936107927700985e-05, "loss": 0.2125, "step": 2622 }, { "epoch": 16.292834890965732, "grad_norm": 3.5, "learning_rate": 4.936052571352122e-05, "loss": 0.2716, "step": 2623 }, { "epoch": 16.299065420560748, "grad_norm": 3.28125, "learning_rate": 4.93599719134383e-05, "loss": 0.3342, "step": 2624 }, { "epoch": 16.305295950155763, "grad_norm": 4.0625, "learning_rate": 4.935941787676647e-05, "loss": 0.305, "step": 2625 }, { "epoch": 16.31152647975078, "grad_norm": 3.203125, "learning_rate": 4.9358863603511095e-05, "loss": 0.314, "step": 2626 }, { "epoch": 16.317757009345794, "grad_norm": 2.921875, "learning_rate": 4.935830909367757e-05, "loss": 0.2842, "step": 2627 }, { "epoch": 16.32398753894081, "grad_norm": 2.625, "learning_rate": 4.935775434727128e-05, "loss": 0.1879, "step": 2628 }, { "epoch": 16.330218068535824, "grad_norm": 4.40625, "learning_rate": 4.9357199364297615e-05, "loss": 0.5024, "step": 2629 }, { "epoch": 16.33644859813084, "grad_norm": 4.375, "learning_rate": 4.935664414476196e-05, "loss": 0.5419, "step": 2630 }, { "epoch": 16.342679127725855, "grad_norm": 2.734375, "learning_rate": 4.9356088688669696e-05, "loss": 0.19, "step": 2631 }, { "epoch": 16.34890965732087, "grad_norm": 3.4375, "learning_rate": 4.9355532996026246e-05, "loss": 0.4606, "step": 2632 }, { "epoch": 16.35514018691589, "grad_norm": 1.5, "learning_rate": 4.9354977066836986e-05, "loss": 0.1849, "step": 2633 }, { "epoch": 16.361370716510905, "grad_norm": 3.0625, "learning_rate": 4.935442090110731e-05, "loss": 0.2495, "step": 2634 }, { "epoch": 16.36760124610592, "grad_norm": 1.578125, "learning_rate": 4.9353864498842636e-05, "loss": 0.1501, "step": 2635 }, { "epoch": 16.373831775700936, "grad_norm": 2.75, "learning_rate": 4.935330786004837e-05, "loss": 0.2151, "step": 2636 }, { "epoch": 16.38006230529595, "grad_norm": 2.875, "learning_rate": 4.93527509847299e-05, "loss": 0.2509, "step": 2637 }, { "epoch": 16.386292834890966, "grad_norm": 3.46875, "learning_rate": 4.935219387289265e-05, "loss": 0.405, "step": 2638 }, { "epoch": 16.39252336448598, "grad_norm": 4.65625, "learning_rate": 4.935163652454202e-05, "loss": 0.3343, "step": 2639 }, { "epoch": 16.398753894080997, "grad_norm": 3.625, "learning_rate": 4.9351078939683426e-05, "loss": 0.3095, "step": 2640 }, { "epoch": 16.404984423676012, "grad_norm": 3.40625, "learning_rate": 4.935052111832229e-05, "loss": 0.2837, "step": 2641 }, { "epoch": 16.411214953271028, "grad_norm": 2.6875, "learning_rate": 4.934996306046403e-05, "loss": 0.1927, "step": 2642 }, { "epoch": 16.417445482866043, "grad_norm": 4.84375, "learning_rate": 4.9349404766114063e-05, "loss": 0.3004, "step": 2643 }, { "epoch": 16.42367601246106, "grad_norm": 3.609375, "learning_rate": 4.9348846235277804e-05, "loss": 0.4334, "step": 2644 }, { "epoch": 16.429906542056074, "grad_norm": 2.921875, "learning_rate": 4.934828746796067e-05, "loss": 0.2579, "step": 2645 }, { "epoch": 16.43613707165109, "grad_norm": 4.75, "learning_rate": 4.934772846416812e-05, "loss": 0.3587, "step": 2646 }, { "epoch": 16.442367601246104, "grad_norm": 3.03125, "learning_rate": 4.934716922390556e-05, "loss": 0.3096, "step": 2647 }, { "epoch": 16.44859813084112, "grad_norm": 3.453125, "learning_rate": 4.9346609747178426e-05, "loss": 0.24, "step": 2648 }, { "epoch": 16.45482866043614, "grad_norm": 2.703125, "learning_rate": 4.9346050033992145e-05, "loss": 0.3318, "step": 2649 }, { "epoch": 16.461059190031154, "grad_norm": 2.484375, "learning_rate": 4.9345490084352165e-05, "loss": 0.2675, "step": 2650 }, { "epoch": 16.46728971962617, "grad_norm": 3.1875, "learning_rate": 4.9344929898263914e-05, "loss": 0.3438, "step": 2651 }, { "epoch": 16.473520249221185, "grad_norm": 3.84375, "learning_rate": 4.9344369475732846e-05, "loss": 0.2118, "step": 2652 }, { "epoch": 16.4797507788162, "grad_norm": 2.390625, "learning_rate": 4.934380881676439e-05, "loss": 0.1759, "step": 2653 }, { "epoch": 16.485981308411215, "grad_norm": 2.234375, "learning_rate": 4.934324792136399e-05, "loss": 0.2005, "step": 2654 }, { "epoch": 16.49221183800623, "grad_norm": 3.5, "learning_rate": 4.934268678953711e-05, "loss": 0.36, "step": 2655 }, { "epoch": 16.498442367601246, "grad_norm": 4.625, "learning_rate": 4.934212542128919e-05, "loss": 0.2233, "step": 2656 }, { "epoch": 16.50467289719626, "grad_norm": 3.0625, "learning_rate": 4.934156381662567e-05, "loss": 0.2592, "step": 2657 }, { "epoch": 16.510903426791277, "grad_norm": 3.609375, "learning_rate": 4.9341001975552027e-05, "loss": 0.209, "step": 2658 }, { "epoch": 16.517133956386292, "grad_norm": 3.09375, "learning_rate": 4.934043989807371e-05, "loss": 0.2364, "step": 2659 }, { "epoch": 16.523364485981308, "grad_norm": 5.3125, "learning_rate": 4.933987758419617e-05, "loss": 0.2205, "step": 2660 }, { "epoch": 16.529595015576323, "grad_norm": 4.46875, "learning_rate": 4.9339315033924874e-05, "loss": 0.2359, "step": 2661 }, { "epoch": 16.53582554517134, "grad_norm": 4.625, "learning_rate": 4.933875224726529e-05, "loss": 0.3235, "step": 2662 }, { "epoch": 16.542056074766354, "grad_norm": 3.625, "learning_rate": 4.933818922422287e-05, "loss": 0.2059, "step": 2663 }, { "epoch": 16.54828660436137, "grad_norm": 3.5, "learning_rate": 4.933762596480309e-05, "loss": 0.4024, "step": 2664 }, { "epoch": 16.554517133956388, "grad_norm": 4.3125, "learning_rate": 4.933706246901143e-05, "loss": 0.2762, "step": 2665 }, { "epoch": 16.560747663551403, "grad_norm": 4.34375, "learning_rate": 4.9336498736853347e-05, "loss": 0.2363, "step": 2666 }, { "epoch": 16.56697819314642, "grad_norm": 3.859375, "learning_rate": 4.9335934768334326e-05, "loss": 0.2733, "step": 2667 }, { "epoch": 16.573208722741434, "grad_norm": 2.734375, "learning_rate": 4.933537056345985e-05, "loss": 0.2201, "step": 2668 }, { "epoch": 16.57943925233645, "grad_norm": 3.15625, "learning_rate": 4.9334806122235376e-05, "loss": 0.3443, "step": 2669 }, { "epoch": 16.585669781931465, "grad_norm": 5.59375, "learning_rate": 4.933424144466641e-05, "loss": 0.3055, "step": 2670 }, { "epoch": 16.59190031152648, "grad_norm": 6.09375, "learning_rate": 4.933367653075843e-05, "loss": 0.3328, "step": 2671 }, { "epoch": 16.598130841121495, "grad_norm": 4.3125, "learning_rate": 4.9333111380516914e-05, "loss": 0.2235, "step": 2672 }, { "epoch": 16.60436137071651, "grad_norm": 3.78125, "learning_rate": 4.9332545993947355e-05, "loss": 0.2929, "step": 2673 }, { "epoch": 16.610591900311526, "grad_norm": 2.78125, "learning_rate": 4.9331980371055244e-05, "loss": 0.2201, "step": 2674 }, { "epoch": 16.61682242990654, "grad_norm": 4.0625, "learning_rate": 4.933141451184608e-05, "loss": 0.3419, "step": 2675 }, { "epoch": 16.623052959501557, "grad_norm": 2.421875, "learning_rate": 4.933084841632535e-05, "loss": 0.2217, "step": 2676 }, { "epoch": 16.629283489096572, "grad_norm": 3.15625, "learning_rate": 4.933028208449856e-05, "loss": 0.3528, "step": 2677 }, { "epoch": 16.635514018691588, "grad_norm": 2.609375, "learning_rate": 4.932971551637121e-05, "loss": 0.3351, "step": 2678 }, { "epoch": 16.641744548286603, "grad_norm": 2.046875, "learning_rate": 4.932914871194879e-05, "loss": 0.1828, "step": 2679 }, { "epoch": 16.64797507788162, "grad_norm": 2.15625, "learning_rate": 4.9328581671236824e-05, "loss": 0.1753, "step": 2680 }, { "epoch": 16.654205607476637, "grad_norm": 4.1875, "learning_rate": 4.9328014394240804e-05, "loss": 0.3926, "step": 2681 }, { "epoch": 16.660436137071652, "grad_norm": 1.4609375, "learning_rate": 4.932744688096625e-05, "loss": 0.1832, "step": 2682 }, { "epoch": 16.666666666666668, "grad_norm": 2.375, "learning_rate": 4.9326879131418666e-05, "loss": 0.2308, "step": 2683 }, { "epoch": 16.672897196261683, "grad_norm": 2.578125, "learning_rate": 4.932631114560357e-05, "loss": 0.1615, "step": 2684 }, { "epoch": 16.6791277258567, "grad_norm": 3.21875, "learning_rate": 4.9325742923526476e-05, "loss": 0.3325, "step": 2685 }, { "epoch": 16.685358255451714, "grad_norm": 3.078125, "learning_rate": 4.93251744651929e-05, "loss": 0.308, "step": 2686 }, { "epoch": 16.69158878504673, "grad_norm": 3.453125, "learning_rate": 4.9324605770608375e-05, "loss": 0.3455, "step": 2687 }, { "epoch": 16.697819314641745, "grad_norm": 3.015625, "learning_rate": 4.932403683977842e-05, "loss": 0.2414, "step": 2688 }, { "epoch": 16.70404984423676, "grad_norm": 2.5, "learning_rate": 4.9323467672708545e-05, "loss": 0.2001, "step": 2689 }, { "epoch": 16.710280373831775, "grad_norm": 2.09375, "learning_rate": 4.9322898269404293e-05, "loss": 0.2032, "step": 2690 }, { "epoch": 16.71651090342679, "grad_norm": 2.40625, "learning_rate": 4.932232862987119e-05, "loss": 0.2223, "step": 2691 }, { "epoch": 16.722741433021806, "grad_norm": 2.765625, "learning_rate": 4.932175875411478e-05, "loss": 0.2558, "step": 2692 }, { "epoch": 16.72897196261682, "grad_norm": 2.25, "learning_rate": 4.9321188642140575e-05, "loss": 0.2414, "step": 2693 }, { "epoch": 16.735202492211837, "grad_norm": 2.828125, "learning_rate": 4.932061829395413e-05, "loss": 0.3136, "step": 2694 }, { "epoch": 16.741433021806852, "grad_norm": 2.703125, "learning_rate": 4.932004770956098e-05, "loss": 0.268, "step": 2695 }, { "epoch": 16.747663551401867, "grad_norm": 3.6875, "learning_rate": 4.931947688896666e-05, "loss": 0.3009, "step": 2696 }, { "epoch": 16.753894080996886, "grad_norm": 2.90625, "learning_rate": 4.931890583217672e-05, "loss": 0.2336, "step": 2697 }, { "epoch": 16.7601246105919, "grad_norm": 2.171875, "learning_rate": 4.931833453919671e-05, "loss": 0.1967, "step": 2698 }, { "epoch": 16.766355140186917, "grad_norm": 3.28125, "learning_rate": 4.931776301003217e-05, "loss": 0.3468, "step": 2699 }, { "epoch": 16.772585669781932, "grad_norm": 3.140625, "learning_rate": 4.931719124468865e-05, "loss": 0.1854, "step": 2700 }, { "epoch": 16.778816199376948, "grad_norm": 4.0, "learning_rate": 4.9316619243171714e-05, "loss": 0.2077, "step": 2701 }, { "epoch": 16.785046728971963, "grad_norm": 2.640625, "learning_rate": 4.931604700548691e-05, "loss": 0.2355, "step": 2702 }, { "epoch": 16.79127725856698, "grad_norm": 4.21875, "learning_rate": 4.93154745316398e-05, "loss": 0.3227, "step": 2703 }, { "epoch": 16.797507788161994, "grad_norm": 4.3125, "learning_rate": 4.931490182163593e-05, "loss": 0.3373, "step": 2704 }, { "epoch": 16.80373831775701, "grad_norm": 2.5625, "learning_rate": 4.931432887548089e-05, "loss": 0.332, "step": 2705 }, { "epoch": 16.809968847352025, "grad_norm": 4.09375, "learning_rate": 4.931375569318021e-05, "loss": 0.3916, "step": 2706 }, { "epoch": 16.81619937694704, "grad_norm": 2.59375, "learning_rate": 4.931318227473949e-05, "loss": 0.2215, "step": 2707 }, { "epoch": 16.822429906542055, "grad_norm": 1.9453125, "learning_rate": 4.931260862016427e-05, "loss": 0.2202, "step": 2708 }, { "epoch": 16.82866043613707, "grad_norm": 2.65625, "learning_rate": 4.9312034729460144e-05, "loss": 0.1828, "step": 2709 }, { "epoch": 16.834890965732086, "grad_norm": 2.5, "learning_rate": 4.931146060263268e-05, "loss": 0.1975, "step": 2710 }, { "epoch": 16.8411214953271, "grad_norm": 2.65625, "learning_rate": 4.931088623968744e-05, "loss": 0.298, "step": 2711 }, { "epoch": 16.847352024922117, "grad_norm": 3.609375, "learning_rate": 4.9310311640630017e-05, "loss": 0.3084, "step": 2712 }, { "epoch": 16.853582554517136, "grad_norm": 2.5625, "learning_rate": 4.930973680546599e-05, "loss": 0.208, "step": 2713 }, { "epoch": 16.85981308411215, "grad_norm": 2.21875, "learning_rate": 4.9309161734200934e-05, "loss": 0.2015, "step": 2714 }, { "epoch": 16.866043613707166, "grad_norm": 3.453125, "learning_rate": 4.9308586426840454e-05, "loss": 0.2851, "step": 2715 }, { "epoch": 16.87227414330218, "grad_norm": 3.71875, "learning_rate": 4.930801088339011e-05, "loss": 0.336, "step": 2716 }, { "epoch": 16.878504672897197, "grad_norm": 2.421875, "learning_rate": 4.9307435103855507e-05, "loss": 0.1921, "step": 2717 }, { "epoch": 16.884735202492212, "grad_norm": 2.71875, "learning_rate": 4.930685908824224e-05, "loss": 0.2339, "step": 2718 }, { "epoch": 16.890965732087228, "grad_norm": 2.84375, "learning_rate": 4.930628283655589e-05, "loss": 0.2511, "step": 2719 }, { "epoch": 16.897196261682243, "grad_norm": 3.34375, "learning_rate": 4.930570634880207e-05, "loss": 0.2649, "step": 2720 }, { "epoch": 16.90342679127726, "grad_norm": 2.609375, "learning_rate": 4.930512962498638e-05, "loss": 0.2109, "step": 2721 }, { "epoch": 16.909657320872274, "grad_norm": 3.015625, "learning_rate": 4.93045526651144e-05, "loss": 0.2105, "step": 2722 }, { "epoch": 16.91588785046729, "grad_norm": 3.484375, "learning_rate": 4.930397546919174e-05, "loss": 0.304, "step": 2723 }, { "epoch": 16.922118380062305, "grad_norm": 3.0, "learning_rate": 4.930339803722402e-05, "loss": 0.4596, "step": 2724 }, { "epoch": 16.92834890965732, "grad_norm": 3.828125, "learning_rate": 4.930282036921685e-05, "loss": 0.4747, "step": 2725 }, { "epoch": 16.934579439252335, "grad_norm": 1.8515625, "learning_rate": 4.9302242465175816e-05, "loss": 0.1835, "step": 2726 }, { "epoch": 16.94080996884735, "grad_norm": 4.1875, "learning_rate": 4.930166432510656e-05, "loss": 0.3725, "step": 2727 }, { "epoch": 16.947040498442366, "grad_norm": 4.375, "learning_rate": 4.930108594901467e-05, "loss": 0.2455, "step": 2728 }, { "epoch": 16.953271028037385, "grad_norm": 4.0625, "learning_rate": 4.9300507336905775e-05, "loss": 0.1543, "step": 2729 }, { "epoch": 16.9595015576324, "grad_norm": 2.21875, "learning_rate": 4.92999284887855e-05, "loss": 0.1527, "step": 2730 }, { "epoch": 16.965732087227416, "grad_norm": 4.8125, "learning_rate": 4.929934940465946e-05, "loss": 0.1851, "step": 2731 }, { "epoch": 16.97196261682243, "grad_norm": 5.5, "learning_rate": 4.9298770084533286e-05, "loss": 0.1962, "step": 2732 }, { "epoch": 16.978193146417446, "grad_norm": 5.65625, "learning_rate": 4.9298190528412594e-05, "loss": 0.2961, "step": 2733 }, { "epoch": 16.98442367601246, "grad_norm": 2.25, "learning_rate": 4.929761073630302e-05, "loss": 0.178, "step": 2734 }, { "epoch": 16.990654205607477, "grad_norm": 3.875, "learning_rate": 4.929703070821019e-05, "loss": 0.2865, "step": 2735 }, { "epoch": 16.996884735202492, "grad_norm": 5.5, "learning_rate": 4.929645044413975e-05, "loss": 0.2261, "step": 2736 }, { "epoch": 17.0, "grad_norm": 2.4375, "learning_rate": 4.929586994409732e-05, "loss": 0.1148, "step": 2737 }, { "epoch": 17.006230529595015, "grad_norm": 4.53125, "learning_rate": 4.929528920808854e-05, "loss": 0.2594, "step": 2738 }, { "epoch": 17.01246105919003, "grad_norm": 2.875, "learning_rate": 4.929470823611906e-05, "loss": 0.2087, "step": 2739 }, { "epoch": 17.018691588785046, "grad_norm": 3.265625, "learning_rate": 4.929412702819453e-05, "loss": 0.1993, "step": 2740 }, { "epoch": 17.02492211838006, "grad_norm": 3.359375, "learning_rate": 4.9293545584320564e-05, "loss": 0.3048, "step": 2741 }, { "epoch": 17.031152647975077, "grad_norm": 3.515625, "learning_rate": 4.9292963904502834e-05, "loss": 0.1861, "step": 2742 }, { "epoch": 17.037383177570092, "grad_norm": 3.390625, "learning_rate": 4.929238198874698e-05, "loss": 0.2261, "step": 2743 }, { "epoch": 17.043613707165107, "grad_norm": 3.390625, "learning_rate": 4.929179983705866e-05, "loss": 0.199, "step": 2744 }, { "epoch": 17.049844236760123, "grad_norm": 4.40625, "learning_rate": 4.9291217449443515e-05, "loss": 0.2962, "step": 2745 }, { "epoch": 17.05607476635514, "grad_norm": 1.4921875, "learning_rate": 4.929063482590722e-05, "loss": 0.1632, "step": 2746 }, { "epoch": 17.062305295950157, "grad_norm": 4.1875, "learning_rate": 4.929005196645542e-05, "loss": 0.174, "step": 2747 }, { "epoch": 17.068535825545172, "grad_norm": 6.0, "learning_rate": 4.928946887109378e-05, "loss": 0.4278, "step": 2748 }, { "epoch": 17.074766355140188, "grad_norm": 5.53125, "learning_rate": 4.928888553982797e-05, "loss": 0.3085, "step": 2749 }, { "epoch": 17.080996884735203, "grad_norm": 3.609375, "learning_rate": 4.928830197266364e-05, "loss": 0.2487, "step": 2750 }, { "epoch": 17.08722741433022, "grad_norm": 3.84375, "learning_rate": 4.9287718169606464e-05, "loss": 0.2853, "step": 2751 }, { "epoch": 17.093457943925234, "grad_norm": 2.65625, "learning_rate": 4.9287134130662114e-05, "loss": 0.2625, "step": 2752 }, { "epoch": 17.09968847352025, "grad_norm": 5.375, "learning_rate": 4.928654985583627e-05, "loss": 0.2023, "step": 2753 }, { "epoch": 17.105919003115265, "grad_norm": 3.1875, "learning_rate": 4.92859653451346e-05, "loss": 0.2105, "step": 2754 }, { "epoch": 17.11214953271028, "grad_norm": 2.234375, "learning_rate": 4.928538059856277e-05, "loss": 0.2505, "step": 2755 }, { "epoch": 17.118380062305295, "grad_norm": 4.8125, "learning_rate": 4.928479561612648e-05, "loss": 0.3854, "step": 2756 }, { "epoch": 17.12461059190031, "grad_norm": 5.46875, "learning_rate": 4.9284210397831384e-05, "loss": 0.2923, "step": 2757 }, { "epoch": 17.130841121495326, "grad_norm": 4.84375, "learning_rate": 4.9283624943683204e-05, "loss": 0.2542, "step": 2758 }, { "epoch": 17.13707165109034, "grad_norm": 4.78125, "learning_rate": 4.928303925368759e-05, "loss": 0.4462, "step": 2759 }, { "epoch": 17.143302180685357, "grad_norm": 3.09375, "learning_rate": 4.928245332785025e-05, "loss": 0.2582, "step": 2760 }, { "epoch": 17.149532710280372, "grad_norm": 3.34375, "learning_rate": 4.928186716617686e-05, "loss": 0.3571, "step": 2761 }, { "epoch": 17.15576323987539, "grad_norm": 4.09375, "learning_rate": 4.9281280768673134e-05, "loss": 0.2719, "step": 2762 }, { "epoch": 17.161993769470406, "grad_norm": 2.609375, "learning_rate": 4.928069413534475e-05, "loss": 0.2414, "step": 2763 }, { "epoch": 17.16822429906542, "grad_norm": 3.9375, "learning_rate": 4.928010726619742e-05, "loss": 0.3002, "step": 2764 }, { "epoch": 17.174454828660437, "grad_norm": 2.53125, "learning_rate": 4.9279520161236824e-05, "loss": 0.2021, "step": 2765 }, { "epoch": 17.180685358255452, "grad_norm": 4.03125, "learning_rate": 4.9278932820468674e-05, "loss": 0.3305, "step": 2766 }, { "epoch": 17.186915887850468, "grad_norm": 2.390625, "learning_rate": 4.927834524389868e-05, "loss": 0.2314, "step": 2767 }, { "epoch": 17.193146417445483, "grad_norm": 2.46875, "learning_rate": 4.927775743153254e-05, "loss": 0.2553, "step": 2768 }, { "epoch": 17.1993769470405, "grad_norm": 2.453125, "learning_rate": 4.9277169383375974e-05, "loss": 0.2552, "step": 2769 }, { "epoch": 17.205607476635514, "grad_norm": 4.125, "learning_rate": 4.9276581099434685e-05, "loss": 0.2273, "step": 2770 }, { "epoch": 17.21183800623053, "grad_norm": 3.65625, "learning_rate": 4.927599257971438e-05, "loss": 0.3656, "step": 2771 }, { "epoch": 17.218068535825545, "grad_norm": 2.875, "learning_rate": 4.927540382422079e-05, "loss": 0.219, "step": 2772 }, { "epoch": 17.22429906542056, "grad_norm": 2.75, "learning_rate": 4.9274814832959627e-05, "loss": 0.2649, "step": 2773 }, { "epoch": 17.230529595015575, "grad_norm": 2.171875, "learning_rate": 4.92742256059366e-05, "loss": 0.1836, "step": 2774 }, { "epoch": 17.23676012461059, "grad_norm": 1.7890625, "learning_rate": 4.9273636143157445e-05, "loss": 0.171, "step": 2775 }, { "epoch": 17.242990654205606, "grad_norm": 3.109375, "learning_rate": 4.927304644462789e-05, "loss": 0.2393, "step": 2776 }, { "epoch": 17.24922118380062, "grad_norm": 1.8203125, "learning_rate": 4.927245651035365e-05, "loss": 0.1812, "step": 2777 }, { "epoch": 17.25545171339564, "grad_norm": 2.78125, "learning_rate": 4.927186634034046e-05, "loss": 0.2884, "step": 2778 }, { "epoch": 17.261682242990656, "grad_norm": 3.265625, "learning_rate": 4.927127593459405e-05, "loss": 0.2654, "step": 2779 }, { "epoch": 17.26791277258567, "grad_norm": 3.609375, "learning_rate": 4.9270685293120164e-05, "loss": 0.2865, "step": 2780 }, { "epoch": 17.274143302180686, "grad_norm": 2.390625, "learning_rate": 4.927009441592453e-05, "loss": 0.201, "step": 2781 }, { "epoch": 17.2803738317757, "grad_norm": 3.984375, "learning_rate": 4.926950330301288e-05, "loss": 0.3265, "step": 2782 }, { "epoch": 17.286604361370717, "grad_norm": 2.015625, "learning_rate": 4.9268911954390974e-05, "loss": 0.2013, "step": 2783 }, { "epoch": 17.292834890965732, "grad_norm": 2.765625, "learning_rate": 4.926832037006453e-05, "loss": 0.1693, "step": 2784 }, { "epoch": 17.299065420560748, "grad_norm": 3.09375, "learning_rate": 4.9267728550039316e-05, "loss": 0.29, "step": 2785 }, { "epoch": 17.305295950155763, "grad_norm": 4.96875, "learning_rate": 4.926713649432107e-05, "loss": 0.3409, "step": 2786 }, { "epoch": 17.31152647975078, "grad_norm": 4.09375, "learning_rate": 4.9266544202915545e-05, "loss": 0.184, "step": 2787 }, { "epoch": 17.317757009345794, "grad_norm": 4.09375, "learning_rate": 4.926595167582849e-05, "loss": 0.2838, "step": 2788 }, { "epoch": 17.32398753894081, "grad_norm": 2.1875, "learning_rate": 4.926535891306566e-05, "loss": 0.2304, "step": 2789 }, { "epoch": 17.330218068535824, "grad_norm": 2.671875, "learning_rate": 4.926476591463282e-05, "loss": 0.2587, "step": 2790 }, { "epoch": 17.33644859813084, "grad_norm": 4.75, "learning_rate": 4.926417268053571e-05, "loss": 0.4267, "step": 2791 }, { "epoch": 17.342679127725855, "grad_norm": 3.453125, "learning_rate": 4.926357921078012e-05, "loss": 0.2332, "step": 2792 }, { "epoch": 17.34890965732087, "grad_norm": 2.328125, "learning_rate": 4.926298550537179e-05, "loss": 0.2214, "step": 2793 }, { "epoch": 17.35514018691589, "grad_norm": 3.640625, "learning_rate": 4.92623915643165e-05, "loss": 0.3728, "step": 2794 }, { "epoch": 17.361370716510905, "grad_norm": 3.421875, "learning_rate": 4.926179738762e-05, "loss": 0.3565, "step": 2795 }, { "epoch": 17.36760124610592, "grad_norm": 3.5, "learning_rate": 4.926120297528809e-05, "loss": 0.212, "step": 2796 }, { "epoch": 17.373831775700936, "grad_norm": 3.21875, "learning_rate": 4.926060832732652e-05, "loss": 0.2472, "step": 2797 }, { "epoch": 17.38006230529595, "grad_norm": 3.28125, "learning_rate": 4.9260013443741074e-05, "loss": 0.4059, "step": 2798 }, { "epoch": 17.386292834890966, "grad_norm": 2.40625, "learning_rate": 4.9259418324537523e-05, "loss": 0.3218, "step": 2799 }, { "epoch": 17.39252336448598, "grad_norm": 4.375, "learning_rate": 4.9258822969721655e-05, "loss": 0.3358, "step": 2800 }, { "epoch": 17.398753894080997, "grad_norm": 3.25, "learning_rate": 4.925822737929925e-05, "loss": 0.1924, "step": 2801 }, { "epoch": 17.404984423676012, "grad_norm": 3.34375, "learning_rate": 4.925763155327609e-05, "loss": 0.2721, "step": 2802 }, { "epoch": 17.411214953271028, "grad_norm": 2.421875, "learning_rate": 4.9257035491657965e-05, "loss": 0.2275, "step": 2803 }, { "epoch": 17.417445482866043, "grad_norm": 2.828125, "learning_rate": 4.925643919445067e-05, "loss": 0.1885, "step": 2804 }, { "epoch": 17.42367601246106, "grad_norm": 2.59375, "learning_rate": 4.9255842661659975e-05, "loss": 0.2534, "step": 2805 }, { "epoch": 17.429906542056074, "grad_norm": 2.90625, "learning_rate": 4.925524589329169e-05, "loss": 0.3022, "step": 2806 }, { "epoch": 17.43613707165109, "grad_norm": 3.625, "learning_rate": 4.925464888935162e-05, "loss": 0.3033, "step": 2807 }, { "epoch": 17.442367601246104, "grad_norm": 3.453125, "learning_rate": 4.925405164984553e-05, "loss": 0.4108, "step": 2808 }, { "epoch": 17.44859813084112, "grad_norm": 3.15625, "learning_rate": 4.9253454174779265e-05, "loss": 0.2479, "step": 2809 }, { "epoch": 17.45482866043614, "grad_norm": 3.921875, "learning_rate": 4.9252856464158595e-05, "loss": 0.2201, "step": 2810 }, { "epoch": 17.461059190031154, "grad_norm": 2.78125, "learning_rate": 4.925225851798934e-05, "loss": 0.2933, "step": 2811 }, { "epoch": 17.46728971962617, "grad_norm": 3.203125, "learning_rate": 4.9251660336277296e-05, "loss": 0.2286, "step": 2812 }, { "epoch": 17.473520249221185, "grad_norm": 3.984375, "learning_rate": 4.9251061919028275e-05, "loss": 0.2681, "step": 2813 }, { "epoch": 17.4797507788162, "grad_norm": 3.9375, "learning_rate": 4.92504632662481e-05, "loss": 0.2766, "step": 2814 }, { "epoch": 17.485981308411215, "grad_norm": 3.484375, "learning_rate": 4.9249864377942576e-05, "loss": 0.2323, "step": 2815 }, { "epoch": 17.49221183800623, "grad_norm": 3.390625, "learning_rate": 4.924926525411752e-05, "loss": 0.1777, "step": 2816 }, { "epoch": 17.498442367601246, "grad_norm": 3.484375, "learning_rate": 4.924866589477876e-05, "loss": 0.2737, "step": 2817 }, { "epoch": 17.50467289719626, "grad_norm": 2.765625, "learning_rate": 4.92480662999321e-05, "loss": 0.2336, "step": 2818 }, { "epoch": 17.510903426791277, "grad_norm": 2.53125, "learning_rate": 4.924746646958338e-05, "loss": 0.2695, "step": 2819 }, { "epoch": 17.517133956386292, "grad_norm": 2.90625, "learning_rate": 4.9246866403738415e-05, "loss": 0.2278, "step": 2820 }, { "epoch": 17.523364485981308, "grad_norm": 4.6875, "learning_rate": 4.924626610240304e-05, "loss": 0.2649, "step": 2821 }, { "epoch": 17.529595015576323, "grad_norm": 3.734375, "learning_rate": 4.9245665565583075e-05, "loss": 0.2589, "step": 2822 }, { "epoch": 17.53582554517134, "grad_norm": 2.640625, "learning_rate": 4.9245064793284365e-05, "loss": 0.2107, "step": 2823 }, { "epoch": 17.542056074766354, "grad_norm": 2.15625, "learning_rate": 4.924446378551274e-05, "loss": 0.2089, "step": 2824 }, { "epoch": 17.54828660436137, "grad_norm": 3.359375, "learning_rate": 4.924386254227403e-05, "loss": 0.3595, "step": 2825 }, { "epoch": 17.554517133956388, "grad_norm": 4.28125, "learning_rate": 4.924326106357409e-05, "loss": 0.2409, "step": 2826 }, { "epoch": 17.560747663551403, "grad_norm": 3.015625, "learning_rate": 4.9242659349418746e-05, "loss": 0.2148, "step": 2827 }, { "epoch": 17.56697819314642, "grad_norm": 2.484375, "learning_rate": 4.9242057399813845e-05, "loss": 0.2376, "step": 2828 }, { "epoch": 17.573208722741434, "grad_norm": 2.5, "learning_rate": 4.924145521476525e-05, "loss": 0.1874, "step": 2829 }, { "epoch": 17.57943925233645, "grad_norm": 5.25, "learning_rate": 4.924085279427879e-05, "loss": 0.2422, "step": 2830 }, { "epoch": 17.585669781931465, "grad_norm": 5.125, "learning_rate": 4.924025013836031e-05, "loss": 0.3137, "step": 2831 }, { "epoch": 17.59190031152648, "grad_norm": 4.15625, "learning_rate": 4.923964724701568e-05, "loss": 0.3996, "step": 2832 }, { "epoch": 17.598130841121495, "grad_norm": 4.5, "learning_rate": 4.923904412025077e-05, "loss": 0.3911, "step": 2833 }, { "epoch": 17.60436137071651, "grad_norm": 4.875, "learning_rate": 4.92384407580714e-05, "loss": 0.2267, "step": 2834 }, { "epoch": 17.610591900311526, "grad_norm": 4.4375, "learning_rate": 4.923783716048345e-05, "loss": 0.2573, "step": 2835 }, { "epoch": 17.61682242990654, "grad_norm": 4.375, "learning_rate": 4.923723332749278e-05, "loss": 0.2836, "step": 2836 }, { "epoch": 17.623052959501557, "grad_norm": 2.375, "learning_rate": 4.923662925910526e-05, "loss": 0.1793, "step": 2837 }, { "epoch": 17.629283489096572, "grad_norm": 3.5625, "learning_rate": 4.923602495532675e-05, "loss": 0.2995, "step": 2838 }, { "epoch": 17.635514018691588, "grad_norm": 3.234375, "learning_rate": 4.923542041616312e-05, "loss": 0.3845, "step": 2839 }, { "epoch": 17.641744548286603, "grad_norm": 2.90625, "learning_rate": 4.923481564162025e-05, "loss": 0.2516, "step": 2840 }, { "epoch": 17.64797507788162, "grad_norm": 3.015625, "learning_rate": 4.923421063170399e-05, "loss": 0.296, "step": 2841 }, { "epoch": 17.654205607476637, "grad_norm": 2.96875, "learning_rate": 4.923360538642025e-05, "loss": 0.2057, "step": 2842 }, { "epoch": 17.660436137071652, "grad_norm": 2.5, "learning_rate": 4.923299990577488e-05, "loss": 0.2092, "step": 2843 }, { "epoch": 17.666666666666668, "grad_norm": 2.953125, "learning_rate": 4.9232394189773764e-05, "loss": 0.3477, "step": 2844 }, { "epoch": 17.672897196261683, "grad_norm": 2.765625, "learning_rate": 4.9231788238422806e-05, "loss": 0.262, "step": 2845 }, { "epoch": 17.6791277258567, "grad_norm": 2.34375, "learning_rate": 4.923118205172786e-05, "loss": 0.1704, "step": 2846 }, { "epoch": 17.685358255451714, "grad_norm": 3.484375, "learning_rate": 4.9230575629694844e-05, "loss": 0.4988, "step": 2847 }, { "epoch": 17.69158878504673, "grad_norm": 3.421875, "learning_rate": 4.922996897232963e-05, "loss": 0.2841, "step": 2848 }, { "epoch": 17.697819314641745, "grad_norm": 1.671875, "learning_rate": 4.922936207963812e-05, "loss": 0.1717, "step": 2849 }, { "epoch": 17.70404984423676, "grad_norm": 2.4375, "learning_rate": 4.9228754951626195e-05, "loss": 0.2229, "step": 2850 }, { "epoch": 17.710280373831775, "grad_norm": 3.015625, "learning_rate": 4.922814758829975e-05, "loss": 0.236, "step": 2851 }, { "epoch": 17.71651090342679, "grad_norm": 4.1875, "learning_rate": 4.9227539989664706e-05, "loss": 0.4678, "step": 2852 }, { "epoch": 17.722741433021806, "grad_norm": 3.296875, "learning_rate": 4.922693215572695e-05, "loss": 0.2051, "step": 2853 }, { "epoch": 17.72897196261682, "grad_norm": 2.859375, "learning_rate": 4.922632408649238e-05, "loss": 0.3035, "step": 2854 }, { "epoch": 17.735202492211837, "grad_norm": 2.796875, "learning_rate": 4.922571578196692e-05, "loss": 0.2543, "step": 2855 }, { "epoch": 17.741433021806852, "grad_norm": 1.7578125, "learning_rate": 4.922510724215646e-05, "loss": 0.1922, "step": 2856 }, { "epoch": 17.747663551401867, "grad_norm": 2.703125, "learning_rate": 4.922449846706691e-05, "loss": 0.3394, "step": 2857 }, { "epoch": 17.753894080996886, "grad_norm": 1.984375, "learning_rate": 4.9223889456704185e-05, "loss": 0.1883, "step": 2858 }, { "epoch": 17.7601246105919, "grad_norm": 2.984375, "learning_rate": 4.9223280211074206e-05, "loss": 0.2075, "step": 2859 }, { "epoch": 17.766355140186917, "grad_norm": 3.109375, "learning_rate": 4.92226707301829e-05, "loss": 0.2581, "step": 2860 }, { "epoch": 17.772585669781932, "grad_norm": 2.203125, "learning_rate": 4.922206101403616e-05, "loss": 0.1965, "step": 2861 }, { "epoch": 17.778816199376948, "grad_norm": 3.265625, "learning_rate": 4.9221451062639925e-05, "loss": 0.4263, "step": 2862 }, { "epoch": 17.785046728971963, "grad_norm": 2.9375, "learning_rate": 4.9220840876000116e-05, "loss": 0.3165, "step": 2863 }, { "epoch": 17.79127725856698, "grad_norm": 2.734375, "learning_rate": 4.922023045412266e-05, "loss": 0.2207, "step": 2864 }, { "epoch": 17.797507788161994, "grad_norm": 4.71875, "learning_rate": 4.921961979701348e-05, "loss": 0.2642, "step": 2865 }, { "epoch": 17.80373831775701, "grad_norm": 3.109375, "learning_rate": 4.921900890467852e-05, "loss": 0.201, "step": 2866 }, { "epoch": 17.809968847352025, "grad_norm": 2.5625, "learning_rate": 4.92183977771237e-05, "loss": 0.1998, "step": 2867 }, { "epoch": 17.81619937694704, "grad_norm": 3.046875, "learning_rate": 4.921778641435496e-05, "loss": 0.3943, "step": 2868 }, { "epoch": 17.822429906542055, "grad_norm": 5.40625, "learning_rate": 4.921717481637823e-05, "loss": 0.2095, "step": 2869 }, { "epoch": 17.82866043613707, "grad_norm": 4.46875, "learning_rate": 4.9216562983199465e-05, "loss": 0.2598, "step": 2870 }, { "epoch": 17.834890965732086, "grad_norm": 4.125, "learning_rate": 4.921595091482461e-05, "loss": 0.34, "step": 2871 }, { "epoch": 17.8411214953271, "grad_norm": 2.765625, "learning_rate": 4.921533861125958e-05, "loss": 0.2603, "step": 2872 }, { "epoch": 17.847352024922117, "grad_norm": 3.84375, "learning_rate": 4.921472607251034e-05, "loss": 0.2473, "step": 2873 }, { "epoch": 17.853582554517136, "grad_norm": 5.09375, "learning_rate": 4.921411329858285e-05, "loss": 0.3327, "step": 2874 }, { "epoch": 17.85981308411215, "grad_norm": 3.046875, "learning_rate": 4.921350028948305e-05, "loss": 0.2884, "step": 2875 }, { "epoch": 17.866043613707166, "grad_norm": 2.859375, "learning_rate": 4.921288704521689e-05, "loss": 0.3332, "step": 2876 }, { "epoch": 17.87227414330218, "grad_norm": 4.125, "learning_rate": 4.9212273565790344e-05, "loss": 0.3603, "step": 2877 }, { "epoch": 17.878504672897197, "grad_norm": 4.8125, "learning_rate": 4.9211659851209344e-05, "loss": 0.4616, "step": 2878 }, { "epoch": 17.884735202492212, "grad_norm": 2.75, "learning_rate": 4.921104590147987e-05, "loss": 0.2946, "step": 2879 }, { "epoch": 17.890965732087228, "grad_norm": 2.9375, "learning_rate": 4.9210431716607875e-05, "loss": 0.223, "step": 2880 }, { "epoch": 17.897196261682243, "grad_norm": 3.25, "learning_rate": 4.920981729659933e-05, "loss": 0.3195, "step": 2881 }, { "epoch": 17.90342679127726, "grad_norm": 3.5625, "learning_rate": 4.92092026414602e-05, "loss": 0.1549, "step": 2882 }, { "epoch": 17.909657320872274, "grad_norm": 3.125, "learning_rate": 4.9208587751196454e-05, "loss": 0.2079, "step": 2883 }, { "epoch": 17.91588785046729, "grad_norm": 2.859375, "learning_rate": 4.920797262581406e-05, "loss": 0.321, "step": 2884 }, { "epoch": 17.922118380062305, "grad_norm": 2.125, "learning_rate": 4.9207357265319005e-05, "loss": 0.1685, "step": 2885 }, { "epoch": 17.92834890965732, "grad_norm": 5.21875, "learning_rate": 4.920674166971725e-05, "loss": 0.1873, "step": 2886 }, { "epoch": 17.934579439252335, "grad_norm": 4.90625, "learning_rate": 4.920612583901479e-05, "loss": 0.2968, "step": 2887 }, { "epoch": 17.94080996884735, "grad_norm": 4.0, "learning_rate": 4.920550977321759e-05, "loss": 0.2625, "step": 2888 }, { "epoch": 17.947040498442366, "grad_norm": 2.140625, "learning_rate": 4.920489347233164e-05, "loss": 0.2059, "step": 2889 }, { "epoch": 17.953271028037385, "grad_norm": 2.421875, "learning_rate": 4.9204276936362926e-05, "loss": 0.2128, "step": 2890 }, { "epoch": 17.9595015576324, "grad_norm": 3.546875, "learning_rate": 4.920366016531744e-05, "loss": 0.1785, "step": 2891 }, { "epoch": 17.965732087227416, "grad_norm": 5.0625, "learning_rate": 4.9203043159201164e-05, "loss": 0.4622, "step": 2892 }, { "epoch": 17.97196261682243, "grad_norm": 3.0625, "learning_rate": 4.92024259180201e-05, "loss": 0.2589, "step": 2893 }, { "epoch": 17.978193146417446, "grad_norm": 3.640625, "learning_rate": 4.920180844178024e-05, "loss": 0.369, "step": 2894 }, { "epoch": 17.98442367601246, "grad_norm": 2.84375, "learning_rate": 4.920119073048757e-05, "loss": 0.1833, "step": 2895 }, { "epoch": 17.990654205607477, "grad_norm": 4.0625, "learning_rate": 4.9200572784148104e-05, "loss": 0.409, "step": 2896 }, { "epoch": 17.996884735202492, "grad_norm": 1.796875, "learning_rate": 4.9199954602767826e-05, "loss": 0.1594, "step": 2897 }, { "epoch": 18.0, "grad_norm": 2.125, "learning_rate": 4.9199336186352764e-05, "loss": 0.2071, "step": 2898 }, { "epoch": 18.006230529595015, "grad_norm": 3.65625, "learning_rate": 4.919871753490891e-05, "loss": 0.3127, "step": 2899 }, { "epoch": 18.01246105919003, "grad_norm": 3.453125, "learning_rate": 4.919809864844227e-05, "loss": 0.2632, "step": 2900 }, { "epoch": 18.018691588785046, "grad_norm": 2.9375, "learning_rate": 4.919747952695886e-05, "loss": 0.1978, "step": 2901 }, { "epoch": 18.02492211838006, "grad_norm": 2.421875, "learning_rate": 4.91968601704647e-05, "loss": 0.1649, "step": 2902 }, { "epoch": 18.031152647975077, "grad_norm": 4.6875, "learning_rate": 4.919624057896579e-05, "loss": 0.2198, "step": 2903 }, { "epoch": 18.037383177570092, "grad_norm": 5.15625, "learning_rate": 4.919562075246815e-05, "loss": 0.3934, "step": 2904 }, { "epoch": 18.043613707165107, "grad_norm": 3.34375, "learning_rate": 4.919500069097781e-05, "loss": 0.2859, "step": 2905 }, { "epoch": 18.049844236760123, "grad_norm": 2.875, "learning_rate": 4.919438039450078e-05, "loss": 0.3524, "step": 2906 }, { "epoch": 18.05607476635514, "grad_norm": 4.0, "learning_rate": 4.91937598630431e-05, "loss": 0.3872, "step": 2907 }, { "epoch": 18.062305295950157, "grad_norm": 3.078125, "learning_rate": 4.9193139096610784e-05, "loss": 0.2265, "step": 2908 }, { "epoch": 18.068535825545172, "grad_norm": 2.71875, "learning_rate": 4.9192518095209874e-05, "loss": 0.268, "step": 2909 }, { "epoch": 18.074766355140188, "grad_norm": 2.03125, "learning_rate": 4.9191896858846375e-05, "loss": 0.1643, "step": 2910 }, { "epoch": 18.080996884735203, "grad_norm": 2.375, "learning_rate": 4.9191275387526345e-05, "loss": 0.3074, "step": 2911 }, { "epoch": 18.08722741433022, "grad_norm": 2.6875, "learning_rate": 4.919065368125582e-05, "loss": 0.1921, "step": 2912 }, { "epoch": 18.093457943925234, "grad_norm": 2.890625, "learning_rate": 4.919003174004083e-05, "loss": 0.2554, "step": 2913 }, { "epoch": 18.09968847352025, "grad_norm": 2.421875, "learning_rate": 4.918940956388741e-05, "loss": 0.2085, "step": 2914 }, { "epoch": 18.105919003115265, "grad_norm": 2.734375, "learning_rate": 4.918878715280161e-05, "loss": 0.2107, "step": 2915 }, { "epoch": 18.11214953271028, "grad_norm": 2.390625, "learning_rate": 4.9188164506789475e-05, "loss": 0.2349, "step": 2916 }, { "epoch": 18.118380062305295, "grad_norm": 2.578125, "learning_rate": 4.918754162585705e-05, "loss": 0.2781, "step": 2917 }, { "epoch": 18.12461059190031, "grad_norm": 3.046875, "learning_rate": 4.918691851001039e-05, "loss": 0.2659, "step": 2918 }, { "epoch": 18.130841121495326, "grad_norm": 2.71875, "learning_rate": 4.9186295159255537e-05, "loss": 0.3332, "step": 2919 }, { "epoch": 18.13707165109034, "grad_norm": 3.8125, "learning_rate": 4.9185671573598555e-05, "loss": 0.3785, "step": 2920 }, { "epoch": 18.143302180685357, "grad_norm": 2.921875, "learning_rate": 4.9185047753045486e-05, "loss": 0.2174, "step": 2921 }, { "epoch": 18.149532710280372, "grad_norm": 4.5625, "learning_rate": 4.918442369760241e-05, "loss": 0.362, "step": 2922 }, { "epoch": 18.15576323987539, "grad_norm": 3.421875, "learning_rate": 4.9183799407275375e-05, "loss": 0.271, "step": 2923 }, { "epoch": 18.161993769470406, "grad_norm": 3.25, "learning_rate": 4.918317488207045e-05, "loss": 0.3347, "step": 2924 }, { "epoch": 18.16822429906542, "grad_norm": 4.25, "learning_rate": 4.9182550121993685e-05, "loss": 0.3114, "step": 2925 }, { "epoch": 18.174454828660437, "grad_norm": 4.40625, "learning_rate": 4.918192512705117e-05, "loss": 0.2701, "step": 2926 }, { "epoch": 18.180685358255452, "grad_norm": 2.609375, "learning_rate": 4.918129989724896e-05, "loss": 0.2887, "step": 2927 }, { "epoch": 18.186915887850468, "grad_norm": 3.234375, "learning_rate": 4.918067443259313e-05, "loss": 0.1823, "step": 2928 }, { "epoch": 18.193146417445483, "grad_norm": 2.484375, "learning_rate": 4.9180048733089755e-05, "loss": 0.2061, "step": 2929 }, { "epoch": 18.1993769470405, "grad_norm": 2.703125, "learning_rate": 4.917942279874491e-05, "loss": 0.1884, "step": 2930 }, { "epoch": 18.205607476635514, "grad_norm": 2.65625, "learning_rate": 4.917879662956469e-05, "loss": 0.2231, "step": 2931 }, { "epoch": 18.21183800623053, "grad_norm": 3.0, "learning_rate": 4.917817022555515e-05, "loss": 0.2475, "step": 2932 }, { "epoch": 18.218068535825545, "grad_norm": 1.9609375, "learning_rate": 4.9177543586722395e-05, "loss": 0.2122, "step": 2933 }, { "epoch": 18.22429906542056, "grad_norm": 1.8828125, "learning_rate": 4.917691671307251e-05, "loss": 0.1879, "step": 2934 }, { "epoch": 18.230529595015575, "grad_norm": 2.5625, "learning_rate": 4.917628960461157e-05, "loss": 0.4146, "step": 2935 }, { "epoch": 18.23676012461059, "grad_norm": 2.71875, "learning_rate": 4.917566226134567e-05, "loss": 0.2922, "step": 2936 }, { "epoch": 18.242990654205606, "grad_norm": 2.484375, "learning_rate": 4.917503468328091e-05, "loss": 0.1831, "step": 2937 }, { "epoch": 18.24922118380062, "grad_norm": 5.0, "learning_rate": 4.917440687042338e-05, "loss": 0.3492, "step": 2938 }, { "epoch": 18.25545171339564, "grad_norm": 4.0, "learning_rate": 4.917377882277917e-05, "loss": 0.1795, "step": 2939 }, { "epoch": 18.261682242990656, "grad_norm": 3.796875, "learning_rate": 4.91731505403544e-05, "loss": 0.377, "step": 2940 }, { "epoch": 18.26791277258567, "grad_norm": 2.015625, "learning_rate": 4.9172522023155154e-05, "loss": 0.2424, "step": 2941 }, { "epoch": 18.274143302180686, "grad_norm": 3.234375, "learning_rate": 4.917189327118754e-05, "loss": 0.2713, "step": 2942 }, { "epoch": 18.2803738317757, "grad_norm": 2.984375, "learning_rate": 4.9171264284457675e-05, "loss": 0.3027, "step": 2943 }, { "epoch": 18.286604361370717, "grad_norm": 3.03125, "learning_rate": 4.917063506297165e-05, "loss": 0.2797, "step": 2944 }, { "epoch": 18.292834890965732, "grad_norm": 2.3125, "learning_rate": 4.9170005606735594e-05, "loss": 0.1817, "step": 2945 }, { "epoch": 18.299065420560748, "grad_norm": 3.90625, "learning_rate": 4.9169375915755606e-05, "loss": 0.4777, "step": 2946 }, { "epoch": 18.305295950155763, "grad_norm": 2.5625, "learning_rate": 4.916874599003781e-05, "loss": 0.2283, "step": 2947 }, { "epoch": 18.31152647975078, "grad_norm": 3.1875, "learning_rate": 4.9168115829588326e-05, "loss": 0.3718, "step": 2948 }, { "epoch": 18.317757009345794, "grad_norm": 2.5625, "learning_rate": 4.916748543441326e-05, "loss": 0.191, "step": 2949 }, { "epoch": 18.32398753894081, "grad_norm": 3.078125, "learning_rate": 4.916685480451875e-05, "loss": 0.3864, "step": 2950 }, { "epoch": 18.330218068535824, "grad_norm": 2.609375, "learning_rate": 4.9166223939910916e-05, "loss": 0.2551, "step": 2951 }, { "epoch": 18.33644859813084, "grad_norm": 1.8046875, "learning_rate": 4.9165592840595884e-05, "loss": 0.1738, "step": 2952 }, { "epoch": 18.342679127725855, "grad_norm": 3.140625, "learning_rate": 4.916496150657978e-05, "loss": 0.1974, "step": 2953 }, { "epoch": 18.34890965732087, "grad_norm": 2.421875, "learning_rate": 4.916432993786874e-05, "loss": 0.2419, "step": 2954 }, { "epoch": 18.35514018691589, "grad_norm": 4.0625, "learning_rate": 4.9163698134468906e-05, "loss": 0.4168, "step": 2955 }, { "epoch": 18.361370716510905, "grad_norm": 1.9375, "learning_rate": 4.91630660963864e-05, "loss": 0.1581, "step": 2956 }, { "epoch": 18.36760124610592, "grad_norm": 3.640625, "learning_rate": 4.9162433823627366e-05, "loss": 0.4956, "step": 2957 }, { "epoch": 18.373831775700936, "grad_norm": 2.90625, "learning_rate": 4.9161801316197945e-05, "loss": 0.3581, "step": 2958 }, { "epoch": 18.38006230529595, "grad_norm": 3.1875, "learning_rate": 4.916116857410427e-05, "loss": 0.2187, "step": 2959 }, { "epoch": 18.386292834890966, "grad_norm": 4.1875, "learning_rate": 4.916053559735251e-05, "loss": 0.3659, "step": 2960 }, { "epoch": 18.39252336448598, "grad_norm": 3.53125, "learning_rate": 4.915990238594879e-05, "loss": 0.3584, "step": 2961 }, { "epoch": 18.398753894080997, "grad_norm": 2.796875, "learning_rate": 4.915926893989927e-05, "loss": 0.4133, "step": 2962 }, { "epoch": 18.404984423676012, "grad_norm": 4.59375, "learning_rate": 4.91586352592101e-05, "loss": 0.2957, "step": 2963 }, { "epoch": 18.411214953271028, "grad_norm": 2.65625, "learning_rate": 4.915800134388744e-05, "loss": 0.1869, "step": 2964 }, { "epoch": 18.417445482866043, "grad_norm": 3.75, "learning_rate": 4.9157367193937444e-05, "loss": 0.2517, "step": 2965 }, { "epoch": 18.42367601246106, "grad_norm": 3.1875, "learning_rate": 4.9156732809366264e-05, "loss": 0.1954, "step": 2966 }, { "epoch": 18.429906542056074, "grad_norm": 1.9375, "learning_rate": 4.915609819018007e-05, "loss": 0.1807, "step": 2967 }, { "epoch": 18.43613707165109, "grad_norm": 2.3125, "learning_rate": 4.915546333638501e-05, "loss": 0.2569, "step": 2968 }, { "epoch": 18.442367601246104, "grad_norm": 4.125, "learning_rate": 4.9154828247987275e-05, "loss": 0.3599, "step": 2969 }, { "epoch": 18.44859813084112, "grad_norm": 1.9453125, "learning_rate": 4.9154192924993015e-05, "loss": 0.1647, "step": 2970 }, { "epoch": 18.45482866043614, "grad_norm": 2.265625, "learning_rate": 4.915355736740841e-05, "loss": 0.3072, "step": 2971 }, { "epoch": 18.461059190031154, "grad_norm": 2.53125, "learning_rate": 4.9152921575239616e-05, "loss": 0.1734, "step": 2972 }, { "epoch": 18.46728971962617, "grad_norm": 3.078125, "learning_rate": 4.915228554849283e-05, "loss": 0.2458, "step": 2973 }, { "epoch": 18.473520249221185, "grad_norm": 4.3125, "learning_rate": 4.915164928717421e-05, "loss": 0.263, "step": 2974 }, { "epoch": 18.4797507788162, "grad_norm": 2.546875, "learning_rate": 4.915101279128995e-05, "loss": 0.2047, "step": 2975 }, { "epoch": 18.485981308411215, "grad_norm": 3.0625, "learning_rate": 4.915037606084622e-05, "loss": 0.2857, "step": 2976 }, { "epoch": 18.49221183800623, "grad_norm": 3.28125, "learning_rate": 4.914973909584922e-05, "loss": 0.2937, "step": 2977 }, { "epoch": 18.498442367601246, "grad_norm": 4.90625, "learning_rate": 4.9149101896305114e-05, "loss": 0.3266, "step": 2978 }, { "epoch": 18.50467289719626, "grad_norm": 2.546875, "learning_rate": 4.914846446222012e-05, "loss": 0.3072, "step": 2979 }, { "epoch": 18.510903426791277, "grad_norm": 3.1875, "learning_rate": 4.9147826793600396e-05, "loss": 0.2843, "step": 2980 }, { "epoch": 18.517133956386292, "grad_norm": 2.828125, "learning_rate": 4.9147188890452164e-05, "loss": 0.2027, "step": 2981 }, { "epoch": 18.523364485981308, "grad_norm": 3.78125, "learning_rate": 4.91465507527816e-05, "loss": 0.2264, "step": 2982 }, { "epoch": 18.529595015576323, "grad_norm": 1.9765625, "learning_rate": 4.914591238059491e-05, "loss": 0.1737, "step": 2983 }, { "epoch": 18.53582554517134, "grad_norm": 3.421875, "learning_rate": 4.9145273773898284e-05, "loss": 0.2718, "step": 2984 }, { "epoch": 18.542056074766354, "grad_norm": 1.9140625, "learning_rate": 4.914463493269794e-05, "loss": 0.1724, "step": 2985 }, { "epoch": 18.54828660436137, "grad_norm": 3.03125, "learning_rate": 4.9143995857000074e-05, "loss": 0.248, "step": 2986 }, { "epoch": 18.554517133956388, "grad_norm": 3.40625, "learning_rate": 4.9143356546810895e-05, "loss": 0.2368, "step": 2987 }, { "epoch": 18.560747663551403, "grad_norm": 3.046875, "learning_rate": 4.914271700213661e-05, "loss": 0.2802, "step": 2988 }, { "epoch": 18.56697819314642, "grad_norm": 2.890625, "learning_rate": 4.914207722298343e-05, "loss": 0.2362, "step": 2989 }, { "epoch": 18.573208722741434, "grad_norm": 2.359375, "learning_rate": 4.9141437209357574e-05, "loss": 0.2001, "step": 2990 }, { "epoch": 18.57943925233645, "grad_norm": 2.625, "learning_rate": 4.914079696126526e-05, "loss": 0.2751, "step": 2991 }, { "epoch": 18.585669781931465, "grad_norm": 3.046875, "learning_rate": 4.9140156478712685e-05, "loss": 0.173, "step": 2992 }, { "epoch": 18.59190031152648, "grad_norm": 2.96875, "learning_rate": 4.913951576170609e-05, "loss": 0.2186, "step": 2993 }, { "epoch": 18.598130841121495, "grad_norm": 5.1875, "learning_rate": 4.91388748102517e-05, "loss": 0.4864, "step": 2994 }, { "epoch": 18.60436137071651, "grad_norm": 2.359375, "learning_rate": 4.9138233624355724e-05, "loss": 0.227, "step": 2995 }, { "epoch": 18.610591900311526, "grad_norm": 3.234375, "learning_rate": 4.91375922040244e-05, "loss": 0.3451, "step": 2996 }, { "epoch": 18.61682242990654, "grad_norm": 3.875, "learning_rate": 4.913695054926396e-05, "loss": 0.3827, "step": 2997 }, { "epoch": 18.623052959501557, "grad_norm": 3.25, "learning_rate": 4.913630866008063e-05, "loss": 0.2321, "step": 2998 }, { "epoch": 18.629283489096572, "grad_norm": 2.953125, "learning_rate": 4.913566653648064e-05, "loss": 0.287, "step": 2999 }, { "epoch": 18.635514018691588, "grad_norm": 3.390625, "learning_rate": 4.913502417847024e-05, "loss": 0.403, "step": 3000 }, { "epoch": 18.641744548286603, "grad_norm": 2.609375, "learning_rate": 4.9134381586055655e-05, "loss": 0.165, "step": 3001 }, { "epoch": 18.64797507788162, "grad_norm": 3.96875, "learning_rate": 4.913373875924313e-05, "loss": 0.2816, "step": 3002 }, { "epoch": 18.654205607476637, "grad_norm": 2.625, "learning_rate": 4.913309569803891e-05, "loss": 0.2639, "step": 3003 }, { "epoch": 18.660436137071652, "grad_norm": 4.5, "learning_rate": 4.9132452402449246e-05, "loss": 0.5255, "step": 3004 }, { "epoch": 18.666666666666668, "grad_norm": 3.59375, "learning_rate": 4.9131808872480376e-05, "loss": 0.2493, "step": 3005 }, { "epoch": 18.672897196261683, "grad_norm": 3.28125, "learning_rate": 4.9131165108138554e-05, "loss": 0.2822, "step": 3006 }, { "epoch": 18.6791277258567, "grad_norm": 4.0625, "learning_rate": 4.913052110943004e-05, "loss": 0.296, "step": 3007 }, { "epoch": 18.685358255451714, "grad_norm": 3.796875, "learning_rate": 4.9129876876361076e-05, "loss": 0.5536, "step": 3008 }, { "epoch": 18.69158878504673, "grad_norm": 2.765625, "learning_rate": 4.9129232408937917e-05, "loss": 0.2331, "step": 3009 }, { "epoch": 18.697819314641745, "grad_norm": 3.671875, "learning_rate": 4.912858770716684e-05, "loss": 0.2556, "step": 3010 }, { "epoch": 18.70404984423676, "grad_norm": 2.953125, "learning_rate": 4.9127942771054095e-05, "loss": 0.2954, "step": 3011 }, { "epoch": 18.710280373831775, "grad_norm": 2.125, "learning_rate": 4.912729760060595e-05, "loss": 0.1803, "step": 3012 }, { "epoch": 18.71651090342679, "grad_norm": 3.1875, "learning_rate": 4.912665219582866e-05, "loss": 0.2831, "step": 3013 }, { "epoch": 18.722741433021806, "grad_norm": 4.3125, "learning_rate": 4.91260065567285e-05, "loss": 0.2552, "step": 3014 }, { "epoch": 18.72897196261682, "grad_norm": 2.953125, "learning_rate": 4.9125360683311744e-05, "loss": 0.1543, "step": 3015 }, { "epoch": 18.735202492211837, "grad_norm": 4.5, "learning_rate": 4.9124714575584664e-05, "loss": 0.495, "step": 3016 }, { "epoch": 18.741433021806852, "grad_norm": 3.59375, "learning_rate": 4.912406823355353e-05, "loss": 0.3161, "step": 3017 }, { "epoch": 18.747663551401867, "grad_norm": 3.546875, "learning_rate": 4.9123421657224636e-05, "loss": 0.1965, "step": 3018 }, { "epoch": 18.753894080996886, "grad_norm": 2.578125, "learning_rate": 4.912277484660424e-05, "loss": 0.2182, "step": 3019 }, { "epoch": 18.7601246105919, "grad_norm": 2.65625, "learning_rate": 4.912212780169863e-05, "loss": 0.2618, "step": 3020 }, { "epoch": 18.766355140186917, "grad_norm": 3.09375, "learning_rate": 4.9121480522514085e-05, "loss": 0.3474, "step": 3021 }, { "epoch": 18.772585669781932, "grad_norm": 2.734375, "learning_rate": 4.912083300905691e-05, "loss": 0.2822, "step": 3022 }, { "epoch": 18.778816199376948, "grad_norm": 2.46875, "learning_rate": 4.9120185261333385e-05, "loss": 0.2252, "step": 3023 }, { "epoch": 18.785046728971963, "grad_norm": 4.65625, "learning_rate": 4.911953727934979e-05, "loss": 0.2888, "step": 3024 }, { "epoch": 18.79127725856698, "grad_norm": 2.953125, "learning_rate": 4.9118889063112437e-05, "loss": 0.2881, "step": 3025 }, { "epoch": 18.797507788161994, "grad_norm": 3.0625, "learning_rate": 4.911824061262761e-05, "loss": 0.2903, "step": 3026 }, { "epoch": 18.80373831775701, "grad_norm": 2.46875, "learning_rate": 4.911759192790161e-05, "loss": 0.1872, "step": 3027 }, { "epoch": 18.809968847352025, "grad_norm": 3.734375, "learning_rate": 4.911694300894073e-05, "loss": 0.3924, "step": 3028 }, { "epoch": 18.81619937694704, "grad_norm": 2.828125, "learning_rate": 4.9116293855751285e-05, "loss": 0.2005, "step": 3029 }, { "epoch": 18.822429906542055, "grad_norm": 4.21875, "learning_rate": 4.9115644468339575e-05, "loss": 0.3651, "step": 3030 }, { "epoch": 18.82866043613707, "grad_norm": 2.90625, "learning_rate": 4.91149948467119e-05, "loss": 0.3127, "step": 3031 }, { "epoch": 18.834890965732086, "grad_norm": 2.359375, "learning_rate": 4.911434499087457e-05, "loss": 0.1762, "step": 3032 }, { "epoch": 18.8411214953271, "grad_norm": 5.28125, "learning_rate": 4.9113694900833906e-05, "loss": 0.3746, "step": 3033 }, { "epoch": 18.847352024922117, "grad_norm": 5.96875, "learning_rate": 4.911304457659622e-05, "loss": 0.3736, "step": 3034 }, { "epoch": 18.853582554517136, "grad_norm": 3.21875, "learning_rate": 4.9112394018167825e-05, "loss": 0.2458, "step": 3035 }, { "epoch": 18.85981308411215, "grad_norm": 1.7421875, "learning_rate": 4.911174322555504e-05, "loss": 0.1357, "step": 3036 }, { "epoch": 18.866043613707166, "grad_norm": 3.625, "learning_rate": 4.911109219876417e-05, "loss": 0.3464, "step": 3037 }, { "epoch": 18.87227414330218, "grad_norm": 3.328125, "learning_rate": 4.911044093780157e-05, "loss": 0.1841, "step": 3038 }, { "epoch": 18.878504672897197, "grad_norm": 2.734375, "learning_rate": 4.9109789442673545e-05, "loss": 0.217, "step": 3039 }, { "epoch": 18.884735202492212, "grad_norm": 3.390625, "learning_rate": 4.9109137713386424e-05, "loss": 0.3025, "step": 3040 }, { "epoch": 18.890965732087228, "grad_norm": 2.734375, "learning_rate": 4.9108485749946544e-05, "loss": 0.2448, "step": 3041 }, { "epoch": 18.897196261682243, "grad_norm": 2.390625, "learning_rate": 4.9107833552360215e-05, "loss": 0.2209, "step": 3042 }, { "epoch": 18.90342679127726, "grad_norm": 2.46875, "learning_rate": 4.910718112063381e-05, "loss": 0.251, "step": 3043 }, { "epoch": 18.909657320872274, "grad_norm": 2.71875, "learning_rate": 4.910652845477363e-05, "loss": 0.3247, "step": 3044 }, { "epoch": 18.91588785046729, "grad_norm": 1.6953125, "learning_rate": 4.910587555478602e-05, "loss": 0.1416, "step": 3045 }, { "epoch": 18.922118380062305, "grad_norm": 3.359375, "learning_rate": 4.910522242067733e-05, "loss": 0.3163, "step": 3046 }, { "epoch": 18.92834890965732, "grad_norm": 3.84375, "learning_rate": 4.910456905245391e-05, "loss": 0.4872, "step": 3047 }, { "epoch": 18.934579439252335, "grad_norm": 3.5, "learning_rate": 4.9103915450122094e-05, "loss": 0.2888, "step": 3048 }, { "epoch": 18.94080996884735, "grad_norm": 3.84375, "learning_rate": 4.910326161368823e-05, "loss": 0.3063, "step": 3049 }, { "epoch": 18.947040498442366, "grad_norm": 5.0, "learning_rate": 4.910260754315868e-05, "loss": 0.2125, "step": 3050 }, { "epoch": 18.953271028037385, "grad_norm": 5.1875, "learning_rate": 4.9101953238539775e-05, "loss": 0.5048, "step": 3051 }, { "epoch": 18.9595015576324, "grad_norm": 3.75, "learning_rate": 4.910129869983789e-05, "loss": 0.2099, "step": 3052 }, { "epoch": 18.965732087227416, "grad_norm": 3.203125, "learning_rate": 4.910064392705937e-05, "loss": 0.2839, "step": 3053 }, { "epoch": 18.97196261682243, "grad_norm": 3.765625, "learning_rate": 4.9099988920210585e-05, "loss": 0.2307, "step": 3054 }, { "epoch": 18.978193146417446, "grad_norm": 4.25, "learning_rate": 4.909933367929789e-05, "loss": 0.362, "step": 3055 }, { "epoch": 18.98442367601246, "grad_norm": 4.40625, "learning_rate": 4.909867820432764e-05, "loss": 0.4904, "step": 3056 }, { "epoch": 18.990654205607477, "grad_norm": 3.59375, "learning_rate": 4.909802249530622e-05, "loss": 0.2392, "step": 3057 }, { "epoch": 18.996884735202492, "grad_norm": 4.21875, "learning_rate": 4.909736655223998e-05, "loss": 0.2383, "step": 3058 }, { "epoch": 19.0, "grad_norm": 3.125, "learning_rate": 4.90967103751353e-05, "loss": 0.2975, "step": 3059 }, { "epoch": 19.006230529595015, "grad_norm": 5.125, "learning_rate": 4.909605396399856e-05, "loss": 0.2667, "step": 3060 }, { "epoch": 19.01246105919003, "grad_norm": 2.9375, "learning_rate": 4.909539731883611e-05, "loss": 0.2142, "step": 3061 }, { "epoch": 19.018691588785046, "grad_norm": 2.71875, "learning_rate": 4.909474043965436e-05, "loss": 0.173, "step": 3062 }, { "epoch": 19.02492211838006, "grad_norm": 4.0625, "learning_rate": 4.909408332645967e-05, "loss": 0.3063, "step": 3063 }, { "epoch": 19.031152647975077, "grad_norm": 2.96875, "learning_rate": 4.909342597925843e-05, "loss": 0.1918, "step": 3064 }, { "epoch": 19.037383177570092, "grad_norm": 2.4375, "learning_rate": 4.909276839805702e-05, "loss": 0.2397, "step": 3065 }, { "epoch": 19.043613707165107, "grad_norm": 1.8828125, "learning_rate": 4.909211058286182e-05, "loss": 0.1636, "step": 3066 }, { "epoch": 19.049844236760123, "grad_norm": 1.25, "learning_rate": 4.909145253367924e-05, "loss": 0.1518, "step": 3067 }, { "epoch": 19.05607476635514, "grad_norm": 2.390625, "learning_rate": 4.9090794250515643e-05, "loss": 0.3124, "step": 3068 }, { "epoch": 19.062305295950157, "grad_norm": 3.328125, "learning_rate": 4.909013573337744e-05, "loss": 0.3117, "step": 3069 }, { "epoch": 19.068535825545172, "grad_norm": 3.265625, "learning_rate": 4.9089476982271026e-05, "loss": 0.2121, "step": 3070 }, { "epoch": 19.074766355140188, "grad_norm": 2.171875, "learning_rate": 4.90888179972028e-05, "loss": 0.1594, "step": 3071 }, { "epoch": 19.080996884735203, "grad_norm": 3.84375, "learning_rate": 4.908815877817915e-05, "loss": 0.2906, "step": 3072 }, { "epoch": 19.08722741433022, "grad_norm": 1.9140625, "learning_rate": 4.9087499325206494e-05, "loss": 0.2148, "step": 3073 }, { "epoch": 19.093457943925234, "grad_norm": 2.078125, "learning_rate": 4.9086839638291224e-05, "loss": 0.1585, "step": 3074 }, { "epoch": 19.09968847352025, "grad_norm": 2.703125, "learning_rate": 4.908617971743976e-05, "loss": 0.2525, "step": 3075 }, { "epoch": 19.105919003115265, "grad_norm": 2.390625, "learning_rate": 4.9085519562658496e-05, "loss": 0.1505, "step": 3076 }, { "epoch": 19.11214953271028, "grad_norm": 3.078125, "learning_rate": 4.908485917395385e-05, "loss": 0.2155, "step": 3077 }, { "epoch": 19.118380062305295, "grad_norm": 3.578125, "learning_rate": 4.9084198551332236e-05, "loss": 0.3814, "step": 3078 }, { "epoch": 19.12461059190031, "grad_norm": 3.875, "learning_rate": 4.9083537694800086e-05, "loss": 0.3142, "step": 3079 }, { "epoch": 19.130841121495326, "grad_norm": 2.796875, "learning_rate": 4.908287660436379e-05, "loss": 0.2386, "step": 3080 }, { "epoch": 19.13707165109034, "grad_norm": 3.046875, "learning_rate": 4.908221528002979e-05, "loss": 0.3296, "step": 3081 }, { "epoch": 19.143302180685357, "grad_norm": 3.03125, "learning_rate": 4.90815537218045e-05, "loss": 0.3576, "step": 3082 }, { "epoch": 19.149532710280372, "grad_norm": 2.140625, "learning_rate": 4.908089192969434e-05, "loss": 0.2233, "step": 3083 }, { "epoch": 19.15576323987539, "grad_norm": 3.703125, "learning_rate": 4.908022990370575e-05, "loss": 0.4476, "step": 3084 }, { "epoch": 19.161993769470406, "grad_norm": 3.453125, "learning_rate": 4.9079567643845146e-05, "loss": 0.2888, "step": 3085 }, { "epoch": 19.16822429906542, "grad_norm": 3.359375, "learning_rate": 4.907890515011898e-05, "loss": 0.2332, "step": 3086 }, { "epoch": 19.174454828660437, "grad_norm": 3.765625, "learning_rate": 4.907824242253367e-05, "loss": 0.3619, "step": 3087 }, { "epoch": 19.180685358255452, "grad_norm": 2.953125, "learning_rate": 4.907757946109565e-05, "loss": 0.3148, "step": 3088 }, { "epoch": 19.186915887850468, "grad_norm": 2.953125, "learning_rate": 4.907691626581137e-05, "loss": 0.3961, "step": 3089 }, { "epoch": 19.193146417445483, "grad_norm": 3.875, "learning_rate": 4.9076252836687265e-05, "loss": 0.4918, "step": 3090 }, { "epoch": 19.1993769470405, "grad_norm": 2.859375, "learning_rate": 4.907558917372978e-05, "loss": 0.276, "step": 3091 }, { "epoch": 19.205607476635514, "grad_norm": 3.140625, "learning_rate": 4.907492527694536e-05, "loss": 0.2792, "step": 3092 }, { "epoch": 19.21183800623053, "grad_norm": 2.96875, "learning_rate": 4.907426114634046e-05, "loss": 0.2226, "step": 3093 }, { "epoch": 19.218068535825545, "grad_norm": 2.671875, "learning_rate": 4.907359678192152e-05, "loss": 0.3387, "step": 3094 }, { "epoch": 19.22429906542056, "grad_norm": 3.84375, "learning_rate": 4.907293218369499e-05, "loss": 0.3505, "step": 3095 }, { "epoch": 19.230529595015575, "grad_norm": 3.1875, "learning_rate": 4.907226735166733e-05, "loss": 0.2482, "step": 3096 }, { "epoch": 19.23676012461059, "grad_norm": 1.8671875, "learning_rate": 4.9071602285844995e-05, "loss": 0.1964, "step": 3097 }, { "epoch": 19.242990654205606, "grad_norm": 3.390625, "learning_rate": 4.907093698623446e-05, "loss": 0.1602, "step": 3098 }, { "epoch": 19.24922118380062, "grad_norm": 3.0, "learning_rate": 4.907027145284217e-05, "loss": 0.3094, "step": 3099 }, { "epoch": 19.25545171339564, "grad_norm": 2.78125, "learning_rate": 4.906960568567458e-05, "loss": 0.1831, "step": 3100 }, { "epoch": 19.261682242990656, "grad_norm": 3.28125, "learning_rate": 4.906893968473818e-05, "loss": 0.3772, "step": 3101 }, { "epoch": 19.26791277258567, "grad_norm": 2.734375, "learning_rate": 4.906827345003942e-05, "loss": 0.2361, "step": 3102 }, { "epoch": 19.274143302180686, "grad_norm": 3.390625, "learning_rate": 4.906760698158478e-05, "loss": 0.3612, "step": 3103 }, { "epoch": 19.2803738317757, "grad_norm": 3.84375, "learning_rate": 4.906694027938074e-05, "loss": 0.3152, "step": 3104 }, { "epoch": 19.286604361370717, "grad_norm": 2.34375, "learning_rate": 4.9066273343433745e-05, "loss": 0.2179, "step": 3105 }, { "epoch": 19.292834890965732, "grad_norm": 2.875, "learning_rate": 4.90656061737503e-05, "loss": 0.2598, "step": 3106 }, { "epoch": 19.299065420560748, "grad_norm": 2.328125, "learning_rate": 4.906493877033689e-05, "loss": 0.2962, "step": 3107 }, { "epoch": 19.305295950155763, "grad_norm": 2.140625, "learning_rate": 4.906427113319997e-05, "loss": 0.1802, "step": 3108 }, { "epoch": 19.31152647975078, "grad_norm": 2.78125, "learning_rate": 4.9063603262346043e-05, "loss": 0.1506, "step": 3109 }, { "epoch": 19.317757009345794, "grad_norm": 2.78125, "learning_rate": 4.906293515778159e-05, "loss": 0.1762, "step": 3110 }, { "epoch": 19.32398753894081, "grad_norm": 4.25, "learning_rate": 4.9062266819513104e-05, "loss": 0.3501, "step": 3111 }, { "epoch": 19.330218068535824, "grad_norm": 2.859375, "learning_rate": 4.9061598247547065e-05, "loss": 0.1877, "step": 3112 }, { "epoch": 19.33644859813084, "grad_norm": 3.359375, "learning_rate": 4.906092944188998e-05, "loss": 0.3526, "step": 3113 }, { "epoch": 19.342679127725855, "grad_norm": 3.28125, "learning_rate": 4.9060260402548335e-05, "loss": 0.2614, "step": 3114 }, { "epoch": 19.34890965732087, "grad_norm": 4.15625, "learning_rate": 4.905959112952864e-05, "loss": 0.3874, "step": 3115 }, { "epoch": 19.35514018691589, "grad_norm": 2.25, "learning_rate": 4.905892162283737e-05, "loss": 0.1889, "step": 3116 }, { "epoch": 19.361370716510905, "grad_norm": 3.203125, "learning_rate": 4.905825188248106e-05, "loss": 0.2606, "step": 3117 }, { "epoch": 19.36760124610592, "grad_norm": 2.09375, "learning_rate": 4.905758190846619e-05, "loss": 0.1775, "step": 3118 }, { "epoch": 19.373831775700936, "grad_norm": 3.046875, "learning_rate": 4.905691170079928e-05, "loss": 0.1741, "step": 3119 }, { "epoch": 19.38006230529595, "grad_norm": 3.609375, "learning_rate": 4.9056241259486834e-05, "loss": 0.2542, "step": 3120 }, { "epoch": 19.386292834890966, "grad_norm": 3.9375, "learning_rate": 4.905557058453536e-05, "loss": 0.4012, "step": 3121 }, { "epoch": 19.39252336448598, "grad_norm": 3.8125, "learning_rate": 4.905489967595138e-05, "loss": 0.36, "step": 3122 }, { "epoch": 19.398753894080997, "grad_norm": 2.515625, "learning_rate": 4.905422853374141e-05, "loss": 0.2138, "step": 3123 }, { "epoch": 19.404984423676012, "grad_norm": 4.25, "learning_rate": 4.9053557157911956e-05, "loss": 0.2211, "step": 3124 }, { "epoch": 19.411214953271028, "grad_norm": 2.5, "learning_rate": 4.905288554846955e-05, "loss": 0.1589, "step": 3125 }, { "epoch": 19.417445482866043, "grad_norm": 3.3125, "learning_rate": 4.9052213705420714e-05, "loss": 0.2433, "step": 3126 }, { "epoch": 19.42367601246106, "grad_norm": 3.359375, "learning_rate": 4.9051541628771966e-05, "loss": 0.3605, "step": 3127 }, { "epoch": 19.429906542056074, "grad_norm": 4.21875, "learning_rate": 4.905086931852983e-05, "loss": 0.1968, "step": 3128 }, { "epoch": 19.43613707165109, "grad_norm": 5.40625, "learning_rate": 4.905019677470086e-05, "loss": 0.2688, "step": 3129 }, { "epoch": 19.442367601246104, "grad_norm": 3.546875, "learning_rate": 4.9049523997291566e-05, "loss": 0.4281, "step": 3130 }, { "epoch": 19.44859813084112, "grad_norm": 2.375, "learning_rate": 4.904885098630848e-05, "loss": 0.2377, "step": 3131 }, { "epoch": 19.45482866043614, "grad_norm": 3.328125, "learning_rate": 4.9048177741758156e-05, "loss": 0.2117, "step": 3132 }, { "epoch": 19.461059190031154, "grad_norm": 4.75, "learning_rate": 4.904750426364712e-05, "loss": 0.3542, "step": 3133 }, { "epoch": 19.46728971962617, "grad_norm": 4.21875, "learning_rate": 4.904683055198192e-05, "loss": 0.2113, "step": 3134 }, { "epoch": 19.473520249221185, "grad_norm": 2.34375, "learning_rate": 4.904615660676908e-05, "loss": 0.1973, "step": 3135 }, { "epoch": 19.4797507788162, "grad_norm": 4.0625, "learning_rate": 4.9045482428015175e-05, "loss": 0.203, "step": 3136 }, { "epoch": 19.485981308411215, "grad_norm": 2.15625, "learning_rate": 4.904480801572673e-05, "loss": 0.1692, "step": 3137 }, { "epoch": 19.49221183800623, "grad_norm": 2.6875, "learning_rate": 4.9044133369910303e-05, "loss": 0.238, "step": 3138 }, { "epoch": 19.498442367601246, "grad_norm": 2.625, "learning_rate": 4.9043458490572444e-05, "loss": 0.2145, "step": 3139 }, { "epoch": 19.50467289719626, "grad_norm": 2.484375, "learning_rate": 4.904278337771972e-05, "loss": 0.1962, "step": 3140 }, { "epoch": 19.510903426791277, "grad_norm": 2.8125, "learning_rate": 4.9042108031358675e-05, "loss": 0.2008, "step": 3141 }, { "epoch": 19.517133956386292, "grad_norm": 3.515625, "learning_rate": 4.9041432451495864e-05, "loss": 0.2027, "step": 3142 }, { "epoch": 19.523364485981308, "grad_norm": 3.515625, "learning_rate": 4.9040756638137864e-05, "loss": 0.3571, "step": 3143 }, { "epoch": 19.529595015576323, "grad_norm": 1.8515625, "learning_rate": 4.904008059129123e-05, "loss": 0.1644, "step": 3144 }, { "epoch": 19.53582554517134, "grad_norm": 2.109375, "learning_rate": 4.903940431096252e-05, "loss": 0.1999, "step": 3145 }, { "epoch": 19.542056074766354, "grad_norm": 2.65625, "learning_rate": 4.9038727797158315e-05, "loss": 0.2853, "step": 3146 }, { "epoch": 19.54828660436137, "grad_norm": 2.75, "learning_rate": 4.903805104988518e-05, "loss": 0.2212, "step": 3147 }, { "epoch": 19.554517133956388, "grad_norm": 3.015625, "learning_rate": 4.903737406914968e-05, "loss": 0.3264, "step": 3148 }, { "epoch": 19.560747663551403, "grad_norm": 2.65625, "learning_rate": 4.903669685495841e-05, "loss": 0.2419, "step": 3149 }, { "epoch": 19.56697819314642, "grad_norm": 2.1875, "learning_rate": 4.9036019407317925e-05, "loss": 0.2386, "step": 3150 }, { "epoch": 19.573208722741434, "grad_norm": 2.296875, "learning_rate": 4.903534172623482e-05, "loss": 0.2133, "step": 3151 }, { "epoch": 19.57943925233645, "grad_norm": 3.1875, "learning_rate": 4.903466381171568e-05, "loss": 0.2937, "step": 3152 }, { "epoch": 19.585669781931465, "grad_norm": 2.609375, "learning_rate": 4.9033985663767067e-05, "loss": 0.347, "step": 3153 }, { "epoch": 19.59190031152648, "grad_norm": 2.234375, "learning_rate": 4.903330728239558e-05, "loss": 0.2167, "step": 3154 }, { "epoch": 19.598130841121495, "grad_norm": 2.0, "learning_rate": 4.9032628667607813e-05, "loss": 0.1958, "step": 3155 }, { "epoch": 19.60436137071651, "grad_norm": 1.9765625, "learning_rate": 4.9031949819410364e-05, "loss": 0.1614, "step": 3156 }, { "epoch": 19.610591900311526, "grad_norm": 1.6796875, "learning_rate": 4.90312707378098e-05, "loss": 0.2898, "step": 3157 }, { "epoch": 19.61682242990654, "grad_norm": 2.75, "learning_rate": 4.903059142281273e-05, "loss": 0.2484, "step": 3158 }, { "epoch": 19.623052959501557, "grad_norm": 2.125, "learning_rate": 4.902991187442576e-05, "loss": 0.1861, "step": 3159 }, { "epoch": 19.629283489096572, "grad_norm": 2.078125, "learning_rate": 4.9029232092655486e-05, "loss": 0.1869, "step": 3160 }, { "epoch": 19.635514018691588, "grad_norm": 2.234375, "learning_rate": 4.90285520775085e-05, "loss": 0.2283, "step": 3161 }, { "epoch": 19.641744548286603, "grad_norm": 2.453125, "learning_rate": 4.902787182899141e-05, "loss": 0.2545, "step": 3162 }, { "epoch": 19.64797507788162, "grad_norm": 2.25, "learning_rate": 4.9027191347110835e-05, "loss": 0.2364, "step": 3163 }, { "epoch": 19.654205607476637, "grad_norm": 2.453125, "learning_rate": 4.902651063187337e-05, "loss": 0.188, "step": 3164 }, { "epoch": 19.660436137071652, "grad_norm": 3.5625, "learning_rate": 4.902582968328563e-05, "loss": 0.2918, "step": 3165 }, { "epoch": 19.666666666666668, "grad_norm": 3.375, "learning_rate": 4.9025148501354225e-05, "loss": 0.2463, "step": 3166 }, { "epoch": 19.672897196261683, "grad_norm": 4.125, "learning_rate": 4.902446708608578e-05, "loss": 0.371, "step": 3167 }, { "epoch": 19.6791277258567, "grad_norm": 2.40625, "learning_rate": 4.902378543748691e-05, "loss": 0.2177, "step": 3168 }, { "epoch": 19.685358255451714, "grad_norm": 2.0625, "learning_rate": 4.902310355556423e-05, "loss": 0.2102, "step": 3169 }, { "epoch": 19.69158878504673, "grad_norm": 4.9375, "learning_rate": 4.902242144032437e-05, "loss": 0.4494, "step": 3170 }, { "epoch": 19.697819314641745, "grad_norm": 4.40625, "learning_rate": 4.9021739091773945e-05, "loss": 0.2002, "step": 3171 }, { "epoch": 19.70404984423676, "grad_norm": 2.15625, "learning_rate": 4.902105650991959e-05, "loss": 0.2199, "step": 3172 }, { "epoch": 19.710280373831775, "grad_norm": 2.578125, "learning_rate": 4.902037369476794e-05, "loss": 0.1912, "step": 3173 }, { "epoch": 19.71651090342679, "grad_norm": 2.421875, "learning_rate": 4.901969064632561e-05, "loss": 0.1454, "step": 3174 }, { "epoch": 19.722741433021806, "grad_norm": 3.15625, "learning_rate": 4.9019007364599246e-05, "loss": 0.2423, "step": 3175 }, { "epoch": 19.72897196261682, "grad_norm": 3.46875, "learning_rate": 4.901832384959548e-05, "loss": 0.4535, "step": 3176 }, { "epoch": 19.735202492211837, "grad_norm": 3.578125, "learning_rate": 4.9017640101320955e-05, "loss": 0.187, "step": 3177 }, { "epoch": 19.741433021806852, "grad_norm": 3.203125, "learning_rate": 4.901695611978231e-05, "loss": 0.1921, "step": 3178 }, { "epoch": 19.747663551401867, "grad_norm": 1.515625, "learning_rate": 4.901627190498618e-05, "loss": 0.1572, "step": 3179 }, { "epoch": 19.753894080996886, "grad_norm": 2.359375, "learning_rate": 4.901558745693922e-05, "loss": 0.2943, "step": 3180 }, { "epoch": 19.7601246105919, "grad_norm": 3.875, "learning_rate": 4.9014902775648076e-05, "loss": 0.3035, "step": 3181 }, { "epoch": 19.766355140186917, "grad_norm": 2.671875, "learning_rate": 4.9014217861119385e-05, "loss": 0.2826, "step": 3182 }, { "epoch": 19.772585669781932, "grad_norm": 2.625, "learning_rate": 4.901353271335982e-05, "loss": 0.2041, "step": 3183 }, { "epoch": 19.778816199376948, "grad_norm": 2.5, "learning_rate": 4.901284733237602e-05, "loss": 0.2349, "step": 3184 }, { "epoch": 19.785046728971963, "grad_norm": 3.390625, "learning_rate": 4.9012161718174657e-05, "loss": 0.1876, "step": 3185 }, { "epoch": 19.79127725856698, "grad_norm": 3.5, "learning_rate": 4.901147587076237e-05, "loss": 0.4627, "step": 3186 }, { "epoch": 19.797507788161994, "grad_norm": 3.8125, "learning_rate": 4.9010789790145827e-05, "loss": 0.2834, "step": 3187 }, { "epoch": 19.80373831775701, "grad_norm": 1.59375, "learning_rate": 4.9010103476331704e-05, "loss": 0.1371, "step": 3188 }, { "epoch": 19.809968847352025, "grad_norm": 2.34375, "learning_rate": 4.9009416929326646e-05, "loss": 0.2568, "step": 3189 }, { "epoch": 19.81619937694704, "grad_norm": 3.0625, "learning_rate": 4.900873014913734e-05, "loss": 0.232, "step": 3190 }, { "epoch": 19.822429906542055, "grad_norm": 2.171875, "learning_rate": 4.9008043135770444e-05, "loss": 0.1609, "step": 3191 }, { "epoch": 19.82866043613707, "grad_norm": 2.015625, "learning_rate": 4.900735588923263e-05, "loss": 0.1803, "step": 3192 }, { "epoch": 19.834890965732086, "grad_norm": 1.859375, "learning_rate": 4.900666840953059e-05, "loss": 0.1417, "step": 3193 }, { "epoch": 19.8411214953271, "grad_norm": 2.890625, "learning_rate": 4.9005980696670975e-05, "loss": 0.3864, "step": 3194 }, { "epoch": 19.847352024922117, "grad_norm": 3.21875, "learning_rate": 4.9005292750660474e-05, "loss": 0.3011, "step": 3195 }, { "epoch": 19.853582554517136, "grad_norm": 1.84375, "learning_rate": 4.900460457150578e-05, "loss": 0.1852, "step": 3196 }, { "epoch": 19.85981308411215, "grad_norm": 2.5, "learning_rate": 4.900391615921356e-05, "loss": 0.2146, "step": 3197 }, { "epoch": 19.866043613707166, "grad_norm": 2.640625, "learning_rate": 4.900322751379052e-05, "loss": 0.2921, "step": 3198 }, { "epoch": 19.87227414330218, "grad_norm": 2.28125, "learning_rate": 4.9002538635243324e-05, "loss": 0.2239, "step": 3199 }, { "epoch": 19.878504672897197, "grad_norm": 3.28125, "learning_rate": 4.900184952357868e-05, "loss": 0.388, "step": 3200 }, { "epoch": 19.884735202492212, "grad_norm": 3.125, "learning_rate": 4.900116017880328e-05, "loss": 0.3077, "step": 3201 }, { "epoch": 19.890965732087228, "grad_norm": 3.78125, "learning_rate": 4.900047060092381e-05, "loss": 0.2593, "step": 3202 }, { "epoch": 19.897196261682243, "grad_norm": 3.875, "learning_rate": 4.899978078994697e-05, "loss": 0.3309, "step": 3203 }, { "epoch": 19.90342679127726, "grad_norm": 2.953125, "learning_rate": 4.899909074587946e-05, "loss": 0.1719, "step": 3204 }, { "epoch": 19.909657320872274, "grad_norm": 3.046875, "learning_rate": 4.899840046872799e-05, "loss": 0.343, "step": 3205 }, { "epoch": 19.91588785046729, "grad_norm": 3.75, "learning_rate": 4.8997709958499256e-05, "loss": 0.249, "step": 3206 }, { "epoch": 19.922118380062305, "grad_norm": 3.453125, "learning_rate": 4.8997019215199966e-05, "loss": 0.1886, "step": 3207 }, { "epoch": 19.92834890965732, "grad_norm": 3.3125, "learning_rate": 4.8996328238836823e-05, "loss": 0.3185, "step": 3208 }, { "epoch": 19.934579439252335, "grad_norm": 4.15625, "learning_rate": 4.899563702941655e-05, "loss": 0.4896, "step": 3209 }, { "epoch": 19.94080996884735, "grad_norm": 2.625, "learning_rate": 4.8994945586945854e-05, "loss": 0.2065, "step": 3210 }, { "epoch": 19.947040498442366, "grad_norm": 3.046875, "learning_rate": 4.8994253911431444e-05, "loss": 0.1521, "step": 3211 }, { "epoch": 19.953271028037385, "grad_norm": 5.0625, "learning_rate": 4.899356200288004e-05, "loss": 0.1821, "step": 3212 }, { "epoch": 19.9595015576324, "grad_norm": 3.515625, "learning_rate": 4.8992869861298376e-05, "loss": 0.1795, "step": 3213 }, { "epoch": 19.965732087227416, "grad_norm": 3.046875, "learning_rate": 4.8992177486693157e-05, "loss": 0.2941, "step": 3214 }, { "epoch": 19.97196261682243, "grad_norm": 2.546875, "learning_rate": 4.899148487907112e-05, "loss": 0.2657, "step": 3215 }, { "epoch": 19.978193146417446, "grad_norm": 4.40625, "learning_rate": 4.899079203843898e-05, "loss": 0.386, "step": 3216 }, { "epoch": 19.98442367601246, "grad_norm": 5.21875, "learning_rate": 4.899009896480348e-05, "loss": 0.251, "step": 3217 }, { "epoch": 19.990654205607477, "grad_norm": 2.546875, "learning_rate": 4.898940565817134e-05, "loss": 0.1601, "step": 3218 }, { "epoch": 19.996884735202492, "grad_norm": 2.296875, "learning_rate": 4.898871211854929e-05, "loss": 0.2104, "step": 3219 }, { "epoch": 20.0, "grad_norm": 1.703125, "learning_rate": 4.898801834594407e-05, "loss": 0.1144, "step": 3220 }, { "epoch": 20.006230529595015, "grad_norm": 5.0625, "learning_rate": 4.898732434036244e-05, "loss": 0.2472, "step": 3221 }, { "epoch": 20.01246105919003, "grad_norm": 5.40625, "learning_rate": 4.89866301018111e-05, "loss": 0.2146, "step": 3222 }, { "epoch": 20.018691588785046, "grad_norm": 5.1875, "learning_rate": 4.898593563029683e-05, "loss": 0.3083, "step": 3223 }, { "epoch": 20.02492211838006, "grad_norm": 2.890625, "learning_rate": 4.898524092582635e-05, "loss": 0.2753, "step": 3224 }, { "epoch": 20.031152647975077, "grad_norm": 4.84375, "learning_rate": 4.8984545988406414e-05, "loss": 0.1724, "step": 3225 }, { "epoch": 20.037383177570092, "grad_norm": 6.65625, "learning_rate": 4.898385081804378e-05, "loss": 0.2902, "step": 3226 }, { "epoch": 20.043613707165107, "grad_norm": 6.03125, "learning_rate": 4.898315541474519e-05, "loss": 0.1972, "step": 3227 }, { "epoch": 20.049844236760123, "grad_norm": 4.28125, "learning_rate": 4.8982459778517396e-05, "loss": 0.3022, "step": 3228 }, { "epoch": 20.05607476635514, "grad_norm": 3.875, "learning_rate": 4.8981763909367164e-05, "loss": 0.265, "step": 3229 }, { "epoch": 20.062305295950157, "grad_norm": 4.78125, "learning_rate": 4.898106780730125e-05, "loss": 0.2067, "step": 3230 }, { "epoch": 20.068535825545172, "grad_norm": 4.40625, "learning_rate": 4.8980371472326405e-05, "loss": 0.163, "step": 3231 }, { "epoch": 20.074766355140188, "grad_norm": 5.75, "learning_rate": 4.8979674904449404e-05, "loss": 0.2827, "step": 3232 }, { "epoch": 20.080996884735203, "grad_norm": 4.96875, "learning_rate": 4.8978978103676995e-05, "loss": 0.2162, "step": 3233 }, { "epoch": 20.08722741433022, "grad_norm": 3.21875, "learning_rate": 4.897828107001597e-05, "loss": 0.2788, "step": 3234 }, { "epoch": 20.093457943925234, "grad_norm": 4.5625, "learning_rate": 4.897758380347308e-05, "loss": 0.2857, "step": 3235 }, { "epoch": 20.09968847352025, "grad_norm": 5.15625, "learning_rate": 4.8976886304055104e-05, "loss": 0.3114, "step": 3236 }, { "epoch": 20.105919003115265, "grad_norm": 6.84375, "learning_rate": 4.897618857176882e-05, "loss": 0.2392, "step": 3237 }, { "epoch": 20.11214953271028, "grad_norm": 3.484375, "learning_rate": 4.8975490606620994e-05, "loss": 0.204, "step": 3238 }, { "epoch": 20.118380062305295, "grad_norm": 2.828125, "learning_rate": 4.897479240861841e-05, "loss": 0.2827, "step": 3239 }, { "epoch": 20.12461059190031, "grad_norm": 1.3125, "learning_rate": 4.897409397776785e-05, "loss": 0.1486, "step": 3240 }, { "epoch": 20.130841121495326, "grad_norm": 3.59375, "learning_rate": 4.8973395314076104e-05, "loss": 0.2473, "step": 3241 }, { "epoch": 20.13707165109034, "grad_norm": 3.609375, "learning_rate": 4.897269641754995e-05, "loss": 0.2595, "step": 3242 }, { "epoch": 20.143302180685357, "grad_norm": 2.6875, "learning_rate": 4.897199728819617e-05, "loss": 0.349, "step": 3243 }, { "epoch": 20.149532710280372, "grad_norm": 2.40625, "learning_rate": 4.897129792602156e-05, "loss": 0.2063, "step": 3244 }, { "epoch": 20.15576323987539, "grad_norm": 3.75, "learning_rate": 4.897059833103291e-05, "loss": 0.2153, "step": 3245 }, { "epoch": 20.161993769470406, "grad_norm": 2.078125, "learning_rate": 4.8969898503237034e-05, "loss": 0.1996, "step": 3246 }, { "epoch": 20.16822429906542, "grad_norm": 3.546875, "learning_rate": 4.8969198442640695e-05, "loss": 0.2079, "step": 3247 }, { "epoch": 20.174454828660437, "grad_norm": 2.828125, "learning_rate": 4.896849814925072e-05, "loss": 0.2401, "step": 3248 }, { "epoch": 20.180685358255452, "grad_norm": 2.796875, "learning_rate": 4.896779762307389e-05, "loss": 0.2154, "step": 3249 }, { "epoch": 20.186915887850468, "grad_norm": 3.453125, "learning_rate": 4.896709686411702e-05, "loss": 0.1862, "step": 3250 }, { "epoch": 20.193146417445483, "grad_norm": 1.984375, "learning_rate": 4.896639587238692e-05, "loss": 0.1939, "step": 3251 }, { "epoch": 20.1993769470405, "grad_norm": 2.484375, "learning_rate": 4.8965694647890393e-05, "loss": 0.2555, "step": 3252 }, { "epoch": 20.205607476635514, "grad_norm": 3.765625, "learning_rate": 4.896499319063424e-05, "loss": 0.5106, "step": 3253 }, { "epoch": 20.21183800623053, "grad_norm": 4.0, "learning_rate": 4.8964291500625295e-05, "loss": 0.3443, "step": 3254 }, { "epoch": 20.218068535825545, "grad_norm": 3.390625, "learning_rate": 4.8963589577870356e-05, "loss": 0.2906, "step": 3255 }, { "epoch": 20.22429906542056, "grad_norm": 2.46875, "learning_rate": 4.896288742237624e-05, "loss": 0.1946, "step": 3256 }, { "epoch": 20.230529595015575, "grad_norm": 3.234375, "learning_rate": 4.8962185034149776e-05, "loss": 0.2539, "step": 3257 }, { "epoch": 20.23676012461059, "grad_norm": 2.53125, "learning_rate": 4.896148241319778e-05, "loss": 0.1938, "step": 3258 }, { "epoch": 20.242990654205606, "grad_norm": 4.4375, "learning_rate": 4.896077955952708e-05, "loss": 0.2428, "step": 3259 }, { "epoch": 20.24922118380062, "grad_norm": 2.109375, "learning_rate": 4.8960076473144486e-05, "loss": 0.1606, "step": 3260 }, { "epoch": 20.25545171339564, "grad_norm": 2.5, "learning_rate": 4.895937315405685e-05, "loss": 0.2111, "step": 3261 }, { "epoch": 20.261682242990656, "grad_norm": 1.5, "learning_rate": 4.8958669602271e-05, "loss": 0.1508, "step": 3262 }, { "epoch": 20.26791277258567, "grad_norm": 3.359375, "learning_rate": 4.895796581779375e-05, "loss": 0.2634, "step": 3263 }, { "epoch": 20.274143302180686, "grad_norm": 3.203125, "learning_rate": 4.8957261800631953e-05, "loss": 0.1663, "step": 3264 }, { "epoch": 20.2803738317757, "grad_norm": 3.125, "learning_rate": 4.895655755079244e-05, "loss": 0.1659, "step": 3265 }, { "epoch": 20.286604361370717, "grad_norm": 4.15625, "learning_rate": 4.895585306828205e-05, "loss": 0.3859, "step": 3266 }, { "epoch": 20.292834890965732, "grad_norm": 2.296875, "learning_rate": 4.8955148353107625e-05, "loss": 0.1836, "step": 3267 }, { "epoch": 20.299065420560748, "grad_norm": 2.609375, "learning_rate": 4.895444340527601e-05, "loss": 0.2099, "step": 3268 }, { "epoch": 20.305295950155763, "grad_norm": 2.46875, "learning_rate": 4.8953738224794056e-05, "loss": 0.1798, "step": 3269 }, { "epoch": 20.31152647975078, "grad_norm": 2.515625, "learning_rate": 4.89530328116686e-05, "loss": 0.2437, "step": 3270 }, { "epoch": 20.317757009345794, "grad_norm": 3.90625, "learning_rate": 4.895232716590651e-05, "loss": 0.2852, "step": 3271 }, { "epoch": 20.32398753894081, "grad_norm": 4.5625, "learning_rate": 4.8951621287514626e-05, "loss": 0.229, "step": 3272 }, { "epoch": 20.330218068535824, "grad_norm": 2.859375, "learning_rate": 4.895091517649981e-05, "loss": 0.2869, "step": 3273 }, { "epoch": 20.33644859813084, "grad_norm": 4.0, "learning_rate": 4.8950208832868916e-05, "loss": 0.3576, "step": 3274 }, { "epoch": 20.342679127725855, "grad_norm": 3.125, "learning_rate": 4.894950225662881e-05, "loss": 0.1831, "step": 3275 }, { "epoch": 20.34890965732087, "grad_norm": 2.40625, "learning_rate": 4.894879544778635e-05, "loss": 0.3379, "step": 3276 }, { "epoch": 20.35514018691589, "grad_norm": 2.546875, "learning_rate": 4.8948088406348404e-05, "loss": 0.2342, "step": 3277 }, { "epoch": 20.361370716510905, "grad_norm": 2.015625, "learning_rate": 4.8947381132321826e-05, "loss": 0.2011, "step": 3278 }, { "epoch": 20.36760124610592, "grad_norm": 3.8125, "learning_rate": 4.89466736257135e-05, "loss": 0.3793, "step": 3279 }, { "epoch": 20.373831775700936, "grad_norm": 2.34375, "learning_rate": 4.8945965886530296e-05, "loss": 0.2084, "step": 3280 }, { "epoch": 20.38006230529595, "grad_norm": 3.234375, "learning_rate": 4.8945257914779086e-05, "loss": 0.2007, "step": 3281 }, { "epoch": 20.386292834890966, "grad_norm": 2.421875, "learning_rate": 4.894454971046674e-05, "loss": 0.1939, "step": 3282 }, { "epoch": 20.39252336448598, "grad_norm": 1.9140625, "learning_rate": 4.894384127360014e-05, "loss": 0.1757, "step": 3283 }, { "epoch": 20.398753894080997, "grad_norm": 1.96875, "learning_rate": 4.894313260418617e-05, "loss": 0.1628, "step": 3284 }, { "epoch": 20.404984423676012, "grad_norm": 2.46875, "learning_rate": 4.894242370223171e-05, "loss": 0.3531, "step": 3285 }, { "epoch": 20.411214953271028, "grad_norm": 2.734375, "learning_rate": 4.8941714567743644e-05, "loss": 0.2213, "step": 3286 }, { "epoch": 20.417445482866043, "grad_norm": 2.53125, "learning_rate": 4.894100520072886e-05, "loss": 0.2321, "step": 3287 }, { "epoch": 20.42367601246106, "grad_norm": 3.390625, "learning_rate": 4.8940295601194254e-05, "loss": 0.3302, "step": 3288 }, { "epoch": 20.429906542056074, "grad_norm": 2.6875, "learning_rate": 4.893958576914671e-05, "loss": 0.3206, "step": 3289 }, { "epoch": 20.43613707165109, "grad_norm": 3.265625, "learning_rate": 4.893887570459312e-05, "loss": 0.3167, "step": 3290 }, { "epoch": 20.442367601246104, "grad_norm": 2.671875, "learning_rate": 4.8938165407540385e-05, "loss": 0.2373, "step": 3291 }, { "epoch": 20.44859813084112, "grad_norm": 2.140625, "learning_rate": 4.89374548779954e-05, "loss": 0.1927, "step": 3292 }, { "epoch": 20.45482866043614, "grad_norm": 3.546875, "learning_rate": 4.893674411596507e-05, "loss": 0.2746, "step": 3293 }, { "epoch": 20.461059190031154, "grad_norm": 3.140625, "learning_rate": 4.8936033121456296e-05, "loss": 0.2338, "step": 3294 }, { "epoch": 20.46728971962617, "grad_norm": 4.34375, "learning_rate": 4.8935321894475984e-05, "loss": 0.4454, "step": 3295 }, { "epoch": 20.473520249221185, "grad_norm": 3.25, "learning_rate": 4.893461043503104e-05, "loss": 0.172, "step": 3296 }, { "epoch": 20.4797507788162, "grad_norm": 3.09375, "learning_rate": 4.893389874312837e-05, "loss": 0.2372, "step": 3297 }, { "epoch": 20.485981308411215, "grad_norm": 3.453125, "learning_rate": 4.89331868187749e-05, "loss": 0.1694, "step": 3298 }, { "epoch": 20.49221183800623, "grad_norm": 2.859375, "learning_rate": 4.8932474661977534e-05, "loss": 0.2115, "step": 3299 }, { "epoch": 20.498442367601246, "grad_norm": 2.765625, "learning_rate": 4.893176227274319e-05, "loss": 0.2714, "step": 3300 }, { "epoch": 20.50467289719626, "grad_norm": 2.578125, "learning_rate": 4.8931049651078784e-05, "loss": 0.2863, "step": 3301 }, { "epoch": 20.510903426791277, "grad_norm": 2.828125, "learning_rate": 4.893033679699124e-05, "loss": 0.2755, "step": 3302 }, { "epoch": 20.517133956386292, "grad_norm": 3.390625, "learning_rate": 4.8929623710487485e-05, "loss": 0.3349, "step": 3303 }, { "epoch": 20.523364485981308, "grad_norm": 4.34375, "learning_rate": 4.892891039157443e-05, "loss": 0.369, "step": 3304 }, { "epoch": 20.529595015576323, "grad_norm": 4.15625, "learning_rate": 4.8928196840259034e-05, "loss": 0.3344, "step": 3305 }, { "epoch": 20.53582554517134, "grad_norm": 2.8125, "learning_rate": 4.892748305654819e-05, "loss": 0.2282, "step": 3306 }, { "epoch": 20.542056074766354, "grad_norm": 3.484375, "learning_rate": 4.8926769040448855e-05, "loss": 0.3164, "step": 3307 }, { "epoch": 20.54828660436137, "grad_norm": 3.640625, "learning_rate": 4.892605479196796e-05, "loss": 0.4667, "step": 3308 }, { "epoch": 20.554517133956388, "grad_norm": 5.15625, "learning_rate": 4.892534031111243e-05, "loss": 0.3724, "step": 3309 }, { "epoch": 20.560747663551403, "grad_norm": 1.546875, "learning_rate": 4.8924625597889215e-05, "loss": 0.1778, "step": 3310 }, { "epoch": 20.56697819314642, "grad_norm": 3.4375, "learning_rate": 4.892391065230525e-05, "loss": 0.2197, "step": 3311 }, { "epoch": 20.573208722741434, "grad_norm": 3.328125, "learning_rate": 4.8923195474367486e-05, "loss": 0.1972, "step": 3312 }, { "epoch": 20.57943925233645, "grad_norm": 2.078125, "learning_rate": 4.8922480064082864e-05, "loss": 0.1505, "step": 3313 }, { "epoch": 20.585669781931465, "grad_norm": 2.75, "learning_rate": 4.892176442145834e-05, "loss": 0.2019, "step": 3314 }, { "epoch": 20.59190031152648, "grad_norm": 2.96875, "learning_rate": 4.892104854650085e-05, "loss": 0.2899, "step": 3315 }, { "epoch": 20.598130841121495, "grad_norm": 3.953125, "learning_rate": 4.892033243921736e-05, "loss": 0.3131, "step": 3316 }, { "epoch": 20.60436137071651, "grad_norm": 3.484375, "learning_rate": 4.891961609961482e-05, "loss": 0.2424, "step": 3317 }, { "epoch": 20.610591900311526, "grad_norm": 2.5, "learning_rate": 4.8918899527700185e-05, "loss": 0.153, "step": 3318 }, { "epoch": 20.61682242990654, "grad_norm": 3.1875, "learning_rate": 4.891818272348042e-05, "loss": 0.2149, "step": 3319 }, { "epoch": 20.623052959501557, "grad_norm": 2.234375, "learning_rate": 4.8917465686962475e-05, "loss": 0.1649, "step": 3320 }, { "epoch": 20.629283489096572, "grad_norm": 3.96875, "learning_rate": 4.891674841815333e-05, "loss": 0.2948, "step": 3321 }, { "epoch": 20.635514018691588, "grad_norm": 3.296875, "learning_rate": 4.891603091705994e-05, "loss": 0.2146, "step": 3322 }, { "epoch": 20.641744548286603, "grad_norm": 3.015625, "learning_rate": 4.891531318368928e-05, "loss": 0.1852, "step": 3323 }, { "epoch": 20.64797507788162, "grad_norm": 4.96875, "learning_rate": 4.891459521804831e-05, "loss": 0.2451, "step": 3324 }, { "epoch": 20.654205607476637, "grad_norm": 3.484375, "learning_rate": 4.891387702014402e-05, "loss": 0.2591, "step": 3325 }, { "epoch": 20.660436137071652, "grad_norm": 2.9375, "learning_rate": 4.8913158589983374e-05, "loss": 0.2098, "step": 3326 }, { "epoch": 20.666666666666668, "grad_norm": 3.078125, "learning_rate": 4.891243992757335e-05, "loss": 0.1971, "step": 3327 }, { "epoch": 20.672897196261683, "grad_norm": 4.84375, "learning_rate": 4.8911721032920924e-05, "loss": 0.2352, "step": 3328 }, { "epoch": 20.6791277258567, "grad_norm": 1.6953125, "learning_rate": 4.891100190603309e-05, "loss": 0.1536, "step": 3329 }, { "epoch": 20.685358255451714, "grad_norm": 2.140625, "learning_rate": 4.891028254691682e-05, "loss": 0.2018, "step": 3330 }, { "epoch": 20.69158878504673, "grad_norm": 3.0625, "learning_rate": 4.8909562955579114e-05, "loss": 0.216, "step": 3331 }, { "epoch": 20.697819314641745, "grad_norm": 3.609375, "learning_rate": 4.890884313202695e-05, "loss": 0.3257, "step": 3332 }, { "epoch": 20.70404984423676, "grad_norm": 3.15625, "learning_rate": 4.8908123076267324e-05, "loss": 0.2069, "step": 3333 }, { "epoch": 20.710280373831775, "grad_norm": 3.703125, "learning_rate": 4.890740278830723e-05, "loss": 0.5516, "step": 3334 }, { "epoch": 20.71651090342679, "grad_norm": 2.453125, "learning_rate": 4.890668226815365e-05, "loss": 0.2724, "step": 3335 }, { "epoch": 20.722741433021806, "grad_norm": 3.375, "learning_rate": 4.8905961515813604e-05, "loss": 0.3129, "step": 3336 }, { "epoch": 20.72897196261682, "grad_norm": 2.046875, "learning_rate": 4.890524053129408e-05, "loss": 0.1899, "step": 3337 }, { "epoch": 20.735202492211837, "grad_norm": 1.546875, "learning_rate": 4.8904519314602075e-05, "loss": 0.1667, "step": 3338 }, { "epoch": 20.741433021806852, "grad_norm": 1.5546875, "learning_rate": 4.8903797865744606e-05, "loss": 0.1681, "step": 3339 }, { "epoch": 20.747663551401867, "grad_norm": 3.359375, "learning_rate": 4.890307618472867e-05, "loss": 0.21, "step": 3340 }, { "epoch": 20.753894080996886, "grad_norm": 2.5625, "learning_rate": 4.890235427156128e-05, "loss": 0.288, "step": 3341 }, { "epoch": 20.7601246105919, "grad_norm": 2.65625, "learning_rate": 4.8901632126249455e-05, "loss": 0.1873, "step": 3342 }, { "epoch": 20.766355140186917, "grad_norm": 2.125, "learning_rate": 4.890090974880019e-05, "loss": 0.2048, "step": 3343 }, { "epoch": 20.772585669781932, "grad_norm": 2.140625, "learning_rate": 4.890018713922052e-05, "loss": 0.1677, "step": 3344 }, { "epoch": 20.778816199376948, "grad_norm": 2.046875, "learning_rate": 4.8899464297517446e-05, "loss": 0.1247, "step": 3345 }, { "epoch": 20.785046728971963, "grad_norm": 3.375, "learning_rate": 4.8898741223698005e-05, "loss": 0.2857, "step": 3346 }, { "epoch": 20.79127725856698, "grad_norm": 3.671875, "learning_rate": 4.889801791776921e-05, "loss": 0.4177, "step": 3347 }, { "epoch": 20.797507788161994, "grad_norm": 3.0, "learning_rate": 4.8897294379738084e-05, "loss": 0.2315, "step": 3348 }, { "epoch": 20.80373831775701, "grad_norm": 3.46875, "learning_rate": 4.889657060961166e-05, "loss": 0.338, "step": 3349 }, { "epoch": 20.809968847352025, "grad_norm": 2.8125, "learning_rate": 4.8895846607396965e-05, "loss": 0.1961, "step": 3350 }, { "epoch": 20.81619937694704, "grad_norm": 4.09375, "learning_rate": 4.889512237310104e-05, "loss": 0.3245, "step": 3351 }, { "epoch": 20.822429906542055, "grad_norm": 2.546875, "learning_rate": 4.88943979067309e-05, "loss": 0.2706, "step": 3352 }, { "epoch": 20.82866043613707, "grad_norm": 2.234375, "learning_rate": 4.8893673208293596e-05, "loss": 0.1789, "step": 3353 }, { "epoch": 20.834890965732086, "grad_norm": 2.625, "learning_rate": 4.8892948277796156e-05, "loss": 0.196, "step": 3354 }, { "epoch": 20.8411214953271, "grad_norm": 3.6875, "learning_rate": 4.8892223115245624e-05, "loss": 0.3247, "step": 3355 }, { "epoch": 20.847352024922117, "grad_norm": 1.953125, "learning_rate": 4.889149772064906e-05, "loss": 0.1582, "step": 3356 }, { "epoch": 20.853582554517136, "grad_norm": 3.375, "learning_rate": 4.889077209401348e-05, "loss": 0.3276, "step": 3357 }, { "epoch": 20.85981308411215, "grad_norm": 3.375, "learning_rate": 4.889004623534596e-05, "loss": 0.3103, "step": 3358 }, { "epoch": 20.866043613707166, "grad_norm": 2.515625, "learning_rate": 4.888932014465352e-05, "loss": 0.2409, "step": 3359 }, { "epoch": 20.87227414330218, "grad_norm": 2.484375, "learning_rate": 4.8888593821943234e-05, "loss": 0.2077, "step": 3360 }, { "epoch": 20.878504672897197, "grad_norm": 2.015625, "learning_rate": 4.888786726722216e-05, "loss": 0.1958, "step": 3361 }, { "epoch": 20.884735202492212, "grad_norm": 2.640625, "learning_rate": 4.8887140480497326e-05, "loss": 0.2112, "step": 3362 }, { "epoch": 20.890965732087228, "grad_norm": 2.875, "learning_rate": 4.8886413461775816e-05, "loss": 0.433, "step": 3363 }, { "epoch": 20.897196261682243, "grad_norm": 3.328125, "learning_rate": 4.8885686211064686e-05, "loss": 0.2487, "step": 3364 }, { "epoch": 20.90342679127726, "grad_norm": 2.75, "learning_rate": 4.8884958728370986e-05, "loss": 0.2748, "step": 3365 }, { "epoch": 20.909657320872274, "grad_norm": 3.03125, "learning_rate": 4.888423101370181e-05, "loss": 0.3471, "step": 3366 }, { "epoch": 20.91588785046729, "grad_norm": 3.046875, "learning_rate": 4.8883503067064195e-05, "loss": 0.3551, "step": 3367 }, { "epoch": 20.922118380062305, "grad_norm": 3.0625, "learning_rate": 4.8882774888465226e-05, "loss": 0.2384, "step": 3368 }, { "epoch": 20.92834890965732, "grad_norm": 3.625, "learning_rate": 4.888204647791198e-05, "loss": 0.3785, "step": 3369 }, { "epoch": 20.934579439252335, "grad_norm": 1.9453125, "learning_rate": 4.8881317835411515e-05, "loss": 0.1391, "step": 3370 }, { "epoch": 20.94080996884735, "grad_norm": 3.75, "learning_rate": 4.888058896097092e-05, "loss": 0.2241, "step": 3371 }, { "epoch": 20.947040498442366, "grad_norm": 3.484375, "learning_rate": 4.887985985459728e-05, "loss": 0.2494, "step": 3372 }, { "epoch": 20.953271028037385, "grad_norm": 4.4375, "learning_rate": 4.887913051629766e-05, "loss": 0.3109, "step": 3373 }, { "epoch": 20.9595015576324, "grad_norm": 2.25, "learning_rate": 4.887840094607915e-05, "loss": 0.3076, "step": 3374 }, { "epoch": 20.965732087227416, "grad_norm": 3.765625, "learning_rate": 4.887767114394884e-05, "loss": 0.1931, "step": 3375 }, { "epoch": 20.97196261682243, "grad_norm": 2.078125, "learning_rate": 4.887694110991381e-05, "loss": 0.1919, "step": 3376 }, { "epoch": 20.978193146417446, "grad_norm": 2.609375, "learning_rate": 4.887621084398116e-05, "loss": 0.2654, "step": 3377 }, { "epoch": 20.98442367601246, "grad_norm": 3.546875, "learning_rate": 4.887548034615797e-05, "loss": 0.5526, "step": 3378 }, { "epoch": 20.990654205607477, "grad_norm": 3.390625, "learning_rate": 4.887474961645136e-05, "loss": 0.1927, "step": 3379 }, { "epoch": 20.996884735202492, "grad_norm": 2.953125, "learning_rate": 4.8874018654868394e-05, "loss": 0.3601, "step": 3380 }, { "epoch": 21.0, "grad_norm": 2.484375, "learning_rate": 4.887328746141619e-05, "loss": 0.1453, "step": 3381 }, { "epoch": 21.006230529595015, "grad_norm": 2.90625, "learning_rate": 4.887255603610185e-05, "loss": 0.2682, "step": 3382 }, { "epoch": 21.01246105919003, "grad_norm": 2.109375, "learning_rate": 4.8871824378932475e-05, "loss": 0.2105, "step": 3383 }, { "epoch": 21.018691588785046, "grad_norm": 1.8671875, "learning_rate": 4.887109248991516e-05, "loss": 0.1946, "step": 3384 }, { "epoch": 21.02492211838006, "grad_norm": 2.671875, "learning_rate": 4.887036036905703e-05, "loss": 0.2618, "step": 3385 }, { "epoch": 21.031152647975077, "grad_norm": 3.484375, "learning_rate": 4.8869628016365185e-05, "loss": 0.2958, "step": 3386 }, { "epoch": 21.037383177570092, "grad_norm": 2.734375, "learning_rate": 4.886889543184674e-05, "loss": 0.1742, "step": 3387 }, { "epoch": 21.043613707165107, "grad_norm": 2.546875, "learning_rate": 4.8868162615508814e-05, "loss": 0.2203, "step": 3388 }, { "epoch": 21.049844236760123, "grad_norm": 2.40625, "learning_rate": 4.886742956735852e-05, "loss": 0.1996, "step": 3389 }, { "epoch": 21.05607476635514, "grad_norm": 2.71875, "learning_rate": 4.886669628740299e-05, "loss": 0.2903, "step": 3390 }, { "epoch": 21.062305295950157, "grad_norm": 1.8046875, "learning_rate": 4.886596277564932e-05, "loss": 0.1989, "step": 3391 }, { "epoch": 21.068535825545172, "grad_norm": 3.15625, "learning_rate": 4.8865229032104656e-05, "loss": 0.2042, "step": 3392 }, { "epoch": 21.074766355140188, "grad_norm": 2.6875, "learning_rate": 4.8864495056776116e-05, "loss": 0.1988, "step": 3393 }, { "epoch": 21.080996884735203, "grad_norm": 2.375, "learning_rate": 4.8863760849670827e-05, "loss": 0.1612, "step": 3394 }, { "epoch": 21.08722741433022, "grad_norm": 3.234375, "learning_rate": 4.886302641079592e-05, "loss": 0.2878, "step": 3395 }, { "epoch": 21.093457943925234, "grad_norm": 2.515625, "learning_rate": 4.886229174015854e-05, "loss": 0.2407, "step": 3396 }, { "epoch": 21.09968847352025, "grad_norm": 3.0625, "learning_rate": 4.886155683776581e-05, "loss": 0.259, "step": 3397 }, { "epoch": 21.105919003115265, "grad_norm": 2.859375, "learning_rate": 4.8860821703624864e-05, "loss": 0.2667, "step": 3398 }, { "epoch": 21.11214953271028, "grad_norm": 1.890625, "learning_rate": 4.886008633774285e-05, "loss": 0.1763, "step": 3399 }, { "epoch": 21.118380062305295, "grad_norm": 4.0625, "learning_rate": 4.8859350740126905e-05, "loss": 0.4643, "step": 3400 }, { "epoch": 21.12461059190031, "grad_norm": 2.21875, "learning_rate": 4.885861491078418e-05, "loss": 0.2414, "step": 3401 }, { "epoch": 21.130841121495326, "grad_norm": 2.828125, "learning_rate": 4.885787884972183e-05, "loss": 0.1778, "step": 3402 }, { "epoch": 21.13707165109034, "grad_norm": 3.25, "learning_rate": 4.885714255694698e-05, "loss": 0.2682, "step": 3403 }, { "epoch": 21.143302180685357, "grad_norm": 2.34375, "learning_rate": 4.8856406032466785e-05, "loss": 0.2395, "step": 3404 }, { "epoch": 21.149532710280372, "grad_norm": 2.953125, "learning_rate": 4.885566927628842e-05, "loss": 0.1749, "step": 3405 }, { "epoch": 21.15576323987539, "grad_norm": 3.078125, "learning_rate": 4.8854932288419014e-05, "loss": 0.2494, "step": 3406 }, { "epoch": 21.161993769470406, "grad_norm": 2.25, "learning_rate": 4.885419506886574e-05, "loss": 0.1303, "step": 3407 }, { "epoch": 21.16822429906542, "grad_norm": 1.6953125, "learning_rate": 4.885345761763576e-05, "loss": 0.1647, "step": 3408 }, { "epoch": 21.174454828660437, "grad_norm": 4.1875, "learning_rate": 4.885271993473623e-05, "loss": 0.4889, "step": 3409 }, { "epoch": 21.180685358255452, "grad_norm": 2.734375, "learning_rate": 4.8851982020174316e-05, "loss": 0.3084, "step": 3410 }, { "epoch": 21.186915887850468, "grad_norm": 2.140625, "learning_rate": 4.8851243873957186e-05, "loss": 0.2094, "step": 3411 }, { "epoch": 21.193146417445483, "grad_norm": 2.71875, "learning_rate": 4.8850505496092e-05, "loss": 0.2649, "step": 3412 }, { "epoch": 21.1993769470405, "grad_norm": 1.8203125, "learning_rate": 4.8849766886585945e-05, "loss": 0.1392, "step": 3413 }, { "epoch": 21.205607476635514, "grad_norm": 2.515625, "learning_rate": 4.884902804544619e-05, "loss": 0.1694, "step": 3414 }, { "epoch": 21.21183800623053, "grad_norm": 2.5625, "learning_rate": 4.8848288972679895e-05, "loss": 0.1583, "step": 3415 }, { "epoch": 21.218068535825545, "grad_norm": 2.703125, "learning_rate": 4.8847549668294254e-05, "loss": 0.2253, "step": 3416 }, { "epoch": 21.22429906542056, "grad_norm": 2.140625, "learning_rate": 4.884681013229645e-05, "loss": 0.2105, "step": 3417 }, { "epoch": 21.230529595015575, "grad_norm": 5.65625, "learning_rate": 4.884607036469365e-05, "loss": 0.2703, "step": 3418 }, { "epoch": 21.23676012461059, "grad_norm": 5.34375, "learning_rate": 4.884533036549306e-05, "loss": 0.3291, "step": 3419 }, { "epoch": 21.242990654205606, "grad_norm": 5.71875, "learning_rate": 4.884459013470184e-05, "loss": 0.2593, "step": 3420 }, { "epoch": 21.24922118380062, "grad_norm": 2.359375, "learning_rate": 4.8843849672327213e-05, "loss": 0.2439, "step": 3421 }, { "epoch": 21.25545171339564, "grad_norm": 3.5625, "learning_rate": 4.884310897837634e-05, "loss": 0.3727, "step": 3422 }, { "epoch": 21.261682242990656, "grad_norm": 4.09375, "learning_rate": 4.8842368052856426e-05, "loss": 0.1833, "step": 3423 }, { "epoch": 21.26791277258567, "grad_norm": 5.625, "learning_rate": 4.8841626895774667e-05, "loss": 0.2189, "step": 3424 }, { "epoch": 21.274143302180686, "grad_norm": 4.59375, "learning_rate": 4.884088550713827e-05, "loss": 0.3129, "step": 3425 }, { "epoch": 21.2803738317757, "grad_norm": 5.6875, "learning_rate": 4.8840143886954414e-05, "loss": 0.4351, "step": 3426 }, { "epoch": 21.286604361370717, "grad_norm": 2.828125, "learning_rate": 4.883940203523032e-05, "loss": 0.2953, "step": 3427 }, { "epoch": 21.292834890965732, "grad_norm": 3.109375, "learning_rate": 4.883865995197319e-05, "loss": 0.3219, "step": 3428 }, { "epoch": 21.299065420560748, "grad_norm": 4.0625, "learning_rate": 4.883791763719023e-05, "loss": 0.2208, "step": 3429 }, { "epoch": 21.305295950155763, "grad_norm": 3.4375, "learning_rate": 4.8837175090888653e-05, "loss": 0.2127, "step": 3430 }, { "epoch": 21.31152647975078, "grad_norm": 2.625, "learning_rate": 4.8836432313075656e-05, "loss": 0.2484, "step": 3431 }, { "epoch": 21.317757009345794, "grad_norm": 1.859375, "learning_rate": 4.883568930375847e-05, "loss": 0.1723, "step": 3432 }, { "epoch": 21.32398753894081, "grad_norm": 4.21875, "learning_rate": 4.883494606294431e-05, "loss": 0.362, "step": 3433 }, { "epoch": 21.330218068535824, "grad_norm": 4.03125, "learning_rate": 4.883420259064038e-05, "loss": 0.1742, "step": 3434 }, { "epoch": 21.33644859813084, "grad_norm": 4.25, "learning_rate": 4.883345888685391e-05, "loss": 0.3653, "step": 3435 }, { "epoch": 21.342679127725855, "grad_norm": 2.1875, "learning_rate": 4.8832714951592127e-05, "loss": 0.1835, "step": 3436 }, { "epoch": 21.34890965732087, "grad_norm": 2.921875, "learning_rate": 4.883197078486226e-05, "loss": 0.207, "step": 3437 }, { "epoch": 21.35514018691589, "grad_norm": 3.125, "learning_rate": 4.8831226386671515e-05, "loss": 0.3358, "step": 3438 }, { "epoch": 21.361370716510905, "grad_norm": 2.359375, "learning_rate": 4.883048175702714e-05, "loss": 0.1781, "step": 3439 }, { "epoch": 21.36760124610592, "grad_norm": 2.515625, "learning_rate": 4.8829736895936364e-05, "loss": 0.1761, "step": 3440 }, { "epoch": 21.373831775700936, "grad_norm": 2.828125, "learning_rate": 4.882899180340642e-05, "loss": 0.3051, "step": 3441 }, { "epoch": 21.38006230529595, "grad_norm": 3.625, "learning_rate": 4.882824647944454e-05, "loss": 0.3534, "step": 3442 }, { "epoch": 21.386292834890966, "grad_norm": 4.625, "learning_rate": 4.8827500924057976e-05, "loss": 0.3004, "step": 3443 }, { "epoch": 21.39252336448598, "grad_norm": 4.65625, "learning_rate": 4.882675513725395e-05, "loss": 0.2649, "step": 3444 }, { "epoch": 21.398753894080997, "grad_norm": 4.78125, "learning_rate": 4.882600911903973e-05, "loss": 0.2398, "step": 3445 }, { "epoch": 21.404984423676012, "grad_norm": 4.53125, "learning_rate": 4.882526286942253e-05, "loss": 0.3462, "step": 3446 }, { "epoch": 21.411214953271028, "grad_norm": 4.40625, "learning_rate": 4.8824516388409625e-05, "loss": 0.2988, "step": 3447 }, { "epoch": 21.417445482866043, "grad_norm": 2.546875, "learning_rate": 4.882376967600825e-05, "loss": 0.2204, "step": 3448 }, { "epoch": 21.42367601246106, "grad_norm": 3.6875, "learning_rate": 4.8823022732225665e-05, "loss": 0.4311, "step": 3449 }, { "epoch": 21.429906542056074, "grad_norm": 3.15625, "learning_rate": 4.882227555706912e-05, "loss": 0.2644, "step": 3450 }, { "epoch": 21.43613707165109, "grad_norm": 3.4375, "learning_rate": 4.882152815054587e-05, "loss": 0.2084, "step": 3451 }, { "epoch": 21.442367601246104, "grad_norm": 3.0, "learning_rate": 4.882078051266318e-05, "loss": 0.2291, "step": 3452 }, { "epoch": 21.44859813084112, "grad_norm": 3.046875, "learning_rate": 4.88200326434283e-05, "loss": 0.3013, "step": 3453 }, { "epoch": 21.45482866043614, "grad_norm": 2.984375, "learning_rate": 4.8819284542848506e-05, "loss": 0.2878, "step": 3454 }, { "epoch": 21.461059190031154, "grad_norm": 3.6875, "learning_rate": 4.8818536210931064e-05, "loss": 0.3104, "step": 3455 }, { "epoch": 21.46728971962617, "grad_norm": 3.1875, "learning_rate": 4.881778764768323e-05, "loss": 0.2899, "step": 3456 }, { "epoch": 21.473520249221185, "grad_norm": 2.625, "learning_rate": 4.881703885311229e-05, "loss": 0.223, "step": 3457 }, { "epoch": 21.4797507788162, "grad_norm": 3.71875, "learning_rate": 4.88162898272255e-05, "loss": 0.2086, "step": 3458 }, { "epoch": 21.485981308411215, "grad_norm": 4.375, "learning_rate": 4.881554057003015e-05, "loss": 0.3687, "step": 3459 }, { "epoch": 21.49221183800623, "grad_norm": 2.828125, "learning_rate": 4.88147910815335e-05, "loss": 0.2305, "step": 3460 }, { "epoch": 21.498442367601246, "grad_norm": 3.171875, "learning_rate": 4.8814041361742845e-05, "loss": 0.2501, "step": 3461 }, { "epoch": 21.50467289719626, "grad_norm": 3.09375, "learning_rate": 4.881329141066546e-05, "loss": 0.3099, "step": 3462 }, { "epoch": 21.510903426791277, "grad_norm": 2.875, "learning_rate": 4.881254122830862e-05, "loss": 0.142, "step": 3463 }, { "epoch": 21.517133956386292, "grad_norm": 3.625, "learning_rate": 4.881179081467963e-05, "loss": 0.2883, "step": 3464 }, { "epoch": 21.523364485981308, "grad_norm": 2.8125, "learning_rate": 4.881104016978576e-05, "loss": 0.2491, "step": 3465 }, { "epoch": 21.529595015576323, "grad_norm": 2.96875, "learning_rate": 4.881028929363432e-05, "loss": 0.1992, "step": 3466 }, { "epoch": 21.53582554517134, "grad_norm": 3.796875, "learning_rate": 4.880953818623258e-05, "loss": 0.2001, "step": 3467 }, { "epoch": 21.542056074766354, "grad_norm": 5.28125, "learning_rate": 4.880878684758785e-05, "loss": 0.3121, "step": 3468 }, { "epoch": 21.54828660436137, "grad_norm": 5.71875, "learning_rate": 4.880803527770742e-05, "loss": 0.2717, "step": 3469 }, { "epoch": 21.554517133956388, "grad_norm": 4.78125, "learning_rate": 4.8807283476598595e-05, "loss": 0.4163, "step": 3470 }, { "epoch": 21.560747663551403, "grad_norm": 2.53125, "learning_rate": 4.8806531444268665e-05, "loss": 0.2333, "step": 3471 }, { "epoch": 21.56697819314642, "grad_norm": 2.0625, "learning_rate": 4.880577918072495e-05, "loss": 0.223, "step": 3472 }, { "epoch": 21.573208722741434, "grad_norm": 4.28125, "learning_rate": 4.880502668597475e-05, "loss": 0.3529, "step": 3473 }, { "epoch": 21.57943925233645, "grad_norm": 4.46875, "learning_rate": 4.8804273960025376e-05, "loss": 0.2861, "step": 3474 }, { "epoch": 21.585669781931465, "grad_norm": 3.515625, "learning_rate": 4.880352100288413e-05, "loss": 0.3058, "step": 3475 }, { "epoch": 21.59190031152648, "grad_norm": 3.0, "learning_rate": 4.880276781455833e-05, "loss": 0.2321, "step": 3476 }, { "epoch": 21.598130841121495, "grad_norm": 2.765625, "learning_rate": 4.880201439505529e-05, "loss": 0.2138, "step": 3477 }, { "epoch": 21.60436137071651, "grad_norm": 4.6875, "learning_rate": 4.8801260744382326e-05, "loss": 0.2472, "step": 3478 }, { "epoch": 21.610591900311526, "grad_norm": 3.8125, "learning_rate": 4.880050686254677e-05, "loss": 0.1485, "step": 3479 }, { "epoch": 21.61682242990654, "grad_norm": 2.296875, "learning_rate": 4.879975274955593e-05, "loss": 0.1532, "step": 3480 }, { "epoch": 21.623052959501557, "grad_norm": 3.8125, "learning_rate": 4.879899840541713e-05, "loss": 0.3092, "step": 3481 }, { "epoch": 21.629283489096572, "grad_norm": 3.5625, "learning_rate": 4.87982438301377e-05, "loss": 0.211, "step": 3482 }, { "epoch": 21.635514018691588, "grad_norm": 3.625, "learning_rate": 4.8797489023724964e-05, "loss": 0.2103, "step": 3483 }, { "epoch": 21.641744548286603, "grad_norm": 3.421875, "learning_rate": 4.879673398618627e-05, "loss": 0.2241, "step": 3484 }, { "epoch": 21.64797507788162, "grad_norm": 2.890625, "learning_rate": 4.8795978717528926e-05, "loss": 0.2175, "step": 3485 }, { "epoch": 21.654205607476637, "grad_norm": 3.703125, "learning_rate": 4.879522321776029e-05, "loss": 0.2334, "step": 3486 }, { "epoch": 21.660436137071652, "grad_norm": 2.0625, "learning_rate": 4.879446748688768e-05, "loss": 0.2306, "step": 3487 }, { "epoch": 21.666666666666668, "grad_norm": 2.484375, "learning_rate": 4.879371152491845e-05, "loss": 0.249, "step": 3488 }, { "epoch": 21.672897196261683, "grad_norm": 3.359375, "learning_rate": 4.879295533185995e-05, "loss": 0.2909, "step": 3489 }, { "epoch": 21.6791277258567, "grad_norm": 1.703125, "learning_rate": 4.879219890771949e-05, "loss": 0.1505, "step": 3490 }, { "epoch": 21.685358255451714, "grad_norm": 2.875, "learning_rate": 4.879144225250445e-05, "loss": 0.2685, "step": 3491 }, { "epoch": 21.69158878504673, "grad_norm": 3.453125, "learning_rate": 4.879068536622217e-05, "loss": 0.3771, "step": 3492 }, { "epoch": 21.697819314641745, "grad_norm": 3.28125, "learning_rate": 4.878992824887999e-05, "loss": 0.2577, "step": 3493 }, { "epoch": 21.70404984423676, "grad_norm": 2.90625, "learning_rate": 4.878917090048527e-05, "loss": 0.3102, "step": 3494 }, { "epoch": 21.710280373831775, "grad_norm": 3.34375, "learning_rate": 4.878841332104538e-05, "loss": 0.3254, "step": 3495 }, { "epoch": 21.71651090342679, "grad_norm": 1.5859375, "learning_rate": 4.878765551056765e-05, "loss": 0.1326, "step": 3496 }, { "epoch": 21.722741433021806, "grad_norm": 2.65625, "learning_rate": 4.878689746905946e-05, "loss": 0.1742, "step": 3497 }, { "epoch": 21.72897196261682, "grad_norm": 2.609375, "learning_rate": 4.878613919652817e-05, "loss": 0.256, "step": 3498 }, { "epoch": 21.735202492211837, "grad_norm": 2.609375, "learning_rate": 4.878538069298114e-05, "loss": 0.2801, "step": 3499 }, { "epoch": 21.741433021806852, "grad_norm": 3.09375, "learning_rate": 4.878462195842573e-05, "loss": 0.3501, "step": 3500 }, { "epoch": 21.747663551401867, "grad_norm": 2.640625, "learning_rate": 4.8783862992869336e-05, "loss": 0.219, "step": 3501 }, { "epoch": 21.753894080996886, "grad_norm": 2.140625, "learning_rate": 4.8783103796319294e-05, "loss": 0.1994, "step": 3502 }, { "epoch": 21.7601246105919, "grad_norm": 2.078125, "learning_rate": 4.8782344368782996e-05, "loss": 0.1456, "step": 3503 }, { "epoch": 21.766355140186917, "grad_norm": 2.734375, "learning_rate": 4.878158471026782e-05, "loss": 0.2526, "step": 3504 }, { "epoch": 21.772585669781932, "grad_norm": 2.3125, "learning_rate": 4.878082482078114e-05, "loss": 0.1971, "step": 3505 }, { "epoch": 21.778816199376948, "grad_norm": 2.390625, "learning_rate": 4.878006470033033e-05, "loss": 0.2872, "step": 3506 }, { "epoch": 21.785046728971963, "grad_norm": 2.3125, "learning_rate": 4.877930434892278e-05, "loss": 0.3292, "step": 3507 }, { "epoch": 21.79127725856698, "grad_norm": 2.40625, "learning_rate": 4.877854376656588e-05, "loss": 0.2431, "step": 3508 }, { "epoch": 21.797507788161994, "grad_norm": 2.59375, "learning_rate": 4.8777782953266996e-05, "loss": 0.2154, "step": 3509 }, { "epoch": 21.80373831775701, "grad_norm": 2.40625, "learning_rate": 4.877702190903354e-05, "loss": 0.2681, "step": 3510 }, { "epoch": 21.809968847352025, "grad_norm": 2.90625, "learning_rate": 4.87762606338729e-05, "loss": 0.2655, "step": 3511 }, { "epoch": 21.81619937694704, "grad_norm": 3.75, "learning_rate": 4.877549912779245e-05, "loss": 0.3904, "step": 3512 }, { "epoch": 21.822429906542055, "grad_norm": 2.015625, "learning_rate": 4.877473739079961e-05, "loss": 0.186, "step": 3513 }, { "epoch": 21.82866043613707, "grad_norm": 2.21875, "learning_rate": 4.877397542290176e-05, "loss": 0.3096, "step": 3514 }, { "epoch": 21.834890965732086, "grad_norm": 2.3125, "learning_rate": 4.8773213224106315e-05, "loss": 0.1414, "step": 3515 }, { "epoch": 21.8411214953271, "grad_norm": 1.890625, "learning_rate": 4.8772450794420666e-05, "loss": 0.1811, "step": 3516 }, { "epoch": 21.847352024922117, "grad_norm": 3.375, "learning_rate": 4.8771688133852225e-05, "loss": 0.2144, "step": 3517 }, { "epoch": 21.853582554517136, "grad_norm": 2.921875, "learning_rate": 4.8770925242408394e-05, "loss": 0.302, "step": 3518 }, { "epoch": 21.85981308411215, "grad_norm": 2.296875, "learning_rate": 4.87701621200966e-05, "loss": 0.2359, "step": 3519 }, { "epoch": 21.866043613707166, "grad_norm": 3.203125, "learning_rate": 4.8769398766924226e-05, "loss": 0.2048, "step": 3520 }, { "epoch": 21.87227414330218, "grad_norm": 2.78125, "learning_rate": 4.8768635182898694e-05, "loss": 0.2858, "step": 3521 }, { "epoch": 21.878504672897197, "grad_norm": 2.46875, "learning_rate": 4.876787136802744e-05, "loss": 0.218, "step": 3522 }, { "epoch": 21.884735202492212, "grad_norm": 3.796875, "learning_rate": 4.8767107322317865e-05, "loss": 0.498, "step": 3523 }, { "epoch": 21.890965732087228, "grad_norm": 1.71875, "learning_rate": 4.8766343045777386e-05, "loss": 0.1787, "step": 3524 }, { "epoch": 21.897196261682243, "grad_norm": 3.15625, "learning_rate": 4.8765578538413436e-05, "loss": 0.3539, "step": 3525 }, { "epoch": 21.90342679127726, "grad_norm": 3.171875, "learning_rate": 4.876481380023344e-05, "loss": 0.3104, "step": 3526 }, { "epoch": 21.909657320872274, "grad_norm": 1.9921875, "learning_rate": 4.8764048831244816e-05, "loss": 0.1992, "step": 3527 }, { "epoch": 21.91588785046729, "grad_norm": 2.5625, "learning_rate": 4.8763283631455007e-05, "loss": 0.2171, "step": 3528 }, { "epoch": 21.922118380062305, "grad_norm": 2.5, "learning_rate": 4.876251820087143e-05, "loss": 0.216, "step": 3529 }, { "epoch": 21.92834890965732, "grad_norm": 2.71875, "learning_rate": 4.876175253950153e-05, "loss": 0.3522, "step": 3530 }, { "epoch": 21.934579439252335, "grad_norm": 2.03125, "learning_rate": 4.8760986647352734e-05, "loss": 0.1933, "step": 3531 }, { "epoch": 21.94080996884735, "grad_norm": 4.03125, "learning_rate": 4.8760220524432497e-05, "loss": 0.2116, "step": 3532 }, { "epoch": 21.947040498442366, "grad_norm": 1.9765625, "learning_rate": 4.875945417074824e-05, "loss": 0.1903, "step": 3533 }, { "epoch": 21.953271028037385, "grad_norm": 2.921875, "learning_rate": 4.875868758630742e-05, "loss": 0.1981, "step": 3534 }, { "epoch": 21.9595015576324, "grad_norm": 4.8125, "learning_rate": 4.8757920771117475e-05, "loss": 0.2132, "step": 3535 }, { "epoch": 21.965732087227416, "grad_norm": 3.171875, "learning_rate": 4.875715372518585e-05, "loss": 0.2941, "step": 3536 }, { "epoch": 21.97196261682243, "grad_norm": 2.40625, "learning_rate": 4.875638644852e-05, "loss": 0.1911, "step": 3537 }, { "epoch": 21.978193146417446, "grad_norm": 4.34375, "learning_rate": 4.875561894112738e-05, "loss": 0.2409, "step": 3538 }, { "epoch": 21.98442367601246, "grad_norm": 4.1875, "learning_rate": 4.8754851203015436e-05, "loss": 0.2879, "step": 3539 }, { "epoch": 21.990654205607477, "grad_norm": 3.734375, "learning_rate": 4.8754083234191635e-05, "loss": 0.2563, "step": 3540 }, { "epoch": 21.996884735202492, "grad_norm": 2.28125, "learning_rate": 4.875331503466343e-05, "loss": 0.1747, "step": 3541 }, { "epoch": 22.0, "grad_norm": 2.125, "learning_rate": 4.875254660443827e-05, "loss": 0.2122, "step": 3542 }, { "epoch": 22.006230529595015, "grad_norm": 3.484375, "learning_rate": 4.8751777943523634e-05, "loss": 0.3462, "step": 3543 }, { "epoch": 22.01246105919003, "grad_norm": 2.34375, "learning_rate": 4.875100905192699e-05, "loss": 0.1499, "step": 3544 }, { "epoch": 22.018691588785046, "grad_norm": 3.375, "learning_rate": 4.875023992965579e-05, "loss": 0.3474, "step": 3545 }, { "epoch": 22.02492211838006, "grad_norm": 2.375, "learning_rate": 4.8749470576717516e-05, "loss": 0.2257, "step": 3546 }, { "epoch": 22.031152647975077, "grad_norm": 3.296875, "learning_rate": 4.874870099311963e-05, "loss": 0.3402, "step": 3547 }, { "epoch": 22.037383177570092, "grad_norm": 2.609375, "learning_rate": 4.874793117886962e-05, "loss": 0.2454, "step": 3548 }, { "epoch": 22.043613707165107, "grad_norm": 3.671875, "learning_rate": 4.874716113397496e-05, "loss": 0.2666, "step": 3549 }, { "epoch": 22.049844236760123, "grad_norm": 3.5625, "learning_rate": 4.8746390858443115e-05, "loss": 0.2864, "step": 3550 }, { "epoch": 22.05607476635514, "grad_norm": 2.65625, "learning_rate": 4.8745620352281576e-05, "loss": 0.3414, "step": 3551 }, { "epoch": 22.062305295950157, "grad_norm": 1.4765625, "learning_rate": 4.874484961549783e-05, "loss": 0.1446, "step": 3552 }, { "epoch": 22.068535825545172, "grad_norm": 2.78125, "learning_rate": 4.874407864809936e-05, "loss": 0.2686, "step": 3553 }, { "epoch": 22.074766355140188, "grad_norm": 2.8125, "learning_rate": 4.874330745009365e-05, "loss": 0.286, "step": 3554 }, { "epoch": 22.080996884735203, "grad_norm": 3.171875, "learning_rate": 4.8742536021488184e-05, "loss": 0.2962, "step": 3555 }, { "epoch": 22.08722741433022, "grad_norm": 3.0625, "learning_rate": 4.874176436229046e-05, "loss": 0.3718, "step": 3556 }, { "epoch": 22.093457943925234, "grad_norm": 1.46875, "learning_rate": 4.874099247250798e-05, "loss": 0.1427, "step": 3557 }, { "epoch": 22.09968847352025, "grad_norm": 1.359375, "learning_rate": 4.874022035214824e-05, "loss": 0.1463, "step": 3558 }, { "epoch": 22.105919003115265, "grad_norm": 2.859375, "learning_rate": 4.873944800121873e-05, "loss": 0.23, "step": 3559 }, { "epoch": 22.11214953271028, "grad_norm": 2.546875, "learning_rate": 4.8738675419726956e-05, "loss": 0.1764, "step": 3560 }, { "epoch": 22.118380062305295, "grad_norm": 3.53125, "learning_rate": 4.873790260768043e-05, "loss": 0.3899, "step": 3561 }, { "epoch": 22.12461059190031, "grad_norm": 2.59375, "learning_rate": 4.873712956508664e-05, "loss": 0.3126, "step": 3562 }, { "epoch": 22.130841121495326, "grad_norm": 3.640625, "learning_rate": 4.87363562919531e-05, "loss": 0.384, "step": 3563 }, { "epoch": 22.13707165109034, "grad_norm": 3.75, "learning_rate": 4.873558278828733e-05, "loss": 0.3091, "step": 3564 }, { "epoch": 22.143302180685357, "grad_norm": 3.84375, "learning_rate": 4.873480905409683e-05, "loss": 0.3469, "step": 3565 }, { "epoch": 22.149532710280372, "grad_norm": 2.0625, "learning_rate": 4.8734035089389115e-05, "loss": 0.1427, "step": 3566 }, { "epoch": 22.15576323987539, "grad_norm": 2.046875, "learning_rate": 4.873326089417172e-05, "loss": 0.2012, "step": 3567 }, { "epoch": 22.161993769470406, "grad_norm": 3.375, "learning_rate": 4.8732486468452144e-05, "loss": 0.2379, "step": 3568 }, { "epoch": 22.16822429906542, "grad_norm": 2.09375, "learning_rate": 4.8731711812237914e-05, "loss": 0.2165, "step": 3569 }, { "epoch": 22.174454828660437, "grad_norm": 2.375, "learning_rate": 4.873093692553656e-05, "loss": 0.239, "step": 3570 }, { "epoch": 22.180685358255452, "grad_norm": 2.6875, "learning_rate": 4.87301618083556e-05, "loss": 0.1541, "step": 3571 }, { "epoch": 22.186915887850468, "grad_norm": 2.859375, "learning_rate": 4.8729386460702565e-05, "loss": 0.2441, "step": 3572 }, { "epoch": 22.193146417445483, "grad_norm": 2.9375, "learning_rate": 4.872861088258499e-05, "loss": 0.1504, "step": 3573 }, { "epoch": 22.1993769470405, "grad_norm": 2.515625, "learning_rate": 4.872783507401039e-05, "loss": 0.1943, "step": 3574 }, { "epoch": 22.205607476635514, "grad_norm": 3.25, "learning_rate": 4.872705903498632e-05, "loss": 0.3805, "step": 3575 }, { "epoch": 22.21183800623053, "grad_norm": 1.7421875, "learning_rate": 4.8726282765520316e-05, "loss": 0.1553, "step": 3576 }, { "epoch": 22.218068535825545, "grad_norm": 2.328125, "learning_rate": 4.87255062656199e-05, "loss": 0.2942, "step": 3577 }, { "epoch": 22.22429906542056, "grad_norm": 2.359375, "learning_rate": 4.872472953529263e-05, "loss": 0.2356, "step": 3578 }, { "epoch": 22.230529595015575, "grad_norm": 3.234375, "learning_rate": 4.872395257454605e-05, "loss": 0.4242, "step": 3579 }, { "epoch": 22.23676012461059, "grad_norm": 1.4921875, "learning_rate": 4.8723175383387684e-05, "loss": 0.1518, "step": 3580 }, { "epoch": 22.242990654205606, "grad_norm": 3.46875, "learning_rate": 4.872239796182512e-05, "loss": 0.1994, "step": 3581 }, { "epoch": 22.24922118380062, "grad_norm": 4.5625, "learning_rate": 4.872162030986587e-05, "loss": 0.3901, "step": 3582 }, { "epoch": 22.25545171339564, "grad_norm": 3.96875, "learning_rate": 4.8720842427517496e-05, "loss": 0.3584, "step": 3583 }, { "epoch": 22.261682242990656, "grad_norm": 1.9765625, "learning_rate": 4.872006431478757e-05, "loss": 0.2951, "step": 3584 }, { "epoch": 22.26791277258567, "grad_norm": 3.0625, "learning_rate": 4.8719285971683636e-05, "loss": 0.2558, "step": 3585 }, { "epoch": 22.274143302180686, "grad_norm": 4.25, "learning_rate": 4.871850739821325e-05, "loss": 0.1748, "step": 3586 }, { "epoch": 22.2803738317757, "grad_norm": 3.046875, "learning_rate": 4.8717728594383984e-05, "loss": 0.2365, "step": 3587 }, { "epoch": 22.286604361370717, "grad_norm": 2.03125, "learning_rate": 4.8716949560203395e-05, "loss": 0.2217, "step": 3588 }, { "epoch": 22.292834890965732, "grad_norm": 2.625, "learning_rate": 4.8716170295679053e-05, "loss": 0.3566, "step": 3589 }, { "epoch": 22.299065420560748, "grad_norm": 3.0, "learning_rate": 4.8715390800818515e-05, "loss": 0.1684, "step": 3590 }, { "epoch": 22.305295950155763, "grad_norm": 3.484375, "learning_rate": 4.8714611075629376e-05, "loss": 0.3894, "step": 3591 }, { "epoch": 22.31152647975078, "grad_norm": 1.6328125, "learning_rate": 4.871383112011918e-05, "loss": 0.1529, "step": 3592 }, { "epoch": 22.317757009345794, "grad_norm": 2.265625, "learning_rate": 4.8713050934295524e-05, "loss": 0.2974, "step": 3593 }, { "epoch": 22.32398753894081, "grad_norm": 1.9921875, "learning_rate": 4.871227051816598e-05, "loss": 0.2858, "step": 3594 }, { "epoch": 22.330218068535824, "grad_norm": 3.59375, "learning_rate": 4.871148987173811e-05, "loss": 0.2649, "step": 3595 }, { "epoch": 22.33644859813084, "grad_norm": 2.96875, "learning_rate": 4.8710708995019527e-05, "loss": 0.2126, "step": 3596 }, { "epoch": 22.342679127725855, "grad_norm": 2.171875, "learning_rate": 4.8709927888017795e-05, "loss": 0.1857, "step": 3597 }, { "epoch": 22.34890965732087, "grad_norm": 2.671875, "learning_rate": 4.87091465507405e-05, "loss": 0.2149, "step": 3598 }, { "epoch": 22.35514018691589, "grad_norm": 4.09375, "learning_rate": 4.870836498319523e-05, "loss": 0.2945, "step": 3599 }, { "epoch": 22.361370716510905, "grad_norm": 2.53125, "learning_rate": 4.870758318538959e-05, "loss": 0.1729, "step": 3600 }, { "epoch": 22.36760124610592, "grad_norm": 3.0, "learning_rate": 4.8706801157331154e-05, "loss": 0.3302, "step": 3601 }, { "epoch": 22.373831775700936, "grad_norm": 4.34375, "learning_rate": 4.870601889902753e-05, "loss": 0.441, "step": 3602 }, { "epoch": 22.38006230529595, "grad_norm": 2.734375, "learning_rate": 4.870523641048631e-05, "loss": 0.2198, "step": 3603 }, { "epoch": 22.386292834890966, "grad_norm": 2.84375, "learning_rate": 4.87044536917151e-05, "loss": 0.3091, "step": 3604 }, { "epoch": 22.39252336448598, "grad_norm": 2.875, "learning_rate": 4.8703670742721494e-05, "loss": 0.2488, "step": 3605 }, { "epoch": 22.398753894080997, "grad_norm": 3.0625, "learning_rate": 4.87028875635131e-05, "loss": 0.1841, "step": 3606 }, { "epoch": 22.404984423676012, "grad_norm": 5.09375, "learning_rate": 4.8702104154097526e-05, "loss": 0.3635, "step": 3607 }, { "epoch": 22.411214953271028, "grad_norm": 2.515625, "learning_rate": 4.870132051448237e-05, "loss": 0.1285, "step": 3608 }, { "epoch": 22.417445482866043, "grad_norm": 3.296875, "learning_rate": 4.870053664467526e-05, "loss": 0.2058, "step": 3609 }, { "epoch": 22.42367601246106, "grad_norm": 3.625, "learning_rate": 4.8699752544683804e-05, "loss": 0.1513, "step": 3610 }, { "epoch": 22.429906542056074, "grad_norm": 2.265625, "learning_rate": 4.869896821451561e-05, "loss": 0.2658, "step": 3611 }, { "epoch": 22.43613707165109, "grad_norm": 1.828125, "learning_rate": 4.86981836541783e-05, "loss": 0.1673, "step": 3612 }, { "epoch": 22.442367601246104, "grad_norm": 3.859375, "learning_rate": 4.869739886367949e-05, "loss": 0.1427, "step": 3613 }, { "epoch": 22.44859813084112, "grad_norm": 4.125, "learning_rate": 4.8696613843026806e-05, "loss": 0.1896, "step": 3614 }, { "epoch": 22.45482866043614, "grad_norm": 3.5, "learning_rate": 4.8695828592227865e-05, "loss": 0.4158, "step": 3615 }, { "epoch": 22.461059190031154, "grad_norm": 2.203125, "learning_rate": 4.869504311129031e-05, "loss": 0.2116, "step": 3616 }, { "epoch": 22.46728971962617, "grad_norm": 4.40625, "learning_rate": 4.8694257400221754e-05, "loss": 0.215, "step": 3617 }, { "epoch": 22.473520249221185, "grad_norm": 5.1875, "learning_rate": 4.869347145902984e-05, "loss": 0.2393, "step": 3618 }, { "epoch": 22.4797507788162, "grad_norm": 4.0625, "learning_rate": 4.869268528772219e-05, "loss": 0.3052, "step": 3619 }, { "epoch": 22.485981308411215, "grad_norm": 2.546875, "learning_rate": 4.8691898886306445e-05, "loss": 0.1894, "step": 3620 }, { "epoch": 22.49221183800623, "grad_norm": 5.9375, "learning_rate": 4.8691112254790246e-05, "loss": 0.4056, "step": 3621 }, { "epoch": 22.498442367601246, "grad_norm": 4.8125, "learning_rate": 4.869032539318123e-05, "loss": 0.2178, "step": 3622 }, { "epoch": 22.50467289719626, "grad_norm": 4.84375, "learning_rate": 4.8689538301487025e-05, "loss": 0.16, "step": 3623 }, { "epoch": 22.510903426791277, "grad_norm": 3.296875, "learning_rate": 4.86887509797153e-05, "loss": 0.2684, "step": 3624 }, { "epoch": 22.517133956386292, "grad_norm": 2.53125, "learning_rate": 4.868796342787368e-05, "loss": 0.26, "step": 3625 }, { "epoch": 22.523364485981308, "grad_norm": 2.734375, "learning_rate": 4.868717564596984e-05, "loss": 0.2388, "step": 3626 }, { "epoch": 22.529595015576323, "grad_norm": 4.125, "learning_rate": 4.8686387634011405e-05, "loss": 0.285, "step": 3627 }, { "epoch": 22.53582554517134, "grad_norm": 5.125, "learning_rate": 4.8685599392006046e-05, "loss": 0.232, "step": 3628 }, { "epoch": 22.542056074766354, "grad_norm": 3.515625, "learning_rate": 4.86848109199614e-05, "loss": 0.265, "step": 3629 }, { "epoch": 22.54828660436137, "grad_norm": 3.328125, "learning_rate": 4.868402221788514e-05, "loss": 0.2908, "step": 3630 }, { "epoch": 22.554517133956388, "grad_norm": 3.40625, "learning_rate": 4.8683233285784924e-05, "loss": 0.2741, "step": 3631 }, { "epoch": 22.560747663551403, "grad_norm": 4.96875, "learning_rate": 4.8682444123668417e-05, "loss": 0.2705, "step": 3632 }, { "epoch": 22.56697819314642, "grad_norm": 3.75, "learning_rate": 4.868165473154327e-05, "loss": 0.3651, "step": 3633 }, { "epoch": 22.573208722741434, "grad_norm": 3.515625, "learning_rate": 4.868086510941716e-05, "loss": 0.3328, "step": 3634 }, { "epoch": 22.57943925233645, "grad_norm": 2.265625, "learning_rate": 4.868007525729775e-05, "loss": 0.1943, "step": 3635 }, { "epoch": 22.585669781931465, "grad_norm": 2.453125, "learning_rate": 4.8679285175192725e-05, "loss": 0.1857, "step": 3636 }, { "epoch": 22.59190031152648, "grad_norm": 2.8125, "learning_rate": 4.867849486310975e-05, "loss": 0.2408, "step": 3637 }, { "epoch": 22.598130841121495, "grad_norm": 2.34375, "learning_rate": 4.8677704321056496e-05, "loss": 0.1564, "step": 3638 }, { "epoch": 22.60436137071651, "grad_norm": 3.234375, "learning_rate": 4.8676913549040634e-05, "loss": 0.2964, "step": 3639 }, { "epoch": 22.610591900311526, "grad_norm": 2.296875, "learning_rate": 4.867612254706987e-05, "loss": 0.1851, "step": 3640 }, { "epoch": 22.61682242990654, "grad_norm": 2.390625, "learning_rate": 4.8675331315151865e-05, "loss": 0.1748, "step": 3641 }, { "epoch": 22.623052959501557, "grad_norm": 3.203125, "learning_rate": 4.867453985329431e-05, "loss": 0.2627, "step": 3642 }, { "epoch": 22.629283489096572, "grad_norm": 2.140625, "learning_rate": 4.867374816150489e-05, "loss": 0.1774, "step": 3643 }, { "epoch": 22.635514018691588, "grad_norm": 2.453125, "learning_rate": 4.867295623979129e-05, "loss": 0.1535, "step": 3644 }, { "epoch": 22.641744548286603, "grad_norm": 2.40625, "learning_rate": 4.867216408816122e-05, "loss": 0.1751, "step": 3645 }, { "epoch": 22.64797507788162, "grad_norm": 3.15625, "learning_rate": 4.867137170662235e-05, "loss": 0.1859, "step": 3646 }, { "epoch": 22.654205607476637, "grad_norm": 4.0625, "learning_rate": 4.86705790951824e-05, "loss": 0.2256, "step": 3647 }, { "epoch": 22.660436137071652, "grad_norm": 3.25, "learning_rate": 4.866978625384904e-05, "loss": 0.2348, "step": 3648 }, { "epoch": 22.666666666666668, "grad_norm": 2.3125, "learning_rate": 4.8668993182629985e-05, "loss": 0.1554, "step": 3649 }, { "epoch": 22.672897196261683, "grad_norm": 3.15625, "learning_rate": 4.866819988153294e-05, "loss": 0.1808, "step": 3650 }, { "epoch": 22.6791277258567, "grad_norm": 3.0625, "learning_rate": 4.8667406350565616e-05, "loss": 0.3271, "step": 3651 }, { "epoch": 22.685358255451714, "grad_norm": 2.9375, "learning_rate": 4.86666125897357e-05, "loss": 0.3034, "step": 3652 }, { "epoch": 22.69158878504673, "grad_norm": 2.8125, "learning_rate": 4.8665818599050906e-05, "loss": 0.2493, "step": 3653 }, { "epoch": 22.697819314641745, "grad_norm": 4.09375, "learning_rate": 4.866502437851896e-05, "loss": 0.3043, "step": 3654 }, { "epoch": 22.70404984423676, "grad_norm": 2.171875, "learning_rate": 4.866422992814757e-05, "loss": 0.2363, "step": 3655 }, { "epoch": 22.710280373831775, "grad_norm": 2.71875, "learning_rate": 4.866343524794444e-05, "loss": 0.2641, "step": 3656 }, { "epoch": 22.71651090342679, "grad_norm": 2.4375, "learning_rate": 4.8662640337917296e-05, "loss": 0.2174, "step": 3657 }, { "epoch": 22.722741433021806, "grad_norm": 5.5, "learning_rate": 4.866184519807387e-05, "loss": 0.2634, "step": 3658 }, { "epoch": 22.72897196261682, "grad_norm": 4.15625, "learning_rate": 4.8661049828421866e-05, "loss": 0.3661, "step": 3659 }, { "epoch": 22.735202492211837, "grad_norm": 3.1875, "learning_rate": 4.866025422896902e-05, "loss": 0.2599, "step": 3660 }, { "epoch": 22.741433021806852, "grad_norm": 2.4375, "learning_rate": 4.865945839972304e-05, "loss": 0.2155, "step": 3661 }, { "epoch": 22.747663551401867, "grad_norm": 3.15625, "learning_rate": 4.865866234069169e-05, "loss": 0.2839, "step": 3662 }, { "epoch": 22.753894080996886, "grad_norm": 3.578125, "learning_rate": 4.865786605188268e-05, "loss": 0.3206, "step": 3663 }, { "epoch": 22.7601246105919, "grad_norm": 5.4375, "learning_rate": 4.865706953330374e-05, "loss": 0.3327, "step": 3664 }, { "epoch": 22.766355140186917, "grad_norm": 2.609375, "learning_rate": 4.865627278496261e-05, "loss": 0.1987, "step": 3665 }, { "epoch": 22.772585669781932, "grad_norm": 2.09375, "learning_rate": 4.865547580686703e-05, "loss": 0.2249, "step": 3666 }, { "epoch": 22.778816199376948, "grad_norm": 2.765625, "learning_rate": 4.865467859902475e-05, "loss": 0.2251, "step": 3667 }, { "epoch": 22.785046728971963, "grad_norm": 2.65625, "learning_rate": 4.865388116144349e-05, "loss": 0.1761, "step": 3668 }, { "epoch": 22.79127725856698, "grad_norm": 1.921875, "learning_rate": 4.865308349413101e-05, "loss": 0.1524, "step": 3669 }, { "epoch": 22.797507788161994, "grad_norm": 2.90625, "learning_rate": 4.865228559709507e-05, "loss": 0.2382, "step": 3670 }, { "epoch": 22.80373831775701, "grad_norm": 2.609375, "learning_rate": 4.865148747034339e-05, "loss": 0.2263, "step": 3671 }, { "epoch": 22.809968847352025, "grad_norm": 2.984375, "learning_rate": 4.8650689113883744e-05, "loss": 0.3107, "step": 3672 }, { "epoch": 22.81619937694704, "grad_norm": 2.015625, "learning_rate": 4.864989052772386e-05, "loss": 0.2019, "step": 3673 }, { "epoch": 22.822429906542055, "grad_norm": 2.625, "learning_rate": 4.864909171187153e-05, "loss": 0.241, "step": 3674 }, { "epoch": 22.82866043613707, "grad_norm": 3.46875, "learning_rate": 4.864829266633449e-05, "loss": 0.2874, "step": 3675 }, { "epoch": 22.834890965732086, "grad_norm": 1.953125, "learning_rate": 4.8647493391120505e-05, "loss": 0.191, "step": 3676 }, { "epoch": 22.8411214953271, "grad_norm": 2.71875, "learning_rate": 4.864669388623733e-05, "loss": 0.2333, "step": 3677 }, { "epoch": 22.847352024922117, "grad_norm": 4.34375, "learning_rate": 4.864589415169274e-05, "loss": 0.2381, "step": 3678 }, { "epoch": 22.853582554517136, "grad_norm": 3.6875, "learning_rate": 4.86450941874945e-05, "loss": 0.2298, "step": 3679 }, { "epoch": 22.85981308411215, "grad_norm": 1.875, "learning_rate": 4.864429399365038e-05, "loss": 0.167, "step": 3680 }, { "epoch": 22.866043613707166, "grad_norm": 3.640625, "learning_rate": 4.864349357016815e-05, "loss": 0.3189, "step": 3681 }, { "epoch": 22.87227414330218, "grad_norm": 2.84375, "learning_rate": 4.864269291705559e-05, "loss": 0.212, "step": 3682 }, { "epoch": 22.878504672897197, "grad_norm": 2.71875, "learning_rate": 4.8641892034320465e-05, "loss": 0.2046, "step": 3683 }, { "epoch": 22.884735202492212, "grad_norm": 2.3125, "learning_rate": 4.864109092197056e-05, "loss": 0.1351, "step": 3684 }, { "epoch": 22.890965732087228, "grad_norm": 2.953125, "learning_rate": 4.864028958001365e-05, "loss": 0.21, "step": 3685 }, { "epoch": 22.897196261682243, "grad_norm": 4.03125, "learning_rate": 4.8639488008457524e-05, "loss": 0.2771, "step": 3686 }, { "epoch": 22.90342679127726, "grad_norm": 3.828125, "learning_rate": 4.8638686207309966e-05, "loss": 0.2772, "step": 3687 }, { "epoch": 22.909657320872274, "grad_norm": 5.40625, "learning_rate": 4.863788417657876e-05, "loss": 0.2687, "step": 3688 }, { "epoch": 22.91588785046729, "grad_norm": 2.71875, "learning_rate": 4.86370819162717e-05, "loss": 0.1836, "step": 3689 }, { "epoch": 22.922118380062305, "grad_norm": 3.265625, "learning_rate": 4.863627942639657e-05, "loss": 0.2225, "step": 3690 }, { "epoch": 22.92834890965732, "grad_norm": 3.90625, "learning_rate": 4.863547670696118e-05, "loss": 0.2703, "step": 3691 }, { "epoch": 22.934579439252335, "grad_norm": 3.890625, "learning_rate": 4.863467375797331e-05, "loss": 0.223, "step": 3692 }, { "epoch": 22.94080996884735, "grad_norm": 7.78125, "learning_rate": 4.8633870579440765e-05, "loss": 0.3165, "step": 3693 }, { "epoch": 22.947040498442366, "grad_norm": 4.125, "learning_rate": 4.8633067171371337e-05, "loss": 0.2271, "step": 3694 }, { "epoch": 22.953271028037385, "grad_norm": 3.0625, "learning_rate": 4.8632263533772846e-05, "loss": 0.3557, "step": 3695 }, { "epoch": 22.9595015576324, "grad_norm": 3.953125, "learning_rate": 4.863145966665309e-05, "loss": 0.2372, "step": 3696 }, { "epoch": 22.965732087227416, "grad_norm": 4.875, "learning_rate": 4.863065557001987e-05, "loss": 0.2318, "step": 3697 }, { "epoch": 22.97196261682243, "grad_norm": 5.875, "learning_rate": 4.8629851243881e-05, "loss": 0.2397, "step": 3698 }, { "epoch": 22.978193146417446, "grad_norm": 3.734375, "learning_rate": 4.8629046688244286e-05, "loss": 0.1747, "step": 3699 }, { "epoch": 22.98442367601246, "grad_norm": 3.84375, "learning_rate": 4.862824190311754e-05, "loss": 0.256, "step": 3700 }, { "epoch": 22.990654205607477, "grad_norm": 4.28125, "learning_rate": 4.8627436888508606e-05, "loss": 0.2206, "step": 3701 }, { "epoch": 22.996884735202492, "grad_norm": 5.25, "learning_rate": 4.8626631644425266e-05, "loss": 0.2593, "step": 3702 }, { "epoch": 23.0, "grad_norm": 2.34375, "learning_rate": 4.862582617087537e-05, "loss": 0.2306, "step": 3703 }, { "epoch": 23.006230529595015, "grad_norm": 1.8515625, "learning_rate": 4.862502046786671e-05, "loss": 0.1898, "step": 3704 }, { "epoch": 23.01246105919003, "grad_norm": 2.234375, "learning_rate": 4.862421453540714e-05, "loss": 0.2238, "step": 3705 }, { "epoch": 23.018691588785046, "grad_norm": 2.34375, "learning_rate": 4.862340837350448e-05, "loss": 0.225, "step": 3706 }, { "epoch": 23.02492211838006, "grad_norm": 4.53125, "learning_rate": 4.862260198216655e-05, "loss": 0.2938, "step": 3707 }, { "epoch": 23.031152647975077, "grad_norm": 2.609375, "learning_rate": 4.8621795361401185e-05, "loss": 0.1911, "step": 3708 }, { "epoch": 23.037383177570092, "grad_norm": 2.25, "learning_rate": 4.862098851121623e-05, "loss": 0.2496, "step": 3709 }, { "epoch": 23.043613707165107, "grad_norm": 2.40625, "learning_rate": 4.86201814316195e-05, "loss": 0.1773, "step": 3710 }, { "epoch": 23.049844236760123, "grad_norm": 3.328125, "learning_rate": 4.8619374122618854e-05, "loss": 0.3133, "step": 3711 }, { "epoch": 23.05607476635514, "grad_norm": 2.5625, "learning_rate": 4.8618566584222127e-05, "loss": 0.3374, "step": 3712 }, { "epoch": 23.062305295950157, "grad_norm": 2.171875, "learning_rate": 4.8617758816437154e-05, "loss": 0.189, "step": 3713 }, { "epoch": 23.068535825545172, "grad_norm": 2.578125, "learning_rate": 4.861695081927179e-05, "loss": 0.1568, "step": 3714 }, { "epoch": 23.074766355140188, "grad_norm": 3.328125, "learning_rate": 4.861614259273388e-05, "loss": 0.2423, "step": 3715 }, { "epoch": 23.080996884735203, "grad_norm": 3.125, "learning_rate": 4.861533413683127e-05, "loss": 0.2039, "step": 3716 }, { "epoch": 23.08722741433022, "grad_norm": 3.28125, "learning_rate": 4.861452545157182e-05, "loss": 0.3428, "step": 3717 }, { "epoch": 23.093457943925234, "grad_norm": 3.0, "learning_rate": 4.861371653696337e-05, "loss": 0.2328, "step": 3718 }, { "epoch": 23.09968847352025, "grad_norm": 2.9375, "learning_rate": 4.861290739301379e-05, "loss": 0.2251, "step": 3719 }, { "epoch": 23.105919003115265, "grad_norm": 4.0625, "learning_rate": 4.861209801973093e-05, "loss": 0.4592, "step": 3720 }, { "epoch": 23.11214953271028, "grad_norm": 2.921875, "learning_rate": 4.8611288417122656e-05, "loss": 0.2358, "step": 3721 }, { "epoch": 23.118380062305295, "grad_norm": 3.265625, "learning_rate": 4.8610478585196826e-05, "loss": 0.4039, "step": 3722 }, { "epoch": 23.12461059190031, "grad_norm": 4.5625, "learning_rate": 4.8609668523961316e-05, "loss": 0.3638, "step": 3723 }, { "epoch": 23.130841121495326, "grad_norm": 3.84375, "learning_rate": 4.8608858233423984e-05, "loss": 0.3132, "step": 3724 }, { "epoch": 23.13707165109034, "grad_norm": 3.78125, "learning_rate": 4.86080477135927e-05, "loss": 0.2967, "step": 3725 }, { "epoch": 23.143302180685357, "grad_norm": 3.15625, "learning_rate": 4.8607236964475335e-05, "loss": 0.2699, "step": 3726 }, { "epoch": 23.149532710280372, "grad_norm": 2.453125, "learning_rate": 4.860642598607976e-05, "loss": 0.2804, "step": 3727 }, { "epoch": 23.15576323987539, "grad_norm": 2.859375, "learning_rate": 4.8605614778413866e-05, "loss": 0.2181, "step": 3728 }, { "epoch": 23.161993769470406, "grad_norm": 3.5, "learning_rate": 4.8604803341485514e-05, "loss": 0.2902, "step": 3729 }, { "epoch": 23.16822429906542, "grad_norm": 2.921875, "learning_rate": 4.860399167530261e-05, "loss": 0.222, "step": 3730 }, { "epoch": 23.174454828660437, "grad_norm": 2.703125, "learning_rate": 4.8603179779873004e-05, "loss": 0.2197, "step": 3731 }, { "epoch": 23.180685358255452, "grad_norm": 4.25, "learning_rate": 4.860236765520461e-05, "loss": 0.3739, "step": 3732 }, { "epoch": 23.186915887850468, "grad_norm": 2.828125, "learning_rate": 4.860155530130529e-05, "loss": 0.2083, "step": 3733 }, { "epoch": 23.193146417445483, "grad_norm": 2.890625, "learning_rate": 4.860074271818296e-05, "loss": 0.2522, "step": 3734 }, { "epoch": 23.1993769470405, "grad_norm": 2.390625, "learning_rate": 4.859992990584549e-05, "loss": 0.1941, "step": 3735 }, { "epoch": 23.205607476635514, "grad_norm": 2.4375, "learning_rate": 4.859911686430079e-05, "loss": 0.1888, "step": 3736 }, { "epoch": 23.21183800623053, "grad_norm": 3.53125, "learning_rate": 4.859830359355674e-05, "loss": 0.3428, "step": 3737 }, { "epoch": 23.218068535825545, "grad_norm": 2.078125, "learning_rate": 4.8597490093621253e-05, "loss": 0.1771, "step": 3738 }, { "epoch": 23.22429906542056, "grad_norm": 3.234375, "learning_rate": 4.859667636450223e-05, "loss": 0.4278, "step": 3739 }, { "epoch": 23.230529595015575, "grad_norm": 3.96875, "learning_rate": 4.859586240620757e-05, "loss": 0.2676, "step": 3740 }, { "epoch": 23.23676012461059, "grad_norm": 2.203125, "learning_rate": 4.8595048218745164e-05, "loss": 0.1552, "step": 3741 }, { "epoch": 23.242990654205606, "grad_norm": 2.953125, "learning_rate": 4.859423380212295e-05, "loss": 0.3059, "step": 3742 }, { "epoch": 23.24922118380062, "grad_norm": 1.984375, "learning_rate": 4.8593419156348804e-05, "loss": 0.1425, "step": 3743 }, { "epoch": 23.25545171339564, "grad_norm": 2.390625, "learning_rate": 4.859260428143067e-05, "loss": 0.1683, "step": 3744 }, { "epoch": 23.261682242990656, "grad_norm": 3.640625, "learning_rate": 4.8591789177376446e-05, "loss": 0.2518, "step": 3745 }, { "epoch": 23.26791277258567, "grad_norm": 4.3125, "learning_rate": 4.859097384419404e-05, "loss": 0.2936, "step": 3746 }, { "epoch": 23.274143302180686, "grad_norm": 3.90625, "learning_rate": 4.8590158281891385e-05, "loss": 0.3042, "step": 3747 }, { "epoch": 23.2803738317757, "grad_norm": 2.875, "learning_rate": 4.8589342490476394e-05, "loss": 0.2371, "step": 3748 }, { "epoch": 23.286604361370717, "grad_norm": 4.75, "learning_rate": 4.8588526469957e-05, "loss": 0.3499, "step": 3749 }, { "epoch": 23.292834890965732, "grad_norm": 2.21875, "learning_rate": 4.858771022034112e-05, "loss": 0.1623, "step": 3750 }, { "epoch": 23.299065420560748, "grad_norm": 3.3125, "learning_rate": 4.8586893741636687e-05, "loss": 0.2338, "step": 3751 }, { "epoch": 23.305295950155763, "grad_norm": 2.390625, "learning_rate": 4.858607703385162e-05, "loss": 0.2798, "step": 3752 }, { "epoch": 23.31152647975078, "grad_norm": 3.25, "learning_rate": 4.858526009699386e-05, "loss": 0.4853, "step": 3753 }, { "epoch": 23.317757009345794, "grad_norm": 3.796875, "learning_rate": 4.858444293107134e-05, "loss": 0.2095, "step": 3754 }, { "epoch": 23.32398753894081, "grad_norm": 4.0625, "learning_rate": 4.858362553609199e-05, "loss": 0.2923, "step": 3755 }, { "epoch": 23.330218068535824, "grad_norm": 3.75, "learning_rate": 4.8582807912063764e-05, "loss": 0.1661, "step": 3756 }, { "epoch": 23.33644859813084, "grad_norm": 3.453125, "learning_rate": 4.858199005899459e-05, "loss": 0.2667, "step": 3757 }, { "epoch": 23.342679127725855, "grad_norm": 2.421875, "learning_rate": 4.858117197689242e-05, "loss": 0.1511, "step": 3758 }, { "epoch": 23.34890965732087, "grad_norm": 5.4375, "learning_rate": 4.858035366576518e-05, "loss": 0.1955, "step": 3759 }, { "epoch": 23.35514018691589, "grad_norm": 5.5625, "learning_rate": 4.857953512562084e-05, "loss": 0.3204, "step": 3760 }, { "epoch": 23.361370716510905, "grad_norm": 4.21875, "learning_rate": 4.857871635646734e-05, "loss": 0.2818, "step": 3761 }, { "epoch": 23.36760124610592, "grad_norm": 3.53125, "learning_rate": 4.857789735831263e-05, "loss": 0.1795, "step": 3762 }, { "epoch": 23.373831775700936, "grad_norm": 3.015625, "learning_rate": 4.857707813116468e-05, "loss": 0.182, "step": 3763 }, { "epoch": 23.38006230529595, "grad_norm": 4.0625, "learning_rate": 4.8576258675031424e-05, "loss": 0.2389, "step": 3764 }, { "epoch": 23.386292834890966, "grad_norm": 4.34375, "learning_rate": 4.857543898992083e-05, "loss": 0.2335, "step": 3765 }, { "epoch": 23.39252336448598, "grad_norm": 3.375, "learning_rate": 4.857461907584086e-05, "loss": 0.2177, "step": 3766 }, { "epoch": 23.398753894080997, "grad_norm": 2.5, "learning_rate": 4.8573798932799485e-05, "loss": 0.1798, "step": 3767 }, { "epoch": 23.404984423676012, "grad_norm": 3.15625, "learning_rate": 4.8572978560804655e-05, "loss": 0.2615, "step": 3768 }, { "epoch": 23.411214953271028, "grad_norm": 3.8125, "learning_rate": 4.857215795986435e-05, "loss": 0.2684, "step": 3769 }, { "epoch": 23.417445482866043, "grad_norm": 2.625, "learning_rate": 4.8571337129986525e-05, "loss": 0.1769, "step": 3770 }, { "epoch": 23.42367601246106, "grad_norm": 3.1875, "learning_rate": 4.8570516071179176e-05, "loss": 0.2048, "step": 3771 }, { "epoch": 23.429906542056074, "grad_norm": 2.1875, "learning_rate": 4.856969478345026e-05, "loss": 0.179, "step": 3772 }, { "epoch": 23.43613707165109, "grad_norm": 1.3984375, "learning_rate": 4.856887326680774e-05, "loss": 0.1612, "step": 3773 }, { "epoch": 23.442367601246104, "grad_norm": 3.515625, "learning_rate": 4.856805152125963e-05, "loss": 0.2223, "step": 3774 }, { "epoch": 23.44859813084112, "grad_norm": 1.546875, "learning_rate": 4.8567229546813894e-05, "loss": 0.1551, "step": 3775 }, { "epoch": 23.45482866043614, "grad_norm": 3.53125, "learning_rate": 4.85664073434785e-05, "loss": 0.2864, "step": 3776 }, { "epoch": 23.461059190031154, "grad_norm": 4.96875, "learning_rate": 4.856558491126146e-05, "loss": 0.1646, "step": 3777 }, { "epoch": 23.46728971962617, "grad_norm": 2.453125, "learning_rate": 4.856476225017074e-05, "loss": 0.1534, "step": 3778 }, { "epoch": 23.473520249221185, "grad_norm": 3.015625, "learning_rate": 4.8563939360214345e-05, "loss": 0.2323, "step": 3779 }, { "epoch": 23.4797507788162, "grad_norm": 2.359375, "learning_rate": 4.856311624140025e-05, "loss": 0.2755, "step": 3780 }, { "epoch": 23.485981308411215, "grad_norm": 4.15625, "learning_rate": 4.856229289373647e-05, "loss": 0.2613, "step": 3781 }, { "epoch": 23.49221183800623, "grad_norm": 2.625, "learning_rate": 4.856146931723099e-05, "loss": 0.1457, "step": 3782 }, { "epoch": 23.498442367601246, "grad_norm": 1.9296875, "learning_rate": 4.856064551189181e-05, "loss": 0.1526, "step": 3783 }, { "epoch": 23.50467289719626, "grad_norm": 2.453125, "learning_rate": 4.855982147772693e-05, "loss": 0.258, "step": 3784 }, { "epoch": 23.510903426791277, "grad_norm": 3.375, "learning_rate": 4.855899721474435e-05, "loss": 0.3029, "step": 3785 }, { "epoch": 23.517133956386292, "grad_norm": 1.7890625, "learning_rate": 4.855817272295209e-05, "loss": 0.1822, "step": 3786 }, { "epoch": 23.523364485981308, "grad_norm": 4.1875, "learning_rate": 4.855734800235814e-05, "loss": 0.2484, "step": 3787 }, { "epoch": 23.529595015576323, "grad_norm": 2.5, "learning_rate": 4.855652305297052e-05, "loss": 0.2392, "step": 3788 }, { "epoch": 23.53582554517134, "grad_norm": 2.09375, "learning_rate": 4.8555697874797236e-05, "loss": 0.1552, "step": 3789 }, { "epoch": 23.542056074766354, "grad_norm": 2.671875, "learning_rate": 4.8554872467846304e-05, "loss": 0.2612, "step": 3790 }, { "epoch": 23.54828660436137, "grad_norm": 2.015625, "learning_rate": 4.8554046832125746e-05, "loss": 0.1257, "step": 3791 }, { "epoch": 23.554517133956388, "grad_norm": 2.28125, "learning_rate": 4.855322096764358e-05, "loss": 0.172, "step": 3792 }, { "epoch": 23.560747663551403, "grad_norm": 2.46875, "learning_rate": 4.8552394874407815e-05, "loss": 0.3605, "step": 3793 }, { "epoch": 23.56697819314642, "grad_norm": 3.1875, "learning_rate": 4.855156855242648e-05, "loss": 0.246, "step": 3794 }, { "epoch": 23.573208722741434, "grad_norm": 3.09375, "learning_rate": 4.8550742001707616e-05, "loss": 0.2805, "step": 3795 }, { "epoch": 23.57943925233645, "grad_norm": 2.390625, "learning_rate": 4.854991522225923e-05, "loss": 0.1631, "step": 3796 }, { "epoch": 23.585669781931465, "grad_norm": 3.109375, "learning_rate": 4.8549088214089366e-05, "loss": 0.1756, "step": 3797 }, { "epoch": 23.59190031152648, "grad_norm": 3.625, "learning_rate": 4.854826097720604e-05, "loss": 0.2168, "step": 3798 }, { "epoch": 23.598130841121495, "grad_norm": 3.40625, "learning_rate": 4.8547433511617305e-05, "loss": 0.1934, "step": 3799 }, { "epoch": 23.60436137071651, "grad_norm": 4.09375, "learning_rate": 4.8546605817331196e-05, "loss": 0.4304, "step": 3800 }, { "epoch": 23.610591900311526, "grad_norm": 2.9375, "learning_rate": 4.8545777894355725e-05, "loss": 0.33, "step": 3801 }, { "epoch": 23.61682242990654, "grad_norm": 2.796875, "learning_rate": 4.854494974269896e-05, "loss": 0.2995, "step": 3802 }, { "epoch": 23.623052959501557, "grad_norm": 2.5, "learning_rate": 4.854412136236894e-05, "loss": 0.2689, "step": 3803 }, { "epoch": 23.629283489096572, "grad_norm": 3.859375, "learning_rate": 4.854329275337371e-05, "loss": 0.3302, "step": 3804 }, { "epoch": 23.635514018691588, "grad_norm": 3.65625, "learning_rate": 4.854246391572131e-05, "loss": 0.2732, "step": 3805 }, { "epoch": 23.641744548286603, "grad_norm": 2.15625, "learning_rate": 4.8541634849419794e-05, "loss": 0.1856, "step": 3806 }, { "epoch": 23.64797507788162, "grad_norm": 4.1875, "learning_rate": 4.854080555447722e-05, "loss": 0.3511, "step": 3807 }, { "epoch": 23.654205607476637, "grad_norm": 5.71875, "learning_rate": 4.853997603090163e-05, "loss": 0.2646, "step": 3808 }, { "epoch": 23.660436137071652, "grad_norm": 2.65625, "learning_rate": 4.8539146278701084e-05, "loss": 0.3009, "step": 3809 }, { "epoch": 23.666666666666668, "grad_norm": 3.34375, "learning_rate": 4.853831629788365e-05, "loss": 0.2525, "step": 3810 }, { "epoch": 23.672897196261683, "grad_norm": 2.421875, "learning_rate": 4.853748608845739e-05, "loss": 0.2716, "step": 3811 }, { "epoch": 23.6791277258567, "grad_norm": 4.46875, "learning_rate": 4.853665565043035e-05, "loss": 0.25, "step": 3812 }, { "epoch": 23.685358255451714, "grad_norm": 5.34375, "learning_rate": 4.8535824983810604e-05, "loss": 0.2683, "step": 3813 }, { "epoch": 23.69158878504673, "grad_norm": 3.703125, "learning_rate": 4.8534994088606225e-05, "loss": 0.4103, "step": 3814 }, { "epoch": 23.697819314641745, "grad_norm": 2.8125, "learning_rate": 4.853416296482528e-05, "loss": 0.2152, "step": 3815 }, { "epoch": 23.70404984423676, "grad_norm": 3.171875, "learning_rate": 4.853333161247583e-05, "loss": 0.2008, "step": 3816 }, { "epoch": 23.710280373831775, "grad_norm": 2.5625, "learning_rate": 4.8532500031565974e-05, "loss": 0.2102, "step": 3817 }, { "epoch": 23.71651090342679, "grad_norm": 3.625, "learning_rate": 4.853166822210376e-05, "loss": 0.2111, "step": 3818 }, { "epoch": 23.722741433021806, "grad_norm": 2.703125, "learning_rate": 4.8530836184097297e-05, "loss": 0.3231, "step": 3819 }, { "epoch": 23.72897196261682, "grad_norm": 2.09375, "learning_rate": 4.853000391755463e-05, "loss": 0.1897, "step": 3820 }, { "epoch": 23.735202492211837, "grad_norm": 1.40625, "learning_rate": 4.8529171422483876e-05, "loss": 0.1588, "step": 3821 }, { "epoch": 23.741433021806852, "grad_norm": 4.1875, "learning_rate": 4.8528338698893104e-05, "loss": 0.2871, "step": 3822 }, { "epoch": 23.747663551401867, "grad_norm": 3.875, "learning_rate": 4.85275057467904e-05, "loss": 0.4092, "step": 3823 }, { "epoch": 23.753894080996886, "grad_norm": 3.96875, "learning_rate": 4.852667256618385e-05, "loss": 0.2064, "step": 3824 }, { "epoch": 23.7601246105919, "grad_norm": 4.03125, "learning_rate": 4.8525839157081564e-05, "loss": 0.2107, "step": 3825 }, { "epoch": 23.766355140186917, "grad_norm": 4.40625, "learning_rate": 4.852500551949162e-05, "loss": 0.2808, "step": 3826 }, { "epoch": 23.772585669781932, "grad_norm": 2.625, "learning_rate": 4.8524171653422126e-05, "loss": 0.1808, "step": 3827 }, { "epoch": 23.778816199376948, "grad_norm": 3.453125, "learning_rate": 4.8523337558881166e-05, "loss": 0.2127, "step": 3828 }, { "epoch": 23.785046728971963, "grad_norm": 1.9296875, "learning_rate": 4.852250323587686e-05, "loss": 0.1839, "step": 3829 }, { "epoch": 23.79127725856698, "grad_norm": 3.078125, "learning_rate": 4.8521668684417284e-05, "loss": 0.2187, "step": 3830 }, { "epoch": 23.797507788161994, "grad_norm": 3.8125, "learning_rate": 4.852083390451058e-05, "loss": 0.3456, "step": 3831 }, { "epoch": 23.80373831775701, "grad_norm": 3.171875, "learning_rate": 4.851999889616482e-05, "loss": 0.284, "step": 3832 }, { "epoch": 23.809968847352025, "grad_norm": 2.9375, "learning_rate": 4.851916365938813e-05, "loss": 0.2092, "step": 3833 }, { "epoch": 23.81619937694704, "grad_norm": 2.734375, "learning_rate": 4.8518328194188635e-05, "loss": 0.2003, "step": 3834 }, { "epoch": 23.822429906542055, "grad_norm": 3.265625, "learning_rate": 4.8517492500574424e-05, "loss": 0.3109, "step": 3835 }, { "epoch": 23.82866043613707, "grad_norm": 1.9296875, "learning_rate": 4.8516656578553626e-05, "loss": 0.1779, "step": 3836 }, { "epoch": 23.834890965732086, "grad_norm": 3.4375, "learning_rate": 4.8515820428134364e-05, "loss": 0.1558, "step": 3837 }, { "epoch": 23.8411214953271, "grad_norm": 4.96875, "learning_rate": 4.851498404932474e-05, "loss": 0.2321, "step": 3838 }, { "epoch": 23.847352024922117, "grad_norm": 3.265625, "learning_rate": 4.851414744213291e-05, "loss": 0.2968, "step": 3839 }, { "epoch": 23.853582554517136, "grad_norm": 3.734375, "learning_rate": 4.8513310606566964e-05, "loss": 0.3211, "step": 3840 }, { "epoch": 23.85981308411215, "grad_norm": 1.6328125, "learning_rate": 4.851247354263505e-05, "loss": 0.1527, "step": 3841 }, { "epoch": 23.866043613707166, "grad_norm": 2.921875, "learning_rate": 4.8511636250345294e-05, "loss": 0.3131, "step": 3842 }, { "epoch": 23.87227414330218, "grad_norm": 2.84375, "learning_rate": 4.851079872970582e-05, "loss": 0.2452, "step": 3843 }, { "epoch": 23.878504672897197, "grad_norm": 2.859375, "learning_rate": 4.850996098072478e-05, "loss": 0.2507, "step": 3844 }, { "epoch": 23.884735202492212, "grad_norm": 2.296875, "learning_rate": 4.850912300341029e-05, "loss": 0.1646, "step": 3845 }, { "epoch": 23.890965732087228, "grad_norm": 1.890625, "learning_rate": 4.85082847977705e-05, "loss": 0.2812, "step": 3846 }, { "epoch": 23.897196261682243, "grad_norm": 2.8125, "learning_rate": 4.8507446363813554e-05, "loss": 0.2463, "step": 3847 }, { "epoch": 23.90342679127726, "grad_norm": 3.046875, "learning_rate": 4.8506607701547585e-05, "loss": 0.3616, "step": 3848 }, { "epoch": 23.909657320872274, "grad_norm": 2.6875, "learning_rate": 4.850576881098074e-05, "loss": 0.1669, "step": 3849 }, { "epoch": 23.91588785046729, "grad_norm": 3.375, "learning_rate": 4.850492969212118e-05, "loss": 0.4007, "step": 3850 }, { "epoch": 23.922118380062305, "grad_norm": 2.71875, "learning_rate": 4.850409034497704e-05, "loss": 0.1948, "step": 3851 }, { "epoch": 23.92834890965732, "grad_norm": 2.671875, "learning_rate": 4.8503250769556473e-05, "loss": 0.2905, "step": 3852 }, { "epoch": 23.934579439252335, "grad_norm": 3.765625, "learning_rate": 4.850241096586764e-05, "loss": 0.3883, "step": 3853 }, { "epoch": 23.94080996884735, "grad_norm": 2.71875, "learning_rate": 4.850157093391869e-05, "loss": 0.2125, "step": 3854 }, { "epoch": 23.947040498442366, "grad_norm": 2.328125, "learning_rate": 4.8500730673717785e-05, "loss": 0.2045, "step": 3855 }, { "epoch": 23.953271028037385, "grad_norm": 2.28125, "learning_rate": 4.849989018527309e-05, "loss": 0.191, "step": 3856 }, { "epoch": 23.9595015576324, "grad_norm": 2.859375, "learning_rate": 4.849904946859276e-05, "loss": 0.3389, "step": 3857 }, { "epoch": 23.965732087227416, "grad_norm": 2.28125, "learning_rate": 4.849820852368497e-05, "loss": 0.1709, "step": 3858 }, { "epoch": 23.97196261682243, "grad_norm": 2.453125, "learning_rate": 4.849736735055788e-05, "loss": 0.1963, "step": 3859 }, { "epoch": 23.978193146417446, "grad_norm": 2.3125, "learning_rate": 4.849652594921965e-05, "loss": 0.2104, "step": 3860 }, { "epoch": 23.98442367601246, "grad_norm": 3.140625, "learning_rate": 4.849568431967848e-05, "loss": 0.1514, "step": 3861 }, { "epoch": 23.990654205607477, "grad_norm": 3.34375, "learning_rate": 4.849484246194251e-05, "loss": 0.2347, "step": 3862 }, { "epoch": 23.996884735202492, "grad_norm": 2.515625, "learning_rate": 4.849400037601995e-05, "loss": 0.194, "step": 3863 }, { "epoch": 24.0, "grad_norm": 1.640625, "learning_rate": 4.849315806191895e-05, "loss": 0.0925, "step": 3864 }, { "epoch": 24.006230529595015, "grad_norm": 2.234375, "learning_rate": 4.849231551964771e-05, "loss": 0.1896, "step": 3865 }, { "epoch": 24.01246105919003, "grad_norm": 2.484375, "learning_rate": 4.849147274921441e-05, "loss": 0.2694, "step": 3866 }, { "epoch": 24.018691588785046, "grad_norm": 3.109375, "learning_rate": 4.849062975062722e-05, "loss": 0.3584, "step": 3867 }, { "epoch": 24.02492211838006, "grad_norm": 2.65625, "learning_rate": 4.848978652389434e-05, "loss": 0.3251, "step": 3868 }, { "epoch": 24.031152647975077, "grad_norm": 2.59375, "learning_rate": 4.848894306902396e-05, "loss": 0.2847, "step": 3869 }, { "epoch": 24.037383177570092, "grad_norm": 2.03125, "learning_rate": 4.848809938602427e-05, "loss": 0.1496, "step": 3870 }, { "epoch": 24.043613707165107, "grad_norm": 2.625, "learning_rate": 4.8487255474903465e-05, "loss": 0.3154, "step": 3871 }, { "epoch": 24.049844236760123, "grad_norm": 3.03125, "learning_rate": 4.848641133566973e-05, "loss": 0.246, "step": 3872 }, { "epoch": 24.05607476635514, "grad_norm": 3.1875, "learning_rate": 4.8485566968331286e-05, "loss": 0.2166, "step": 3873 }, { "epoch": 24.062305295950157, "grad_norm": 3.40625, "learning_rate": 4.848472237289632e-05, "loss": 0.3476, "step": 3874 }, { "epoch": 24.068535825545172, "grad_norm": 2.859375, "learning_rate": 4.848387754937303e-05, "loss": 0.2274, "step": 3875 }, { "epoch": 24.074766355140188, "grad_norm": 3.3125, "learning_rate": 4.848303249776963e-05, "loss": 0.1921, "step": 3876 }, { "epoch": 24.080996884735203, "grad_norm": 2.796875, "learning_rate": 4.848218721809432e-05, "loss": 0.3016, "step": 3877 }, { "epoch": 24.08722741433022, "grad_norm": 2.921875, "learning_rate": 4.848134171035532e-05, "loss": 0.1777, "step": 3878 }, { "epoch": 24.093457943925234, "grad_norm": 2.84375, "learning_rate": 4.8480495974560836e-05, "loss": 0.1864, "step": 3879 }, { "epoch": 24.09968847352025, "grad_norm": 2.640625, "learning_rate": 4.847965001071907e-05, "loss": 0.2472, "step": 3880 }, { "epoch": 24.105919003115265, "grad_norm": 3.703125, "learning_rate": 4.847880381883826e-05, "loss": 0.499, "step": 3881 }, { "epoch": 24.11214953271028, "grad_norm": 2.5625, "learning_rate": 4.847795739892661e-05, "loss": 0.2393, "step": 3882 }, { "epoch": 24.118380062305295, "grad_norm": 2.140625, "learning_rate": 4.847711075099235e-05, "loss": 0.1484, "step": 3883 }, { "epoch": 24.12461059190031, "grad_norm": 2.53125, "learning_rate": 4.847626387504369e-05, "loss": 0.1669, "step": 3884 }, { "epoch": 24.130841121495326, "grad_norm": 3.0625, "learning_rate": 4.8475416771088865e-05, "loss": 0.3511, "step": 3885 }, { "epoch": 24.13707165109034, "grad_norm": 2.296875, "learning_rate": 4.84745694391361e-05, "loss": 0.2196, "step": 3886 }, { "epoch": 24.143302180685357, "grad_norm": 2.078125, "learning_rate": 4.847372187919362e-05, "loss": 0.167, "step": 3887 }, { "epoch": 24.149532710280372, "grad_norm": 2.34375, "learning_rate": 4.8472874091269674e-05, "loss": 0.1962, "step": 3888 }, { "epoch": 24.15576323987539, "grad_norm": 3.546875, "learning_rate": 4.847202607537247e-05, "loss": 0.3097, "step": 3889 }, { "epoch": 24.161993769470406, "grad_norm": 4.28125, "learning_rate": 4.8471177831510264e-05, "loss": 0.3306, "step": 3890 }, { "epoch": 24.16822429906542, "grad_norm": 5.0, "learning_rate": 4.847032935969128e-05, "loss": 0.3557, "step": 3891 }, { "epoch": 24.174454828660437, "grad_norm": 2.375, "learning_rate": 4.846948065992377e-05, "loss": 0.1706, "step": 3892 }, { "epoch": 24.180685358255452, "grad_norm": 3.75, "learning_rate": 4.8468631732215975e-05, "loss": 0.198, "step": 3893 }, { "epoch": 24.186915887850468, "grad_norm": 2.265625, "learning_rate": 4.8467782576576135e-05, "loss": 0.2766, "step": 3894 }, { "epoch": 24.193146417445483, "grad_norm": 1.9375, "learning_rate": 4.8466933193012495e-05, "loss": 0.1591, "step": 3895 }, { "epoch": 24.1993769470405, "grad_norm": 2.03125, "learning_rate": 4.8466083581533316e-05, "loss": 0.1551, "step": 3896 }, { "epoch": 24.205607476635514, "grad_norm": 4.15625, "learning_rate": 4.8465233742146845e-05, "loss": 0.3925, "step": 3897 }, { "epoch": 24.21183800623053, "grad_norm": 4.34375, "learning_rate": 4.846438367486133e-05, "loss": 0.3048, "step": 3898 }, { "epoch": 24.218068535825545, "grad_norm": 2.890625, "learning_rate": 4.8463533379685024e-05, "loss": 0.1903, "step": 3899 }, { "epoch": 24.22429906542056, "grad_norm": 4.09375, "learning_rate": 4.846268285662619e-05, "loss": 0.3602, "step": 3900 }, { "epoch": 24.230529595015575, "grad_norm": 3.5625, "learning_rate": 4.8461832105693095e-05, "loss": 0.3212, "step": 3901 }, { "epoch": 24.23676012461059, "grad_norm": 2.953125, "learning_rate": 4.8460981126894e-05, "loss": 0.2745, "step": 3902 }, { "epoch": 24.242990654205606, "grad_norm": 3.96875, "learning_rate": 4.846012992023716e-05, "loss": 0.3367, "step": 3903 }, { "epoch": 24.24922118380062, "grad_norm": 2.296875, "learning_rate": 4.845927848573085e-05, "loss": 0.1477, "step": 3904 }, { "epoch": 24.25545171339564, "grad_norm": 2.625, "learning_rate": 4.845842682338334e-05, "loss": 0.2273, "step": 3905 }, { "epoch": 24.261682242990656, "grad_norm": 2.109375, "learning_rate": 4.845757493320291e-05, "loss": 0.2107, "step": 3906 }, { "epoch": 24.26791277258567, "grad_norm": 3.84375, "learning_rate": 4.845672281519781e-05, "loss": 0.3259, "step": 3907 }, { "epoch": 24.274143302180686, "grad_norm": 1.859375, "learning_rate": 4.845587046937633e-05, "loss": 0.2482, "step": 3908 }, { "epoch": 24.2803738317757, "grad_norm": 2.015625, "learning_rate": 4.845501789574674e-05, "loss": 0.1329, "step": 3909 }, { "epoch": 24.286604361370717, "grad_norm": 3.109375, "learning_rate": 4.845416509431734e-05, "loss": 0.2781, "step": 3910 }, { "epoch": 24.292834890965732, "grad_norm": 3.578125, "learning_rate": 4.84533120650964e-05, "loss": 0.2474, "step": 3911 }, { "epoch": 24.299065420560748, "grad_norm": 3.234375, "learning_rate": 4.84524588080922e-05, "loss": 0.2363, "step": 3912 }, { "epoch": 24.305295950155763, "grad_norm": 3.09375, "learning_rate": 4.8451605323313034e-05, "loss": 0.219, "step": 3913 }, { "epoch": 24.31152647975078, "grad_norm": 2.890625, "learning_rate": 4.8450751610767194e-05, "loss": 0.2897, "step": 3914 }, { "epoch": 24.317757009345794, "grad_norm": 2.28125, "learning_rate": 4.844989767046296e-05, "loss": 0.3018, "step": 3915 }, { "epoch": 24.32398753894081, "grad_norm": 2.171875, "learning_rate": 4.844904350240863e-05, "loss": 0.2067, "step": 3916 }, { "epoch": 24.330218068535824, "grad_norm": 3.125, "learning_rate": 4.8448189106612505e-05, "loss": 0.2313, "step": 3917 }, { "epoch": 24.33644859813084, "grad_norm": 1.7890625, "learning_rate": 4.8447334483082886e-05, "loss": 0.1433, "step": 3918 }, { "epoch": 24.342679127725855, "grad_norm": 2.5, "learning_rate": 4.8446479631828066e-05, "loss": 0.2056, "step": 3919 }, { "epoch": 24.34890965732087, "grad_norm": 2.9375, "learning_rate": 4.8445624552856346e-05, "loss": 0.1983, "step": 3920 }, { "epoch": 24.35514018691589, "grad_norm": 1.609375, "learning_rate": 4.844476924617603e-05, "loss": 0.1328, "step": 3921 }, { "epoch": 24.361370716510905, "grad_norm": 2.09375, "learning_rate": 4.844391371179543e-05, "loss": 0.1728, "step": 3922 }, { "epoch": 24.36760124610592, "grad_norm": 1.2890625, "learning_rate": 4.844305794972286e-05, "loss": 0.146, "step": 3923 }, { "epoch": 24.373831775700936, "grad_norm": 3.71875, "learning_rate": 4.844220195996662e-05, "loss": 0.3091, "step": 3924 }, { "epoch": 24.38006230529595, "grad_norm": 3.390625, "learning_rate": 4.8441345742535024e-05, "loss": 0.2899, "step": 3925 }, { "epoch": 24.386292834890966, "grad_norm": 2.53125, "learning_rate": 4.84404892974364e-05, "loss": 0.1446, "step": 3926 }, { "epoch": 24.39252336448598, "grad_norm": 3.484375, "learning_rate": 4.843963262467906e-05, "loss": 0.3171, "step": 3927 }, { "epoch": 24.398753894080997, "grad_norm": 1.4296875, "learning_rate": 4.8438775724271316e-05, "loss": 0.147, "step": 3928 }, { "epoch": 24.404984423676012, "grad_norm": 3.9375, "learning_rate": 4.84379185962215e-05, "loss": 0.3188, "step": 3929 }, { "epoch": 24.411214953271028, "grad_norm": 2.53125, "learning_rate": 4.8437061240537934e-05, "loss": 0.1362, "step": 3930 }, { "epoch": 24.417445482866043, "grad_norm": 1.5234375, "learning_rate": 4.843620365722894e-05, "loss": 0.1576, "step": 3931 }, { "epoch": 24.42367601246106, "grad_norm": 2.734375, "learning_rate": 4.843534584630286e-05, "loss": 0.2844, "step": 3932 }, { "epoch": 24.429906542056074, "grad_norm": 3.296875, "learning_rate": 4.8434487807768006e-05, "loss": 0.2565, "step": 3933 }, { "epoch": 24.43613707165109, "grad_norm": 4.40625, "learning_rate": 4.843362954163273e-05, "loss": 0.3394, "step": 3934 }, { "epoch": 24.442367601246104, "grad_norm": 3.1875, "learning_rate": 4.843277104790535e-05, "loss": 0.2114, "step": 3935 }, { "epoch": 24.44859813084112, "grad_norm": 2.5625, "learning_rate": 4.8431912326594226e-05, "loss": 0.2112, "step": 3936 }, { "epoch": 24.45482866043614, "grad_norm": 1.828125, "learning_rate": 4.843105337770768e-05, "loss": 0.196, "step": 3937 }, { "epoch": 24.461059190031154, "grad_norm": 5.59375, "learning_rate": 4.843019420125405e-05, "loss": 0.2452, "step": 3938 }, { "epoch": 24.46728971962617, "grad_norm": 5.40625, "learning_rate": 4.842933479724171e-05, "loss": 0.3382, "step": 3939 }, { "epoch": 24.473520249221185, "grad_norm": 4.6875, "learning_rate": 4.842847516567898e-05, "loss": 0.375, "step": 3940 }, { "epoch": 24.4797507788162, "grad_norm": 4.28125, "learning_rate": 4.84276153065742e-05, "loss": 0.3069, "step": 3941 }, { "epoch": 24.485981308411215, "grad_norm": 4.375, "learning_rate": 4.8426755219935756e-05, "loss": 0.2721, "step": 3942 }, { "epoch": 24.49221183800623, "grad_norm": 4.75, "learning_rate": 4.8425894905771975e-05, "loss": 0.3004, "step": 3943 }, { "epoch": 24.498442367601246, "grad_norm": 5.875, "learning_rate": 4.842503436409122e-05, "loss": 0.2394, "step": 3944 }, { "epoch": 24.50467289719626, "grad_norm": 3.203125, "learning_rate": 4.842417359490185e-05, "loss": 0.4061, "step": 3945 }, { "epoch": 24.510903426791277, "grad_norm": 2.234375, "learning_rate": 4.842331259821222e-05, "loss": 0.2039, "step": 3946 }, { "epoch": 24.517133956386292, "grad_norm": 2.609375, "learning_rate": 4.84224513740307e-05, "loss": 0.3308, "step": 3947 }, { "epoch": 24.523364485981308, "grad_norm": 3.640625, "learning_rate": 4.842158992236565e-05, "loss": 0.3125, "step": 3948 }, { "epoch": 24.529595015576323, "grad_norm": 3.828125, "learning_rate": 4.8420728243225435e-05, "loss": 0.4212, "step": 3949 }, { "epoch": 24.53582554517134, "grad_norm": 2.015625, "learning_rate": 4.841986633661842e-05, "loss": 0.1752, "step": 3950 }, { "epoch": 24.542056074766354, "grad_norm": 2.34375, "learning_rate": 4.8419004202553e-05, "loss": 0.1787, "step": 3951 }, { "epoch": 24.54828660436137, "grad_norm": 2.625, "learning_rate": 4.8418141841037515e-05, "loss": 0.22, "step": 3952 }, { "epoch": 24.554517133956388, "grad_norm": 2.890625, "learning_rate": 4.8417279252080363e-05, "loss": 0.2873, "step": 3953 }, { "epoch": 24.560747663551403, "grad_norm": 1.296875, "learning_rate": 4.8416416435689906e-05, "loss": 0.1223, "step": 3954 }, { "epoch": 24.56697819314642, "grad_norm": 3.890625, "learning_rate": 4.841555339187454e-05, "loss": 0.2535, "step": 3955 }, { "epoch": 24.573208722741434, "grad_norm": 2.515625, "learning_rate": 4.841469012064263e-05, "loss": 0.2551, "step": 3956 }, { "epoch": 24.57943925233645, "grad_norm": 2.234375, "learning_rate": 4.841382662200257e-05, "loss": 0.1908, "step": 3957 }, { "epoch": 24.585669781931465, "grad_norm": 2.375, "learning_rate": 4.841296289596274e-05, "loss": 0.1757, "step": 3958 }, { "epoch": 24.59190031152648, "grad_norm": 3.125, "learning_rate": 4.8412098942531546e-05, "loss": 0.3824, "step": 3959 }, { "epoch": 24.598130841121495, "grad_norm": 1.625, "learning_rate": 4.8411234761717364e-05, "loss": 0.1576, "step": 3960 }, { "epoch": 24.60436137071651, "grad_norm": 3.046875, "learning_rate": 4.841037035352859e-05, "loss": 0.3452, "step": 3961 }, { "epoch": 24.610591900311526, "grad_norm": 2.96875, "learning_rate": 4.8409505717973625e-05, "loss": 0.2098, "step": 3962 }, { "epoch": 24.61682242990654, "grad_norm": 3.34375, "learning_rate": 4.8408640855060846e-05, "loss": 0.3904, "step": 3963 }, { "epoch": 24.623052959501557, "grad_norm": 2.375, "learning_rate": 4.840777576479868e-05, "loss": 0.2154, "step": 3964 }, { "epoch": 24.629283489096572, "grad_norm": 2.109375, "learning_rate": 4.8406910447195516e-05, "loss": 0.1874, "step": 3965 }, { "epoch": 24.635514018691588, "grad_norm": 3.34375, "learning_rate": 4.840604490225976e-05, "loss": 0.3661, "step": 3966 }, { "epoch": 24.641744548286603, "grad_norm": 2.234375, "learning_rate": 4.840517912999981e-05, "loss": 0.1584, "step": 3967 }, { "epoch": 24.64797507788162, "grad_norm": 3.515625, "learning_rate": 4.840431313042408e-05, "loss": 0.4264, "step": 3968 }, { "epoch": 24.654205607476637, "grad_norm": 2.953125, "learning_rate": 4.840344690354099e-05, "loss": 0.2394, "step": 3969 }, { "epoch": 24.660436137071652, "grad_norm": 2.9375, "learning_rate": 4.840258044935895e-05, "loss": 0.2335, "step": 3970 }, { "epoch": 24.666666666666668, "grad_norm": 3.046875, "learning_rate": 4.840171376788636e-05, "loss": 0.5393, "step": 3971 }, { "epoch": 24.672897196261683, "grad_norm": 2.84375, "learning_rate": 4.8400846859131646e-05, "loss": 0.2247, "step": 3972 }, { "epoch": 24.6791277258567, "grad_norm": 1.9375, "learning_rate": 4.839997972310324e-05, "loss": 0.1655, "step": 3973 }, { "epoch": 24.685358255451714, "grad_norm": 4.53125, "learning_rate": 4.839911235980955e-05, "loss": 0.362, "step": 3974 }, { "epoch": 24.69158878504673, "grad_norm": 1.7265625, "learning_rate": 4.8398244769259e-05, "loss": 0.1694, "step": 3975 }, { "epoch": 24.697819314641745, "grad_norm": 2.203125, "learning_rate": 4.839737695146002e-05, "loss": 0.232, "step": 3976 }, { "epoch": 24.70404984423676, "grad_norm": 3.015625, "learning_rate": 4.839650890642104e-05, "loss": 0.2655, "step": 3977 }, { "epoch": 24.710280373831775, "grad_norm": 2.9375, "learning_rate": 4.8395640634150485e-05, "loss": 0.265, "step": 3978 }, { "epoch": 24.71651090342679, "grad_norm": 2.6875, "learning_rate": 4.8394772134656795e-05, "loss": 0.2032, "step": 3979 }, { "epoch": 24.722741433021806, "grad_norm": 3.171875, "learning_rate": 4.839390340794841e-05, "loss": 0.2854, "step": 3980 }, { "epoch": 24.72897196261682, "grad_norm": 2.53125, "learning_rate": 4.8393034454033745e-05, "loss": 0.1904, "step": 3981 }, { "epoch": 24.735202492211837, "grad_norm": 3.828125, "learning_rate": 4.839216527292126e-05, "loss": 0.3314, "step": 3982 }, { "epoch": 24.741433021806852, "grad_norm": 5.125, "learning_rate": 4.839129586461938e-05, "loss": 0.2799, "step": 3983 }, { "epoch": 24.747663551401867, "grad_norm": 5.71875, "learning_rate": 4.839042622913657e-05, "loss": 0.2543, "step": 3984 }, { "epoch": 24.753894080996886, "grad_norm": 2.6875, "learning_rate": 4.838955636648126e-05, "loss": 0.1551, "step": 3985 }, { "epoch": 24.7601246105919, "grad_norm": 1.765625, "learning_rate": 4.838868627666191e-05, "loss": 0.1722, "step": 3986 }, { "epoch": 24.766355140186917, "grad_norm": 4.46875, "learning_rate": 4.838781595968695e-05, "loss": 0.2786, "step": 3987 }, { "epoch": 24.772585669781932, "grad_norm": 4.71875, "learning_rate": 4.838694541556486e-05, "loss": 0.1866, "step": 3988 }, { "epoch": 24.778816199376948, "grad_norm": 5.03125, "learning_rate": 4.838607464430408e-05, "loss": 0.2253, "step": 3989 }, { "epoch": 24.785046728971963, "grad_norm": 2.796875, "learning_rate": 4.838520364591306e-05, "loss": 0.1394, "step": 3990 }, { "epoch": 24.79127725856698, "grad_norm": 3.578125, "learning_rate": 4.838433242040027e-05, "loss": 0.3372, "step": 3991 }, { "epoch": 24.797507788161994, "grad_norm": 4.90625, "learning_rate": 4.8383460967774166e-05, "loss": 0.2044, "step": 3992 }, { "epoch": 24.80373831775701, "grad_norm": 3.96875, "learning_rate": 4.838258928804322e-05, "loss": 0.2155, "step": 3993 }, { "epoch": 24.809968847352025, "grad_norm": 4.28125, "learning_rate": 4.83817173812159e-05, "loss": 0.1546, "step": 3994 }, { "epoch": 24.81619937694704, "grad_norm": 3.375, "learning_rate": 4.838084524730065e-05, "loss": 0.2863, "step": 3995 }, { "epoch": 24.822429906542055, "grad_norm": 3.265625, "learning_rate": 4.8379972886305966e-05, "loss": 0.227, "step": 3996 }, { "epoch": 24.82866043613707, "grad_norm": 3.171875, "learning_rate": 4.837910029824031e-05, "loss": 0.2454, "step": 3997 }, { "epoch": 24.834890965732086, "grad_norm": 3.6875, "learning_rate": 4.837822748311217e-05, "loss": 0.3773, "step": 3998 }, { "epoch": 24.8411214953271, "grad_norm": 1.953125, "learning_rate": 4.8377354440930003e-05, "loss": 0.1471, "step": 3999 }, { "epoch": 24.847352024922117, "grad_norm": 2.125, "learning_rate": 4.8376481171702294e-05, "loss": 0.1515, "step": 4000 }, { "epoch": 24.853582554517136, "grad_norm": 2.21875, "learning_rate": 4.837560767543753e-05, "loss": 0.206, "step": 4001 }, { "epoch": 24.85981308411215, "grad_norm": 3.59375, "learning_rate": 4.8374733952144186e-05, "loss": 0.2463, "step": 4002 }, { "epoch": 24.866043613707166, "grad_norm": 2.8125, "learning_rate": 4.8373860001830755e-05, "loss": 0.1872, "step": 4003 }, { "epoch": 24.87227414330218, "grad_norm": 3.8125, "learning_rate": 4.837298582450573e-05, "loss": 0.1736, "step": 4004 }, { "epoch": 24.878504672897197, "grad_norm": 2.25, "learning_rate": 4.837211142017759e-05, "loss": 0.193, "step": 4005 }, { "epoch": 24.884735202492212, "grad_norm": 3.484375, "learning_rate": 4.8371236788854835e-05, "loss": 0.3578, "step": 4006 }, { "epoch": 24.890965732087228, "grad_norm": 4.6875, "learning_rate": 4.837036193054595e-05, "loss": 0.1488, "step": 4007 }, { "epoch": 24.897196261682243, "grad_norm": 3.171875, "learning_rate": 4.8369486845259436e-05, "loss": 0.2902, "step": 4008 }, { "epoch": 24.90342679127726, "grad_norm": 2.921875, "learning_rate": 4.83686115330038e-05, "loss": 0.1912, "step": 4009 }, { "epoch": 24.909657320872274, "grad_norm": 2.34375, "learning_rate": 4.836773599378754e-05, "loss": 0.2035, "step": 4010 }, { "epoch": 24.91588785046729, "grad_norm": 2.96875, "learning_rate": 4.836686022761916e-05, "loss": 0.1494, "step": 4011 }, { "epoch": 24.922118380062305, "grad_norm": 3.828125, "learning_rate": 4.836598423450716e-05, "loss": 0.1863, "step": 4012 }, { "epoch": 24.92834890965732, "grad_norm": 4.6875, "learning_rate": 4.836510801446004e-05, "loss": 0.2902, "step": 4013 }, { "epoch": 24.934579439252335, "grad_norm": 2.5625, "learning_rate": 4.836423156748633e-05, "loss": 0.203, "step": 4014 }, { "epoch": 24.94080996884735, "grad_norm": 2.171875, "learning_rate": 4.836335489359453e-05, "loss": 0.1933, "step": 4015 }, { "epoch": 24.947040498442366, "grad_norm": 2.296875, "learning_rate": 4.8362477992793155e-05, "loss": 0.2052, "step": 4016 }, { "epoch": 24.953271028037385, "grad_norm": 3.578125, "learning_rate": 4.836160086509073e-05, "loss": 0.2079, "step": 4017 }, { "epoch": 24.9595015576324, "grad_norm": 4.4375, "learning_rate": 4.836072351049576e-05, "loss": 0.2795, "step": 4018 }, { "epoch": 24.965732087227416, "grad_norm": 4.0625, "learning_rate": 4.835984592901678e-05, "loss": 0.3119, "step": 4019 }, { "epoch": 24.97196261682243, "grad_norm": 2.828125, "learning_rate": 4.8358968120662305e-05, "loss": 0.2265, "step": 4020 }, { "epoch": 24.978193146417446, "grad_norm": 2.484375, "learning_rate": 4.8358090085440864e-05, "loss": 0.1799, "step": 4021 }, { "epoch": 24.98442367601246, "grad_norm": 2.90625, "learning_rate": 4.8357211823360974e-05, "loss": 0.2099, "step": 4022 }, { "epoch": 24.990654205607477, "grad_norm": 1.7734375, "learning_rate": 4.835633333443118e-05, "loss": 0.1996, "step": 4023 }, { "epoch": 24.996884735202492, "grad_norm": 2.140625, "learning_rate": 4.835545461866001e-05, "loss": 0.1692, "step": 4024 }, { "epoch": 25.0, "grad_norm": 1.7890625, "learning_rate": 4.835457567605599e-05, "loss": 0.1461, "step": 4025 }, { "epoch": 25.006230529595015, "grad_norm": 2.1875, "learning_rate": 4.835369650662767e-05, "loss": 0.2209, "step": 4026 }, { "epoch": 25.01246105919003, "grad_norm": 3.90625, "learning_rate": 4.8352817110383574e-05, "loss": 0.307, "step": 4027 }, { "epoch": 25.018691588785046, "grad_norm": 2.03125, "learning_rate": 4.8351937487332256e-05, "loss": 0.1428, "step": 4028 }, { "epoch": 25.02492211838006, "grad_norm": 3.03125, "learning_rate": 4.8351057637482255e-05, "loss": 0.2877, "step": 4029 }, { "epoch": 25.031152647975077, "grad_norm": 2.0625, "learning_rate": 4.835017756084211e-05, "loss": 0.241, "step": 4030 }, { "epoch": 25.037383177570092, "grad_norm": 1.8984375, "learning_rate": 4.8349297257420367e-05, "loss": 0.1917, "step": 4031 }, { "epoch": 25.043613707165107, "grad_norm": 1.9453125, "learning_rate": 4.8348416727225595e-05, "loss": 0.1747, "step": 4032 }, { "epoch": 25.049844236760123, "grad_norm": 3.265625, "learning_rate": 4.834753597026632e-05, "loss": 0.3358, "step": 4033 }, { "epoch": 25.05607476635514, "grad_norm": 3.296875, "learning_rate": 4.8346654986551115e-05, "loss": 0.2664, "step": 4034 }, { "epoch": 25.062305295950157, "grad_norm": 2.640625, "learning_rate": 4.834577377608853e-05, "loss": 0.2513, "step": 4035 }, { "epoch": 25.068535825545172, "grad_norm": 2.65625, "learning_rate": 4.834489233888712e-05, "loss": 0.2228, "step": 4036 }, { "epoch": 25.074766355140188, "grad_norm": 3.3125, "learning_rate": 4.834401067495545e-05, "loss": 0.3459, "step": 4037 }, { "epoch": 25.080996884735203, "grad_norm": 2.484375, "learning_rate": 4.8343128784302085e-05, "loss": 0.2211, "step": 4038 }, { "epoch": 25.08722741433022, "grad_norm": 2.171875, "learning_rate": 4.834224666693559e-05, "loss": 0.216, "step": 4039 }, { "epoch": 25.093457943925234, "grad_norm": 2.546875, "learning_rate": 4.8341364322864523e-05, "loss": 0.2223, "step": 4040 }, { "epoch": 25.09968847352025, "grad_norm": 2.46875, "learning_rate": 4.834048175209746e-05, "loss": 0.2144, "step": 4041 }, { "epoch": 25.105919003115265, "grad_norm": 1.671875, "learning_rate": 4.8339598954642976e-05, "loss": 0.1499, "step": 4042 }, { "epoch": 25.11214953271028, "grad_norm": 2.265625, "learning_rate": 4.8338715930509636e-05, "loss": 0.175, "step": 4043 }, { "epoch": 25.118380062305295, "grad_norm": 2.609375, "learning_rate": 4.8337832679706024e-05, "loss": 0.231, "step": 4044 }, { "epoch": 25.12461059190031, "grad_norm": 3.09375, "learning_rate": 4.833694920224072e-05, "loss": 0.3172, "step": 4045 }, { "epoch": 25.130841121495326, "grad_norm": 2.828125, "learning_rate": 4.83360654981223e-05, "loss": 0.2139, "step": 4046 }, { "epoch": 25.13707165109034, "grad_norm": 1.9921875, "learning_rate": 4.833518156735934e-05, "loss": 0.1284, "step": 4047 }, { "epoch": 25.143302180685357, "grad_norm": 3.046875, "learning_rate": 4.833429740996044e-05, "loss": 0.3139, "step": 4048 }, { "epoch": 25.149532710280372, "grad_norm": 3.125, "learning_rate": 4.833341302593417e-05, "loss": 0.2919, "step": 4049 }, { "epoch": 25.15576323987539, "grad_norm": 2.71875, "learning_rate": 4.8332528415289136e-05, "loss": 0.2635, "step": 4050 }, { "epoch": 25.161993769470406, "grad_norm": 1.796875, "learning_rate": 4.833164357803392e-05, "loss": 0.1756, "step": 4051 }, { "epoch": 25.16822429906542, "grad_norm": 2.84375, "learning_rate": 4.8330758514177114e-05, "loss": 0.2422, "step": 4052 }, { "epoch": 25.174454828660437, "grad_norm": 4.0625, "learning_rate": 4.8329873223727326e-05, "loss": 0.2887, "step": 4053 }, { "epoch": 25.180685358255452, "grad_norm": 2.34375, "learning_rate": 4.832898770669314e-05, "loss": 0.2898, "step": 4054 }, { "epoch": 25.186915887850468, "grad_norm": 3.109375, "learning_rate": 4.832810196308316e-05, "loss": 0.3248, "step": 4055 }, { "epoch": 25.193146417445483, "grad_norm": 2.234375, "learning_rate": 4.8327215992905995e-05, "loss": 0.1969, "step": 4056 }, { "epoch": 25.1993769470405, "grad_norm": 1.9609375, "learning_rate": 4.832632979617024e-05, "loss": 0.1933, "step": 4057 }, { "epoch": 25.205607476635514, "grad_norm": 2.65625, "learning_rate": 4.832544337288452e-05, "loss": 0.2108, "step": 4058 }, { "epoch": 25.21183800623053, "grad_norm": 2.96875, "learning_rate": 4.832455672305741e-05, "loss": 0.3195, "step": 4059 }, { "epoch": 25.218068535825545, "grad_norm": 2.078125, "learning_rate": 4.8323669846697564e-05, "loss": 0.2648, "step": 4060 }, { "epoch": 25.22429906542056, "grad_norm": 2.625, "learning_rate": 4.832278274381356e-05, "loss": 0.2628, "step": 4061 }, { "epoch": 25.230529595015575, "grad_norm": 2.90625, "learning_rate": 4.832189541441403e-05, "loss": 0.1908, "step": 4062 }, { "epoch": 25.23676012461059, "grad_norm": 3.21875, "learning_rate": 4.8321007858507594e-05, "loss": 0.3498, "step": 4063 }, { "epoch": 25.242990654205606, "grad_norm": 3.609375, "learning_rate": 4.832012007610287e-05, "loss": 0.2644, "step": 4064 }, { "epoch": 25.24922118380062, "grad_norm": 1.5390625, "learning_rate": 4.831923206720848e-05, "loss": 0.1374, "step": 4065 }, { "epoch": 25.25545171339564, "grad_norm": 2.8125, "learning_rate": 4.8318343831833035e-05, "loss": 0.204, "step": 4066 }, { "epoch": 25.261682242990656, "grad_norm": 1.5546875, "learning_rate": 4.8317455369985174e-05, "loss": 0.1468, "step": 4067 }, { "epoch": 25.26791277258567, "grad_norm": 3.328125, "learning_rate": 4.831656668167353e-05, "loss": 0.2338, "step": 4068 }, { "epoch": 25.274143302180686, "grad_norm": 4.0625, "learning_rate": 4.831567776690674e-05, "loss": 0.3726, "step": 4069 }, { "epoch": 25.2803738317757, "grad_norm": 3.21875, "learning_rate": 4.831478862569342e-05, "loss": 0.2763, "step": 4070 }, { "epoch": 25.286604361370717, "grad_norm": 1.9765625, "learning_rate": 4.8313899258042206e-05, "loss": 0.2347, "step": 4071 }, { "epoch": 25.292834890965732, "grad_norm": 2.640625, "learning_rate": 4.8313009663961746e-05, "loss": 0.1761, "step": 4072 }, { "epoch": 25.299065420560748, "grad_norm": 2.9375, "learning_rate": 4.831211984346067e-05, "loss": 0.2241, "step": 4073 }, { "epoch": 25.305295950155763, "grad_norm": 2.828125, "learning_rate": 4.8311229796547634e-05, "loss": 0.224, "step": 4074 }, { "epoch": 25.31152647975078, "grad_norm": 3.515625, "learning_rate": 4.8310339523231276e-05, "loss": 0.189, "step": 4075 }, { "epoch": 25.317757009345794, "grad_norm": 1.9765625, "learning_rate": 4.830944902352024e-05, "loss": 0.2463, "step": 4076 }, { "epoch": 25.32398753894081, "grad_norm": 3.53125, "learning_rate": 4.830855829742317e-05, "loss": 0.3409, "step": 4077 }, { "epoch": 25.330218068535824, "grad_norm": 2.734375, "learning_rate": 4.830766734494873e-05, "loss": 0.2703, "step": 4078 }, { "epoch": 25.33644859813084, "grad_norm": 4.6875, "learning_rate": 4.830677616610556e-05, "loss": 0.303, "step": 4079 }, { "epoch": 25.342679127725855, "grad_norm": 5.34375, "learning_rate": 4.830588476090232e-05, "loss": 0.4538, "step": 4080 }, { "epoch": 25.34890965732087, "grad_norm": 1.25, "learning_rate": 4.8304993129347664e-05, "loss": 0.1393, "step": 4081 }, { "epoch": 25.35514018691589, "grad_norm": 1.7734375, "learning_rate": 4.830410127145026e-05, "loss": 0.2075, "step": 4082 }, { "epoch": 25.361370716510905, "grad_norm": 4.375, "learning_rate": 4.830320918721878e-05, "loss": 0.1837, "step": 4083 }, { "epoch": 25.36760124610592, "grad_norm": 3.40625, "learning_rate": 4.830231687666186e-05, "loss": 0.1819, "step": 4084 }, { "epoch": 25.373831775700936, "grad_norm": 2.84375, "learning_rate": 4.830142433978818e-05, "loss": 0.1953, "step": 4085 }, { "epoch": 25.38006230529595, "grad_norm": 3.453125, "learning_rate": 4.830053157660641e-05, "loss": 0.3821, "step": 4086 }, { "epoch": 25.386292834890966, "grad_norm": 4.25, "learning_rate": 4.829963858712523e-05, "loss": 0.2432, "step": 4087 }, { "epoch": 25.39252336448598, "grad_norm": 3.28125, "learning_rate": 4.829874537135329e-05, "loss": 0.2306, "step": 4088 }, { "epoch": 25.398753894080997, "grad_norm": 2.953125, "learning_rate": 4.8297851929299284e-05, "loss": 0.3973, "step": 4089 }, { "epoch": 25.404984423676012, "grad_norm": 2.421875, "learning_rate": 4.829695826097188e-05, "loss": 0.28, "step": 4090 }, { "epoch": 25.411214953271028, "grad_norm": 1.765625, "learning_rate": 4.829606436637976e-05, "loss": 0.1505, "step": 4091 }, { "epoch": 25.417445482866043, "grad_norm": 3.28125, "learning_rate": 4.8295170245531605e-05, "loss": 0.2812, "step": 4092 }, { "epoch": 25.42367601246106, "grad_norm": 3.359375, "learning_rate": 4.8294275898436105e-05, "loss": 0.303, "step": 4093 }, { "epoch": 25.429906542056074, "grad_norm": 3.6875, "learning_rate": 4.829338132510194e-05, "loss": 0.3126, "step": 4094 }, { "epoch": 25.43613707165109, "grad_norm": 2.515625, "learning_rate": 4.829248652553779e-05, "loss": 0.1786, "step": 4095 }, { "epoch": 25.442367601246104, "grad_norm": 2.96875, "learning_rate": 4.8291591499752365e-05, "loss": 0.2912, "step": 4096 }, { "epoch": 25.44859813084112, "grad_norm": 2.234375, "learning_rate": 4.829069624775435e-05, "loss": 0.2125, "step": 4097 }, { "epoch": 25.45482866043614, "grad_norm": 2.9375, "learning_rate": 4.828980076955243e-05, "loss": 0.1926, "step": 4098 }, { "epoch": 25.461059190031154, "grad_norm": 2.265625, "learning_rate": 4.828890506515531e-05, "loss": 0.1635, "step": 4099 }, { "epoch": 25.46728971962617, "grad_norm": 2.703125, "learning_rate": 4.828800913457169e-05, "loss": 0.2471, "step": 4100 }, { "epoch": 25.473520249221185, "grad_norm": 2.90625, "learning_rate": 4.8287112977810274e-05, "loss": 0.2444, "step": 4101 }, { "epoch": 25.4797507788162, "grad_norm": 3.265625, "learning_rate": 4.828621659487976e-05, "loss": 0.2892, "step": 4102 }, { "epoch": 25.485981308411215, "grad_norm": 2.71875, "learning_rate": 4.828531998578885e-05, "loss": 0.1782, "step": 4103 }, { "epoch": 25.49221183800623, "grad_norm": 2.3125, "learning_rate": 4.828442315054627e-05, "loss": 0.1786, "step": 4104 }, { "epoch": 25.498442367601246, "grad_norm": 1.3203125, "learning_rate": 4.8283526089160705e-05, "loss": 0.1305, "step": 4105 }, { "epoch": 25.50467289719626, "grad_norm": 4.3125, "learning_rate": 4.8282628801640886e-05, "loss": 0.2982, "step": 4106 }, { "epoch": 25.510903426791277, "grad_norm": 3.84375, "learning_rate": 4.828173128799551e-05, "loss": 0.4291, "step": 4107 }, { "epoch": 25.517133956386292, "grad_norm": 3.046875, "learning_rate": 4.8280833548233324e-05, "loss": 0.2148, "step": 4108 }, { "epoch": 25.523364485981308, "grad_norm": 3.671875, "learning_rate": 4.827993558236302e-05, "loss": 0.2714, "step": 4109 }, { "epoch": 25.529595015576323, "grad_norm": 2.03125, "learning_rate": 4.827903739039333e-05, "loss": 0.211, "step": 4110 }, { "epoch": 25.53582554517134, "grad_norm": 2.34375, "learning_rate": 4.827813897233298e-05, "loss": 0.1616, "step": 4111 }, { "epoch": 25.542056074766354, "grad_norm": 4.3125, "learning_rate": 4.8277240328190676e-05, "loss": 0.3285, "step": 4112 }, { "epoch": 25.54828660436137, "grad_norm": 3.671875, "learning_rate": 4.827634145797517e-05, "loss": 0.1585, "step": 4113 }, { "epoch": 25.554517133956388, "grad_norm": 2.515625, "learning_rate": 4.827544236169519e-05, "loss": 0.1751, "step": 4114 }, { "epoch": 25.560747663551403, "grad_norm": 3.875, "learning_rate": 4.827454303935946e-05, "loss": 0.1748, "step": 4115 }, { "epoch": 25.56697819314642, "grad_norm": 3.828125, "learning_rate": 4.827364349097671e-05, "loss": 0.2258, "step": 4116 }, { "epoch": 25.573208722741434, "grad_norm": 2.828125, "learning_rate": 4.8272743716555674e-05, "loss": 0.2166, "step": 4117 }, { "epoch": 25.57943925233645, "grad_norm": 2.4375, "learning_rate": 4.827184371610511e-05, "loss": 0.1433, "step": 4118 }, { "epoch": 25.585669781931465, "grad_norm": 3.484375, "learning_rate": 4.827094348963375e-05, "loss": 0.2704, "step": 4119 }, { "epoch": 25.59190031152648, "grad_norm": 4.375, "learning_rate": 4.8270043037150324e-05, "loss": 0.29, "step": 4120 }, { "epoch": 25.598130841121495, "grad_norm": 2.484375, "learning_rate": 4.8269142358663596e-05, "loss": 0.2731, "step": 4121 }, { "epoch": 25.60436137071651, "grad_norm": 1.5546875, "learning_rate": 4.8268241454182304e-05, "loss": 0.1227, "step": 4122 }, { "epoch": 25.610591900311526, "grad_norm": 3.75, "learning_rate": 4.8267340323715203e-05, "loss": 0.2936, "step": 4123 }, { "epoch": 25.61682242990654, "grad_norm": 2.578125, "learning_rate": 4.8266438967271036e-05, "loss": 0.1934, "step": 4124 }, { "epoch": 25.623052959501557, "grad_norm": 2.71875, "learning_rate": 4.826553738485857e-05, "loss": 0.176, "step": 4125 }, { "epoch": 25.629283489096572, "grad_norm": 3.15625, "learning_rate": 4.826463557648655e-05, "loss": 0.2092, "step": 4126 }, { "epoch": 25.635514018691588, "grad_norm": 2.6875, "learning_rate": 4.826373354216373e-05, "loss": 0.2399, "step": 4127 }, { "epoch": 25.641744548286603, "grad_norm": 2.359375, "learning_rate": 4.8262831281898893e-05, "loss": 0.2838, "step": 4128 }, { "epoch": 25.64797507788162, "grad_norm": 2.375, "learning_rate": 4.826192879570078e-05, "loss": 0.1749, "step": 4129 }, { "epoch": 25.654205607476637, "grad_norm": 3.65625, "learning_rate": 4.826102608357817e-05, "loss": 0.2197, "step": 4130 }, { "epoch": 25.660436137071652, "grad_norm": 2.765625, "learning_rate": 4.826012314553983e-05, "loss": 0.2354, "step": 4131 }, { "epoch": 25.666666666666668, "grad_norm": 2.65625, "learning_rate": 4.825921998159452e-05, "loss": 0.2259, "step": 4132 }, { "epoch": 25.672897196261683, "grad_norm": 2.390625, "learning_rate": 4.825831659175101e-05, "loss": 0.2219, "step": 4133 }, { "epoch": 25.6791277258567, "grad_norm": 3.265625, "learning_rate": 4.8257412976018084e-05, "loss": 0.2159, "step": 4134 }, { "epoch": 25.685358255451714, "grad_norm": 2.5625, "learning_rate": 4.825650913440452e-05, "loss": 0.2088, "step": 4135 }, { "epoch": 25.69158878504673, "grad_norm": 2.75, "learning_rate": 4.825560506691908e-05, "loss": 0.2283, "step": 4136 }, { "epoch": 25.697819314641745, "grad_norm": 3.25, "learning_rate": 4.825470077357056e-05, "loss": 0.2716, "step": 4137 }, { "epoch": 25.70404984423676, "grad_norm": 1.984375, "learning_rate": 4.825379625436773e-05, "loss": 0.1515, "step": 4138 }, { "epoch": 25.710280373831775, "grad_norm": 2.140625, "learning_rate": 4.8252891509319395e-05, "loss": 0.1506, "step": 4139 }, { "epoch": 25.71651090342679, "grad_norm": 1.890625, "learning_rate": 4.825198653843432e-05, "loss": 0.1216, "step": 4140 }, { "epoch": 25.722741433021806, "grad_norm": 3.03125, "learning_rate": 4.825108134172131e-05, "loss": 0.1963, "step": 4141 }, { "epoch": 25.72897196261682, "grad_norm": 2.53125, "learning_rate": 4.825017591918915e-05, "loss": 0.2236, "step": 4142 }, { "epoch": 25.735202492211837, "grad_norm": 2.15625, "learning_rate": 4.824927027084663e-05, "loss": 0.2192, "step": 4143 }, { "epoch": 25.741433021806852, "grad_norm": 2.046875, "learning_rate": 4.824836439670255e-05, "loss": 0.227, "step": 4144 }, { "epoch": 25.747663551401867, "grad_norm": 2.34375, "learning_rate": 4.824745829676571e-05, "loss": 0.1908, "step": 4145 }, { "epoch": 25.753894080996886, "grad_norm": 1.8125, "learning_rate": 4.8246551971044906e-05, "loss": 0.1968, "step": 4146 }, { "epoch": 25.7601246105919, "grad_norm": 2.140625, "learning_rate": 4.8245645419548945e-05, "loss": 0.1974, "step": 4147 }, { "epoch": 25.766355140186917, "grad_norm": 3.0625, "learning_rate": 4.824473864228663e-05, "loss": 0.3448, "step": 4148 }, { "epoch": 25.772585669781932, "grad_norm": 3.0, "learning_rate": 4.824383163926676e-05, "loss": 0.1822, "step": 4149 }, { "epoch": 25.778816199376948, "grad_norm": 3.25, "learning_rate": 4.824292441049816e-05, "loss": 0.1543, "step": 4150 }, { "epoch": 25.785046728971963, "grad_norm": 2.1875, "learning_rate": 4.824201695598963e-05, "loss": 0.2074, "step": 4151 }, { "epoch": 25.79127725856698, "grad_norm": 5.21875, "learning_rate": 4.8241109275749976e-05, "loss": 0.2024, "step": 4152 }, { "epoch": 25.797507788161994, "grad_norm": 3.53125, "learning_rate": 4.824020136978803e-05, "loss": 0.2855, "step": 4153 }, { "epoch": 25.80373831775701, "grad_norm": 3.59375, "learning_rate": 4.82392932381126e-05, "loss": 0.2518, "step": 4154 }, { "epoch": 25.809968847352025, "grad_norm": 2.25, "learning_rate": 4.823838488073251e-05, "loss": 0.1656, "step": 4155 }, { "epoch": 25.81619937694704, "grad_norm": 2.390625, "learning_rate": 4.823747629765659e-05, "loss": 0.1913, "step": 4156 }, { "epoch": 25.822429906542055, "grad_norm": 2.15625, "learning_rate": 4.823656748889364e-05, "loss": 0.1622, "step": 4157 }, { "epoch": 25.82866043613707, "grad_norm": 4.15625, "learning_rate": 4.823565845445251e-05, "loss": 0.2439, "step": 4158 }, { "epoch": 25.834890965732086, "grad_norm": 3.125, "learning_rate": 4.823474919434202e-05, "loss": 0.2527, "step": 4159 }, { "epoch": 25.8411214953271, "grad_norm": 2.65625, "learning_rate": 4.8233839708571e-05, "loss": 0.2993, "step": 4160 }, { "epoch": 25.847352024922117, "grad_norm": 2.578125, "learning_rate": 4.8232929997148276e-05, "loss": 0.2843, "step": 4161 }, { "epoch": 25.853582554517136, "grad_norm": 3.21875, "learning_rate": 4.8232020060082706e-05, "loss": 0.3185, "step": 4162 }, { "epoch": 25.85981308411215, "grad_norm": 2.40625, "learning_rate": 4.823110989738311e-05, "loss": 0.1663, "step": 4163 }, { "epoch": 25.866043613707166, "grad_norm": 2.359375, "learning_rate": 4.8230199509058326e-05, "loss": 0.1881, "step": 4164 }, { "epoch": 25.87227414330218, "grad_norm": 2.25, "learning_rate": 4.82292888951172e-05, "loss": 0.1935, "step": 4165 }, { "epoch": 25.878504672897197, "grad_norm": 3.15625, "learning_rate": 4.822837805556858e-05, "loss": 0.1472, "step": 4166 }, { "epoch": 25.884735202492212, "grad_norm": 3.8125, "learning_rate": 4.822746699042131e-05, "loss": 0.348, "step": 4167 }, { "epoch": 25.890965732087228, "grad_norm": 2.984375, "learning_rate": 4.822655569968424e-05, "loss": 0.228, "step": 4168 }, { "epoch": 25.897196261682243, "grad_norm": 2.5625, "learning_rate": 4.8225644183366216e-05, "loss": 0.1748, "step": 4169 }, { "epoch": 25.90342679127726, "grad_norm": 4.4375, "learning_rate": 4.8224732441476086e-05, "loss": 0.2482, "step": 4170 }, { "epoch": 25.909657320872274, "grad_norm": 1.9921875, "learning_rate": 4.8223820474022716e-05, "loss": 0.1526, "step": 4171 }, { "epoch": 25.91588785046729, "grad_norm": 2.03125, "learning_rate": 4.822290828101496e-05, "loss": 0.1428, "step": 4172 }, { "epoch": 25.922118380062305, "grad_norm": 3.265625, "learning_rate": 4.822199586246168e-05, "loss": 0.1863, "step": 4173 }, { "epoch": 25.92834890965732, "grad_norm": 2.0625, "learning_rate": 4.822108321837173e-05, "loss": 0.2521, "step": 4174 }, { "epoch": 25.934579439252335, "grad_norm": 3.765625, "learning_rate": 4.822017034875398e-05, "loss": 0.1658, "step": 4175 }, { "epoch": 25.94080996884735, "grad_norm": 3.296875, "learning_rate": 4.821925725361729e-05, "loss": 0.3191, "step": 4176 }, { "epoch": 25.947040498442366, "grad_norm": 3.390625, "learning_rate": 4.8218343932970535e-05, "loss": 0.2976, "step": 4177 }, { "epoch": 25.953271028037385, "grad_norm": 2.671875, "learning_rate": 4.8217430386822585e-05, "loss": 0.1234, "step": 4178 }, { "epoch": 25.9595015576324, "grad_norm": 3.671875, "learning_rate": 4.821651661518231e-05, "loss": 0.2349, "step": 4179 }, { "epoch": 25.965732087227416, "grad_norm": 1.4296875, "learning_rate": 4.821560261805858e-05, "loss": 0.1155, "step": 4180 }, { "epoch": 25.97196261682243, "grad_norm": 1.6640625, "learning_rate": 4.821468839546028e-05, "loss": 0.1745, "step": 4181 }, { "epoch": 25.978193146417446, "grad_norm": 4.3125, "learning_rate": 4.8213773947396284e-05, "loss": 0.2627, "step": 4182 }, { "epoch": 25.98442367601246, "grad_norm": 4.53125, "learning_rate": 4.8212859273875474e-05, "loss": 0.4228, "step": 4183 }, { "epoch": 25.990654205607477, "grad_norm": 4.5, "learning_rate": 4.8211944374906735e-05, "loss": 0.2879, "step": 4184 }, { "epoch": 25.996884735202492, "grad_norm": 2.125, "learning_rate": 4.821102925049896e-05, "loss": 0.2256, "step": 4185 }, { "epoch": 26.0, "grad_norm": 1.453125, "learning_rate": 4.821011390066102e-05, "loss": 0.1276, "step": 4186 }, { "epoch": 26.006230529595015, "grad_norm": 3.71875, "learning_rate": 4.8209198325401815e-05, "loss": 0.211, "step": 4187 }, { "epoch": 26.01246105919003, "grad_norm": 3.046875, "learning_rate": 4.8208282524730244e-05, "loss": 0.1594, "step": 4188 }, { "epoch": 26.018691588785046, "grad_norm": 3.828125, "learning_rate": 4.820736649865518e-05, "loss": 0.2208, "step": 4189 }, { "epoch": 26.02492211838006, "grad_norm": 3.125, "learning_rate": 4.820645024718555e-05, "loss": 0.2754, "step": 4190 }, { "epoch": 26.031152647975077, "grad_norm": 2.703125, "learning_rate": 4.8205533770330226e-05, "loss": 0.1459, "step": 4191 }, { "epoch": 26.037383177570092, "grad_norm": 2.109375, "learning_rate": 4.8204617068098124e-05, "loss": 0.1536, "step": 4192 }, { "epoch": 26.043613707165107, "grad_norm": 3.28125, "learning_rate": 4.8203700140498135e-05, "loss": 0.2202, "step": 4193 }, { "epoch": 26.049844236760123, "grad_norm": 2.75, "learning_rate": 4.820278298753918e-05, "loss": 0.1792, "step": 4194 }, { "epoch": 26.05607476635514, "grad_norm": 1.59375, "learning_rate": 4.8201865609230156e-05, "loss": 0.1266, "step": 4195 }, { "epoch": 26.062305295950157, "grad_norm": 1.2265625, "learning_rate": 4.8200948005579974e-05, "loss": 0.1295, "step": 4196 }, { "epoch": 26.068535825545172, "grad_norm": 3.765625, "learning_rate": 4.820003017659755e-05, "loss": 0.3194, "step": 4197 }, { "epoch": 26.074766355140188, "grad_norm": 3.0, "learning_rate": 4.819911212229179e-05, "loss": 0.2123, "step": 4198 }, { "epoch": 26.080996884735203, "grad_norm": 2.46875, "learning_rate": 4.8198193842671625e-05, "loss": 0.2313, "step": 4199 }, { "epoch": 26.08722741433022, "grad_norm": 3.6875, "learning_rate": 4.819727533774595e-05, "loss": 0.3561, "step": 4200 }, { "epoch": 26.093457943925234, "grad_norm": 1.8984375, "learning_rate": 4.8196356607523713e-05, "loss": 0.1722, "step": 4201 }, { "epoch": 26.09968847352025, "grad_norm": 2.703125, "learning_rate": 4.819543765201382e-05, "loss": 0.2561, "step": 4202 }, { "epoch": 26.105919003115265, "grad_norm": 2.96875, "learning_rate": 4.8194518471225194e-05, "loss": 0.185, "step": 4203 }, { "epoch": 26.11214953271028, "grad_norm": 2.640625, "learning_rate": 4.8193599065166774e-05, "loss": 0.1259, "step": 4204 }, { "epoch": 26.118380062305295, "grad_norm": 2.0625, "learning_rate": 4.819267943384748e-05, "loss": 0.22, "step": 4205 }, { "epoch": 26.12461059190031, "grad_norm": 2.953125, "learning_rate": 4.819175957727625e-05, "loss": 0.2954, "step": 4206 }, { "epoch": 26.130841121495326, "grad_norm": 3.765625, "learning_rate": 4.8190839495462015e-05, "loss": 0.2288, "step": 4207 }, { "epoch": 26.13707165109034, "grad_norm": 2.4375, "learning_rate": 4.8189919188413714e-05, "loss": 0.1484, "step": 4208 }, { "epoch": 26.143302180685357, "grad_norm": 1.9921875, "learning_rate": 4.818899865614028e-05, "loss": 0.1493, "step": 4209 }, { "epoch": 26.149532710280372, "grad_norm": 3.46875, "learning_rate": 4.818807789865065e-05, "loss": 0.2403, "step": 4210 }, { "epoch": 26.15576323987539, "grad_norm": 3.59375, "learning_rate": 4.818715691595378e-05, "loss": 0.2679, "step": 4211 }, { "epoch": 26.161993769470406, "grad_norm": 2.34375, "learning_rate": 4.8186235708058604e-05, "loss": 0.1703, "step": 4212 }, { "epoch": 26.16822429906542, "grad_norm": 2.8125, "learning_rate": 4.818531427497407e-05, "loss": 0.1923, "step": 4213 }, { "epoch": 26.174454828660437, "grad_norm": 3.171875, "learning_rate": 4.818439261670913e-05, "loss": 0.2383, "step": 4214 }, { "epoch": 26.180685358255452, "grad_norm": 2.25, "learning_rate": 4.818347073327274e-05, "loss": 0.2848, "step": 4215 }, { "epoch": 26.186915887850468, "grad_norm": 3.1875, "learning_rate": 4.818254862467384e-05, "loss": 0.1999, "step": 4216 }, { "epoch": 26.193146417445483, "grad_norm": 2.234375, "learning_rate": 4.81816262909214e-05, "loss": 0.1686, "step": 4217 }, { "epoch": 26.1993769470405, "grad_norm": 3.109375, "learning_rate": 4.818070373202437e-05, "loss": 0.238, "step": 4218 }, { "epoch": 26.205607476635514, "grad_norm": 3.171875, "learning_rate": 4.8179780947991707e-05, "loss": 0.1783, "step": 4219 }, { "epoch": 26.21183800623053, "grad_norm": 2.828125, "learning_rate": 4.8178857938832375e-05, "loss": 0.2703, "step": 4220 }, { "epoch": 26.218068535825545, "grad_norm": 2.171875, "learning_rate": 4.817793470455535e-05, "loss": 0.174, "step": 4221 }, { "epoch": 26.22429906542056, "grad_norm": 2.15625, "learning_rate": 4.817701124516959e-05, "loss": 0.2456, "step": 4222 }, { "epoch": 26.230529595015575, "grad_norm": 2.015625, "learning_rate": 4.817608756068406e-05, "loss": 0.1523, "step": 4223 }, { "epoch": 26.23676012461059, "grad_norm": 3.03125, "learning_rate": 4.817516365110773e-05, "loss": 0.2035, "step": 4224 }, { "epoch": 26.242990654205606, "grad_norm": 2.75, "learning_rate": 4.817423951644959e-05, "loss": 0.2002, "step": 4225 }, { "epoch": 26.24922118380062, "grad_norm": 2.65625, "learning_rate": 4.81733151567186e-05, "loss": 0.2111, "step": 4226 }, { "epoch": 26.25545171339564, "grad_norm": 3.171875, "learning_rate": 4.817239057192374e-05, "loss": 0.3706, "step": 4227 }, { "epoch": 26.261682242990656, "grad_norm": 2.125, "learning_rate": 4.8171465762073986e-05, "loss": 0.2067, "step": 4228 }, { "epoch": 26.26791277258567, "grad_norm": 1.7734375, "learning_rate": 4.8170540727178326e-05, "loss": 0.1449, "step": 4229 }, { "epoch": 26.274143302180686, "grad_norm": 2.53125, "learning_rate": 4.8169615467245744e-05, "loss": 0.3467, "step": 4230 }, { "epoch": 26.2803738317757, "grad_norm": 2.359375, "learning_rate": 4.816868998228523e-05, "loss": 0.1806, "step": 4231 }, { "epoch": 26.286604361370717, "grad_norm": 2.03125, "learning_rate": 4.816776427230576e-05, "loss": 0.1768, "step": 4232 }, { "epoch": 26.292834890965732, "grad_norm": 2.453125, "learning_rate": 4.8166838337316334e-05, "loss": 0.1829, "step": 4233 }, { "epoch": 26.299065420560748, "grad_norm": 2.515625, "learning_rate": 4.8165912177325946e-05, "loss": 0.1738, "step": 4234 }, { "epoch": 26.305295950155763, "grad_norm": 2.875, "learning_rate": 4.8164985792343583e-05, "loss": 0.3271, "step": 4235 }, { "epoch": 26.31152647975078, "grad_norm": 2.90625, "learning_rate": 4.816405918237825e-05, "loss": 0.2049, "step": 4236 }, { "epoch": 26.317757009345794, "grad_norm": 1.6328125, "learning_rate": 4.816313234743895e-05, "loss": 0.1204, "step": 4237 }, { "epoch": 26.32398753894081, "grad_norm": 3.46875, "learning_rate": 4.816220528753467e-05, "loss": 0.2743, "step": 4238 }, { "epoch": 26.330218068535824, "grad_norm": 2.46875, "learning_rate": 4.8161278002674424e-05, "loss": 0.1604, "step": 4239 }, { "epoch": 26.33644859813084, "grad_norm": 3.046875, "learning_rate": 4.8160350492867225e-05, "loss": 0.2139, "step": 4240 }, { "epoch": 26.342679127725855, "grad_norm": 4.15625, "learning_rate": 4.8159422758122066e-05, "loss": 0.2545, "step": 4241 }, { "epoch": 26.34890965732087, "grad_norm": 3.359375, "learning_rate": 4.815849479844796e-05, "loss": 0.2656, "step": 4242 }, { "epoch": 26.35514018691589, "grad_norm": 2.015625, "learning_rate": 4.815756661385392e-05, "loss": 0.1708, "step": 4243 }, { "epoch": 26.361370716510905, "grad_norm": 1.5078125, "learning_rate": 4.815663820434897e-05, "loss": 0.1422, "step": 4244 }, { "epoch": 26.36760124610592, "grad_norm": 3.09375, "learning_rate": 4.815570956994212e-05, "loss": 0.2563, "step": 4245 }, { "epoch": 26.373831775700936, "grad_norm": 2.875, "learning_rate": 4.8154780710642395e-05, "loss": 0.2519, "step": 4246 }, { "epoch": 26.38006230529595, "grad_norm": 2.59375, "learning_rate": 4.8153851626458804e-05, "loss": 0.2022, "step": 4247 }, { "epoch": 26.386292834890966, "grad_norm": 2.421875, "learning_rate": 4.815292231740039e-05, "loss": 0.171, "step": 4248 }, { "epoch": 26.39252336448598, "grad_norm": 2.078125, "learning_rate": 4.815199278347615e-05, "loss": 0.1298, "step": 4249 }, { "epoch": 26.398753894080997, "grad_norm": 2.875, "learning_rate": 4.815106302469513e-05, "loss": 0.2098, "step": 4250 }, { "epoch": 26.404984423676012, "grad_norm": 2.640625, "learning_rate": 4.815013304106637e-05, "loss": 0.2, "step": 4251 }, { "epoch": 26.411214953271028, "grad_norm": 2.625, "learning_rate": 4.8149202832598884e-05, "loss": 0.1445, "step": 4252 }, { "epoch": 26.417445482866043, "grad_norm": 3.640625, "learning_rate": 4.814827239930171e-05, "loss": 0.3074, "step": 4253 }, { "epoch": 26.42367601246106, "grad_norm": 3.171875, "learning_rate": 4.814734174118388e-05, "loss": 0.2151, "step": 4254 }, { "epoch": 26.429906542056074, "grad_norm": 3.0625, "learning_rate": 4.814641085825445e-05, "loss": 0.2834, "step": 4255 }, { "epoch": 26.43613707165109, "grad_norm": 2.703125, "learning_rate": 4.814547975052245e-05, "loss": 0.1894, "step": 4256 }, { "epoch": 26.442367601246104, "grad_norm": 1.828125, "learning_rate": 4.814454841799693e-05, "loss": 0.144, "step": 4257 }, { "epoch": 26.44859813084112, "grad_norm": 2.984375, "learning_rate": 4.814361686068692e-05, "loss": 0.3853, "step": 4258 }, { "epoch": 26.45482866043614, "grad_norm": 3.453125, "learning_rate": 4.814268507860148e-05, "loss": 0.3309, "step": 4259 }, { "epoch": 26.461059190031154, "grad_norm": 3.5, "learning_rate": 4.814175307174966e-05, "loss": 0.2561, "step": 4260 }, { "epoch": 26.46728971962617, "grad_norm": 2.296875, "learning_rate": 4.81408208401405e-05, "loss": 0.124, "step": 4261 }, { "epoch": 26.473520249221185, "grad_norm": 4.5, "learning_rate": 4.813988838378307e-05, "loss": 0.4667, "step": 4262 }, { "epoch": 26.4797507788162, "grad_norm": 4.0625, "learning_rate": 4.8138955702686415e-05, "loss": 0.2996, "step": 4263 }, { "epoch": 26.485981308411215, "grad_norm": 3.296875, "learning_rate": 4.8138022796859593e-05, "loss": 0.2158, "step": 4264 }, { "epoch": 26.49221183800623, "grad_norm": 3.8125, "learning_rate": 4.813708966631167e-05, "loss": 0.2842, "step": 4265 }, { "epoch": 26.498442367601246, "grad_norm": 2.875, "learning_rate": 4.8136156311051716e-05, "loss": 0.2934, "step": 4266 }, { "epoch": 26.50467289719626, "grad_norm": 1.953125, "learning_rate": 4.813522273108878e-05, "loss": 0.1566, "step": 4267 }, { "epoch": 26.510903426791277, "grad_norm": 2.375, "learning_rate": 4.813428892643194e-05, "loss": 0.1717, "step": 4268 }, { "epoch": 26.517133956386292, "grad_norm": 3.0, "learning_rate": 4.813335489709025e-05, "loss": 0.2037, "step": 4269 }, { "epoch": 26.523364485981308, "grad_norm": 2.421875, "learning_rate": 4.813242064307281e-05, "loss": 0.164, "step": 4270 }, { "epoch": 26.529595015576323, "grad_norm": 2.34375, "learning_rate": 4.8131486164388665e-05, "loss": 0.1383, "step": 4271 }, { "epoch": 26.53582554517134, "grad_norm": 2.078125, "learning_rate": 4.81305514610469e-05, "loss": 0.2159, "step": 4272 }, { "epoch": 26.542056074766354, "grad_norm": 1.921875, "learning_rate": 4.81296165330566e-05, "loss": 0.146, "step": 4273 }, { "epoch": 26.54828660436137, "grad_norm": 2.890625, "learning_rate": 4.8128681380426844e-05, "loss": 0.2169, "step": 4274 }, { "epoch": 26.554517133956388, "grad_norm": 2.875, "learning_rate": 4.8127746003166706e-05, "loss": 0.1691, "step": 4275 }, { "epoch": 26.560747663551403, "grad_norm": 3.1875, "learning_rate": 4.812681040128528e-05, "loss": 0.1981, "step": 4276 }, { "epoch": 26.56697819314642, "grad_norm": 4.09375, "learning_rate": 4.8125874574791654e-05, "loss": 0.3833, "step": 4277 }, { "epoch": 26.573208722741434, "grad_norm": 4.34375, "learning_rate": 4.81249385236949e-05, "loss": 0.3071, "step": 4278 }, { "epoch": 26.57943925233645, "grad_norm": 2.59375, "learning_rate": 4.8124002248004126e-05, "loss": 0.1824, "step": 4279 }, { "epoch": 26.585669781931465, "grad_norm": 3.0, "learning_rate": 4.8123065747728415e-05, "loss": 0.2542, "step": 4280 }, { "epoch": 26.59190031152648, "grad_norm": 2.9375, "learning_rate": 4.812212902287687e-05, "loss": 0.24, "step": 4281 }, { "epoch": 26.598130841121495, "grad_norm": 3.578125, "learning_rate": 4.812119207345859e-05, "loss": 0.2912, "step": 4282 }, { "epoch": 26.60436137071651, "grad_norm": 2.625, "learning_rate": 4.8120254899482665e-05, "loss": 0.2295, "step": 4283 }, { "epoch": 26.610591900311526, "grad_norm": 2.078125, "learning_rate": 4.811931750095821e-05, "loss": 0.1601, "step": 4284 }, { "epoch": 26.61682242990654, "grad_norm": 2.578125, "learning_rate": 4.8118379877894314e-05, "loss": 0.1568, "step": 4285 }, { "epoch": 26.623052959501557, "grad_norm": 2.71875, "learning_rate": 4.811744203030009e-05, "loss": 0.331, "step": 4286 }, { "epoch": 26.629283489096572, "grad_norm": 2.40625, "learning_rate": 4.811650395818466e-05, "loss": 0.1562, "step": 4287 }, { "epoch": 26.635514018691588, "grad_norm": 3.09375, "learning_rate": 4.811556566155711e-05, "loss": 0.2542, "step": 4288 }, { "epoch": 26.641744548286603, "grad_norm": 3.375, "learning_rate": 4.811462714042657e-05, "loss": 0.2265, "step": 4289 }, { "epoch": 26.64797507788162, "grad_norm": 2.125, "learning_rate": 4.8113688394802145e-05, "loss": 0.2257, "step": 4290 }, { "epoch": 26.654205607476637, "grad_norm": 2.734375, "learning_rate": 4.8112749424692974e-05, "loss": 0.2476, "step": 4291 }, { "epoch": 26.660436137071652, "grad_norm": 2.046875, "learning_rate": 4.8111810230108145e-05, "loss": 0.1926, "step": 4292 }, { "epoch": 26.666666666666668, "grad_norm": 2.734375, "learning_rate": 4.81108708110568e-05, "loss": 0.2884, "step": 4293 }, { "epoch": 26.672897196261683, "grad_norm": 2.84375, "learning_rate": 4.810993116754806e-05, "loss": 0.3984, "step": 4294 }, { "epoch": 26.6791277258567, "grad_norm": 3.5625, "learning_rate": 4.810899129959105e-05, "loss": 0.3946, "step": 4295 }, { "epoch": 26.685358255451714, "grad_norm": 3.578125, "learning_rate": 4.810805120719489e-05, "loss": 0.348, "step": 4296 }, { "epoch": 26.69158878504673, "grad_norm": 2.65625, "learning_rate": 4.8107110890368725e-05, "loss": 0.1985, "step": 4297 }, { "epoch": 26.697819314641745, "grad_norm": 3.15625, "learning_rate": 4.810617034912168e-05, "loss": 0.3615, "step": 4298 }, { "epoch": 26.70404984423676, "grad_norm": 2.671875, "learning_rate": 4.8105229583462885e-05, "loss": 0.3404, "step": 4299 }, { "epoch": 26.710280373831775, "grad_norm": 3.421875, "learning_rate": 4.810428859340148e-05, "loss": 0.2046, "step": 4300 }, { "epoch": 26.71651090342679, "grad_norm": 1.7890625, "learning_rate": 4.810334737894661e-05, "loss": 0.1383, "step": 4301 }, { "epoch": 26.722741433021806, "grad_norm": 3.46875, "learning_rate": 4.810240594010742e-05, "loss": 0.3612, "step": 4302 }, { "epoch": 26.72897196261682, "grad_norm": 3.765625, "learning_rate": 4.8101464276893026e-05, "loss": 0.3771, "step": 4303 }, { "epoch": 26.735202492211837, "grad_norm": 1.53125, "learning_rate": 4.810052238931261e-05, "loss": 0.1533, "step": 4304 }, { "epoch": 26.741433021806852, "grad_norm": 2.265625, "learning_rate": 4.809958027737529e-05, "loss": 0.2191, "step": 4305 }, { "epoch": 26.747663551401867, "grad_norm": 3.703125, "learning_rate": 4.809863794109024e-05, "loss": 0.3429, "step": 4306 }, { "epoch": 26.753894080996886, "grad_norm": 3.375, "learning_rate": 4.80976953804666e-05, "loss": 0.2708, "step": 4307 }, { "epoch": 26.7601246105919, "grad_norm": 1.5234375, "learning_rate": 4.809675259551352e-05, "loss": 0.15, "step": 4308 }, { "epoch": 26.766355140186917, "grad_norm": 4.40625, "learning_rate": 4.809580958624016e-05, "loss": 0.1531, "step": 4309 }, { "epoch": 26.772585669781932, "grad_norm": 2.359375, "learning_rate": 4.809486635265569e-05, "loss": 0.1345, "step": 4310 }, { "epoch": 26.778816199376948, "grad_norm": 2.453125, "learning_rate": 4.809392289476925e-05, "loss": 0.1728, "step": 4311 }, { "epoch": 26.785046728971963, "grad_norm": 4.03125, "learning_rate": 4.809297921259002e-05, "loss": 0.1882, "step": 4312 }, { "epoch": 26.79127725856698, "grad_norm": 2.671875, "learning_rate": 4.809203530612716e-05, "loss": 0.2086, "step": 4313 }, { "epoch": 26.797507788161994, "grad_norm": 2.578125, "learning_rate": 4.8091091175389833e-05, "loss": 0.2326, "step": 4314 }, { "epoch": 26.80373831775701, "grad_norm": 2.015625, "learning_rate": 4.8090146820387223e-05, "loss": 0.177, "step": 4315 }, { "epoch": 26.809968847352025, "grad_norm": 1.5234375, "learning_rate": 4.808920224112849e-05, "loss": 0.1189, "step": 4316 }, { "epoch": 26.81619937694704, "grad_norm": 1.9296875, "learning_rate": 4.80882574376228e-05, "loss": 0.138, "step": 4317 }, { "epoch": 26.822429906542055, "grad_norm": 4.71875, "learning_rate": 4.808731240987934e-05, "loss": 0.3662, "step": 4318 }, { "epoch": 26.82866043613707, "grad_norm": 4.46875, "learning_rate": 4.808636715790729e-05, "loss": 0.1757, "step": 4319 }, { "epoch": 26.834890965732086, "grad_norm": 3.578125, "learning_rate": 4.808542168171582e-05, "loss": 0.1778, "step": 4320 }, { "epoch": 26.8411214953271, "grad_norm": 3.046875, "learning_rate": 4.808447598131414e-05, "loss": 0.3024, "step": 4321 }, { "epoch": 26.847352024922117, "grad_norm": 3.265625, "learning_rate": 4.8083530056711394e-05, "loss": 0.1967, "step": 4322 }, { "epoch": 26.853582554517136, "grad_norm": 4.4375, "learning_rate": 4.80825839079168e-05, "loss": 0.4271, "step": 4323 }, { "epoch": 26.85981308411215, "grad_norm": 3.734375, "learning_rate": 4.8081637534939536e-05, "loss": 0.2163, "step": 4324 }, { "epoch": 26.866043613707166, "grad_norm": 2.796875, "learning_rate": 4.808069093778879e-05, "loss": 0.1776, "step": 4325 }, { "epoch": 26.87227414330218, "grad_norm": 3.078125, "learning_rate": 4.8079744116473766e-05, "loss": 0.2281, "step": 4326 }, { "epoch": 26.878504672897197, "grad_norm": 2.421875, "learning_rate": 4.8078797071003644e-05, "loss": 0.2057, "step": 4327 }, { "epoch": 26.884735202492212, "grad_norm": 2.5, "learning_rate": 4.807784980138764e-05, "loss": 0.2007, "step": 4328 }, { "epoch": 26.890965732087228, "grad_norm": 3.046875, "learning_rate": 4.807690230763494e-05, "loss": 0.2103, "step": 4329 }, { "epoch": 26.897196261682243, "grad_norm": 5.09375, "learning_rate": 4.807595458975476e-05, "loss": 0.2703, "step": 4330 }, { "epoch": 26.90342679127726, "grad_norm": 2.03125, "learning_rate": 4.8075006647756296e-05, "loss": 0.171, "step": 4331 }, { "epoch": 26.909657320872274, "grad_norm": 2.15625, "learning_rate": 4.8074058481648744e-05, "loss": 0.1626, "step": 4332 }, { "epoch": 26.91588785046729, "grad_norm": 3.84375, "learning_rate": 4.807311009144133e-05, "loss": 0.4179, "step": 4333 }, { "epoch": 26.922118380062305, "grad_norm": 4.875, "learning_rate": 4.807216147714326e-05, "loss": 0.382, "step": 4334 }, { "epoch": 26.92834890965732, "grad_norm": 2.40625, "learning_rate": 4.807121263876374e-05, "loss": 0.1648, "step": 4335 }, { "epoch": 26.934579439252335, "grad_norm": 2.5, "learning_rate": 4.8070263576312e-05, "loss": 0.1578, "step": 4336 }, { "epoch": 26.94080996884735, "grad_norm": 2.171875, "learning_rate": 4.806931428979724e-05, "loss": 0.1729, "step": 4337 }, { "epoch": 26.947040498442366, "grad_norm": 3.03125, "learning_rate": 4.806836477922869e-05, "loss": 0.3472, "step": 4338 }, { "epoch": 26.953271028037385, "grad_norm": 3.859375, "learning_rate": 4.806741504461557e-05, "loss": 0.131, "step": 4339 }, { "epoch": 26.9595015576324, "grad_norm": 3.90625, "learning_rate": 4.8066465085967104e-05, "loss": 0.3222, "step": 4340 }, { "epoch": 26.965732087227416, "grad_norm": 1.8984375, "learning_rate": 4.8065514903292524e-05, "loss": 0.1606, "step": 4341 }, { "epoch": 26.97196261682243, "grad_norm": 2.0, "learning_rate": 4.806456449660105e-05, "loss": 0.1784, "step": 4342 }, { "epoch": 26.978193146417446, "grad_norm": 2.8125, "learning_rate": 4.8063613865901915e-05, "loss": 0.275, "step": 4343 }, { "epoch": 26.98442367601246, "grad_norm": 1.8828125, "learning_rate": 4.8062663011204344e-05, "loss": 0.1473, "step": 4344 }, { "epoch": 26.990654205607477, "grad_norm": 2.1875, "learning_rate": 4.806171193251759e-05, "loss": 0.2095, "step": 4345 }, { "epoch": 26.996884735202492, "grad_norm": 2.171875, "learning_rate": 4.806076062985088e-05, "loss": 0.1583, "step": 4346 }, { "epoch": 27.0, "grad_norm": 2.125, "learning_rate": 4.8059809103213446e-05, "loss": 0.1393, "step": 4347 }, { "epoch": 27.006230529595015, "grad_norm": 2.78125, "learning_rate": 4.805885735261454e-05, "loss": 0.1776, "step": 4348 }, { "epoch": 27.01246105919003, "grad_norm": 1.90625, "learning_rate": 4.80579053780634e-05, "loss": 0.1805, "step": 4349 }, { "epoch": 27.018691588785046, "grad_norm": 1.8984375, "learning_rate": 4.805695317956927e-05, "loss": 0.134, "step": 4350 }, { "epoch": 27.02492211838006, "grad_norm": 2.75, "learning_rate": 4.805600075714141e-05, "loss": 0.1874, "step": 4351 }, { "epoch": 27.031152647975077, "grad_norm": 3.0, "learning_rate": 4.805504811078906e-05, "loss": 0.1598, "step": 4352 }, { "epoch": 27.037383177570092, "grad_norm": 3.40625, "learning_rate": 4.805409524052147e-05, "loss": 0.1592, "step": 4353 }, { "epoch": 27.043613707165107, "grad_norm": 2.078125, "learning_rate": 4.8053142146347904e-05, "loss": 0.1483, "step": 4354 }, { "epoch": 27.049844236760123, "grad_norm": 2.78125, "learning_rate": 4.805218882827761e-05, "loss": 0.2776, "step": 4355 }, { "epoch": 27.05607476635514, "grad_norm": 2.140625, "learning_rate": 4.805123528631985e-05, "loss": 0.1519, "step": 4356 }, { "epoch": 27.062305295950157, "grad_norm": 2.90625, "learning_rate": 4.805028152048389e-05, "loss": 0.2183, "step": 4357 }, { "epoch": 27.068535825545172, "grad_norm": 3.484375, "learning_rate": 4.804932753077898e-05, "loss": 0.268, "step": 4358 }, { "epoch": 27.074766355140188, "grad_norm": 2.8125, "learning_rate": 4.804837331721439e-05, "loss": 0.1924, "step": 4359 }, { "epoch": 27.080996884735203, "grad_norm": 2.546875, "learning_rate": 4.8047418879799407e-05, "loss": 0.2164, "step": 4360 }, { "epoch": 27.08722741433022, "grad_norm": 2.234375, "learning_rate": 4.804646421854327e-05, "loss": 0.2088, "step": 4361 }, { "epoch": 27.093457943925234, "grad_norm": 2.859375, "learning_rate": 4.804550933345527e-05, "loss": 0.1728, "step": 4362 }, { "epoch": 27.09968847352025, "grad_norm": 3.15625, "learning_rate": 4.8044554224544685e-05, "loss": 0.2281, "step": 4363 }, { "epoch": 27.105919003115265, "grad_norm": 2.8125, "learning_rate": 4.804359889182077e-05, "loss": 0.2204, "step": 4364 }, { "epoch": 27.11214953271028, "grad_norm": 3.109375, "learning_rate": 4.8042643335292826e-05, "loss": 0.2454, "step": 4365 }, { "epoch": 27.118380062305295, "grad_norm": 3.59375, "learning_rate": 4.804168755497013e-05, "loss": 0.2613, "step": 4366 }, { "epoch": 27.12461059190031, "grad_norm": 3.453125, "learning_rate": 4.804073155086194e-05, "loss": 0.1652, "step": 4367 }, { "epoch": 27.130841121495326, "grad_norm": 2.90625, "learning_rate": 4.803977532297757e-05, "loss": 0.4108, "step": 4368 }, { "epoch": 27.13707165109034, "grad_norm": 4.0, "learning_rate": 4.8038818871326293e-05, "loss": 0.2879, "step": 4369 }, { "epoch": 27.143302180685357, "grad_norm": 1.796875, "learning_rate": 4.8037862195917405e-05, "loss": 0.1616, "step": 4370 }, { "epoch": 27.149532710280372, "grad_norm": 2.609375, "learning_rate": 4.803690529676019e-05, "loss": 0.1616, "step": 4371 }, { "epoch": 27.15576323987539, "grad_norm": 3.796875, "learning_rate": 4.803594817386395e-05, "loss": 0.407, "step": 4372 }, { "epoch": 27.161993769470406, "grad_norm": 3.140625, "learning_rate": 4.803499082723798e-05, "loss": 0.3402, "step": 4373 }, { "epoch": 27.16822429906542, "grad_norm": 2.84375, "learning_rate": 4.8034033256891566e-05, "loss": 0.3454, "step": 4374 }, { "epoch": 27.174454828660437, "grad_norm": 1.7578125, "learning_rate": 4.803307546283402e-05, "loss": 0.1535, "step": 4375 }, { "epoch": 27.180685358255452, "grad_norm": 2.375, "learning_rate": 4.8032117445074645e-05, "loss": 0.3072, "step": 4376 }, { "epoch": 27.186915887850468, "grad_norm": 2.515625, "learning_rate": 4.803115920362273e-05, "loss": 0.2453, "step": 4377 }, { "epoch": 27.193146417445483, "grad_norm": 1.5390625, "learning_rate": 4.8030200738487606e-05, "loss": 0.1386, "step": 4378 }, { "epoch": 27.1993769470405, "grad_norm": 1.6875, "learning_rate": 4.802924204967856e-05, "loss": 0.1702, "step": 4379 }, { "epoch": 27.205607476635514, "grad_norm": 2.484375, "learning_rate": 4.802828313720491e-05, "loss": 0.199, "step": 4380 }, { "epoch": 27.21183800623053, "grad_norm": 2.71875, "learning_rate": 4.802732400107598e-05, "loss": 0.217, "step": 4381 }, { "epoch": 27.218068535825545, "grad_norm": 3.125, "learning_rate": 4.802636464130107e-05, "loss": 0.2218, "step": 4382 }, { "epoch": 27.22429906542056, "grad_norm": 1.75, "learning_rate": 4.802540505788951e-05, "loss": 0.1526, "step": 4383 }, { "epoch": 27.230529595015575, "grad_norm": 1.8671875, "learning_rate": 4.8024445250850613e-05, "loss": 0.1697, "step": 4384 }, { "epoch": 27.23676012461059, "grad_norm": 2.078125, "learning_rate": 4.80234852201937e-05, "loss": 0.2265, "step": 4385 }, { "epoch": 27.242990654205606, "grad_norm": 2.109375, "learning_rate": 4.8022524965928084e-05, "loss": 0.1928, "step": 4386 }, { "epoch": 27.24922118380062, "grad_norm": 3.640625, "learning_rate": 4.802156448806312e-05, "loss": 0.4256, "step": 4387 }, { "epoch": 27.25545171339564, "grad_norm": 2.171875, "learning_rate": 4.802060378660811e-05, "loss": 0.1881, "step": 4388 }, { "epoch": 27.261682242990656, "grad_norm": 2.875, "learning_rate": 4.80196428615724e-05, "loss": 0.1541, "step": 4389 }, { "epoch": 27.26791277258567, "grad_norm": 1.5234375, "learning_rate": 4.801868171296531e-05, "loss": 0.1339, "step": 4390 }, { "epoch": 27.274143302180686, "grad_norm": 2.953125, "learning_rate": 4.801772034079619e-05, "loss": 0.214, "step": 4391 }, { "epoch": 27.2803738317757, "grad_norm": 3.265625, "learning_rate": 4.801675874507436e-05, "loss": 0.2889, "step": 4392 }, { "epoch": 27.286604361370717, "grad_norm": 1.9609375, "learning_rate": 4.801579692580918e-05, "loss": 0.1668, "step": 4393 }, { "epoch": 27.292834890965732, "grad_norm": 3.3125, "learning_rate": 4.8014834883009966e-05, "loss": 0.302, "step": 4394 }, { "epoch": 27.299065420560748, "grad_norm": 3.125, "learning_rate": 4.8013872616686086e-05, "loss": 0.1942, "step": 4395 }, { "epoch": 27.305295950155763, "grad_norm": 2.84375, "learning_rate": 4.8012910126846866e-05, "loss": 0.1963, "step": 4396 }, { "epoch": 27.31152647975078, "grad_norm": 2.8125, "learning_rate": 4.8011947413501666e-05, "loss": 0.2268, "step": 4397 }, { "epoch": 27.317757009345794, "grad_norm": 2.171875, "learning_rate": 4.801098447665984e-05, "loss": 0.2484, "step": 4398 }, { "epoch": 27.32398753894081, "grad_norm": 2.234375, "learning_rate": 4.801002131633072e-05, "loss": 0.1795, "step": 4399 }, { "epoch": 27.330218068535824, "grad_norm": 2.59375, "learning_rate": 4.800905793252368e-05, "loss": 0.2404, "step": 4400 }, { "epoch": 27.33644859813084, "grad_norm": 2.375, "learning_rate": 4.800809432524808e-05, "loss": 0.1669, "step": 4401 }, { "epoch": 27.342679127725855, "grad_norm": 3.3125, "learning_rate": 4.8007130494513256e-05, "loss": 0.2323, "step": 4402 }, { "epoch": 27.34890965732087, "grad_norm": 2.921875, "learning_rate": 4.800616644032858e-05, "loss": 0.3158, "step": 4403 }, { "epoch": 27.35514018691589, "grad_norm": 2.03125, "learning_rate": 4.800520216270342e-05, "loss": 0.1639, "step": 4404 }, { "epoch": 27.361370716510905, "grad_norm": 1.84375, "learning_rate": 4.800423766164714e-05, "loss": 0.1622, "step": 4405 }, { "epoch": 27.36760124610592, "grad_norm": 2.90625, "learning_rate": 4.800327293716911e-05, "loss": 0.2903, "step": 4406 }, { "epoch": 27.373831775700936, "grad_norm": 1.71875, "learning_rate": 4.8002307989278685e-05, "loss": 0.1535, "step": 4407 }, { "epoch": 27.38006230529595, "grad_norm": 2.421875, "learning_rate": 4.800134281798525e-05, "loss": 0.205, "step": 4408 }, { "epoch": 27.386292834890966, "grad_norm": 1.875, "learning_rate": 4.8000377423298175e-05, "loss": 0.189, "step": 4409 }, { "epoch": 27.39252336448598, "grad_norm": 3.421875, "learning_rate": 4.799941180522683e-05, "loss": 0.2026, "step": 4410 }, { "epoch": 27.398753894080997, "grad_norm": 3.09375, "learning_rate": 4.799844596378061e-05, "loss": 0.1545, "step": 4411 }, { "epoch": 27.404984423676012, "grad_norm": 2.6875, "learning_rate": 4.7997479898968886e-05, "loss": 0.1715, "step": 4412 }, { "epoch": 27.411214953271028, "grad_norm": 4.125, "learning_rate": 4.7996513610801044e-05, "loss": 0.131, "step": 4413 }, { "epoch": 27.417445482866043, "grad_norm": 3.765625, "learning_rate": 4.799554709928646e-05, "loss": 0.1407, "step": 4414 }, { "epoch": 27.42367601246106, "grad_norm": 2.375, "learning_rate": 4.7994580364434524e-05, "loss": 0.2416, "step": 4415 }, { "epoch": 27.429906542056074, "grad_norm": 3.046875, "learning_rate": 4.799361340625463e-05, "loss": 0.207, "step": 4416 }, { "epoch": 27.43613707165109, "grad_norm": 4.34375, "learning_rate": 4.799264622475616e-05, "loss": 0.1739, "step": 4417 }, { "epoch": 27.442367601246104, "grad_norm": 3.375, "learning_rate": 4.7991678819948516e-05, "loss": 0.2707, "step": 4418 }, { "epoch": 27.44859813084112, "grad_norm": 3.28125, "learning_rate": 4.7990711191841095e-05, "loss": 0.1265, "step": 4419 }, { "epoch": 27.45482866043614, "grad_norm": 2.890625, "learning_rate": 4.7989743340443294e-05, "loss": 0.216, "step": 4420 }, { "epoch": 27.461059190031154, "grad_norm": 3.203125, "learning_rate": 4.7988775265764504e-05, "loss": 0.1661, "step": 4421 }, { "epoch": 27.46728971962617, "grad_norm": 2.78125, "learning_rate": 4.798780696781413e-05, "loss": 0.2075, "step": 4422 }, { "epoch": 27.473520249221185, "grad_norm": 3.4375, "learning_rate": 4.798683844660159e-05, "loss": 0.2647, "step": 4423 }, { "epoch": 27.4797507788162, "grad_norm": 3.1875, "learning_rate": 4.798586970213627e-05, "loss": 0.3302, "step": 4424 }, { "epoch": 27.485981308411215, "grad_norm": 2.265625, "learning_rate": 4.7984900734427596e-05, "loss": 0.1714, "step": 4425 }, { "epoch": 27.49221183800623, "grad_norm": 2.671875, "learning_rate": 4.7983931543484967e-05, "loss": 0.1726, "step": 4426 }, { "epoch": 27.498442367601246, "grad_norm": 3.09375, "learning_rate": 4.79829621293178e-05, "loss": 0.324, "step": 4427 }, { "epoch": 27.50467289719626, "grad_norm": 2.296875, "learning_rate": 4.798199249193551e-05, "loss": 0.1868, "step": 4428 }, { "epoch": 27.510903426791277, "grad_norm": 2.71875, "learning_rate": 4.798102263134752e-05, "loss": 0.1808, "step": 4429 }, { "epoch": 27.517133956386292, "grad_norm": 1.8125, "learning_rate": 4.798005254756325e-05, "loss": 0.1314, "step": 4430 }, { "epoch": 27.523364485981308, "grad_norm": 2.3125, "learning_rate": 4.79790822405921e-05, "loss": 0.2478, "step": 4431 }, { "epoch": 27.529595015576323, "grad_norm": 2.125, "learning_rate": 4.797811171044352e-05, "loss": 0.1608, "step": 4432 }, { "epoch": 27.53582554517134, "grad_norm": 2.21875, "learning_rate": 4.797714095712692e-05, "loss": 0.1609, "step": 4433 }, { "epoch": 27.542056074766354, "grad_norm": 2.78125, "learning_rate": 4.797616998065174e-05, "loss": 0.2668, "step": 4434 }, { "epoch": 27.54828660436137, "grad_norm": 1.8671875, "learning_rate": 4.7975198781027406e-05, "loss": 0.1306, "step": 4435 }, { "epoch": 27.554517133956388, "grad_norm": 2.328125, "learning_rate": 4.797422735826334e-05, "loss": 0.184, "step": 4436 }, { "epoch": 27.560747663551403, "grad_norm": 1.6640625, "learning_rate": 4.7973255712368994e-05, "loss": 0.1477, "step": 4437 }, { "epoch": 27.56697819314642, "grad_norm": 2.734375, "learning_rate": 4.797228384335379e-05, "loss": 0.2492, "step": 4438 }, { "epoch": 27.573208722741434, "grad_norm": 1.6171875, "learning_rate": 4.7971311751227174e-05, "loss": 0.1689, "step": 4439 }, { "epoch": 27.57943925233645, "grad_norm": 2.375, "learning_rate": 4.797033943599859e-05, "loss": 0.2433, "step": 4440 }, { "epoch": 27.585669781931465, "grad_norm": 3.03125, "learning_rate": 4.796936689767747e-05, "loss": 0.2188, "step": 4441 }, { "epoch": 27.59190031152648, "grad_norm": 1.875, "learning_rate": 4.796839413627327e-05, "loss": 0.1253, "step": 4442 }, { "epoch": 27.598130841121495, "grad_norm": 3.21875, "learning_rate": 4.796742115179544e-05, "loss": 0.2384, "step": 4443 }, { "epoch": 27.60436137071651, "grad_norm": 2.953125, "learning_rate": 4.796644794425342e-05, "loss": 0.1955, "step": 4444 }, { "epoch": 27.610591900311526, "grad_norm": 3.109375, "learning_rate": 4.796547451365667e-05, "loss": 0.1999, "step": 4445 }, { "epoch": 27.61682242990654, "grad_norm": 1.953125, "learning_rate": 4.796450086001465e-05, "loss": 0.1532, "step": 4446 }, { "epoch": 27.623052959501557, "grad_norm": 3.296875, "learning_rate": 4.796352698333679e-05, "loss": 0.1723, "step": 4447 }, { "epoch": 27.629283489096572, "grad_norm": 2.28125, "learning_rate": 4.796255288363257e-05, "loss": 0.2164, "step": 4448 }, { "epoch": 27.635514018691588, "grad_norm": 2.3125, "learning_rate": 4.796157856091146e-05, "loss": 0.3783, "step": 4449 }, { "epoch": 27.641744548286603, "grad_norm": 4.0, "learning_rate": 4.796060401518289e-05, "loss": 0.3741, "step": 4450 }, { "epoch": 27.64797507788162, "grad_norm": 3.515625, "learning_rate": 4.795962924645636e-05, "loss": 0.2942, "step": 4451 }, { "epoch": 27.654205607476637, "grad_norm": 3.1875, "learning_rate": 4.795865425474131e-05, "loss": 0.1581, "step": 4452 }, { "epoch": 27.660436137071652, "grad_norm": 3.359375, "learning_rate": 4.795767904004722e-05, "loss": 0.1849, "step": 4453 }, { "epoch": 27.666666666666668, "grad_norm": 3.078125, "learning_rate": 4.795670360238357e-05, "loss": 0.2463, "step": 4454 }, { "epoch": 27.672897196261683, "grad_norm": 2.9375, "learning_rate": 4.795572794175982e-05, "loss": 0.3854, "step": 4455 }, { "epoch": 27.6791277258567, "grad_norm": 2.75, "learning_rate": 4.795475205818545e-05, "loss": 0.235, "step": 4456 }, { "epoch": 27.685358255451714, "grad_norm": 1.59375, "learning_rate": 4.795377595166994e-05, "loss": 0.1498, "step": 4457 }, { "epoch": 27.69158878504673, "grad_norm": 2.859375, "learning_rate": 4.7952799622222776e-05, "loss": 0.122, "step": 4458 }, { "epoch": 27.697819314641745, "grad_norm": 2.515625, "learning_rate": 4.795182306985343e-05, "loss": 0.1721, "step": 4459 }, { "epoch": 27.70404984423676, "grad_norm": 1.703125, "learning_rate": 4.7950846294571395e-05, "loss": 0.1762, "step": 4460 }, { "epoch": 27.710280373831775, "grad_norm": 5.3125, "learning_rate": 4.794986929638614e-05, "loss": 0.2767, "step": 4461 }, { "epoch": 27.71651090342679, "grad_norm": 6.03125, "learning_rate": 4.794889207530718e-05, "loss": 0.3226, "step": 4462 }, { "epoch": 27.722741433021806, "grad_norm": 3.171875, "learning_rate": 4.794791463134399e-05, "loss": 0.1557, "step": 4463 }, { "epoch": 27.72897196261682, "grad_norm": 2.84375, "learning_rate": 4.794693696450606e-05, "loss": 0.1652, "step": 4464 }, { "epoch": 27.735202492211837, "grad_norm": 3.421875, "learning_rate": 4.7945959074802895e-05, "loss": 0.28, "step": 4465 }, { "epoch": 27.741433021806852, "grad_norm": 3.28125, "learning_rate": 4.794498096224399e-05, "loss": 0.251, "step": 4466 }, { "epoch": 27.747663551401867, "grad_norm": 2.546875, "learning_rate": 4.7944002626838844e-05, "loss": 0.1354, "step": 4467 }, { "epoch": 27.753894080996886, "grad_norm": 3.640625, "learning_rate": 4.7943024068596954e-05, "loss": 0.2715, "step": 4468 }, { "epoch": 27.7601246105919, "grad_norm": 2.703125, "learning_rate": 4.794204528752783e-05, "loss": 0.1585, "step": 4469 }, { "epoch": 27.766355140186917, "grad_norm": 3.65625, "learning_rate": 4.794106628364098e-05, "loss": 0.3257, "step": 4470 }, { "epoch": 27.772585669781932, "grad_norm": 4.25, "learning_rate": 4.7940087056945906e-05, "loss": 0.3312, "step": 4471 }, { "epoch": 27.778816199376948, "grad_norm": 5.1875, "learning_rate": 4.793910760745211e-05, "loss": 0.1514, "step": 4472 }, { "epoch": 27.785046728971963, "grad_norm": 4.5625, "learning_rate": 4.793812793516913e-05, "loss": 0.2379, "step": 4473 }, { "epoch": 27.79127725856698, "grad_norm": 2.453125, "learning_rate": 4.793714804010646e-05, "loss": 0.3346, "step": 4474 }, { "epoch": 27.797507788161994, "grad_norm": 2.71875, "learning_rate": 4.7936167922273624e-05, "loss": 0.199, "step": 4475 }, { "epoch": 27.80373831775701, "grad_norm": 2.671875, "learning_rate": 4.793518758168014e-05, "loss": 0.215, "step": 4476 }, { "epoch": 27.809968847352025, "grad_norm": 4.3125, "learning_rate": 4.793420701833553e-05, "loss": 0.1609, "step": 4477 }, { "epoch": 27.81619937694704, "grad_norm": 3.03125, "learning_rate": 4.793322623224932e-05, "loss": 0.2378, "step": 4478 }, { "epoch": 27.822429906542055, "grad_norm": 2.0625, "learning_rate": 4.7932245223431026e-05, "loss": 0.2639, "step": 4479 }, { "epoch": 27.82866043613707, "grad_norm": 3.125, "learning_rate": 4.793126399189018e-05, "loss": 0.2692, "step": 4480 }, { "epoch": 27.834890965732086, "grad_norm": 3.25, "learning_rate": 4.793028253763633e-05, "loss": 0.2362, "step": 4481 }, { "epoch": 27.8411214953271, "grad_norm": 3.515625, "learning_rate": 4.792930086067898e-05, "loss": 0.2583, "step": 4482 }, { "epoch": 27.847352024922117, "grad_norm": 2.828125, "learning_rate": 4.792831896102768e-05, "loss": 0.3094, "step": 4483 }, { "epoch": 27.853582554517136, "grad_norm": 1.46875, "learning_rate": 4.792733683869196e-05, "loss": 0.1514, "step": 4484 }, { "epoch": 27.85981308411215, "grad_norm": 2.25, "learning_rate": 4.7926354493681366e-05, "loss": 0.2028, "step": 4485 }, { "epoch": 27.866043613707166, "grad_norm": 2.859375, "learning_rate": 4.7925371926005435e-05, "loss": 0.2456, "step": 4486 }, { "epoch": 27.87227414330218, "grad_norm": 2.21875, "learning_rate": 4.79243891356737e-05, "loss": 0.1644, "step": 4487 }, { "epoch": 27.878504672897197, "grad_norm": 2.296875, "learning_rate": 4.792340612269572e-05, "loss": 0.184, "step": 4488 }, { "epoch": 27.884735202492212, "grad_norm": 1.2578125, "learning_rate": 4.7922422887081045e-05, "loss": 0.1224, "step": 4489 }, { "epoch": 27.890965732087228, "grad_norm": 4.21875, "learning_rate": 4.792143942883921e-05, "loss": 0.1601, "step": 4490 }, { "epoch": 27.897196261682243, "grad_norm": 1.8046875, "learning_rate": 4.7920455747979776e-05, "loss": 0.1496, "step": 4491 }, { "epoch": 27.90342679127726, "grad_norm": 2.71875, "learning_rate": 4.791947184451229e-05, "loss": 0.2693, "step": 4492 }, { "epoch": 27.909657320872274, "grad_norm": 3.828125, "learning_rate": 4.7918487718446315e-05, "loss": 0.1628, "step": 4493 }, { "epoch": 27.91588785046729, "grad_norm": 3.328125, "learning_rate": 4.791750336979141e-05, "loss": 0.2335, "step": 4494 }, { "epoch": 27.922118380062305, "grad_norm": 1.71875, "learning_rate": 4.791651879855712e-05, "loss": 0.1337, "step": 4495 }, { "epoch": 27.92834890965732, "grad_norm": 1.9453125, "learning_rate": 4.7915534004753026e-05, "loss": 0.1709, "step": 4496 }, { "epoch": 27.934579439252335, "grad_norm": 1.84375, "learning_rate": 4.791454898838869e-05, "loss": 0.1349, "step": 4497 }, { "epoch": 27.94080996884735, "grad_norm": 2.875, "learning_rate": 4.791356374947366e-05, "loss": 0.1383, "step": 4498 }, { "epoch": 27.947040498442366, "grad_norm": 1.9609375, "learning_rate": 4.791257828801752e-05, "loss": 0.1605, "step": 4499 }, { "epoch": 27.953271028037385, "grad_norm": 3.734375, "learning_rate": 4.7911592604029856e-05, "loss": 0.2626, "step": 4500 }, { "epoch": 27.9595015576324, "grad_norm": 3.890625, "learning_rate": 4.791060669752021e-05, "loss": 0.285, "step": 4501 }, { "epoch": 27.965732087227416, "grad_norm": 2.171875, "learning_rate": 4.790962056849818e-05, "loss": 0.1804, "step": 4502 }, { "epoch": 27.97196261682243, "grad_norm": 2.9375, "learning_rate": 4.790863421697333e-05, "loss": 0.2285, "step": 4503 }, { "epoch": 27.978193146417446, "grad_norm": 2.828125, "learning_rate": 4.790764764295524e-05, "loss": 0.3226, "step": 4504 }, { "epoch": 27.98442367601246, "grad_norm": 2.5, "learning_rate": 4.790666084645351e-05, "loss": 0.1464, "step": 4505 }, { "epoch": 27.990654205607477, "grad_norm": 3.9375, "learning_rate": 4.7905673827477705e-05, "loss": 0.2761, "step": 4506 }, { "epoch": 27.996884735202492, "grad_norm": 3.140625, "learning_rate": 4.7904686586037414e-05, "loss": 0.2682, "step": 4507 }, { "epoch": 28.0, "grad_norm": 2.109375, "learning_rate": 4.790369912214223e-05, "loss": 0.1076, "step": 4508 }, { "epoch": 28.006230529595015, "grad_norm": 4.46875, "learning_rate": 4.790271143580174e-05, "loss": 0.4618, "step": 4509 }, { "epoch": 28.01246105919003, "grad_norm": 6.6875, "learning_rate": 4.7901723527025545e-05, "loss": 0.3197, "step": 4510 }, { "epoch": 28.018691588785046, "grad_norm": 6.0625, "learning_rate": 4.790073539582323e-05, "loss": 0.3568, "step": 4511 }, { "epoch": 28.02492211838006, "grad_norm": 4.78125, "learning_rate": 4.789974704220439e-05, "loss": 0.1866, "step": 4512 }, { "epoch": 28.031152647975077, "grad_norm": 2.8125, "learning_rate": 4.789875846617864e-05, "loss": 0.1554, "step": 4513 }, { "epoch": 28.037383177570092, "grad_norm": 4.21875, "learning_rate": 4.789776966775556e-05, "loss": 0.1527, "step": 4514 }, { "epoch": 28.043613707165107, "grad_norm": 4.5, "learning_rate": 4.789678064694477e-05, "loss": 0.1236, "step": 4515 }, { "epoch": 28.049844236760123, "grad_norm": 3.734375, "learning_rate": 4.7895791403755864e-05, "loss": 0.231, "step": 4516 }, { "epoch": 28.05607476635514, "grad_norm": 4.4375, "learning_rate": 4.7894801938198457e-05, "loss": 0.2483, "step": 4517 }, { "epoch": 28.062305295950157, "grad_norm": 3.375, "learning_rate": 4.789381225028216e-05, "loss": 0.2323, "step": 4518 }, { "epoch": 28.068535825545172, "grad_norm": 3.984375, "learning_rate": 4.789282234001658e-05, "loss": 0.1276, "step": 4519 }, { "epoch": 28.074766355140188, "grad_norm": 4.15625, "learning_rate": 4.7891832207411334e-05, "loss": 0.1843, "step": 4520 }, { "epoch": 28.080996884735203, "grad_norm": 4.84375, "learning_rate": 4.789084185247604e-05, "loss": 0.2214, "step": 4521 }, { "epoch": 28.08722741433022, "grad_norm": 5.03125, "learning_rate": 4.788985127522031e-05, "loss": 0.2518, "step": 4522 }, { "epoch": 28.093457943925234, "grad_norm": 2.65625, "learning_rate": 4.788886047565377e-05, "loss": 0.1545, "step": 4523 }, { "epoch": 28.09968847352025, "grad_norm": 4.875, "learning_rate": 4.788786945378604e-05, "loss": 0.316, "step": 4524 }, { "epoch": 28.105919003115265, "grad_norm": 6.0625, "learning_rate": 4.788687820962675e-05, "loss": 0.3833, "step": 4525 }, { "epoch": 28.11214953271028, "grad_norm": 5.875, "learning_rate": 4.788588674318552e-05, "loss": 0.2915, "step": 4526 }, { "epoch": 28.118380062305295, "grad_norm": 5.28125, "learning_rate": 4.788489505447199e-05, "loss": 0.2313, "step": 4527 }, { "epoch": 28.12461059190031, "grad_norm": 3.96875, "learning_rate": 4.7883903143495776e-05, "loss": 0.3829, "step": 4528 }, { "epoch": 28.130841121495326, "grad_norm": 1.7734375, "learning_rate": 4.7882911010266526e-05, "loss": 0.1472, "step": 4529 }, { "epoch": 28.13707165109034, "grad_norm": 2.65625, "learning_rate": 4.788191865479387e-05, "loss": 0.2615, "step": 4530 }, { "epoch": 28.143302180685357, "grad_norm": 4.71875, "learning_rate": 4.788092607708744e-05, "loss": 0.2895, "step": 4531 }, { "epoch": 28.149532710280372, "grad_norm": 3.484375, "learning_rate": 4.7879933277156884e-05, "loss": 0.2525, "step": 4532 }, { "epoch": 28.15576323987539, "grad_norm": 2.765625, "learning_rate": 4.787894025501185e-05, "loss": 0.2953, "step": 4533 }, { "epoch": 28.161993769470406, "grad_norm": 3.59375, "learning_rate": 4.787794701066197e-05, "loss": 0.2536, "step": 4534 }, { "epoch": 28.16822429906542, "grad_norm": 2.609375, "learning_rate": 4.7876953544116897e-05, "loss": 0.2717, "step": 4535 }, { "epoch": 28.174454828660437, "grad_norm": 2.6875, "learning_rate": 4.787595985538628e-05, "loss": 0.1384, "step": 4536 }, { "epoch": 28.180685358255452, "grad_norm": 2.0625, "learning_rate": 4.7874965944479754e-05, "loss": 0.1556, "step": 4537 }, { "epoch": 28.186915887850468, "grad_norm": 2.953125, "learning_rate": 4.7873971811407e-05, "loss": 0.2582, "step": 4538 }, { "epoch": 28.193146417445483, "grad_norm": 3.9375, "learning_rate": 4.787297745617766e-05, "loss": 0.1837, "step": 4539 }, { "epoch": 28.1993769470405, "grad_norm": 3.53125, "learning_rate": 4.787198287880138e-05, "loss": 0.3257, "step": 4540 }, { "epoch": 28.205607476635514, "grad_norm": 2.796875, "learning_rate": 4.787098807928784e-05, "loss": 0.185, "step": 4541 }, { "epoch": 28.21183800623053, "grad_norm": 3.46875, "learning_rate": 4.786999305764669e-05, "loss": 0.223, "step": 4542 }, { "epoch": 28.218068535825545, "grad_norm": 2.671875, "learning_rate": 4.7868997813887596e-05, "loss": 0.2099, "step": 4543 }, { "epoch": 28.22429906542056, "grad_norm": 3.828125, "learning_rate": 4.786800234802022e-05, "loss": 0.2321, "step": 4544 }, { "epoch": 28.230529595015575, "grad_norm": 2.078125, "learning_rate": 4.7867006660054237e-05, "loss": 0.1586, "step": 4545 }, { "epoch": 28.23676012461059, "grad_norm": 2.953125, "learning_rate": 4.786601074999931e-05, "loss": 0.2051, "step": 4546 }, { "epoch": 28.242990654205606, "grad_norm": 1.8671875, "learning_rate": 4.786501461786512e-05, "loss": 0.1492, "step": 4547 }, { "epoch": 28.24922118380062, "grad_norm": 1.96875, "learning_rate": 4.7864018263661336e-05, "loss": 0.1636, "step": 4548 }, { "epoch": 28.25545171339564, "grad_norm": 3.0, "learning_rate": 4.786302168739764e-05, "loss": 0.1906, "step": 4549 }, { "epoch": 28.261682242990656, "grad_norm": 2.21875, "learning_rate": 4.7862024889083704e-05, "loss": 0.2065, "step": 4550 }, { "epoch": 28.26791277258567, "grad_norm": 3.40625, "learning_rate": 4.786102786872921e-05, "loss": 0.1938, "step": 4551 }, { "epoch": 28.274143302180686, "grad_norm": 2.421875, "learning_rate": 4.786003062634385e-05, "loss": 0.1599, "step": 4552 }, { "epoch": 28.2803738317757, "grad_norm": 3.15625, "learning_rate": 4.78590331619373e-05, "loss": 0.2547, "step": 4553 }, { "epoch": 28.286604361370717, "grad_norm": 3.375, "learning_rate": 4.7858035475519246e-05, "loss": 0.1783, "step": 4554 }, { "epoch": 28.292834890965732, "grad_norm": 3.09375, "learning_rate": 4.785703756709939e-05, "loss": 0.4335, "step": 4555 }, { "epoch": 28.299065420560748, "grad_norm": 1.859375, "learning_rate": 4.785603943668741e-05, "loss": 0.1282, "step": 4556 }, { "epoch": 28.305295950155763, "grad_norm": 2.390625, "learning_rate": 4.7855041084293006e-05, "loss": 0.2027, "step": 4557 }, { "epoch": 28.31152647975078, "grad_norm": 2.734375, "learning_rate": 4.785404250992588e-05, "loss": 0.1458, "step": 4558 }, { "epoch": 28.317757009345794, "grad_norm": 3.71875, "learning_rate": 4.785304371359572e-05, "loss": 0.303, "step": 4559 }, { "epoch": 28.32398753894081, "grad_norm": 3.390625, "learning_rate": 4.785204469531223e-05, "loss": 0.2297, "step": 4560 }, { "epoch": 28.330218068535824, "grad_norm": 2.734375, "learning_rate": 4.785104545508512e-05, "loss": 0.2591, "step": 4561 }, { "epoch": 28.33644859813084, "grad_norm": 4.0625, "learning_rate": 4.785004599292408e-05, "loss": 0.3797, "step": 4562 }, { "epoch": 28.342679127725855, "grad_norm": 2.671875, "learning_rate": 4.784904630883883e-05, "loss": 0.1732, "step": 4563 }, { "epoch": 28.34890965732087, "grad_norm": 2.859375, "learning_rate": 4.784804640283908e-05, "loss": 0.2462, "step": 4564 }, { "epoch": 28.35514018691589, "grad_norm": 3.515625, "learning_rate": 4.7847046274934525e-05, "loss": 0.311, "step": 4565 }, { "epoch": 28.361370716510905, "grad_norm": 3.296875, "learning_rate": 4.78460459251349e-05, "loss": 0.1899, "step": 4566 }, { "epoch": 28.36760124610592, "grad_norm": 1.9609375, "learning_rate": 4.7845045353449905e-05, "loss": 0.1703, "step": 4567 }, { "epoch": 28.373831775700936, "grad_norm": 2.375, "learning_rate": 4.7844044559889264e-05, "loss": 0.224, "step": 4568 }, { "epoch": 28.38006230529595, "grad_norm": 3.203125, "learning_rate": 4.7843043544462696e-05, "loss": 0.2757, "step": 4569 }, { "epoch": 28.386292834890966, "grad_norm": 1.46875, "learning_rate": 4.784204230717992e-05, "loss": 0.1676, "step": 4570 }, { "epoch": 28.39252336448598, "grad_norm": 2.984375, "learning_rate": 4.784104084805067e-05, "loss": 0.3189, "step": 4571 }, { "epoch": 28.398753894080997, "grad_norm": 1.578125, "learning_rate": 4.784003916708466e-05, "loss": 0.1566, "step": 4572 }, { "epoch": 28.404984423676012, "grad_norm": 2.9375, "learning_rate": 4.783903726429163e-05, "loss": 0.2754, "step": 4573 }, { "epoch": 28.411214953271028, "grad_norm": 2.875, "learning_rate": 4.78380351396813e-05, "loss": 0.2702, "step": 4574 }, { "epoch": 28.417445482866043, "grad_norm": 2.8125, "learning_rate": 4.78370327932634e-05, "loss": 0.4116, "step": 4575 }, { "epoch": 28.42367601246106, "grad_norm": 2.609375, "learning_rate": 4.7836030225047685e-05, "loss": 0.2031, "step": 4576 }, { "epoch": 28.429906542056074, "grad_norm": 2.03125, "learning_rate": 4.783502743504387e-05, "loss": 0.1902, "step": 4577 }, { "epoch": 28.43613707165109, "grad_norm": 1.9609375, "learning_rate": 4.7834024423261715e-05, "loss": 0.162, "step": 4578 }, { "epoch": 28.442367601246104, "grad_norm": 2.859375, "learning_rate": 4.7833021189710937e-05, "loss": 0.252, "step": 4579 }, { "epoch": 28.44859813084112, "grad_norm": 2.796875, "learning_rate": 4.78320177344013e-05, "loss": 0.1932, "step": 4580 }, { "epoch": 28.45482866043614, "grad_norm": 2.75, "learning_rate": 4.7831014057342546e-05, "loss": 0.1575, "step": 4581 }, { "epoch": 28.461059190031154, "grad_norm": 2.515625, "learning_rate": 4.783001015854442e-05, "loss": 0.1817, "step": 4582 }, { "epoch": 28.46728971962617, "grad_norm": 2.375, "learning_rate": 4.7829006038016664e-05, "loss": 0.1868, "step": 4583 }, { "epoch": 28.473520249221185, "grad_norm": 1.9453125, "learning_rate": 4.7828001695769044e-05, "loss": 0.1704, "step": 4584 }, { "epoch": 28.4797507788162, "grad_norm": 2.53125, "learning_rate": 4.78269971318113e-05, "loss": 0.2887, "step": 4585 }, { "epoch": 28.485981308411215, "grad_norm": 3.75, "learning_rate": 4.7825992346153205e-05, "loss": 0.3787, "step": 4586 }, { "epoch": 28.49221183800623, "grad_norm": 3.15625, "learning_rate": 4.782498733880451e-05, "loss": 0.3507, "step": 4587 }, { "epoch": 28.498442367601246, "grad_norm": 3.265625, "learning_rate": 4.782398210977497e-05, "loss": 0.2639, "step": 4588 }, { "epoch": 28.50467289719626, "grad_norm": 2.921875, "learning_rate": 4.782297665907436e-05, "loss": 0.1975, "step": 4589 }, { "epoch": 28.510903426791277, "grad_norm": 1.578125, "learning_rate": 4.782197098671243e-05, "loss": 0.1393, "step": 4590 }, { "epoch": 28.517133956386292, "grad_norm": 1.7734375, "learning_rate": 4.7820965092698967e-05, "loss": 0.1655, "step": 4591 }, { "epoch": 28.523364485981308, "grad_norm": 2.796875, "learning_rate": 4.781995897704372e-05, "loss": 0.2588, "step": 4592 }, { "epoch": 28.529595015576323, "grad_norm": 2.46875, "learning_rate": 4.781895263975647e-05, "loss": 0.1769, "step": 4593 }, { "epoch": 28.53582554517134, "grad_norm": 2.328125, "learning_rate": 4.781794608084699e-05, "loss": 0.1367, "step": 4594 }, { "epoch": 28.542056074766354, "grad_norm": 3.5625, "learning_rate": 4.781693930032506e-05, "loss": 0.1847, "step": 4595 }, { "epoch": 28.54828660436137, "grad_norm": 4.46875, "learning_rate": 4.7815932298200455e-05, "loss": 0.2222, "step": 4596 }, { "epoch": 28.554517133956388, "grad_norm": 5.15625, "learning_rate": 4.781492507448295e-05, "loss": 0.2168, "step": 4597 }, { "epoch": 28.560747663551403, "grad_norm": 1.8984375, "learning_rate": 4.7813917629182334e-05, "loss": 0.1302, "step": 4598 }, { "epoch": 28.56697819314642, "grad_norm": 2.609375, "learning_rate": 4.7812909962308394e-05, "loss": 0.226, "step": 4599 }, { "epoch": 28.573208722741434, "grad_norm": 2.953125, "learning_rate": 4.781190207387091e-05, "loss": 0.2745, "step": 4600 }, { "epoch": 28.57943925233645, "grad_norm": 3.625, "learning_rate": 4.781089396387968e-05, "loss": 0.2104, "step": 4601 }, { "epoch": 28.585669781931465, "grad_norm": 2.109375, "learning_rate": 4.7809885632344475e-05, "loss": 0.1841, "step": 4602 }, { "epoch": 28.59190031152648, "grad_norm": 2.640625, "learning_rate": 4.780887707927511e-05, "loss": 0.2442, "step": 4603 }, { "epoch": 28.598130841121495, "grad_norm": 1.4375, "learning_rate": 4.780786830468136e-05, "loss": 0.1295, "step": 4604 }, { "epoch": 28.60436137071651, "grad_norm": 2.21875, "learning_rate": 4.780685930857305e-05, "loss": 0.2124, "step": 4605 }, { "epoch": 28.610591900311526, "grad_norm": 1.7890625, "learning_rate": 4.7805850090959953e-05, "loss": 0.1489, "step": 4606 }, { "epoch": 28.61682242990654, "grad_norm": 1.71875, "learning_rate": 4.780484065185188e-05, "loss": 0.1383, "step": 4607 }, { "epoch": 28.623052959501557, "grad_norm": 2.203125, "learning_rate": 4.780383099125864e-05, "loss": 0.1457, "step": 4608 }, { "epoch": 28.629283489096572, "grad_norm": 3.453125, "learning_rate": 4.7802821109190035e-05, "loss": 0.1731, "step": 4609 }, { "epoch": 28.635514018691588, "grad_norm": 3.34375, "learning_rate": 4.780181100565587e-05, "loss": 0.2057, "step": 4610 }, { "epoch": 28.641744548286603, "grad_norm": 2.296875, "learning_rate": 4.7800800680665964e-05, "loss": 0.2414, "step": 4611 }, { "epoch": 28.64797507788162, "grad_norm": 3.28125, "learning_rate": 4.779979013423012e-05, "loss": 0.2422, "step": 4612 }, { "epoch": 28.654205607476637, "grad_norm": 3.546875, "learning_rate": 4.779877936635816e-05, "loss": 0.228, "step": 4613 }, { "epoch": 28.660436137071652, "grad_norm": 2.40625, "learning_rate": 4.779776837705989e-05, "loss": 0.1733, "step": 4614 }, { "epoch": 28.666666666666668, "grad_norm": 3.546875, "learning_rate": 4.779675716634515e-05, "loss": 0.1994, "step": 4615 }, { "epoch": 28.672897196261683, "grad_norm": 2.59375, "learning_rate": 4.7795745734223734e-05, "loss": 0.1882, "step": 4616 }, { "epoch": 28.6791277258567, "grad_norm": 3.453125, "learning_rate": 4.779473408070548e-05, "loss": 0.3435, "step": 4617 }, { "epoch": 28.685358255451714, "grad_norm": 5.09375, "learning_rate": 4.779372220580022e-05, "loss": 0.1968, "step": 4618 }, { "epoch": 28.69158878504673, "grad_norm": 2.875, "learning_rate": 4.779271010951777e-05, "loss": 0.3114, "step": 4619 }, { "epoch": 28.697819314641745, "grad_norm": 3.359375, "learning_rate": 4.7791697791867965e-05, "loss": 0.4557, "step": 4620 }, { "epoch": 28.70404984423676, "grad_norm": 2.46875, "learning_rate": 4.779068525286063e-05, "loss": 0.1684, "step": 4621 }, { "epoch": 28.710280373831775, "grad_norm": 2.59375, "learning_rate": 4.7789672492505604e-05, "loss": 0.2076, "step": 4622 }, { "epoch": 28.71651090342679, "grad_norm": 3.171875, "learning_rate": 4.7788659510812726e-05, "loss": 0.2385, "step": 4623 }, { "epoch": 28.722741433021806, "grad_norm": 2.734375, "learning_rate": 4.778764630779183e-05, "loss": 0.2286, "step": 4624 }, { "epoch": 28.72897196261682, "grad_norm": 2.734375, "learning_rate": 4.778663288345276e-05, "loss": 0.2548, "step": 4625 }, { "epoch": 28.735202492211837, "grad_norm": 2.53125, "learning_rate": 4.778561923780536e-05, "loss": 0.2128, "step": 4626 }, { "epoch": 28.741433021806852, "grad_norm": 2.5, "learning_rate": 4.778460537085946e-05, "loss": 0.2047, "step": 4627 }, { "epoch": 28.747663551401867, "grad_norm": 2.90625, "learning_rate": 4.7783591282624926e-05, "loss": 0.182, "step": 4628 }, { "epoch": 28.753894080996886, "grad_norm": 2.9375, "learning_rate": 4.7782576973111606e-05, "loss": 0.2893, "step": 4629 }, { "epoch": 28.7601246105919, "grad_norm": 4.03125, "learning_rate": 4.778156244232933e-05, "loss": 0.2325, "step": 4630 }, { "epoch": 28.766355140186917, "grad_norm": 2.09375, "learning_rate": 4.778054769028797e-05, "loss": 0.1391, "step": 4631 }, { "epoch": 28.772585669781932, "grad_norm": 2.171875, "learning_rate": 4.777953271699738e-05, "loss": 0.1853, "step": 4632 }, { "epoch": 28.778816199376948, "grad_norm": 3.25, "learning_rate": 4.7778517522467415e-05, "loss": 0.3983, "step": 4633 }, { "epoch": 28.785046728971963, "grad_norm": 2.296875, "learning_rate": 4.777750210670793e-05, "loss": 0.1476, "step": 4634 }, { "epoch": 28.79127725856698, "grad_norm": 2.609375, "learning_rate": 4.7776486469728796e-05, "loss": 0.1584, "step": 4635 }, { "epoch": 28.797507788161994, "grad_norm": 3.796875, "learning_rate": 4.777547061153987e-05, "loss": 0.1665, "step": 4636 }, { "epoch": 28.80373831775701, "grad_norm": 2.4375, "learning_rate": 4.7774454532151026e-05, "loss": 0.1522, "step": 4637 }, { "epoch": 28.809968847352025, "grad_norm": 2.140625, "learning_rate": 4.777343823157212e-05, "loss": 0.2208, "step": 4638 }, { "epoch": 28.81619937694704, "grad_norm": 2.3125, "learning_rate": 4.777242170981303e-05, "loss": 0.1762, "step": 4639 }, { "epoch": 28.822429906542055, "grad_norm": 3.875, "learning_rate": 4.777140496688363e-05, "loss": 0.2667, "step": 4640 }, { "epoch": 28.82866043613707, "grad_norm": 3.734375, "learning_rate": 4.77703880027938e-05, "loss": 0.2204, "step": 4641 }, { "epoch": 28.834890965732086, "grad_norm": 3.4375, "learning_rate": 4.77693708175534e-05, "loss": 0.1492, "step": 4642 }, { "epoch": 28.8411214953271, "grad_norm": 2.796875, "learning_rate": 4.776835341117232e-05, "loss": 0.1611, "step": 4643 }, { "epoch": 28.847352024922117, "grad_norm": 4.21875, "learning_rate": 4.7767335783660444e-05, "loss": 0.1703, "step": 4644 }, { "epoch": 28.853582554517136, "grad_norm": 3.90625, "learning_rate": 4.7766317935027646e-05, "loss": 0.1799, "step": 4645 }, { "epoch": 28.85981308411215, "grad_norm": 2.671875, "learning_rate": 4.776529986528383e-05, "loss": 0.1522, "step": 4646 }, { "epoch": 28.866043613707166, "grad_norm": 3.3125, "learning_rate": 4.776428157443886e-05, "loss": 0.2555, "step": 4647 }, { "epoch": 28.87227414330218, "grad_norm": 3.96875, "learning_rate": 4.776326306250264e-05, "loss": 0.2783, "step": 4648 }, { "epoch": 28.878504672897197, "grad_norm": 3.921875, "learning_rate": 4.7762244329485064e-05, "loss": 0.1879, "step": 4649 }, { "epoch": 28.884735202492212, "grad_norm": 3.546875, "learning_rate": 4.776122537539601e-05, "loss": 0.1866, "step": 4650 }, { "epoch": 28.890965732087228, "grad_norm": 2.53125, "learning_rate": 4.776020620024539e-05, "loss": 0.2062, "step": 4651 }, { "epoch": 28.897196261682243, "grad_norm": 2.34375, "learning_rate": 4.77591868040431e-05, "loss": 0.1855, "step": 4652 }, { "epoch": 28.90342679127726, "grad_norm": 2.53125, "learning_rate": 4.775816718679904e-05, "loss": 0.333, "step": 4653 }, { "epoch": 28.909657320872274, "grad_norm": 3.421875, "learning_rate": 4.775714734852311e-05, "loss": 0.1931, "step": 4654 }, { "epoch": 28.91588785046729, "grad_norm": 2.625, "learning_rate": 4.775612728922521e-05, "loss": 0.2631, "step": 4655 }, { "epoch": 28.922118380062305, "grad_norm": 1.4921875, "learning_rate": 4.775510700891526e-05, "loss": 0.1283, "step": 4656 }, { "epoch": 28.92834890965732, "grad_norm": 2.34375, "learning_rate": 4.7754086507603155e-05, "loss": 0.1527, "step": 4657 }, { "epoch": 28.934579439252335, "grad_norm": 4.21875, "learning_rate": 4.775306578529882e-05, "loss": 0.1659, "step": 4658 }, { "epoch": 28.94080996884735, "grad_norm": 5.625, "learning_rate": 4.775204484201216e-05, "loss": 0.4256, "step": 4659 }, { "epoch": 28.947040498442366, "grad_norm": 3.09375, "learning_rate": 4.7751023677753095e-05, "loss": 0.1928, "step": 4660 }, { "epoch": 28.953271028037385, "grad_norm": 4.3125, "learning_rate": 4.7750002292531537e-05, "loss": 0.2237, "step": 4661 }, { "epoch": 28.9595015576324, "grad_norm": 2.84375, "learning_rate": 4.77489806863574e-05, "loss": 0.22, "step": 4662 }, { "epoch": 28.965732087227416, "grad_norm": 3.140625, "learning_rate": 4.7747958859240626e-05, "loss": 0.2017, "step": 4663 }, { "epoch": 28.97196261682243, "grad_norm": 3.390625, "learning_rate": 4.7746936811191135e-05, "loss": 0.2765, "step": 4664 }, { "epoch": 28.978193146417446, "grad_norm": 3.3125, "learning_rate": 4.7745914542218835e-05, "loss": 0.2116, "step": 4665 }, { "epoch": 28.98442367601246, "grad_norm": 2.9375, "learning_rate": 4.774489205233367e-05, "loss": 0.2369, "step": 4666 }, { "epoch": 28.990654205607477, "grad_norm": 2.578125, "learning_rate": 4.774386934154557e-05, "loss": 0.1703, "step": 4667 }, { "epoch": 28.996884735202492, "grad_norm": 2.90625, "learning_rate": 4.7742846409864456e-05, "loss": 0.2233, "step": 4668 }, { "epoch": 29.0, "grad_norm": 2.28125, "learning_rate": 4.774182325730028e-05, "loss": 0.1239, "step": 4669 }, { "epoch": 29.006230529595015, "grad_norm": 2.5, "learning_rate": 4.774079988386296e-05, "loss": 0.2045, "step": 4670 }, { "epoch": 29.01246105919003, "grad_norm": 4.8125, "learning_rate": 4.7739776289562456e-05, "loss": 0.3893, "step": 4671 }, { "epoch": 29.018691588785046, "grad_norm": 2.328125, "learning_rate": 4.773875247440869e-05, "loss": 0.2503, "step": 4672 }, { "epoch": 29.02492211838006, "grad_norm": 2.421875, "learning_rate": 4.773772843841162e-05, "loss": 0.185, "step": 4673 }, { "epoch": 29.031152647975077, "grad_norm": 3.140625, "learning_rate": 4.7736704181581195e-05, "loss": 0.3087, "step": 4674 }, { "epoch": 29.037383177570092, "grad_norm": 2.296875, "learning_rate": 4.773567970392734e-05, "loss": 0.1615, "step": 4675 }, { "epoch": 29.043613707165107, "grad_norm": 3.359375, "learning_rate": 4.773465500546003e-05, "loss": 0.4085, "step": 4676 }, { "epoch": 29.049844236760123, "grad_norm": 3.265625, "learning_rate": 4.7733630086189196e-05, "loss": 0.1801, "step": 4677 }, { "epoch": 29.05607476635514, "grad_norm": 1.9765625, "learning_rate": 4.7732604946124806e-05, "loss": 0.1519, "step": 4678 }, { "epoch": 29.062305295950157, "grad_norm": 3.34375, "learning_rate": 4.773157958527681e-05, "loss": 0.2038, "step": 4679 }, { "epoch": 29.068535825545172, "grad_norm": 3.21875, "learning_rate": 4.773055400365518e-05, "loss": 0.2856, "step": 4680 }, { "epoch": 29.074766355140188, "grad_norm": 4.375, "learning_rate": 4.772952820126985e-05, "loss": 0.1864, "step": 4681 }, { "epoch": 29.080996884735203, "grad_norm": 6.875, "learning_rate": 4.7728502178130806e-05, "loss": 0.4035, "step": 4682 }, { "epoch": 29.08722741433022, "grad_norm": 4.6875, "learning_rate": 4.772747593424801e-05, "loss": 0.3028, "step": 4683 }, { "epoch": 29.093457943925234, "grad_norm": 2.46875, "learning_rate": 4.772644946963142e-05, "loss": 0.1411, "step": 4684 }, { "epoch": 29.09968847352025, "grad_norm": 1.921875, "learning_rate": 4.772542278429101e-05, "loss": 0.1252, "step": 4685 }, { "epoch": 29.105919003115265, "grad_norm": 3.796875, "learning_rate": 4.7724395878236746e-05, "loss": 0.2237, "step": 4686 }, { "epoch": 29.11214953271028, "grad_norm": 2.421875, "learning_rate": 4.7723368751478615e-05, "loss": 0.1783, "step": 4687 }, { "epoch": 29.118380062305295, "grad_norm": 3.203125, "learning_rate": 4.7722341404026585e-05, "loss": 0.2356, "step": 4688 }, { "epoch": 29.12461059190031, "grad_norm": 2.484375, "learning_rate": 4.772131383589063e-05, "loss": 0.1907, "step": 4689 }, { "epoch": 29.130841121495326, "grad_norm": 2.65625, "learning_rate": 4.7720286047080733e-05, "loss": 0.1447, "step": 4690 }, { "epoch": 29.13707165109034, "grad_norm": 3.0, "learning_rate": 4.7719258037606875e-05, "loss": 0.195, "step": 4691 }, { "epoch": 29.143302180685357, "grad_norm": 2.71875, "learning_rate": 4.771822980747904e-05, "loss": 0.154, "step": 4692 }, { "epoch": 29.149532710280372, "grad_norm": 2.8125, "learning_rate": 4.771720135670722e-05, "loss": 0.2764, "step": 4693 }, { "epoch": 29.15576323987539, "grad_norm": 1.34375, "learning_rate": 4.77161726853014e-05, "loss": 0.1326, "step": 4694 }, { "epoch": 29.161993769470406, "grad_norm": 2.4375, "learning_rate": 4.771514379327157e-05, "loss": 0.2071, "step": 4695 }, { "epoch": 29.16822429906542, "grad_norm": 2.203125, "learning_rate": 4.771411468062772e-05, "loss": 0.2019, "step": 4696 }, { "epoch": 29.174454828660437, "grad_norm": 3.75, "learning_rate": 4.771308534737985e-05, "loss": 0.3163, "step": 4697 }, { "epoch": 29.180685358255452, "grad_norm": 2.953125, "learning_rate": 4.7712055793537955e-05, "loss": 0.3427, "step": 4698 }, { "epoch": 29.186915887850468, "grad_norm": 2.203125, "learning_rate": 4.771102601911203e-05, "loss": 0.1503, "step": 4699 }, { "epoch": 29.193146417445483, "grad_norm": 2.03125, "learning_rate": 4.770999602411208e-05, "loss": 0.2545, "step": 4700 }, { "epoch": 29.1993769470405, "grad_norm": 2.609375, "learning_rate": 4.770896580854811e-05, "loss": 0.1805, "step": 4701 }, { "epoch": 29.205607476635514, "grad_norm": 2.5625, "learning_rate": 4.770793537243013e-05, "loss": 0.2794, "step": 4702 }, { "epoch": 29.21183800623053, "grad_norm": 2.390625, "learning_rate": 4.770690471576814e-05, "loss": 0.2302, "step": 4703 }, { "epoch": 29.218068535825545, "grad_norm": 3.03125, "learning_rate": 4.770587383857215e-05, "loss": 0.1411, "step": 4704 }, { "epoch": 29.22429906542056, "grad_norm": 1.25, "learning_rate": 4.7704842740852176e-05, "loss": 0.1448, "step": 4705 }, { "epoch": 29.230529595015575, "grad_norm": 2.796875, "learning_rate": 4.7703811422618225e-05, "loss": 0.2587, "step": 4706 }, { "epoch": 29.23676012461059, "grad_norm": 2.5625, "learning_rate": 4.770277988388033e-05, "loss": 0.2498, "step": 4707 }, { "epoch": 29.242990654205606, "grad_norm": 2.609375, "learning_rate": 4.770174812464849e-05, "loss": 0.2298, "step": 4708 }, { "epoch": 29.24922118380062, "grad_norm": 1.734375, "learning_rate": 4.7700716144932744e-05, "loss": 0.1452, "step": 4709 }, { "epoch": 29.25545171339564, "grad_norm": 2.96875, "learning_rate": 4.76996839447431e-05, "loss": 0.2612, "step": 4710 }, { "epoch": 29.261682242990656, "grad_norm": 3.796875, "learning_rate": 4.769865152408958e-05, "loss": 0.2177, "step": 4711 }, { "epoch": 29.26791277258567, "grad_norm": 3.609375, "learning_rate": 4.769761888298222e-05, "loss": 0.2548, "step": 4712 }, { "epoch": 29.274143302180686, "grad_norm": 2.0625, "learning_rate": 4.769658602143106e-05, "loss": 0.1322, "step": 4713 }, { "epoch": 29.2803738317757, "grad_norm": 3.0, "learning_rate": 4.769555293944611e-05, "loss": 0.3553, "step": 4714 }, { "epoch": 29.286604361370717, "grad_norm": 5.5, "learning_rate": 4.7694519637037414e-05, "loss": 0.4367, "step": 4715 }, { "epoch": 29.292834890965732, "grad_norm": 3.8125, "learning_rate": 4.7693486114215015e-05, "loss": 0.1578, "step": 4716 }, { "epoch": 29.299065420560748, "grad_norm": 4.59375, "learning_rate": 4.769245237098894e-05, "loss": 0.1901, "step": 4717 }, { "epoch": 29.305295950155763, "grad_norm": 2.859375, "learning_rate": 4.7691418407369225e-05, "loss": 0.1743, "step": 4718 }, { "epoch": 29.31152647975078, "grad_norm": 3.078125, "learning_rate": 4.7690384223365925e-05, "loss": 0.1514, "step": 4719 }, { "epoch": 29.317757009345794, "grad_norm": 3.46875, "learning_rate": 4.7689349818989074e-05, "loss": 0.1811, "step": 4720 }, { "epoch": 29.32398753894081, "grad_norm": 4.15625, "learning_rate": 4.768831519424872e-05, "loss": 0.162, "step": 4721 }, { "epoch": 29.330218068535824, "grad_norm": 2.78125, "learning_rate": 4.7687280349154914e-05, "loss": 0.1838, "step": 4722 }, { "epoch": 29.33644859813084, "grad_norm": 3.96875, "learning_rate": 4.768624528371771e-05, "loss": 0.1896, "step": 4723 }, { "epoch": 29.342679127725855, "grad_norm": 2.71875, "learning_rate": 4.7685209997947154e-05, "loss": 0.2107, "step": 4724 }, { "epoch": 29.34890965732087, "grad_norm": 1.9140625, "learning_rate": 4.7684174491853305e-05, "loss": 0.2521, "step": 4725 }, { "epoch": 29.35514018691589, "grad_norm": 2.59375, "learning_rate": 4.768313876544622e-05, "loss": 0.218, "step": 4726 }, { "epoch": 29.361370716510905, "grad_norm": 2.359375, "learning_rate": 4.768210281873595e-05, "loss": 0.1978, "step": 4727 }, { "epoch": 29.36760124610592, "grad_norm": 2.84375, "learning_rate": 4.7681066651732574e-05, "loss": 0.1953, "step": 4728 }, { "epoch": 29.373831775700936, "grad_norm": 2.5, "learning_rate": 4.768003026444614e-05, "loss": 0.2141, "step": 4729 }, { "epoch": 29.38006230529595, "grad_norm": 2.5, "learning_rate": 4.767899365688672e-05, "loss": 0.1714, "step": 4730 }, { "epoch": 29.386292834890966, "grad_norm": 2.796875, "learning_rate": 4.7677956829064376e-05, "loss": 0.2402, "step": 4731 }, { "epoch": 29.39252336448598, "grad_norm": 2.25, "learning_rate": 4.767691978098918e-05, "loss": 0.1454, "step": 4732 }, { "epoch": 29.398753894080997, "grad_norm": 2.59375, "learning_rate": 4.767588251267121e-05, "loss": 0.2068, "step": 4733 }, { "epoch": 29.404984423676012, "grad_norm": 2.40625, "learning_rate": 4.767484502412054e-05, "loss": 0.2011, "step": 4734 }, { "epoch": 29.411214953271028, "grad_norm": 3.96875, "learning_rate": 4.767380731534724e-05, "loss": 0.2593, "step": 4735 }, { "epoch": 29.417445482866043, "grad_norm": 2.78125, "learning_rate": 4.767276938636139e-05, "loss": 0.2226, "step": 4736 }, { "epoch": 29.42367601246106, "grad_norm": 2.828125, "learning_rate": 4.7671731237173065e-05, "loss": 0.2443, "step": 4737 }, { "epoch": 29.429906542056074, "grad_norm": 1.4765625, "learning_rate": 4.7670692867792365e-05, "loss": 0.1197, "step": 4738 }, { "epoch": 29.43613707165109, "grad_norm": 1.265625, "learning_rate": 4.766965427822936e-05, "loss": 0.1263, "step": 4739 }, { "epoch": 29.442367601246104, "grad_norm": 2.5, "learning_rate": 4.766861546849414e-05, "loss": 0.1709, "step": 4740 }, { "epoch": 29.44859813084112, "grad_norm": 3.375, "learning_rate": 4.7667576438596796e-05, "loss": 0.2307, "step": 4741 }, { "epoch": 29.45482866043614, "grad_norm": 1.90625, "learning_rate": 4.766653718854742e-05, "loss": 0.188, "step": 4742 }, { "epoch": 29.461059190031154, "grad_norm": 1.8828125, "learning_rate": 4.76654977183561e-05, "loss": 0.1789, "step": 4743 }, { "epoch": 29.46728971962617, "grad_norm": 2.796875, "learning_rate": 4.7664458028032944e-05, "loss": 0.1892, "step": 4744 }, { "epoch": 29.473520249221185, "grad_norm": 2.546875, "learning_rate": 4.766341811758803e-05, "loss": 0.1767, "step": 4745 }, { "epoch": 29.4797507788162, "grad_norm": 3.0, "learning_rate": 4.766237798703148e-05, "loss": 0.3716, "step": 4746 }, { "epoch": 29.485981308411215, "grad_norm": 1.8359375, "learning_rate": 4.766133763637338e-05, "loss": 0.162, "step": 4747 }, { "epoch": 29.49221183800623, "grad_norm": 4.03125, "learning_rate": 4.766029706562384e-05, "loss": 0.477, "step": 4748 }, { "epoch": 29.498442367601246, "grad_norm": 4.5625, "learning_rate": 4.765925627479296e-05, "loss": 0.1866, "step": 4749 }, { "epoch": 29.50467289719626, "grad_norm": 1.734375, "learning_rate": 4.7658215263890856e-05, "loss": 0.1407, "step": 4750 }, { "epoch": 29.510903426791277, "grad_norm": 2.21875, "learning_rate": 4.765717403292763e-05, "loss": 0.188, "step": 4751 }, { "epoch": 29.517133956386292, "grad_norm": 3.03125, "learning_rate": 4.765613258191342e-05, "loss": 0.2616, "step": 4752 }, { "epoch": 29.523364485981308, "grad_norm": 2.421875, "learning_rate": 4.76550909108583e-05, "loss": 0.1999, "step": 4753 }, { "epoch": 29.529595015576323, "grad_norm": 2.8125, "learning_rate": 4.765404901977242e-05, "loss": 0.1987, "step": 4754 }, { "epoch": 29.53582554517134, "grad_norm": 2.1875, "learning_rate": 4.765300690866588e-05, "loss": 0.161, "step": 4755 }, { "epoch": 29.542056074766354, "grad_norm": 2.59375, "learning_rate": 4.765196457754881e-05, "loss": 0.186, "step": 4756 }, { "epoch": 29.54828660436137, "grad_norm": 1.7578125, "learning_rate": 4.7650922026431346e-05, "loss": 0.118, "step": 4757 }, { "epoch": 29.554517133956388, "grad_norm": 2.53125, "learning_rate": 4.764987925532358e-05, "loss": 0.2691, "step": 4758 }, { "epoch": 29.560747663551403, "grad_norm": 3.96875, "learning_rate": 4.764883626423568e-05, "loss": 0.292, "step": 4759 }, { "epoch": 29.56697819314642, "grad_norm": 2.359375, "learning_rate": 4.7647793053177734e-05, "loss": 0.1647, "step": 4760 }, { "epoch": 29.573208722741434, "grad_norm": 2.78125, "learning_rate": 4.76467496221599e-05, "loss": 0.2434, "step": 4761 }, { "epoch": 29.57943925233645, "grad_norm": 2.140625, "learning_rate": 4.7645705971192315e-05, "loss": 0.148, "step": 4762 }, { "epoch": 29.585669781931465, "grad_norm": 4.1875, "learning_rate": 4.76446621002851e-05, "loss": 0.1341, "step": 4763 }, { "epoch": 29.59190031152648, "grad_norm": 5.6875, "learning_rate": 4.7643618009448403e-05, "loss": 0.2412, "step": 4764 }, { "epoch": 29.598130841121495, "grad_norm": 4.3125, "learning_rate": 4.764257369869236e-05, "loss": 0.1954, "step": 4765 }, { "epoch": 29.60436137071651, "grad_norm": 3.375, "learning_rate": 4.764152916802712e-05, "loss": 0.1902, "step": 4766 }, { "epoch": 29.610591900311526, "grad_norm": 1.2421875, "learning_rate": 4.764048441746282e-05, "loss": 0.1211, "step": 4767 }, { "epoch": 29.61682242990654, "grad_norm": 3.96875, "learning_rate": 4.763943944700961e-05, "loss": 0.2843, "step": 4768 }, { "epoch": 29.623052959501557, "grad_norm": 3.171875, "learning_rate": 4.763839425667763e-05, "loss": 0.1706, "step": 4769 }, { "epoch": 29.629283489096572, "grad_norm": 3.90625, "learning_rate": 4.7637348846477046e-05, "loss": 0.1197, "step": 4770 }, { "epoch": 29.635514018691588, "grad_norm": 2.5625, "learning_rate": 4.763630321641801e-05, "loss": 0.2351, "step": 4771 }, { "epoch": 29.641744548286603, "grad_norm": 2.65625, "learning_rate": 4.7635257366510665e-05, "loss": 0.1645, "step": 4772 }, { "epoch": 29.64797507788162, "grad_norm": 3.6875, "learning_rate": 4.763421129676517e-05, "loss": 0.336, "step": 4773 }, { "epoch": 29.654205607476637, "grad_norm": 2.53125, "learning_rate": 4.7633165007191704e-05, "loss": 0.1628, "step": 4774 }, { "epoch": 29.660436137071652, "grad_norm": 2.515625, "learning_rate": 4.763211849780041e-05, "loss": 0.2149, "step": 4775 }, { "epoch": 29.666666666666668, "grad_norm": 4.34375, "learning_rate": 4.7631071768601457e-05, "loss": 0.304, "step": 4776 }, { "epoch": 29.672897196261683, "grad_norm": 4.21875, "learning_rate": 4.763002481960501e-05, "loss": 0.2265, "step": 4777 }, { "epoch": 29.6791277258567, "grad_norm": 3.921875, "learning_rate": 4.762897765082124e-05, "loss": 0.2956, "step": 4778 }, { "epoch": 29.685358255451714, "grad_norm": 2.40625, "learning_rate": 4.7627930262260314e-05, "loss": 0.1655, "step": 4779 }, { "epoch": 29.69158878504673, "grad_norm": 1.8203125, "learning_rate": 4.76268826539324e-05, "loss": 0.1423, "step": 4780 }, { "epoch": 29.697819314641745, "grad_norm": 3.65625, "learning_rate": 4.762583482584769e-05, "loss": 0.1975, "step": 4781 }, { "epoch": 29.70404984423676, "grad_norm": 4.09375, "learning_rate": 4.762478677801635e-05, "loss": 0.2468, "step": 4782 }, { "epoch": 29.710280373831775, "grad_norm": 4.90625, "learning_rate": 4.762373851044855e-05, "loss": 0.2114, "step": 4783 }, { "epoch": 29.71651090342679, "grad_norm": 3.6875, "learning_rate": 4.762269002315448e-05, "loss": 0.2332, "step": 4784 }, { "epoch": 29.722741433021806, "grad_norm": 3.234375, "learning_rate": 4.7621641316144325e-05, "loss": 0.3915, "step": 4785 }, { "epoch": 29.72897196261682, "grad_norm": 3.765625, "learning_rate": 4.762059238942827e-05, "loss": 0.2228, "step": 4786 }, { "epoch": 29.735202492211837, "grad_norm": 4.1875, "learning_rate": 4.7619543243016495e-05, "loss": 0.3571, "step": 4787 }, { "epoch": 29.741433021806852, "grad_norm": 4.03125, "learning_rate": 4.76184938769192e-05, "loss": 0.1625, "step": 4788 }, { "epoch": 29.747663551401867, "grad_norm": 1.6953125, "learning_rate": 4.7617444291146565e-05, "loss": 0.1626, "step": 4789 }, { "epoch": 29.753894080996886, "grad_norm": 2.5625, "learning_rate": 4.7616394485708794e-05, "loss": 0.2225, "step": 4790 }, { "epoch": 29.7601246105919, "grad_norm": 2.90625, "learning_rate": 4.7615344460616075e-05, "loss": 0.3312, "step": 4791 }, { "epoch": 29.766355140186917, "grad_norm": 3.21875, "learning_rate": 4.761429421587861e-05, "loss": 0.2696, "step": 4792 }, { "epoch": 29.772585669781932, "grad_norm": 3.6875, "learning_rate": 4.761324375150661e-05, "loss": 0.1937, "step": 4793 }, { "epoch": 29.778816199376948, "grad_norm": 2.828125, "learning_rate": 4.761219306751025e-05, "loss": 0.194, "step": 4794 }, { "epoch": 29.785046728971963, "grad_norm": 2.6875, "learning_rate": 4.761114216389976e-05, "loss": 0.2416, "step": 4795 }, { "epoch": 29.79127725856698, "grad_norm": 2.984375, "learning_rate": 4.761009104068533e-05, "loss": 0.1874, "step": 4796 }, { "epoch": 29.797507788161994, "grad_norm": 3.75, "learning_rate": 4.760903969787719e-05, "loss": 0.3368, "step": 4797 }, { "epoch": 29.80373831775701, "grad_norm": 3.390625, "learning_rate": 4.7607988135485525e-05, "loss": 0.1464, "step": 4798 }, { "epoch": 29.809968847352025, "grad_norm": 3.4375, "learning_rate": 4.760693635352057e-05, "loss": 0.4013, "step": 4799 }, { "epoch": 29.81619937694704, "grad_norm": 2.765625, "learning_rate": 4.7605884351992524e-05, "loss": 0.1837, "step": 4800 }, { "epoch": 29.822429906542055, "grad_norm": 4.34375, "learning_rate": 4.760483213091161e-05, "loss": 0.3221, "step": 4801 }, { "epoch": 29.82866043613707, "grad_norm": 2.125, "learning_rate": 4.7603779690288046e-05, "loss": 0.1703, "step": 4802 }, { "epoch": 29.834890965732086, "grad_norm": 3.140625, "learning_rate": 4.760272703013206e-05, "loss": 0.3053, "step": 4803 }, { "epoch": 29.8411214953271, "grad_norm": 2.3125, "learning_rate": 4.7601674150453866e-05, "loss": 0.2034, "step": 4804 }, { "epoch": 29.847352024922117, "grad_norm": 3.5625, "learning_rate": 4.760062105126369e-05, "loss": 0.2462, "step": 4805 }, { "epoch": 29.853582554517136, "grad_norm": 2.640625, "learning_rate": 4.759956773257178e-05, "loss": 0.1527, "step": 4806 }, { "epoch": 29.85981308411215, "grad_norm": 1.6640625, "learning_rate": 4.759851419438834e-05, "loss": 0.1615, "step": 4807 }, { "epoch": 29.866043613707166, "grad_norm": 2.25, "learning_rate": 4.759746043672362e-05, "loss": 0.1491, "step": 4808 }, { "epoch": 29.87227414330218, "grad_norm": 4.53125, "learning_rate": 4.759640645958784e-05, "loss": 0.3829, "step": 4809 }, { "epoch": 29.878504672897197, "grad_norm": 5.40625, "learning_rate": 4.7595352262991245e-05, "loss": 0.3291, "step": 4810 }, { "epoch": 29.884735202492212, "grad_norm": 3.40625, "learning_rate": 4.759429784694407e-05, "loss": 0.1441, "step": 4811 }, { "epoch": 29.890965732087228, "grad_norm": 4.84375, "learning_rate": 4.759324321145656e-05, "loss": 0.1502, "step": 4812 }, { "epoch": 29.897196261682243, "grad_norm": 2.484375, "learning_rate": 4.759218835653896e-05, "loss": 0.1831, "step": 4813 }, { "epoch": 29.90342679127726, "grad_norm": 3.9375, "learning_rate": 4.7591133282201504e-05, "loss": 0.2373, "step": 4814 }, { "epoch": 29.909657320872274, "grad_norm": 4.40625, "learning_rate": 4.759007798845445e-05, "loss": 0.1875, "step": 4815 }, { "epoch": 29.91588785046729, "grad_norm": 3.328125, "learning_rate": 4.758902247530804e-05, "loss": 0.1507, "step": 4816 }, { "epoch": 29.922118380062305, "grad_norm": 3.71875, "learning_rate": 4.758796674277253e-05, "loss": 0.2852, "step": 4817 }, { "epoch": 29.92834890965732, "grad_norm": 3.0, "learning_rate": 4.7586910790858174e-05, "loss": 0.1631, "step": 4818 }, { "epoch": 29.934579439252335, "grad_norm": 2.9375, "learning_rate": 4.758585461957522e-05, "loss": 0.2059, "step": 4819 }, { "epoch": 29.94080996884735, "grad_norm": 2.5, "learning_rate": 4.758479822893394e-05, "loss": 0.2411, "step": 4820 }, { "epoch": 29.947040498442366, "grad_norm": 2.875, "learning_rate": 4.7583741618944575e-05, "loss": 0.2674, "step": 4821 }, { "epoch": 29.953271028037385, "grad_norm": 2.484375, "learning_rate": 4.75826847896174e-05, "loss": 0.2078, "step": 4822 }, { "epoch": 29.9595015576324, "grad_norm": 2.4375, "learning_rate": 4.7581627740962676e-05, "loss": 0.3135, "step": 4823 }, { "epoch": 29.965732087227416, "grad_norm": 1.78125, "learning_rate": 4.758057047299067e-05, "loss": 0.1679, "step": 4824 }, { "epoch": 29.97196261682243, "grad_norm": 2.21875, "learning_rate": 4.757951298571165e-05, "loss": 0.2014, "step": 4825 }, { "epoch": 29.978193146417446, "grad_norm": 2.75, "learning_rate": 4.7578455279135895e-05, "loss": 0.2344, "step": 4826 }, { "epoch": 29.98442367601246, "grad_norm": 2.984375, "learning_rate": 4.757739735327366e-05, "loss": 0.2921, "step": 4827 }, { "epoch": 29.990654205607477, "grad_norm": 3.046875, "learning_rate": 4.757633920813522e-05, "loss": 0.1886, "step": 4828 }, { "epoch": 29.996884735202492, "grad_norm": 1.9921875, "learning_rate": 4.757528084373087e-05, "loss": 0.1343, "step": 4829 }, { "epoch": 30.0, "grad_norm": 1.1796875, "learning_rate": 4.7574222260070885e-05, "loss": 0.0894, "step": 4830 }, { "epoch": 30.006230529595015, "grad_norm": 4.4375, "learning_rate": 4.7573163457165534e-05, "loss": 0.1246, "step": 4831 }, { "epoch": 30.01246105919003, "grad_norm": 1.59375, "learning_rate": 4.757210443502511e-05, "loss": 0.1262, "step": 4832 }, { "epoch": 30.018691588785046, "grad_norm": 3.71875, "learning_rate": 4.757104519365989e-05, "loss": 0.1891, "step": 4833 }, { "epoch": 30.02492211838006, "grad_norm": 3.015625, "learning_rate": 4.7569985733080174e-05, "loss": 0.2726, "step": 4834 }, { "epoch": 30.031152647975077, "grad_norm": 2.65625, "learning_rate": 4.7568926053296244e-05, "loss": 0.2172, "step": 4835 }, { "epoch": 30.037383177570092, "grad_norm": 2.796875, "learning_rate": 4.7567866154318394e-05, "loss": 0.237, "step": 4836 }, { "epoch": 30.043613707165107, "grad_norm": 2.828125, "learning_rate": 4.7566806036156906e-05, "loss": 0.3544, "step": 4837 }, { "epoch": 30.049844236760123, "grad_norm": 2.34375, "learning_rate": 4.75657456988221e-05, "loss": 0.1328, "step": 4838 }, { "epoch": 30.05607476635514, "grad_norm": 3.0, "learning_rate": 4.756468514232426e-05, "loss": 0.1707, "step": 4839 }, { "epoch": 30.062305295950157, "grad_norm": 3.109375, "learning_rate": 4.756362436667368e-05, "loss": 0.1936, "step": 4840 }, { "epoch": 30.068535825545172, "grad_norm": 4.0, "learning_rate": 4.756256337188067e-05, "loss": 0.1661, "step": 4841 }, { "epoch": 30.074766355140188, "grad_norm": 4.65625, "learning_rate": 4.756150215795554e-05, "loss": 0.3407, "step": 4842 }, { "epoch": 30.080996884735203, "grad_norm": 1.5390625, "learning_rate": 4.7560440724908586e-05, "loss": 0.1299, "step": 4843 }, { "epoch": 30.08722741433022, "grad_norm": 2.484375, "learning_rate": 4.755937907275012e-05, "loss": 0.1944, "step": 4844 }, { "epoch": 30.093457943925234, "grad_norm": 5.5, "learning_rate": 4.7558317201490464e-05, "loss": 0.2543, "step": 4845 }, { "epoch": 30.09968847352025, "grad_norm": 3.890625, "learning_rate": 4.755725511113992e-05, "loss": 0.2107, "step": 4846 }, { "epoch": 30.105919003115265, "grad_norm": 3.171875, "learning_rate": 4.75561928017088e-05, "loss": 0.2664, "step": 4847 }, { "epoch": 30.11214953271028, "grad_norm": 3.0, "learning_rate": 4.7555130273207425e-05, "loss": 0.1841, "step": 4848 }, { "epoch": 30.118380062305295, "grad_norm": 3.3125, "learning_rate": 4.7554067525646125e-05, "loss": 0.377, "step": 4849 }, { "epoch": 30.12461059190031, "grad_norm": 3.59375, "learning_rate": 4.755300455903521e-05, "loss": 0.2064, "step": 4850 }, { "epoch": 30.130841121495326, "grad_norm": 3.859375, "learning_rate": 4.7551941373385e-05, "loss": 0.2071, "step": 4851 }, { "epoch": 30.13707165109034, "grad_norm": 4.96875, "learning_rate": 4.755087796870583e-05, "loss": 0.2355, "step": 4852 }, { "epoch": 30.143302180685357, "grad_norm": 1.8203125, "learning_rate": 4.754981434500803e-05, "loss": 0.157, "step": 4853 }, { "epoch": 30.149532710280372, "grad_norm": 3.40625, "learning_rate": 4.754875050230192e-05, "loss": 0.1673, "step": 4854 }, { "epoch": 30.15576323987539, "grad_norm": 3.6875, "learning_rate": 4.754768644059784e-05, "loss": 0.1336, "step": 4855 }, { "epoch": 30.161993769470406, "grad_norm": 6.625, "learning_rate": 4.754662215990613e-05, "loss": 0.232, "step": 4856 }, { "epoch": 30.16822429906542, "grad_norm": 4.4375, "learning_rate": 4.754555766023711e-05, "loss": 0.1954, "step": 4857 }, { "epoch": 30.174454828660437, "grad_norm": 3.40625, "learning_rate": 4.754449294160112e-05, "loss": 0.2747, "step": 4858 }, { "epoch": 30.180685358255452, "grad_norm": 3.875, "learning_rate": 4.754342800400852e-05, "loss": 0.2532, "step": 4859 }, { "epoch": 30.186915887850468, "grad_norm": 5.375, "learning_rate": 4.7542362847469633e-05, "loss": 0.171, "step": 4860 }, { "epoch": 30.193146417445483, "grad_norm": 3.34375, "learning_rate": 4.7541297471994815e-05, "loss": 0.1441, "step": 4861 }, { "epoch": 30.1993769470405, "grad_norm": 2.78125, "learning_rate": 4.754023187759442e-05, "loss": 0.2079, "step": 4862 }, { "epoch": 30.205607476635514, "grad_norm": 1.828125, "learning_rate": 4.753916606427877e-05, "loss": 0.1245, "step": 4863 }, { "epoch": 30.21183800623053, "grad_norm": 2.203125, "learning_rate": 4.753810003205825e-05, "loss": 0.2758, "step": 4864 }, { "epoch": 30.218068535825545, "grad_norm": 3.703125, "learning_rate": 4.753703378094319e-05, "loss": 0.1941, "step": 4865 }, { "epoch": 30.22429906542056, "grad_norm": 3.9375, "learning_rate": 4.7535967310943955e-05, "loss": 0.2627, "step": 4866 }, { "epoch": 30.230529595015575, "grad_norm": 2.03125, "learning_rate": 4.753490062207089e-05, "loss": 0.203, "step": 4867 }, { "epoch": 30.23676012461059, "grad_norm": 1.8203125, "learning_rate": 4.753383371433437e-05, "loss": 0.146, "step": 4868 }, { "epoch": 30.242990654205606, "grad_norm": 3.0625, "learning_rate": 4.753276658774476e-05, "loss": 0.1478, "step": 4869 }, { "epoch": 30.24922118380062, "grad_norm": 3.296875, "learning_rate": 4.753169924231241e-05, "loss": 0.199, "step": 4870 }, { "epoch": 30.25545171339564, "grad_norm": 3.359375, "learning_rate": 4.75306316780477e-05, "loss": 0.2828, "step": 4871 }, { "epoch": 30.261682242990656, "grad_norm": 2.59375, "learning_rate": 4.7529563894960984e-05, "loss": 0.1788, "step": 4872 }, { "epoch": 30.26791277258567, "grad_norm": 2.765625, "learning_rate": 4.752849589306264e-05, "loss": 0.2368, "step": 4873 }, { "epoch": 30.274143302180686, "grad_norm": 2.640625, "learning_rate": 4.7527427672363046e-05, "loss": 0.2248, "step": 4874 }, { "epoch": 30.2803738317757, "grad_norm": 3.28125, "learning_rate": 4.752635923287257e-05, "loss": 0.3025, "step": 4875 }, { "epoch": 30.286604361370717, "grad_norm": 3.171875, "learning_rate": 4.752529057460159e-05, "loss": 0.4119, "step": 4876 }, { "epoch": 30.292834890965732, "grad_norm": 4.28125, "learning_rate": 4.752422169756048e-05, "loss": 0.2873, "step": 4877 }, { "epoch": 30.299065420560748, "grad_norm": 2.875, "learning_rate": 4.752315260175963e-05, "loss": 0.1381, "step": 4878 }, { "epoch": 30.305295950155763, "grad_norm": 3.46875, "learning_rate": 4.7522083287209415e-05, "loss": 0.2219, "step": 4879 }, { "epoch": 30.31152647975078, "grad_norm": 3.546875, "learning_rate": 4.752101375392023e-05, "loss": 0.3652, "step": 4880 }, { "epoch": 30.317757009345794, "grad_norm": 1.8671875, "learning_rate": 4.751994400190246e-05, "loss": 0.131, "step": 4881 }, { "epoch": 30.32398753894081, "grad_norm": 2.71875, "learning_rate": 4.7518874031166495e-05, "loss": 0.2555, "step": 4882 }, { "epoch": 30.330218068535824, "grad_norm": 1.6171875, "learning_rate": 4.751780384172272e-05, "loss": 0.1491, "step": 4883 }, { "epoch": 30.33644859813084, "grad_norm": 3.6875, "learning_rate": 4.751673343358153e-05, "loss": 0.2859, "step": 4884 }, { "epoch": 30.342679127725855, "grad_norm": 3.03125, "learning_rate": 4.7515662806753325e-05, "loss": 0.284, "step": 4885 }, { "epoch": 30.34890965732087, "grad_norm": 1.5, "learning_rate": 4.75145919612485e-05, "loss": 0.133, "step": 4886 }, { "epoch": 30.35514018691589, "grad_norm": 2.671875, "learning_rate": 4.7513520897077456e-05, "loss": 0.3111, "step": 4887 }, { "epoch": 30.361370716510905, "grad_norm": 2.5, "learning_rate": 4.751244961425061e-05, "loss": 0.2623, "step": 4888 }, { "epoch": 30.36760124610592, "grad_norm": 1.6875, "learning_rate": 4.751137811277834e-05, "loss": 0.1295, "step": 4889 }, { "epoch": 30.373831775700936, "grad_norm": 2.265625, "learning_rate": 4.7510306392671065e-05, "loss": 0.1837, "step": 4890 }, { "epoch": 30.38006230529595, "grad_norm": 3.703125, "learning_rate": 4.75092344539392e-05, "loss": 0.316, "step": 4891 }, { "epoch": 30.386292834890966, "grad_norm": 2.671875, "learning_rate": 4.750816229659315e-05, "loss": 0.2097, "step": 4892 }, { "epoch": 30.39252336448598, "grad_norm": 2.953125, "learning_rate": 4.750708992064333e-05, "loss": 0.2377, "step": 4893 }, { "epoch": 30.398753894080997, "grad_norm": 3.640625, "learning_rate": 4.750601732610015e-05, "loss": 0.2942, "step": 4894 }, { "epoch": 30.404984423676012, "grad_norm": 1.5078125, "learning_rate": 4.7504944512974036e-05, "loss": 0.1403, "step": 4895 }, { "epoch": 30.411214953271028, "grad_norm": 2.375, "learning_rate": 4.75038714812754e-05, "loss": 0.2265, "step": 4896 }, { "epoch": 30.417445482866043, "grad_norm": 2.21875, "learning_rate": 4.750279823101466e-05, "loss": 0.2137, "step": 4897 }, { "epoch": 30.42367601246106, "grad_norm": 3.4375, "learning_rate": 4.750172476220225e-05, "loss": 0.2216, "step": 4898 }, { "epoch": 30.429906542056074, "grad_norm": 2.6875, "learning_rate": 4.750065107484859e-05, "loss": 0.1698, "step": 4899 }, { "epoch": 30.43613707165109, "grad_norm": 3.265625, "learning_rate": 4.749957716896412e-05, "loss": 0.2778, "step": 4900 }, { "epoch": 30.442367601246104, "grad_norm": 3.0625, "learning_rate": 4.749850304455924e-05, "loss": 0.3, "step": 4901 }, { "epoch": 30.44859813084112, "grad_norm": 2.921875, "learning_rate": 4.749742870164441e-05, "loss": 0.2611, "step": 4902 }, { "epoch": 30.45482866043614, "grad_norm": 2.90625, "learning_rate": 4.749635414023006e-05, "loss": 0.2203, "step": 4903 }, { "epoch": 30.461059190031154, "grad_norm": 2.703125, "learning_rate": 4.749527936032662e-05, "loss": 0.2189, "step": 4904 }, { "epoch": 30.46728971962617, "grad_norm": 3.484375, "learning_rate": 4.749420436194452e-05, "loss": 0.2121, "step": 4905 }, { "epoch": 30.473520249221185, "grad_norm": 1.578125, "learning_rate": 4.749312914509422e-05, "loss": 0.1275, "step": 4906 }, { "epoch": 30.4797507788162, "grad_norm": 3.671875, "learning_rate": 4.749205370978615e-05, "loss": 0.2369, "step": 4907 }, { "epoch": 30.485981308411215, "grad_norm": 5.09375, "learning_rate": 4.749097805603076e-05, "loss": 0.2562, "step": 4908 }, { "epoch": 30.49221183800623, "grad_norm": 2.5625, "learning_rate": 4.74899021838385e-05, "loss": 0.2104, "step": 4909 }, { "epoch": 30.498442367601246, "grad_norm": 3.5, "learning_rate": 4.74888260932198e-05, "loss": 0.2571, "step": 4910 }, { "epoch": 30.50467289719626, "grad_norm": 2.328125, "learning_rate": 4.748774978418514e-05, "loss": 0.1844, "step": 4911 }, { "epoch": 30.510903426791277, "grad_norm": 3.71875, "learning_rate": 4.748667325674495e-05, "loss": 0.2354, "step": 4912 }, { "epoch": 30.517133956386292, "grad_norm": 3.34375, "learning_rate": 4.74855965109097e-05, "loss": 0.2363, "step": 4913 }, { "epoch": 30.523364485981308, "grad_norm": 1.90625, "learning_rate": 4.7484519546689844e-05, "loss": 0.1349, "step": 4914 }, { "epoch": 30.529595015576323, "grad_norm": 2.625, "learning_rate": 4.748344236409583e-05, "loss": 0.1988, "step": 4915 }, { "epoch": 30.53582554517134, "grad_norm": 2.90625, "learning_rate": 4.748236496313814e-05, "loss": 0.2534, "step": 4916 }, { "epoch": 30.542056074766354, "grad_norm": 2.828125, "learning_rate": 4.748128734382722e-05, "loss": 0.2134, "step": 4917 }, { "epoch": 30.54828660436137, "grad_norm": 2.421875, "learning_rate": 4.7480209506173545e-05, "loss": 0.1618, "step": 4918 }, { "epoch": 30.554517133956388, "grad_norm": 2.0625, "learning_rate": 4.747913145018758e-05, "loss": 0.1862, "step": 4919 }, { "epoch": 30.560747663551403, "grad_norm": 3.75, "learning_rate": 4.74780531758798e-05, "loss": 0.3579, "step": 4920 }, { "epoch": 30.56697819314642, "grad_norm": 2.203125, "learning_rate": 4.7476974683260674e-05, "loss": 0.3206, "step": 4921 }, { "epoch": 30.573208722741434, "grad_norm": 3.3125, "learning_rate": 4.747589597234068e-05, "loss": 0.2952, "step": 4922 }, { "epoch": 30.57943925233645, "grad_norm": 3.171875, "learning_rate": 4.747481704313028e-05, "loss": 0.1732, "step": 4923 }, { "epoch": 30.585669781931465, "grad_norm": 1.9921875, "learning_rate": 4.7473737895639974e-05, "loss": 0.1327, "step": 4924 }, { "epoch": 30.59190031152648, "grad_norm": 3.671875, "learning_rate": 4.747265852988023e-05, "loss": 0.2391, "step": 4925 }, { "epoch": 30.598130841121495, "grad_norm": 2.984375, "learning_rate": 4.747157894586154e-05, "loss": 0.3133, "step": 4926 }, { "epoch": 30.60436137071651, "grad_norm": 2.453125, "learning_rate": 4.7470499143594375e-05, "loss": 0.2323, "step": 4927 }, { "epoch": 30.610591900311526, "grad_norm": 2.5625, "learning_rate": 4.746941912308923e-05, "loss": 0.1848, "step": 4928 }, { "epoch": 30.61682242990654, "grad_norm": 2.46875, "learning_rate": 4.74683388843566e-05, "loss": 0.2471, "step": 4929 }, { "epoch": 30.623052959501557, "grad_norm": 1.796875, "learning_rate": 4.7467258427406974e-05, "loss": 0.1558, "step": 4930 }, { "epoch": 30.629283489096572, "grad_norm": 2.296875, "learning_rate": 4.746617775225084e-05, "loss": 0.2183, "step": 4931 }, { "epoch": 30.635514018691588, "grad_norm": 2.453125, "learning_rate": 4.7465096858898696e-05, "loss": 0.2084, "step": 4932 }, { "epoch": 30.641744548286603, "grad_norm": 2.953125, "learning_rate": 4.746401574736104e-05, "loss": 0.3596, "step": 4933 }, { "epoch": 30.64797507788162, "grad_norm": 3.0625, "learning_rate": 4.7462934417648374e-05, "loss": 0.3247, "step": 4934 }, { "epoch": 30.654205607476637, "grad_norm": 3.109375, "learning_rate": 4.74618528697712e-05, "loss": 0.1772, "step": 4935 }, { "epoch": 30.660436137071652, "grad_norm": 3.40625, "learning_rate": 4.746077110374002e-05, "loss": 0.1919, "step": 4936 }, { "epoch": 30.666666666666668, "grad_norm": 1.4609375, "learning_rate": 4.745968911956534e-05, "loss": 0.1607, "step": 4937 }, { "epoch": 30.672897196261683, "grad_norm": 3.359375, "learning_rate": 4.745860691725767e-05, "loss": 0.3382, "step": 4938 }, { "epoch": 30.6791277258567, "grad_norm": 3.046875, "learning_rate": 4.7457524496827524e-05, "loss": 0.2494, "step": 4939 }, { "epoch": 30.685358255451714, "grad_norm": 2.109375, "learning_rate": 4.74564418582854e-05, "loss": 0.164, "step": 4940 }, { "epoch": 30.69158878504673, "grad_norm": 3.109375, "learning_rate": 4.745535900164183e-05, "loss": 0.1517, "step": 4941 }, { "epoch": 30.697819314641745, "grad_norm": 1.9765625, "learning_rate": 4.745427592690733e-05, "loss": 0.1431, "step": 4942 }, { "epoch": 30.70404984423676, "grad_norm": 5.125, "learning_rate": 4.74531926340924e-05, "loss": 0.2312, "step": 4943 }, { "epoch": 30.710280373831775, "grad_norm": 5.09375, "learning_rate": 4.745210912320759e-05, "loss": 0.1508, "step": 4944 }, { "epoch": 30.71651090342679, "grad_norm": 7.0, "learning_rate": 4.74510253942634e-05, "loss": 0.2457, "step": 4945 }, { "epoch": 30.722741433021806, "grad_norm": 4.625, "learning_rate": 4.744994144727036e-05, "loss": 0.2437, "step": 4946 }, { "epoch": 30.72897196261682, "grad_norm": 2.328125, "learning_rate": 4.744885728223901e-05, "loss": 0.2186, "step": 4947 }, { "epoch": 30.735202492211837, "grad_norm": 2.734375, "learning_rate": 4.7447772899179873e-05, "loss": 0.1384, "step": 4948 }, { "epoch": 30.741433021806852, "grad_norm": 3.53125, "learning_rate": 4.744668829810347e-05, "loss": 0.1418, "step": 4949 }, { "epoch": 30.747663551401867, "grad_norm": 2.234375, "learning_rate": 4.744560347902035e-05, "loss": 0.2169, "step": 4950 }, { "epoch": 30.753894080996886, "grad_norm": 3.296875, "learning_rate": 4.744451844194103e-05, "loss": 0.3255, "step": 4951 }, { "epoch": 30.7601246105919, "grad_norm": 2.40625, "learning_rate": 4.744343318687607e-05, "loss": 0.2089, "step": 4952 }, { "epoch": 30.766355140186917, "grad_norm": 2.96875, "learning_rate": 4.7442347713836e-05, "loss": 0.2914, "step": 4953 }, { "epoch": 30.772585669781932, "grad_norm": 3.640625, "learning_rate": 4.7441262022831365e-05, "loss": 0.2626, "step": 4954 }, { "epoch": 30.778816199376948, "grad_norm": 3.078125, "learning_rate": 4.7440176113872705e-05, "loss": 0.2402, "step": 4955 }, { "epoch": 30.785046728971963, "grad_norm": 1.9609375, "learning_rate": 4.7439089986970564e-05, "loss": 0.1509, "step": 4956 }, { "epoch": 30.79127725856698, "grad_norm": 1.359375, "learning_rate": 4.74380036421355e-05, "loss": 0.1514, "step": 4957 }, { "epoch": 30.797507788161994, "grad_norm": 2.53125, "learning_rate": 4.7436917079378066e-05, "loss": 0.1683, "step": 4958 }, { "epoch": 30.80373831775701, "grad_norm": 2.765625, "learning_rate": 4.74358302987088e-05, "loss": 0.2637, "step": 4959 }, { "epoch": 30.809968847352025, "grad_norm": 2.796875, "learning_rate": 4.7434743300138264e-05, "loss": 0.1312, "step": 4960 }, { "epoch": 30.81619937694704, "grad_norm": 3.0625, "learning_rate": 4.7433656083677015e-05, "loss": 0.2528, "step": 4961 }, { "epoch": 30.822429906542055, "grad_norm": 3.78125, "learning_rate": 4.7432568649335616e-05, "loss": 0.3139, "step": 4962 }, { "epoch": 30.82866043613707, "grad_norm": 2.28125, "learning_rate": 4.743148099712463e-05, "loss": 0.1867, "step": 4963 }, { "epoch": 30.834890965732086, "grad_norm": 3.15625, "learning_rate": 4.743039312705462e-05, "loss": 0.2503, "step": 4964 }, { "epoch": 30.8411214953271, "grad_norm": 3.9375, "learning_rate": 4.742930503913614e-05, "loss": 0.1935, "step": 4965 }, { "epoch": 30.847352024922117, "grad_norm": 3.53125, "learning_rate": 4.7428216733379765e-05, "loss": 0.3226, "step": 4966 }, { "epoch": 30.853582554517136, "grad_norm": 2.53125, "learning_rate": 4.742712820979607e-05, "loss": 0.1617, "step": 4967 }, { "epoch": 30.85981308411215, "grad_norm": 2.703125, "learning_rate": 4.742603946839562e-05, "loss": 0.2398, "step": 4968 }, { "epoch": 30.866043613707166, "grad_norm": 2.265625, "learning_rate": 4.7424950509188995e-05, "loss": 0.1259, "step": 4969 }, { "epoch": 30.87227414330218, "grad_norm": 3.828125, "learning_rate": 4.742386133218677e-05, "loss": 0.295, "step": 4970 }, { "epoch": 30.878504672897197, "grad_norm": 3.765625, "learning_rate": 4.7422771937399514e-05, "loss": 0.2414, "step": 4971 }, { "epoch": 30.884735202492212, "grad_norm": 3.484375, "learning_rate": 4.742168232483782e-05, "loss": 0.253, "step": 4972 }, { "epoch": 30.890965732087228, "grad_norm": 2.859375, "learning_rate": 4.7420592494512264e-05, "loss": 0.216, "step": 4973 }, { "epoch": 30.897196261682243, "grad_norm": 2.609375, "learning_rate": 4.7419502446433425e-05, "loss": 0.2366, "step": 4974 }, { "epoch": 30.90342679127726, "grad_norm": 3.828125, "learning_rate": 4.7418412180611906e-05, "loss": 0.3219, "step": 4975 }, { "epoch": 30.909657320872274, "grad_norm": 4.28125, "learning_rate": 4.741732169705829e-05, "loss": 0.3032, "step": 4976 }, { "epoch": 30.91588785046729, "grad_norm": 2.90625, "learning_rate": 4.7416230995783165e-05, "loss": 0.1501, "step": 4977 }, { "epoch": 30.922118380062305, "grad_norm": 1.03125, "learning_rate": 4.741514007679711e-05, "loss": 0.1163, "step": 4978 }, { "epoch": 30.92834890965732, "grad_norm": 3.390625, "learning_rate": 4.741404894011074e-05, "loss": 0.1679, "step": 4979 }, { "epoch": 30.934579439252335, "grad_norm": 3.03125, "learning_rate": 4.741295758573465e-05, "loss": 0.1737, "step": 4980 }, { "epoch": 30.94080996884735, "grad_norm": 3.703125, "learning_rate": 4.7411866013679426e-05, "loss": 0.231, "step": 4981 }, { "epoch": 30.947040498442366, "grad_norm": 2.203125, "learning_rate": 4.7410774223955686e-05, "loss": 0.2196, "step": 4982 }, { "epoch": 30.953271028037385, "grad_norm": 3.1875, "learning_rate": 4.740968221657404e-05, "loss": 0.1881, "step": 4983 }, { "epoch": 30.9595015576324, "grad_norm": 3.8125, "learning_rate": 4.740858999154506e-05, "loss": 0.2797, "step": 4984 }, { "epoch": 30.965732087227416, "grad_norm": 2.359375, "learning_rate": 4.7407497548879384e-05, "loss": 0.1508, "step": 4985 }, { "epoch": 30.97196261682243, "grad_norm": 1.9140625, "learning_rate": 4.740640488858761e-05, "loss": 0.1309, "step": 4986 }, { "epoch": 30.978193146417446, "grad_norm": 1.5390625, "learning_rate": 4.740531201068036e-05, "loss": 0.1362, "step": 4987 }, { "epoch": 30.98442367601246, "grad_norm": 4.15625, "learning_rate": 4.740421891516823e-05, "loss": 0.191, "step": 4988 }, { "epoch": 30.990654205607477, "grad_norm": 4.09375, "learning_rate": 4.7403125602061846e-05, "loss": 0.1348, "step": 4989 }, { "epoch": 30.996884735202492, "grad_norm": 3.25, "learning_rate": 4.7402032071371835e-05, "loss": 0.2483, "step": 4990 }, { "epoch": 31.0, "grad_norm": 1.28125, "learning_rate": 4.7400938323108815e-05, "loss": 0.0883, "step": 4991 }, { "epoch": 31.006230529595015, "grad_norm": 2.078125, "learning_rate": 4.7399844357283397e-05, "loss": 0.2162, "step": 4992 }, { "epoch": 31.01246105919003, "grad_norm": 1.59375, "learning_rate": 4.739875017390621e-05, "loss": 0.1498, "step": 4993 }, { "epoch": 31.018691588785046, "grad_norm": 3.671875, "learning_rate": 4.7397655772987894e-05, "loss": 0.1467, "step": 4994 }, { "epoch": 31.02492211838006, "grad_norm": 3.890625, "learning_rate": 4.739656115453907e-05, "loss": 0.1924, "step": 4995 }, { "epoch": 31.031152647975077, "grad_norm": 4.375, "learning_rate": 4.7395466318570356e-05, "loss": 0.1495, "step": 4996 }, { "epoch": 31.037383177570092, "grad_norm": 2.375, "learning_rate": 4.7394371265092394e-05, "loss": 0.1703, "step": 4997 }, { "epoch": 31.043613707165107, "grad_norm": 2.59375, "learning_rate": 4.739327599411583e-05, "loss": 0.2071, "step": 4998 }, { "epoch": 31.049844236760123, "grad_norm": 3.125, "learning_rate": 4.7392180505651295e-05, "loss": 0.2531, "step": 4999 }, { "epoch": 31.05607476635514, "grad_norm": 5.53125, "learning_rate": 4.739108479970942e-05, "loss": 0.2008, "step": 5000 }, { "epoch": 31.062305295950157, "grad_norm": 4.0, "learning_rate": 4.738998887630086e-05, "loss": 0.1564, "step": 5001 }, { "epoch": 31.068535825545172, "grad_norm": 3.015625, "learning_rate": 4.738889273543625e-05, "loss": 0.2226, "step": 5002 }, { "epoch": 31.074766355140188, "grad_norm": 3.78125, "learning_rate": 4.738779637712624e-05, "loss": 0.3273, "step": 5003 }, { "epoch": 31.080996884735203, "grad_norm": 5.59375, "learning_rate": 4.7386699801381465e-05, "loss": 0.2377, "step": 5004 }, { "epoch": 31.08722741433022, "grad_norm": 5.5, "learning_rate": 4.73856030082126e-05, "loss": 0.1689, "step": 5005 }, { "epoch": 31.093457943925234, "grad_norm": 3.59375, "learning_rate": 4.7384505997630276e-05, "loss": 0.3206, "step": 5006 }, { "epoch": 31.09968847352025, "grad_norm": 2.625, "learning_rate": 4.738340876964516e-05, "loss": 0.1825, "step": 5007 }, { "epoch": 31.105919003115265, "grad_norm": 2.796875, "learning_rate": 4.7382311324267894e-05, "loss": 0.219, "step": 5008 }, { "epoch": 31.11214953271028, "grad_norm": 2.703125, "learning_rate": 4.738121366150916e-05, "loss": 0.1757, "step": 5009 }, { "epoch": 31.118380062305295, "grad_norm": 2.71875, "learning_rate": 4.73801157813796e-05, "loss": 0.2079, "step": 5010 }, { "epoch": 31.12461059190031, "grad_norm": 3.84375, "learning_rate": 4.737901768388988e-05, "loss": 0.1998, "step": 5011 }, { "epoch": 31.130841121495326, "grad_norm": 2.796875, "learning_rate": 4.737791936905067e-05, "loss": 0.2534, "step": 5012 }, { "epoch": 31.13707165109034, "grad_norm": 3.328125, "learning_rate": 4.737682083687263e-05, "loss": 0.155, "step": 5013 }, { "epoch": 31.143302180685357, "grad_norm": 4.375, "learning_rate": 4.737572208736643e-05, "loss": 0.2321, "step": 5014 }, { "epoch": 31.149532710280372, "grad_norm": 2.5625, "learning_rate": 4.737462312054275e-05, "loss": 0.2804, "step": 5015 }, { "epoch": 31.15576323987539, "grad_norm": 3.09375, "learning_rate": 4.7373523936412256e-05, "loss": 0.376, "step": 5016 }, { "epoch": 31.161993769470406, "grad_norm": 3.875, "learning_rate": 4.737242453498563e-05, "loss": 0.261, "step": 5017 }, { "epoch": 31.16822429906542, "grad_norm": 2.171875, "learning_rate": 4.737132491627354e-05, "loss": 0.2297, "step": 5018 }, { "epoch": 31.174454828660437, "grad_norm": 2.828125, "learning_rate": 4.7370225080286666e-05, "loss": 0.1423, "step": 5019 }, { "epoch": 31.180685358255452, "grad_norm": 3.375, "learning_rate": 4.73691250270357e-05, "loss": 0.1841, "step": 5020 }, { "epoch": 31.186915887850468, "grad_norm": 2.640625, "learning_rate": 4.7368024756531316e-05, "loss": 0.1776, "step": 5021 }, { "epoch": 31.193146417445483, "grad_norm": 2.703125, "learning_rate": 4.736692426878421e-05, "loss": 0.3129, "step": 5022 }, { "epoch": 31.1993769470405, "grad_norm": 1.609375, "learning_rate": 4.736582356380506e-05, "loss": 0.156, "step": 5023 }, { "epoch": 31.205607476635514, "grad_norm": 3.234375, "learning_rate": 4.7364722641604565e-05, "loss": 0.2138, "step": 5024 }, { "epoch": 31.21183800623053, "grad_norm": 2.5625, "learning_rate": 4.736362150219341e-05, "loss": 0.3184, "step": 5025 }, { "epoch": 31.218068535825545, "grad_norm": 2.59375, "learning_rate": 4.736252014558229e-05, "loss": 0.1484, "step": 5026 }, { "epoch": 31.22429906542056, "grad_norm": 2.15625, "learning_rate": 4.736141857178191e-05, "loss": 0.1768, "step": 5027 }, { "epoch": 31.230529595015575, "grad_norm": 2.25, "learning_rate": 4.736031678080296e-05, "loss": 0.1598, "step": 5028 }, { "epoch": 31.23676012461059, "grad_norm": 1.984375, "learning_rate": 4.735921477265615e-05, "loss": 0.1759, "step": 5029 }, { "epoch": 31.242990654205606, "grad_norm": 2.703125, "learning_rate": 4.7358112547352165e-05, "loss": 0.2, "step": 5030 }, { "epoch": 31.24922118380062, "grad_norm": 1.3984375, "learning_rate": 4.7357010104901725e-05, "loss": 0.1618, "step": 5031 }, { "epoch": 31.25545171339564, "grad_norm": 2.359375, "learning_rate": 4.735590744531554e-05, "loss": 0.2206, "step": 5032 }, { "epoch": 31.261682242990656, "grad_norm": 3.359375, "learning_rate": 4.735480456860431e-05, "loss": 0.4582, "step": 5033 }, { "epoch": 31.26791277258567, "grad_norm": 3.25, "learning_rate": 4.735370147477874e-05, "loss": 0.2731, "step": 5034 }, { "epoch": 31.274143302180686, "grad_norm": 1.9296875, "learning_rate": 4.735259816384957e-05, "loss": 0.1335, "step": 5035 }, { "epoch": 31.2803738317757, "grad_norm": 2.953125, "learning_rate": 4.735149463582749e-05, "loss": 0.3837, "step": 5036 }, { "epoch": 31.286604361370717, "grad_norm": 3.875, "learning_rate": 4.7350390890723224e-05, "loss": 0.2538, "step": 5037 }, { "epoch": 31.292834890965732, "grad_norm": 3.03125, "learning_rate": 4.7349286928547494e-05, "loss": 0.1843, "step": 5038 }, { "epoch": 31.299065420560748, "grad_norm": 3.1875, "learning_rate": 4.734818274931102e-05, "loss": 0.243, "step": 5039 }, { "epoch": 31.305295950155763, "grad_norm": 2.125, "learning_rate": 4.7347078353024534e-05, "loss": 0.1343, "step": 5040 }, { "epoch": 31.31152647975078, "grad_norm": 2.21875, "learning_rate": 4.734597373969876e-05, "loss": 0.1377, "step": 5041 }, { "epoch": 31.317757009345794, "grad_norm": 3.78125, "learning_rate": 4.734486890934441e-05, "loss": 0.3313, "step": 5042 }, { "epoch": 31.32398753894081, "grad_norm": 2.953125, "learning_rate": 4.7343763861972234e-05, "loss": 0.1404, "step": 5043 }, { "epoch": 31.330218068535824, "grad_norm": 2.21875, "learning_rate": 4.7342658597592954e-05, "loss": 0.2054, "step": 5044 }, { "epoch": 31.33644859813084, "grad_norm": 1.84375, "learning_rate": 4.734155311621731e-05, "loss": 0.1604, "step": 5045 }, { "epoch": 31.342679127725855, "grad_norm": 2.75, "learning_rate": 4.734044741785603e-05, "loss": 0.2524, "step": 5046 }, { "epoch": 31.34890965732087, "grad_norm": 2.546875, "learning_rate": 4.733934150251986e-05, "loss": 0.1827, "step": 5047 }, { "epoch": 31.35514018691589, "grad_norm": 2.609375, "learning_rate": 4.733823537021955e-05, "loss": 0.1642, "step": 5048 }, { "epoch": 31.361370716510905, "grad_norm": 3.015625, "learning_rate": 4.7337129020965824e-05, "loss": 0.1725, "step": 5049 }, { "epoch": 31.36760124610592, "grad_norm": 2.46875, "learning_rate": 4.733602245476944e-05, "loss": 0.1369, "step": 5050 }, { "epoch": 31.373831775700936, "grad_norm": 2.5, "learning_rate": 4.7334915671641136e-05, "loss": 0.1365, "step": 5051 }, { "epoch": 31.38006230529595, "grad_norm": 3.09375, "learning_rate": 4.7333808671591674e-05, "loss": 0.2846, "step": 5052 }, { "epoch": 31.386292834890966, "grad_norm": 2.671875, "learning_rate": 4.7332701454631786e-05, "loss": 0.1788, "step": 5053 }, { "epoch": 31.39252336448598, "grad_norm": 1.890625, "learning_rate": 4.733159402077225e-05, "loss": 0.1981, "step": 5054 }, { "epoch": 31.398753894080997, "grad_norm": 3.4375, "learning_rate": 4.73304863700238e-05, "loss": 0.2816, "step": 5055 }, { "epoch": 31.404984423676012, "grad_norm": 1.9296875, "learning_rate": 4.7329378502397204e-05, "loss": 0.202, "step": 5056 }, { "epoch": 31.411214953271028, "grad_norm": 3.78125, "learning_rate": 4.7328270417903224e-05, "loss": 0.2047, "step": 5057 }, { "epoch": 31.417445482866043, "grad_norm": 1.8515625, "learning_rate": 4.7327162116552614e-05, "loss": 0.1546, "step": 5058 }, { "epoch": 31.42367601246106, "grad_norm": 2.125, "learning_rate": 4.732605359835615e-05, "loss": 0.1683, "step": 5059 }, { "epoch": 31.429906542056074, "grad_norm": 3.21875, "learning_rate": 4.732494486332458e-05, "loss": 0.3011, "step": 5060 }, { "epoch": 31.43613707165109, "grad_norm": 2.1875, "learning_rate": 4.732383591146869e-05, "loss": 0.1964, "step": 5061 }, { "epoch": 31.442367601246104, "grad_norm": 1.75, "learning_rate": 4.7322726742799246e-05, "loss": 0.1388, "step": 5062 }, { "epoch": 31.44859813084112, "grad_norm": 2.265625, "learning_rate": 4.7321617357327006e-05, "loss": 0.1747, "step": 5063 }, { "epoch": 31.45482866043614, "grad_norm": 3.625, "learning_rate": 4.732050775506277e-05, "loss": 0.2565, "step": 5064 }, { "epoch": 31.461059190031154, "grad_norm": 2.78125, "learning_rate": 4.7319397936017294e-05, "loss": 0.2996, "step": 5065 }, { "epoch": 31.46728971962617, "grad_norm": 2.8125, "learning_rate": 4.731828790020136e-05, "loss": 0.1495, "step": 5066 }, { "epoch": 31.473520249221185, "grad_norm": 2.15625, "learning_rate": 4.731717764762576e-05, "loss": 0.1573, "step": 5067 }, { "epoch": 31.4797507788162, "grad_norm": 1.5390625, "learning_rate": 4.7316067178301264e-05, "loss": 0.1297, "step": 5068 }, { "epoch": 31.485981308411215, "grad_norm": 2.359375, "learning_rate": 4.7314956492238665e-05, "loss": 0.1462, "step": 5069 }, { "epoch": 31.49221183800623, "grad_norm": 3.046875, "learning_rate": 4.731384558944875e-05, "loss": 0.2732, "step": 5070 }, { "epoch": 31.498442367601246, "grad_norm": 1.34375, "learning_rate": 4.73127344699423e-05, "loss": 0.1212, "step": 5071 }, { "epoch": 31.50467289719626, "grad_norm": 2.921875, "learning_rate": 4.731162313373012e-05, "loss": 0.1244, "step": 5072 }, { "epoch": 31.510903426791277, "grad_norm": 1.5625, "learning_rate": 4.731051158082299e-05, "loss": 0.1383, "step": 5073 }, { "epoch": 31.517133956386292, "grad_norm": 3.640625, "learning_rate": 4.730939981123171e-05, "loss": 0.2785, "step": 5074 }, { "epoch": 31.523364485981308, "grad_norm": 3.59375, "learning_rate": 4.730828782496709e-05, "loss": 0.2583, "step": 5075 }, { "epoch": 31.529595015576323, "grad_norm": 2.15625, "learning_rate": 4.7307175622039914e-05, "loss": 0.1277, "step": 5076 }, { "epoch": 31.53582554517134, "grad_norm": 1.84375, "learning_rate": 4.730606320246099e-05, "loss": 0.1071, "step": 5077 }, { "epoch": 31.542056074766354, "grad_norm": 4.0, "learning_rate": 4.730495056624112e-05, "loss": 0.2585, "step": 5078 }, { "epoch": 31.54828660436137, "grad_norm": 5.28125, "learning_rate": 4.7303837713391105e-05, "loss": 0.1959, "step": 5079 }, { "epoch": 31.554517133956388, "grad_norm": 6.3125, "learning_rate": 4.730272464392177e-05, "loss": 0.3489, "step": 5080 }, { "epoch": 31.560747663551403, "grad_norm": 5.03125, "learning_rate": 4.730161135784391e-05, "loss": 0.3942, "step": 5081 }, { "epoch": 31.56697819314642, "grad_norm": 3.34375, "learning_rate": 4.7300497855168344e-05, "loss": 0.1934, "step": 5082 }, { "epoch": 31.573208722741434, "grad_norm": 2.171875, "learning_rate": 4.729938413590588e-05, "loss": 0.1747, "step": 5083 }, { "epoch": 31.57943925233645, "grad_norm": 4.84375, "learning_rate": 4.729827020006735e-05, "loss": 0.1914, "step": 5084 }, { "epoch": 31.585669781931465, "grad_norm": 4.40625, "learning_rate": 4.729715604766355e-05, "loss": 0.1757, "step": 5085 }, { "epoch": 31.59190031152648, "grad_norm": 3.265625, "learning_rate": 4.729604167870533e-05, "loss": 0.2883, "step": 5086 }, { "epoch": 31.598130841121495, "grad_norm": 3.140625, "learning_rate": 4.729492709320348e-05, "loss": 0.263, "step": 5087 }, { "epoch": 31.60436137071651, "grad_norm": 2.4375, "learning_rate": 4.729381229116885e-05, "loss": 0.1922, "step": 5088 }, { "epoch": 31.610591900311526, "grad_norm": 3.4375, "learning_rate": 4.729269727261225e-05, "loss": 0.3171, "step": 5089 }, { "epoch": 31.61682242990654, "grad_norm": 3.125, "learning_rate": 4.729158203754453e-05, "loss": 0.2916, "step": 5090 }, { "epoch": 31.623052959501557, "grad_norm": 2.640625, "learning_rate": 4.72904665859765e-05, "loss": 0.1429, "step": 5091 }, { "epoch": 31.629283489096572, "grad_norm": 2.96875, "learning_rate": 4.7289350917919005e-05, "loss": 0.1823, "step": 5092 }, { "epoch": 31.635514018691588, "grad_norm": 2.15625, "learning_rate": 4.7288235033382884e-05, "loss": 0.197, "step": 5093 }, { "epoch": 31.641744548286603, "grad_norm": 1.25, "learning_rate": 4.728711893237896e-05, "loss": 0.1186, "step": 5094 }, { "epoch": 31.64797507788162, "grad_norm": 2.90625, "learning_rate": 4.728600261491809e-05, "loss": 0.3459, "step": 5095 }, { "epoch": 31.654205607476637, "grad_norm": 3.734375, "learning_rate": 4.7284886081011105e-05, "loss": 0.2535, "step": 5096 }, { "epoch": 31.660436137071652, "grad_norm": 2.6875, "learning_rate": 4.728376933066885e-05, "loss": 0.1971, "step": 5097 }, { "epoch": 31.666666666666668, "grad_norm": 2.421875, "learning_rate": 4.728265236390217e-05, "loss": 0.1492, "step": 5098 }, { "epoch": 31.672897196261683, "grad_norm": 3.0, "learning_rate": 4.728153518072193e-05, "loss": 0.1798, "step": 5099 }, { "epoch": 31.6791277258567, "grad_norm": 2.59375, "learning_rate": 4.728041778113895e-05, "loss": 0.2716, "step": 5100 }, { "epoch": 31.685358255451714, "grad_norm": 2.125, "learning_rate": 4.72793001651641e-05, "loss": 0.1432, "step": 5101 }, { "epoch": 31.69158878504673, "grad_norm": 3.328125, "learning_rate": 4.727818233280824e-05, "loss": 0.3046, "step": 5102 }, { "epoch": 31.697819314641745, "grad_norm": 3.328125, "learning_rate": 4.7277064284082216e-05, "loss": 0.334, "step": 5103 }, { "epoch": 31.70404984423676, "grad_norm": 3.671875, "learning_rate": 4.7275946018996894e-05, "loss": 0.3441, "step": 5104 }, { "epoch": 31.710280373831775, "grad_norm": 2.21875, "learning_rate": 4.727482753756313e-05, "loss": 0.2731, "step": 5105 }, { "epoch": 31.71651090342679, "grad_norm": 4.28125, "learning_rate": 4.727370883979179e-05, "loss": 0.2697, "step": 5106 }, { "epoch": 31.722741433021806, "grad_norm": 3.0, "learning_rate": 4.7272589925693735e-05, "loss": 0.1635, "step": 5107 }, { "epoch": 31.72897196261682, "grad_norm": 3.21875, "learning_rate": 4.727147079527983e-05, "loss": 0.2187, "step": 5108 }, { "epoch": 31.735202492211837, "grad_norm": 2.5625, "learning_rate": 4.7270351448560954e-05, "loss": 0.3282, "step": 5109 }, { "epoch": 31.741433021806852, "grad_norm": 3.359375, "learning_rate": 4.726923188554797e-05, "loss": 0.2521, "step": 5110 }, { "epoch": 31.747663551401867, "grad_norm": 2.765625, "learning_rate": 4.726811210625176e-05, "loss": 0.3152, "step": 5111 }, { "epoch": 31.753894080996886, "grad_norm": 3.828125, "learning_rate": 4.7266992110683196e-05, "loss": 0.3511, "step": 5112 }, { "epoch": 31.7601246105919, "grad_norm": 2.953125, "learning_rate": 4.726587189885314e-05, "loss": 0.2264, "step": 5113 }, { "epoch": 31.766355140186917, "grad_norm": 2.515625, "learning_rate": 4.72647514707725e-05, "loss": 0.217, "step": 5114 }, { "epoch": 31.772585669781932, "grad_norm": 2.4375, "learning_rate": 4.7263630826452135e-05, "loss": 0.2135, "step": 5115 }, { "epoch": 31.778816199376948, "grad_norm": 3.125, "learning_rate": 4.726250996590294e-05, "loss": 0.1883, "step": 5116 }, { "epoch": 31.785046728971963, "grad_norm": 5.125, "learning_rate": 4.726138888913579e-05, "loss": 0.2803, "step": 5117 }, { "epoch": 31.79127725856698, "grad_norm": 4.46875, "learning_rate": 4.726026759616159e-05, "loss": 0.3403, "step": 5118 }, { "epoch": 31.797507788161994, "grad_norm": 3.765625, "learning_rate": 4.725914608699122e-05, "loss": 0.303, "step": 5119 }, { "epoch": 31.80373831775701, "grad_norm": 2.65625, "learning_rate": 4.725802436163558e-05, "loss": 0.228, "step": 5120 }, { "epoch": 31.809968847352025, "grad_norm": 2.3125, "learning_rate": 4.725690242010554e-05, "loss": 0.127, "step": 5121 }, { "epoch": 31.81619937694704, "grad_norm": 2.59375, "learning_rate": 4.725578026241203e-05, "loss": 0.2752, "step": 5122 }, { "epoch": 31.822429906542055, "grad_norm": 1.6953125, "learning_rate": 4.725465788856593e-05, "loss": 0.1369, "step": 5123 }, { "epoch": 31.82866043613707, "grad_norm": 2.125, "learning_rate": 4.7253535298578145e-05, "loss": 0.1876, "step": 5124 }, { "epoch": 31.834890965732086, "grad_norm": 2.796875, "learning_rate": 4.7252412492459576e-05, "loss": 0.2304, "step": 5125 }, { "epoch": 31.8411214953271, "grad_norm": 1.6640625, "learning_rate": 4.725128947022113e-05, "loss": 0.1477, "step": 5126 }, { "epoch": 31.847352024922117, "grad_norm": 3.765625, "learning_rate": 4.725016623187371e-05, "loss": 0.2598, "step": 5127 }, { "epoch": 31.853582554517136, "grad_norm": 2.421875, "learning_rate": 4.724904277742822e-05, "loss": 0.1787, "step": 5128 }, { "epoch": 31.85981308411215, "grad_norm": 1.515625, "learning_rate": 4.724791910689559e-05, "loss": 0.1228, "step": 5129 }, { "epoch": 31.866043613707166, "grad_norm": 2.484375, "learning_rate": 4.724679522028672e-05, "loss": 0.2807, "step": 5130 }, { "epoch": 31.87227414330218, "grad_norm": 2.359375, "learning_rate": 4.7245671117612525e-05, "loss": 0.2857, "step": 5131 }, { "epoch": 31.878504672897197, "grad_norm": 1.15625, "learning_rate": 4.7244546798883926e-05, "loss": 0.1525, "step": 5132 }, { "epoch": 31.884735202492212, "grad_norm": 2.796875, "learning_rate": 4.724342226411185e-05, "loss": 0.2561, "step": 5133 }, { "epoch": 31.890965732087228, "grad_norm": 2.828125, "learning_rate": 4.7242297513307196e-05, "loss": 0.2225, "step": 5134 }, { "epoch": 31.897196261682243, "grad_norm": 3.078125, "learning_rate": 4.724117254648092e-05, "loss": 0.2131, "step": 5135 }, { "epoch": 31.90342679127726, "grad_norm": 2.78125, "learning_rate": 4.724004736364391e-05, "loss": 0.2567, "step": 5136 }, { "epoch": 31.909657320872274, "grad_norm": 3.390625, "learning_rate": 4.723892196480713e-05, "loss": 0.185, "step": 5137 }, { "epoch": 31.91588785046729, "grad_norm": 3.484375, "learning_rate": 4.723779634998149e-05, "loss": 0.2204, "step": 5138 }, { "epoch": 31.922118380062305, "grad_norm": 2.921875, "learning_rate": 4.723667051917792e-05, "loss": 0.2328, "step": 5139 }, { "epoch": 31.92834890965732, "grad_norm": 2.09375, "learning_rate": 4.723554447240737e-05, "loss": 0.1728, "step": 5140 }, { "epoch": 31.934579439252335, "grad_norm": 1.1640625, "learning_rate": 4.723441820968076e-05, "loss": 0.1232, "step": 5141 }, { "epoch": 31.94080996884735, "grad_norm": 5.25, "learning_rate": 4.723329173100905e-05, "loss": 0.1777, "step": 5142 }, { "epoch": 31.947040498442366, "grad_norm": 5.40625, "learning_rate": 4.7232165036403154e-05, "loss": 0.1957, "step": 5143 }, { "epoch": 31.953271028037385, "grad_norm": 2.25, "learning_rate": 4.723103812587403e-05, "loss": 0.1846, "step": 5144 }, { "epoch": 31.9595015576324, "grad_norm": 4.3125, "learning_rate": 4.722991099943261e-05, "loss": 0.2763, "step": 5145 }, { "epoch": 31.965732087227416, "grad_norm": 2.15625, "learning_rate": 4.7228783657089865e-05, "loss": 0.164, "step": 5146 }, { "epoch": 31.97196261682243, "grad_norm": 4.71875, "learning_rate": 4.722765609885672e-05, "loss": 0.2665, "step": 5147 }, { "epoch": 31.978193146417446, "grad_norm": 4.6875, "learning_rate": 4.722652832474415e-05, "loss": 0.1506, "step": 5148 }, { "epoch": 31.98442367601246, "grad_norm": 5.15625, "learning_rate": 4.722540033476308e-05, "loss": 0.3722, "step": 5149 }, { "epoch": 31.990654205607477, "grad_norm": 3.546875, "learning_rate": 4.722427212892448e-05, "loss": 0.1189, "step": 5150 }, { "epoch": 31.996884735202492, "grad_norm": 2.71875, "learning_rate": 4.722314370723931e-05, "loss": 0.1996, "step": 5151 }, { "epoch": 32.0, "grad_norm": 3.203125, "learning_rate": 4.722201506971853e-05, "loss": 0.2339, "step": 5152 }, { "epoch": 32.006230529595015, "grad_norm": 5.6875, "learning_rate": 4.722088621637309e-05, "loss": 0.1994, "step": 5153 }, { "epoch": 32.01246105919003, "grad_norm": 6.5625, "learning_rate": 4.721975714721396e-05, "loss": 0.1591, "step": 5154 }, { "epoch": 32.018691588785046, "grad_norm": 6.0, "learning_rate": 4.721862786225212e-05, "loss": 0.403, "step": 5155 }, { "epoch": 32.02492211838006, "grad_norm": 4.46875, "learning_rate": 4.721749836149851e-05, "loss": 0.1692, "step": 5156 }, { "epoch": 32.03115264797508, "grad_norm": 3.71875, "learning_rate": 4.721636864496413e-05, "loss": 0.3821, "step": 5157 }, { "epoch": 32.03738317757009, "grad_norm": 5.0625, "learning_rate": 4.721523871265992e-05, "loss": 0.253, "step": 5158 }, { "epoch": 32.04361370716511, "grad_norm": 4.34375, "learning_rate": 4.721410856459688e-05, "loss": 0.1847, "step": 5159 }, { "epoch": 32.04984423676012, "grad_norm": 6.1875, "learning_rate": 4.721297820078597e-05, "loss": 0.2948, "step": 5160 }, { "epoch": 32.05607476635514, "grad_norm": 3.296875, "learning_rate": 4.7211847621238185e-05, "loss": 0.1571, "step": 5161 }, { "epoch": 32.06230529595015, "grad_norm": 2.734375, "learning_rate": 4.7210716825964496e-05, "loss": 0.1648, "step": 5162 }, { "epoch": 32.06853582554517, "grad_norm": 4.15625, "learning_rate": 4.7209585814975876e-05, "loss": 0.2149, "step": 5163 }, { "epoch": 32.074766355140184, "grad_norm": 2.59375, "learning_rate": 4.720845458828332e-05, "loss": 0.1392, "step": 5164 }, { "epoch": 32.0809968847352, "grad_norm": 3.5, "learning_rate": 4.720732314589783e-05, "loss": 0.2167, "step": 5165 }, { "epoch": 32.087227414330215, "grad_norm": 3.578125, "learning_rate": 4.720619148783036e-05, "loss": 0.2682, "step": 5166 }, { "epoch": 32.09345794392523, "grad_norm": 3.9375, "learning_rate": 4.720505961409193e-05, "loss": 0.1404, "step": 5167 }, { "epoch": 32.099688473520246, "grad_norm": 2.234375, "learning_rate": 4.720392752469351e-05, "loss": 0.1735, "step": 5168 }, { "epoch": 32.10591900311527, "grad_norm": 3.828125, "learning_rate": 4.720279521964612e-05, "loss": 0.1499, "step": 5169 }, { "epoch": 32.11214953271028, "grad_norm": 3.71875, "learning_rate": 4.720166269896074e-05, "loss": 0.1503, "step": 5170 }, { "epoch": 32.1183800623053, "grad_norm": 3.21875, "learning_rate": 4.720052996264838e-05, "loss": 0.2824, "step": 5171 }, { "epoch": 32.124610591900314, "grad_norm": 3.34375, "learning_rate": 4.719939701072003e-05, "loss": 0.2994, "step": 5172 }, { "epoch": 32.13084112149533, "grad_norm": 2.3125, "learning_rate": 4.719826384318669e-05, "loss": 0.2415, "step": 5173 }, { "epoch": 32.137071651090345, "grad_norm": 2.390625, "learning_rate": 4.719713046005938e-05, "loss": 0.1746, "step": 5174 }, { "epoch": 32.14330218068536, "grad_norm": 3.703125, "learning_rate": 4.719599686134911e-05, "loss": 0.2241, "step": 5175 }, { "epoch": 32.149532710280376, "grad_norm": 3.34375, "learning_rate": 4.719486304706687e-05, "loss": 0.2274, "step": 5176 }, { "epoch": 32.15576323987539, "grad_norm": 4.40625, "learning_rate": 4.719372901722369e-05, "loss": 0.2543, "step": 5177 }, { "epoch": 32.161993769470406, "grad_norm": 2.796875, "learning_rate": 4.719259477183058e-05, "loss": 0.1926, "step": 5178 }, { "epoch": 32.16822429906542, "grad_norm": 3.125, "learning_rate": 4.719146031089855e-05, "loss": 0.1888, "step": 5179 }, { "epoch": 32.17445482866044, "grad_norm": 4.15625, "learning_rate": 4.719032563443862e-05, "loss": 0.2434, "step": 5180 }, { "epoch": 32.18068535825545, "grad_norm": 2.671875, "learning_rate": 4.718919074246181e-05, "loss": 0.2855, "step": 5181 }, { "epoch": 32.18691588785047, "grad_norm": 2.8125, "learning_rate": 4.718805563497914e-05, "loss": 0.1468, "step": 5182 }, { "epoch": 32.19314641744548, "grad_norm": 1.953125, "learning_rate": 4.7186920312001654e-05, "loss": 0.1833, "step": 5183 }, { "epoch": 32.1993769470405, "grad_norm": 2.734375, "learning_rate": 4.718578477354035e-05, "loss": 0.265, "step": 5184 }, { "epoch": 32.205607476635514, "grad_norm": 3.21875, "learning_rate": 4.7184649019606275e-05, "loss": 0.118, "step": 5185 }, { "epoch": 32.21183800623053, "grad_norm": 3.34375, "learning_rate": 4.7183513050210455e-05, "loss": 0.2761, "step": 5186 }, { "epoch": 32.218068535825545, "grad_norm": 2.390625, "learning_rate": 4.718237686536392e-05, "loss": 0.1722, "step": 5187 }, { "epoch": 32.22429906542056, "grad_norm": 3.96875, "learning_rate": 4.71812404650777e-05, "loss": 0.3862, "step": 5188 }, { "epoch": 32.230529595015575, "grad_norm": 2.671875, "learning_rate": 4.718010384936285e-05, "loss": 0.2912, "step": 5189 }, { "epoch": 32.23676012461059, "grad_norm": 2.75, "learning_rate": 4.717896701823039e-05, "loss": 0.2733, "step": 5190 }, { "epoch": 32.242990654205606, "grad_norm": 3.640625, "learning_rate": 4.7177829971691376e-05, "loss": 0.3513, "step": 5191 }, { "epoch": 32.24922118380062, "grad_norm": 2.484375, "learning_rate": 4.7176692709756835e-05, "loss": 0.2667, "step": 5192 }, { "epoch": 32.25545171339564, "grad_norm": 2.796875, "learning_rate": 4.717555523243783e-05, "loss": 0.2473, "step": 5193 }, { "epoch": 32.26168224299065, "grad_norm": 2.03125, "learning_rate": 4.717441753974539e-05, "loss": 0.1375, "step": 5194 }, { "epoch": 32.26791277258567, "grad_norm": 1.296875, "learning_rate": 4.717327963169058e-05, "loss": 0.1487, "step": 5195 }, { "epoch": 32.27414330218068, "grad_norm": 1.6953125, "learning_rate": 4.717214150828445e-05, "loss": 0.1415, "step": 5196 }, { "epoch": 32.2803738317757, "grad_norm": 2.59375, "learning_rate": 4.717100316953804e-05, "loss": 0.2573, "step": 5197 }, { "epoch": 32.28660436137071, "grad_norm": 1.921875, "learning_rate": 4.7169864615462425e-05, "loss": 0.1898, "step": 5198 }, { "epoch": 32.29283489096573, "grad_norm": 2.640625, "learning_rate": 4.716872584606865e-05, "loss": 0.204, "step": 5199 }, { "epoch": 32.299065420560744, "grad_norm": 1.59375, "learning_rate": 4.7167586861367774e-05, "loss": 0.134, "step": 5200 }, { "epoch": 32.30529595015577, "grad_norm": 4.375, "learning_rate": 4.7166447661370865e-05, "loss": 0.1564, "step": 5201 }, { "epoch": 32.31152647975078, "grad_norm": 3.421875, "learning_rate": 4.716530824608899e-05, "loss": 0.2392, "step": 5202 }, { "epoch": 32.3177570093458, "grad_norm": 2.640625, "learning_rate": 4.71641686155332e-05, "loss": 0.1519, "step": 5203 }, { "epoch": 32.32398753894081, "grad_norm": 2.859375, "learning_rate": 4.7163028769714586e-05, "loss": 0.2571, "step": 5204 }, { "epoch": 32.33021806853583, "grad_norm": 3.0, "learning_rate": 4.7161888708644197e-05, "loss": 0.2418, "step": 5205 }, { "epoch": 32.33644859813084, "grad_norm": 2.671875, "learning_rate": 4.7160748432333126e-05, "loss": 0.2051, "step": 5206 }, { "epoch": 32.34267912772586, "grad_norm": 2.8125, "learning_rate": 4.715960794079243e-05, "loss": 0.2238, "step": 5207 }, { "epoch": 32.348909657320874, "grad_norm": 2.75, "learning_rate": 4.7158467234033186e-05, "loss": 0.2294, "step": 5208 }, { "epoch": 32.35514018691589, "grad_norm": 2.9375, "learning_rate": 4.715732631206649e-05, "loss": 0.2188, "step": 5209 }, { "epoch": 32.361370716510905, "grad_norm": 1.6640625, "learning_rate": 4.71561851749034e-05, "loss": 0.1189, "step": 5210 }, { "epoch": 32.36760124610592, "grad_norm": 2.765625, "learning_rate": 4.715504382255502e-05, "loss": 0.254, "step": 5211 }, { "epoch": 32.373831775700936, "grad_norm": 3.59375, "learning_rate": 4.715390225503242e-05, "loss": 0.339, "step": 5212 }, { "epoch": 32.38006230529595, "grad_norm": 1.78125, "learning_rate": 4.7152760472346694e-05, "loss": 0.1456, "step": 5213 }, { "epoch": 32.386292834890966, "grad_norm": 3.578125, "learning_rate": 4.715161847450893e-05, "loss": 0.2779, "step": 5214 }, { "epoch": 32.39252336448598, "grad_norm": 3.203125, "learning_rate": 4.715047626153022e-05, "loss": 0.239, "step": 5215 }, { "epoch": 32.398753894081, "grad_norm": 1.75, "learning_rate": 4.714933383342166e-05, "loss": 0.1581, "step": 5216 }, { "epoch": 32.40498442367601, "grad_norm": 2.703125, "learning_rate": 4.7148191190194336e-05, "loss": 0.2857, "step": 5217 }, { "epoch": 32.41121495327103, "grad_norm": 4.03125, "learning_rate": 4.7147048331859356e-05, "loss": 0.1566, "step": 5218 }, { "epoch": 32.41744548286604, "grad_norm": 2.984375, "learning_rate": 4.714590525842781e-05, "loss": 0.2233, "step": 5219 }, { "epoch": 32.42367601246106, "grad_norm": 3.734375, "learning_rate": 4.7144761969910814e-05, "loss": 0.3076, "step": 5220 }, { "epoch": 32.429906542056074, "grad_norm": 3.125, "learning_rate": 4.7143618466319464e-05, "loss": 0.1419, "step": 5221 }, { "epoch": 32.43613707165109, "grad_norm": 2.1875, "learning_rate": 4.7142474747664856e-05, "loss": 0.1453, "step": 5222 }, { "epoch": 32.442367601246104, "grad_norm": 3.21875, "learning_rate": 4.714133081395811e-05, "loss": 0.2084, "step": 5223 }, { "epoch": 32.44859813084112, "grad_norm": 3.421875, "learning_rate": 4.714018666521034e-05, "loss": 0.3714, "step": 5224 }, { "epoch": 32.454828660436135, "grad_norm": 2.46875, "learning_rate": 4.713904230143264e-05, "loss": 0.1749, "step": 5225 }, { "epoch": 32.46105919003115, "grad_norm": 3.15625, "learning_rate": 4.7137897722636144e-05, "loss": 0.2027, "step": 5226 }, { "epoch": 32.467289719626166, "grad_norm": 2.375, "learning_rate": 4.713675292883196e-05, "loss": 0.1944, "step": 5227 }, { "epoch": 32.47352024922118, "grad_norm": 2.0625, "learning_rate": 4.71356079200312e-05, "loss": 0.1606, "step": 5228 }, { "epoch": 32.4797507788162, "grad_norm": 2.6875, "learning_rate": 4.7134462696245e-05, "loss": 0.1963, "step": 5229 }, { "epoch": 32.48598130841121, "grad_norm": 1.96875, "learning_rate": 4.713331725748447e-05, "loss": 0.1244, "step": 5230 }, { "epoch": 32.49221183800623, "grad_norm": 3.359375, "learning_rate": 4.713217160376073e-05, "loss": 0.2315, "step": 5231 }, { "epoch": 32.49844236760124, "grad_norm": 2.65625, "learning_rate": 4.713102573508492e-05, "loss": 0.1566, "step": 5232 }, { "epoch": 32.504672897196265, "grad_norm": 2.3125, "learning_rate": 4.7129879651468164e-05, "loss": 0.146, "step": 5233 }, { "epoch": 32.51090342679128, "grad_norm": 2.453125, "learning_rate": 4.71287333529216e-05, "loss": 0.2152, "step": 5234 }, { "epoch": 32.517133956386296, "grad_norm": 2.34375, "learning_rate": 4.712758683945635e-05, "loss": 0.1501, "step": 5235 }, { "epoch": 32.52336448598131, "grad_norm": 3.875, "learning_rate": 4.712644011108355e-05, "loss": 0.2594, "step": 5236 }, { "epoch": 32.52959501557633, "grad_norm": 2.265625, "learning_rate": 4.7125293167814345e-05, "loss": 0.1453, "step": 5237 }, { "epoch": 32.53582554517134, "grad_norm": 3.4375, "learning_rate": 4.712414600965987e-05, "loss": 0.1896, "step": 5238 }, { "epoch": 32.54205607476636, "grad_norm": 2.6875, "learning_rate": 4.712299863663127e-05, "loss": 0.2697, "step": 5239 }, { "epoch": 32.54828660436137, "grad_norm": 3.953125, "learning_rate": 4.712185104873967e-05, "loss": 0.3648, "step": 5240 }, { "epoch": 32.55451713395639, "grad_norm": 2.84375, "learning_rate": 4.712070324599624e-05, "loss": 0.2748, "step": 5241 }, { "epoch": 32.5607476635514, "grad_norm": 2.640625, "learning_rate": 4.711955522841213e-05, "loss": 0.2683, "step": 5242 }, { "epoch": 32.56697819314642, "grad_norm": 2.546875, "learning_rate": 4.7118406995998466e-05, "loss": 0.1599, "step": 5243 }, { "epoch": 32.573208722741434, "grad_norm": 2.59375, "learning_rate": 4.7117258548766404e-05, "loss": 0.1563, "step": 5244 }, { "epoch": 32.57943925233645, "grad_norm": 3.09375, "learning_rate": 4.711610988672712e-05, "loss": 0.1413, "step": 5245 }, { "epoch": 32.585669781931465, "grad_norm": 3.609375, "learning_rate": 4.711496100989176e-05, "loss": 0.3167, "step": 5246 }, { "epoch": 32.59190031152648, "grad_norm": 3.375, "learning_rate": 4.711381191827146e-05, "loss": 0.2269, "step": 5247 }, { "epoch": 32.598130841121495, "grad_norm": 2.515625, "learning_rate": 4.7112662611877414e-05, "loss": 0.2584, "step": 5248 }, { "epoch": 32.60436137071651, "grad_norm": 2.140625, "learning_rate": 4.711151309072075e-05, "loss": 0.1367, "step": 5249 }, { "epoch": 32.610591900311526, "grad_norm": 2.84375, "learning_rate": 4.711036335481267e-05, "loss": 0.2016, "step": 5250 }, { "epoch": 32.61682242990654, "grad_norm": 1.359375, "learning_rate": 4.710921340416431e-05, "loss": 0.1218, "step": 5251 }, { "epoch": 32.62305295950156, "grad_norm": 2.9375, "learning_rate": 4.710806323878685e-05, "loss": 0.2165, "step": 5252 }, { "epoch": 32.62928348909657, "grad_norm": 3.015625, "learning_rate": 4.710691285869146e-05, "loss": 0.2247, "step": 5253 }, { "epoch": 32.63551401869159, "grad_norm": 2.8125, "learning_rate": 4.710576226388932e-05, "loss": 0.1547, "step": 5254 }, { "epoch": 32.6417445482866, "grad_norm": 2.3125, "learning_rate": 4.7104611454391604e-05, "loss": 0.1962, "step": 5255 }, { "epoch": 32.64797507788162, "grad_norm": 1.875, "learning_rate": 4.710346043020947e-05, "loss": 0.1357, "step": 5256 }, { "epoch": 32.654205607476634, "grad_norm": 2.6875, "learning_rate": 4.710230919135412e-05, "loss": 0.1891, "step": 5257 }, { "epoch": 32.66043613707165, "grad_norm": 2.84375, "learning_rate": 4.710115773783671e-05, "loss": 0.1668, "step": 5258 }, { "epoch": 32.666666666666664, "grad_norm": 2.109375, "learning_rate": 4.710000606966845e-05, "loss": 0.1401, "step": 5259 }, { "epoch": 32.67289719626168, "grad_norm": 2.234375, "learning_rate": 4.709885418686051e-05, "loss": 0.2056, "step": 5260 }, { "epoch": 32.679127725856695, "grad_norm": 2.5625, "learning_rate": 4.709770208942408e-05, "loss": 0.211, "step": 5261 }, { "epoch": 32.68535825545171, "grad_norm": 1.796875, "learning_rate": 4.7096549777370355e-05, "loss": 0.1529, "step": 5262 }, { "epoch": 32.691588785046726, "grad_norm": 1.9296875, "learning_rate": 4.709539725071052e-05, "loss": 0.1428, "step": 5263 }, { "epoch": 32.69781931464174, "grad_norm": 2.125, "learning_rate": 4.7094244509455766e-05, "loss": 0.1453, "step": 5264 }, { "epoch": 32.70404984423676, "grad_norm": 2.703125, "learning_rate": 4.709309155361729e-05, "loss": 0.222, "step": 5265 }, { "epoch": 32.71028037383178, "grad_norm": 2.671875, "learning_rate": 4.7091938383206304e-05, "loss": 0.2145, "step": 5266 }, { "epoch": 32.716510903426794, "grad_norm": 2.0, "learning_rate": 4.709078499823398e-05, "loss": 0.1343, "step": 5267 }, { "epoch": 32.72274143302181, "grad_norm": 3.046875, "learning_rate": 4.7089631398711556e-05, "loss": 0.3087, "step": 5268 }, { "epoch": 32.728971962616825, "grad_norm": 2.1875, "learning_rate": 4.7088477584650206e-05, "loss": 0.2519, "step": 5269 }, { "epoch": 32.73520249221184, "grad_norm": 3.1875, "learning_rate": 4.708732355606114e-05, "loss": 0.235, "step": 5270 }, { "epoch": 32.741433021806856, "grad_norm": 4.84375, "learning_rate": 4.708616931295558e-05, "loss": 0.3816, "step": 5271 }, { "epoch": 32.74766355140187, "grad_norm": 1.9140625, "learning_rate": 4.7085014855344725e-05, "loss": 0.1481, "step": 5272 }, { "epoch": 32.753894080996886, "grad_norm": 2.796875, "learning_rate": 4.708386018323979e-05, "loss": 0.1958, "step": 5273 }, { "epoch": 32.7601246105919, "grad_norm": 1.578125, "learning_rate": 4.7082705296651995e-05, "loss": 0.1415, "step": 5274 }, { "epoch": 32.76635514018692, "grad_norm": 1.71875, "learning_rate": 4.7081550195592555e-05, "loss": 0.1313, "step": 5275 }, { "epoch": 32.77258566978193, "grad_norm": 1.9921875, "learning_rate": 4.708039488007268e-05, "loss": 0.1645, "step": 5276 }, { "epoch": 32.77881619937695, "grad_norm": 3.171875, "learning_rate": 4.70792393501036e-05, "loss": 0.2501, "step": 5277 }, { "epoch": 32.78504672897196, "grad_norm": 2.078125, "learning_rate": 4.707808360569653e-05, "loss": 0.2763, "step": 5278 }, { "epoch": 32.79127725856698, "grad_norm": 2.46875, "learning_rate": 4.70769276468627e-05, "loss": 0.2286, "step": 5279 }, { "epoch": 32.797507788161994, "grad_norm": 3.65625, "learning_rate": 4.707577147361333e-05, "loss": 0.1588, "step": 5280 }, { "epoch": 32.80373831775701, "grad_norm": 2.6875, "learning_rate": 4.707461508595966e-05, "loss": 0.1551, "step": 5281 }, { "epoch": 32.809968847352025, "grad_norm": 1.828125, "learning_rate": 4.7073458483912926e-05, "loss": 0.1424, "step": 5282 }, { "epoch": 32.81619937694704, "grad_norm": 3.0625, "learning_rate": 4.7072301667484334e-05, "loss": 0.2729, "step": 5283 }, { "epoch": 32.822429906542055, "grad_norm": 2.5625, "learning_rate": 4.707114463668515e-05, "loss": 0.1473, "step": 5284 }, { "epoch": 32.82866043613707, "grad_norm": 2.171875, "learning_rate": 4.706998739152658e-05, "loss": 0.1852, "step": 5285 }, { "epoch": 32.834890965732086, "grad_norm": 2.140625, "learning_rate": 4.7068829932019895e-05, "loss": 0.1899, "step": 5286 }, { "epoch": 32.8411214953271, "grad_norm": 2.40625, "learning_rate": 4.706767225817631e-05, "loss": 0.1443, "step": 5287 }, { "epoch": 32.84735202492212, "grad_norm": 3.09375, "learning_rate": 4.70665143700071e-05, "loss": 0.16, "step": 5288 }, { "epoch": 32.85358255451713, "grad_norm": 3.328125, "learning_rate": 4.7065356267523476e-05, "loss": 0.2003, "step": 5289 }, { "epoch": 32.85981308411215, "grad_norm": 1.921875, "learning_rate": 4.70641979507367e-05, "loss": 0.1524, "step": 5290 }, { "epoch": 32.86604361370716, "grad_norm": 1.5625, "learning_rate": 4.7063039419658035e-05, "loss": 0.1577, "step": 5291 }, { "epoch": 32.87227414330218, "grad_norm": 2.3125, "learning_rate": 4.706188067429871e-05, "loss": 0.1924, "step": 5292 }, { "epoch": 32.87850467289719, "grad_norm": 2.140625, "learning_rate": 4.7060721714669994e-05, "loss": 0.1373, "step": 5293 }, { "epoch": 32.88473520249221, "grad_norm": 1.6015625, "learning_rate": 4.705956254078314e-05, "loss": 0.1335, "step": 5294 }, { "epoch": 32.890965732087224, "grad_norm": 2.265625, "learning_rate": 4.70584031526494e-05, "loss": 0.1846, "step": 5295 }, { "epoch": 32.89719626168224, "grad_norm": 2.140625, "learning_rate": 4.705724355028005e-05, "loss": 0.1956, "step": 5296 }, { "epoch": 32.90342679127726, "grad_norm": 2.71875, "learning_rate": 4.705608373368633e-05, "loss": 0.1912, "step": 5297 }, { "epoch": 32.90965732087228, "grad_norm": 2.90625, "learning_rate": 4.705492370287952e-05, "loss": 0.2095, "step": 5298 }, { "epoch": 32.91588785046729, "grad_norm": 1.8125, "learning_rate": 4.705376345787089e-05, "loss": 0.1706, "step": 5299 }, { "epoch": 32.92211838006231, "grad_norm": 2.109375, "learning_rate": 4.705260299867169e-05, "loss": 0.1569, "step": 5300 }, { "epoch": 32.92834890965732, "grad_norm": 2.5625, "learning_rate": 4.70514423252932e-05, "loss": 0.1718, "step": 5301 }, { "epoch": 32.93457943925234, "grad_norm": 2.09375, "learning_rate": 4.70502814377467e-05, "loss": 0.2654, "step": 5302 }, { "epoch": 32.940809968847354, "grad_norm": 3.125, "learning_rate": 4.704912033604346e-05, "loss": 0.2086, "step": 5303 }, { "epoch": 32.94704049844237, "grad_norm": 3.8125, "learning_rate": 4.704795902019475e-05, "loss": 0.2172, "step": 5304 }, { "epoch": 32.953271028037385, "grad_norm": 3.078125, "learning_rate": 4.7046797490211865e-05, "loss": 0.291, "step": 5305 }, { "epoch": 32.9595015576324, "grad_norm": 2.359375, "learning_rate": 4.704563574610606e-05, "loss": 0.2049, "step": 5306 }, { "epoch": 32.965732087227416, "grad_norm": 4.53125, "learning_rate": 4.7044473787888654e-05, "loss": 0.3794, "step": 5307 }, { "epoch": 32.97196261682243, "grad_norm": 3.03125, "learning_rate": 4.7043311615570895e-05, "loss": 0.2318, "step": 5308 }, { "epoch": 32.978193146417446, "grad_norm": 2.859375, "learning_rate": 4.7042149229164095e-05, "loss": 0.1745, "step": 5309 }, { "epoch": 32.98442367601246, "grad_norm": 2.609375, "learning_rate": 4.7040986628679537e-05, "loss": 0.1685, "step": 5310 }, { "epoch": 32.99065420560748, "grad_norm": 2.734375, "learning_rate": 4.70398238141285e-05, "loss": 0.2309, "step": 5311 }, { "epoch": 32.99688473520249, "grad_norm": 3.140625, "learning_rate": 4.703866078552231e-05, "loss": 0.1687, "step": 5312 }, { "epoch": 33.0, "grad_norm": 1.890625, "learning_rate": 4.7037497542872215e-05, "loss": 0.0924, "step": 5313 }, { "epoch": 33.006230529595015, "grad_norm": 3.890625, "learning_rate": 4.7036334086189555e-05, "loss": 0.3491, "step": 5314 }, { "epoch": 33.01246105919003, "grad_norm": 2.734375, "learning_rate": 4.703517041548561e-05, "loss": 0.1873, "step": 5315 }, { "epoch": 33.018691588785046, "grad_norm": 3.296875, "learning_rate": 4.703400653077168e-05, "loss": 0.2767, "step": 5316 }, { "epoch": 33.02492211838006, "grad_norm": 2.609375, "learning_rate": 4.7032842432059075e-05, "loss": 0.3601, "step": 5317 }, { "epoch": 33.03115264797508, "grad_norm": 3.109375, "learning_rate": 4.7031678119359104e-05, "loss": 0.4086, "step": 5318 }, { "epoch": 33.03738317757009, "grad_norm": 1.7421875, "learning_rate": 4.7030513592683074e-05, "loss": 0.1404, "step": 5319 }, { "epoch": 33.04361370716511, "grad_norm": 3.296875, "learning_rate": 4.7029348852042285e-05, "loss": 0.4118, "step": 5320 }, { "epoch": 33.04984423676012, "grad_norm": 1.53125, "learning_rate": 4.702818389744806e-05, "loss": 0.135, "step": 5321 }, { "epoch": 33.05607476635514, "grad_norm": 2.625, "learning_rate": 4.7027018728911714e-05, "loss": 0.1306, "step": 5322 }, { "epoch": 33.06230529595015, "grad_norm": 4.21875, "learning_rate": 4.702585334644455e-05, "loss": 0.4496, "step": 5323 }, { "epoch": 33.06853582554517, "grad_norm": 2.65625, "learning_rate": 4.702468775005791e-05, "loss": 0.2583, "step": 5324 }, { "epoch": 33.074766355140184, "grad_norm": 2.125, "learning_rate": 4.7023521939763084e-05, "loss": 0.1684, "step": 5325 }, { "epoch": 33.0809968847352, "grad_norm": 1.9921875, "learning_rate": 4.7022355915571414e-05, "loss": 0.2018, "step": 5326 }, { "epoch": 33.087227414330215, "grad_norm": 3.265625, "learning_rate": 4.702118967749423e-05, "loss": 0.2718, "step": 5327 }, { "epoch": 33.09345794392523, "grad_norm": 4.28125, "learning_rate": 4.702002322554284e-05, "loss": 0.2231, "step": 5328 }, { "epoch": 33.099688473520246, "grad_norm": 2.671875, "learning_rate": 4.7018856559728585e-05, "loss": 0.2278, "step": 5329 }, { "epoch": 33.10591900311527, "grad_norm": 2.46875, "learning_rate": 4.7017689680062796e-05, "loss": 0.1898, "step": 5330 }, { "epoch": 33.11214953271028, "grad_norm": 1.828125, "learning_rate": 4.70165225865568e-05, "loss": 0.1597, "step": 5331 }, { "epoch": 33.1183800623053, "grad_norm": 3.46875, "learning_rate": 4.701535527922194e-05, "loss": 0.321, "step": 5332 }, { "epoch": 33.124610591900314, "grad_norm": 2.015625, "learning_rate": 4.701418775806955e-05, "loss": 0.1487, "step": 5333 }, { "epoch": 33.13084112149533, "grad_norm": 2.171875, "learning_rate": 4.701302002311096e-05, "loss": 0.1828, "step": 5334 }, { "epoch": 33.137071651090345, "grad_norm": 3.4375, "learning_rate": 4.7011852074357524e-05, "loss": 0.3464, "step": 5335 }, { "epoch": 33.14330218068536, "grad_norm": 3.921875, "learning_rate": 4.701068391182059e-05, "loss": 0.2837, "step": 5336 }, { "epoch": 33.149532710280376, "grad_norm": 2.6875, "learning_rate": 4.700951553551148e-05, "loss": 0.2378, "step": 5337 }, { "epoch": 33.15576323987539, "grad_norm": 2.890625, "learning_rate": 4.7008346945441555e-05, "loss": 0.3439, "step": 5338 }, { "epoch": 33.161993769470406, "grad_norm": 2.1875, "learning_rate": 4.700717814162218e-05, "loss": 0.198, "step": 5339 }, { "epoch": 33.16822429906542, "grad_norm": 3.53125, "learning_rate": 4.700600912406467e-05, "loss": 0.2228, "step": 5340 }, { "epoch": 33.17445482866044, "grad_norm": 2.671875, "learning_rate": 4.700483989278042e-05, "loss": 0.1886, "step": 5341 }, { "epoch": 33.18068535825545, "grad_norm": 1.8046875, "learning_rate": 4.700367044778076e-05, "loss": 0.1554, "step": 5342 }, { "epoch": 33.18691588785047, "grad_norm": 3.84375, "learning_rate": 4.7002500789077045e-05, "loss": 0.4717, "step": 5343 }, { "epoch": 33.19314641744548, "grad_norm": 3.125, "learning_rate": 4.7001330916680654e-05, "loss": 0.2391, "step": 5344 }, { "epoch": 33.1993769470405, "grad_norm": 2.328125, "learning_rate": 4.7000160830602936e-05, "loss": 0.216, "step": 5345 }, { "epoch": 33.205607476635514, "grad_norm": 2.078125, "learning_rate": 4.699899053085526e-05, "loss": 0.1516, "step": 5346 }, { "epoch": 33.21183800623053, "grad_norm": 3.328125, "learning_rate": 4.699782001744899e-05, "loss": 0.2622, "step": 5347 }, { "epoch": 33.218068535825545, "grad_norm": 2.046875, "learning_rate": 4.69966492903955e-05, "loss": 0.1716, "step": 5348 }, { "epoch": 33.22429906542056, "grad_norm": 2.578125, "learning_rate": 4.6995478349706145e-05, "loss": 0.2527, "step": 5349 }, { "epoch": 33.230529595015575, "grad_norm": 2.234375, "learning_rate": 4.699430719539231e-05, "loss": 0.1723, "step": 5350 }, { "epoch": 33.23676012461059, "grad_norm": 2.375, "learning_rate": 4.699313582746537e-05, "loss": 0.1792, "step": 5351 }, { "epoch": 33.242990654205606, "grad_norm": 2.953125, "learning_rate": 4.6991964245936695e-05, "loss": 0.2793, "step": 5352 }, { "epoch": 33.24922118380062, "grad_norm": 2.671875, "learning_rate": 4.6990792450817676e-05, "loss": 0.1615, "step": 5353 }, { "epoch": 33.25545171339564, "grad_norm": 1.3984375, "learning_rate": 4.698962044211968e-05, "loss": 0.1236, "step": 5354 }, { "epoch": 33.26168224299065, "grad_norm": 1.9765625, "learning_rate": 4.6988448219854086e-05, "loss": 0.1936, "step": 5355 }, { "epoch": 33.26791277258567, "grad_norm": 1.671875, "learning_rate": 4.69872757840323e-05, "loss": 0.1284, "step": 5356 }, { "epoch": 33.27414330218068, "grad_norm": 2.25, "learning_rate": 4.6986103134665684e-05, "loss": 0.1847, "step": 5357 }, { "epoch": 33.2803738317757, "grad_norm": 2.546875, "learning_rate": 4.698493027176565e-05, "loss": 0.2143, "step": 5358 }, { "epoch": 33.28660436137071, "grad_norm": 3.109375, "learning_rate": 4.698375719534358e-05, "loss": 0.231, "step": 5359 }, { "epoch": 33.29283489096573, "grad_norm": 1.6328125, "learning_rate": 4.698258390541086e-05, "loss": 0.1627, "step": 5360 }, { "epoch": 33.299065420560744, "grad_norm": 2.34375, "learning_rate": 4.698141040197889e-05, "loss": 0.1296, "step": 5361 }, { "epoch": 33.30529595015577, "grad_norm": 1.953125, "learning_rate": 4.6980236685059065e-05, "loss": 0.1686, "step": 5362 }, { "epoch": 33.31152647975078, "grad_norm": 1.7578125, "learning_rate": 4.697906275466279e-05, "loss": 0.1535, "step": 5363 }, { "epoch": 33.3177570093458, "grad_norm": 1.8203125, "learning_rate": 4.6977888610801466e-05, "loss": 0.1578, "step": 5364 }, { "epoch": 33.32398753894081, "grad_norm": 2.5, "learning_rate": 4.69767142534865e-05, "loss": 0.3714, "step": 5365 }, { "epoch": 33.33021806853583, "grad_norm": 2.734375, "learning_rate": 4.697553968272929e-05, "loss": 0.3043, "step": 5366 }, { "epoch": 33.33644859813084, "grad_norm": 2.171875, "learning_rate": 4.697436489854124e-05, "loss": 0.1835, "step": 5367 }, { "epoch": 33.34267912772586, "grad_norm": 2.875, "learning_rate": 4.697318990093377e-05, "loss": 0.4234, "step": 5368 }, { "epoch": 33.348909657320874, "grad_norm": 1.859375, "learning_rate": 4.6972014689918285e-05, "loss": 0.1594, "step": 5369 }, { "epoch": 33.35514018691589, "grad_norm": 1.5859375, "learning_rate": 4.69708392655062e-05, "loss": 0.1354, "step": 5370 }, { "epoch": 33.361370716510905, "grad_norm": 2.5, "learning_rate": 4.6969663627708935e-05, "loss": 0.2407, "step": 5371 }, { "epoch": 33.36760124610592, "grad_norm": 2.671875, "learning_rate": 4.6968487776537906e-05, "loss": 0.2098, "step": 5372 }, { "epoch": 33.373831775700936, "grad_norm": 2.125, "learning_rate": 4.696731171200452e-05, "loss": 0.1878, "step": 5373 }, { "epoch": 33.38006230529595, "grad_norm": 2.71875, "learning_rate": 4.696613543412023e-05, "loss": 0.1406, "step": 5374 }, { "epoch": 33.386292834890966, "grad_norm": 2.671875, "learning_rate": 4.696495894289643e-05, "loss": 0.1897, "step": 5375 }, { "epoch": 33.39252336448598, "grad_norm": 3.59375, "learning_rate": 4.6963782238344565e-05, "loss": 0.2908, "step": 5376 }, { "epoch": 33.398753894081, "grad_norm": 3.296875, "learning_rate": 4.696260532047605e-05, "loss": 0.299, "step": 5377 }, { "epoch": 33.40498442367601, "grad_norm": 2.0625, "learning_rate": 4.696142818930233e-05, "loss": 0.1431, "step": 5378 }, { "epoch": 33.41121495327103, "grad_norm": 2.203125, "learning_rate": 4.696025084483482e-05, "loss": 0.2016, "step": 5379 }, { "epoch": 33.41744548286604, "grad_norm": 3.015625, "learning_rate": 4.6959073287084974e-05, "loss": 0.1489, "step": 5380 }, { "epoch": 33.42367601246106, "grad_norm": 1.84375, "learning_rate": 4.695789551606421e-05, "loss": 0.1245, "step": 5381 }, { "epoch": 33.429906542056074, "grad_norm": 2.484375, "learning_rate": 4.695671753178398e-05, "loss": 0.1817, "step": 5382 }, { "epoch": 33.43613707165109, "grad_norm": 1.0234375, "learning_rate": 4.6955539334255716e-05, "loss": 0.1051, "step": 5383 }, { "epoch": 33.442367601246104, "grad_norm": 3.734375, "learning_rate": 4.695436092349087e-05, "loss": 0.2754, "step": 5384 }, { "epoch": 33.44859813084112, "grad_norm": 2.296875, "learning_rate": 4.6953182299500876e-05, "loss": 0.1733, "step": 5385 }, { "epoch": 33.454828660436135, "grad_norm": 1.9765625, "learning_rate": 4.6952003462297195e-05, "loss": 0.1423, "step": 5386 }, { "epoch": 33.46105919003115, "grad_norm": 2.765625, "learning_rate": 4.695082441189126e-05, "loss": 0.1947, "step": 5387 }, { "epoch": 33.467289719626166, "grad_norm": 2.875, "learning_rate": 4.694964514829454e-05, "loss": 0.2291, "step": 5388 }, { "epoch": 33.47352024922118, "grad_norm": 3.484375, "learning_rate": 4.694846567151847e-05, "loss": 0.2789, "step": 5389 }, { "epoch": 33.4797507788162, "grad_norm": 3.78125, "learning_rate": 4.694728598157452e-05, "loss": 0.2731, "step": 5390 }, { "epoch": 33.48598130841121, "grad_norm": 2.40625, "learning_rate": 4.694610607847414e-05, "loss": 0.1932, "step": 5391 }, { "epoch": 33.49221183800623, "grad_norm": 2.53125, "learning_rate": 4.6944925962228784e-05, "loss": 0.1697, "step": 5392 }, { "epoch": 33.49844236760124, "grad_norm": 3.546875, "learning_rate": 4.694374563284993e-05, "loss": 0.491, "step": 5393 }, { "epoch": 33.504672897196265, "grad_norm": 2.40625, "learning_rate": 4.6942565090349024e-05, "loss": 0.1956, "step": 5394 }, { "epoch": 33.51090342679128, "grad_norm": 1.7421875, "learning_rate": 4.694138433473755e-05, "loss": 0.1691, "step": 5395 }, { "epoch": 33.517133956386296, "grad_norm": 1.859375, "learning_rate": 4.6940203366026955e-05, "loss": 0.1646, "step": 5396 }, { "epoch": 33.52336448598131, "grad_norm": 2.25, "learning_rate": 4.693902218422872e-05, "loss": 0.149, "step": 5397 }, { "epoch": 33.52959501557633, "grad_norm": 3.21875, "learning_rate": 4.6937840789354326e-05, "loss": 0.2441, "step": 5398 }, { "epoch": 33.53582554517134, "grad_norm": 2.65625, "learning_rate": 4.693665918141523e-05, "loss": 0.1742, "step": 5399 }, { "epoch": 33.54205607476636, "grad_norm": 2.609375, "learning_rate": 4.6935477360422916e-05, "loss": 0.26, "step": 5400 }, { "epoch": 33.54828660436137, "grad_norm": 3.328125, "learning_rate": 4.6934295326388854e-05, "loss": 0.2151, "step": 5401 }, { "epoch": 33.55451713395639, "grad_norm": 2.9375, "learning_rate": 4.693311307932454e-05, "loss": 0.3104, "step": 5402 }, { "epoch": 33.5607476635514, "grad_norm": 3.25, "learning_rate": 4.6931930619241446e-05, "loss": 0.2734, "step": 5403 }, { "epoch": 33.56697819314642, "grad_norm": 2.390625, "learning_rate": 4.693074794615105e-05, "loss": 0.2566, "step": 5404 }, { "epoch": 33.573208722741434, "grad_norm": 3.25, "learning_rate": 4.6929565060064864e-05, "loss": 0.2753, "step": 5405 }, { "epoch": 33.57943925233645, "grad_norm": 2.140625, "learning_rate": 4.6928381960994336e-05, "loss": 0.192, "step": 5406 }, { "epoch": 33.585669781931465, "grad_norm": 2.203125, "learning_rate": 4.6927198648951e-05, "loss": 0.2211, "step": 5407 }, { "epoch": 33.59190031152648, "grad_norm": 3.25, "learning_rate": 4.692601512394631e-05, "loss": 0.1829, "step": 5408 }, { "epoch": 33.598130841121495, "grad_norm": 2.859375, "learning_rate": 4.692483138599178e-05, "loss": 0.1743, "step": 5409 }, { "epoch": 33.60436137071651, "grad_norm": 2.71875, "learning_rate": 4.692364743509892e-05, "loss": 0.1916, "step": 5410 }, { "epoch": 33.610591900311526, "grad_norm": 2.59375, "learning_rate": 4.6922463271279204e-05, "loss": 0.2434, "step": 5411 }, { "epoch": 33.61682242990654, "grad_norm": 1.9453125, "learning_rate": 4.692127889454414e-05, "loss": 0.2046, "step": 5412 }, { "epoch": 33.62305295950156, "grad_norm": 2.65625, "learning_rate": 4.692009430490524e-05, "loss": 0.3397, "step": 5413 }, { "epoch": 33.62928348909657, "grad_norm": 2.3125, "learning_rate": 4.6918909502373994e-05, "loss": 0.2623, "step": 5414 }, { "epoch": 33.63551401869159, "grad_norm": 2.171875, "learning_rate": 4.691772448696193e-05, "loss": 0.2042, "step": 5415 }, { "epoch": 33.6417445482866, "grad_norm": 3.125, "learning_rate": 4.691653925868054e-05, "loss": 0.2541, "step": 5416 }, { "epoch": 33.64797507788162, "grad_norm": 3.171875, "learning_rate": 4.691535381754133e-05, "loss": 0.1345, "step": 5417 }, { "epoch": 33.654205607476634, "grad_norm": 1.984375, "learning_rate": 4.6914168163555826e-05, "loss": 0.1634, "step": 5418 }, { "epoch": 33.66043613707165, "grad_norm": 2.734375, "learning_rate": 4.6912982296735544e-05, "loss": 0.204, "step": 5419 }, { "epoch": 33.666666666666664, "grad_norm": 2.828125, "learning_rate": 4.6911796217092e-05, "loss": 0.2203, "step": 5420 }, { "epoch": 33.67289719626168, "grad_norm": 2.765625, "learning_rate": 4.6910609924636706e-05, "loss": 0.1893, "step": 5421 }, { "epoch": 33.679127725856695, "grad_norm": 2.125, "learning_rate": 4.6909423419381196e-05, "loss": 0.157, "step": 5422 }, { "epoch": 33.68535825545171, "grad_norm": 3.640625, "learning_rate": 4.6908236701336976e-05, "loss": 0.3169, "step": 5423 }, { "epoch": 33.691588785046726, "grad_norm": 2.859375, "learning_rate": 4.6907049770515586e-05, "loss": 0.278, "step": 5424 }, { "epoch": 33.69781931464174, "grad_norm": 3.859375, "learning_rate": 4.690586262692855e-05, "loss": 0.246, "step": 5425 }, { "epoch": 33.70404984423676, "grad_norm": 2.1875, "learning_rate": 4.69046752705874e-05, "loss": 0.1939, "step": 5426 }, { "epoch": 33.71028037383178, "grad_norm": 1.484375, "learning_rate": 4.690348770150366e-05, "loss": 0.1255, "step": 5427 }, { "epoch": 33.716510903426794, "grad_norm": 2.875, "learning_rate": 4.6902299919688876e-05, "loss": 0.1943, "step": 5428 }, { "epoch": 33.72274143302181, "grad_norm": 2.703125, "learning_rate": 4.690111192515457e-05, "loss": 0.3092, "step": 5429 }, { "epoch": 33.728971962616825, "grad_norm": 2.640625, "learning_rate": 4.6899923717912286e-05, "loss": 0.144, "step": 5430 }, { "epoch": 33.73520249221184, "grad_norm": 2.640625, "learning_rate": 4.6898735297973575e-05, "loss": 0.2372, "step": 5431 }, { "epoch": 33.741433021806856, "grad_norm": 2.078125, "learning_rate": 4.689754666534996e-05, "loss": 0.254, "step": 5432 }, { "epoch": 33.74766355140187, "grad_norm": 3.25, "learning_rate": 4.689635782005299e-05, "loss": 0.3061, "step": 5433 }, { "epoch": 33.753894080996886, "grad_norm": 2.546875, "learning_rate": 4.6895168762094235e-05, "loss": 0.2845, "step": 5434 }, { "epoch": 33.7601246105919, "grad_norm": 3.0625, "learning_rate": 4.689397949148521e-05, "loss": 0.1375, "step": 5435 }, { "epoch": 33.76635514018692, "grad_norm": 2.875, "learning_rate": 4.689279000823748e-05, "loss": 0.1304, "step": 5436 }, { "epoch": 33.77258566978193, "grad_norm": 2.875, "learning_rate": 4.689160031236261e-05, "loss": 0.1683, "step": 5437 }, { "epoch": 33.77881619937695, "grad_norm": 2.890625, "learning_rate": 4.689041040387212e-05, "loss": 0.2349, "step": 5438 }, { "epoch": 33.78504672897196, "grad_norm": 3.484375, "learning_rate": 4.68892202827776e-05, "loss": 0.1595, "step": 5439 }, { "epoch": 33.79127725856698, "grad_norm": 2.671875, "learning_rate": 4.68880299490906e-05, "loss": 0.227, "step": 5440 }, { "epoch": 33.797507788161994, "grad_norm": 1.671875, "learning_rate": 4.688683940282267e-05, "loss": 0.1454, "step": 5441 }, { "epoch": 33.80373831775701, "grad_norm": 3.25, "learning_rate": 4.688564864398539e-05, "loss": 0.3275, "step": 5442 }, { "epoch": 33.809968847352025, "grad_norm": 3.828125, "learning_rate": 4.688445767259031e-05, "loss": 0.1898, "step": 5443 }, { "epoch": 33.81619937694704, "grad_norm": 1.6484375, "learning_rate": 4.6883266488649004e-05, "loss": 0.1615, "step": 5444 }, { "epoch": 33.822429906542055, "grad_norm": 3.125, "learning_rate": 4.6882075092173036e-05, "loss": 0.1789, "step": 5445 }, { "epoch": 33.82866043613707, "grad_norm": 2.921875, "learning_rate": 4.688088348317399e-05, "loss": 0.2069, "step": 5446 }, { "epoch": 33.834890965732086, "grad_norm": 1.421875, "learning_rate": 4.687969166166342e-05, "loss": 0.1179, "step": 5447 }, { "epoch": 33.8411214953271, "grad_norm": 2.984375, "learning_rate": 4.687849962765291e-05, "loss": 0.2942, "step": 5448 }, { "epoch": 33.84735202492212, "grad_norm": 2.984375, "learning_rate": 4.6877307381154044e-05, "loss": 0.1553, "step": 5449 }, { "epoch": 33.85358255451713, "grad_norm": 4.3125, "learning_rate": 4.687611492217839e-05, "loss": 0.2831, "step": 5450 }, { "epoch": 33.85981308411215, "grad_norm": 4.46875, "learning_rate": 4.687492225073754e-05, "loss": 0.1817, "step": 5451 }, { "epoch": 33.86604361370716, "grad_norm": 3.65625, "learning_rate": 4.6873729366843075e-05, "loss": 0.3686, "step": 5452 }, { "epoch": 33.87227414330218, "grad_norm": 2.671875, "learning_rate": 4.687253627050657e-05, "loss": 0.2603, "step": 5453 }, { "epoch": 33.87850467289719, "grad_norm": 3.578125, "learning_rate": 4.687134296173963e-05, "loss": 0.2104, "step": 5454 }, { "epoch": 33.88473520249221, "grad_norm": 2.25, "learning_rate": 4.687014944055382e-05, "loss": 0.1575, "step": 5455 }, { "epoch": 33.890965732087224, "grad_norm": 2.984375, "learning_rate": 4.686895570696076e-05, "loss": 0.2034, "step": 5456 }, { "epoch": 33.89719626168224, "grad_norm": 1.796875, "learning_rate": 4.686776176097203e-05, "loss": 0.1629, "step": 5457 }, { "epoch": 33.90342679127726, "grad_norm": 2.34375, "learning_rate": 4.686656760259922e-05, "loss": 0.2684, "step": 5458 }, { "epoch": 33.90965732087228, "grad_norm": 2.65625, "learning_rate": 4.686537323185394e-05, "loss": 0.1901, "step": 5459 }, { "epoch": 33.91588785046729, "grad_norm": 4.09375, "learning_rate": 4.686417864874778e-05, "loss": 0.3578, "step": 5460 }, { "epoch": 33.92211838006231, "grad_norm": 2.03125, "learning_rate": 4.686298385329235e-05, "loss": 0.1271, "step": 5461 }, { "epoch": 33.92834890965732, "grad_norm": 3.03125, "learning_rate": 4.686178884549925e-05, "loss": 0.3397, "step": 5462 }, { "epoch": 33.93457943925234, "grad_norm": 2.78125, "learning_rate": 4.686059362538008e-05, "loss": 0.1904, "step": 5463 }, { "epoch": 33.940809968847354, "grad_norm": 3.5625, "learning_rate": 4.685939819294647e-05, "loss": 0.2509, "step": 5464 }, { "epoch": 33.94704049844237, "grad_norm": 3.34375, "learning_rate": 4.685820254821e-05, "loss": 0.2917, "step": 5465 }, { "epoch": 33.953271028037385, "grad_norm": 1.7109375, "learning_rate": 4.6857006691182305e-05, "loss": 0.1109, "step": 5466 }, { "epoch": 33.9595015576324, "grad_norm": 3.015625, "learning_rate": 4.685581062187498e-05, "loss": 0.1987, "step": 5467 }, { "epoch": 33.965732087227416, "grad_norm": 3.328125, "learning_rate": 4.6854614340299676e-05, "loss": 0.242, "step": 5468 }, { "epoch": 33.97196261682243, "grad_norm": 2.46875, "learning_rate": 4.685341784646797e-05, "loss": 0.1585, "step": 5469 }, { "epoch": 33.978193146417446, "grad_norm": 2.734375, "learning_rate": 4.6852221140391506e-05, "loss": 0.1622, "step": 5470 }, { "epoch": 33.98442367601246, "grad_norm": 2.546875, "learning_rate": 4.685102422208191e-05, "loss": 0.2087, "step": 5471 }, { "epoch": 33.99065420560748, "grad_norm": 2.0625, "learning_rate": 4.684982709155079e-05, "loss": 0.1774, "step": 5472 }, { "epoch": 33.99688473520249, "grad_norm": 2.046875, "learning_rate": 4.6848629748809784e-05, "loss": 0.1371, "step": 5473 }, { "epoch": 34.0, "grad_norm": 1.1171875, "learning_rate": 4.684743219387052e-05, "loss": 0.0868, "step": 5474 }, { "epoch": 34.006230529595015, "grad_norm": 2.625, "learning_rate": 4.684623442674463e-05, "loss": 0.1707, "step": 5475 }, { "epoch": 34.01246105919003, "grad_norm": 3.03125, "learning_rate": 4.684503644744373e-05, "loss": 0.2609, "step": 5476 }, { "epoch": 34.018691588785046, "grad_norm": 2.921875, "learning_rate": 4.684383825597949e-05, "loss": 0.2685, "step": 5477 }, { "epoch": 34.02492211838006, "grad_norm": 1.8125, "learning_rate": 4.684263985236351e-05, "loss": 0.157, "step": 5478 }, { "epoch": 34.03115264797508, "grad_norm": 3.65625, "learning_rate": 4.684144123660745e-05, "loss": 0.3836, "step": 5479 }, { "epoch": 34.03738317757009, "grad_norm": 3.203125, "learning_rate": 4.6840242408722956e-05, "loss": 0.3093, "step": 5480 }, { "epoch": 34.04361370716511, "grad_norm": 3.6875, "learning_rate": 4.683904336872165e-05, "loss": 0.2413, "step": 5481 }, { "epoch": 34.04984423676012, "grad_norm": 2.015625, "learning_rate": 4.6837844116615194e-05, "loss": 0.1454, "step": 5482 }, { "epoch": 34.05607476635514, "grad_norm": 2.71875, "learning_rate": 4.683664465241523e-05, "loss": 0.1491, "step": 5483 }, { "epoch": 34.06230529595015, "grad_norm": 2.671875, "learning_rate": 4.683544497613341e-05, "loss": 0.2079, "step": 5484 }, { "epoch": 34.06853582554517, "grad_norm": 3.140625, "learning_rate": 4.683424508778138e-05, "loss": 0.2612, "step": 5485 }, { "epoch": 34.074766355140184, "grad_norm": 2.21875, "learning_rate": 4.68330449873708e-05, "loss": 0.1738, "step": 5486 }, { "epoch": 34.0809968847352, "grad_norm": 3.25, "learning_rate": 4.683184467491332e-05, "loss": 0.4629, "step": 5487 }, { "epoch": 34.087227414330215, "grad_norm": 2.3125, "learning_rate": 4.68306441504206e-05, "loss": 0.1816, "step": 5488 }, { "epoch": 34.09345794392523, "grad_norm": 2.234375, "learning_rate": 4.68294434139043e-05, "loss": 0.1677, "step": 5489 }, { "epoch": 34.099688473520246, "grad_norm": 2.734375, "learning_rate": 4.6828242465376085e-05, "loss": 0.1485, "step": 5490 }, { "epoch": 34.10591900311527, "grad_norm": 1.6171875, "learning_rate": 4.682704130484762e-05, "loss": 0.1449, "step": 5491 }, { "epoch": 34.11214953271028, "grad_norm": 3.375, "learning_rate": 4.682583993233056e-05, "loss": 0.2927, "step": 5492 }, { "epoch": 34.1183800623053, "grad_norm": 1.09375, "learning_rate": 4.682463834783658e-05, "loss": 0.114, "step": 5493 }, { "epoch": 34.124610591900314, "grad_norm": 1.5859375, "learning_rate": 4.682343655137735e-05, "loss": 0.1554, "step": 5494 }, { "epoch": 34.13084112149533, "grad_norm": 2.5625, "learning_rate": 4.6822234542964545e-05, "loss": 0.1903, "step": 5495 }, { "epoch": 34.137071651090345, "grad_norm": 1.328125, "learning_rate": 4.682103232260984e-05, "loss": 0.1307, "step": 5496 }, { "epoch": 34.14330218068536, "grad_norm": 2.734375, "learning_rate": 4.68198298903249e-05, "loss": 0.2009, "step": 5497 }, { "epoch": 34.149532710280376, "grad_norm": 1.9140625, "learning_rate": 4.681862724612141e-05, "loss": 0.1769, "step": 5498 }, { "epoch": 34.15576323987539, "grad_norm": 1.71875, "learning_rate": 4.681742439001105e-05, "loss": 0.1826, "step": 5499 }, { "epoch": 34.161993769470406, "grad_norm": 1.484375, "learning_rate": 4.68162213220055e-05, "loss": 0.1167, "step": 5500 }, { "epoch": 34.16822429906542, "grad_norm": 2.140625, "learning_rate": 4.681501804211645e-05, "loss": 0.2119, "step": 5501 }, { "epoch": 34.17445482866044, "grad_norm": 2.3125, "learning_rate": 4.681381455035559e-05, "loss": 0.1675, "step": 5502 }, { "epoch": 34.18068535825545, "grad_norm": 2.8125, "learning_rate": 4.68126108467346e-05, "loss": 0.2602, "step": 5503 }, { "epoch": 34.18691588785047, "grad_norm": 2.53125, "learning_rate": 4.681140693126518e-05, "loss": 0.1654, "step": 5504 }, { "epoch": 34.19314641744548, "grad_norm": 3.15625, "learning_rate": 4.6810202803959004e-05, "loss": 0.2515, "step": 5505 }, { "epoch": 34.1993769470405, "grad_norm": 2.609375, "learning_rate": 4.680899846482778e-05, "loss": 0.1984, "step": 5506 }, { "epoch": 34.205607476635514, "grad_norm": 2.015625, "learning_rate": 4.680779391388321e-05, "loss": 0.1427, "step": 5507 }, { "epoch": 34.21183800623053, "grad_norm": 1.984375, "learning_rate": 4.680658915113698e-05, "loss": 0.1586, "step": 5508 }, { "epoch": 34.218068535825545, "grad_norm": 2.703125, "learning_rate": 4.68053841766008e-05, "loss": 0.1552, "step": 5509 }, { "epoch": 34.22429906542056, "grad_norm": 3.234375, "learning_rate": 4.680417899028637e-05, "loss": 0.2298, "step": 5510 }, { "epoch": 34.230529595015575, "grad_norm": 1.390625, "learning_rate": 4.6802973592205394e-05, "loss": 0.1096, "step": 5511 }, { "epoch": 34.23676012461059, "grad_norm": 2.53125, "learning_rate": 4.680176798236958e-05, "loss": 0.2452, "step": 5512 }, { "epoch": 34.242990654205606, "grad_norm": 2.03125, "learning_rate": 4.680056216079064e-05, "loss": 0.1725, "step": 5513 }, { "epoch": 34.24922118380062, "grad_norm": 3.0, "learning_rate": 4.679935612748028e-05, "loss": 0.2374, "step": 5514 }, { "epoch": 34.25545171339564, "grad_norm": 2.140625, "learning_rate": 4.6798149882450216e-05, "loss": 0.209, "step": 5515 }, { "epoch": 34.26168224299065, "grad_norm": 2.78125, "learning_rate": 4.679694342571216e-05, "loss": 0.1591, "step": 5516 }, { "epoch": 34.26791277258567, "grad_norm": 2.390625, "learning_rate": 4.679573675727784e-05, "loss": 0.214, "step": 5517 }, { "epoch": 34.27414330218068, "grad_norm": 3.34375, "learning_rate": 4.679452987715897e-05, "loss": 0.2489, "step": 5518 }, { "epoch": 34.2803738317757, "grad_norm": 3.96875, "learning_rate": 4.679332278536725e-05, "loss": 0.2529, "step": 5519 }, { "epoch": 34.28660436137071, "grad_norm": 3.640625, "learning_rate": 4.6792115481914445e-05, "loss": 0.2189, "step": 5520 }, { "epoch": 34.29283489096573, "grad_norm": 2.5625, "learning_rate": 4.679090796681225e-05, "loss": 0.1512, "step": 5521 }, { "epoch": 34.299065420560744, "grad_norm": 2.328125, "learning_rate": 4.67897002400724e-05, "loss": 0.1574, "step": 5522 }, { "epoch": 34.30529595015577, "grad_norm": 2.21875, "learning_rate": 4.678849230170662e-05, "loss": 0.2312, "step": 5523 }, { "epoch": 34.31152647975078, "grad_norm": 2.09375, "learning_rate": 4.678728415172665e-05, "loss": 0.1566, "step": 5524 }, { "epoch": 34.3177570093458, "grad_norm": 2.984375, "learning_rate": 4.678607579014422e-05, "loss": 0.1761, "step": 5525 }, { "epoch": 34.32398753894081, "grad_norm": 1.5390625, "learning_rate": 4.678486721697107e-05, "loss": 0.1248, "step": 5526 }, { "epoch": 34.33021806853583, "grad_norm": 3.71875, "learning_rate": 4.678365843221894e-05, "loss": 0.3024, "step": 5527 }, { "epoch": 34.33644859813084, "grad_norm": 2.515625, "learning_rate": 4.678244943589956e-05, "loss": 0.1755, "step": 5528 }, { "epoch": 34.34267912772586, "grad_norm": 2.859375, "learning_rate": 4.678124022802468e-05, "loss": 0.1832, "step": 5529 }, { "epoch": 34.348909657320874, "grad_norm": 3.265625, "learning_rate": 4.6780030808606035e-05, "loss": 0.2864, "step": 5530 }, { "epoch": 34.35514018691589, "grad_norm": 3.109375, "learning_rate": 4.677882117765539e-05, "loss": 0.1547, "step": 5531 }, { "epoch": 34.361370716510905, "grad_norm": 3.921875, "learning_rate": 4.6777611335184464e-05, "loss": 0.2739, "step": 5532 }, { "epoch": 34.36760124610592, "grad_norm": 2.8125, "learning_rate": 4.677640128120503e-05, "loss": 0.1828, "step": 5533 }, { "epoch": 34.373831775700936, "grad_norm": 1.8359375, "learning_rate": 4.6775191015728834e-05, "loss": 0.1211, "step": 5534 }, { "epoch": 34.38006230529595, "grad_norm": 2.484375, "learning_rate": 4.677398053876763e-05, "loss": 0.1882, "step": 5535 }, { "epoch": 34.386292834890966, "grad_norm": 1.921875, "learning_rate": 4.677276985033317e-05, "loss": 0.2567, "step": 5536 }, { "epoch": 34.39252336448598, "grad_norm": 2.5625, "learning_rate": 4.677155895043722e-05, "loss": 0.2382, "step": 5537 }, { "epoch": 34.398753894081, "grad_norm": 1.8984375, "learning_rate": 4.677034783909154e-05, "loss": 0.1941, "step": 5538 }, { "epoch": 34.40498442367601, "grad_norm": 2.125, "learning_rate": 4.676913651630789e-05, "loss": 0.1904, "step": 5539 }, { "epoch": 34.41121495327103, "grad_norm": 2.078125, "learning_rate": 4.676792498209803e-05, "loss": 0.1548, "step": 5540 }, { "epoch": 34.41744548286604, "grad_norm": 2.28125, "learning_rate": 4.676671323647373e-05, "loss": 0.1908, "step": 5541 }, { "epoch": 34.42367601246106, "grad_norm": 1.8046875, "learning_rate": 4.6765501279446775e-05, "loss": 0.1496, "step": 5542 }, { "epoch": 34.429906542056074, "grad_norm": 1.53125, "learning_rate": 4.67642891110289e-05, "loss": 0.1283, "step": 5543 }, { "epoch": 34.43613707165109, "grad_norm": 2.703125, "learning_rate": 4.6763076731231916e-05, "loss": 0.2315, "step": 5544 }, { "epoch": 34.442367601246104, "grad_norm": 2.796875, "learning_rate": 4.676186414006757e-05, "loss": 0.2719, "step": 5545 }, { "epoch": 34.44859813084112, "grad_norm": 3.046875, "learning_rate": 4.676065133754766e-05, "loss": 0.2957, "step": 5546 }, { "epoch": 34.454828660436135, "grad_norm": 3.234375, "learning_rate": 4.675943832368394e-05, "loss": 0.2158, "step": 5547 }, { "epoch": 34.46105919003115, "grad_norm": 3.09375, "learning_rate": 4.675822509848821e-05, "loss": 0.154, "step": 5548 }, { "epoch": 34.467289719626166, "grad_norm": 2.625, "learning_rate": 4.675701166197225e-05, "loss": 0.1925, "step": 5549 }, { "epoch": 34.47352024922118, "grad_norm": 2.96875, "learning_rate": 4.675579801414785e-05, "loss": 0.2322, "step": 5550 }, { "epoch": 34.4797507788162, "grad_norm": 2.078125, "learning_rate": 4.675458415502678e-05, "loss": 0.1308, "step": 5551 }, { "epoch": 34.48598130841121, "grad_norm": 3.015625, "learning_rate": 4.6753370084620844e-05, "loss": 0.2054, "step": 5552 }, { "epoch": 34.49221183800623, "grad_norm": 2.671875, "learning_rate": 4.675215580294183e-05, "loss": 0.1504, "step": 5553 }, { "epoch": 34.49844236760124, "grad_norm": 2.84375, "learning_rate": 4.6750941310001535e-05, "loss": 0.2174, "step": 5554 }, { "epoch": 34.504672897196265, "grad_norm": 1.375, "learning_rate": 4.674972660581174e-05, "loss": 0.1285, "step": 5555 }, { "epoch": 34.51090342679128, "grad_norm": 3.265625, "learning_rate": 4.6748511690384265e-05, "loss": 0.3434, "step": 5556 }, { "epoch": 34.517133956386296, "grad_norm": 3.0, "learning_rate": 4.6747296563730894e-05, "loss": 0.1933, "step": 5557 }, { "epoch": 34.52336448598131, "grad_norm": 2.0, "learning_rate": 4.674608122586343e-05, "loss": 0.1594, "step": 5558 }, { "epoch": 34.52959501557633, "grad_norm": 1.6171875, "learning_rate": 4.674486567679367e-05, "loss": 0.1146, "step": 5559 }, { "epoch": 34.53582554517134, "grad_norm": 1.828125, "learning_rate": 4.6743649916533435e-05, "loss": 0.1777, "step": 5560 }, { "epoch": 34.54205607476636, "grad_norm": 2.515625, "learning_rate": 4.674243394509452e-05, "loss": 0.2767, "step": 5561 }, { "epoch": 34.54828660436137, "grad_norm": 2.21875, "learning_rate": 4.674121776248875e-05, "loss": 0.1732, "step": 5562 }, { "epoch": 34.55451713395639, "grad_norm": 2.046875, "learning_rate": 4.674000136872792e-05, "loss": 0.1296, "step": 5563 }, { "epoch": 34.5607476635514, "grad_norm": 2.296875, "learning_rate": 4.673878476382385e-05, "loss": 0.169, "step": 5564 }, { "epoch": 34.56697819314642, "grad_norm": 3.0, "learning_rate": 4.6737567947788364e-05, "loss": 0.2192, "step": 5565 }, { "epoch": 34.573208722741434, "grad_norm": 2.453125, "learning_rate": 4.6736350920633265e-05, "loss": 0.1819, "step": 5566 }, { "epoch": 34.57943925233645, "grad_norm": 3.046875, "learning_rate": 4.673513368237039e-05, "loss": 0.3079, "step": 5567 }, { "epoch": 34.585669781931465, "grad_norm": 3.265625, "learning_rate": 4.673391623301154e-05, "loss": 0.1767, "step": 5568 }, { "epoch": 34.59190031152648, "grad_norm": 2.796875, "learning_rate": 4.6732698572568565e-05, "loss": 0.1448, "step": 5569 }, { "epoch": 34.598130841121495, "grad_norm": 3.015625, "learning_rate": 4.6731480701053266e-05, "loss": 0.2259, "step": 5570 }, { "epoch": 34.60436137071651, "grad_norm": 3.421875, "learning_rate": 4.673026261847748e-05, "loss": 0.3535, "step": 5571 }, { "epoch": 34.610591900311526, "grad_norm": 3.859375, "learning_rate": 4.672904432485305e-05, "loss": 0.3305, "step": 5572 }, { "epoch": 34.61682242990654, "grad_norm": 3.703125, "learning_rate": 4.672782582019179e-05, "loss": 0.1422, "step": 5573 }, { "epoch": 34.62305295950156, "grad_norm": 2.546875, "learning_rate": 4.672660710450555e-05, "loss": 0.1777, "step": 5574 }, { "epoch": 34.62928348909657, "grad_norm": 2.828125, "learning_rate": 4.672538817780615e-05, "loss": 0.2261, "step": 5575 }, { "epoch": 34.63551401869159, "grad_norm": 3.0, "learning_rate": 4.672416904010545e-05, "loss": 0.2415, "step": 5576 }, { "epoch": 34.6417445482866, "grad_norm": 3.390625, "learning_rate": 4.672294969141527e-05, "loss": 0.2387, "step": 5577 }, { "epoch": 34.64797507788162, "grad_norm": 3.609375, "learning_rate": 4.6721730131747456e-05, "loss": 0.2292, "step": 5578 }, { "epoch": 34.654205607476634, "grad_norm": 2.4375, "learning_rate": 4.6720510361113864e-05, "loss": 0.2679, "step": 5579 }, { "epoch": 34.66043613707165, "grad_norm": 3.03125, "learning_rate": 4.6719290379526323e-05, "loss": 0.2896, "step": 5580 }, { "epoch": 34.666666666666664, "grad_norm": 1.875, "learning_rate": 4.6718070186996706e-05, "loss": 0.1525, "step": 5581 }, { "epoch": 34.67289719626168, "grad_norm": 3.21875, "learning_rate": 4.6716849783536844e-05, "loss": 0.1837, "step": 5582 }, { "epoch": 34.679127725856695, "grad_norm": 4.3125, "learning_rate": 4.6715629169158595e-05, "loss": 0.2018, "step": 5583 }, { "epoch": 34.68535825545171, "grad_norm": 3.40625, "learning_rate": 4.671440834387382e-05, "loss": 0.1752, "step": 5584 }, { "epoch": 34.691588785046726, "grad_norm": 2.4375, "learning_rate": 4.6713187307694364e-05, "loss": 0.2492, "step": 5585 }, { "epoch": 34.69781931464174, "grad_norm": 2.296875, "learning_rate": 4.671196606063209e-05, "loss": 0.1952, "step": 5586 }, { "epoch": 34.70404984423676, "grad_norm": 3.6875, "learning_rate": 4.671074460269887e-05, "loss": 0.1634, "step": 5587 }, { "epoch": 34.71028037383178, "grad_norm": 1.8203125, "learning_rate": 4.670952293390655e-05, "loss": 0.1355, "step": 5588 }, { "epoch": 34.716510903426794, "grad_norm": 2.046875, "learning_rate": 4.6708301054267006e-05, "loss": 0.1065, "step": 5589 }, { "epoch": 34.72274143302181, "grad_norm": 2.09375, "learning_rate": 4.670707896379211e-05, "loss": 0.1201, "step": 5590 }, { "epoch": 34.728971962616825, "grad_norm": 2.625, "learning_rate": 4.670585666249371e-05, "loss": 0.2098, "step": 5591 }, { "epoch": 34.73520249221184, "grad_norm": 3.15625, "learning_rate": 4.67046341503837e-05, "loss": 0.1425, "step": 5592 }, { "epoch": 34.741433021806856, "grad_norm": 2.421875, "learning_rate": 4.670341142747395e-05, "loss": 0.1918, "step": 5593 }, { "epoch": 34.74766355140187, "grad_norm": 2.390625, "learning_rate": 4.670218849377632e-05, "loss": 0.1934, "step": 5594 }, { "epoch": 34.753894080996886, "grad_norm": 1.1328125, "learning_rate": 4.67009653493027e-05, "loss": 0.1095, "step": 5595 }, { "epoch": 34.7601246105919, "grad_norm": 2.5625, "learning_rate": 4.669974199406497e-05, "loss": 0.1993, "step": 5596 }, { "epoch": 34.76635514018692, "grad_norm": 3.171875, "learning_rate": 4.6698518428075004e-05, "loss": 0.2864, "step": 5597 }, { "epoch": 34.77258566978193, "grad_norm": 2.171875, "learning_rate": 4.669729465134469e-05, "loss": 0.1473, "step": 5598 }, { "epoch": 34.77881619937695, "grad_norm": 3.015625, "learning_rate": 4.669607066388592e-05, "loss": 0.1947, "step": 5599 }, { "epoch": 34.78504672897196, "grad_norm": 3.796875, "learning_rate": 4.669484646571057e-05, "loss": 0.1911, "step": 5600 }, { "epoch": 34.79127725856698, "grad_norm": 1.9296875, "learning_rate": 4.6693622056830534e-05, "loss": 0.1775, "step": 5601 }, { "epoch": 34.797507788161994, "grad_norm": 2.640625, "learning_rate": 4.66923974372577e-05, "loss": 0.2002, "step": 5602 }, { "epoch": 34.80373831775701, "grad_norm": 1.84375, "learning_rate": 4.669117260700397e-05, "loss": 0.2227, "step": 5603 }, { "epoch": 34.809968847352025, "grad_norm": 2.828125, "learning_rate": 4.6689947566081236e-05, "loss": 0.2011, "step": 5604 }, { "epoch": 34.81619937694704, "grad_norm": 2.3125, "learning_rate": 4.6688722314501395e-05, "loss": 0.1998, "step": 5605 }, { "epoch": 34.822429906542055, "grad_norm": 2.21875, "learning_rate": 4.6687496852276344e-05, "loss": 0.149, "step": 5606 }, { "epoch": 34.82866043613707, "grad_norm": 2.421875, "learning_rate": 4.6686271179418e-05, "loss": 0.2163, "step": 5607 }, { "epoch": 34.834890965732086, "grad_norm": 4.125, "learning_rate": 4.668504529593824e-05, "loss": 0.406, "step": 5608 }, { "epoch": 34.8411214953271, "grad_norm": 2.140625, "learning_rate": 4.668381920184899e-05, "loss": 0.1866, "step": 5609 }, { "epoch": 34.84735202492212, "grad_norm": 2.953125, "learning_rate": 4.668259289716216e-05, "loss": 0.2679, "step": 5610 }, { "epoch": 34.85358255451713, "grad_norm": 2.25, "learning_rate": 4.6681366381889646e-05, "loss": 0.1665, "step": 5611 }, { "epoch": 34.85981308411215, "grad_norm": 1.6796875, "learning_rate": 4.668013965604337e-05, "loss": 0.1354, "step": 5612 }, { "epoch": 34.86604361370716, "grad_norm": 3.21875, "learning_rate": 4.6678912719635246e-05, "loss": 0.1773, "step": 5613 }, { "epoch": 34.87227414330218, "grad_norm": 3.59375, "learning_rate": 4.667768557267719e-05, "loss": 0.3359, "step": 5614 }, { "epoch": 34.87850467289719, "grad_norm": 1.90625, "learning_rate": 4.667645821518111e-05, "loss": 0.1511, "step": 5615 }, { "epoch": 34.88473520249221, "grad_norm": 3.109375, "learning_rate": 4.667523064715894e-05, "loss": 0.2474, "step": 5616 }, { "epoch": 34.890965732087224, "grad_norm": 2.53125, "learning_rate": 4.66740028686226e-05, "loss": 0.1893, "step": 5617 }, { "epoch": 34.89719626168224, "grad_norm": 2.328125, "learning_rate": 4.6672774879584004e-05, "loss": 0.2181, "step": 5618 }, { "epoch": 34.90342679127726, "grad_norm": 2.703125, "learning_rate": 4.6671546680055094e-05, "loss": 0.2131, "step": 5619 }, { "epoch": 34.90965732087228, "grad_norm": 1.484375, "learning_rate": 4.667031827004779e-05, "loss": 0.121, "step": 5620 }, { "epoch": 34.91588785046729, "grad_norm": 2.0625, "learning_rate": 4.6669089649574016e-05, "loss": 0.2219, "step": 5621 }, { "epoch": 34.92211838006231, "grad_norm": 4.03125, "learning_rate": 4.6667860818645714e-05, "loss": 0.3349, "step": 5622 }, { "epoch": 34.92834890965732, "grad_norm": 3.28125, "learning_rate": 4.666663177727482e-05, "loss": 0.1139, "step": 5623 }, { "epoch": 34.93457943925234, "grad_norm": 2.625, "learning_rate": 4.666540252547327e-05, "loss": 0.2201, "step": 5624 }, { "epoch": 34.940809968847354, "grad_norm": 2.390625, "learning_rate": 4.666417306325299e-05, "loss": 0.1628, "step": 5625 }, { "epoch": 34.94704049844237, "grad_norm": 4.25, "learning_rate": 4.6662943390625935e-05, "loss": 0.2896, "step": 5626 }, { "epoch": 34.953271028037385, "grad_norm": 1.921875, "learning_rate": 4.6661713507604046e-05, "loss": 0.1947, "step": 5627 }, { "epoch": 34.9595015576324, "grad_norm": 2.484375, "learning_rate": 4.6660483414199265e-05, "loss": 0.262, "step": 5628 }, { "epoch": 34.965732087227416, "grad_norm": 3.671875, "learning_rate": 4.665925311042354e-05, "loss": 0.1611, "step": 5629 }, { "epoch": 34.97196261682243, "grad_norm": 3.984375, "learning_rate": 4.665802259628882e-05, "loss": 0.3566, "step": 5630 }, { "epoch": 34.978193146417446, "grad_norm": 2.78125, "learning_rate": 4.665679187180705e-05, "loss": 0.1253, "step": 5631 }, { "epoch": 34.98442367601246, "grad_norm": 3.390625, "learning_rate": 4.6655560936990196e-05, "loss": 0.2186, "step": 5632 }, { "epoch": 34.99065420560748, "grad_norm": 2.4375, "learning_rate": 4.665432979185019e-05, "loss": 0.1646, "step": 5633 }, { "epoch": 34.99688473520249, "grad_norm": 4.4375, "learning_rate": 4.665309843639902e-05, "loss": 0.2897, "step": 5634 }, { "epoch": 35.0, "grad_norm": 2.578125, "learning_rate": 4.6651866870648626e-05, "loss": 0.1194, "step": 5635 }, { "epoch": 35.006230529595015, "grad_norm": 3.0, "learning_rate": 4.665063509461097e-05, "loss": 0.1226, "step": 5636 }, { "epoch": 35.01246105919003, "grad_norm": 3.53125, "learning_rate": 4.664940310829802e-05, "loss": 0.2173, "step": 5637 }, { "epoch": 35.018691588785046, "grad_norm": 1.5703125, "learning_rate": 4.664817091172173e-05, "loss": 0.1063, "step": 5638 }, { "epoch": 35.02492211838006, "grad_norm": 4.46875, "learning_rate": 4.6646938504894086e-05, "loss": 0.1579, "step": 5639 }, { "epoch": 35.03115264797508, "grad_norm": 5.375, "learning_rate": 4.664570588782705e-05, "loss": 0.1966, "step": 5640 }, { "epoch": 35.03738317757009, "grad_norm": 6.28125, "learning_rate": 4.664447306053259e-05, "loss": 0.1275, "step": 5641 }, { "epoch": 35.04361370716511, "grad_norm": 5.28125, "learning_rate": 4.6643240023022684e-05, "loss": 0.1692, "step": 5642 }, { "epoch": 35.04984423676012, "grad_norm": 3.546875, "learning_rate": 4.66420067753093e-05, "loss": 0.2082, "step": 5643 }, { "epoch": 35.05607476635514, "grad_norm": 4.84375, "learning_rate": 4.6640773317404416e-05, "loss": 0.1793, "step": 5644 }, { "epoch": 35.06230529595015, "grad_norm": 6.03125, "learning_rate": 4.663953964932002e-05, "loss": 0.1161, "step": 5645 }, { "epoch": 35.06853582554517, "grad_norm": 6.71875, "learning_rate": 4.663830577106809e-05, "loss": 0.1939, "step": 5646 }, { "epoch": 35.074766355140184, "grad_norm": 3.21875, "learning_rate": 4.663707168266061e-05, "loss": 0.1802, "step": 5647 }, { "epoch": 35.0809968847352, "grad_norm": 2.78125, "learning_rate": 4.663583738410956e-05, "loss": 0.1899, "step": 5648 }, { "epoch": 35.087227414330215, "grad_norm": 3.234375, "learning_rate": 4.6634602875426936e-05, "loss": 0.3087, "step": 5649 }, { "epoch": 35.09345794392523, "grad_norm": 3.453125, "learning_rate": 4.6633368156624726e-05, "loss": 0.1175, "step": 5650 }, { "epoch": 35.099688473520246, "grad_norm": 2.71875, "learning_rate": 4.6632133227714915e-05, "loss": 0.1364, "step": 5651 }, { "epoch": 35.10591900311527, "grad_norm": 2.328125, "learning_rate": 4.6630898088709505e-05, "loss": 0.1723, "step": 5652 }, { "epoch": 35.11214953271028, "grad_norm": 2.953125, "learning_rate": 4.662966273962049e-05, "loss": 0.2068, "step": 5653 }, { "epoch": 35.1183800623053, "grad_norm": 1.65625, "learning_rate": 4.6628427180459864e-05, "loss": 0.165, "step": 5654 }, { "epoch": 35.124610591900314, "grad_norm": 2.5, "learning_rate": 4.6627191411239634e-05, "loss": 0.1327, "step": 5655 }, { "epoch": 35.13084112149533, "grad_norm": 3.28125, "learning_rate": 4.662595543197179e-05, "loss": 0.295, "step": 5656 }, { "epoch": 35.137071651090345, "grad_norm": 2.40625, "learning_rate": 4.662471924266835e-05, "loss": 0.2001, "step": 5657 }, { "epoch": 35.14330218068536, "grad_norm": 2.84375, "learning_rate": 4.6623482843341316e-05, "loss": 0.324, "step": 5658 }, { "epoch": 35.149532710280376, "grad_norm": 2.46875, "learning_rate": 4.6622246234002686e-05, "loss": 0.2308, "step": 5659 }, { "epoch": 35.15576323987539, "grad_norm": 3.34375, "learning_rate": 4.662100941466448e-05, "loss": 0.2361, "step": 5660 }, { "epoch": 35.161993769470406, "grad_norm": 2.0625, "learning_rate": 4.6619772385338714e-05, "loss": 0.1613, "step": 5661 }, { "epoch": 35.16822429906542, "grad_norm": 2.71875, "learning_rate": 4.66185351460374e-05, "loss": 0.2165, "step": 5662 }, { "epoch": 35.17445482866044, "grad_norm": 2.546875, "learning_rate": 4.6617297696772545e-05, "loss": 0.1935, "step": 5663 }, { "epoch": 35.18068535825545, "grad_norm": 1.6171875, "learning_rate": 4.6616060037556173e-05, "loss": 0.1744, "step": 5664 }, { "epoch": 35.18691588785047, "grad_norm": 2.609375, "learning_rate": 4.661482216840031e-05, "loss": 0.1911, "step": 5665 }, { "epoch": 35.19314641744548, "grad_norm": 1.734375, "learning_rate": 4.661358408931696e-05, "loss": 0.1185, "step": 5666 }, { "epoch": 35.1993769470405, "grad_norm": 3.734375, "learning_rate": 4.661234580031818e-05, "loss": 0.3243, "step": 5667 }, { "epoch": 35.205607476635514, "grad_norm": 2.359375, "learning_rate": 4.661110730141597e-05, "loss": 0.1697, "step": 5668 }, { "epoch": 35.21183800623053, "grad_norm": 3.734375, "learning_rate": 4.660986859262236e-05, "loss": 0.2525, "step": 5669 }, { "epoch": 35.218068535825545, "grad_norm": 4.65625, "learning_rate": 4.660862967394939e-05, "loss": 0.1881, "step": 5670 }, { "epoch": 35.22429906542056, "grad_norm": 2.46875, "learning_rate": 4.660739054540909e-05, "loss": 0.1233, "step": 5671 }, { "epoch": 35.230529595015575, "grad_norm": 3.078125, "learning_rate": 4.660615120701349e-05, "loss": 0.2193, "step": 5672 }, { "epoch": 35.23676012461059, "grad_norm": 3.265625, "learning_rate": 4.660491165877464e-05, "loss": 0.3748, "step": 5673 }, { "epoch": 35.242990654205606, "grad_norm": 1.4765625, "learning_rate": 4.660367190070456e-05, "loss": 0.1186, "step": 5674 }, { "epoch": 35.24922118380062, "grad_norm": 2.953125, "learning_rate": 4.6602431932815305e-05, "loss": 0.2173, "step": 5675 }, { "epoch": 35.25545171339564, "grad_norm": 3.421875, "learning_rate": 4.6601191755118914e-05, "loss": 0.3376, "step": 5676 }, { "epoch": 35.26168224299065, "grad_norm": 2.53125, "learning_rate": 4.659995136762742e-05, "loss": 0.1293, "step": 5677 }, { "epoch": 35.26791277258567, "grad_norm": 2.46875, "learning_rate": 4.6598710770352897e-05, "loss": 0.1584, "step": 5678 }, { "epoch": 35.27414330218068, "grad_norm": 1.234375, "learning_rate": 4.659746996330736e-05, "loss": 0.1387, "step": 5679 }, { "epoch": 35.2803738317757, "grad_norm": 3.21875, "learning_rate": 4.659622894650288e-05, "loss": 0.1977, "step": 5680 }, { "epoch": 35.28660436137071, "grad_norm": 3.578125, "learning_rate": 4.6594987719951525e-05, "loss": 0.2523, "step": 5681 }, { "epoch": 35.29283489096573, "grad_norm": 3.828125, "learning_rate": 4.659374628366532e-05, "loss": 0.225, "step": 5682 }, { "epoch": 35.299065420560744, "grad_norm": 1.3359375, "learning_rate": 4.659250463765633e-05, "loss": 0.1065, "step": 5683 }, { "epoch": 35.30529595015577, "grad_norm": 1.46875, "learning_rate": 4.659126278193663e-05, "loss": 0.141, "step": 5684 }, { "epoch": 35.31152647975078, "grad_norm": 2.8125, "learning_rate": 4.659002071651826e-05, "loss": 0.1275, "step": 5685 }, { "epoch": 35.3177570093458, "grad_norm": 2.828125, "learning_rate": 4.65887784414133e-05, "loss": 0.1964, "step": 5686 }, { "epoch": 35.32398753894081, "grad_norm": 1.796875, "learning_rate": 4.65875359566338e-05, "loss": 0.1238, "step": 5687 }, { "epoch": 35.33021806853583, "grad_norm": 3.5, "learning_rate": 4.658629326219185e-05, "loss": 0.3957, "step": 5688 }, { "epoch": 35.33644859813084, "grad_norm": 2.546875, "learning_rate": 4.6585050358099494e-05, "loss": 0.217, "step": 5689 }, { "epoch": 35.34267912772586, "grad_norm": 2.609375, "learning_rate": 4.6583807244368814e-05, "loss": 0.1705, "step": 5690 }, { "epoch": 35.348909657320874, "grad_norm": 3.96875, "learning_rate": 4.658256392101189e-05, "loss": 0.3227, "step": 5691 }, { "epoch": 35.35514018691589, "grad_norm": 2.484375, "learning_rate": 4.6581320388040784e-05, "loss": 0.291, "step": 5692 }, { "epoch": 35.361370716510905, "grad_norm": 2.34375, "learning_rate": 4.6580076645467586e-05, "loss": 0.172, "step": 5693 }, { "epoch": 35.36760124610592, "grad_norm": 1.9453125, "learning_rate": 4.6578832693304365e-05, "loss": 0.1583, "step": 5694 }, { "epoch": 35.373831775700936, "grad_norm": 3.359375, "learning_rate": 4.657758853156321e-05, "loss": 0.2306, "step": 5695 }, { "epoch": 35.38006230529595, "grad_norm": 3.28125, "learning_rate": 4.65763441602562e-05, "loss": 0.2214, "step": 5696 }, { "epoch": 35.386292834890966, "grad_norm": 2.484375, "learning_rate": 4.657509957939542e-05, "loss": 0.1408, "step": 5697 }, { "epoch": 35.39252336448598, "grad_norm": 3.046875, "learning_rate": 4.657385478899297e-05, "loss": 0.1521, "step": 5698 }, { "epoch": 35.398753894081, "grad_norm": 2.21875, "learning_rate": 4.6572609789060914e-05, "loss": 0.1393, "step": 5699 }, { "epoch": 35.40498442367601, "grad_norm": 1.765625, "learning_rate": 4.657136457961138e-05, "loss": 0.1435, "step": 5700 }, { "epoch": 35.41121495327103, "grad_norm": 1.6953125, "learning_rate": 4.657011916065642e-05, "loss": 0.1392, "step": 5701 }, { "epoch": 35.41744548286604, "grad_norm": 2.921875, "learning_rate": 4.656887353220816e-05, "loss": 0.2085, "step": 5702 }, { "epoch": 35.42367601246106, "grad_norm": 2.515625, "learning_rate": 4.656762769427869e-05, "loss": 0.1424, "step": 5703 }, { "epoch": 35.429906542056074, "grad_norm": 2.25, "learning_rate": 4.656638164688011e-05, "loss": 0.1413, "step": 5704 }, { "epoch": 35.43613707165109, "grad_norm": 3.0625, "learning_rate": 4.6565135390024515e-05, "loss": 0.2329, "step": 5705 }, { "epoch": 35.442367601246104, "grad_norm": 2.296875, "learning_rate": 4.6563888923724026e-05, "loss": 0.1378, "step": 5706 }, { "epoch": 35.44859813084112, "grad_norm": 4.15625, "learning_rate": 4.656264224799073e-05, "loss": 0.3638, "step": 5707 }, { "epoch": 35.454828660436135, "grad_norm": 2.421875, "learning_rate": 4.656139536283674e-05, "loss": 0.1934, "step": 5708 }, { "epoch": 35.46105919003115, "grad_norm": 3.8125, "learning_rate": 4.6560148268274165e-05, "loss": 0.3368, "step": 5709 }, { "epoch": 35.467289719626166, "grad_norm": 3.21875, "learning_rate": 4.655890096431513e-05, "loss": 0.1807, "step": 5710 }, { "epoch": 35.47352024922118, "grad_norm": 1.2578125, "learning_rate": 4.6557653450971726e-05, "loss": 0.1133, "step": 5711 }, { "epoch": 35.4797507788162, "grad_norm": 2.84375, "learning_rate": 4.655640572825609e-05, "loss": 0.2062, "step": 5712 }, { "epoch": 35.48598130841121, "grad_norm": 3.609375, "learning_rate": 4.6555157796180335e-05, "loss": 0.2225, "step": 5713 }, { "epoch": 35.49221183800623, "grad_norm": 2.015625, "learning_rate": 4.6553909654756576e-05, "loss": 0.1196, "step": 5714 }, { "epoch": 35.49844236760124, "grad_norm": 3.015625, "learning_rate": 4.6552661303996946e-05, "loss": 0.2893, "step": 5715 }, { "epoch": 35.504672897196265, "grad_norm": 2.0, "learning_rate": 4.6551412743913546e-05, "loss": 0.1719, "step": 5716 }, { "epoch": 35.51090342679128, "grad_norm": 4.09375, "learning_rate": 4.6550163974518534e-05, "loss": 0.4588, "step": 5717 }, { "epoch": 35.517133956386296, "grad_norm": 1.9296875, "learning_rate": 4.654891499582401e-05, "loss": 0.1238, "step": 5718 }, { "epoch": 35.52336448598131, "grad_norm": 2.03125, "learning_rate": 4.6547665807842114e-05, "loss": 0.1258, "step": 5719 }, { "epoch": 35.52959501557633, "grad_norm": 1.8671875, "learning_rate": 4.6546416410584994e-05, "loss": 0.1374, "step": 5720 }, { "epoch": 35.53582554517134, "grad_norm": 2.578125, "learning_rate": 4.654516680406476e-05, "loss": 0.1113, "step": 5721 }, { "epoch": 35.54205607476636, "grad_norm": 4.40625, "learning_rate": 4.654391698829356e-05, "loss": 0.2513, "step": 5722 }, { "epoch": 35.54828660436137, "grad_norm": 4.40625, "learning_rate": 4.6542666963283535e-05, "loss": 0.3177, "step": 5723 }, { "epoch": 35.55451713395639, "grad_norm": 3.78125, "learning_rate": 4.654141672904682e-05, "loss": 0.2989, "step": 5724 }, { "epoch": 35.5607476635514, "grad_norm": 3.765625, "learning_rate": 4.6540166285595555e-05, "loss": 0.3837, "step": 5725 }, { "epoch": 35.56697819314642, "grad_norm": 3.625, "learning_rate": 4.653891563294189e-05, "loss": 0.2108, "step": 5726 }, { "epoch": 35.573208722741434, "grad_norm": 3.6875, "learning_rate": 4.6537664771097977e-05, "loss": 0.135, "step": 5727 }, { "epoch": 35.57943925233645, "grad_norm": 2.515625, "learning_rate": 4.653641370007596e-05, "loss": 0.1346, "step": 5728 }, { "epoch": 35.585669781931465, "grad_norm": 1.6015625, "learning_rate": 4.6535162419887985e-05, "loss": 0.1162, "step": 5729 }, { "epoch": 35.59190031152648, "grad_norm": 2.375, "learning_rate": 4.653391093054621e-05, "loss": 0.2024, "step": 5730 }, { "epoch": 35.598130841121495, "grad_norm": 4.4375, "learning_rate": 4.653265923206278e-05, "loss": 0.1315, "step": 5731 }, { "epoch": 35.60436137071651, "grad_norm": 5.28125, "learning_rate": 4.6531407324449864e-05, "loss": 0.2834, "step": 5732 }, { "epoch": 35.610591900311526, "grad_norm": 4.5625, "learning_rate": 4.653015520771962e-05, "loss": 0.2437, "step": 5733 }, { "epoch": 35.61682242990654, "grad_norm": 2.59375, "learning_rate": 4.652890288188419e-05, "loss": 0.2187, "step": 5734 }, { "epoch": 35.62305295950156, "grad_norm": 1.140625, "learning_rate": 4.652765034695576e-05, "loss": 0.1111, "step": 5735 }, { "epoch": 35.62928348909657, "grad_norm": 2.21875, "learning_rate": 4.65263976029465e-05, "loss": 0.1272, "step": 5736 }, { "epoch": 35.63551401869159, "grad_norm": 4.15625, "learning_rate": 4.6525144649868546e-05, "loss": 0.2191, "step": 5737 }, { "epoch": 35.6417445482866, "grad_norm": 3.140625, "learning_rate": 4.652389148773409e-05, "loss": 0.135, "step": 5738 }, { "epoch": 35.64797507788162, "grad_norm": 3.15625, "learning_rate": 4.652263811655531e-05, "loss": 0.129, "step": 5739 }, { "epoch": 35.654205607476634, "grad_norm": 2.59375, "learning_rate": 4.6521384536344346e-05, "loss": 0.1847, "step": 5740 }, { "epoch": 35.66043613707165, "grad_norm": 4.46875, "learning_rate": 4.65201307471134e-05, "loss": 0.219, "step": 5741 }, { "epoch": 35.666666666666664, "grad_norm": 4.28125, "learning_rate": 4.651887674887464e-05, "loss": 0.2656, "step": 5742 }, { "epoch": 35.67289719626168, "grad_norm": 4.5625, "learning_rate": 4.651762254164026e-05, "loss": 0.3272, "step": 5743 }, { "epoch": 35.679127725856695, "grad_norm": 3.21875, "learning_rate": 4.651636812542242e-05, "loss": 0.3238, "step": 5744 }, { "epoch": 35.68535825545171, "grad_norm": 2.4375, "learning_rate": 4.651511350023331e-05, "loss": 0.2207, "step": 5745 }, { "epoch": 35.691588785046726, "grad_norm": 2.765625, "learning_rate": 4.651385866608512e-05, "loss": 0.1517, "step": 5746 }, { "epoch": 35.69781931464174, "grad_norm": 2.140625, "learning_rate": 4.651260362299003e-05, "loss": 0.1392, "step": 5747 }, { "epoch": 35.70404984423676, "grad_norm": 2.0625, "learning_rate": 4.651134837096024e-05, "loss": 0.1936, "step": 5748 }, { "epoch": 35.71028037383178, "grad_norm": 2.0, "learning_rate": 4.651009291000793e-05, "loss": 0.1398, "step": 5749 }, { "epoch": 35.716510903426794, "grad_norm": 1.609375, "learning_rate": 4.65088372401453e-05, "loss": 0.1151, "step": 5750 }, { "epoch": 35.72274143302181, "grad_norm": 3.40625, "learning_rate": 4.6507581361384537e-05, "loss": 0.2426, "step": 5751 }, { "epoch": 35.728971962616825, "grad_norm": 2.296875, "learning_rate": 4.650632527373785e-05, "loss": 0.1203, "step": 5752 }, { "epoch": 35.73520249221184, "grad_norm": 3.71875, "learning_rate": 4.6505068977217434e-05, "loss": 0.2103, "step": 5753 }, { "epoch": 35.741433021806856, "grad_norm": 1.9140625, "learning_rate": 4.650381247183549e-05, "loss": 0.1344, "step": 5754 }, { "epoch": 35.74766355140187, "grad_norm": 2.953125, "learning_rate": 4.650255575760421e-05, "loss": 0.2342, "step": 5755 }, { "epoch": 35.753894080996886, "grad_norm": 1.84375, "learning_rate": 4.6501298834535826e-05, "loss": 0.1607, "step": 5756 }, { "epoch": 35.7601246105919, "grad_norm": 2.421875, "learning_rate": 4.6500041702642525e-05, "loss": 0.1919, "step": 5757 }, { "epoch": 35.76635514018692, "grad_norm": 2.859375, "learning_rate": 4.649878436193651e-05, "loss": 0.1606, "step": 5758 }, { "epoch": 35.77258566978193, "grad_norm": 1.5625, "learning_rate": 4.6497526812430016e-05, "loss": 0.1582, "step": 5759 }, { "epoch": 35.77881619937695, "grad_norm": 2.59375, "learning_rate": 4.6496269054135244e-05, "loss": 0.2892, "step": 5760 }, { "epoch": 35.78504672897196, "grad_norm": 2.640625, "learning_rate": 4.649501108706441e-05, "loss": 0.2185, "step": 5761 }, { "epoch": 35.79127725856698, "grad_norm": 3.671875, "learning_rate": 4.6493752911229724e-05, "loss": 0.2613, "step": 5762 }, { "epoch": 35.797507788161994, "grad_norm": 2.90625, "learning_rate": 4.649249452664341e-05, "loss": 0.2305, "step": 5763 }, { "epoch": 35.80373831775701, "grad_norm": 2.765625, "learning_rate": 4.649123593331771e-05, "loss": 0.1568, "step": 5764 }, { "epoch": 35.809968847352025, "grad_norm": 2.421875, "learning_rate": 4.648997713126482e-05, "loss": 0.1504, "step": 5765 }, { "epoch": 35.81619937694704, "grad_norm": 3.40625, "learning_rate": 4.648871812049698e-05, "loss": 0.3384, "step": 5766 }, { "epoch": 35.822429906542055, "grad_norm": 2.546875, "learning_rate": 4.64874589010264e-05, "loss": 0.1695, "step": 5767 }, { "epoch": 35.82866043613707, "grad_norm": 2.796875, "learning_rate": 4.6486199472865344e-05, "loss": 0.2299, "step": 5768 }, { "epoch": 35.834890965732086, "grad_norm": 3.15625, "learning_rate": 4.648493983602601e-05, "loss": 0.1215, "step": 5769 }, { "epoch": 35.8411214953271, "grad_norm": 2.34375, "learning_rate": 4.648367999052065e-05, "loss": 0.1554, "step": 5770 }, { "epoch": 35.84735202492212, "grad_norm": 1.5546875, "learning_rate": 4.64824199363615e-05, "loss": 0.1373, "step": 5771 }, { "epoch": 35.85358255451713, "grad_norm": 3.078125, "learning_rate": 4.648115967356079e-05, "loss": 0.2594, "step": 5772 }, { "epoch": 35.85981308411215, "grad_norm": 2.8125, "learning_rate": 4.647989920213076e-05, "loss": 0.2095, "step": 5773 }, { "epoch": 35.86604361370716, "grad_norm": 2.8125, "learning_rate": 4.6478638522083654e-05, "loss": 0.2611, "step": 5774 }, { "epoch": 35.87227414330218, "grad_norm": 2.640625, "learning_rate": 4.647737763343172e-05, "loss": 0.2104, "step": 5775 }, { "epoch": 35.87850467289719, "grad_norm": 2.375, "learning_rate": 4.647611653618721e-05, "loss": 0.1586, "step": 5776 }, { "epoch": 35.88473520249221, "grad_norm": 2.796875, "learning_rate": 4.6474855230362355e-05, "loss": 0.1332, "step": 5777 }, { "epoch": 35.890965732087224, "grad_norm": 2.59375, "learning_rate": 4.647359371596941e-05, "loss": 0.179, "step": 5778 }, { "epoch": 35.89719626168224, "grad_norm": 1.9609375, "learning_rate": 4.647233199302064e-05, "loss": 0.175, "step": 5779 }, { "epoch": 35.90342679127726, "grad_norm": 1.296875, "learning_rate": 4.647107006152828e-05, "loss": 0.1134, "step": 5780 }, { "epoch": 35.90965732087228, "grad_norm": 2.25, "learning_rate": 4.646980792150459e-05, "loss": 0.1477, "step": 5781 }, { "epoch": 35.91588785046729, "grad_norm": 2.515625, "learning_rate": 4.6468545572961844e-05, "loss": 0.2126, "step": 5782 }, { "epoch": 35.92211838006231, "grad_norm": 1.5859375, "learning_rate": 4.6467283015912295e-05, "loss": 0.1459, "step": 5783 }, { "epoch": 35.92834890965732, "grad_norm": 3.140625, "learning_rate": 4.6466020250368194e-05, "loss": 0.2685, "step": 5784 }, { "epoch": 35.93457943925234, "grad_norm": 2.0625, "learning_rate": 4.646475727634182e-05, "loss": 0.1771, "step": 5785 }, { "epoch": 35.940809968847354, "grad_norm": 2.5, "learning_rate": 4.6463494093845427e-05, "loss": 0.2125, "step": 5786 }, { "epoch": 35.94704049844237, "grad_norm": 2.984375, "learning_rate": 4.6462230702891297e-05, "loss": 0.2064, "step": 5787 }, { "epoch": 35.953271028037385, "grad_norm": 2.671875, "learning_rate": 4.646096710349168e-05, "loss": 0.1549, "step": 5788 }, { "epoch": 35.9595015576324, "grad_norm": 2.96875, "learning_rate": 4.6459703295658876e-05, "loss": 0.3156, "step": 5789 }, { "epoch": 35.965732087227416, "grad_norm": 2.796875, "learning_rate": 4.645843927940513e-05, "loss": 0.1974, "step": 5790 }, { "epoch": 35.97196261682243, "grad_norm": 2.140625, "learning_rate": 4.645717505474274e-05, "loss": 0.1702, "step": 5791 }, { "epoch": 35.978193146417446, "grad_norm": 1.3515625, "learning_rate": 4.6455910621683976e-05, "loss": 0.1392, "step": 5792 }, { "epoch": 35.98442367601246, "grad_norm": 4.03125, "learning_rate": 4.645464598024111e-05, "loss": 0.2819, "step": 5793 }, { "epoch": 35.99065420560748, "grad_norm": 2.5, "learning_rate": 4.6453381130426444e-05, "loss": 0.1736, "step": 5794 }, { "epoch": 35.99688473520249, "grad_norm": 2.984375, "learning_rate": 4.645211607225225e-05, "loss": 0.3359, "step": 5795 }, { "epoch": 36.0, "grad_norm": 1.265625, "learning_rate": 4.645085080573082e-05, "loss": 0.1044, "step": 5796 }, { "epoch": 36.006230529595015, "grad_norm": 2.28125, "learning_rate": 4.644958533087443e-05, "loss": 0.1663, "step": 5797 }, { "epoch": 36.01246105919003, "grad_norm": 2.390625, "learning_rate": 4.6448319647695384e-05, "loss": 0.1721, "step": 5798 }, { "epoch": 36.018691588785046, "grad_norm": 1.90625, "learning_rate": 4.6447053756205966e-05, "loss": 0.1445, "step": 5799 }, { "epoch": 36.02492211838006, "grad_norm": 2.03125, "learning_rate": 4.644578765641848e-05, "loss": 0.1571, "step": 5800 }, { "epoch": 36.03115264797508, "grad_norm": 1.9765625, "learning_rate": 4.6444521348345216e-05, "loss": 0.2114, "step": 5801 }, { "epoch": 36.03738317757009, "grad_norm": 2.296875, "learning_rate": 4.6443254831998475e-05, "loss": 0.1694, "step": 5802 }, { "epoch": 36.04361370716511, "grad_norm": 1.8359375, "learning_rate": 4.644198810739056e-05, "loss": 0.1646, "step": 5803 }, { "epoch": 36.04984423676012, "grad_norm": 2.953125, "learning_rate": 4.644072117453376e-05, "loss": 0.2339, "step": 5804 }, { "epoch": 36.05607476635514, "grad_norm": 3.921875, "learning_rate": 4.64394540334404e-05, "loss": 0.2555, "step": 5805 }, { "epoch": 36.06230529595015, "grad_norm": 3.640625, "learning_rate": 4.643818668412277e-05, "loss": 0.2253, "step": 5806 }, { "epoch": 36.06853582554517, "grad_norm": 3.640625, "learning_rate": 4.6436919126593195e-05, "loss": 0.3139, "step": 5807 }, { "epoch": 36.074766355140184, "grad_norm": 1.9609375, "learning_rate": 4.6435651360863967e-05, "loss": 0.1837, "step": 5808 }, { "epoch": 36.0809968847352, "grad_norm": 2.9375, "learning_rate": 4.64343833869474e-05, "loss": 0.2572, "step": 5809 }, { "epoch": 36.087227414330215, "grad_norm": 3.828125, "learning_rate": 4.6433115204855835e-05, "loss": 0.2726, "step": 5810 }, { "epoch": 36.09345794392523, "grad_norm": 3.40625, "learning_rate": 4.6431846814601564e-05, "loss": 0.3173, "step": 5811 }, { "epoch": 36.099688473520246, "grad_norm": 2.234375, "learning_rate": 4.6430578216196915e-05, "loss": 0.1763, "step": 5812 }, { "epoch": 36.10591900311527, "grad_norm": 1.9140625, "learning_rate": 4.64293094096542e-05, "loss": 0.1589, "step": 5813 }, { "epoch": 36.11214953271028, "grad_norm": 4.25, "learning_rate": 4.642804039498575e-05, "loss": 0.2892, "step": 5814 }, { "epoch": 36.1183800623053, "grad_norm": 1.7890625, "learning_rate": 4.6426771172203893e-05, "loss": 0.1712, "step": 5815 }, { "epoch": 36.124610591900314, "grad_norm": 2.28125, "learning_rate": 4.6425501741320946e-05, "loss": 0.115, "step": 5816 }, { "epoch": 36.13084112149533, "grad_norm": 2.5625, "learning_rate": 4.6424232102349243e-05, "loss": 0.1691, "step": 5817 }, { "epoch": 36.137071651090345, "grad_norm": 2.234375, "learning_rate": 4.642296225530112e-05, "loss": 0.1778, "step": 5818 }, { "epoch": 36.14330218068536, "grad_norm": 2.890625, "learning_rate": 4.6421692200188906e-05, "loss": 0.2132, "step": 5819 }, { "epoch": 36.149532710280376, "grad_norm": 4.75, "learning_rate": 4.642042193702493e-05, "loss": 0.5643, "step": 5820 }, { "epoch": 36.15576323987539, "grad_norm": 3.5, "learning_rate": 4.641915146582154e-05, "loss": 0.246, "step": 5821 }, { "epoch": 36.161993769470406, "grad_norm": 3.25, "learning_rate": 4.641788078659106e-05, "loss": 0.2878, "step": 5822 }, { "epoch": 36.16822429906542, "grad_norm": 2.421875, "learning_rate": 4.641660989934585e-05, "loss": 0.224, "step": 5823 }, { "epoch": 36.17445482866044, "grad_norm": 3.9375, "learning_rate": 4.641533880409824e-05, "loss": 0.3751, "step": 5824 }, { "epoch": 36.18068535825545, "grad_norm": 3.375, "learning_rate": 4.641406750086058e-05, "loss": 0.272, "step": 5825 }, { "epoch": 36.18691588785047, "grad_norm": 2.984375, "learning_rate": 4.641279598964521e-05, "loss": 0.2314, "step": 5826 }, { "epoch": 36.19314641744548, "grad_norm": 3.265625, "learning_rate": 4.641152427046449e-05, "loss": 0.3101, "step": 5827 }, { "epoch": 36.1993769470405, "grad_norm": 2.625, "learning_rate": 4.6410252343330765e-05, "loss": 0.3125, "step": 5828 }, { "epoch": 36.205607476635514, "grad_norm": 2.796875, "learning_rate": 4.6408980208256394e-05, "loss": 0.1866, "step": 5829 }, { "epoch": 36.21183800623053, "grad_norm": 2.671875, "learning_rate": 4.640770786525372e-05, "loss": 0.4048, "step": 5830 }, { "epoch": 36.218068535825545, "grad_norm": 2.5, "learning_rate": 4.640643531433512e-05, "loss": 0.1532, "step": 5831 }, { "epoch": 36.22429906542056, "grad_norm": 2.5625, "learning_rate": 4.640516255551293e-05, "loss": 0.3823, "step": 5832 }, { "epoch": 36.230529595015575, "grad_norm": 2.46875, "learning_rate": 4.640388958879952e-05, "loss": 0.2093, "step": 5833 }, { "epoch": 36.23676012461059, "grad_norm": 2.921875, "learning_rate": 4.640261641420726e-05, "loss": 0.2476, "step": 5834 }, { "epoch": 36.242990654205606, "grad_norm": 1.75, "learning_rate": 4.640134303174851e-05, "loss": 0.1097, "step": 5835 }, { "epoch": 36.24922118380062, "grad_norm": 3.125, "learning_rate": 4.6400069441435644e-05, "loss": 0.1729, "step": 5836 }, { "epoch": 36.25545171339564, "grad_norm": 4.3125, "learning_rate": 4.639879564328102e-05, "loss": 0.2965, "step": 5837 }, { "epoch": 36.26168224299065, "grad_norm": 3.421875, "learning_rate": 4.639752163729703e-05, "loss": 0.1252, "step": 5838 }, { "epoch": 36.26791277258567, "grad_norm": 2.234375, "learning_rate": 4.6396247423496006e-05, "loss": 0.1309, "step": 5839 }, { "epoch": 36.27414330218068, "grad_norm": 3.75, "learning_rate": 4.639497300189037e-05, "loss": 0.3085, "step": 5840 }, { "epoch": 36.2803738317757, "grad_norm": 2.53125, "learning_rate": 4.639369837249247e-05, "loss": 0.2374, "step": 5841 }, { "epoch": 36.28660436137071, "grad_norm": 2.875, "learning_rate": 4.63924235353147e-05, "loss": 0.2015, "step": 5842 }, { "epoch": 36.29283489096573, "grad_norm": 3.171875, "learning_rate": 4.639114849036944e-05, "loss": 0.2131, "step": 5843 }, { "epoch": 36.299065420560744, "grad_norm": 3.359375, "learning_rate": 4.638987323766906e-05, "loss": 0.2463, "step": 5844 }, { "epoch": 36.30529595015577, "grad_norm": 2.28125, "learning_rate": 4.6388597777225964e-05, "loss": 0.1977, "step": 5845 }, { "epoch": 36.31152647975078, "grad_norm": 3.359375, "learning_rate": 4.638732210905252e-05, "loss": 0.2955, "step": 5846 }, { "epoch": 36.3177570093458, "grad_norm": 2.140625, "learning_rate": 4.638604623316114e-05, "loss": 0.176, "step": 5847 }, { "epoch": 36.32398753894081, "grad_norm": 1.8203125, "learning_rate": 4.63847701495642e-05, "loss": 0.1717, "step": 5848 }, { "epoch": 36.33021806853583, "grad_norm": 1.7265625, "learning_rate": 4.63834938582741e-05, "loss": 0.1471, "step": 5849 }, { "epoch": 36.33644859813084, "grad_norm": 2.9375, "learning_rate": 4.638221735930323e-05, "loss": 0.2234, "step": 5850 }, { "epoch": 36.34267912772586, "grad_norm": 1.8984375, "learning_rate": 4.638094065266398e-05, "loss": 0.1418, "step": 5851 }, { "epoch": 36.348909657320874, "grad_norm": 1.6328125, "learning_rate": 4.6379663738368776e-05, "loss": 0.1349, "step": 5852 }, { "epoch": 36.35514018691589, "grad_norm": 1.9140625, "learning_rate": 4.637838661643e-05, "loss": 0.1137, "step": 5853 }, { "epoch": 36.361370716510905, "grad_norm": 3.3125, "learning_rate": 4.637710928686006e-05, "loss": 0.3129, "step": 5854 }, { "epoch": 36.36760124610592, "grad_norm": 3.34375, "learning_rate": 4.6375831749671364e-05, "loss": 0.2719, "step": 5855 }, { "epoch": 36.373831775700936, "grad_norm": 2.75, "learning_rate": 4.637455400487632e-05, "loss": 0.2608, "step": 5856 }, { "epoch": 36.38006230529595, "grad_norm": 2.046875, "learning_rate": 4.637327605248732e-05, "loss": 0.1371, "step": 5857 }, { "epoch": 36.386292834890966, "grad_norm": 3.25, "learning_rate": 4.6371997892516795e-05, "loss": 0.2038, "step": 5858 }, { "epoch": 36.39252336448598, "grad_norm": 3.3125, "learning_rate": 4.6370719524977155e-05, "loss": 0.2177, "step": 5859 }, { "epoch": 36.398753894081, "grad_norm": 3.125, "learning_rate": 4.6369440949880826e-05, "loss": 0.1752, "step": 5860 }, { "epoch": 36.40498442367601, "grad_norm": 1.796875, "learning_rate": 4.63681621672402e-05, "loss": 0.1242, "step": 5861 }, { "epoch": 36.41121495327103, "grad_norm": 3.046875, "learning_rate": 4.636688317706772e-05, "loss": 0.2125, "step": 5862 }, { "epoch": 36.41744548286604, "grad_norm": 2.15625, "learning_rate": 4.6365603979375804e-05, "loss": 0.2238, "step": 5863 }, { "epoch": 36.42367601246106, "grad_norm": 1.390625, "learning_rate": 4.636432457417687e-05, "loss": 0.1543, "step": 5864 }, { "epoch": 36.429906542056074, "grad_norm": 4.5625, "learning_rate": 4.636304496148334e-05, "loss": 0.3637, "step": 5865 }, { "epoch": 36.43613707165109, "grad_norm": 2.703125, "learning_rate": 4.6361765141307645e-05, "loss": 0.2354, "step": 5866 }, { "epoch": 36.442367601246104, "grad_norm": 2.578125, "learning_rate": 4.6360485113662216e-05, "loss": 0.2205, "step": 5867 }, { "epoch": 36.44859813084112, "grad_norm": 2.1875, "learning_rate": 4.635920487855949e-05, "loss": 0.1105, "step": 5868 }, { "epoch": 36.454828660436135, "grad_norm": 1.7421875, "learning_rate": 4.635792443601189e-05, "loss": 0.1416, "step": 5869 }, { "epoch": 36.46105919003115, "grad_norm": 2.015625, "learning_rate": 4.635664378603187e-05, "loss": 0.2487, "step": 5870 }, { "epoch": 36.467289719626166, "grad_norm": 2.078125, "learning_rate": 4.635536292863185e-05, "loss": 0.1397, "step": 5871 }, { "epoch": 36.47352024922118, "grad_norm": 2.140625, "learning_rate": 4.635408186382427e-05, "loss": 0.1939, "step": 5872 }, { "epoch": 36.4797507788162, "grad_norm": 2.75, "learning_rate": 4.6352800591621595e-05, "loss": 0.1905, "step": 5873 }, { "epoch": 36.48598130841121, "grad_norm": 1.8671875, "learning_rate": 4.635151911203624e-05, "loss": 0.2569, "step": 5874 }, { "epoch": 36.49221183800623, "grad_norm": 2.546875, "learning_rate": 4.635023742508067e-05, "loss": 0.2732, "step": 5875 }, { "epoch": 36.49844236760124, "grad_norm": 1.59375, "learning_rate": 4.6348955530767314e-05, "loss": 0.1083, "step": 5876 }, { "epoch": 36.504672897196265, "grad_norm": 1.4140625, "learning_rate": 4.634767342910864e-05, "loss": 0.1296, "step": 5877 }, { "epoch": 36.51090342679128, "grad_norm": 3.21875, "learning_rate": 4.634639112011711e-05, "loss": 0.2391, "step": 5878 }, { "epoch": 36.517133956386296, "grad_norm": 2.78125, "learning_rate": 4.6345108603805146e-05, "loss": 0.1666, "step": 5879 }, { "epoch": 36.52336448598131, "grad_norm": 2.09375, "learning_rate": 4.6343825880185224e-05, "loss": 0.1521, "step": 5880 }, { "epoch": 36.52959501557633, "grad_norm": 1.3828125, "learning_rate": 4.63425429492698e-05, "loss": 0.1404, "step": 5881 }, { "epoch": 36.53582554517134, "grad_norm": 2.671875, "learning_rate": 4.634125981107133e-05, "loss": 0.1959, "step": 5882 }, { "epoch": 36.54205607476636, "grad_norm": 1.90625, "learning_rate": 4.6339976465602276e-05, "loss": 0.1506, "step": 5883 }, { "epoch": 36.54828660436137, "grad_norm": 2.78125, "learning_rate": 4.6338692912875104e-05, "loss": 0.214, "step": 5884 }, { "epoch": 36.55451713395639, "grad_norm": 2.203125, "learning_rate": 4.633740915290229e-05, "loss": 0.1332, "step": 5885 }, { "epoch": 36.5607476635514, "grad_norm": 2.96875, "learning_rate": 4.633612518569629e-05, "loss": 0.2542, "step": 5886 }, { "epoch": 36.56697819314642, "grad_norm": 2.4375, "learning_rate": 4.633484101126957e-05, "loss": 0.2003, "step": 5887 }, { "epoch": 36.573208722741434, "grad_norm": 2.34375, "learning_rate": 4.6333556629634616e-05, "loss": 0.2522, "step": 5888 }, { "epoch": 36.57943925233645, "grad_norm": 2.59375, "learning_rate": 4.6332272040803895e-05, "loss": 0.1371, "step": 5889 }, { "epoch": 36.585669781931465, "grad_norm": 2.015625, "learning_rate": 4.633098724478988e-05, "loss": 0.1734, "step": 5890 }, { "epoch": 36.59190031152648, "grad_norm": 1.5625, "learning_rate": 4.632970224160505e-05, "loss": 0.1384, "step": 5891 }, { "epoch": 36.598130841121495, "grad_norm": 3.59375, "learning_rate": 4.632841703126191e-05, "loss": 0.439, "step": 5892 }, { "epoch": 36.60436137071651, "grad_norm": 2.234375, "learning_rate": 4.63271316137729e-05, "loss": 0.1837, "step": 5893 }, { "epoch": 36.610591900311526, "grad_norm": 2.859375, "learning_rate": 4.6325845989150526e-05, "loss": 0.2098, "step": 5894 }, { "epoch": 36.61682242990654, "grad_norm": 3.546875, "learning_rate": 4.632456015740727e-05, "loss": 0.3212, "step": 5895 }, { "epoch": 36.62305295950156, "grad_norm": 2.5, "learning_rate": 4.632327411855563e-05, "loss": 0.1205, "step": 5896 }, { "epoch": 36.62928348909657, "grad_norm": 3.015625, "learning_rate": 4.632198787260809e-05, "loss": 0.1669, "step": 5897 }, { "epoch": 36.63551401869159, "grad_norm": 2.78125, "learning_rate": 4.632070141957714e-05, "loss": 0.1527, "step": 5898 }, { "epoch": 36.6417445482866, "grad_norm": 2.625, "learning_rate": 4.6319414759475274e-05, "loss": 0.2194, "step": 5899 }, { "epoch": 36.64797507788162, "grad_norm": 2.90625, "learning_rate": 4.6318127892315e-05, "loss": 0.3155, "step": 5900 }, { "epoch": 36.654205607476634, "grad_norm": 1.7265625, "learning_rate": 4.631684081810879e-05, "loss": 0.1653, "step": 5901 }, { "epoch": 36.66043613707165, "grad_norm": 2.015625, "learning_rate": 4.631555353686917e-05, "loss": 0.1601, "step": 5902 }, { "epoch": 36.666666666666664, "grad_norm": 2.421875, "learning_rate": 4.631426604860864e-05, "loss": 0.2088, "step": 5903 }, { "epoch": 36.67289719626168, "grad_norm": 1.1875, "learning_rate": 4.631297835333969e-05, "loss": 0.1415, "step": 5904 }, { "epoch": 36.679127725856695, "grad_norm": 2.734375, "learning_rate": 4.631169045107483e-05, "loss": 0.1655, "step": 5905 }, { "epoch": 36.68535825545171, "grad_norm": 3.46875, "learning_rate": 4.631040234182658e-05, "loss": 0.1264, "step": 5906 }, { "epoch": 36.691588785046726, "grad_norm": 2.890625, "learning_rate": 4.630911402560743e-05, "loss": 0.2112, "step": 5907 }, { "epoch": 36.69781931464174, "grad_norm": 3.25, "learning_rate": 4.630782550242991e-05, "loss": 0.2722, "step": 5908 }, { "epoch": 36.70404984423676, "grad_norm": 3.21875, "learning_rate": 4.630653677230654e-05, "loss": 0.2159, "step": 5909 }, { "epoch": 36.71028037383178, "grad_norm": 3.296875, "learning_rate": 4.630524783524982e-05, "loss": 0.1719, "step": 5910 }, { "epoch": 36.716510903426794, "grad_norm": 2.0, "learning_rate": 4.630395869127226e-05, "loss": 0.1417, "step": 5911 }, { "epoch": 36.72274143302181, "grad_norm": 2.75, "learning_rate": 4.630266934038642e-05, "loss": 0.2162, "step": 5912 }, { "epoch": 36.728971962616825, "grad_norm": 2.546875, "learning_rate": 4.630137978260478e-05, "loss": 0.1518, "step": 5913 }, { "epoch": 36.73520249221184, "grad_norm": 3.703125, "learning_rate": 4.630009001793989e-05, "loss": 0.2325, "step": 5914 }, { "epoch": 36.741433021806856, "grad_norm": 2.96875, "learning_rate": 4.629880004640426e-05, "loss": 0.2266, "step": 5915 }, { "epoch": 36.74766355140187, "grad_norm": 1.984375, "learning_rate": 4.629750986801043e-05, "loss": 0.1535, "step": 5916 }, { "epoch": 36.753894080996886, "grad_norm": 1.8359375, "learning_rate": 4.6296219482770925e-05, "loss": 0.1462, "step": 5917 }, { "epoch": 36.7601246105919, "grad_norm": 1.9140625, "learning_rate": 4.629492889069829e-05, "loss": 0.2322, "step": 5918 }, { "epoch": 36.76635514018692, "grad_norm": 4.0, "learning_rate": 4.629363809180503e-05, "loss": 0.4212, "step": 5919 }, { "epoch": 36.77258566978193, "grad_norm": 2.953125, "learning_rate": 4.629234708610372e-05, "loss": 0.2838, "step": 5920 }, { "epoch": 36.77881619937695, "grad_norm": 3.109375, "learning_rate": 4.6291055873606856e-05, "loss": 0.323, "step": 5921 }, { "epoch": 36.78504672897196, "grad_norm": 2.828125, "learning_rate": 4.628976445432702e-05, "loss": 0.2343, "step": 5922 }, { "epoch": 36.79127725856698, "grad_norm": 2.234375, "learning_rate": 4.628847282827673e-05, "loss": 0.1694, "step": 5923 }, { "epoch": 36.797507788161994, "grad_norm": 2.46875, "learning_rate": 4.628718099546854e-05, "loss": 0.1633, "step": 5924 }, { "epoch": 36.80373831775701, "grad_norm": 3.859375, "learning_rate": 4.628588895591499e-05, "loss": 0.3108, "step": 5925 }, { "epoch": 36.809968847352025, "grad_norm": 2.984375, "learning_rate": 4.628459670962863e-05, "loss": 0.1661, "step": 5926 }, { "epoch": 36.81619937694704, "grad_norm": 2.03125, "learning_rate": 4.628330425662202e-05, "loss": 0.2208, "step": 5927 }, { "epoch": 36.822429906542055, "grad_norm": 2.75, "learning_rate": 4.6282011596907694e-05, "loss": 0.1856, "step": 5928 }, { "epoch": 36.82866043613707, "grad_norm": 2.171875, "learning_rate": 4.628071873049822e-05, "loss": 0.185, "step": 5929 }, { "epoch": 36.834890965732086, "grad_norm": 2.984375, "learning_rate": 4.627942565740615e-05, "loss": 0.172, "step": 5930 }, { "epoch": 36.8411214953271, "grad_norm": 2.453125, "learning_rate": 4.627813237764405e-05, "loss": 0.2416, "step": 5931 }, { "epoch": 36.84735202492212, "grad_norm": 2.5, "learning_rate": 4.627683889122447e-05, "loss": 0.329, "step": 5932 }, { "epoch": 36.85358255451713, "grad_norm": 2.859375, "learning_rate": 4.627554519815999e-05, "loss": 0.3174, "step": 5933 }, { "epoch": 36.85981308411215, "grad_norm": 2.234375, "learning_rate": 4.627425129846314e-05, "loss": 0.1495, "step": 5934 }, { "epoch": 36.86604361370716, "grad_norm": 2.53125, "learning_rate": 4.627295719214653e-05, "loss": 0.2078, "step": 5935 }, { "epoch": 36.87227414330218, "grad_norm": 2.0, "learning_rate": 4.6271662879222695e-05, "loss": 0.171, "step": 5936 }, { "epoch": 36.87850467289719, "grad_norm": 1.3203125, "learning_rate": 4.627036835970422e-05, "loss": 0.126, "step": 5937 }, { "epoch": 36.88473520249221, "grad_norm": 2.8125, "learning_rate": 4.626907363360368e-05, "loss": 0.2531, "step": 5938 }, { "epoch": 36.890965732087224, "grad_norm": 1.9296875, "learning_rate": 4.626777870093364e-05, "loss": 0.1598, "step": 5939 }, { "epoch": 36.89719626168224, "grad_norm": 2.140625, "learning_rate": 4.626648356170669e-05, "loss": 0.1559, "step": 5940 }, { "epoch": 36.90342679127726, "grad_norm": 2.515625, "learning_rate": 4.626518821593538e-05, "loss": 0.1894, "step": 5941 }, { "epoch": 36.90965732087228, "grad_norm": 1.7265625, "learning_rate": 4.626389266363234e-05, "loss": 0.2325, "step": 5942 }, { "epoch": 36.91588785046729, "grad_norm": 2.125, "learning_rate": 4.626259690481011e-05, "loss": 0.1966, "step": 5943 }, { "epoch": 36.92211838006231, "grad_norm": 2.8125, "learning_rate": 4.626130093948128e-05, "loss": 0.1529, "step": 5944 }, { "epoch": 36.92834890965732, "grad_norm": 4.21875, "learning_rate": 4.626000476765846e-05, "loss": 0.4714, "step": 5945 }, { "epoch": 36.93457943925234, "grad_norm": 2.875, "learning_rate": 4.625870838935421e-05, "loss": 0.2372, "step": 5946 }, { "epoch": 36.940809968847354, "grad_norm": 1.515625, "learning_rate": 4.625741180458114e-05, "loss": 0.1671, "step": 5947 }, { "epoch": 36.94704049844237, "grad_norm": 2.375, "learning_rate": 4.625611501335184e-05, "loss": 0.19, "step": 5948 }, { "epoch": 36.953271028037385, "grad_norm": 2.515625, "learning_rate": 4.6254818015678904e-05, "loss": 0.2509, "step": 5949 }, { "epoch": 36.9595015576324, "grad_norm": 3.828125, "learning_rate": 4.625352081157492e-05, "loss": 0.225, "step": 5950 }, { "epoch": 36.965732087227416, "grad_norm": 2.484375, "learning_rate": 4.62522234010525e-05, "loss": 0.1451, "step": 5951 }, { "epoch": 36.97196261682243, "grad_norm": 2.53125, "learning_rate": 4.6250925784124233e-05, "loss": 0.1993, "step": 5952 }, { "epoch": 36.978193146417446, "grad_norm": 2.046875, "learning_rate": 4.624962796080273e-05, "loss": 0.1648, "step": 5953 }, { "epoch": 36.98442367601246, "grad_norm": 2.609375, "learning_rate": 4.6248329931100596e-05, "loss": 0.313, "step": 5954 }, { "epoch": 36.99065420560748, "grad_norm": 2.453125, "learning_rate": 4.624703169503043e-05, "loss": 0.2258, "step": 5955 }, { "epoch": 36.99688473520249, "grad_norm": 2.5625, "learning_rate": 4.624573325260484e-05, "loss": 0.2422, "step": 5956 }, { "epoch": 37.0, "grad_norm": 1.1875, "learning_rate": 4.624443460383645e-05, "loss": 0.0829, "step": 5957 }, { "epoch": 37.006230529595015, "grad_norm": 3.359375, "learning_rate": 4.6243135748737864e-05, "loss": 0.2859, "step": 5958 }, { "epoch": 37.01246105919003, "grad_norm": 2.28125, "learning_rate": 4.6241836687321694e-05, "loss": 0.1339, "step": 5959 }, { "epoch": 37.018691588785046, "grad_norm": 2.59375, "learning_rate": 4.624053741960056e-05, "loss": 0.2014, "step": 5960 }, { "epoch": 37.02492211838006, "grad_norm": 2.375, "learning_rate": 4.623923794558708e-05, "loss": 0.1455, "step": 5961 }, { "epoch": 37.03115264797508, "grad_norm": 1.96875, "learning_rate": 4.623793826529388e-05, "loss": 0.2181, "step": 5962 }, { "epoch": 37.03738317757009, "grad_norm": 1.703125, "learning_rate": 4.623663837873357e-05, "loss": 0.1506, "step": 5963 }, { "epoch": 37.04361370716511, "grad_norm": 2.15625, "learning_rate": 4.623533828591879e-05, "loss": 0.1323, "step": 5964 }, { "epoch": 37.04984423676012, "grad_norm": 2.046875, "learning_rate": 4.6234037986862155e-05, "loss": 0.1706, "step": 5965 }, { "epoch": 37.05607476635514, "grad_norm": 3.890625, "learning_rate": 4.6232737481576296e-05, "loss": 0.217, "step": 5966 }, { "epoch": 37.06230529595015, "grad_norm": 3.171875, "learning_rate": 4.6231436770073856e-05, "loss": 0.231, "step": 5967 }, { "epoch": 37.06853582554517, "grad_norm": 2.890625, "learning_rate": 4.623013585236745e-05, "loss": 0.2451, "step": 5968 }, { "epoch": 37.074766355140184, "grad_norm": 3.375, "learning_rate": 4.622883472846973e-05, "loss": 0.2413, "step": 5969 }, { "epoch": 37.0809968847352, "grad_norm": 1.4375, "learning_rate": 4.6227533398393315e-05, "loss": 0.1297, "step": 5970 }, { "epoch": 37.087227414330215, "grad_norm": 2.296875, "learning_rate": 4.622623186215085e-05, "loss": 0.2729, "step": 5971 }, { "epoch": 37.09345794392523, "grad_norm": 3.265625, "learning_rate": 4.622493011975498e-05, "loss": 0.2586, "step": 5972 }, { "epoch": 37.099688473520246, "grad_norm": 2.28125, "learning_rate": 4.622362817121835e-05, "loss": 0.1435, "step": 5973 }, { "epoch": 37.10591900311527, "grad_norm": 2.125, "learning_rate": 4.62223260165536e-05, "loss": 0.1516, "step": 5974 }, { "epoch": 37.11214953271028, "grad_norm": 3.59375, "learning_rate": 4.6221023655773375e-05, "loss": 0.1963, "step": 5975 }, { "epoch": 37.1183800623053, "grad_norm": 2.515625, "learning_rate": 4.621972108889033e-05, "loss": 0.1559, "step": 5976 }, { "epoch": 37.124610591900314, "grad_norm": 2.421875, "learning_rate": 4.621841831591711e-05, "loss": 0.2374, "step": 5977 }, { "epoch": 37.13084112149533, "grad_norm": 2.296875, "learning_rate": 4.6217115336866375e-05, "loss": 0.1392, "step": 5978 }, { "epoch": 37.137071651090345, "grad_norm": 2.75, "learning_rate": 4.621581215175077e-05, "loss": 0.1995, "step": 5979 }, { "epoch": 37.14330218068536, "grad_norm": 2.625, "learning_rate": 4.621450876058295e-05, "loss": 0.1662, "step": 5980 }, { "epoch": 37.149532710280376, "grad_norm": 3.0625, "learning_rate": 4.6213205163375586e-05, "loss": 0.2681, "step": 5981 }, { "epoch": 37.15576323987539, "grad_norm": 3.046875, "learning_rate": 4.621190136014134e-05, "loss": 0.2508, "step": 5982 }, { "epoch": 37.161993769470406, "grad_norm": 2.484375, "learning_rate": 4.621059735089286e-05, "loss": 0.1809, "step": 5983 }, { "epoch": 37.16822429906542, "grad_norm": 1.7578125, "learning_rate": 4.620929313564282e-05, "loss": 0.1381, "step": 5984 }, { "epoch": 37.17445482866044, "grad_norm": 3.0, "learning_rate": 4.620798871440388e-05, "loss": 0.2761, "step": 5985 }, { "epoch": 37.18068535825545, "grad_norm": 4.1875, "learning_rate": 4.620668408718872e-05, "loss": 0.1834, "step": 5986 }, { "epoch": 37.18691588785047, "grad_norm": 1.4296875, "learning_rate": 4.620537925401001e-05, "loss": 0.1321, "step": 5987 }, { "epoch": 37.19314641744548, "grad_norm": 3.125, "learning_rate": 4.620407421488041e-05, "loss": 0.2411, "step": 5988 }, { "epoch": 37.1993769470405, "grad_norm": 2.390625, "learning_rate": 4.6202768969812606e-05, "loss": 0.1425, "step": 5989 }, { "epoch": 37.205607476635514, "grad_norm": 1.2890625, "learning_rate": 4.620146351881927e-05, "loss": 0.1213, "step": 5990 }, { "epoch": 37.21183800623053, "grad_norm": 2.65625, "learning_rate": 4.620015786191308e-05, "loss": 0.1635, "step": 5991 }, { "epoch": 37.218068535825545, "grad_norm": 1.8203125, "learning_rate": 4.619885199910672e-05, "loss": 0.1468, "step": 5992 }, { "epoch": 37.22429906542056, "grad_norm": 2.6875, "learning_rate": 4.6197545930412874e-05, "loss": 0.1777, "step": 5993 }, { "epoch": 37.230529595015575, "grad_norm": 2.625, "learning_rate": 4.6196239655844225e-05, "loss": 0.1266, "step": 5994 }, { "epoch": 37.23676012461059, "grad_norm": 2.859375, "learning_rate": 4.619493317541346e-05, "loss": 0.1764, "step": 5995 }, { "epoch": 37.242990654205606, "grad_norm": 2.5625, "learning_rate": 4.6193626489133265e-05, "loss": 0.2153, "step": 5996 }, { "epoch": 37.24922118380062, "grad_norm": 1.9453125, "learning_rate": 4.619231959701633e-05, "loss": 0.136, "step": 5997 }, { "epoch": 37.25545171339564, "grad_norm": 2.109375, "learning_rate": 4.6191012499075355e-05, "loss": 0.2163, "step": 5998 }, { "epoch": 37.26168224299065, "grad_norm": 2.0, "learning_rate": 4.618970519532303e-05, "loss": 0.175, "step": 5999 }, { "epoch": 37.26791277258567, "grad_norm": 2.28125, "learning_rate": 4.618839768577205e-05, "loss": 0.2113, "step": 6000 }, { "epoch": 37.27414330218068, "grad_norm": 3.46875, "learning_rate": 4.6187089970435124e-05, "loss": 0.2237, "step": 6001 }, { "epoch": 37.2803738317757, "grad_norm": 3.015625, "learning_rate": 4.6185782049324936e-05, "loss": 0.2623, "step": 6002 }, { "epoch": 37.28660436137071, "grad_norm": 2.53125, "learning_rate": 4.6184473922454195e-05, "loss": 0.2182, "step": 6003 }, { "epoch": 37.29283489096573, "grad_norm": 2.375, "learning_rate": 4.618316558983562e-05, "loss": 0.1869, "step": 6004 }, { "epoch": 37.299065420560744, "grad_norm": 4.0, "learning_rate": 4.6181857051481905e-05, "loss": 0.2598, "step": 6005 }, { "epoch": 37.30529595015577, "grad_norm": 2.484375, "learning_rate": 4.618054830740576e-05, "loss": 0.2134, "step": 6006 }, { "epoch": 37.31152647975078, "grad_norm": 2.71875, "learning_rate": 4.6179239357619885e-05, "loss": 0.1722, "step": 6007 }, { "epoch": 37.3177570093458, "grad_norm": 1.8359375, "learning_rate": 4.617793020213701e-05, "loss": 0.1316, "step": 6008 }, { "epoch": 37.32398753894081, "grad_norm": 3.1875, "learning_rate": 4.617662084096985e-05, "loss": 0.2911, "step": 6009 }, { "epoch": 37.33021806853583, "grad_norm": 5.1875, "learning_rate": 4.617531127413111e-05, "loss": 0.2764, "step": 6010 }, { "epoch": 37.33644859813084, "grad_norm": 2.03125, "learning_rate": 4.617400150163351e-05, "loss": 0.1582, "step": 6011 }, { "epoch": 37.34267912772586, "grad_norm": 2.65625, "learning_rate": 4.6172691523489775e-05, "loss": 0.2345, "step": 6012 }, { "epoch": 37.348909657320874, "grad_norm": 3.15625, "learning_rate": 4.617138133971264e-05, "loss": 0.3373, "step": 6013 }, { "epoch": 37.35514018691589, "grad_norm": 2.453125, "learning_rate": 4.61700709503148e-05, "loss": 0.1256, "step": 6014 }, { "epoch": 37.361370716510905, "grad_norm": 2.875, "learning_rate": 4.616876035530901e-05, "loss": 0.3643, "step": 6015 }, { "epoch": 37.36760124610592, "grad_norm": 2.0625, "learning_rate": 4.616744955470799e-05, "loss": 0.1647, "step": 6016 }, { "epoch": 37.373831775700936, "grad_norm": 4.4375, "learning_rate": 4.6166138548524465e-05, "loss": 0.2182, "step": 6017 }, { "epoch": 37.38006230529595, "grad_norm": 3.859375, "learning_rate": 4.616482733677117e-05, "loss": 0.2356, "step": 6018 }, { "epoch": 37.386292834890966, "grad_norm": 3.078125, "learning_rate": 4.616351591946084e-05, "loss": 0.1899, "step": 6019 }, { "epoch": 37.39252336448598, "grad_norm": 4.0, "learning_rate": 4.616220429660621e-05, "loss": 0.3035, "step": 6020 }, { "epoch": 37.398753894081, "grad_norm": 2.1875, "learning_rate": 4.616089246822003e-05, "loss": 0.1762, "step": 6021 }, { "epoch": 37.40498442367601, "grad_norm": 3.203125, "learning_rate": 4.615958043431503e-05, "loss": 0.2289, "step": 6022 }, { "epoch": 37.41121495327103, "grad_norm": 2.234375, "learning_rate": 4.6158268194903956e-05, "loss": 0.1237, "step": 6023 }, { "epoch": 37.41744548286604, "grad_norm": 2.171875, "learning_rate": 4.6156955749999545e-05, "loss": 0.2218, "step": 6024 }, { "epoch": 37.42367601246106, "grad_norm": 1.8359375, "learning_rate": 4.615564309961455e-05, "loss": 0.1495, "step": 6025 }, { "epoch": 37.429906542056074, "grad_norm": 2.71875, "learning_rate": 4.615433024376173e-05, "loss": 0.2801, "step": 6026 }, { "epoch": 37.43613707165109, "grad_norm": 1.046875, "learning_rate": 4.6153017182453814e-05, "loss": 0.13, "step": 6027 }, { "epoch": 37.442367601246104, "grad_norm": 2.375, "learning_rate": 4.6151703915703584e-05, "loss": 0.2526, "step": 6028 }, { "epoch": 37.44859813084112, "grad_norm": 2.578125, "learning_rate": 4.615039044352376e-05, "loss": 0.2449, "step": 6029 }, { "epoch": 37.454828660436135, "grad_norm": 1.3671875, "learning_rate": 4.6149076765927126e-05, "loss": 0.1237, "step": 6030 }, { "epoch": 37.46105919003115, "grad_norm": 2.796875, "learning_rate": 4.614776288292643e-05, "loss": 0.2761, "step": 6031 }, { "epoch": 37.467289719626166, "grad_norm": 1.578125, "learning_rate": 4.6146448794534435e-05, "loss": 0.1327, "step": 6032 }, { "epoch": 37.47352024922118, "grad_norm": 1.6875, "learning_rate": 4.61451345007639e-05, "loss": 0.137, "step": 6033 }, { "epoch": 37.4797507788162, "grad_norm": 2.515625, "learning_rate": 4.614382000162759e-05, "loss": 0.2037, "step": 6034 }, { "epoch": 37.48598130841121, "grad_norm": 1.7890625, "learning_rate": 4.614250529713827e-05, "loss": 0.1308, "step": 6035 }, { "epoch": 37.49221183800623, "grad_norm": 2.9375, "learning_rate": 4.614119038730872e-05, "loss": 0.2361, "step": 6036 }, { "epoch": 37.49844236760124, "grad_norm": 1.3671875, "learning_rate": 4.61398752721517e-05, "loss": 0.1151, "step": 6037 }, { "epoch": 37.504672897196265, "grad_norm": 2.953125, "learning_rate": 4.6138559951679985e-05, "loss": 0.3032, "step": 6038 }, { "epoch": 37.51090342679128, "grad_norm": 3.375, "learning_rate": 4.613724442590635e-05, "loss": 0.2957, "step": 6039 }, { "epoch": 37.517133956386296, "grad_norm": 2.421875, "learning_rate": 4.613592869484358e-05, "loss": 0.1692, "step": 6040 }, { "epoch": 37.52336448598131, "grad_norm": 2.453125, "learning_rate": 4.613461275850443e-05, "loss": 0.1669, "step": 6041 }, { "epoch": 37.52959501557633, "grad_norm": 2.765625, "learning_rate": 4.613329661690171e-05, "loss": 0.2525, "step": 6042 }, { "epoch": 37.53582554517134, "grad_norm": 4.03125, "learning_rate": 4.613198027004818e-05, "loss": 0.3588, "step": 6043 }, { "epoch": 37.54205607476636, "grad_norm": 2.421875, "learning_rate": 4.613066371795663e-05, "loss": 0.2289, "step": 6044 }, { "epoch": 37.54828660436137, "grad_norm": 1.6328125, "learning_rate": 4.6129346960639864e-05, "loss": 0.136, "step": 6045 }, { "epoch": 37.55451713395639, "grad_norm": 1.921875, "learning_rate": 4.612802999811065e-05, "loss": 0.1455, "step": 6046 }, { "epoch": 37.5607476635514, "grad_norm": 2.890625, "learning_rate": 4.6126712830381777e-05, "loss": 0.3307, "step": 6047 }, { "epoch": 37.56697819314642, "grad_norm": 2.140625, "learning_rate": 4.612539545746606e-05, "loss": 0.1391, "step": 6048 }, { "epoch": 37.573208722741434, "grad_norm": 1.484375, "learning_rate": 4.612407787937627e-05, "loss": 0.1411, "step": 6049 }, { "epoch": 37.57943925233645, "grad_norm": 3.171875, "learning_rate": 4.612276009612522e-05, "loss": 0.241, "step": 6050 }, { "epoch": 37.585669781931465, "grad_norm": 3.03125, "learning_rate": 4.6121442107725694e-05, "loss": 0.3058, "step": 6051 }, { "epoch": 37.59190031152648, "grad_norm": 3.6875, "learning_rate": 4.6120123914190505e-05, "loss": 0.4112, "step": 6052 }, { "epoch": 37.598130841121495, "grad_norm": 3.8125, "learning_rate": 4.611880551553245e-05, "loss": 0.2561, "step": 6053 }, { "epoch": 37.60436137071651, "grad_norm": 3.453125, "learning_rate": 4.611748691176433e-05, "loss": 0.1648, "step": 6054 }, { "epoch": 37.610591900311526, "grad_norm": 2.609375, "learning_rate": 4.611616810289896e-05, "loss": 0.1725, "step": 6055 }, { "epoch": 37.61682242990654, "grad_norm": 2.125, "learning_rate": 4.611484908894914e-05, "loss": 0.1282, "step": 6056 }, { "epoch": 37.62305295950156, "grad_norm": 3.671875, "learning_rate": 4.6113529869927696e-05, "loss": 0.1901, "step": 6057 }, { "epoch": 37.62928348909657, "grad_norm": 3.765625, "learning_rate": 4.611221044584742e-05, "loss": 0.2516, "step": 6058 }, { "epoch": 37.63551401869159, "grad_norm": 3.78125, "learning_rate": 4.611089081672113e-05, "loss": 0.2938, "step": 6059 }, { "epoch": 37.6417445482866, "grad_norm": 2.203125, "learning_rate": 4.610957098256166e-05, "loss": 0.1262, "step": 6060 }, { "epoch": 37.64797507788162, "grad_norm": 2.234375, "learning_rate": 4.610825094338182e-05, "loss": 0.1898, "step": 6061 }, { "epoch": 37.654205607476634, "grad_norm": 2.90625, "learning_rate": 4.610693069919442e-05, "loss": 0.1182, "step": 6062 }, { "epoch": 37.66043613707165, "grad_norm": 4.375, "learning_rate": 4.6105610250012286e-05, "loss": 0.2315, "step": 6063 }, { "epoch": 37.666666666666664, "grad_norm": 4.15625, "learning_rate": 4.610428959584825e-05, "loss": 0.204, "step": 6064 }, { "epoch": 37.67289719626168, "grad_norm": 3.546875, "learning_rate": 4.6102968736715134e-05, "loss": 0.195, "step": 6065 }, { "epoch": 37.679127725856695, "grad_norm": 3.671875, "learning_rate": 4.610164767262577e-05, "loss": 0.3311, "step": 6066 }, { "epoch": 37.68535825545171, "grad_norm": 4.09375, "learning_rate": 4.6100326403592986e-05, "loss": 0.25, "step": 6067 }, { "epoch": 37.691588785046726, "grad_norm": 4.3125, "learning_rate": 4.6099004929629615e-05, "loss": 0.2619, "step": 6068 }, { "epoch": 37.69781931464174, "grad_norm": 2.4375, "learning_rate": 4.609768325074848e-05, "loss": 0.2158, "step": 6069 }, { "epoch": 37.70404984423676, "grad_norm": 1.5234375, "learning_rate": 4.6096361366962436e-05, "loss": 0.1113, "step": 6070 }, { "epoch": 37.71028037383178, "grad_norm": 2.984375, "learning_rate": 4.609503927828432e-05, "loss": 0.3139, "step": 6071 }, { "epoch": 37.716510903426794, "grad_norm": 4.21875, "learning_rate": 4.609371698472696e-05, "loss": 0.141, "step": 6072 }, { "epoch": 37.72274143302181, "grad_norm": 4.625, "learning_rate": 4.60923944863032e-05, "loss": 0.2067, "step": 6073 }, { "epoch": 37.728971962616825, "grad_norm": 4.0625, "learning_rate": 4.609107178302589e-05, "loss": 0.1784, "step": 6074 }, { "epoch": 37.73520249221184, "grad_norm": 2.296875, "learning_rate": 4.608974887490788e-05, "loss": 0.1223, "step": 6075 }, { "epoch": 37.741433021806856, "grad_norm": 2.640625, "learning_rate": 4.6088425761962005e-05, "loss": 0.16, "step": 6076 }, { "epoch": 37.74766355140187, "grad_norm": 4.4375, "learning_rate": 4.6087102444201133e-05, "loss": 0.1511, "step": 6077 }, { "epoch": 37.753894080996886, "grad_norm": 7.15625, "learning_rate": 4.60857789216381e-05, "loss": 0.2528, "step": 6078 }, { "epoch": 37.7601246105919, "grad_norm": 5.125, "learning_rate": 4.608445519428577e-05, "loss": 0.3091, "step": 6079 }, { "epoch": 37.76635514018692, "grad_norm": 2.9375, "learning_rate": 4.6083131262157e-05, "loss": 0.1992, "step": 6080 }, { "epoch": 37.77258566978193, "grad_norm": 2.5, "learning_rate": 4.608180712526463e-05, "loss": 0.2136, "step": 6081 }, { "epoch": 37.77881619937695, "grad_norm": 3.9375, "learning_rate": 4.608048278362155e-05, "loss": 0.3084, "step": 6082 }, { "epoch": 37.78504672897196, "grad_norm": 3.28125, "learning_rate": 4.60791582372406e-05, "loss": 0.1552, "step": 6083 }, { "epoch": 37.79127725856698, "grad_norm": 5.34375, "learning_rate": 4.6077833486134656e-05, "loss": 0.2332, "step": 6084 }, { "epoch": 37.797507788161994, "grad_norm": 2.671875, "learning_rate": 4.6076508530316575e-05, "loss": 0.1662, "step": 6085 }, { "epoch": 37.80373831775701, "grad_norm": 1.390625, "learning_rate": 4.607518336979923e-05, "loss": 0.1301, "step": 6086 }, { "epoch": 37.809968847352025, "grad_norm": 2.328125, "learning_rate": 4.60738580045955e-05, "loss": 0.1677, "step": 6087 }, { "epoch": 37.81619937694704, "grad_norm": 1.984375, "learning_rate": 4.607253243471824e-05, "loss": 0.14, "step": 6088 }, { "epoch": 37.822429906542055, "grad_norm": 2.25, "learning_rate": 4.6071206660180335e-05, "loss": 0.1471, "step": 6089 }, { "epoch": 37.82866043613707, "grad_norm": 2.953125, "learning_rate": 4.606988068099466e-05, "loss": 0.2339, "step": 6090 }, { "epoch": 37.834890965732086, "grad_norm": 2.359375, "learning_rate": 4.6068554497174095e-05, "loss": 0.1684, "step": 6091 }, { "epoch": 37.8411214953271, "grad_norm": 3.140625, "learning_rate": 4.606722810873151e-05, "loss": 0.1912, "step": 6092 }, { "epoch": 37.84735202492212, "grad_norm": 2.6875, "learning_rate": 4.6065901515679796e-05, "loss": 0.1954, "step": 6093 }, { "epoch": 37.85358255451713, "grad_norm": 1.640625, "learning_rate": 4.606457471803184e-05, "loss": 0.1284, "step": 6094 }, { "epoch": 37.85981308411215, "grad_norm": 2.203125, "learning_rate": 4.6063247715800525e-05, "loss": 0.2012, "step": 6095 }, { "epoch": 37.86604361370716, "grad_norm": 2.296875, "learning_rate": 4.6061920508998735e-05, "loss": 0.1957, "step": 6096 }, { "epoch": 37.87227414330218, "grad_norm": 2.671875, "learning_rate": 4.6060593097639356e-05, "loss": 0.1339, "step": 6097 }, { "epoch": 37.87850467289719, "grad_norm": 2.578125, "learning_rate": 4.605926548173529e-05, "loss": 0.1247, "step": 6098 }, { "epoch": 37.88473520249221, "grad_norm": 2.109375, "learning_rate": 4.605793766129943e-05, "loss": 0.1763, "step": 6099 }, { "epoch": 37.890965732087224, "grad_norm": 2.359375, "learning_rate": 4.605660963634467e-05, "loss": 0.1577, "step": 6100 }, { "epoch": 37.89719626168224, "grad_norm": 2.109375, "learning_rate": 4.605528140688391e-05, "loss": 0.1808, "step": 6101 }, { "epoch": 37.90342679127726, "grad_norm": 3.0, "learning_rate": 4.605395297293005e-05, "loss": 0.2461, "step": 6102 }, { "epoch": 37.90965732087228, "grad_norm": 2.453125, "learning_rate": 4.605262433449599e-05, "loss": 0.205, "step": 6103 }, { "epoch": 37.91588785046729, "grad_norm": 3.25, "learning_rate": 4.605129549159462e-05, "loss": 0.1462, "step": 6104 }, { "epoch": 37.92211838006231, "grad_norm": 2.640625, "learning_rate": 4.604996644423888e-05, "loss": 0.2622, "step": 6105 }, { "epoch": 37.92834890965732, "grad_norm": 2.953125, "learning_rate": 4.604863719244164e-05, "loss": 0.1752, "step": 6106 }, { "epoch": 37.93457943925234, "grad_norm": 3.875, "learning_rate": 4.6047307736215845e-05, "loss": 0.2235, "step": 6107 }, { "epoch": 37.940809968847354, "grad_norm": 1.796875, "learning_rate": 4.604597807557438e-05, "loss": 0.2348, "step": 6108 }, { "epoch": 37.94704049844237, "grad_norm": 3.515625, "learning_rate": 4.6044648210530175e-05, "loss": 0.2363, "step": 6109 }, { "epoch": 37.953271028037385, "grad_norm": 2.015625, "learning_rate": 4.6043318141096135e-05, "loss": 0.1558, "step": 6110 }, { "epoch": 37.9595015576324, "grad_norm": 3.96875, "learning_rate": 4.604198786728518e-05, "loss": 0.1284, "step": 6111 }, { "epoch": 37.965732087227416, "grad_norm": 2.75, "learning_rate": 4.6040657389110236e-05, "loss": 0.2112, "step": 6112 }, { "epoch": 37.97196261682243, "grad_norm": 3.375, "learning_rate": 4.603932670658422e-05, "loss": 0.1519, "step": 6113 }, { "epoch": 37.978193146417446, "grad_norm": 3.90625, "learning_rate": 4.603799581972007e-05, "loss": 0.332, "step": 6114 }, { "epoch": 37.98442367601246, "grad_norm": 3.109375, "learning_rate": 4.603666472853068e-05, "loss": 0.2223, "step": 6115 }, { "epoch": 37.99065420560748, "grad_norm": 3.75, "learning_rate": 4.6035333433029006e-05, "loss": 0.3945, "step": 6116 }, { "epoch": 37.99688473520249, "grad_norm": 2.6875, "learning_rate": 4.6034001933227965e-05, "loss": 0.225, "step": 6117 }, { "epoch": 38.0, "grad_norm": 1.4375, "learning_rate": 4.603267022914049e-05, "loss": 0.0507, "step": 6118 }, { "epoch": 38.006230529595015, "grad_norm": 2.296875, "learning_rate": 4.6031338320779534e-05, "loss": 0.156, "step": 6119 }, { "epoch": 38.01246105919003, "grad_norm": 2.890625, "learning_rate": 4.6030006208158e-05, "loss": 0.233, "step": 6120 }, { "epoch": 38.018691588785046, "grad_norm": 4.8125, "learning_rate": 4.602867389128885e-05, "loss": 0.4863, "step": 6121 }, { "epoch": 38.02492211838006, "grad_norm": 4.125, "learning_rate": 4.6027341370185014e-05, "loss": 0.2313, "step": 6122 }, { "epoch": 38.03115264797508, "grad_norm": 2.609375, "learning_rate": 4.602600864485943e-05, "loss": 0.1718, "step": 6123 }, { "epoch": 38.03738317757009, "grad_norm": 2.25, "learning_rate": 4.602467571532505e-05, "loss": 0.1407, "step": 6124 }, { "epoch": 38.04361370716511, "grad_norm": 2.34375, "learning_rate": 4.602334258159482e-05, "loss": 0.1833, "step": 6125 }, { "epoch": 38.04984423676012, "grad_norm": 2.671875, "learning_rate": 4.602200924368169e-05, "loss": 0.2421, "step": 6126 }, { "epoch": 38.05607476635514, "grad_norm": 2.859375, "learning_rate": 4.602067570159859e-05, "loss": 0.2103, "step": 6127 }, { "epoch": 38.06230529595015, "grad_norm": 2.21875, "learning_rate": 4.6019341955358494e-05, "loss": 0.2205, "step": 6128 }, { "epoch": 38.06853582554517, "grad_norm": 2.109375, "learning_rate": 4.601800800497435e-05, "loss": 0.1418, "step": 6129 }, { "epoch": 38.074766355140184, "grad_norm": 2.40625, "learning_rate": 4.60166738504591e-05, "loss": 0.1568, "step": 6130 }, { "epoch": 38.0809968847352, "grad_norm": 2.84375, "learning_rate": 4.6015339491825716e-05, "loss": 0.2084, "step": 6131 }, { "epoch": 38.087227414330215, "grad_norm": 3.25, "learning_rate": 4.601400492908716e-05, "loss": 0.3412, "step": 6132 }, { "epoch": 38.09345794392523, "grad_norm": 3.0625, "learning_rate": 4.6012670162256376e-05, "loss": 0.1979, "step": 6133 }, { "epoch": 38.099688473520246, "grad_norm": 2.765625, "learning_rate": 4.601133519134635e-05, "loss": 0.1654, "step": 6134 }, { "epoch": 38.10591900311527, "grad_norm": 2.671875, "learning_rate": 4.601000001637004e-05, "loss": 0.1954, "step": 6135 }, { "epoch": 38.11214953271028, "grad_norm": 3.09375, "learning_rate": 4.60086646373404e-05, "loss": 0.2187, "step": 6136 }, { "epoch": 38.1183800623053, "grad_norm": 3.25, "learning_rate": 4.600732905427041e-05, "loss": 0.1152, "step": 6137 }, { "epoch": 38.124610591900314, "grad_norm": 3.359375, "learning_rate": 4.6005993267173044e-05, "loss": 0.2522, "step": 6138 }, { "epoch": 38.13084112149533, "grad_norm": 2.640625, "learning_rate": 4.600465727606128e-05, "loss": 0.123, "step": 6139 }, { "epoch": 38.137071651090345, "grad_norm": 3.03125, "learning_rate": 4.600332108094807e-05, "loss": 0.2237, "step": 6140 }, { "epoch": 38.14330218068536, "grad_norm": 4.1875, "learning_rate": 4.600198468184642e-05, "loss": 0.5504, "step": 6141 }, { "epoch": 38.149532710280376, "grad_norm": 2.625, "learning_rate": 4.600064807876929e-05, "loss": 0.1557, "step": 6142 }, { "epoch": 38.15576323987539, "grad_norm": 2.984375, "learning_rate": 4.599931127172967e-05, "loss": 0.1916, "step": 6143 }, { "epoch": 38.161993769470406, "grad_norm": 2.375, "learning_rate": 4.599797426074054e-05, "loss": 0.1512, "step": 6144 }, { "epoch": 38.16822429906542, "grad_norm": 3.21875, "learning_rate": 4.599663704581488e-05, "loss": 0.3585, "step": 6145 }, { "epoch": 38.17445482866044, "grad_norm": 1.890625, "learning_rate": 4.59952996269657e-05, "loss": 0.154, "step": 6146 }, { "epoch": 38.18068535825545, "grad_norm": 3.46875, "learning_rate": 4.599396200420596e-05, "loss": 0.202, "step": 6147 }, { "epoch": 38.18691588785047, "grad_norm": 2.34375, "learning_rate": 4.599262417754867e-05, "loss": 0.2373, "step": 6148 }, { "epoch": 38.19314641744548, "grad_norm": 3.3125, "learning_rate": 4.599128614700682e-05, "loss": 0.2832, "step": 6149 }, { "epoch": 38.1993769470405, "grad_norm": 3.390625, "learning_rate": 4.59899479125934e-05, "loss": 0.305, "step": 6150 }, { "epoch": 38.205607476635514, "grad_norm": 2.53125, "learning_rate": 4.598860947432141e-05, "loss": 0.1882, "step": 6151 }, { "epoch": 38.21183800623053, "grad_norm": 2.15625, "learning_rate": 4.5987270832203844e-05, "loss": 0.1834, "step": 6152 }, { "epoch": 38.218068535825545, "grad_norm": 2.40625, "learning_rate": 4.5985931986253716e-05, "loss": 0.2375, "step": 6153 }, { "epoch": 38.22429906542056, "grad_norm": 1.734375, "learning_rate": 4.5984592936484016e-05, "loss": 0.1391, "step": 6154 }, { "epoch": 38.230529595015575, "grad_norm": 2.40625, "learning_rate": 4.598325368290776e-05, "loss": 0.1701, "step": 6155 }, { "epoch": 38.23676012461059, "grad_norm": 2.734375, "learning_rate": 4.5981914225537944e-05, "loss": 0.24, "step": 6156 }, { "epoch": 38.242990654205606, "grad_norm": 2.828125, "learning_rate": 4.598057456438759e-05, "loss": 0.187, "step": 6157 }, { "epoch": 38.24922118380062, "grad_norm": 1.96875, "learning_rate": 4.59792346994697e-05, "loss": 0.1709, "step": 6158 }, { "epoch": 38.25545171339564, "grad_norm": 2.328125, "learning_rate": 4.597789463079728e-05, "loss": 0.2039, "step": 6159 }, { "epoch": 38.26168224299065, "grad_norm": 3.609375, "learning_rate": 4.5976554358383364e-05, "loss": 0.3938, "step": 6160 }, { "epoch": 38.26791277258567, "grad_norm": 1.7109375, "learning_rate": 4.597521388224095e-05, "loss": 0.1553, "step": 6161 }, { "epoch": 38.27414330218068, "grad_norm": 2.875, "learning_rate": 4.5973873202383076e-05, "loss": 0.2139, "step": 6162 }, { "epoch": 38.2803738317757, "grad_norm": 2.703125, "learning_rate": 4.597253231882275e-05, "loss": 0.1976, "step": 6163 }, { "epoch": 38.28660436137071, "grad_norm": 3.71875, "learning_rate": 4.5971191231572994e-05, "loss": 0.2292, "step": 6164 }, { "epoch": 38.29283489096573, "grad_norm": 2.546875, "learning_rate": 4.5969849940646834e-05, "loss": 0.1497, "step": 6165 }, { "epoch": 38.299065420560744, "grad_norm": 1.453125, "learning_rate": 4.5968508446057315e-05, "loss": 0.1416, "step": 6166 }, { "epoch": 38.30529595015577, "grad_norm": 1.71875, "learning_rate": 4.596716674781743e-05, "loss": 0.1534, "step": 6167 }, { "epoch": 38.31152647975078, "grad_norm": 2.34375, "learning_rate": 4.596582484594024e-05, "loss": 0.1447, "step": 6168 }, { "epoch": 38.3177570093458, "grad_norm": 2.484375, "learning_rate": 4.596448274043877e-05, "loss": 0.1897, "step": 6169 }, { "epoch": 38.32398753894081, "grad_norm": 2.953125, "learning_rate": 4.596314043132606e-05, "loss": 0.2673, "step": 6170 }, { "epoch": 38.33021806853583, "grad_norm": 3.078125, "learning_rate": 4.596179791861512e-05, "loss": 0.1647, "step": 6171 }, { "epoch": 38.33644859813084, "grad_norm": 2.359375, "learning_rate": 4.596045520231902e-05, "loss": 0.2157, "step": 6172 }, { "epoch": 38.34267912772586, "grad_norm": 3.578125, "learning_rate": 4.595911228245079e-05, "loss": 0.1592, "step": 6173 }, { "epoch": 38.348909657320874, "grad_norm": 3.125, "learning_rate": 4.5957769159023474e-05, "loss": 0.1919, "step": 6174 }, { "epoch": 38.35514018691589, "grad_norm": 3.515625, "learning_rate": 4.595642583205011e-05, "loss": 0.2061, "step": 6175 }, { "epoch": 38.361370716510905, "grad_norm": 4.15625, "learning_rate": 4.595508230154375e-05, "loss": 0.4068, "step": 6176 }, { "epoch": 38.36760124610592, "grad_norm": 2.375, "learning_rate": 4.5953738567517446e-05, "loss": 0.1231, "step": 6177 }, { "epoch": 38.373831775700936, "grad_norm": 4.25, "learning_rate": 4.595239462998424e-05, "loss": 0.2883, "step": 6178 }, { "epoch": 38.38006230529595, "grad_norm": 5.15625, "learning_rate": 4.5951050488957187e-05, "loss": 0.247, "step": 6179 }, { "epoch": 38.386292834890966, "grad_norm": 1.8125, "learning_rate": 4.594970614444935e-05, "loss": 0.1526, "step": 6180 }, { "epoch": 38.39252336448598, "grad_norm": 1.5234375, "learning_rate": 4.594836159647377e-05, "loss": 0.1301, "step": 6181 }, { "epoch": 38.398753894081, "grad_norm": 3.296875, "learning_rate": 4.594701684504352e-05, "loss": 0.2062, "step": 6182 }, { "epoch": 38.40498442367601, "grad_norm": 2.9375, "learning_rate": 4.594567189017166e-05, "loss": 0.2544, "step": 6183 }, { "epoch": 38.41121495327103, "grad_norm": 2.375, "learning_rate": 4.594432673187124e-05, "loss": 0.1629, "step": 6184 }, { "epoch": 38.41744548286604, "grad_norm": 2.4375, "learning_rate": 4.5942981370155334e-05, "loss": 0.1958, "step": 6185 }, { "epoch": 38.42367601246106, "grad_norm": 2.15625, "learning_rate": 4.594163580503701e-05, "loss": 0.1592, "step": 6186 }, { "epoch": 38.429906542056074, "grad_norm": 2.203125, "learning_rate": 4.594029003652933e-05, "loss": 0.1631, "step": 6187 }, { "epoch": 38.43613707165109, "grad_norm": 3.359375, "learning_rate": 4.593894406464537e-05, "loss": 0.374, "step": 6188 }, { "epoch": 38.442367601246104, "grad_norm": 2.921875, "learning_rate": 4.593759788939819e-05, "loss": 0.2255, "step": 6189 }, { "epoch": 38.44859813084112, "grad_norm": 2.109375, "learning_rate": 4.593625151080089e-05, "loss": 0.1464, "step": 6190 }, { "epoch": 38.454828660436135, "grad_norm": 2.46875, "learning_rate": 4.593490492886652e-05, "loss": 0.2327, "step": 6191 }, { "epoch": 38.46105919003115, "grad_norm": 2.84375, "learning_rate": 4.593355814360817e-05, "loss": 0.2545, "step": 6192 }, { "epoch": 38.467289719626166, "grad_norm": 3.1875, "learning_rate": 4.5932211155038916e-05, "loss": 0.2403, "step": 6193 }, { "epoch": 38.47352024922118, "grad_norm": 3.109375, "learning_rate": 4.593086396317185e-05, "loss": 0.1523, "step": 6194 }, { "epoch": 38.4797507788162, "grad_norm": 3.28125, "learning_rate": 4.5929516568020035e-05, "loss": 0.3429, "step": 6195 }, { "epoch": 38.48598130841121, "grad_norm": 3.359375, "learning_rate": 4.5928168969596586e-05, "loss": 0.3162, "step": 6196 }, { "epoch": 38.49221183800623, "grad_norm": 1.75, "learning_rate": 4.592682116791457e-05, "loss": 0.1355, "step": 6197 }, { "epoch": 38.49844236760124, "grad_norm": 2.15625, "learning_rate": 4.592547316298709e-05, "loss": 0.1292, "step": 6198 }, { "epoch": 38.504672897196265, "grad_norm": 3.4375, "learning_rate": 4.592412495482722e-05, "loss": 0.2071, "step": 6199 }, { "epoch": 38.51090342679128, "grad_norm": 2.390625, "learning_rate": 4.592277654344807e-05, "loss": 0.1924, "step": 6200 }, { "epoch": 38.517133956386296, "grad_norm": 1.578125, "learning_rate": 4.5921427928862745e-05, "loss": 0.132, "step": 6201 }, { "epoch": 38.52336448598131, "grad_norm": 2.921875, "learning_rate": 4.5920079111084316e-05, "loss": 0.2327, "step": 6202 }, { "epoch": 38.52959501557633, "grad_norm": 2.546875, "learning_rate": 4.5918730090125905e-05, "loss": 0.1716, "step": 6203 }, { "epoch": 38.53582554517134, "grad_norm": 2.234375, "learning_rate": 4.5917380866000604e-05, "loss": 0.1243, "step": 6204 }, { "epoch": 38.54205607476636, "grad_norm": 2.75, "learning_rate": 4.5916031438721505e-05, "loss": 0.178, "step": 6205 }, { "epoch": 38.54828660436137, "grad_norm": 2.15625, "learning_rate": 4.591468180830174e-05, "loss": 0.1527, "step": 6206 }, { "epoch": 38.55451713395639, "grad_norm": 2.578125, "learning_rate": 4.59133319747544e-05, "loss": 0.2405, "step": 6207 }, { "epoch": 38.5607476635514, "grad_norm": 2.609375, "learning_rate": 4.5911981938092606e-05, "loss": 0.2145, "step": 6208 }, { "epoch": 38.56697819314642, "grad_norm": 1.84375, "learning_rate": 4.591063169832947e-05, "loss": 0.178, "step": 6209 }, { "epoch": 38.573208722741434, "grad_norm": 3.40625, "learning_rate": 4.590928125547808e-05, "loss": 0.2479, "step": 6210 }, { "epoch": 38.57943925233645, "grad_norm": 1.765625, "learning_rate": 4.5907930609551584e-05, "loss": 0.1651, "step": 6211 }, { "epoch": 38.585669781931465, "grad_norm": 3.578125, "learning_rate": 4.590657976056308e-05, "loss": 0.3394, "step": 6212 }, { "epoch": 38.59190031152648, "grad_norm": 3.84375, "learning_rate": 4.590522870852569e-05, "loss": 0.3083, "step": 6213 }, { "epoch": 38.598130841121495, "grad_norm": 2.3125, "learning_rate": 4.590387745345255e-05, "loss": 0.1911, "step": 6214 }, { "epoch": 38.60436137071651, "grad_norm": 3.21875, "learning_rate": 4.590252599535677e-05, "loss": 0.2581, "step": 6215 }, { "epoch": 38.610591900311526, "grad_norm": 1.671875, "learning_rate": 4.590117433425148e-05, "loss": 0.1428, "step": 6216 }, { "epoch": 38.61682242990654, "grad_norm": 3.359375, "learning_rate": 4.589982247014981e-05, "loss": 0.2986, "step": 6217 }, { "epoch": 38.62305295950156, "grad_norm": 2.484375, "learning_rate": 4.5898470403064876e-05, "loss": 0.1876, "step": 6218 }, { "epoch": 38.62928348909657, "grad_norm": 3.078125, "learning_rate": 4.589711813300983e-05, "loss": 0.32, "step": 6219 }, { "epoch": 38.63551401869159, "grad_norm": 3.84375, "learning_rate": 4.589576565999779e-05, "loss": 0.3478, "step": 6220 }, { "epoch": 38.6417445482866, "grad_norm": 2.828125, "learning_rate": 4.58944129840419e-05, "loss": 0.1911, "step": 6221 }, { "epoch": 38.64797507788162, "grad_norm": 1.9765625, "learning_rate": 4.589306010515529e-05, "loss": 0.1843, "step": 6222 }, { "epoch": 38.654205607476634, "grad_norm": 2.390625, "learning_rate": 4.5891707023351105e-05, "loss": 0.1996, "step": 6223 }, { "epoch": 38.66043613707165, "grad_norm": 2.78125, "learning_rate": 4.5890353738642486e-05, "loss": 0.288, "step": 6224 }, { "epoch": 38.666666666666664, "grad_norm": 2.53125, "learning_rate": 4.588900025104258e-05, "loss": 0.1836, "step": 6225 }, { "epoch": 38.67289719626168, "grad_norm": 1.3671875, "learning_rate": 4.5887646560564524e-05, "loss": 0.1318, "step": 6226 }, { "epoch": 38.679127725856695, "grad_norm": 1.9375, "learning_rate": 4.588629266722147e-05, "loss": 0.1443, "step": 6227 }, { "epoch": 38.68535825545171, "grad_norm": 3.046875, "learning_rate": 4.588493857102657e-05, "loss": 0.1826, "step": 6228 }, { "epoch": 38.691588785046726, "grad_norm": 3.1875, "learning_rate": 4.5883584271992964e-05, "loss": 0.202, "step": 6229 }, { "epoch": 38.69781931464174, "grad_norm": 3.328125, "learning_rate": 4.588222977013382e-05, "loss": 0.2444, "step": 6230 }, { "epoch": 38.70404984423676, "grad_norm": 2.953125, "learning_rate": 4.588087506546228e-05, "loss": 0.2385, "step": 6231 }, { "epoch": 38.71028037383178, "grad_norm": 3.15625, "learning_rate": 4.587952015799151e-05, "loss": 0.2283, "step": 6232 }, { "epoch": 38.716510903426794, "grad_norm": 2.546875, "learning_rate": 4.5878165047734665e-05, "loss": 0.2527, "step": 6233 }, { "epoch": 38.72274143302181, "grad_norm": 2.4375, "learning_rate": 4.587680973470491e-05, "loss": 0.2877, "step": 6234 }, { "epoch": 38.728971962616825, "grad_norm": 4.4375, "learning_rate": 4.587545421891541e-05, "loss": 0.2574, "step": 6235 }, { "epoch": 38.73520249221184, "grad_norm": 2.5625, "learning_rate": 4.587409850037932e-05, "loss": 0.3099, "step": 6236 }, { "epoch": 38.741433021806856, "grad_norm": 3.265625, "learning_rate": 4.587274257910981e-05, "loss": 0.2653, "step": 6237 }, { "epoch": 38.74766355140187, "grad_norm": 1.4765625, "learning_rate": 4.5871386455120057e-05, "loss": 0.1403, "step": 6238 }, { "epoch": 38.753894080996886, "grad_norm": 2.890625, "learning_rate": 4.587003012842323e-05, "loss": 0.2107, "step": 6239 }, { "epoch": 38.7601246105919, "grad_norm": 2.375, "learning_rate": 4.5868673599032484e-05, "loss": 0.2372, "step": 6240 }, { "epoch": 38.76635514018692, "grad_norm": 2.40625, "learning_rate": 4.586731686696102e-05, "loss": 0.1961, "step": 6241 }, { "epoch": 38.77258566978193, "grad_norm": 2.578125, "learning_rate": 4.5865959932222004e-05, "loss": 0.182, "step": 6242 }, { "epoch": 38.77881619937695, "grad_norm": 1.765625, "learning_rate": 4.586460279482861e-05, "loss": 0.1117, "step": 6243 }, { "epoch": 38.78504672897196, "grad_norm": 2.71875, "learning_rate": 4.586324545479403e-05, "loss": 0.241, "step": 6244 }, { "epoch": 38.79127725856698, "grad_norm": 3.6875, "learning_rate": 4.586188791213143e-05, "loss": 0.1599, "step": 6245 }, { "epoch": 38.797507788161994, "grad_norm": 3.8125, "learning_rate": 4.586053016685401e-05, "loss": 0.1681, "step": 6246 }, { "epoch": 38.80373831775701, "grad_norm": 2.453125, "learning_rate": 4.585917221897495e-05, "loss": 0.1714, "step": 6247 }, { "epoch": 38.809968847352025, "grad_norm": 2.765625, "learning_rate": 4.585781406850744e-05, "loss": 0.1774, "step": 6248 }, { "epoch": 38.81619937694704, "grad_norm": 2.828125, "learning_rate": 4.585645571546467e-05, "loss": 0.3124, "step": 6249 }, { "epoch": 38.822429906542055, "grad_norm": 2.390625, "learning_rate": 4.585509715985983e-05, "loss": 0.2731, "step": 6250 }, { "epoch": 38.82866043613707, "grad_norm": 2.171875, "learning_rate": 4.585373840170612e-05, "loss": 0.1686, "step": 6251 }, { "epoch": 38.834890965732086, "grad_norm": 3.171875, "learning_rate": 4.5852379441016734e-05, "loss": 0.1762, "step": 6252 }, { "epoch": 38.8411214953271, "grad_norm": 1.9296875, "learning_rate": 4.5851020277804865e-05, "loss": 0.1643, "step": 6253 }, { "epoch": 38.84735202492212, "grad_norm": 1.5859375, "learning_rate": 4.584966091208373e-05, "loss": 0.1146, "step": 6254 }, { "epoch": 38.85358255451713, "grad_norm": 2.828125, "learning_rate": 4.584830134386652e-05, "loss": 0.1762, "step": 6255 }, { "epoch": 38.85981308411215, "grad_norm": 2.640625, "learning_rate": 4.584694157316643e-05, "loss": 0.177, "step": 6256 }, { "epoch": 38.86604361370716, "grad_norm": 1.625, "learning_rate": 4.584558159999668e-05, "loss": 0.1569, "step": 6257 }, { "epoch": 38.87227414330218, "grad_norm": 2.328125, "learning_rate": 4.584422142437048e-05, "loss": 0.1611, "step": 6258 }, { "epoch": 38.87850467289719, "grad_norm": 1.3671875, "learning_rate": 4.584286104630103e-05, "loss": 0.1459, "step": 6259 }, { "epoch": 38.88473520249221, "grad_norm": 2.953125, "learning_rate": 4.584150046580155e-05, "loss": 0.1814, "step": 6260 }, { "epoch": 38.890965732087224, "grad_norm": 2.3125, "learning_rate": 4.584013968288524e-05, "loss": 0.1785, "step": 6261 }, { "epoch": 38.89719626168224, "grad_norm": 1.8515625, "learning_rate": 4.583877869756534e-05, "loss": 0.1879, "step": 6262 }, { "epoch": 38.90342679127726, "grad_norm": 1.6875, "learning_rate": 4.583741750985505e-05, "loss": 0.1734, "step": 6263 }, { "epoch": 38.90965732087228, "grad_norm": 1.953125, "learning_rate": 4.58360561197676e-05, "loss": 0.1647, "step": 6264 }, { "epoch": 38.91588785046729, "grad_norm": 2.6875, "learning_rate": 4.583469452731621e-05, "loss": 0.253, "step": 6265 }, { "epoch": 38.92211838006231, "grad_norm": 2.578125, "learning_rate": 4.58333327325141e-05, "loss": 0.2541, "step": 6266 }, { "epoch": 38.92834890965732, "grad_norm": 1.53125, "learning_rate": 4.583197073537449e-05, "loss": 0.1412, "step": 6267 }, { "epoch": 38.93457943925234, "grad_norm": 1.8203125, "learning_rate": 4.583060853591063e-05, "loss": 0.2028, "step": 6268 }, { "epoch": 38.940809968847354, "grad_norm": 1.953125, "learning_rate": 4.582924613413572e-05, "loss": 0.1545, "step": 6269 }, { "epoch": 38.94704049844237, "grad_norm": 1.8125, "learning_rate": 4.582788353006302e-05, "loss": 0.1672, "step": 6270 }, { "epoch": 38.953271028037385, "grad_norm": 3.265625, "learning_rate": 4.5826520723705743e-05, "loss": 0.1766, "step": 6271 }, { "epoch": 38.9595015576324, "grad_norm": 3.375, "learning_rate": 4.582515771507714e-05, "loss": 0.2874, "step": 6272 }, { "epoch": 38.965732087227416, "grad_norm": 3.28125, "learning_rate": 4.582379450419044e-05, "loss": 0.3512, "step": 6273 }, { "epoch": 38.97196261682243, "grad_norm": 3.296875, "learning_rate": 4.582243109105889e-05, "loss": 0.2821, "step": 6274 }, { "epoch": 38.978193146417446, "grad_norm": 2.09375, "learning_rate": 4.582106747569572e-05, "loss": 0.1631, "step": 6275 }, { "epoch": 38.98442367601246, "grad_norm": 3.328125, "learning_rate": 4.581970365811419e-05, "loss": 0.3265, "step": 6276 }, { "epoch": 38.99065420560748, "grad_norm": 2.390625, "learning_rate": 4.581833963832752e-05, "loss": 0.1715, "step": 6277 }, { "epoch": 38.99688473520249, "grad_norm": 1.8125, "learning_rate": 4.5816975416348986e-05, "loss": 0.146, "step": 6278 }, { "epoch": 39.0, "grad_norm": 1.6171875, "learning_rate": 4.5815610992191826e-05, "loss": 0.1169, "step": 6279 }, { "epoch": 39.006230529595015, "grad_norm": 3.28125, "learning_rate": 4.581424636586929e-05, "loss": 0.1716, "step": 6280 }, { "epoch": 39.01246105919003, "grad_norm": 2.328125, "learning_rate": 4.581288153739462e-05, "loss": 0.2665, "step": 6281 }, { "epoch": 39.018691588785046, "grad_norm": 3.484375, "learning_rate": 4.5811516506781104e-05, "loss": 0.2731, "step": 6282 }, { "epoch": 39.02492211838006, "grad_norm": 1.625, "learning_rate": 4.581015127404197e-05, "loss": 0.1759, "step": 6283 }, { "epoch": 39.03115264797508, "grad_norm": 3.1875, "learning_rate": 4.5808785839190486e-05, "loss": 0.1975, "step": 6284 }, { "epoch": 39.03738317757009, "grad_norm": 2.71875, "learning_rate": 4.580742020223992e-05, "loss": 0.1191, "step": 6285 }, { "epoch": 39.04361370716511, "grad_norm": 2.390625, "learning_rate": 4.580605436320352e-05, "loss": 0.2579, "step": 6286 }, { "epoch": 39.04984423676012, "grad_norm": 3.625, "learning_rate": 4.580468832209457e-05, "loss": 0.2069, "step": 6287 }, { "epoch": 39.05607476635514, "grad_norm": 1.8515625, "learning_rate": 4.580332207892632e-05, "loss": 0.2607, "step": 6288 }, { "epoch": 39.06230529595015, "grad_norm": 3.09375, "learning_rate": 4.580195563371206e-05, "loss": 0.1958, "step": 6289 }, { "epoch": 39.06853582554517, "grad_norm": 3.9375, "learning_rate": 4.580058898646503e-05, "loss": 0.221, "step": 6290 }, { "epoch": 39.074766355140184, "grad_norm": 4.25, "learning_rate": 4.579922213719854e-05, "loss": 0.2131, "step": 6291 }, { "epoch": 39.0809968847352, "grad_norm": 4.28125, "learning_rate": 4.5797855085925835e-05, "loss": 0.3417, "step": 6292 }, { "epoch": 39.087227414330215, "grad_norm": 2.5625, "learning_rate": 4.579648783266021e-05, "loss": 0.1677, "step": 6293 }, { "epoch": 39.09345794392523, "grad_norm": 3.984375, "learning_rate": 4.5795120377414936e-05, "loss": 0.1196, "step": 6294 }, { "epoch": 39.099688473520246, "grad_norm": 5.1875, "learning_rate": 4.57937527202033e-05, "loss": 0.284, "step": 6295 }, { "epoch": 39.10591900311527, "grad_norm": 5.0625, "learning_rate": 4.579238486103857e-05, "loss": 0.1561, "step": 6296 }, { "epoch": 39.11214953271028, "grad_norm": 3.578125, "learning_rate": 4.579101679993405e-05, "loss": 0.2018, "step": 6297 }, { "epoch": 39.1183800623053, "grad_norm": 3.15625, "learning_rate": 4.5789648536903025e-05, "loss": 0.1139, "step": 6298 }, { "epoch": 39.124610591900314, "grad_norm": 1.671875, "learning_rate": 4.5788280071958764e-05, "loss": 0.1603, "step": 6299 }, { "epoch": 39.13084112149533, "grad_norm": 6.4375, "learning_rate": 4.5786911405114585e-05, "loss": 0.3642, "step": 6300 }, { "epoch": 39.137071651090345, "grad_norm": 6.40625, "learning_rate": 4.578554253638376e-05, "loss": 0.1844, "step": 6301 }, { "epoch": 39.14330218068536, "grad_norm": 5.8125, "learning_rate": 4.578417346577959e-05, "loss": 0.2257, "step": 6302 }, { "epoch": 39.149532710280376, "grad_norm": 2.703125, "learning_rate": 4.578280419331538e-05, "loss": 0.1386, "step": 6303 }, { "epoch": 39.15576323987539, "grad_norm": 3.28125, "learning_rate": 4.5781434719004413e-05, "loss": 0.2474, "step": 6304 }, { "epoch": 39.161993769470406, "grad_norm": 2.671875, "learning_rate": 4.578006504286e-05, "loss": 0.1874, "step": 6305 }, { "epoch": 39.16822429906542, "grad_norm": 4.375, "learning_rate": 4.5778695164895437e-05, "loss": 0.203, "step": 6306 }, { "epoch": 39.17445482866044, "grad_norm": 3.203125, "learning_rate": 4.577732508512404e-05, "loss": 0.1945, "step": 6307 }, { "epoch": 39.18068535825545, "grad_norm": 3.609375, "learning_rate": 4.577595480355911e-05, "loss": 0.1691, "step": 6308 }, { "epoch": 39.18691588785047, "grad_norm": 2.359375, "learning_rate": 4.5774584320213945e-05, "loss": 0.1949, "step": 6309 }, { "epoch": 39.19314641744548, "grad_norm": 3.046875, "learning_rate": 4.5773213635101875e-05, "loss": 0.1572, "step": 6310 }, { "epoch": 39.1993769470405, "grad_norm": 3.359375, "learning_rate": 4.577184274823619e-05, "loss": 0.295, "step": 6311 }, { "epoch": 39.205607476635514, "grad_norm": 3.953125, "learning_rate": 4.577047165963022e-05, "loss": 0.2748, "step": 6312 }, { "epoch": 39.21183800623053, "grad_norm": 3.265625, "learning_rate": 4.576910036929728e-05, "loss": 0.1459, "step": 6313 }, { "epoch": 39.218068535825545, "grad_norm": 2.21875, "learning_rate": 4.5767728877250694e-05, "loss": 0.1383, "step": 6314 }, { "epoch": 39.22429906542056, "grad_norm": 2.546875, "learning_rate": 4.5766357183503764e-05, "loss": 0.1268, "step": 6315 }, { "epoch": 39.230529595015575, "grad_norm": 3.90625, "learning_rate": 4.576498528806982e-05, "loss": 0.1658, "step": 6316 }, { "epoch": 39.23676012461059, "grad_norm": 3.4375, "learning_rate": 4.576361319096219e-05, "loss": 0.2812, "step": 6317 }, { "epoch": 39.242990654205606, "grad_norm": 4.09375, "learning_rate": 4.57622408921942e-05, "loss": 0.2387, "step": 6318 }, { "epoch": 39.24922118380062, "grad_norm": 3.109375, "learning_rate": 4.576086839177917e-05, "loss": 0.1676, "step": 6319 }, { "epoch": 39.25545171339564, "grad_norm": 2.59375, "learning_rate": 4.5759495689730446e-05, "loss": 0.2252, "step": 6320 }, { "epoch": 39.26168224299065, "grad_norm": 3.671875, "learning_rate": 4.575812278606134e-05, "loss": 0.1748, "step": 6321 }, { "epoch": 39.26791277258567, "grad_norm": 5.84375, "learning_rate": 4.5756749680785204e-05, "loss": 0.1349, "step": 6322 }, { "epoch": 39.27414330218068, "grad_norm": 5.5625, "learning_rate": 4.575537637391536e-05, "loss": 0.1763, "step": 6323 }, { "epoch": 39.2803738317757, "grad_norm": 5.0, "learning_rate": 4.5754002865465165e-05, "loss": 0.1899, "step": 6324 }, { "epoch": 39.28660436137071, "grad_norm": 1.71875, "learning_rate": 4.5752629155447926e-05, "loss": 0.1396, "step": 6325 }, { "epoch": 39.29283489096573, "grad_norm": 3.171875, "learning_rate": 4.5751255243877015e-05, "loss": 0.2156, "step": 6326 }, { "epoch": 39.299065420560744, "grad_norm": 5.3125, "learning_rate": 4.574988113076577e-05, "loss": 0.1351, "step": 6327 }, { "epoch": 39.30529595015577, "grad_norm": 5.4375, "learning_rate": 4.574850681612752e-05, "loss": 0.199, "step": 6328 }, { "epoch": 39.31152647975078, "grad_norm": 3.984375, "learning_rate": 4.574713229997563e-05, "loss": 0.225, "step": 6329 }, { "epoch": 39.3177570093458, "grad_norm": 2.71875, "learning_rate": 4.574575758232344e-05, "loss": 0.1456, "step": 6330 }, { "epoch": 39.32398753894081, "grad_norm": 2.75, "learning_rate": 4.574438266318432e-05, "loss": 0.1735, "step": 6331 }, { "epoch": 39.33021806853583, "grad_norm": 1.9296875, "learning_rate": 4.5743007542571594e-05, "loss": 0.1137, "step": 6332 }, { "epoch": 39.33644859813084, "grad_norm": 4.46875, "learning_rate": 4.574163222049863e-05, "loss": 0.1906, "step": 6333 }, { "epoch": 39.34267912772586, "grad_norm": 3.328125, "learning_rate": 4.5740256696978804e-05, "loss": 0.1509, "step": 6334 }, { "epoch": 39.348909657320874, "grad_norm": 2.875, "learning_rate": 4.573888097202544e-05, "loss": 0.1955, "step": 6335 }, { "epoch": 39.35514018691589, "grad_norm": 2.875, "learning_rate": 4.573750504565193e-05, "loss": 0.1616, "step": 6336 }, { "epoch": 39.361370716510905, "grad_norm": 3.046875, "learning_rate": 4.573612891787162e-05, "loss": 0.2945, "step": 6337 }, { "epoch": 39.36760124610592, "grad_norm": 4.59375, "learning_rate": 4.573475258869788e-05, "loss": 0.2681, "step": 6338 }, { "epoch": 39.373831775700936, "grad_norm": 3.375, "learning_rate": 4.573337605814408e-05, "loss": 0.3064, "step": 6339 }, { "epoch": 39.38006230529595, "grad_norm": 4.25, "learning_rate": 4.57319993262236e-05, "loss": 0.284, "step": 6340 }, { "epoch": 39.386292834890966, "grad_norm": 2.0625, "learning_rate": 4.5730622392949784e-05, "loss": 0.1546, "step": 6341 }, { "epoch": 39.39252336448598, "grad_norm": 2.796875, "learning_rate": 4.5729245258336015e-05, "loss": 0.2358, "step": 6342 }, { "epoch": 39.398753894081, "grad_norm": 3.78125, "learning_rate": 4.572786792239569e-05, "loss": 0.1373, "step": 6343 }, { "epoch": 39.40498442367601, "grad_norm": 4.46875, "learning_rate": 4.572649038514216e-05, "loss": 0.2251, "step": 6344 }, { "epoch": 39.41121495327103, "grad_norm": 2.796875, "learning_rate": 4.572511264658881e-05, "loss": 0.148, "step": 6345 }, { "epoch": 39.41744548286604, "grad_norm": 1.8125, "learning_rate": 4.572373470674902e-05, "loss": 0.1017, "step": 6346 }, { "epoch": 39.42367601246106, "grad_norm": 2.078125, "learning_rate": 4.572235656563618e-05, "loss": 0.1699, "step": 6347 }, { "epoch": 39.429906542056074, "grad_norm": 3.375, "learning_rate": 4.5720978223263666e-05, "loss": 0.2901, "step": 6348 }, { "epoch": 39.43613707165109, "grad_norm": 4.1875, "learning_rate": 4.571959967964488e-05, "loss": 0.4715, "step": 6349 }, { "epoch": 39.442367601246104, "grad_norm": 2.953125, "learning_rate": 4.571822093479319e-05, "loss": 0.2528, "step": 6350 }, { "epoch": 39.44859813084112, "grad_norm": 2.96875, "learning_rate": 4.5716841988722004e-05, "loss": 0.1904, "step": 6351 }, { "epoch": 39.454828660436135, "grad_norm": 2.671875, "learning_rate": 4.57154628414447e-05, "loss": 0.2099, "step": 6352 }, { "epoch": 39.46105919003115, "grad_norm": 2.515625, "learning_rate": 4.571408349297468e-05, "loss": 0.2212, "step": 6353 }, { "epoch": 39.467289719626166, "grad_norm": 2.859375, "learning_rate": 4.5712703943325355e-05, "loss": 0.2066, "step": 6354 }, { "epoch": 39.47352024922118, "grad_norm": 1.4609375, "learning_rate": 4.5711324192510094e-05, "loss": 0.1788, "step": 6355 }, { "epoch": 39.4797507788162, "grad_norm": 1.84375, "learning_rate": 4.570994424054232e-05, "loss": 0.1297, "step": 6356 }, { "epoch": 39.48598130841121, "grad_norm": 2.40625, "learning_rate": 4.570856408743542e-05, "loss": 0.2237, "step": 6357 }, { "epoch": 39.49221183800623, "grad_norm": 2.65625, "learning_rate": 4.5707183733202814e-05, "loss": 0.2482, "step": 6358 }, { "epoch": 39.49844236760124, "grad_norm": 2.90625, "learning_rate": 4.57058031778579e-05, "loss": 0.2425, "step": 6359 }, { "epoch": 39.504672897196265, "grad_norm": 1.84375, "learning_rate": 4.570442242141408e-05, "loss": 0.1211, "step": 6360 }, { "epoch": 39.51090342679128, "grad_norm": 3.71875, "learning_rate": 4.5703041463884775e-05, "loss": 0.2973, "step": 6361 }, { "epoch": 39.517133956386296, "grad_norm": 1.59375, "learning_rate": 4.570166030528339e-05, "loss": 0.1268, "step": 6362 }, { "epoch": 39.52336448598131, "grad_norm": 2.71875, "learning_rate": 4.570027894562334e-05, "loss": 0.278, "step": 6363 }, { "epoch": 39.52959501557633, "grad_norm": 2.34375, "learning_rate": 4.569889738491805e-05, "loss": 0.1802, "step": 6364 }, { "epoch": 39.53582554517134, "grad_norm": 2.046875, "learning_rate": 4.569751562318093e-05, "loss": 0.1554, "step": 6365 }, { "epoch": 39.54205607476636, "grad_norm": 3.6875, "learning_rate": 4.56961336604254e-05, "loss": 0.1675, "step": 6366 }, { "epoch": 39.54828660436137, "grad_norm": 4.625, "learning_rate": 4.569475149666488e-05, "loss": 0.2459, "step": 6367 }, { "epoch": 39.55451713395639, "grad_norm": 3.03125, "learning_rate": 4.569336913191279e-05, "loss": 0.1403, "step": 6368 }, { "epoch": 39.5607476635514, "grad_norm": 2.859375, "learning_rate": 4.5691986566182576e-05, "loss": 0.1951, "step": 6369 }, { "epoch": 39.56697819314642, "grad_norm": 2.359375, "learning_rate": 4.569060379948765e-05, "loss": 0.195, "step": 6370 }, { "epoch": 39.573208722741434, "grad_norm": 3.953125, "learning_rate": 4.568922083184144e-05, "loss": 0.1761, "step": 6371 }, { "epoch": 39.57943925233645, "grad_norm": 3.078125, "learning_rate": 4.568783766325738e-05, "loss": 0.1735, "step": 6372 }, { "epoch": 39.585669781931465, "grad_norm": 2.75, "learning_rate": 4.568645429374891e-05, "loss": 0.172, "step": 6373 }, { "epoch": 39.59190031152648, "grad_norm": 3.75, "learning_rate": 4.568507072332946e-05, "loss": 0.2547, "step": 6374 }, { "epoch": 39.598130841121495, "grad_norm": 3.015625, "learning_rate": 4.568368695201246e-05, "loss": 0.1921, "step": 6375 }, { "epoch": 39.60436137071651, "grad_norm": 3.6875, "learning_rate": 4.5682302979811366e-05, "loss": 0.2725, "step": 6376 }, { "epoch": 39.610591900311526, "grad_norm": 2.921875, "learning_rate": 4.568091880673961e-05, "loss": 0.1451, "step": 6377 }, { "epoch": 39.61682242990654, "grad_norm": 2.15625, "learning_rate": 4.5679534432810636e-05, "loss": 0.1454, "step": 6378 }, { "epoch": 39.62305295950156, "grad_norm": 3.859375, "learning_rate": 4.567814985803789e-05, "loss": 0.1258, "step": 6379 }, { "epoch": 39.62928348909657, "grad_norm": 3.71875, "learning_rate": 4.567676508243482e-05, "loss": 0.2779, "step": 6380 }, { "epoch": 39.63551401869159, "grad_norm": 1.9765625, "learning_rate": 4.567538010601487e-05, "loss": 0.157, "step": 6381 }, { "epoch": 39.6417445482866, "grad_norm": 1.9375, "learning_rate": 4.56739949287915e-05, "loss": 0.1664, "step": 6382 }, { "epoch": 39.64797507788162, "grad_norm": 2.34375, "learning_rate": 4.567260955077816e-05, "loss": 0.1483, "step": 6383 }, { "epoch": 39.654205607476634, "grad_norm": 2.421875, "learning_rate": 4.56712239719883e-05, "loss": 0.2341, "step": 6384 }, { "epoch": 39.66043613707165, "grad_norm": 3.203125, "learning_rate": 4.566983819243538e-05, "loss": 0.294, "step": 6385 }, { "epoch": 39.666666666666664, "grad_norm": 3.1875, "learning_rate": 4.566845221213286e-05, "loss": 0.2367, "step": 6386 }, { "epoch": 39.67289719626168, "grad_norm": 3.203125, "learning_rate": 4.566706603109421e-05, "loss": 0.3415, "step": 6387 }, { "epoch": 39.679127725856695, "grad_norm": 2.9375, "learning_rate": 4.566567964933287e-05, "loss": 0.2159, "step": 6388 }, { "epoch": 39.68535825545171, "grad_norm": 2.46875, "learning_rate": 4.5664293066862324e-05, "loss": 0.1739, "step": 6389 }, { "epoch": 39.691588785046726, "grad_norm": 1.90625, "learning_rate": 4.566290628369603e-05, "loss": 0.1307, "step": 6390 }, { "epoch": 39.69781931464174, "grad_norm": 3.578125, "learning_rate": 4.566151929984746e-05, "loss": 0.2084, "step": 6391 }, { "epoch": 39.70404984423676, "grad_norm": 3.390625, "learning_rate": 4.566013211533009e-05, "loss": 0.148, "step": 6392 }, { "epoch": 39.71028037383178, "grad_norm": 1.84375, "learning_rate": 4.565874473015739e-05, "loss": 0.1602, "step": 6393 }, { "epoch": 39.716510903426794, "grad_norm": 2.109375, "learning_rate": 4.565735714434282e-05, "loss": 0.1731, "step": 6394 }, { "epoch": 39.72274143302181, "grad_norm": 2.8125, "learning_rate": 4.5655969357899874e-05, "loss": 0.3012, "step": 6395 }, { "epoch": 39.728971962616825, "grad_norm": 2.984375, "learning_rate": 4.5654581370842025e-05, "loss": 0.3193, "step": 6396 }, { "epoch": 39.73520249221184, "grad_norm": 1.59375, "learning_rate": 4.565319318318275e-05, "loss": 0.1198, "step": 6397 }, { "epoch": 39.741433021806856, "grad_norm": 2.171875, "learning_rate": 4.5651804794935535e-05, "loss": 0.1893, "step": 6398 }, { "epoch": 39.74766355140187, "grad_norm": 3.140625, "learning_rate": 4.565041620611386e-05, "loss": 0.1956, "step": 6399 }, { "epoch": 39.753894080996886, "grad_norm": 4.09375, "learning_rate": 4.564902741673123e-05, "loss": 0.4735, "step": 6400 }, { "epoch": 39.7601246105919, "grad_norm": 2.546875, "learning_rate": 4.56476384268011e-05, "loss": 0.25, "step": 6401 }, { "epoch": 39.76635514018692, "grad_norm": 4.5625, "learning_rate": 4.564624923633699e-05, "loss": 0.3026, "step": 6402 }, { "epoch": 39.77258566978193, "grad_norm": 2.0625, "learning_rate": 4.5644859845352375e-05, "loss": 0.1758, "step": 6403 }, { "epoch": 39.77881619937695, "grad_norm": 2.578125, "learning_rate": 4.564347025386076e-05, "loss": 0.192, "step": 6404 }, { "epoch": 39.78504672897196, "grad_norm": 2.375, "learning_rate": 4.564208046187563e-05, "loss": 0.2344, "step": 6405 }, { "epoch": 39.79127725856698, "grad_norm": 2.5625, "learning_rate": 4.564069046941049e-05, "loss": 0.2605, "step": 6406 }, { "epoch": 39.797507788161994, "grad_norm": 3.203125, "learning_rate": 4.5639300276478834e-05, "loss": 0.235, "step": 6407 }, { "epoch": 39.80373831775701, "grad_norm": 3.515625, "learning_rate": 4.563790988309418e-05, "loss": 0.2344, "step": 6408 }, { "epoch": 39.809968847352025, "grad_norm": 2.90625, "learning_rate": 4.5636519289270006e-05, "loss": 0.1232, "step": 6409 }, { "epoch": 39.81619937694704, "grad_norm": 2.71875, "learning_rate": 4.563512849501983e-05, "loss": 0.2375, "step": 6410 }, { "epoch": 39.822429906542055, "grad_norm": 2.3125, "learning_rate": 4.563373750035718e-05, "loss": 0.2307, "step": 6411 }, { "epoch": 39.82866043613707, "grad_norm": 3.734375, "learning_rate": 4.563234630529553e-05, "loss": 0.3603, "step": 6412 }, { "epoch": 39.834890965732086, "grad_norm": 1.8671875, "learning_rate": 4.5630954909848414e-05, "loss": 0.1422, "step": 6413 }, { "epoch": 39.8411214953271, "grad_norm": 3.625, "learning_rate": 4.562956331402934e-05, "loss": 0.1531, "step": 6414 }, { "epoch": 39.84735202492212, "grad_norm": 2.40625, "learning_rate": 4.562817151785182e-05, "loss": 0.1754, "step": 6415 }, { "epoch": 39.85358255451713, "grad_norm": 2.9375, "learning_rate": 4.5626779521329376e-05, "loss": 0.2419, "step": 6416 }, { "epoch": 39.85981308411215, "grad_norm": 2.046875, "learning_rate": 4.562538732447553e-05, "loss": 0.1387, "step": 6417 }, { "epoch": 39.86604361370716, "grad_norm": 3.546875, "learning_rate": 4.562399492730379e-05, "loss": 0.3183, "step": 6418 }, { "epoch": 39.87227414330218, "grad_norm": 2.890625, "learning_rate": 4.56226023298277e-05, "loss": 0.2094, "step": 6419 }, { "epoch": 39.87850467289719, "grad_norm": 2.609375, "learning_rate": 4.562120953206077e-05, "loss": 0.18, "step": 6420 }, { "epoch": 39.88473520249221, "grad_norm": 2.0625, "learning_rate": 4.5619816534016526e-05, "loss": 0.1642, "step": 6421 }, { "epoch": 39.890965732087224, "grad_norm": 3.171875, "learning_rate": 4.561842333570851e-05, "loss": 0.2701, "step": 6422 }, { "epoch": 39.89719626168224, "grad_norm": 1.234375, "learning_rate": 4.561702993715023e-05, "loss": 0.1305, "step": 6423 }, { "epoch": 39.90342679127726, "grad_norm": 3.3125, "learning_rate": 4.5615636338355255e-05, "loss": 0.3198, "step": 6424 }, { "epoch": 39.90965732087228, "grad_norm": 2.171875, "learning_rate": 4.5614242539337085e-05, "loss": 0.164, "step": 6425 }, { "epoch": 39.91588785046729, "grad_norm": 2.3125, "learning_rate": 4.5612848540109274e-05, "loss": 0.2669, "step": 6426 }, { "epoch": 39.92211838006231, "grad_norm": 3.234375, "learning_rate": 4.561145434068536e-05, "loss": 0.2492, "step": 6427 }, { "epoch": 39.92834890965732, "grad_norm": 2.390625, "learning_rate": 4.561005994107888e-05, "loss": 0.2607, "step": 6428 }, { "epoch": 39.93457943925234, "grad_norm": 2.5625, "learning_rate": 4.560866534130338e-05, "loss": 0.18, "step": 6429 }, { "epoch": 39.940809968847354, "grad_norm": 3.328125, "learning_rate": 4.56072705413724e-05, "loss": 0.176, "step": 6430 }, { "epoch": 39.94704049844237, "grad_norm": 3.890625, "learning_rate": 4.560587554129949e-05, "loss": 0.2713, "step": 6431 }, { "epoch": 39.953271028037385, "grad_norm": 2.25, "learning_rate": 4.56044803410982e-05, "loss": 0.1612, "step": 6432 }, { "epoch": 39.9595015576324, "grad_norm": 2.015625, "learning_rate": 4.560308494078207e-05, "loss": 0.1533, "step": 6433 }, { "epoch": 39.965732087227416, "grad_norm": 1.453125, "learning_rate": 4.560168934036466e-05, "loss": 0.1166, "step": 6434 }, { "epoch": 39.97196261682243, "grad_norm": 2.046875, "learning_rate": 4.560029353985954e-05, "loss": 0.2658, "step": 6435 }, { "epoch": 39.978193146417446, "grad_norm": 1.890625, "learning_rate": 4.5598897539280236e-05, "loss": 0.1687, "step": 6436 }, { "epoch": 39.98442367601246, "grad_norm": 2.84375, "learning_rate": 4.5597501338640324e-05, "loss": 0.1556, "step": 6437 }, { "epoch": 39.99065420560748, "grad_norm": 3.609375, "learning_rate": 4.559610493795337e-05, "loss": 0.2546, "step": 6438 }, { "epoch": 39.99688473520249, "grad_norm": 3.890625, "learning_rate": 4.559470833723292e-05, "loss": 0.2194, "step": 6439 }, { "epoch": 40.0, "grad_norm": 1.25, "learning_rate": 4.559331153649255e-05, "loss": 0.0581, "step": 6440 }, { "epoch": 40.006230529595015, "grad_norm": 1.578125, "learning_rate": 4.559191453574582e-05, "loss": 0.1292, "step": 6441 }, { "epoch": 40.01246105919003, "grad_norm": 2.890625, "learning_rate": 4.55905173350063e-05, "loss": 0.3025, "step": 6442 }, { "epoch": 40.018691588785046, "grad_norm": 3.109375, "learning_rate": 4.558911993428756e-05, "loss": 0.1645, "step": 6443 }, { "epoch": 40.02492211838006, "grad_norm": 2.984375, "learning_rate": 4.558772233360317e-05, "loss": 0.2018, "step": 6444 }, { "epoch": 40.03115264797508, "grad_norm": 2.0625, "learning_rate": 4.558632453296671e-05, "loss": 0.1475, "step": 6445 }, { "epoch": 40.03738317757009, "grad_norm": 2.1875, "learning_rate": 4.5584926532391746e-05, "loss": 0.2522, "step": 6446 }, { "epoch": 40.04361370716511, "grad_norm": 3.0625, "learning_rate": 4.558352833189186e-05, "loss": 0.2011, "step": 6447 }, { "epoch": 40.04984423676012, "grad_norm": 3.234375, "learning_rate": 4.558212993148064e-05, "loss": 0.3073, "step": 6448 }, { "epoch": 40.05607476635514, "grad_norm": 1.9375, "learning_rate": 4.5580731331171655e-05, "loss": 0.175, "step": 6449 }, { "epoch": 40.06230529595015, "grad_norm": 2.40625, "learning_rate": 4.557933253097849e-05, "loss": 0.1992, "step": 6450 }, { "epoch": 40.06853582554517, "grad_norm": 2.25, "learning_rate": 4.557793353091474e-05, "loss": 0.1236, "step": 6451 }, { "epoch": 40.074766355140184, "grad_norm": 2.40625, "learning_rate": 4.557653433099398e-05, "loss": 0.172, "step": 6452 }, { "epoch": 40.0809968847352, "grad_norm": 2.390625, "learning_rate": 4.557513493122981e-05, "loss": 0.1854, "step": 6453 }, { "epoch": 40.087227414330215, "grad_norm": 2.390625, "learning_rate": 4.557373533163581e-05, "loss": 0.1528, "step": 6454 }, { "epoch": 40.09345794392523, "grad_norm": 2.34375, "learning_rate": 4.557233553222558e-05, "loss": 0.2407, "step": 6455 }, { "epoch": 40.099688473520246, "grad_norm": 3.015625, "learning_rate": 4.5570935533012724e-05, "loss": 0.2214, "step": 6456 }, { "epoch": 40.10591900311527, "grad_norm": 1.5234375, "learning_rate": 4.556953533401083e-05, "loss": 0.1163, "step": 6457 }, { "epoch": 40.11214953271028, "grad_norm": 1.359375, "learning_rate": 4.5568134935233487e-05, "loss": 0.1178, "step": 6458 }, { "epoch": 40.1183800623053, "grad_norm": 2.46875, "learning_rate": 4.5566734336694314e-05, "loss": 0.2142, "step": 6459 }, { "epoch": 40.124610591900314, "grad_norm": 1.796875, "learning_rate": 4.5565333538406904e-05, "loss": 0.1759, "step": 6460 }, { "epoch": 40.13084112149533, "grad_norm": 2.15625, "learning_rate": 4.556393254038486e-05, "loss": 0.1404, "step": 6461 }, { "epoch": 40.137071651090345, "grad_norm": 3.171875, "learning_rate": 4.55625313426418e-05, "loss": 0.2477, "step": 6462 }, { "epoch": 40.14330218068536, "grad_norm": 2.171875, "learning_rate": 4.556112994519132e-05, "loss": 0.2048, "step": 6463 }, { "epoch": 40.149532710280376, "grad_norm": 2.546875, "learning_rate": 4.555972834804704e-05, "loss": 0.235, "step": 6464 }, { "epoch": 40.15576323987539, "grad_norm": 2.296875, "learning_rate": 4.5558326551222565e-05, "loss": 0.1897, "step": 6465 }, { "epoch": 40.161993769470406, "grad_norm": 3.453125, "learning_rate": 4.555692455473151e-05, "loss": 0.3458, "step": 6466 }, { "epoch": 40.16822429906542, "grad_norm": 3.15625, "learning_rate": 4.5555522358587496e-05, "loss": 0.3038, "step": 6467 }, { "epoch": 40.17445482866044, "grad_norm": 2.609375, "learning_rate": 4.555411996280414e-05, "loss": 0.2414, "step": 6468 }, { "epoch": 40.18068535825545, "grad_norm": 3.734375, "learning_rate": 4.555271736739507e-05, "loss": 0.3393, "step": 6469 }, { "epoch": 40.18691588785047, "grad_norm": 1.1640625, "learning_rate": 4.555131457237389e-05, "loss": 0.1143, "step": 6470 }, { "epoch": 40.19314641744548, "grad_norm": 2.5625, "learning_rate": 4.554991157775425e-05, "loss": 0.1223, "step": 6471 }, { "epoch": 40.1993769470405, "grad_norm": 2.359375, "learning_rate": 4.554850838354974e-05, "loss": 0.1657, "step": 6472 }, { "epoch": 40.205607476635514, "grad_norm": 3.4375, "learning_rate": 4.5547104989774025e-05, "loss": 0.2779, "step": 6473 }, { "epoch": 40.21183800623053, "grad_norm": 1.8046875, "learning_rate": 4.554570139644071e-05, "loss": 0.1506, "step": 6474 }, { "epoch": 40.218068535825545, "grad_norm": 1.7734375, "learning_rate": 4.554429760356343e-05, "loss": 0.1552, "step": 6475 }, { "epoch": 40.22429906542056, "grad_norm": 1.03125, "learning_rate": 4.554289361115585e-05, "loss": 0.107, "step": 6476 }, { "epoch": 40.230529595015575, "grad_norm": 1.3203125, "learning_rate": 4.554148941923156e-05, "loss": 0.1103, "step": 6477 }, { "epoch": 40.23676012461059, "grad_norm": 3.265625, "learning_rate": 4.554008502780422e-05, "loss": 0.1774, "step": 6478 }, { "epoch": 40.242990654205606, "grad_norm": 2.421875, "learning_rate": 4.553868043688747e-05, "loss": 0.149, "step": 6479 }, { "epoch": 40.24922118380062, "grad_norm": 1.9921875, "learning_rate": 4.5537275646494956e-05, "loss": 0.1555, "step": 6480 }, { "epoch": 40.25545171339564, "grad_norm": 2.3125, "learning_rate": 4.5535870656640314e-05, "loss": 0.159, "step": 6481 }, { "epoch": 40.26168224299065, "grad_norm": 2.265625, "learning_rate": 4.553446546733718e-05, "loss": 0.1374, "step": 6482 }, { "epoch": 40.26791277258567, "grad_norm": 2.65625, "learning_rate": 4.5533060078599226e-05, "loss": 0.1959, "step": 6483 }, { "epoch": 40.27414330218068, "grad_norm": 2.21875, "learning_rate": 4.553165449044008e-05, "loss": 0.2413, "step": 6484 }, { "epoch": 40.2803738317757, "grad_norm": 1.6640625, "learning_rate": 4.553024870287341e-05, "loss": 0.1173, "step": 6485 }, { "epoch": 40.28660436137071, "grad_norm": 2.46875, "learning_rate": 4.5528842715912855e-05, "loss": 0.1821, "step": 6486 }, { "epoch": 40.29283489096573, "grad_norm": 3.15625, "learning_rate": 4.552743652957208e-05, "loss": 0.1873, "step": 6487 }, { "epoch": 40.299065420560744, "grad_norm": 2.484375, "learning_rate": 4.552603014386473e-05, "loss": 0.1436, "step": 6488 }, { "epoch": 40.30529595015577, "grad_norm": 2.46875, "learning_rate": 4.552462355880448e-05, "loss": 0.2322, "step": 6489 }, { "epoch": 40.31152647975078, "grad_norm": 3.59375, "learning_rate": 4.5523216774404976e-05, "loss": 0.2193, "step": 6490 }, { "epoch": 40.3177570093458, "grad_norm": 2.109375, "learning_rate": 4.552180979067989e-05, "loss": 0.189, "step": 6491 }, { "epoch": 40.32398753894081, "grad_norm": 2.640625, "learning_rate": 4.552040260764289e-05, "loss": 0.1737, "step": 6492 }, { "epoch": 40.33021806853583, "grad_norm": 1.984375, "learning_rate": 4.5518995225307635e-05, "loss": 0.1445, "step": 6493 }, { "epoch": 40.33644859813084, "grad_norm": 1.1875, "learning_rate": 4.55175876436878e-05, "loss": 0.1177, "step": 6494 }, { "epoch": 40.34267912772586, "grad_norm": 4.3125, "learning_rate": 4.551617986279705e-05, "loss": 0.1352, "step": 6495 }, { "epoch": 40.348909657320874, "grad_norm": 3.546875, "learning_rate": 4.551477188264906e-05, "loss": 0.1291, "step": 6496 }, { "epoch": 40.35514018691589, "grad_norm": 4.625, "learning_rate": 4.5513363703257496e-05, "loss": 0.1866, "step": 6497 }, { "epoch": 40.361370716510905, "grad_norm": 3.0, "learning_rate": 4.5511955324636054e-05, "loss": 0.1671, "step": 6498 }, { "epoch": 40.36760124610592, "grad_norm": 1.953125, "learning_rate": 4.55105467467984e-05, "loss": 0.1624, "step": 6499 }, { "epoch": 40.373831775700936, "grad_norm": 3.796875, "learning_rate": 4.550913796975821e-05, "loss": 0.2017, "step": 6500 }, { "epoch": 40.38006230529595, "grad_norm": 3.578125, "learning_rate": 4.550772899352918e-05, "loss": 0.2211, "step": 6501 }, { "epoch": 40.386292834890966, "grad_norm": 3.484375, "learning_rate": 4.550631981812499e-05, "loss": 0.1518, "step": 6502 }, { "epoch": 40.39252336448598, "grad_norm": 2.34375, "learning_rate": 4.550491044355931e-05, "loss": 0.1511, "step": 6503 }, { "epoch": 40.398753894081, "grad_norm": 2.015625, "learning_rate": 4.550350086984585e-05, "loss": 0.1313, "step": 6504 }, { "epoch": 40.40498442367601, "grad_norm": 3.5625, "learning_rate": 4.550209109699829e-05, "loss": 0.2025, "step": 6505 }, { "epoch": 40.41121495327103, "grad_norm": 3.421875, "learning_rate": 4.550068112503032e-05, "loss": 0.1256, "step": 6506 }, { "epoch": 40.41744548286604, "grad_norm": 4.71875, "learning_rate": 4.5499270953955635e-05, "loss": 0.1708, "step": 6507 }, { "epoch": 40.42367601246106, "grad_norm": 3.046875, "learning_rate": 4.5497860583787936e-05, "loss": 0.1728, "step": 6508 }, { "epoch": 40.429906542056074, "grad_norm": 2.5, "learning_rate": 4.549645001454092e-05, "loss": 0.174, "step": 6509 }, { "epoch": 40.43613707165109, "grad_norm": 1.7265625, "learning_rate": 4.5495039246228274e-05, "loss": 0.1246, "step": 6510 }, { "epoch": 40.442367601246104, "grad_norm": 2.4375, "learning_rate": 4.549362827886372e-05, "loss": 0.2023, "step": 6511 }, { "epoch": 40.44859813084112, "grad_norm": 3.171875, "learning_rate": 4.5492217112460955e-05, "loss": 0.1901, "step": 6512 }, { "epoch": 40.454828660436135, "grad_norm": 1.796875, "learning_rate": 4.549080574703367e-05, "loss": 0.1782, "step": 6513 }, { "epoch": 40.46105919003115, "grad_norm": 2.046875, "learning_rate": 4.548939418259559e-05, "loss": 0.1621, "step": 6514 }, { "epoch": 40.467289719626166, "grad_norm": 2.140625, "learning_rate": 4.548798241916042e-05, "loss": 0.1378, "step": 6515 }, { "epoch": 40.47352024922118, "grad_norm": 3.046875, "learning_rate": 4.5486570456741867e-05, "loss": 0.1992, "step": 6516 }, { "epoch": 40.4797507788162, "grad_norm": 3.953125, "learning_rate": 4.548515829535365e-05, "loss": 0.3058, "step": 6517 }, { "epoch": 40.48598130841121, "grad_norm": 1.8125, "learning_rate": 4.548374593500947e-05, "loss": 0.1605, "step": 6518 }, { "epoch": 40.49221183800623, "grad_norm": 1.640625, "learning_rate": 4.548233337572306e-05, "loss": 0.1289, "step": 6519 }, { "epoch": 40.49844236760124, "grad_norm": 5.75, "learning_rate": 4.5480920617508135e-05, "loss": 0.2221, "step": 6520 }, { "epoch": 40.504672897196265, "grad_norm": 3.109375, "learning_rate": 4.547950766037842e-05, "loss": 0.1358, "step": 6521 }, { "epoch": 40.51090342679128, "grad_norm": 2.328125, "learning_rate": 4.5478094504347626e-05, "loss": 0.2202, "step": 6522 }, { "epoch": 40.517133956386296, "grad_norm": 2.34375, "learning_rate": 4.5476681149429486e-05, "loss": 0.2592, "step": 6523 }, { "epoch": 40.52336448598131, "grad_norm": 2.125, "learning_rate": 4.547526759563773e-05, "loss": 0.2681, "step": 6524 }, { "epoch": 40.52959501557633, "grad_norm": 3.4375, "learning_rate": 4.547385384298608e-05, "loss": 0.2889, "step": 6525 }, { "epoch": 40.53582554517134, "grad_norm": 2.0625, "learning_rate": 4.547243989148826e-05, "loss": 0.1775, "step": 6526 }, { "epoch": 40.54205607476636, "grad_norm": 2.0625, "learning_rate": 4.5471025741158024e-05, "loss": 0.1589, "step": 6527 }, { "epoch": 40.54828660436137, "grad_norm": 1.921875, "learning_rate": 4.5469611392009083e-05, "loss": 0.2471, "step": 6528 }, { "epoch": 40.55451713395639, "grad_norm": 2.859375, "learning_rate": 4.546819684405519e-05, "loss": 0.2132, "step": 6529 }, { "epoch": 40.5607476635514, "grad_norm": 3.953125, "learning_rate": 4.546678209731008e-05, "loss": 0.4547, "step": 6530 }, { "epoch": 40.56697819314642, "grad_norm": 3.5, "learning_rate": 4.546536715178749e-05, "loss": 0.3338, "step": 6531 }, { "epoch": 40.573208722741434, "grad_norm": 2.0625, "learning_rate": 4.546395200750116e-05, "loss": 0.2511, "step": 6532 }, { "epoch": 40.57943925233645, "grad_norm": 3.90625, "learning_rate": 4.546253666446484e-05, "loss": 0.2923, "step": 6533 }, { "epoch": 40.585669781931465, "grad_norm": 3.09375, "learning_rate": 4.546112112269228e-05, "loss": 0.25, "step": 6534 }, { "epoch": 40.59190031152648, "grad_norm": 3.46875, "learning_rate": 4.5459705382197205e-05, "loss": 0.2287, "step": 6535 }, { "epoch": 40.598130841121495, "grad_norm": 2.59375, "learning_rate": 4.5458289442993395e-05, "loss": 0.1677, "step": 6536 }, { "epoch": 40.60436137071651, "grad_norm": 2.15625, "learning_rate": 4.545687330509458e-05, "loss": 0.1837, "step": 6537 }, { "epoch": 40.610591900311526, "grad_norm": 1.984375, "learning_rate": 4.545545696851453e-05, "loss": 0.118, "step": 6538 }, { "epoch": 40.61682242990654, "grad_norm": 2.609375, "learning_rate": 4.5454040433266987e-05, "loss": 0.2134, "step": 6539 }, { "epoch": 40.62305295950156, "grad_norm": 3.796875, "learning_rate": 4.545262369936572e-05, "loss": 0.237, "step": 6540 }, { "epoch": 40.62928348909657, "grad_norm": 1.8671875, "learning_rate": 4.5451206766824476e-05, "loss": 0.1495, "step": 6541 }, { "epoch": 40.63551401869159, "grad_norm": 2.453125, "learning_rate": 4.544978963565703e-05, "loss": 0.2026, "step": 6542 }, { "epoch": 40.6417445482866, "grad_norm": 3.125, "learning_rate": 4.544837230587714e-05, "loss": 0.1244, "step": 6543 }, { "epoch": 40.64797507788162, "grad_norm": 2.625, "learning_rate": 4.5446954777498564e-05, "loss": 0.16, "step": 6544 }, { "epoch": 40.654205607476634, "grad_norm": 1.875, "learning_rate": 4.544553705053508e-05, "loss": 0.1561, "step": 6545 }, { "epoch": 40.66043613707165, "grad_norm": 2.09375, "learning_rate": 4.544411912500045e-05, "loss": 0.1328, "step": 6546 }, { "epoch": 40.666666666666664, "grad_norm": 2.875, "learning_rate": 4.544270100090846e-05, "loss": 0.1772, "step": 6547 }, { "epoch": 40.67289719626168, "grad_norm": 3.1875, "learning_rate": 4.544128267827286e-05, "loss": 0.1705, "step": 6548 }, { "epoch": 40.679127725856695, "grad_norm": 2.46875, "learning_rate": 4.5439864157107434e-05, "loss": 0.1923, "step": 6549 }, { "epoch": 40.68535825545171, "grad_norm": 2.875, "learning_rate": 4.543844543742597e-05, "loss": 0.2501, "step": 6550 }, { "epoch": 40.691588785046726, "grad_norm": 2.9375, "learning_rate": 4.5437026519242245e-05, "loss": 0.2068, "step": 6551 }, { "epoch": 40.69781931464174, "grad_norm": 2.3125, "learning_rate": 4.5435607402570015e-05, "loss": 0.1999, "step": 6552 }, { "epoch": 40.70404984423676, "grad_norm": 3.03125, "learning_rate": 4.543418808742309e-05, "loss": 0.1759, "step": 6553 }, { "epoch": 40.71028037383178, "grad_norm": 2.734375, "learning_rate": 4.543276857381525e-05, "loss": 0.289, "step": 6554 }, { "epoch": 40.716510903426794, "grad_norm": 2.84375, "learning_rate": 4.543134886176027e-05, "loss": 0.3705, "step": 6555 }, { "epoch": 40.72274143302181, "grad_norm": 2.578125, "learning_rate": 4.542992895127195e-05, "loss": 0.151, "step": 6556 }, { "epoch": 40.728971962616825, "grad_norm": 2.96875, "learning_rate": 4.542850884236407e-05, "loss": 0.1531, "step": 6557 }, { "epoch": 40.73520249221184, "grad_norm": 2.890625, "learning_rate": 4.542708853505043e-05, "loss": 0.2447, "step": 6558 }, { "epoch": 40.741433021806856, "grad_norm": 1.390625, "learning_rate": 4.5425668029344827e-05, "loss": 0.1088, "step": 6559 }, { "epoch": 40.74766355140187, "grad_norm": 2.625, "learning_rate": 4.542424732526105e-05, "loss": 0.2891, "step": 6560 }, { "epoch": 40.753894080996886, "grad_norm": 2.84375, "learning_rate": 4.54228264228129e-05, "loss": 0.1682, "step": 6561 }, { "epoch": 40.7601246105919, "grad_norm": 3.28125, "learning_rate": 4.542140532201419e-05, "loss": 0.1468, "step": 6562 }, { "epoch": 40.76635514018692, "grad_norm": 3.46875, "learning_rate": 4.541998402287869e-05, "loss": 0.319, "step": 6563 }, { "epoch": 40.77258566978193, "grad_norm": 3.609375, "learning_rate": 4.541856252542024e-05, "loss": 0.1784, "step": 6564 }, { "epoch": 40.77881619937695, "grad_norm": 2.984375, "learning_rate": 4.5417140829652616e-05, "loss": 0.2047, "step": 6565 }, { "epoch": 40.78504672897196, "grad_norm": 2.78125, "learning_rate": 4.541571893558965e-05, "loss": 0.1565, "step": 6566 }, { "epoch": 40.79127725856698, "grad_norm": 2.15625, "learning_rate": 4.541429684324513e-05, "loss": 0.1747, "step": 6567 }, { "epoch": 40.797507788161994, "grad_norm": 2.53125, "learning_rate": 4.541287455263288e-05, "loss": 0.3982, "step": 6568 }, { "epoch": 40.80373831775701, "grad_norm": 4.34375, "learning_rate": 4.541145206376671e-05, "loss": 0.1463, "step": 6569 }, { "epoch": 40.809968847352025, "grad_norm": 2.078125, "learning_rate": 4.541002937666045e-05, "loss": 0.118, "step": 6570 }, { "epoch": 40.81619937694704, "grad_norm": 2.25, "learning_rate": 4.540860649132789e-05, "loss": 0.1506, "step": 6571 }, { "epoch": 40.822429906542055, "grad_norm": 1.953125, "learning_rate": 4.540718340778287e-05, "loss": 0.1187, "step": 6572 }, { "epoch": 40.82866043613707, "grad_norm": 1.734375, "learning_rate": 4.540576012603921e-05, "loss": 0.121, "step": 6573 }, { "epoch": 40.834890965732086, "grad_norm": 2.59375, "learning_rate": 4.540433664611073e-05, "loss": 0.2074, "step": 6574 }, { "epoch": 40.8411214953271, "grad_norm": 2.71875, "learning_rate": 4.5402912968011246e-05, "loss": 0.1749, "step": 6575 }, { "epoch": 40.84735202492212, "grad_norm": 2.53125, "learning_rate": 4.540148909175459e-05, "loss": 0.1272, "step": 6576 }, { "epoch": 40.85358255451713, "grad_norm": 4.125, "learning_rate": 4.540006501735461e-05, "loss": 0.2098, "step": 6577 }, { "epoch": 40.85981308411215, "grad_norm": 1.859375, "learning_rate": 4.5398640744825106e-05, "loss": 0.1293, "step": 6578 }, { "epoch": 40.86604361370716, "grad_norm": 1.71875, "learning_rate": 4.5397216274179934e-05, "loss": 0.1673, "step": 6579 }, { "epoch": 40.87227414330218, "grad_norm": 2.5, "learning_rate": 4.539579160543291e-05, "loss": 0.1344, "step": 6580 }, { "epoch": 40.87850467289719, "grad_norm": 2.8125, "learning_rate": 4.539436673859789e-05, "loss": 0.1404, "step": 6581 }, { "epoch": 40.88473520249221, "grad_norm": 4.71875, "learning_rate": 4.5392941673688694e-05, "loss": 0.474, "step": 6582 }, { "epoch": 40.890965732087224, "grad_norm": 2.859375, "learning_rate": 4.539151641071918e-05, "loss": 0.2128, "step": 6583 }, { "epoch": 40.89719626168224, "grad_norm": 3.15625, "learning_rate": 4.5390090949703177e-05, "loss": 0.2462, "step": 6584 }, { "epoch": 40.90342679127726, "grad_norm": 1.4140625, "learning_rate": 4.5388665290654544e-05, "loss": 0.1508, "step": 6585 }, { "epoch": 40.90965732087228, "grad_norm": 3.578125, "learning_rate": 4.538723943358712e-05, "loss": 0.2499, "step": 6586 }, { "epoch": 40.91588785046729, "grad_norm": 3.859375, "learning_rate": 4.538581337851474e-05, "loss": 0.2284, "step": 6587 }, { "epoch": 40.92211838006231, "grad_norm": 3.375, "learning_rate": 4.538438712545127e-05, "loss": 0.2084, "step": 6588 }, { "epoch": 40.92834890965732, "grad_norm": 2.21875, "learning_rate": 4.538296067441056e-05, "loss": 0.2285, "step": 6589 }, { "epoch": 40.93457943925234, "grad_norm": 2.5625, "learning_rate": 4.5381534025406454e-05, "loss": 0.2083, "step": 6590 }, { "epoch": 40.940809968847354, "grad_norm": 3.203125, "learning_rate": 4.5380107178452815e-05, "loss": 0.1524, "step": 6591 }, { "epoch": 40.94704049844237, "grad_norm": 3.796875, "learning_rate": 4.537868013356351e-05, "loss": 0.264, "step": 6592 }, { "epoch": 40.953271028037385, "grad_norm": 3.75, "learning_rate": 4.537725289075239e-05, "loss": 0.2199, "step": 6593 }, { "epoch": 40.9595015576324, "grad_norm": 1.40625, "learning_rate": 4.53758254500333e-05, "loss": 0.1277, "step": 6594 }, { "epoch": 40.965732087227416, "grad_norm": 3.984375, "learning_rate": 4.5374397811420134e-05, "loss": 0.3369, "step": 6595 }, { "epoch": 40.97196261682243, "grad_norm": 2.0625, "learning_rate": 4.5372969974926746e-05, "loss": 0.1607, "step": 6596 }, { "epoch": 40.978193146417446, "grad_norm": 2.21875, "learning_rate": 4.5371541940567e-05, "loss": 0.1143, "step": 6597 }, { "epoch": 40.98442367601246, "grad_norm": 1.7578125, "learning_rate": 4.537011370835476e-05, "loss": 0.1449, "step": 6598 }, { "epoch": 40.99065420560748, "grad_norm": 1.6640625, "learning_rate": 4.5368685278303914e-05, "loss": 0.1416, "step": 6599 }, { "epoch": 40.99688473520249, "grad_norm": 2.09375, "learning_rate": 4.5367256650428314e-05, "loss": 0.1171, "step": 6600 }, { "epoch": 41.0, "grad_norm": 1.625, "learning_rate": 4.5365827824741845e-05, "loss": 0.1107, "step": 6601 }, { "epoch": 41.006230529595015, "grad_norm": 2.921875, "learning_rate": 4.5364398801258396e-05, "loss": 0.1876, "step": 6602 }, { "epoch": 41.01246105919003, "grad_norm": 2.84375, "learning_rate": 4.536296957999183e-05, "loss": 0.2429, "step": 6603 }, { "epoch": 41.018691588785046, "grad_norm": 0.9921875, "learning_rate": 4.536154016095603e-05, "loss": 0.1071, "step": 6604 }, { "epoch": 41.02492211838006, "grad_norm": 1.7421875, "learning_rate": 4.5360110544164886e-05, "loss": 0.1531, "step": 6605 }, { "epoch": 41.03115264797508, "grad_norm": 2.359375, "learning_rate": 4.535868072963227e-05, "loss": 0.1651, "step": 6606 }, { "epoch": 41.03738317757009, "grad_norm": 3.34375, "learning_rate": 4.535725071737209e-05, "loss": 0.2642, "step": 6607 }, { "epoch": 41.04361370716511, "grad_norm": 1.6484375, "learning_rate": 4.535582050739822e-05, "loss": 0.1417, "step": 6608 }, { "epoch": 41.04984423676012, "grad_norm": 3.890625, "learning_rate": 4.535439009972454e-05, "loss": 0.1894, "step": 6609 }, { "epoch": 41.05607476635514, "grad_norm": 3.703125, "learning_rate": 4.5352959494364964e-05, "loss": 0.1597, "step": 6610 }, { "epoch": 41.06230529595015, "grad_norm": 3.859375, "learning_rate": 4.535152869133338e-05, "loss": 0.1387, "step": 6611 }, { "epoch": 41.06853582554517, "grad_norm": 2.71875, "learning_rate": 4.535009769064368e-05, "loss": 0.1855, "step": 6612 }, { "epoch": 41.074766355140184, "grad_norm": 2.34375, "learning_rate": 4.5348666492309754e-05, "loss": 0.1869, "step": 6613 }, { "epoch": 41.0809968847352, "grad_norm": 3.53125, "learning_rate": 4.534723509634552e-05, "loss": 0.1797, "step": 6614 }, { "epoch": 41.087227414330215, "grad_norm": 4.34375, "learning_rate": 4.534580350276487e-05, "loss": 0.1507, "step": 6615 }, { "epoch": 41.09345794392523, "grad_norm": 3.921875, "learning_rate": 4.53443717115817e-05, "loss": 0.1394, "step": 6616 }, { "epoch": 41.099688473520246, "grad_norm": 2.046875, "learning_rate": 4.5342939722809944e-05, "loss": 0.1827, "step": 6617 }, { "epoch": 41.10591900311527, "grad_norm": 1.40625, "learning_rate": 4.534150753646348e-05, "loss": 0.1012, "step": 6618 }, { "epoch": 41.11214953271028, "grad_norm": 3.4375, "learning_rate": 4.5340075152556226e-05, "loss": 0.2765, "step": 6619 }, { "epoch": 41.1183800623053, "grad_norm": 4.21875, "learning_rate": 4.53386425711021e-05, "loss": 0.178, "step": 6620 }, { "epoch": 41.124610591900314, "grad_norm": 2.84375, "learning_rate": 4.533720979211501e-05, "loss": 0.2592, "step": 6621 }, { "epoch": 41.13084112149533, "grad_norm": 2.0, "learning_rate": 4.533577681560888e-05, "loss": 0.1416, "step": 6622 }, { "epoch": 41.137071651090345, "grad_norm": 1.90625, "learning_rate": 4.533434364159761e-05, "loss": 0.1394, "step": 6623 }, { "epoch": 41.14330218068536, "grad_norm": 1.0703125, "learning_rate": 4.533291027009513e-05, "loss": 0.1078, "step": 6624 }, { "epoch": 41.149532710280376, "grad_norm": 2.671875, "learning_rate": 4.5331476701115366e-05, "loss": 0.1182, "step": 6625 }, { "epoch": 41.15576323987539, "grad_norm": 1.7578125, "learning_rate": 4.533004293467224e-05, "loss": 0.1158, "step": 6626 }, { "epoch": 41.161993769470406, "grad_norm": 2.3125, "learning_rate": 4.5328608970779663e-05, "loss": 0.1565, "step": 6627 }, { "epoch": 41.16822429906542, "grad_norm": 2.8125, "learning_rate": 4.532717480945158e-05, "loss": 0.2138, "step": 6628 }, { "epoch": 41.17445482866044, "grad_norm": 2.484375, "learning_rate": 4.532574045070191e-05, "loss": 0.1715, "step": 6629 }, { "epoch": 41.18068535825545, "grad_norm": 2.234375, "learning_rate": 4.532430589454458e-05, "loss": 0.1435, "step": 6630 }, { "epoch": 41.18691588785047, "grad_norm": 1.71875, "learning_rate": 4.532287114099354e-05, "loss": 0.1555, "step": 6631 }, { "epoch": 41.19314641744548, "grad_norm": 2.75, "learning_rate": 4.53214361900627e-05, "loss": 0.2483, "step": 6632 }, { "epoch": 41.1993769470405, "grad_norm": 2.265625, "learning_rate": 4.532000104176602e-05, "loss": 0.161, "step": 6633 }, { "epoch": 41.205607476635514, "grad_norm": 3.515625, "learning_rate": 4.531856569611742e-05, "loss": 0.2959, "step": 6634 }, { "epoch": 41.21183800623053, "grad_norm": 1.4609375, "learning_rate": 4.531713015313086e-05, "loss": 0.1314, "step": 6635 }, { "epoch": 41.218068535825545, "grad_norm": 2.40625, "learning_rate": 4.5315694412820264e-05, "loss": 0.303, "step": 6636 }, { "epoch": 41.22429906542056, "grad_norm": 1.796875, "learning_rate": 4.531425847519958e-05, "loss": 0.1699, "step": 6637 }, { "epoch": 41.230529595015575, "grad_norm": 2.78125, "learning_rate": 4.531282234028276e-05, "loss": 0.268, "step": 6638 }, { "epoch": 41.23676012461059, "grad_norm": 2.46875, "learning_rate": 4.5311386008083747e-05, "loss": 0.1624, "step": 6639 }, { "epoch": 41.242990654205606, "grad_norm": 2.015625, "learning_rate": 4.5309949478616494e-05, "loss": 0.1417, "step": 6640 }, { "epoch": 41.24922118380062, "grad_norm": 3.109375, "learning_rate": 4.5308512751894946e-05, "loss": 0.2017, "step": 6641 }, { "epoch": 41.25545171339564, "grad_norm": 2.8125, "learning_rate": 4.5307075827933076e-05, "loss": 0.184, "step": 6642 }, { "epoch": 41.26168224299065, "grad_norm": 2.859375, "learning_rate": 4.5305638706744816e-05, "loss": 0.2408, "step": 6643 }, { "epoch": 41.26791277258567, "grad_norm": 3.171875, "learning_rate": 4.530420138834414e-05, "loss": 0.1591, "step": 6644 }, { "epoch": 41.27414330218068, "grad_norm": 2.109375, "learning_rate": 4.530276387274499e-05, "loss": 0.1231, "step": 6645 }, { "epoch": 41.2803738317757, "grad_norm": 2.75, "learning_rate": 4.530132615996135e-05, "loss": 0.1666, "step": 6646 }, { "epoch": 41.28660436137071, "grad_norm": 1.8984375, "learning_rate": 4.529988825000717e-05, "loss": 0.1372, "step": 6647 }, { "epoch": 41.29283489096573, "grad_norm": 2.21875, "learning_rate": 4.529845014289642e-05, "loss": 0.1709, "step": 6648 }, { "epoch": 41.299065420560744, "grad_norm": 2.9375, "learning_rate": 4.529701183864305e-05, "loss": 0.2368, "step": 6649 }, { "epoch": 41.30529595015577, "grad_norm": 1.4609375, "learning_rate": 4.529557333726106e-05, "loss": 0.1443, "step": 6650 }, { "epoch": 41.31152647975078, "grad_norm": 4.5, "learning_rate": 4.5294134638764395e-05, "loss": 0.3933, "step": 6651 }, { "epoch": 41.3177570093458, "grad_norm": 2.6875, "learning_rate": 4.5292695743167044e-05, "loss": 0.2414, "step": 6652 }, { "epoch": 41.32398753894081, "grad_norm": 2.140625, "learning_rate": 4.5291256650482974e-05, "loss": 0.1778, "step": 6653 }, { "epoch": 41.33021806853583, "grad_norm": 1.0234375, "learning_rate": 4.528981736072616e-05, "loss": 0.1178, "step": 6654 }, { "epoch": 41.33644859813084, "grad_norm": 2.40625, "learning_rate": 4.528837787391059e-05, "loss": 0.1615, "step": 6655 }, { "epoch": 41.34267912772586, "grad_norm": 0.8671875, "learning_rate": 4.528693819005024e-05, "loss": 0.0995, "step": 6656 }, { "epoch": 41.348909657320874, "grad_norm": 2.875, "learning_rate": 4.528549830915908e-05, "loss": 0.234, "step": 6657 }, { "epoch": 41.35514018691589, "grad_norm": 1.90625, "learning_rate": 4.5284058231251115e-05, "loss": 0.1497, "step": 6658 }, { "epoch": 41.361370716510905, "grad_norm": 2.3125, "learning_rate": 4.528261795634032e-05, "loss": 0.1831, "step": 6659 }, { "epoch": 41.36760124610592, "grad_norm": 1.8984375, "learning_rate": 4.528117748444069e-05, "loss": 0.1256, "step": 6660 }, { "epoch": 41.373831775700936, "grad_norm": 2.515625, "learning_rate": 4.52797368155662e-05, "loss": 0.1753, "step": 6661 }, { "epoch": 41.38006230529595, "grad_norm": 1.3828125, "learning_rate": 4.527829594973086e-05, "loss": 0.1215, "step": 6662 }, { "epoch": 41.386292834890966, "grad_norm": 1.4921875, "learning_rate": 4.5276854886948644e-05, "loss": 0.1196, "step": 6663 }, { "epoch": 41.39252336448598, "grad_norm": 2.734375, "learning_rate": 4.527541362723357e-05, "loss": 0.3042, "step": 6664 }, { "epoch": 41.398753894081, "grad_norm": 2.859375, "learning_rate": 4.527397217059962e-05, "loss": 0.2715, "step": 6665 }, { "epoch": 41.40498442367601, "grad_norm": 2.65625, "learning_rate": 4.5272530517060806e-05, "loss": 0.1917, "step": 6666 }, { "epoch": 41.41121495327103, "grad_norm": 2.0625, "learning_rate": 4.5271088666631126e-05, "loss": 0.1062, "step": 6667 }, { "epoch": 41.41744548286604, "grad_norm": 1.046875, "learning_rate": 4.526964661932457e-05, "loss": 0.1045, "step": 6668 }, { "epoch": 41.42367601246106, "grad_norm": 1.9375, "learning_rate": 4.526820437515516e-05, "loss": 0.1484, "step": 6669 }, { "epoch": 41.429906542056074, "grad_norm": 2.765625, "learning_rate": 4.5266761934136894e-05, "loss": 0.2401, "step": 6670 }, { "epoch": 41.43613707165109, "grad_norm": 1.4375, "learning_rate": 4.526531929628379e-05, "loss": 0.1184, "step": 6671 }, { "epoch": 41.442367601246104, "grad_norm": 2.046875, "learning_rate": 4.5263876461609856e-05, "loss": 0.1622, "step": 6672 }, { "epoch": 41.44859813084112, "grad_norm": 2.90625, "learning_rate": 4.526243343012909e-05, "loss": 0.2113, "step": 6673 }, { "epoch": 41.454828660436135, "grad_norm": 2.578125, "learning_rate": 4.526099020185553e-05, "loss": 0.2368, "step": 6674 }, { "epoch": 41.46105919003115, "grad_norm": 2.546875, "learning_rate": 4.5259546776803184e-05, "loss": 0.3263, "step": 6675 }, { "epoch": 41.467289719626166, "grad_norm": 4.1875, "learning_rate": 4.525810315498608e-05, "loss": 0.1642, "step": 6676 }, { "epoch": 41.47352024922118, "grad_norm": 3.609375, "learning_rate": 4.525665933641821e-05, "loss": 0.3232, "step": 6677 }, { "epoch": 41.4797507788162, "grad_norm": 3.0, "learning_rate": 4.525521532111362e-05, "loss": 0.1954, "step": 6678 }, { "epoch": 41.48598130841121, "grad_norm": 2.015625, "learning_rate": 4.5253771109086335e-05, "loss": 0.144, "step": 6679 }, { "epoch": 41.49221183800623, "grad_norm": 2.75, "learning_rate": 4.525232670035037e-05, "loss": 0.2883, "step": 6680 }, { "epoch": 41.49844236760124, "grad_norm": 3.875, "learning_rate": 4.525088209491977e-05, "loss": 0.3956, "step": 6681 }, { "epoch": 41.504672897196265, "grad_norm": 4.40625, "learning_rate": 4.5249437292808545e-05, "loss": 0.2178, "step": 6682 }, { "epoch": 41.51090342679128, "grad_norm": 2.15625, "learning_rate": 4.524799229403074e-05, "loss": 0.1504, "step": 6683 }, { "epoch": 41.517133956386296, "grad_norm": 1.8359375, "learning_rate": 4.524654709860039e-05, "loss": 0.1334, "step": 6684 }, { "epoch": 41.52336448598131, "grad_norm": 3.3125, "learning_rate": 4.524510170653152e-05, "loss": 0.2092, "step": 6685 }, { "epoch": 41.52959501557633, "grad_norm": 4.5625, "learning_rate": 4.524365611783818e-05, "loss": 0.2593, "step": 6686 }, { "epoch": 41.53582554517134, "grad_norm": 4.1875, "learning_rate": 4.524221033253441e-05, "loss": 0.2736, "step": 6687 }, { "epoch": 41.54205607476636, "grad_norm": 3.6875, "learning_rate": 4.524076435063423e-05, "loss": 0.1554, "step": 6688 }, { "epoch": 41.54828660436137, "grad_norm": 3.515625, "learning_rate": 4.5239318172151716e-05, "loss": 0.1524, "step": 6689 }, { "epoch": 41.55451713395639, "grad_norm": 2.984375, "learning_rate": 4.5237871797100894e-05, "loss": 0.1846, "step": 6690 }, { "epoch": 41.5607476635514, "grad_norm": 3.96875, "learning_rate": 4.523642522549582e-05, "loss": 0.1635, "step": 6691 }, { "epoch": 41.56697819314642, "grad_norm": 3.609375, "learning_rate": 4.523497845735054e-05, "loss": 0.1927, "step": 6692 }, { "epoch": 41.573208722741434, "grad_norm": 2.46875, "learning_rate": 4.523353149267909e-05, "loss": 0.1514, "step": 6693 }, { "epoch": 41.57943925233645, "grad_norm": 2.96875, "learning_rate": 4.523208433149555e-05, "loss": 0.1565, "step": 6694 }, { "epoch": 41.585669781931465, "grad_norm": 2.890625, "learning_rate": 4.5230636973813964e-05, "loss": 0.2065, "step": 6695 }, { "epoch": 41.59190031152648, "grad_norm": 3.921875, "learning_rate": 4.522918941964838e-05, "loss": 0.1279, "step": 6696 }, { "epoch": 41.598130841121495, "grad_norm": 4.53125, "learning_rate": 4.5227741669012866e-05, "loss": 0.2833, "step": 6697 }, { "epoch": 41.60436137071651, "grad_norm": 5.0, "learning_rate": 4.5226293721921483e-05, "loss": 0.1638, "step": 6698 }, { "epoch": 41.610591900311526, "grad_norm": 3.8125, "learning_rate": 4.52248455783883e-05, "loss": 0.2311, "step": 6699 }, { "epoch": 41.61682242990654, "grad_norm": 2.125, "learning_rate": 4.5223397238427364e-05, "loss": 0.2683, "step": 6700 }, { "epoch": 41.62305295950156, "grad_norm": 3.515625, "learning_rate": 4.5221948702052755e-05, "loss": 0.1191, "step": 6701 }, { "epoch": 41.62928348909657, "grad_norm": 4.96875, "learning_rate": 4.5220499969278534e-05, "loss": 0.2605, "step": 6702 }, { "epoch": 41.63551401869159, "grad_norm": 4.75, "learning_rate": 4.5219051040118775e-05, "loss": 0.2988, "step": 6703 }, { "epoch": 41.6417445482866, "grad_norm": 3.25, "learning_rate": 4.521760191458756e-05, "loss": 0.1358, "step": 6704 }, { "epoch": 41.64797507788162, "grad_norm": 2.8125, "learning_rate": 4.521615259269894e-05, "loss": 0.2352, "step": 6705 }, { "epoch": 41.654205607476634, "grad_norm": 4.59375, "learning_rate": 4.521470307446701e-05, "loss": 0.1087, "step": 6706 }, { "epoch": 41.66043613707165, "grad_norm": 4.1875, "learning_rate": 4.521325335990585e-05, "loss": 0.1722, "step": 6707 }, { "epoch": 41.666666666666664, "grad_norm": 5.09375, "learning_rate": 4.5211803449029534e-05, "loss": 0.1739, "step": 6708 }, { "epoch": 41.67289719626168, "grad_norm": 3.125, "learning_rate": 4.521035334185213e-05, "loss": 0.1255, "step": 6709 }, { "epoch": 41.679127725856695, "grad_norm": 1.9609375, "learning_rate": 4.5208903038387745e-05, "loss": 0.1191, "step": 6710 }, { "epoch": 41.68535825545171, "grad_norm": 3.34375, "learning_rate": 4.520745253865045e-05, "loss": 0.2915, "step": 6711 }, { "epoch": 41.691588785046726, "grad_norm": 3.34375, "learning_rate": 4.520600184265433e-05, "loss": 0.2449, "step": 6712 }, { "epoch": 41.69781931464174, "grad_norm": 4.28125, "learning_rate": 4.5204550950413495e-05, "loss": 0.2225, "step": 6713 }, { "epoch": 41.70404984423676, "grad_norm": 4.6875, "learning_rate": 4.520309986194201e-05, "loss": 0.2526, "step": 6714 }, { "epoch": 41.71028037383178, "grad_norm": 3.1875, "learning_rate": 4.5201648577253986e-05, "loss": 0.1787, "step": 6715 }, { "epoch": 41.716510903426794, "grad_norm": 1.765625, "learning_rate": 4.5200197096363516e-05, "loss": 0.1892, "step": 6716 }, { "epoch": 41.72274143302181, "grad_norm": 3.234375, "learning_rate": 4.519874541928469e-05, "loss": 0.1726, "step": 6717 }, { "epoch": 41.728971962616825, "grad_norm": 5.25, "learning_rate": 4.519729354603161e-05, "loss": 0.351, "step": 6718 }, { "epoch": 41.73520249221184, "grad_norm": 3.265625, "learning_rate": 4.519584147661838e-05, "loss": 0.1934, "step": 6719 }, { "epoch": 41.741433021806856, "grad_norm": 5.375, "learning_rate": 4.51943892110591e-05, "loss": 0.3275, "step": 6720 }, { "epoch": 41.74766355140187, "grad_norm": 1.8125, "learning_rate": 4.519293674936788e-05, "loss": 0.1351, "step": 6721 }, { "epoch": 41.753894080996886, "grad_norm": 2.84375, "learning_rate": 4.519148409155881e-05, "loss": 0.1888, "step": 6722 }, { "epoch": 41.7601246105919, "grad_norm": 3.125, "learning_rate": 4.519003123764602e-05, "loss": 0.2092, "step": 6723 }, { "epoch": 41.76635514018692, "grad_norm": 1.6796875, "learning_rate": 4.5188578187643605e-05, "loss": 0.1636, "step": 6724 }, { "epoch": 41.77258566978193, "grad_norm": 3.484375, "learning_rate": 4.5187124941565685e-05, "loss": 0.2214, "step": 6725 }, { "epoch": 41.77881619937695, "grad_norm": 1.7890625, "learning_rate": 4.518567149942637e-05, "loss": 0.1296, "step": 6726 }, { "epoch": 41.78504672897196, "grad_norm": 2.515625, "learning_rate": 4.5184217861239784e-05, "loss": 0.1717, "step": 6727 }, { "epoch": 41.79127725856698, "grad_norm": 2.71875, "learning_rate": 4.518276402702004e-05, "loss": 0.3033, "step": 6728 }, { "epoch": 41.797507788161994, "grad_norm": 3.734375, "learning_rate": 4.518130999678125e-05, "loss": 0.2594, "step": 6729 }, { "epoch": 41.80373831775701, "grad_norm": 3.15625, "learning_rate": 4.517985577053755e-05, "loss": 0.1741, "step": 6730 }, { "epoch": 41.809968847352025, "grad_norm": 2.9375, "learning_rate": 4.517840134830305e-05, "loss": 0.1332, "step": 6731 }, { "epoch": 41.81619937694704, "grad_norm": 2.28125, "learning_rate": 4.517694673009189e-05, "loss": 0.1332, "step": 6732 }, { "epoch": 41.822429906542055, "grad_norm": 2.046875, "learning_rate": 4.517549191591818e-05, "loss": 0.1792, "step": 6733 }, { "epoch": 41.82866043613707, "grad_norm": 2.359375, "learning_rate": 4.517403690579607e-05, "loss": 0.171, "step": 6734 }, { "epoch": 41.834890965732086, "grad_norm": 2.09375, "learning_rate": 4.517258169973968e-05, "loss": 0.1971, "step": 6735 }, { "epoch": 41.8411214953271, "grad_norm": 3.234375, "learning_rate": 4.517112629776314e-05, "loss": 0.192, "step": 6736 }, { "epoch": 41.84735202492212, "grad_norm": 2.171875, "learning_rate": 4.516967069988059e-05, "loss": 0.2436, "step": 6737 }, { "epoch": 41.85358255451713, "grad_norm": 2.28125, "learning_rate": 4.516821490610617e-05, "loss": 0.2234, "step": 6738 }, { "epoch": 41.85981308411215, "grad_norm": 1.828125, "learning_rate": 4.516675891645401e-05, "loss": 0.1513, "step": 6739 }, { "epoch": 41.86604361370716, "grad_norm": 2.59375, "learning_rate": 4.516530273093825e-05, "loss": 0.1888, "step": 6740 }, { "epoch": 41.87227414330218, "grad_norm": 3.125, "learning_rate": 4.516384634957305e-05, "loss": 0.2116, "step": 6741 }, { "epoch": 41.87850467289719, "grad_norm": 2.53125, "learning_rate": 4.516238977237254e-05, "loss": 0.197, "step": 6742 }, { "epoch": 41.88473520249221, "grad_norm": 3.890625, "learning_rate": 4.516093299935087e-05, "loss": 0.2778, "step": 6743 }, { "epoch": 41.890965732087224, "grad_norm": 2.890625, "learning_rate": 4.515947603052219e-05, "loss": 0.2721, "step": 6744 }, { "epoch": 41.89719626168224, "grad_norm": 3.40625, "learning_rate": 4.515801886590065e-05, "loss": 0.2855, "step": 6745 }, { "epoch": 41.90342679127726, "grad_norm": 3.390625, "learning_rate": 4.51565615055004e-05, "loss": 0.4505, "step": 6746 }, { "epoch": 41.90965732087228, "grad_norm": 1.671875, "learning_rate": 4.51551039493356e-05, "loss": 0.1514, "step": 6747 }, { "epoch": 41.91588785046729, "grad_norm": 3.0, "learning_rate": 4.515364619742039e-05, "loss": 0.2898, "step": 6748 }, { "epoch": 41.92211838006231, "grad_norm": 2.65625, "learning_rate": 4.515218824976895e-05, "loss": 0.1797, "step": 6749 }, { "epoch": 41.92834890965732, "grad_norm": 3.078125, "learning_rate": 4.5150730106395426e-05, "loss": 0.2267, "step": 6750 }, { "epoch": 41.93457943925234, "grad_norm": 1.9375, "learning_rate": 4.514927176731398e-05, "loss": 0.2777, "step": 6751 }, { "epoch": 41.940809968847354, "grad_norm": 3.25, "learning_rate": 4.5147813232538785e-05, "loss": 0.2333, "step": 6752 }, { "epoch": 41.94704049844237, "grad_norm": 2.265625, "learning_rate": 4.514635450208399e-05, "loss": 0.2158, "step": 6753 }, { "epoch": 41.953271028037385, "grad_norm": 2.046875, "learning_rate": 4.514489557596378e-05, "loss": 0.2546, "step": 6754 }, { "epoch": 41.9595015576324, "grad_norm": 3.65625, "learning_rate": 4.514343645419232e-05, "loss": 0.251, "step": 6755 }, { "epoch": 41.965732087227416, "grad_norm": 1.625, "learning_rate": 4.514197713678377e-05, "loss": 0.1429, "step": 6756 }, { "epoch": 41.97196261682243, "grad_norm": 3.046875, "learning_rate": 4.5140517623752324e-05, "loss": 0.1995, "step": 6757 }, { "epoch": 41.978193146417446, "grad_norm": 2.140625, "learning_rate": 4.5139057915112135e-05, "loss": 0.166, "step": 6758 }, { "epoch": 41.98442367601246, "grad_norm": 2.03125, "learning_rate": 4.513759801087739e-05, "loss": 0.1347, "step": 6759 }, { "epoch": 41.99065420560748, "grad_norm": 3.390625, "learning_rate": 4.513613791106227e-05, "loss": 0.3018, "step": 6760 }, { "epoch": 41.99688473520249, "grad_norm": 2.921875, "learning_rate": 4.513467761568095e-05, "loss": 0.1909, "step": 6761 }, { "epoch": 42.0, "grad_norm": 2.328125, "learning_rate": 4.513321712474763e-05, "loss": 0.2597, "step": 6762 }, { "epoch": 42.006230529595015, "grad_norm": 3.109375, "learning_rate": 4.513175643827647e-05, "loss": 0.1523, "step": 6763 }, { "epoch": 42.01246105919003, "grad_norm": 2.890625, "learning_rate": 4.513029555628167e-05, "loss": 0.1941, "step": 6764 }, { "epoch": 42.018691588785046, "grad_norm": 2.15625, "learning_rate": 4.512883447877741e-05, "loss": 0.1368, "step": 6765 }, { "epoch": 42.02492211838006, "grad_norm": 2.75, "learning_rate": 4.512737320577789e-05, "loss": 0.2624, "step": 6766 }, { "epoch": 42.03115264797508, "grad_norm": 2.203125, "learning_rate": 4.51259117372973e-05, "loss": 0.097, "step": 6767 }, { "epoch": 42.03738317757009, "grad_norm": 1.7421875, "learning_rate": 4.512445007334983e-05, "loss": 0.1272, "step": 6768 }, { "epoch": 42.04361370716511, "grad_norm": 1.90625, "learning_rate": 4.5122988213949676e-05, "loss": 0.1512, "step": 6769 }, { "epoch": 42.04984423676012, "grad_norm": 2.296875, "learning_rate": 4.5121526159111035e-05, "loss": 0.2203, "step": 6770 }, { "epoch": 42.05607476635514, "grad_norm": 1.84375, "learning_rate": 4.512006390884812e-05, "loss": 0.1322, "step": 6771 }, { "epoch": 42.06230529595015, "grad_norm": 2.546875, "learning_rate": 4.511860146317511e-05, "loss": 0.2368, "step": 6772 }, { "epoch": 42.06853582554517, "grad_norm": 3.515625, "learning_rate": 4.511713882210623e-05, "loss": 0.2095, "step": 6773 }, { "epoch": 42.074766355140184, "grad_norm": 4.84375, "learning_rate": 4.511567598565567e-05, "loss": 0.3668, "step": 6774 }, { "epoch": 42.0809968847352, "grad_norm": 2.125, "learning_rate": 4.511421295383765e-05, "loss": 0.1624, "step": 6775 }, { "epoch": 42.087227414330215, "grad_norm": 3.703125, "learning_rate": 4.5112749726666374e-05, "loss": 0.287, "step": 6776 }, { "epoch": 42.09345794392523, "grad_norm": 2.875, "learning_rate": 4.511128630415605e-05, "loss": 0.1808, "step": 6777 }, { "epoch": 42.099688473520246, "grad_norm": 2.734375, "learning_rate": 4.510982268632089e-05, "loss": 0.4194, "step": 6778 }, { "epoch": 42.10591900311527, "grad_norm": 2.625, "learning_rate": 4.510835887317512e-05, "loss": 0.2554, "step": 6779 }, { "epoch": 42.11214953271028, "grad_norm": 1.5234375, "learning_rate": 4.510689486473294e-05, "loss": 0.1451, "step": 6780 }, { "epoch": 42.1183800623053, "grad_norm": 1.890625, "learning_rate": 4.510543066100859e-05, "loss": 0.1505, "step": 6781 }, { "epoch": 42.124610591900314, "grad_norm": 2.9375, "learning_rate": 4.510396626201626e-05, "loss": 0.2813, "step": 6782 }, { "epoch": 42.13084112149533, "grad_norm": 2.859375, "learning_rate": 4.51025016677702e-05, "loss": 0.3243, "step": 6783 }, { "epoch": 42.137071651090345, "grad_norm": 1.8671875, "learning_rate": 4.5101036878284634e-05, "loss": 0.1539, "step": 6784 }, { "epoch": 42.14330218068536, "grad_norm": 2.484375, "learning_rate": 4.509957189357377e-05, "loss": 0.2435, "step": 6785 }, { "epoch": 42.149532710280376, "grad_norm": 1.6640625, "learning_rate": 4.5098106713651846e-05, "loss": 0.1836, "step": 6786 }, { "epoch": 42.15576323987539, "grad_norm": 3.515625, "learning_rate": 4.50966413385331e-05, "loss": 0.2455, "step": 6787 }, { "epoch": 42.161993769470406, "grad_norm": 2.984375, "learning_rate": 4.5095175768231744e-05, "loss": 0.2823, "step": 6788 }, { "epoch": 42.16822429906542, "grad_norm": 1.4609375, "learning_rate": 4.509371000276203e-05, "loss": 0.1277, "step": 6789 }, { "epoch": 42.17445482866044, "grad_norm": 2.359375, "learning_rate": 4.509224404213819e-05, "loss": 0.1437, "step": 6790 }, { "epoch": 42.18068535825545, "grad_norm": 2.890625, "learning_rate": 4.509077788637446e-05, "loss": 0.1987, "step": 6791 }, { "epoch": 42.18691588785047, "grad_norm": 2.890625, "learning_rate": 4.508931153548507e-05, "loss": 0.2217, "step": 6792 }, { "epoch": 42.19314641744548, "grad_norm": 2.265625, "learning_rate": 4.5087844989484284e-05, "loss": 0.1579, "step": 6793 }, { "epoch": 42.1993769470405, "grad_norm": 2.390625, "learning_rate": 4.5086378248386316e-05, "loss": 0.1914, "step": 6794 }, { "epoch": 42.205607476635514, "grad_norm": 2.265625, "learning_rate": 4.5084911312205444e-05, "loss": 0.1762, "step": 6795 }, { "epoch": 42.21183800623053, "grad_norm": 2.109375, "learning_rate": 4.5083444180955894e-05, "loss": 0.1205, "step": 6796 }, { "epoch": 42.218068535825545, "grad_norm": 3.453125, "learning_rate": 4.508197685465191e-05, "loss": 0.1832, "step": 6797 }, { "epoch": 42.22429906542056, "grad_norm": 4.125, "learning_rate": 4.5080509333307764e-05, "loss": 0.378, "step": 6798 }, { "epoch": 42.230529595015575, "grad_norm": 2.984375, "learning_rate": 4.507904161693769e-05, "loss": 0.1678, "step": 6799 }, { "epoch": 42.23676012461059, "grad_norm": 4.6875, "learning_rate": 4.507757370555595e-05, "loss": 0.347, "step": 6800 }, { "epoch": 42.242990654205606, "grad_norm": 2.578125, "learning_rate": 4.50761055991768e-05, "loss": 0.1485, "step": 6801 }, { "epoch": 42.24922118380062, "grad_norm": 3.078125, "learning_rate": 4.5074637297814506e-05, "loss": 0.1842, "step": 6802 }, { "epoch": 42.25545171339564, "grad_norm": 2.5, "learning_rate": 4.507316880148332e-05, "loss": 0.2078, "step": 6803 }, { "epoch": 42.26168224299065, "grad_norm": 1.515625, "learning_rate": 4.50717001101975e-05, "loss": 0.1244, "step": 6804 }, { "epoch": 42.26791277258567, "grad_norm": 2.8125, "learning_rate": 4.5070231223971326e-05, "loss": 0.1872, "step": 6805 }, { "epoch": 42.27414330218068, "grad_norm": 2.46875, "learning_rate": 4.5068762142819046e-05, "loss": 0.1458, "step": 6806 }, { "epoch": 42.2803738317757, "grad_norm": 2.171875, "learning_rate": 4.5067292866754937e-05, "loss": 0.1557, "step": 6807 }, { "epoch": 42.28660436137071, "grad_norm": 2.859375, "learning_rate": 4.506582339579327e-05, "loss": 0.1757, "step": 6808 }, { "epoch": 42.29283489096573, "grad_norm": 1.953125, "learning_rate": 4.5064353729948315e-05, "loss": 0.1804, "step": 6809 }, { "epoch": 42.299065420560744, "grad_norm": 3.234375, "learning_rate": 4.506288386923435e-05, "loss": 0.2549, "step": 6810 }, { "epoch": 42.30529595015577, "grad_norm": 2.578125, "learning_rate": 4.506141381366564e-05, "loss": 0.2049, "step": 6811 }, { "epoch": 42.31152647975078, "grad_norm": 3.96875, "learning_rate": 4.5059943563256476e-05, "loss": 0.4585, "step": 6812 }, { "epoch": 42.3177570093458, "grad_norm": 2.25, "learning_rate": 4.505847311802112e-05, "loss": 0.1464, "step": 6813 }, { "epoch": 42.32398753894081, "grad_norm": 2.390625, "learning_rate": 4.505700247797387e-05, "loss": 0.1508, "step": 6814 }, { "epoch": 42.33021806853583, "grad_norm": 2.6875, "learning_rate": 4.5055531643129004e-05, "loss": 0.2065, "step": 6815 }, { "epoch": 42.33644859813084, "grad_norm": 3.84375, "learning_rate": 4.5054060613500804e-05, "loss": 0.2112, "step": 6816 }, { "epoch": 42.34267912772586, "grad_norm": 3.375, "learning_rate": 4.5052589389103564e-05, "loss": 0.4356, "step": 6817 }, { "epoch": 42.348909657320874, "grad_norm": 3.515625, "learning_rate": 4.5051117969951556e-05, "loss": 0.3064, "step": 6818 }, { "epoch": 42.35514018691589, "grad_norm": 2.6875, "learning_rate": 4.504964635605909e-05, "loss": 0.163, "step": 6819 }, { "epoch": 42.361370716510905, "grad_norm": 3.109375, "learning_rate": 4.5048174547440445e-05, "loss": 0.266, "step": 6820 }, { "epoch": 42.36760124610592, "grad_norm": 2.484375, "learning_rate": 4.5046702544109926e-05, "loss": 0.2066, "step": 6821 }, { "epoch": 42.373831775700936, "grad_norm": 3.0625, "learning_rate": 4.504523034608182e-05, "loss": 0.1567, "step": 6822 }, { "epoch": 42.38006230529595, "grad_norm": 3.015625, "learning_rate": 4.504375795337044e-05, "loss": 0.2354, "step": 6823 }, { "epoch": 42.386292834890966, "grad_norm": 2.453125, "learning_rate": 4.504228536599007e-05, "loss": 0.1114, "step": 6824 }, { "epoch": 42.39252336448598, "grad_norm": 2.8125, "learning_rate": 4.5040812583955014e-05, "loss": 0.1837, "step": 6825 }, { "epoch": 42.398753894081, "grad_norm": 1.953125, "learning_rate": 4.503933960727959e-05, "loss": 0.1418, "step": 6826 }, { "epoch": 42.40498442367601, "grad_norm": 2.625, "learning_rate": 4.503786643597808e-05, "loss": 0.1935, "step": 6827 }, { "epoch": 42.41121495327103, "grad_norm": 2.546875, "learning_rate": 4.503639307006481e-05, "loss": 0.1823, "step": 6828 }, { "epoch": 42.41744548286604, "grad_norm": 2.765625, "learning_rate": 4.50349195095541e-05, "loss": 0.258, "step": 6829 }, { "epoch": 42.42367601246106, "grad_norm": 3.546875, "learning_rate": 4.5033445754460226e-05, "loss": 0.3422, "step": 6830 }, { "epoch": 42.429906542056074, "grad_norm": 3.390625, "learning_rate": 4.503197180479753e-05, "loss": 0.1993, "step": 6831 }, { "epoch": 42.43613707165109, "grad_norm": 2.671875, "learning_rate": 4.503049766058033e-05, "loss": 0.2058, "step": 6832 }, { "epoch": 42.442367601246104, "grad_norm": 1.53125, "learning_rate": 4.502902332182292e-05, "loss": 0.1193, "step": 6833 }, { "epoch": 42.44859813084112, "grad_norm": 2.96875, "learning_rate": 4.502754878853963e-05, "loss": 0.2103, "step": 6834 }, { "epoch": 42.454828660436135, "grad_norm": 3.390625, "learning_rate": 4.502607406074479e-05, "loss": 0.1938, "step": 6835 }, { "epoch": 42.46105919003115, "grad_norm": 2.6875, "learning_rate": 4.502459913845272e-05, "loss": 0.1866, "step": 6836 }, { "epoch": 42.467289719626166, "grad_norm": 1.4375, "learning_rate": 4.502312402167773e-05, "loss": 0.1133, "step": 6837 }, { "epoch": 42.47352024922118, "grad_norm": 2.671875, "learning_rate": 4.502164871043416e-05, "loss": 0.2621, "step": 6838 }, { "epoch": 42.4797507788162, "grad_norm": 1.8984375, "learning_rate": 4.502017320473634e-05, "loss": 0.1563, "step": 6839 }, { "epoch": 42.48598130841121, "grad_norm": 2.9375, "learning_rate": 4.5018697504598596e-05, "loss": 0.2397, "step": 6840 }, { "epoch": 42.49221183800623, "grad_norm": 2.34375, "learning_rate": 4.5017221610035254e-05, "loss": 0.2049, "step": 6841 }, { "epoch": 42.49844236760124, "grad_norm": 2.609375, "learning_rate": 4.501574552106066e-05, "loss": 0.2036, "step": 6842 }, { "epoch": 42.504672897196265, "grad_norm": 1.9609375, "learning_rate": 4.5014269237689136e-05, "loss": 0.2021, "step": 6843 }, { "epoch": 42.51090342679128, "grad_norm": 2.65625, "learning_rate": 4.501279275993504e-05, "loss": 0.19, "step": 6844 }, { "epoch": 42.517133956386296, "grad_norm": 3.25, "learning_rate": 4.501131608781269e-05, "loss": 0.2462, "step": 6845 }, { "epoch": 42.52336448598131, "grad_norm": 3.515625, "learning_rate": 4.500983922133644e-05, "loss": 0.3555, "step": 6846 }, { "epoch": 42.52959501557633, "grad_norm": 2.453125, "learning_rate": 4.5008362160520635e-05, "loss": 0.1813, "step": 6847 }, { "epoch": 42.53582554517134, "grad_norm": 3.859375, "learning_rate": 4.500688490537961e-05, "loss": 0.3211, "step": 6848 }, { "epoch": 42.54205607476636, "grad_norm": 3.5625, "learning_rate": 4.500540745592773e-05, "loss": 0.2089, "step": 6849 }, { "epoch": 42.54828660436137, "grad_norm": 2.484375, "learning_rate": 4.500392981217933e-05, "loss": 0.1378, "step": 6850 }, { "epoch": 42.55451713395639, "grad_norm": 3.609375, "learning_rate": 4.500245197414876e-05, "loss": 0.3175, "step": 6851 }, { "epoch": 42.5607476635514, "grad_norm": 2.09375, "learning_rate": 4.500097394185039e-05, "loss": 0.1763, "step": 6852 }, { "epoch": 42.56697819314642, "grad_norm": 1.5234375, "learning_rate": 4.499949571529855e-05, "loss": 0.1175, "step": 6853 }, { "epoch": 42.573208722741434, "grad_norm": 2.25, "learning_rate": 4.4998017294507615e-05, "loss": 0.1563, "step": 6854 }, { "epoch": 42.57943925233645, "grad_norm": 3.046875, "learning_rate": 4.499653867949194e-05, "loss": 0.2924, "step": 6855 }, { "epoch": 42.585669781931465, "grad_norm": 3.03125, "learning_rate": 4.499505987026588e-05, "loss": 0.2094, "step": 6856 }, { "epoch": 42.59190031152648, "grad_norm": 2.28125, "learning_rate": 4.499358086684381e-05, "loss": 0.1624, "step": 6857 }, { "epoch": 42.598130841121495, "grad_norm": 2.28125, "learning_rate": 4.4992101669240076e-05, "loss": 0.1769, "step": 6858 }, { "epoch": 42.60436137071651, "grad_norm": 2.796875, "learning_rate": 4.4990622277469057e-05, "loss": 0.2649, "step": 6859 }, { "epoch": 42.610591900311526, "grad_norm": 2.234375, "learning_rate": 4.498914269154512e-05, "loss": 0.1965, "step": 6860 }, { "epoch": 42.61682242990654, "grad_norm": 2.53125, "learning_rate": 4.498766291148263e-05, "loss": 0.1079, "step": 6861 }, { "epoch": 42.62305295950156, "grad_norm": 3.03125, "learning_rate": 4.4986182937295974e-05, "loss": 0.1563, "step": 6862 }, { "epoch": 42.62928348909657, "grad_norm": 4.0, "learning_rate": 4.49847027689995e-05, "loss": 0.2025, "step": 6863 }, { "epoch": 42.63551401869159, "grad_norm": 1.328125, "learning_rate": 4.498322240660761e-05, "loss": 0.1139, "step": 6864 }, { "epoch": 42.6417445482866, "grad_norm": 4.1875, "learning_rate": 4.498174185013466e-05, "loss": 0.154, "step": 6865 }, { "epoch": 42.64797507788162, "grad_norm": 3.875, "learning_rate": 4.498026109959504e-05, "loss": 0.2946, "step": 6866 }, { "epoch": 42.654205607476634, "grad_norm": 3.515625, "learning_rate": 4.497878015500313e-05, "loss": 0.1567, "step": 6867 }, { "epoch": 42.66043613707165, "grad_norm": 2.96875, "learning_rate": 4.497729901637332e-05, "loss": 0.2515, "step": 6868 }, { "epoch": 42.666666666666664, "grad_norm": 2.84375, "learning_rate": 4.497581768371998e-05, "loss": 0.219, "step": 6869 }, { "epoch": 42.67289719626168, "grad_norm": 1.9140625, "learning_rate": 4.497433615705751e-05, "loss": 0.1263, "step": 6870 }, { "epoch": 42.679127725856695, "grad_norm": 2.46875, "learning_rate": 4.49728544364003e-05, "loss": 0.2369, "step": 6871 }, { "epoch": 42.68535825545171, "grad_norm": 2.875, "learning_rate": 4.4971372521762725e-05, "loss": 0.1437, "step": 6872 }, { "epoch": 42.691588785046726, "grad_norm": 1.9921875, "learning_rate": 4.496989041315919e-05, "loss": 0.1282, "step": 6873 }, { "epoch": 42.69781931464174, "grad_norm": 3.1875, "learning_rate": 4.4968408110604095e-05, "loss": 0.191, "step": 6874 }, { "epoch": 42.70404984423676, "grad_norm": 3.015625, "learning_rate": 4.496692561411182e-05, "loss": 0.2698, "step": 6875 }, { "epoch": 42.71028037383178, "grad_norm": 2.6875, "learning_rate": 4.496544292369678e-05, "loss": 0.1677, "step": 6876 }, { "epoch": 42.716510903426794, "grad_norm": 1.8359375, "learning_rate": 4.496396003937336e-05, "loss": 0.1072, "step": 6877 }, { "epoch": 42.72274143302181, "grad_norm": 3.109375, "learning_rate": 4.4962476961155976e-05, "loss": 0.2709, "step": 6878 }, { "epoch": 42.728971962616825, "grad_norm": 1.4296875, "learning_rate": 4.496099368905902e-05, "loss": 0.1201, "step": 6879 }, { "epoch": 42.73520249221184, "grad_norm": 2.546875, "learning_rate": 4.49595102230969e-05, "loss": 0.197, "step": 6880 }, { "epoch": 42.741433021806856, "grad_norm": 2.265625, "learning_rate": 4.4958026563284035e-05, "loss": 0.1396, "step": 6881 }, { "epoch": 42.74766355140187, "grad_norm": 2.609375, "learning_rate": 4.495654270963482e-05, "loss": 0.1926, "step": 6882 }, { "epoch": 42.753894080996886, "grad_norm": 2.234375, "learning_rate": 4.4955058662163674e-05, "loss": 0.1023, "step": 6883 }, { "epoch": 42.7601246105919, "grad_norm": 3.4375, "learning_rate": 4.495357442088501e-05, "loss": 0.1607, "step": 6884 }, { "epoch": 42.76635514018692, "grad_norm": 2.640625, "learning_rate": 4.495208998581325e-05, "loss": 0.1489, "step": 6885 }, { "epoch": 42.77258566978193, "grad_norm": 3.328125, "learning_rate": 4.4950605356962794e-05, "loss": 0.2484, "step": 6886 }, { "epoch": 42.77881619937695, "grad_norm": 2.0625, "learning_rate": 4.494912053434808e-05, "loss": 0.1324, "step": 6887 }, { "epoch": 42.78504672897196, "grad_norm": 2.703125, "learning_rate": 4.494763551798351e-05, "loss": 0.1228, "step": 6888 }, { "epoch": 42.79127725856698, "grad_norm": 2.65625, "learning_rate": 4.4946150307883523e-05, "loss": 0.2293, "step": 6889 }, { "epoch": 42.797507788161994, "grad_norm": 3.0, "learning_rate": 4.4944664904062536e-05, "loss": 0.1548, "step": 6890 }, { "epoch": 42.80373831775701, "grad_norm": 3.59375, "learning_rate": 4.494317930653497e-05, "loss": 0.1827, "step": 6891 }, { "epoch": 42.809968847352025, "grad_norm": 2.546875, "learning_rate": 4.4941693515315266e-05, "loss": 0.1131, "step": 6892 }, { "epoch": 42.81619937694704, "grad_norm": 2.546875, "learning_rate": 4.494020753041785e-05, "loss": 0.1746, "step": 6893 }, { "epoch": 42.822429906542055, "grad_norm": 3.734375, "learning_rate": 4.493872135185715e-05, "loss": 0.2439, "step": 6894 }, { "epoch": 42.82866043613707, "grad_norm": 3.609375, "learning_rate": 4.4937234979647604e-05, "loss": 0.1642, "step": 6895 }, { "epoch": 42.834890965732086, "grad_norm": 4.375, "learning_rate": 4.4935748413803644e-05, "loss": 0.1595, "step": 6896 }, { "epoch": 42.8411214953271, "grad_norm": 4.46875, "learning_rate": 4.493426165433971e-05, "loss": 0.1639, "step": 6897 }, { "epoch": 42.84735202492212, "grad_norm": 2.734375, "learning_rate": 4.493277470127024e-05, "loss": 0.2609, "step": 6898 }, { "epoch": 42.85358255451713, "grad_norm": 2.40625, "learning_rate": 4.493128755460968e-05, "loss": 0.1658, "step": 6899 }, { "epoch": 42.85981308411215, "grad_norm": 4.90625, "learning_rate": 4.492980021437248e-05, "loss": 0.3705, "step": 6900 }, { "epoch": 42.86604361370716, "grad_norm": 5.53125, "learning_rate": 4.4928312680573064e-05, "loss": 0.3557, "step": 6901 }, { "epoch": 42.87227414330218, "grad_norm": 3.171875, "learning_rate": 4.49268249532259e-05, "loss": 0.1626, "step": 6902 }, { "epoch": 42.87850467289719, "grad_norm": 2.25, "learning_rate": 4.492533703234542e-05, "loss": 0.1819, "step": 6903 }, { "epoch": 42.88473520249221, "grad_norm": 2.0625, "learning_rate": 4.492384891794609e-05, "loss": 0.1364, "step": 6904 }, { "epoch": 42.890965732087224, "grad_norm": 3.0625, "learning_rate": 4.492236061004236e-05, "loss": 0.2326, "step": 6905 }, { "epoch": 42.89719626168224, "grad_norm": 1.203125, "learning_rate": 4.492087210864867e-05, "loss": 0.1139, "step": 6906 }, { "epoch": 42.90342679127726, "grad_norm": 1.9765625, "learning_rate": 4.4919383413779494e-05, "loss": 0.1694, "step": 6907 }, { "epoch": 42.90965732087228, "grad_norm": 2.421875, "learning_rate": 4.491789452544929e-05, "loss": 0.1988, "step": 6908 }, { "epoch": 42.91588785046729, "grad_norm": 3.921875, "learning_rate": 4.4916405443672506e-05, "loss": 0.1884, "step": 6909 }, { "epoch": 42.92211838006231, "grad_norm": 2.015625, "learning_rate": 4.4914916168463614e-05, "loss": 0.1181, "step": 6910 }, { "epoch": 42.92834890965732, "grad_norm": 3.328125, "learning_rate": 4.4913426699837066e-05, "loss": 0.1917, "step": 6911 }, { "epoch": 42.93457943925234, "grad_norm": 2.609375, "learning_rate": 4.491193703780735e-05, "loss": 0.1808, "step": 6912 }, { "epoch": 42.940809968847354, "grad_norm": 1.5859375, "learning_rate": 4.491044718238892e-05, "loss": 0.1461, "step": 6913 }, { "epoch": 42.94704049844237, "grad_norm": 3.078125, "learning_rate": 4.490895713359624e-05, "loss": 0.2859, "step": 6914 }, { "epoch": 42.953271028037385, "grad_norm": 2.078125, "learning_rate": 4.4907466891443796e-05, "loss": 0.1501, "step": 6915 }, { "epoch": 42.9595015576324, "grad_norm": 4.09375, "learning_rate": 4.490597645594605e-05, "loss": 0.2168, "step": 6916 }, { "epoch": 42.965732087227416, "grad_norm": 2.15625, "learning_rate": 4.4904485827117484e-05, "loss": 0.1456, "step": 6917 }, { "epoch": 42.97196261682243, "grad_norm": 1.9296875, "learning_rate": 4.490299500497257e-05, "loss": 0.2205, "step": 6918 }, { "epoch": 42.978193146417446, "grad_norm": 3.21875, "learning_rate": 4.490150398952579e-05, "loss": 0.2372, "step": 6919 }, { "epoch": 42.98442367601246, "grad_norm": 1.578125, "learning_rate": 4.490001278079162e-05, "loss": 0.101, "step": 6920 }, { "epoch": 42.99065420560748, "grad_norm": 3.15625, "learning_rate": 4.4898521378784557e-05, "loss": 0.2348, "step": 6921 }, { "epoch": 42.99688473520249, "grad_norm": 2.15625, "learning_rate": 4.4897029783519076e-05, "loss": 0.1339, "step": 6922 }, { "epoch": 43.0, "grad_norm": 1.0546875, "learning_rate": 4.489553799500966e-05, "loss": 0.0459, "step": 6923 }, { "epoch": 43.006230529595015, "grad_norm": 1.9296875, "learning_rate": 4.489404601327081e-05, "loss": 0.2584, "step": 6924 }, { "epoch": 43.01246105919003, "grad_norm": 1.1484375, "learning_rate": 4.4892553838316994e-05, "loss": 0.1223, "step": 6925 }, { "epoch": 43.018691588785046, "grad_norm": 2.71875, "learning_rate": 4.4891061470162735e-05, "loss": 0.2567, "step": 6926 }, { "epoch": 43.02492211838006, "grad_norm": 3.0625, "learning_rate": 4.48895689088225e-05, "loss": 0.2118, "step": 6927 }, { "epoch": 43.03115264797508, "grad_norm": 3.0625, "learning_rate": 4.48880761543108e-05, "loss": 0.1944, "step": 6928 }, { "epoch": 43.03738317757009, "grad_norm": 2.359375, "learning_rate": 4.488658320664212e-05, "loss": 0.1629, "step": 6929 }, { "epoch": 43.04361370716511, "grad_norm": 2.1875, "learning_rate": 4.488509006583098e-05, "loss": 0.2362, "step": 6930 }, { "epoch": 43.04984423676012, "grad_norm": 2.890625, "learning_rate": 4.488359673189186e-05, "loss": 0.2341, "step": 6931 }, { "epoch": 43.05607476635514, "grad_norm": 2.359375, "learning_rate": 4.488210320483929e-05, "loss": 0.1662, "step": 6932 }, { "epoch": 43.06230529595015, "grad_norm": 3.484375, "learning_rate": 4.4880609484687746e-05, "loss": 0.278, "step": 6933 }, { "epoch": 43.06853582554517, "grad_norm": 2.171875, "learning_rate": 4.487911557145175e-05, "loss": 0.1787, "step": 6934 }, { "epoch": 43.074766355140184, "grad_norm": 3.90625, "learning_rate": 4.4877621465145815e-05, "loss": 0.3371, "step": 6935 }, { "epoch": 43.0809968847352, "grad_norm": 2.25, "learning_rate": 4.487612716578444e-05, "loss": 0.1501, "step": 6936 }, { "epoch": 43.087227414330215, "grad_norm": 3.328125, "learning_rate": 4.487463267338215e-05, "loss": 0.2584, "step": 6937 }, { "epoch": 43.09345794392523, "grad_norm": 3.125, "learning_rate": 4.487313798795346e-05, "loss": 0.2772, "step": 6938 }, { "epoch": 43.099688473520246, "grad_norm": 2.140625, "learning_rate": 4.487164310951287e-05, "loss": 0.1522, "step": 6939 }, { "epoch": 43.10591900311527, "grad_norm": 2.390625, "learning_rate": 4.487014803807492e-05, "loss": 0.226, "step": 6940 }, { "epoch": 43.11214953271028, "grad_norm": 2.609375, "learning_rate": 4.486865277365411e-05, "loss": 0.1966, "step": 6941 }, { "epoch": 43.1183800623053, "grad_norm": 1.4609375, "learning_rate": 4.486715731626498e-05, "loss": 0.1451, "step": 6942 }, { "epoch": 43.124610591900314, "grad_norm": 2.921875, "learning_rate": 4.486566166592204e-05, "loss": 0.1588, "step": 6943 }, { "epoch": 43.13084112149533, "grad_norm": 2.28125, "learning_rate": 4.486416582263983e-05, "loss": 0.1167, "step": 6944 }, { "epoch": 43.137071651090345, "grad_norm": 2.875, "learning_rate": 4.4862669786432865e-05, "loss": 0.2014, "step": 6945 }, { "epoch": 43.14330218068536, "grad_norm": 1.3203125, "learning_rate": 4.4861173557315685e-05, "loss": 0.1155, "step": 6946 }, { "epoch": 43.149532710280376, "grad_norm": 2.953125, "learning_rate": 4.485967713530281e-05, "loss": 0.3106, "step": 6947 }, { "epoch": 43.15576323987539, "grad_norm": 2.4375, "learning_rate": 4.485818052040879e-05, "loss": 0.1826, "step": 6948 }, { "epoch": 43.161993769470406, "grad_norm": 2.640625, "learning_rate": 4.485668371264815e-05, "loss": 0.2459, "step": 6949 }, { "epoch": 43.16822429906542, "grad_norm": 3.390625, "learning_rate": 4.4855186712035423e-05, "loss": 0.2339, "step": 6950 }, { "epoch": 43.17445482866044, "grad_norm": 1.3203125, "learning_rate": 4.4853689518585165e-05, "loss": 0.1251, "step": 6951 }, { "epoch": 43.18068535825545, "grad_norm": 1.984375, "learning_rate": 4.48521921323119e-05, "loss": 0.1439, "step": 6952 }, { "epoch": 43.18691588785047, "grad_norm": 3.0625, "learning_rate": 4.485069455323017e-05, "loss": 0.242, "step": 6953 }, { "epoch": 43.19314641744548, "grad_norm": 2.984375, "learning_rate": 4.484919678135453e-05, "loss": 0.3036, "step": 6954 }, { "epoch": 43.1993769470405, "grad_norm": 3.359375, "learning_rate": 4.484769881669952e-05, "loss": 0.2552, "step": 6955 }, { "epoch": 43.205607476635514, "grad_norm": 3.25, "learning_rate": 4.48462006592797e-05, "loss": 0.1421, "step": 6956 }, { "epoch": 43.21183800623053, "grad_norm": 1.890625, "learning_rate": 4.484470230910961e-05, "loss": 0.1407, "step": 6957 }, { "epoch": 43.218068535825545, "grad_norm": 1.921875, "learning_rate": 4.48432037662038e-05, "loss": 0.1485, "step": 6958 }, { "epoch": 43.22429906542056, "grad_norm": 1.875, "learning_rate": 4.484170503057683e-05, "loss": 0.1623, "step": 6959 }, { "epoch": 43.230529595015575, "grad_norm": 2.28125, "learning_rate": 4.4840206102243254e-05, "loss": 0.1367, "step": 6960 }, { "epoch": 43.23676012461059, "grad_norm": 1.6171875, "learning_rate": 4.483870698121763e-05, "loss": 0.0999, "step": 6961 }, { "epoch": 43.242990654205606, "grad_norm": 2.25, "learning_rate": 4.4837207667514516e-05, "loss": 0.1692, "step": 6962 }, { "epoch": 43.24922118380062, "grad_norm": 2.953125, "learning_rate": 4.483570816114848e-05, "loss": 0.1631, "step": 6963 }, { "epoch": 43.25545171339564, "grad_norm": 3.234375, "learning_rate": 4.483420846213408e-05, "loss": 0.2728, "step": 6964 }, { "epoch": 43.26168224299065, "grad_norm": 1.59375, "learning_rate": 4.483270857048588e-05, "loss": 0.1489, "step": 6965 }, { "epoch": 43.26791277258567, "grad_norm": 3.5625, "learning_rate": 4.4831208486218454e-05, "loss": 0.3387, "step": 6966 }, { "epoch": 43.27414330218068, "grad_norm": 1.8046875, "learning_rate": 4.4829708209346365e-05, "loss": 0.2412, "step": 6967 }, { "epoch": 43.2803738317757, "grad_norm": 2.53125, "learning_rate": 4.4828207739884185e-05, "loss": 0.136, "step": 6968 }, { "epoch": 43.28660436137071, "grad_norm": 1.2265625, "learning_rate": 4.48267070778465e-05, "loss": 0.1055, "step": 6969 }, { "epoch": 43.29283489096573, "grad_norm": 1.953125, "learning_rate": 4.4825206223247855e-05, "loss": 0.1437, "step": 6970 }, { "epoch": 43.299065420560744, "grad_norm": 2.1875, "learning_rate": 4.4823705176102855e-05, "loss": 0.1465, "step": 6971 }, { "epoch": 43.30529595015577, "grad_norm": 2.65625, "learning_rate": 4.482220393642606e-05, "loss": 0.1712, "step": 6972 }, { "epoch": 43.31152647975078, "grad_norm": 3.015625, "learning_rate": 4.482070250423206e-05, "loss": 0.2829, "step": 6973 }, { "epoch": 43.3177570093458, "grad_norm": 1.6015625, "learning_rate": 4.481920087953544e-05, "loss": 0.1218, "step": 6974 }, { "epoch": 43.32398753894081, "grad_norm": 2.296875, "learning_rate": 4.481769906235078e-05, "loss": 0.1475, "step": 6975 }, { "epoch": 43.33021806853583, "grad_norm": 1.6328125, "learning_rate": 4.481619705269266e-05, "loss": 0.1156, "step": 6976 }, { "epoch": 43.33644859813084, "grad_norm": 2.421875, "learning_rate": 4.481469485057567e-05, "loss": 0.2512, "step": 6977 }, { "epoch": 43.34267912772586, "grad_norm": 3.3125, "learning_rate": 4.481319245601441e-05, "loss": 0.31, "step": 6978 }, { "epoch": 43.348909657320874, "grad_norm": 4.40625, "learning_rate": 4.481168986902347e-05, "loss": 0.273, "step": 6979 }, { "epoch": 43.35514018691589, "grad_norm": 3.671875, "learning_rate": 4.481018708961742e-05, "loss": 0.2246, "step": 6980 }, { "epoch": 43.361370716510905, "grad_norm": 1.6875, "learning_rate": 4.4808684117810885e-05, "loss": 0.1037, "step": 6981 }, { "epoch": 43.36760124610592, "grad_norm": 1.9609375, "learning_rate": 4.480718095361844e-05, "loss": 0.1595, "step": 6982 }, { "epoch": 43.373831775700936, "grad_norm": 2.953125, "learning_rate": 4.480567759705471e-05, "loss": 0.1152, "step": 6983 }, { "epoch": 43.38006230529595, "grad_norm": 2.8125, "learning_rate": 4.4804174048134276e-05, "loss": 0.1812, "step": 6984 }, { "epoch": 43.386292834890966, "grad_norm": 4.03125, "learning_rate": 4.480267030687174e-05, "loss": 0.3871, "step": 6985 }, { "epoch": 43.39252336448598, "grad_norm": 2.765625, "learning_rate": 4.480116637328171e-05, "loss": 0.2029, "step": 6986 }, { "epoch": 43.398753894081, "grad_norm": 2.78125, "learning_rate": 4.479966224737879e-05, "loss": 0.1819, "step": 6987 }, { "epoch": 43.40498442367601, "grad_norm": 2.453125, "learning_rate": 4.4798157929177596e-05, "loss": 0.3071, "step": 6988 }, { "epoch": 43.41121495327103, "grad_norm": 3.375, "learning_rate": 4.4796653418692745e-05, "loss": 0.1653, "step": 6989 }, { "epoch": 43.41744548286604, "grad_norm": 2.59375, "learning_rate": 4.479514871593883e-05, "loss": 0.1267, "step": 6990 }, { "epoch": 43.42367601246106, "grad_norm": 2.203125, "learning_rate": 4.479364382093048e-05, "loss": 0.1641, "step": 6991 }, { "epoch": 43.429906542056074, "grad_norm": 3.65625, "learning_rate": 4.479213873368229e-05, "loss": 0.147, "step": 6992 }, { "epoch": 43.43613707165109, "grad_norm": 4.5, "learning_rate": 4.4790633454208904e-05, "loss": 0.1933, "step": 6993 }, { "epoch": 43.442367601246104, "grad_norm": 3.21875, "learning_rate": 4.478912798252493e-05, "loss": 0.1612, "step": 6994 }, { "epoch": 43.44859813084112, "grad_norm": 3.140625, "learning_rate": 4.4787622318644984e-05, "loss": 0.2689, "step": 6995 }, { "epoch": 43.454828660436135, "grad_norm": 2.421875, "learning_rate": 4.478611646258369e-05, "loss": 0.2223, "step": 6996 }, { "epoch": 43.46105919003115, "grad_norm": 2.625, "learning_rate": 4.478461041435569e-05, "loss": 0.1764, "step": 6997 }, { "epoch": 43.467289719626166, "grad_norm": 2.46875, "learning_rate": 4.47831041739756e-05, "loss": 0.2342, "step": 6998 }, { "epoch": 43.47352024922118, "grad_norm": 2.765625, "learning_rate": 4.478159774145804e-05, "loss": 0.1642, "step": 6999 }, { "epoch": 43.4797507788162, "grad_norm": 1.9453125, "learning_rate": 4.4780091116817645e-05, "loss": 0.1432, "step": 7000 }, { "epoch": 43.48598130841121, "grad_norm": 2.328125, "learning_rate": 4.477858430006906e-05, "loss": 0.2665, "step": 7001 }, { "epoch": 43.49221183800623, "grad_norm": 2.53125, "learning_rate": 4.477707729122691e-05, "loss": 0.2234, "step": 7002 }, { "epoch": 43.49844236760124, "grad_norm": 1.9375, "learning_rate": 4.4775570090305825e-05, "loss": 0.1362, "step": 7003 }, { "epoch": 43.504672897196265, "grad_norm": 2.203125, "learning_rate": 4.477406269732045e-05, "loss": 0.2563, "step": 7004 }, { "epoch": 43.51090342679128, "grad_norm": 3.078125, "learning_rate": 4.477255511228543e-05, "loss": 0.3276, "step": 7005 }, { "epoch": 43.517133956386296, "grad_norm": 4.4375, "learning_rate": 4.47710473352154e-05, "loss": 0.3102, "step": 7006 }, { "epoch": 43.52336448598131, "grad_norm": 2.140625, "learning_rate": 4.476953936612501e-05, "loss": 0.1436, "step": 7007 }, { "epoch": 43.52959501557633, "grad_norm": 3.328125, "learning_rate": 4.47680312050289e-05, "loss": 0.2562, "step": 7008 }, { "epoch": 43.53582554517134, "grad_norm": 3.96875, "learning_rate": 4.476652285194172e-05, "loss": 0.1913, "step": 7009 }, { "epoch": 43.54205607476636, "grad_norm": 3.546875, "learning_rate": 4.476501430687812e-05, "loss": 0.2332, "step": 7010 }, { "epoch": 43.54828660436137, "grad_norm": 3.96875, "learning_rate": 4.4763505569852745e-05, "loss": 0.3588, "step": 7011 }, { "epoch": 43.55451713395639, "grad_norm": 3.59375, "learning_rate": 4.476199664088026e-05, "loss": 0.2662, "step": 7012 }, { "epoch": 43.5607476635514, "grad_norm": 3.328125, "learning_rate": 4.47604875199753e-05, "loss": 0.2528, "step": 7013 }, { "epoch": 43.56697819314642, "grad_norm": 3.421875, "learning_rate": 4.475897820715255e-05, "loss": 0.1977, "step": 7014 }, { "epoch": 43.573208722741434, "grad_norm": 2.046875, "learning_rate": 4.475746870242664e-05, "loss": 0.1682, "step": 7015 }, { "epoch": 43.57943925233645, "grad_norm": 1.8984375, "learning_rate": 4.4755959005812256e-05, "loss": 0.1772, "step": 7016 }, { "epoch": 43.585669781931465, "grad_norm": 2.515625, "learning_rate": 4.475444911732405e-05, "loss": 0.1339, "step": 7017 }, { "epoch": 43.59190031152648, "grad_norm": 3.96875, "learning_rate": 4.4752939036976684e-05, "loss": 0.2733, "step": 7018 }, { "epoch": 43.598130841121495, "grad_norm": 2.546875, "learning_rate": 4.475142876478482e-05, "loss": 0.1333, "step": 7019 }, { "epoch": 43.60436137071651, "grad_norm": 2.609375, "learning_rate": 4.474991830076313e-05, "loss": 0.1765, "step": 7020 }, { "epoch": 43.610591900311526, "grad_norm": 2.765625, "learning_rate": 4.474840764492629e-05, "loss": 0.2262, "step": 7021 }, { "epoch": 43.61682242990654, "grad_norm": 2.609375, "learning_rate": 4.4746896797288965e-05, "loss": 0.2563, "step": 7022 }, { "epoch": 43.62305295950156, "grad_norm": 1.9296875, "learning_rate": 4.4745385757865836e-05, "loss": 0.1321, "step": 7023 }, { "epoch": 43.62928348909657, "grad_norm": 1.2734375, "learning_rate": 4.474387452667157e-05, "loss": 0.1309, "step": 7024 }, { "epoch": 43.63551401869159, "grad_norm": 1.6015625, "learning_rate": 4.474236310372085e-05, "loss": 0.1092, "step": 7025 }, { "epoch": 43.6417445482866, "grad_norm": 2.59375, "learning_rate": 4.474085148902835e-05, "loss": 0.2385, "step": 7026 }, { "epoch": 43.64797507788162, "grad_norm": 1.1015625, "learning_rate": 4.473933968260876e-05, "loss": 0.1099, "step": 7027 }, { "epoch": 43.654205607476634, "grad_norm": 2.859375, "learning_rate": 4.473782768447675e-05, "loss": 0.2619, "step": 7028 }, { "epoch": 43.66043613707165, "grad_norm": 3.765625, "learning_rate": 4.4736315494647016e-05, "loss": 0.3316, "step": 7029 }, { "epoch": 43.666666666666664, "grad_norm": 3.625, "learning_rate": 4.473480311313424e-05, "loss": 0.3325, "step": 7030 }, { "epoch": 43.67289719626168, "grad_norm": 2.625, "learning_rate": 4.473329053995311e-05, "loss": 0.1744, "step": 7031 }, { "epoch": 43.679127725856695, "grad_norm": 2.953125, "learning_rate": 4.473177777511832e-05, "loss": 0.1836, "step": 7032 }, { "epoch": 43.68535825545171, "grad_norm": 3.171875, "learning_rate": 4.473026481864456e-05, "loss": 0.3017, "step": 7033 }, { "epoch": 43.691588785046726, "grad_norm": 1.4765625, "learning_rate": 4.472875167054652e-05, "loss": 0.1662, "step": 7034 }, { "epoch": 43.69781931464174, "grad_norm": 2.3125, "learning_rate": 4.4727238330838904e-05, "loss": 0.2122, "step": 7035 }, { "epoch": 43.70404984423676, "grad_norm": 3.28125, "learning_rate": 4.4725724799536405e-05, "loss": 0.2809, "step": 7036 }, { "epoch": 43.71028037383178, "grad_norm": 3.484375, "learning_rate": 4.472421107665372e-05, "loss": 0.2295, "step": 7037 }, { "epoch": 43.716510903426794, "grad_norm": 3.21875, "learning_rate": 4.4722697162205566e-05, "loss": 0.1347, "step": 7038 }, { "epoch": 43.72274143302181, "grad_norm": 2.328125, "learning_rate": 4.4721183056206614e-05, "loss": 0.147, "step": 7039 }, { "epoch": 43.728971962616825, "grad_norm": 2.375, "learning_rate": 4.4719668758671606e-05, "loss": 0.1457, "step": 7040 }, { "epoch": 43.73520249221184, "grad_norm": 1.8984375, "learning_rate": 4.471815426961523e-05, "loss": 0.137, "step": 7041 }, { "epoch": 43.741433021806856, "grad_norm": 2.390625, "learning_rate": 4.47166395890522e-05, "loss": 0.2224, "step": 7042 }, { "epoch": 43.74766355140187, "grad_norm": 1.3984375, "learning_rate": 4.471512471699722e-05, "loss": 0.1274, "step": 7043 }, { "epoch": 43.753894080996886, "grad_norm": 2.3125, "learning_rate": 4.471360965346501e-05, "loss": 0.1977, "step": 7044 }, { "epoch": 43.7601246105919, "grad_norm": 3.265625, "learning_rate": 4.4712094398470284e-05, "loss": 0.1778, "step": 7045 }, { "epoch": 43.76635514018692, "grad_norm": 2.265625, "learning_rate": 4.4710578952027755e-05, "loss": 0.1511, "step": 7046 }, { "epoch": 43.77258566978193, "grad_norm": 3.609375, "learning_rate": 4.470906331415215e-05, "loss": 0.1623, "step": 7047 }, { "epoch": 43.77881619937695, "grad_norm": 1.4609375, "learning_rate": 4.470754748485817e-05, "loss": 0.1259, "step": 7048 }, { "epoch": 43.78504672897196, "grad_norm": 2.734375, "learning_rate": 4.470603146416056e-05, "loss": 0.2995, "step": 7049 }, { "epoch": 43.79127725856698, "grad_norm": 2.390625, "learning_rate": 4.4704515252074016e-05, "loss": 0.2139, "step": 7050 }, { "epoch": 43.797507788161994, "grad_norm": 2.25, "learning_rate": 4.47029988486133e-05, "loss": 0.1637, "step": 7051 }, { "epoch": 43.80373831775701, "grad_norm": 2.734375, "learning_rate": 4.470148225379312e-05, "loss": 0.1804, "step": 7052 }, { "epoch": 43.809968847352025, "grad_norm": 2.140625, "learning_rate": 4.4699965467628194e-05, "loss": 0.1845, "step": 7053 }, { "epoch": 43.81619937694704, "grad_norm": 2.671875, "learning_rate": 4.4698448490133274e-05, "loss": 0.2532, "step": 7054 }, { "epoch": 43.822429906542055, "grad_norm": 1.59375, "learning_rate": 4.469693132132308e-05, "loss": 0.1148, "step": 7055 }, { "epoch": 43.82866043613707, "grad_norm": 1.9296875, "learning_rate": 4.469541396121236e-05, "loss": 0.1236, "step": 7056 }, { "epoch": 43.834890965732086, "grad_norm": 2.953125, "learning_rate": 4.4693896409815836e-05, "loss": 0.3123, "step": 7057 }, { "epoch": 43.8411214953271, "grad_norm": 3.75, "learning_rate": 4.469237866714825e-05, "loss": 0.2714, "step": 7058 }, { "epoch": 43.84735202492212, "grad_norm": 2.390625, "learning_rate": 4.469086073322435e-05, "loss": 0.1266, "step": 7059 }, { "epoch": 43.85358255451713, "grad_norm": 1.796875, "learning_rate": 4.468934260805888e-05, "loss": 0.1129, "step": 7060 }, { "epoch": 43.85981308411215, "grad_norm": 2.671875, "learning_rate": 4.468782429166657e-05, "loss": 0.2252, "step": 7061 }, { "epoch": 43.86604361370716, "grad_norm": 2.09375, "learning_rate": 4.468630578406218e-05, "loss": 0.1216, "step": 7062 }, { "epoch": 43.87227414330218, "grad_norm": 2.84375, "learning_rate": 4.4684787085260446e-05, "loss": 0.286, "step": 7063 }, { "epoch": 43.87850467289719, "grad_norm": 1.84375, "learning_rate": 4.4683268195276126e-05, "loss": 0.134, "step": 7064 }, { "epoch": 43.88473520249221, "grad_norm": 1.5625, "learning_rate": 4.4681749114123985e-05, "loss": 0.1355, "step": 7065 }, { "epoch": 43.890965732087224, "grad_norm": 2.140625, "learning_rate": 4.468022984181875e-05, "loss": 0.1397, "step": 7066 }, { "epoch": 43.89719626168224, "grad_norm": 1.8359375, "learning_rate": 4.467871037837519e-05, "loss": 0.162, "step": 7067 }, { "epoch": 43.90342679127726, "grad_norm": 2.703125, "learning_rate": 4.467719072380806e-05, "loss": 0.144, "step": 7068 }, { "epoch": 43.90965732087228, "grad_norm": 2.59375, "learning_rate": 4.467567087813212e-05, "loss": 0.289, "step": 7069 }, { "epoch": 43.91588785046729, "grad_norm": 2.8125, "learning_rate": 4.467415084136214e-05, "loss": 0.2438, "step": 7070 }, { "epoch": 43.92211838006231, "grad_norm": 1.9140625, "learning_rate": 4.467263061351287e-05, "loss": 0.1629, "step": 7071 }, { "epoch": 43.92834890965732, "grad_norm": 1.2265625, "learning_rate": 4.467111019459908e-05, "loss": 0.126, "step": 7072 }, { "epoch": 43.93457943925234, "grad_norm": 2.5, "learning_rate": 4.466958958463552e-05, "loss": 0.1447, "step": 7073 }, { "epoch": 43.940809968847354, "grad_norm": 2.75, "learning_rate": 4.466806878363699e-05, "loss": 0.2097, "step": 7074 }, { "epoch": 43.94704049844237, "grad_norm": 2.546875, "learning_rate": 4.466654779161824e-05, "loss": 0.2826, "step": 7075 }, { "epoch": 43.953271028037385, "grad_norm": 1.8046875, "learning_rate": 4.466502660859405e-05, "loss": 0.1704, "step": 7076 }, { "epoch": 43.9595015576324, "grad_norm": 1.8984375, "learning_rate": 4.4663505234579185e-05, "loss": 0.1474, "step": 7077 }, { "epoch": 43.965732087227416, "grad_norm": 2.359375, "learning_rate": 4.466198366958843e-05, "loss": 0.1432, "step": 7078 }, { "epoch": 43.97196261682243, "grad_norm": 1.8671875, "learning_rate": 4.466046191363656e-05, "loss": 0.1434, "step": 7079 }, { "epoch": 43.978193146417446, "grad_norm": 2.71875, "learning_rate": 4.4658939966738354e-05, "loss": 0.2064, "step": 7080 }, { "epoch": 43.98442367601246, "grad_norm": 2.859375, "learning_rate": 4.465741782890859e-05, "loss": 0.289, "step": 7081 }, { "epoch": 43.99065420560748, "grad_norm": 2.03125, "learning_rate": 4.465589550016206e-05, "loss": 0.1371, "step": 7082 }, { "epoch": 43.99688473520249, "grad_norm": 2.953125, "learning_rate": 4.4654372980513536e-05, "loss": 0.2055, "step": 7083 }, { "epoch": 44.0, "grad_norm": 1.5703125, "learning_rate": 4.465285026997782e-05, "loss": 0.0746, "step": 7084 }, { "epoch": 44.006230529595015, "grad_norm": 1.296875, "learning_rate": 4.465132736856969e-05, "loss": 0.0957, "step": 7085 }, { "epoch": 44.01246105919003, "grad_norm": 2.359375, "learning_rate": 4.464980427630394e-05, "loss": 0.2958, "step": 7086 }, { "epoch": 44.018691588785046, "grad_norm": 3.65625, "learning_rate": 4.464828099319536e-05, "loss": 0.2306, "step": 7087 }, { "epoch": 44.02492211838006, "grad_norm": 2.5625, "learning_rate": 4.464675751925875e-05, "loss": 0.1537, "step": 7088 }, { "epoch": 44.03115264797508, "grad_norm": 2.328125, "learning_rate": 4.464523385450891e-05, "loss": 0.2829, "step": 7089 }, { "epoch": 44.03738317757009, "grad_norm": 3.0, "learning_rate": 4.4643709998960624e-05, "loss": 0.3318, "step": 7090 }, { "epoch": 44.04361370716511, "grad_norm": 2.328125, "learning_rate": 4.4642185952628706e-05, "loss": 0.1434, "step": 7091 }, { "epoch": 44.04984423676012, "grad_norm": 2.6875, "learning_rate": 4.464066171552795e-05, "loss": 0.1406, "step": 7092 }, { "epoch": 44.05607476635514, "grad_norm": 1.4453125, "learning_rate": 4.4639137287673164e-05, "loss": 0.1175, "step": 7093 }, { "epoch": 44.06230529595015, "grad_norm": 1.328125, "learning_rate": 4.4637612669079145e-05, "loss": 0.1292, "step": 7094 }, { "epoch": 44.06853582554517, "grad_norm": 2.25, "learning_rate": 4.4636087859760704e-05, "loss": 0.2359, "step": 7095 }, { "epoch": 44.074766355140184, "grad_norm": 3.84375, "learning_rate": 4.463456285973265e-05, "loss": 0.2489, "step": 7096 }, { "epoch": 44.0809968847352, "grad_norm": 2.4375, "learning_rate": 4.4633037669009816e-05, "loss": 0.1191, "step": 7097 }, { "epoch": 44.087227414330215, "grad_norm": 1.7890625, "learning_rate": 4.463151228760698e-05, "loss": 0.172, "step": 7098 }, { "epoch": 44.09345794392523, "grad_norm": 3.09375, "learning_rate": 4.4629986715538976e-05, "loss": 0.2741, "step": 7099 }, { "epoch": 44.099688473520246, "grad_norm": 3.9375, "learning_rate": 4.462846095282062e-05, "loss": 0.4083, "step": 7100 }, { "epoch": 44.10591900311527, "grad_norm": 1.9609375, "learning_rate": 4.4626934999466715e-05, "loss": 0.163, "step": 7101 }, { "epoch": 44.11214953271028, "grad_norm": 3.1875, "learning_rate": 4.46254088554921e-05, "loss": 0.2975, "step": 7102 }, { "epoch": 44.1183800623053, "grad_norm": 4.0625, "learning_rate": 4.46238825209116e-05, "loss": 0.2686, "step": 7103 }, { "epoch": 44.124610591900314, "grad_norm": 2.109375, "learning_rate": 4.4622355995740015e-05, "loss": 0.1518, "step": 7104 }, { "epoch": 44.13084112149533, "grad_norm": 1.921875, "learning_rate": 4.4620829279992207e-05, "loss": 0.1169, "step": 7105 }, { "epoch": 44.137071651090345, "grad_norm": 3.125, "learning_rate": 4.461930237368297e-05, "loss": 0.2741, "step": 7106 }, { "epoch": 44.14330218068536, "grad_norm": 2.15625, "learning_rate": 4.461777527682714e-05, "loss": 0.2051, "step": 7107 }, { "epoch": 44.149532710280376, "grad_norm": 3.125, "learning_rate": 4.4616247989439565e-05, "loss": 0.1951, "step": 7108 }, { "epoch": 44.15576323987539, "grad_norm": 3.046875, "learning_rate": 4.461472051153507e-05, "loss": 0.1853, "step": 7109 }, { "epoch": 44.161993769470406, "grad_norm": 1.9921875, "learning_rate": 4.4613192843128484e-05, "loss": 0.1284, "step": 7110 }, { "epoch": 44.16822429906542, "grad_norm": 1.9765625, "learning_rate": 4.4611664984234646e-05, "loss": 0.167, "step": 7111 }, { "epoch": 44.17445482866044, "grad_norm": 2.859375, "learning_rate": 4.4610136934868394e-05, "loss": 0.2728, "step": 7112 }, { "epoch": 44.18068535825545, "grad_norm": 2.875, "learning_rate": 4.460860869504459e-05, "loss": 0.226, "step": 7113 }, { "epoch": 44.18691588785047, "grad_norm": 2.640625, "learning_rate": 4.460708026477804e-05, "loss": 0.2413, "step": 7114 }, { "epoch": 44.19314641744548, "grad_norm": 2.703125, "learning_rate": 4.4605551644083614e-05, "loss": 0.1912, "step": 7115 }, { "epoch": 44.1993769470405, "grad_norm": 2.4375, "learning_rate": 4.460402283297615e-05, "loss": 0.1895, "step": 7116 }, { "epoch": 44.205607476635514, "grad_norm": 2.5625, "learning_rate": 4.4602493831470505e-05, "loss": 0.247, "step": 7117 }, { "epoch": 44.21183800623053, "grad_norm": 3.484375, "learning_rate": 4.460096463958151e-05, "loss": 0.2983, "step": 7118 }, { "epoch": 44.218068535825545, "grad_norm": 1.8984375, "learning_rate": 4.459943525732404e-05, "loss": 0.1654, "step": 7119 }, { "epoch": 44.22429906542056, "grad_norm": 2.0, "learning_rate": 4.459790568471293e-05, "loss": 0.1736, "step": 7120 }, { "epoch": 44.230529595015575, "grad_norm": 1.8828125, "learning_rate": 4.459637592176304e-05, "loss": 0.1559, "step": 7121 }, { "epoch": 44.23676012461059, "grad_norm": 2.640625, "learning_rate": 4.459484596848923e-05, "loss": 0.2116, "step": 7122 }, { "epoch": 44.242990654205606, "grad_norm": 2.59375, "learning_rate": 4.4593315824906366e-05, "loss": 0.1879, "step": 7123 }, { "epoch": 44.24922118380062, "grad_norm": 2.0, "learning_rate": 4.45917854910293e-05, "loss": 0.1806, "step": 7124 }, { "epoch": 44.25545171339564, "grad_norm": 2.21875, "learning_rate": 4.459025496687289e-05, "loss": 0.1942, "step": 7125 }, { "epoch": 44.26168224299065, "grad_norm": 2.484375, "learning_rate": 4.458872425245201e-05, "loss": 0.2408, "step": 7126 }, { "epoch": 44.26791277258567, "grad_norm": 2.34375, "learning_rate": 4.458719334778153e-05, "loss": 0.1829, "step": 7127 }, { "epoch": 44.27414330218068, "grad_norm": 2.265625, "learning_rate": 4.458566225287631e-05, "loss": 0.282, "step": 7128 }, { "epoch": 44.2803738317757, "grad_norm": 3.578125, "learning_rate": 4.458413096775122e-05, "loss": 0.3425, "step": 7129 }, { "epoch": 44.28660436137071, "grad_norm": 2.40625, "learning_rate": 4.458259949242114e-05, "loss": 0.1572, "step": 7130 }, { "epoch": 44.29283489096573, "grad_norm": 2.078125, "learning_rate": 4.458106782690094e-05, "loss": 0.1627, "step": 7131 }, { "epoch": 44.299065420560744, "grad_norm": 2.015625, "learning_rate": 4.4579535971205495e-05, "loss": 0.1502, "step": 7132 }, { "epoch": 44.30529595015577, "grad_norm": 2.296875, "learning_rate": 4.4578003925349675e-05, "loss": 0.1233, "step": 7133 }, { "epoch": 44.31152647975078, "grad_norm": 3.671875, "learning_rate": 4.4576471689348375e-05, "loss": 0.3374, "step": 7134 }, { "epoch": 44.3177570093458, "grad_norm": 1.6640625, "learning_rate": 4.457493926321646e-05, "loss": 0.1679, "step": 7135 }, { "epoch": 44.32398753894081, "grad_norm": 1.3671875, "learning_rate": 4.457340664696883e-05, "loss": 0.1232, "step": 7136 }, { "epoch": 44.33021806853583, "grad_norm": 2.03125, "learning_rate": 4.4571873840620364e-05, "loss": 0.1929, "step": 7137 }, { "epoch": 44.33644859813084, "grad_norm": 2.703125, "learning_rate": 4.457034084418594e-05, "loss": 0.1815, "step": 7138 }, { "epoch": 44.34267912772586, "grad_norm": 2.9375, "learning_rate": 4.456880765768046e-05, "loss": 0.2629, "step": 7139 }, { "epoch": 44.348909657320874, "grad_norm": 1.890625, "learning_rate": 4.456727428111881e-05, "loss": 0.1232, "step": 7140 }, { "epoch": 44.35514018691589, "grad_norm": 2.671875, "learning_rate": 4.456574071451587e-05, "loss": 0.1975, "step": 7141 }, { "epoch": 44.361370716510905, "grad_norm": 3.296875, "learning_rate": 4.4564206957886554e-05, "loss": 0.2549, "step": 7142 }, { "epoch": 44.36760124610592, "grad_norm": 2.578125, "learning_rate": 4.456267301124575e-05, "loss": 0.164, "step": 7143 }, { "epoch": 44.373831775700936, "grad_norm": 2.859375, "learning_rate": 4.456113887460835e-05, "loss": 0.3997, "step": 7144 }, { "epoch": 44.38006230529595, "grad_norm": 2.078125, "learning_rate": 4.4559604547989256e-05, "loss": 0.1851, "step": 7145 }, { "epoch": 44.386292834890966, "grad_norm": 1.8515625, "learning_rate": 4.455807003140338e-05, "loss": 0.1718, "step": 7146 }, { "epoch": 44.39252336448598, "grad_norm": 2.890625, "learning_rate": 4.455653532486561e-05, "loss": 0.2643, "step": 7147 }, { "epoch": 44.398753894081, "grad_norm": 2.890625, "learning_rate": 4.4555000428390866e-05, "loss": 0.2044, "step": 7148 }, { "epoch": 44.40498442367601, "grad_norm": 1.984375, "learning_rate": 4.455346534199405e-05, "loss": 0.1532, "step": 7149 }, { "epoch": 44.41121495327103, "grad_norm": 3.09375, "learning_rate": 4.455193006569007e-05, "loss": 0.2792, "step": 7150 }, { "epoch": 44.41744548286604, "grad_norm": 2.4375, "learning_rate": 4.455039459949384e-05, "loss": 0.1647, "step": 7151 }, { "epoch": 44.42367601246106, "grad_norm": 1.96875, "learning_rate": 4.4548858943420257e-05, "loss": 0.1555, "step": 7152 }, { "epoch": 44.429906542056074, "grad_norm": 3.75, "learning_rate": 4.454732309748426e-05, "loss": 0.2996, "step": 7153 }, { "epoch": 44.43613707165109, "grad_norm": 2.921875, "learning_rate": 4.454578706170075e-05, "loss": 0.2684, "step": 7154 }, { "epoch": 44.442367601246104, "grad_norm": 3.046875, "learning_rate": 4.454425083608464e-05, "loss": 0.2586, "step": 7155 }, { "epoch": 44.44859813084112, "grad_norm": 3.78125, "learning_rate": 4.4542714420650874e-05, "loss": 0.3384, "step": 7156 }, { "epoch": 44.454828660436135, "grad_norm": 2.921875, "learning_rate": 4.454117781541435e-05, "loss": 0.1895, "step": 7157 }, { "epoch": 44.46105919003115, "grad_norm": 2.96875, "learning_rate": 4.453964102039e-05, "loss": 0.1716, "step": 7158 }, { "epoch": 44.467289719626166, "grad_norm": 1.578125, "learning_rate": 4.453810403559276e-05, "loss": 0.143, "step": 7159 }, { "epoch": 44.47352024922118, "grad_norm": 2.6875, "learning_rate": 4.453656686103754e-05, "loss": 0.258, "step": 7160 }, { "epoch": 44.4797507788162, "grad_norm": 3.125, "learning_rate": 4.453502949673928e-05, "loss": 0.1486, "step": 7161 }, { "epoch": 44.48598130841121, "grad_norm": 4.625, "learning_rate": 4.4533491942712907e-05, "loss": 0.2976, "step": 7162 }, { "epoch": 44.49221183800623, "grad_norm": 2.46875, "learning_rate": 4.453195419897336e-05, "loss": 0.1404, "step": 7163 }, { "epoch": 44.49844236760124, "grad_norm": 2.328125, "learning_rate": 4.4530416265535566e-05, "loss": 0.1521, "step": 7164 }, { "epoch": 44.504672897196265, "grad_norm": 2.515625, "learning_rate": 4.452887814241446e-05, "loss": 0.1935, "step": 7165 }, { "epoch": 44.51090342679128, "grad_norm": 1.9765625, "learning_rate": 4.4527339829624994e-05, "loss": 0.1258, "step": 7166 }, { "epoch": 44.517133956386296, "grad_norm": 3.171875, "learning_rate": 4.452580132718209e-05, "loss": 0.2085, "step": 7167 }, { "epoch": 44.52336448598131, "grad_norm": 2.71875, "learning_rate": 4.452426263510071e-05, "loss": 0.1392, "step": 7168 }, { "epoch": 44.52959501557633, "grad_norm": 2.734375, "learning_rate": 4.452272375339579e-05, "loss": 0.245, "step": 7169 }, { "epoch": 44.53582554517134, "grad_norm": 2.515625, "learning_rate": 4.4521184682082266e-05, "loss": 0.2163, "step": 7170 }, { "epoch": 44.54205607476636, "grad_norm": 3.359375, "learning_rate": 4.45196454211751e-05, "loss": 0.3182, "step": 7171 }, { "epoch": 44.54828660436137, "grad_norm": 2.40625, "learning_rate": 4.451810597068923e-05, "loss": 0.1519, "step": 7172 }, { "epoch": 44.55451713395639, "grad_norm": 2.265625, "learning_rate": 4.451656633063962e-05, "loss": 0.1668, "step": 7173 }, { "epoch": 44.5607476635514, "grad_norm": 2.125, "learning_rate": 4.451502650104121e-05, "loss": 0.1239, "step": 7174 }, { "epoch": 44.56697819314642, "grad_norm": 3.4375, "learning_rate": 4.4513486481908975e-05, "loss": 0.2102, "step": 7175 }, { "epoch": 44.573208722741434, "grad_norm": 2.96875, "learning_rate": 4.4511946273257846e-05, "loss": 0.2648, "step": 7176 }, { "epoch": 44.57943925233645, "grad_norm": 2.828125, "learning_rate": 4.451040587510279e-05, "loss": 0.1754, "step": 7177 }, { "epoch": 44.585669781931465, "grad_norm": 2.40625, "learning_rate": 4.4508865287458785e-05, "loss": 0.1714, "step": 7178 }, { "epoch": 44.59190031152648, "grad_norm": 2.9375, "learning_rate": 4.450732451034078e-05, "loss": 0.3708, "step": 7179 }, { "epoch": 44.598130841121495, "grad_norm": 2.5625, "learning_rate": 4.450578354376373e-05, "loss": 0.1809, "step": 7180 }, { "epoch": 44.60436137071651, "grad_norm": 1.9765625, "learning_rate": 4.4504242387742615e-05, "loss": 0.1929, "step": 7181 }, { "epoch": 44.610591900311526, "grad_norm": 3.765625, "learning_rate": 4.45027010422924e-05, "loss": 0.1412, "step": 7182 }, { "epoch": 44.61682242990654, "grad_norm": 3.6875, "learning_rate": 4.450115950742805e-05, "loss": 0.2027, "step": 7183 }, { "epoch": 44.62305295950156, "grad_norm": 3.453125, "learning_rate": 4.449961778316455e-05, "loss": 0.256, "step": 7184 }, { "epoch": 44.62928348909657, "grad_norm": 2.203125, "learning_rate": 4.449807586951685e-05, "loss": 0.1276, "step": 7185 }, { "epoch": 44.63551401869159, "grad_norm": 3.3125, "learning_rate": 4.449653376649996e-05, "loss": 0.1768, "step": 7186 }, { "epoch": 44.6417445482866, "grad_norm": 2.734375, "learning_rate": 4.449499147412882e-05, "loss": 0.1719, "step": 7187 }, { "epoch": 44.64797507788162, "grad_norm": 4.03125, "learning_rate": 4.449344899241843e-05, "loss": 0.242, "step": 7188 }, { "epoch": 44.654205607476634, "grad_norm": 2.0625, "learning_rate": 4.449190632138377e-05, "loss": 0.1259, "step": 7189 }, { "epoch": 44.66043613707165, "grad_norm": 1.359375, "learning_rate": 4.449036346103982e-05, "loss": 0.1304, "step": 7190 }, { "epoch": 44.666666666666664, "grad_norm": 3.125, "learning_rate": 4.448882041140155e-05, "loss": 0.282, "step": 7191 }, { "epoch": 44.67289719626168, "grad_norm": 2.640625, "learning_rate": 4.4487277172483976e-05, "loss": 0.2289, "step": 7192 }, { "epoch": 44.679127725856695, "grad_norm": 2.796875, "learning_rate": 4.448573374430207e-05, "loss": 0.1233, "step": 7193 }, { "epoch": 44.68535825545171, "grad_norm": 2.765625, "learning_rate": 4.448419012687082e-05, "loss": 0.2584, "step": 7194 }, { "epoch": 44.691588785046726, "grad_norm": 1.578125, "learning_rate": 4.4482646320205226e-05, "loss": 0.1695, "step": 7195 }, { "epoch": 44.69781931464174, "grad_norm": 2.53125, "learning_rate": 4.4481102324320264e-05, "loss": 0.1911, "step": 7196 }, { "epoch": 44.70404984423676, "grad_norm": 2.484375, "learning_rate": 4.447955813923095e-05, "loss": 0.289, "step": 7197 }, { "epoch": 44.71028037383178, "grad_norm": 1.8046875, "learning_rate": 4.447801376495227e-05, "loss": 0.134, "step": 7198 }, { "epoch": 44.716510903426794, "grad_norm": 2.546875, "learning_rate": 4.4476469201499227e-05, "loss": 0.1643, "step": 7199 }, { "epoch": 44.72274143302181, "grad_norm": 3.203125, "learning_rate": 4.447492444888682e-05, "loss": 0.2619, "step": 7200 }, { "epoch": 44.728971962616825, "grad_norm": 2.015625, "learning_rate": 4.4473379507130067e-05, "loss": 0.1669, "step": 7201 }, { "epoch": 44.73520249221184, "grad_norm": 1.875, "learning_rate": 4.447183437624395e-05, "loss": 0.1285, "step": 7202 }, { "epoch": 44.741433021806856, "grad_norm": 2.65625, "learning_rate": 4.447028905624348e-05, "loss": 0.2024, "step": 7203 }, { "epoch": 44.74766355140187, "grad_norm": 3.796875, "learning_rate": 4.446874354714368e-05, "loss": 0.2748, "step": 7204 }, { "epoch": 44.753894080996886, "grad_norm": 3.109375, "learning_rate": 4.446719784895954e-05, "loss": 0.2207, "step": 7205 }, { "epoch": 44.7601246105919, "grad_norm": 2.359375, "learning_rate": 4.4465651961706087e-05, "loss": 0.1236, "step": 7206 }, { "epoch": 44.76635514018692, "grad_norm": 2.96875, "learning_rate": 4.4464105885398336e-05, "loss": 0.2224, "step": 7207 }, { "epoch": 44.77258566978193, "grad_norm": 2.53125, "learning_rate": 4.4462559620051304e-05, "loss": 0.1328, "step": 7208 }, { "epoch": 44.77881619937695, "grad_norm": 2.953125, "learning_rate": 4.4461013165679986e-05, "loss": 0.2598, "step": 7209 }, { "epoch": 44.78504672897196, "grad_norm": 2.953125, "learning_rate": 4.445946652229943e-05, "loss": 0.1693, "step": 7210 }, { "epoch": 44.79127725856698, "grad_norm": 3.078125, "learning_rate": 4.445791968992464e-05, "loss": 0.2494, "step": 7211 }, { "epoch": 44.797507788161994, "grad_norm": 1.75, "learning_rate": 4.445637266857065e-05, "loss": 0.1714, "step": 7212 }, { "epoch": 44.80373831775701, "grad_norm": 1.3046875, "learning_rate": 4.445482545825247e-05, "loss": 0.1232, "step": 7213 }, { "epoch": 44.809968847352025, "grad_norm": 3.78125, "learning_rate": 4.4453278058985135e-05, "loss": 0.2608, "step": 7214 }, { "epoch": 44.81619937694704, "grad_norm": 4.28125, "learning_rate": 4.4451730470783686e-05, "loss": 0.2309, "step": 7215 }, { "epoch": 44.822429906542055, "grad_norm": 4.1875, "learning_rate": 4.445018269366314e-05, "loss": 0.3741, "step": 7216 }, { "epoch": 44.82866043613707, "grad_norm": 2.78125, "learning_rate": 4.444863472763853e-05, "loss": 0.2312, "step": 7217 }, { "epoch": 44.834890965732086, "grad_norm": 2.546875, "learning_rate": 4.444708657272489e-05, "loss": 0.1881, "step": 7218 }, { "epoch": 44.8411214953271, "grad_norm": 2.1875, "learning_rate": 4.4445538228937255e-05, "loss": 0.153, "step": 7219 }, { "epoch": 44.84735202492212, "grad_norm": 4.28125, "learning_rate": 4.4443989696290675e-05, "loss": 0.2344, "step": 7220 }, { "epoch": 44.85358255451713, "grad_norm": 3.8125, "learning_rate": 4.444244097480017e-05, "loss": 0.0981, "step": 7221 }, { "epoch": 44.85981308411215, "grad_norm": 3.90625, "learning_rate": 4.44408920644808e-05, "loss": 0.2738, "step": 7222 }, { "epoch": 44.86604361370716, "grad_norm": 2.421875, "learning_rate": 4.4439342965347595e-05, "loss": 0.1105, "step": 7223 }, { "epoch": 44.87227414330218, "grad_norm": 4.78125, "learning_rate": 4.443779367741561e-05, "loss": 0.2356, "step": 7224 }, { "epoch": 44.87850467289719, "grad_norm": 4.21875, "learning_rate": 4.4436244200699876e-05, "loss": 0.1386, "step": 7225 }, { "epoch": 44.88473520249221, "grad_norm": 4.96875, "learning_rate": 4.443469453521547e-05, "loss": 0.2369, "step": 7226 }, { "epoch": 44.890965732087224, "grad_norm": 3.5625, "learning_rate": 4.443314468097741e-05, "loss": 0.1733, "step": 7227 }, { "epoch": 44.89719626168224, "grad_norm": 3.359375, "learning_rate": 4.4431594638000774e-05, "loss": 0.2433, "step": 7228 }, { "epoch": 44.90342679127726, "grad_norm": 3.375, "learning_rate": 4.443004440630061e-05, "loss": 0.1757, "step": 7229 }, { "epoch": 44.90965732087228, "grad_norm": 4.0625, "learning_rate": 4.442849398589196e-05, "loss": 0.145, "step": 7230 }, { "epoch": 44.91588785046729, "grad_norm": 3.5625, "learning_rate": 4.44269433767899e-05, "loss": 0.1971, "step": 7231 }, { "epoch": 44.92211838006231, "grad_norm": 2.96875, "learning_rate": 4.442539257900947e-05, "loss": 0.1657, "step": 7232 }, { "epoch": 44.92834890965732, "grad_norm": 2.546875, "learning_rate": 4.442384159256576e-05, "loss": 0.1783, "step": 7233 }, { "epoch": 44.93457943925234, "grad_norm": 1.953125, "learning_rate": 4.442229041747381e-05, "loss": 0.1578, "step": 7234 }, { "epoch": 44.940809968847354, "grad_norm": 2.84375, "learning_rate": 4.442073905374869e-05, "loss": 0.1737, "step": 7235 }, { "epoch": 44.94704049844237, "grad_norm": 4.875, "learning_rate": 4.441918750140548e-05, "loss": 0.2975, "step": 7236 }, { "epoch": 44.953271028037385, "grad_norm": 4.59375, "learning_rate": 4.441763576045923e-05, "loss": 0.2034, "step": 7237 }, { "epoch": 44.9595015576324, "grad_norm": 3.375, "learning_rate": 4.441608383092503e-05, "loss": 0.2048, "step": 7238 }, { "epoch": 44.965732087227416, "grad_norm": 3.34375, "learning_rate": 4.441453171281793e-05, "loss": 0.1653, "step": 7239 }, { "epoch": 44.97196261682243, "grad_norm": 2.65625, "learning_rate": 4.441297940615303e-05, "loss": 0.1517, "step": 7240 }, { "epoch": 44.978193146417446, "grad_norm": 3.78125, "learning_rate": 4.441142691094539e-05, "loss": 0.166, "step": 7241 }, { "epoch": 44.98442367601246, "grad_norm": 3.328125, "learning_rate": 4.440987422721009e-05, "loss": 0.1539, "step": 7242 }, { "epoch": 44.99065420560748, "grad_norm": 3.6875, "learning_rate": 4.44083213549622e-05, "loss": 0.2523, "step": 7243 }, { "epoch": 44.99688473520249, "grad_norm": 3.625, "learning_rate": 4.440676829421683e-05, "loss": 0.2276, "step": 7244 }, { "epoch": 45.0, "grad_norm": 1.0546875, "learning_rate": 4.440521504498904e-05, "loss": 0.0692, "step": 7245 }, { "epoch": 45.006230529595015, "grad_norm": 5.9375, "learning_rate": 4.440366160729392e-05, "loss": 0.1883, "step": 7246 }, { "epoch": 45.01246105919003, "grad_norm": 6.46875, "learning_rate": 4.4402107981146566e-05, "loss": 0.1515, "step": 7247 }, { "epoch": 45.018691588785046, "grad_norm": 6.75, "learning_rate": 4.440055416656206e-05, "loss": 0.2474, "step": 7248 }, { "epoch": 45.02492211838006, "grad_norm": 6.6875, "learning_rate": 4.439900016355549e-05, "loss": 0.3472, "step": 7249 }, { "epoch": 45.03115264797508, "grad_norm": 3.203125, "learning_rate": 4.4397445972141956e-05, "loss": 0.1781, "step": 7250 }, { "epoch": 45.03738317757009, "grad_norm": 1.328125, "learning_rate": 4.4395891592336545e-05, "loss": 0.1208, "step": 7251 }, { "epoch": 45.04361370716511, "grad_norm": 4.5, "learning_rate": 4.439433702415436e-05, "loss": 0.1796, "step": 7252 }, { "epoch": 45.04984423676012, "grad_norm": 4.25, "learning_rate": 4.43927822676105e-05, "loss": 0.1294, "step": 7253 }, { "epoch": 45.05607476635514, "grad_norm": 3.359375, "learning_rate": 4.439122732272006e-05, "loss": 0.1529, "step": 7254 }, { "epoch": 45.06230529595015, "grad_norm": 1.90625, "learning_rate": 4.4389672189498146e-05, "loss": 0.1677, "step": 7255 }, { "epoch": 45.06853582554517, "grad_norm": 3.609375, "learning_rate": 4.438811686795985e-05, "loss": 0.3703, "step": 7256 }, { "epoch": 45.074766355140184, "grad_norm": 3.34375, "learning_rate": 4.438656135812029e-05, "loss": 0.1516, "step": 7257 }, { "epoch": 45.0809968847352, "grad_norm": 4.21875, "learning_rate": 4.4385005659994584e-05, "loss": 0.1615, "step": 7258 }, { "epoch": 45.087227414330215, "grad_norm": 4.875, "learning_rate": 4.438344977359782e-05, "loss": 0.1155, "step": 7259 }, { "epoch": 45.09345794392523, "grad_norm": 3.375, "learning_rate": 4.438189369894511e-05, "loss": 0.2369, "step": 7260 }, { "epoch": 45.099688473520246, "grad_norm": 1.890625, "learning_rate": 4.4380337436051586e-05, "loss": 0.1552, "step": 7261 }, { "epoch": 45.10591900311527, "grad_norm": 3.53125, "learning_rate": 4.4378780984932345e-05, "loss": 0.1886, "step": 7262 }, { "epoch": 45.11214953271028, "grad_norm": 4.34375, "learning_rate": 4.4377224345602505e-05, "loss": 0.1323, "step": 7263 }, { "epoch": 45.1183800623053, "grad_norm": 4.5, "learning_rate": 4.437566751807719e-05, "loss": 0.2176, "step": 7264 }, { "epoch": 45.124610591900314, "grad_norm": 3.640625, "learning_rate": 4.437411050237153e-05, "loss": 0.2117, "step": 7265 }, { "epoch": 45.13084112149533, "grad_norm": 1.7578125, "learning_rate": 4.437255329850063e-05, "loss": 0.191, "step": 7266 }, { "epoch": 45.137071651090345, "grad_norm": 1.7421875, "learning_rate": 4.4370995906479616e-05, "loss": 0.1359, "step": 7267 }, { "epoch": 45.14330218068536, "grad_norm": 3.671875, "learning_rate": 4.436943832632362e-05, "loss": 0.3605, "step": 7268 }, { "epoch": 45.149532710280376, "grad_norm": 3.984375, "learning_rate": 4.436788055804777e-05, "loss": 0.3596, "step": 7269 }, { "epoch": 45.15576323987539, "grad_norm": 3.453125, "learning_rate": 4.436632260166718e-05, "loss": 0.1995, "step": 7270 }, { "epoch": 45.161993769470406, "grad_norm": 2.015625, "learning_rate": 4.4364764457197005e-05, "loss": 0.0911, "step": 7271 }, { "epoch": 45.16822429906542, "grad_norm": 2.265625, "learning_rate": 4.436320612465237e-05, "loss": 0.276, "step": 7272 }, { "epoch": 45.17445482866044, "grad_norm": 3.890625, "learning_rate": 4.4361647604048394e-05, "loss": 0.2098, "step": 7273 }, { "epoch": 45.18068535825545, "grad_norm": 3.546875, "learning_rate": 4.436008889540023e-05, "loss": 0.2325, "step": 7274 }, { "epoch": 45.18691588785047, "grad_norm": 2.265625, "learning_rate": 4.435852999872302e-05, "loss": 0.1462, "step": 7275 }, { "epoch": 45.19314641744548, "grad_norm": 2.125, "learning_rate": 4.435697091403189e-05, "loss": 0.1686, "step": 7276 }, { "epoch": 45.1993769470405, "grad_norm": 2.328125, "learning_rate": 4.4355411641341996e-05, "loss": 0.1941, "step": 7277 }, { "epoch": 45.205607476635514, "grad_norm": 2.296875, "learning_rate": 4.435385218066847e-05, "loss": 0.157, "step": 7278 }, { "epoch": 45.21183800623053, "grad_norm": 2.5, "learning_rate": 4.435229253202646e-05, "loss": 0.1633, "step": 7279 }, { "epoch": 45.218068535825545, "grad_norm": 3.09375, "learning_rate": 4.435073269543113e-05, "loss": 0.1229, "step": 7280 }, { "epoch": 45.22429906542056, "grad_norm": 2.359375, "learning_rate": 4.43491726708976e-05, "loss": 0.1553, "step": 7281 }, { "epoch": 45.230529595015575, "grad_norm": 2.453125, "learning_rate": 4.4347612458441046e-05, "loss": 0.1815, "step": 7282 }, { "epoch": 45.23676012461059, "grad_norm": 2.90625, "learning_rate": 4.434605205807661e-05, "loss": 0.1748, "step": 7283 }, { "epoch": 45.242990654205606, "grad_norm": 2.515625, "learning_rate": 4.4344491469819446e-05, "loss": 0.1554, "step": 7284 }, { "epoch": 45.24922118380062, "grad_norm": 3.640625, "learning_rate": 4.4342930693684714e-05, "loss": 0.5125, "step": 7285 }, { "epoch": 45.25545171339564, "grad_norm": 3.296875, "learning_rate": 4.4341369729687575e-05, "loss": 0.1623, "step": 7286 }, { "epoch": 45.26168224299065, "grad_norm": 1.59375, "learning_rate": 4.433980857784319e-05, "loss": 0.153, "step": 7287 }, { "epoch": 45.26791277258567, "grad_norm": 2.40625, "learning_rate": 4.433824723816672e-05, "loss": 0.179, "step": 7288 }, { "epoch": 45.27414330218068, "grad_norm": 3.453125, "learning_rate": 4.433668571067332e-05, "loss": 0.175, "step": 7289 }, { "epoch": 45.2803738317757, "grad_norm": 2.578125, "learning_rate": 4.433512399537817e-05, "loss": 0.1652, "step": 7290 }, { "epoch": 45.28660436137071, "grad_norm": 2.25, "learning_rate": 4.4333562092296425e-05, "loss": 0.1144, "step": 7291 }, { "epoch": 45.29283489096573, "grad_norm": 2.703125, "learning_rate": 4.433200000144326e-05, "loss": 0.2883, "step": 7292 }, { "epoch": 45.299065420560744, "grad_norm": 2.625, "learning_rate": 4.433043772283385e-05, "loss": 0.1932, "step": 7293 }, { "epoch": 45.30529595015577, "grad_norm": 1.7265625, "learning_rate": 4.432887525648336e-05, "loss": 0.1096, "step": 7294 }, { "epoch": 45.31152647975078, "grad_norm": 2.703125, "learning_rate": 4.432731260240697e-05, "loss": 0.1836, "step": 7295 }, { "epoch": 45.3177570093458, "grad_norm": 2.6875, "learning_rate": 4.432574976061986e-05, "loss": 0.1373, "step": 7296 }, { "epoch": 45.32398753894081, "grad_norm": 1.8671875, "learning_rate": 4.432418673113721e-05, "loss": 0.1926, "step": 7297 }, { "epoch": 45.33021806853583, "grad_norm": 2.765625, "learning_rate": 4.432262351397419e-05, "loss": 0.2083, "step": 7298 }, { "epoch": 45.33644859813084, "grad_norm": 2.359375, "learning_rate": 4.4321060109145984e-05, "loss": 0.178, "step": 7299 }, { "epoch": 45.34267912772586, "grad_norm": 2.578125, "learning_rate": 4.431949651666779e-05, "loss": 0.2033, "step": 7300 }, { "epoch": 45.348909657320874, "grad_norm": 2.640625, "learning_rate": 4.431793273655477e-05, "loss": 0.2007, "step": 7301 }, { "epoch": 45.35514018691589, "grad_norm": 2.4375, "learning_rate": 4.4316368768822134e-05, "loss": 0.1385, "step": 7302 }, { "epoch": 45.361370716510905, "grad_norm": 2.140625, "learning_rate": 4.431480461348507e-05, "loss": 0.1704, "step": 7303 }, { "epoch": 45.36760124610592, "grad_norm": 1.4921875, "learning_rate": 4.431324027055875e-05, "loss": 0.1277, "step": 7304 }, { "epoch": 45.373831775700936, "grad_norm": 2.0625, "learning_rate": 4.431167574005839e-05, "loss": 0.1529, "step": 7305 }, { "epoch": 45.38006230529595, "grad_norm": 3.375, "learning_rate": 4.4310111021999165e-05, "loss": 0.222, "step": 7306 }, { "epoch": 45.386292834890966, "grad_norm": 2.171875, "learning_rate": 4.430854611639629e-05, "loss": 0.137, "step": 7307 }, { "epoch": 45.39252336448598, "grad_norm": 2.9375, "learning_rate": 4.430698102326495e-05, "loss": 0.2125, "step": 7308 }, { "epoch": 45.398753894081, "grad_norm": 2.90625, "learning_rate": 4.430541574262035e-05, "loss": 0.2266, "step": 7309 }, { "epoch": 45.40498442367601, "grad_norm": 1.84375, "learning_rate": 4.43038502744777e-05, "loss": 0.1637, "step": 7310 }, { "epoch": 45.41121495327103, "grad_norm": 4.21875, "learning_rate": 4.43022846188522e-05, "loss": 0.2881, "step": 7311 }, { "epoch": 45.41744548286604, "grad_norm": 3.78125, "learning_rate": 4.4300718775759045e-05, "loss": 0.1484, "step": 7312 }, { "epoch": 45.42367601246106, "grad_norm": 3.203125, "learning_rate": 4.4299152745213454e-05, "loss": 0.3379, "step": 7313 }, { "epoch": 45.429906542056074, "grad_norm": 2.75, "learning_rate": 4.4297586527230626e-05, "loss": 0.2059, "step": 7314 }, { "epoch": 45.43613707165109, "grad_norm": 3.15625, "learning_rate": 4.429602012182579e-05, "loss": 0.1142, "step": 7315 }, { "epoch": 45.442367601246104, "grad_norm": 3.03125, "learning_rate": 4.429445352901415e-05, "loss": 0.1753, "step": 7316 }, { "epoch": 45.44859813084112, "grad_norm": 4.53125, "learning_rate": 4.429288674881091e-05, "loss": 0.3865, "step": 7317 }, { "epoch": 45.454828660436135, "grad_norm": 1.953125, "learning_rate": 4.4291319781231314e-05, "loss": 0.1841, "step": 7318 }, { "epoch": 45.46105919003115, "grad_norm": 2.0625, "learning_rate": 4.428975262629055e-05, "loss": 0.1678, "step": 7319 }, { "epoch": 45.467289719626166, "grad_norm": 2.21875, "learning_rate": 4.428818528400386e-05, "loss": 0.1321, "step": 7320 }, { "epoch": 45.47352024922118, "grad_norm": 2.28125, "learning_rate": 4.428661775438646e-05, "loss": 0.2825, "step": 7321 }, { "epoch": 45.4797507788162, "grad_norm": 1.5078125, "learning_rate": 4.4285050037453565e-05, "loss": 0.1408, "step": 7322 }, { "epoch": 45.48598130841121, "grad_norm": 2.0, "learning_rate": 4.428348213322041e-05, "loss": 0.1196, "step": 7323 }, { "epoch": 45.49221183800623, "grad_norm": 1.015625, "learning_rate": 4.428191404170223e-05, "loss": 0.1077, "step": 7324 }, { "epoch": 45.49844236760124, "grad_norm": 2.78125, "learning_rate": 4.428034576291424e-05, "loss": 0.1488, "step": 7325 }, { "epoch": 45.504672897196265, "grad_norm": 3.328125, "learning_rate": 4.4278777296871676e-05, "loss": 0.2479, "step": 7326 }, { "epoch": 45.51090342679128, "grad_norm": 2.40625, "learning_rate": 4.4277208643589774e-05, "loss": 0.2298, "step": 7327 }, { "epoch": 45.517133956386296, "grad_norm": 3.90625, "learning_rate": 4.4275639803083776e-05, "loss": 0.135, "step": 7328 }, { "epoch": 45.52336448598131, "grad_norm": 3.640625, "learning_rate": 4.4274070775368906e-05, "loss": 0.1256, "step": 7329 }, { "epoch": 45.52959501557633, "grad_norm": 3.9375, "learning_rate": 4.4272501560460406e-05, "loss": 0.2485, "step": 7330 }, { "epoch": 45.53582554517134, "grad_norm": 1.265625, "learning_rate": 4.4270932158373515e-05, "loss": 0.1165, "step": 7331 }, { "epoch": 45.54205607476636, "grad_norm": 2.421875, "learning_rate": 4.4269362569123486e-05, "loss": 0.2151, "step": 7332 }, { "epoch": 45.54828660436137, "grad_norm": 2.578125, "learning_rate": 4.426779279272555e-05, "loss": 0.1578, "step": 7333 }, { "epoch": 45.55451713395639, "grad_norm": 3.5625, "learning_rate": 4.426622282919497e-05, "loss": 0.1405, "step": 7334 }, { "epoch": 45.5607476635514, "grad_norm": 2.859375, "learning_rate": 4.426465267854697e-05, "loss": 0.1477, "step": 7335 }, { "epoch": 45.56697819314642, "grad_norm": 1.859375, "learning_rate": 4.426308234079682e-05, "loss": 0.1521, "step": 7336 }, { "epoch": 45.573208722741434, "grad_norm": 2.859375, "learning_rate": 4.426151181595976e-05, "loss": 0.3208, "step": 7337 }, { "epoch": 45.57943925233645, "grad_norm": 4.1875, "learning_rate": 4.425994110405105e-05, "loss": 0.2296, "step": 7338 }, { "epoch": 45.585669781931465, "grad_norm": 3.875, "learning_rate": 4.425837020508594e-05, "loss": 0.117, "step": 7339 }, { "epoch": 45.59190031152648, "grad_norm": 5.65625, "learning_rate": 4.425679911907968e-05, "loss": 0.1976, "step": 7340 }, { "epoch": 45.598130841121495, "grad_norm": 1.7890625, "learning_rate": 4.425522784604755e-05, "loss": 0.1183, "step": 7341 }, { "epoch": 45.60436137071651, "grad_norm": 1.3828125, "learning_rate": 4.425365638600479e-05, "loss": 0.1465, "step": 7342 }, { "epoch": 45.610591900311526, "grad_norm": 5.5625, "learning_rate": 4.425208473896668e-05, "loss": 0.2277, "step": 7343 }, { "epoch": 45.61682242990654, "grad_norm": 4.4375, "learning_rate": 4.425051290494847e-05, "loss": 0.1439, "step": 7344 }, { "epoch": 45.62305295950156, "grad_norm": 3.3125, "learning_rate": 4.4248940883965425e-05, "loss": 0.1428, "step": 7345 }, { "epoch": 45.62928348909657, "grad_norm": 3.703125, "learning_rate": 4.424736867603283e-05, "loss": 0.2267, "step": 7346 }, { "epoch": 45.63551401869159, "grad_norm": 3.21875, "learning_rate": 4.424579628116594e-05, "loss": 0.2447, "step": 7347 }, { "epoch": 45.6417445482866, "grad_norm": 2.34375, "learning_rate": 4.424422369938002e-05, "loss": 0.2377, "step": 7348 }, { "epoch": 45.64797507788162, "grad_norm": 4.0625, "learning_rate": 4.424265093069036e-05, "loss": 0.2351, "step": 7349 }, { "epoch": 45.654205607476634, "grad_norm": 4.53125, "learning_rate": 4.4241077975112226e-05, "loss": 0.1566, "step": 7350 }, { "epoch": 45.66043613707165, "grad_norm": 4.75, "learning_rate": 4.423950483266089e-05, "loss": 0.1975, "step": 7351 }, { "epoch": 45.666666666666664, "grad_norm": 4.21875, "learning_rate": 4.423793150335165e-05, "loss": 0.175, "step": 7352 }, { "epoch": 45.67289719626168, "grad_norm": 2.9375, "learning_rate": 4.423635798719977e-05, "loss": 0.2131, "step": 7353 }, { "epoch": 45.679127725856695, "grad_norm": 3.5, "learning_rate": 4.423478428422053e-05, "loss": 0.3229, "step": 7354 }, { "epoch": 45.68535825545171, "grad_norm": 5.21875, "learning_rate": 4.423321039442922e-05, "loss": 0.2424, "step": 7355 }, { "epoch": 45.691588785046726, "grad_norm": 4.40625, "learning_rate": 4.423163631784113e-05, "loss": 0.2501, "step": 7356 }, { "epoch": 45.69781931464174, "grad_norm": 3.0, "learning_rate": 4.423006205447154e-05, "loss": 0.1606, "step": 7357 }, { "epoch": 45.70404984423676, "grad_norm": 3.515625, "learning_rate": 4.422848760433576e-05, "loss": 0.159, "step": 7358 }, { "epoch": 45.71028037383178, "grad_norm": 1.8359375, "learning_rate": 4.422691296744904e-05, "loss": 0.1357, "step": 7359 }, { "epoch": 45.716510903426794, "grad_norm": 2.109375, "learning_rate": 4.422533814382671e-05, "loss": 0.1628, "step": 7360 }, { "epoch": 45.72274143302181, "grad_norm": 4.8125, "learning_rate": 4.422376313348405e-05, "loss": 0.2061, "step": 7361 }, { "epoch": 45.728971962616825, "grad_norm": 4.78125, "learning_rate": 4.422218793643636e-05, "loss": 0.2108, "step": 7362 }, { "epoch": 45.73520249221184, "grad_norm": 3.609375, "learning_rate": 4.4220612552698947e-05, "loss": 0.2232, "step": 7363 }, { "epoch": 45.741433021806856, "grad_norm": 3.21875, "learning_rate": 4.421903698228709e-05, "loss": 0.1415, "step": 7364 }, { "epoch": 45.74766355140187, "grad_norm": 3.796875, "learning_rate": 4.4217461225216115e-05, "loss": 0.2782, "step": 7365 }, { "epoch": 45.753894080996886, "grad_norm": 3.140625, "learning_rate": 4.421588528150131e-05, "loss": 0.1851, "step": 7366 }, { "epoch": 45.7601246105919, "grad_norm": 2.734375, "learning_rate": 4.421430915115798e-05, "loss": 0.1725, "step": 7367 }, { "epoch": 45.76635514018692, "grad_norm": 3.828125, "learning_rate": 4.421273283420145e-05, "loss": 0.2356, "step": 7368 }, { "epoch": 45.77258566978193, "grad_norm": 3.65625, "learning_rate": 4.421115633064701e-05, "loss": 0.2703, "step": 7369 }, { "epoch": 45.77881619937695, "grad_norm": 1.28125, "learning_rate": 4.420957964050998e-05, "loss": 0.1267, "step": 7370 }, { "epoch": 45.78504672897196, "grad_norm": 2.078125, "learning_rate": 4.420800276380568e-05, "loss": 0.107, "step": 7371 }, { "epoch": 45.79127725856698, "grad_norm": 2.46875, "learning_rate": 4.4206425700549406e-05, "loss": 0.3162, "step": 7372 }, { "epoch": 45.797507788161994, "grad_norm": 2.28125, "learning_rate": 4.4204848450756496e-05, "loss": 0.1586, "step": 7373 }, { "epoch": 45.80373831775701, "grad_norm": 2.25, "learning_rate": 4.4203271014442255e-05, "loss": 0.1637, "step": 7374 }, { "epoch": 45.809968847352025, "grad_norm": 2.921875, "learning_rate": 4.4201693391622005e-05, "loss": 0.1354, "step": 7375 }, { "epoch": 45.81619937694704, "grad_norm": 3.109375, "learning_rate": 4.420011558231107e-05, "loss": 0.2108, "step": 7376 }, { "epoch": 45.822429906542055, "grad_norm": 3.0625, "learning_rate": 4.419853758652478e-05, "loss": 0.3401, "step": 7377 }, { "epoch": 45.82866043613707, "grad_norm": 2.171875, "learning_rate": 4.4196959404278456e-05, "loss": 0.2746, "step": 7378 }, { "epoch": 45.834890965732086, "grad_norm": 3.265625, "learning_rate": 4.419538103558742e-05, "loss": 0.2793, "step": 7379 }, { "epoch": 45.8411214953271, "grad_norm": 2.3125, "learning_rate": 4.419380248046701e-05, "loss": 0.205, "step": 7380 }, { "epoch": 45.84735202492212, "grad_norm": 2.75, "learning_rate": 4.4192223738932546e-05, "loss": 0.2479, "step": 7381 }, { "epoch": 45.85358255451713, "grad_norm": 2.03125, "learning_rate": 4.419064481099938e-05, "loss": 0.2192, "step": 7382 }, { "epoch": 45.85981308411215, "grad_norm": 1.6953125, "learning_rate": 4.418906569668282e-05, "loss": 0.1428, "step": 7383 }, { "epoch": 45.86604361370716, "grad_norm": 2.671875, "learning_rate": 4.4187486395998235e-05, "loss": 0.2286, "step": 7384 }, { "epoch": 45.87227414330218, "grad_norm": 2.328125, "learning_rate": 4.418590690896094e-05, "loss": 0.1526, "step": 7385 }, { "epoch": 45.87850467289719, "grad_norm": 2.0625, "learning_rate": 4.4184327235586273e-05, "loss": 0.1236, "step": 7386 }, { "epoch": 45.88473520249221, "grad_norm": 3.0625, "learning_rate": 4.41827473758896e-05, "loss": 0.3318, "step": 7387 }, { "epoch": 45.890965732087224, "grad_norm": 1.9609375, "learning_rate": 4.4181167329886244e-05, "loss": 0.1945, "step": 7388 }, { "epoch": 45.89719626168224, "grad_norm": 3.171875, "learning_rate": 4.4179587097591556e-05, "loss": 0.2798, "step": 7389 }, { "epoch": 45.90342679127726, "grad_norm": 2.734375, "learning_rate": 4.417800667902088e-05, "loss": 0.2862, "step": 7390 }, { "epoch": 45.90965732087228, "grad_norm": 3.125, "learning_rate": 4.417642607418957e-05, "loss": 0.2684, "step": 7391 }, { "epoch": 45.91588785046729, "grad_norm": 2.84375, "learning_rate": 4.417484528311299e-05, "loss": 0.1964, "step": 7392 }, { "epoch": 45.92211838006231, "grad_norm": 1.8203125, "learning_rate": 4.417326430580646e-05, "loss": 0.1102, "step": 7393 }, { "epoch": 45.92834890965732, "grad_norm": 2.15625, "learning_rate": 4.417168314228537e-05, "loss": 0.1386, "step": 7394 }, { "epoch": 45.93457943925234, "grad_norm": 3.828125, "learning_rate": 4.417010179256506e-05, "loss": 0.2478, "step": 7395 }, { "epoch": 45.940809968847354, "grad_norm": 2.84375, "learning_rate": 4.416852025666088e-05, "loss": 0.1566, "step": 7396 }, { "epoch": 45.94704049844237, "grad_norm": 2.390625, "learning_rate": 4.4166938534588205e-05, "loss": 0.2933, "step": 7397 }, { "epoch": 45.953271028037385, "grad_norm": 2.375, "learning_rate": 4.416535662636239e-05, "loss": 0.161, "step": 7398 }, { "epoch": 45.9595015576324, "grad_norm": 3.71875, "learning_rate": 4.41637745319988e-05, "loss": 0.2738, "step": 7399 }, { "epoch": 45.965732087227416, "grad_norm": 2.25, "learning_rate": 4.41621922515128e-05, "loss": 0.1381, "step": 7400 }, { "epoch": 45.97196261682243, "grad_norm": 4.0625, "learning_rate": 4.416060978491976e-05, "loss": 0.2903, "step": 7401 }, { "epoch": 45.978193146417446, "grad_norm": 1.8671875, "learning_rate": 4.4159027132235046e-05, "loss": 0.2153, "step": 7402 }, { "epoch": 45.98442367601246, "grad_norm": 2.109375, "learning_rate": 4.415744429347404e-05, "loss": 0.2134, "step": 7403 }, { "epoch": 45.99065420560748, "grad_norm": 2.1875, "learning_rate": 4.415586126865209e-05, "loss": 0.1748, "step": 7404 }, { "epoch": 45.99688473520249, "grad_norm": 2.421875, "learning_rate": 4.41542780577846e-05, "loss": 0.1991, "step": 7405 }, { "epoch": 46.0, "grad_norm": 1.3984375, "learning_rate": 4.4152694660886925e-05, "loss": 0.0838, "step": 7406 }, { "epoch": 46.006230529595015, "grad_norm": 2.65625, "learning_rate": 4.415111107797445e-05, "loss": 0.1846, "step": 7407 }, { "epoch": 46.01246105919003, "grad_norm": 2.625, "learning_rate": 4.414952730906256e-05, "loss": 0.2065, "step": 7408 }, { "epoch": 46.018691588785046, "grad_norm": 2.171875, "learning_rate": 4.414794335416663e-05, "loss": 0.1514, "step": 7409 }, { "epoch": 46.02492211838006, "grad_norm": 2.328125, "learning_rate": 4.4146359213302055e-05, "loss": 0.2241, "step": 7410 }, { "epoch": 46.03115264797508, "grad_norm": 3.390625, "learning_rate": 4.4144774886484205e-05, "loss": 0.1433, "step": 7411 }, { "epoch": 46.03738317757009, "grad_norm": 2.46875, "learning_rate": 4.4143190373728474e-05, "loss": 0.2479, "step": 7412 }, { "epoch": 46.04361370716511, "grad_norm": 2.6875, "learning_rate": 4.414160567505025e-05, "loss": 0.1876, "step": 7413 }, { "epoch": 46.04984423676012, "grad_norm": 2.21875, "learning_rate": 4.414002079046493e-05, "loss": 0.1703, "step": 7414 }, { "epoch": 46.05607476635514, "grad_norm": 1.65625, "learning_rate": 4.41384357199879e-05, "loss": 0.138, "step": 7415 }, { "epoch": 46.06230529595015, "grad_norm": 2.40625, "learning_rate": 4.413685046363455e-05, "loss": 0.1493, "step": 7416 }, { "epoch": 46.06853582554517, "grad_norm": 2.421875, "learning_rate": 4.413526502142029e-05, "loss": 0.1538, "step": 7417 }, { "epoch": 46.074766355140184, "grad_norm": 1.6640625, "learning_rate": 4.413367939336051e-05, "loss": 0.1334, "step": 7418 }, { "epoch": 46.0809968847352, "grad_norm": 2.609375, "learning_rate": 4.413209357947061e-05, "loss": 0.1819, "step": 7419 }, { "epoch": 46.087227414330215, "grad_norm": 2.390625, "learning_rate": 4.4130507579766e-05, "loss": 0.1661, "step": 7420 }, { "epoch": 46.09345794392523, "grad_norm": 2.109375, "learning_rate": 4.412892139426207e-05, "loss": 0.1609, "step": 7421 }, { "epoch": 46.099688473520246, "grad_norm": 2.15625, "learning_rate": 4.412733502297423e-05, "loss": 0.1512, "step": 7422 }, { "epoch": 46.10591900311527, "grad_norm": 2.921875, "learning_rate": 4.4125748465917894e-05, "loss": 0.3704, "step": 7423 }, { "epoch": 46.11214953271028, "grad_norm": 2.484375, "learning_rate": 4.4124161723108456e-05, "loss": 0.1026, "step": 7424 }, { "epoch": 46.1183800623053, "grad_norm": 2.515625, "learning_rate": 4.412257479456134e-05, "loss": 0.2341, "step": 7425 }, { "epoch": 46.124610591900314, "grad_norm": 2.15625, "learning_rate": 4.412098768029196e-05, "loss": 0.1751, "step": 7426 }, { "epoch": 46.13084112149533, "grad_norm": 3.640625, "learning_rate": 4.411940038031572e-05, "loss": 0.1517, "step": 7427 }, { "epoch": 46.137071651090345, "grad_norm": 1.4609375, "learning_rate": 4.411781289464805e-05, "loss": 0.1317, "step": 7428 }, { "epoch": 46.14330218068536, "grad_norm": 3.71875, "learning_rate": 4.411622522330435e-05, "loss": 0.3552, "step": 7429 }, { "epoch": 46.149532710280376, "grad_norm": 2.5, "learning_rate": 4.411463736630006e-05, "loss": 0.206, "step": 7430 }, { "epoch": 46.15576323987539, "grad_norm": 2.0, "learning_rate": 4.411304932365059e-05, "loss": 0.1669, "step": 7431 }, { "epoch": 46.161993769470406, "grad_norm": 2.296875, "learning_rate": 4.4111461095371356e-05, "loss": 0.1701, "step": 7432 }, { "epoch": 46.16822429906542, "grad_norm": 1.4921875, "learning_rate": 4.4109872681477796e-05, "loss": 0.1407, "step": 7433 }, { "epoch": 46.17445482866044, "grad_norm": 2.578125, "learning_rate": 4.410828408198533e-05, "loss": 0.203, "step": 7434 }, { "epoch": 46.18068535825545, "grad_norm": 2.5625, "learning_rate": 4.4106695296909384e-05, "loss": 0.2641, "step": 7435 }, { "epoch": 46.18691588785047, "grad_norm": 2.125, "learning_rate": 4.4105106326265404e-05, "loss": 0.2562, "step": 7436 }, { "epoch": 46.19314641744548, "grad_norm": 1.125, "learning_rate": 4.410351717006881e-05, "loss": 0.1044, "step": 7437 }, { "epoch": 46.1993769470405, "grad_norm": 1.5546875, "learning_rate": 4.410192782833503e-05, "loss": 0.1354, "step": 7438 }, { "epoch": 46.205607476635514, "grad_norm": 3.859375, "learning_rate": 4.410033830107953e-05, "loss": 0.489, "step": 7439 }, { "epoch": 46.21183800623053, "grad_norm": 2.546875, "learning_rate": 4.409874858831771e-05, "loss": 0.2374, "step": 7440 }, { "epoch": 46.218068535825545, "grad_norm": 2.0625, "learning_rate": 4.4097158690065023e-05, "loss": 0.1699, "step": 7441 }, { "epoch": 46.22429906542056, "grad_norm": 1.5, "learning_rate": 4.409556860633692e-05, "loss": 0.1376, "step": 7442 }, { "epoch": 46.230529595015575, "grad_norm": 1.4921875, "learning_rate": 4.4093978337148835e-05, "loss": 0.1038, "step": 7443 }, { "epoch": 46.23676012461059, "grad_norm": 1.8125, "learning_rate": 4.409238788251622e-05, "loss": 0.1363, "step": 7444 }, { "epoch": 46.242990654205606, "grad_norm": 1.7734375, "learning_rate": 4.409079724245451e-05, "loss": 0.1357, "step": 7445 }, { "epoch": 46.24922118380062, "grad_norm": 2.078125, "learning_rate": 4.408920641697917e-05, "loss": 0.1282, "step": 7446 }, { "epoch": 46.25545171339564, "grad_norm": 3.625, "learning_rate": 4.408761540610563e-05, "loss": 0.3276, "step": 7447 }, { "epoch": 46.26168224299065, "grad_norm": 3.40625, "learning_rate": 4.408602420984937e-05, "loss": 0.3922, "step": 7448 }, { "epoch": 46.26791277258567, "grad_norm": 2.5625, "learning_rate": 4.408443282822581e-05, "loss": 0.1111, "step": 7449 }, { "epoch": 46.27414330218068, "grad_norm": 2.328125, "learning_rate": 4.408284126125043e-05, "loss": 0.1747, "step": 7450 }, { "epoch": 46.2803738317757, "grad_norm": 3.28125, "learning_rate": 4.4081249508938684e-05, "loss": 0.1928, "step": 7451 }, { "epoch": 46.28660436137071, "grad_norm": 3.453125, "learning_rate": 4.407965757130602e-05, "loss": 0.2168, "step": 7452 }, { "epoch": 46.29283489096573, "grad_norm": 1.703125, "learning_rate": 4.407806544836792e-05, "loss": 0.184, "step": 7453 }, { "epoch": 46.299065420560744, "grad_norm": 3.875, "learning_rate": 4.407647314013983e-05, "loss": 0.1623, "step": 7454 }, { "epoch": 46.30529595015577, "grad_norm": 2.96875, "learning_rate": 4.4074880646637216e-05, "loss": 0.2134, "step": 7455 }, { "epoch": 46.31152647975078, "grad_norm": 2.546875, "learning_rate": 4.4073287967875555e-05, "loss": 0.2093, "step": 7456 }, { "epoch": 46.3177570093458, "grad_norm": 1.5390625, "learning_rate": 4.40716951038703e-05, "loss": 0.1179, "step": 7457 }, { "epoch": 46.32398753894081, "grad_norm": 2.234375, "learning_rate": 4.4070102054636934e-05, "loss": 0.1277, "step": 7458 }, { "epoch": 46.33021806853583, "grad_norm": 4.0625, "learning_rate": 4.406850882019092e-05, "loss": 0.1282, "step": 7459 }, { "epoch": 46.33644859813084, "grad_norm": 2.265625, "learning_rate": 4.406691540054775e-05, "loss": 0.1762, "step": 7460 }, { "epoch": 46.34267912772586, "grad_norm": 2.515625, "learning_rate": 4.406532179572287e-05, "loss": 0.2912, "step": 7461 }, { "epoch": 46.348909657320874, "grad_norm": 2.59375, "learning_rate": 4.4063728005731783e-05, "loss": 0.2087, "step": 7462 }, { "epoch": 46.35514018691589, "grad_norm": 2.765625, "learning_rate": 4.4062134030589964e-05, "loss": 0.2096, "step": 7463 }, { "epoch": 46.361370716510905, "grad_norm": 1.7734375, "learning_rate": 4.406053987031288e-05, "loss": 0.1288, "step": 7464 }, { "epoch": 46.36760124610592, "grad_norm": 2.375, "learning_rate": 4.405894552491603e-05, "loss": 0.252, "step": 7465 }, { "epoch": 46.373831775700936, "grad_norm": 2.578125, "learning_rate": 4.4057350994414884e-05, "loss": 0.1331, "step": 7466 }, { "epoch": 46.38006230529595, "grad_norm": 1.8359375, "learning_rate": 4.4055756278824936e-05, "loss": 0.154, "step": 7467 }, { "epoch": 46.386292834890966, "grad_norm": 2.578125, "learning_rate": 4.405416137816168e-05, "loss": 0.1954, "step": 7468 }, { "epoch": 46.39252336448598, "grad_norm": 1.25, "learning_rate": 4.405256629244059e-05, "loss": 0.1378, "step": 7469 }, { "epoch": 46.398753894081, "grad_norm": 1.4140625, "learning_rate": 4.405097102167718e-05, "loss": 0.1609, "step": 7470 }, { "epoch": 46.40498442367601, "grad_norm": 2.484375, "learning_rate": 4.404937556588692e-05, "loss": 0.2431, "step": 7471 }, { "epoch": 46.41121495327103, "grad_norm": 1.6171875, "learning_rate": 4.404777992508532e-05, "loss": 0.1165, "step": 7472 }, { "epoch": 46.41744548286604, "grad_norm": 2.3125, "learning_rate": 4.404618409928788e-05, "loss": 0.1843, "step": 7473 }, { "epoch": 46.42367601246106, "grad_norm": 2.296875, "learning_rate": 4.404458808851009e-05, "loss": 0.1597, "step": 7474 }, { "epoch": 46.429906542056074, "grad_norm": 2.78125, "learning_rate": 4.404299189276744e-05, "loss": 0.2265, "step": 7475 }, { "epoch": 46.43613707165109, "grad_norm": 2.265625, "learning_rate": 4.4041395512075464e-05, "loss": 0.1762, "step": 7476 }, { "epoch": 46.442367601246104, "grad_norm": 1.359375, "learning_rate": 4.403979894644964e-05, "loss": 0.1141, "step": 7477 }, { "epoch": 46.44859813084112, "grad_norm": 2.953125, "learning_rate": 4.403820219590548e-05, "loss": 0.302, "step": 7478 }, { "epoch": 46.454828660436135, "grad_norm": 1.6796875, "learning_rate": 4.40366052604585e-05, "loss": 0.1493, "step": 7479 }, { "epoch": 46.46105919003115, "grad_norm": 1.3984375, "learning_rate": 4.40350081401242e-05, "loss": 0.131, "step": 7480 }, { "epoch": 46.467289719626166, "grad_norm": 3.15625, "learning_rate": 4.4033410834918104e-05, "loss": 0.1729, "step": 7481 }, { "epoch": 46.47352024922118, "grad_norm": 2.875, "learning_rate": 4.403181334485571e-05, "loss": 0.166, "step": 7482 }, { "epoch": 46.4797507788162, "grad_norm": 3.6875, "learning_rate": 4.403021566995254e-05, "loss": 0.2164, "step": 7483 }, { "epoch": 46.48598130841121, "grad_norm": 2.078125, "learning_rate": 4.402861781022412e-05, "loss": 0.1594, "step": 7484 }, { "epoch": 46.49221183800623, "grad_norm": 3.515625, "learning_rate": 4.402701976568596e-05, "loss": 0.2909, "step": 7485 }, { "epoch": 46.49844236760124, "grad_norm": 2.96875, "learning_rate": 4.402542153635357e-05, "loss": 0.139, "step": 7486 }, { "epoch": 46.504672897196265, "grad_norm": 2.109375, "learning_rate": 4.4023823122242494e-05, "loss": 0.1019, "step": 7487 }, { "epoch": 46.51090342679128, "grad_norm": 3.0625, "learning_rate": 4.4022224523368233e-05, "loss": 0.2389, "step": 7488 }, { "epoch": 46.517133956386296, "grad_norm": 2.734375, "learning_rate": 4.4020625739746336e-05, "loss": 0.2294, "step": 7489 }, { "epoch": 46.52336448598131, "grad_norm": 3.4375, "learning_rate": 4.401902677139231e-05, "loss": 0.3396, "step": 7490 }, { "epoch": 46.52959501557633, "grad_norm": 2.578125, "learning_rate": 4.40174276183217e-05, "loss": 0.1993, "step": 7491 }, { "epoch": 46.53582554517134, "grad_norm": 2.328125, "learning_rate": 4.4015828280550034e-05, "loss": 0.265, "step": 7492 }, { "epoch": 46.54205607476636, "grad_norm": 2.1875, "learning_rate": 4.4014228758092834e-05, "loss": 0.1508, "step": 7493 }, { "epoch": 46.54828660436137, "grad_norm": 2.421875, "learning_rate": 4.4012629050965646e-05, "loss": 0.1558, "step": 7494 }, { "epoch": 46.55451713395639, "grad_norm": 1.8671875, "learning_rate": 4.4011029159184015e-05, "loss": 0.1264, "step": 7495 }, { "epoch": 46.5607476635514, "grad_norm": 1.3828125, "learning_rate": 4.400942908276345e-05, "loss": 0.1015, "step": 7496 }, { "epoch": 46.56697819314642, "grad_norm": 2.1875, "learning_rate": 4.400782882171952e-05, "loss": 0.2552, "step": 7497 }, { "epoch": 46.573208722741434, "grad_norm": 4.09375, "learning_rate": 4.400622837606776e-05, "loss": 0.2823, "step": 7498 }, { "epoch": 46.57943925233645, "grad_norm": 2.53125, "learning_rate": 4.400462774582371e-05, "loss": 0.1643, "step": 7499 }, { "epoch": 46.585669781931465, "grad_norm": 1.7890625, "learning_rate": 4.400302693100291e-05, "loss": 0.1159, "step": 7500 }, { "epoch": 46.59190031152648, "grad_norm": 1.328125, "learning_rate": 4.400142593162092e-05, "loss": 0.1508, "step": 7501 }, { "epoch": 46.598130841121495, "grad_norm": 1.5703125, "learning_rate": 4.399982474769328e-05, "loss": 0.1429, "step": 7502 }, { "epoch": 46.60436137071651, "grad_norm": 2.4375, "learning_rate": 4.399822337923555e-05, "loss": 0.2664, "step": 7503 }, { "epoch": 46.610591900311526, "grad_norm": 1.2890625, "learning_rate": 4.399662182626327e-05, "loss": 0.1259, "step": 7504 }, { "epoch": 46.61682242990654, "grad_norm": 1.703125, "learning_rate": 4.3995020088792e-05, "loss": 0.0969, "step": 7505 }, { "epoch": 46.62305295950156, "grad_norm": 2.765625, "learning_rate": 4.3993418166837295e-05, "loss": 0.1897, "step": 7506 }, { "epoch": 46.62928348909657, "grad_norm": 3.484375, "learning_rate": 4.3991816060414725e-05, "loss": 0.3161, "step": 7507 }, { "epoch": 46.63551401869159, "grad_norm": 1.6640625, "learning_rate": 4.399021376953985e-05, "loss": 0.1231, "step": 7508 }, { "epoch": 46.6417445482866, "grad_norm": 2.546875, "learning_rate": 4.39886112942282e-05, "loss": 0.2238, "step": 7509 }, { "epoch": 46.64797507788162, "grad_norm": 2.3125, "learning_rate": 4.398700863449537e-05, "loss": 0.2234, "step": 7510 }, { "epoch": 46.654205607476634, "grad_norm": 2.46875, "learning_rate": 4.3985405790356924e-05, "loss": 0.164, "step": 7511 }, { "epoch": 46.66043613707165, "grad_norm": 2.859375, "learning_rate": 4.398380276182842e-05, "loss": 0.2205, "step": 7512 }, { "epoch": 46.666666666666664, "grad_norm": 2.6875, "learning_rate": 4.398219954892542e-05, "loss": 0.1102, "step": 7513 }, { "epoch": 46.67289719626168, "grad_norm": 2.359375, "learning_rate": 4.398059615166351e-05, "loss": 0.1756, "step": 7514 }, { "epoch": 46.679127725856695, "grad_norm": 3.109375, "learning_rate": 4.3978992570058254e-05, "loss": 0.1886, "step": 7515 }, { "epoch": 46.68535825545171, "grad_norm": 2.203125, "learning_rate": 4.397738880412523e-05, "loss": 0.1449, "step": 7516 }, { "epoch": 46.691588785046726, "grad_norm": 3.28125, "learning_rate": 4.3975784853880005e-05, "loss": 0.306, "step": 7517 }, { "epoch": 46.69781931464174, "grad_norm": 2.21875, "learning_rate": 4.397418071933818e-05, "loss": 0.1446, "step": 7518 }, { "epoch": 46.70404984423676, "grad_norm": 3.234375, "learning_rate": 4.397257640051531e-05, "loss": 0.1317, "step": 7519 }, { "epoch": 46.71028037383178, "grad_norm": 3.09375, "learning_rate": 4.397097189742698e-05, "loss": 0.2306, "step": 7520 }, { "epoch": 46.716510903426794, "grad_norm": 3.03125, "learning_rate": 4.3969367210088785e-05, "loss": 0.3446, "step": 7521 }, { "epoch": 46.72274143302181, "grad_norm": 2.640625, "learning_rate": 4.39677623385163e-05, "loss": 0.1715, "step": 7522 }, { "epoch": 46.728971962616825, "grad_norm": 3.5625, "learning_rate": 4.396615728272513e-05, "loss": 0.2412, "step": 7523 }, { "epoch": 46.73520249221184, "grad_norm": 2.25, "learning_rate": 4.396455204273083e-05, "loss": 0.1354, "step": 7524 }, { "epoch": 46.741433021806856, "grad_norm": 1.234375, "learning_rate": 4.396294661854902e-05, "loss": 0.1149, "step": 7525 }, { "epoch": 46.74766355140187, "grad_norm": 2.5, "learning_rate": 4.3961341010195276e-05, "loss": 0.1968, "step": 7526 }, { "epoch": 46.753894080996886, "grad_norm": 3.765625, "learning_rate": 4.3959735217685205e-05, "loss": 0.118, "step": 7527 }, { "epoch": 46.7601246105919, "grad_norm": 4.0, "learning_rate": 4.3958129241034394e-05, "loss": 0.2299, "step": 7528 }, { "epoch": 46.76635514018692, "grad_norm": 3.53125, "learning_rate": 4.395652308025844e-05, "loss": 0.1538, "step": 7529 }, { "epoch": 46.77258566978193, "grad_norm": 2.421875, "learning_rate": 4.395491673537294e-05, "loss": 0.1452, "step": 7530 }, { "epoch": 46.77881619937695, "grad_norm": 1.9921875, "learning_rate": 4.395331020639351e-05, "loss": 0.1656, "step": 7531 }, { "epoch": 46.78504672897196, "grad_norm": 3.765625, "learning_rate": 4.3951703493335735e-05, "loss": 0.1216, "step": 7532 }, { "epoch": 46.79127725856698, "grad_norm": 4.21875, "learning_rate": 4.395009659621523e-05, "loss": 0.3083, "step": 7533 }, { "epoch": 46.797507788161994, "grad_norm": 3.90625, "learning_rate": 4.39484895150476e-05, "loss": 0.1432, "step": 7534 }, { "epoch": 46.80373831775701, "grad_norm": 3.625, "learning_rate": 4.394688224984844e-05, "loss": 0.1944, "step": 7535 }, { "epoch": 46.809968847352025, "grad_norm": 2.640625, "learning_rate": 4.394527480063339e-05, "loss": 0.262, "step": 7536 }, { "epoch": 46.81619937694704, "grad_norm": 2.84375, "learning_rate": 4.3943667167418034e-05, "loss": 0.1373, "step": 7537 }, { "epoch": 46.822429906542055, "grad_norm": 2.75, "learning_rate": 4.3942059350218e-05, "loss": 0.1896, "step": 7538 }, { "epoch": 46.82866043613707, "grad_norm": 4.09375, "learning_rate": 4.39404513490489e-05, "loss": 0.4241, "step": 7539 }, { "epoch": 46.834890965732086, "grad_norm": 2.859375, "learning_rate": 4.393884316392634e-05, "loss": 0.2357, "step": 7540 }, { "epoch": 46.8411214953271, "grad_norm": 3.3125, "learning_rate": 4.393723479486595e-05, "loss": 0.2426, "step": 7541 }, { "epoch": 46.84735202492212, "grad_norm": 1.375, "learning_rate": 4.393562624188337e-05, "loss": 0.1283, "step": 7542 }, { "epoch": 46.85358255451713, "grad_norm": 2.25, "learning_rate": 4.393401750499418e-05, "loss": 0.1154, "step": 7543 }, { "epoch": 46.85981308411215, "grad_norm": 3.640625, "learning_rate": 4.393240858421405e-05, "loss": 0.2662, "step": 7544 }, { "epoch": 46.86604361370716, "grad_norm": 1.7421875, "learning_rate": 4.393079947955856e-05, "loss": 0.1072, "step": 7545 }, { "epoch": 46.87227414330218, "grad_norm": 3.09375, "learning_rate": 4.3929190191043374e-05, "loss": 0.2629, "step": 7546 }, { "epoch": 46.87850467289719, "grad_norm": 1.53125, "learning_rate": 4.3927580718684106e-05, "loss": 0.1464, "step": 7547 }, { "epoch": 46.88473520249221, "grad_norm": 2.796875, "learning_rate": 4.392597106249639e-05, "loss": 0.3127, "step": 7548 }, { "epoch": 46.890965732087224, "grad_norm": 3.21875, "learning_rate": 4.3924361222495866e-05, "loss": 0.3017, "step": 7549 }, { "epoch": 46.89719626168224, "grad_norm": 3.203125, "learning_rate": 4.392275119869815e-05, "loss": 0.2884, "step": 7550 }, { "epoch": 46.90342679127726, "grad_norm": 2.5, "learning_rate": 4.39211409911189e-05, "loss": 0.185, "step": 7551 }, { "epoch": 46.90965732087228, "grad_norm": 2.734375, "learning_rate": 4.391953059977375e-05, "loss": 0.1381, "step": 7552 }, { "epoch": 46.91588785046729, "grad_norm": 2.578125, "learning_rate": 4.391792002467833e-05, "loss": 0.3518, "step": 7553 }, { "epoch": 46.92211838006231, "grad_norm": 2.75, "learning_rate": 4.391630926584829e-05, "loss": 0.1113, "step": 7554 }, { "epoch": 46.92834890965732, "grad_norm": 2.171875, "learning_rate": 4.391469832329927e-05, "loss": 0.1222, "step": 7555 }, { "epoch": 46.93457943925234, "grad_norm": 3.75, "learning_rate": 4.391308719704693e-05, "loss": 0.3908, "step": 7556 }, { "epoch": 46.940809968847354, "grad_norm": 2.96875, "learning_rate": 4.39114758871069e-05, "loss": 0.2373, "step": 7557 }, { "epoch": 46.94704049844237, "grad_norm": 1.328125, "learning_rate": 4.390986439349484e-05, "loss": 0.1304, "step": 7558 }, { "epoch": 46.953271028037385, "grad_norm": 2.109375, "learning_rate": 4.39082527162264e-05, "loss": 0.167, "step": 7559 }, { "epoch": 46.9595015576324, "grad_norm": 2.09375, "learning_rate": 4.3906640855317214e-05, "loss": 0.1867, "step": 7560 }, { "epoch": 46.965732087227416, "grad_norm": 3.546875, "learning_rate": 4.390502881078296e-05, "loss": 0.2567, "step": 7561 }, { "epoch": 46.97196261682243, "grad_norm": 2.78125, "learning_rate": 4.390341658263929e-05, "loss": 0.2547, "step": 7562 }, { "epoch": 46.978193146417446, "grad_norm": 1.765625, "learning_rate": 4.390180417090186e-05, "loss": 0.122, "step": 7563 }, { "epoch": 46.98442367601246, "grad_norm": 3.15625, "learning_rate": 4.3900191575586324e-05, "loss": 0.2773, "step": 7564 }, { "epoch": 46.99065420560748, "grad_norm": 2.890625, "learning_rate": 4.3898578796708355e-05, "loss": 0.2234, "step": 7565 }, { "epoch": 46.99688473520249, "grad_norm": 3.21875, "learning_rate": 4.389696583428361e-05, "loss": 0.2235, "step": 7566 }, { "epoch": 47.0, "grad_norm": 0.9921875, "learning_rate": 4.389535268832776e-05, "loss": 0.0503, "step": 7567 }, { "epoch": 47.006230529595015, "grad_norm": 3.796875, "learning_rate": 4.389373935885646e-05, "loss": 0.3209, "step": 7568 }, { "epoch": 47.01246105919003, "grad_norm": 3.4375, "learning_rate": 4.389212584588539e-05, "loss": 0.2791, "step": 7569 }, { "epoch": 47.018691588785046, "grad_norm": 2.890625, "learning_rate": 4.389051214943022e-05, "loss": 0.2384, "step": 7570 }, { "epoch": 47.02492211838006, "grad_norm": 2.828125, "learning_rate": 4.388889826950661e-05, "loss": 0.2135, "step": 7571 }, { "epoch": 47.03115264797508, "grad_norm": 2.703125, "learning_rate": 4.388728420613025e-05, "loss": 0.2108, "step": 7572 }, { "epoch": 47.03738317757009, "grad_norm": 2.140625, "learning_rate": 4.388566995931681e-05, "loss": 0.1273, "step": 7573 }, { "epoch": 47.04361370716511, "grad_norm": 2.46875, "learning_rate": 4.388405552908197e-05, "loss": 0.1435, "step": 7574 }, { "epoch": 47.04984423676012, "grad_norm": 5.0, "learning_rate": 4.38824409154414e-05, "loss": 0.2362, "step": 7575 }, { "epoch": 47.05607476635514, "grad_norm": 1.6015625, "learning_rate": 4.38808261184108e-05, "loss": 0.1471, "step": 7576 }, { "epoch": 47.06230529595015, "grad_norm": 1.765625, "learning_rate": 4.387921113800584e-05, "loss": 0.1458, "step": 7577 }, { "epoch": 47.06853582554517, "grad_norm": 3.25, "learning_rate": 4.38775959742422e-05, "loss": 0.2611, "step": 7578 }, { "epoch": 47.074766355140184, "grad_norm": 3.515625, "learning_rate": 4.387598062713558e-05, "loss": 0.1951, "step": 7579 }, { "epoch": 47.0809968847352, "grad_norm": 2.671875, "learning_rate": 4.387436509670167e-05, "loss": 0.2097, "step": 7580 }, { "epoch": 47.087227414330215, "grad_norm": 2.5, "learning_rate": 4.387274938295614e-05, "loss": 0.1578, "step": 7581 }, { "epoch": 47.09345794392523, "grad_norm": 1.9140625, "learning_rate": 4.3871133485914705e-05, "loss": 0.1369, "step": 7582 }, { "epoch": 47.099688473520246, "grad_norm": 1.6171875, "learning_rate": 4.386951740559305e-05, "loss": 0.1165, "step": 7583 }, { "epoch": 47.10591900311527, "grad_norm": 3.078125, "learning_rate": 4.386790114200686e-05, "loss": 0.2027, "step": 7584 }, { "epoch": 47.11214953271028, "grad_norm": 4.9375, "learning_rate": 4.386628469517185e-05, "loss": 0.3022, "step": 7585 }, { "epoch": 47.1183800623053, "grad_norm": 1.78125, "learning_rate": 4.386466806510372e-05, "loss": 0.1135, "step": 7586 }, { "epoch": 47.124610591900314, "grad_norm": 2.265625, "learning_rate": 4.386305125181815e-05, "loss": 0.1688, "step": 7587 }, { "epoch": 47.13084112149533, "grad_norm": 2.78125, "learning_rate": 4.386143425533086e-05, "loss": 0.1272, "step": 7588 }, { "epoch": 47.137071651090345, "grad_norm": 3.71875, "learning_rate": 4.385981707565754e-05, "loss": 0.1111, "step": 7589 }, { "epoch": 47.14330218068536, "grad_norm": 2.1875, "learning_rate": 4.385819971281392e-05, "loss": 0.118, "step": 7590 }, { "epoch": 47.149532710280376, "grad_norm": 2.625, "learning_rate": 4.385658216681569e-05, "loss": 0.1748, "step": 7591 }, { "epoch": 47.15576323987539, "grad_norm": 2.40625, "learning_rate": 4.385496443767857e-05, "loss": 0.1223, "step": 7592 }, { "epoch": 47.161993769470406, "grad_norm": 2.28125, "learning_rate": 4.385334652541826e-05, "loss": 0.1296, "step": 7593 }, { "epoch": 47.16822429906542, "grad_norm": 2.109375, "learning_rate": 4.3851728430050474e-05, "loss": 0.1473, "step": 7594 }, { "epoch": 47.17445482866044, "grad_norm": 2.765625, "learning_rate": 4.385011015159095e-05, "loss": 0.1686, "step": 7595 }, { "epoch": 47.18068535825545, "grad_norm": 3.15625, "learning_rate": 4.384849169005537e-05, "loss": 0.3117, "step": 7596 }, { "epoch": 47.18691588785047, "grad_norm": 1.40625, "learning_rate": 4.384687304545948e-05, "loss": 0.1454, "step": 7597 }, { "epoch": 47.19314641744548, "grad_norm": 2.375, "learning_rate": 4.3845254217818994e-05, "loss": 0.2069, "step": 7598 }, { "epoch": 47.1993769470405, "grad_norm": 3.15625, "learning_rate": 4.384363520714962e-05, "loss": 0.2774, "step": 7599 }, { "epoch": 47.205607476635514, "grad_norm": 2.515625, "learning_rate": 4.3842016013467114e-05, "loss": 0.2193, "step": 7600 }, { "epoch": 47.21183800623053, "grad_norm": 2.265625, "learning_rate": 4.384039663678716e-05, "loss": 0.1266, "step": 7601 }, { "epoch": 47.218068535825545, "grad_norm": 1.421875, "learning_rate": 4.383877707712552e-05, "loss": 0.1326, "step": 7602 }, { "epoch": 47.22429906542056, "grad_norm": 2.90625, "learning_rate": 4.3837157334497915e-05, "loss": 0.2677, "step": 7603 }, { "epoch": 47.230529595015575, "grad_norm": 2.09375, "learning_rate": 4.383553740892007e-05, "loss": 0.1899, "step": 7604 }, { "epoch": 47.23676012461059, "grad_norm": 1.7734375, "learning_rate": 4.3833917300407716e-05, "loss": 0.1365, "step": 7605 }, { "epoch": 47.242990654205606, "grad_norm": 1.2734375, "learning_rate": 4.38322970089766e-05, "loss": 0.1304, "step": 7606 }, { "epoch": 47.24922118380062, "grad_norm": 1.09375, "learning_rate": 4.383067653464244e-05, "loss": 0.1232, "step": 7607 }, { "epoch": 47.25545171339564, "grad_norm": 2.8125, "learning_rate": 4.3829055877420987e-05, "loss": 0.2834, "step": 7608 }, { "epoch": 47.26168224299065, "grad_norm": 2.296875, "learning_rate": 4.3827435037327985e-05, "loss": 0.1923, "step": 7609 }, { "epoch": 47.26791277258567, "grad_norm": 3.46875, "learning_rate": 4.382581401437917e-05, "loss": 0.2718, "step": 7610 }, { "epoch": 47.27414330218068, "grad_norm": 1.578125, "learning_rate": 4.382419280859028e-05, "loss": 0.1658, "step": 7611 }, { "epoch": 47.2803738317757, "grad_norm": 1.5546875, "learning_rate": 4.382257141997708e-05, "loss": 0.1605, "step": 7612 }, { "epoch": 47.28660436137071, "grad_norm": 2.75, "learning_rate": 4.3820949848555296e-05, "loss": 0.2707, "step": 7613 }, { "epoch": 47.29283489096573, "grad_norm": 2.265625, "learning_rate": 4.381932809434068e-05, "loss": 0.1671, "step": 7614 }, { "epoch": 47.299065420560744, "grad_norm": 2.46875, "learning_rate": 4.381770615734899e-05, "loss": 0.1652, "step": 7615 }, { "epoch": 47.30529595015577, "grad_norm": 3.046875, "learning_rate": 4.3816084037595976e-05, "loss": 0.256, "step": 7616 }, { "epoch": 47.31152647975078, "grad_norm": 2.03125, "learning_rate": 4.38144617350974e-05, "loss": 0.1491, "step": 7617 }, { "epoch": 47.3177570093458, "grad_norm": 2.421875, "learning_rate": 4.3812839249868995e-05, "loss": 0.1519, "step": 7618 }, { "epoch": 47.32398753894081, "grad_norm": 2.0625, "learning_rate": 4.381121658192655e-05, "loss": 0.1643, "step": 7619 }, { "epoch": 47.33021806853583, "grad_norm": 2.8125, "learning_rate": 4.38095937312858e-05, "loss": 0.2866, "step": 7620 }, { "epoch": 47.33644859813084, "grad_norm": 2.5, "learning_rate": 4.380797069796252e-05, "loss": 0.1829, "step": 7621 }, { "epoch": 47.34267912772586, "grad_norm": 1.5546875, "learning_rate": 4.380634748197247e-05, "loss": 0.1363, "step": 7622 }, { "epoch": 47.348909657320874, "grad_norm": 1.8203125, "learning_rate": 4.380472408333141e-05, "loss": 0.1878, "step": 7623 }, { "epoch": 47.35514018691589, "grad_norm": 3.453125, "learning_rate": 4.380310050205512e-05, "loss": 0.2344, "step": 7624 }, { "epoch": 47.361370716510905, "grad_norm": 1.875, "learning_rate": 4.3801476738159356e-05, "loss": 0.228, "step": 7625 }, { "epoch": 47.36760124610592, "grad_norm": 3.421875, "learning_rate": 4.379985279165988e-05, "loss": 0.2315, "step": 7626 }, { "epoch": 47.373831775700936, "grad_norm": 2.25, "learning_rate": 4.379822866257248e-05, "loss": 0.1583, "step": 7627 }, { "epoch": 47.38006230529595, "grad_norm": 2.328125, "learning_rate": 4.379660435091293e-05, "loss": 0.1233, "step": 7628 }, { "epoch": 47.386292834890966, "grad_norm": 1.6171875, "learning_rate": 4.3794979856697e-05, "loss": 0.1284, "step": 7629 }, { "epoch": 47.39252336448598, "grad_norm": 2.03125, "learning_rate": 4.379335517994048e-05, "loss": 0.1721, "step": 7630 }, { "epoch": 47.398753894081, "grad_norm": 3.453125, "learning_rate": 4.379173032065912e-05, "loss": 0.1737, "step": 7631 }, { "epoch": 47.40498442367601, "grad_norm": 3.421875, "learning_rate": 4.379010527886872e-05, "loss": 0.1777, "step": 7632 }, { "epoch": 47.41121495327103, "grad_norm": 3.203125, "learning_rate": 4.378848005458507e-05, "loss": 0.1405, "step": 7633 }, { "epoch": 47.41744548286604, "grad_norm": 3.15625, "learning_rate": 4.3786854647823944e-05, "loss": 0.3298, "step": 7634 }, { "epoch": 47.42367601246106, "grad_norm": 2.65625, "learning_rate": 4.378522905860113e-05, "loss": 0.1678, "step": 7635 }, { "epoch": 47.429906542056074, "grad_norm": 2.296875, "learning_rate": 4.378360328693241e-05, "loss": 0.1624, "step": 7636 }, { "epoch": 47.43613707165109, "grad_norm": 4.90625, "learning_rate": 4.378197733283359e-05, "loss": 0.1445, "step": 7637 }, { "epoch": 47.442367601246104, "grad_norm": 4.90625, "learning_rate": 4.378035119632045e-05, "loss": 0.1257, "step": 7638 }, { "epoch": 47.44859813084112, "grad_norm": 3.015625, "learning_rate": 4.3778724877408776e-05, "loss": 0.1462, "step": 7639 }, { "epoch": 47.454828660436135, "grad_norm": 2.84375, "learning_rate": 4.377709837611438e-05, "loss": 0.1869, "step": 7640 }, { "epoch": 47.46105919003115, "grad_norm": 2.390625, "learning_rate": 4.3775471692453055e-05, "loss": 0.1802, "step": 7641 }, { "epoch": 47.467289719626166, "grad_norm": 4.25, "learning_rate": 4.3773844826440585e-05, "loss": 0.1023, "step": 7642 }, { "epoch": 47.47352024922118, "grad_norm": 5.5, "learning_rate": 4.377221777809279e-05, "loss": 0.3019, "step": 7643 }, { "epoch": 47.4797507788162, "grad_norm": 3.484375, "learning_rate": 4.377059054742546e-05, "loss": 0.1139, "step": 7644 }, { "epoch": 47.48598130841121, "grad_norm": 3.234375, "learning_rate": 4.37689631344544e-05, "loss": 0.2195, "step": 7645 }, { "epoch": 47.49221183800623, "grad_norm": 3.21875, "learning_rate": 4.376733553919542e-05, "loss": 0.2302, "step": 7646 }, { "epoch": 47.49844236760124, "grad_norm": 3.1875, "learning_rate": 4.3765707761664324e-05, "loss": 0.2413, "step": 7647 }, { "epoch": 47.504672897196265, "grad_norm": 4.0, "learning_rate": 4.376407980187693e-05, "loss": 0.26, "step": 7648 }, { "epoch": 47.51090342679128, "grad_norm": 4.5625, "learning_rate": 4.376245165984903e-05, "loss": 0.3106, "step": 7649 }, { "epoch": 47.517133956386296, "grad_norm": 4.4375, "learning_rate": 4.376082333559646e-05, "loss": 0.2903, "step": 7650 }, { "epoch": 47.52336448598131, "grad_norm": 4.03125, "learning_rate": 4.375919482913502e-05, "loss": 0.236, "step": 7651 }, { "epoch": 47.52959501557633, "grad_norm": 2.609375, "learning_rate": 4.375756614048053e-05, "loss": 0.1762, "step": 7652 }, { "epoch": 47.53582554517134, "grad_norm": 3.375, "learning_rate": 4.3755937269648805e-05, "loss": 0.2557, "step": 7653 }, { "epoch": 47.54205607476636, "grad_norm": 4.53125, "learning_rate": 4.375430821665568e-05, "loss": 0.2081, "step": 7654 }, { "epoch": 47.54828660436137, "grad_norm": 4.6875, "learning_rate": 4.375267898151695e-05, "loss": 0.1918, "step": 7655 }, { "epoch": 47.55451713395639, "grad_norm": 3.9375, "learning_rate": 4.375104956424846e-05, "loss": 0.1795, "step": 7656 }, { "epoch": 47.5607476635514, "grad_norm": 3.078125, "learning_rate": 4.374941996486602e-05, "loss": 0.2128, "step": 7657 }, { "epoch": 47.56697819314642, "grad_norm": 2.375, "learning_rate": 4.374779018338548e-05, "loss": 0.1549, "step": 7658 }, { "epoch": 47.573208722741434, "grad_norm": 3.375, "learning_rate": 4.374616021982264e-05, "loss": 0.1365, "step": 7659 }, { "epoch": 47.57943925233645, "grad_norm": 3.875, "learning_rate": 4.374453007419336e-05, "loss": 0.2608, "step": 7660 }, { "epoch": 47.585669781931465, "grad_norm": 4.46875, "learning_rate": 4.374289974651344e-05, "loss": 0.1915, "step": 7661 }, { "epoch": 47.59190031152648, "grad_norm": 4.65625, "learning_rate": 4.374126923679874e-05, "loss": 0.135, "step": 7662 }, { "epoch": 47.598130841121495, "grad_norm": 3.171875, "learning_rate": 4.3739638545065086e-05, "loss": 0.1054, "step": 7663 }, { "epoch": 47.60436137071651, "grad_norm": 4.25, "learning_rate": 4.3738007671328315e-05, "loss": 0.209, "step": 7664 }, { "epoch": 47.610591900311526, "grad_norm": 6.1875, "learning_rate": 4.373637661560428e-05, "loss": 0.3371, "step": 7665 }, { "epoch": 47.61682242990654, "grad_norm": 4.84375, "learning_rate": 4.3734745377908794e-05, "loss": 0.2188, "step": 7666 }, { "epoch": 47.62305295950156, "grad_norm": 5.0, "learning_rate": 4.373311395825772e-05, "loss": 0.2187, "step": 7667 }, { "epoch": 47.62928348909657, "grad_norm": 4.3125, "learning_rate": 4.37314823566669e-05, "loss": 0.2732, "step": 7668 }, { "epoch": 47.63551401869159, "grad_norm": 2.78125, "learning_rate": 4.372985057315218e-05, "loss": 0.18, "step": 7669 }, { "epoch": 47.6417445482866, "grad_norm": 2.328125, "learning_rate": 4.372821860772941e-05, "loss": 0.1672, "step": 7670 }, { "epoch": 47.64797507788162, "grad_norm": 2.8125, "learning_rate": 4.372658646041443e-05, "loss": 0.2329, "step": 7671 }, { "epoch": 47.654205607476634, "grad_norm": 2.9375, "learning_rate": 4.37249541312231e-05, "loss": 0.2271, "step": 7672 }, { "epoch": 47.66043613707165, "grad_norm": 3.421875, "learning_rate": 4.372332162017128e-05, "loss": 0.2613, "step": 7673 }, { "epoch": 47.666666666666664, "grad_norm": 3.5, "learning_rate": 4.37216889272748e-05, "loss": 0.221, "step": 7674 }, { "epoch": 47.67289719626168, "grad_norm": 3.953125, "learning_rate": 4.372005605254956e-05, "loss": 0.2392, "step": 7675 }, { "epoch": 47.679127725856695, "grad_norm": 2.796875, "learning_rate": 4.371842299601138e-05, "loss": 0.2826, "step": 7676 }, { "epoch": 47.68535825545171, "grad_norm": 2.4375, "learning_rate": 4.371678975767613e-05, "loss": 0.1931, "step": 7677 }, { "epoch": 47.691588785046726, "grad_norm": 1.5703125, "learning_rate": 4.371515633755968e-05, "loss": 0.123, "step": 7678 }, { "epoch": 47.69781931464174, "grad_norm": 1.6875, "learning_rate": 4.371352273567789e-05, "loss": 0.1331, "step": 7679 }, { "epoch": 47.70404984423676, "grad_norm": 2.953125, "learning_rate": 4.371188895204663e-05, "loss": 0.2411, "step": 7680 }, { "epoch": 47.71028037383178, "grad_norm": 1.8984375, "learning_rate": 4.3710254986681766e-05, "loss": 0.1331, "step": 7681 }, { "epoch": 47.716510903426794, "grad_norm": 2.390625, "learning_rate": 4.370862083959916e-05, "loss": 0.1192, "step": 7682 }, { "epoch": 47.72274143302181, "grad_norm": 1.78125, "learning_rate": 4.370698651081469e-05, "loss": 0.1518, "step": 7683 }, { "epoch": 47.728971962616825, "grad_norm": 1.8671875, "learning_rate": 4.370535200034423e-05, "loss": 0.1171, "step": 7684 }, { "epoch": 47.73520249221184, "grad_norm": 1.9140625, "learning_rate": 4.370371730820365e-05, "loss": 0.1296, "step": 7685 }, { "epoch": 47.741433021806856, "grad_norm": 2.421875, "learning_rate": 4.370208243440883e-05, "loss": 0.125, "step": 7686 }, { "epoch": 47.74766355140187, "grad_norm": 2.21875, "learning_rate": 4.370044737897565e-05, "loss": 0.1652, "step": 7687 }, { "epoch": 47.753894080996886, "grad_norm": 1.8203125, "learning_rate": 4.369881214191997e-05, "loss": 0.1919, "step": 7688 }, { "epoch": 47.7601246105919, "grad_norm": 2.421875, "learning_rate": 4.369717672325771e-05, "loss": 0.2512, "step": 7689 }, { "epoch": 47.76635514018692, "grad_norm": 1.4609375, "learning_rate": 4.369554112300472e-05, "loss": 0.1262, "step": 7690 }, { "epoch": 47.77258566978193, "grad_norm": 2.0, "learning_rate": 4.369390534117691e-05, "loss": 0.192, "step": 7691 }, { "epoch": 47.77881619937695, "grad_norm": 3.671875, "learning_rate": 4.369226937779014e-05, "loss": 0.3273, "step": 7692 }, { "epoch": 47.78504672897196, "grad_norm": 2.859375, "learning_rate": 4.3690633232860316e-05, "loss": 0.2292, "step": 7693 }, { "epoch": 47.79127725856698, "grad_norm": 2.609375, "learning_rate": 4.3688996906403333e-05, "loss": 0.161, "step": 7694 }, { "epoch": 47.797507788161994, "grad_norm": 2.203125, "learning_rate": 4.368736039843507e-05, "loss": 0.158, "step": 7695 }, { "epoch": 47.80373831775701, "grad_norm": 1.9140625, "learning_rate": 4.3685723708971426e-05, "loss": 0.1368, "step": 7696 }, { "epoch": 47.809968847352025, "grad_norm": 2.703125, "learning_rate": 4.36840868380283e-05, "loss": 0.3073, "step": 7697 }, { "epoch": 47.81619937694704, "grad_norm": 2.203125, "learning_rate": 4.3682449785621596e-05, "loss": 0.1759, "step": 7698 }, { "epoch": 47.822429906542055, "grad_norm": 1.7265625, "learning_rate": 4.368081255176719e-05, "loss": 0.1162, "step": 7699 }, { "epoch": 47.82866043613707, "grad_norm": 2.203125, "learning_rate": 4.367917513648101e-05, "loss": 0.1202, "step": 7700 }, { "epoch": 47.834890965732086, "grad_norm": 2.546875, "learning_rate": 4.3677537539778946e-05, "loss": 0.2219, "step": 7701 }, { "epoch": 47.8411214953271, "grad_norm": 2.390625, "learning_rate": 4.3675899761676906e-05, "loss": 0.1737, "step": 7702 }, { "epoch": 47.84735202492212, "grad_norm": 2.984375, "learning_rate": 4.367426180219079e-05, "loss": 0.1885, "step": 7703 }, { "epoch": 47.85358255451713, "grad_norm": 1.7578125, "learning_rate": 4.367262366133651e-05, "loss": 0.1223, "step": 7704 }, { "epoch": 47.85981308411215, "grad_norm": 2.671875, "learning_rate": 4.367098533912997e-05, "loss": 0.2949, "step": 7705 }, { "epoch": 47.86604361370716, "grad_norm": 2.296875, "learning_rate": 4.36693468355871e-05, "loss": 0.2885, "step": 7706 }, { "epoch": 47.87227414330218, "grad_norm": 2.390625, "learning_rate": 4.36677081507238e-05, "loss": 0.158, "step": 7707 }, { "epoch": 47.87850467289719, "grad_norm": 1.5703125, "learning_rate": 4.3666069284555976e-05, "loss": 0.1329, "step": 7708 }, { "epoch": 47.88473520249221, "grad_norm": 1.5, "learning_rate": 4.366443023709956e-05, "loss": 0.1467, "step": 7709 }, { "epoch": 47.890965732087224, "grad_norm": 3.6875, "learning_rate": 4.3662791008370475e-05, "loss": 0.3359, "step": 7710 }, { "epoch": 47.89719626168224, "grad_norm": 2.9375, "learning_rate": 4.3661151598384624e-05, "loss": 0.1696, "step": 7711 }, { "epoch": 47.90342679127726, "grad_norm": 1.984375, "learning_rate": 4.365951200715794e-05, "loss": 0.1434, "step": 7712 }, { "epoch": 47.90965732087228, "grad_norm": 1.8125, "learning_rate": 4.3657872234706356e-05, "loss": 0.1177, "step": 7713 }, { "epoch": 47.91588785046729, "grad_norm": 1.59375, "learning_rate": 4.3656232281045774e-05, "loss": 0.1221, "step": 7714 }, { "epoch": 47.92211838006231, "grad_norm": 1.4140625, "learning_rate": 4.365459214619214e-05, "loss": 0.1392, "step": 7715 }, { "epoch": 47.92834890965732, "grad_norm": 2.421875, "learning_rate": 4.365295183016138e-05, "loss": 0.1876, "step": 7716 }, { "epoch": 47.93457943925234, "grad_norm": 1.8984375, "learning_rate": 4.365131133296942e-05, "loss": 0.1279, "step": 7717 }, { "epoch": 47.940809968847354, "grad_norm": 2.703125, "learning_rate": 4.36496706546322e-05, "loss": 0.187, "step": 7718 }, { "epoch": 47.94704049844237, "grad_norm": 2.3125, "learning_rate": 4.364802979516564e-05, "loss": 0.2331, "step": 7719 }, { "epoch": 47.953271028037385, "grad_norm": 2.03125, "learning_rate": 4.36463887545857e-05, "loss": 0.2617, "step": 7720 }, { "epoch": 47.9595015576324, "grad_norm": 1.71875, "learning_rate": 4.36447475329083e-05, "loss": 0.1239, "step": 7721 }, { "epoch": 47.965732087227416, "grad_norm": 1.921875, "learning_rate": 4.364310613014938e-05, "loss": 0.1386, "step": 7722 }, { "epoch": 47.97196261682243, "grad_norm": 2.109375, "learning_rate": 4.3641464546324894e-05, "loss": 0.1442, "step": 7723 }, { "epoch": 47.978193146417446, "grad_norm": 2.75, "learning_rate": 4.363982278145077e-05, "loss": 0.1756, "step": 7724 }, { "epoch": 47.98442367601246, "grad_norm": 2.203125, "learning_rate": 4.363818083554297e-05, "loss": 0.1808, "step": 7725 }, { "epoch": 47.99065420560748, "grad_norm": 2.296875, "learning_rate": 4.363653870861742e-05, "loss": 0.1799, "step": 7726 }, { "epoch": 47.99688473520249, "grad_norm": 2.734375, "learning_rate": 4.3634896400690095e-05, "loss": 0.1159, "step": 7727 }, { "epoch": 48.0, "grad_norm": 0.85546875, "learning_rate": 4.3633253911776926e-05, "loss": 0.0539, "step": 7728 }, { "epoch": 48.006230529595015, "grad_norm": 3.09375, "learning_rate": 4.3631611241893874e-05, "loss": 0.2184, "step": 7729 }, { "epoch": 48.01246105919003, "grad_norm": 2.34375, "learning_rate": 4.3629968391056883e-05, "loss": 0.2474, "step": 7730 }, { "epoch": 48.018691588785046, "grad_norm": 2.9375, "learning_rate": 4.362832535928191e-05, "loss": 0.236, "step": 7731 }, { "epoch": 48.02492211838006, "grad_norm": 3.125, "learning_rate": 4.362668214658493e-05, "loss": 0.147, "step": 7732 }, { "epoch": 48.03115264797508, "grad_norm": 3.453125, "learning_rate": 4.3625038752981886e-05, "loss": 0.2678, "step": 7733 }, { "epoch": 48.03738317757009, "grad_norm": 2.015625, "learning_rate": 4.3623395178488734e-05, "loss": 0.1445, "step": 7734 }, { "epoch": 48.04361370716511, "grad_norm": 3.53125, "learning_rate": 4.362175142312145e-05, "loss": 0.309, "step": 7735 }, { "epoch": 48.04984423676012, "grad_norm": 1.921875, "learning_rate": 4.3620107486896e-05, "loss": 0.1737, "step": 7736 }, { "epoch": 48.05607476635514, "grad_norm": 2.75, "learning_rate": 4.3618463369828335e-05, "loss": 0.1955, "step": 7737 }, { "epoch": 48.06230529595015, "grad_norm": 2.875, "learning_rate": 4.361681907193444e-05, "loss": 0.3539, "step": 7738 }, { "epoch": 48.06853582554517, "grad_norm": 3.515625, "learning_rate": 4.361517459323027e-05, "loss": 0.2128, "step": 7739 }, { "epoch": 48.074766355140184, "grad_norm": 1.8046875, "learning_rate": 4.361352993373181e-05, "loss": 0.1622, "step": 7740 }, { "epoch": 48.0809968847352, "grad_norm": 2.71875, "learning_rate": 4.3611885093455016e-05, "loss": 0.2233, "step": 7741 }, { "epoch": 48.087227414330215, "grad_norm": 2.984375, "learning_rate": 4.361024007241589e-05, "loss": 0.1815, "step": 7742 }, { "epoch": 48.09345794392523, "grad_norm": 1.4453125, "learning_rate": 4.3608594870630375e-05, "loss": 0.134, "step": 7743 }, { "epoch": 48.099688473520246, "grad_norm": 2.4375, "learning_rate": 4.360694948811447e-05, "loss": 0.1802, "step": 7744 }, { "epoch": 48.10591900311527, "grad_norm": 2.734375, "learning_rate": 4.360530392488416e-05, "loss": 0.384, "step": 7745 }, { "epoch": 48.11214953271028, "grad_norm": 2.671875, "learning_rate": 4.360365818095542e-05, "loss": 0.2092, "step": 7746 }, { "epoch": 48.1183800623053, "grad_norm": 3.140625, "learning_rate": 4.360201225634423e-05, "loss": 0.2794, "step": 7747 }, { "epoch": 48.124610591900314, "grad_norm": 2.15625, "learning_rate": 4.3600366151066575e-05, "loss": 0.176, "step": 7748 }, { "epoch": 48.13084112149533, "grad_norm": 1.25, "learning_rate": 4.3598719865138444e-05, "loss": 0.1551, "step": 7749 }, { "epoch": 48.137071651090345, "grad_norm": 1.875, "learning_rate": 4.3597073398575835e-05, "loss": 0.1396, "step": 7750 }, { "epoch": 48.14330218068536, "grad_norm": 2.109375, "learning_rate": 4.359542675139473e-05, "loss": 0.1657, "step": 7751 }, { "epoch": 48.149532710280376, "grad_norm": 2.484375, "learning_rate": 4.3593779923611114e-05, "loss": 0.219, "step": 7752 }, { "epoch": 48.15576323987539, "grad_norm": 1.9375, "learning_rate": 4.3592132915241e-05, "loss": 0.1502, "step": 7753 }, { "epoch": 48.161993769470406, "grad_norm": 2.921875, "learning_rate": 4.3590485726300375e-05, "loss": 0.2347, "step": 7754 }, { "epoch": 48.16822429906542, "grad_norm": 2.546875, "learning_rate": 4.3588838356805225e-05, "loss": 0.1494, "step": 7755 }, { "epoch": 48.17445482866044, "grad_norm": 2.046875, "learning_rate": 4.3587190806771575e-05, "loss": 0.2236, "step": 7756 }, { "epoch": 48.18068535825545, "grad_norm": 1.9453125, "learning_rate": 4.358554307621541e-05, "loss": 0.1513, "step": 7757 }, { "epoch": 48.18691588785047, "grad_norm": 1.125, "learning_rate": 4.358389516515273e-05, "loss": 0.0997, "step": 7758 }, { "epoch": 48.19314641744548, "grad_norm": 2.734375, "learning_rate": 4.358224707359955e-05, "loss": 0.2139, "step": 7759 }, { "epoch": 48.1993769470405, "grad_norm": 2.875, "learning_rate": 4.358059880157187e-05, "loss": 0.2036, "step": 7760 }, { "epoch": 48.205607476635514, "grad_norm": 3.796875, "learning_rate": 4.3578950349085704e-05, "loss": 0.1102, "step": 7761 }, { "epoch": 48.21183800623053, "grad_norm": 3.328125, "learning_rate": 4.3577301716157056e-05, "loss": 0.1961, "step": 7762 }, { "epoch": 48.218068535825545, "grad_norm": 4.3125, "learning_rate": 4.3575652902801934e-05, "loss": 0.3765, "step": 7763 }, { "epoch": 48.22429906542056, "grad_norm": 3.5625, "learning_rate": 4.3574003909036364e-05, "loss": 0.3204, "step": 7764 }, { "epoch": 48.230529595015575, "grad_norm": 3.65625, "learning_rate": 4.357235473487635e-05, "loss": 0.3109, "step": 7765 }, { "epoch": 48.23676012461059, "grad_norm": 2.3125, "learning_rate": 4.3570705380337925e-05, "loss": 0.2656, "step": 7766 }, { "epoch": 48.242990654205606, "grad_norm": 2.140625, "learning_rate": 4.356905584543709e-05, "loss": 0.1241, "step": 7767 }, { "epoch": 48.24922118380062, "grad_norm": 2.3125, "learning_rate": 4.356740613018987e-05, "loss": 0.1656, "step": 7768 }, { "epoch": 48.25545171339564, "grad_norm": 2.53125, "learning_rate": 4.35657562346123e-05, "loss": 0.2409, "step": 7769 }, { "epoch": 48.26168224299065, "grad_norm": 2.3125, "learning_rate": 4.356410615872038e-05, "loss": 0.114, "step": 7770 }, { "epoch": 48.26791277258567, "grad_norm": 2.046875, "learning_rate": 4.356245590253016e-05, "loss": 0.1421, "step": 7771 }, { "epoch": 48.27414330218068, "grad_norm": 2.671875, "learning_rate": 4.356080546605765e-05, "loss": 0.1643, "step": 7772 }, { "epoch": 48.2803738317757, "grad_norm": 2.171875, "learning_rate": 4.3559154849318895e-05, "loss": 0.1839, "step": 7773 }, { "epoch": 48.28660436137071, "grad_norm": 3.078125, "learning_rate": 4.355750405232992e-05, "loss": 0.2716, "step": 7774 }, { "epoch": 48.29283489096573, "grad_norm": 3.734375, "learning_rate": 4.355585307510675e-05, "loss": 0.3377, "step": 7775 }, { "epoch": 48.299065420560744, "grad_norm": 2.75, "learning_rate": 4.3554201917665425e-05, "loss": 0.2021, "step": 7776 }, { "epoch": 48.30529595015577, "grad_norm": 3.125, "learning_rate": 4.355255058002199e-05, "loss": 0.2841, "step": 7777 }, { "epoch": 48.31152647975078, "grad_norm": 2.6875, "learning_rate": 4.3550899062192466e-05, "loss": 0.159, "step": 7778 }, { "epoch": 48.3177570093458, "grad_norm": 1.75, "learning_rate": 4.35492473641929e-05, "loss": 0.12, "step": 7779 }, { "epoch": 48.32398753894081, "grad_norm": 1.53125, "learning_rate": 4.354759548603935e-05, "loss": 0.1408, "step": 7780 }, { "epoch": 48.33021806853583, "grad_norm": 2.578125, "learning_rate": 4.354594342774784e-05, "loss": 0.1641, "step": 7781 }, { "epoch": 48.33644859813084, "grad_norm": 2.0625, "learning_rate": 4.3544291189334406e-05, "loss": 0.1418, "step": 7782 }, { "epoch": 48.34267912772586, "grad_norm": 2.734375, "learning_rate": 4.354263877081512e-05, "loss": 0.1434, "step": 7783 }, { "epoch": 48.348909657320874, "grad_norm": 2.3125, "learning_rate": 4.3540986172206024e-05, "loss": 0.1601, "step": 7784 }, { "epoch": 48.35514018691589, "grad_norm": 1.6796875, "learning_rate": 4.353933339352316e-05, "loss": 0.1546, "step": 7785 }, { "epoch": 48.361370716510905, "grad_norm": 2.09375, "learning_rate": 4.353768043478259e-05, "loss": 0.1523, "step": 7786 }, { "epoch": 48.36760124610592, "grad_norm": 2.484375, "learning_rate": 4.353602729600035e-05, "loss": 0.2224, "step": 7787 }, { "epoch": 48.373831775700936, "grad_norm": 3.40625, "learning_rate": 4.353437397719251e-05, "loss": 0.1678, "step": 7788 }, { "epoch": 48.38006230529595, "grad_norm": 1.8515625, "learning_rate": 4.353272047837513e-05, "loss": 0.1261, "step": 7789 }, { "epoch": 48.386292834890966, "grad_norm": 1.9140625, "learning_rate": 4.353106679956426e-05, "loss": 0.1542, "step": 7790 }, { "epoch": 48.39252336448598, "grad_norm": 2.953125, "learning_rate": 4.352941294077597e-05, "loss": 0.3037, "step": 7791 }, { "epoch": 48.398753894081, "grad_norm": 2.671875, "learning_rate": 4.3527758902026316e-05, "loss": 0.1599, "step": 7792 }, { "epoch": 48.40498442367601, "grad_norm": 1.890625, "learning_rate": 4.3526104683331356e-05, "loss": 0.1095, "step": 7793 }, { "epoch": 48.41121495327103, "grad_norm": 2.3125, "learning_rate": 4.352445028470718e-05, "loss": 0.1874, "step": 7794 }, { "epoch": 48.41744548286604, "grad_norm": 1.578125, "learning_rate": 4.3522795706169826e-05, "loss": 0.1444, "step": 7795 }, { "epoch": 48.42367601246106, "grad_norm": 2.21875, "learning_rate": 4.352114094773538e-05, "loss": 0.199, "step": 7796 }, { "epoch": 48.429906542056074, "grad_norm": 2.796875, "learning_rate": 4.351948600941991e-05, "loss": 0.2334, "step": 7797 }, { "epoch": 48.43613707165109, "grad_norm": 1.84375, "learning_rate": 4.3517830891239496e-05, "loss": 0.1365, "step": 7798 }, { "epoch": 48.442367601246104, "grad_norm": 1.640625, "learning_rate": 4.351617559321021e-05, "loss": 0.1225, "step": 7799 }, { "epoch": 48.44859813084112, "grad_norm": 1.6953125, "learning_rate": 4.351452011534811e-05, "loss": 0.1482, "step": 7800 }, { "epoch": 48.454828660436135, "grad_norm": 3.5625, "learning_rate": 4.35128644576693e-05, "loss": 0.334, "step": 7801 }, { "epoch": 48.46105919003115, "grad_norm": 3.34375, "learning_rate": 4.3511208620189846e-05, "loss": 0.2634, "step": 7802 }, { "epoch": 48.467289719626166, "grad_norm": 2.28125, "learning_rate": 4.350955260292584e-05, "loss": 0.1355, "step": 7803 }, { "epoch": 48.47352024922118, "grad_norm": 2.078125, "learning_rate": 4.3507896405893357e-05, "loss": 0.1861, "step": 7804 }, { "epoch": 48.4797507788162, "grad_norm": 2.734375, "learning_rate": 4.350624002910848e-05, "loss": 0.2907, "step": 7805 }, { "epoch": 48.48598130841121, "grad_norm": 2.328125, "learning_rate": 4.35045834725873e-05, "loss": 0.1859, "step": 7806 }, { "epoch": 48.49221183800623, "grad_norm": 2.125, "learning_rate": 4.350292673634591e-05, "loss": 0.1141, "step": 7807 }, { "epoch": 48.49844236760124, "grad_norm": 2.453125, "learning_rate": 4.35012698204004e-05, "loss": 0.2209, "step": 7808 }, { "epoch": 48.504672897196265, "grad_norm": 4.125, "learning_rate": 4.349961272476685e-05, "loss": 0.4341, "step": 7809 }, { "epoch": 48.51090342679128, "grad_norm": 2.71875, "learning_rate": 4.349795544946137e-05, "loss": 0.1751, "step": 7810 }, { "epoch": 48.517133956386296, "grad_norm": 3.59375, "learning_rate": 4.3496297994500046e-05, "loss": 0.2462, "step": 7811 }, { "epoch": 48.52336448598131, "grad_norm": 3.53125, "learning_rate": 4.3494640359898985e-05, "loss": 0.3669, "step": 7812 }, { "epoch": 48.52959501557633, "grad_norm": 2.015625, "learning_rate": 4.3492982545674274e-05, "loss": 0.1996, "step": 7813 }, { "epoch": 48.53582554517134, "grad_norm": 2.140625, "learning_rate": 4.349132455184202e-05, "loss": 0.1515, "step": 7814 }, { "epoch": 48.54205607476636, "grad_norm": 2.484375, "learning_rate": 4.3489666378418324e-05, "loss": 0.1986, "step": 7815 }, { "epoch": 48.54828660436137, "grad_norm": 2.359375, "learning_rate": 4.34880080254193e-05, "loss": 0.152, "step": 7816 }, { "epoch": 48.55451713395639, "grad_norm": 2.796875, "learning_rate": 4.348634949286103e-05, "loss": 0.1172, "step": 7817 }, { "epoch": 48.5607476635514, "grad_norm": 3.484375, "learning_rate": 4.348469078075966e-05, "loss": 0.3052, "step": 7818 }, { "epoch": 48.56697819314642, "grad_norm": 1.1171875, "learning_rate": 4.3483031889131266e-05, "loss": 0.0901, "step": 7819 }, { "epoch": 48.573208722741434, "grad_norm": 3.546875, "learning_rate": 4.348137281799197e-05, "loss": 0.2354, "step": 7820 }, { "epoch": 48.57943925233645, "grad_norm": 3.484375, "learning_rate": 4.3479713567357886e-05, "loss": 0.1034, "step": 7821 }, { "epoch": 48.585669781931465, "grad_norm": 3.609375, "learning_rate": 4.347805413724514e-05, "loss": 0.1909, "step": 7822 }, { "epoch": 48.59190031152648, "grad_norm": 3.953125, "learning_rate": 4.3476394527669826e-05, "loss": 0.2382, "step": 7823 }, { "epoch": 48.598130841121495, "grad_norm": 2.0, "learning_rate": 4.347473473864808e-05, "loss": 0.1458, "step": 7824 }, { "epoch": 48.60436137071651, "grad_norm": 2.09375, "learning_rate": 4.347307477019602e-05, "loss": 0.2541, "step": 7825 }, { "epoch": 48.610591900311526, "grad_norm": 2.65625, "learning_rate": 4.347141462232977e-05, "loss": 0.2086, "step": 7826 }, { "epoch": 48.61682242990654, "grad_norm": 3.609375, "learning_rate": 4.3469754295065434e-05, "loss": 0.1814, "step": 7827 }, { "epoch": 48.62305295950156, "grad_norm": 3.6875, "learning_rate": 4.346809378841916e-05, "loss": 0.2682, "step": 7828 }, { "epoch": 48.62928348909657, "grad_norm": 2.734375, "learning_rate": 4.3466433102407074e-05, "loss": 0.252, "step": 7829 }, { "epoch": 48.63551401869159, "grad_norm": 2.515625, "learning_rate": 4.3464772237045295e-05, "loss": 0.1573, "step": 7830 }, { "epoch": 48.6417445482866, "grad_norm": 3.421875, "learning_rate": 4.3463111192349945e-05, "loss": 0.1717, "step": 7831 }, { "epoch": 48.64797507788162, "grad_norm": 1.75, "learning_rate": 4.3461449968337184e-05, "loss": 0.1307, "step": 7832 }, { "epoch": 48.654205607476634, "grad_norm": 2.390625, "learning_rate": 4.345978856502313e-05, "loss": 0.212, "step": 7833 }, { "epoch": 48.66043613707165, "grad_norm": 2.609375, "learning_rate": 4.345812698242391e-05, "loss": 0.1347, "step": 7834 }, { "epoch": 48.666666666666664, "grad_norm": 1.6640625, "learning_rate": 4.3456465220555674e-05, "loss": 0.1304, "step": 7835 }, { "epoch": 48.67289719626168, "grad_norm": 2.3125, "learning_rate": 4.3454803279434554e-05, "loss": 0.238, "step": 7836 }, { "epoch": 48.679127725856695, "grad_norm": 2.625, "learning_rate": 4.34531411590767e-05, "loss": 0.2256, "step": 7837 }, { "epoch": 48.68535825545171, "grad_norm": 3.046875, "learning_rate": 4.345147885949825e-05, "loss": 0.1911, "step": 7838 }, { "epoch": 48.691588785046726, "grad_norm": 3.5625, "learning_rate": 4.344981638071535e-05, "loss": 0.2221, "step": 7839 }, { "epoch": 48.69781931464174, "grad_norm": 3.453125, "learning_rate": 4.3448153722744144e-05, "loss": 0.2922, "step": 7840 }, { "epoch": 48.70404984423676, "grad_norm": 2.328125, "learning_rate": 4.344649088560079e-05, "loss": 0.1349, "step": 7841 }, { "epoch": 48.71028037383178, "grad_norm": 3.21875, "learning_rate": 4.344482786930142e-05, "loss": 0.2071, "step": 7842 }, { "epoch": 48.716510903426794, "grad_norm": 1.4609375, "learning_rate": 4.344316467386219e-05, "loss": 0.1189, "step": 7843 }, { "epoch": 48.72274143302181, "grad_norm": 2.53125, "learning_rate": 4.344150129929927e-05, "loss": 0.2375, "step": 7844 }, { "epoch": 48.728971962616825, "grad_norm": 2.9375, "learning_rate": 4.34398377456288e-05, "loss": 0.253, "step": 7845 }, { "epoch": 48.73520249221184, "grad_norm": 3.53125, "learning_rate": 4.3438174012866934e-05, "loss": 0.1702, "step": 7846 }, { "epoch": 48.741433021806856, "grad_norm": 2.21875, "learning_rate": 4.343651010102984e-05, "loss": 0.1293, "step": 7847 }, { "epoch": 48.74766355140187, "grad_norm": 2.734375, "learning_rate": 4.343484601013367e-05, "loss": 0.2555, "step": 7848 }, { "epoch": 48.753894080996886, "grad_norm": 5.21875, "learning_rate": 4.343318174019459e-05, "loss": 0.1736, "step": 7849 }, { "epoch": 48.7601246105919, "grad_norm": 4.4375, "learning_rate": 4.343151729122877e-05, "loss": 0.1407, "step": 7850 }, { "epoch": 48.76635514018692, "grad_norm": 3.671875, "learning_rate": 4.3429852663252376e-05, "loss": 0.1337, "step": 7851 }, { "epoch": 48.77258566978193, "grad_norm": 3.71875, "learning_rate": 4.3428187856281554e-05, "loss": 0.2788, "step": 7852 }, { "epoch": 48.77881619937695, "grad_norm": 3.046875, "learning_rate": 4.3426522870332495e-05, "loss": 0.2168, "step": 7853 }, { "epoch": 48.78504672897196, "grad_norm": 5.53125, "learning_rate": 4.3424857705421365e-05, "loss": 0.4656, "step": 7854 }, { "epoch": 48.79127725856698, "grad_norm": 2.609375, "learning_rate": 4.3423192361564334e-05, "loss": 0.1462, "step": 7855 }, { "epoch": 48.797507788161994, "grad_norm": 2.671875, "learning_rate": 4.342152683877757e-05, "loss": 0.1847, "step": 7856 }, { "epoch": 48.80373831775701, "grad_norm": 3.03125, "learning_rate": 4.341986113707726e-05, "loss": 0.1879, "step": 7857 }, { "epoch": 48.809968847352025, "grad_norm": 2.453125, "learning_rate": 4.3418195256479574e-05, "loss": 0.2546, "step": 7858 }, { "epoch": 48.81619937694704, "grad_norm": 1.625, "learning_rate": 4.34165291970007e-05, "loss": 0.1422, "step": 7859 }, { "epoch": 48.822429906542055, "grad_norm": 1.0859375, "learning_rate": 4.3414862958656805e-05, "loss": 0.1011, "step": 7860 }, { "epoch": 48.82866043613707, "grad_norm": 2.703125, "learning_rate": 4.3413196541464086e-05, "loss": 0.2037, "step": 7861 }, { "epoch": 48.834890965732086, "grad_norm": 1.90625, "learning_rate": 4.341152994543871e-05, "loss": 0.1446, "step": 7862 }, { "epoch": 48.8411214953271, "grad_norm": 1.78125, "learning_rate": 4.3409863170596886e-05, "loss": 0.1341, "step": 7863 }, { "epoch": 48.84735202492212, "grad_norm": 2.90625, "learning_rate": 4.340819621695479e-05, "loss": 0.174, "step": 7864 }, { "epoch": 48.85358255451713, "grad_norm": 3.03125, "learning_rate": 4.340652908452861e-05, "loss": 0.2195, "step": 7865 }, { "epoch": 48.85981308411215, "grad_norm": 2.484375, "learning_rate": 4.340486177333454e-05, "loss": 0.1844, "step": 7866 }, { "epoch": 48.86604361370716, "grad_norm": 1.7578125, "learning_rate": 4.340319428338877e-05, "loss": 0.1821, "step": 7867 }, { "epoch": 48.87227414330218, "grad_norm": 2.328125, "learning_rate": 4.34015266147075e-05, "loss": 0.2188, "step": 7868 }, { "epoch": 48.87850467289719, "grad_norm": 1.984375, "learning_rate": 4.3399858767306936e-05, "loss": 0.1677, "step": 7869 }, { "epoch": 48.88473520249221, "grad_norm": 2.0625, "learning_rate": 4.3398190741203255e-05, "loss": 0.1435, "step": 7870 }, { "epoch": 48.890965732087224, "grad_norm": 3.3125, "learning_rate": 4.339652253641267e-05, "loss": 0.2697, "step": 7871 }, { "epoch": 48.89719626168224, "grad_norm": 0.984375, "learning_rate": 4.339485415295138e-05, "loss": 0.099, "step": 7872 }, { "epoch": 48.90342679127726, "grad_norm": 3.8125, "learning_rate": 4.339318559083559e-05, "loss": 0.3364, "step": 7873 }, { "epoch": 48.90965732087228, "grad_norm": 4.4375, "learning_rate": 4.3391516850081503e-05, "loss": 0.2156, "step": 7874 }, { "epoch": 48.91588785046729, "grad_norm": 3.75, "learning_rate": 4.338984793070533e-05, "loss": 0.2137, "step": 7875 }, { "epoch": 48.92211838006231, "grad_norm": 1.9765625, "learning_rate": 4.338817883272328e-05, "loss": 0.1592, "step": 7876 }, { "epoch": 48.92834890965732, "grad_norm": 3.734375, "learning_rate": 4.3386509556151554e-05, "loss": 0.4222, "step": 7877 }, { "epoch": 48.93457943925234, "grad_norm": 2.6875, "learning_rate": 4.338484010100638e-05, "loss": 0.2027, "step": 7878 }, { "epoch": 48.940809968847354, "grad_norm": 4.15625, "learning_rate": 4.3383170467303957e-05, "loss": 0.1505, "step": 7879 }, { "epoch": 48.94704049844237, "grad_norm": 2.75, "learning_rate": 4.338150065506052e-05, "loss": 0.1488, "step": 7880 }, { "epoch": 48.953271028037385, "grad_norm": 1.359375, "learning_rate": 4.3379830664292267e-05, "loss": 0.1434, "step": 7881 }, { "epoch": 48.9595015576324, "grad_norm": 2.921875, "learning_rate": 4.337816049501542e-05, "loss": 0.1787, "step": 7882 }, { "epoch": 48.965732087227416, "grad_norm": 2.453125, "learning_rate": 4.337649014724621e-05, "loss": 0.1466, "step": 7883 }, { "epoch": 48.97196261682243, "grad_norm": 2.25, "learning_rate": 4.337481962100085e-05, "loss": 0.2037, "step": 7884 }, { "epoch": 48.978193146417446, "grad_norm": 2.6875, "learning_rate": 4.337314891629557e-05, "loss": 0.2252, "step": 7885 }, { "epoch": 48.98442367601246, "grad_norm": 3.21875, "learning_rate": 4.33714780331466e-05, "loss": 0.1957, "step": 7886 }, { "epoch": 48.99065420560748, "grad_norm": 2.3125, "learning_rate": 4.336980697157015e-05, "loss": 0.2627, "step": 7887 }, { "epoch": 48.99688473520249, "grad_norm": 2.84375, "learning_rate": 4.3368135731582477e-05, "loss": 0.1165, "step": 7888 }, { "epoch": 49.0, "grad_norm": 2.015625, "learning_rate": 4.336646431319979e-05, "loss": 0.1641, "step": 7889 }, { "epoch": 49.006230529595015, "grad_norm": 3.75, "learning_rate": 4.336479271643833e-05, "loss": 0.3449, "step": 7890 }, { "epoch": 49.01246105919003, "grad_norm": 2.328125, "learning_rate": 4.336312094131434e-05, "loss": 0.1501, "step": 7891 }, { "epoch": 49.018691588785046, "grad_norm": 2.140625, "learning_rate": 4.336144898784404e-05, "loss": 0.158, "step": 7892 }, { "epoch": 49.02492211838006, "grad_norm": 2.265625, "learning_rate": 4.335977685604368e-05, "loss": 0.1601, "step": 7893 }, { "epoch": 49.03115264797508, "grad_norm": 3.078125, "learning_rate": 4.3358104545929504e-05, "loss": 0.1612, "step": 7894 }, { "epoch": 49.03738317757009, "grad_norm": 4.28125, "learning_rate": 4.3356432057517736e-05, "loss": 0.2581, "step": 7895 }, { "epoch": 49.04361370716511, "grad_norm": 2.984375, "learning_rate": 4.335475939082464e-05, "loss": 0.1506, "step": 7896 }, { "epoch": 49.04984423676012, "grad_norm": 3.1875, "learning_rate": 4.335308654586644e-05, "loss": 0.1457, "step": 7897 }, { "epoch": 49.05607476635514, "grad_norm": 2.34375, "learning_rate": 4.3351413522659405e-05, "loss": 0.195, "step": 7898 }, { "epoch": 49.06230529595015, "grad_norm": 3.125, "learning_rate": 4.334974032121977e-05, "loss": 0.2247, "step": 7899 }, { "epoch": 49.06853582554517, "grad_norm": 4.78125, "learning_rate": 4.334806694156379e-05, "loss": 0.245, "step": 7900 }, { "epoch": 49.074766355140184, "grad_norm": 3.578125, "learning_rate": 4.334639338370772e-05, "loss": 0.1943, "step": 7901 }, { "epoch": 49.0809968847352, "grad_norm": 3.828125, "learning_rate": 4.3344719647667796e-05, "loss": 0.1232, "step": 7902 }, { "epoch": 49.087227414330215, "grad_norm": 1.8828125, "learning_rate": 4.3343045733460294e-05, "loss": 0.1761, "step": 7903 }, { "epoch": 49.09345794392523, "grad_norm": 3.59375, "learning_rate": 4.3341371641101476e-05, "loss": 0.2577, "step": 7904 }, { "epoch": 49.099688473520246, "grad_norm": 2.40625, "learning_rate": 4.3339697370607577e-05, "loss": 0.1713, "step": 7905 }, { "epoch": 49.10591900311527, "grad_norm": 2.71875, "learning_rate": 4.333802292199488e-05, "loss": 0.1434, "step": 7906 }, { "epoch": 49.11214953271028, "grad_norm": 1.9296875, "learning_rate": 4.333634829527964e-05, "loss": 0.1271, "step": 7907 }, { "epoch": 49.1183800623053, "grad_norm": 2.359375, "learning_rate": 4.333467349047811e-05, "loss": 0.2922, "step": 7908 }, { "epoch": 49.124610591900314, "grad_norm": 1.59375, "learning_rate": 4.333299850760657e-05, "loss": 0.1345, "step": 7909 }, { "epoch": 49.13084112149533, "grad_norm": 2.9375, "learning_rate": 4.333132334668129e-05, "loss": 0.2443, "step": 7910 }, { "epoch": 49.137071651090345, "grad_norm": 3.578125, "learning_rate": 4.332964800771853e-05, "loss": 0.3175, "step": 7911 }, { "epoch": 49.14330218068536, "grad_norm": 2.71875, "learning_rate": 4.332797249073456e-05, "loss": 0.2004, "step": 7912 }, { "epoch": 49.149532710280376, "grad_norm": 2.609375, "learning_rate": 4.332629679574566e-05, "loss": 0.2118, "step": 7913 }, { "epoch": 49.15576323987539, "grad_norm": 2.0625, "learning_rate": 4.33246209227681e-05, "loss": 0.1361, "step": 7914 }, { "epoch": 49.161993769470406, "grad_norm": 1.828125, "learning_rate": 4.332294487181816e-05, "loss": 0.1198, "step": 7915 }, { "epoch": 49.16822429906542, "grad_norm": 1.703125, "learning_rate": 4.332126864291212e-05, "loss": 0.1139, "step": 7916 }, { "epoch": 49.17445482866044, "grad_norm": 3.109375, "learning_rate": 4.3319592236066255e-05, "loss": 0.3782, "step": 7917 }, { "epoch": 49.18068535825545, "grad_norm": 3.84375, "learning_rate": 4.331791565129685e-05, "loss": 0.2075, "step": 7918 }, { "epoch": 49.18691588785047, "grad_norm": 2.984375, "learning_rate": 4.3316238888620186e-05, "loss": 0.2377, "step": 7919 }, { "epoch": 49.19314641744548, "grad_norm": 2.9375, "learning_rate": 4.331456194805254e-05, "loss": 0.1994, "step": 7920 }, { "epoch": 49.1993769470405, "grad_norm": 3.109375, "learning_rate": 4.3312884829610224e-05, "loss": 0.2143, "step": 7921 }, { "epoch": 49.205607476635514, "grad_norm": 3.5625, "learning_rate": 4.331120753330949e-05, "loss": 0.3452, "step": 7922 }, { "epoch": 49.21183800623053, "grad_norm": 3.078125, "learning_rate": 4.3309530059166666e-05, "loss": 0.1967, "step": 7923 }, { "epoch": 49.218068535825545, "grad_norm": 2.625, "learning_rate": 4.330785240719801e-05, "loss": 0.1583, "step": 7924 }, { "epoch": 49.22429906542056, "grad_norm": 2.65625, "learning_rate": 4.3306174577419844e-05, "loss": 0.1837, "step": 7925 }, { "epoch": 49.230529595015575, "grad_norm": 2.171875, "learning_rate": 4.330449656984845e-05, "loss": 0.2252, "step": 7926 }, { "epoch": 49.23676012461059, "grad_norm": 2.875, "learning_rate": 4.330281838450012e-05, "loss": 0.1735, "step": 7927 }, { "epoch": 49.242990654205606, "grad_norm": 2.25, "learning_rate": 4.330114002139116e-05, "loss": 0.1638, "step": 7928 }, { "epoch": 49.24922118380062, "grad_norm": 2.828125, "learning_rate": 4.329946148053787e-05, "loss": 0.2172, "step": 7929 }, { "epoch": 49.25545171339564, "grad_norm": 2.265625, "learning_rate": 4.329778276195656e-05, "loss": 0.2188, "step": 7930 }, { "epoch": 49.26168224299065, "grad_norm": 3.1875, "learning_rate": 4.329610386566352e-05, "loss": 0.1534, "step": 7931 }, { "epoch": 49.26791277258567, "grad_norm": 2.453125, "learning_rate": 4.3294424791675064e-05, "loss": 0.1654, "step": 7932 }, { "epoch": 49.27414330218068, "grad_norm": 3.0, "learning_rate": 4.3292745540007495e-05, "loss": 0.1661, "step": 7933 }, { "epoch": 49.2803738317757, "grad_norm": 2.859375, "learning_rate": 4.329106611067712e-05, "loss": 0.256, "step": 7934 }, { "epoch": 49.28660436137071, "grad_norm": 2.515625, "learning_rate": 4.3289386503700266e-05, "loss": 0.2282, "step": 7935 }, { "epoch": 49.29283489096573, "grad_norm": 2.453125, "learning_rate": 4.328770671909323e-05, "loss": 0.1771, "step": 7936 }, { "epoch": 49.299065420560744, "grad_norm": 2.921875, "learning_rate": 4.328602675687233e-05, "loss": 0.1744, "step": 7937 }, { "epoch": 49.30529595015577, "grad_norm": 5.28125, "learning_rate": 4.328434661705388e-05, "loss": 0.1637, "step": 7938 }, { "epoch": 49.31152647975078, "grad_norm": 3.359375, "learning_rate": 4.328266629965421e-05, "loss": 0.1608, "step": 7939 }, { "epoch": 49.3177570093458, "grad_norm": 3.0, "learning_rate": 4.3280985804689614e-05, "loss": 0.2309, "step": 7940 }, { "epoch": 49.32398753894081, "grad_norm": 2.53125, "learning_rate": 4.327930513217644e-05, "loss": 0.1211, "step": 7941 }, { "epoch": 49.33021806853583, "grad_norm": 3.296875, "learning_rate": 4.3277624282131e-05, "loss": 0.1876, "step": 7942 }, { "epoch": 49.33644859813084, "grad_norm": 2.453125, "learning_rate": 4.3275943254569614e-05, "loss": 0.1848, "step": 7943 }, { "epoch": 49.34267912772586, "grad_norm": 2.4375, "learning_rate": 4.3274262049508625e-05, "loss": 0.1893, "step": 7944 }, { "epoch": 49.348909657320874, "grad_norm": 2.390625, "learning_rate": 4.327258066696435e-05, "loss": 0.1436, "step": 7945 }, { "epoch": 49.35514018691589, "grad_norm": 2.96875, "learning_rate": 4.3270899106953105e-05, "loss": 0.1545, "step": 7946 }, { "epoch": 49.361370716510905, "grad_norm": 2.15625, "learning_rate": 4.3269217369491244e-05, "loss": 0.1748, "step": 7947 }, { "epoch": 49.36760124610592, "grad_norm": 2.484375, "learning_rate": 4.326753545459509e-05, "loss": 0.1313, "step": 7948 }, { "epoch": 49.373831775700936, "grad_norm": 4.03125, "learning_rate": 4.3265853362280986e-05, "loss": 0.1805, "step": 7949 }, { "epoch": 49.38006230529595, "grad_norm": 2.953125, "learning_rate": 4.326417109256525e-05, "loss": 0.2795, "step": 7950 }, { "epoch": 49.386292834890966, "grad_norm": 3.078125, "learning_rate": 4.326248864546425e-05, "loss": 0.2459, "step": 7951 }, { "epoch": 49.39252336448598, "grad_norm": 2.453125, "learning_rate": 4.3260806020994295e-05, "loss": 0.182, "step": 7952 }, { "epoch": 49.398753894081, "grad_norm": 3.171875, "learning_rate": 4.325912321917175e-05, "loss": 0.3199, "step": 7953 }, { "epoch": 49.40498442367601, "grad_norm": 2.453125, "learning_rate": 4.3257440240012945e-05, "loss": 0.104, "step": 7954 }, { "epoch": 49.41121495327103, "grad_norm": 1.96875, "learning_rate": 4.325575708353423e-05, "loss": 0.1496, "step": 7955 }, { "epoch": 49.41744548286604, "grad_norm": 2.46875, "learning_rate": 4.325407374975197e-05, "loss": 0.1853, "step": 7956 }, { "epoch": 49.42367601246106, "grad_norm": 3.90625, "learning_rate": 4.325239023868247e-05, "loss": 0.3736, "step": 7957 }, { "epoch": 49.429906542056074, "grad_norm": 3.015625, "learning_rate": 4.3250706550342126e-05, "loss": 0.2634, "step": 7958 }, { "epoch": 49.43613707165109, "grad_norm": 3.859375, "learning_rate": 4.324902268474727e-05, "loss": 0.1528, "step": 7959 }, { "epoch": 49.442367601246104, "grad_norm": 3.796875, "learning_rate": 4.324733864191425e-05, "loss": 0.2254, "step": 7960 }, { "epoch": 49.44859813084112, "grad_norm": 2.765625, "learning_rate": 4.324565442185944e-05, "loss": 0.2716, "step": 7961 }, { "epoch": 49.454828660436135, "grad_norm": 3.90625, "learning_rate": 4.324397002459918e-05, "loss": 0.3712, "step": 7962 }, { "epoch": 49.46105919003115, "grad_norm": 4.0625, "learning_rate": 4.3242285450149836e-05, "loss": 0.1306, "step": 7963 }, { "epoch": 49.467289719626166, "grad_norm": 1.953125, "learning_rate": 4.3240600698527764e-05, "loss": 0.1228, "step": 7964 }, { "epoch": 49.47352024922118, "grad_norm": 1.9609375, "learning_rate": 4.323891576974935e-05, "loss": 0.1495, "step": 7965 }, { "epoch": 49.4797507788162, "grad_norm": 2.203125, "learning_rate": 4.323723066383092e-05, "loss": 0.2333, "step": 7966 }, { "epoch": 49.48598130841121, "grad_norm": 3.5, "learning_rate": 4.3235545380788865e-05, "loss": 0.3027, "step": 7967 }, { "epoch": 49.49221183800623, "grad_norm": 3.1875, "learning_rate": 4.3233859920639555e-05, "loss": 0.1634, "step": 7968 }, { "epoch": 49.49844236760124, "grad_norm": 3.515625, "learning_rate": 4.323217428339935e-05, "loss": 0.2275, "step": 7969 }, { "epoch": 49.504672897196265, "grad_norm": 3.25, "learning_rate": 4.323048846908462e-05, "loss": 0.1819, "step": 7970 }, { "epoch": 49.51090342679128, "grad_norm": 1.3984375, "learning_rate": 4.322880247771175e-05, "loss": 0.1027, "step": 7971 }, { "epoch": 49.517133956386296, "grad_norm": 1.5625, "learning_rate": 4.32271163092971e-05, "loss": 0.1306, "step": 7972 }, { "epoch": 49.52336448598131, "grad_norm": 3.3125, "learning_rate": 4.322542996385705e-05, "loss": 0.1674, "step": 7973 }, { "epoch": 49.52959501557633, "grad_norm": 5.75, "learning_rate": 4.322374344140799e-05, "loss": 0.162, "step": 7974 }, { "epoch": 49.53582554517134, "grad_norm": 5.09375, "learning_rate": 4.322205674196629e-05, "loss": 0.1237, "step": 7975 }, { "epoch": 49.54205607476636, "grad_norm": 2.921875, "learning_rate": 4.322036986554833e-05, "loss": 0.2338, "step": 7976 }, { "epoch": 49.54828660436137, "grad_norm": 2.28125, "learning_rate": 4.3218682812170485e-05, "loss": 0.2345, "step": 7977 }, { "epoch": 49.55451713395639, "grad_norm": 1.5625, "learning_rate": 4.321699558184916e-05, "loss": 0.17, "step": 7978 }, { "epoch": 49.5607476635514, "grad_norm": 2.140625, "learning_rate": 4.321530817460073e-05, "loss": 0.134, "step": 7979 }, { "epoch": 49.56697819314642, "grad_norm": 2.65625, "learning_rate": 4.321362059044159e-05, "loss": 0.2058, "step": 7980 }, { "epoch": 49.573208722741434, "grad_norm": 1.9609375, "learning_rate": 4.3211932829388126e-05, "loss": 0.1447, "step": 7981 }, { "epoch": 49.57943925233645, "grad_norm": 2.359375, "learning_rate": 4.321024489145673e-05, "loss": 0.2053, "step": 7982 }, { "epoch": 49.585669781931465, "grad_norm": 2.046875, "learning_rate": 4.320855677666379e-05, "loss": 0.2522, "step": 7983 }, { "epoch": 49.59190031152648, "grad_norm": 2.265625, "learning_rate": 4.3206868485025716e-05, "loss": 0.2184, "step": 7984 }, { "epoch": 49.598130841121495, "grad_norm": 1.65625, "learning_rate": 4.320518001655889e-05, "loss": 0.1631, "step": 7985 }, { "epoch": 49.60436137071651, "grad_norm": 2.625, "learning_rate": 4.320349137127972e-05, "loss": 0.2335, "step": 7986 }, { "epoch": 49.610591900311526, "grad_norm": 3.140625, "learning_rate": 4.32018025492046e-05, "loss": 0.2334, "step": 7987 }, { "epoch": 49.61682242990654, "grad_norm": 4.21875, "learning_rate": 4.320011355034993e-05, "loss": 0.2462, "step": 7988 }, { "epoch": 49.62305295950156, "grad_norm": 2.40625, "learning_rate": 4.319842437473212e-05, "loss": 0.1709, "step": 7989 }, { "epoch": 49.62928348909657, "grad_norm": 2.359375, "learning_rate": 4.319673502236758e-05, "loss": 0.2026, "step": 7990 }, { "epoch": 49.63551401869159, "grad_norm": 3.171875, "learning_rate": 4.319504549327271e-05, "loss": 0.2693, "step": 7991 }, { "epoch": 49.6417445482866, "grad_norm": 1.3203125, "learning_rate": 4.319335578746392e-05, "loss": 0.1298, "step": 7992 }, { "epoch": 49.64797507788162, "grad_norm": 2.625, "learning_rate": 4.319166590495762e-05, "loss": 0.216, "step": 7993 }, { "epoch": 49.654205607476634, "grad_norm": 3.5, "learning_rate": 4.3189975845770234e-05, "loss": 0.1936, "step": 7994 }, { "epoch": 49.66043613707165, "grad_norm": 1.9921875, "learning_rate": 4.318828560991816e-05, "loss": 0.0971, "step": 7995 }, { "epoch": 49.666666666666664, "grad_norm": 2.96875, "learning_rate": 4.3186595197417824e-05, "loss": 0.159, "step": 7996 }, { "epoch": 49.67289719626168, "grad_norm": 2.625, "learning_rate": 4.318490460828563e-05, "loss": 0.2117, "step": 7997 }, { "epoch": 49.679127725856695, "grad_norm": 2.015625, "learning_rate": 4.318321384253802e-05, "loss": 0.1483, "step": 7998 }, { "epoch": 49.68535825545171, "grad_norm": 2.234375, "learning_rate": 4.318152290019139e-05, "loss": 0.181, "step": 7999 }, { "epoch": 49.691588785046726, "grad_norm": 3.71875, "learning_rate": 4.317983178126218e-05, "loss": 0.1093, "step": 8000 }, { "epoch": 49.69781931464174, "grad_norm": 5.25, "learning_rate": 4.317814048576682e-05, "loss": 0.2892, "step": 8001 }, { "epoch": 49.70404984423676, "grad_norm": 3.28125, "learning_rate": 4.3176449013721706e-05, "loss": 0.1561, "step": 8002 }, { "epoch": 49.71028037383178, "grad_norm": 2.359375, "learning_rate": 4.31747573651433e-05, "loss": 0.149, "step": 8003 }, { "epoch": 49.716510903426794, "grad_norm": 2.28125, "learning_rate": 4.317306554004802e-05, "loss": 0.1451, "step": 8004 }, { "epoch": 49.72274143302181, "grad_norm": 3.0625, "learning_rate": 4.317137353845229e-05, "loss": 0.1227, "step": 8005 }, { "epoch": 49.728971962616825, "grad_norm": 3.234375, "learning_rate": 4.316968136037254e-05, "loss": 0.1972, "step": 8006 }, { "epoch": 49.73520249221184, "grad_norm": 4.125, "learning_rate": 4.316798900582523e-05, "loss": 0.2115, "step": 8007 }, { "epoch": 49.741433021806856, "grad_norm": 4.03125, "learning_rate": 4.3166296474826764e-05, "loss": 0.3734, "step": 8008 }, { "epoch": 49.74766355140187, "grad_norm": 3.203125, "learning_rate": 4.31646037673936e-05, "loss": 0.2273, "step": 8009 }, { "epoch": 49.753894080996886, "grad_norm": 2.9375, "learning_rate": 4.316291088354217e-05, "loss": 0.1882, "step": 8010 }, { "epoch": 49.7601246105919, "grad_norm": 3.921875, "learning_rate": 4.316121782328892e-05, "loss": 0.2278, "step": 8011 }, { "epoch": 49.76635514018692, "grad_norm": 5.21875, "learning_rate": 4.3159524586650294e-05, "loss": 0.1583, "step": 8012 }, { "epoch": 49.77258566978193, "grad_norm": 2.234375, "learning_rate": 4.315783117364274e-05, "loss": 0.1487, "step": 8013 }, { "epoch": 49.77881619937695, "grad_norm": 1.1015625, "learning_rate": 4.315613758428269e-05, "loss": 0.106, "step": 8014 }, { "epoch": 49.78504672897196, "grad_norm": 2.359375, "learning_rate": 4.315444381858661e-05, "loss": 0.1831, "step": 8015 }, { "epoch": 49.79127725856698, "grad_norm": 2.921875, "learning_rate": 4.315274987657094e-05, "loss": 0.2642, "step": 8016 }, { "epoch": 49.797507788161994, "grad_norm": 2.609375, "learning_rate": 4.3151055758252135e-05, "loss": 0.2094, "step": 8017 }, { "epoch": 49.80373831775701, "grad_norm": 2.828125, "learning_rate": 4.314936146364665e-05, "loss": 0.2194, "step": 8018 }, { "epoch": 49.809968847352025, "grad_norm": 3.140625, "learning_rate": 4.314766699277093e-05, "loss": 0.3369, "step": 8019 }, { "epoch": 49.81619937694704, "grad_norm": 2.5, "learning_rate": 4.314597234564145e-05, "loss": 0.1323, "step": 8020 }, { "epoch": 49.822429906542055, "grad_norm": 2.765625, "learning_rate": 4.314427752227466e-05, "loss": 0.2123, "step": 8021 }, { "epoch": 49.82866043613707, "grad_norm": 3.09375, "learning_rate": 4.3142582522687017e-05, "loss": 0.1903, "step": 8022 }, { "epoch": 49.834890965732086, "grad_norm": 3.375, "learning_rate": 4.3140887346894974e-05, "loss": 0.2461, "step": 8023 }, { "epoch": 49.8411214953271, "grad_norm": 2.15625, "learning_rate": 4.313919199491502e-05, "loss": 0.1693, "step": 8024 }, { "epoch": 49.84735202492212, "grad_norm": 2.578125, "learning_rate": 4.3137496466763605e-05, "loss": 0.1907, "step": 8025 }, { "epoch": 49.85358255451713, "grad_norm": 2.5, "learning_rate": 4.3135800762457203e-05, "loss": 0.2307, "step": 8026 }, { "epoch": 49.85981308411215, "grad_norm": 3.015625, "learning_rate": 4.313410488201227e-05, "loss": 0.2124, "step": 8027 }, { "epoch": 49.86604361370716, "grad_norm": 2.75, "learning_rate": 4.313240882544529e-05, "loss": 0.1708, "step": 8028 }, { "epoch": 49.87227414330218, "grad_norm": 3.109375, "learning_rate": 4.3130712592772735e-05, "loss": 0.1589, "step": 8029 }, { "epoch": 49.87850467289719, "grad_norm": 3.28125, "learning_rate": 4.312901618401107e-05, "loss": 0.2189, "step": 8030 }, { "epoch": 49.88473520249221, "grad_norm": 2.359375, "learning_rate": 4.312731959917678e-05, "loss": 0.2503, "step": 8031 }, { "epoch": 49.890965732087224, "grad_norm": 1.859375, "learning_rate": 4.312562283828633e-05, "loss": 0.1145, "step": 8032 }, { "epoch": 49.89719626168224, "grad_norm": 2.6875, "learning_rate": 4.312392590135621e-05, "loss": 0.131, "step": 8033 }, { "epoch": 49.90342679127726, "grad_norm": 1.8125, "learning_rate": 4.31222287884029e-05, "loss": 0.1276, "step": 8034 }, { "epoch": 49.90965732087228, "grad_norm": 2.578125, "learning_rate": 4.312053149944289e-05, "loss": 0.1768, "step": 8035 }, { "epoch": 49.91588785046729, "grad_norm": 2.96875, "learning_rate": 4.311883403449264e-05, "loss": 0.2432, "step": 8036 }, { "epoch": 49.92211838006231, "grad_norm": 2.59375, "learning_rate": 4.311713639356866e-05, "loss": 0.1663, "step": 8037 }, { "epoch": 49.92834890965732, "grad_norm": 1.7421875, "learning_rate": 4.3115438576687435e-05, "loss": 0.1468, "step": 8038 }, { "epoch": 49.93457943925234, "grad_norm": 3.53125, "learning_rate": 4.3113740583865445e-05, "loss": 0.26, "step": 8039 }, { "epoch": 49.940809968847354, "grad_norm": 2.25, "learning_rate": 4.3112042415119184e-05, "loss": 0.2097, "step": 8040 }, { "epoch": 49.94704049844237, "grad_norm": 2.40625, "learning_rate": 4.3110344070465145e-05, "loss": 0.1237, "step": 8041 }, { "epoch": 49.953271028037385, "grad_norm": 2.421875, "learning_rate": 4.3108645549919836e-05, "loss": 0.1932, "step": 8042 }, { "epoch": 49.9595015576324, "grad_norm": 2.40625, "learning_rate": 4.310694685349973e-05, "loss": 0.1388, "step": 8043 }, { "epoch": 49.965732087227416, "grad_norm": 3.046875, "learning_rate": 4.310524798122134e-05, "loss": 0.2793, "step": 8044 }, { "epoch": 49.97196261682243, "grad_norm": 1.828125, "learning_rate": 4.3103548933101154e-05, "loss": 0.1226, "step": 8045 }, { "epoch": 49.978193146417446, "grad_norm": 2.71875, "learning_rate": 4.31018497091557e-05, "loss": 0.2063, "step": 8046 }, { "epoch": 49.98442367601246, "grad_norm": 2.46875, "learning_rate": 4.310015030940145e-05, "loss": 0.1413, "step": 8047 }, { "epoch": 49.99065420560748, "grad_norm": 3.296875, "learning_rate": 4.309845073385492e-05, "loss": 0.2532, "step": 8048 }, { "epoch": 49.99688473520249, "grad_norm": 1.5078125, "learning_rate": 4.309675098253263e-05, "loss": 0.133, "step": 8049 }, { "epoch": 50.0, "grad_norm": 1.2265625, "learning_rate": 4.309505105545107e-05, "loss": 0.0774, "step": 8050 }, { "epoch": 50.006230529595015, "grad_norm": 1.9296875, "learning_rate": 4.309335095262676e-05, "loss": 0.1664, "step": 8051 }, { "epoch": 50.01246105919003, "grad_norm": 2.40625, "learning_rate": 4.30916506740762e-05, "loss": 0.1755, "step": 8052 }, { "epoch": 50.018691588785046, "grad_norm": 3.625, "learning_rate": 4.3089950219815925e-05, "loss": 0.2159, "step": 8053 }, { "epoch": 50.02492211838006, "grad_norm": 2.484375, "learning_rate": 4.308824958986243e-05, "loss": 0.162, "step": 8054 }, { "epoch": 50.03115264797508, "grad_norm": 2.796875, "learning_rate": 4.308654878423224e-05, "loss": 0.2462, "step": 8055 }, { "epoch": 50.03738317757009, "grad_norm": 3.75, "learning_rate": 4.3084847802941873e-05, "loss": 0.1622, "step": 8056 }, { "epoch": 50.04361370716511, "grad_norm": 3.265625, "learning_rate": 4.3083146646007844e-05, "loss": 0.2021, "step": 8057 }, { "epoch": 50.04984423676012, "grad_norm": 2.6875, "learning_rate": 4.308144531344669e-05, "loss": 0.1762, "step": 8058 }, { "epoch": 50.05607476635514, "grad_norm": 2.15625, "learning_rate": 4.307974380527492e-05, "loss": 0.1486, "step": 8059 }, { "epoch": 50.06230529595015, "grad_norm": 2.75, "learning_rate": 4.307804212150906e-05, "loss": 0.2034, "step": 8060 }, { "epoch": 50.06853582554517, "grad_norm": 3.59375, "learning_rate": 4.307634026216565e-05, "loss": 0.2377, "step": 8061 }, { "epoch": 50.074766355140184, "grad_norm": 2.796875, "learning_rate": 4.3074638227261196e-05, "loss": 0.1369, "step": 8062 }, { "epoch": 50.0809968847352, "grad_norm": 3.15625, "learning_rate": 4.307293601681226e-05, "loss": 0.2163, "step": 8063 }, { "epoch": 50.087227414330215, "grad_norm": 2.203125, "learning_rate": 4.307123363083534e-05, "loss": 0.1619, "step": 8064 }, { "epoch": 50.09345794392523, "grad_norm": 2.78125, "learning_rate": 4.3069531069347004e-05, "loss": 0.2565, "step": 8065 }, { "epoch": 50.099688473520246, "grad_norm": 4.15625, "learning_rate": 4.306782833236376e-05, "loss": 0.2482, "step": 8066 }, { "epoch": 50.10591900311527, "grad_norm": 2.546875, "learning_rate": 4.3066125419902156e-05, "loss": 0.142, "step": 8067 }, { "epoch": 50.11214953271028, "grad_norm": 3.34375, "learning_rate": 4.306442233197873e-05, "loss": 0.1933, "step": 8068 }, { "epoch": 50.1183800623053, "grad_norm": 2.09375, "learning_rate": 4.306271906861002e-05, "loss": 0.1592, "step": 8069 }, { "epoch": 50.124610591900314, "grad_norm": 2.703125, "learning_rate": 4.306101562981258e-05, "loss": 0.2483, "step": 8070 }, { "epoch": 50.13084112149533, "grad_norm": 2.265625, "learning_rate": 4.3059312015602946e-05, "loss": 0.1924, "step": 8071 }, { "epoch": 50.137071651090345, "grad_norm": 2.46875, "learning_rate": 4.305760822599766e-05, "loss": 0.1628, "step": 8072 }, { "epoch": 50.14330218068536, "grad_norm": 2.5625, "learning_rate": 4.305590426101327e-05, "loss": 0.2776, "step": 8073 }, { "epoch": 50.149532710280376, "grad_norm": 2.4375, "learning_rate": 4.3054200120666334e-05, "loss": 0.209, "step": 8074 }, { "epoch": 50.15576323987539, "grad_norm": 1.8984375, "learning_rate": 4.305249580497339e-05, "loss": 0.1316, "step": 8075 }, { "epoch": 50.161993769470406, "grad_norm": 3.125, "learning_rate": 4.305079131395101e-05, "loss": 0.3529, "step": 8076 }, { "epoch": 50.16822429906542, "grad_norm": 1.8359375, "learning_rate": 4.3049086647615726e-05, "loss": 0.1574, "step": 8077 }, { "epoch": 50.17445482866044, "grad_norm": 2.15625, "learning_rate": 4.304738180598411e-05, "loss": 0.1721, "step": 8078 }, { "epoch": 50.18068535825545, "grad_norm": 2.375, "learning_rate": 4.3045676789072707e-05, "loss": 0.1665, "step": 8079 }, { "epoch": 50.18691588785047, "grad_norm": 2.296875, "learning_rate": 4.304397159689809e-05, "loss": 0.1451, "step": 8080 }, { "epoch": 50.19314641744548, "grad_norm": 3.140625, "learning_rate": 4.304226622947681e-05, "loss": 0.3262, "step": 8081 }, { "epoch": 50.1993769470405, "grad_norm": 1.1640625, "learning_rate": 4.304056068682544e-05, "loss": 0.1344, "step": 8082 }, { "epoch": 50.205607476635514, "grad_norm": 2.265625, "learning_rate": 4.303885496896054e-05, "loss": 0.2087, "step": 8083 }, { "epoch": 50.21183800623053, "grad_norm": 1.3671875, "learning_rate": 4.303714907589866e-05, "loss": 0.1134, "step": 8084 }, { "epoch": 50.218068535825545, "grad_norm": 3.375, "learning_rate": 4.30354430076564e-05, "loss": 0.3396, "step": 8085 }, { "epoch": 50.22429906542056, "grad_norm": 2.328125, "learning_rate": 4.30337367642503e-05, "loss": 0.181, "step": 8086 }, { "epoch": 50.230529595015575, "grad_norm": 1.9609375, "learning_rate": 4.303203034569695e-05, "loss": 0.1361, "step": 8087 }, { "epoch": 50.23676012461059, "grad_norm": 2.46875, "learning_rate": 4.3030323752012916e-05, "loss": 0.2305, "step": 8088 }, { "epoch": 50.242990654205606, "grad_norm": 2.015625, "learning_rate": 4.302861698321478e-05, "loss": 0.1621, "step": 8089 }, { "epoch": 50.24922118380062, "grad_norm": 3.859375, "learning_rate": 4.30269100393191e-05, "loss": 0.4271, "step": 8090 }, { "epoch": 50.25545171339564, "grad_norm": 1.8359375, "learning_rate": 4.302520292034247e-05, "loss": 0.1554, "step": 8091 }, { "epoch": 50.26168224299065, "grad_norm": 1.3671875, "learning_rate": 4.3023495626301466e-05, "loss": 0.1177, "step": 8092 }, { "epoch": 50.26791277258567, "grad_norm": 2.34375, "learning_rate": 4.302178815721267e-05, "loss": 0.1815, "step": 8093 }, { "epoch": 50.27414330218068, "grad_norm": 2.171875, "learning_rate": 4.302008051309267e-05, "loss": 0.1524, "step": 8094 }, { "epoch": 50.2803738317757, "grad_norm": 2.546875, "learning_rate": 4.3018372693958046e-05, "loss": 0.2202, "step": 8095 }, { "epoch": 50.28660436137071, "grad_norm": 2.859375, "learning_rate": 4.3016664699825376e-05, "loss": 0.2684, "step": 8096 }, { "epoch": 50.29283489096573, "grad_norm": 2.59375, "learning_rate": 4.301495653071126e-05, "loss": 0.2188, "step": 8097 }, { "epoch": 50.299065420560744, "grad_norm": 2.734375, "learning_rate": 4.3013248186632287e-05, "loss": 0.302, "step": 8098 }, { "epoch": 50.30529595015577, "grad_norm": 2.625, "learning_rate": 4.301153966760505e-05, "loss": 0.2426, "step": 8099 }, { "epoch": 50.31152647975078, "grad_norm": 1.6171875, "learning_rate": 4.3009830973646136e-05, "loss": 0.1001, "step": 8100 }, { "epoch": 50.3177570093458, "grad_norm": 3.65625, "learning_rate": 4.300812210477214e-05, "loss": 0.3358, "step": 8101 }, { "epoch": 50.32398753894081, "grad_norm": 2.90625, "learning_rate": 4.3006413060999665e-05, "loss": 0.2423, "step": 8102 }, { "epoch": 50.33021806853583, "grad_norm": 3.703125, "learning_rate": 4.3004703842345304e-05, "loss": 0.2247, "step": 8103 }, { "epoch": 50.33644859813084, "grad_norm": 3.46875, "learning_rate": 4.300299444882567e-05, "loss": 0.1383, "step": 8104 }, { "epoch": 50.34267912772586, "grad_norm": 4.3125, "learning_rate": 4.300128488045735e-05, "loss": 0.1248, "step": 8105 }, { "epoch": 50.348909657320874, "grad_norm": 2.328125, "learning_rate": 4.2999575137256945e-05, "loss": 0.1624, "step": 8106 }, { "epoch": 50.35514018691589, "grad_norm": 4.125, "learning_rate": 4.2997865219241076e-05, "loss": 0.2169, "step": 8107 }, { "epoch": 50.361370716510905, "grad_norm": 2.59375, "learning_rate": 4.299615512642634e-05, "loss": 0.2222, "step": 8108 }, { "epoch": 50.36760124610592, "grad_norm": 4.28125, "learning_rate": 4.2994444858829355e-05, "loss": 0.2057, "step": 8109 }, { "epoch": 50.373831775700936, "grad_norm": 1.921875, "learning_rate": 4.299273441646672e-05, "loss": 0.1228, "step": 8110 }, { "epoch": 50.38006230529595, "grad_norm": 3.515625, "learning_rate": 4.299102379935505e-05, "loss": 0.1471, "step": 8111 }, { "epoch": 50.386292834890966, "grad_norm": 2.6875, "learning_rate": 4.2989313007510954e-05, "loss": 0.1409, "step": 8112 }, { "epoch": 50.39252336448598, "grad_norm": 2.609375, "learning_rate": 4.298760204095106e-05, "loss": 0.1645, "step": 8113 }, { "epoch": 50.398753894081, "grad_norm": 2.859375, "learning_rate": 4.298589089969198e-05, "loss": 0.1857, "step": 8114 }, { "epoch": 50.40498442367601, "grad_norm": 2.359375, "learning_rate": 4.298417958375034e-05, "loss": 0.133, "step": 8115 }, { "epoch": 50.41121495327103, "grad_norm": 2.703125, "learning_rate": 4.298246809314274e-05, "loss": 0.1888, "step": 8116 }, { "epoch": 50.41744548286604, "grad_norm": 2.875, "learning_rate": 4.298075642788583e-05, "loss": 0.1638, "step": 8117 }, { "epoch": 50.42367601246106, "grad_norm": 2.015625, "learning_rate": 4.29790445879962e-05, "loss": 0.1229, "step": 8118 }, { "epoch": 50.429906542056074, "grad_norm": 2.40625, "learning_rate": 4.29773325734905e-05, "loss": 0.1533, "step": 8119 }, { "epoch": 50.43613707165109, "grad_norm": 2.109375, "learning_rate": 4.2975620384385364e-05, "loss": 0.2396, "step": 8120 }, { "epoch": 50.442367601246104, "grad_norm": 2.59375, "learning_rate": 4.2973908020697406e-05, "loss": 0.1978, "step": 8121 }, { "epoch": 50.44859813084112, "grad_norm": 2.484375, "learning_rate": 4.2972195482443256e-05, "loss": 0.2057, "step": 8122 }, { "epoch": 50.454828660436135, "grad_norm": 2.5, "learning_rate": 4.2970482769639556e-05, "loss": 0.2157, "step": 8123 }, { "epoch": 50.46105919003115, "grad_norm": 3.03125, "learning_rate": 4.296876988230293e-05, "loss": 0.1907, "step": 8124 }, { "epoch": 50.467289719626166, "grad_norm": 1.9921875, "learning_rate": 4.296705682045002e-05, "loss": 0.1474, "step": 8125 }, { "epoch": 50.47352024922118, "grad_norm": 2.9375, "learning_rate": 4.296534358409746e-05, "loss": 0.3053, "step": 8126 }, { "epoch": 50.4797507788162, "grad_norm": 2.96875, "learning_rate": 4.2963630173261896e-05, "loss": 0.1846, "step": 8127 }, { "epoch": 50.48598130841121, "grad_norm": 3.15625, "learning_rate": 4.296191658795996e-05, "loss": 0.2577, "step": 8128 }, { "epoch": 50.49221183800623, "grad_norm": 1.8828125, "learning_rate": 4.2960202828208305e-05, "loss": 0.1473, "step": 8129 }, { "epoch": 50.49844236760124, "grad_norm": 2.078125, "learning_rate": 4.2958488894023576e-05, "loss": 0.1504, "step": 8130 }, { "epoch": 50.504672897196265, "grad_norm": 3.328125, "learning_rate": 4.29567747854224e-05, "loss": 0.2517, "step": 8131 }, { "epoch": 50.51090342679128, "grad_norm": 2.890625, "learning_rate": 4.2955060502421444e-05, "loss": 0.1785, "step": 8132 }, { "epoch": 50.517133956386296, "grad_norm": 2.34375, "learning_rate": 4.2953346045037345e-05, "loss": 0.1394, "step": 8133 }, { "epoch": 50.52336448598131, "grad_norm": 2.625, "learning_rate": 4.295163141328677e-05, "loss": 0.1442, "step": 8134 }, { "epoch": 50.52959501557633, "grad_norm": 1.5234375, "learning_rate": 4.2949916607186357e-05, "loss": 0.1117, "step": 8135 }, { "epoch": 50.53582554517134, "grad_norm": 2.640625, "learning_rate": 4.294820162675277e-05, "loss": 0.2148, "step": 8136 }, { "epoch": 50.54205607476636, "grad_norm": 3.328125, "learning_rate": 4.294648647200265e-05, "loss": 0.3305, "step": 8137 }, { "epoch": 50.54828660436137, "grad_norm": 1.7578125, "learning_rate": 4.2944771142952676e-05, "loss": 0.1196, "step": 8138 }, { "epoch": 50.55451713395639, "grad_norm": 3.109375, "learning_rate": 4.29430556396195e-05, "loss": 0.3233, "step": 8139 }, { "epoch": 50.5607476635514, "grad_norm": 2.765625, "learning_rate": 4.294133996201978e-05, "loss": 0.2489, "step": 8140 }, { "epoch": 50.56697819314642, "grad_norm": 2.125, "learning_rate": 4.293962411017017e-05, "loss": 0.2278, "step": 8141 }, { "epoch": 50.573208722741434, "grad_norm": 3.65625, "learning_rate": 4.2937908084087356e-05, "loss": 0.3445, "step": 8142 }, { "epoch": 50.57943925233645, "grad_norm": 2.21875, "learning_rate": 4.293619188378798e-05, "loss": 0.1366, "step": 8143 }, { "epoch": 50.585669781931465, "grad_norm": 2.84375, "learning_rate": 4.2934475509288736e-05, "loss": 0.1796, "step": 8144 }, { "epoch": 50.59190031152648, "grad_norm": 2.78125, "learning_rate": 4.293275896060627e-05, "loss": 0.2859, "step": 8145 }, { "epoch": 50.598130841121495, "grad_norm": 2.59375, "learning_rate": 4.293104223775728e-05, "loss": 0.2065, "step": 8146 }, { "epoch": 50.60436137071651, "grad_norm": 2.625, "learning_rate": 4.292932534075841e-05, "loss": 0.2272, "step": 8147 }, { "epoch": 50.610591900311526, "grad_norm": 1.34375, "learning_rate": 4.292760826962635e-05, "loss": 0.113, "step": 8148 }, { "epoch": 50.61682242990654, "grad_norm": 3.765625, "learning_rate": 4.292589102437778e-05, "loss": 0.2785, "step": 8149 }, { "epoch": 50.62305295950156, "grad_norm": 1.9296875, "learning_rate": 4.2924173605029364e-05, "loss": 0.1353, "step": 8150 }, { "epoch": 50.62928348909657, "grad_norm": 2.703125, "learning_rate": 4.29224560115978e-05, "loss": 0.2538, "step": 8151 }, { "epoch": 50.63551401869159, "grad_norm": 1.6875, "learning_rate": 4.292073824409976e-05, "loss": 0.1425, "step": 8152 }, { "epoch": 50.6417445482866, "grad_norm": 3.421875, "learning_rate": 4.2919020302551924e-05, "loss": 0.2243, "step": 8153 }, { "epoch": 50.64797507788162, "grad_norm": 2.390625, "learning_rate": 4.291730218697098e-05, "loss": 0.1927, "step": 8154 }, { "epoch": 50.654205607476634, "grad_norm": 2.4375, "learning_rate": 4.291558389737362e-05, "loss": 0.243, "step": 8155 }, { "epoch": 50.66043613707165, "grad_norm": 2.890625, "learning_rate": 4.291386543377652e-05, "loss": 0.2371, "step": 8156 }, { "epoch": 50.666666666666664, "grad_norm": 1.25, "learning_rate": 4.2912146796196386e-05, "loss": 0.1281, "step": 8157 }, { "epoch": 50.67289719626168, "grad_norm": 2.15625, "learning_rate": 4.29104279846499e-05, "loss": 0.1861, "step": 8158 }, { "epoch": 50.679127725856695, "grad_norm": 2.390625, "learning_rate": 4.2908708999153756e-05, "loss": 0.2923, "step": 8159 }, { "epoch": 50.68535825545171, "grad_norm": 2.09375, "learning_rate": 4.2906989839724656e-05, "loss": 0.1149, "step": 8160 }, { "epoch": 50.691588785046726, "grad_norm": 2.75, "learning_rate": 4.290527050637928e-05, "loss": 0.2755, "step": 8161 }, { "epoch": 50.69781931464174, "grad_norm": 1.8515625, "learning_rate": 4.2903550999134346e-05, "loss": 0.1716, "step": 8162 }, { "epoch": 50.70404984423676, "grad_norm": 2.5625, "learning_rate": 4.290183131800654e-05, "loss": 0.1918, "step": 8163 }, { "epoch": 50.71028037383178, "grad_norm": 3.0625, "learning_rate": 4.290011146301257e-05, "loss": 0.143, "step": 8164 }, { "epoch": 50.716510903426794, "grad_norm": 2.171875, "learning_rate": 4.289839143416915e-05, "loss": 0.1447, "step": 8165 }, { "epoch": 50.72274143302181, "grad_norm": 2.828125, "learning_rate": 4.2896671231492966e-05, "loss": 0.2545, "step": 8166 }, { "epoch": 50.728971962616825, "grad_norm": 1.2421875, "learning_rate": 4.2894950855000734e-05, "loss": 0.1049, "step": 8167 }, { "epoch": 50.73520249221184, "grad_norm": 3.21875, "learning_rate": 4.2893230304709166e-05, "loss": 0.1881, "step": 8168 }, { "epoch": 50.741433021806856, "grad_norm": 2.71875, "learning_rate": 4.289150958063496e-05, "loss": 0.2071, "step": 8169 }, { "epoch": 50.74766355140187, "grad_norm": 2.421875, "learning_rate": 4.288978868279484e-05, "loss": 0.1427, "step": 8170 }, { "epoch": 50.753894080996886, "grad_norm": 2.3125, "learning_rate": 4.288806761120552e-05, "loss": 0.1344, "step": 8171 }, { "epoch": 50.7601246105919, "grad_norm": 2.125, "learning_rate": 4.288634636588371e-05, "loss": 0.1428, "step": 8172 }, { "epoch": 50.76635514018692, "grad_norm": 2.015625, "learning_rate": 4.288462494684613e-05, "loss": 0.1471, "step": 8173 }, { "epoch": 50.77258566978193, "grad_norm": 2.515625, "learning_rate": 4.288290335410949e-05, "loss": 0.1767, "step": 8174 }, { "epoch": 50.77881619937695, "grad_norm": 3.4375, "learning_rate": 4.288118158769052e-05, "loss": 0.1091, "step": 8175 }, { "epoch": 50.78504672897196, "grad_norm": 3.1875, "learning_rate": 4.287945964760595e-05, "loss": 0.2453, "step": 8176 }, { "epoch": 50.79127725856698, "grad_norm": 2.15625, "learning_rate": 4.2877737533872485e-05, "loss": 0.1564, "step": 8177 }, { "epoch": 50.797507788161994, "grad_norm": 2.890625, "learning_rate": 4.2876015246506864e-05, "loss": 0.1653, "step": 8178 }, { "epoch": 50.80373831775701, "grad_norm": 3.328125, "learning_rate": 4.287429278552581e-05, "loss": 0.2323, "step": 8179 }, { "epoch": 50.809968847352025, "grad_norm": 1.7890625, "learning_rate": 4.2872570150946046e-05, "loss": 0.2127, "step": 8180 }, { "epoch": 50.81619937694704, "grad_norm": 2.046875, "learning_rate": 4.287084734278431e-05, "loss": 0.1867, "step": 8181 }, { "epoch": 50.822429906542055, "grad_norm": 2.484375, "learning_rate": 4.286912436105733e-05, "loss": 0.1415, "step": 8182 }, { "epoch": 50.82866043613707, "grad_norm": 1.5, "learning_rate": 4.286740120578185e-05, "loss": 0.1181, "step": 8183 }, { "epoch": 50.834890965732086, "grad_norm": 1.8203125, "learning_rate": 4.286567787697459e-05, "loss": 0.1635, "step": 8184 }, { "epoch": 50.8411214953271, "grad_norm": 3.3125, "learning_rate": 4.2863954374652304e-05, "loss": 0.2708, "step": 8185 }, { "epoch": 50.84735202492212, "grad_norm": 2.578125, "learning_rate": 4.286223069883172e-05, "loss": 0.2633, "step": 8186 }, { "epoch": 50.85358255451713, "grad_norm": 3.28125, "learning_rate": 4.286050684952957e-05, "loss": 0.3282, "step": 8187 }, { "epoch": 50.85981308411215, "grad_norm": 3.03125, "learning_rate": 4.2858782826762616e-05, "loss": 0.1901, "step": 8188 }, { "epoch": 50.86604361370716, "grad_norm": 2.09375, "learning_rate": 4.2857058630547594e-05, "loss": 0.1458, "step": 8189 }, { "epoch": 50.87227414330218, "grad_norm": 2.5625, "learning_rate": 4.285533426090125e-05, "loss": 0.1316, "step": 8190 }, { "epoch": 50.87850467289719, "grad_norm": 4.15625, "learning_rate": 4.285360971784033e-05, "loss": 0.2696, "step": 8191 }, { "epoch": 50.88473520249221, "grad_norm": 1.6484375, "learning_rate": 4.285188500138157e-05, "loss": 0.1157, "step": 8192 }, { "epoch": 50.890965732087224, "grad_norm": 2.609375, "learning_rate": 4.2850160111541746e-05, "loss": 0.1478, "step": 8193 }, { "epoch": 50.89719626168224, "grad_norm": 2.21875, "learning_rate": 4.28484350483376e-05, "loss": 0.1994, "step": 8194 }, { "epoch": 50.90342679127726, "grad_norm": 2.171875, "learning_rate": 4.284670981178588e-05, "loss": 0.1331, "step": 8195 }, { "epoch": 50.90965732087228, "grad_norm": 1.96875, "learning_rate": 4.2844984401903354e-05, "loss": 0.1452, "step": 8196 }, { "epoch": 50.91588785046729, "grad_norm": 1.8046875, "learning_rate": 4.284325881870677e-05, "loss": 0.163, "step": 8197 }, { "epoch": 50.92211838006231, "grad_norm": 1.8359375, "learning_rate": 4.284153306221289e-05, "loss": 0.1242, "step": 8198 }, { "epoch": 50.92834890965732, "grad_norm": 1.7578125, "learning_rate": 4.283980713243847e-05, "loss": 0.1296, "step": 8199 }, { "epoch": 50.93457943925234, "grad_norm": 1.4921875, "learning_rate": 4.283808102940028e-05, "loss": 0.1166, "step": 8200 }, { "epoch": 50.940809968847354, "grad_norm": 1.7421875, "learning_rate": 4.283635475311508e-05, "loss": 0.1998, "step": 8201 }, { "epoch": 50.94704049844237, "grad_norm": 2.84375, "learning_rate": 4.2834628303599635e-05, "loss": 0.207, "step": 8202 }, { "epoch": 50.953271028037385, "grad_norm": 2.421875, "learning_rate": 4.283290168087072e-05, "loss": 0.2783, "step": 8203 }, { "epoch": 50.9595015576324, "grad_norm": 2.125, "learning_rate": 4.28311748849451e-05, "loss": 0.1423, "step": 8204 }, { "epoch": 50.965732087227416, "grad_norm": 3.015625, "learning_rate": 4.2829447915839536e-05, "loss": 0.1915, "step": 8205 }, { "epoch": 50.97196261682243, "grad_norm": 2.453125, "learning_rate": 4.282772077357082e-05, "loss": 0.2128, "step": 8206 }, { "epoch": 50.978193146417446, "grad_norm": 1.6953125, "learning_rate": 4.282599345815571e-05, "loss": 0.172, "step": 8207 }, { "epoch": 50.98442367601246, "grad_norm": 2.484375, "learning_rate": 4.282426596961099e-05, "loss": 0.206, "step": 8208 }, { "epoch": 50.99065420560748, "grad_norm": 3.765625, "learning_rate": 4.282253830795344e-05, "loss": 0.3354, "step": 8209 }, { "epoch": 50.99688473520249, "grad_norm": 3.28125, "learning_rate": 4.282081047319983e-05, "loss": 0.1923, "step": 8210 }, { "epoch": 51.0, "grad_norm": 2.0625, "learning_rate": 4.2819082465366944e-05, "loss": 0.0881, "step": 8211 }, { "epoch": 51.006230529595015, "grad_norm": 3.75, "learning_rate": 4.281735428447158e-05, "loss": 0.2886, "step": 8212 }, { "epoch": 51.01246105919003, "grad_norm": 3.125, "learning_rate": 4.281562593053049e-05, "loss": 0.2321, "step": 8213 }, { "epoch": 51.018691588785046, "grad_norm": 1.84375, "learning_rate": 4.28138974035605e-05, "loss": 0.1418, "step": 8214 }, { "epoch": 51.02492211838006, "grad_norm": 1.5703125, "learning_rate": 4.281216870357836e-05, "loss": 0.11, "step": 8215 }, { "epoch": 51.03115264797508, "grad_norm": 2.171875, "learning_rate": 4.2810439830600894e-05, "loss": 0.1475, "step": 8216 }, { "epoch": 51.03738317757009, "grad_norm": 2.65625, "learning_rate": 4.2808710784644854e-05, "loss": 0.271, "step": 8217 }, { "epoch": 51.04361370716511, "grad_norm": 2.78125, "learning_rate": 4.280698156572707e-05, "loss": 0.2569, "step": 8218 }, { "epoch": 51.04984423676012, "grad_norm": 3.1875, "learning_rate": 4.280525217386432e-05, "loss": 0.1905, "step": 8219 }, { "epoch": 51.05607476635514, "grad_norm": 3.890625, "learning_rate": 4.2803522609073404e-05, "loss": 0.2922, "step": 8220 }, { "epoch": 51.06230529595015, "grad_norm": 2.640625, "learning_rate": 4.2801792871371116e-05, "loss": 0.1899, "step": 8221 }, { "epoch": 51.06853582554517, "grad_norm": 2.5, "learning_rate": 4.280006296077426e-05, "loss": 0.2406, "step": 8222 }, { "epoch": 51.074766355140184, "grad_norm": 2.671875, "learning_rate": 4.2798332877299624e-05, "loss": 0.2097, "step": 8223 }, { "epoch": 51.0809968847352, "grad_norm": 2.046875, "learning_rate": 4.279660262096403e-05, "loss": 0.1455, "step": 8224 }, { "epoch": 51.087227414330215, "grad_norm": 3.65625, "learning_rate": 4.279487219178427e-05, "loss": 0.187, "step": 8225 }, { "epoch": 51.09345794392523, "grad_norm": 3.28125, "learning_rate": 4.279314158977715e-05, "loss": 0.2075, "step": 8226 }, { "epoch": 51.099688473520246, "grad_norm": 2.40625, "learning_rate": 4.2791410814959495e-05, "loss": 0.1942, "step": 8227 }, { "epoch": 51.10591900311527, "grad_norm": 2.6875, "learning_rate": 4.278967986734809e-05, "loss": 0.2322, "step": 8228 }, { "epoch": 51.11214953271028, "grad_norm": 4.15625, "learning_rate": 4.278794874695976e-05, "loss": 0.159, "step": 8229 }, { "epoch": 51.1183800623053, "grad_norm": 3.5, "learning_rate": 4.278621745381132e-05, "loss": 0.2638, "step": 8230 }, { "epoch": 51.124610591900314, "grad_norm": 3.109375, "learning_rate": 4.2784485987919574e-05, "loss": 0.2835, "step": 8231 }, { "epoch": 51.13084112149533, "grad_norm": 2.171875, "learning_rate": 4.278275434930136e-05, "loss": 0.1229, "step": 8232 }, { "epoch": 51.137071651090345, "grad_norm": 2.4375, "learning_rate": 4.278102253797347e-05, "loss": 0.1814, "step": 8233 }, { "epoch": 51.14330218068536, "grad_norm": 1.7109375, "learning_rate": 4.2779290553952735e-05, "loss": 0.1489, "step": 8234 }, { "epoch": 51.149532710280376, "grad_norm": 4.0, "learning_rate": 4.277755839725598e-05, "loss": 0.2248, "step": 8235 }, { "epoch": 51.15576323987539, "grad_norm": 2.890625, "learning_rate": 4.2775826067900013e-05, "loss": 0.2585, "step": 8236 }, { "epoch": 51.161993769470406, "grad_norm": 4.5, "learning_rate": 4.277409356590168e-05, "loss": 0.2309, "step": 8237 }, { "epoch": 51.16822429906542, "grad_norm": 2.109375, "learning_rate": 4.2772360891277796e-05, "loss": 0.1842, "step": 8238 }, { "epoch": 51.17445482866044, "grad_norm": 1.796875, "learning_rate": 4.277062804404519e-05, "loss": 0.129, "step": 8239 }, { "epoch": 51.18068535825545, "grad_norm": 3.171875, "learning_rate": 4.276889502422069e-05, "loss": 0.2524, "step": 8240 }, { "epoch": 51.18691588785047, "grad_norm": 3.25, "learning_rate": 4.2767161831821136e-05, "loss": 0.2352, "step": 8241 }, { "epoch": 51.19314641744548, "grad_norm": 3.234375, "learning_rate": 4.276542846686334e-05, "loss": 0.2089, "step": 8242 }, { "epoch": 51.1993769470405, "grad_norm": 2.859375, "learning_rate": 4.2763694929364166e-05, "loss": 0.1789, "step": 8243 }, { "epoch": 51.205607476635514, "grad_norm": 2.609375, "learning_rate": 4.276196121934044e-05, "loss": 0.1518, "step": 8244 }, { "epoch": 51.21183800623053, "grad_norm": 3.9375, "learning_rate": 4.276022733680898e-05, "loss": 0.2978, "step": 8245 }, { "epoch": 51.218068535825545, "grad_norm": 1.671875, "learning_rate": 4.2758493281786647e-05, "loss": 0.159, "step": 8246 }, { "epoch": 51.22429906542056, "grad_norm": 1.9609375, "learning_rate": 4.2756759054290284e-05, "loss": 0.1718, "step": 8247 }, { "epoch": 51.230529595015575, "grad_norm": 1.953125, "learning_rate": 4.2755024654336725e-05, "loss": 0.138, "step": 8248 }, { "epoch": 51.23676012461059, "grad_norm": 3.5, "learning_rate": 4.2753290081942816e-05, "loss": 0.2088, "step": 8249 }, { "epoch": 51.242990654205606, "grad_norm": 2.78125, "learning_rate": 4.2751555337125404e-05, "loss": 0.1834, "step": 8250 }, { "epoch": 51.24922118380062, "grad_norm": 1.90625, "learning_rate": 4.274982041990134e-05, "loss": 0.1477, "step": 8251 }, { "epoch": 51.25545171339564, "grad_norm": 1.7890625, "learning_rate": 4.274808533028747e-05, "loss": 0.1129, "step": 8252 }, { "epoch": 51.26168224299065, "grad_norm": 1.484375, "learning_rate": 4.274635006830064e-05, "loss": 0.1086, "step": 8253 }, { "epoch": 51.26791277258567, "grad_norm": 1.9296875, "learning_rate": 4.2744614633957726e-05, "loss": 0.2023, "step": 8254 }, { "epoch": 51.27414330218068, "grad_norm": 3.109375, "learning_rate": 4.274287902727556e-05, "loss": 0.3348, "step": 8255 }, { "epoch": 51.2803738317757, "grad_norm": 2.359375, "learning_rate": 4.274114324827101e-05, "loss": 0.1512, "step": 8256 }, { "epoch": 51.28660436137071, "grad_norm": 2.3125, "learning_rate": 4.273940729696093e-05, "loss": 0.1467, "step": 8257 }, { "epoch": 51.29283489096573, "grad_norm": 4.5, "learning_rate": 4.273767117336217e-05, "loss": 0.2216, "step": 8258 }, { "epoch": 51.299065420560744, "grad_norm": 2.609375, "learning_rate": 4.273593487749161e-05, "loss": 0.2123, "step": 8259 }, { "epoch": 51.30529595015577, "grad_norm": 2.5, "learning_rate": 4.2734198409366113e-05, "loss": 0.1724, "step": 8260 }, { "epoch": 51.31152647975078, "grad_norm": 3.03125, "learning_rate": 4.273246176900252e-05, "loss": 0.1517, "step": 8261 }, { "epoch": 51.3177570093458, "grad_norm": 1.8671875, "learning_rate": 4.2730724956417725e-05, "loss": 0.1691, "step": 8262 }, { "epoch": 51.32398753894081, "grad_norm": 2.546875, "learning_rate": 4.2728987971628575e-05, "loss": 0.2256, "step": 8263 }, { "epoch": 51.33021806853583, "grad_norm": 3.359375, "learning_rate": 4.272725081465196e-05, "loss": 0.3655, "step": 8264 }, { "epoch": 51.33644859813084, "grad_norm": 2.5625, "learning_rate": 4.272551348550474e-05, "loss": 0.1691, "step": 8265 }, { "epoch": 51.34267912772586, "grad_norm": 5.03125, "learning_rate": 4.2723775984203784e-05, "loss": 0.2311, "step": 8266 }, { "epoch": 51.348909657320874, "grad_norm": 2.71875, "learning_rate": 4.272203831076598e-05, "loss": 0.176, "step": 8267 }, { "epoch": 51.35514018691589, "grad_norm": 1.859375, "learning_rate": 4.272030046520819e-05, "loss": 0.1261, "step": 8268 }, { "epoch": 51.361370716510905, "grad_norm": 3.15625, "learning_rate": 4.2718562447547304e-05, "loss": 0.3077, "step": 8269 }, { "epoch": 51.36760124610592, "grad_norm": 4.1875, "learning_rate": 4.27168242578002e-05, "loss": 0.2557, "step": 8270 }, { "epoch": 51.373831775700936, "grad_norm": 2.625, "learning_rate": 4.271508589598375e-05, "loss": 0.1691, "step": 8271 }, { "epoch": 51.38006230529595, "grad_norm": 1.1015625, "learning_rate": 4.271334736211485e-05, "loss": 0.1194, "step": 8272 }, { "epoch": 51.386292834890966, "grad_norm": 2.375, "learning_rate": 4.271160865621038e-05, "loss": 0.132, "step": 8273 }, { "epoch": 51.39252336448598, "grad_norm": 2.890625, "learning_rate": 4.270986977828722e-05, "loss": 0.1569, "step": 8274 }, { "epoch": 51.398753894081, "grad_norm": 1.515625, "learning_rate": 4.270813072836227e-05, "loss": 0.1513, "step": 8275 }, { "epoch": 51.40498442367601, "grad_norm": 4.3125, "learning_rate": 4.2706391506452414e-05, "loss": 0.3518, "step": 8276 }, { "epoch": 51.41121495327103, "grad_norm": 2.6875, "learning_rate": 4.270465211257455e-05, "loss": 0.159, "step": 8277 }, { "epoch": 51.41744548286604, "grad_norm": 2.578125, "learning_rate": 4.270291254674556e-05, "loss": 0.1884, "step": 8278 }, { "epoch": 51.42367601246106, "grad_norm": 2.546875, "learning_rate": 4.270117280898234e-05, "loss": 0.1377, "step": 8279 }, { "epoch": 51.429906542056074, "grad_norm": 2.375, "learning_rate": 4.269943289930179e-05, "loss": 0.1356, "step": 8280 }, { "epoch": 51.43613707165109, "grad_norm": 3.0625, "learning_rate": 4.269769281772082e-05, "loss": 0.331, "step": 8281 }, { "epoch": 51.442367601246104, "grad_norm": 1.5859375, "learning_rate": 4.269595256425631e-05, "loss": 0.1485, "step": 8282 }, { "epoch": 51.44859813084112, "grad_norm": 2.71875, "learning_rate": 4.269421213892518e-05, "loss": 0.1878, "step": 8283 }, { "epoch": 51.454828660436135, "grad_norm": 2.0, "learning_rate": 4.269247154174432e-05, "loss": 0.1341, "step": 8284 }, { "epoch": 51.46105919003115, "grad_norm": 3.140625, "learning_rate": 4.269073077273064e-05, "loss": 0.3415, "step": 8285 }, { "epoch": 51.467289719626166, "grad_norm": 2.03125, "learning_rate": 4.268898983190105e-05, "loss": 0.1761, "step": 8286 }, { "epoch": 51.47352024922118, "grad_norm": 2.484375, "learning_rate": 4.268724871927245e-05, "loss": 0.1534, "step": 8287 }, { "epoch": 51.4797507788162, "grad_norm": 1.8359375, "learning_rate": 4.268550743486176e-05, "loss": 0.1163, "step": 8288 }, { "epoch": 51.48598130841121, "grad_norm": 3.203125, "learning_rate": 4.268376597868587e-05, "loss": 0.2241, "step": 8289 }, { "epoch": 51.49221183800623, "grad_norm": 1.921875, "learning_rate": 4.268202435076173e-05, "loss": 0.1701, "step": 8290 }, { "epoch": 51.49844236760124, "grad_norm": 2.921875, "learning_rate": 4.2680282551106235e-05, "loss": 0.2154, "step": 8291 }, { "epoch": 51.504672897196265, "grad_norm": 3.609375, "learning_rate": 4.267854057973629e-05, "loss": 0.1402, "step": 8292 }, { "epoch": 51.51090342679128, "grad_norm": 1.984375, "learning_rate": 4.267679843666883e-05, "loss": 0.1368, "step": 8293 }, { "epoch": 51.517133956386296, "grad_norm": 2.734375, "learning_rate": 4.2675056121920764e-05, "loss": 0.241, "step": 8294 }, { "epoch": 51.52336448598131, "grad_norm": 1.78125, "learning_rate": 4.2673313635509026e-05, "loss": 0.2113, "step": 8295 }, { "epoch": 51.52959501557633, "grad_norm": 4.0625, "learning_rate": 4.267157097745054e-05, "loss": 0.238, "step": 8296 }, { "epoch": 51.53582554517134, "grad_norm": 4.8125, "learning_rate": 4.2669828147762213e-05, "loss": 0.2633, "step": 8297 }, { "epoch": 51.54205607476636, "grad_norm": 2.8125, "learning_rate": 4.2668085146460987e-05, "loss": 0.1501, "step": 8298 }, { "epoch": 51.54828660436137, "grad_norm": 2.984375, "learning_rate": 4.2666341973563787e-05, "loss": 0.2415, "step": 8299 }, { "epoch": 51.55451713395639, "grad_norm": 2.03125, "learning_rate": 4.266459862908754e-05, "loss": 0.135, "step": 8300 }, { "epoch": 51.5607476635514, "grad_norm": 3.328125, "learning_rate": 4.266285511304918e-05, "loss": 0.2457, "step": 8301 }, { "epoch": 51.56697819314642, "grad_norm": 1.5703125, "learning_rate": 4.266111142546564e-05, "loss": 0.138, "step": 8302 }, { "epoch": 51.573208722741434, "grad_norm": 3.1875, "learning_rate": 4.265936756635386e-05, "loss": 0.2751, "step": 8303 }, { "epoch": 51.57943925233645, "grad_norm": 2.21875, "learning_rate": 4.265762353573077e-05, "loss": 0.1862, "step": 8304 }, { "epoch": 51.585669781931465, "grad_norm": 3.0625, "learning_rate": 4.2655879333613307e-05, "loss": 0.2698, "step": 8305 }, { "epoch": 51.59190031152648, "grad_norm": 3.671875, "learning_rate": 4.265413496001842e-05, "loss": 0.36, "step": 8306 }, { "epoch": 51.598130841121495, "grad_norm": 2.53125, "learning_rate": 4.265239041496304e-05, "loss": 0.1688, "step": 8307 }, { "epoch": 51.60436137071651, "grad_norm": 2.546875, "learning_rate": 4.265064569846412e-05, "loss": 0.1404, "step": 8308 }, { "epoch": 51.610591900311526, "grad_norm": 2.203125, "learning_rate": 4.264890081053859e-05, "loss": 0.2328, "step": 8309 }, { "epoch": 51.61682242990654, "grad_norm": 2.4375, "learning_rate": 4.264715575120342e-05, "loss": 0.1617, "step": 8310 }, { "epoch": 51.62305295950156, "grad_norm": 3.25, "learning_rate": 4.2645410520475546e-05, "loss": 0.2349, "step": 8311 }, { "epoch": 51.62928348909657, "grad_norm": 2.125, "learning_rate": 4.264366511837191e-05, "loss": 0.1802, "step": 8312 }, { "epoch": 51.63551401869159, "grad_norm": 2.3125, "learning_rate": 4.264191954490948e-05, "loss": 0.1829, "step": 8313 }, { "epoch": 51.6417445482866, "grad_norm": 3.0625, "learning_rate": 4.26401738001052e-05, "loss": 0.2539, "step": 8314 }, { "epoch": 51.64797507788162, "grad_norm": 2.359375, "learning_rate": 4.263842788397602e-05, "loss": 0.169, "step": 8315 }, { "epoch": 51.654205607476634, "grad_norm": 1.421875, "learning_rate": 4.26366817965389e-05, "loss": 0.119, "step": 8316 }, { "epoch": 51.66043613707165, "grad_norm": 2.140625, "learning_rate": 4.2634935537810807e-05, "loss": 0.1768, "step": 8317 }, { "epoch": 51.666666666666664, "grad_norm": 2.328125, "learning_rate": 4.2633189107808694e-05, "loss": 0.1399, "step": 8318 }, { "epoch": 51.67289719626168, "grad_norm": 2.21875, "learning_rate": 4.263144250654952e-05, "loss": 0.1404, "step": 8319 }, { "epoch": 51.679127725856695, "grad_norm": 2.46875, "learning_rate": 4.2629695734050254e-05, "loss": 0.19, "step": 8320 }, { "epoch": 51.68535825545171, "grad_norm": 1.171875, "learning_rate": 4.262794879032786e-05, "loss": 0.1154, "step": 8321 }, { "epoch": 51.691588785046726, "grad_norm": 1.0859375, "learning_rate": 4.26262016753993e-05, "loss": 0.1218, "step": 8322 }, { "epoch": 51.69781931464174, "grad_norm": 2.3125, "learning_rate": 4.2624454389281556e-05, "loss": 0.183, "step": 8323 }, { "epoch": 51.70404984423676, "grad_norm": 2.375, "learning_rate": 4.262270693199158e-05, "loss": 0.1857, "step": 8324 }, { "epoch": 51.71028037383178, "grad_norm": 2.140625, "learning_rate": 4.262095930354635e-05, "loss": 0.1576, "step": 8325 }, { "epoch": 51.716510903426794, "grad_norm": 2.25, "learning_rate": 4.2619211503962855e-05, "loss": 0.1203, "step": 8326 }, { "epoch": 51.72274143302181, "grad_norm": 2.8125, "learning_rate": 4.261746353325804e-05, "loss": 0.1835, "step": 8327 }, { "epoch": 51.728971962616825, "grad_norm": 1.2265625, "learning_rate": 4.26157153914489e-05, "loss": 0.0989, "step": 8328 }, { "epoch": 51.73520249221184, "grad_norm": 1.8046875, "learning_rate": 4.261396707855241e-05, "loss": 0.1497, "step": 8329 }, { "epoch": 51.741433021806856, "grad_norm": 2.71875, "learning_rate": 4.261221859458556e-05, "loss": 0.1714, "step": 8330 }, { "epoch": 51.74766355140187, "grad_norm": 1.3203125, "learning_rate": 4.261046993956531e-05, "loss": 0.1187, "step": 8331 }, { "epoch": 51.753894080996886, "grad_norm": 2.53125, "learning_rate": 4.2608721113508664e-05, "loss": 0.1638, "step": 8332 }, { "epoch": 51.7601246105919, "grad_norm": 2.25, "learning_rate": 4.26069721164326e-05, "loss": 0.1606, "step": 8333 }, { "epoch": 51.76635514018692, "grad_norm": 1.921875, "learning_rate": 4.26052229483541e-05, "loss": 0.1161, "step": 8334 }, { "epoch": 51.77258566978193, "grad_norm": 1.5234375, "learning_rate": 4.2603473609290144e-05, "loss": 0.1404, "step": 8335 }, { "epoch": 51.77881619937695, "grad_norm": 2.0625, "learning_rate": 4.260172409925775e-05, "loss": 0.1379, "step": 8336 }, { "epoch": 51.78504672897196, "grad_norm": 1.7578125, "learning_rate": 4.259997441827389e-05, "loss": 0.1204, "step": 8337 }, { "epoch": 51.79127725856698, "grad_norm": 2.0, "learning_rate": 4.259822456635556e-05, "loss": 0.1411, "step": 8338 }, { "epoch": 51.797507788161994, "grad_norm": 2.640625, "learning_rate": 4.259647454351975e-05, "loss": 0.1566, "step": 8339 }, { "epoch": 51.80373831775701, "grad_norm": 2.5, "learning_rate": 4.259472434978347e-05, "loss": 0.1713, "step": 8340 }, { "epoch": 51.809968847352025, "grad_norm": 3.03125, "learning_rate": 4.2592973985163695e-05, "loss": 0.134, "step": 8341 }, { "epoch": 51.81619937694704, "grad_norm": 3.59375, "learning_rate": 4.2591223449677455e-05, "loss": 0.2418, "step": 8342 }, { "epoch": 51.822429906542055, "grad_norm": 2.6875, "learning_rate": 4.258947274334173e-05, "loss": 0.2059, "step": 8343 }, { "epoch": 51.82866043613707, "grad_norm": 1.609375, "learning_rate": 4.258772186617353e-05, "loss": 0.1494, "step": 8344 }, { "epoch": 51.834890965732086, "grad_norm": 1.28125, "learning_rate": 4.2585970818189856e-05, "loss": 0.1146, "step": 8345 }, { "epoch": 51.8411214953271, "grad_norm": 3.671875, "learning_rate": 4.258421959940773e-05, "loss": 0.3877, "step": 8346 }, { "epoch": 51.84735202492212, "grad_norm": 2.90625, "learning_rate": 4.258246820984414e-05, "loss": 0.2349, "step": 8347 }, { "epoch": 51.85358255451713, "grad_norm": 2.15625, "learning_rate": 4.25807166495161e-05, "loss": 0.1612, "step": 8348 }, { "epoch": 51.85981308411215, "grad_norm": 1.9921875, "learning_rate": 4.257896491844063e-05, "loss": 0.1545, "step": 8349 }, { "epoch": 51.86604361370716, "grad_norm": 2.640625, "learning_rate": 4.257721301663474e-05, "loss": 0.2313, "step": 8350 }, { "epoch": 51.87227414330218, "grad_norm": 2.71875, "learning_rate": 4.257546094411544e-05, "loss": 0.1639, "step": 8351 }, { "epoch": 51.87850467289719, "grad_norm": 2.796875, "learning_rate": 4.257370870089976e-05, "loss": 0.2112, "step": 8352 }, { "epoch": 51.88473520249221, "grad_norm": 2.265625, "learning_rate": 4.25719562870047e-05, "loss": 0.1472, "step": 8353 }, { "epoch": 51.890965732087224, "grad_norm": 2.578125, "learning_rate": 4.2570203702447285e-05, "loss": 0.2034, "step": 8354 }, { "epoch": 51.89719626168224, "grad_norm": 2.609375, "learning_rate": 4.256845094724454e-05, "loss": 0.1964, "step": 8355 }, { "epoch": 51.90342679127726, "grad_norm": 1.171875, "learning_rate": 4.2566698021413496e-05, "loss": 0.1199, "step": 8356 }, { "epoch": 51.90965732087228, "grad_norm": 2.296875, "learning_rate": 4.256494492497116e-05, "loss": 0.1709, "step": 8357 }, { "epoch": 51.91588785046729, "grad_norm": 2.359375, "learning_rate": 4.256319165793458e-05, "loss": 0.1767, "step": 8358 }, { "epoch": 51.92211838006231, "grad_norm": 1.8359375, "learning_rate": 4.256143822032076e-05, "loss": 0.1451, "step": 8359 }, { "epoch": 51.92834890965732, "grad_norm": 2.734375, "learning_rate": 4.255968461214674e-05, "loss": 0.426, "step": 8360 }, { "epoch": 51.93457943925234, "grad_norm": 2.625, "learning_rate": 4.255793083342956e-05, "loss": 0.1273, "step": 8361 }, { "epoch": 51.940809968847354, "grad_norm": 2.40625, "learning_rate": 4.255617688418625e-05, "loss": 0.1429, "step": 8362 }, { "epoch": 51.94704049844237, "grad_norm": 1.1484375, "learning_rate": 4.255442276443383e-05, "loss": 0.1, "step": 8363 }, { "epoch": 51.953271028037385, "grad_norm": 2.375, "learning_rate": 4.255266847418935e-05, "loss": 0.1341, "step": 8364 }, { "epoch": 51.9595015576324, "grad_norm": 3.5625, "learning_rate": 4.255091401346985e-05, "loss": 0.2875, "step": 8365 }, { "epoch": 51.965732087227416, "grad_norm": 2.203125, "learning_rate": 4.254915938229236e-05, "loss": 0.2217, "step": 8366 }, { "epoch": 51.97196261682243, "grad_norm": 2.765625, "learning_rate": 4.2547404580673926e-05, "loss": 0.1678, "step": 8367 }, { "epoch": 51.978193146417446, "grad_norm": 1.9453125, "learning_rate": 4.2545649608631596e-05, "loss": 0.1449, "step": 8368 }, { "epoch": 51.98442367601246, "grad_norm": 2.765625, "learning_rate": 4.2543894466182405e-05, "loss": 0.2459, "step": 8369 }, { "epoch": 51.99065420560748, "grad_norm": 2.546875, "learning_rate": 4.254213915334341e-05, "loss": 0.218, "step": 8370 }, { "epoch": 51.99688473520249, "grad_norm": 2.71875, "learning_rate": 4.2540383670131645e-05, "loss": 0.2089, "step": 8371 }, { "epoch": 52.0, "grad_norm": 1.3515625, "learning_rate": 4.253862801656417e-05, "loss": 0.0668, "step": 8372 }, { "epoch": 52.006230529595015, "grad_norm": 3.71875, "learning_rate": 4.2536872192658036e-05, "loss": 0.2126, "step": 8373 }, { "epoch": 52.01246105919003, "grad_norm": 2.390625, "learning_rate": 4.253511619843029e-05, "loss": 0.134, "step": 8374 }, { "epoch": 52.018691588785046, "grad_norm": 1.890625, "learning_rate": 4.253336003389799e-05, "loss": 0.1815, "step": 8375 }, { "epoch": 52.02492211838006, "grad_norm": 3.640625, "learning_rate": 4.25316036990782e-05, "loss": 0.2128, "step": 8376 }, { "epoch": 52.03115264797508, "grad_norm": 4.09375, "learning_rate": 4.252984719398796e-05, "loss": 0.2345, "step": 8377 }, { "epoch": 52.03738317757009, "grad_norm": 5.0625, "learning_rate": 4.252809051864435e-05, "loss": 0.2635, "step": 8378 }, { "epoch": 52.04361370716511, "grad_norm": 4.375, "learning_rate": 4.252633367306442e-05, "loss": 0.1889, "step": 8379 }, { "epoch": 52.04984423676012, "grad_norm": 2.390625, "learning_rate": 4.252457665726523e-05, "loss": 0.1548, "step": 8380 }, { "epoch": 52.05607476635514, "grad_norm": 2.984375, "learning_rate": 4.252281947126385e-05, "loss": 0.2039, "step": 8381 }, { "epoch": 52.06230529595015, "grad_norm": 2.296875, "learning_rate": 4.252106211507734e-05, "loss": 0.1316, "step": 8382 }, { "epoch": 52.06853582554517, "grad_norm": 3.203125, "learning_rate": 4.251930458872278e-05, "loss": 0.1721, "step": 8383 }, { "epoch": 52.074766355140184, "grad_norm": 4.0, "learning_rate": 4.251754689221722e-05, "loss": 0.3915, "step": 8384 }, { "epoch": 52.0809968847352, "grad_norm": 3.578125, "learning_rate": 4.251578902557775e-05, "loss": 0.227, "step": 8385 }, { "epoch": 52.087227414330215, "grad_norm": 3.34375, "learning_rate": 4.251403098882144e-05, "loss": 0.2709, "step": 8386 }, { "epoch": 52.09345794392523, "grad_norm": 4.25, "learning_rate": 4.251227278196536e-05, "loss": 0.1942, "step": 8387 }, { "epoch": 52.099688473520246, "grad_norm": 4.25, "learning_rate": 4.251051440502658e-05, "loss": 0.1435, "step": 8388 }, { "epoch": 52.10591900311527, "grad_norm": 2.8125, "learning_rate": 4.250875585802219e-05, "loss": 0.1342, "step": 8389 }, { "epoch": 52.11214953271028, "grad_norm": 2.046875, "learning_rate": 4.250699714096925e-05, "loss": 0.2432, "step": 8390 }, { "epoch": 52.1183800623053, "grad_norm": 3.4375, "learning_rate": 4.250523825388486e-05, "loss": 0.2962, "step": 8391 }, { "epoch": 52.124610591900314, "grad_norm": 2.53125, "learning_rate": 4.25034791967861e-05, "loss": 0.1413, "step": 8392 }, { "epoch": 52.13084112149533, "grad_norm": 3.109375, "learning_rate": 4.250171996969006e-05, "loss": 0.1997, "step": 8393 }, { "epoch": 52.137071651090345, "grad_norm": 1.34375, "learning_rate": 4.249996057261381e-05, "loss": 0.1251, "step": 8394 }, { "epoch": 52.14330218068536, "grad_norm": 2.515625, "learning_rate": 4.249820100557444e-05, "loss": 0.2544, "step": 8395 }, { "epoch": 52.149532710280376, "grad_norm": 2.578125, "learning_rate": 4.2496441268589046e-05, "loss": 0.2165, "step": 8396 }, { "epoch": 52.15576323987539, "grad_norm": 1.46875, "learning_rate": 4.249468136167472e-05, "loss": 0.1028, "step": 8397 }, { "epoch": 52.161993769470406, "grad_norm": 1.71875, "learning_rate": 4.249292128484855e-05, "loss": 0.2102, "step": 8398 }, { "epoch": 52.16822429906542, "grad_norm": 1.734375, "learning_rate": 4.249116103812763e-05, "loss": 0.1527, "step": 8399 }, { "epoch": 52.17445482866044, "grad_norm": 4.0, "learning_rate": 4.248940062152906e-05, "loss": 0.4477, "step": 8400 }, { "epoch": 52.18068535825545, "grad_norm": 1.78125, "learning_rate": 4.248764003506993e-05, "loss": 0.1269, "step": 8401 }, { "epoch": 52.18691588785047, "grad_norm": 1.8984375, "learning_rate": 4.248587927876735e-05, "loss": 0.1598, "step": 8402 }, { "epoch": 52.19314641744548, "grad_norm": 1.953125, "learning_rate": 4.248411835263842e-05, "loss": 0.1508, "step": 8403 }, { "epoch": 52.1993769470405, "grad_norm": 2.25, "learning_rate": 4.248235725670023e-05, "loss": 0.1891, "step": 8404 }, { "epoch": 52.205607476635514, "grad_norm": 2.21875, "learning_rate": 4.248059599096989e-05, "loss": 0.1567, "step": 8405 }, { "epoch": 52.21183800623053, "grad_norm": 3.625, "learning_rate": 4.2478834555464505e-05, "loss": 0.1817, "step": 8406 }, { "epoch": 52.218068535825545, "grad_norm": 3.265625, "learning_rate": 4.24770729502012e-05, "loss": 0.2863, "step": 8407 }, { "epoch": 52.22429906542056, "grad_norm": 1.8203125, "learning_rate": 4.2475311175197056e-05, "loss": 0.1017, "step": 8408 }, { "epoch": 52.230529595015575, "grad_norm": 1.8046875, "learning_rate": 4.24735492304692e-05, "loss": 0.1494, "step": 8409 }, { "epoch": 52.23676012461059, "grad_norm": 2.1875, "learning_rate": 4.247178711603474e-05, "loss": 0.1503, "step": 8410 }, { "epoch": 52.242990654205606, "grad_norm": 2.796875, "learning_rate": 4.247002483191079e-05, "loss": 0.1633, "step": 8411 }, { "epoch": 52.24922118380062, "grad_norm": 1.234375, "learning_rate": 4.246826237811447e-05, "loss": 0.1364, "step": 8412 }, { "epoch": 52.25545171339564, "grad_norm": 2.375, "learning_rate": 4.2466499754662894e-05, "loss": 0.1916, "step": 8413 }, { "epoch": 52.26168224299065, "grad_norm": 2.453125, "learning_rate": 4.246473696157317e-05, "loss": 0.1854, "step": 8414 }, { "epoch": 52.26791277258567, "grad_norm": 2.328125, "learning_rate": 4.246297399886244e-05, "loss": 0.214, "step": 8415 }, { "epoch": 52.27414330218068, "grad_norm": 2.9375, "learning_rate": 4.246121086654782e-05, "loss": 0.2273, "step": 8416 }, { "epoch": 52.2803738317757, "grad_norm": 2.75, "learning_rate": 4.2459447564646424e-05, "loss": 0.1973, "step": 8417 }, { "epoch": 52.28660436137071, "grad_norm": 2.078125, "learning_rate": 4.245768409317538e-05, "loss": 0.215, "step": 8418 }, { "epoch": 52.29283489096573, "grad_norm": 2.28125, "learning_rate": 4.245592045215182e-05, "loss": 0.1631, "step": 8419 }, { "epoch": 52.299065420560744, "grad_norm": 2.40625, "learning_rate": 4.245415664159288e-05, "loss": 0.1562, "step": 8420 }, { "epoch": 52.30529595015577, "grad_norm": 1.5078125, "learning_rate": 4.2452392661515674e-05, "loss": 0.1221, "step": 8421 }, { "epoch": 52.31152647975078, "grad_norm": 2.59375, "learning_rate": 4.245062851193734e-05, "loss": 0.2366, "step": 8422 }, { "epoch": 52.3177570093458, "grad_norm": 2.46875, "learning_rate": 4.2448864192875014e-05, "loss": 0.1979, "step": 8423 }, { "epoch": 52.32398753894081, "grad_norm": 1.7109375, "learning_rate": 4.2447099704345835e-05, "loss": 0.1425, "step": 8424 }, { "epoch": 52.33021806853583, "grad_norm": 1.7421875, "learning_rate": 4.244533504636693e-05, "loss": 0.1714, "step": 8425 }, { "epoch": 52.33644859813084, "grad_norm": 2.21875, "learning_rate": 4.2443570218955455e-05, "loss": 0.1659, "step": 8426 }, { "epoch": 52.34267912772586, "grad_norm": 1.9453125, "learning_rate": 4.244180522212853e-05, "loss": 0.1212, "step": 8427 }, { "epoch": 52.348909657320874, "grad_norm": 1.4296875, "learning_rate": 4.244004005590331e-05, "loss": 0.1243, "step": 8428 }, { "epoch": 52.35514018691589, "grad_norm": 1.4140625, "learning_rate": 4.243827472029693e-05, "loss": 0.1372, "step": 8429 }, { "epoch": 52.361370716510905, "grad_norm": 1.4765625, "learning_rate": 4.243650921532653e-05, "loss": 0.1496, "step": 8430 }, { "epoch": 52.36760124610592, "grad_norm": 3.578125, "learning_rate": 4.2434743541009294e-05, "loss": 0.2711, "step": 8431 }, { "epoch": 52.373831775700936, "grad_norm": 2.359375, "learning_rate": 4.243297769736232e-05, "loss": 0.2144, "step": 8432 }, { "epoch": 52.38006230529595, "grad_norm": 2.609375, "learning_rate": 4.2431211684402796e-05, "loss": 0.1444, "step": 8433 }, { "epoch": 52.386292834890966, "grad_norm": 1.8828125, "learning_rate": 4.242944550214785e-05, "loss": 0.1507, "step": 8434 }, { "epoch": 52.39252336448598, "grad_norm": 1.984375, "learning_rate": 4.242767915061465e-05, "loss": 0.1371, "step": 8435 }, { "epoch": 52.398753894081, "grad_norm": 2.09375, "learning_rate": 4.242591262982035e-05, "loss": 0.1978, "step": 8436 }, { "epoch": 52.40498442367601, "grad_norm": 1.296875, "learning_rate": 4.24241459397821e-05, "loss": 0.151, "step": 8437 }, { "epoch": 52.41121495327103, "grad_norm": 2.78125, "learning_rate": 4.242237908051706e-05, "loss": 0.2261, "step": 8438 }, { "epoch": 52.41744548286604, "grad_norm": 1.9765625, "learning_rate": 4.242061205204239e-05, "loss": 0.165, "step": 8439 }, { "epoch": 52.42367601246106, "grad_norm": 2.28125, "learning_rate": 4.241884485437526e-05, "loss": 0.2439, "step": 8440 }, { "epoch": 52.429906542056074, "grad_norm": 1.90625, "learning_rate": 4.241707748753283e-05, "loss": 0.133, "step": 8441 }, { "epoch": 52.43613707165109, "grad_norm": 2.140625, "learning_rate": 4.241530995153225e-05, "loss": 0.1417, "step": 8442 }, { "epoch": 52.442367601246104, "grad_norm": 2.015625, "learning_rate": 4.24135422463907e-05, "loss": 0.1416, "step": 8443 }, { "epoch": 52.44859813084112, "grad_norm": 3.15625, "learning_rate": 4.241177437212536e-05, "loss": 0.2181, "step": 8444 }, { "epoch": 52.454828660436135, "grad_norm": 1.6640625, "learning_rate": 4.2410006328753374e-05, "loss": 0.1375, "step": 8445 }, { "epoch": 52.46105919003115, "grad_norm": 1.703125, "learning_rate": 4.240823811629192e-05, "loss": 0.1809, "step": 8446 }, { "epoch": 52.467289719626166, "grad_norm": 1.8359375, "learning_rate": 4.240646973475819e-05, "loss": 0.1465, "step": 8447 }, { "epoch": 52.47352024922118, "grad_norm": 2.640625, "learning_rate": 4.240470118416935e-05, "loss": 0.1837, "step": 8448 }, { "epoch": 52.4797507788162, "grad_norm": 2.828125, "learning_rate": 4.240293246454256e-05, "loss": 0.1731, "step": 8449 }, { "epoch": 52.48598130841121, "grad_norm": 2.5, "learning_rate": 4.240116357589502e-05, "loss": 0.1216, "step": 8450 }, { "epoch": 52.49221183800623, "grad_norm": 3.078125, "learning_rate": 4.23993945182439e-05, "loss": 0.2225, "step": 8451 }, { "epoch": 52.49844236760124, "grad_norm": 1.53125, "learning_rate": 4.239762529160638e-05, "loss": 0.1353, "step": 8452 }, { "epoch": 52.504672897196265, "grad_norm": 2.0, "learning_rate": 4.2395855895999646e-05, "loss": 0.1795, "step": 8453 }, { "epoch": 52.51090342679128, "grad_norm": 3.46875, "learning_rate": 4.2394086331440885e-05, "loss": 0.2575, "step": 8454 }, { "epoch": 52.517133956386296, "grad_norm": 3.78125, "learning_rate": 4.239231659794728e-05, "loss": 0.254, "step": 8455 }, { "epoch": 52.52336448598131, "grad_norm": 2.921875, "learning_rate": 4.239054669553601e-05, "loss": 0.1725, "step": 8456 }, { "epoch": 52.52959501557633, "grad_norm": 2.703125, "learning_rate": 4.238877662422429e-05, "loss": 0.1377, "step": 8457 }, { "epoch": 52.53582554517134, "grad_norm": 2.078125, "learning_rate": 4.2387006384029284e-05, "loss": 0.2053, "step": 8458 }, { "epoch": 52.54205607476636, "grad_norm": 1.5234375, "learning_rate": 4.23852359749682e-05, "loss": 0.1254, "step": 8459 }, { "epoch": 52.54828660436137, "grad_norm": 2.546875, "learning_rate": 4.238346539705823e-05, "loss": 0.1583, "step": 8460 }, { "epoch": 52.55451713395639, "grad_norm": 2.359375, "learning_rate": 4.2381694650316563e-05, "loss": 0.1743, "step": 8461 }, { "epoch": 52.5607476635514, "grad_norm": 2.125, "learning_rate": 4.2379923734760404e-05, "loss": 0.1444, "step": 8462 }, { "epoch": 52.56697819314642, "grad_norm": 1.359375, "learning_rate": 4.237815265040696e-05, "loss": 0.1122, "step": 8463 }, { "epoch": 52.573208722741434, "grad_norm": 1.9453125, "learning_rate": 4.2376381397273425e-05, "loss": 0.1497, "step": 8464 }, { "epoch": 52.57943925233645, "grad_norm": 3.21875, "learning_rate": 4.237460997537699e-05, "loss": 0.233, "step": 8465 }, { "epoch": 52.585669781931465, "grad_norm": 2.5, "learning_rate": 4.237283838473487e-05, "loss": 0.1457, "step": 8466 }, { "epoch": 52.59190031152648, "grad_norm": 3.375, "learning_rate": 4.2371066625364274e-05, "loss": 0.3435, "step": 8467 }, { "epoch": 52.598130841121495, "grad_norm": 1.6015625, "learning_rate": 4.236929469728241e-05, "loss": 0.1304, "step": 8468 }, { "epoch": 52.60436137071651, "grad_norm": 2.53125, "learning_rate": 4.236752260050648e-05, "loss": 0.1174, "step": 8469 }, { "epoch": 52.610591900311526, "grad_norm": 1.8359375, "learning_rate": 4.23657503350537e-05, "loss": 0.1306, "step": 8470 }, { "epoch": 52.61682242990654, "grad_norm": 2.40625, "learning_rate": 4.236397790094128e-05, "loss": 0.2512, "step": 8471 }, { "epoch": 52.62305295950156, "grad_norm": 1.96875, "learning_rate": 4.236220529818644e-05, "loss": 0.1593, "step": 8472 }, { "epoch": 52.62928348909657, "grad_norm": 1.8125, "learning_rate": 4.236043252680639e-05, "loss": 0.1574, "step": 8473 }, { "epoch": 52.63551401869159, "grad_norm": 1.6484375, "learning_rate": 4.235865958681835e-05, "loss": 0.1255, "step": 8474 }, { "epoch": 52.6417445482866, "grad_norm": 1.7890625, "learning_rate": 4.2356886478239535e-05, "loss": 0.1557, "step": 8475 }, { "epoch": 52.64797507788162, "grad_norm": 1.8515625, "learning_rate": 4.235511320108717e-05, "loss": 0.1625, "step": 8476 }, { "epoch": 52.654205607476634, "grad_norm": 2.359375, "learning_rate": 4.235333975537847e-05, "loss": 0.1373, "step": 8477 }, { "epoch": 52.66043613707165, "grad_norm": 2.03125, "learning_rate": 4.2351566141130675e-05, "loss": 0.163, "step": 8478 }, { "epoch": 52.666666666666664, "grad_norm": 1.890625, "learning_rate": 4.2349792358361e-05, "loss": 0.1579, "step": 8479 }, { "epoch": 52.67289719626168, "grad_norm": 2.375, "learning_rate": 4.234801840708667e-05, "loss": 0.1344, "step": 8480 }, { "epoch": 52.679127725856695, "grad_norm": 2.4375, "learning_rate": 4.2346244287324924e-05, "loss": 0.1853, "step": 8481 }, { "epoch": 52.68535825545171, "grad_norm": 3.140625, "learning_rate": 4.2344469999092976e-05, "loss": 0.2134, "step": 8482 }, { "epoch": 52.691588785046726, "grad_norm": 1.484375, "learning_rate": 4.2342695542408075e-05, "loss": 0.1177, "step": 8483 }, { "epoch": 52.69781931464174, "grad_norm": 2.4375, "learning_rate": 4.234092091728744e-05, "loss": 0.2699, "step": 8484 }, { "epoch": 52.70404984423676, "grad_norm": 1.8828125, "learning_rate": 4.2339146123748324e-05, "loss": 0.1509, "step": 8485 }, { "epoch": 52.71028037383178, "grad_norm": 1.109375, "learning_rate": 4.233737116180795e-05, "loss": 0.1082, "step": 8486 }, { "epoch": 52.716510903426794, "grad_norm": 1.859375, "learning_rate": 4.233559603148357e-05, "loss": 0.1513, "step": 8487 }, { "epoch": 52.72274143302181, "grad_norm": 3.1875, "learning_rate": 4.233382073279241e-05, "loss": 0.2316, "step": 8488 }, { "epoch": 52.728971962616825, "grad_norm": 1.890625, "learning_rate": 4.233204526575171e-05, "loss": 0.2084, "step": 8489 }, { "epoch": 52.73520249221184, "grad_norm": 2.875, "learning_rate": 4.233026963037873e-05, "loss": 0.1745, "step": 8490 }, { "epoch": 52.741433021806856, "grad_norm": 2.1875, "learning_rate": 4.2328493826690706e-05, "loss": 0.2125, "step": 8491 }, { "epoch": 52.74766355140187, "grad_norm": 2.203125, "learning_rate": 4.2326717854704886e-05, "loss": 0.1581, "step": 8492 }, { "epoch": 52.753894080996886, "grad_norm": 3.0, "learning_rate": 4.232494171443852e-05, "loss": 0.1871, "step": 8493 }, { "epoch": 52.7601246105919, "grad_norm": 2.78125, "learning_rate": 4.232316540590885e-05, "loss": 0.2407, "step": 8494 }, { "epoch": 52.76635514018692, "grad_norm": 2.515625, "learning_rate": 4.232138892913315e-05, "loss": 0.1806, "step": 8495 }, { "epoch": 52.77258566978193, "grad_norm": 1.5234375, "learning_rate": 4.2319612284128644e-05, "loss": 0.118, "step": 8496 }, { "epoch": 52.77881619937695, "grad_norm": 2.65625, "learning_rate": 4.231783547091261e-05, "loss": 0.2165, "step": 8497 }, { "epoch": 52.78504672897196, "grad_norm": 2.34375, "learning_rate": 4.231605848950229e-05, "loss": 0.1706, "step": 8498 }, { "epoch": 52.79127725856698, "grad_norm": 2.21875, "learning_rate": 4.231428133991495e-05, "loss": 0.2325, "step": 8499 }, { "epoch": 52.797507788161994, "grad_norm": 3.78125, "learning_rate": 4.231250402216785e-05, "loss": 0.2774, "step": 8500 }, { "epoch": 52.80373831775701, "grad_norm": 2.984375, "learning_rate": 4.2310726536278256e-05, "loss": 0.3175, "step": 8501 }, { "epoch": 52.809968847352025, "grad_norm": 2.21875, "learning_rate": 4.2308948882263415e-05, "loss": 0.1241, "step": 8502 }, { "epoch": 52.81619937694704, "grad_norm": 2.921875, "learning_rate": 4.230717106014061e-05, "loss": 0.1907, "step": 8503 }, { "epoch": 52.822429906542055, "grad_norm": 4.1875, "learning_rate": 4.2305393069927094e-05, "loss": 0.2438, "step": 8504 }, { "epoch": 52.82866043613707, "grad_norm": 2.046875, "learning_rate": 4.2303614911640146e-05, "loss": 0.1112, "step": 8505 }, { "epoch": 52.834890965732086, "grad_norm": 3.03125, "learning_rate": 4.230183658529703e-05, "loss": 0.2844, "step": 8506 }, { "epoch": 52.8411214953271, "grad_norm": 2.25, "learning_rate": 4.2300058090915004e-05, "loss": 0.1932, "step": 8507 }, { "epoch": 52.84735202492212, "grad_norm": 2.453125, "learning_rate": 4.2298279428511377e-05, "loss": 0.1642, "step": 8508 }, { "epoch": 52.85358255451713, "grad_norm": 3.484375, "learning_rate": 4.229650059810339e-05, "loss": 0.3006, "step": 8509 }, { "epoch": 52.85981308411215, "grad_norm": 2.59375, "learning_rate": 4.2294721599708343e-05, "loss": 0.1821, "step": 8510 }, { "epoch": 52.86604361370716, "grad_norm": 2.75, "learning_rate": 4.22929424333435e-05, "loss": 0.1866, "step": 8511 }, { "epoch": 52.87227414330218, "grad_norm": 2.890625, "learning_rate": 4.229116309902613e-05, "loss": 0.1953, "step": 8512 }, { "epoch": 52.87850467289719, "grad_norm": 3.125, "learning_rate": 4.228938359677354e-05, "loss": 0.1846, "step": 8513 }, { "epoch": 52.88473520249221, "grad_norm": 1.671875, "learning_rate": 4.2287603926603e-05, "loss": 0.1566, "step": 8514 }, { "epoch": 52.890965732087224, "grad_norm": 2.328125, "learning_rate": 4.228582408853179e-05, "loss": 0.2532, "step": 8515 }, { "epoch": 52.89719626168224, "grad_norm": 2.46875, "learning_rate": 4.22840440825772e-05, "loss": 0.1708, "step": 8516 }, { "epoch": 52.90342679127726, "grad_norm": 3.453125, "learning_rate": 4.228226390875653e-05, "loss": 0.2197, "step": 8517 }, { "epoch": 52.90965732087228, "grad_norm": 3.328125, "learning_rate": 4.228048356708705e-05, "loss": 0.2686, "step": 8518 }, { "epoch": 52.91588785046729, "grad_norm": 3.03125, "learning_rate": 4.227870305758606e-05, "loss": 0.237, "step": 8519 }, { "epoch": 52.92211838006231, "grad_norm": 2.890625, "learning_rate": 4.227692238027084e-05, "loss": 0.3154, "step": 8520 }, { "epoch": 52.92834890965732, "grad_norm": 2.15625, "learning_rate": 4.227514153515871e-05, "loss": 0.1358, "step": 8521 }, { "epoch": 52.93457943925234, "grad_norm": 1.9765625, "learning_rate": 4.2273360522266954e-05, "loss": 0.1537, "step": 8522 }, { "epoch": 52.940809968847354, "grad_norm": 1.875, "learning_rate": 4.2271579341612866e-05, "loss": 0.1206, "step": 8523 }, { "epoch": 52.94704049844237, "grad_norm": 1.9609375, "learning_rate": 4.226979799321374e-05, "loss": 0.1525, "step": 8524 }, { "epoch": 52.953271028037385, "grad_norm": 2.890625, "learning_rate": 4.226801647708689e-05, "loss": 0.2302, "step": 8525 }, { "epoch": 52.9595015576324, "grad_norm": 4.875, "learning_rate": 4.226623479324961e-05, "loss": 0.4133, "step": 8526 }, { "epoch": 52.965732087227416, "grad_norm": 5.1875, "learning_rate": 4.2264452941719213e-05, "loss": 0.2167, "step": 8527 }, { "epoch": 52.97196261682243, "grad_norm": 2.15625, "learning_rate": 4.2262670922512984e-05, "loss": 0.1387, "step": 8528 }, { "epoch": 52.978193146417446, "grad_norm": 1.6015625, "learning_rate": 4.226088873564825e-05, "loss": 0.1478, "step": 8529 }, { "epoch": 52.98442367601246, "grad_norm": 2.59375, "learning_rate": 4.225910638114232e-05, "loss": 0.1734, "step": 8530 }, { "epoch": 52.99065420560748, "grad_norm": 2.34375, "learning_rate": 4.2257323859012496e-05, "loss": 0.2485, "step": 8531 }, { "epoch": 52.99688473520249, "grad_norm": 2.421875, "learning_rate": 4.225554116927609e-05, "loss": 0.2822, "step": 8532 }, { "epoch": 53.0, "grad_norm": 1.65625, "learning_rate": 4.225375831195042e-05, "loss": 0.0681, "step": 8533 }, { "epoch": 53.006230529595015, "grad_norm": 3.03125, "learning_rate": 4.2251975287052804e-05, "loss": 0.1913, "step": 8534 }, { "epoch": 53.01246105919003, "grad_norm": 3.8125, "learning_rate": 4.225019209460055e-05, "loss": 0.241, "step": 8535 }, { "epoch": 53.018691588785046, "grad_norm": 3.203125, "learning_rate": 4.224840873461099e-05, "loss": 0.3061, "step": 8536 }, { "epoch": 53.02492211838006, "grad_norm": 2.765625, "learning_rate": 4.224662520710143e-05, "loss": 0.2796, "step": 8537 }, { "epoch": 53.03115264797508, "grad_norm": 2.09375, "learning_rate": 4.2244841512089196e-05, "loss": 0.1144, "step": 8538 }, { "epoch": 53.03738317757009, "grad_norm": 3.09375, "learning_rate": 4.224305764959162e-05, "loss": 0.1779, "step": 8539 }, { "epoch": 53.04361370716511, "grad_norm": 2.65625, "learning_rate": 4.224127361962602e-05, "loss": 0.133, "step": 8540 }, { "epoch": 53.04984423676012, "grad_norm": 2.234375, "learning_rate": 4.223948942220972e-05, "loss": 0.1544, "step": 8541 }, { "epoch": 53.05607476635514, "grad_norm": 1.9375, "learning_rate": 4.223770505736005e-05, "loss": 0.1065, "step": 8542 }, { "epoch": 53.06230529595015, "grad_norm": 3.078125, "learning_rate": 4.223592052509435e-05, "loss": 0.1926, "step": 8543 }, { "epoch": 53.06853582554517, "grad_norm": 3.03125, "learning_rate": 4.2234135825429946e-05, "loss": 0.2207, "step": 8544 }, { "epoch": 53.074766355140184, "grad_norm": 2.046875, "learning_rate": 4.223235095838417e-05, "loss": 0.1397, "step": 8545 }, { "epoch": 53.0809968847352, "grad_norm": 1.8671875, "learning_rate": 4.223056592397435e-05, "loss": 0.0987, "step": 8546 }, { "epoch": 53.087227414330215, "grad_norm": 2.59375, "learning_rate": 4.2228780722217835e-05, "loss": 0.2042, "step": 8547 }, { "epoch": 53.09345794392523, "grad_norm": 3.859375, "learning_rate": 4.222699535313195e-05, "loss": 0.1896, "step": 8548 }, { "epoch": 53.099688473520246, "grad_norm": 2.5, "learning_rate": 4.222520981673406e-05, "loss": 0.1316, "step": 8549 }, { "epoch": 53.10591900311527, "grad_norm": 5.1875, "learning_rate": 4.222342411304147e-05, "loss": 0.248, "step": 8550 }, { "epoch": 53.11214953271028, "grad_norm": 1.5078125, "learning_rate": 4.222163824207155e-05, "loss": 0.1402, "step": 8551 }, { "epoch": 53.1183800623053, "grad_norm": 2.984375, "learning_rate": 4.221985220384164e-05, "loss": 0.1521, "step": 8552 }, { "epoch": 53.124610591900314, "grad_norm": 3.890625, "learning_rate": 4.221806599836907e-05, "loss": 0.2136, "step": 8553 }, { "epoch": 53.13084112149533, "grad_norm": 2.859375, "learning_rate": 4.221627962567122e-05, "loss": 0.1506, "step": 8554 }, { "epoch": 53.137071651090345, "grad_norm": 2.625, "learning_rate": 4.221449308576541e-05, "loss": 0.1279, "step": 8555 }, { "epoch": 53.14330218068536, "grad_norm": 2.0625, "learning_rate": 4.2212706378669e-05, "loss": 0.1395, "step": 8556 }, { "epoch": 53.149532710280376, "grad_norm": 2.109375, "learning_rate": 4.221091950439935e-05, "loss": 0.3588, "step": 8557 }, { "epoch": 53.15576323987539, "grad_norm": 2.109375, "learning_rate": 4.2209132462973806e-05, "loss": 0.1118, "step": 8558 }, { "epoch": 53.161993769470406, "grad_norm": 2.625, "learning_rate": 4.220734525440973e-05, "loss": 0.1792, "step": 8559 }, { "epoch": 53.16822429906542, "grad_norm": 3.09375, "learning_rate": 4.220555787872448e-05, "loss": 0.1768, "step": 8560 }, { "epoch": 53.17445482866044, "grad_norm": 3.359375, "learning_rate": 4.2203770335935403e-05, "loss": 0.1872, "step": 8561 }, { "epoch": 53.18068535825545, "grad_norm": 2.515625, "learning_rate": 4.220198262605988e-05, "loss": 0.1875, "step": 8562 }, { "epoch": 53.18691588785047, "grad_norm": 2.171875, "learning_rate": 4.2200194749115254e-05, "loss": 0.1988, "step": 8563 }, { "epoch": 53.19314641744548, "grad_norm": 1.453125, "learning_rate": 4.21984067051189e-05, "loss": 0.1323, "step": 8564 }, { "epoch": 53.1993769470405, "grad_norm": 2.828125, "learning_rate": 4.2196618494088194e-05, "loss": 0.1238, "step": 8565 }, { "epoch": 53.205607476635514, "grad_norm": 1.9921875, "learning_rate": 4.219483011604048e-05, "loss": 0.1185, "step": 8566 }, { "epoch": 53.21183800623053, "grad_norm": 2.59375, "learning_rate": 4.219304157099314e-05, "loss": 0.2082, "step": 8567 }, { "epoch": 53.218068535825545, "grad_norm": 2.515625, "learning_rate": 4.219125285896355e-05, "loss": 0.1634, "step": 8568 }, { "epoch": 53.22429906542056, "grad_norm": 3.859375, "learning_rate": 4.218946397996907e-05, "loss": 0.1879, "step": 8569 }, { "epoch": 53.230529595015575, "grad_norm": 4.03125, "learning_rate": 4.218767493402708e-05, "loss": 0.3736, "step": 8570 }, { "epoch": 53.23676012461059, "grad_norm": 3.703125, "learning_rate": 4.218588572115495e-05, "loss": 0.1373, "step": 8571 }, { "epoch": 53.242990654205606, "grad_norm": 2.15625, "learning_rate": 4.218409634137007e-05, "loss": 0.1413, "step": 8572 }, { "epoch": 53.24922118380062, "grad_norm": 2.359375, "learning_rate": 4.218230679468981e-05, "loss": 0.1294, "step": 8573 }, { "epoch": 53.25545171339564, "grad_norm": 2.140625, "learning_rate": 4.218051708113155e-05, "loss": 0.1818, "step": 8574 }, { "epoch": 53.26168224299065, "grad_norm": 2.8125, "learning_rate": 4.2178727200712673e-05, "loss": 0.191, "step": 8575 }, { "epoch": 53.26791277258567, "grad_norm": 2.46875, "learning_rate": 4.217693715345057e-05, "loss": 0.1647, "step": 8576 }, { "epoch": 53.27414330218068, "grad_norm": 1.703125, "learning_rate": 4.217514693936261e-05, "loss": 0.1285, "step": 8577 }, { "epoch": 53.2803738317757, "grad_norm": 3.65625, "learning_rate": 4.21733565584662e-05, "loss": 0.143, "step": 8578 }, { "epoch": 53.28660436137071, "grad_norm": 2.8125, "learning_rate": 4.217156601077871e-05, "loss": 0.1808, "step": 8579 }, { "epoch": 53.29283489096573, "grad_norm": 1.8671875, "learning_rate": 4.2169775296317545e-05, "loss": 0.1443, "step": 8580 }, { "epoch": 53.299065420560744, "grad_norm": 2.984375, "learning_rate": 4.2167984415100086e-05, "loss": 0.3025, "step": 8581 }, { "epoch": 53.30529595015577, "grad_norm": 2.921875, "learning_rate": 4.216619336714373e-05, "loss": 0.1936, "step": 8582 }, { "epoch": 53.31152647975078, "grad_norm": 2.609375, "learning_rate": 4.2164402152465874e-05, "loss": 0.192, "step": 8583 }, { "epoch": 53.3177570093458, "grad_norm": 3.796875, "learning_rate": 4.216261077108391e-05, "loss": 0.2395, "step": 8584 }, { "epoch": 53.32398753894081, "grad_norm": 2.515625, "learning_rate": 4.216081922301525e-05, "loss": 0.1576, "step": 8585 }, { "epoch": 53.33021806853583, "grad_norm": 2.765625, "learning_rate": 4.215902750827727e-05, "loss": 0.1491, "step": 8586 }, { "epoch": 53.33644859813084, "grad_norm": 2.40625, "learning_rate": 4.215723562688739e-05, "loss": 0.1411, "step": 8587 }, { "epoch": 53.34267912772586, "grad_norm": 3.921875, "learning_rate": 4.215544357886301e-05, "loss": 0.2479, "step": 8588 }, { "epoch": 53.348909657320874, "grad_norm": 3.0, "learning_rate": 4.215365136422153e-05, "loss": 0.1122, "step": 8589 }, { "epoch": 53.35514018691589, "grad_norm": 2.09375, "learning_rate": 4.2151858982980364e-05, "loss": 0.1517, "step": 8590 }, { "epoch": 53.361370716510905, "grad_norm": 2.78125, "learning_rate": 4.215006643515691e-05, "loss": 0.1226, "step": 8591 }, { "epoch": 53.36760124610592, "grad_norm": 3.09375, "learning_rate": 4.214827372076858e-05, "loss": 0.2451, "step": 8592 }, { "epoch": 53.373831775700936, "grad_norm": 1.7265625, "learning_rate": 4.2146480839832794e-05, "loss": 0.1526, "step": 8593 }, { "epoch": 53.38006230529595, "grad_norm": 2.625, "learning_rate": 4.214468779236695e-05, "loss": 0.1658, "step": 8594 }, { "epoch": 53.386292834890966, "grad_norm": 2.15625, "learning_rate": 4.214289457838849e-05, "loss": 0.1362, "step": 8595 }, { "epoch": 53.39252336448598, "grad_norm": 1.953125, "learning_rate": 4.214110119791479e-05, "loss": 0.1637, "step": 8596 }, { "epoch": 53.398753894081, "grad_norm": 1.8046875, "learning_rate": 4.21393076509633e-05, "loss": 0.1391, "step": 8597 }, { "epoch": 53.40498442367601, "grad_norm": 4.125, "learning_rate": 4.2137513937551425e-05, "loss": 0.2548, "step": 8598 }, { "epoch": 53.41121495327103, "grad_norm": 3.296875, "learning_rate": 4.213572005769659e-05, "loss": 0.1975, "step": 8599 }, { "epoch": 53.41744548286604, "grad_norm": 3.734375, "learning_rate": 4.213392601141621e-05, "loss": 0.2721, "step": 8600 }, { "epoch": 53.42367601246106, "grad_norm": 2.46875, "learning_rate": 4.213213179872772e-05, "loss": 0.1718, "step": 8601 }, { "epoch": 53.429906542056074, "grad_norm": 2.796875, "learning_rate": 4.213033741964855e-05, "loss": 0.1978, "step": 8602 }, { "epoch": 53.43613707165109, "grad_norm": 2.484375, "learning_rate": 4.212854287419611e-05, "loss": 0.1785, "step": 8603 }, { "epoch": 53.442367601246104, "grad_norm": 1.7421875, "learning_rate": 4.212674816238783e-05, "loss": 0.1272, "step": 8604 }, { "epoch": 53.44859813084112, "grad_norm": 3.421875, "learning_rate": 4.212495328424116e-05, "loss": 0.2458, "step": 8605 }, { "epoch": 53.454828660436135, "grad_norm": 3.03125, "learning_rate": 4.212315823977352e-05, "loss": 0.2635, "step": 8606 }, { "epoch": 53.46105919003115, "grad_norm": 3.8125, "learning_rate": 4.212136302900234e-05, "loss": 0.2253, "step": 8607 }, { "epoch": 53.467289719626166, "grad_norm": 3.984375, "learning_rate": 4.211956765194506e-05, "loss": 0.2682, "step": 8608 }, { "epoch": 53.47352024922118, "grad_norm": 2.671875, "learning_rate": 4.2117772108619116e-05, "loss": 0.2362, "step": 8609 }, { "epoch": 53.4797507788162, "grad_norm": 1.421875, "learning_rate": 4.2115976399041954e-05, "loss": 0.1535, "step": 8610 }, { "epoch": 53.48598130841121, "grad_norm": 2.84375, "learning_rate": 4.2114180523231004e-05, "loss": 0.2187, "step": 8611 }, { "epoch": 53.49221183800623, "grad_norm": 2.515625, "learning_rate": 4.211238448120371e-05, "loss": 0.1265, "step": 8612 }, { "epoch": 53.49844236760124, "grad_norm": 2.71875, "learning_rate": 4.2110588272977526e-05, "loss": 0.1518, "step": 8613 }, { "epoch": 53.504672897196265, "grad_norm": 3.90625, "learning_rate": 4.2108791898569875e-05, "loss": 0.1325, "step": 8614 }, { "epoch": 53.51090342679128, "grad_norm": 3.046875, "learning_rate": 4.2106995357998226e-05, "loss": 0.2935, "step": 8615 }, { "epoch": 53.517133956386296, "grad_norm": 1.8203125, "learning_rate": 4.210519865128001e-05, "loss": 0.1677, "step": 8616 }, { "epoch": 53.52336448598131, "grad_norm": 3.40625, "learning_rate": 4.2103401778432694e-05, "loss": 0.2268, "step": 8617 }, { "epoch": 53.52959501557633, "grad_norm": 4.125, "learning_rate": 4.2101604739473716e-05, "loss": 0.1763, "step": 8618 }, { "epoch": 53.53582554517134, "grad_norm": 2.15625, "learning_rate": 4.2099807534420536e-05, "loss": 0.1186, "step": 8619 }, { "epoch": 53.54205607476636, "grad_norm": 1.90625, "learning_rate": 4.2098010163290615e-05, "loss": 0.1426, "step": 8620 }, { "epoch": 53.54828660436137, "grad_norm": 1.3671875, "learning_rate": 4.209621262610138e-05, "loss": 0.0994, "step": 8621 }, { "epoch": 53.55451713395639, "grad_norm": 3.03125, "learning_rate": 4.209441492287033e-05, "loss": 0.2377, "step": 8622 }, { "epoch": 53.5607476635514, "grad_norm": 2.890625, "learning_rate": 4.20926170536149e-05, "loss": 0.202, "step": 8623 }, { "epoch": 53.56697819314642, "grad_norm": 3.875, "learning_rate": 4.209081901835255e-05, "loss": 0.2209, "step": 8624 }, { "epoch": 53.573208722741434, "grad_norm": 4.09375, "learning_rate": 4.208902081710075e-05, "loss": 0.1497, "step": 8625 }, { "epoch": 53.57943925233645, "grad_norm": 2.09375, "learning_rate": 4.208722244987698e-05, "loss": 0.2407, "step": 8626 }, { "epoch": 53.585669781931465, "grad_norm": 1.5859375, "learning_rate": 4.2085423916698674e-05, "loss": 0.1352, "step": 8627 }, { "epoch": 53.59190031152648, "grad_norm": 3.09375, "learning_rate": 4.208362521758332e-05, "loss": 0.2183, "step": 8628 }, { "epoch": 53.598130841121495, "grad_norm": 3.09375, "learning_rate": 4.2081826352548386e-05, "loss": 0.1387, "step": 8629 }, { "epoch": 53.60436137071651, "grad_norm": 3.765625, "learning_rate": 4.208002732161134e-05, "loss": 0.172, "step": 8630 }, { "epoch": 53.610591900311526, "grad_norm": 2.359375, "learning_rate": 4.207822812478965e-05, "loss": 0.1323, "step": 8631 }, { "epoch": 53.61682242990654, "grad_norm": 2.296875, "learning_rate": 4.20764287621008e-05, "loss": 0.1861, "step": 8632 }, { "epoch": 53.62305295950156, "grad_norm": 2.828125, "learning_rate": 4.207462923356226e-05, "loss": 0.3126, "step": 8633 }, { "epoch": 53.62928348909657, "grad_norm": 3.9375, "learning_rate": 4.2072829539191495e-05, "loss": 0.1079, "step": 8634 }, { "epoch": 53.63551401869159, "grad_norm": 2.96875, "learning_rate": 4.2071029679006015e-05, "loss": 0.1692, "step": 8635 }, { "epoch": 53.6417445482866, "grad_norm": 6.8125, "learning_rate": 4.206922965302327e-05, "loss": 0.1746, "step": 8636 }, { "epoch": 53.64797507788162, "grad_norm": 2.8125, "learning_rate": 4.206742946126075e-05, "loss": 0.2238, "step": 8637 }, { "epoch": 53.654205607476634, "grad_norm": 2.390625, "learning_rate": 4.206562910373596e-05, "loss": 0.1362, "step": 8638 }, { "epoch": 53.66043613707165, "grad_norm": 2.59375, "learning_rate": 4.206382858046636e-05, "loss": 0.1537, "step": 8639 }, { "epoch": 53.666666666666664, "grad_norm": 5.125, "learning_rate": 4.206202789146943e-05, "loss": 0.1478, "step": 8640 }, { "epoch": 53.67289719626168, "grad_norm": 2.90625, "learning_rate": 4.20602270367627e-05, "loss": 0.1277, "step": 8641 }, { "epoch": 53.679127725856695, "grad_norm": 4.625, "learning_rate": 4.205842601636363e-05, "loss": 0.2184, "step": 8642 }, { "epoch": 53.68535825545171, "grad_norm": 2.421875, "learning_rate": 4.205662483028971e-05, "loss": 0.1459, "step": 8643 }, { "epoch": 53.691588785046726, "grad_norm": 1.421875, "learning_rate": 4.2054823478558435e-05, "loss": 0.1078, "step": 8644 }, { "epoch": 53.69781931464174, "grad_norm": 3.0, "learning_rate": 4.2053021961187316e-05, "loss": 0.1667, "step": 8645 }, { "epoch": 53.70404984423676, "grad_norm": 3.28125, "learning_rate": 4.205122027819384e-05, "loss": 0.2553, "step": 8646 }, { "epoch": 53.71028037383178, "grad_norm": 3.578125, "learning_rate": 4.204941842959549e-05, "loss": 0.1831, "step": 8647 }, { "epoch": 53.716510903426794, "grad_norm": 4.09375, "learning_rate": 4.2047616415409797e-05, "loss": 0.4186, "step": 8648 }, { "epoch": 53.72274143302181, "grad_norm": 2.1875, "learning_rate": 4.204581423565424e-05, "loss": 0.1804, "step": 8649 }, { "epoch": 53.728971962616825, "grad_norm": 2.15625, "learning_rate": 4.204401189034632e-05, "loss": 0.1081, "step": 8650 }, { "epoch": 53.73520249221184, "grad_norm": 2.546875, "learning_rate": 4.204220937950356e-05, "loss": 0.231, "step": 8651 }, { "epoch": 53.741433021806856, "grad_norm": 2.875, "learning_rate": 4.2040406703143443e-05, "loss": 0.2026, "step": 8652 }, { "epoch": 53.74766355140187, "grad_norm": 2.75, "learning_rate": 4.20386038612835e-05, "loss": 0.2291, "step": 8653 }, { "epoch": 53.753894080996886, "grad_norm": 2.390625, "learning_rate": 4.203680085394123e-05, "loss": 0.1154, "step": 8654 }, { "epoch": 53.7601246105919, "grad_norm": 3.125, "learning_rate": 4.2034997681134136e-05, "loss": 0.1765, "step": 8655 }, { "epoch": 53.76635514018692, "grad_norm": 2.125, "learning_rate": 4.203319434287975e-05, "loss": 0.1017, "step": 8656 }, { "epoch": 53.77258566978193, "grad_norm": 2.890625, "learning_rate": 4.203139083919557e-05, "loss": 0.2066, "step": 8657 }, { "epoch": 53.77881619937695, "grad_norm": 2.15625, "learning_rate": 4.202958717009912e-05, "loss": 0.1399, "step": 8658 }, { "epoch": 53.78504672897196, "grad_norm": 2.484375, "learning_rate": 4.2027783335607916e-05, "loss": 0.227, "step": 8659 }, { "epoch": 53.79127725856698, "grad_norm": 2.515625, "learning_rate": 4.2025979335739476e-05, "loss": 0.1507, "step": 8660 }, { "epoch": 53.797507788161994, "grad_norm": 3.53125, "learning_rate": 4.202417517051132e-05, "loss": 0.1412, "step": 8661 }, { "epoch": 53.80373831775701, "grad_norm": 2.984375, "learning_rate": 4.2022370839940975e-05, "loss": 0.159, "step": 8662 }, { "epoch": 53.809968847352025, "grad_norm": 2.96875, "learning_rate": 4.2020566344045954e-05, "loss": 0.1359, "step": 8663 }, { "epoch": 53.81619937694704, "grad_norm": 2.921875, "learning_rate": 4.2018761682843794e-05, "loss": 0.1516, "step": 8664 }, { "epoch": 53.822429906542055, "grad_norm": 2.421875, "learning_rate": 4.201695685635202e-05, "loss": 0.138, "step": 8665 }, { "epoch": 53.82866043613707, "grad_norm": 3.578125, "learning_rate": 4.2015151864588156e-05, "loss": 0.1501, "step": 8666 }, { "epoch": 53.834890965732086, "grad_norm": 4.53125, "learning_rate": 4.201334670756974e-05, "loss": 0.1629, "step": 8667 }, { "epoch": 53.8411214953271, "grad_norm": 3.578125, "learning_rate": 4.20115413853143e-05, "loss": 0.1449, "step": 8668 }, { "epoch": 53.84735202492212, "grad_norm": 2.578125, "learning_rate": 4.200973589783937e-05, "loss": 0.1513, "step": 8669 }, { "epoch": 53.85358255451713, "grad_norm": 3.046875, "learning_rate": 4.200793024516248e-05, "loss": 0.352, "step": 8670 }, { "epoch": 53.85981308411215, "grad_norm": 1.9609375, "learning_rate": 4.2006124427301176e-05, "loss": 0.1914, "step": 8671 }, { "epoch": 53.86604361370716, "grad_norm": 3.796875, "learning_rate": 4.2004318444272985e-05, "loss": 0.1823, "step": 8672 }, { "epoch": 53.87227414330218, "grad_norm": 2.578125, "learning_rate": 4.200251229609547e-05, "loss": 0.1335, "step": 8673 }, { "epoch": 53.87850467289719, "grad_norm": 3.140625, "learning_rate": 4.200070598278614e-05, "loss": 0.2121, "step": 8674 }, { "epoch": 53.88473520249221, "grad_norm": 2.78125, "learning_rate": 4.199889950436256e-05, "loss": 0.2288, "step": 8675 }, { "epoch": 53.890965732087224, "grad_norm": 1.2734375, "learning_rate": 4.199709286084227e-05, "loss": 0.116, "step": 8676 }, { "epoch": 53.89719626168224, "grad_norm": 1.9296875, "learning_rate": 4.199528605224282e-05, "loss": 0.1521, "step": 8677 }, { "epoch": 53.90342679127726, "grad_norm": 2.828125, "learning_rate": 4.199347907858175e-05, "loss": 0.2541, "step": 8678 }, { "epoch": 53.90965732087228, "grad_norm": 3.09375, "learning_rate": 4.199167193987662e-05, "loss": 0.2484, "step": 8679 }, { "epoch": 53.91588785046729, "grad_norm": 1.6328125, "learning_rate": 4.198986463614497e-05, "loss": 0.1135, "step": 8680 }, { "epoch": 53.92211838006231, "grad_norm": 1.7109375, "learning_rate": 4.198805716740436e-05, "loss": 0.129, "step": 8681 }, { "epoch": 53.92834890965732, "grad_norm": 1.4609375, "learning_rate": 4.1986249533672345e-05, "loss": 0.124, "step": 8682 }, { "epoch": 53.93457943925234, "grad_norm": 3.265625, "learning_rate": 4.1984441734966486e-05, "loss": 0.1143, "step": 8683 }, { "epoch": 53.940809968847354, "grad_norm": 1.03125, "learning_rate": 4.1982633771304316e-05, "loss": 0.102, "step": 8684 }, { "epoch": 53.94704049844237, "grad_norm": 1.6953125, "learning_rate": 4.1980825642703425e-05, "loss": 0.1217, "step": 8685 }, { "epoch": 53.953271028037385, "grad_norm": 2.703125, "learning_rate": 4.197901734918136e-05, "loss": 0.2203, "step": 8686 }, { "epoch": 53.9595015576324, "grad_norm": 3.71875, "learning_rate": 4.197720889075568e-05, "loss": 0.1581, "step": 8687 }, { "epoch": 53.965732087227416, "grad_norm": 3.046875, "learning_rate": 4.1975400267443956e-05, "loss": 0.2185, "step": 8688 }, { "epoch": 53.97196261682243, "grad_norm": 1.9453125, "learning_rate": 4.1973591479263753e-05, "loss": 0.1519, "step": 8689 }, { "epoch": 53.978193146417446, "grad_norm": 1.359375, "learning_rate": 4.197178252623263e-05, "loss": 0.1097, "step": 8690 }, { "epoch": 53.98442367601246, "grad_norm": 2.484375, "learning_rate": 4.196997340836817e-05, "loss": 0.1789, "step": 8691 }, { "epoch": 53.99065420560748, "grad_norm": 1.890625, "learning_rate": 4.1968164125687937e-05, "loss": 0.1096, "step": 8692 }, { "epoch": 53.99688473520249, "grad_norm": 2.421875, "learning_rate": 4.1966354678209495e-05, "loss": 0.1737, "step": 8693 }, { "epoch": 54.0, "grad_norm": 1.8671875, "learning_rate": 4.196454506595043e-05, "loss": 0.1626, "step": 8694 }, { "epoch": 54.006230529595015, "grad_norm": 2.9375, "learning_rate": 4.1962735288928305e-05, "loss": 0.2097, "step": 8695 }, { "epoch": 54.01246105919003, "grad_norm": 2.765625, "learning_rate": 4.196092534716071e-05, "loss": 0.2108, "step": 8696 }, { "epoch": 54.018691588785046, "grad_norm": 2.046875, "learning_rate": 4.195911524066521e-05, "loss": 0.154, "step": 8697 }, { "epoch": 54.02492211838006, "grad_norm": 2.984375, "learning_rate": 4.195730496945941e-05, "loss": 0.2404, "step": 8698 }, { "epoch": 54.03115264797508, "grad_norm": 2.109375, "learning_rate": 4.1955494533560865e-05, "loss": 0.1864, "step": 8699 }, { "epoch": 54.03738317757009, "grad_norm": 1.78125, "learning_rate": 4.1953683932987165e-05, "loss": 0.1234, "step": 8700 }, { "epoch": 54.04361370716511, "grad_norm": 2.578125, "learning_rate": 4.19518731677559e-05, "loss": 0.1533, "step": 8701 }, { "epoch": 54.04984423676012, "grad_norm": 1.8203125, "learning_rate": 4.1950062237884656e-05, "loss": 0.1214, "step": 8702 }, { "epoch": 54.05607476635514, "grad_norm": 1.7265625, "learning_rate": 4.1948251143391015e-05, "loss": 0.1939, "step": 8703 }, { "epoch": 54.06230529595015, "grad_norm": 4.0625, "learning_rate": 4.1946439884292573e-05, "loss": 0.2156, "step": 8704 }, { "epoch": 54.06853582554517, "grad_norm": 3.53125, "learning_rate": 4.1944628460606916e-05, "loss": 0.3732, "step": 8705 }, { "epoch": 54.074766355140184, "grad_norm": 1.9453125, "learning_rate": 4.194281687235165e-05, "loss": 0.15, "step": 8706 }, { "epoch": 54.0809968847352, "grad_norm": 2.859375, "learning_rate": 4.194100511954435e-05, "loss": 0.2332, "step": 8707 }, { "epoch": 54.087227414330215, "grad_norm": 1.9375, "learning_rate": 4.193919320220263e-05, "loss": 0.1434, "step": 8708 }, { "epoch": 54.09345794392523, "grad_norm": 2.125, "learning_rate": 4.1937381120344074e-05, "loss": 0.2034, "step": 8709 }, { "epoch": 54.099688473520246, "grad_norm": 2.546875, "learning_rate": 4.1935568873986286e-05, "loss": 0.1822, "step": 8710 }, { "epoch": 54.10591900311527, "grad_norm": 2.203125, "learning_rate": 4.193375646314687e-05, "loss": 0.1279, "step": 8711 }, { "epoch": 54.11214953271028, "grad_norm": 3.25, "learning_rate": 4.193194388784342e-05, "loss": 0.3016, "step": 8712 }, { "epoch": 54.1183800623053, "grad_norm": 2.359375, "learning_rate": 4.193013114809355e-05, "loss": 0.2128, "step": 8713 }, { "epoch": 54.124610591900314, "grad_norm": 3.34375, "learning_rate": 4.1928318243914866e-05, "loss": 0.2043, "step": 8714 }, { "epoch": 54.13084112149533, "grad_norm": 2.265625, "learning_rate": 4.192650517532497e-05, "loss": 0.1664, "step": 8715 }, { "epoch": 54.137071651090345, "grad_norm": 1.828125, "learning_rate": 4.192469194234148e-05, "loss": 0.135, "step": 8716 }, { "epoch": 54.14330218068536, "grad_norm": 1.8046875, "learning_rate": 4.192287854498198e-05, "loss": 0.1288, "step": 8717 }, { "epoch": 54.149532710280376, "grad_norm": 1.53125, "learning_rate": 4.192106498326411e-05, "loss": 0.146, "step": 8718 }, { "epoch": 54.15576323987539, "grad_norm": 1.984375, "learning_rate": 4.191925125720548e-05, "loss": 0.1583, "step": 8719 }, { "epoch": 54.161993769470406, "grad_norm": 2.515625, "learning_rate": 4.191743736682369e-05, "loss": 0.2341, "step": 8720 }, { "epoch": 54.16822429906542, "grad_norm": 1.6171875, "learning_rate": 4.1915623312136376e-05, "loss": 0.1333, "step": 8721 }, { "epoch": 54.17445482866044, "grad_norm": 1.2109375, "learning_rate": 4.191380909316114e-05, "loss": 0.1253, "step": 8722 }, { "epoch": 54.18068535825545, "grad_norm": 2.90625, "learning_rate": 4.191199470991561e-05, "loss": 0.2028, "step": 8723 }, { "epoch": 54.18691588785047, "grad_norm": 2.9375, "learning_rate": 4.191018016241741e-05, "loss": 0.142, "step": 8724 }, { "epoch": 54.19314641744548, "grad_norm": 2.65625, "learning_rate": 4.190836545068416e-05, "loss": 0.1584, "step": 8725 }, { "epoch": 54.1993769470405, "grad_norm": 3.375, "learning_rate": 4.1906550574733484e-05, "loss": 0.3149, "step": 8726 }, { "epoch": 54.205607476635514, "grad_norm": 2.265625, "learning_rate": 4.1904735534583006e-05, "loss": 0.183, "step": 8727 }, { "epoch": 54.21183800623053, "grad_norm": 2.5625, "learning_rate": 4.190292033025036e-05, "loss": 0.205, "step": 8728 }, { "epoch": 54.218068535825545, "grad_norm": 2.0625, "learning_rate": 4.1901104961753174e-05, "loss": 0.1113, "step": 8729 }, { "epoch": 54.22429906542056, "grad_norm": 1.9296875, "learning_rate": 4.189928942910908e-05, "loss": 0.1114, "step": 8730 }, { "epoch": 54.230529595015575, "grad_norm": 1.09375, "learning_rate": 4.1897473732335704e-05, "loss": 0.1002, "step": 8731 }, { "epoch": 54.23676012461059, "grad_norm": 1.8359375, "learning_rate": 4.1895657871450686e-05, "loss": 0.1253, "step": 8732 }, { "epoch": 54.242990654205606, "grad_norm": 3.25, "learning_rate": 4.1893841846471657e-05, "loss": 0.1346, "step": 8733 }, { "epoch": 54.24922118380062, "grad_norm": 4.03125, "learning_rate": 4.1892025657416273e-05, "loss": 0.2166, "step": 8734 }, { "epoch": 54.25545171339564, "grad_norm": 3.3125, "learning_rate": 4.189020930430215e-05, "loss": 0.2454, "step": 8735 }, { "epoch": 54.26168224299065, "grad_norm": 2.78125, "learning_rate": 4.1888392787146934e-05, "loss": 0.2173, "step": 8736 }, { "epoch": 54.26791277258567, "grad_norm": 2.640625, "learning_rate": 4.188657610596828e-05, "loss": 0.2075, "step": 8737 }, { "epoch": 54.27414330218068, "grad_norm": 2.359375, "learning_rate": 4.1884759260783816e-05, "loss": 0.1876, "step": 8738 }, { "epoch": 54.2803738317757, "grad_norm": 2.921875, "learning_rate": 4.18829422516112e-05, "loss": 0.2169, "step": 8739 }, { "epoch": 54.28660436137071, "grad_norm": 3.0625, "learning_rate": 4.188112507846808e-05, "loss": 0.186, "step": 8740 }, { "epoch": 54.29283489096573, "grad_norm": 1.484375, "learning_rate": 4.187930774137209e-05, "loss": 0.1116, "step": 8741 }, { "epoch": 54.299065420560744, "grad_norm": 2.0625, "learning_rate": 4.187749024034088e-05, "loss": 0.1605, "step": 8742 }, { "epoch": 54.30529595015577, "grad_norm": 2.3125, "learning_rate": 4.187567257539213e-05, "loss": 0.1828, "step": 8743 }, { "epoch": 54.31152647975078, "grad_norm": 2.171875, "learning_rate": 4.187385474654346e-05, "loss": 0.128, "step": 8744 }, { "epoch": 54.3177570093458, "grad_norm": 1.6953125, "learning_rate": 4.1872036753812547e-05, "loss": 0.1473, "step": 8745 }, { "epoch": 54.32398753894081, "grad_norm": 3.078125, "learning_rate": 4.187021859721704e-05, "loss": 0.3694, "step": 8746 }, { "epoch": 54.33021806853583, "grad_norm": 2.140625, "learning_rate": 4.1868400276774596e-05, "loss": 0.1816, "step": 8747 }, { "epoch": 54.33644859813084, "grad_norm": 4.40625, "learning_rate": 4.1866581792502875e-05, "loss": 0.1798, "step": 8748 }, { "epoch": 54.34267912772586, "grad_norm": 2.265625, "learning_rate": 4.1864763144419536e-05, "loss": 0.1975, "step": 8749 }, { "epoch": 54.348909657320874, "grad_norm": 3.390625, "learning_rate": 4.186294433254226e-05, "loss": 0.2383, "step": 8750 }, { "epoch": 54.35514018691589, "grad_norm": 3.640625, "learning_rate": 4.1861125356888686e-05, "loss": 0.2114, "step": 8751 }, { "epoch": 54.361370716510905, "grad_norm": 2.640625, "learning_rate": 4.18593062174765e-05, "loss": 0.2253, "step": 8752 }, { "epoch": 54.36760124610592, "grad_norm": 2.453125, "learning_rate": 4.185748691432336e-05, "loss": 0.1861, "step": 8753 }, { "epoch": 54.373831775700936, "grad_norm": 2.5, "learning_rate": 4.185566744744693e-05, "loss": 0.12, "step": 8754 }, { "epoch": 54.38006230529595, "grad_norm": 2.234375, "learning_rate": 4.18538478168649e-05, "loss": 0.1641, "step": 8755 }, { "epoch": 54.386292834890966, "grad_norm": 3.40625, "learning_rate": 4.1852028022594924e-05, "loss": 0.2203, "step": 8756 }, { "epoch": 54.39252336448598, "grad_norm": 2.875, "learning_rate": 4.1850208064654685e-05, "loss": 0.1835, "step": 8757 }, { "epoch": 54.398753894081, "grad_norm": 3.5, "learning_rate": 4.1848387943061854e-05, "loss": 0.2379, "step": 8758 }, { "epoch": 54.40498442367601, "grad_norm": 3.640625, "learning_rate": 4.1846567657834115e-05, "loss": 0.1435, "step": 8759 }, { "epoch": 54.41121495327103, "grad_norm": 2.328125, "learning_rate": 4.1844747208989135e-05, "loss": 0.126, "step": 8760 }, { "epoch": 54.41744548286604, "grad_norm": 2.296875, "learning_rate": 4.1842926596544607e-05, "loss": 0.158, "step": 8761 }, { "epoch": 54.42367601246106, "grad_norm": 1.53125, "learning_rate": 4.1841105820518216e-05, "loss": 0.1127, "step": 8762 }, { "epoch": 54.429906542056074, "grad_norm": 1.65625, "learning_rate": 4.183928488092763e-05, "loss": 0.1329, "step": 8763 }, { "epoch": 54.43613707165109, "grad_norm": 2.484375, "learning_rate": 4.183746377779055e-05, "loss": 0.1697, "step": 8764 }, { "epoch": 54.442367601246104, "grad_norm": 3.140625, "learning_rate": 4.1835642511124656e-05, "loss": 0.2277, "step": 8765 }, { "epoch": 54.44859813084112, "grad_norm": 2.59375, "learning_rate": 4.1833821080947635e-05, "loss": 0.1855, "step": 8766 }, { "epoch": 54.454828660436135, "grad_norm": 3.734375, "learning_rate": 4.1831999487277176e-05, "loss": 0.2673, "step": 8767 }, { "epoch": 54.46105919003115, "grad_norm": 3.375, "learning_rate": 4.183017773013098e-05, "loss": 0.145, "step": 8768 }, { "epoch": 54.467289719626166, "grad_norm": 2.5625, "learning_rate": 4.1828355809526726e-05, "loss": 0.1794, "step": 8769 }, { "epoch": 54.47352024922118, "grad_norm": 2.890625, "learning_rate": 4.182653372548212e-05, "loss": 0.1941, "step": 8770 }, { "epoch": 54.4797507788162, "grad_norm": 1.2578125, "learning_rate": 4.182471147801485e-05, "loss": 0.12, "step": 8771 }, { "epoch": 54.48598130841121, "grad_norm": 2.625, "learning_rate": 4.182288906714262e-05, "loss": 0.1632, "step": 8772 }, { "epoch": 54.49221183800623, "grad_norm": 1.7890625, "learning_rate": 4.182106649288314e-05, "loss": 0.1513, "step": 8773 }, { "epoch": 54.49844236760124, "grad_norm": 2.046875, "learning_rate": 4.181924375525409e-05, "loss": 0.1717, "step": 8774 }, { "epoch": 54.504672897196265, "grad_norm": 1.4375, "learning_rate": 4.181742085427318e-05, "loss": 0.1189, "step": 8775 }, { "epoch": 54.51090342679128, "grad_norm": 1.9765625, "learning_rate": 4.1815597789958124e-05, "loss": 0.1742, "step": 8776 }, { "epoch": 54.517133956386296, "grad_norm": 1.3515625, "learning_rate": 4.181377456232661e-05, "loss": 0.1272, "step": 8777 }, { "epoch": 54.52336448598131, "grad_norm": 2.09375, "learning_rate": 4.181195117139637e-05, "loss": 0.2095, "step": 8778 }, { "epoch": 54.52959501557633, "grad_norm": 2.15625, "learning_rate": 4.18101276171851e-05, "loss": 0.1917, "step": 8779 }, { "epoch": 54.53582554517134, "grad_norm": 1.90625, "learning_rate": 4.18083038997105e-05, "loss": 0.1744, "step": 8780 }, { "epoch": 54.54205607476636, "grad_norm": 2.15625, "learning_rate": 4.180648001899029e-05, "loss": 0.1404, "step": 8781 }, { "epoch": 54.54828660436137, "grad_norm": 3.59375, "learning_rate": 4.18046559750422e-05, "loss": 0.2783, "step": 8782 }, { "epoch": 54.55451713395639, "grad_norm": 2.890625, "learning_rate": 4.180283176788392e-05, "loss": 0.2265, "step": 8783 }, { "epoch": 54.5607476635514, "grad_norm": 1.6484375, "learning_rate": 4.180100739753319e-05, "loss": 0.1202, "step": 8784 }, { "epoch": 54.56697819314642, "grad_norm": 3.046875, "learning_rate": 4.179918286400771e-05, "loss": 0.2393, "step": 8785 }, { "epoch": 54.573208722741434, "grad_norm": 2.390625, "learning_rate": 4.1797358167325205e-05, "loss": 0.2129, "step": 8786 }, { "epoch": 54.57943925233645, "grad_norm": 2.296875, "learning_rate": 4.179553330750341e-05, "loss": 0.191, "step": 8787 }, { "epoch": 54.585669781931465, "grad_norm": 2.203125, "learning_rate": 4.179370828456003e-05, "loss": 0.1984, "step": 8788 }, { "epoch": 54.59190031152648, "grad_norm": 3.296875, "learning_rate": 4.179188309851281e-05, "loss": 0.2533, "step": 8789 }, { "epoch": 54.598130841121495, "grad_norm": 2.265625, "learning_rate": 4.179005774937945e-05, "loss": 0.1759, "step": 8790 }, { "epoch": 54.60436137071651, "grad_norm": 1.5, "learning_rate": 4.17882322371777e-05, "loss": 0.1444, "step": 8791 }, { "epoch": 54.610591900311526, "grad_norm": 2.46875, "learning_rate": 4.1786406561925276e-05, "loss": 0.1557, "step": 8792 }, { "epoch": 54.61682242990654, "grad_norm": 2.6875, "learning_rate": 4.1784580723639923e-05, "loss": 0.3139, "step": 8793 }, { "epoch": 54.62305295950156, "grad_norm": 3.578125, "learning_rate": 4.178275472233937e-05, "loss": 0.2743, "step": 8794 }, { "epoch": 54.62928348909657, "grad_norm": 3.03125, "learning_rate": 4.178092855804134e-05, "loss": 0.1946, "step": 8795 }, { "epoch": 54.63551401869159, "grad_norm": 2.234375, "learning_rate": 4.1779102230763576e-05, "loss": 0.1506, "step": 8796 }, { "epoch": 54.6417445482866, "grad_norm": 1.9296875, "learning_rate": 4.177727574052382e-05, "loss": 0.1507, "step": 8797 }, { "epoch": 54.64797507788162, "grad_norm": 2.5, "learning_rate": 4.177544908733981e-05, "loss": 0.0995, "step": 8798 }, { "epoch": 54.654205607476634, "grad_norm": 2.015625, "learning_rate": 4.1773622271229283e-05, "loss": 0.1505, "step": 8799 }, { "epoch": 54.66043613707165, "grad_norm": 3.09375, "learning_rate": 4.1771795292209994e-05, "loss": 0.1477, "step": 8800 }, { "epoch": 54.666666666666664, "grad_norm": 2.859375, "learning_rate": 4.176996815029967e-05, "loss": 0.2464, "step": 8801 }, { "epoch": 54.67289719626168, "grad_norm": 1.78125, "learning_rate": 4.176814084551606e-05, "loss": 0.1524, "step": 8802 }, { "epoch": 54.679127725856695, "grad_norm": 1.890625, "learning_rate": 4.176631337787691e-05, "loss": 0.2562, "step": 8803 }, { "epoch": 54.68535825545171, "grad_norm": 2.03125, "learning_rate": 4.176448574739999e-05, "loss": 0.1132, "step": 8804 }, { "epoch": 54.691588785046726, "grad_norm": 1.765625, "learning_rate": 4.1762657954103016e-05, "loss": 0.1401, "step": 8805 }, { "epoch": 54.69781931464174, "grad_norm": 1.8046875, "learning_rate": 4.176082999800376e-05, "loss": 0.1435, "step": 8806 }, { "epoch": 54.70404984423676, "grad_norm": 2.5, "learning_rate": 4.175900187911997e-05, "loss": 0.2898, "step": 8807 }, { "epoch": 54.71028037383178, "grad_norm": 2.921875, "learning_rate": 4.1757173597469414e-05, "loss": 0.2189, "step": 8808 }, { "epoch": 54.716510903426794, "grad_norm": 2.9375, "learning_rate": 4.175534515306983e-05, "loss": 0.2282, "step": 8809 }, { "epoch": 54.72274143302181, "grad_norm": 3.421875, "learning_rate": 4.175351654593899e-05, "loss": 0.2657, "step": 8810 }, { "epoch": 54.728971962616825, "grad_norm": 4.6875, "learning_rate": 4.175168777609465e-05, "loss": 0.3383, "step": 8811 }, { "epoch": 54.73520249221184, "grad_norm": 2.765625, "learning_rate": 4.174985884355456e-05, "loss": 0.1636, "step": 8812 }, { "epoch": 54.741433021806856, "grad_norm": 2.453125, "learning_rate": 4.1748029748336506e-05, "loss": 0.2113, "step": 8813 }, { "epoch": 54.74766355140187, "grad_norm": 1.8203125, "learning_rate": 4.174620049045823e-05, "loss": 0.1732, "step": 8814 }, { "epoch": 54.753894080996886, "grad_norm": 4.1875, "learning_rate": 4.1744371069937505e-05, "loss": 0.2108, "step": 8815 }, { "epoch": 54.7601246105919, "grad_norm": 2.03125, "learning_rate": 4.1742541486792106e-05, "loss": 0.1642, "step": 8816 }, { "epoch": 54.76635514018692, "grad_norm": 3.453125, "learning_rate": 4.174071174103979e-05, "loss": 0.256, "step": 8817 }, { "epoch": 54.77258566978193, "grad_norm": 2.046875, "learning_rate": 4.1738881832698354e-05, "loss": 0.1667, "step": 8818 }, { "epoch": 54.77881619937695, "grad_norm": 2.859375, "learning_rate": 4.1737051761785535e-05, "loss": 0.2237, "step": 8819 }, { "epoch": 54.78504672897196, "grad_norm": 3.578125, "learning_rate": 4.173522152831913e-05, "loss": 0.3068, "step": 8820 }, { "epoch": 54.79127725856698, "grad_norm": 1.7890625, "learning_rate": 4.17333911323169e-05, "loss": 0.2307, "step": 8821 }, { "epoch": 54.797507788161994, "grad_norm": 2.171875, "learning_rate": 4.1731560573796635e-05, "loss": 0.1686, "step": 8822 }, { "epoch": 54.80373831775701, "grad_norm": 2.453125, "learning_rate": 4.1729729852776116e-05, "loss": 0.1293, "step": 8823 }, { "epoch": 54.809968847352025, "grad_norm": 2.921875, "learning_rate": 4.172789896927311e-05, "loss": 0.2466, "step": 8824 }, { "epoch": 54.81619937694704, "grad_norm": 2.53125, "learning_rate": 4.1726067923305415e-05, "loss": 0.2508, "step": 8825 }, { "epoch": 54.822429906542055, "grad_norm": 3.359375, "learning_rate": 4.172423671489079e-05, "loss": 0.4194, "step": 8826 }, { "epoch": 54.82866043613707, "grad_norm": 3.0, "learning_rate": 4.172240534404704e-05, "loss": 0.3345, "step": 8827 }, { "epoch": 54.834890965732086, "grad_norm": 3.5, "learning_rate": 4.172057381079195e-05, "loss": 0.2998, "step": 8828 }, { "epoch": 54.8411214953271, "grad_norm": 2.75, "learning_rate": 4.1718742115143307e-05, "loss": 0.1207, "step": 8829 }, { "epoch": 54.84735202492212, "grad_norm": 3.15625, "learning_rate": 4.171691025711889e-05, "loss": 0.3183, "step": 8830 }, { "epoch": 54.85358255451713, "grad_norm": 1.2421875, "learning_rate": 4.17150782367365e-05, "loss": 0.1028, "step": 8831 }, { "epoch": 54.85981308411215, "grad_norm": 3.828125, "learning_rate": 4.1713246054013934e-05, "loss": 0.2683, "step": 8832 }, { "epoch": 54.86604361370716, "grad_norm": 1.921875, "learning_rate": 4.171141370896898e-05, "loss": 0.1302, "step": 8833 }, { "epoch": 54.87227414330218, "grad_norm": 3.234375, "learning_rate": 4.1709581201619435e-05, "loss": 0.2809, "step": 8834 }, { "epoch": 54.87850467289719, "grad_norm": 1.515625, "learning_rate": 4.170774853198309e-05, "loss": 0.1192, "step": 8835 }, { "epoch": 54.88473520249221, "grad_norm": 2.859375, "learning_rate": 4.170591570007777e-05, "loss": 0.2309, "step": 8836 }, { "epoch": 54.890965732087224, "grad_norm": 2.4375, "learning_rate": 4.1704082705921246e-05, "loss": 0.1388, "step": 8837 }, { "epoch": 54.89719626168224, "grad_norm": 2.03125, "learning_rate": 4.170224954953132e-05, "loss": 0.1417, "step": 8838 }, { "epoch": 54.90342679127726, "grad_norm": 3.015625, "learning_rate": 4.170041623092583e-05, "loss": 0.1718, "step": 8839 }, { "epoch": 54.90965732087228, "grad_norm": 1.8828125, "learning_rate": 4.1698582750122544e-05, "loss": 0.1559, "step": 8840 }, { "epoch": 54.91588785046729, "grad_norm": 2.0625, "learning_rate": 4.169674910713929e-05, "loss": 0.1643, "step": 8841 }, { "epoch": 54.92211838006231, "grad_norm": 1.78125, "learning_rate": 4.169491530199388e-05, "loss": 0.1094, "step": 8842 }, { "epoch": 54.92834890965732, "grad_norm": 2.140625, "learning_rate": 4.16930813347041e-05, "loss": 0.1855, "step": 8843 }, { "epoch": 54.93457943925234, "grad_norm": 2.828125, "learning_rate": 4.169124720528777e-05, "loss": 0.1965, "step": 8844 }, { "epoch": 54.940809968847354, "grad_norm": 3.109375, "learning_rate": 4.168941291376273e-05, "loss": 0.2675, "step": 8845 }, { "epoch": 54.94704049844237, "grad_norm": 2.984375, "learning_rate": 4.168757846014677e-05, "loss": 0.1929, "step": 8846 }, { "epoch": 54.953271028037385, "grad_norm": 2.34375, "learning_rate": 4.168574384445771e-05, "loss": 0.171, "step": 8847 }, { "epoch": 54.9595015576324, "grad_norm": 2.6875, "learning_rate": 4.168390906671337e-05, "loss": 0.1611, "step": 8848 }, { "epoch": 54.965732087227416, "grad_norm": 2.75, "learning_rate": 4.168207412693157e-05, "loss": 0.2653, "step": 8849 }, { "epoch": 54.97196261682243, "grad_norm": 1.8203125, "learning_rate": 4.1680239025130134e-05, "loss": 0.1129, "step": 8850 }, { "epoch": 54.978193146417446, "grad_norm": 1.71875, "learning_rate": 4.167840376132688e-05, "loss": 0.1423, "step": 8851 }, { "epoch": 54.98442367601246, "grad_norm": 1.140625, "learning_rate": 4.167656833553964e-05, "loss": 0.1071, "step": 8852 }, { "epoch": 54.99065420560748, "grad_norm": 1.8203125, "learning_rate": 4.167473274778623e-05, "loss": 0.1282, "step": 8853 }, { "epoch": 54.99688473520249, "grad_norm": 1.671875, "learning_rate": 4.167289699808448e-05, "loss": 0.1402, "step": 8854 }, { "epoch": 55.0, "grad_norm": 1.2578125, "learning_rate": 4.167106108645223e-05, "loss": 0.0595, "step": 8855 }, { "epoch": 55.006230529595015, "grad_norm": 2.578125, "learning_rate": 4.166922501290729e-05, "loss": 0.1452, "step": 8856 }, { "epoch": 55.01246105919003, "grad_norm": 3.1875, "learning_rate": 4.166738877746751e-05, "loss": 0.3761, "step": 8857 }, { "epoch": 55.018691588785046, "grad_norm": 3.171875, "learning_rate": 4.166555238015072e-05, "loss": 0.3196, "step": 8858 }, { "epoch": 55.02492211838006, "grad_norm": 2.328125, "learning_rate": 4.166371582097476e-05, "loss": 0.1149, "step": 8859 }, { "epoch": 55.03115264797508, "grad_norm": 2.203125, "learning_rate": 4.1661879099957454e-05, "loss": 0.1659, "step": 8860 }, { "epoch": 55.03738317757009, "grad_norm": 3.109375, "learning_rate": 4.166004221711665e-05, "loss": 0.2858, "step": 8861 }, { "epoch": 55.04361370716511, "grad_norm": 3.515625, "learning_rate": 4.165820517247018e-05, "loss": 0.2806, "step": 8862 }, { "epoch": 55.04984423676012, "grad_norm": 3.03125, "learning_rate": 4.165636796603589e-05, "loss": 0.2246, "step": 8863 }, { "epoch": 55.05607476635514, "grad_norm": 2.3125, "learning_rate": 4.1654530597831634e-05, "loss": 0.2082, "step": 8864 }, { "epoch": 55.06230529595015, "grad_norm": 1.828125, "learning_rate": 4.1652693067875246e-05, "loss": 0.2289, "step": 8865 }, { "epoch": 55.06853582554517, "grad_norm": 1.578125, "learning_rate": 4.165085537618457e-05, "loss": 0.1117, "step": 8866 }, { "epoch": 55.074766355140184, "grad_norm": 2.84375, "learning_rate": 4.164901752277746e-05, "loss": 0.2052, "step": 8867 }, { "epoch": 55.0809968847352, "grad_norm": 3.5625, "learning_rate": 4.1647179507671764e-05, "loss": 0.2303, "step": 8868 }, { "epoch": 55.087227414330215, "grad_norm": 3.109375, "learning_rate": 4.1645341330885335e-05, "loss": 0.2727, "step": 8869 }, { "epoch": 55.09345794392523, "grad_norm": 2.65625, "learning_rate": 4.164350299243601e-05, "loss": 0.1071, "step": 8870 }, { "epoch": 55.099688473520246, "grad_norm": 1.7890625, "learning_rate": 4.164166449234168e-05, "loss": 0.131, "step": 8871 }, { "epoch": 55.10591900311527, "grad_norm": 1.8515625, "learning_rate": 4.163982583062016e-05, "loss": 0.1232, "step": 8872 }, { "epoch": 55.11214953271028, "grad_norm": 2.734375, "learning_rate": 4.1637987007289336e-05, "loss": 0.2321, "step": 8873 }, { "epoch": 55.1183800623053, "grad_norm": 3.40625, "learning_rate": 4.163614802236705e-05, "loss": 0.1302, "step": 8874 }, { "epoch": 55.124610591900314, "grad_norm": 2.046875, "learning_rate": 4.163430887587117e-05, "loss": 0.0974, "step": 8875 }, { "epoch": 55.13084112149533, "grad_norm": 2.359375, "learning_rate": 4.1632469567819555e-05, "loss": 0.1384, "step": 8876 }, { "epoch": 55.137071651090345, "grad_norm": 2.265625, "learning_rate": 4.1630630098230075e-05, "loss": 0.196, "step": 8877 }, { "epoch": 55.14330218068536, "grad_norm": 3.0625, "learning_rate": 4.1628790467120594e-05, "loss": 0.2237, "step": 8878 }, { "epoch": 55.149532710280376, "grad_norm": 2.0625, "learning_rate": 4.162695067450897e-05, "loss": 0.1797, "step": 8879 }, { "epoch": 55.15576323987539, "grad_norm": 2.171875, "learning_rate": 4.162511072041308e-05, "loss": 0.1645, "step": 8880 }, { "epoch": 55.161993769470406, "grad_norm": 2.921875, "learning_rate": 4.162327060485078e-05, "loss": 0.2227, "step": 8881 }, { "epoch": 55.16822429906542, "grad_norm": 1.203125, "learning_rate": 4.162143032783997e-05, "loss": 0.0894, "step": 8882 }, { "epoch": 55.17445482866044, "grad_norm": 2.53125, "learning_rate": 4.16195898893985e-05, "loss": 0.2684, "step": 8883 }, { "epoch": 55.18068535825545, "grad_norm": 1.7421875, "learning_rate": 4.161774928954425e-05, "loss": 0.1406, "step": 8884 }, { "epoch": 55.18691588785047, "grad_norm": 1.5, "learning_rate": 4.16159085282951e-05, "loss": 0.1352, "step": 8885 }, { "epoch": 55.19314641744548, "grad_norm": 3.796875, "learning_rate": 4.161406760566892e-05, "loss": 0.3807, "step": 8886 }, { "epoch": 55.1993769470405, "grad_norm": 2.84375, "learning_rate": 4.161222652168359e-05, "loss": 0.1685, "step": 8887 }, { "epoch": 55.205607476635514, "grad_norm": 2.234375, "learning_rate": 4.161038527635701e-05, "loss": 0.1191, "step": 8888 }, { "epoch": 55.21183800623053, "grad_norm": 2.03125, "learning_rate": 4.160854386970704e-05, "loss": 0.1606, "step": 8889 }, { "epoch": 55.218068535825545, "grad_norm": 2.203125, "learning_rate": 4.1606702301751565e-05, "loss": 0.1656, "step": 8890 }, { "epoch": 55.22429906542056, "grad_norm": 1.140625, "learning_rate": 4.160486057250849e-05, "loss": 0.123, "step": 8891 }, { "epoch": 55.230529595015575, "grad_norm": 3.109375, "learning_rate": 4.1603018681995676e-05, "loss": 0.2764, "step": 8892 }, { "epoch": 55.23676012461059, "grad_norm": 2.8125, "learning_rate": 4.160117663023104e-05, "loss": 0.1475, "step": 8893 }, { "epoch": 55.242990654205606, "grad_norm": 3.078125, "learning_rate": 4.159933441723245e-05, "loss": 0.1534, "step": 8894 }, { "epoch": 55.24922118380062, "grad_norm": 2.625, "learning_rate": 4.15974920430178e-05, "loss": 0.1684, "step": 8895 }, { "epoch": 55.25545171339564, "grad_norm": 2.359375, "learning_rate": 4.1595649507605005e-05, "loss": 0.2739, "step": 8896 }, { "epoch": 55.26168224299065, "grad_norm": 3.109375, "learning_rate": 4.1593806811011934e-05, "loss": 0.1887, "step": 8897 }, { "epoch": 55.26791277258567, "grad_norm": 3.578125, "learning_rate": 4.15919639532565e-05, "loss": 0.2948, "step": 8898 }, { "epoch": 55.27414330218068, "grad_norm": 2.5625, "learning_rate": 4.159012093435659e-05, "loss": 0.2537, "step": 8899 }, { "epoch": 55.2803738317757, "grad_norm": 2.671875, "learning_rate": 4.158827775433012e-05, "loss": 0.2096, "step": 8900 }, { "epoch": 55.28660436137071, "grad_norm": 2.40625, "learning_rate": 4.158643441319496e-05, "loss": 0.2636, "step": 8901 }, { "epoch": 55.29283489096573, "grad_norm": 3.359375, "learning_rate": 4.1584590910969055e-05, "loss": 0.2363, "step": 8902 }, { "epoch": 55.299065420560744, "grad_norm": 3.546875, "learning_rate": 4.158274724767028e-05, "loss": 0.2113, "step": 8903 }, { "epoch": 55.30529595015577, "grad_norm": 2.0, "learning_rate": 4.158090342331654e-05, "loss": 0.1852, "step": 8904 }, { "epoch": 55.31152647975078, "grad_norm": 3.359375, "learning_rate": 4.157905943792576e-05, "loss": 0.1615, "step": 8905 }, { "epoch": 55.3177570093458, "grad_norm": 2.96875, "learning_rate": 4.157721529151584e-05, "loss": 0.2421, "step": 8906 }, { "epoch": 55.32398753894081, "grad_norm": 3.59375, "learning_rate": 4.157537098410469e-05, "loss": 0.3656, "step": 8907 }, { "epoch": 55.33021806853583, "grad_norm": 2.921875, "learning_rate": 4.1573526515710225e-05, "loss": 0.2282, "step": 8908 }, { "epoch": 55.33644859813084, "grad_norm": 1.953125, "learning_rate": 4.157168188635036e-05, "loss": 0.1534, "step": 8909 }, { "epoch": 55.34267912772586, "grad_norm": 1.703125, "learning_rate": 4.156983709604301e-05, "loss": 0.1486, "step": 8910 }, { "epoch": 55.348909657320874, "grad_norm": 1.3203125, "learning_rate": 4.156799214480608e-05, "loss": 0.1285, "step": 8911 }, { "epoch": 55.35514018691589, "grad_norm": 1.796875, "learning_rate": 4.15661470326575e-05, "loss": 0.1385, "step": 8912 }, { "epoch": 55.361370716510905, "grad_norm": 2.09375, "learning_rate": 4.15643017596152e-05, "loss": 0.1587, "step": 8913 }, { "epoch": 55.36760124610592, "grad_norm": 2.765625, "learning_rate": 4.156245632569708e-05, "loss": 0.2294, "step": 8914 }, { "epoch": 55.373831775700936, "grad_norm": 1.7265625, "learning_rate": 4.156061073092107e-05, "loss": 0.1321, "step": 8915 }, { "epoch": 55.38006230529595, "grad_norm": 3.703125, "learning_rate": 4.155876497530511e-05, "loss": 0.2357, "step": 8916 }, { "epoch": 55.386292834890966, "grad_norm": 4.03125, "learning_rate": 4.15569190588671e-05, "loss": 0.3631, "step": 8917 }, { "epoch": 55.39252336448598, "grad_norm": 2.0625, "learning_rate": 4.1555072981625e-05, "loss": 0.1391, "step": 8918 }, { "epoch": 55.398753894081, "grad_norm": 1.203125, "learning_rate": 4.1553226743596704e-05, "loss": 0.1158, "step": 8919 }, { "epoch": 55.40498442367601, "grad_norm": 1.203125, "learning_rate": 4.155138034480017e-05, "loss": 0.1223, "step": 8920 }, { "epoch": 55.41121495327103, "grad_norm": 2.421875, "learning_rate": 4.154953378525333e-05, "loss": 0.1323, "step": 8921 }, { "epoch": 55.41744548286604, "grad_norm": 1.8046875, "learning_rate": 4.1547687064974095e-05, "loss": 0.1245, "step": 8922 }, { "epoch": 55.42367601246106, "grad_norm": 2.015625, "learning_rate": 4.154584018398042e-05, "loss": 0.1111, "step": 8923 }, { "epoch": 55.429906542056074, "grad_norm": 2.953125, "learning_rate": 4.154399314229024e-05, "loss": 0.2069, "step": 8924 }, { "epoch": 55.43613707165109, "grad_norm": 1.65625, "learning_rate": 4.154214593992149e-05, "loss": 0.1496, "step": 8925 }, { "epoch": 55.442367601246104, "grad_norm": 2.671875, "learning_rate": 4.154029857689211e-05, "loss": 0.156, "step": 8926 }, { "epoch": 55.44859813084112, "grad_norm": 2.890625, "learning_rate": 4.1538451053220044e-05, "loss": 0.1215, "step": 8927 }, { "epoch": 55.454828660436135, "grad_norm": 3.484375, "learning_rate": 4.153660336892324e-05, "loss": 0.1338, "step": 8928 }, { "epoch": 55.46105919003115, "grad_norm": 2.40625, "learning_rate": 4.1534755524019634e-05, "loss": 0.1587, "step": 8929 }, { "epoch": 55.467289719626166, "grad_norm": 2.640625, "learning_rate": 4.153290751852717e-05, "loss": 0.2435, "step": 8930 }, { "epoch": 55.47352024922118, "grad_norm": 3.59375, "learning_rate": 4.153105935246383e-05, "loss": 0.3088, "step": 8931 }, { "epoch": 55.4797507788162, "grad_norm": 2.9375, "learning_rate": 4.1529211025847514e-05, "loss": 0.2653, "step": 8932 }, { "epoch": 55.48598130841121, "grad_norm": 3.390625, "learning_rate": 4.1527362538696204e-05, "loss": 0.1787, "step": 8933 }, { "epoch": 55.49221183800623, "grad_norm": 1.3359375, "learning_rate": 4.152551389102784e-05, "loss": 0.1033, "step": 8934 }, { "epoch": 55.49844236760124, "grad_norm": 2.5, "learning_rate": 4.152366508286038e-05, "loss": 0.2727, "step": 8935 }, { "epoch": 55.504672897196265, "grad_norm": 2.578125, "learning_rate": 4.1521816114211785e-05, "loss": 0.2077, "step": 8936 }, { "epoch": 55.51090342679128, "grad_norm": 2.5, "learning_rate": 4.151996698510002e-05, "loss": 0.116, "step": 8937 }, { "epoch": 55.517133956386296, "grad_norm": 3.859375, "learning_rate": 4.151811769554303e-05, "loss": 0.3054, "step": 8938 }, { "epoch": 55.52336448598131, "grad_norm": 2.734375, "learning_rate": 4.151626824555877e-05, "loss": 0.1665, "step": 8939 }, { "epoch": 55.52959501557633, "grad_norm": 1.859375, "learning_rate": 4.1514418635165205e-05, "loss": 0.1386, "step": 8940 }, { "epoch": 55.53582554517134, "grad_norm": 2.765625, "learning_rate": 4.1512568864380316e-05, "loss": 0.1758, "step": 8941 }, { "epoch": 55.54205607476636, "grad_norm": 3.453125, "learning_rate": 4.151071893322206e-05, "loss": 0.1837, "step": 8942 }, { "epoch": 55.54828660436137, "grad_norm": 3.046875, "learning_rate": 4.1508868841708394e-05, "loss": 0.1675, "step": 8943 }, { "epoch": 55.55451713395639, "grad_norm": 3.546875, "learning_rate": 4.1507018589857295e-05, "loss": 0.1737, "step": 8944 }, { "epoch": 55.5607476635514, "grad_norm": 3.0625, "learning_rate": 4.150516817768674e-05, "loss": 0.1503, "step": 8945 }, { "epoch": 55.56697819314642, "grad_norm": 3.9375, "learning_rate": 4.150331760521468e-05, "loss": 0.2535, "step": 8946 }, { "epoch": 55.573208722741434, "grad_norm": 1.8515625, "learning_rate": 4.1501466872459104e-05, "loss": 0.1508, "step": 8947 }, { "epoch": 55.57943925233645, "grad_norm": 3.296875, "learning_rate": 4.1499615979437986e-05, "loss": 0.1669, "step": 8948 }, { "epoch": 55.585669781931465, "grad_norm": 3.0, "learning_rate": 4.14977649261693e-05, "loss": 0.2017, "step": 8949 }, { "epoch": 55.59190031152648, "grad_norm": 1.828125, "learning_rate": 4.149591371267103e-05, "loss": 0.1753, "step": 8950 }, { "epoch": 55.598130841121495, "grad_norm": 1.6875, "learning_rate": 4.1494062338961134e-05, "loss": 0.1449, "step": 8951 }, { "epoch": 55.60436137071651, "grad_norm": 3.875, "learning_rate": 4.1492210805057616e-05, "loss": 0.2181, "step": 8952 }, { "epoch": 55.610591900311526, "grad_norm": 3.34375, "learning_rate": 4.149035911097845e-05, "loss": 0.2795, "step": 8953 }, { "epoch": 55.61682242990654, "grad_norm": 1.90625, "learning_rate": 4.1488507256741616e-05, "loss": 0.1341, "step": 8954 }, { "epoch": 55.62305295950156, "grad_norm": 1.3984375, "learning_rate": 4.14866552423651e-05, "loss": 0.1408, "step": 8955 }, { "epoch": 55.62928348909657, "grad_norm": 1.03125, "learning_rate": 4.14848030678669e-05, "loss": 0.1234, "step": 8956 }, { "epoch": 55.63551401869159, "grad_norm": 2.734375, "learning_rate": 4.1482950733265003e-05, "loss": 0.2512, "step": 8957 }, { "epoch": 55.6417445482866, "grad_norm": 2.234375, "learning_rate": 4.1481098238577387e-05, "loss": 0.1823, "step": 8958 }, { "epoch": 55.64797507788162, "grad_norm": 2.9375, "learning_rate": 4.1479245583822044e-05, "loss": 0.1303, "step": 8959 }, { "epoch": 55.654205607476634, "grad_norm": 2.375, "learning_rate": 4.147739276901698e-05, "loss": 0.1796, "step": 8960 }, { "epoch": 55.66043613707165, "grad_norm": 2.84375, "learning_rate": 4.1475539794180185e-05, "loss": 0.3176, "step": 8961 }, { "epoch": 55.666666666666664, "grad_norm": 2.1875, "learning_rate": 4.1473686659329644e-05, "loss": 0.1226, "step": 8962 }, { "epoch": 55.67289719626168, "grad_norm": 3.484375, "learning_rate": 4.147183336448338e-05, "loss": 0.2383, "step": 8963 }, { "epoch": 55.679127725856695, "grad_norm": 1.7109375, "learning_rate": 4.1469979909659375e-05, "loss": 0.1165, "step": 8964 }, { "epoch": 55.68535825545171, "grad_norm": 1.8125, "learning_rate": 4.146812629487562e-05, "loss": 0.1617, "step": 8965 }, { "epoch": 55.691588785046726, "grad_norm": 3.4375, "learning_rate": 4.146627252015014e-05, "loss": 0.2582, "step": 8966 }, { "epoch": 55.69781931464174, "grad_norm": 2.109375, "learning_rate": 4.146441858550093e-05, "loss": 0.1636, "step": 8967 }, { "epoch": 55.70404984423676, "grad_norm": 3.90625, "learning_rate": 4.1462564490946e-05, "loss": 0.1895, "step": 8968 }, { "epoch": 55.71028037383178, "grad_norm": 2.859375, "learning_rate": 4.146071023650335e-05, "loss": 0.1691, "step": 8969 }, { "epoch": 55.716510903426794, "grad_norm": 2.78125, "learning_rate": 4.145885582219098e-05, "loss": 0.2067, "step": 8970 }, { "epoch": 55.72274143302181, "grad_norm": 3.09375, "learning_rate": 4.145700124802693e-05, "loss": 0.2836, "step": 8971 }, { "epoch": 55.728971962616825, "grad_norm": 2.828125, "learning_rate": 4.145514651402918e-05, "loss": 0.2284, "step": 8972 }, { "epoch": 55.73520249221184, "grad_norm": 1.7578125, "learning_rate": 4.1453291620215766e-05, "loss": 0.1382, "step": 8973 }, { "epoch": 55.741433021806856, "grad_norm": 1.796875, "learning_rate": 4.14514365666047e-05, "loss": 0.1036, "step": 8974 }, { "epoch": 55.74766355140187, "grad_norm": 3.109375, "learning_rate": 4.144958135321397e-05, "loss": 0.209, "step": 8975 }, { "epoch": 55.753894080996886, "grad_norm": 2.125, "learning_rate": 4.144772598006165e-05, "loss": 0.1595, "step": 8976 }, { "epoch": 55.7601246105919, "grad_norm": 1.078125, "learning_rate": 4.14458704471657e-05, "loss": 0.0959, "step": 8977 }, { "epoch": 55.76635514018692, "grad_norm": 2.5625, "learning_rate": 4.144401475454418e-05, "loss": 0.2285, "step": 8978 }, { "epoch": 55.77258566978193, "grad_norm": 2.5, "learning_rate": 4.1442158902215104e-05, "loss": 0.2011, "step": 8979 }, { "epoch": 55.77881619937695, "grad_norm": 1.5546875, "learning_rate": 4.1440302890196495e-05, "loss": 0.1137, "step": 8980 }, { "epoch": 55.78504672897196, "grad_norm": 2.34375, "learning_rate": 4.1438446718506376e-05, "loss": 0.18, "step": 8981 }, { "epoch": 55.79127725856698, "grad_norm": 3.296875, "learning_rate": 4.143659038716276e-05, "loss": 0.3299, "step": 8982 }, { "epoch": 55.797507788161994, "grad_norm": 3.78125, "learning_rate": 4.1434733896183716e-05, "loss": 0.3089, "step": 8983 }, { "epoch": 55.80373831775701, "grad_norm": 2.640625, "learning_rate": 4.143287724558724e-05, "loss": 0.2075, "step": 8984 }, { "epoch": 55.809968847352025, "grad_norm": 2.328125, "learning_rate": 4.143102043539138e-05, "loss": 0.1637, "step": 8985 }, { "epoch": 55.81619937694704, "grad_norm": 2.625, "learning_rate": 4.1429163465614165e-05, "loss": 0.2162, "step": 8986 }, { "epoch": 55.822429906542055, "grad_norm": 2.171875, "learning_rate": 4.1427306336273624e-05, "loss": 0.1373, "step": 8987 }, { "epoch": 55.82866043613707, "grad_norm": 1.5703125, "learning_rate": 4.142544904738781e-05, "loss": 0.1372, "step": 8988 }, { "epoch": 55.834890965732086, "grad_norm": 3.078125, "learning_rate": 4.1423591598974744e-05, "loss": 0.2844, "step": 8989 }, { "epoch": 55.8411214953271, "grad_norm": 2.484375, "learning_rate": 4.142173399105247e-05, "loss": 0.1699, "step": 8990 }, { "epoch": 55.84735202492212, "grad_norm": 3.359375, "learning_rate": 4.141987622363904e-05, "loss": 0.2119, "step": 8991 }, { "epoch": 55.85358255451713, "grad_norm": 2.4375, "learning_rate": 4.141801829675249e-05, "loss": 0.1911, "step": 8992 }, { "epoch": 55.85981308411215, "grad_norm": 2.015625, "learning_rate": 4.141616021041086e-05, "loss": 0.1376, "step": 8993 }, { "epoch": 55.86604361370716, "grad_norm": 2.546875, "learning_rate": 4.141430196463221e-05, "loss": 0.19, "step": 8994 }, { "epoch": 55.87227414330218, "grad_norm": 2.140625, "learning_rate": 4.1412443559434566e-05, "loss": 0.1702, "step": 8995 }, { "epoch": 55.87850467289719, "grad_norm": 2.4375, "learning_rate": 4.141058499483599e-05, "loss": 0.1842, "step": 8996 }, { "epoch": 55.88473520249221, "grad_norm": 1.3359375, "learning_rate": 4.140872627085454e-05, "loss": 0.1139, "step": 8997 }, { "epoch": 55.890965732087224, "grad_norm": 2.390625, "learning_rate": 4.140686738750826e-05, "loss": 0.1675, "step": 8998 }, { "epoch": 55.89719626168224, "grad_norm": 3.609375, "learning_rate": 4.1405008344815196e-05, "loss": 0.4247, "step": 8999 }, { "epoch": 55.90342679127726, "grad_norm": 3.5625, "learning_rate": 4.140314914279342e-05, "loss": 0.1887, "step": 9000 }, { "epoch": 55.90965732087228, "grad_norm": 1.796875, "learning_rate": 4.140128978146098e-05, "loss": 0.116, "step": 9001 }, { "epoch": 55.91588785046729, "grad_norm": 1.6328125, "learning_rate": 4.1399430260835926e-05, "loss": 0.1331, "step": 9002 }, { "epoch": 55.92211838006231, "grad_norm": 2.546875, "learning_rate": 4.139757058093634e-05, "loss": 0.163, "step": 9003 }, { "epoch": 55.92834890965732, "grad_norm": 5.15625, "learning_rate": 4.1395710741780255e-05, "loss": 0.2915, "step": 9004 }, { "epoch": 55.93457943925234, "grad_norm": 2.453125, "learning_rate": 4.139385074338577e-05, "loss": 0.1492, "step": 9005 }, { "epoch": 55.940809968847354, "grad_norm": 3.359375, "learning_rate": 4.139199058577091e-05, "loss": 0.2231, "step": 9006 }, { "epoch": 55.94704049844237, "grad_norm": 3.3125, "learning_rate": 4.139013026895377e-05, "loss": 0.1756, "step": 9007 }, { "epoch": 55.953271028037385, "grad_norm": 3.15625, "learning_rate": 4.1388269792952414e-05, "loss": 0.215, "step": 9008 }, { "epoch": 55.9595015576324, "grad_norm": 2.484375, "learning_rate": 4.1386409157784896e-05, "loss": 0.1334, "step": 9009 }, { "epoch": 55.965732087227416, "grad_norm": 3.0, "learning_rate": 4.13845483634693e-05, "loss": 0.1755, "step": 9010 }, { "epoch": 55.97196261682243, "grad_norm": 2.515625, "learning_rate": 4.13826874100237e-05, "loss": 0.1644, "step": 9011 }, { "epoch": 55.978193146417446, "grad_norm": 1.828125, "learning_rate": 4.138082629746616e-05, "loss": 0.1209, "step": 9012 }, { "epoch": 55.98442367601246, "grad_norm": 2.171875, "learning_rate": 4.1378965025814754e-05, "loss": 0.1592, "step": 9013 }, { "epoch": 55.99065420560748, "grad_norm": 1.953125, "learning_rate": 4.137710359508757e-05, "loss": 0.113, "step": 9014 }, { "epoch": 55.99688473520249, "grad_norm": 3.046875, "learning_rate": 4.137524200530269e-05, "loss": 0.301, "step": 9015 }, { "epoch": 56.0, "grad_norm": 2.359375, "learning_rate": 4.137338025647818e-05, "loss": 0.1291, "step": 9016 }, { "epoch": 56.006230529595015, "grad_norm": 3.578125, "learning_rate": 4.137151834863213e-05, "loss": 0.2271, "step": 9017 }, { "epoch": 56.01246105919003, "grad_norm": 2.78125, "learning_rate": 4.136965628178262e-05, "loss": 0.2141, "step": 9018 }, { "epoch": 56.018691588785046, "grad_norm": 2.125, "learning_rate": 4.1367794055947737e-05, "loss": 0.1275, "step": 9019 }, { "epoch": 56.02492211838006, "grad_norm": 1.703125, "learning_rate": 4.136593167114556e-05, "loss": 0.1189, "step": 9020 }, { "epoch": 56.03115264797508, "grad_norm": 1.921875, "learning_rate": 4.13640691273942e-05, "loss": 0.2038, "step": 9021 }, { "epoch": 56.03738317757009, "grad_norm": 2.875, "learning_rate": 4.136220642471172e-05, "loss": 0.151, "step": 9022 }, { "epoch": 56.04361370716511, "grad_norm": 2.609375, "learning_rate": 4.136034356311621e-05, "loss": 0.1543, "step": 9023 }, { "epoch": 56.04984423676012, "grad_norm": 2.75, "learning_rate": 4.1358480542625785e-05, "loss": 0.2863, "step": 9024 }, { "epoch": 56.05607476635514, "grad_norm": 2.421875, "learning_rate": 4.135661736325852e-05, "loss": 0.2536, "step": 9025 }, { "epoch": 56.06230529595015, "grad_norm": 2.203125, "learning_rate": 4.1354754025032514e-05, "loss": 0.1689, "step": 9026 }, { "epoch": 56.06853582554517, "grad_norm": 4.5, "learning_rate": 4.135289052796588e-05, "loss": 0.1242, "step": 9027 }, { "epoch": 56.074766355140184, "grad_norm": 1.375, "learning_rate": 4.13510268720767e-05, "loss": 0.1236, "step": 9028 }, { "epoch": 56.0809968847352, "grad_norm": 2.5, "learning_rate": 4.134916305738308e-05, "loss": 0.2198, "step": 9029 }, { "epoch": 56.087227414330215, "grad_norm": 5.78125, "learning_rate": 4.134729908390311e-05, "loss": 0.2012, "step": 9030 }, { "epoch": 56.09345794392523, "grad_norm": 2.5, "learning_rate": 4.134543495165492e-05, "loss": 0.1387, "step": 9031 }, { "epoch": 56.099688473520246, "grad_norm": 3.484375, "learning_rate": 4.134357066065658e-05, "loss": 0.4548, "step": 9032 }, { "epoch": 56.10591900311527, "grad_norm": 3.1875, "learning_rate": 4.1341706210926234e-05, "loss": 0.1797, "step": 9033 }, { "epoch": 56.11214953271028, "grad_norm": 3.015625, "learning_rate": 4.133984160248196e-05, "loss": 0.2699, "step": 9034 }, { "epoch": 56.1183800623053, "grad_norm": 4.71875, "learning_rate": 4.133797683534188e-05, "loss": 0.1728, "step": 9035 }, { "epoch": 56.124610591900314, "grad_norm": 3.59375, "learning_rate": 4.13361119095241e-05, "loss": 0.2772, "step": 9036 }, { "epoch": 56.13084112149533, "grad_norm": 3.5625, "learning_rate": 4.1334246825046744e-05, "loss": 0.1426, "step": 9037 }, { "epoch": 56.137071651090345, "grad_norm": 3.53125, "learning_rate": 4.1332381581927903e-05, "loss": 0.1586, "step": 9038 }, { "epoch": 56.14330218068536, "grad_norm": 2.40625, "learning_rate": 4.133051618018572e-05, "loss": 0.1876, "step": 9039 }, { "epoch": 56.149532710280376, "grad_norm": 2.859375, "learning_rate": 4.132865061983829e-05, "loss": 0.3908, "step": 9040 }, { "epoch": 56.15576323987539, "grad_norm": 2.890625, "learning_rate": 4.132678490090375e-05, "loss": 0.2149, "step": 9041 }, { "epoch": 56.161993769470406, "grad_norm": 3.703125, "learning_rate": 4.132491902340021e-05, "loss": 0.191, "step": 9042 }, { "epoch": 56.16822429906542, "grad_norm": 3.421875, "learning_rate": 4.132305298734579e-05, "loss": 0.116, "step": 9043 }, { "epoch": 56.17445482866044, "grad_norm": 3.5, "learning_rate": 4.1321186792758615e-05, "loss": 0.2899, "step": 9044 }, { "epoch": 56.18068535825545, "grad_norm": 3.296875, "learning_rate": 4.131932043965681e-05, "loss": 0.2191, "step": 9045 }, { "epoch": 56.18691588785047, "grad_norm": 1.421875, "learning_rate": 4.1317453928058517e-05, "loss": 0.139, "step": 9046 }, { "epoch": 56.19314641744548, "grad_norm": 2.5625, "learning_rate": 4.1315587257981833e-05, "loss": 0.1898, "step": 9047 }, { "epoch": 56.1993769470405, "grad_norm": 3.453125, "learning_rate": 4.131372042944491e-05, "loss": 0.2105, "step": 9048 }, { "epoch": 56.205607476635514, "grad_norm": 3.84375, "learning_rate": 4.131185344246587e-05, "loss": 0.2855, "step": 9049 }, { "epoch": 56.21183800623053, "grad_norm": 2.609375, "learning_rate": 4.1309986297062855e-05, "loss": 0.1986, "step": 9050 }, { "epoch": 56.218068535825545, "grad_norm": 2.21875, "learning_rate": 4.130811899325398e-05, "loss": 0.1945, "step": 9051 }, { "epoch": 56.22429906542056, "grad_norm": 1.9765625, "learning_rate": 4.130625153105742e-05, "loss": 0.1184, "step": 9052 }, { "epoch": 56.230529595015575, "grad_norm": 2.359375, "learning_rate": 4.1304383910491264e-05, "loss": 0.1896, "step": 9053 }, { "epoch": 56.23676012461059, "grad_norm": 2.328125, "learning_rate": 4.130251613157368e-05, "loss": 0.1468, "step": 9054 }, { "epoch": 56.242990654205606, "grad_norm": 3.25, "learning_rate": 4.1300648194322796e-05, "loss": 0.248, "step": 9055 }, { "epoch": 56.24922118380062, "grad_norm": 2.46875, "learning_rate": 4.1298780098756754e-05, "loss": 0.1333, "step": 9056 }, { "epoch": 56.25545171339564, "grad_norm": 2.703125, "learning_rate": 4.129691184489371e-05, "loss": 0.152, "step": 9057 }, { "epoch": 56.26168224299065, "grad_norm": 2.953125, "learning_rate": 4.1295043432751804e-05, "loss": 0.2285, "step": 9058 }, { "epoch": 56.26791277258567, "grad_norm": 2.171875, "learning_rate": 4.129317486234917e-05, "loss": 0.2014, "step": 9059 }, { "epoch": 56.27414330218068, "grad_norm": 2.796875, "learning_rate": 4.1291306133703966e-05, "loss": 0.2387, "step": 9060 }, { "epoch": 56.2803738317757, "grad_norm": 2.4375, "learning_rate": 4.128943724683434e-05, "loss": 0.1589, "step": 9061 }, { "epoch": 56.28660436137071, "grad_norm": 1.9375, "learning_rate": 4.128756820175845e-05, "loss": 0.1199, "step": 9062 }, { "epoch": 56.29283489096573, "grad_norm": 2.90625, "learning_rate": 4.128569899849443e-05, "loss": 0.1968, "step": 9063 }, { "epoch": 56.299065420560744, "grad_norm": 1.53125, "learning_rate": 4.128382963706046e-05, "loss": 0.1403, "step": 9064 }, { "epoch": 56.30529595015577, "grad_norm": 2.40625, "learning_rate": 4.1281960117474674e-05, "loss": 0.2063, "step": 9065 }, { "epoch": 56.31152647975078, "grad_norm": 3.171875, "learning_rate": 4.128009043975523e-05, "loss": 0.1885, "step": 9066 }, { "epoch": 56.3177570093458, "grad_norm": 3.890625, "learning_rate": 4.1278220603920304e-05, "loss": 0.1276, "step": 9067 }, { "epoch": 56.32398753894081, "grad_norm": 2.515625, "learning_rate": 4.127635060998804e-05, "loss": 0.1995, "step": 9068 }, { "epoch": 56.33021806853583, "grad_norm": 2.640625, "learning_rate": 4.1274480457976604e-05, "loss": 0.242, "step": 9069 }, { "epoch": 56.33644859813084, "grad_norm": 3.53125, "learning_rate": 4.1272610147904166e-05, "loss": 0.3044, "step": 9070 }, { "epoch": 56.34267912772586, "grad_norm": 4.21875, "learning_rate": 4.127073967978889e-05, "loss": 0.1469, "step": 9071 }, { "epoch": 56.348909657320874, "grad_norm": 1.7890625, "learning_rate": 4.126886905364893e-05, "loss": 0.1571, "step": 9072 }, { "epoch": 56.35514018691589, "grad_norm": 2.59375, "learning_rate": 4.1266998269502456e-05, "loss": 0.129, "step": 9073 }, { "epoch": 56.361370716510905, "grad_norm": 1.9921875, "learning_rate": 4.1265127327367656e-05, "loss": 0.1849, "step": 9074 }, { "epoch": 56.36760124610592, "grad_norm": 3.328125, "learning_rate": 4.126325622726268e-05, "loss": 0.308, "step": 9075 }, { "epoch": 56.373831775700936, "grad_norm": 2.796875, "learning_rate": 4.126138496920571e-05, "loss": 0.1818, "step": 9076 }, { "epoch": 56.38006230529595, "grad_norm": 4.03125, "learning_rate": 4.125951355321492e-05, "loss": 0.2258, "step": 9077 }, { "epoch": 56.386292834890966, "grad_norm": 2.484375, "learning_rate": 4.125764197930848e-05, "loss": 0.2369, "step": 9078 }, { "epoch": 56.39252336448598, "grad_norm": 1.796875, "learning_rate": 4.1255770247504576e-05, "loss": 0.1262, "step": 9079 }, { "epoch": 56.398753894081, "grad_norm": 1.9765625, "learning_rate": 4.125389835782138e-05, "loss": 0.1862, "step": 9080 }, { "epoch": 56.40498442367601, "grad_norm": 4.15625, "learning_rate": 4.1252026310277075e-05, "loss": 0.1993, "step": 9081 }, { "epoch": 56.41121495327103, "grad_norm": 3.390625, "learning_rate": 4.125015410488984e-05, "loss": 0.161, "step": 9082 }, { "epoch": 56.41744548286604, "grad_norm": 3.828125, "learning_rate": 4.124828174167786e-05, "loss": 0.1826, "step": 9083 }, { "epoch": 56.42367601246106, "grad_norm": 2.25, "learning_rate": 4.124640922065932e-05, "loss": 0.1582, "step": 9084 }, { "epoch": 56.429906542056074, "grad_norm": 3.015625, "learning_rate": 4.124453654185241e-05, "loss": 0.2057, "step": 9085 }, { "epoch": 56.43613707165109, "grad_norm": 3.375, "learning_rate": 4.124266370527531e-05, "loss": 0.2503, "step": 9086 }, { "epoch": 56.442367601246104, "grad_norm": 1.6484375, "learning_rate": 4.1240790710946214e-05, "loss": 0.1044, "step": 9087 }, { "epoch": 56.44859813084112, "grad_norm": 3.265625, "learning_rate": 4.123891755888331e-05, "loss": 0.2194, "step": 9088 }, { "epoch": 56.454828660436135, "grad_norm": 2.171875, "learning_rate": 4.123704424910479e-05, "loss": 0.1283, "step": 9089 }, { "epoch": 56.46105919003115, "grad_norm": 2.421875, "learning_rate": 4.1235170781628864e-05, "loss": 0.2612, "step": 9090 }, { "epoch": 56.467289719626166, "grad_norm": 1.3828125, "learning_rate": 4.1233297156473706e-05, "loss": 0.1339, "step": 9091 }, { "epoch": 56.47352024922118, "grad_norm": 1.3359375, "learning_rate": 4.123142337365752e-05, "loss": 0.1, "step": 9092 }, { "epoch": 56.4797507788162, "grad_norm": 2.640625, "learning_rate": 4.1229549433198504e-05, "loss": 0.3207, "step": 9093 }, { "epoch": 56.48598130841121, "grad_norm": 1.1875, "learning_rate": 4.122767533511486e-05, "loss": 0.1208, "step": 9094 }, { "epoch": 56.49221183800623, "grad_norm": 2.828125, "learning_rate": 4.122580107942479e-05, "loss": 0.1722, "step": 9095 }, { "epoch": 56.49844236760124, "grad_norm": 1.6484375, "learning_rate": 4.1223926666146494e-05, "loss": 0.1293, "step": 9096 }, { "epoch": 56.504672897196265, "grad_norm": 2.421875, "learning_rate": 4.1222052095298195e-05, "loss": 0.2283, "step": 9097 }, { "epoch": 56.51090342679128, "grad_norm": 1.7421875, "learning_rate": 4.122017736689806e-05, "loss": 0.1314, "step": 9098 }, { "epoch": 56.517133956386296, "grad_norm": 2.109375, "learning_rate": 4.121830248096433e-05, "loss": 0.175, "step": 9099 }, { "epoch": 56.52336448598131, "grad_norm": 1.9140625, "learning_rate": 4.121642743751521e-05, "loss": 0.1479, "step": 9100 }, { "epoch": 56.52959501557633, "grad_norm": 2.5, "learning_rate": 4.12145522365689e-05, "loss": 0.1453, "step": 9101 }, { "epoch": 56.53582554517134, "grad_norm": 2.765625, "learning_rate": 4.121267687814362e-05, "loss": 0.2188, "step": 9102 }, { "epoch": 56.54205607476636, "grad_norm": 3.125, "learning_rate": 4.121080136225758e-05, "loss": 0.2435, "step": 9103 }, { "epoch": 56.54828660436137, "grad_norm": 3.15625, "learning_rate": 4.1208925688928996e-05, "loss": 0.2926, "step": 9104 }, { "epoch": 56.55451713395639, "grad_norm": 2.234375, "learning_rate": 4.120704985817608e-05, "loss": 0.1511, "step": 9105 }, { "epoch": 56.5607476635514, "grad_norm": 2.453125, "learning_rate": 4.120517387001707e-05, "loss": 0.1755, "step": 9106 }, { "epoch": 56.56697819314642, "grad_norm": 2.640625, "learning_rate": 4.120329772447016e-05, "loss": 0.2048, "step": 9107 }, { "epoch": 56.573208722741434, "grad_norm": 2.53125, "learning_rate": 4.1201421421553586e-05, "loss": 0.1247, "step": 9108 }, { "epoch": 56.57943925233645, "grad_norm": 2.75, "learning_rate": 4.1199544961285574e-05, "loss": 0.2076, "step": 9109 }, { "epoch": 56.585669781931465, "grad_norm": 3.953125, "learning_rate": 4.1197668343684336e-05, "loss": 0.3129, "step": 9110 }, { "epoch": 56.59190031152648, "grad_norm": 2.75, "learning_rate": 4.119579156876811e-05, "loss": 0.1504, "step": 9111 }, { "epoch": 56.598130841121495, "grad_norm": 2.40625, "learning_rate": 4.119391463655511e-05, "loss": 0.1331, "step": 9112 }, { "epoch": 56.60436137071651, "grad_norm": 3.09375, "learning_rate": 4.1192037547063584e-05, "loss": 0.1592, "step": 9113 }, { "epoch": 56.610591900311526, "grad_norm": 3.046875, "learning_rate": 4.1190160300311745e-05, "loss": 0.1378, "step": 9114 }, { "epoch": 56.61682242990654, "grad_norm": 3.21875, "learning_rate": 4.118828289631784e-05, "loss": 0.1612, "step": 9115 }, { "epoch": 56.62305295950156, "grad_norm": 1.140625, "learning_rate": 4.1186405335100094e-05, "loss": 0.1125, "step": 9116 }, { "epoch": 56.62928348909657, "grad_norm": 2.65625, "learning_rate": 4.118452761667674e-05, "loss": 0.1911, "step": 9117 }, { "epoch": 56.63551401869159, "grad_norm": 5.0, "learning_rate": 4.1182649741066024e-05, "loss": 0.177, "step": 9118 }, { "epoch": 56.6417445482866, "grad_norm": 4.15625, "learning_rate": 4.118077170828618e-05, "loss": 0.1295, "step": 9119 }, { "epoch": 56.64797507788162, "grad_norm": 3.921875, "learning_rate": 4.117889351835544e-05, "loss": 0.239, "step": 9120 }, { "epoch": 56.654205607476634, "grad_norm": 3.828125, "learning_rate": 4.117701517129205e-05, "loss": 0.2077, "step": 9121 }, { "epoch": 56.66043613707165, "grad_norm": 2.296875, "learning_rate": 4.117513666711426e-05, "loss": 0.2325, "step": 9122 }, { "epoch": 56.666666666666664, "grad_norm": 3.171875, "learning_rate": 4.117325800584031e-05, "loss": 0.1738, "step": 9123 }, { "epoch": 56.67289719626168, "grad_norm": 3.25, "learning_rate": 4.1171379187488445e-05, "loss": 0.1784, "step": 9124 }, { "epoch": 56.679127725856695, "grad_norm": 4.15625, "learning_rate": 4.116950021207691e-05, "loss": 0.1159, "step": 9125 }, { "epoch": 56.68535825545171, "grad_norm": 2.953125, "learning_rate": 4.1167621079623956e-05, "loss": 0.1503, "step": 9126 }, { "epoch": 56.691588785046726, "grad_norm": 2.765625, "learning_rate": 4.116574179014784e-05, "loss": 0.1741, "step": 9127 }, { "epoch": 56.69781931464174, "grad_norm": 2.90625, "learning_rate": 4.11638623436668e-05, "loss": 0.1272, "step": 9128 }, { "epoch": 56.70404984423676, "grad_norm": 3.296875, "learning_rate": 4.1161982740199115e-05, "loss": 0.2956, "step": 9129 }, { "epoch": 56.71028037383178, "grad_norm": 2.765625, "learning_rate": 4.1160102979763e-05, "loss": 0.1155, "step": 9130 }, { "epoch": 56.716510903426794, "grad_norm": 3.640625, "learning_rate": 4.115822306237675e-05, "loss": 0.1738, "step": 9131 }, { "epoch": 56.72274143302181, "grad_norm": 1.6171875, "learning_rate": 4.1156342988058606e-05, "loss": 0.1346, "step": 9132 }, { "epoch": 56.728971962616825, "grad_norm": 2.6875, "learning_rate": 4.115446275682682e-05, "loss": 0.2035, "step": 9133 }, { "epoch": 56.73520249221184, "grad_norm": 2.0, "learning_rate": 4.115258236869967e-05, "loss": 0.1101, "step": 9134 }, { "epoch": 56.741433021806856, "grad_norm": 1.59375, "learning_rate": 4.1150701823695416e-05, "loss": 0.1272, "step": 9135 }, { "epoch": 56.74766355140187, "grad_norm": 1.5078125, "learning_rate": 4.1148821121832305e-05, "loss": 0.1321, "step": 9136 }, { "epoch": 56.753894080996886, "grad_norm": 1.59375, "learning_rate": 4.114694026312862e-05, "loss": 0.1341, "step": 9137 }, { "epoch": 56.7601246105919, "grad_norm": 2.953125, "learning_rate": 4.114505924760263e-05, "loss": 0.1786, "step": 9138 }, { "epoch": 56.76635514018692, "grad_norm": 2.46875, "learning_rate": 4.1143178075272595e-05, "loss": 0.2018, "step": 9139 }, { "epoch": 56.77258566978193, "grad_norm": 2.140625, "learning_rate": 4.114129674615678e-05, "loss": 0.1857, "step": 9140 }, { "epoch": 56.77881619937695, "grad_norm": 1.8359375, "learning_rate": 4.1139415260273474e-05, "loss": 0.1801, "step": 9141 }, { "epoch": 56.78504672897196, "grad_norm": 2.328125, "learning_rate": 4.1137533617640935e-05, "loss": 0.1528, "step": 9142 }, { "epoch": 56.79127725856698, "grad_norm": 2.40625, "learning_rate": 4.1135651818277445e-05, "loss": 0.1045, "step": 9143 }, { "epoch": 56.797507788161994, "grad_norm": 1.546875, "learning_rate": 4.113376986220128e-05, "loss": 0.1537, "step": 9144 }, { "epoch": 56.80373831775701, "grad_norm": 3.609375, "learning_rate": 4.113188774943071e-05, "loss": 0.2193, "step": 9145 }, { "epoch": 56.809968847352025, "grad_norm": 3.328125, "learning_rate": 4.113000547998403e-05, "loss": 0.1753, "step": 9146 }, { "epoch": 56.81619937694704, "grad_norm": 1.9375, "learning_rate": 4.112812305387951e-05, "loss": 0.105, "step": 9147 }, { "epoch": 56.822429906542055, "grad_norm": 3.328125, "learning_rate": 4.1126240471135435e-05, "loss": 0.2367, "step": 9148 }, { "epoch": 56.82866043613707, "grad_norm": 2.734375, "learning_rate": 4.112435773177009e-05, "loss": 0.2913, "step": 9149 }, { "epoch": 56.834890965732086, "grad_norm": 1.96875, "learning_rate": 4.112247483580176e-05, "loss": 0.1295, "step": 9150 }, { "epoch": 56.8411214953271, "grad_norm": 2.359375, "learning_rate": 4.112059178324873e-05, "loss": 0.1805, "step": 9151 }, { "epoch": 56.84735202492212, "grad_norm": 2.671875, "learning_rate": 4.111870857412929e-05, "loss": 0.1361, "step": 9152 }, { "epoch": 56.85358255451713, "grad_norm": 2.953125, "learning_rate": 4.111682520846173e-05, "loss": 0.2564, "step": 9153 }, { "epoch": 56.85981308411215, "grad_norm": 2.140625, "learning_rate": 4.111494168626434e-05, "loss": 0.215, "step": 9154 }, { "epoch": 56.86604361370716, "grad_norm": 2.671875, "learning_rate": 4.1113058007555415e-05, "loss": 0.2176, "step": 9155 }, { "epoch": 56.87227414330218, "grad_norm": 2.546875, "learning_rate": 4.111117417235325e-05, "loss": 0.1069, "step": 9156 }, { "epoch": 56.87850467289719, "grad_norm": 1.359375, "learning_rate": 4.110929018067615e-05, "loss": 0.1199, "step": 9157 }, { "epoch": 56.88473520249221, "grad_norm": 2.578125, "learning_rate": 4.11074060325424e-05, "loss": 0.1833, "step": 9158 }, { "epoch": 56.890965732087224, "grad_norm": 2.609375, "learning_rate": 4.110552172797029e-05, "loss": 0.2574, "step": 9159 }, { "epoch": 56.89719626168224, "grad_norm": 3.21875, "learning_rate": 4.110363726697814e-05, "loss": 0.2848, "step": 9160 }, { "epoch": 56.90342679127726, "grad_norm": 2.25, "learning_rate": 4.110175264958425e-05, "loss": 0.1211, "step": 9161 }, { "epoch": 56.90965732087228, "grad_norm": 2.203125, "learning_rate": 4.109986787580692e-05, "loss": 0.1801, "step": 9162 }, { "epoch": 56.91588785046729, "grad_norm": 2.421875, "learning_rate": 4.109798294566446e-05, "loss": 0.2119, "step": 9163 }, { "epoch": 56.92211838006231, "grad_norm": 1.296875, "learning_rate": 4.109609785917516e-05, "loss": 0.1245, "step": 9164 }, { "epoch": 56.92834890965732, "grad_norm": 2.28125, "learning_rate": 4.1094212616357344e-05, "loss": 0.1745, "step": 9165 }, { "epoch": 56.93457943925234, "grad_norm": 2.140625, "learning_rate": 4.109232721722932e-05, "loss": 0.0928, "step": 9166 }, { "epoch": 56.940809968847354, "grad_norm": 2.59375, "learning_rate": 4.10904416618094e-05, "loss": 0.1656, "step": 9167 }, { "epoch": 56.94704049844237, "grad_norm": 2.5625, "learning_rate": 4.108855595011588e-05, "loss": 0.1749, "step": 9168 }, { "epoch": 56.953271028037385, "grad_norm": 3.34375, "learning_rate": 4.10866700821671e-05, "loss": 0.2677, "step": 9169 }, { "epoch": 56.9595015576324, "grad_norm": 2.828125, "learning_rate": 4.108478405798137e-05, "loss": 0.1935, "step": 9170 }, { "epoch": 56.965732087227416, "grad_norm": 3.0625, "learning_rate": 4.1082897877576994e-05, "loss": 0.1533, "step": 9171 }, { "epoch": 56.97196261682243, "grad_norm": 2.5, "learning_rate": 4.1081011540972295e-05, "loss": 0.1859, "step": 9172 }, { "epoch": 56.978193146417446, "grad_norm": 3.0625, "learning_rate": 4.107912504818561e-05, "loss": 0.2041, "step": 9173 }, { "epoch": 56.98442367601246, "grad_norm": 1.859375, "learning_rate": 4.1077238399235235e-05, "loss": 0.1258, "step": 9174 }, { "epoch": 56.99065420560748, "grad_norm": 3.96875, "learning_rate": 4.107535159413951e-05, "loss": 0.2036, "step": 9175 }, { "epoch": 56.99688473520249, "grad_norm": 2.65625, "learning_rate": 4.107346463291676e-05, "loss": 0.1683, "step": 9176 }, { "epoch": 57.0, "grad_norm": 1.7578125, "learning_rate": 4.1071577515585315e-05, "loss": 0.061, "step": 9177 }, { "epoch": 57.006230529595015, "grad_norm": 2.3125, "learning_rate": 4.1069690242163484e-05, "loss": 0.1406, "step": 9178 }, { "epoch": 57.01246105919003, "grad_norm": 4.0, "learning_rate": 4.106780281266962e-05, "loss": 0.2068, "step": 9179 }, { "epoch": 57.018691588785046, "grad_norm": 3.515625, "learning_rate": 4.106591522712203e-05, "loss": 0.1584, "step": 9180 }, { "epoch": 57.02492211838006, "grad_norm": 4.3125, "learning_rate": 4.106402748553907e-05, "loss": 0.1698, "step": 9181 }, { "epoch": 57.03115264797508, "grad_norm": 3.09375, "learning_rate": 4.106213958793906e-05, "loss": 0.2692, "step": 9182 }, { "epoch": 57.03738317757009, "grad_norm": 2.53125, "learning_rate": 4.106025153434034e-05, "loss": 0.1877, "step": 9183 }, { "epoch": 57.04361370716511, "grad_norm": 2.578125, "learning_rate": 4.105836332476125e-05, "loss": 0.1221, "step": 9184 }, { "epoch": 57.04984423676012, "grad_norm": 2.890625, "learning_rate": 4.105647495922013e-05, "loss": 0.1559, "step": 9185 }, { "epoch": 57.05607476635514, "grad_norm": 2.640625, "learning_rate": 4.105458643773531e-05, "loss": 0.216, "step": 9186 }, { "epoch": 57.06230529595015, "grad_norm": 3.03125, "learning_rate": 4.105269776032513e-05, "loss": 0.2838, "step": 9187 }, { "epoch": 57.06853582554517, "grad_norm": 2.828125, "learning_rate": 4.105080892700795e-05, "loss": 0.1419, "step": 9188 }, { "epoch": 57.074766355140184, "grad_norm": 2.734375, "learning_rate": 4.10489199378021e-05, "loss": 0.2467, "step": 9189 }, { "epoch": 57.0809968847352, "grad_norm": 2.609375, "learning_rate": 4.1047030792725935e-05, "loss": 0.2198, "step": 9190 }, { "epoch": 57.087227414330215, "grad_norm": 2.765625, "learning_rate": 4.1045141491797795e-05, "loss": 0.2128, "step": 9191 }, { "epoch": 57.09345794392523, "grad_norm": 2.296875, "learning_rate": 4.1043252035036036e-05, "loss": 0.159, "step": 9192 }, { "epoch": 57.099688473520246, "grad_norm": 2.21875, "learning_rate": 4.104136242245901e-05, "loss": 0.1498, "step": 9193 }, { "epoch": 57.10591900311527, "grad_norm": 1.1796875, "learning_rate": 4.103947265408505e-05, "loss": 0.1283, "step": 9194 }, { "epoch": 57.11214953271028, "grad_norm": 2.75, "learning_rate": 4.1037582729932545e-05, "loss": 0.3224, "step": 9195 }, { "epoch": 57.1183800623053, "grad_norm": 1.640625, "learning_rate": 4.103569265001982e-05, "loss": 0.1545, "step": 9196 }, { "epoch": 57.124610591900314, "grad_norm": 2.703125, "learning_rate": 4.1033802414365244e-05, "loss": 0.1852, "step": 9197 }, { "epoch": 57.13084112149533, "grad_norm": 2.15625, "learning_rate": 4.103191202298718e-05, "loss": 0.1558, "step": 9198 }, { "epoch": 57.137071651090345, "grad_norm": 1.9375, "learning_rate": 4.103002147590398e-05, "loss": 0.1712, "step": 9199 }, { "epoch": 57.14330218068536, "grad_norm": 3.140625, "learning_rate": 4.1028130773134e-05, "loss": 0.2232, "step": 9200 }, { "epoch": 57.149532710280376, "grad_norm": 3.015625, "learning_rate": 4.1026239914695617e-05, "loss": 0.2236, "step": 9201 }, { "epoch": 57.15576323987539, "grad_norm": 1.8203125, "learning_rate": 4.102434890060718e-05, "loss": 0.1586, "step": 9202 }, { "epoch": 57.161993769470406, "grad_norm": 2.09375, "learning_rate": 4.102245773088707e-05, "loss": 0.1473, "step": 9203 }, { "epoch": 57.16822429906542, "grad_norm": 1.9453125, "learning_rate": 4.102056640555365e-05, "loss": 0.1628, "step": 9204 }, { "epoch": 57.17445482866044, "grad_norm": 2.796875, "learning_rate": 4.101867492462528e-05, "loss": 0.1984, "step": 9205 }, { "epoch": 57.18068535825545, "grad_norm": 2.671875, "learning_rate": 4.101678328812034e-05, "loss": 0.166, "step": 9206 }, { "epoch": 57.18691588785047, "grad_norm": 2.875, "learning_rate": 4.1014891496057197e-05, "loss": 0.1812, "step": 9207 }, { "epoch": 57.19314641744548, "grad_norm": 1.390625, "learning_rate": 4.101299954845423e-05, "loss": 0.1208, "step": 9208 }, { "epoch": 57.1993769470405, "grad_norm": 2.859375, "learning_rate": 4.101110744532981e-05, "loss": 0.1778, "step": 9209 }, { "epoch": 57.205607476635514, "grad_norm": 3.640625, "learning_rate": 4.100921518670231e-05, "loss": 0.1751, "step": 9210 }, { "epoch": 57.21183800623053, "grad_norm": 1.9296875, "learning_rate": 4.100732277259012e-05, "loss": 0.1206, "step": 9211 }, { "epoch": 57.218068535825545, "grad_norm": 1.90625, "learning_rate": 4.10054302030116e-05, "loss": 0.1311, "step": 9212 }, { "epoch": 57.22429906542056, "grad_norm": 2.671875, "learning_rate": 4.1003537477985154e-05, "loss": 0.1323, "step": 9213 }, { "epoch": 57.230529595015575, "grad_norm": 3.34375, "learning_rate": 4.1001644597529145e-05, "loss": 0.2866, "step": 9214 }, { "epoch": 57.23676012461059, "grad_norm": 2.65625, "learning_rate": 4.099975156166197e-05, "loss": 0.1246, "step": 9215 }, { "epoch": 57.242990654205606, "grad_norm": 3.4375, "learning_rate": 4.0997858370402007e-05, "loss": 0.2669, "step": 9216 }, { "epoch": 57.24922118380062, "grad_norm": 3.09375, "learning_rate": 4.0995965023767646e-05, "loss": 0.1158, "step": 9217 }, { "epoch": 57.25545171339564, "grad_norm": 1.9296875, "learning_rate": 4.099407152177728e-05, "loss": 0.1593, "step": 9218 }, { "epoch": 57.26168224299065, "grad_norm": 2.8125, "learning_rate": 4.099217786444928e-05, "loss": 0.2004, "step": 9219 }, { "epoch": 57.26791277258567, "grad_norm": 1.8046875, "learning_rate": 4.0990284051802064e-05, "loss": 0.15, "step": 9220 }, { "epoch": 57.27414330218068, "grad_norm": 3.6875, "learning_rate": 4.0988390083854014e-05, "loss": 0.2168, "step": 9221 }, { "epoch": 57.2803738317757, "grad_norm": 2.15625, "learning_rate": 4.098649596062352e-05, "loss": 0.123, "step": 9222 }, { "epoch": 57.28660436137071, "grad_norm": 1.65625, "learning_rate": 4.098460168212898e-05, "loss": 0.1623, "step": 9223 }, { "epoch": 57.29283489096573, "grad_norm": 1.515625, "learning_rate": 4.098270724838879e-05, "loss": 0.1185, "step": 9224 }, { "epoch": 57.299065420560744, "grad_norm": 4.0625, "learning_rate": 4.0980812659421355e-05, "loss": 0.1766, "step": 9225 }, { "epoch": 57.30529595015577, "grad_norm": 4.21875, "learning_rate": 4.0978917915245074e-05, "loss": 0.2163, "step": 9226 }, { "epoch": 57.31152647975078, "grad_norm": 3.046875, "learning_rate": 4.097702301587835e-05, "loss": 0.2313, "step": 9227 }, { "epoch": 57.3177570093458, "grad_norm": 3.265625, "learning_rate": 4.097512796133959e-05, "loss": 0.2929, "step": 9228 }, { "epoch": 57.32398753894081, "grad_norm": 3.09375, "learning_rate": 4.097323275164718e-05, "loss": 0.2177, "step": 9229 }, { "epoch": 57.33021806853583, "grad_norm": 2.421875, "learning_rate": 4.097133738681955e-05, "loss": 0.1215, "step": 9230 }, { "epoch": 57.33644859813084, "grad_norm": 3.21875, "learning_rate": 4.0969441866875094e-05, "loss": 0.2046, "step": 9231 }, { "epoch": 57.34267912772586, "grad_norm": 2.78125, "learning_rate": 4.096754619183223e-05, "loss": 0.1486, "step": 9232 }, { "epoch": 57.348909657320874, "grad_norm": 3.375, "learning_rate": 4.0965650361709363e-05, "loss": 0.3945, "step": 9233 }, { "epoch": 57.35514018691589, "grad_norm": 1.9453125, "learning_rate": 4.096375437652491e-05, "loss": 0.1506, "step": 9234 }, { "epoch": 57.361370716510905, "grad_norm": 2.40625, "learning_rate": 4.096185823629728e-05, "loss": 0.2241, "step": 9235 }, { "epoch": 57.36760124610592, "grad_norm": 1.953125, "learning_rate": 4.09599619410449e-05, "loss": 0.2214, "step": 9236 }, { "epoch": 57.373831775700936, "grad_norm": 3.046875, "learning_rate": 4.095806549078617e-05, "loss": 0.3342, "step": 9237 }, { "epoch": 57.38006230529595, "grad_norm": 2.953125, "learning_rate": 4.0956168885539524e-05, "loss": 0.2092, "step": 9238 }, { "epoch": 57.386292834890966, "grad_norm": 2.4375, "learning_rate": 4.0954272125323375e-05, "loss": 0.1311, "step": 9239 }, { "epoch": 57.39252336448598, "grad_norm": 3.71875, "learning_rate": 4.095237521015615e-05, "loss": 0.3466, "step": 9240 }, { "epoch": 57.398753894081, "grad_norm": 1.453125, "learning_rate": 4.095047814005626e-05, "loss": 0.119, "step": 9241 }, { "epoch": 57.40498442367601, "grad_norm": 2.390625, "learning_rate": 4.094858091504214e-05, "loss": 0.1566, "step": 9242 }, { "epoch": 57.41121495327103, "grad_norm": 1.3515625, "learning_rate": 4.094668353513221e-05, "loss": 0.1013, "step": 9243 }, { "epoch": 57.41744548286604, "grad_norm": 1.265625, "learning_rate": 4.0944786000344915e-05, "loss": 0.1149, "step": 9244 }, { "epoch": 57.42367601246106, "grad_norm": 2.28125, "learning_rate": 4.0942888310698656e-05, "loss": 0.1924, "step": 9245 }, { "epoch": 57.429906542056074, "grad_norm": 2.75, "learning_rate": 4.0940990466211894e-05, "loss": 0.2027, "step": 9246 }, { "epoch": 57.43613707165109, "grad_norm": 1.765625, "learning_rate": 4.093909246690303e-05, "loss": 0.0945, "step": 9247 }, { "epoch": 57.442367601246104, "grad_norm": 1.46875, "learning_rate": 4.0937194312790526e-05, "loss": 0.1164, "step": 9248 }, { "epoch": 57.44859813084112, "grad_norm": 2.59375, "learning_rate": 4.0935296003892804e-05, "loss": 0.1756, "step": 9249 }, { "epoch": 57.454828660436135, "grad_norm": 2.109375, "learning_rate": 4.0933397540228305e-05, "loss": 0.1696, "step": 9250 }, { "epoch": 57.46105919003115, "grad_norm": 2.65625, "learning_rate": 4.093149892181545e-05, "loss": 0.1379, "step": 9251 }, { "epoch": 57.467289719626166, "grad_norm": 2.125, "learning_rate": 4.0929600148672695e-05, "loss": 0.2411, "step": 9252 }, { "epoch": 57.47352024922118, "grad_norm": 3.515625, "learning_rate": 4.0927701220818495e-05, "loss": 0.3118, "step": 9253 }, { "epoch": 57.4797507788162, "grad_norm": 1.921875, "learning_rate": 4.0925802138271263e-05, "loss": 0.1292, "step": 9254 }, { "epoch": 57.48598130841121, "grad_norm": 1.375, "learning_rate": 4.092390290104946e-05, "loss": 0.1123, "step": 9255 }, { "epoch": 57.49221183800623, "grad_norm": 2.28125, "learning_rate": 4.092200350917153e-05, "loss": 0.1703, "step": 9256 }, { "epoch": 57.49844236760124, "grad_norm": 3.765625, "learning_rate": 4.092010396265592e-05, "loss": 0.1755, "step": 9257 }, { "epoch": 57.504672897196265, "grad_norm": 2.84375, "learning_rate": 4.091820426152108e-05, "loss": 0.2515, "step": 9258 }, { "epoch": 57.51090342679128, "grad_norm": 3.03125, "learning_rate": 4.0916304405785456e-05, "loss": 0.262, "step": 9259 }, { "epoch": 57.517133956386296, "grad_norm": 2.625, "learning_rate": 4.09144043954675e-05, "loss": 0.1761, "step": 9260 }, { "epoch": 57.52336448598131, "grad_norm": 2.75, "learning_rate": 4.091250423058567e-05, "loss": 0.2159, "step": 9261 }, { "epoch": 57.52959501557633, "grad_norm": 2.75, "learning_rate": 4.091060391115841e-05, "loss": 0.1436, "step": 9262 }, { "epoch": 57.53582554517134, "grad_norm": 2.796875, "learning_rate": 4.090870343720419e-05, "loss": 0.1691, "step": 9263 }, { "epoch": 57.54205607476636, "grad_norm": 2.203125, "learning_rate": 4.090680280874146e-05, "loss": 0.2988, "step": 9264 }, { "epoch": 57.54828660436137, "grad_norm": 1.96875, "learning_rate": 4.090490202578869e-05, "loss": 0.1578, "step": 9265 }, { "epoch": 57.55451713395639, "grad_norm": 2.9375, "learning_rate": 4.090300108836432e-05, "loss": 0.2459, "step": 9266 }, { "epoch": 57.5607476635514, "grad_norm": 4.71875, "learning_rate": 4.090109999648683e-05, "loss": 0.1638, "step": 9267 }, { "epoch": 57.56697819314642, "grad_norm": 3.09375, "learning_rate": 4.0899198750174687e-05, "loss": 0.1889, "step": 9268 }, { "epoch": 57.573208722741434, "grad_norm": 1.2578125, "learning_rate": 4.089729734944634e-05, "loss": 0.1198, "step": 9269 }, { "epoch": 57.57943925233645, "grad_norm": 1.84375, "learning_rate": 4.0895395794320265e-05, "loss": 0.1013, "step": 9270 }, { "epoch": 57.585669781931465, "grad_norm": 2.984375, "learning_rate": 4.089349408481492e-05, "loss": 0.2303, "step": 9271 }, { "epoch": 57.59190031152648, "grad_norm": 2.09375, "learning_rate": 4.089159222094879e-05, "loss": 0.1119, "step": 9272 }, { "epoch": 57.598130841121495, "grad_norm": 3.6875, "learning_rate": 4.0889690202740346e-05, "loss": 0.1659, "step": 9273 }, { "epoch": 57.60436137071651, "grad_norm": 3.96875, "learning_rate": 4.088778803020805e-05, "loss": 0.3773, "step": 9274 }, { "epoch": 57.610591900311526, "grad_norm": 2.421875, "learning_rate": 4.088588570337037e-05, "loss": 0.1576, "step": 9275 }, { "epoch": 57.61682242990654, "grad_norm": 2.234375, "learning_rate": 4.088398322224581e-05, "loss": 0.1377, "step": 9276 }, { "epoch": 57.62305295950156, "grad_norm": 2.5, "learning_rate": 4.088208058685282e-05, "loss": 0.1638, "step": 9277 }, { "epoch": 57.62928348909657, "grad_norm": 2.828125, "learning_rate": 4.088017779720989e-05, "loss": 0.2398, "step": 9278 }, { "epoch": 57.63551401869159, "grad_norm": 4.125, "learning_rate": 4.0878274853335496e-05, "loss": 0.2124, "step": 9279 }, { "epoch": 57.6417445482866, "grad_norm": 4.25, "learning_rate": 4.0876371755248126e-05, "loss": 0.168, "step": 9280 }, { "epoch": 57.64797507788162, "grad_norm": 3.828125, "learning_rate": 4.087446850296626e-05, "loss": 0.2383, "step": 9281 }, { "epoch": 57.654205607476634, "grad_norm": 2.078125, "learning_rate": 4.087256509650839e-05, "loss": 0.1319, "step": 9282 }, { "epoch": 57.66043613707165, "grad_norm": 2.5, "learning_rate": 4.087066153589298e-05, "loss": 0.1865, "step": 9283 }, { "epoch": 57.666666666666664, "grad_norm": 2.625, "learning_rate": 4.086875782113854e-05, "loss": 0.1339, "step": 9284 }, { "epoch": 57.67289719626168, "grad_norm": 2.390625, "learning_rate": 4.086685395226355e-05, "loss": 0.1956, "step": 9285 }, { "epoch": 57.679127725856695, "grad_norm": 2.515625, "learning_rate": 4.0864949929286504e-05, "loss": 0.1298, "step": 9286 }, { "epoch": 57.68535825545171, "grad_norm": 3.21875, "learning_rate": 4.086304575222589e-05, "loss": 0.1442, "step": 9287 }, { "epoch": 57.691588785046726, "grad_norm": 3.765625, "learning_rate": 4.0861141421100214e-05, "loss": 0.2104, "step": 9288 }, { "epoch": 57.69781931464174, "grad_norm": 3.734375, "learning_rate": 4.085923693592796e-05, "loss": 0.326, "step": 9289 }, { "epoch": 57.70404984423676, "grad_norm": 2.25, "learning_rate": 4.0857332296727615e-05, "loss": 0.1521, "step": 9290 }, { "epoch": 57.71028037383178, "grad_norm": 1.6328125, "learning_rate": 4.08554275035177e-05, "loss": 0.1449, "step": 9291 }, { "epoch": 57.716510903426794, "grad_norm": 3.046875, "learning_rate": 4.0853522556316696e-05, "loss": 0.2915, "step": 9292 }, { "epoch": 57.72274143302181, "grad_norm": 2.671875, "learning_rate": 4.085161745514312e-05, "loss": 0.1124, "step": 9293 }, { "epoch": 57.728971962616825, "grad_norm": 2.609375, "learning_rate": 4.084971220001547e-05, "loss": 0.1604, "step": 9294 }, { "epoch": 57.73520249221184, "grad_norm": 3.328125, "learning_rate": 4.084780679095223e-05, "loss": 0.2105, "step": 9295 }, { "epoch": 57.741433021806856, "grad_norm": 1.1015625, "learning_rate": 4.0845901227971936e-05, "loss": 0.1106, "step": 9296 }, { "epoch": 57.74766355140187, "grad_norm": 1.609375, "learning_rate": 4.084399551109308e-05, "loss": 0.1232, "step": 9297 }, { "epoch": 57.753894080996886, "grad_norm": 2.375, "learning_rate": 4.084208964033417e-05, "loss": 0.166, "step": 9298 }, { "epoch": 57.7601246105919, "grad_norm": 2.984375, "learning_rate": 4.0840183615713714e-05, "loss": 0.2024, "step": 9299 }, { "epoch": 57.76635514018692, "grad_norm": 2.875, "learning_rate": 4.083827743725024e-05, "loss": 0.1828, "step": 9300 }, { "epoch": 57.77258566978193, "grad_norm": 2.40625, "learning_rate": 4.083637110496224e-05, "loss": 0.1542, "step": 9301 }, { "epoch": 57.77881619937695, "grad_norm": 2.359375, "learning_rate": 4.0834464618868236e-05, "loss": 0.128, "step": 9302 }, { "epoch": 57.78504672897196, "grad_norm": 2.84375, "learning_rate": 4.083255797898675e-05, "loss": 0.2188, "step": 9303 }, { "epoch": 57.79127725856698, "grad_norm": 2.15625, "learning_rate": 4.08306511853363e-05, "loss": 0.1458, "step": 9304 }, { "epoch": 57.797507788161994, "grad_norm": 2.734375, "learning_rate": 4.0828744237935395e-05, "loss": 0.1941, "step": 9305 }, { "epoch": 57.80373831775701, "grad_norm": 2.890625, "learning_rate": 4.0826837136802565e-05, "loss": 0.3106, "step": 9306 }, { "epoch": 57.809968847352025, "grad_norm": 2.734375, "learning_rate": 4.082492988195633e-05, "loss": 0.2775, "step": 9307 }, { "epoch": 57.81619937694704, "grad_norm": 3.15625, "learning_rate": 4.082302247341521e-05, "loss": 0.2262, "step": 9308 }, { "epoch": 57.822429906542055, "grad_norm": 2.421875, "learning_rate": 4.082111491119774e-05, "loss": 0.156, "step": 9309 }, { "epoch": 57.82866043613707, "grad_norm": 1.078125, "learning_rate": 4.0819207195322426e-05, "loss": 0.0943, "step": 9310 }, { "epoch": 57.834890965732086, "grad_norm": 2.578125, "learning_rate": 4.0817299325807825e-05, "loss": 0.1548, "step": 9311 }, { "epoch": 57.8411214953271, "grad_norm": 2.921875, "learning_rate": 4.081539130267244e-05, "loss": 0.2014, "step": 9312 }, { "epoch": 57.84735202492212, "grad_norm": 3.859375, "learning_rate": 4.0813483125934824e-05, "loss": 0.2159, "step": 9313 }, { "epoch": 57.85358255451713, "grad_norm": 2.9375, "learning_rate": 4.081157479561349e-05, "loss": 0.192, "step": 9314 }, { "epoch": 57.85981308411215, "grad_norm": 1.8203125, "learning_rate": 4.080966631172698e-05, "loss": 0.1176, "step": 9315 }, { "epoch": 57.86604361370716, "grad_norm": 1.0703125, "learning_rate": 4.0807757674293834e-05, "loss": 0.1138, "step": 9316 }, { "epoch": 57.87227414330218, "grad_norm": 1.6015625, "learning_rate": 4.0805848883332584e-05, "loss": 0.1669, "step": 9317 }, { "epoch": 57.87850467289719, "grad_norm": 1.625, "learning_rate": 4.0803939938861765e-05, "loss": 0.1524, "step": 9318 }, { "epoch": 57.88473520249221, "grad_norm": 3.0, "learning_rate": 4.0802030840899936e-05, "loss": 0.293, "step": 9319 }, { "epoch": 57.890965732087224, "grad_norm": 2.421875, "learning_rate": 4.080012158946561e-05, "loss": 0.2046, "step": 9320 }, { "epoch": 57.89719626168224, "grad_norm": 1.640625, "learning_rate": 4.0798212184577354e-05, "loss": 0.1363, "step": 9321 }, { "epoch": 57.90342679127726, "grad_norm": 2.796875, "learning_rate": 4.07963026262537e-05, "loss": 0.2278, "step": 9322 }, { "epoch": 57.90965732087228, "grad_norm": 1.9140625, "learning_rate": 4.0794392914513204e-05, "loss": 0.1291, "step": 9323 }, { "epoch": 57.91588785046729, "grad_norm": 1.671875, "learning_rate": 4.07924830493744e-05, "loss": 0.1541, "step": 9324 }, { "epoch": 57.92211838006231, "grad_norm": 1.6171875, "learning_rate": 4.079057303085584e-05, "loss": 0.1022, "step": 9325 }, { "epoch": 57.92834890965732, "grad_norm": 3.734375, "learning_rate": 4.078866285897609e-05, "loss": 0.3387, "step": 9326 }, { "epoch": 57.93457943925234, "grad_norm": 2.6875, "learning_rate": 4.078675253375368e-05, "loss": 0.1813, "step": 9327 }, { "epoch": 57.940809968847354, "grad_norm": 2.765625, "learning_rate": 4.078484205520717e-05, "loss": 0.2329, "step": 9328 }, { "epoch": 57.94704049844237, "grad_norm": 2.375, "learning_rate": 4.078293142335513e-05, "loss": 0.3176, "step": 9329 }, { "epoch": 57.953271028037385, "grad_norm": 1.4609375, "learning_rate": 4.0781020638216106e-05, "loss": 0.0879, "step": 9330 }, { "epoch": 57.9595015576324, "grad_norm": 2.421875, "learning_rate": 4.077910969980865e-05, "loss": 0.1885, "step": 9331 }, { "epoch": 57.965732087227416, "grad_norm": 1.3203125, "learning_rate": 4.077719860815132e-05, "loss": 0.0983, "step": 9332 }, { "epoch": 57.97196261682243, "grad_norm": 3.5, "learning_rate": 4.077528736326269e-05, "loss": 0.2132, "step": 9333 }, { "epoch": 57.978193146417446, "grad_norm": 3.5, "learning_rate": 4.077337596516132e-05, "loss": 0.1351, "step": 9334 }, { "epoch": 57.98442367601246, "grad_norm": 4.21875, "learning_rate": 4.077146441386576e-05, "loss": 0.2456, "step": 9335 }, { "epoch": 57.99065420560748, "grad_norm": 3.390625, "learning_rate": 4.076955270939459e-05, "loss": 0.137, "step": 9336 }, { "epoch": 57.99688473520249, "grad_norm": 3.484375, "learning_rate": 4.076764085176637e-05, "loss": 0.3315, "step": 9337 }, { "epoch": 58.0, "grad_norm": 1.625, "learning_rate": 4.076572884099967e-05, "loss": 0.0472, "step": 9338 }, { "epoch": 58.006230529595015, "grad_norm": 4.78125, "learning_rate": 4.0763816677113064e-05, "loss": 0.3681, "step": 9339 }, { "epoch": 58.01246105919003, "grad_norm": 4.0625, "learning_rate": 4.0761904360125115e-05, "loss": 0.2914, "step": 9340 }, { "epoch": 58.018691588785046, "grad_norm": 3.9375, "learning_rate": 4.0759991890054406e-05, "loss": 0.2029, "step": 9341 }, { "epoch": 58.02492211838006, "grad_norm": 2.90625, "learning_rate": 4.0758079266919504e-05, "loss": 0.2691, "step": 9342 }, { "epoch": 58.03115264797508, "grad_norm": 2.609375, "learning_rate": 4.075616649073898e-05, "loss": 0.2937, "step": 9343 }, { "epoch": 58.03738317757009, "grad_norm": 3.265625, "learning_rate": 4.075425356153142e-05, "loss": 0.3314, "step": 9344 }, { "epoch": 58.04361370716511, "grad_norm": 3.875, "learning_rate": 4.07523404793154e-05, "loss": 0.1532, "step": 9345 }, { "epoch": 58.04984423676012, "grad_norm": 3.796875, "learning_rate": 4.0750427244109504e-05, "loss": 0.1788, "step": 9346 }, { "epoch": 58.05607476635514, "grad_norm": 3.0625, "learning_rate": 4.074851385593231e-05, "loss": 0.1747, "step": 9347 }, { "epoch": 58.06230529595015, "grad_norm": 2.609375, "learning_rate": 4.07466003148024e-05, "loss": 0.1734, "step": 9348 }, { "epoch": 58.06853582554517, "grad_norm": 3.1875, "learning_rate": 4.0744686620738356e-05, "loss": 0.2681, "step": 9349 }, { "epoch": 58.074766355140184, "grad_norm": 3.046875, "learning_rate": 4.074277277375877e-05, "loss": 0.2516, "step": 9350 }, { "epoch": 58.0809968847352, "grad_norm": 2.78125, "learning_rate": 4.0740858773882226e-05, "loss": 0.1717, "step": 9351 }, { "epoch": 58.087227414330215, "grad_norm": 3.6875, "learning_rate": 4.073894462112732e-05, "loss": 0.1865, "step": 9352 }, { "epoch": 58.09345794392523, "grad_norm": 1.8984375, "learning_rate": 4.073703031551264e-05, "loss": 0.1424, "step": 9353 }, { "epoch": 58.099688473520246, "grad_norm": 2.390625, "learning_rate": 4.073511585705677e-05, "loss": 0.1315, "step": 9354 }, { "epoch": 58.10591900311527, "grad_norm": 3.09375, "learning_rate": 4.0733201245778306e-05, "loss": 0.2049, "step": 9355 }, { "epoch": 58.11214953271028, "grad_norm": 3.0, "learning_rate": 4.073128648169585e-05, "loss": 0.1455, "step": 9356 }, { "epoch": 58.1183800623053, "grad_norm": 2.8125, "learning_rate": 4.0729371564828e-05, "loss": 0.2493, "step": 9357 }, { "epoch": 58.124610591900314, "grad_norm": 2.859375, "learning_rate": 4.072745649519335e-05, "loss": 0.2353, "step": 9358 }, { "epoch": 58.13084112149533, "grad_norm": 3.015625, "learning_rate": 4.0725541272810495e-05, "loss": 0.2055, "step": 9359 }, { "epoch": 58.137071651090345, "grad_norm": 1.578125, "learning_rate": 4.072362589769804e-05, "loss": 0.1108, "step": 9360 }, { "epoch": 58.14330218068536, "grad_norm": 1.359375, "learning_rate": 4.072171036987459e-05, "loss": 0.1203, "step": 9361 }, { "epoch": 58.149532710280376, "grad_norm": 2.359375, "learning_rate": 4.071979468935874e-05, "loss": 0.2831, "step": 9362 }, { "epoch": 58.15576323987539, "grad_norm": 2.40625, "learning_rate": 4.07178788561691e-05, "loss": 0.2696, "step": 9363 }, { "epoch": 58.161993769470406, "grad_norm": 3.484375, "learning_rate": 4.0715962870324284e-05, "loss": 0.2274, "step": 9364 }, { "epoch": 58.16822429906542, "grad_norm": 3.296875, "learning_rate": 4.071404673184289e-05, "loss": 0.1553, "step": 9365 }, { "epoch": 58.17445482866044, "grad_norm": 2.546875, "learning_rate": 4.071213044074354e-05, "loss": 0.1362, "step": 9366 }, { "epoch": 58.18068535825545, "grad_norm": 2.734375, "learning_rate": 4.071021399704484e-05, "loss": 0.2039, "step": 9367 }, { "epoch": 58.18691588785047, "grad_norm": 2.078125, "learning_rate": 4.0708297400765395e-05, "loss": 0.213, "step": 9368 }, { "epoch": 58.19314641744548, "grad_norm": 2.578125, "learning_rate": 4.0706380651923826e-05, "loss": 0.2426, "step": 9369 }, { "epoch": 58.1993769470405, "grad_norm": 2.953125, "learning_rate": 4.070446375053875e-05, "loss": 0.2193, "step": 9370 }, { "epoch": 58.205607476635514, "grad_norm": 3.578125, "learning_rate": 4.070254669662879e-05, "loss": 0.2446, "step": 9371 }, { "epoch": 58.21183800623053, "grad_norm": 1.828125, "learning_rate": 4.070062949021254e-05, "loss": 0.1489, "step": 9372 }, { "epoch": 58.218068535825545, "grad_norm": 1.203125, "learning_rate": 4.069871213130865e-05, "loss": 0.1086, "step": 9373 }, { "epoch": 58.22429906542056, "grad_norm": 1.7890625, "learning_rate": 4.0696794619935726e-05, "loss": 0.1255, "step": 9374 }, { "epoch": 58.230529595015575, "grad_norm": 1.6484375, "learning_rate": 4.06948769561124e-05, "loss": 0.1242, "step": 9375 }, { "epoch": 58.23676012461059, "grad_norm": 1.28125, "learning_rate": 4.069295913985728e-05, "loss": 0.0992, "step": 9376 }, { "epoch": 58.242990654205606, "grad_norm": 1.5, "learning_rate": 4.069104117118902e-05, "loss": 0.1205, "step": 9377 }, { "epoch": 58.24922118380062, "grad_norm": 2.90625, "learning_rate": 4.068912305012622e-05, "loss": 0.1863, "step": 9378 }, { "epoch": 58.25545171339564, "grad_norm": 1.7734375, "learning_rate": 4.068720477668753e-05, "loss": 0.1577, "step": 9379 }, { "epoch": 58.26168224299065, "grad_norm": 2.0625, "learning_rate": 4.068528635089156e-05, "loss": 0.1584, "step": 9380 }, { "epoch": 58.26791277258567, "grad_norm": 2.359375, "learning_rate": 4.068336777275696e-05, "loss": 0.1527, "step": 9381 }, { "epoch": 58.27414330218068, "grad_norm": 2.765625, "learning_rate": 4.068144904230236e-05, "loss": 0.2031, "step": 9382 }, { "epoch": 58.2803738317757, "grad_norm": 1.515625, "learning_rate": 4.0679530159546385e-05, "loss": 0.1278, "step": 9383 }, { "epoch": 58.28660436137071, "grad_norm": 2.25, "learning_rate": 4.0677611124507685e-05, "loss": 0.1957, "step": 9384 }, { "epoch": 58.29283489096573, "grad_norm": 2.109375, "learning_rate": 4.0675691937204886e-05, "loss": 0.1463, "step": 9385 }, { "epoch": 58.299065420560744, "grad_norm": 2.875, "learning_rate": 4.067377259765664e-05, "loss": 0.171, "step": 9386 }, { "epoch": 58.30529595015577, "grad_norm": 2.03125, "learning_rate": 4.067185310588158e-05, "loss": 0.1434, "step": 9387 }, { "epoch": 58.31152647975078, "grad_norm": 1.6171875, "learning_rate": 4.066993346189835e-05, "loss": 0.1334, "step": 9388 }, { "epoch": 58.3177570093458, "grad_norm": 1.9296875, "learning_rate": 4.066801366572559e-05, "loss": 0.1113, "step": 9389 }, { "epoch": 58.32398753894081, "grad_norm": 2.453125, "learning_rate": 4.0666093717381945e-05, "loss": 0.2749, "step": 9390 }, { "epoch": 58.33021806853583, "grad_norm": 3.265625, "learning_rate": 4.0664173616886075e-05, "loss": 0.3489, "step": 9391 }, { "epoch": 58.33644859813084, "grad_norm": 4.1875, "learning_rate": 4.066225336425662e-05, "loss": 0.4474, "step": 9392 }, { "epoch": 58.34267912772586, "grad_norm": 3.421875, "learning_rate": 4.066033295951223e-05, "loss": 0.2358, "step": 9393 }, { "epoch": 58.348909657320874, "grad_norm": 3.359375, "learning_rate": 4.0658412402671545e-05, "loss": 0.2234, "step": 9394 }, { "epoch": 58.35514018691589, "grad_norm": 3.390625, "learning_rate": 4.065649169375324e-05, "loss": 0.1038, "step": 9395 }, { "epoch": 58.361370716510905, "grad_norm": 2.671875, "learning_rate": 4.065457083277595e-05, "loss": 0.2223, "step": 9396 }, { "epoch": 58.36760124610592, "grad_norm": 1.921875, "learning_rate": 4.065264981975835e-05, "loss": 0.109, "step": 9397 }, { "epoch": 58.373831775700936, "grad_norm": 2.015625, "learning_rate": 4.0650728654719074e-05, "loss": 0.1468, "step": 9398 }, { "epoch": 58.38006230529595, "grad_norm": 1.71875, "learning_rate": 4.0648807337676795e-05, "loss": 0.1122, "step": 9399 }, { "epoch": 58.386292834890966, "grad_norm": 1.9453125, "learning_rate": 4.0646885868650174e-05, "loss": 0.1442, "step": 9400 }, { "epoch": 58.39252336448598, "grad_norm": 2.0, "learning_rate": 4.064496424765787e-05, "loss": 0.1921, "step": 9401 }, { "epoch": 58.398753894081, "grad_norm": 2.0, "learning_rate": 4.064304247471855e-05, "loss": 0.1223, "step": 9402 }, { "epoch": 58.40498442367601, "grad_norm": 1.390625, "learning_rate": 4.064112054985087e-05, "loss": 0.1374, "step": 9403 }, { "epoch": 58.41121495327103, "grad_norm": 1.8203125, "learning_rate": 4.0639198473073495e-05, "loss": 0.137, "step": 9404 }, { "epoch": 58.41744548286604, "grad_norm": 1.4609375, "learning_rate": 4.063727624440511e-05, "loss": 0.1456, "step": 9405 }, { "epoch": 58.42367601246106, "grad_norm": 1.8203125, "learning_rate": 4.0635353863864365e-05, "loss": 0.1177, "step": 9406 }, { "epoch": 58.429906542056074, "grad_norm": 2.546875, "learning_rate": 4.063343133146994e-05, "loss": 0.2143, "step": 9407 }, { "epoch": 58.43613707165109, "grad_norm": 3.28125, "learning_rate": 4.06315086472405e-05, "loss": 0.1393, "step": 9408 }, { "epoch": 58.442367601246104, "grad_norm": 1.9765625, "learning_rate": 4.0629585811194726e-05, "loss": 0.2284, "step": 9409 }, { "epoch": 58.44859813084112, "grad_norm": 1.9609375, "learning_rate": 4.0627662823351295e-05, "loss": 0.172, "step": 9410 }, { "epoch": 58.454828660436135, "grad_norm": 3.21875, "learning_rate": 4.062573968372887e-05, "loss": 0.1244, "step": 9411 }, { "epoch": 58.46105919003115, "grad_norm": 2.8125, "learning_rate": 4.062381639234615e-05, "loss": 0.3028, "step": 9412 }, { "epoch": 58.467289719626166, "grad_norm": 5.09375, "learning_rate": 4.062189294922179e-05, "loss": 0.304, "step": 9413 }, { "epoch": 58.47352024922118, "grad_norm": 3.328125, "learning_rate": 4.0619969354374485e-05, "loss": 0.1121, "step": 9414 }, { "epoch": 58.4797507788162, "grad_norm": 3.96875, "learning_rate": 4.061804560782292e-05, "loss": 0.1767, "step": 9415 }, { "epoch": 58.48598130841121, "grad_norm": 3.6875, "learning_rate": 4.0616121709585775e-05, "loss": 0.1605, "step": 9416 }, { "epoch": 58.49221183800623, "grad_norm": 2.328125, "learning_rate": 4.061419765968173e-05, "loss": 0.1362, "step": 9417 }, { "epoch": 58.49844236760124, "grad_norm": 2.265625, "learning_rate": 4.061227345812948e-05, "loss": 0.2381, "step": 9418 }, { "epoch": 58.504672897196265, "grad_norm": 4.09375, "learning_rate": 4.061034910494771e-05, "loss": 0.282, "step": 9419 }, { "epoch": 58.51090342679128, "grad_norm": 2.609375, "learning_rate": 4.0608424600155106e-05, "loss": 0.1464, "step": 9420 }, { "epoch": 58.517133956386296, "grad_norm": 4.9375, "learning_rate": 4.060649994377036e-05, "loss": 0.2718, "step": 9421 }, { "epoch": 58.52336448598131, "grad_norm": 3.828125, "learning_rate": 4.0604575135812164e-05, "loss": 0.1543, "step": 9422 }, { "epoch": 58.52959501557633, "grad_norm": 3.53125, "learning_rate": 4.0602650176299215e-05, "loss": 0.2131, "step": 9423 }, { "epoch": 58.53582554517134, "grad_norm": 3.34375, "learning_rate": 4.060072506525022e-05, "loss": 0.2288, "step": 9424 }, { "epoch": 58.54205607476636, "grad_norm": 2.5625, "learning_rate": 4.059879980268386e-05, "loss": 0.2031, "step": 9425 }, { "epoch": 58.54828660436137, "grad_norm": 2.796875, "learning_rate": 4.059687438861883e-05, "loss": 0.1094, "step": 9426 }, { "epoch": 58.55451713395639, "grad_norm": 2.609375, "learning_rate": 4.0594948823073846e-05, "loss": 0.1209, "step": 9427 }, { "epoch": 58.5607476635514, "grad_norm": 2.46875, "learning_rate": 4.0593023106067606e-05, "loss": 0.1589, "step": 9428 }, { "epoch": 58.56697819314642, "grad_norm": 2.703125, "learning_rate": 4.05910972376188e-05, "loss": 0.2341, "step": 9429 }, { "epoch": 58.573208722741434, "grad_norm": 3.890625, "learning_rate": 4.0589171217746145e-05, "loss": 0.2543, "step": 9430 }, { "epoch": 58.57943925233645, "grad_norm": 3.453125, "learning_rate": 4.058724504646834e-05, "loss": 0.2806, "step": 9431 }, { "epoch": 58.585669781931465, "grad_norm": 2.8125, "learning_rate": 4.05853187238041e-05, "loss": 0.1261, "step": 9432 }, { "epoch": 58.59190031152648, "grad_norm": 1.7421875, "learning_rate": 4.058339224977212e-05, "loss": 0.111, "step": 9433 }, { "epoch": 58.598130841121495, "grad_norm": 1.765625, "learning_rate": 4.058146562439113e-05, "loss": 0.1234, "step": 9434 }, { "epoch": 58.60436137071651, "grad_norm": 1.6171875, "learning_rate": 4.057953884767983e-05, "loss": 0.1137, "step": 9435 }, { "epoch": 58.610591900311526, "grad_norm": 2.859375, "learning_rate": 4.057761191965693e-05, "loss": 0.2217, "step": 9436 }, { "epoch": 58.61682242990654, "grad_norm": 3.15625, "learning_rate": 4.057568484034115e-05, "loss": 0.2181, "step": 9437 }, { "epoch": 58.62305295950156, "grad_norm": 2.484375, "learning_rate": 4.0573757609751206e-05, "loss": 0.1923, "step": 9438 }, { "epoch": 58.62928348909657, "grad_norm": 2.875, "learning_rate": 4.057183022790581e-05, "loss": 0.2337, "step": 9439 }, { "epoch": 58.63551401869159, "grad_norm": 3.15625, "learning_rate": 4.056990269482369e-05, "loss": 0.222, "step": 9440 }, { "epoch": 58.6417445482866, "grad_norm": 2.046875, "learning_rate": 4.0567975010523565e-05, "loss": 0.1705, "step": 9441 }, { "epoch": 58.64797507788162, "grad_norm": 3.421875, "learning_rate": 4.056604717502416e-05, "loss": 0.1894, "step": 9442 }, { "epoch": 58.654205607476634, "grad_norm": 2.359375, "learning_rate": 4.056411918834418e-05, "loss": 0.1528, "step": 9443 }, { "epoch": 58.66043613707165, "grad_norm": 1.953125, "learning_rate": 4.056219105050237e-05, "loss": 0.1129, "step": 9444 }, { "epoch": 58.666666666666664, "grad_norm": 1.3125, "learning_rate": 4.056026276151745e-05, "loss": 0.1141, "step": 9445 }, { "epoch": 58.67289719626168, "grad_norm": 3.03125, "learning_rate": 4.055833432140814e-05, "loss": 0.2331, "step": 9446 }, { "epoch": 58.679127725856695, "grad_norm": 2.40625, "learning_rate": 4.055640573019319e-05, "loss": 0.2044, "step": 9447 }, { "epoch": 58.68535825545171, "grad_norm": 1.5703125, "learning_rate": 4.055447698789131e-05, "loss": 0.1408, "step": 9448 }, { "epoch": 58.691588785046726, "grad_norm": 3.125, "learning_rate": 4.055254809452124e-05, "loss": 0.21, "step": 9449 }, { "epoch": 58.69781931464174, "grad_norm": 2.484375, "learning_rate": 4.0550619050101724e-05, "loss": 0.2092, "step": 9450 }, { "epoch": 58.70404984423676, "grad_norm": 2.953125, "learning_rate": 4.054868985465148e-05, "loss": 0.2756, "step": 9451 }, { "epoch": 58.71028037383178, "grad_norm": 2.3125, "learning_rate": 4.054676050818924e-05, "loss": 0.1675, "step": 9452 }, { "epoch": 58.716510903426794, "grad_norm": 2.6875, "learning_rate": 4.0544831010733774e-05, "loss": 0.2315, "step": 9453 }, { "epoch": 58.72274143302181, "grad_norm": 1.9140625, "learning_rate": 4.054290136230379e-05, "loss": 0.1284, "step": 9454 }, { "epoch": 58.728971962616825, "grad_norm": 2.9375, "learning_rate": 4.054097156291804e-05, "loss": 0.2713, "step": 9455 }, { "epoch": 58.73520249221184, "grad_norm": 2.984375, "learning_rate": 4.053904161259528e-05, "loss": 0.2349, "step": 9456 }, { "epoch": 58.741433021806856, "grad_norm": 2.765625, "learning_rate": 4.053711151135423e-05, "loss": 0.2636, "step": 9457 }, { "epoch": 58.74766355140187, "grad_norm": 3.359375, "learning_rate": 4.053518125921365e-05, "loss": 0.2524, "step": 9458 }, { "epoch": 58.753894080996886, "grad_norm": 2.171875, "learning_rate": 4.053325085619229e-05, "loss": 0.1342, "step": 9459 }, { "epoch": 58.7601246105919, "grad_norm": 2.328125, "learning_rate": 4.053132030230888e-05, "loss": 0.1804, "step": 9460 }, { "epoch": 58.76635514018692, "grad_norm": 2.734375, "learning_rate": 4.0529389597582193e-05, "loss": 0.1655, "step": 9461 }, { "epoch": 58.77258566978193, "grad_norm": 2.78125, "learning_rate": 4.052745874203097e-05, "loss": 0.1485, "step": 9462 }, { "epoch": 58.77881619937695, "grad_norm": 3.921875, "learning_rate": 4.0525527735673963e-05, "loss": 0.2787, "step": 9463 }, { "epoch": 58.78504672897196, "grad_norm": 2.8125, "learning_rate": 4.052359657852992e-05, "loss": 0.2725, "step": 9464 }, { "epoch": 58.79127725856698, "grad_norm": 1.7421875, "learning_rate": 4.0521665270617614e-05, "loss": 0.1254, "step": 9465 }, { "epoch": 58.797507788161994, "grad_norm": 3.109375, "learning_rate": 4.051973381195578e-05, "loss": 0.2405, "step": 9466 }, { "epoch": 58.80373831775701, "grad_norm": 3.0, "learning_rate": 4.0517802202563205e-05, "loss": 0.2139, "step": 9467 }, { "epoch": 58.809968847352025, "grad_norm": 2.71875, "learning_rate": 4.051587044245862e-05, "loss": 0.1738, "step": 9468 }, { "epoch": 58.81619937694704, "grad_norm": 1.4765625, "learning_rate": 4.051393853166081e-05, "loss": 0.1158, "step": 9469 }, { "epoch": 58.822429906542055, "grad_norm": 2.53125, "learning_rate": 4.0512006470188515e-05, "loss": 0.1598, "step": 9470 }, { "epoch": 58.82866043613707, "grad_norm": 1.640625, "learning_rate": 4.051007425806052e-05, "loss": 0.1666, "step": 9471 }, { "epoch": 58.834890965732086, "grad_norm": 2.09375, "learning_rate": 4.050814189529558e-05, "loss": 0.145, "step": 9472 }, { "epoch": 58.8411214953271, "grad_norm": 3.28125, "learning_rate": 4.050620938191247e-05, "loss": 0.1584, "step": 9473 }, { "epoch": 58.84735202492212, "grad_norm": 2.5, "learning_rate": 4.050427671792996e-05, "loss": 0.2095, "step": 9474 }, { "epoch": 58.85358255451713, "grad_norm": 1.8984375, "learning_rate": 4.05023439033668e-05, "loss": 0.1638, "step": 9475 }, { "epoch": 58.85981308411215, "grad_norm": 3.09375, "learning_rate": 4.050041093824179e-05, "loss": 0.1811, "step": 9476 }, { "epoch": 58.86604361370716, "grad_norm": 3.53125, "learning_rate": 4.049847782257369e-05, "loss": 0.3851, "step": 9477 }, { "epoch": 58.87227414330218, "grad_norm": 2.703125, "learning_rate": 4.0496544556381263e-05, "loss": 0.1665, "step": 9478 }, { "epoch": 58.87850467289719, "grad_norm": 2.40625, "learning_rate": 4.0494611139683316e-05, "loss": 0.1454, "step": 9479 }, { "epoch": 58.88473520249221, "grad_norm": 1.5, "learning_rate": 4.0492677572498595e-05, "loss": 0.1369, "step": 9480 }, { "epoch": 58.890965732087224, "grad_norm": 2.609375, "learning_rate": 4.049074385484589e-05, "loss": 0.168, "step": 9481 }, { "epoch": 58.89719626168224, "grad_norm": 2.359375, "learning_rate": 4.0488809986743994e-05, "loss": 0.1708, "step": 9482 }, { "epoch": 58.90342679127726, "grad_norm": 2.359375, "learning_rate": 4.0486875968211675e-05, "loss": 0.1573, "step": 9483 }, { "epoch": 58.90965732087228, "grad_norm": 1.453125, "learning_rate": 4.048494179926772e-05, "loss": 0.1241, "step": 9484 }, { "epoch": 58.91588785046729, "grad_norm": 2.75, "learning_rate": 4.048300747993092e-05, "loss": 0.2203, "step": 9485 }, { "epoch": 58.92211838006231, "grad_norm": 2.546875, "learning_rate": 4.048107301022005e-05, "loss": 0.1519, "step": 9486 }, { "epoch": 58.92834890965732, "grad_norm": 1.84375, "learning_rate": 4.0479138390153907e-05, "loss": 0.1249, "step": 9487 }, { "epoch": 58.93457943925234, "grad_norm": 2.046875, "learning_rate": 4.047720361975128e-05, "loss": 0.1306, "step": 9488 }, { "epoch": 58.940809968847354, "grad_norm": 3.390625, "learning_rate": 4.0475268699030956e-05, "loss": 0.2433, "step": 9489 }, { "epoch": 58.94704049844237, "grad_norm": 3.71875, "learning_rate": 4.047333362801172e-05, "loss": 0.1924, "step": 9490 }, { "epoch": 58.953271028037385, "grad_norm": 3.328125, "learning_rate": 4.047139840671239e-05, "loss": 0.2341, "step": 9491 }, { "epoch": 58.9595015576324, "grad_norm": 2.828125, "learning_rate": 4.046946303515173e-05, "loss": 0.2663, "step": 9492 }, { "epoch": 58.965732087227416, "grad_norm": 3.28125, "learning_rate": 4.046752751334857e-05, "loss": 0.2076, "step": 9493 }, { "epoch": 58.97196261682243, "grad_norm": 1.953125, "learning_rate": 4.0465591841321685e-05, "loss": 0.1388, "step": 9494 }, { "epoch": 58.978193146417446, "grad_norm": 1.6640625, "learning_rate": 4.0463656019089876e-05, "loss": 0.141, "step": 9495 }, { "epoch": 58.98442367601246, "grad_norm": 2.203125, "learning_rate": 4.046172004667195e-05, "loss": 0.1784, "step": 9496 }, { "epoch": 58.99065420560748, "grad_norm": 2.4375, "learning_rate": 4.04597839240867e-05, "loss": 0.148, "step": 9497 }, { "epoch": 58.99688473520249, "grad_norm": 2.296875, "learning_rate": 4.0457847651352954e-05, "loss": 0.1429, "step": 9498 }, { "epoch": 59.0, "grad_norm": 1.015625, "learning_rate": 4.045591122848949e-05, "loss": 0.0482, "step": 9499 }, { "epoch": 59.006230529595015, "grad_norm": 2.21875, "learning_rate": 4.045397465551513e-05, "loss": 0.166, "step": 9500 }, { "epoch": 59.01246105919003, "grad_norm": 2.203125, "learning_rate": 4.0452037932448684e-05, "loss": 0.1532, "step": 9501 }, { "epoch": 59.018691588785046, "grad_norm": 2.203125, "learning_rate": 4.045010105930895e-05, "loss": 0.1424, "step": 9502 }, { "epoch": 59.02492211838006, "grad_norm": 2.34375, "learning_rate": 4.044816403611475e-05, "loss": 0.1609, "step": 9503 }, { "epoch": 59.03115264797508, "grad_norm": 2.703125, "learning_rate": 4.0446226862884895e-05, "loss": 0.1428, "step": 9504 }, { "epoch": 59.03738317757009, "grad_norm": 1.6875, "learning_rate": 4.044428953963819e-05, "loss": 0.1087, "step": 9505 }, { "epoch": 59.04361370716511, "grad_norm": 2.578125, "learning_rate": 4.0442352066393464e-05, "loss": 0.1784, "step": 9506 }, { "epoch": 59.04984423676012, "grad_norm": 1.7578125, "learning_rate": 4.044041444316953e-05, "loss": 0.1462, "step": 9507 }, { "epoch": 59.05607476635514, "grad_norm": 1.9921875, "learning_rate": 4.0438476669985196e-05, "loss": 0.2695, "step": 9508 }, { "epoch": 59.06230529595015, "grad_norm": 2.0, "learning_rate": 4.0436538746859296e-05, "loss": 0.15, "step": 9509 }, { "epoch": 59.06853582554517, "grad_norm": 2.078125, "learning_rate": 4.043460067381064e-05, "loss": 0.1843, "step": 9510 }, { "epoch": 59.074766355140184, "grad_norm": 3.71875, "learning_rate": 4.043266245085806e-05, "loss": 0.2147, "step": 9511 }, { "epoch": 59.0809968847352, "grad_norm": 2.453125, "learning_rate": 4.043072407802038e-05, "loss": 0.1858, "step": 9512 }, { "epoch": 59.087227414330215, "grad_norm": 2.65625, "learning_rate": 4.042878555531642e-05, "loss": 0.1865, "step": 9513 }, { "epoch": 59.09345794392523, "grad_norm": 2.671875, "learning_rate": 4.0426846882765015e-05, "loss": 0.2265, "step": 9514 }, { "epoch": 59.099688473520246, "grad_norm": 2.53125, "learning_rate": 4.042490806038499e-05, "loss": 0.242, "step": 9515 }, { "epoch": 59.10591900311527, "grad_norm": 2.6875, "learning_rate": 4.042296908819517e-05, "loss": 0.1289, "step": 9516 }, { "epoch": 59.11214953271028, "grad_norm": 3.0, "learning_rate": 4.04210299662144e-05, "loss": 0.1808, "step": 9517 }, { "epoch": 59.1183800623053, "grad_norm": 2.046875, "learning_rate": 4.041909069446149e-05, "loss": 0.1845, "step": 9518 }, { "epoch": 59.124610591900314, "grad_norm": 2.984375, "learning_rate": 4.041715127295529e-05, "loss": 0.2056, "step": 9519 }, { "epoch": 59.13084112149533, "grad_norm": 3.40625, "learning_rate": 4.041521170171464e-05, "loss": 0.2066, "step": 9520 }, { "epoch": 59.137071651090345, "grad_norm": 2.125, "learning_rate": 4.041327198075838e-05, "loss": 0.1454, "step": 9521 }, { "epoch": 59.14330218068536, "grad_norm": 1.328125, "learning_rate": 4.041133211010533e-05, "loss": 0.1213, "step": 9522 }, { "epoch": 59.149532710280376, "grad_norm": 1.7890625, "learning_rate": 4.040939208977435e-05, "loss": 0.1462, "step": 9523 }, { "epoch": 59.15576323987539, "grad_norm": 1.890625, "learning_rate": 4.040745191978427e-05, "loss": 0.175, "step": 9524 }, { "epoch": 59.161993769470406, "grad_norm": 2.84375, "learning_rate": 4.0405511600153934e-05, "loss": 0.1981, "step": 9525 }, { "epoch": 59.16822429906542, "grad_norm": 1.171875, "learning_rate": 4.0403571130902194e-05, "loss": 0.0889, "step": 9526 }, { "epoch": 59.17445482866044, "grad_norm": 2.6875, "learning_rate": 4.0401630512047895e-05, "loss": 0.1913, "step": 9527 }, { "epoch": 59.18068535825545, "grad_norm": 4.09375, "learning_rate": 4.039968974360988e-05, "loss": 0.3318, "step": 9528 }, { "epoch": 59.18691588785047, "grad_norm": 3.015625, "learning_rate": 4.0397748825607e-05, "loss": 0.2551, "step": 9529 }, { "epoch": 59.19314641744548, "grad_norm": 2.734375, "learning_rate": 4.03958077580581e-05, "loss": 0.2866, "step": 9530 }, { "epoch": 59.1993769470405, "grad_norm": 3.0625, "learning_rate": 4.039386654098204e-05, "loss": 0.3356, "step": 9531 }, { "epoch": 59.205607476635514, "grad_norm": 1.9921875, "learning_rate": 4.039192517439767e-05, "loss": 0.1326, "step": 9532 }, { "epoch": 59.21183800623053, "grad_norm": 2.546875, "learning_rate": 4.038998365832385e-05, "loss": 0.1041, "step": 9533 }, { "epoch": 59.218068535825545, "grad_norm": 2.9375, "learning_rate": 4.038804199277943e-05, "loss": 0.2435, "step": 9534 }, { "epoch": 59.22429906542056, "grad_norm": 2.21875, "learning_rate": 4.038610017778327e-05, "loss": 0.2428, "step": 9535 }, { "epoch": 59.230529595015575, "grad_norm": 2.40625, "learning_rate": 4.038415821335423e-05, "loss": 0.1172, "step": 9536 }, { "epoch": 59.23676012461059, "grad_norm": 1.9140625, "learning_rate": 4.038221609951117e-05, "loss": 0.1294, "step": 9537 }, { "epoch": 59.242990654205606, "grad_norm": 3.75, "learning_rate": 4.0380273836272955e-05, "loss": 0.2205, "step": 9538 }, { "epoch": 59.24922118380062, "grad_norm": 2.171875, "learning_rate": 4.037833142365845e-05, "loss": 0.0935, "step": 9539 }, { "epoch": 59.25545171339564, "grad_norm": 2.015625, "learning_rate": 4.03763888616865e-05, "loss": 0.1849, "step": 9540 }, { "epoch": 59.26168224299065, "grad_norm": 1.78125, "learning_rate": 4.0374446150376e-05, "loss": 0.1251, "step": 9541 }, { "epoch": 59.26791277258567, "grad_norm": 1.9765625, "learning_rate": 4.03725032897458e-05, "loss": 0.1399, "step": 9542 }, { "epoch": 59.27414330218068, "grad_norm": 2.46875, "learning_rate": 4.037056027981478e-05, "loss": 0.1889, "step": 9543 }, { "epoch": 59.2803738317757, "grad_norm": 1.859375, "learning_rate": 4.03686171206018e-05, "loss": 0.1173, "step": 9544 }, { "epoch": 59.28660436137071, "grad_norm": 1.8671875, "learning_rate": 4.0366673812125744e-05, "loss": 0.1492, "step": 9545 }, { "epoch": 59.29283489096573, "grad_norm": 3.171875, "learning_rate": 4.0364730354405475e-05, "loss": 0.3055, "step": 9546 }, { "epoch": 59.299065420560744, "grad_norm": 2.609375, "learning_rate": 4.036278674745988e-05, "loss": 0.2729, "step": 9547 }, { "epoch": 59.30529595015577, "grad_norm": 3.5, "learning_rate": 4.036084299130782e-05, "loss": 0.2036, "step": 9548 }, { "epoch": 59.31152647975078, "grad_norm": 1.734375, "learning_rate": 4.0358899085968194e-05, "loss": 0.1475, "step": 9549 }, { "epoch": 59.3177570093458, "grad_norm": 2.765625, "learning_rate": 4.0356955031459864e-05, "loss": 0.2404, "step": 9550 }, { "epoch": 59.32398753894081, "grad_norm": 3.375, "learning_rate": 4.035501082780172e-05, "loss": 0.118, "step": 9551 }, { "epoch": 59.33021806853583, "grad_norm": 1.578125, "learning_rate": 4.035306647501264e-05, "loss": 0.1381, "step": 9552 }, { "epoch": 59.33644859813084, "grad_norm": 2.046875, "learning_rate": 4.03511219731115e-05, "loss": 0.1495, "step": 9553 }, { "epoch": 59.34267912772586, "grad_norm": 2.421875, "learning_rate": 4.034917732211721e-05, "loss": 0.2659, "step": 9554 }, { "epoch": 59.348909657320874, "grad_norm": 2.609375, "learning_rate": 4.034723252204864e-05, "loss": 0.2029, "step": 9555 }, { "epoch": 59.35514018691589, "grad_norm": 1.890625, "learning_rate": 4.0345287572924675e-05, "loss": 0.161, "step": 9556 }, { "epoch": 59.361370716510905, "grad_norm": 1.3671875, "learning_rate": 4.03433424747642e-05, "loss": 0.127, "step": 9557 }, { "epoch": 59.36760124610592, "grad_norm": 1.546875, "learning_rate": 4.034139722758613e-05, "loss": 0.1345, "step": 9558 }, { "epoch": 59.373831775700936, "grad_norm": 3.328125, "learning_rate": 4.033945183140934e-05, "loss": 0.2884, "step": 9559 }, { "epoch": 59.38006230529595, "grad_norm": 2.5625, "learning_rate": 4.0337506286252726e-05, "loss": 0.1708, "step": 9560 }, { "epoch": 59.386292834890966, "grad_norm": 3.3125, "learning_rate": 4.0335560592135185e-05, "loss": 0.2446, "step": 9561 }, { "epoch": 59.39252336448598, "grad_norm": 3.21875, "learning_rate": 4.033361474907561e-05, "loss": 0.245, "step": 9562 }, { "epoch": 59.398753894081, "grad_norm": 2.3125, "learning_rate": 4.033166875709291e-05, "loss": 0.2185, "step": 9563 }, { "epoch": 59.40498442367601, "grad_norm": 2.4375, "learning_rate": 4.032972261620598e-05, "loss": 0.1651, "step": 9564 }, { "epoch": 59.41121495327103, "grad_norm": 2.234375, "learning_rate": 4.0327776326433716e-05, "loss": 0.1209, "step": 9565 }, { "epoch": 59.41744548286604, "grad_norm": 2.515625, "learning_rate": 4.032582988779502e-05, "loss": 0.1548, "step": 9566 }, { "epoch": 59.42367601246106, "grad_norm": 2.421875, "learning_rate": 4.03238833003088e-05, "loss": 0.1251, "step": 9567 }, { "epoch": 59.429906542056074, "grad_norm": 3.09375, "learning_rate": 4.032193656399397e-05, "loss": 0.188, "step": 9568 }, { "epoch": 59.43613707165109, "grad_norm": 2.765625, "learning_rate": 4.0319989678869426e-05, "loss": 0.2485, "step": 9569 }, { "epoch": 59.442367601246104, "grad_norm": 2.25, "learning_rate": 4.031804264495408e-05, "loss": 0.1573, "step": 9570 }, { "epoch": 59.44859813084112, "grad_norm": 2.015625, "learning_rate": 4.031609546226685e-05, "loss": 0.19, "step": 9571 }, { "epoch": 59.454828660436135, "grad_norm": 2.0625, "learning_rate": 4.031414813082662e-05, "loss": 0.2062, "step": 9572 }, { "epoch": 59.46105919003115, "grad_norm": 2.6875, "learning_rate": 4.031220065065233e-05, "loss": 0.2042, "step": 9573 }, { "epoch": 59.467289719626166, "grad_norm": 2.46875, "learning_rate": 4.0310253021762896e-05, "loss": 0.1895, "step": 9574 }, { "epoch": 59.47352024922118, "grad_norm": 2.53125, "learning_rate": 4.030830524417721e-05, "loss": 0.3338, "step": 9575 }, { "epoch": 59.4797507788162, "grad_norm": 3.15625, "learning_rate": 4.030635731791421e-05, "loss": 0.2183, "step": 9576 }, { "epoch": 59.48598130841121, "grad_norm": 3.59375, "learning_rate": 4.030440924299281e-05, "loss": 0.2304, "step": 9577 }, { "epoch": 59.49221183800623, "grad_norm": 1.6328125, "learning_rate": 4.0302461019431934e-05, "loss": 0.1154, "step": 9578 }, { "epoch": 59.49844236760124, "grad_norm": 3.078125, "learning_rate": 4.030051264725049e-05, "loss": 0.1874, "step": 9579 }, { "epoch": 59.504672897196265, "grad_norm": 1.5859375, "learning_rate": 4.029856412646741e-05, "loss": 0.1088, "step": 9580 }, { "epoch": 59.51090342679128, "grad_norm": 3.203125, "learning_rate": 4.029661545710162e-05, "loss": 0.1368, "step": 9581 }, { "epoch": 59.517133956386296, "grad_norm": 2.96875, "learning_rate": 4.029466663917203e-05, "loss": 0.2116, "step": 9582 }, { "epoch": 59.52336448598131, "grad_norm": 2.9375, "learning_rate": 4.029271767269759e-05, "loss": 0.1729, "step": 9583 }, { "epoch": 59.52959501557633, "grad_norm": 3.53125, "learning_rate": 4.029076855769722e-05, "loss": 0.2708, "step": 9584 }, { "epoch": 59.53582554517134, "grad_norm": 2.890625, "learning_rate": 4.028881929418985e-05, "loss": 0.1781, "step": 9585 }, { "epoch": 59.54205607476636, "grad_norm": 1.9453125, "learning_rate": 4.0286869882194414e-05, "loss": 0.1544, "step": 9586 }, { "epoch": 59.54828660436137, "grad_norm": 1.8828125, "learning_rate": 4.028492032172983e-05, "loss": 0.127, "step": 9587 }, { "epoch": 59.55451713395639, "grad_norm": 2.65625, "learning_rate": 4.0282970612815054e-05, "loss": 0.1186, "step": 9588 }, { "epoch": 59.5607476635514, "grad_norm": 3.3125, "learning_rate": 4.028102075546901e-05, "loss": 0.171, "step": 9589 }, { "epoch": 59.56697819314642, "grad_norm": 2.90625, "learning_rate": 4.0279070749710634e-05, "loss": 0.1463, "step": 9590 }, { "epoch": 59.573208722741434, "grad_norm": 3.234375, "learning_rate": 4.027712059555887e-05, "loss": 0.3116, "step": 9591 }, { "epoch": 59.57943925233645, "grad_norm": 2.703125, "learning_rate": 4.027517029303266e-05, "loss": 0.1038, "step": 9592 }, { "epoch": 59.585669781931465, "grad_norm": 2.78125, "learning_rate": 4.0273219842150945e-05, "loss": 0.3505, "step": 9593 }, { "epoch": 59.59190031152648, "grad_norm": 2.046875, "learning_rate": 4.0271269242932665e-05, "loss": 0.1672, "step": 9594 }, { "epoch": 59.598130841121495, "grad_norm": 2.015625, "learning_rate": 4.026931849539676e-05, "loss": 0.1086, "step": 9595 }, { "epoch": 59.60436137071651, "grad_norm": 2.65625, "learning_rate": 4.0267367599562185e-05, "loss": 0.1889, "step": 9596 }, { "epoch": 59.610591900311526, "grad_norm": 2.296875, "learning_rate": 4.026541655544788e-05, "loss": 0.1564, "step": 9597 }, { "epoch": 59.61682242990654, "grad_norm": 2.015625, "learning_rate": 4.02634653630728e-05, "loss": 0.1541, "step": 9598 }, { "epoch": 59.62305295950156, "grad_norm": 1.3203125, "learning_rate": 4.026151402245589e-05, "loss": 0.1319, "step": 9599 }, { "epoch": 59.62928348909657, "grad_norm": 1.9765625, "learning_rate": 4.025956253361611e-05, "loss": 0.1331, "step": 9600 }, { "epoch": 59.63551401869159, "grad_norm": 2.09375, "learning_rate": 4.0257610896572414e-05, "loss": 0.1291, "step": 9601 }, { "epoch": 59.6417445482866, "grad_norm": 2.734375, "learning_rate": 4.0255659111343744e-05, "loss": 0.1227, "step": 9602 }, { "epoch": 59.64797507788162, "grad_norm": 1.546875, "learning_rate": 4.025370717794906e-05, "loss": 0.1045, "step": 9603 }, { "epoch": 59.654205607476634, "grad_norm": 1.2734375, "learning_rate": 4.025175509640733e-05, "loss": 0.0997, "step": 9604 }, { "epoch": 59.66043613707165, "grad_norm": 2.125, "learning_rate": 4.024980286673751e-05, "loss": 0.1656, "step": 9605 }, { "epoch": 59.666666666666664, "grad_norm": 2.78125, "learning_rate": 4.0247850488958545e-05, "loss": 0.2455, "step": 9606 }, { "epoch": 59.67289719626168, "grad_norm": 3.296875, "learning_rate": 4.024589796308941e-05, "loss": 0.2018, "step": 9607 }, { "epoch": 59.679127725856695, "grad_norm": 3.875, "learning_rate": 4.024394528914907e-05, "loss": 0.1681, "step": 9608 }, { "epoch": 59.68535825545171, "grad_norm": 3.09375, "learning_rate": 4.024199246715649e-05, "loss": 0.2372, "step": 9609 }, { "epoch": 59.691588785046726, "grad_norm": 2.078125, "learning_rate": 4.0240039497130624e-05, "loss": 0.1555, "step": 9610 }, { "epoch": 59.69781931464174, "grad_norm": 1.4921875, "learning_rate": 4.023808637909045e-05, "loss": 0.0987, "step": 9611 }, { "epoch": 59.70404984423676, "grad_norm": 2.109375, "learning_rate": 4.023613311305493e-05, "loss": 0.106, "step": 9612 }, { "epoch": 59.71028037383178, "grad_norm": 4.09375, "learning_rate": 4.023417969904305e-05, "loss": 0.3621, "step": 9613 }, { "epoch": 59.716510903426794, "grad_norm": 2.234375, "learning_rate": 4.0232226137073765e-05, "loss": 0.1048, "step": 9614 }, { "epoch": 59.72274143302181, "grad_norm": 3.25, "learning_rate": 4.023027242716606e-05, "loss": 0.1975, "step": 9615 }, { "epoch": 59.728971962616825, "grad_norm": 1.65625, "learning_rate": 4.02283185693389e-05, "loss": 0.1249, "step": 9616 }, { "epoch": 59.73520249221184, "grad_norm": 2.640625, "learning_rate": 4.0226364563611264e-05, "loss": 0.1825, "step": 9617 }, { "epoch": 59.741433021806856, "grad_norm": 2.21875, "learning_rate": 4.022441041000213e-05, "loss": 0.1912, "step": 9618 }, { "epoch": 59.74766355140187, "grad_norm": 2.546875, "learning_rate": 4.022245610853049e-05, "loss": 0.1096, "step": 9619 }, { "epoch": 59.753894080996886, "grad_norm": 1.3828125, "learning_rate": 4.0220501659215306e-05, "loss": 0.1267, "step": 9620 }, { "epoch": 59.7601246105919, "grad_norm": 2.453125, "learning_rate": 4.0218547062075565e-05, "loss": 0.2289, "step": 9621 }, { "epoch": 59.76635514018692, "grad_norm": 2.734375, "learning_rate": 4.021659231713026e-05, "loss": 0.3083, "step": 9622 }, { "epoch": 59.77258566978193, "grad_norm": 2.6875, "learning_rate": 4.0214637424398364e-05, "loss": 0.1831, "step": 9623 }, { "epoch": 59.77881619937695, "grad_norm": 1.8515625, "learning_rate": 4.021268238389887e-05, "loss": 0.1574, "step": 9624 }, { "epoch": 59.78504672897196, "grad_norm": 2.171875, "learning_rate": 4.0210727195650765e-05, "loss": 0.2003, "step": 9625 }, { "epoch": 59.79127725856698, "grad_norm": 1.5, "learning_rate": 4.020877185967303e-05, "loss": 0.1311, "step": 9626 }, { "epoch": 59.797507788161994, "grad_norm": 3.390625, "learning_rate": 4.020681637598467e-05, "loss": 0.1729, "step": 9627 }, { "epoch": 59.80373831775701, "grad_norm": 3.03125, "learning_rate": 4.020486074460467e-05, "loss": 0.2739, "step": 9628 }, { "epoch": 59.809968847352025, "grad_norm": 2.90625, "learning_rate": 4.020290496555202e-05, "loss": 0.2126, "step": 9629 }, { "epoch": 59.81619937694704, "grad_norm": 3.6875, "learning_rate": 4.0200949038845714e-05, "loss": 0.1052, "step": 9630 }, { "epoch": 59.822429906542055, "grad_norm": 3.796875, "learning_rate": 4.019899296450476e-05, "loss": 0.2467, "step": 9631 }, { "epoch": 59.82866043613707, "grad_norm": 1.71875, "learning_rate": 4.019703674254815e-05, "loss": 0.1222, "step": 9632 }, { "epoch": 59.834890965732086, "grad_norm": 2.1875, "learning_rate": 4.0195080372994864e-05, "loss": 0.1517, "step": 9633 }, { "epoch": 59.8411214953271, "grad_norm": 1.578125, "learning_rate": 4.019312385586393e-05, "loss": 0.1166, "step": 9634 }, { "epoch": 59.84735202492212, "grad_norm": 2.15625, "learning_rate": 4.019116719117435e-05, "loss": 0.1538, "step": 9635 }, { "epoch": 59.85358255451713, "grad_norm": 4.53125, "learning_rate": 4.018921037894511e-05, "loss": 0.1055, "step": 9636 }, { "epoch": 59.85981308411215, "grad_norm": 3.375, "learning_rate": 4.018725341919522e-05, "loss": 0.1254, "step": 9637 }, { "epoch": 59.86604361370716, "grad_norm": 4.21875, "learning_rate": 4.018529631194369e-05, "loss": 0.1449, "step": 9638 }, { "epoch": 59.87227414330218, "grad_norm": 3.0, "learning_rate": 4.018333905720953e-05, "loss": 0.1874, "step": 9639 }, { "epoch": 59.87850467289719, "grad_norm": 3.171875, "learning_rate": 4.018138165501174e-05, "loss": 0.2065, "step": 9640 }, { "epoch": 59.88473520249221, "grad_norm": 2.421875, "learning_rate": 4.017942410536934e-05, "loss": 0.2026, "step": 9641 }, { "epoch": 59.890965732087224, "grad_norm": 3.265625, "learning_rate": 4.017746640830134e-05, "loss": 0.1794, "step": 9642 }, { "epoch": 59.89719626168224, "grad_norm": 2.65625, "learning_rate": 4.0175508563826746e-05, "loss": 0.2402, "step": 9643 }, { "epoch": 59.90342679127726, "grad_norm": 3.484375, "learning_rate": 4.017355057196459e-05, "loss": 0.2463, "step": 9644 }, { "epoch": 59.90965732087228, "grad_norm": 3.1875, "learning_rate": 4.017159243273387e-05, "loss": 0.1702, "step": 9645 }, { "epoch": 59.91588785046729, "grad_norm": 2.796875, "learning_rate": 4.016963414615361e-05, "loss": 0.2027, "step": 9646 }, { "epoch": 59.92211838006231, "grad_norm": 1.828125, "learning_rate": 4.016767571224284e-05, "loss": 0.1481, "step": 9647 }, { "epoch": 59.92834890965732, "grad_norm": 3.03125, "learning_rate": 4.016571713102056e-05, "loss": 0.2552, "step": 9648 }, { "epoch": 59.93457943925234, "grad_norm": 2.734375, "learning_rate": 4.0163758402505816e-05, "loss": 0.2819, "step": 9649 }, { "epoch": 59.940809968847354, "grad_norm": 3.28125, "learning_rate": 4.0161799526717616e-05, "loss": 0.2345, "step": 9650 }, { "epoch": 59.94704049844237, "grad_norm": 2.390625, "learning_rate": 4.015984050367499e-05, "loss": 0.1499, "step": 9651 }, { "epoch": 59.953271028037385, "grad_norm": 3.59375, "learning_rate": 4.015788133339695e-05, "loss": 0.2615, "step": 9652 }, { "epoch": 59.9595015576324, "grad_norm": 2.4375, "learning_rate": 4.015592201590255e-05, "loss": 0.1521, "step": 9653 }, { "epoch": 59.965732087227416, "grad_norm": 1.40625, "learning_rate": 4.01539625512108e-05, "loss": 0.0999, "step": 9654 }, { "epoch": 59.97196261682243, "grad_norm": 1.84375, "learning_rate": 4.015200293934073e-05, "loss": 0.1612, "step": 9655 }, { "epoch": 59.978193146417446, "grad_norm": 2.40625, "learning_rate": 4.0150043180311384e-05, "loss": 0.1829, "step": 9656 }, { "epoch": 59.98442367601246, "grad_norm": 1.703125, "learning_rate": 4.0148083274141796e-05, "loss": 0.147, "step": 9657 }, { "epoch": 59.99065420560748, "grad_norm": 1.7734375, "learning_rate": 4.014612322085099e-05, "loss": 0.1603, "step": 9658 }, { "epoch": 59.99688473520249, "grad_norm": 2.703125, "learning_rate": 4.014416302045801e-05, "loss": 0.1541, "step": 9659 }, { "epoch": 60.0, "grad_norm": 0.80859375, "learning_rate": 4.014220267298189e-05, "loss": 0.0571, "step": 9660 }, { "epoch": 60.006230529595015, "grad_norm": 1.796875, "learning_rate": 4.014024217844167e-05, "loss": 0.1197, "step": 9661 }, { "epoch": 60.01246105919003, "grad_norm": 2.078125, "learning_rate": 4.013828153685639e-05, "loss": 0.1095, "step": 9662 }, { "epoch": 60.018691588785046, "grad_norm": 1.515625, "learning_rate": 4.01363207482451e-05, "loss": 0.157, "step": 9663 }, { "epoch": 60.02492211838006, "grad_norm": 3.140625, "learning_rate": 4.013435981262683e-05, "loss": 0.1992, "step": 9664 }, { "epoch": 60.03115264797508, "grad_norm": 2.3125, "learning_rate": 4.0132398730020634e-05, "loss": 0.1085, "step": 9665 }, { "epoch": 60.03738317757009, "grad_norm": 2.0, "learning_rate": 4.0130437500445563e-05, "loss": 0.1135, "step": 9666 }, { "epoch": 60.04361370716511, "grad_norm": 3.140625, "learning_rate": 4.012847612392065e-05, "loss": 0.2569, "step": 9667 }, { "epoch": 60.04984423676012, "grad_norm": 1.9296875, "learning_rate": 4.012651460046495e-05, "loss": 0.1808, "step": 9668 }, { "epoch": 60.05607476635514, "grad_norm": 3.671875, "learning_rate": 4.0124552930097525e-05, "loss": 0.1227, "step": 9669 }, { "epoch": 60.06230529595015, "grad_norm": 3.28125, "learning_rate": 4.0122591112837416e-05, "loss": 0.1859, "step": 9670 }, { "epoch": 60.06853582554517, "grad_norm": 1.8828125, "learning_rate": 4.0120629148703684e-05, "loss": 0.1506, "step": 9671 }, { "epoch": 60.074766355140184, "grad_norm": 4.0625, "learning_rate": 4.011866703771537e-05, "loss": 0.3366, "step": 9672 }, { "epoch": 60.0809968847352, "grad_norm": 2.65625, "learning_rate": 4.0116704779891545e-05, "loss": 0.1509, "step": 9673 }, { "epoch": 60.087227414330215, "grad_norm": 2.296875, "learning_rate": 4.011474237525126e-05, "loss": 0.1565, "step": 9674 }, { "epoch": 60.09345794392523, "grad_norm": 3.8125, "learning_rate": 4.0112779823813574e-05, "loss": 0.3915, "step": 9675 }, { "epoch": 60.099688473520246, "grad_norm": 1.7578125, "learning_rate": 4.011081712559755e-05, "loss": 0.1973, "step": 9676 }, { "epoch": 60.10591900311527, "grad_norm": 1.5, "learning_rate": 4.010885428062225e-05, "loss": 0.1172, "step": 9677 }, { "epoch": 60.11214953271028, "grad_norm": 2.03125, "learning_rate": 4.0106891288906746e-05, "loss": 0.1286, "step": 9678 }, { "epoch": 60.1183800623053, "grad_norm": 1.328125, "learning_rate": 4.010492815047009e-05, "loss": 0.1282, "step": 9679 }, { "epoch": 60.124610591900314, "grad_norm": 2.109375, "learning_rate": 4.010296486533134e-05, "loss": 0.2518, "step": 9680 }, { "epoch": 60.13084112149533, "grad_norm": 1.5, "learning_rate": 4.010100143350959e-05, "loss": 0.1414, "step": 9681 }, { "epoch": 60.137071651090345, "grad_norm": 2.59375, "learning_rate": 4.009903785502389e-05, "loss": 0.1979, "step": 9682 }, { "epoch": 60.14330218068536, "grad_norm": 2.171875, "learning_rate": 4.009707412989331e-05, "loss": 0.1517, "step": 9683 }, { "epoch": 60.149532710280376, "grad_norm": 1.7890625, "learning_rate": 4.009511025813694e-05, "loss": 0.1333, "step": 9684 }, { "epoch": 60.15576323987539, "grad_norm": 2.125, "learning_rate": 4.0093146239773834e-05, "loss": 0.2144, "step": 9685 }, { "epoch": 60.161993769470406, "grad_norm": 2.78125, "learning_rate": 4.009118207482307e-05, "loss": 0.2304, "step": 9686 }, { "epoch": 60.16822429906542, "grad_norm": 2.515625, "learning_rate": 4.008921776330374e-05, "loss": 0.2293, "step": 9687 }, { "epoch": 60.17445482866044, "grad_norm": 3.390625, "learning_rate": 4.0087253305234905e-05, "loss": 0.1964, "step": 9688 }, { "epoch": 60.18068535825545, "grad_norm": 3.75, "learning_rate": 4.0085288700635657e-05, "loss": 0.3213, "step": 9689 }, { "epoch": 60.18691588785047, "grad_norm": 2.671875, "learning_rate": 4.0083323949525065e-05, "loss": 0.2377, "step": 9690 }, { "epoch": 60.19314641744548, "grad_norm": 2.703125, "learning_rate": 4.0081359051922205e-05, "loss": 0.1586, "step": 9691 }, { "epoch": 60.1993769470405, "grad_norm": 2.875, "learning_rate": 4.0079394007846186e-05, "loss": 0.2978, "step": 9692 }, { "epoch": 60.205607476635514, "grad_norm": 1.5546875, "learning_rate": 4.007742881731608e-05, "loss": 0.1194, "step": 9693 }, { "epoch": 60.21183800623053, "grad_norm": 3.03125, "learning_rate": 4.007546348035095e-05, "loss": 0.3783, "step": 9694 }, { "epoch": 60.218068535825545, "grad_norm": 2.375, "learning_rate": 4.0073497996969936e-05, "loss": 0.1577, "step": 9695 }, { "epoch": 60.22429906542056, "grad_norm": 1.7265625, "learning_rate": 4.0071532367192074e-05, "loss": 0.1409, "step": 9696 }, { "epoch": 60.230529595015575, "grad_norm": 2.859375, "learning_rate": 4.006956659103648e-05, "loss": 0.1861, "step": 9697 }, { "epoch": 60.23676012461059, "grad_norm": 3.015625, "learning_rate": 4.006760066852225e-05, "loss": 0.2673, "step": 9698 }, { "epoch": 60.242990654205606, "grad_norm": 2.484375, "learning_rate": 4.0065634599668465e-05, "loss": 0.3452, "step": 9699 }, { "epoch": 60.24922118380062, "grad_norm": 2.90625, "learning_rate": 4.006366838449422e-05, "loss": 0.3073, "step": 9700 }, { "epoch": 60.25545171339564, "grad_norm": 2.59375, "learning_rate": 4.006170202301863e-05, "loss": 0.2771, "step": 9701 }, { "epoch": 60.26168224299065, "grad_norm": 3.65625, "learning_rate": 4.005973551526077e-05, "loss": 0.1971, "step": 9702 }, { "epoch": 60.26791277258567, "grad_norm": 3.140625, "learning_rate": 4.005776886123975e-05, "loss": 0.2043, "step": 9703 }, { "epoch": 60.27414330218068, "grad_norm": 3.40625, "learning_rate": 4.0055802060974665e-05, "loss": 0.3273, "step": 9704 }, { "epoch": 60.2803738317757, "grad_norm": 1.625, "learning_rate": 4.005383511448463e-05, "loss": 0.1061, "step": 9705 }, { "epoch": 60.28660436137071, "grad_norm": 1.7734375, "learning_rate": 4.005186802178873e-05, "loss": 0.1247, "step": 9706 }, { "epoch": 60.29283489096573, "grad_norm": 1.828125, "learning_rate": 4.0049900782906086e-05, "loss": 0.115, "step": 9707 }, { "epoch": 60.299065420560744, "grad_norm": 3.328125, "learning_rate": 4.00479333978558e-05, "loss": 0.1567, "step": 9708 }, { "epoch": 60.30529595015577, "grad_norm": 5.09375, "learning_rate": 4.0045965866656966e-05, "loss": 0.1594, "step": 9709 }, { "epoch": 60.31152647975078, "grad_norm": 3.8125, "learning_rate": 4.004399818932871e-05, "loss": 0.1565, "step": 9710 }, { "epoch": 60.3177570093458, "grad_norm": 3.140625, "learning_rate": 4.0042030365890135e-05, "loss": 0.2188, "step": 9711 }, { "epoch": 60.32398753894081, "grad_norm": 1.9609375, "learning_rate": 4.004006239636036e-05, "loss": 0.1634, "step": 9712 }, { "epoch": 60.33021806853583, "grad_norm": 2.03125, "learning_rate": 4.003809428075848e-05, "loss": 0.1101, "step": 9713 }, { "epoch": 60.33644859813084, "grad_norm": 3.28125, "learning_rate": 4.0036126019103634e-05, "loss": 0.2157, "step": 9714 }, { "epoch": 60.34267912772586, "grad_norm": 2.34375, "learning_rate": 4.0034157611414917e-05, "loss": 0.1577, "step": 9715 }, { "epoch": 60.348909657320874, "grad_norm": 3.03125, "learning_rate": 4.003218905771146e-05, "loss": 0.1605, "step": 9716 }, { "epoch": 60.35514018691589, "grad_norm": 3.03125, "learning_rate": 4.003022035801238e-05, "loss": 0.1291, "step": 9717 }, { "epoch": 60.361370716510905, "grad_norm": 1.4921875, "learning_rate": 4.002825151233679e-05, "loss": 0.1307, "step": 9718 }, { "epoch": 60.36760124610592, "grad_norm": 2.40625, "learning_rate": 4.0026282520703825e-05, "loss": 0.2669, "step": 9719 }, { "epoch": 60.373831775700936, "grad_norm": 1.703125, "learning_rate": 4.00243133831326e-05, "loss": 0.1651, "step": 9720 }, { "epoch": 60.38006230529595, "grad_norm": 2.4375, "learning_rate": 4.002234409964224e-05, "loss": 0.2027, "step": 9721 }, { "epoch": 60.386292834890966, "grad_norm": 2.203125, "learning_rate": 4.002037467025186e-05, "loss": 0.1558, "step": 9722 }, { "epoch": 60.39252336448598, "grad_norm": 3.6875, "learning_rate": 4.001840509498061e-05, "loss": 0.3799, "step": 9723 }, { "epoch": 60.398753894081, "grad_norm": 2.359375, "learning_rate": 4.001643537384761e-05, "loss": 0.2068, "step": 9724 }, { "epoch": 60.40498442367601, "grad_norm": 2.625, "learning_rate": 4.001446550687198e-05, "loss": 0.1253, "step": 9725 }, { "epoch": 60.41121495327103, "grad_norm": 1.5390625, "learning_rate": 4.001249549407286e-05, "loss": 0.1097, "step": 9726 }, { "epoch": 60.41744548286604, "grad_norm": 2.375, "learning_rate": 4.0010525335469385e-05, "loss": 0.1874, "step": 9727 }, { "epoch": 60.42367601246106, "grad_norm": 3.59375, "learning_rate": 4.000855503108069e-05, "loss": 0.2367, "step": 9728 }, { "epoch": 60.429906542056074, "grad_norm": 2.96875, "learning_rate": 4.00065845809259e-05, "loss": 0.2272, "step": 9729 }, { "epoch": 60.43613707165109, "grad_norm": 1.9921875, "learning_rate": 4.000461398502418e-05, "loss": 0.1164, "step": 9730 }, { "epoch": 60.442367601246104, "grad_norm": 1.46875, "learning_rate": 4.0002643243394626e-05, "loss": 0.1492, "step": 9731 }, { "epoch": 60.44859813084112, "grad_norm": 1.9609375, "learning_rate": 4.000067235605641e-05, "loss": 0.1778, "step": 9732 }, { "epoch": 60.454828660436135, "grad_norm": 2.984375, "learning_rate": 3.999870132302867e-05, "loss": 0.2146, "step": 9733 }, { "epoch": 60.46105919003115, "grad_norm": 1.453125, "learning_rate": 3.999673014433054e-05, "loss": 0.106, "step": 9734 }, { "epoch": 60.467289719626166, "grad_norm": 2.078125, "learning_rate": 3.999475881998117e-05, "loss": 0.1424, "step": 9735 }, { "epoch": 60.47352024922118, "grad_norm": 2.859375, "learning_rate": 3.999278734999972e-05, "loss": 0.2261, "step": 9736 }, { "epoch": 60.4797507788162, "grad_norm": 1.6640625, "learning_rate": 3.99908157344053e-05, "loss": 0.1154, "step": 9737 }, { "epoch": 60.48598130841121, "grad_norm": 1.703125, "learning_rate": 3.998884397321709e-05, "loss": 0.1603, "step": 9738 }, { "epoch": 60.49221183800623, "grad_norm": 1.2421875, "learning_rate": 3.9986872066454226e-05, "loss": 0.0991, "step": 9739 }, { "epoch": 60.49844236760124, "grad_norm": 2.296875, "learning_rate": 3.998490001413587e-05, "loss": 0.1757, "step": 9740 }, { "epoch": 60.504672897196265, "grad_norm": 3.921875, "learning_rate": 3.9982927816281166e-05, "loss": 0.1954, "step": 9741 }, { "epoch": 60.51090342679128, "grad_norm": 2.953125, "learning_rate": 3.9980955472909275e-05, "loss": 0.191, "step": 9742 }, { "epoch": 60.517133956386296, "grad_norm": 2.484375, "learning_rate": 3.9978982984039356e-05, "loss": 0.1812, "step": 9743 }, { "epoch": 60.52336448598131, "grad_norm": 3.6875, "learning_rate": 3.9977010349690547e-05, "loss": 0.3071, "step": 9744 }, { "epoch": 60.52959501557633, "grad_norm": 3.09375, "learning_rate": 3.997503756988203e-05, "loss": 0.3164, "step": 9745 }, { "epoch": 60.53582554517134, "grad_norm": 0.87890625, "learning_rate": 3.9973064644632954e-05, "loss": 0.0998, "step": 9746 }, { "epoch": 60.54205607476636, "grad_norm": 2.21875, "learning_rate": 3.997109157396248e-05, "loss": 0.1917, "step": 9747 }, { "epoch": 60.54828660436137, "grad_norm": 2.59375, "learning_rate": 3.996911835788977e-05, "loss": 0.1336, "step": 9748 }, { "epoch": 60.55451713395639, "grad_norm": 3.640625, "learning_rate": 3.996714499643399e-05, "loss": 0.1984, "step": 9749 }, { "epoch": 60.5607476635514, "grad_norm": 2.53125, "learning_rate": 3.9965171489614306e-05, "loss": 0.2186, "step": 9750 }, { "epoch": 60.56697819314642, "grad_norm": 2.78125, "learning_rate": 3.996319783744988e-05, "loss": 0.2229, "step": 9751 }, { "epoch": 60.573208722741434, "grad_norm": 1.4921875, "learning_rate": 3.9961224039959896e-05, "loss": 0.1236, "step": 9752 }, { "epoch": 60.57943925233645, "grad_norm": 2.765625, "learning_rate": 3.99592500971635e-05, "loss": 0.1646, "step": 9753 }, { "epoch": 60.585669781931465, "grad_norm": 2.3125, "learning_rate": 3.9957276009079894e-05, "loss": 0.1602, "step": 9754 }, { "epoch": 60.59190031152648, "grad_norm": 2.71875, "learning_rate": 3.995530177572822e-05, "loss": 0.253, "step": 9755 }, { "epoch": 60.598130841121495, "grad_norm": 3.40625, "learning_rate": 3.9953327397127673e-05, "loss": 0.1948, "step": 9756 }, { "epoch": 60.60436137071651, "grad_norm": 2.265625, "learning_rate": 3.995135287329741e-05, "loss": 0.1896, "step": 9757 }, { "epoch": 60.610591900311526, "grad_norm": 2.28125, "learning_rate": 3.994937820425663e-05, "loss": 0.218, "step": 9758 }, { "epoch": 60.61682242990654, "grad_norm": 1.5078125, "learning_rate": 3.9947403390024504e-05, "loss": 0.1296, "step": 9759 }, { "epoch": 60.62305295950156, "grad_norm": 1.109375, "learning_rate": 3.99454284306202e-05, "loss": 0.1286, "step": 9760 }, { "epoch": 60.62928348909657, "grad_norm": 2.0, "learning_rate": 3.9943453326062907e-05, "loss": 0.229, "step": 9761 }, { "epoch": 60.63551401869159, "grad_norm": 2.1875, "learning_rate": 3.9941478076371814e-05, "loss": 0.157, "step": 9762 }, { "epoch": 60.6417445482866, "grad_norm": 3.0625, "learning_rate": 3.9939502681566096e-05, "loss": 0.2424, "step": 9763 }, { "epoch": 60.64797507788162, "grad_norm": 2.625, "learning_rate": 3.993752714166494e-05, "loss": 0.2956, "step": 9764 }, { "epoch": 60.654205607476634, "grad_norm": 2.4375, "learning_rate": 3.9935551456687534e-05, "loss": 0.2001, "step": 9765 }, { "epoch": 60.66043613707165, "grad_norm": 1.0859375, "learning_rate": 3.993357562665307e-05, "loss": 0.1211, "step": 9766 }, { "epoch": 60.666666666666664, "grad_norm": 1.859375, "learning_rate": 3.993159965158073e-05, "loss": 0.1496, "step": 9767 }, { "epoch": 60.67289719626168, "grad_norm": 2.734375, "learning_rate": 3.992962353148972e-05, "loss": 0.1884, "step": 9768 }, { "epoch": 60.679127725856695, "grad_norm": 1.4921875, "learning_rate": 3.9927647266399216e-05, "loss": 0.1156, "step": 9769 }, { "epoch": 60.68535825545171, "grad_norm": 2.21875, "learning_rate": 3.992567085632841e-05, "loss": 0.1457, "step": 9770 }, { "epoch": 60.691588785046726, "grad_norm": 2.65625, "learning_rate": 3.992369430129651e-05, "loss": 0.1902, "step": 9771 }, { "epoch": 60.69781931464174, "grad_norm": 2.09375, "learning_rate": 3.992171760132272e-05, "loss": 0.2373, "step": 9772 }, { "epoch": 60.70404984423676, "grad_norm": 3.265625, "learning_rate": 3.991974075642621e-05, "loss": 0.2091, "step": 9773 }, { "epoch": 60.71028037383178, "grad_norm": 2.625, "learning_rate": 3.99177637666262e-05, "loss": 0.2367, "step": 9774 }, { "epoch": 60.716510903426794, "grad_norm": 1.71875, "learning_rate": 3.991578663194189e-05, "loss": 0.1604, "step": 9775 }, { "epoch": 60.72274143302181, "grad_norm": 1.9765625, "learning_rate": 3.9913809352392474e-05, "loss": 0.1811, "step": 9776 }, { "epoch": 60.728971962616825, "grad_norm": 1.9453125, "learning_rate": 3.991183192799716e-05, "loss": 0.1364, "step": 9777 }, { "epoch": 60.73520249221184, "grad_norm": 2.9375, "learning_rate": 3.990985435877515e-05, "loss": 0.1778, "step": 9778 }, { "epoch": 60.741433021806856, "grad_norm": 3.546875, "learning_rate": 3.990787664474567e-05, "loss": 0.1873, "step": 9779 }, { "epoch": 60.74766355140187, "grad_norm": 1.578125, "learning_rate": 3.99058987859279e-05, "loss": 0.1377, "step": 9780 }, { "epoch": 60.753894080996886, "grad_norm": 2.0, "learning_rate": 3.990392078234106e-05, "loss": 0.1382, "step": 9781 }, { "epoch": 60.7601246105919, "grad_norm": 3.4375, "learning_rate": 3.9901942634004364e-05, "loss": 0.3614, "step": 9782 }, { "epoch": 60.76635514018692, "grad_norm": 3.140625, "learning_rate": 3.9899964340937025e-05, "loss": 0.2224, "step": 9783 }, { "epoch": 60.77258566978193, "grad_norm": 3.21875, "learning_rate": 3.9897985903158254e-05, "loss": 0.3052, "step": 9784 }, { "epoch": 60.77881619937695, "grad_norm": 1.03125, "learning_rate": 3.989600732068726e-05, "loss": 0.1128, "step": 9785 }, { "epoch": 60.78504672897196, "grad_norm": 1.34375, "learning_rate": 3.9894028593543265e-05, "loss": 0.1209, "step": 9786 }, { "epoch": 60.79127725856698, "grad_norm": 1.8203125, "learning_rate": 3.9892049721745496e-05, "loss": 0.1451, "step": 9787 }, { "epoch": 60.797507788161994, "grad_norm": 2.609375, "learning_rate": 3.989007070531316e-05, "loss": 0.1251, "step": 9788 }, { "epoch": 60.80373831775701, "grad_norm": 1.828125, "learning_rate": 3.9888091544265476e-05, "loss": 0.1192, "step": 9789 }, { "epoch": 60.809968847352025, "grad_norm": 3.03125, "learning_rate": 3.9886112238621674e-05, "loss": 0.2598, "step": 9790 }, { "epoch": 60.81619937694704, "grad_norm": 2.171875, "learning_rate": 3.9884132788400975e-05, "loss": 0.2085, "step": 9791 }, { "epoch": 60.822429906542055, "grad_norm": 1.34375, "learning_rate": 3.98821531936226e-05, "loss": 0.1039, "step": 9792 }, { "epoch": 60.82866043613707, "grad_norm": 3.546875, "learning_rate": 3.988017345430578e-05, "loss": 0.1865, "step": 9793 }, { "epoch": 60.834890965732086, "grad_norm": 3.140625, "learning_rate": 3.9878193570469746e-05, "loss": 0.2241, "step": 9794 }, { "epoch": 60.8411214953271, "grad_norm": 2.703125, "learning_rate": 3.987621354213372e-05, "loss": 0.1164, "step": 9795 }, { "epoch": 60.84735202492212, "grad_norm": 3.453125, "learning_rate": 3.9874233369316925e-05, "loss": 0.2145, "step": 9796 }, { "epoch": 60.85358255451713, "grad_norm": 2.4375, "learning_rate": 3.9872253052038613e-05, "loss": 0.1399, "step": 9797 }, { "epoch": 60.85981308411215, "grad_norm": 2.25, "learning_rate": 3.9870272590318e-05, "loss": 0.1245, "step": 9798 }, { "epoch": 60.86604361370716, "grad_norm": 1.8125, "learning_rate": 3.986829198417433e-05, "loss": 0.1329, "step": 9799 }, { "epoch": 60.87227414330218, "grad_norm": 3.421875, "learning_rate": 3.9866311233626844e-05, "loss": 0.193, "step": 9800 }, { "epoch": 60.87850467289719, "grad_norm": 2.578125, "learning_rate": 3.9864330338694756e-05, "loss": 0.2008, "step": 9801 }, { "epoch": 60.88473520249221, "grad_norm": 3.109375, "learning_rate": 3.986234929939733e-05, "loss": 0.2371, "step": 9802 }, { "epoch": 60.890965732087224, "grad_norm": 2.78125, "learning_rate": 3.9860368115753796e-05, "loss": 0.2121, "step": 9803 }, { "epoch": 60.89719626168224, "grad_norm": 3.8125, "learning_rate": 3.98583867877834e-05, "loss": 0.1907, "step": 9804 }, { "epoch": 60.90342679127726, "grad_norm": 1.765625, "learning_rate": 3.9856405315505374e-05, "loss": 0.1315, "step": 9805 }, { "epoch": 60.90965732087228, "grad_norm": 2.734375, "learning_rate": 3.985442369893897e-05, "loss": 0.2076, "step": 9806 }, { "epoch": 60.91588785046729, "grad_norm": 3.6875, "learning_rate": 3.985244193810343e-05, "loss": 0.3422, "step": 9807 }, { "epoch": 60.92211838006231, "grad_norm": 1.8984375, "learning_rate": 3.985046003301801e-05, "loss": 0.1245, "step": 9808 }, { "epoch": 60.92834890965732, "grad_norm": 2.703125, "learning_rate": 3.984847798370196e-05, "loss": 0.2329, "step": 9809 }, { "epoch": 60.93457943925234, "grad_norm": 3.296875, "learning_rate": 3.984649579017452e-05, "loss": 0.3295, "step": 9810 }, { "epoch": 60.940809968847354, "grad_norm": 1.4453125, "learning_rate": 3.984451345245494e-05, "loss": 0.1268, "step": 9811 }, { "epoch": 60.94704049844237, "grad_norm": 2.734375, "learning_rate": 3.984253097056248e-05, "loss": 0.2333, "step": 9812 }, { "epoch": 60.953271028037385, "grad_norm": 2.59375, "learning_rate": 3.984054834451639e-05, "loss": 0.2873, "step": 9813 }, { "epoch": 60.9595015576324, "grad_norm": 1.7421875, "learning_rate": 3.983856557433594e-05, "loss": 0.1521, "step": 9814 }, { "epoch": 60.965732087227416, "grad_norm": 1.4453125, "learning_rate": 3.983658266004036e-05, "loss": 0.132, "step": 9815 }, { "epoch": 60.97196261682243, "grad_norm": 2.921875, "learning_rate": 3.9834599601648936e-05, "loss": 0.2148, "step": 9816 }, { "epoch": 60.978193146417446, "grad_norm": 2.234375, "learning_rate": 3.9832616399180905e-05, "loss": 0.2777, "step": 9817 }, { "epoch": 60.98442367601246, "grad_norm": 3.640625, "learning_rate": 3.983063305265554e-05, "loss": 0.2555, "step": 9818 }, { "epoch": 60.99065420560748, "grad_norm": 1.453125, "learning_rate": 3.9828649562092115e-05, "loss": 0.1045, "step": 9819 }, { "epoch": 60.99688473520249, "grad_norm": 1.2421875, "learning_rate": 3.9826665927509866e-05, "loss": 0.1222, "step": 9820 }, { "epoch": 61.0, "grad_norm": 0.98046875, "learning_rate": 3.982468214892808e-05, "loss": 0.0677, "step": 9821 }, { "epoch": 61.006230529595015, "grad_norm": 3.25, "learning_rate": 3.982269822636602e-05, "loss": 0.2575, "step": 9822 }, { "epoch": 61.01246105919003, "grad_norm": 2.40625, "learning_rate": 3.9820714159842945e-05, "loss": 0.1881, "step": 9823 }, { "epoch": 61.018691588785046, "grad_norm": 2.34375, "learning_rate": 3.9818729949378134e-05, "loss": 0.167, "step": 9824 }, { "epoch": 61.02492211838006, "grad_norm": 1.6171875, "learning_rate": 3.981674559499086e-05, "loss": 0.1363, "step": 9825 }, { "epoch": 61.03115264797508, "grad_norm": 2.75, "learning_rate": 3.981476109670038e-05, "loss": 0.2299, "step": 9826 }, { "epoch": 61.03738317757009, "grad_norm": 3.421875, "learning_rate": 3.9812776454525984e-05, "loss": 0.2743, "step": 9827 }, { "epoch": 61.04361370716511, "grad_norm": 3.0625, "learning_rate": 3.9810791668486944e-05, "loss": 0.1769, "step": 9828 }, { "epoch": 61.04984423676012, "grad_norm": 2.875, "learning_rate": 3.980880673860254e-05, "loss": 0.2048, "step": 9829 }, { "epoch": 61.05607476635514, "grad_norm": 2.0, "learning_rate": 3.9806821664892035e-05, "loss": 0.1363, "step": 9830 }, { "epoch": 61.06230529595015, "grad_norm": 3.484375, "learning_rate": 3.980483644737473e-05, "loss": 0.2932, "step": 9831 }, { "epoch": 61.06853582554517, "grad_norm": 3.421875, "learning_rate": 3.980285108606988e-05, "loss": 0.2186, "step": 9832 }, { "epoch": 61.074766355140184, "grad_norm": 2.984375, "learning_rate": 3.9800865580996796e-05, "loss": 0.1918, "step": 9833 }, { "epoch": 61.0809968847352, "grad_norm": 2.78125, "learning_rate": 3.979887993217473e-05, "loss": 0.1593, "step": 9834 }, { "epoch": 61.087227414330215, "grad_norm": 3.5, "learning_rate": 3.9796894139622995e-05, "loss": 0.1899, "step": 9835 }, { "epoch": 61.09345794392523, "grad_norm": 3.234375, "learning_rate": 3.979490820336086e-05, "loss": 0.1177, "step": 9836 }, { "epoch": 61.099688473520246, "grad_norm": 2.234375, "learning_rate": 3.979292212340763e-05, "loss": 0.2034, "step": 9837 }, { "epoch": 61.10591900311527, "grad_norm": 3.859375, "learning_rate": 3.9790935899782576e-05, "loss": 0.2627, "step": 9838 }, { "epoch": 61.11214953271028, "grad_norm": 1.2421875, "learning_rate": 3.978894953250499e-05, "loss": 0.1193, "step": 9839 }, { "epoch": 61.1183800623053, "grad_norm": 1.8984375, "learning_rate": 3.978696302159418e-05, "loss": 0.1488, "step": 9840 }, { "epoch": 61.124610591900314, "grad_norm": 3.5, "learning_rate": 3.9784976367069426e-05, "loss": 0.2274, "step": 9841 }, { "epoch": 61.13084112149533, "grad_norm": 3.140625, "learning_rate": 3.978298956895002e-05, "loss": 0.412, "step": 9842 }, { "epoch": 61.137071651090345, "grad_norm": 1.9375, "learning_rate": 3.978100262725527e-05, "loss": 0.152, "step": 9843 }, { "epoch": 61.14330218068536, "grad_norm": 2.15625, "learning_rate": 3.977901554200447e-05, "loss": 0.1434, "step": 9844 }, { "epoch": 61.149532710280376, "grad_norm": 1.6328125, "learning_rate": 3.977702831321692e-05, "loss": 0.1267, "step": 9845 }, { "epoch": 61.15576323987539, "grad_norm": 1.5390625, "learning_rate": 3.9775040940911907e-05, "loss": 0.1377, "step": 9846 }, { "epoch": 61.161993769470406, "grad_norm": 1.5234375, "learning_rate": 3.9773053425108745e-05, "loss": 0.1035, "step": 9847 }, { "epoch": 61.16822429906542, "grad_norm": 1.6796875, "learning_rate": 3.977106576582674e-05, "loss": 0.1275, "step": 9848 }, { "epoch": 61.17445482866044, "grad_norm": 3.515625, "learning_rate": 3.976907796308519e-05, "loss": 0.2925, "step": 9849 }, { "epoch": 61.18068535825545, "grad_norm": 1.484375, "learning_rate": 3.9767090016903396e-05, "loss": 0.1352, "step": 9850 }, { "epoch": 61.18691588785047, "grad_norm": 2.6875, "learning_rate": 3.976510192730068e-05, "loss": 0.0907, "step": 9851 }, { "epoch": 61.19314641744548, "grad_norm": 1.7421875, "learning_rate": 3.976311369429634e-05, "loss": 0.1621, "step": 9852 }, { "epoch": 61.1993769470405, "grad_norm": 2.296875, "learning_rate": 3.976112531790969e-05, "loss": 0.2217, "step": 9853 }, { "epoch": 61.205607476635514, "grad_norm": 2.109375, "learning_rate": 3.975913679816004e-05, "loss": 0.1865, "step": 9854 }, { "epoch": 61.21183800623053, "grad_norm": 2.640625, "learning_rate": 3.97571481350667e-05, "loss": 0.2003, "step": 9855 }, { "epoch": 61.218068535825545, "grad_norm": 2.46875, "learning_rate": 3.9755159328649e-05, "loss": 0.1567, "step": 9856 }, { "epoch": 61.22429906542056, "grad_norm": 3.0, "learning_rate": 3.975317037892623e-05, "loss": 0.2616, "step": 9857 }, { "epoch": 61.230529595015575, "grad_norm": 1.6953125, "learning_rate": 3.9751181285917714e-05, "loss": 0.1259, "step": 9858 }, { "epoch": 61.23676012461059, "grad_norm": 4.3125, "learning_rate": 3.9749192049642785e-05, "loss": 0.2012, "step": 9859 }, { "epoch": 61.242990654205606, "grad_norm": 2.359375, "learning_rate": 3.974720267012076e-05, "loss": 0.1303, "step": 9860 }, { "epoch": 61.24922118380062, "grad_norm": 3.40625, "learning_rate": 3.974521314737095e-05, "loss": 0.1173, "step": 9861 }, { "epoch": 61.25545171339564, "grad_norm": 2.515625, "learning_rate": 3.974322348141268e-05, "loss": 0.1463, "step": 9862 }, { "epoch": 61.26168224299065, "grad_norm": 2.78125, "learning_rate": 3.974123367226528e-05, "loss": 0.2574, "step": 9863 }, { "epoch": 61.26791277258567, "grad_norm": 4.40625, "learning_rate": 3.973924371994807e-05, "loss": 0.153, "step": 9864 }, { "epoch": 61.27414330218068, "grad_norm": 3.40625, "learning_rate": 3.9737253624480375e-05, "loss": 0.1399, "step": 9865 }, { "epoch": 61.2803738317757, "grad_norm": 2.84375, "learning_rate": 3.973526338588153e-05, "loss": 0.1624, "step": 9866 }, { "epoch": 61.28660436137071, "grad_norm": 2.359375, "learning_rate": 3.9733273004170865e-05, "loss": 0.1136, "step": 9867 }, { "epoch": 61.29283489096573, "grad_norm": 2.046875, "learning_rate": 3.9731282479367706e-05, "loss": 0.1766, "step": 9868 }, { "epoch": 61.299065420560744, "grad_norm": 2.546875, "learning_rate": 3.9729291811491385e-05, "loss": 0.1873, "step": 9869 }, { "epoch": 61.30529595015577, "grad_norm": 2.609375, "learning_rate": 3.972730100056124e-05, "loss": 0.1944, "step": 9870 }, { "epoch": 61.31152647975078, "grad_norm": 1.7890625, "learning_rate": 3.9725310046596595e-05, "loss": 0.1195, "step": 9871 }, { "epoch": 61.3177570093458, "grad_norm": 3.53125, "learning_rate": 3.9723318949616805e-05, "loss": 0.4138, "step": 9872 }, { "epoch": 61.32398753894081, "grad_norm": 2.328125, "learning_rate": 3.97213277096412e-05, "loss": 0.1601, "step": 9873 }, { "epoch": 61.33021806853583, "grad_norm": 2.921875, "learning_rate": 3.971933632668911e-05, "loss": 0.2061, "step": 9874 }, { "epoch": 61.33644859813084, "grad_norm": 2.9375, "learning_rate": 3.971734480077989e-05, "loss": 0.1455, "step": 9875 }, { "epoch": 61.34267912772586, "grad_norm": 1.5234375, "learning_rate": 3.971535313193287e-05, "loss": 0.1288, "step": 9876 }, { "epoch": 61.348909657320874, "grad_norm": 1.765625, "learning_rate": 3.97133613201674e-05, "loss": 0.1367, "step": 9877 }, { "epoch": 61.35514018691589, "grad_norm": 4.03125, "learning_rate": 3.9711369365502824e-05, "loss": 0.2682, "step": 9878 }, { "epoch": 61.361370716510905, "grad_norm": 2.796875, "learning_rate": 3.9709377267958493e-05, "loss": 0.1537, "step": 9879 }, { "epoch": 61.36760124610592, "grad_norm": 3.125, "learning_rate": 3.970738502755374e-05, "loss": 0.1927, "step": 9880 }, { "epoch": 61.373831775700936, "grad_norm": 2.4375, "learning_rate": 3.970539264430792e-05, "loss": 0.2316, "step": 9881 }, { "epoch": 61.38006230529595, "grad_norm": 3.421875, "learning_rate": 3.97034001182404e-05, "loss": 0.2121, "step": 9882 }, { "epoch": 61.386292834890966, "grad_norm": 2.59375, "learning_rate": 3.970140744937052e-05, "loss": 0.1918, "step": 9883 }, { "epoch": 61.39252336448598, "grad_norm": 4.84375, "learning_rate": 3.969941463771763e-05, "loss": 0.2697, "step": 9884 }, { "epoch": 61.398753894081, "grad_norm": 3.1875, "learning_rate": 3.969742168330108e-05, "loss": 0.1915, "step": 9885 }, { "epoch": 61.40498442367601, "grad_norm": 3.078125, "learning_rate": 3.9695428586140236e-05, "loss": 0.1229, "step": 9886 }, { "epoch": 61.41121495327103, "grad_norm": 3.84375, "learning_rate": 3.969343534625445e-05, "loss": 0.2202, "step": 9887 }, { "epoch": 61.41744548286604, "grad_norm": 1.6796875, "learning_rate": 3.9691441963663085e-05, "loss": 0.1516, "step": 9888 }, { "epoch": 61.42367601246106, "grad_norm": 2.265625, "learning_rate": 3.9689448438385505e-05, "loss": 0.112, "step": 9889 }, { "epoch": 61.429906542056074, "grad_norm": 3.734375, "learning_rate": 3.968745477044106e-05, "loss": 0.2931, "step": 9890 }, { "epoch": 61.43613707165109, "grad_norm": 3.640625, "learning_rate": 3.9685460959849105e-05, "loss": 0.2373, "step": 9891 }, { "epoch": 61.442367601246104, "grad_norm": 3.09375, "learning_rate": 3.968346700662904e-05, "loss": 0.1419, "step": 9892 }, { "epoch": 61.44859813084112, "grad_norm": 2.234375, "learning_rate": 3.9681472910800196e-05, "loss": 0.1204, "step": 9893 }, { "epoch": 61.454828660436135, "grad_norm": 2.890625, "learning_rate": 3.9679478672381955e-05, "loss": 0.1584, "step": 9894 }, { "epoch": 61.46105919003115, "grad_norm": 2.859375, "learning_rate": 3.967748429139368e-05, "loss": 0.232, "step": 9895 }, { "epoch": 61.467289719626166, "grad_norm": 3.71875, "learning_rate": 3.9675489767854747e-05, "loss": 0.247, "step": 9896 }, { "epoch": 61.47352024922118, "grad_norm": 2.4375, "learning_rate": 3.967349510178452e-05, "loss": 0.2227, "step": 9897 }, { "epoch": 61.4797507788162, "grad_norm": 1.9609375, "learning_rate": 3.967150029320238e-05, "loss": 0.1337, "step": 9898 }, { "epoch": 61.48598130841121, "grad_norm": 2.203125, "learning_rate": 3.96695053421277e-05, "loss": 0.1766, "step": 9899 }, { "epoch": 61.49221183800623, "grad_norm": 1.828125, "learning_rate": 3.966751024857984e-05, "loss": 0.1181, "step": 9900 }, { "epoch": 61.49844236760124, "grad_norm": 1.1953125, "learning_rate": 3.966551501257819e-05, "loss": 0.119, "step": 9901 }, { "epoch": 61.504672897196265, "grad_norm": 1.1796875, "learning_rate": 3.9663519634142135e-05, "loss": 0.0719, "step": 9902 }, { "epoch": 61.51090342679128, "grad_norm": 2.5, "learning_rate": 3.966152411329105e-05, "loss": 0.1212, "step": 9903 }, { "epoch": 61.517133956386296, "grad_norm": 2.21875, "learning_rate": 3.96595284500443e-05, "loss": 0.185, "step": 9904 }, { "epoch": 61.52336448598131, "grad_norm": 2.140625, "learning_rate": 3.965753264442128e-05, "loss": 0.1702, "step": 9905 }, { "epoch": 61.52959501557633, "grad_norm": 1.890625, "learning_rate": 3.9655536696441384e-05, "loss": 0.1433, "step": 9906 }, { "epoch": 61.53582554517134, "grad_norm": 2.21875, "learning_rate": 3.9653540606123975e-05, "loss": 0.1856, "step": 9907 }, { "epoch": 61.54205607476636, "grad_norm": 2.921875, "learning_rate": 3.9651544373488456e-05, "loss": 0.2484, "step": 9908 }, { "epoch": 61.54828660436137, "grad_norm": 3.421875, "learning_rate": 3.96495479985542e-05, "loss": 0.4184, "step": 9909 }, { "epoch": 61.55451713395639, "grad_norm": 2.46875, "learning_rate": 3.964755148134062e-05, "loss": 0.1535, "step": 9910 }, { "epoch": 61.5607476635514, "grad_norm": 2.8125, "learning_rate": 3.964555482186708e-05, "loss": 0.2296, "step": 9911 }, { "epoch": 61.56697819314642, "grad_norm": 1.0, "learning_rate": 3.9643558020152994e-05, "loss": 0.1041, "step": 9912 }, { "epoch": 61.573208722741434, "grad_norm": 3.203125, "learning_rate": 3.9641561076217736e-05, "loss": 0.2404, "step": 9913 }, { "epoch": 61.57943925233645, "grad_norm": 2.328125, "learning_rate": 3.9639563990080716e-05, "loss": 0.1272, "step": 9914 }, { "epoch": 61.585669781931465, "grad_norm": 1.5859375, "learning_rate": 3.963756676176132e-05, "loss": 0.0863, "step": 9915 }, { "epoch": 61.59190031152648, "grad_norm": 3.40625, "learning_rate": 3.963556939127896e-05, "loss": 0.2724, "step": 9916 }, { "epoch": 61.598130841121495, "grad_norm": 3.25, "learning_rate": 3.9633571878653006e-05, "loss": 0.2825, "step": 9917 }, { "epoch": 61.60436137071651, "grad_norm": 2.625, "learning_rate": 3.96315742239029e-05, "loss": 0.1918, "step": 9918 }, { "epoch": 61.610591900311526, "grad_norm": 2.09375, "learning_rate": 3.9629576427047996e-05, "loss": 0.1591, "step": 9919 }, { "epoch": 61.61682242990654, "grad_norm": 4.0, "learning_rate": 3.9627578488107736e-05, "loss": 0.1155, "step": 9920 }, { "epoch": 61.62305295950156, "grad_norm": 2.9375, "learning_rate": 3.962558040710151e-05, "loss": 0.1638, "step": 9921 }, { "epoch": 61.62928348909657, "grad_norm": 2.75, "learning_rate": 3.962358218404872e-05, "loss": 0.2924, "step": 9922 }, { "epoch": 61.63551401869159, "grad_norm": 2.3125, "learning_rate": 3.9621583818968774e-05, "loss": 0.1679, "step": 9923 }, { "epoch": 61.6417445482866, "grad_norm": 2.875, "learning_rate": 3.961958531188109e-05, "loss": 0.1875, "step": 9924 }, { "epoch": 61.64797507788162, "grad_norm": 2.65625, "learning_rate": 3.961758666280506e-05, "loss": 0.2553, "step": 9925 }, { "epoch": 61.654205607476634, "grad_norm": 2.953125, "learning_rate": 3.961558787176012e-05, "loss": 0.1549, "step": 9926 }, { "epoch": 61.66043613707165, "grad_norm": 2.265625, "learning_rate": 3.9613588938765664e-05, "loss": 0.18, "step": 9927 }, { "epoch": 61.666666666666664, "grad_norm": 2.125, "learning_rate": 3.961158986384111e-05, "loss": 0.1609, "step": 9928 }, { "epoch": 61.67289719626168, "grad_norm": 2.75, "learning_rate": 3.9609590647005866e-05, "loss": 0.1684, "step": 9929 }, { "epoch": 61.679127725856695, "grad_norm": 3.1875, "learning_rate": 3.960759128827937e-05, "loss": 0.2654, "step": 9930 }, { "epoch": 61.68535825545171, "grad_norm": 2.09375, "learning_rate": 3.9605591787681025e-05, "loss": 0.1477, "step": 9931 }, { "epoch": 61.691588785046726, "grad_norm": 2.875, "learning_rate": 3.960359214523025e-05, "loss": 0.2165, "step": 9932 }, { "epoch": 61.69781931464174, "grad_norm": 2.484375, "learning_rate": 3.9601592360946463e-05, "loss": 0.1958, "step": 9933 }, { "epoch": 61.70404984423676, "grad_norm": 2.140625, "learning_rate": 3.95995924348491e-05, "loss": 0.1489, "step": 9934 }, { "epoch": 61.71028037383178, "grad_norm": 1.921875, "learning_rate": 3.9597592366957576e-05, "loss": 0.1345, "step": 9935 }, { "epoch": 61.716510903426794, "grad_norm": 2.53125, "learning_rate": 3.959559215729132e-05, "loss": 0.1672, "step": 9936 }, { "epoch": 61.72274143302181, "grad_norm": 3.0625, "learning_rate": 3.959359180586975e-05, "loss": 0.3125, "step": 9937 }, { "epoch": 61.728971962616825, "grad_norm": 1.7578125, "learning_rate": 3.95915913127123e-05, "loss": 0.1186, "step": 9938 }, { "epoch": 61.73520249221184, "grad_norm": 2.609375, "learning_rate": 3.95895906778384e-05, "loss": 0.1773, "step": 9939 }, { "epoch": 61.741433021806856, "grad_norm": 2.484375, "learning_rate": 3.9587589901267474e-05, "loss": 0.1962, "step": 9940 }, { "epoch": 61.74766355140187, "grad_norm": 3.046875, "learning_rate": 3.958558898301897e-05, "loss": 0.1321, "step": 9941 }, { "epoch": 61.753894080996886, "grad_norm": 2.578125, "learning_rate": 3.95835879231123e-05, "loss": 0.1306, "step": 9942 }, { "epoch": 61.7601246105919, "grad_norm": 2.15625, "learning_rate": 3.958158672156691e-05, "loss": 0.1115, "step": 9943 }, { "epoch": 61.76635514018692, "grad_norm": 2.125, "learning_rate": 3.957958537840224e-05, "loss": 0.1088, "step": 9944 }, { "epoch": 61.77258566978193, "grad_norm": 3.484375, "learning_rate": 3.9577583893637715e-05, "loss": 0.128, "step": 9945 }, { "epoch": 61.77881619937695, "grad_norm": 3.78125, "learning_rate": 3.9575582267292785e-05, "loss": 0.1582, "step": 9946 }, { "epoch": 61.78504672897196, "grad_norm": 3.28125, "learning_rate": 3.957358049938689e-05, "loss": 0.2573, "step": 9947 }, { "epoch": 61.79127725856698, "grad_norm": 1.8984375, "learning_rate": 3.957157858993946e-05, "loss": 0.1278, "step": 9948 }, { "epoch": 61.797507788161994, "grad_norm": 1.78125, "learning_rate": 3.9569576538969953e-05, "loss": 0.1289, "step": 9949 }, { "epoch": 61.80373831775701, "grad_norm": 4.21875, "learning_rate": 3.956757434649781e-05, "loss": 0.1431, "step": 9950 }, { "epoch": 61.809968847352025, "grad_norm": 2.578125, "learning_rate": 3.956557201254247e-05, "loss": 0.1226, "step": 9951 }, { "epoch": 61.81619937694704, "grad_norm": 4.96875, "learning_rate": 3.956356953712338e-05, "loss": 0.1663, "step": 9952 }, { "epoch": 61.822429906542055, "grad_norm": 4.375, "learning_rate": 3.9561566920259994e-05, "loss": 0.2216, "step": 9953 }, { "epoch": 61.82866043613707, "grad_norm": 2.4375, "learning_rate": 3.955956416197175e-05, "loss": 0.3202, "step": 9954 }, { "epoch": 61.834890965732086, "grad_norm": 1.46875, "learning_rate": 3.955756126227812e-05, "loss": 0.1197, "step": 9955 }, { "epoch": 61.8411214953271, "grad_norm": 3.6875, "learning_rate": 3.9555558221198544e-05, "loss": 0.3475, "step": 9956 }, { "epoch": 61.84735202492212, "grad_norm": 3.609375, "learning_rate": 3.9553555038752466e-05, "loss": 0.1374, "step": 9957 }, { "epoch": 61.85358255451713, "grad_norm": 3.09375, "learning_rate": 3.955155171495936e-05, "loss": 0.1141, "step": 9958 }, { "epoch": 61.85981308411215, "grad_norm": 3.109375, "learning_rate": 3.954954824983867e-05, "loss": 0.2225, "step": 9959 }, { "epoch": 61.86604361370716, "grad_norm": 2.484375, "learning_rate": 3.954754464340987e-05, "loss": 0.2837, "step": 9960 }, { "epoch": 61.87227414330218, "grad_norm": 3.109375, "learning_rate": 3.95455408956924e-05, "loss": 0.2558, "step": 9961 }, { "epoch": 61.87850467289719, "grad_norm": 3.3125, "learning_rate": 3.9543537006705736e-05, "loss": 0.1349, "step": 9962 }, { "epoch": 61.88473520249221, "grad_norm": 2.828125, "learning_rate": 3.954153297646933e-05, "loss": 0.1374, "step": 9963 }, { "epoch": 61.890965732087224, "grad_norm": 4.15625, "learning_rate": 3.9539528805002644e-05, "loss": 0.1745, "step": 9964 }, { "epoch": 61.89719626168224, "grad_norm": 1.6796875, "learning_rate": 3.953752449232515e-05, "loss": 0.1392, "step": 9965 }, { "epoch": 61.90342679127726, "grad_norm": 2.4375, "learning_rate": 3.953552003845632e-05, "loss": 0.2108, "step": 9966 }, { "epoch": 61.90965732087228, "grad_norm": 1.484375, "learning_rate": 3.95335154434156e-05, "loss": 0.1344, "step": 9967 }, { "epoch": 61.91588785046729, "grad_norm": 4.0625, "learning_rate": 3.953151070722248e-05, "loss": 0.2585, "step": 9968 }, { "epoch": 61.92211838006231, "grad_norm": 4.40625, "learning_rate": 3.952950582989643e-05, "loss": 0.3011, "step": 9969 }, { "epoch": 61.92834890965732, "grad_norm": 2.015625, "learning_rate": 3.95275008114569e-05, "loss": 0.1207, "step": 9970 }, { "epoch": 61.93457943925234, "grad_norm": 2.40625, "learning_rate": 3.9525495651923386e-05, "loss": 0.1438, "step": 9971 }, { "epoch": 61.940809968847354, "grad_norm": 2.40625, "learning_rate": 3.952349035131536e-05, "loss": 0.2419, "step": 9972 }, { "epoch": 61.94704049844237, "grad_norm": 3.5625, "learning_rate": 3.9521484909652284e-05, "loss": 0.2476, "step": 9973 }, { "epoch": 61.953271028037385, "grad_norm": 5.03125, "learning_rate": 3.951947932695365e-05, "loss": 0.1477, "step": 9974 }, { "epoch": 61.9595015576324, "grad_norm": 1.734375, "learning_rate": 3.951747360323893e-05, "loss": 0.1431, "step": 9975 }, { "epoch": 61.965732087227416, "grad_norm": 2.625, "learning_rate": 3.95154677385276e-05, "loss": 0.1971, "step": 9976 }, { "epoch": 61.97196261682243, "grad_norm": 2.390625, "learning_rate": 3.951346173283915e-05, "loss": 0.1978, "step": 9977 }, { "epoch": 61.978193146417446, "grad_norm": 4.1875, "learning_rate": 3.951145558619306e-05, "loss": 0.2082, "step": 9978 }, { "epoch": 61.98442367601246, "grad_norm": 4.15625, "learning_rate": 3.95094492986088e-05, "loss": 0.24, "step": 9979 }, { "epoch": 61.99065420560748, "grad_norm": 3.046875, "learning_rate": 3.950744287010588e-05, "loss": 0.2379, "step": 9980 }, { "epoch": 61.99688473520249, "grad_norm": 2.671875, "learning_rate": 3.950543630070378e-05, "loss": 0.3577, "step": 9981 }, { "epoch": 62.0, "grad_norm": 1.421875, "learning_rate": 3.950342959042197e-05, "loss": 0.0825, "step": 9982 }, { "epoch": 62.006230529595015, "grad_norm": 1.34375, "learning_rate": 3.9501422739279956e-05, "loss": 0.1418, "step": 9983 }, { "epoch": 62.01246105919003, "grad_norm": 3.328125, "learning_rate": 3.949941574729723e-05, "loss": 0.2908, "step": 9984 }, { "epoch": 62.018691588785046, "grad_norm": 2.0625, "learning_rate": 3.949740861449327e-05, "loss": 0.1346, "step": 9985 }, { "epoch": 62.02492211838006, "grad_norm": 1.484375, "learning_rate": 3.9495401340887594e-05, "loss": 0.1197, "step": 9986 }, { "epoch": 62.03115264797508, "grad_norm": 1.203125, "learning_rate": 3.949339392649967e-05, "loss": 0.1114, "step": 9987 }, { "epoch": 62.03738317757009, "grad_norm": 2.765625, "learning_rate": 3.9491386371349017e-05, "loss": 0.2111, "step": 9988 }, { "epoch": 62.04361370716511, "grad_norm": 2.375, "learning_rate": 3.9489378675455116e-05, "loss": 0.1331, "step": 9989 }, { "epoch": 62.04984423676012, "grad_norm": 2.453125, "learning_rate": 3.948737083883748e-05, "loss": 0.1763, "step": 9990 }, { "epoch": 62.05607476635514, "grad_norm": 2.4375, "learning_rate": 3.9485362861515594e-05, "loss": 0.2183, "step": 9991 }, { "epoch": 62.06230529595015, "grad_norm": 1.984375, "learning_rate": 3.948335474350897e-05, "loss": 0.1249, "step": 9992 }, { "epoch": 62.06853582554517, "grad_norm": 2.625, "learning_rate": 3.948134648483711e-05, "loss": 0.169, "step": 9993 }, { "epoch": 62.074766355140184, "grad_norm": 2.8125, "learning_rate": 3.947933808551953e-05, "loss": 0.237, "step": 9994 }, { "epoch": 62.0809968847352, "grad_norm": 2.859375, "learning_rate": 3.9477329545575705e-05, "loss": 0.2642, "step": 9995 }, { "epoch": 62.087227414330215, "grad_norm": 1.0859375, "learning_rate": 3.9475320865025166e-05, "loss": 0.103, "step": 9996 }, { "epoch": 62.09345794392523, "grad_norm": 2.125, "learning_rate": 3.9473312043887426e-05, "loss": 0.1528, "step": 9997 }, { "epoch": 62.099688473520246, "grad_norm": 2.203125, "learning_rate": 3.947130308218198e-05, "loss": 0.1774, "step": 9998 }, { "epoch": 62.10591900311527, "grad_norm": 1.8828125, "learning_rate": 3.946929397992835e-05, "loss": 0.1141, "step": 9999 }, { "epoch": 62.11214953271028, "grad_norm": 2.765625, "learning_rate": 3.946728473714604e-05, "loss": 0.2351, "step": 10000 }, { "epoch": 62.1183800623053, "grad_norm": 3.234375, "learning_rate": 3.9465275353854575e-05, "loss": 0.2203, "step": 10001 }, { "epoch": 62.124610591900314, "grad_norm": 1.015625, "learning_rate": 3.946326583007346e-05, "loss": 0.0889, "step": 10002 }, { "epoch": 62.13084112149533, "grad_norm": 1.609375, "learning_rate": 3.946125616582221e-05, "loss": 0.0958, "step": 10003 }, { "epoch": 62.137071651090345, "grad_norm": 3.640625, "learning_rate": 3.945924636112036e-05, "loss": 0.2943, "step": 10004 }, { "epoch": 62.14330218068536, "grad_norm": 2.984375, "learning_rate": 3.945723641598741e-05, "loss": 0.2127, "step": 10005 }, { "epoch": 62.149532710280376, "grad_norm": 3.828125, "learning_rate": 3.945522633044289e-05, "loss": 0.1382, "step": 10006 }, { "epoch": 62.15576323987539, "grad_norm": 2.296875, "learning_rate": 3.9453216104506336e-05, "loss": 0.1103, "step": 10007 }, { "epoch": 62.161993769470406, "grad_norm": 3.46875, "learning_rate": 3.945120573819725e-05, "loss": 0.2913, "step": 10008 }, { "epoch": 62.16822429906542, "grad_norm": 1.671875, "learning_rate": 3.9449195231535156e-05, "loss": 0.1384, "step": 10009 }, { "epoch": 62.17445482866044, "grad_norm": 3.46875, "learning_rate": 3.94471845845396e-05, "loss": 0.2553, "step": 10010 }, { "epoch": 62.18068535825545, "grad_norm": 2.328125, "learning_rate": 3.944517379723009e-05, "loss": 0.1937, "step": 10011 }, { "epoch": 62.18691588785047, "grad_norm": 2.828125, "learning_rate": 3.944316286962618e-05, "loss": 0.184, "step": 10012 }, { "epoch": 62.19314641744548, "grad_norm": 2.96875, "learning_rate": 3.944115180174738e-05, "loss": 0.223, "step": 10013 }, { "epoch": 62.1993769470405, "grad_norm": 2.078125, "learning_rate": 3.943914059361322e-05, "loss": 0.1597, "step": 10014 }, { "epoch": 62.205607476635514, "grad_norm": 2.5, "learning_rate": 3.943712924524324e-05, "loss": 0.1915, "step": 10015 }, { "epoch": 62.21183800623053, "grad_norm": 2.171875, "learning_rate": 3.943511775665698e-05, "loss": 0.1589, "step": 10016 }, { "epoch": 62.218068535825545, "grad_norm": 3.359375, "learning_rate": 3.943310612787397e-05, "loss": 0.2316, "step": 10017 }, { "epoch": 62.22429906542056, "grad_norm": 2.453125, "learning_rate": 3.9431094358913745e-05, "loss": 0.1219, "step": 10018 }, { "epoch": 62.230529595015575, "grad_norm": 3.203125, "learning_rate": 3.942908244979585e-05, "loss": 0.2497, "step": 10019 }, { "epoch": 62.23676012461059, "grad_norm": 1.9140625, "learning_rate": 3.942707040053982e-05, "loss": 0.1177, "step": 10020 }, { "epoch": 62.242990654205606, "grad_norm": 4.6875, "learning_rate": 3.94250582111652e-05, "loss": 0.3016, "step": 10021 }, { "epoch": 62.24922118380062, "grad_norm": 2.265625, "learning_rate": 3.942304588169153e-05, "loss": 0.1524, "step": 10022 }, { "epoch": 62.25545171339564, "grad_norm": 2.8125, "learning_rate": 3.9421033412138366e-05, "loss": 0.1802, "step": 10023 }, { "epoch": 62.26168224299065, "grad_norm": 2.578125, "learning_rate": 3.941902080252523e-05, "loss": 0.1976, "step": 10024 }, { "epoch": 62.26791277258567, "grad_norm": 2.0625, "learning_rate": 3.941700805287168e-05, "loss": 0.2181, "step": 10025 }, { "epoch": 62.27414330218068, "grad_norm": 2.125, "learning_rate": 3.9414995163197276e-05, "loss": 0.1385, "step": 10026 }, { "epoch": 62.2803738317757, "grad_norm": 2.359375, "learning_rate": 3.9412982133521545e-05, "loss": 0.232, "step": 10027 }, { "epoch": 62.28660436137071, "grad_norm": 2.109375, "learning_rate": 3.941096896386406e-05, "loss": 0.1192, "step": 10028 }, { "epoch": 62.29283489096573, "grad_norm": 3.46875, "learning_rate": 3.9408955654244364e-05, "loss": 0.1801, "step": 10029 }, { "epoch": 62.299065420560744, "grad_norm": 1.859375, "learning_rate": 3.940694220468201e-05, "loss": 0.144, "step": 10030 }, { "epoch": 62.30529595015577, "grad_norm": 3.515625, "learning_rate": 3.940492861519654e-05, "loss": 0.3367, "step": 10031 }, { "epoch": 62.31152647975078, "grad_norm": 2.171875, "learning_rate": 3.940291488580754e-05, "loss": 0.1997, "step": 10032 }, { "epoch": 62.3177570093458, "grad_norm": 2.515625, "learning_rate": 3.9400901016534544e-05, "loss": 0.2606, "step": 10033 }, { "epoch": 62.32398753894081, "grad_norm": 2.484375, "learning_rate": 3.939888700739712e-05, "loss": 0.1456, "step": 10034 }, { "epoch": 62.33021806853583, "grad_norm": 2.59375, "learning_rate": 3.939687285841483e-05, "loss": 0.1326, "step": 10035 }, { "epoch": 62.33644859813084, "grad_norm": 2.703125, "learning_rate": 3.939485856960723e-05, "loss": 0.1467, "step": 10036 }, { "epoch": 62.34267912772586, "grad_norm": 2.859375, "learning_rate": 3.939284414099389e-05, "loss": 0.2289, "step": 10037 }, { "epoch": 62.348909657320874, "grad_norm": 1.7109375, "learning_rate": 3.939082957259437e-05, "loss": 0.1248, "step": 10038 }, { "epoch": 62.35514018691589, "grad_norm": 2.328125, "learning_rate": 3.938881486442823e-05, "loss": 0.1589, "step": 10039 }, { "epoch": 62.361370716510905, "grad_norm": 3.078125, "learning_rate": 3.938680001651505e-05, "loss": 0.2084, "step": 10040 }, { "epoch": 62.36760124610592, "grad_norm": 3.203125, "learning_rate": 3.938478502887438e-05, "loss": 0.1769, "step": 10041 }, { "epoch": 62.373831775700936, "grad_norm": 1.5703125, "learning_rate": 3.9382769901525816e-05, "loss": 0.153, "step": 10042 }, { "epoch": 62.38006230529595, "grad_norm": 2.484375, "learning_rate": 3.9380754634488916e-05, "loss": 0.2937, "step": 10043 }, { "epoch": 62.386292834890966, "grad_norm": 2.375, "learning_rate": 3.937873922778324e-05, "loss": 0.1939, "step": 10044 }, { "epoch": 62.39252336448598, "grad_norm": 1.90625, "learning_rate": 3.93767236814284e-05, "loss": 0.1331, "step": 10045 }, { "epoch": 62.398753894081, "grad_norm": 1.71875, "learning_rate": 3.9374707995443914e-05, "loss": 0.1272, "step": 10046 }, { "epoch": 62.40498442367601, "grad_norm": 3.15625, "learning_rate": 3.937269216984941e-05, "loss": 0.3046, "step": 10047 }, { "epoch": 62.41121495327103, "grad_norm": 3.3125, "learning_rate": 3.9370676204664444e-05, "loss": 0.3076, "step": 10048 }, { "epoch": 62.41744548286604, "grad_norm": 2.359375, "learning_rate": 3.936866009990859e-05, "loss": 0.1981, "step": 10049 }, { "epoch": 62.42367601246106, "grad_norm": 1.9375, "learning_rate": 3.9366643855601446e-05, "loss": 0.148, "step": 10050 }, { "epoch": 62.429906542056074, "grad_norm": 2.234375, "learning_rate": 3.936462747176258e-05, "loss": 0.1713, "step": 10051 }, { "epoch": 62.43613707165109, "grad_norm": 2.5, "learning_rate": 3.9362610948411585e-05, "loss": 0.1563, "step": 10052 }, { "epoch": 62.442367601246104, "grad_norm": 2.59375, "learning_rate": 3.9360594285568045e-05, "loss": 0.1597, "step": 10053 }, { "epoch": 62.44859813084112, "grad_norm": 2.953125, "learning_rate": 3.935857748325153e-05, "loss": 0.2718, "step": 10054 }, { "epoch": 62.454828660436135, "grad_norm": 3.21875, "learning_rate": 3.935656054148165e-05, "loss": 0.3456, "step": 10055 }, { "epoch": 62.46105919003115, "grad_norm": 2.4375, "learning_rate": 3.9354543460277976e-05, "loss": 0.2006, "step": 10056 }, { "epoch": 62.467289719626166, "grad_norm": 1.484375, "learning_rate": 3.9352526239660116e-05, "loss": 0.1282, "step": 10057 }, { "epoch": 62.47352024922118, "grad_norm": 1.53125, "learning_rate": 3.935050887964764e-05, "loss": 0.1619, "step": 10058 }, { "epoch": 62.4797507788162, "grad_norm": 1.5390625, "learning_rate": 3.934849138026017e-05, "loss": 0.1198, "step": 10059 }, { "epoch": 62.48598130841121, "grad_norm": 1.90625, "learning_rate": 3.934647374151727e-05, "loss": 0.1829, "step": 10060 }, { "epoch": 62.49221183800623, "grad_norm": 1.0703125, "learning_rate": 3.934445596343856e-05, "loss": 0.1065, "step": 10061 }, { "epoch": 62.49844236760124, "grad_norm": 2.296875, "learning_rate": 3.9342438046043615e-05, "loss": 0.1952, "step": 10062 }, { "epoch": 62.504672897196265, "grad_norm": 2.078125, "learning_rate": 3.9340419989352045e-05, "loss": 0.1182, "step": 10063 }, { "epoch": 62.51090342679128, "grad_norm": 2.28125, "learning_rate": 3.9338401793383465e-05, "loss": 0.1522, "step": 10064 }, { "epoch": 62.517133956386296, "grad_norm": 2.484375, "learning_rate": 3.933638345815744e-05, "loss": 0.1739, "step": 10065 }, { "epoch": 62.52336448598131, "grad_norm": 3.15625, "learning_rate": 3.9334364983693604e-05, "loss": 0.367, "step": 10066 }, { "epoch": 62.52959501557633, "grad_norm": 2.640625, "learning_rate": 3.933234637001155e-05, "loss": 0.1855, "step": 10067 }, { "epoch": 62.53582554517134, "grad_norm": 1.359375, "learning_rate": 3.933032761713088e-05, "loss": 0.0882, "step": 10068 }, { "epoch": 62.54205607476636, "grad_norm": 1.9140625, "learning_rate": 3.9328308725071204e-05, "loss": 0.185, "step": 10069 }, { "epoch": 62.54828660436137, "grad_norm": 3.015625, "learning_rate": 3.932628969385213e-05, "loss": 0.2088, "step": 10070 }, { "epoch": 62.55451713395639, "grad_norm": 2.859375, "learning_rate": 3.932427052349327e-05, "loss": 0.207, "step": 10071 }, { "epoch": 62.5607476635514, "grad_norm": 3.703125, "learning_rate": 3.9322251214014225e-05, "loss": 0.3106, "step": 10072 }, { "epoch": 62.56697819314642, "grad_norm": 2.03125, "learning_rate": 3.932023176543462e-05, "loss": 0.1033, "step": 10073 }, { "epoch": 62.573208722741434, "grad_norm": 2.515625, "learning_rate": 3.931821217777406e-05, "loss": 0.1842, "step": 10074 }, { "epoch": 62.57943925233645, "grad_norm": 2.8125, "learning_rate": 3.931619245105216e-05, "loss": 0.1816, "step": 10075 }, { "epoch": 62.585669781931465, "grad_norm": 2.625, "learning_rate": 3.931417258528853e-05, "loss": 0.1951, "step": 10076 }, { "epoch": 62.59190031152648, "grad_norm": 1.9609375, "learning_rate": 3.9312152580502804e-05, "loss": 0.0863, "step": 10077 }, { "epoch": 62.598130841121495, "grad_norm": 2.96875, "learning_rate": 3.9310132436714584e-05, "loss": 0.1945, "step": 10078 }, { "epoch": 62.60436137071651, "grad_norm": 1.5859375, "learning_rate": 3.9308112153943505e-05, "loss": 0.1353, "step": 10079 }, { "epoch": 62.610591900311526, "grad_norm": 1.734375, "learning_rate": 3.930609173220918e-05, "loss": 0.1217, "step": 10080 }, { "epoch": 62.61682242990654, "grad_norm": 3.453125, "learning_rate": 3.930407117153123e-05, "loss": 0.2628, "step": 10081 }, { "epoch": 62.62305295950156, "grad_norm": 4.0, "learning_rate": 3.930205047192927e-05, "loss": 0.1983, "step": 10082 }, { "epoch": 62.62928348909657, "grad_norm": 3.328125, "learning_rate": 3.930002963342296e-05, "loss": 0.2893, "step": 10083 }, { "epoch": 62.63551401869159, "grad_norm": 1.546875, "learning_rate": 3.9298008656031894e-05, "loss": 0.121, "step": 10084 }, { "epoch": 62.6417445482866, "grad_norm": 2.640625, "learning_rate": 3.9295987539775705e-05, "loss": 0.2012, "step": 10085 }, { "epoch": 62.64797507788162, "grad_norm": 1.375, "learning_rate": 3.9293966284674025e-05, "loss": 0.1053, "step": 10086 }, { "epoch": 62.654205607476634, "grad_norm": 5.1875, "learning_rate": 3.92919448907465e-05, "loss": 0.2297, "step": 10087 }, { "epoch": 62.66043613707165, "grad_norm": 3.71875, "learning_rate": 3.928992335801274e-05, "loss": 0.1821, "step": 10088 }, { "epoch": 62.666666666666664, "grad_norm": 3.53125, "learning_rate": 3.9287901686492384e-05, "loss": 0.2053, "step": 10089 }, { "epoch": 62.67289719626168, "grad_norm": 3.21875, "learning_rate": 3.928587987620508e-05, "loss": 0.158, "step": 10090 }, { "epoch": 62.679127725856695, "grad_norm": 2.109375, "learning_rate": 3.928385792717045e-05, "loss": 0.1813, "step": 10091 }, { "epoch": 62.68535825545171, "grad_norm": 3.53125, "learning_rate": 3.928183583940813e-05, "loss": 0.1166, "step": 10092 }, { "epoch": 62.691588785046726, "grad_norm": 2.8125, "learning_rate": 3.927981361293778e-05, "loss": 0.1959, "step": 10093 }, { "epoch": 62.69781931464174, "grad_norm": 3.40625, "learning_rate": 3.927779124777902e-05, "loss": 0.3647, "step": 10094 }, { "epoch": 62.70404984423676, "grad_norm": 2.25, "learning_rate": 3.927576874395149e-05, "loss": 0.152, "step": 10095 }, { "epoch": 62.71028037383178, "grad_norm": 2.5, "learning_rate": 3.927374610147485e-05, "loss": 0.1146, "step": 10096 }, { "epoch": 62.716510903426794, "grad_norm": 2.03125, "learning_rate": 3.927172332036873e-05, "loss": 0.1724, "step": 10097 }, { "epoch": 62.72274143302181, "grad_norm": 2.296875, "learning_rate": 3.926970040065278e-05, "loss": 0.1307, "step": 10098 }, { "epoch": 62.728971962616825, "grad_norm": 2.359375, "learning_rate": 3.9267677342346655e-05, "loss": 0.1098, "step": 10099 }, { "epoch": 62.73520249221184, "grad_norm": 2.03125, "learning_rate": 3.926565414546998e-05, "loss": 0.1911, "step": 10100 }, { "epoch": 62.741433021806856, "grad_norm": 2.109375, "learning_rate": 3.9263630810042426e-05, "loss": 0.1593, "step": 10101 }, { "epoch": 62.74766355140187, "grad_norm": 2.921875, "learning_rate": 3.926160733608365e-05, "loss": 0.2921, "step": 10102 }, { "epoch": 62.753894080996886, "grad_norm": 2.34375, "learning_rate": 3.925958372361328e-05, "loss": 0.1805, "step": 10103 }, { "epoch": 62.7601246105919, "grad_norm": 2.296875, "learning_rate": 3.9257559972650986e-05, "loss": 0.1435, "step": 10104 }, { "epoch": 62.76635514018692, "grad_norm": 1.2578125, "learning_rate": 3.9255536083216415e-05, "loss": 0.11, "step": 10105 }, { "epoch": 62.77258566978193, "grad_norm": 1.4296875, "learning_rate": 3.925351205532924e-05, "loss": 0.1148, "step": 10106 }, { "epoch": 62.77881619937695, "grad_norm": 3.40625, "learning_rate": 3.9251487889009094e-05, "loss": 0.177, "step": 10107 }, { "epoch": 62.78504672897196, "grad_norm": 3.53125, "learning_rate": 3.924946358427566e-05, "loss": 0.2898, "step": 10108 }, { "epoch": 62.79127725856698, "grad_norm": 3.515625, "learning_rate": 3.924743914114858e-05, "loss": 0.2079, "step": 10109 }, { "epoch": 62.797507788161994, "grad_norm": 2.96875, "learning_rate": 3.924541455964753e-05, "loss": 0.2172, "step": 10110 }, { "epoch": 62.80373831775701, "grad_norm": 1.5703125, "learning_rate": 3.924338983979216e-05, "loss": 0.106, "step": 10111 }, { "epoch": 62.809968847352025, "grad_norm": 2.46875, "learning_rate": 3.9241364981602144e-05, "loss": 0.1644, "step": 10112 }, { "epoch": 62.81619937694704, "grad_norm": 2.828125, "learning_rate": 3.923933998509713e-05, "loss": 0.144, "step": 10113 }, { "epoch": 62.822429906542055, "grad_norm": 2.6875, "learning_rate": 3.923731485029682e-05, "loss": 0.2831, "step": 10114 }, { "epoch": 62.82866043613707, "grad_norm": 2.65625, "learning_rate": 3.923528957722086e-05, "loss": 0.1681, "step": 10115 }, { "epoch": 62.834890965732086, "grad_norm": 2.484375, "learning_rate": 3.9233264165888906e-05, "loss": 0.2066, "step": 10116 }, { "epoch": 62.8411214953271, "grad_norm": 2.5625, "learning_rate": 3.9231238616320656e-05, "loss": 0.1497, "step": 10117 }, { "epoch": 62.84735202492212, "grad_norm": 3.296875, "learning_rate": 3.922921292853577e-05, "loss": 0.182, "step": 10118 }, { "epoch": 62.85358255451713, "grad_norm": 3.515625, "learning_rate": 3.922718710255393e-05, "loss": 0.1689, "step": 10119 }, { "epoch": 62.85981308411215, "grad_norm": 2.640625, "learning_rate": 3.92251611383948e-05, "loss": 0.1616, "step": 10120 }, { "epoch": 62.86604361370716, "grad_norm": 2.40625, "learning_rate": 3.9223135036078064e-05, "loss": 0.2711, "step": 10121 }, { "epoch": 62.87227414330218, "grad_norm": 2.796875, "learning_rate": 3.92211087956234e-05, "loss": 0.1332, "step": 10122 }, { "epoch": 62.87850467289719, "grad_norm": 3.578125, "learning_rate": 3.921908241705047e-05, "loss": 0.1994, "step": 10123 }, { "epoch": 62.88473520249221, "grad_norm": 3.453125, "learning_rate": 3.921705590037897e-05, "loss": 0.1833, "step": 10124 }, { "epoch": 62.890965732087224, "grad_norm": 3.40625, "learning_rate": 3.92150292456286e-05, "loss": 0.1438, "step": 10125 }, { "epoch": 62.89719626168224, "grad_norm": 1.859375, "learning_rate": 3.9213002452819014e-05, "loss": 0.1692, "step": 10126 }, { "epoch": 62.90342679127726, "grad_norm": 2.4375, "learning_rate": 3.9210975521969904e-05, "loss": 0.1442, "step": 10127 }, { "epoch": 62.90965732087228, "grad_norm": 1.3671875, "learning_rate": 3.920894845310097e-05, "loss": 0.1281, "step": 10128 }, { "epoch": 62.91588785046729, "grad_norm": 3.5625, "learning_rate": 3.920692124623188e-05, "loss": 0.2382, "step": 10129 }, { "epoch": 62.92211838006231, "grad_norm": 2.21875, "learning_rate": 3.9204893901382324e-05, "loss": 0.1457, "step": 10130 }, { "epoch": 62.92834890965732, "grad_norm": 2.171875, "learning_rate": 3.920286641857201e-05, "loss": 0.1941, "step": 10131 }, { "epoch": 62.93457943925234, "grad_norm": 2.921875, "learning_rate": 3.920083879782061e-05, "loss": 0.177, "step": 10132 }, { "epoch": 62.940809968847354, "grad_norm": 2.484375, "learning_rate": 3.919881103914783e-05, "loss": 0.1593, "step": 10133 }, { "epoch": 62.94704049844237, "grad_norm": 2.125, "learning_rate": 3.9196783142573355e-05, "loss": 0.1917, "step": 10134 }, { "epoch": 62.953271028037385, "grad_norm": 2.921875, "learning_rate": 3.9194755108116895e-05, "loss": 0.1527, "step": 10135 }, { "epoch": 62.9595015576324, "grad_norm": 1.65625, "learning_rate": 3.919272693579813e-05, "loss": 0.1244, "step": 10136 }, { "epoch": 62.965732087227416, "grad_norm": 2.859375, "learning_rate": 3.919069862563676e-05, "loss": 0.1867, "step": 10137 }, { "epoch": 62.97196261682243, "grad_norm": 2.546875, "learning_rate": 3.918867017765249e-05, "loss": 0.2587, "step": 10138 }, { "epoch": 62.978193146417446, "grad_norm": 1.8984375, "learning_rate": 3.918664159186502e-05, "loss": 0.104, "step": 10139 }, { "epoch": 62.98442367601246, "grad_norm": 2.59375, "learning_rate": 3.918461286829405e-05, "loss": 0.2081, "step": 10140 }, { "epoch": 62.99065420560748, "grad_norm": 0.9296875, "learning_rate": 3.9182584006959287e-05, "loss": 0.1171, "step": 10141 }, { "epoch": 62.99688473520249, "grad_norm": 2.28125, "learning_rate": 3.918055500788043e-05, "loss": 0.1375, "step": 10142 }, { "epoch": 63.0, "grad_norm": 1.078125, "learning_rate": 3.917852587107719e-05, "loss": 0.0487, "step": 10143 }, { "epoch": 63.006230529595015, "grad_norm": 1.5703125, "learning_rate": 3.9176496596569265e-05, "loss": 0.1373, "step": 10144 }, { "epoch": 63.01246105919003, "grad_norm": 2.8125, "learning_rate": 3.9174467184376384e-05, "loss": 0.2556, "step": 10145 }, { "epoch": 63.018691588785046, "grad_norm": 6.125, "learning_rate": 3.917243763451824e-05, "loss": 0.388, "step": 10146 }, { "epoch": 63.02492211838006, "grad_norm": 2.171875, "learning_rate": 3.917040794701455e-05, "loss": 0.2492, "step": 10147 }, { "epoch": 63.03115264797508, "grad_norm": 2.390625, "learning_rate": 3.916837812188502e-05, "loss": 0.1146, "step": 10148 }, { "epoch": 63.03738317757009, "grad_norm": 2.765625, "learning_rate": 3.916634815914937e-05, "loss": 0.1533, "step": 10149 }, { "epoch": 63.04361370716511, "grad_norm": 2.671875, "learning_rate": 3.916431805882732e-05, "loss": 0.1944, "step": 10150 }, { "epoch": 63.04984423676012, "grad_norm": 2.9375, "learning_rate": 3.916228782093857e-05, "loss": 0.2409, "step": 10151 }, { "epoch": 63.05607476635514, "grad_norm": 1.90625, "learning_rate": 3.916025744550286e-05, "loss": 0.187, "step": 10152 }, { "epoch": 63.06230529595015, "grad_norm": 2.890625, "learning_rate": 3.91582269325399e-05, "loss": 0.2353, "step": 10153 }, { "epoch": 63.06853582554517, "grad_norm": 3.328125, "learning_rate": 3.915619628206941e-05, "loss": 0.1364, "step": 10154 }, { "epoch": 63.074766355140184, "grad_norm": 1.2734375, "learning_rate": 3.91541654941111e-05, "loss": 0.1051, "step": 10155 }, { "epoch": 63.0809968847352, "grad_norm": 2.515625, "learning_rate": 3.915213456868472e-05, "loss": 0.1867, "step": 10156 }, { "epoch": 63.087227414330215, "grad_norm": 3.25, "learning_rate": 3.915010350580998e-05, "loss": 0.1428, "step": 10157 }, { "epoch": 63.09345794392523, "grad_norm": 1.4453125, "learning_rate": 3.91480723055066e-05, "loss": 0.1677, "step": 10158 }, { "epoch": 63.099688473520246, "grad_norm": 1.796875, "learning_rate": 3.914604096779431e-05, "loss": 0.099, "step": 10159 }, { "epoch": 63.10591900311527, "grad_norm": 1.5234375, "learning_rate": 3.914400949269284e-05, "loss": 0.1115, "step": 10160 }, { "epoch": 63.11214953271028, "grad_norm": 3.703125, "learning_rate": 3.914197788022194e-05, "loss": 0.2627, "step": 10161 }, { "epoch": 63.1183800623053, "grad_norm": 1.3359375, "learning_rate": 3.9139946130401305e-05, "loss": 0.1171, "step": 10162 }, { "epoch": 63.124610591900314, "grad_norm": 1.9765625, "learning_rate": 3.91379142432507e-05, "loss": 0.149, "step": 10163 }, { "epoch": 63.13084112149533, "grad_norm": 1.671875, "learning_rate": 3.913588221878984e-05, "loss": 0.1348, "step": 10164 }, { "epoch": 63.137071651090345, "grad_norm": 2.984375, "learning_rate": 3.9133850057038465e-05, "loss": 0.211, "step": 10165 }, { "epoch": 63.14330218068536, "grad_norm": 2.78125, "learning_rate": 3.913181775801632e-05, "loss": 0.1551, "step": 10166 }, { "epoch": 63.149532710280376, "grad_norm": 1.2734375, "learning_rate": 3.9129785321743125e-05, "loss": 0.1122, "step": 10167 }, { "epoch": 63.15576323987539, "grad_norm": 2.203125, "learning_rate": 3.912775274823864e-05, "loss": 0.1311, "step": 10168 }, { "epoch": 63.161993769470406, "grad_norm": 2.515625, "learning_rate": 3.9125720037522596e-05, "loss": 0.1371, "step": 10169 }, { "epoch": 63.16822429906542, "grad_norm": 1.796875, "learning_rate": 3.9123687189614726e-05, "loss": 0.1216, "step": 10170 }, { "epoch": 63.17445482866044, "grad_norm": 3.453125, "learning_rate": 3.912165420453478e-05, "loss": 0.1843, "step": 10171 }, { "epoch": 63.18068535825545, "grad_norm": 2.515625, "learning_rate": 3.911962108230253e-05, "loss": 0.1357, "step": 10172 }, { "epoch": 63.18691588785047, "grad_norm": 3.28125, "learning_rate": 3.9117587822937676e-05, "loss": 0.2556, "step": 10173 }, { "epoch": 63.19314641744548, "grad_norm": 3.15625, "learning_rate": 3.911555442645999e-05, "loss": 0.3838, "step": 10174 }, { "epoch": 63.1993769470405, "grad_norm": 2.296875, "learning_rate": 3.911352089288922e-05, "loss": 0.1714, "step": 10175 }, { "epoch": 63.205607476635514, "grad_norm": 2.4375, "learning_rate": 3.911148722224511e-05, "loss": 0.2136, "step": 10176 }, { "epoch": 63.21183800623053, "grad_norm": 1.6640625, "learning_rate": 3.9109453414547425e-05, "loss": 0.1467, "step": 10177 }, { "epoch": 63.218068535825545, "grad_norm": 1.609375, "learning_rate": 3.91074194698159e-05, "loss": 0.1409, "step": 10178 }, { "epoch": 63.22429906542056, "grad_norm": 2.640625, "learning_rate": 3.9105385388070305e-05, "loss": 0.2202, "step": 10179 }, { "epoch": 63.230529595015575, "grad_norm": 1.953125, "learning_rate": 3.910335116933038e-05, "loss": 0.1591, "step": 10180 }, { "epoch": 63.23676012461059, "grad_norm": 1.640625, "learning_rate": 3.910131681361589e-05, "loss": 0.1191, "step": 10181 }, { "epoch": 63.242990654205606, "grad_norm": 1.140625, "learning_rate": 3.9099282320946595e-05, "loss": 0.1157, "step": 10182 }, { "epoch": 63.24922118380062, "grad_norm": 3.71875, "learning_rate": 3.909724769134225e-05, "loss": 0.1974, "step": 10183 }, { "epoch": 63.25545171339564, "grad_norm": 3.40625, "learning_rate": 3.9095212924822614e-05, "loss": 0.3156, "step": 10184 }, { "epoch": 63.26168224299065, "grad_norm": 4.15625, "learning_rate": 3.9093178021407464e-05, "loss": 0.1267, "step": 10185 }, { "epoch": 63.26791277258567, "grad_norm": 4.3125, "learning_rate": 3.909114298111654e-05, "loss": 0.2359, "step": 10186 }, { "epoch": 63.27414330218068, "grad_norm": 1.9375, "learning_rate": 3.908910780396962e-05, "loss": 0.1272, "step": 10187 }, { "epoch": 63.2803738317757, "grad_norm": 3.03125, "learning_rate": 3.908707248998647e-05, "loss": 0.1146, "step": 10188 }, { "epoch": 63.28660436137071, "grad_norm": 4.03125, "learning_rate": 3.908503703918686e-05, "loss": 0.1909, "step": 10189 }, { "epoch": 63.29283489096573, "grad_norm": 3.5625, "learning_rate": 3.908300145159055e-05, "loss": 0.2291, "step": 10190 }, { "epoch": 63.299065420560744, "grad_norm": 3.828125, "learning_rate": 3.908096572721732e-05, "loss": 0.216, "step": 10191 }, { "epoch": 63.30529595015577, "grad_norm": 2.171875, "learning_rate": 3.9078929866086935e-05, "loss": 0.1961, "step": 10192 }, { "epoch": 63.31152647975078, "grad_norm": 1.53125, "learning_rate": 3.9076893868219165e-05, "loss": 0.126, "step": 10193 }, { "epoch": 63.3177570093458, "grad_norm": 1.890625, "learning_rate": 3.9074857733633786e-05, "loss": 0.1101, "step": 10194 }, { "epoch": 63.32398753894081, "grad_norm": 2.578125, "learning_rate": 3.907282146235058e-05, "loss": 0.1551, "step": 10195 }, { "epoch": 63.33021806853583, "grad_norm": 2.75, "learning_rate": 3.907078505438932e-05, "loss": 0.1796, "step": 10196 }, { "epoch": 63.33644859813084, "grad_norm": 3.171875, "learning_rate": 3.906874850976978e-05, "loss": 0.1916, "step": 10197 }, { "epoch": 63.34267912772586, "grad_norm": 1.9296875, "learning_rate": 3.9066711828511745e-05, "loss": 0.1685, "step": 10198 }, { "epoch": 63.348909657320874, "grad_norm": 2.90625, "learning_rate": 3.906467501063499e-05, "loss": 0.2988, "step": 10199 }, { "epoch": 63.35514018691589, "grad_norm": 2.8125, "learning_rate": 3.906263805615929e-05, "loss": 0.2851, "step": 10200 }, { "epoch": 63.361370716510905, "grad_norm": 3.515625, "learning_rate": 3.906060096510446e-05, "loss": 0.1755, "step": 10201 }, { "epoch": 63.36760124610592, "grad_norm": 2.59375, "learning_rate": 3.905856373749025e-05, "loss": 0.2194, "step": 10202 }, { "epoch": 63.373831775700936, "grad_norm": 2.359375, "learning_rate": 3.905652637333645e-05, "loss": 0.2443, "step": 10203 }, { "epoch": 63.38006230529595, "grad_norm": 2.90625, "learning_rate": 3.9054488872662874e-05, "loss": 0.1931, "step": 10204 }, { "epoch": 63.386292834890966, "grad_norm": 2.125, "learning_rate": 3.9052451235489286e-05, "loss": 0.1796, "step": 10205 }, { "epoch": 63.39252336448598, "grad_norm": 3.0625, "learning_rate": 3.905041346183548e-05, "loss": 0.2386, "step": 10206 }, { "epoch": 63.398753894081, "grad_norm": 2.96875, "learning_rate": 3.904837555172125e-05, "loss": 0.1896, "step": 10207 }, { "epoch": 63.40498442367601, "grad_norm": 2.5625, "learning_rate": 3.90463375051664e-05, "loss": 0.1948, "step": 10208 }, { "epoch": 63.41121495327103, "grad_norm": 2.234375, "learning_rate": 3.904429932219069e-05, "loss": 0.2512, "step": 10209 }, { "epoch": 63.41744548286604, "grad_norm": 2.546875, "learning_rate": 3.904226100281395e-05, "loss": 0.2209, "step": 10210 }, { "epoch": 63.42367601246106, "grad_norm": 2.421875, "learning_rate": 3.904022254705597e-05, "loss": 0.1356, "step": 10211 }, { "epoch": 63.429906542056074, "grad_norm": 2.453125, "learning_rate": 3.903818395493654e-05, "loss": 0.2212, "step": 10212 }, { "epoch": 63.43613707165109, "grad_norm": 2.90625, "learning_rate": 3.903614522647545e-05, "loss": 0.2272, "step": 10213 }, { "epoch": 63.442367601246104, "grad_norm": 2.09375, "learning_rate": 3.9034106361692524e-05, "loss": 0.1312, "step": 10214 }, { "epoch": 63.44859813084112, "grad_norm": 2.953125, "learning_rate": 3.903206736060754e-05, "loss": 0.1365, "step": 10215 }, { "epoch": 63.454828660436135, "grad_norm": 3.71875, "learning_rate": 3.9030028223240325e-05, "loss": 0.3613, "step": 10216 }, { "epoch": 63.46105919003115, "grad_norm": 2.953125, "learning_rate": 3.9027988949610663e-05, "loss": 0.2655, "step": 10217 }, { "epoch": 63.467289719626166, "grad_norm": 2.21875, "learning_rate": 3.902594953973837e-05, "loss": 0.1435, "step": 10218 }, { "epoch": 63.47352024922118, "grad_norm": 2.078125, "learning_rate": 3.9023909993643246e-05, "loss": 0.1793, "step": 10219 }, { "epoch": 63.4797507788162, "grad_norm": 2.28125, "learning_rate": 3.902187031134512e-05, "loss": 0.2718, "step": 10220 }, { "epoch": 63.48598130841121, "grad_norm": 3.0625, "learning_rate": 3.901983049286378e-05, "loss": 0.1916, "step": 10221 }, { "epoch": 63.49221183800623, "grad_norm": 1.6875, "learning_rate": 3.9017790538219036e-05, "loss": 0.1214, "step": 10222 }, { "epoch": 63.49844236760124, "grad_norm": 1.984375, "learning_rate": 3.901575044743072e-05, "loss": 0.1367, "step": 10223 }, { "epoch": 63.504672897196265, "grad_norm": 3.328125, "learning_rate": 3.901371022051862e-05, "loss": 0.2322, "step": 10224 }, { "epoch": 63.51090342679128, "grad_norm": 2.9375, "learning_rate": 3.901166985750257e-05, "loss": 0.1823, "step": 10225 }, { "epoch": 63.517133956386296, "grad_norm": 2.734375, "learning_rate": 3.900962935840239e-05, "loss": 0.1733, "step": 10226 }, { "epoch": 63.52336448598131, "grad_norm": 2.03125, "learning_rate": 3.900758872323788e-05, "loss": 0.1693, "step": 10227 }, { "epoch": 63.52959501557633, "grad_norm": 1.71875, "learning_rate": 3.900554795202887e-05, "loss": 0.1463, "step": 10228 }, { "epoch": 63.53582554517134, "grad_norm": 2.3125, "learning_rate": 3.900350704479519e-05, "loss": 0.1546, "step": 10229 }, { "epoch": 63.54205607476636, "grad_norm": 2.546875, "learning_rate": 3.900146600155663e-05, "loss": 0.1759, "step": 10230 }, { "epoch": 63.54828660436137, "grad_norm": 3.390625, "learning_rate": 3.899942482233305e-05, "loss": 0.1742, "step": 10231 }, { "epoch": 63.55451713395639, "grad_norm": 2.109375, "learning_rate": 3.899738350714425e-05, "loss": 0.1527, "step": 10232 }, { "epoch": 63.5607476635514, "grad_norm": 2.484375, "learning_rate": 3.8995342056010065e-05, "loss": 0.1594, "step": 10233 }, { "epoch": 63.56697819314642, "grad_norm": 2.046875, "learning_rate": 3.8993300468950316e-05, "loss": 0.1616, "step": 10234 }, { "epoch": 63.573208722741434, "grad_norm": 2.40625, "learning_rate": 3.8991258745984836e-05, "loss": 0.1653, "step": 10235 }, { "epoch": 63.57943925233645, "grad_norm": 2.53125, "learning_rate": 3.898921688713346e-05, "loss": 0.1472, "step": 10236 }, { "epoch": 63.585669781931465, "grad_norm": 3.015625, "learning_rate": 3.898717489241601e-05, "loss": 0.3108, "step": 10237 }, { "epoch": 63.59190031152648, "grad_norm": 2.484375, "learning_rate": 3.898513276185233e-05, "loss": 0.137, "step": 10238 }, { "epoch": 63.598130841121495, "grad_norm": 2.265625, "learning_rate": 3.8983090495462235e-05, "loss": 0.1535, "step": 10239 }, { "epoch": 63.60436137071651, "grad_norm": 1.4921875, "learning_rate": 3.898104809326557e-05, "loss": 0.1317, "step": 10240 }, { "epoch": 63.610591900311526, "grad_norm": 2.828125, "learning_rate": 3.897900555528218e-05, "loss": 0.2108, "step": 10241 }, { "epoch": 63.61682242990654, "grad_norm": 1.8203125, "learning_rate": 3.897696288153189e-05, "loss": 0.1548, "step": 10242 }, { "epoch": 63.62305295950156, "grad_norm": 2.46875, "learning_rate": 3.8974920072034535e-05, "loss": 0.1188, "step": 10243 }, { "epoch": 63.62928348909657, "grad_norm": 2.1875, "learning_rate": 3.897287712680997e-05, "loss": 0.1225, "step": 10244 }, { "epoch": 63.63551401869159, "grad_norm": 3.03125, "learning_rate": 3.8970834045878024e-05, "loss": 0.2417, "step": 10245 }, { "epoch": 63.6417445482866, "grad_norm": 2.34375, "learning_rate": 3.896879082925855e-05, "loss": 0.1719, "step": 10246 }, { "epoch": 63.64797507788162, "grad_norm": 1.640625, "learning_rate": 3.8966747476971384e-05, "loss": 0.1032, "step": 10247 }, { "epoch": 63.654205607476634, "grad_norm": 3.34375, "learning_rate": 3.896470398903638e-05, "loss": 0.2397, "step": 10248 }, { "epoch": 63.66043613707165, "grad_norm": 3.078125, "learning_rate": 3.896266036547337e-05, "loss": 0.2296, "step": 10249 }, { "epoch": 63.666666666666664, "grad_norm": 2.234375, "learning_rate": 3.896061660630221e-05, "loss": 0.2139, "step": 10250 }, { "epoch": 63.67289719626168, "grad_norm": 2.171875, "learning_rate": 3.895857271154276e-05, "loss": 0.1504, "step": 10251 }, { "epoch": 63.679127725856695, "grad_norm": 2.921875, "learning_rate": 3.8956528681214866e-05, "loss": 0.2667, "step": 10252 }, { "epoch": 63.68535825545171, "grad_norm": 2.75, "learning_rate": 3.8954484515338366e-05, "loss": 0.1857, "step": 10253 }, { "epoch": 63.691588785046726, "grad_norm": 5.5625, "learning_rate": 3.8952440213933125e-05, "loss": 0.2903, "step": 10254 }, { "epoch": 63.69781931464174, "grad_norm": 2.671875, "learning_rate": 3.8950395777019e-05, "loss": 0.1413, "step": 10255 }, { "epoch": 63.70404984423676, "grad_norm": 2.625, "learning_rate": 3.894835120461584e-05, "loss": 0.19, "step": 10256 }, { "epoch": 63.71028037383178, "grad_norm": 3.171875, "learning_rate": 3.89463064967435e-05, "loss": 0.2382, "step": 10257 }, { "epoch": 63.716510903426794, "grad_norm": 3.90625, "learning_rate": 3.894426165342185e-05, "loss": 0.1708, "step": 10258 }, { "epoch": 63.72274143302181, "grad_norm": 4.34375, "learning_rate": 3.894221667467074e-05, "loss": 0.4967, "step": 10259 }, { "epoch": 63.728971962616825, "grad_norm": 2.734375, "learning_rate": 3.894017156051003e-05, "loss": 0.1166, "step": 10260 }, { "epoch": 63.73520249221184, "grad_norm": 2.9375, "learning_rate": 3.8938126310959595e-05, "loss": 0.3321, "step": 10261 }, { "epoch": 63.741433021806856, "grad_norm": 2.671875, "learning_rate": 3.893608092603929e-05, "loss": 0.1718, "step": 10262 }, { "epoch": 63.74766355140187, "grad_norm": 1.46875, "learning_rate": 3.893403540576898e-05, "loss": 0.1142, "step": 10263 }, { "epoch": 63.753894080996886, "grad_norm": 3.078125, "learning_rate": 3.893198975016853e-05, "loss": 0.2696, "step": 10264 }, { "epoch": 63.7601246105919, "grad_norm": 2.421875, "learning_rate": 3.892994395925781e-05, "loss": 0.3183, "step": 10265 }, { "epoch": 63.76635514018692, "grad_norm": 3.34375, "learning_rate": 3.89278980330567e-05, "loss": 0.2625, "step": 10266 }, { "epoch": 63.77258566978193, "grad_norm": 2.046875, "learning_rate": 3.892585197158505e-05, "loss": 0.1182, "step": 10267 }, { "epoch": 63.77881619937695, "grad_norm": 1.9765625, "learning_rate": 3.892380577486275e-05, "loss": 0.1509, "step": 10268 }, { "epoch": 63.78504672897196, "grad_norm": 2.390625, "learning_rate": 3.892175944290966e-05, "loss": 0.1908, "step": 10269 }, { "epoch": 63.79127725856698, "grad_norm": 2.40625, "learning_rate": 3.8919712975745654e-05, "loss": 0.185, "step": 10270 }, { "epoch": 63.797507788161994, "grad_norm": 2.375, "learning_rate": 3.891766637339062e-05, "loss": 0.2089, "step": 10271 }, { "epoch": 63.80373831775701, "grad_norm": 3.65625, "learning_rate": 3.891561963586443e-05, "loss": 0.1587, "step": 10272 }, { "epoch": 63.809968847352025, "grad_norm": 2.953125, "learning_rate": 3.8913572763186955e-05, "loss": 0.1554, "step": 10273 }, { "epoch": 63.81619937694704, "grad_norm": 1.3203125, "learning_rate": 3.8911525755378084e-05, "loss": 0.1015, "step": 10274 }, { "epoch": 63.822429906542055, "grad_norm": 1.1484375, "learning_rate": 3.8909478612457686e-05, "loss": 0.1218, "step": 10275 }, { "epoch": 63.82866043613707, "grad_norm": 2.140625, "learning_rate": 3.890743133444566e-05, "loss": 0.1113, "step": 10276 }, { "epoch": 63.834890965732086, "grad_norm": 3.40625, "learning_rate": 3.890538392136188e-05, "loss": 0.2903, "step": 10277 }, { "epoch": 63.8411214953271, "grad_norm": 1.3359375, "learning_rate": 3.8903336373226226e-05, "loss": 0.1065, "step": 10278 }, { "epoch": 63.84735202492212, "grad_norm": 2.578125, "learning_rate": 3.89012886900586e-05, "loss": 0.1119, "step": 10279 }, { "epoch": 63.85358255451713, "grad_norm": 2.8125, "learning_rate": 3.8899240871878875e-05, "loss": 0.2006, "step": 10280 }, { "epoch": 63.85981308411215, "grad_norm": 1.7890625, "learning_rate": 3.889719291870694e-05, "loss": 0.1903, "step": 10281 }, { "epoch": 63.86604361370716, "grad_norm": 2.28125, "learning_rate": 3.889514483056269e-05, "loss": 0.1391, "step": 10282 }, { "epoch": 63.87227414330218, "grad_norm": 1.8671875, "learning_rate": 3.889309660746602e-05, "loss": 0.1683, "step": 10283 }, { "epoch": 63.87850467289719, "grad_norm": 2.421875, "learning_rate": 3.889104824943681e-05, "loss": 0.2223, "step": 10284 }, { "epoch": 63.88473520249221, "grad_norm": 2.515625, "learning_rate": 3.888899975649497e-05, "loss": 0.2331, "step": 10285 }, { "epoch": 63.890965732087224, "grad_norm": 2.1875, "learning_rate": 3.888695112866039e-05, "loss": 0.1379, "step": 10286 }, { "epoch": 63.89719626168224, "grad_norm": 1.875, "learning_rate": 3.8884902365952965e-05, "loss": 0.1296, "step": 10287 }, { "epoch": 63.90342679127726, "grad_norm": 2.1875, "learning_rate": 3.888285346839259e-05, "loss": 0.2024, "step": 10288 }, { "epoch": 63.90965732087228, "grad_norm": 1.9453125, "learning_rate": 3.8880804435999164e-05, "loss": 0.17, "step": 10289 }, { "epoch": 63.91588785046729, "grad_norm": 1.765625, "learning_rate": 3.8878755268792585e-05, "loss": 0.1238, "step": 10290 }, { "epoch": 63.92211838006231, "grad_norm": 2.140625, "learning_rate": 3.8876705966792774e-05, "loss": 0.1832, "step": 10291 }, { "epoch": 63.92834890965732, "grad_norm": 3.125, "learning_rate": 3.887465653001961e-05, "loss": 0.188, "step": 10292 }, { "epoch": 63.93457943925234, "grad_norm": 1.765625, "learning_rate": 3.887260695849302e-05, "loss": 0.11, "step": 10293 }, { "epoch": 63.940809968847354, "grad_norm": 2.03125, "learning_rate": 3.887055725223289e-05, "loss": 0.1481, "step": 10294 }, { "epoch": 63.94704049844237, "grad_norm": 2.3125, "learning_rate": 3.8868507411259136e-05, "loss": 0.2747, "step": 10295 }, { "epoch": 63.953271028037385, "grad_norm": 1.78125, "learning_rate": 3.886645743559166e-05, "loss": 0.1329, "step": 10296 }, { "epoch": 63.9595015576324, "grad_norm": 2.25, "learning_rate": 3.88644073252504e-05, "loss": 0.1111, "step": 10297 }, { "epoch": 63.965732087227416, "grad_norm": 2.015625, "learning_rate": 3.886235708025522e-05, "loss": 0.1821, "step": 10298 }, { "epoch": 63.97196261682243, "grad_norm": 0.9375, "learning_rate": 3.886030670062606e-05, "loss": 0.0999, "step": 10299 }, { "epoch": 63.978193146417446, "grad_norm": 1.9140625, "learning_rate": 3.8858256186382844e-05, "loss": 0.1064, "step": 10300 }, { "epoch": 63.98442367601246, "grad_norm": 2.125, "learning_rate": 3.885620553754547e-05, "loss": 0.1416, "step": 10301 }, { "epoch": 63.99065420560748, "grad_norm": 2.359375, "learning_rate": 3.8854154754133856e-05, "loss": 0.2301, "step": 10302 }, { "epoch": 63.99688473520249, "grad_norm": 2.25, "learning_rate": 3.885210383616792e-05, "loss": 0.218, "step": 10303 }, { "epoch": 64.0, "grad_norm": 1.9609375, "learning_rate": 3.8850052783667576e-05, "loss": 0.104, "step": 10304 }, { "epoch": 64.00623052959502, "grad_norm": 2.390625, "learning_rate": 3.884800159665276e-05, "loss": 0.239, "step": 10305 }, { "epoch": 64.01246105919003, "grad_norm": 1.421875, "learning_rate": 3.884595027514338e-05, "loss": 0.1357, "step": 10306 }, { "epoch": 64.01869158878505, "grad_norm": 2.53125, "learning_rate": 3.884389881915936e-05, "loss": 0.2193, "step": 10307 }, { "epoch": 64.02492211838006, "grad_norm": 2.390625, "learning_rate": 3.884184722872064e-05, "loss": 0.1819, "step": 10308 }, { "epoch": 64.03115264797508, "grad_norm": 1.515625, "learning_rate": 3.883979550384712e-05, "loss": 0.1252, "step": 10309 }, { "epoch": 64.03738317757009, "grad_norm": 1.7109375, "learning_rate": 3.883774364455874e-05, "loss": 0.123, "step": 10310 }, { "epoch": 64.04361370716511, "grad_norm": 2.375, "learning_rate": 3.883569165087544e-05, "loss": 0.2006, "step": 10311 }, { "epoch": 64.04984423676012, "grad_norm": 2.59375, "learning_rate": 3.8833639522817135e-05, "loss": 0.2702, "step": 10312 }, { "epoch": 64.05607476635514, "grad_norm": 1.9921875, "learning_rate": 3.8831587260403746e-05, "loss": 0.1913, "step": 10313 }, { "epoch": 64.06230529595015, "grad_norm": 1.6484375, "learning_rate": 3.882953486365523e-05, "loss": 0.1434, "step": 10314 }, { "epoch": 64.06853582554517, "grad_norm": 2.734375, "learning_rate": 3.882748233259149e-05, "loss": 0.1925, "step": 10315 }, { "epoch": 64.07476635514018, "grad_norm": 1.265625, "learning_rate": 3.882542966723249e-05, "loss": 0.1018, "step": 10316 }, { "epoch": 64.0809968847352, "grad_norm": 2.421875, "learning_rate": 3.882337686759815e-05, "loss": 0.1733, "step": 10317 }, { "epoch": 64.08722741433021, "grad_norm": 2.4375, "learning_rate": 3.882132393370841e-05, "loss": 0.1376, "step": 10318 }, { "epoch": 64.09345794392523, "grad_norm": 2.109375, "learning_rate": 3.8819270865583214e-05, "loss": 0.159, "step": 10319 }, { "epoch": 64.09968847352025, "grad_norm": 2.125, "learning_rate": 3.881721766324249e-05, "loss": 0.2275, "step": 10320 }, { "epoch": 64.10591900311526, "grad_norm": 1.1171875, "learning_rate": 3.881516432670619e-05, "loss": 0.0985, "step": 10321 }, { "epoch": 64.11214953271028, "grad_norm": 1.9375, "learning_rate": 3.881311085599425e-05, "loss": 0.1432, "step": 10322 }, { "epoch": 64.11838006230529, "grad_norm": 1.4140625, "learning_rate": 3.8811057251126615e-05, "loss": 0.115, "step": 10323 }, { "epoch": 64.1246105919003, "grad_norm": 3.109375, "learning_rate": 3.880900351212323e-05, "loss": 0.326, "step": 10324 }, { "epoch": 64.13084112149532, "grad_norm": 2.578125, "learning_rate": 3.880694963900405e-05, "loss": 0.2358, "step": 10325 }, { "epoch": 64.13707165109034, "grad_norm": 2.390625, "learning_rate": 3.880489563178901e-05, "loss": 0.1658, "step": 10326 }, { "epoch": 64.14330218068535, "grad_norm": 1.34375, "learning_rate": 3.880284149049806e-05, "loss": 0.1136, "step": 10327 }, { "epoch": 64.14953271028037, "grad_norm": 2.359375, "learning_rate": 3.880078721515117e-05, "loss": 0.2165, "step": 10328 }, { "epoch": 64.15576323987538, "grad_norm": 2.421875, "learning_rate": 3.8798732805768266e-05, "loss": 0.1824, "step": 10329 }, { "epoch": 64.1619937694704, "grad_norm": 2.25, "learning_rate": 3.879667826236931e-05, "loss": 0.205, "step": 10330 }, { "epoch": 64.16822429906541, "grad_norm": 3.015625, "learning_rate": 3.8794623584974254e-05, "loss": 0.2704, "step": 10331 }, { "epoch": 64.17445482866043, "grad_norm": 2.484375, "learning_rate": 3.879256877360307e-05, "loss": 0.1605, "step": 10332 }, { "epoch": 64.18068535825545, "grad_norm": 1.7265625, "learning_rate": 3.8790513828275683e-05, "loss": 0.1178, "step": 10333 }, { "epoch": 64.18691588785046, "grad_norm": 2.484375, "learning_rate": 3.8788458749012073e-05, "loss": 0.1878, "step": 10334 }, { "epoch": 64.19314641744548, "grad_norm": 2.984375, "learning_rate": 3.87864035358322e-05, "loss": 0.1859, "step": 10335 }, { "epoch": 64.19937694704049, "grad_norm": 3.1875, "learning_rate": 3.878434818875602e-05, "loss": 0.269, "step": 10336 }, { "epoch": 64.20560747663552, "grad_norm": 2.921875, "learning_rate": 3.878229270780349e-05, "loss": 0.3583, "step": 10337 }, { "epoch": 64.21183800623054, "grad_norm": 1.796875, "learning_rate": 3.878023709299459e-05, "loss": 0.1311, "step": 10338 }, { "epoch": 64.21806853582555, "grad_norm": 1.2890625, "learning_rate": 3.877818134434926e-05, "loss": 0.1237, "step": 10339 }, { "epoch": 64.22429906542057, "grad_norm": 2.859375, "learning_rate": 3.8776125461887484e-05, "loss": 0.2452, "step": 10340 }, { "epoch": 64.23052959501558, "grad_norm": 2.21875, "learning_rate": 3.8774069445629225e-05, "loss": 0.2043, "step": 10341 }, { "epoch": 64.2367601246106, "grad_norm": 1.265625, "learning_rate": 3.8772013295594444e-05, "loss": 0.1159, "step": 10342 }, { "epoch": 64.24299065420561, "grad_norm": 2.5625, "learning_rate": 3.8769957011803125e-05, "loss": 0.1635, "step": 10343 }, { "epoch": 64.24922118380063, "grad_norm": 2.34375, "learning_rate": 3.876790059427523e-05, "loss": 0.2424, "step": 10344 }, { "epoch": 64.25545171339564, "grad_norm": 2.125, "learning_rate": 3.876584404303073e-05, "loss": 0.1717, "step": 10345 }, { "epoch": 64.26168224299066, "grad_norm": 2.765625, "learning_rate": 3.8763787358089596e-05, "loss": 0.2083, "step": 10346 }, { "epoch": 64.26791277258567, "grad_norm": 2.0625, "learning_rate": 3.8761730539471815e-05, "loss": 0.1535, "step": 10347 }, { "epoch": 64.27414330218069, "grad_norm": 2.34375, "learning_rate": 3.875967358719735e-05, "loss": 0.0916, "step": 10348 }, { "epoch": 64.2803738317757, "grad_norm": 4.03125, "learning_rate": 3.875761650128619e-05, "loss": 0.3252, "step": 10349 }, { "epoch": 64.28660436137072, "grad_norm": 2.265625, "learning_rate": 3.87555592817583e-05, "loss": 0.1692, "step": 10350 }, { "epoch": 64.29283489096574, "grad_norm": 2.140625, "learning_rate": 3.875350192863368e-05, "loss": 0.1878, "step": 10351 }, { "epoch": 64.29906542056075, "grad_norm": 3.078125, "learning_rate": 3.8751444441932295e-05, "loss": 0.2188, "step": 10352 }, { "epoch": 64.30529595015577, "grad_norm": 2.015625, "learning_rate": 3.874938682167413e-05, "loss": 0.1304, "step": 10353 }, { "epoch": 64.31152647975078, "grad_norm": 1.921875, "learning_rate": 3.874732906787918e-05, "loss": 0.1358, "step": 10354 }, { "epoch": 64.3177570093458, "grad_norm": 2.0625, "learning_rate": 3.874527118056741e-05, "loss": 0.1426, "step": 10355 }, { "epoch": 64.32398753894081, "grad_norm": 1.515625, "learning_rate": 3.874321315975883e-05, "loss": 0.1007, "step": 10356 }, { "epoch": 64.33021806853583, "grad_norm": 2.09375, "learning_rate": 3.874115500547341e-05, "loss": 0.1666, "step": 10357 }, { "epoch": 64.33644859813084, "grad_norm": 1.2421875, "learning_rate": 3.873909671773115e-05, "loss": 0.0981, "step": 10358 }, { "epoch": 64.34267912772586, "grad_norm": 2.703125, "learning_rate": 3.873703829655203e-05, "loss": 0.1127, "step": 10359 }, { "epoch": 64.34890965732087, "grad_norm": 2.921875, "learning_rate": 3.8734979741956054e-05, "loss": 0.123, "step": 10360 }, { "epoch": 64.35514018691589, "grad_norm": 2.984375, "learning_rate": 3.873292105396321e-05, "loss": 0.3132, "step": 10361 }, { "epoch": 64.3613707165109, "grad_norm": 1.9296875, "learning_rate": 3.873086223259349e-05, "loss": 0.1034, "step": 10362 }, { "epoch": 64.36760124610592, "grad_norm": 2.421875, "learning_rate": 3.872880327786689e-05, "loss": 0.1227, "step": 10363 }, { "epoch": 64.37383177570094, "grad_norm": 1.671875, "learning_rate": 3.8726744189803404e-05, "loss": 0.1277, "step": 10364 }, { "epoch": 64.38006230529595, "grad_norm": 2.109375, "learning_rate": 3.8724684968423036e-05, "loss": 0.1275, "step": 10365 }, { "epoch": 64.38629283489097, "grad_norm": 2.1875, "learning_rate": 3.872262561374579e-05, "loss": 0.1414, "step": 10366 }, { "epoch": 64.39252336448598, "grad_norm": 1.765625, "learning_rate": 3.872056612579166e-05, "loss": 0.1132, "step": 10367 }, { "epoch": 64.398753894081, "grad_norm": 3.890625, "learning_rate": 3.8718506504580646e-05, "loss": 0.1947, "step": 10368 }, { "epoch": 64.40498442367601, "grad_norm": 3.640625, "learning_rate": 3.871644675013275e-05, "loss": 0.2693, "step": 10369 }, { "epoch": 64.41121495327103, "grad_norm": 1.21875, "learning_rate": 3.8714386862467986e-05, "loss": 0.114, "step": 10370 }, { "epoch": 64.41744548286604, "grad_norm": 1.8671875, "learning_rate": 3.871232684160636e-05, "loss": 0.144, "step": 10371 }, { "epoch": 64.42367601246106, "grad_norm": 2.09375, "learning_rate": 3.871026668756786e-05, "loss": 0.1382, "step": 10372 }, { "epoch": 64.42990654205607, "grad_norm": 2.53125, "learning_rate": 3.8708206400372524e-05, "loss": 0.1124, "step": 10373 }, { "epoch": 64.43613707165109, "grad_norm": 2.03125, "learning_rate": 3.8706145980040344e-05, "loss": 0.1116, "step": 10374 }, { "epoch": 64.4423676012461, "grad_norm": 1.9609375, "learning_rate": 3.870408542659134e-05, "loss": 0.1551, "step": 10375 }, { "epoch": 64.44859813084112, "grad_norm": 3.25, "learning_rate": 3.8702024740045505e-05, "loss": 0.2018, "step": 10376 }, { "epoch": 64.45482866043614, "grad_norm": 3.203125, "learning_rate": 3.8699963920422874e-05, "loss": 0.2376, "step": 10377 }, { "epoch": 64.46105919003115, "grad_norm": 2.328125, "learning_rate": 3.869790296774346e-05, "loss": 0.1009, "step": 10378 }, { "epoch": 64.46728971962617, "grad_norm": 2.796875, "learning_rate": 3.8695841882027275e-05, "loss": 0.2739, "step": 10379 }, { "epoch": 64.47352024922118, "grad_norm": 3.796875, "learning_rate": 3.869378066329432e-05, "loss": 0.2535, "step": 10380 }, { "epoch": 64.4797507788162, "grad_norm": 2.921875, "learning_rate": 3.8691719311564644e-05, "loss": 0.2843, "step": 10381 }, { "epoch": 64.48598130841121, "grad_norm": 1.421875, "learning_rate": 3.868965782685825e-05, "loss": 0.1359, "step": 10382 }, { "epoch": 64.49221183800623, "grad_norm": 2.109375, "learning_rate": 3.8687596209195164e-05, "loss": 0.1414, "step": 10383 }, { "epoch": 64.49844236760124, "grad_norm": 2.21875, "learning_rate": 3.86855344585954e-05, "loss": 0.1358, "step": 10384 }, { "epoch": 64.50467289719626, "grad_norm": 2.1875, "learning_rate": 3.868347257507899e-05, "loss": 0.1348, "step": 10385 }, { "epoch": 64.51090342679127, "grad_norm": 2.5625, "learning_rate": 3.8681410558665974e-05, "loss": 0.1651, "step": 10386 }, { "epoch": 64.51713395638629, "grad_norm": 2.84375, "learning_rate": 3.8679348409376355e-05, "loss": 0.1873, "step": 10387 }, { "epoch": 64.5233644859813, "grad_norm": 2.140625, "learning_rate": 3.867728612723016e-05, "loss": 0.2696, "step": 10388 }, { "epoch": 64.52959501557632, "grad_norm": 2.140625, "learning_rate": 3.867522371224744e-05, "loss": 0.2079, "step": 10389 }, { "epoch": 64.53582554517133, "grad_norm": 2.234375, "learning_rate": 3.8673161164448215e-05, "loss": 0.176, "step": 10390 }, { "epoch": 64.54205607476635, "grad_norm": 2.59375, "learning_rate": 3.8671098483852505e-05, "loss": 0.1895, "step": 10391 }, { "epoch": 64.54828660436137, "grad_norm": 2.03125, "learning_rate": 3.866903567048037e-05, "loss": 0.131, "step": 10392 }, { "epoch": 64.55451713395638, "grad_norm": 3.46875, "learning_rate": 3.866697272435181e-05, "loss": 0.1695, "step": 10393 }, { "epoch": 64.5607476635514, "grad_norm": 1.5546875, "learning_rate": 3.866490964548689e-05, "loss": 0.1445, "step": 10394 }, { "epoch": 64.56697819314641, "grad_norm": 2.453125, "learning_rate": 3.866284643390563e-05, "loss": 0.2554, "step": 10395 }, { "epoch": 64.57320872274143, "grad_norm": 1.640625, "learning_rate": 3.866078308962808e-05, "loss": 0.1578, "step": 10396 }, { "epoch": 64.57943925233644, "grad_norm": 2.234375, "learning_rate": 3.865871961267427e-05, "loss": 0.124, "step": 10397 }, { "epoch": 64.58566978193146, "grad_norm": 3.5, "learning_rate": 3.8656656003064245e-05, "loss": 0.1796, "step": 10398 }, { "epoch": 64.59190031152647, "grad_norm": 2.625, "learning_rate": 3.8654592260818045e-05, "loss": 0.141, "step": 10399 }, { "epoch": 64.59813084112149, "grad_norm": 2.53125, "learning_rate": 3.865252838595572e-05, "loss": 0.1458, "step": 10400 }, { "epoch": 64.60436137071652, "grad_norm": 0.94921875, "learning_rate": 3.865046437849731e-05, "loss": 0.0927, "step": 10401 }, { "epoch": 64.61059190031153, "grad_norm": 2.59375, "learning_rate": 3.864840023846286e-05, "loss": 0.1069, "step": 10402 }, { "epoch": 64.61682242990655, "grad_norm": 2.0625, "learning_rate": 3.8646335965872414e-05, "loss": 0.1762, "step": 10403 }, { "epoch": 64.62305295950156, "grad_norm": 3.328125, "learning_rate": 3.864427156074603e-05, "loss": 0.3518, "step": 10404 }, { "epoch": 64.62928348909658, "grad_norm": 3.25, "learning_rate": 3.864220702310376e-05, "loss": 0.154, "step": 10405 }, { "epoch": 64.6355140186916, "grad_norm": 3.546875, "learning_rate": 3.8640142352965634e-05, "loss": 0.2108, "step": 10406 }, { "epoch": 64.64174454828661, "grad_norm": 4.34375, "learning_rate": 3.8638077550351725e-05, "loss": 0.2641, "step": 10407 }, { "epoch": 64.64797507788163, "grad_norm": 3.40625, "learning_rate": 3.863601261528209e-05, "loss": 0.193, "step": 10408 }, { "epoch": 64.65420560747664, "grad_norm": 2.96875, "learning_rate": 3.863394754777676e-05, "loss": 0.1255, "step": 10409 }, { "epoch": 64.66043613707166, "grad_norm": 2.75, "learning_rate": 3.863188234785582e-05, "loss": 0.2021, "step": 10410 }, { "epoch": 64.66666666666667, "grad_norm": 2.921875, "learning_rate": 3.862981701553931e-05, "loss": 0.1981, "step": 10411 }, { "epoch": 64.67289719626169, "grad_norm": 3.8125, "learning_rate": 3.862775155084728e-05, "loss": 0.3191, "step": 10412 }, { "epoch": 64.6791277258567, "grad_norm": 2.796875, "learning_rate": 3.8625685953799815e-05, "loss": 0.1765, "step": 10413 }, { "epoch": 64.68535825545172, "grad_norm": 3.84375, "learning_rate": 3.8623620224416957e-05, "loss": 0.2328, "step": 10414 }, { "epoch": 64.69158878504673, "grad_norm": 3.875, "learning_rate": 3.8621554362718784e-05, "loss": 0.1698, "step": 10415 }, { "epoch": 64.69781931464175, "grad_norm": 3.140625, "learning_rate": 3.861948836872534e-05, "loss": 0.2951, "step": 10416 }, { "epoch": 64.70404984423676, "grad_norm": 2.578125, "learning_rate": 3.861742224245671e-05, "loss": 0.1767, "step": 10417 }, { "epoch": 64.71028037383178, "grad_norm": 3.75, "learning_rate": 3.861535598393296e-05, "loss": 0.2697, "step": 10418 }, { "epoch": 64.7165109034268, "grad_norm": 2.5625, "learning_rate": 3.861328959317414e-05, "loss": 0.1922, "step": 10419 }, { "epoch": 64.72274143302181, "grad_norm": 2.421875, "learning_rate": 3.861122307020034e-05, "loss": 0.2158, "step": 10420 }, { "epoch": 64.72897196261682, "grad_norm": 3.921875, "learning_rate": 3.860915641503161e-05, "loss": 0.2524, "step": 10421 }, { "epoch": 64.73520249221184, "grad_norm": 2.65625, "learning_rate": 3.860708962768804e-05, "loss": 0.1526, "step": 10422 }, { "epoch": 64.74143302180686, "grad_norm": 2.375, "learning_rate": 3.86050227081897e-05, "loss": 0.2073, "step": 10423 }, { "epoch": 64.74766355140187, "grad_norm": 2.015625, "learning_rate": 3.860295565655665e-05, "loss": 0.0924, "step": 10424 }, { "epoch": 64.75389408099689, "grad_norm": 2.28125, "learning_rate": 3.860088847280898e-05, "loss": 0.1455, "step": 10425 }, { "epoch": 64.7601246105919, "grad_norm": 2.453125, "learning_rate": 3.8598821156966757e-05, "loss": 0.1598, "step": 10426 }, { "epoch": 64.76635514018692, "grad_norm": 2.875, "learning_rate": 3.859675370905007e-05, "loss": 0.1058, "step": 10427 }, { "epoch": 64.77258566978193, "grad_norm": 2.875, "learning_rate": 3.859468612907899e-05, "loss": 0.3011, "step": 10428 }, { "epoch": 64.77881619937695, "grad_norm": 3.109375, "learning_rate": 3.85926184170736e-05, "loss": 0.219, "step": 10429 }, { "epoch": 64.78504672897196, "grad_norm": 2.375, "learning_rate": 3.8590550573053994e-05, "loss": 0.2036, "step": 10430 }, { "epoch": 64.79127725856698, "grad_norm": 1.9140625, "learning_rate": 3.8588482597040235e-05, "loss": 0.1719, "step": 10431 }, { "epoch": 64.797507788162, "grad_norm": 3.640625, "learning_rate": 3.8586414489052414e-05, "loss": 0.2784, "step": 10432 }, { "epoch": 64.80373831775701, "grad_norm": 2.34375, "learning_rate": 3.858434624911062e-05, "loss": 0.1309, "step": 10433 }, { "epoch": 64.80996884735202, "grad_norm": 2.125, "learning_rate": 3.8582277877234946e-05, "loss": 0.1522, "step": 10434 }, { "epoch": 64.81619937694704, "grad_norm": 2.59375, "learning_rate": 3.8580209373445474e-05, "loss": 0.1727, "step": 10435 }, { "epoch": 64.82242990654206, "grad_norm": 3.34375, "learning_rate": 3.8578140737762294e-05, "loss": 0.1553, "step": 10436 }, { "epoch": 64.82866043613707, "grad_norm": 3.046875, "learning_rate": 3.857607197020549e-05, "loss": 0.1472, "step": 10437 }, { "epoch": 64.83489096573209, "grad_norm": 2.0, "learning_rate": 3.8574003070795165e-05, "loss": 0.1704, "step": 10438 }, { "epoch": 64.8411214953271, "grad_norm": 3.703125, "learning_rate": 3.8571934039551406e-05, "loss": 0.3057, "step": 10439 }, { "epoch": 64.84735202492212, "grad_norm": 2.1875, "learning_rate": 3.856986487649432e-05, "loss": 0.1371, "step": 10440 }, { "epoch": 64.85358255451713, "grad_norm": 4.21875, "learning_rate": 3.8567795581643986e-05, "loss": 0.2914, "step": 10441 }, { "epoch": 64.85981308411215, "grad_norm": 2.609375, "learning_rate": 3.856572615502051e-05, "loss": 0.18, "step": 10442 }, { "epoch": 64.86604361370716, "grad_norm": 2.46875, "learning_rate": 3.856365659664399e-05, "loss": 0.1446, "step": 10443 }, { "epoch": 64.87227414330218, "grad_norm": 2.515625, "learning_rate": 3.856158690653452e-05, "loss": 0.1771, "step": 10444 }, { "epoch": 64.8785046728972, "grad_norm": 1.8046875, "learning_rate": 3.8559517084712215e-05, "loss": 0.1522, "step": 10445 }, { "epoch": 64.88473520249221, "grad_norm": 2.140625, "learning_rate": 3.855744713119717e-05, "loss": 0.1507, "step": 10446 }, { "epoch": 64.89096573208722, "grad_norm": 3.1875, "learning_rate": 3.855537704600949e-05, "loss": 0.1669, "step": 10447 }, { "epoch": 64.89719626168224, "grad_norm": 3.765625, "learning_rate": 3.8553306829169265e-05, "loss": 0.2501, "step": 10448 }, { "epoch": 64.90342679127725, "grad_norm": 3.40625, "learning_rate": 3.8551236480696635e-05, "loss": 0.2846, "step": 10449 }, { "epoch": 64.90965732087227, "grad_norm": 2.453125, "learning_rate": 3.854916600061167e-05, "loss": 0.1336, "step": 10450 }, { "epoch": 64.91588785046729, "grad_norm": 2.15625, "learning_rate": 3.8547095388934506e-05, "loss": 0.1717, "step": 10451 }, { "epoch": 64.9221183800623, "grad_norm": 3.53125, "learning_rate": 3.8545024645685243e-05, "loss": 0.2879, "step": 10452 }, { "epoch": 64.92834890965732, "grad_norm": 3.5625, "learning_rate": 3.8542953770883986e-05, "loss": 0.2768, "step": 10453 }, { "epoch": 64.93457943925233, "grad_norm": 3.015625, "learning_rate": 3.8540882764550864e-05, "loss": 0.2287, "step": 10454 }, { "epoch": 64.94080996884735, "grad_norm": 2.5, "learning_rate": 3.853881162670597e-05, "loss": 0.1861, "step": 10455 }, { "epoch": 64.94704049844236, "grad_norm": 2.90625, "learning_rate": 3.8536740357369444e-05, "loss": 0.1721, "step": 10456 }, { "epoch": 64.95327102803738, "grad_norm": 2.046875, "learning_rate": 3.853466895656138e-05, "loss": 0.1132, "step": 10457 }, { "epoch": 64.9595015576324, "grad_norm": 3.8125, "learning_rate": 3.853259742430191e-05, "loss": 0.248, "step": 10458 }, { "epoch": 64.96573208722741, "grad_norm": 2.796875, "learning_rate": 3.853052576061115e-05, "loss": 0.2214, "step": 10459 }, { "epoch": 64.97196261682242, "grad_norm": 1.921875, "learning_rate": 3.852845396550922e-05, "loss": 0.2136, "step": 10460 }, { "epoch": 64.97819314641744, "grad_norm": 4.1875, "learning_rate": 3.852638203901624e-05, "loss": 0.3138, "step": 10461 }, { "epoch": 64.98442367601245, "grad_norm": 2.3125, "learning_rate": 3.852430998115233e-05, "loss": 0.1131, "step": 10462 }, { "epoch": 64.99065420560747, "grad_norm": 3.328125, "learning_rate": 3.852223779193762e-05, "loss": 0.1346, "step": 10463 }, { "epoch": 64.99688473520249, "grad_norm": 2.765625, "learning_rate": 3.8520165471392235e-05, "loss": 0.2013, "step": 10464 }, { "epoch": 65.0, "grad_norm": 2.015625, "learning_rate": 3.851809301953631e-05, "loss": 0.1027, "step": 10465 }, { "epoch": 65.00623052959502, "grad_norm": 2.828125, "learning_rate": 3.851602043638994e-05, "loss": 0.1885, "step": 10466 }, { "epoch": 65.01246105919003, "grad_norm": 2.578125, "learning_rate": 3.851394772197329e-05, "loss": 0.1261, "step": 10467 }, { "epoch": 65.01869158878505, "grad_norm": 3.96875, "learning_rate": 3.851187487630649e-05, "loss": 0.331, "step": 10468 }, { "epoch": 65.02492211838006, "grad_norm": 1.46875, "learning_rate": 3.8509801899409634e-05, "loss": 0.1317, "step": 10469 }, { "epoch": 65.03115264797508, "grad_norm": 2.421875, "learning_rate": 3.8507728791302896e-05, "loss": 0.2411, "step": 10470 }, { "epoch": 65.03738317757009, "grad_norm": 1.90625, "learning_rate": 3.8505655552006394e-05, "loss": 0.1409, "step": 10471 }, { "epoch": 65.04361370716511, "grad_norm": 1.734375, "learning_rate": 3.8503582181540275e-05, "loss": 0.1424, "step": 10472 }, { "epoch": 65.04984423676012, "grad_norm": 2.609375, "learning_rate": 3.8501508679924655e-05, "loss": 0.1521, "step": 10473 }, { "epoch": 65.05607476635514, "grad_norm": 3.125, "learning_rate": 3.849943504717969e-05, "loss": 0.256, "step": 10474 }, { "epoch": 65.06230529595015, "grad_norm": 2.984375, "learning_rate": 3.8497361283325516e-05, "loss": 0.2498, "step": 10475 }, { "epoch": 65.06853582554517, "grad_norm": 2.21875, "learning_rate": 3.849528738838226e-05, "loss": 0.2486, "step": 10476 }, { "epoch": 65.07476635514018, "grad_norm": 2.453125, "learning_rate": 3.849321336237008e-05, "loss": 0.1513, "step": 10477 }, { "epoch": 65.0809968847352, "grad_norm": 1.3515625, "learning_rate": 3.8491139205309126e-05, "loss": 0.1341, "step": 10478 }, { "epoch": 65.08722741433021, "grad_norm": 3.25, "learning_rate": 3.848906491721952e-05, "loss": 0.2091, "step": 10479 }, { "epoch": 65.09345794392523, "grad_norm": 2.9375, "learning_rate": 3.848699049812142e-05, "loss": 0.2204, "step": 10480 }, { "epoch": 65.09968847352025, "grad_norm": 2.90625, "learning_rate": 3.8484915948034974e-05, "loss": 0.2022, "step": 10481 }, { "epoch": 65.10591900311526, "grad_norm": 1.5390625, "learning_rate": 3.848284126698032e-05, "loss": 0.1426, "step": 10482 }, { "epoch": 65.11214953271028, "grad_norm": 3.640625, "learning_rate": 3.848076645497763e-05, "loss": 0.2429, "step": 10483 }, { "epoch": 65.11838006230529, "grad_norm": 1.65625, "learning_rate": 3.847869151204704e-05, "loss": 0.0948, "step": 10484 }, { "epoch": 65.1246105919003, "grad_norm": 2.515625, "learning_rate": 3.8476616438208694e-05, "loss": 0.2431, "step": 10485 }, { "epoch": 65.13084112149532, "grad_norm": 2.125, "learning_rate": 3.847454123348276e-05, "loss": 0.2053, "step": 10486 }, { "epoch": 65.13707165109034, "grad_norm": 2.59375, "learning_rate": 3.8472465897889394e-05, "loss": 0.1913, "step": 10487 }, { "epoch": 65.14330218068535, "grad_norm": 2.5, "learning_rate": 3.847039043144873e-05, "loss": 0.2595, "step": 10488 }, { "epoch": 65.14953271028037, "grad_norm": 2.921875, "learning_rate": 3.8468314834180954e-05, "loss": 0.2701, "step": 10489 }, { "epoch": 65.15576323987538, "grad_norm": 1.0234375, "learning_rate": 3.846623910610622e-05, "loss": 0.1042, "step": 10490 }, { "epoch": 65.1619937694704, "grad_norm": 1.609375, "learning_rate": 3.8464163247244664e-05, "loss": 0.1429, "step": 10491 }, { "epoch": 65.16822429906541, "grad_norm": 1.5625, "learning_rate": 3.846208725761647e-05, "loss": 0.1242, "step": 10492 }, { "epoch": 65.17445482866043, "grad_norm": 2.15625, "learning_rate": 3.846001113724179e-05, "loss": 0.159, "step": 10493 }, { "epoch": 65.18068535825545, "grad_norm": 1.9296875, "learning_rate": 3.84579348861408e-05, "loss": 0.147, "step": 10494 }, { "epoch": 65.18691588785046, "grad_norm": 2.140625, "learning_rate": 3.845585850433364e-05, "loss": 0.1455, "step": 10495 }, { "epoch": 65.19314641744548, "grad_norm": 2.234375, "learning_rate": 3.845378199184051e-05, "loss": 0.1568, "step": 10496 }, { "epoch": 65.19937694704049, "grad_norm": 2.3125, "learning_rate": 3.845170534868155e-05, "loss": 0.1563, "step": 10497 }, { "epoch": 65.20560747663552, "grad_norm": 1.9609375, "learning_rate": 3.844962857487694e-05, "loss": 0.1793, "step": 10498 }, { "epoch": 65.21183800623054, "grad_norm": 1.921875, "learning_rate": 3.8447551670446846e-05, "loss": 0.124, "step": 10499 }, { "epoch": 65.21806853582555, "grad_norm": 3.234375, "learning_rate": 3.8445474635411454e-05, "loss": 0.2008, "step": 10500 }, { "epoch": 65.22429906542057, "grad_norm": 3.3125, "learning_rate": 3.844339746979091e-05, "loss": 0.2292, "step": 10501 }, { "epoch": 65.23052959501558, "grad_norm": 3.296875, "learning_rate": 3.844132017360541e-05, "loss": 0.1726, "step": 10502 }, { "epoch": 65.2367601246106, "grad_norm": 1.1875, "learning_rate": 3.843924274687513e-05, "loss": 0.0976, "step": 10503 }, { "epoch": 65.24299065420561, "grad_norm": 2.171875, "learning_rate": 3.843716518962022e-05, "loss": 0.1533, "step": 10504 }, { "epoch": 65.24922118380063, "grad_norm": 1.328125, "learning_rate": 3.843508750186089e-05, "loss": 0.103, "step": 10505 }, { "epoch": 65.25545171339564, "grad_norm": 1.03125, "learning_rate": 3.8433009683617294e-05, "loss": 0.0975, "step": 10506 }, { "epoch": 65.26168224299066, "grad_norm": 3.84375, "learning_rate": 3.8430931734909634e-05, "loss": 0.4214, "step": 10507 }, { "epoch": 65.26791277258567, "grad_norm": 3.125, "learning_rate": 3.8428853655758066e-05, "loss": 0.2329, "step": 10508 }, { "epoch": 65.27414330218069, "grad_norm": 1.703125, "learning_rate": 3.842677544618279e-05, "loss": 0.1227, "step": 10509 }, { "epoch": 65.2803738317757, "grad_norm": 3.671875, "learning_rate": 3.8424697106203996e-05, "loss": 0.2936, "step": 10510 }, { "epoch": 65.28660436137072, "grad_norm": 2.078125, "learning_rate": 3.842261863584186e-05, "loss": 0.1449, "step": 10511 }, { "epoch": 65.29283489096574, "grad_norm": 1.984375, "learning_rate": 3.842054003511656e-05, "loss": 0.1707, "step": 10512 }, { "epoch": 65.29906542056075, "grad_norm": 3.28125, "learning_rate": 3.84184613040483e-05, "loss": 0.1581, "step": 10513 }, { "epoch": 65.30529595015577, "grad_norm": 2.109375, "learning_rate": 3.841638244265726e-05, "loss": 0.133, "step": 10514 }, { "epoch": 65.31152647975078, "grad_norm": 3.515625, "learning_rate": 3.841430345096363e-05, "loss": 0.2148, "step": 10515 }, { "epoch": 65.3177570093458, "grad_norm": 3.125, "learning_rate": 3.8412224328987614e-05, "loss": 0.1792, "step": 10516 }, { "epoch": 65.32398753894081, "grad_norm": 2.59375, "learning_rate": 3.8410145076749374e-05, "loss": 0.1835, "step": 10517 }, { "epoch": 65.33021806853583, "grad_norm": 1.78125, "learning_rate": 3.840806569426913e-05, "loss": 0.1246, "step": 10518 }, { "epoch": 65.33644859813084, "grad_norm": 2.0625, "learning_rate": 3.840598618156708e-05, "loss": 0.1253, "step": 10519 }, { "epoch": 65.34267912772586, "grad_norm": 3.703125, "learning_rate": 3.8403906538663416e-05, "loss": 0.1466, "step": 10520 }, { "epoch": 65.34890965732087, "grad_norm": 1.6875, "learning_rate": 3.840182676557832e-05, "loss": 0.152, "step": 10521 }, { "epoch": 65.35514018691589, "grad_norm": 1.921875, "learning_rate": 3.8399746862332006e-05, "loss": 0.1422, "step": 10522 }, { "epoch": 65.3613707165109, "grad_norm": 2.75, "learning_rate": 3.839766682894468e-05, "loss": 0.1643, "step": 10523 }, { "epoch": 65.36760124610592, "grad_norm": 4.0, "learning_rate": 3.839558666543652e-05, "loss": 0.1721, "step": 10524 }, { "epoch": 65.37383177570094, "grad_norm": 3.328125, "learning_rate": 3.839350637182776e-05, "loss": 0.1125, "step": 10525 }, { "epoch": 65.38006230529595, "grad_norm": 1.9609375, "learning_rate": 3.839142594813858e-05, "loss": 0.1026, "step": 10526 }, { "epoch": 65.38629283489097, "grad_norm": 2.421875, "learning_rate": 3.8389345394389204e-05, "loss": 0.1566, "step": 10527 }, { "epoch": 65.39252336448598, "grad_norm": 3.34375, "learning_rate": 3.838726471059982e-05, "loss": 0.235, "step": 10528 }, { "epoch": 65.398753894081, "grad_norm": 2.546875, "learning_rate": 3.838518389679065e-05, "loss": 0.1936, "step": 10529 }, { "epoch": 65.40498442367601, "grad_norm": 2.234375, "learning_rate": 3.838310295298189e-05, "loss": 0.1881, "step": 10530 }, { "epoch": 65.41121495327103, "grad_norm": 3.390625, "learning_rate": 3.8381021879193766e-05, "loss": 0.17, "step": 10531 }, { "epoch": 65.41744548286604, "grad_norm": 3.546875, "learning_rate": 3.8378940675446483e-05, "loss": 0.2598, "step": 10532 }, { "epoch": 65.42367601246106, "grad_norm": 2.828125, "learning_rate": 3.8376859341760253e-05, "loss": 0.1181, "step": 10533 }, { "epoch": 65.42990654205607, "grad_norm": 1.875, "learning_rate": 3.837477787815529e-05, "loss": 0.1462, "step": 10534 }, { "epoch": 65.43613707165109, "grad_norm": 3.21875, "learning_rate": 3.8372696284651814e-05, "loss": 0.3093, "step": 10535 }, { "epoch": 65.4423676012461, "grad_norm": 3.109375, "learning_rate": 3.837061456127004e-05, "loss": 0.3068, "step": 10536 }, { "epoch": 65.44859813084112, "grad_norm": 1.953125, "learning_rate": 3.8368532708030176e-05, "loss": 0.1086, "step": 10537 }, { "epoch": 65.45482866043614, "grad_norm": 2.28125, "learning_rate": 3.836645072495245e-05, "loss": 0.1333, "step": 10538 }, { "epoch": 65.46105919003115, "grad_norm": 2.59375, "learning_rate": 3.836436861205709e-05, "loss": 0.1615, "step": 10539 }, { "epoch": 65.46728971962617, "grad_norm": 3.03125, "learning_rate": 3.836228636936431e-05, "loss": 0.2555, "step": 10540 }, { "epoch": 65.47352024922118, "grad_norm": 2.734375, "learning_rate": 3.836020399689433e-05, "loss": 0.1255, "step": 10541 }, { "epoch": 65.4797507788162, "grad_norm": 1.8046875, "learning_rate": 3.835812149466738e-05, "loss": 0.1395, "step": 10542 }, { "epoch": 65.48598130841121, "grad_norm": 2.953125, "learning_rate": 3.835603886270368e-05, "loss": 0.1474, "step": 10543 }, { "epoch": 65.49221183800623, "grad_norm": 2.234375, "learning_rate": 3.835395610102347e-05, "loss": 0.1803, "step": 10544 }, { "epoch": 65.49844236760124, "grad_norm": 2.125, "learning_rate": 3.835187320964696e-05, "loss": 0.1475, "step": 10545 }, { "epoch": 65.50467289719626, "grad_norm": 1.8828125, "learning_rate": 3.834979018859439e-05, "loss": 0.1403, "step": 10546 }, { "epoch": 65.51090342679127, "grad_norm": 3.671875, "learning_rate": 3.8347707037885986e-05, "loss": 0.4029, "step": 10547 }, { "epoch": 65.51713395638629, "grad_norm": 1.7734375, "learning_rate": 3.8345623757541996e-05, "loss": 0.1372, "step": 10548 }, { "epoch": 65.5233644859813, "grad_norm": 2.796875, "learning_rate": 3.834354034758263e-05, "loss": 0.3189, "step": 10549 }, { "epoch": 65.52959501557632, "grad_norm": 1.9140625, "learning_rate": 3.834145680802813e-05, "loss": 0.1378, "step": 10550 }, { "epoch": 65.53582554517133, "grad_norm": 1.6640625, "learning_rate": 3.833937313889874e-05, "loss": 0.1516, "step": 10551 }, { "epoch": 65.54205607476635, "grad_norm": 1.71875, "learning_rate": 3.833728934021469e-05, "loss": 0.1278, "step": 10552 }, { "epoch": 65.54828660436137, "grad_norm": 3.109375, "learning_rate": 3.833520541199622e-05, "loss": 0.1394, "step": 10553 }, { "epoch": 65.55451713395638, "grad_norm": 3.234375, "learning_rate": 3.833312135426358e-05, "loss": 0.2561, "step": 10554 }, { "epoch": 65.5607476635514, "grad_norm": 3.0625, "learning_rate": 3.833103716703699e-05, "loss": 0.2842, "step": 10555 }, { "epoch": 65.56697819314641, "grad_norm": 1.578125, "learning_rate": 3.8328952850336695e-05, "loss": 0.1327, "step": 10556 }, { "epoch": 65.57320872274143, "grad_norm": 2.28125, "learning_rate": 3.832686840418296e-05, "loss": 0.1488, "step": 10557 }, { "epoch": 65.57943925233644, "grad_norm": 3.875, "learning_rate": 3.8324783828596e-05, "loss": 0.2689, "step": 10558 }, { "epoch": 65.58566978193146, "grad_norm": 1.8828125, "learning_rate": 3.832269912359608e-05, "loss": 0.1597, "step": 10559 }, { "epoch": 65.59190031152647, "grad_norm": 2.421875, "learning_rate": 3.8320614289203454e-05, "loss": 0.2084, "step": 10560 }, { "epoch": 65.59813084112149, "grad_norm": 2.5, "learning_rate": 3.8318529325438354e-05, "loss": 0.1734, "step": 10561 }, { "epoch": 65.60436137071652, "grad_norm": 3.265625, "learning_rate": 3.831644423232104e-05, "loss": 0.3643, "step": 10562 }, { "epoch": 65.61059190031153, "grad_norm": 3.03125, "learning_rate": 3.8314359009871744e-05, "loss": 0.3014, "step": 10563 }, { "epoch": 65.61682242990655, "grad_norm": 1.953125, "learning_rate": 3.831227365811074e-05, "loss": 0.131, "step": 10564 }, { "epoch": 65.62305295950156, "grad_norm": 1.2578125, "learning_rate": 3.831018817705827e-05, "loss": 0.109, "step": 10565 }, { "epoch": 65.62928348909658, "grad_norm": 2.859375, "learning_rate": 3.830810256673459e-05, "loss": 0.2062, "step": 10566 }, { "epoch": 65.6355140186916, "grad_norm": 1.4765625, "learning_rate": 3.830601682715996e-05, "loss": 0.0947, "step": 10567 }, { "epoch": 65.64174454828661, "grad_norm": 2.40625, "learning_rate": 3.830393095835464e-05, "loss": 0.1205, "step": 10568 }, { "epoch": 65.64797507788163, "grad_norm": 1.3046875, "learning_rate": 3.830184496033888e-05, "loss": 0.131, "step": 10569 }, { "epoch": 65.65420560747664, "grad_norm": 2.578125, "learning_rate": 3.829975883313295e-05, "loss": 0.2031, "step": 10570 }, { "epoch": 65.66043613707166, "grad_norm": 4.4375, "learning_rate": 3.829767257675708e-05, "loss": 0.1666, "step": 10571 }, { "epoch": 65.66666666666667, "grad_norm": 1.765625, "learning_rate": 3.829558619123158e-05, "loss": 0.1524, "step": 10572 }, { "epoch": 65.67289719626169, "grad_norm": 1.3828125, "learning_rate": 3.829349967657668e-05, "loss": 0.1145, "step": 10573 }, { "epoch": 65.6791277258567, "grad_norm": 2.640625, "learning_rate": 3.829141303281265e-05, "loss": 0.2291, "step": 10574 }, { "epoch": 65.68535825545172, "grad_norm": 3.484375, "learning_rate": 3.828932625995977e-05, "loss": 0.188, "step": 10575 }, { "epoch": 65.69158878504673, "grad_norm": 4.15625, "learning_rate": 3.828723935803829e-05, "loss": 0.1764, "step": 10576 }, { "epoch": 65.69781931464175, "grad_norm": 3.171875, "learning_rate": 3.8285152327068484e-05, "loss": 0.2712, "step": 10577 }, { "epoch": 65.70404984423676, "grad_norm": 2.640625, "learning_rate": 3.8283065167070625e-05, "loss": 0.1937, "step": 10578 }, { "epoch": 65.71028037383178, "grad_norm": 3.34375, "learning_rate": 3.828097787806498e-05, "loss": 0.2444, "step": 10579 }, { "epoch": 65.7165109034268, "grad_norm": 2.9375, "learning_rate": 3.8278890460071823e-05, "loss": 0.3118, "step": 10580 }, { "epoch": 65.72274143302181, "grad_norm": 3.140625, "learning_rate": 3.827680291311143e-05, "loss": 0.192, "step": 10581 }, { "epoch": 65.72897196261682, "grad_norm": 2.46875, "learning_rate": 3.827471523720407e-05, "loss": 0.2046, "step": 10582 }, { "epoch": 65.73520249221184, "grad_norm": 2.5, "learning_rate": 3.827262743237003e-05, "loss": 0.1866, "step": 10583 }, { "epoch": 65.74143302180686, "grad_norm": 3.0625, "learning_rate": 3.827053949862958e-05, "loss": 0.1601, "step": 10584 }, { "epoch": 65.74766355140187, "grad_norm": 2.4375, "learning_rate": 3.826845143600299e-05, "loss": 0.1997, "step": 10585 }, { "epoch": 65.75389408099689, "grad_norm": 1.5859375, "learning_rate": 3.826636324451054e-05, "loss": 0.116, "step": 10586 }, { "epoch": 65.7601246105919, "grad_norm": 2.578125, "learning_rate": 3.826427492417253e-05, "loss": 0.2464, "step": 10587 }, { "epoch": 65.76635514018692, "grad_norm": 2.671875, "learning_rate": 3.826218647500923e-05, "loss": 0.1596, "step": 10588 }, { "epoch": 65.77258566978193, "grad_norm": 1.5546875, "learning_rate": 3.826009789704093e-05, "loss": 0.117, "step": 10589 }, { "epoch": 65.77881619937695, "grad_norm": 2.25, "learning_rate": 3.825800919028789e-05, "loss": 0.1758, "step": 10590 }, { "epoch": 65.78504672897196, "grad_norm": 2.3125, "learning_rate": 3.825592035477043e-05, "loss": 0.1658, "step": 10591 }, { "epoch": 65.79127725856698, "grad_norm": 2.375, "learning_rate": 3.825383139050881e-05, "loss": 0.1989, "step": 10592 }, { "epoch": 65.797507788162, "grad_norm": 3.890625, "learning_rate": 3.825174229752334e-05, "loss": 0.1899, "step": 10593 }, { "epoch": 65.80373831775701, "grad_norm": 2.5625, "learning_rate": 3.8249653075834296e-05, "loss": 0.2778, "step": 10594 }, { "epoch": 65.80996884735202, "grad_norm": 1.203125, "learning_rate": 3.824756372546196e-05, "loss": 0.1007, "step": 10595 }, { "epoch": 65.81619937694704, "grad_norm": 2.84375, "learning_rate": 3.8245474246426666e-05, "loss": 0.3005, "step": 10596 }, { "epoch": 65.82242990654206, "grad_norm": 2.578125, "learning_rate": 3.824338463874866e-05, "loss": 0.156, "step": 10597 }, { "epoch": 65.82866043613707, "grad_norm": 2.078125, "learning_rate": 3.8241294902448255e-05, "loss": 0.1653, "step": 10598 }, { "epoch": 65.83489096573209, "grad_norm": 2.265625, "learning_rate": 3.823920503754575e-05, "loss": 0.1836, "step": 10599 }, { "epoch": 65.8411214953271, "grad_norm": 2.34375, "learning_rate": 3.823711504406144e-05, "loss": 0.1593, "step": 10600 }, { "epoch": 65.84735202492212, "grad_norm": 3.5, "learning_rate": 3.823502492201562e-05, "loss": 0.2295, "step": 10601 }, { "epoch": 65.85358255451713, "grad_norm": 2.296875, "learning_rate": 3.82329346714286e-05, "loss": 0.163, "step": 10602 }, { "epoch": 65.85981308411215, "grad_norm": 2.984375, "learning_rate": 3.8230844292320666e-05, "loss": 0.2139, "step": 10603 }, { "epoch": 65.86604361370716, "grad_norm": 2.046875, "learning_rate": 3.8228753784712126e-05, "loss": 0.1586, "step": 10604 }, { "epoch": 65.87227414330218, "grad_norm": 1.9375, "learning_rate": 3.8226663148623284e-05, "loss": 0.1512, "step": 10605 }, { "epoch": 65.8785046728972, "grad_norm": 2.890625, "learning_rate": 3.8224572384074456e-05, "loss": 0.1732, "step": 10606 }, { "epoch": 65.88473520249221, "grad_norm": 1.9609375, "learning_rate": 3.822248149108593e-05, "loss": 0.2024, "step": 10607 }, { "epoch": 65.89096573208722, "grad_norm": 2.671875, "learning_rate": 3.822039046967802e-05, "loss": 0.1588, "step": 10608 }, { "epoch": 65.89719626168224, "grad_norm": 2.796875, "learning_rate": 3.821829931987103e-05, "loss": 0.2109, "step": 10609 }, { "epoch": 65.90342679127725, "grad_norm": 4.0625, "learning_rate": 3.821620804168529e-05, "loss": 0.1001, "step": 10610 }, { "epoch": 65.90965732087227, "grad_norm": 2.96875, "learning_rate": 3.821411663514108e-05, "loss": 0.1951, "step": 10611 }, { "epoch": 65.91588785046729, "grad_norm": 1.8828125, "learning_rate": 3.8212025100258744e-05, "loss": 0.1706, "step": 10612 }, { "epoch": 65.9221183800623, "grad_norm": 3.109375, "learning_rate": 3.820993343705856e-05, "loss": 0.265, "step": 10613 }, { "epoch": 65.92834890965732, "grad_norm": 2.3125, "learning_rate": 3.820784164556087e-05, "loss": 0.1464, "step": 10614 }, { "epoch": 65.93457943925233, "grad_norm": 2.484375, "learning_rate": 3.820574972578599e-05, "loss": 0.1887, "step": 10615 }, { "epoch": 65.94080996884735, "grad_norm": 1.9609375, "learning_rate": 3.8203657677754216e-05, "loss": 0.1755, "step": 10616 }, { "epoch": 65.94704049844236, "grad_norm": 3.03125, "learning_rate": 3.820156550148588e-05, "loss": 0.2007, "step": 10617 }, { "epoch": 65.95327102803738, "grad_norm": 2.71875, "learning_rate": 3.819947319700131e-05, "loss": 0.1732, "step": 10618 }, { "epoch": 65.9595015576324, "grad_norm": 2.125, "learning_rate": 3.819738076432081e-05, "loss": 0.1191, "step": 10619 }, { "epoch": 65.96573208722741, "grad_norm": 1.4765625, "learning_rate": 3.819528820346472e-05, "loss": 0.1405, "step": 10620 }, { "epoch": 65.97196261682242, "grad_norm": 3.71875, "learning_rate": 3.8193195514453347e-05, "loss": 0.1798, "step": 10621 }, { "epoch": 65.97819314641744, "grad_norm": 2.859375, "learning_rate": 3.8191102697307026e-05, "loss": 0.1405, "step": 10622 }, { "epoch": 65.98442367601245, "grad_norm": 2.375, "learning_rate": 3.818900975204607e-05, "loss": 0.1364, "step": 10623 }, { "epoch": 65.99065420560747, "grad_norm": 2.453125, "learning_rate": 3.818691667869082e-05, "loss": 0.27, "step": 10624 }, { "epoch": 65.99688473520249, "grad_norm": 2.421875, "learning_rate": 3.81848234772616e-05, "loss": 0.1819, "step": 10625 }, { "epoch": 66.0, "grad_norm": 0.953125, "learning_rate": 3.818273014777873e-05, "loss": 0.0464, "step": 10626 }, { "epoch": 66.00623052959502, "grad_norm": 3.28125, "learning_rate": 3.818063669026256e-05, "loss": 0.1546, "step": 10627 }, { "epoch": 66.01246105919003, "grad_norm": 2.65625, "learning_rate": 3.817854310473341e-05, "loss": 0.2335, "step": 10628 }, { "epoch": 66.01869158878505, "grad_norm": 3.28125, "learning_rate": 3.8176449391211616e-05, "loss": 0.1043, "step": 10629 }, { "epoch": 66.02492211838006, "grad_norm": 3.796875, "learning_rate": 3.817435554971751e-05, "loss": 0.2146, "step": 10630 }, { "epoch": 66.03115264797508, "grad_norm": 2.1875, "learning_rate": 3.817226158027143e-05, "loss": 0.1797, "step": 10631 }, { "epoch": 66.03738317757009, "grad_norm": 2.671875, "learning_rate": 3.8170167482893706e-05, "loss": 0.2166, "step": 10632 }, { "epoch": 66.04361370716511, "grad_norm": 3.125, "learning_rate": 3.816807325760469e-05, "loss": 0.1794, "step": 10633 }, { "epoch": 66.04984423676012, "grad_norm": 2.921875, "learning_rate": 3.8165978904424715e-05, "loss": 0.1758, "step": 10634 }, { "epoch": 66.05607476635514, "grad_norm": 4.0625, "learning_rate": 3.8163884423374116e-05, "loss": 0.3023, "step": 10635 }, { "epoch": 66.06230529595015, "grad_norm": 2.296875, "learning_rate": 3.8161789814473246e-05, "loss": 0.1143, "step": 10636 }, { "epoch": 66.06853582554517, "grad_norm": 2.109375, "learning_rate": 3.815969507774244e-05, "loss": 0.1339, "step": 10637 }, { "epoch": 66.07476635514018, "grad_norm": 2.34375, "learning_rate": 3.8157600213202044e-05, "loss": 0.173, "step": 10638 }, { "epoch": 66.0809968847352, "grad_norm": 3.765625, "learning_rate": 3.8155505220872404e-05, "loss": 0.2624, "step": 10639 }, { "epoch": 66.08722741433021, "grad_norm": 1.9453125, "learning_rate": 3.815341010077387e-05, "loss": 0.1466, "step": 10640 }, { "epoch": 66.09345794392523, "grad_norm": 2.359375, "learning_rate": 3.815131485292678e-05, "loss": 0.18, "step": 10641 }, { "epoch": 66.09968847352025, "grad_norm": 1.8359375, "learning_rate": 3.814921947735149e-05, "loss": 0.1667, "step": 10642 }, { "epoch": 66.10591900311526, "grad_norm": 2.703125, "learning_rate": 3.814712397406837e-05, "loss": 0.2302, "step": 10643 }, { "epoch": 66.11214953271028, "grad_norm": 3.921875, "learning_rate": 3.814502834309774e-05, "loss": 0.3013, "step": 10644 }, { "epoch": 66.11838006230529, "grad_norm": 2.796875, "learning_rate": 3.814293258445997e-05, "loss": 0.144, "step": 10645 }, { "epoch": 66.1246105919003, "grad_norm": 2.96875, "learning_rate": 3.814083669817541e-05, "loss": 0.2013, "step": 10646 }, { "epoch": 66.13084112149532, "grad_norm": 2.6875, "learning_rate": 3.813874068426442e-05, "loss": 0.226, "step": 10647 }, { "epoch": 66.13707165109034, "grad_norm": 1.125, "learning_rate": 3.813664454274735e-05, "loss": 0.0913, "step": 10648 }, { "epoch": 66.14330218068535, "grad_norm": 2.328125, "learning_rate": 3.813454827364457e-05, "loss": 0.1477, "step": 10649 }, { "epoch": 66.14953271028037, "grad_norm": 2.53125, "learning_rate": 3.813245187697643e-05, "loss": 0.1709, "step": 10650 }, { "epoch": 66.15576323987538, "grad_norm": 2.1875, "learning_rate": 3.813035535276328e-05, "loss": 0.1212, "step": 10651 }, { "epoch": 66.1619937694704, "grad_norm": 2.234375, "learning_rate": 3.81282587010255e-05, "loss": 0.1382, "step": 10652 }, { "epoch": 66.16822429906541, "grad_norm": 2.40625, "learning_rate": 3.812616192178346e-05, "loss": 0.1838, "step": 10653 }, { "epoch": 66.17445482866043, "grad_norm": 2.0625, "learning_rate": 3.8124065015057495e-05, "loss": 0.1478, "step": 10654 }, { "epoch": 66.18068535825545, "grad_norm": 2.328125, "learning_rate": 3.812196798086799e-05, "loss": 0.1883, "step": 10655 }, { "epoch": 66.18691588785046, "grad_norm": 2.296875, "learning_rate": 3.811987081923532e-05, "loss": 0.2061, "step": 10656 }, { "epoch": 66.19314641744548, "grad_norm": 2.75, "learning_rate": 3.811777353017983e-05, "loss": 0.1291, "step": 10657 }, { "epoch": 66.19937694704049, "grad_norm": 1.96875, "learning_rate": 3.811567611372191e-05, "loss": 0.1437, "step": 10658 }, { "epoch": 66.20560747663552, "grad_norm": 3.703125, "learning_rate": 3.811357856988191e-05, "loss": 0.2861, "step": 10659 }, { "epoch": 66.21183800623054, "grad_norm": 1.546875, "learning_rate": 3.811148089868023e-05, "loss": 0.1467, "step": 10660 }, { "epoch": 66.21806853582555, "grad_norm": 2.65625, "learning_rate": 3.8109383100137216e-05, "loss": 0.2136, "step": 10661 }, { "epoch": 66.22429906542057, "grad_norm": 2.015625, "learning_rate": 3.8107285174273254e-05, "loss": 0.1433, "step": 10662 }, { "epoch": 66.23052959501558, "grad_norm": 2.921875, "learning_rate": 3.8105187121108723e-05, "loss": 0.1797, "step": 10663 }, { "epoch": 66.2367601246106, "grad_norm": 2.21875, "learning_rate": 3.810308894066399e-05, "loss": 0.236, "step": 10664 }, { "epoch": 66.24299065420561, "grad_norm": 1.7890625, "learning_rate": 3.810099063295943e-05, "loss": 0.1326, "step": 10665 }, { "epoch": 66.24922118380063, "grad_norm": 1.953125, "learning_rate": 3.8098892198015446e-05, "loss": 0.1833, "step": 10666 }, { "epoch": 66.25545171339564, "grad_norm": 2.140625, "learning_rate": 3.809679363585239e-05, "loss": 0.1936, "step": 10667 }, { "epoch": 66.26168224299066, "grad_norm": 3.125, "learning_rate": 3.809469494649066e-05, "loss": 0.1611, "step": 10668 }, { "epoch": 66.26791277258567, "grad_norm": 1.953125, "learning_rate": 3.809259612995064e-05, "loss": 0.1208, "step": 10669 }, { "epoch": 66.27414330218069, "grad_norm": 2.546875, "learning_rate": 3.809049718625271e-05, "loss": 0.1724, "step": 10670 }, { "epoch": 66.2803738317757, "grad_norm": 2.25, "learning_rate": 3.808839811541725e-05, "loss": 0.1513, "step": 10671 }, { "epoch": 66.28660436137072, "grad_norm": 3.140625, "learning_rate": 3.8086298917464654e-05, "loss": 0.2888, "step": 10672 }, { "epoch": 66.29283489096574, "grad_norm": 1.796875, "learning_rate": 3.8084199592415305e-05, "loss": 0.1361, "step": 10673 }, { "epoch": 66.29906542056075, "grad_norm": 3.15625, "learning_rate": 3.808210014028959e-05, "loss": 0.2556, "step": 10674 }, { "epoch": 66.30529595015577, "grad_norm": 2.1875, "learning_rate": 3.808000056110791e-05, "loss": 0.1314, "step": 10675 }, { "epoch": 66.31152647975078, "grad_norm": 2.390625, "learning_rate": 3.8077900854890645e-05, "loss": 0.1829, "step": 10676 }, { "epoch": 66.3177570093458, "grad_norm": 1.125, "learning_rate": 3.80758010216582e-05, "loss": 0.1331, "step": 10677 }, { "epoch": 66.32398753894081, "grad_norm": 2.984375, "learning_rate": 3.807370106143095e-05, "loss": 0.2269, "step": 10678 }, { "epoch": 66.33021806853583, "grad_norm": 1.65625, "learning_rate": 3.807160097422931e-05, "loss": 0.1047, "step": 10679 }, { "epoch": 66.33644859813084, "grad_norm": 3.0, "learning_rate": 3.806950076007367e-05, "loss": 0.1613, "step": 10680 }, { "epoch": 66.34267912772586, "grad_norm": 3.015625, "learning_rate": 3.806740041898442e-05, "loss": 0.1569, "step": 10681 }, { "epoch": 66.34890965732087, "grad_norm": 2.984375, "learning_rate": 3.806529995098197e-05, "loss": 0.2754, "step": 10682 }, { "epoch": 66.35514018691589, "grad_norm": 1.0078125, "learning_rate": 3.806319935608671e-05, "loss": 0.1159, "step": 10683 }, { "epoch": 66.3613707165109, "grad_norm": 3.0625, "learning_rate": 3.8061098634319056e-05, "loss": 0.1743, "step": 10684 }, { "epoch": 66.36760124610592, "grad_norm": 3.296875, "learning_rate": 3.805899778569939e-05, "loss": 0.1655, "step": 10685 }, { "epoch": 66.37383177570094, "grad_norm": 2.28125, "learning_rate": 3.805689681024813e-05, "loss": 0.1592, "step": 10686 }, { "epoch": 66.38006230529595, "grad_norm": 1.9765625, "learning_rate": 3.805479570798568e-05, "loss": 0.1517, "step": 10687 }, { "epoch": 66.38629283489097, "grad_norm": 2.421875, "learning_rate": 3.805269447893245e-05, "loss": 0.2103, "step": 10688 }, { "epoch": 66.39252336448598, "grad_norm": 2.125, "learning_rate": 3.805059312310884e-05, "loss": 0.1731, "step": 10689 }, { "epoch": 66.398753894081, "grad_norm": 0.96875, "learning_rate": 3.8048491640535255e-05, "loss": 0.0999, "step": 10690 }, { "epoch": 66.40498442367601, "grad_norm": 2.171875, "learning_rate": 3.804639003123211e-05, "loss": 0.1598, "step": 10691 }, { "epoch": 66.41121495327103, "grad_norm": 2.609375, "learning_rate": 3.804428829521982e-05, "loss": 0.2082, "step": 10692 }, { "epoch": 66.41744548286604, "grad_norm": 1.90625, "learning_rate": 3.8042186432518795e-05, "loss": 0.1482, "step": 10693 }, { "epoch": 66.42367601246106, "grad_norm": 2.015625, "learning_rate": 3.8040084443149455e-05, "loss": 0.2193, "step": 10694 }, { "epoch": 66.42990654205607, "grad_norm": 2.859375, "learning_rate": 3.8037982327132196e-05, "loss": 0.2283, "step": 10695 }, { "epoch": 66.43613707165109, "grad_norm": 2.1875, "learning_rate": 3.803588008448745e-05, "loss": 0.1548, "step": 10696 }, { "epoch": 66.4423676012461, "grad_norm": 3.0, "learning_rate": 3.8033777715235636e-05, "loss": 0.3176, "step": 10697 }, { "epoch": 66.44859813084112, "grad_norm": 1.9921875, "learning_rate": 3.803167521939717e-05, "loss": 0.1618, "step": 10698 }, { "epoch": 66.45482866043614, "grad_norm": 2.609375, "learning_rate": 3.8029572596992456e-05, "loss": 0.1291, "step": 10699 }, { "epoch": 66.46105919003115, "grad_norm": 2.859375, "learning_rate": 3.8027469848041936e-05, "loss": 0.2227, "step": 10700 }, { "epoch": 66.46728971962617, "grad_norm": 2.5625, "learning_rate": 3.8025366972566026e-05, "loss": 0.186, "step": 10701 }, { "epoch": 66.47352024922118, "grad_norm": 1.4609375, "learning_rate": 3.802326397058514e-05, "loss": 0.1196, "step": 10702 }, { "epoch": 66.4797507788162, "grad_norm": 1.40625, "learning_rate": 3.802116084211973e-05, "loss": 0.1225, "step": 10703 }, { "epoch": 66.48598130841121, "grad_norm": 2.625, "learning_rate": 3.801905758719019e-05, "loss": 0.2522, "step": 10704 }, { "epoch": 66.49221183800623, "grad_norm": 3.078125, "learning_rate": 3.801695420581696e-05, "loss": 0.2821, "step": 10705 }, { "epoch": 66.49844236760124, "grad_norm": 2.328125, "learning_rate": 3.801485069802046e-05, "loss": 0.1922, "step": 10706 }, { "epoch": 66.50467289719626, "grad_norm": 2.796875, "learning_rate": 3.801274706382115e-05, "loss": 0.1074, "step": 10707 }, { "epoch": 66.51090342679127, "grad_norm": 2.71875, "learning_rate": 3.8010643303239424e-05, "loss": 0.1219, "step": 10708 }, { "epoch": 66.51713395638629, "grad_norm": 2.171875, "learning_rate": 3.800853941629573e-05, "loss": 0.2338, "step": 10709 }, { "epoch": 66.5233644859813, "grad_norm": 1.828125, "learning_rate": 3.800643540301051e-05, "loss": 0.1088, "step": 10710 }, { "epoch": 66.52959501557632, "grad_norm": 2.6875, "learning_rate": 3.800433126340418e-05, "loss": 0.1314, "step": 10711 }, { "epoch": 66.53582554517133, "grad_norm": 1.9375, "learning_rate": 3.8002226997497194e-05, "loss": 0.1395, "step": 10712 }, { "epoch": 66.54205607476635, "grad_norm": 2.796875, "learning_rate": 3.8000122605309974e-05, "loss": 0.3555, "step": 10713 }, { "epoch": 66.54828660436137, "grad_norm": 2.21875, "learning_rate": 3.7998018086862965e-05, "loss": 0.1878, "step": 10714 }, { "epoch": 66.55451713395638, "grad_norm": 3.5625, "learning_rate": 3.79959134421766e-05, "loss": 0.2801, "step": 10715 }, { "epoch": 66.5607476635514, "grad_norm": 1.84375, "learning_rate": 3.799380867127134e-05, "loss": 0.1468, "step": 10716 }, { "epoch": 66.56697819314641, "grad_norm": 1.765625, "learning_rate": 3.7991703774167594e-05, "loss": 0.1338, "step": 10717 }, { "epoch": 66.57320872274143, "grad_norm": 2.21875, "learning_rate": 3.798959875088584e-05, "loss": 0.1443, "step": 10718 }, { "epoch": 66.57943925233644, "grad_norm": 1.765625, "learning_rate": 3.798749360144651e-05, "loss": 0.0993, "step": 10719 }, { "epoch": 66.58566978193146, "grad_norm": 2.296875, "learning_rate": 3.798538832587003e-05, "loss": 0.2121, "step": 10720 }, { "epoch": 66.59190031152647, "grad_norm": 2.125, "learning_rate": 3.7983282924176876e-05, "loss": 0.2108, "step": 10721 }, { "epoch": 66.59813084112149, "grad_norm": 2.53125, "learning_rate": 3.798117739638747e-05, "loss": 0.1476, "step": 10722 }, { "epoch": 66.60436137071652, "grad_norm": 3.359375, "learning_rate": 3.7979071742522286e-05, "loss": 0.173, "step": 10723 }, { "epoch": 66.61059190031153, "grad_norm": 3.109375, "learning_rate": 3.797696596260175e-05, "loss": 0.1433, "step": 10724 }, { "epoch": 66.61682242990655, "grad_norm": 1.8515625, "learning_rate": 3.797486005664634e-05, "loss": 0.1248, "step": 10725 }, { "epoch": 66.62305295950156, "grad_norm": 2.140625, "learning_rate": 3.797275402467649e-05, "loss": 0.2153, "step": 10726 }, { "epoch": 66.62928348909658, "grad_norm": 2.921875, "learning_rate": 3.797064786671266e-05, "loss": 0.2345, "step": 10727 }, { "epoch": 66.6355140186916, "grad_norm": 1.828125, "learning_rate": 3.796854158277531e-05, "loss": 0.1166, "step": 10728 }, { "epoch": 66.64174454828661, "grad_norm": 3.328125, "learning_rate": 3.7966435172884885e-05, "loss": 0.2421, "step": 10729 }, { "epoch": 66.64797507788163, "grad_norm": 3.203125, "learning_rate": 3.796432863706185e-05, "loss": 0.2128, "step": 10730 }, { "epoch": 66.65420560747664, "grad_norm": 2.421875, "learning_rate": 3.796222197532666e-05, "loss": 0.1877, "step": 10731 }, { "epoch": 66.66043613707166, "grad_norm": 3.65625, "learning_rate": 3.796011518769979e-05, "loss": 0.1709, "step": 10732 }, { "epoch": 66.66666666666667, "grad_norm": 2.46875, "learning_rate": 3.795800827420168e-05, "loss": 0.2115, "step": 10733 }, { "epoch": 66.67289719626169, "grad_norm": 2.21875, "learning_rate": 3.795590123485282e-05, "loss": 0.1701, "step": 10734 }, { "epoch": 66.6791277258567, "grad_norm": 2.078125, "learning_rate": 3.7953794069673636e-05, "loss": 0.1623, "step": 10735 }, { "epoch": 66.68535825545172, "grad_norm": 2.078125, "learning_rate": 3.795168677868463e-05, "loss": 0.2744, "step": 10736 }, { "epoch": 66.69158878504673, "grad_norm": 2.5625, "learning_rate": 3.794957936190624e-05, "loss": 0.3623, "step": 10737 }, { "epoch": 66.69781931464175, "grad_norm": 2.34375, "learning_rate": 3.794747181935896e-05, "loss": 0.1725, "step": 10738 }, { "epoch": 66.70404984423676, "grad_norm": 1.75, "learning_rate": 3.7945364151063236e-05, "loss": 0.1338, "step": 10739 }, { "epoch": 66.71028037383178, "grad_norm": 2.671875, "learning_rate": 3.794325635703956e-05, "loss": 0.2229, "step": 10740 }, { "epoch": 66.7165109034268, "grad_norm": 1.15625, "learning_rate": 3.794114843730837e-05, "loss": 0.1038, "step": 10741 }, { "epoch": 66.72274143302181, "grad_norm": 3.484375, "learning_rate": 3.793904039189018e-05, "loss": 0.378, "step": 10742 }, { "epoch": 66.72897196261682, "grad_norm": 1.5390625, "learning_rate": 3.793693222080543e-05, "loss": 0.115, "step": 10743 }, { "epoch": 66.73520249221184, "grad_norm": 2.140625, "learning_rate": 3.7934823924074617e-05, "loss": 0.1772, "step": 10744 }, { "epoch": 66.74143302180686, "grad_norm": 2.140625, "learning_rate": 3.7932715501718206e-05, "loss": 0.1839, "step": 10745 }, { "epoch": 66.74766355140187, "grad_norm": 2.296875, "learning_rate": 3.7930606953756675e-05, "loss": 0.1964, "step": 10746 }, { "epoch": 66.75389408099689, "grad_norm": 1.3125, "learning_rate": 3.792849828021051e-05, "loss": 0.095, "step": 10747 }, { "epoch": 66.7601246105919, "grad_norm": 2.046875, "learning_rate": 3.792638948110018e-05, "loss": 0.1557, "step": 10748 }, { "epoch": 66.76635514018692, "grad_norm": 2.90625, "learning_rate": 3.792428055644617e-05, "loss": 0.2559, "step": 10749 }, { "epoch": 66.77258566978193, "grad_norm": 1.1640625, "learning_rate": 3.792217150626896e-05, "loss": 0.1265, "step": 10750 }, { "epoch": 66.77881619937695, "grad_norm": 2.234375, "learning_rate": 3.7920062330589044e-05, "loss": 0.2215, "step": 10751 }, { "epoch": 66.78504672897196, "grad_norm": 2.640625, "learning_rate": 3.7917953029426904e-05, "loss": 0.2137, "step": 10752 }, { "epoch": 66.79127725856698, "grad_norm": 1.453125, "learning_rate": 3.7915843602803016e-05, "loss": 0.1026, "step": 10753 }, { "epoch": 66.797507788162, "grad_norm": 2.453125, "learning_rate": 3.791373405073787e-05, "loss": 0.2662, "step": 10754 }, { "epoch": 66.80373831775701, "grad_norm": 1.3515625, "learning_rate": 3.7911624373251963e-05, "loss": 0.1205, "step": 10755 }, { "epoch": 66.80996884735202, "grad_norm": 3.15625, "learning_rate": 3.790951457036578e-05, "loss": 0.3263, "step": 10756 }, { "epoch": 66.81619937694704, "grad_norm": 2.546875, "learning_rate": 3.79074046420998e-05, "loss": 0.2546, "step": 10757 }, { "epoch": 66.82242990654206, "grad_norm": 2.140625, "learning_rate": 3.7905294588474536e-05, "loss": 0.1349, "step": 10758 }, { "epoch": 66.82866043613707, "grad_norm": 1.546875, "learning_rate": 3.7903184409510475e-05, "loss": 0.1384, "step": 10759 }, { "epoch": 66.83489096573209, "grad_norm": 2.515625, "learning_rate": 3.790107410522809e-05, "loss": 0.1746, "step": 10760 }, { "epoch": 66.8411214953271, "grad_norm": 2.234375, "learning_rate": 3.7898963675647915e-05, "loss": 0.2019, "step": 10761 }, { "epoch": 66.84735202492212, "grad_norm": 1.84375, "learning_rate": 3.789685312079042e-05, "loss": 0.1559, "step": 10762 }, { "epoch": 66.85358255451713, "grad_norm": 1.6015625, "learning_rate": 3.7894742440676104e-05, "loss": 0.1152, "step": 10763 }, { "epoch": 66.85981308411215, "grad_norm": 1.375, "learning_rate": 3.789263163532548e-05, "loss": 0.1409, "step": 10764 }, { "epoch": 66.86604361370716, "grad_norm": 3.5625, "learning_rate": 3.789052070475903e-05, "loss": 0.4256, "step": 10765 }, { "epoch": 66.87227414330218, "grad_norm": 2.5625, "learning_rate": 3.788840964899727e-05, "loss": 0.2279, "step": 10766 }, { "epoch": 66.8785046728972, "grad_norm": 3.640625, "learning_rate": 3.788629846806071e-05, "loss": 0.3305, "step": 10767 }, { "epoch": 66.88473520249221, "grad_norm": 1.96875, "learning_rate": 3.788418716196983e-05, "loss": 0.1824, "step": 10768 }, { "epoch": 66.89096573208722, "grad_norm": 1.4296875, "learning_rate": 3.7882075730745155e-05, "loss": 0.0994, "step": 10769 }, { "epoch": 66.89719626168224, "grad_norm": 1.2890625, "learning_rate": 3.787996417440718e-05, "loss": 0.1073, "step": 10770 }, { "epoch": 66.90342679127725, "grad_norm": 2.890625, "learning_rate": 3.787785249297643e-05, "loss": 0.1414, "step": 10771 }, { "epoch": 66.90965732087227, "grad_norm": 4.5, "learning_rate": 3.787574068647338e-05, "loss": 0.2903, "step": 10772 }, { "epoch": 66.91588785046729, "grad_norm": 1.9140625, "learning_rate": 3.787362875491858e-05, "loss": 0.1319, "step": 10773 }, { "epoch": 66.9221183800623, "grad_norm": 2.21875, "learning_rate": 3.787151669833252e-05, "loss": 0.142, "step": 10774 }, { "epoch": 66.92834890965732, "grad_norm": 1.5703125, "learning_rate": 3.786940451673572e-05, "loss": 0.1187, "step": 10775 }, { "epoch": 66.93457943925233, "grad_norm": 2.625, "learning_rate": 3.7867292210148685e-05, "loss": 0.2781, "step": 10776 }, { "epoch": 66.94080996884735, "grad_norm": 3.90625, "learning_rate": 3.786517977859195e-05, "loss": 0.1788, "step": 10777 }, { "epoch": 66.94704049844236, "grad_norm": 3.65625, "learning_rate": 3.7863067222086004e-05, "loss": 0.193, "step": 10778 }, { "epoch": 66.95327102803738, "grad_norm": 3.1875, "learning_rate": 3.7860954540651375e-05, "loss": 0.1585, "step": 10779 }, { "epoch": 66.9595015576324, "grad_norm": 2.078125, "learning_rate": 3.785884173430859e-05, "loss": 0.1736, "step": 10780 }, { "epoch": 66.96573208722741, "grad_norm": 2.640625, "learning_rate": 3.785672880307817e-05, "loss": 0.1692, "step": 10781 }, { "epoch": 66.97196261682242, "grad_norm": 2.25, "learning_rate": 3.7854615746980626e-05, "loss": 0.1543, "step": 10782 }, { "epoch": 66.97819314641744, "grad_norm": 3.90625, "learning_rate": 3.785250256603648e-05, "loss": 0.2626, "step": 10783 }, { "epoch": 66.98442367601245, "grad_norm": 4.1875, "learning_rate": 3.7850389260266265e-05, "loss": 0.1708, "step": 10784 }, { "epoch": 66.99065420560747, "grad_norm": 3.453125, "learning_rate": 3.7848275829690496e-05, "loss": 0.2984, "step": 10785 }, { "epoch": 66.99688473520249, "grad_norm": 1.3828125, "learning_rate": 3.78461622743297e-05, "loss": 0.1054, "step": 10786 }, { "epoch": 67.0, "grad_norm": 1.34375, "learning_rate": 3.784404859420442e-05, "loss": 0.0652, "step": 10787 }, { "epoch": 67.00623052959502, "grad_norm": 2.359375, "learning_rate": 3.7841934789335164e-05, "loss": 0.1284, "step": 10788 }, { "epoch": 67.01246105919003, "grad_norm": 2.75, "learning_rate": 3.783982085974247e-05, "loss": 0.1183, "step": 10789 }, { "epoch": 67.01869158878505, "grad_norm": 3.890625, "learning_rate": 3.783770680544687e-05, "loss": 0.2674, "step": 10790 }, { "epoch": 67.02492211838006, "grad_norm": 0.8203125, "learning_rate": 3.7835592626468894e-05, "loss": 0.08, "step": 10791 }, { "epoch": 67.03115264797508, "grad_norm": 2.453125, "learning_rate": 3.783347832282909e-05, "loss": 0.1759, "step": 10792 }, { "epoch": 67.03738317757009, "grad_norm": 2.46875, "learning_rate": 3.783136389454796e-05, "loss": 0.2207, "step": 10793 }, { "epoch": 67.04361370716511, "grad_norm": 1.765625, "learning_rate": 3.7829249341646065e-05, "loss": 0.1309, "step": 10794 }, { "epoch": 67.04984423676012, "grad_norm": 1.7734375, "learning_rate": 3.782713466414395e-05, "loss": 0.1462, "step": 10795 }, { "epoch": 67.05607476635514, "grad_norm": 2.9375, "learning_rate": 3.782501986206212e-05, "loss": 0.26, "step": 10796 }, { "epoch": 67.06230529595015, "grad_norm": 1.828125, "learning_rate": 3.7822904935421145e-05, "loss": 0.1348, "step": 10797 }, { "epoch": 67.06853582554517, "grad_norm": 1.609375, "learning_rate": 3.7820789884241545e-05, "loss": 0.1023, "step": 10798 }, { "epoch": 67.07476635514018, "grad_norm": 3.875, "learning_rate": 3.781867470854388e-05, "loss": 0.1569, "step": 10799 }, { "epoch": 67.0809968847352, "grad_norm": 1.8671875, "learning_rate": 3.7816559408348676e-05, "loss": 0.1517, "step": 10800 }, { "epoch": 67.08722741433021, "grad_norm": 2.203125, "learning_rate": 3.78144439836765e-05, "loss": 0.1734, "step": 10801 }, { "epoch": 67.09345794392523, "grad_norm": 3.921875, "learning_rate": 3.781232843454786e-05, "loss": 0.2572, "step": 10802 }, { "epoch": 67.09968847352025, "grad_norm": 1.8125, "learning_rate": 3.781021276098334e-05, "loss": 0.176, "step": 10803 }, { "epoch": 67.10591900311526, "grad_norm": 1.8828125, "learning_rate": 3.780809696300347e-05, "loss": 0.1639, "step": 10804 }, { "epoch": 67.11214953271028, "grad_norm": 2.1875, "learning_rate": 3.78059810406288e-05, "loss": 0.1458, "step": 10805 }, { "epoch": 67.11838006230529, "grad_norm": 2.671875, "learning_rate": 3.780386499387989e-05, "loss": 0.138, "step": 10806 }, { "epoch": 67.1246105919003, "grad_norm": 2.640625, "learning_rate": 3.780174882277727e-05, "loss": 0.1915, "step": 10807 }, { "epoch": 67.13084112149532, "grad_norm": 3.203125, "learning_rate": 3.7799632527341514e-05, "loss": 0.2528, "step": 10808 }, { "epoch": 67.13707165109034, "grad_norm": 2.96875, "learning_rate": 3.779751610759317e-05, "loss": 0.2638, "step": 10809 }, { "epoch": 67.14330218068535, "grad_norm": 3.03125, "learning_rate": 3.779539956355279e-05, "loss": 0.1751, "step": 10810 }, { "epoch": 67.14953271028037, "grad_norm": 3.875, "learning_rate": 3.7793282895240926e-05, "loss": 0.1923, "step": 10811 }, { "epoch": 67.15576323987538, "grad_norm": 4.0, "learning_rate": 3.779116610267815e-05, "loss": 0.2272, "step": 10812 }, { "epoch": 67.1619937694704, "grad_norm": 2.171875, "learning_rate": 3.7789049185885007e-05, "loss": 0.1198, "step": 10813 }, { "epoch": 67.16822429906541, "grad_norm": 1.1640625, "learning_rate": 3.778693214488206e-05, "loss": 0.1066, "step": 10814 }, { "epoch": 67.17445482866043, "grad_norm": 1.4765625, "learning_rate": 3.778481497968988e-05, "loss": 0.1192, "step": 10815 }, { "epoch": 67.18068535825545, "grad_norm": 3.65625, "learning_rate": 3.778269769032902e-05, "loss": 0.2075, "step": 10816 }, { "epoch": 67.18691588785046, "grad_norm": 3.75, "learning_rate": 3.778058027682004e-05, "loss": 0.2304, "step": 10817 }, { "epoch": 67.19314641744548, "grad_norm": 3.03125, "learning_rate": 3.777846273918351e-05, "loss": 0.1781, "step": 10818 }, { "epoch": 67.19937694704049, "grad_norm": 3.0, "learning_rate": 3.7776345077439997e-05, "loss": 0.1097, "step": 10819 }, { "epoch": 67.20560747663552, "grad_norm": 3.34375, "learning_rate": 3.777422729161007e-05, "loss": 0.213, "step": 10820 }, { "epoch": 67.21183800623054, "grad_norm": 2.734375, "learning_rate": 3.777210938171428e-05, "loss": 0.1229, "step": 10821 }, { "epoch": 67.21806853582555, "grad_norm": 3.828125, "learning_rate": 3.7769991347773223e-05, "loss": 0.1627, "step": 10822 }, { "epoch": 67.22429906542057, "grad_norm": 3.640625, "learning_rate": 3.776787318980745e-05, "loss": 0.1582, "step": 10823 }, { "epoch": 67.23052959501558, "grad_norm": 1.765625, "learning_rate": 3.776575490783755e-05, "loss": 0.1337, "step": 10824 }, { "epoch": 67.2367601246106, "grad_norm": 3.5, "learning_rate": 3.776363650188408e-05, "loss": 0.3561, "step": 10825 }, { "epoch": 67.24299065420561, "grad_norm": 4.34375, "learning_rate": 3.776151797196762e-05, "loss": 0.1706, "step": 10826 }, { "epoch": 67.24922118380063, "grad_norm": 1.7421875, "learning_rate": 3.775939931810875e-05, "loss": 0.1501, "step": 10827 }, { "epoch": 67.25545171339564, "grad_norm": 2.71875, "learning_rate": 3.775728054032805e-05, "loss": 0.3404, "step": 10828 }, { "epoch": 67.26168224299066, "grad_norm": 3.359375, "learning_rate": 3.775516163864607e-05, "loss": 0.2452, "step": 10829 }, { "epoch": 67.26791277258567, "grad_norm": 1.9921875, "learning_rate": 3.7753042613083434e-05, "loss": 0.153, "step": 10830 }, { "epoch": 67.27414330218069, "grad_norm": 1.8671875, "learning_rate": 3.775092346366069e-05, "loss": 0.1833, "step": 10831 }, { "epoch": 67.2803738317757, "grad_norm": 2.546875, "learning_rate": 3.774880419039842e-05, "loss": 0.2426, "step": 10832 }, { "epoch": 67.28660436137072, "grad_norm": 2.4375, "learning_rate": 3.7746684793317227e-05, "loss": 0.1094, "step": 10833 }, { "epoch": 67.29283489096574, "grad_norm": 2.953125, "learning_rate": 3.774456527243768e-05, "loss": 0.1329, "step": 10834 }, { "epoch": 67.29906542056075, "grad_norm": 2.53125, "learning_rate": 3.774244562778036e-05, "loss": 0.1899, "step": 10835 }, { "epoch": 67.30529595015577, "grad_norm": 1.8984375, "learning_rate": 3.774032585936588e-05, "loss": 0.1362, "step": 10836 }, { "epoch": 67.31152647975078, "grad_norm": 2.1875, "learning_rate": 3.7738205967214784e-05, "loss": 0.2337, "step": 10837 }, { "epoch": 67.3177570093458, "grad_norm": 3.34375, "learning_rate": 3.77360859513477e-05, "loss": 0.2373, "step": 10838 }, { "epoch": 67.32398753894081, "grad_norm": 2.0, "learning_rate": 3.77339658117852e-05, "loss": 0.1808, "step": 10839 }, { "epoch": 67.33021806853583, "grad_norm": 2.015625, "learning_rate": 3.773184554854788e-05, "loss": 0.1763, "step": 10840 }, { "epoch": 67.33644859813084, "grad_norm": 1.484375, "learning_rate": 3.7729725161656334e-05, "loss": 0.1284, "step": 10841 }, { "epoch": 67.34267912772586, "grad_norm": 1.7109375, "learning_rate": 3.7727604651131146e-05, "loss": 0.1501, "step": 10842 }, { "epoch": 67.34890965732087, "grad_norm": 2.390625, "learning_rate": 3.772548401699292e-05, "loss": 0.1814, "step": 10843 }, { "epoch": 67.35514018691589, "grad_norm": 2.9375, "learning_rate": 3.7723363259262254e-05, "loss": 0.2721, "step": 10844 }, { "epoch": 67.3613707165109, "grad_norm": 2.625, "learning_rate": 3.772124237795973e-05, "loss": 0.2574, "step": 10845 }, { "epoch": 67.36760124610592, "grad_norm": 2.078125, "learning_rate": 3.7719121373105965e-05, "loss": 0.1652, "step": 10846 }, { "epoch": 67.37383177570094, "grad_norm": 1.7578125, "learning_rate": 3.771700024472156e-05, "loss": 0.1305, "step": 10847 }, { "epoch": 67.38006230529595, "grad_norm": 1.828125, "learning_rate": 3.7714878992827084e-05, "loss": 0.1594, "step": 10848 }, { "epoch": 67.38629283489097, "grad_norm": 2.703125, "learning_rate": 3.7712757617443174e-05, "loss": 0.2199, "step": 10849 }, { "epoch": 67.39252336448598, "grad_norm": 2.4375, "learning_rate": 3.771063611859042e-05, "loss": 0.1442, "step": 10850 }, { "epoch": 67.398753894081, "grad_norm": 2.609375, "learning_rate": 3.770851449628943e-05, "loss": 0.1542, "step": 10851 }, { "epoch": 67.40498442367601, "grad_norm": 2.203125, "learning_rate": 3.77063927505608e-05, "loss": 0.1907, "step": 10852 }, { "epoch": 67.41121495327103, "grad_norm": 3.21875, "learning_rate": 3.770427088142515e-05, "loss": 0.1292, "step": 10853 }, { "epoch": 67.41744548286604, "grad_norm": 2.578125, "learning_rate": 3.7702148888903084e-05, "loss": 0.2066, "step": 10854 }, { "epoch": 67.42367601246106, "grad_norm": 1.890625, "learning_rate": 3.7700026773015196e-05, "loss": 0.1507, "step": 10855 }, { "epoch": 67.42990654205607, "grad_norm": 2.078125, "learning_rate": 3.769790453378212e-05, "loss": 0.1768, "step": 10856 }, { "epoch": 67.43613707165109, "grad_norm": 1.6875, "learning_rate": 3.769578217122446e-05, "loss": 0.1147, "step": 10857 }, { "epoch": 67.4423676012461, "grad_norm": 2.1875, "learning_rate": 3.7693659685362814e-05, "loss": 0.1403, "step": 10858 }, { "epoch": 67.44859813084112, "grad_norm": 2.296875, "learning_rate": 3.769153707621781e-05, "loss": 0.1299, "step": 10859 }, { "epoch": 67.45482866043614, "grad_norm": 1.7734375, "learning_rate": 3.768941434381007e-05, "loss": 0.164, "step": 10860 }, { "epoch": 67.46105919003115, "grad_norm": 4.1875, "learning_rate": 3.7687291488160195e-05, "loss": 0.4097, "step": 10861 }, { "epoch": 67.46728971962617, "grad_norm": 1.5390625, "learning_rate": 3.768516850928881e-05, "loss": 0.116, "step": 10862 }, { "epoch": 67.47352024922118, "grad_norm": 1.7734375, "learning_rate": 3.768304540721653e-05, "loss": 0.1392, "step": 10863 }, { "epoch": 67.4797507788162, "grad_norm": 2.25, "learning_rate": 3.768092218196398e-05, "loss": 0.1436, "step": 10864 }, { "epoch": 67.48598130841121, "grad_norm": 2.59375, "learning_rate": 3.7678798833551785e-05, "loss": 0.1523, "step": 10865 }, { "epoch": 67.49221183800623, "grad_norm": 2.453125, "learning_rate": 3.767667536200056e-05, "loss": 0.1881, "step": 10866 }, { "epoch": 67.49844236760124, "grad_norm": 2.359375, "learning_rate": 3.767455176733092e-05, "loss": 0.141, "step": 10867 }, { "epoch": 67.50467289719626, "grad_norm": 1.75, "learning_rate": 3.76724280495635e-05, "loss": 0.1443, "step": 10868 }, { "epoch": 67.51090342679127, "grad_norm": 2.28125, "learning_rate": 3.767030420871894e-05, "loss": 0.135, "step": 10869 }, { "epoch": 67.51713395638629, "grad_norm": 2.390625, "learning_rate": 3.766818024481784e-05, "loss": 0.1632, "step": 10870 }, { "epoch": 67.5233644859813, "grad_norm": 3.546875, "learning_rate": 3.7666056157880844e-05, "loss": 0.1767, "step": 10871 }, { "epoch": 67.52959501557632, "grad_norm": 2.4375, "learning_rate": 3.766393194792858e-05, "loss": 0.1279, "step": 10872 }, { "epoch": 67.53582554517133, "grad_norm": 3.734375, "learning_rate": 3.766180761498168e-05, "loss": 0.2163, "step": 10873 }, { "epoch": 67.54205607476635, "grad_norm": 3.234375, "learning_rate": 3.765968315906077e-05, "loss": 0.1327, "step": 10874 }, { "epoch": 67.54828660436137, "grad_norm": 1.8515625, "learning_rate": 3.765755858018648e-05, "loss": 0.175, "step": 10875 }, { "epoch": 67.55451713395638, "grad_norm": 3.546875, "learning_rate": 3.765543387837946e-05, "loss": 0.1563, "step": 10876 }, { "epoch": 67.5607476635514, "grad_norm": 2.859375, "learning_rate": 3.7653309053660336e-05, "loss": 0.1664, "step": 10877 }, { "epoch": 67.56697819314641, "grad_norm": 3.140625, "learning_rate": 3.7651184106049744e-05, "loss": 0.1973, "step": 10878 }, { "epoch": 67.57320872274143, "grad_norm": 3.21875, "learning_rate": 3.764905903556832e-05, "loss": 0.3276, "step": 10879 }, { "epoch": 67.57943925233644, "grad_norm": 2.046875, "learning_rate": 3.764693384223671e-05, "loss": 0.1277, "step": 10880 }, { "epoch": 67.58566978193146, "grad_norm": 1.8515625, "learning_rate": 3.7644808526075546e-05, "loss": 0.0934, "step": 10881 }, { "epoch": 67.59190031152647, "grad_norm": 2.890625, "learning_rate": 3.764268308710548e-05, "loss": 0.2084, "step": 10882 }, { "epoch": 67.59813084112149, "grad_norm": 1.828125, "learning_rate": 3.764055752534714e-05, "loss": 0.1153, "step": 10883 }, { "epoch": 67.60436137071652, "grad_norm": 3.015625, "learning_rate": 3.763843184082119e-05, "loss": 0.2507, "step": 10884 }, { "epoch": 67.61059190031153, "grad_norm": 1.578125, "learning_rate": 3.763630603354825e-05, "loss": 0.1131, "step": 10885 }, { "epoch": 67.61682242990655, "grad_norm": 1.4765625, "learning_rate": 3.7634180103549e-05, "loss": 0.1096, "step": 10886 }, { "epoch": 67.62305295950156, "grad_norm": 1.9921875, "learning_rate": 3.763205405084404e-05, "loss": 0.1148, "step": 10887 }, { "epoch": 67.62928348909658, "grad_norm": 2.453125, "learning_rate": 3.762992787545407e-05, "loss": 0.1993, "step": 10888 }, { "epoch": 67.6355140186916, "grad_norm": 2.765625, "learning_rate": 3.76278015773997e-05, "loss": 0.2369, "step": 10889 }, { "epoch": 67.64174454828661, "grad_norm": 2.6875, "learning_rate": 3.76256751567016e-05, "loss": 0.2332, "step": 10890 }, { "epoch": 67.64797507788163, "grad_norm": 2.125, "learning_rate": 3.7623548613380435e-05, "loss": 0.1277, "step": 10891 }, { "epoch": 67.65420560747664, "grad_norm": 1.8984375, "learning_rate": 3.7621421947456824e-05, "loss": 0.15, "step": 10892 }, { "epoch": 67.66043613707166, "grad_norm": 3.046875, "learning_rate": 3.761929515895145e-05, "loss": 0.3298, "step": 10893 }, { "epoch": 67.66666666666667, "grad_norm": 2.84375, "learning_rate": 3.7617168247884946e-05, "loss": 0.2498, "step": 10894 }, { "epoch": 67.67289719626169, "grad_norm": 1.8046875, "learning_rate": 3.7615041214278e-05, "loss": 0.1147, "step": 10895 }, { "epoch": 67.6791277258567, "grad_norm": 1.2578125, "learning_rate": 3.7612914058151236e-05, "loss": 0.1192, "step": 10896 }, { "epoch": 67.68535825545172, "grad_norm": 2.75, "learning_rate": 3.761078677952534e-05, "loss": 0.1267, "step": 10897 }, { "epoch": 67.69158878504673, "grad_norm": 2.015625, "learning_rate": 3.760865937842096e-05, "loss": 0.1531, "step": 10898 }, { "epoch": 67.69781931464175, "grad_norm": 2.828125, "learning_rate": 3.760653185485876e-05, "loss": 0.195, "step": 10899 }, { "epoch": 67.70404984423676, "grad_norm": 2.125, "learning_rate": 3.76044042088594e-05, "loss": 0.1144, "step": 10900 }, { "epoch": 67.71028037383178, "grad_norm": 3.203125, "learning_rate": 3.760227644044354e-05, "loss": 0.2127, "step": 10901 }, { "epoch": 67.7165109034268, "grad_norm": 3.796875, "learning_rate": 3.7600148549631864e-05, "loss": 0.2521, "step": 10902 }, { "epoch": 67.72274143302181, "grad_norm": 2.6875, "learning_rate": 3.7598020536445017e-05, "loss": 0.2462, "step": 10903 }, { "epoch": 67.72897196261682, "grad_norm": 2.265625, "learning_rate": 3.7595892400903684e-05, "loss": 0.1298, "step": 10904 }, { "epoch": 67.73520249221184, "grad_norm": 2.671875, "learning_rate": 3.759376414302852e-05, "loss": 0.1581, "step": 10905 }, { "epoch": 67.74143302180686, "grad_norm": 1.6484375, "learning_rate": 3.759163576284021e-05, "loss": 0.1457, "step": 10906 }, { "epoch": 67.74766355140187, "grad_norm": 2.390625, "learning_rate": 3.7589507260359404e-05, "loss": 0.1722, "step": 10907 }, { "epoch": 67.75389408099689, "grad_norm": 1.96875, "learning_rate": 3.758737863560679e-05, "loss": 0.132, "step": 10908 }, { "epoch": 67.7601246105919, "grad_norm": 2.296875, "learning_rate": 3.758524988860305e-05, "loss": 0.1564, "step": 10909 }, { "epoch": 67.76635514018692, "grad_norm": 2.78125, "learning_rate": 3.758312101936884e-05, "loss": 0.1946, "step": 10910 }, { "epoch": 67.77258566978193, "grad_norm": 1.875, "learning_rate": 3.758099202792484e-05, "loss": 0.1183, "step": 10911 }, { "epoch": 67.77881619937695, "grad_norm": 3.46875, "learning_rate": 3.757886291429174e-05, "loss": 0.2372, "step": 10912 }, { "epoch": 67.78504672897196, "grad_norm": 3.84375, "learning_rate": 3.7576733678490195e-05, "loss": 0.2172, "step": 10913 }, { "epoch": 67.79127725856698, "grad_norm": 2.296875, "learning_rate": 3.7574604320540905e-05, "loss": 0.1737, "step": 10914 }, { "epoch": 67.797507788162, "grad_norm": 2.21875, "learning_rate": 3.757247484046455e-05, "loss": 0.2317, "step": 10915 }, { "epoch": 67.80373831775701, "grad_norm": 4.25, "learning_rate": 3.75703452382818e-05, "loss": 0.2884, "step": 10916 }, { "epoch": 67.80996884735202, "grad_norm": 3.28125, "learning_rate": 3.756821551401335e-05, "loss": 0.1923, "step": 10917 }, { "epoch": 67.81619937694704, "grad_norm": 1.515625, "learning_rate": 3.7566085667679865e-05, "loss": 0.1111, "step": 10918 }, { "epoch": 67.82242990654206, "grad_norm": 3.671875, "learning_rate": 3.756395569930206e-05, "loss": 0.3361, "step": 10919 }, { "epoch": 67.82866043613707, "grad_norm": 2.1875, "learning_rate": 3.7561825608900597e-05, "loss": 0.1591, "step": 10920 }, { "epoch": 67.83489096573209, "grad_norm": 3.25, "learning_rate": 3.7559695396496175e-05, "loss": 0.1102, "step": 10921 }, { "epoch": 67.8411214953271, "grad_norm": 3.546875, "learning_rate": 3.7557565062109475e-05, "loss": 0.2792, "step": 10922 }, { "epoch": 67.84735202492212, "grad_norm": 4.59375, "learning_rate": 3.75554346057612e-05, "loss": 0.4083, "step": 10923 }, { "epoch": 67.85358255451713, "grad_norm": 1.90625, "learning_rate": 3.755330402747204e-05, "loss": 0.1042, "step": 10924 }, { "epoch": 67.85981308411215, "grad_norm": 1.4765625, "learning_rate": 3.7551173327262665e-05, "loss": 0.134, "step": 10925 }, { "epoch": 67.86604361370716, "grad_norm": 2.90625, "learning_rate": 3.75490425051538e-05, "loss": 0.1423, "step": 10926 }, { "epoch": 67.87227414330218, "grad_norm": 3.828125, "learning_rate": 3.754691156116612e-05, "loss": 0.2165, "step": 10927 }, { "epoch": 67.8785046728972, "grad_norm": 3.65625, "learning_rate": 3.754478049532033e-05, "loss": 0.1713, "step": 10928 }, { "epoch": 67.88473520249221, "grad_norm": 2.4375, "learning_rate": 3.754264930763711e-05, "loss": 0.1331, "step": 10929 }, { "epoch": 67.89096573208722, "grad_norm": 2.390625, "learning_rate": 3.754051799813719e-05, "loss": 0.1752, "step": 10930 }, { "epoch": 67.89719626168224, "grad_norm": 1.0625, "learning_rate": 3.7538386566841246e-05, "loss": 0.0964, "step": 10931 }, { "epoch": 67.90342679127725, "grad_norm": 4.5625, "learning_rate": 3.753625501376999e-05, "loss": 0.1745, "step": 10932 }, { "epoch": 67.90965732087227, "grad_norm": 3.78125, "learning_rate": 3.7534123338944117e-05, "loss": 0.372, "step": 10933 }, { "epoch": 67.91588785046729, "grad_norm": 4.65625, "learning_rate": 3.753199154238433e-05, "loss": 0.1875, "step": 10934 }, { "epoch": 67.9221183800623, "grad_norm": 4.375, "learning_rate": 3.752985962411133e-05, "loss": 0.2021, "step": 10935 }, { "epoch": 67.92834890965732, "grad_norm": 1.3515625, "learning_rate": 3.752772758414585e-05, "loss": 0.1103, "step": 10936 }, { "epoch": 67.93457943925233, "grad_norm": 1.0234375, "learning_rate": 3.752559542250855e-05, "loss": 0.0995, "step": 10937 }, { "epoch": 67.94080996884735, "grad_norm": 1.9921875, "learning_rate": 3.752346313922018e-05, "loss": 0.1273, "step": 10938 }, { "epoch": 67.94704049844236, "grad_norm": 2.359375, "learning_rate": 3.7521330734301425e-05, "loss": 0.1142, "step": 10939 }, { "epoch": 67.95327102803738, "grad_norm": 2.140625, "learning_rate": 3.751919820777301e-05, "loss": 0.1172, "step": 10940 }, { "epoch": 67.9595015576324, "grad_norm": 3.03125, "learning_rate": 3.751706555965563e-05, "loss": 0.1147, "step": 10941 }, { "epoch": 67.96573208722741, "grad_norm": 3.46875, "learning_rate": 3.751493278997001e-05, "loss": 0.2435, "step": 10942 }, { "epoch": 67.97196261682242, "grad_norm": 2.109375, "learning_rate": 3.751279989873687e-05, "loss": 0.1467, "step": 10943 }, { "epoch": 67.97819314641744, "grad_norm": 2.21875, "learning_rate": 3.75106668859769e-05, "loss": 0.1837, "step": 10944 }, { "epoch": 67.98442367601245, "grad_norm": 2.828125, "learning_rate": 3.7508533751710846e-05, "loss": 0.1661, "step": 10945 }, { "epoch": 67.99065420560747, "grad_norm": 2.765625, "learning_rate": 3.7506400495959404e-05, "loss": 0.1926, "step": 10946 }, { "epoch": 67.99688473520249, "grad_norm": 2.84375, "learning_rate": 3.750426711874331e-05, "loss": 0.1794, "step": 10947 }, { "epoch": 68.0, "grad_norm": 1.3671875, "learning_rate": 3.750213362008327e-05, "loss": 0.0491, "step": 10948 }, { "epoch": 68.00623052959502, "grad_norm": 3.15625, "learning_rate": 3.7500000000000003e-05, "loss": 0.2999, "step": 10949 }, { "epoch": 68.01246105919003, "grad_norm": 3.0625, "learning_rate": 3.7497866258514245e-05, "loss": 0.249, "step": 10950 }, { "epoch": 68.01869158878505, "grad_norm": 2.875, "learning_rate": 3.74957323956467e-05, "loss": 0.216, "step": 10951 }, { "epoch": 68.02492211838006, "grad_norm": 1.9609375, "learning_rate": 3.7493598411418117e-05, "loss": 0.1631, "step": 10952 }, { "epoch": 68.03115264797508, "grad_norm": 0.78125, "learning_rate": 3.74914643058492e-05, "loss": 0.0905, "step": 10953 }, { "epoch": 68.03738317757009, "grad_norm": 2.328125, "learning_rate": 3.748933007896069e-05, "loss": 0.192, "step": 10954 }, { "epoch": 68.04361370716511, "grad_norm": 2.984375, "learning_rate": 3.7487195730773314e-05, "loss": 0.235, "step": 10955 }, { "epoch": 68.04984423676012, "grad_norm": 1.796875, "learning_rate": 3.7485061261307785e-05, "loss": 0.1394, "step": 10956 }, { "epoch": 68.05607476635514, "grad_norm": 2.71875, "learning_rate": 3.748292667058485e-05, "loss": 0.2032, "step": 10957 }, { "epoch": 68.06230529595015, "grad_norm": 2.6875, "learning_rate": 3.748079195862524e-05, "loss": 0.1624, "step": 10958 }, { "epoch": 68.06853582554517, "grad_norm": 1.734375, "learning_rate": 3.747865712544968e-05, "loss": 0.1202, "step": 10959 }, { "epoch": 68.07476635514018, "grad_norm": 2.390625, "learning_rate": 3.747652217107891e-05, "loss": 0.1855, "step": 10960 }, { "epoch": 68.0809968847352, "grad_norm": 2.34375, "learning_rate": 3.747438709553367e-05, "loss": 0.2707, "step": 10961 }, { "epoch": 68.08722741433021, "grad_norm": 1.5859375, "learning_rate": 3.747225189883467e-05, "loss": 0.1486, "step": 10962 }, { "epoch": 68.09345794392523, "grad_norm": 2.359375, "learning_rate": 3.7470116581002676e-05, "loss": 0.1655, "step": 10963 }, { "epoch": 68.09968847352025, "grad_norm": 2.0, "learning_rate": 3.746798114205842e-05, "loss": 0.1374, "step": 10964 }, { "epoch": 68.10591900311526, "grad_norm": 2.65625, "learning_rate": 3.746584558202264e-05, "loss": 0.2525, "step": 10965 }, { "epoch": 68.11214953271028, "grad_norm": 2.1875, "learning_rate": 3.7463709900916064e-05, "loss": 0.157, "step": 10966 }, { "epoch": 68.11838006230529, "grad_norm": 2.25, "learning_rate": 3.7461574098759464e-05, "loss": 0.1933, "step": 10967 }, { "epoch": 68.1246105919003, "grad_norm": 1.6875, "learning_rate": 3.745943817557355e-05, "loss": 0.1349, "step": 10968 }, { "epoch": 68.13084112149532, "grad_norm": 1.421875, "learning_rate": 3.745730213137909e-05, "loss": 0.1719, "step": 10969 }, { "epoch": 68.13707165109034, "grad_norm": 1.65625, "learning_rate": 3.745516596619681e-05, "loss": 0.1454, "step": 10970 }, { "epoch": 68.14330218068535, "grad_norm": 1.7578125, "learning_rate": 3.745302968004749e-05, "loss": 0.1381, "step": 10971 }, { "epoch": 68.14953271028037, "grad_norm": 1.578125, "learning_rate": 3.745089327295184e-05, "loss": 0.1233, "step": 10972 }, { "epoch": 68.15576323987538, "grad_norm": 2.421875, "learning_rate": 3.744875674493063e-05, "loss": 0.1836, "step": 10973 }, { "epoch": 68.1619937694704, "grad_norm": 2.09375, "learning_rate": 3.74466200960046e-05, "loss": 0.1271, "step": 10974 }, { "epoch": 68.16822429906541, "grad_norm": 1.40625, "learning_rate": 3.744448332619452e-05, "loss": 0.1329, "step": 10975 }, { "epoch": 68.17445482866043, "grad_norm": 3.578125, "learning_rate": 3.7442346435521124e-05, "loss": 0.2348, "step": 10976 }, { "epoch": 68.18068535825545, "grad_norm": 2.59375, "learning_rate": 3.744020942400517e-05, "loss": 0.142, "step": 10977 }, { "epoch": 68.18691588785046, "grad_norm": 3.015625, "learning_rate": 3.743807229166742e-05, "loss": 0.1177, "step": 10978 }, { "epoch": 68.19314641744548, "grad_norm": 1.828125, "learning_rate": 3.7435935038528616e-05, "loss": 0.1273, "step": 10979 }, { "epoch": 68.19937694704049, "grad_norm": 1.296875, "learning_rate": 3.7433797664609534e-05, "loss": 0.094, "step": 10980 }, { "epoch": 68.20560747663552, "grad_norm": 1.6328125, "learning_rate": 3.743166016993092e-05, "loss": 0.0832, "step": 10981 }, { "epoch": 68.21183800623054, "grad_norm": 2.578125, "learning_rate": 3.742952255451354e-05, "loss": 0.2088, "step": 10982 }, { "epoch": 68.21806853582555, "grad_norm": 3.015625, "learning_rate": 3.7427384818378144e-05, "loss": 0.2268, "step": 10983 }, { "epoch": 68.22429906542057, "grad_norm": 1.8984375, "learning_rate": 3.742524696154552e-05, "loss": 0.1312, "step": 10984 }, { "epoch": 68.23052959501558, "grad_norm": 2.375, "learning_rate": 3.74231089840364e-05, "loss": 0.1799, "step": 10985 }, { "epoch": 68.2367601246106, "grad_norm": 2.140625, "learning_rate": 3.742097088587157e-05, "loss": 0.1477, "step": 10986 }, { "epoch": 68.24299065420561, "grad_norm": 1.7265625, "learning_rate": 3.741883266707178e-05, "loss": 0.113, "step": 10987 }, { "epoch": 68.24922118380063, "grad_norm": 2.28125, "learning_rate": 3.741669432765781e-05, "loss": 0.1782, "step": 10988 }, { "epoch": 68.25545171339564, "grad_norm": 3.3125, "learning_rate": 3.7414555867650416e-05, "loss": 0.2816, "step": 10989 }, { "epoch": 68.26168224299066, "grad_norm": 1.34375, "learning_rate": 3.7412417287070377e-05, "loss": 0.1122, "step": 10990 }, { "epoch": 68.26791277258567, "grad_norm": 3.109375, "learning_rate": 3.741027858593846e-05, "loss": 0.2966, "step": 10991 }, { "epoch": 68.27414330218069, "grad_norm": 1.5703125, "learning_rate": 3.740813976427544e-05, "loss": 0.1262, "step": 10992 }, { "epoch": 68.2803738317757, "grad_norm": 1.21875, "learning_rate": 3.7406000822102085e-05, "loss": 0.1025, "step": 10993 }, { "epoch": 68.28660436137072, "grad_norm": 3.484375, "learning_rate": 3.740386175943917e-05, "loss": 0.3067, "step": 10994 }, { "epoch": 68.29283489096574, "grad_norm": 3.234375, "learning_rate": 3.740172257630747e-05, "loss": 0.1855, "step": 10995 }, { "epoch": 68.29906542056075, "grad_norm": 1.9140625, "learning_rate": 3.7399583272727764e-05, "loss": 0.1498, "step": 10996 }, { "epoch": 68.30529595015577, "grad_norm": 1.359375, "learning_rate": 3.7397443848720826e-05, "loss": 0.1268, "step": 10997 }, { "epoch": 68.31152647975078, "grad_norm": 1.703125, "learning_rate": 3.739530430430743e-05, "loss": 0.1243, "step": 10998 }, { "epoch": 68.3177570093458, "grad_norm": 2.265625, "learning_rate": 3.739316463950836e-05, "loss": 0.1806, "step": 10999 }, { "epoch": 68.32398753894081, "grad_norm": 2.75, "learning_rate": 3.73910248543444e-05, "loss": 0.2022, "step": 11000 }, { "epoch": 68.33021806853583, "grad_norm": 2.234375, "learning_rate": 3.738888494883633e-05, "loss": 0.1129, "step": 11001 }, { "epoch": 68.33644859813084, "grad_norm": 3.421875, "learning_rate": 3.738674492300494e-05, "loss": 0.1515, "step": 11002 }, { "epoch": 68.34267912772586, "grad_norm": 2.75, "learning_rate": 3.7384604776871005e-05, "loss": 0.1663, "step": 11003 }, { "epoch": 68.34890965732087, "grad_norm": 3.953125, "learning_rate": 3.73824645104553e-05, "loss": 0.3326, "step": 11004 }, { "epoch": 68.35514018691589, "grad_norm": 1.8046875, "learning_rate": 3.738032412377863e-05, "loss": 0.0971, "step": 11005 }, { "epoch": 68.3613707165109, "grad_norm": 1.8828125, "learning_rate": 3.737818361686179e-05, "loss": 0.1287, "step": 11006 }, { "epoch": 68.36760124610592, "grad_norm": 1.1328125, "learning_rate": 3.737604298972554e-05, "loss": 0.1221, "step": 11007 }, { "epoch": 68.37383177570094, "grad_norm": 2.21875, "learning_rate": 3.73739022423907e-05, "loss": 0.1873, "step": 11008 }, { "epoch": 68.38006230529595, "grad_norm": 1.9296875, "learning_rate": 3.737176137487804e-05, "loss": 0.1091, "step": 11009 }, { "epoch": 68.38629283489097, "grad_norm": 2.234375, "learning_rate": 3.736962038720836e-05, "loss": 0.2486, "step": 11010 }, { "epoch": 68.39252336448598, "grad_norm": 2.0, "learning_rate": 3.736747927940246e-05, "loss": 0.1455, "step": 11011 }, { "epoch": 68.398753894081, "grad_norm": 3.21875, "learning_rate": 3.736533805148112e-05, "loss": 0.1256, "step": 11012 }, { "epoch": 68.40498442367601, "grad_norm": 1.359375, "learning_rate": 3.7363196703465156e-05, "loss": 0.0959, "step": 11013 }, { "epoch": 68.41121495327103, "grad_norm": 3.671875, "learning_rate": 3.736105523537534e-05, "loss": 0.1734, "step": 11014 }, { "epoch": 68.41744548286604, "grad_norm": 1.078125, "learning_rate": 3.73589136472325e-05, "loss": 0.103, "step": 11015 }, { "epoch": 68.42367601246106, "grad_norm": 1.84375, "learning_rate": 3.7356771939057414e-05, "loss": 0.1711, "step": 11016 }, { "epoch": 68.42990654205607, "grad_norm": 3.96875, "learning_rate": 3.7354630110870885e-05, "loss": 0.2944, "step": 11017 }, { "epoch": 68.43613707165109, "grad_norm": 3.828125, "learning_rate": 3.735248816269372e-05, "loss": 0.1179, "step": 11018 }, { "epoch": 68.4423676012461, "grad_norm": 3.515625, "learning_rate": 3.7350346094546725e-05, "loss": 0.194, "step": 11019 }, { "epoch": 68.44859813084112, "grad_norm": 2.65625, "learning_rate": 3.734820390645069e-05, "loss": 0.1038, "step": 11020 }, { "epoch": 68.45482866043614, "grad_norm": 2.484375, "learning_rate": 3.734606159842644e-05, "loss": 0.1725, "step": 11021 }, { "epoch": 68.46105919003115, "grad_norm": 2.40625, "learning_rate": 3.7343919170494775e-05, "loss": 0.2184, "step": 11022 }, { "epoch": 68.46728971962617, "grad_norm": 3.0625, "learning_rate": 3.7341776622676494e-05, "loss": 0.1851, "step": 11023 }, { "epoch": 68.47352024922118, "grad_norm": 3.921875, "learning_rate": 3.733963395499241e-05, "loss": 0.1671, "step": 11024 }, { "epoch": 68.4797507788162, "grad_norm": 3.3125, "learning_rate": 3.7337491167463335e-05, "loss": 0.0891, "step": 11025 }, { "epoch": 68.48598130841121, "grad_norm": 4.0625, "learning_rate": 3.733534826011008e-05, "loss": 0.1257, "step": 11026 }, { "epoch": 68.49221183800623, "grad_norm": 2.03125, "learning_rate": 3.733320523295346e-05, "loss": 0.2665, "step": 11027 }, { "epoch": 68.49844236760124, "grad_norm": 1.6953125, "learning_rate": 3.733106208601428e-05, "loss": 0.1554, "step": 11028 }, { "epoch": 68.50467289719626, "grad_norm": 2.90625, "learning_rate": 3.7328918819313364e-05, "loss": 0.1911, "step": 11029 }, { "epoch": 68.51090342679127, "grad_norm": 4.03125, "learning_rate": 3.732677543287152e-05, "loss": 0.1669, "step": 11030 }, { "epoch": 68.51713395638629, "grad_norm": 3.6875, "learning_rate": 3.732463192670957e-05, "loss": 0.1633, "step": 11031 }, { "epoch": 68.5233644859813, "grad_norm": 2.1875, "learning_rate": 3.7322488300848337e-05, "loss": 0.2511, "step": 11032 }, { "epoch": 68.52959501557632, "grad_norm": 2.046875, "learning_rate": 3.732034455530863e-05, "loss": 0.145, "step": 11033 }, { "epoch": 68.53582554517133, "grad_norm": 2.4375, "learning_rate": 3.731820069011127e-05, "loss": 0.1468, "step": 11034 }, { "epoch": 68.54205607476635, "grad_norm": 2.78125, "learning_rate": 3.731605670527709e-05, "loss": 0.1889, "step": 11035 }, { "epoch": 68.54828660436137, "grad_norm": 3.828125, "learning_rate": 3.7313912600826905e-05, "loss": 0.1114, "step": 11036 }, { "epoch": 68.55451713395638, "grad_norm": 3.46875, "learning_rate": 3.7311768376781536e-05, "loss": 0.1576, "step": 11037 }, { "epoch": 68.5607476635514, "grad_norm": 3.53125, "learning_rate": 3.730962403316181e-05, "loss": 0.1509, "step": 11038 }, { "epoch": 68.56697819314641, "grad_norm": 2.375, "learning_rate": 3.730747956998856e-05, "loss": 0.1661, "step": 11039 }, { "epoch": 68.57320872274143, "grad_norm": 3.078125, "learning_rate": 3.73053349872826e-05, "loss": 0.3209, "step": 11040 }, { "epoch": 68.57943925233644, "grad_norm": 1.3359375, "learning_rate": 3.7303190285064776e-05, "loss": 0.1233, "step": 11041 }, { "epoch": 68.58566978193146, "grad_norm": 3.625, "learning_rate": 3.73010454633559e-05, "loss": 0.1815, "step": 11042 }, { "epoch": 68.59190031152647, "grad_norm": 3.59375, "learning_rate": 3.7298900522176814e-05, "loss": 0.2051, "step": 11043 }, { "epoch": 68.59813084112149, "grad_norm": 5.0, "learning_rate": 3.729675546154835e-05, "loss": 0.1387, "step": 11044 }, { "epoch": 68.60436137071652, "grad_norm": 3.84375, "learning_rate": 3.7294610281491337e-05, "loss": 0.2624, "step": 11045 }, { "epoch": 68.61059190031153, "grad_norm": 2.890625, "learning_rate": 3.729246498202661e-05, "loss": 0.2682, "step": 11046 }, { "epoch": 68.61682242990655, "grad_norm": 2.875, "learning_rate": 3.729031956317501e-05, "loss": 0.1554, "step": 11047 }, { "epoch": 68.62305295950156, "grad_norm": 3.046875, "learning_rate": 3.7288174024957374e-05, "loss": 0.1143, "step": 11048 }, { "epoch": 68.62928348909658, "grad_norm": 5.0625, "learning_rate": 3.728602836739453e-05, "loss": 0.1239, "step": 11049 }, { "epoch": 68.6355140186916, "grad_norm": 3.84375, "learning_rate": 3.728388259050732e-05, "loss": 0.1299, "step": 11050 }, { "epoch": 68.64174454828661, "grad_norm": 3.4375, "learning_rate": 3.728173669431659e-05, "loss": 0.2511, "step": 11051 }, { "epoch": 68.64797507788163, "grad_norm": 2.03125, "learning_rate": 3.727959067884317e-05, "loss": 0.1281, "step": 11052 }, { "epoch": 68.65420560747664, "grad_norm": 2.1875, "learning_rate": 3.727744454410792e-05, "loss": 0.1204, "step": 11053 }, { "epoch": 68.66043613707166, "grad_norm": 4.21875, "learning_rate": 3.7275298290131675e-05, "loss": 0.2525, "step": 11054 }, { "epoch": 68.66666666666667, "grad_norm": 3.390625, "learning_rate": 3.727315191693527e-05, "loss": 0.2232, "step": 11055 }, { "epoch": 68.67289719626169, "grad_norm": 4.46875, "learning_rate": 3.727100542453957e-05, "loss": 0.1526, "step": 11056 }, { "epoch": 68.6791277258567, "grad_norm": 1.7265625, "learning_rate": 3.7268858812965414e-05, "loss": 0.1008, "step": 11057 }, { "epoch": 68.68535825545172, "grad_norm": 2.734375, "learning_rate": 3.726671208223364e-05, "loss": 0.2141, "step": 11058 }, { "epoch": 68.69158878504673, "grad_norm": 2.53125, "learning_rate": 3.726456523236511e-05, "loss": 0.1133, "step": 11059 }, { "epoch": 68.69781931464175, "grad_norm": 3.171875, "learning_rate": 3.7262418263380673e-05, "loss": 0.1118, "step": 11060 }, { "epoch": 68.70404984423676, "grad_norm": 3.859375, "learning_rate": 3.7260271175301173e-05, "loss": 0.1673, "step": 11061 }, { "epoch": 68.71028037383178, "grad_norm": 2.859375, "learning_rate": 3.7258123968147476e-05, "loss": 0.1263, "step": 11062 }, { "epoch": 68.7165109034268, "grad_norm": 3.109375, "learning_rate": 3.7255976641940424e-05, "loss": 0.1407, "step": 11063 }, { "epoch": 68.72274143302181, "grad_norm": 1.8828125, "learning_rate": 3.7253829196700876e-05, "loss": 0.1522, "step": 11064 }, { "epoch": 68.72897196261682, "grad_norm": 3.625, "learning_rate": 3.725168163244969e-05, "loss": 0.1261, "step": 11065 }, { "epoch": 68.73520249221184, "grad_norm": 4.6875, "learning_rate": 3.724953394920772e-05, "loss": 0.3011, "step": 11066 }, { "epoch": 68.74143302180686, "grad_norm": 2.859375, "learning_rate": 3.724738614699583e-05, "loss": 0.1212, "step": 11067 }, { "epoch": 68.74766355140187, "grad_norm": 3.9375, "learning_rate": 3.724523822583488e-05, "loss": 0.1251, "step": 11068 }, { "epoch": 68.75389408099689, "grad_norm": 1.9375, "learning_rate": 3.724309018574573e-05, "loss": 0.1175, "step": 11069 }, { "epoch": 68.7601246105919, "grad_norm": 1.5, "learning_rate": 3.724094202674924e-05, "loss": 0.0986, "step": 11070 }, { "epoch": 68.76635514018692, "grad_norm": 2.546875, "learning_rate": 3.7238793748866275e-05, "loss": 0.1076, "step": 11071 }, { "epoch": 68.77258566978193, "grad_norm": 3.875, "learning_rate": 3.723664535211769e-05, "loss": 0.3265, "step": 11072 }, { "epoch": 68.77881619937695, "grad_norm": 2.40625, "learning_rate": 3.723449683652437e-05, "loss": 0.1079, "step": 11073 }, { "epoch": 68.78504672897196, "grad_norm": 1.984375, "learning_rate": 3.723234820210717e-05, "loss": 0.1142, "step": 11074 }, { "epoch": 68.79127725856698, "grad_norm": 2.6875, "learning_rate": 3.7230199448886945e-05, "loss": 0.205, "step": 11075 }, { "epoch": 68.797507788162, "grad_norm": 2.828125, "learning_rate": 3.7228050576884596e-05, "loss": 0.2823, "step": 11076 }, { "epoch": 68.80373831775701, "grad_norm": 4.0, "learning_rate": 3.722590158612096e-05, "loss": 0.3078, "step": 11077 }, { "epoch": 68.80996884735202, "grad_norm": 2.875, "learning_rate": 3.722375247661694e-05, "loss": 0.1743, "step": 11078 }, { "epoch": 68.81619937694704, "grad_norm": 3.8125, "learning_rate": 3.7221603248393385e-05, "loss": 0.1242, "step": 11079 }, { "epoch": 68.82242990654206, "grad_norm": 1.9765625, "learning_rate": 3.7219453901471175e-05, "loss": 0.1881, "step": 11080 }, { "epoch": 68.82866043613707, "grad_norm": 3.234375, "learning_rate": 3.721730443587118e-05, "loss": 0.2437, "step": 11081 }, { "epoch": 68.83489096573209, "grad_norm": 1.1171875, "learning_rate": 3.721515485161429e-05, "loss": 0.0932, "step": 11082 }, { "epoch": 68.8411214953271, "grad_norm": 2.671875, "learning_rate": 3.721300514872137e-05, "loss": 0.2003, "step": 11083 }, { "epoch": 68.84735202492212, "grad_norm": 2.546875, "learning_rate": 3.721085532721331e-05, "loss": 0.133, "step": 11084 }, { "epoch": 68.85358255451713, "grad_norm": 2.203125, "learning_rate": 3.720870538711097e-05, "loss": 0.2012, "step": 11085 }, { "epoch": 68.85981308411215, "grad_norm": 1.0390625, "learning_rate": 3.720655532843526e-05, "loss": 0.1177, "step": 11086 }, { "epoch": 68.86604361370716, "grad_norm": 1.6953125, "learning_rate": 3.7204405151207036e-05, "loss": 0.0996, "step": 11087 }, { "epoch": 68.87227414330218, "grad_norm": 3.0625, "learning_rate": 3.7202254855447185e-05, "loss": 0.1514, "step": 11088 }, { "epoch": 68.8785046728972, "grad_norm": 2.234375, "learning_rate": 3.720010444117661e-05, "loss": 0.2661, "step": 11089 }, { "epoch": 68.88473520249221, "grad_norm": 4.09375, "learning_rate": 3.719795390841616e-05, "loss": 0.3252, "step": 11090 }, { "epoch": 68.89096573208722, "grad_norm": 2.0625, "learning_rate": 3.719580325718676e-05, "loss": 0.1459, "step": 11091 }, { "epoch": 68.89719626168224, "grad_norm": 3.40625, "learning_rate": 3.7193652487509275e-05, "loss": 0.2501, "step": 11092 }, { "epoch": 68.90342679127725, "grad_norm": 3.0625, "learning_rate": 3.71915015994046e-05, "loss": 0.2591, "step": 11093 }, { "epoch": 68.90965732087227, "grad_norm": 1.953125, "learning_rate": 3.7189350592893624e-05, "loss": 0.1149, "step": 11094 }, { "epoch": 68.91588785046729, "grad_norm": 1.890625, "learning_rate": 3.718719946799724e-05, "loss": 0.1648, "step": 11095 }, { "epoch": 68.9221183800623, "grad_norm": 2.15625, "learning_rate": 3.718504822473634e-05, "loss": 0.1784, "step": 11096 }, { "epoch": 68.92834890965732, "grad_norm": 3.015625, "learning_rate": 3.718289686313181e-05, "loss": 0.2562, "step": 11097 }, { "epoch": 68.93457943925233, "grad_norm": 1.59375, "learning_rate": 3.7180745383204554e-05, "loss": 0.136, "step": 11098 }, { "epoch": 68.94080996884735, "grad_norm": 1.71875, "learning_rate": 3.717859378497547e-05, "loss": 0.1488, "step": 11099 }, { "epoch": 68.94704049844236, "grad_norm": 2.515625, "learning_rate": 3.7176442068465436e-05, "loss": 0.1687, "step": 11100 }, { "epoch": 68.95327102803738, "grad_norm": 1.984375, "learning_rate": 3.717429023369537e-05, "loss": 0.0998, "step": 11101 }, { "epoch": 68.9595015576324, "grad_norm": 2.171875, "learning_rate": 3.7172138280686166e-05, "loss": 0.1806, "step": 11102 }, { "epoch": 68.96573208722741, "grad_norm": 3.1875, "learning_rate": 3.716998620945871e-05, "loss": 0.2131, "step": 11103 }, { "epoch": 68.97196261682242, "grad_norm": 2.671875, "learning_rate": 3.716783402003393e-05, "loss": 0.1927, "step": 11104 }, { "epoch": 68.97819314641744, "grad_norm": 1.984375, "learning_rate": 3.716568171243271e-05, "loss": 0.1442, "step": 11105 }, { "epoch": 68.98442367601245, "grad_norm": 2.109375, "learning_rate": 3.716352928667595e-05, "loss": 0.1556, "step": 11106 }, { "epoch": 68.99065420560747, "grad_norm": 1.9375, "learning_rate": 3.716137674278457e-05, "loss": 0.2206, "step": 11107 }, { "epoch": 68.99688473520249, "grad_norm": 1.828125, "learning_rate": 3.7159224080779466e-05, "loss": 0.1167, "step": 11108 }, { "epoch": 69.0, "grad_norm": 0.9453125, "learning_rate": 3.715707130068154e-05, "loss": 0.0606, "step": 11109 }, { "epoch": 69.00623052959502, "grad_norm": 1.625, "learning_rate": 3.715491840251172e-05, "loss": 0.0912, "step": 11110 }, { "epoch": 69.01246105919003, "grad_norm": 2.109375, "learning_rate": 3.71527653862909e-05, "loss": 0.1494, "step": 11111 }, { "epoch": 69.01869158878505, "grad_norm": 1.8828125, "learning_rate": 3.715061225203998e-05, "loss": 0.1415, "step": 11112 }, { "epoch": 69.02492211838006, "grad_norm": 2.375, "learning_rate": 3.71484589997799e-05, "loss": 0.2096, "step": 11113 }, { "epoch": 69.03115264797508, "grad_norm": 2.53125, "learning_rate": 3.714630562953155e-05, "loss": 0.1293, "step": 11114 }, { "epoch": 69.03738317757009, "grad_norm": 1.3984375, "learning_rate": 3.714415214131586e-05, "loss": 0.0942, "step": 11115 }, { "epoch": 69.04361370716511, "grad_norm": 2.65625, "learning_rate": 3.714199853515373e-05, "loss": 0.1963, "step": 11116 }, { "epoch": 69.04984423676012, "grad_norm": 1.0390625, "learning_rate": 3.7139844811066085e-05, "loss": 0.0895, "step": 11117 }, { "epoch": 69.05607476635514, "grad_norm": 1.8203125, "learning_rate": 3.7137690969073835e-05, "loss": 0.1708, "step": 11118 }, { "epoch": 69.06230529595015, "grad_norm": 1.765625, "learning_rate": 3.713553700919791e-05, "loss": 0.1206, "step": 11119 }, { "epoch": 69.06853582554517, "grad_norm": 2.125, "learning_rate": 3.7133382931459225e-05, "loss": 0.1295, "step": 11120 }, { "epoch": 69.07476635514018, "grad_norm": 2.421875, "learning_rate": 3.7131228735878695e-05, "loss": 0.1329, "step": 11121 }, { "epoch": 69.0809968847352, "grad_norm": 1.71875, "learning_rate": 3.712907442247725e-05, "loss": 0.1374, "step": 11122 }, { "epoch": 69.08722741433021, "grad_norm": 2.71875, "learning_rate": 3.7126919991275805e-05, "loss": 0.1851, "step": 11123 }, { "epoch": 69.09345794392523, "grad_norm": 2.921875, "learning_rate": 3.7124765442295294e-05, "loss": 0.1338, "step": 11124 }, { "epoch": 69.09968847352025, "grad_norm": 3.8125, "learning_rate": 3.712261077555663e-05, "loss": 0.161, "step": 11125 }, { "epoch": 69.10591900311526, "grad_norm": 3.296875, "learning_rate": 3.712045599108076e-05, "loss": 0.19, "step": 11126 }, { "epoch": 69.11214953271028, "grad_norm": 2.234375, "learning_rate": 3.711830108888859e-05, "loss": 0.1638, "step": 11127 }, { "epoch": 69.11838006230529, "grad_norm": 2.046875, "learning_rate": 3.711614606900105e-05, "loss": 0.1715, "step": 11128 }, { "epoch": 69.1246105919003, "grad_norm": 2.9375, "learning_rate": 3.711399093143909e-05, "loss": 0.246, "step": 11129 }, { "epoch": 69.13084112149532, "grad_norm": 1.8984375, "learning_rate": 3.7111835676223626e-05, "loss": 0.1305, "step": 11130 }, { "epoch": 69.13707165109034, "grad_norm": 2.75, "learning_rate": 3.710968030337559e-05, "loss": 0.1598, "step": 11131 }, { "epoch": 69.14330218068535, "grad_norm": 2.265625, "learning_rate": 3.710752481291592e-05, "loss": 0.1232, "step": 11132 }, { "epoch": 69.14953271028037, "grad_norm": 2.671875, "learning_rate": 3.710536920486555e-05, "loss": 0.264, "step": 11133 }, { "epoch": 69.15576323987538, "grad_norm": 3.734375, "learning_rate": 3.710321347924541e-05, "loss": 0.2005, "step": 11134 }, { "epoch": 69.1619937694704, "grad_norm": 1.9140625, "learning_rate": 3.710105763607645e-05, "loss": 0.1083, "step": 11135 }, { "epoch": 69.16822429906541, "grad_norm": 3.375, "learning_rate": 3.709890167537958e-05, "loss": 0.29, "step": 11136 }, { "epoch": 69.17445482866043, "grad_norm": 1.8125, "learning_rate": 3.709674559717578e-05, "loss": 0.1272, "step": 11137 }, { "epoch": 69.18068535825545, "grad_norm": 2.53125, "learning_rate": 3.709458940148596e-05, "loss": 0.2111, "step": 11138 }, { "epoch": 69.18691588785046, "grad_norm": 2.390625, "learning_rate": 3.709243308833107e-05, "loss": 0.1514, "step": 11139 }, { "epoch": 69.19314641744548, "grad_norm": 2.734375, "learning_rate": 3.7090276657732055e-05, "loss": 0.2093, "step": 11140 }, { "epoch": 69.19937694704049, "grad_norm": 2.0625, "learning_rate": 3.708812010970985e-05, "loss": 0.1705, "step": 11141 }, { "epoch": 69.20560747663552, "grad_norm": 2.59375, "learning_rate": 3.708596344428541e-05, "loss": 0.2204, "step": 11142 }, { "epoch": 69.21183800623054, "grad_norm": 2.59375, "learning_rate": 3.708380666147968e-05, "loss": 0.1564, "step": 11143 }, { "epoch": 69.21806853582555, "grad_norm": 1.953125, "learning_rate": 3.70816497613136e-05, "loss": 0.1739, "step": 11144 }, { "epoch": 69.22429906542057, "grad_norm": 1.828125, "learning_rate": 3.7079492743808126e-05, "loss": 0.1159, "step": 11145 }, { "epoch": 69.23052959501558, "grad_norm": 1.6953125, "learning_rate": 3.7077335608984204e-05, "loss": 0.167, "step": 11146 }, { "epoch": 69.2367601246106, "grad_norm": 1.3515625, "learning_rate": 3.707517835686278e-05, "loss": 0.1103, "step": 11147 }, { "epoch": 69.24299065420561, "grad_norm": 1.7890625, "learning_rate": 3.707302098746481e-05, "loss": 0.1713, "step": 11148 }, { "epoch": 69.24922118380063, "grad_norm": 2.484375, "learning_rate": 3.707086350081125e-05, "loss": 0.1661, "step": 11149 }, { "epoch": 69.25545171339564, "grad_norm": 1.890625, "learning_rate": 3.706870589692305e-05, "loss": 0.1496, "step": 11150 }, { "epoch": 69.26168224299066, "grad_norm": 4.53125, "learning_rate": 3.706654817582116e-05, "loss": 0.3502, "step": 11151 }, { "epoch": 69.26791277258567, "grad_norm": 1.7265625, "learning_rate": 3.706439033752655e-05, "loss": 0.1112, "step": 11152 }, { "epoch": 69.27414330218069, "grad_norm": 2.578125, "learning_rate": 3.706223238206017e-05, "loss": 0.2272, "step": 11153 }, { "epoch": 69.2803738317757, "grad_norm": 2.390625, "learning_rate": 3.706007430944298e-05, "loss": 0.1388, "step": 11154 }, { "epoch": 69.28660436137072, "grad_norm": 2.734375, "learning_rate": 3.705791611969593e-05, "loss": 0.2619, "step": 11155 }, { "epoch": 69.29283489096574, "grad_norm": 1.3359375, "learning_rate": 3.705575781283999e-05, "loss": 0.1209, "step": 11156 }, { "epoch": 69.29906542056075, "grad_norm": 2.765625, "learning_rate": 3.705359938889612e-05, "loss": 0.1155, "step": 11157 }, { "epoch": 69.30529595015577, "grad_norm": 1.7734375, "learning_rate": 3.7051440847885284e-05, "loss": 0.1542, "step": 11158 }, { "epoch": 69.31152647975078, "grad_norm": 1.5546875, "learning_rate": 3.704928218982845e-05, "loss": 0.1085, "step": 11159 }, { "epoch": 69.3177570093458, "grad_norm": 1.953125, "learning_rate": 3.704712341474657e-05, "loss": 0.1251, "step": 11160 }, { "epoch": 69.32398753894081, "grad_norm": 2.6875, "learning_rate": 3.7044964522660626e-05, "loss": 0.1723, "step": 11161 }, { "epoch": 69.33021806853583, "grad_norm": 2.375, "learning_rate": 3.7042805513591574e-05, "loss": 0.1443, "step": 11162 }, { "epoch": 69.33644859813084, "grad_norm": 2.78125, "learning_rate": 3.70406463875604e-05, "loss": 0.2249, "step": 11163 }, { "epoch": 69.34267912772586, "grad_norm": 3.921875, "learning_rate": 3.703848714458804e-05, "loss": 0.2697, "step": 11164 }, { "epoch": 69.34890965732087, "grad_norm": 3.359375, "learning_rate": 3.70363277846955e-05, "loss": 0.2108, "step": 11165 }, { "epoch": 69.35514018691589, "grad_norm": 2.484375, "learning_rate": 3.703416830790374e-05, "loss": 0.1283, "step": 11166 }, { "epoch": 69.3613707165109, "grad_norm": 3.015625, "learning_rate": 3.703200871423372e-05, "loss": 0.2325, "step": 11167 }, { "epoch": 69.36760124610592, "grad_norm": 1.5390625, "learning_rate": 3.702984900370644e-05, "loss": 0.1017, "step": 11168 }, { "epoch": 69.37383177570094, "grad_norm": 2.625, "learning_rate": 3.7027689176342855e-05, "loss": 0.1223, "step": 11169 }, { "epoch": 69.38006230529595, "grad_norm": 2.484375, "learning_rate": 3.702552923216395e-05, "loss": 0.121, "step": 11170 }, { "epoch": 69.38629283489097, "grad_norm": 1.8984375, "learning_rate": 3.70233691711907e-05, "loss": 0.1774, "step": 11171 }, { "epoch": 69.39252336448598, "grad_norm": 2.6875, "learning_rate": 3.702120899344409e-05, "loss": 0.135, "step": 11172 }, { "epoch": 69.398753894081, "grad_norm": 2.40625, "learning_rate": 3.701904869894509e-05, "loss": 0.2193, "step": 11173 }, { "epoch": 69.40498442367601, "grad_norm": 3.265625, "learning_rate": 3.701688828771469e-05, "loss": 0.1925, "step": 11174 }, { "epoch": 69.41121495327103, "grad_norm": 2.046875, "learning_rate": 3.7014727759773867e-05, "loss": 0.1253, "step": 11175 }, { "epoch": 69.41744548286604, "grad_norm": 2.71875, "learning_rate": 3.70125671151436e-05, "loss": 0.2311, "step": 11176 }, { "epoch": 69.42367601246106, "grad_norm": 2.265625, "learning_rate": 3.701040635384489e-05, "loss": 0.2291, "step": 11177 }, { "epoch": 69.42990654205607, "grad_norm": 2.109375, "learning_rate": 3.700824547589871e-05, "loss": 0.1121, "step": 11178 }, { "epoch": 69.43613707165109, "grad_norm": 2.96875, "learning_rate": 3.700608448132604e-05, "loss": 0.2039, "step": 11179 }, { "epoch": 69.4423676012461, "grad_norm": 3.0, "learning_rate": 3.700392337014789e-05, "loss": 0.1148, "step": 11180 }, { "epoch": 69.44859813084112, "grad_norm": 2.265625, "learning_rate": 3.7001762142385234e-05, "loss": 0.1647, "step": 11181 }, { "epoch": 69.45482866043614, "grad_norm": 2.265625, "learning_rate": 3.6999600798059065e-05, "loss": 0.1844, "step": 11182 }, { "epoch": 69.46105919003115, "grad_norm": 2.140625, "learning_rate": 3.699743933719038e-05, "loss": 0.1547, "step": 11183 }, { "epoch": 69.46728971962617, "grad_norm": 2.65625, "learning_rate": 3.699527775980015e-05, "loss": 0.1808, "step": 11184 }, { "epoch": 69.47352024922118, "grad_norm": 1.2734375, "learning_rate": 3.6993116065909396e-05, "loss": 0.13, "step": 11185 }, { "epoch": 69.4797507788162, "grad_norm": 1.96875, "learning_rate": 3.699095425553909e-05, "loss": 0.1392, "step": 11186 }, { "epoch": 69.48598130841121, "grad_norm": 3.953125, "learning_rate": 3.6988792328710257e-05, "loss": 0.3726, "step": 11187 }, { "epoch": 69.49221183800623, "grad_norm": 1.921875, "learning_rate": 3.698663028544386e-05, "loss": 0.1619, "step": 11188 }, { "epoch": 69.49844236760124, "grad_norm": 2.328125, "learning_rate": 3.698446812576093e-05, "loss": 0.1636, "step": 11189 }, { "epoch": 69.50467289719626, "grad_norm": 2.0625, "learning_rate": 3.698230584968243e-05, "loss": 0.1463, "step": 11190 }, { "epoch": 69.51090342679127, "grad_norm": 3.34375, "learning_rate": 3.69801434572294e-05, "loss": 0.2965, "step": 11191 }, { "epoch": 69.51713395638629, "grad_norm": 2.421875, "learning_rate": 3.6977980948422814e-05, "loss": 0.2417, "step": 11192 }, { "epoch": 69.5233644859813, "grad_norm": 2.53125, "learning_rate": 3.697581832328368e-05, "loss": 0.1615, "step": 11193 }, { "epoch": 69.52959501557632, "grad_norm": 2.03125, "learning_rate": 3.697365558183301e-05, "loss": 0.1196, "step": 11194 }, { "epoch": 69.53582554517133, "grad_norm": 2.171875, "learning_rate": 3.6971492724091795e-05, "loss": 0.2068, "step": 11195 }, { "epoch": 69.54205607476635, "grad_norm": 2.4375, "learning_rate": 3.696932975008105e-05, "loss": 0.1094, "step": 11196 }, { "epoch": 69.54828660436137, "grad_norm": 2.5, "learning_rate": 3.6967166659821794e-05, "loss": 0.1502, "step": 11197 }, { "epoch": 69.55451713395638, "grad_norm": 2.4375, "learning_rate": 3.696500345333501e-05, "loss": 0.1309, "step": 11198 }, { "epoch": 69.5607476635514, "grad_norm": 2.359375, "learning_rate": 3.696284013064173e-05, "loss": 0.2105, "step": 11199 }, { "epoch": 69.56697819314641, "grad_norm": 2.421875, "learning_rate": 3.696067669176295e-05, "loss": 0.1966, "step": 11200 }, { "epoch": 69.57320872274143, "grad_norm": 3.1875, "learning_rate": 3.695851313671969e-05, "loss": 0.2694, "step": 11201 }, { "epoch": 69.57943925233644, "grad_norm": 3.203125, "learning_rate": 3.695634946553296e-05, "loss": 0.1579, "step": 11202 }, { "epoch": 69.58566978193146, "grad_norm": 3.34375, "learning_rate": 3.695418567822377e-05, "loss": 0.2168, "step": 11203 }, { "epoch": 69.59190031152647, "grad_norm": 3.015625, "learning_rate": 3.695202177481314e-05, "loss": 0.157, "step": 11204 }, { "epoch": 69.59813084112149, "grad_norm": 2.765625, "learning_rate": 3.6949857755322094e-05, "loss": 0.1725, "step": 11205 }, { "epoch": 69.60436137071652, "grad_norm": 1.828125, "learning_rate": 3.694769361977163e-05, "loss": 0.0994, "step": 11206 }, { "epoch": 69.61059190031153, "grad_norm": 3.8125, "learning_rate": 3.6945529368182786e-05, "loss": 0.2395, "step": 11207 }, { "epoch": 69.61682242990655, "grad_norm": 3.0625, "learning_rate": 3.6943365000576565e-05, "loss": 0.2418, "step": 11208 }, { "epoch": 69.62305295950156, "grad_norm": 1.71875, "learning_rate": 3.6941200516973994e-05, "loss": 0.1028, "step": 11209 }, { "epoch": 69.62928348909658, "grad_norm": 2.84375, "learning_rate": 3.6939035917396105e-05, "loss": 0.1519, "step": 11210 }, { "epoch": 69.6355140186916, "grad_norm": 2.15625, "learning_rate": 3.693687120186391e-05, "loss": 0.1609, "step": 11211 }, { "epoch": 69.64174454828661, "grad_norm": 2.09375, "learning_rate": 3.693470637039843e-05, "loss": 0.1406, "step": 11212 }, { "epoch": 69.64797507788163, "grad_norm": 2.671875, "learning_rate": 3.693254142302071e-05, "loss": 0.1343, "step": 11213 }, { "epoch": 69.65420560747664, "grad_norm": 2.640625, "learning_rate": 3.693037635975175e-05, "loss": 0.2335, "step": 11214 }, { "epoch": 69.66043613707166, "grad_norm": 1.2734375, "learning_rate": 3.6928211180612585e-05, "loss": 0.0825, "step": 11215 }, { "epoch": 69.66666666666667, "grad_norm": 2.71875, "learning_rate": 3.692604588562426e-05, "loss": 0.2072, "step": 11216 }, { "epoch": 69.67289719626169, "grad_norm": 2.953125, "learning_rate": 3.6923880474807794e-05, "loss": 0.2326, "step": 11217 }, { "epoch": 69.6791277258567, "grad_norm": 2.0, "learning_rate": 3.692171494818421e-05, "loss": 0.1689, "step": 11218 }, { "epoch": 69.68535825545172, "grad_norm": 3.703125, "learning_rate": 3.6919549305774556e-05, "loss": 0.4028, "step": 11219 }, { "epoch": 69.69158878504673, "grad_norm": 2.6875, "learning_rate": 3.691738354759985e-05, "loss": 0.2262, "step": 11220 }, { "epoch": 69.69781931464175, "grad_norm": 3.625, "learning_rate": 3.6915217673681124e-05, "loss": 0.1725, "step": 11221 }, { "epoch": 69.70404984423676, "grad_norm": 2.078125, "learning_rate": 3.691305168403944e-05, "loss": 0.142, "step": 11222 }, { "epoch": 69.71028037383178, "grad_norm": 2.4375, "learning_rate": 3.691088557869581e-05, "loss": 0.2139, "step": 11223 }, { "epoch": 69.7165109034268, "grad_norm": 2.109375, "learning_rate": 3.6908719357671276e-05, "loss": 0.1836, "step": 11224 }, { "epoch": 69.72274143302181, "grad_norm": 2.359375, "learning_rate": 3.6906553020986876e-05, "loss": 0.1289, "step": 11225 }, { "epoch": 69.72897196261682, "grad_norm": 1.75, "learning_rate": 3.690438656866366e-05, "loss": 0.1267, "step": 11226 }, { "epoch": 69.73520249221184, "grad_norm": 1.9140625, "learning_rate": 3.6902220000722657e-05, "loss": 0.0984, "step": 11227 }, { "epoch": 69.74143302180686, "grad_norm": 1.65625, "learning_rate": 3.690005331718491e-05, "loss": 0.1261, "step": 11228 }, { "epoch": 69.74766355140187, "grad_norm": 2.1875, "learning_rate": 3.689788651807148e-05, "loss": 0.1698, "step": 11229 }, { "epoch": 69.75389408099689, "grad_norm": 1.2890625, "learning_rate": 3.6895719603403387e-05, "loss": 0.1123, "step": 11230 }, { "epoch": 69.7601246105919, "grad_norm": 2.265625, "learning_rate": 3.6893552573201694e-05, "loss": 0.1552, "step": 11231 }, { "epoch": 69.76635514018692, "grad_norm": 3.78125, "learning_rate": 3.6891385427487435e-05, "loss": 0.2858, "step": 11232 }, { "epoch": 69.77258566978193, "grad_norm": 3.234375, "learning_rate": 3.688921816628167e-05, "loss": 0.2045, "step": 11233 }, { "epoch": 69.77881619937695, "grad_norm": 3.0, "learning_rate": 3.688705078960544e-05, "loss": 0.2287, "step": 11234 }, { "epoch": 69.78504672897196, "grad_norm": 2.109375, "learning_rate": 3.68848832974798e-05, "loss": 0.2355, "step": 11235 }, { "epoch": 69.79127725856698, "grad_norm": 2.984375, "learning_rate": 3.6882715689925784e-05, "loss": 0.3157, "step": 11236 }, { "epoch": 69.797507788162, "grad_norm": 2.359375, "learning_rate": 3.688054796696446e-05, "loss": 0.2916, "step": 11237 }, { "epoch": 69.80373831775701, "grad_norm": 2.609375, "learning_rate": 3.687838012861689e-05, "loss": 0.2196, "step": 11238 }, { "epoch": 69.80996884735202, "grad_norm": 4.15625, "learning_rate": 3.6876212174904115e-05, "loss": 0.3413, "step": 11239 }, { "epoch": 69.81619937694704, "grad_norm": 2.484375, "learning_rate": 3.6874044105847195e-05, "loss": 0.1816, "step": 11240 }, { "epoch": 69.82242990654206, "grad_norm": 2.4375, "learning_rate": 3.6871875921467174e-05, "loss": 0.1733, "step": 11241 }, { "epoch": 69.82866043613707, "grad_norm": 2.765625, "learning_rate": 3.686970762178513e-05, "loss": 0.1791, "step": 11242 }, { "epoch": 69.83489096573209, "grad_norm": 3.734375, "learning_rate": 3.686753920682211e-05, "loss": 0.2484, "step": 11243 }, { "epoch": 69.8411214953271, "grad_norm": 2.390625, "learning_rate": 3.6865370676599184e-05, "loss": 0.1654, "step": 11244 }, { "epoch": 69.84735202492212, "grad_norm": 2.96875, "learning_rate": 3.68632020311374e-05, "loss": 0.2448, "step": 11245 }, { "epoch": 69.85358255451713, "grad_norm": 2.734375, "learning_rate": 3.6861033270457825e-05, "loss": 0.1962, "step": 11246 }, { "epoch": 69.85981308411215, "grad_norm": 1.8671875, "learning_rate": 3.685886439458152e-05, "loss": 0.113, "step": 11247 }, { "epoch": 69.86604361370716, "grad_norm": 1.796875, "learning_rate": 3.685669540352957e-05, "loss": 0.1456, "step": 11248 }, { "epoch": 69.87227414330218, "grad_norm": 2.25, "learning_rate": 3.685452629732301e-05, "loss": 0.1391, "step": 11249 }, { "epoch": 69.8785046728972, "grad_norm": 1.6796875, "learning_rate": 3.685235707598292e-05, "loss": 0.153, "step": 11250 }, { "epoch": 69.88473520249221, "grad_norm": 2.671875, "learning_rate": 3.685018773953037e-05, "loss": 0.2455, "step": 11251 }, { "epoch": 69.89096573208722, "grad_norm": 2.140625, "learning_rate": 3.684801828798644e-05, "loss": 0.1701, "step": 11252 }, { "epoch": 69.89719626168224, "grad_norm": 2.21875, "learning_rate": 3.684584872137218e-05, "loss": 0.1411, "step": 11253 }, { "epoch": 69.90342679127725, "grad_norm": 1.3125, "learning_rate": 3.6843679039708664e-05, "loss": 0.1248, "step": 11254 }, { "epoch": 69.90965732087227, "grad_norm": 1.3984375, "learning_rate": 3.684150924301697e-05, "loss": 0.1421, "step": 11255 }, { "epoch": 69.91588785046729, "grad_norm": 1.5625, "learning_rate": 3.683933933131818e-05, "loss": 0.1425, "step": 11256 }, { "epoch": 69.9221183800623, "grad_norm": 3.21875, "learning_rate": 3.683716930463335e-05, "loss": 0.3568, "step": 11257 }, { "epoch": 69.92834890965732, "grad_norm": 2.578125, "learning_rate": 3.683499916298357e-05, "loss": 0.1661, "step": 11258 }, { "epoch": 69.93457943925233, "grad_norm": 4.5625, "learning_rate": 3.6832828906389913e-05, "loss": 0.213, "step": 11259 }, { "epoch": 69.94080996884735, "grad_norm": 3.3125, "learning_rate": 3.683065853487346e-05, "loss": 0.1836, "step": 11260 }, { "epoch": 69.94704049844236, "grad_norm": 3.921875, "learning_rate": 3.682848804845528e-05, "loss": 0.2417, "step": 11261 }, { "epoch": 69.95327102803738, "grad_norm": 1.734375, "learning_rate": 3.682631744715646e-05, "loss": 0.1268, "step": 11262 }, { "epoch": 69.9595015576324, "grad_norm": 3.265625, "learning_rate": 3.682414673099809e-05, "loss": 0.1968, "step": 11263 }, { "epoch": 69.96573208722741, "grad_norm": 3.78125, "learning_rate": 3.682197590000123e-05, "loss": 0.2245, "step": 11264 }, { "epoch": 69.97196261682242, "grad_norm": 2.71875, "learning_rate": 3.6819804954186986e-05, "loss": 0.2, "step": 11265 }, { "epoch": 69.97819314641744, "grad_norm": 1.1796875, "learning_rate": 3.6817633893576434e-05, "loss": 0.1107, "step": 11266 }, { "epoch": 69.98442367601245, "grad_norm": 2.28125, "learning_rate": 3.681546271819065e-05, "loss": 0.1516, "step": 11267 }, { "epoch": 69.99065420560747, "grad_norm": 1.484375, "learning_rate": 3.681329142805075e-05, "loss": 0.1201, "step": 11268 }, { "epoch": 69.99688473520249, "grad_norm": 2.359375, "learning_rate": 3.6811120023177775e-05, "loss": 0.2991, "step": 11269 }, { "epoch": 70.0, "grad_norm": 1.1484375, "learning_rate": 3.680894850359286e-05, "loss": 0.0649, "step": 11270 }, { "epoch": 70.00623052959502, "grad_norm": 3.015625, "learning_rate": 3.680677686931707e-05, "loss": 0.1564, "step": 11271 }, { "epoch": 70.01246105919003, "grad_norm": 2.34375, "learning_rate": 3.6804605120371503e-05, "loss": 0.1778, "step": 11272 }, { "epoch": 70.01869158878505, "grad_norm": 2.53125, "learning_rate": 3.6802433256777254e-05, "loss": 0.1253, "step": 11273 }, { "epoch": 70.02492211838006, "grad_norm": 2.859375, "learning_rate": 3.680026127855541e-05, "loss": 0.2018, "step": 11274 }, { "epoch": 70.03115264797508, "grad_norm": 2.09375, "learning_rate": 3.679808918572708e-05, "loss": 0.25, "step": 11275 }, { "epoch": 70.03738317757009, "grad_norm": 3.453125, "learning_rate": 3.679591697831334e-05, "loss": 0.1504, "step": 11276 }, { "epoch": 70.04361370716511, "grad_norm": 2.96875, "learning_rate": 3.67937446563353e-05, "loss": 0.2081, "step": 11277 }, { "epoch": 70.04984423676012, "grad_norm": 1.9765625, "learning_rate": 3.679157221981405e-05, "loss": 0.1644, "step": 11278 }, { "epoch": 70.05607476635514, "grad_norm": 1.0, "learning_rate": 3.67893996687707e-05, "loss": 0.1075, "step": 11279 }, { "epoch": 70.06230529595015, "grad_norm": 2.65625, "learning_rate": 3.6787227003226335e-05, "loss": 0.2335, "step": 11280 }, { "epoch": 70.06853582554517, "grad_norm": 1.6015625, "learning_rate": 3.678505422320207e-05, "loss": 0.112, "step": 11281 }, { "epoch": 70.07476635514018, "grad_norm": 2.453125, "learning_rate": 3.6782881328719e-05, "loss": 0.1575, "step": 11282 }, { "epoch": 70.0809968847352, "grad_norm": 2.625, "learning_rate": 3.678070831979824e-05, "loss": 0.2011, "step": 11283 }, { "epoch": 70.08722741433021, "grad_norm": 2.140625, "learning_rate": 3.677853519646087e-05, "loss": 0.1747, "step": 11284 }, { "epoch": 70.09345794392523, "grad_norm": 1.7890625, "learning_rate": 3.677636195872802e-05, "loss": 0.1636, "step": 11285 }, { "epoch": 70.09968847352025, "grad_norm": 1.71875, "learning_rate": 3.67741886066208e-05, "loss": 0.1087, "step": 11286 }, { "epoch": 70.10591900311526, "grad_norm": 2.609375, "learning_rate": 3.677201514016029e-05, "loss": 0.1602, "step": 11287 }, { "epoch": 70.11214953271028, "grad_norm": 3.0625, "learning_rate": 3.676984155936762e-05, "loss": 0.114, "step": 11288 }, { "epoch": 70.11838006230529, "grad_norm": 2.453125, "learning_rate": 3.67676678642639e-05, "loss": 0.1413, "step": 11289 }, { "epoch": 70.1246105919003, "grad_norm": 1.46875, "learning_rate": 3.676549405487024e-05, "loss": 0.1214, "step": 11290 }, { "epoch": 70.13084112149532, "grad_norm": 2.296875, "learning_rate": 3.6763320131207744e-05, "loss": 0.1958, "step": 11291 }, { "epoch": 70.13707165109034, "grad_norm": 1.453125, "learning_rate": 3.676114609329753e-05, "loss": 0.1248, "step": 11292 }, { "epoch": 70.14330218068535, "grad_norm": 1.59375, "learning_rate": 3.675897194116072e-05, "loss": 0.2391, "step": 11293 }, { "epoch": 70.14953271028037, "grad_norm": 1.6640625, "learning_rate": 3.675679767481842e-05, "loss": 0.1327, "step": 11294 }, { "epoch": 70.15576323987538, "grad_norm": 1.53125, "learning_rate": 3.675462329429175e-05, "loss": 0.0906, "step": 11295 }, { "epoch": 70.1619937694704, "grad_norm": 2.140625, "learning_rate": 3.6752448799601835e-05, "loss": 0.2595, "step": 11296 }, { "epoch": 70.16822429906541, "grad_norm": 3.0625, "learning_rate": 3.6750274190769785e-05, "loss": 0.2741, "step": 11297 }, { "epoch": 70.17445482866043, "grad_norm": 1.78125, "learning_rate": 3.674809946781672e-05, "loss": 0.1663, "step": 11298 }, { "epoch": 70.18068535825545, "grad_norm": 1.078125, "learning_rate": 3.674592463076377e-05, "loss": 0.1018, "step": 11299 }, { "epoch": 70.18691588785046, "grad_norm": 3.046875, "learning_rate": 3.674374967963204e-05, "loss": 0.2346, "step": 11300 }, { "epoch": 70.19314641744548, "grad_norm": 1.8515625, "learning_rate": 3.6741574614442683e-05, "loss": 0.1365, "step": 11301 }, { "epoch": 70.19937694704049, "grad_norm": 2.359375, "learning_rate": 3.67393994352168e-05, "loss": 0.1795, "step": 11302 }, { "epoch": 70.20560747663552, "grad_norm": 1.5078125, "learning_rate": 3.673722414197552e-05, "loss": 0.1004, "step": 11303 }, { "epoch": 70.21183800623054, "grad_norm": 2.78125, "learning_rate": 3.673504873473997e-05, "loss": 0.113, "step": 11304 }, { "epoch": 70.21806853582555, "grad_norm": 2.515625, "learning_rate": 3.673287321353129e-05, "loss": 0.1299, "step": 11305 }, { "epoch": 70.22429906542057, "grad_norm": 3.734375, "learning_rate": 3.6730697578370585e-05, "loss": 0.2114, "step": 11306 }, { "epoch": 70.23052959501558, "grad_norm": 3.875, "learning_rate": 3.672852182927901e-05, "loss": 0.3068, "step": 11307 }, { "epoch": 70.2367601246106, "grad_norm": 2.921875, "learning_rate": 3.672634596627769e-05, "loss": 0.2555, "step": 11308 }, { "epoch": 70.24299065420561, "grad_norm": 2.640625, "learning_rate": 3.672416998938775e-05, "loss": 0.2701, "step": 11309 }, { "epoch": 70.24922118380063, "grad_norm": 3.875, "learning_rate": 3.6721993898630324e-05, "loss": 0.1053, "step": 11310 }, { "epoch": 70.25545171339564, "grad_norm": 2.03125, "learning_rate": 3.6719817694026556e-05, "loss": 0.1335, "step": 11311 }, { "epoch": 70.26168224299066, "grad_norm": 3.421875, "learning_rate": 3.671764137559757e-05, "loss": 0.2492, "step": 11312 }, { "epoch": 70.26791277258567, "grad_norm": 1.5703125, "learning_rate": 3.671546494336451e-05, "loss": 0.1359, "step": 11313 }, { "epoch": 70.27414330218069, "grad_norm": 3.0625, "learning_rate": 3.671328839734852e-05, "loss": 0.1701, "step": 11314 }, { "epoch": 70.2803738317757, "grad_norm": 2.28125, "learning_rate": 3.671111173757072e-05, "loss": 0.2333, "step": 11315 }, { "epoch": 70.28660436137072, "grad_norm": 1.953125, "learning_rate": 3.670893496405228e-05, "loss": 0.1313, "step": 11316 }, { "epoch": 70.29283489096574, "grad_norm": 2.796875, "learning_rate": 3.67067580768143e-05, "loss": 0.1636, "step": 11317 }, { "epoch": 70.29906542056075, "grad_norm": 3.40625, "learning_rate": 3.6704581075877966e-05, "loss": 0.2096, "step": 11318 }, { "epoch": 70.30529595015577, "grad_norm": 3.65625, "learning_rate": 3.6702403961264384e-05, "loss": 0.1336, "step": 11319 }, { "epoch": 70.31152647975078, "grad_norm": 2.296875, "learning_rate": 3.670022673299472e-05, "loss": 0.1799, "step": 11320 }, { "epoch": 70.3177570093458, "grad_norm": 2.609375, "learning_rate": 3.669804939109012e-05, "loss": 0.1458, "step": 11321 }, { "epoch": 70.32398753894081, "grad_norm": 2.625, "learning_rate": 3.669587193557173e-05, "loss": 0.1197, "step": 11322 }, { "epoch": 70.33021806853583, "grad_norm": 2.8125, "learning_rate": 3.669369436646069e-05, "loss": 0.1192, "step": 11323 }, { "epoch": 70.33644859813084, "grad_norm": 2.3125, "learning_rate": 3.6691516683778144e-05, "loss": 0.1958, "step": 11324 }, { "epoch": 70.34267912772586, "grad_norm": 2.53125, "learning_rate": 3.668933888754526e-05, "loss": 0.157, "step": 11325 }, { "epoch": 70.34890965732087, "grad_norm": 1.90625, "learning_rate": 3.6687160977783176e-05, "loss": 0.1535, "step": 11326 }, { "epoch": 70.35514018691589, "grad_norm": 3.640625, "learning_rate": 3.668498295451305e-05, "loss": 0.2234, "step": 11327 }, { "epoch": 70.3613707165109, "grad_norm": 1.3359375, "learning_rate": 3.668280481775603e-05, "loss": 0.1097, "step": 11328 }, { "epoch": 70.36760124610592, "grad_norm": 3.40625, "learning_rate": 3.6680626567533285e-05, "loss": 0.1108, "step": 11329 }, { "epoch": 70.37383177570094, "grad_norm": 1.953125, "learning_rate": 3.6678448203865945e-05, "loss": 0.1305, "step": 11330 }, { "epoch": 70.38006230529595, "grad_norm": 1.6875, "learning_rate": 3.6676269726775196e-05, "loss": 0.1372, "step": 11331 }, { "epoch": 70.38629283489097, "grad_norm": 2.078125, "learning_rate": 3.667409113628217e-05, "loss": 0.2106, "step": 11332 }, { "epoch": 70.39252336448598, "grad_norm": 3.328125, "learning_rate": 3.667191243240804e-05, "loss": 0.1682, "step": 11333 }, { "epoch": 70.398753894081, "grad_norm": 1.9375, "learning_rate": 3.666973361517397e-05, "loss": 0.1407, "step": 11334 }, { "epoch": 70.40498442367601, "grad_norm": 1.5625, "learning_rate": 3.666755468460111e-05, "loss": 0.1476, "step": 11335 }, { "epoch": 70.41121495327103, "grad_norm": 1.5234375, "learning_rate": 3.6665375640710624e-05, "loss": 0.1344, "step": 11336 }, { "epoch": 70.41744548286604, "grad_norm": 2.953125, "learning_rate": 3.666319648352368e-05, "loss": 0.1079, "step": 11337 }, { "epoch": 70.42367601246106, "grad_norm": 2.296875, "learning_rate": 3.666101721306144e-05, "loss": 0.1518, "step": 11338 }, { "epoch": 70.42990654205607, "grad_norm": 3.046875, "learning_rate": 3.665883782934507e-05, "loss": 0.1408, "step": 11339 }, { "epoch": 70.43613707165109, "grad_norm": 3.59375, "learning_rate": 3.665665833239574e-05, "loss": 0.2662, "step": 11340 }, { "epoch": 70.4423676012461, "grad_norm": 2.40625, "learning_rate": 3.665447872223461e-05, "loss": 0.1828, "step": 11341 }, { "epoch": 70.44859813084112, "grad_norm": 2.625, "learning_rate": 3.665229899888286e-05, "loss": 0.1853, "step": 11342 }, { "epoch": 70.45482866043614, "grad_norm": 3.015625, "learning_rate": 3.6650119162361635e-05, "loss": 0.2516, "step": 11343 }, { "epoch": 70.46105919003115, "grad_norm": 1.6328125, "learning_rate": 3.664793921269214e-05, "loss": 0.1224, "step": 11344 }, { "epoch": 70.46728971962617, "grad_norm": 2.59375, "learning_rate": 3.6645759149895526e-05, "loss": 0.1454, "step": 11345 }, { "epoch": 70.47352024922118, "grad_norm": 1.6796875, "learning_rate": 3.664357897399297e-05, "loss": 0.1037, "step": 11346 }, { "epoch": 70.4797507788162, "grad_norm": 2.984375, "learning_rate": 3.6641398685005654e-05, "loss": 0.1659, "step": 11347 }, { "epoch": 70.48598130841121, "grad_norm": 1.1796875, "learning_rate": 3.663921828295474e-05, "loss": 0.1058, "step": 11348 }, { "epoch": 70.49221183800623, "grad_norm": 3.0625, "learning_rate": 3.663703776786142e-05, "loss": 0.1739, "step": 11349 }, { "epoch": 70.49844236760124, "grad_norm": 3.484375, "learning_rate": 3.6634857139746856e-05, "loss": 0.3001, "step": 11350 }, { "epoch": 70.50467289719626, "grad_norm": 3.921875, "learning_rate": 3.663267639863223e-05, "loss": 0.3808, "step": 11351 }, { "epoch": 70.51090342679127, "grad_norm": 3.53125, "learning_rate": 3.6630495544538726e-05, "loss": 0.1661, "step": 11352 }, { "epoch": 70.51713395638629, "grad_norm": 2.859375, "learning_rate": 3.6628314577487536e-05, "loss": 0.1759, "step": 11353 }, { "epoch": 70.5233644859813, "grad_norm": 3.421875, "learning_rate": 3.6626133497499816e-05, "loss": 0.2634, "step": 11354 }, { "epoch": 70.52959501557632, "grad_norm": 1.6484375, "learning_rate": 3.662395230459677e-05, "loss": 0.1145, "step": 11355 }, { "epoch": 70.53582554517133, "grad_norm": 2.03125, "learning_rate": 3.6621770998799574e-05, "loss": 0.1748, "step": 11356 }, { "epoch": 70.54205607476635, "grad_norm": 2.625, "learning_rate": 3.6619589580129424e-05, "loss": 0.1779, "step": 11357 }, { "epoch": 70.54828660436137, "grad_norm": 2.125, "learning_rate": 3.661740804860748e-05, "loss": 0.1514, "step": 11358 }, { "epoch": 70.55451713395638, "grad_norm": 2.546875, "learning_rate": 3.661522640425496e-05, "loss": 0.1733, "step": 11359 }, { "epoch": 70.5607476635514, "grad_norm": 2.84375, "learning_rate": 3.661304464709304e-05, "loss": 0.3005, "step": 11360 }, { "epoch": 70.56697819314641, "grad_norm": 1.6484375, "learning_rate": 3.66108627771429e-05, "loss": 0.1277, "step": 11361 }, { "epoch": 70.57320872274143, "grad_norm": 2.75, "learning_rate": 3.660868079442575e-05, "loss": 0.2248, "step": 11362 }, { "epoch": 70.57943925233644, "grad_norm": 1.765625, "learning_rate": 3.660649869896276e-05, "loss": 0.1295, "step": 11363 }, { "epoch": 70.58566978193146, "grad_norm": 3.25, "learning_rate": 3.660431649077515e-05, "loss": 0.1518, "step": 11364 }, { "epoch": 70.59190031152647, "grad_norm": 4.1875, "learning_rate": 3.6602134169884085e-05, "loss": 0.317, "step": 11365 }, { "epoch": 70.59813084112149, "grad_norm": 1.84375, "learning_rate": 3.659995173631078e-05, "loss": 0.1243, "step": 11366 }, { "epoch": 70.60436137071652, "grad_norm": 2.890625, "learning_rate": 3.659776919007642e-05, "loss": 0.1161, "step": 11367 }, { "epoch": 70.61059190031153, "grad_norm": 2.546875, "learning_rate": 3.6595586531202206e-05, "loss": 0.181, "step": 11368 }, { "epoch": 70.61682242990655, "grad_norm": 2.75, "learning_rate": 3.659340375970934e-05, "loss": 0.1901, "step": 11369 }, { "epoch": 70.62305295950156, "grad_norm": 2.46875, "learning_rate": 3.659122087561903e-05, "loss": 0.1398, "step": 11370 }, { "epoch": 70.62928348909658, "grad_norm": 1.9921875, "learning_rate": 3.6589037878952455e-05, "loss": 0.1447, "step": 11371 }, { "epoch": 70.6355140186916, "grad_norm": 2.8125, "learning_rate": 3.658685476973083e-05, "loss": 0.2535, "step": 11372 }, { "epoch": 70.64174454828661, "grad_norm": 2.75, "learning_rate": 3.658467154797535e-05, "loss": 0.1407, "step": 11373 }, { "epoch": 70.64797507788163, "grad_norm": 2.75, "learning_rate": 3.658248821370723e-05, "loss": 0.2061, "step": 11374 }, { "epoch": 70.65420560747664, "grad_norm": 2.296875, "learning_rate": 3.658030476694767e-05, "loss": 0.2082, "step": 11375 }, { "epoch": 70.66043613707166, "grad_norm": 2.609375, "learning_rate": 3.6578121207717875e-05, "loss": 0.1568, "step": 11376 }, { "epoch": 70.66666666666667, "grad_norm": 3.828125, "learning_rate": 3.657593753603905e-05, "loss": 0.1841, "step": 11377 }, { "epoch": 70.67289719626169, "grad_norm": 2.75, "learning_rate": 3.657375375193241e-05, "loss": 0.23, "step": 11378 }, { "epoch": 70.6791277258567, "grad_norm": 1.546875, "learning_rate": 3.657156985541916e-05, "loss": 0.1301, "step": 11379 }, { "epoch": 70.68535825545172, "grad_norm": 1.7578125, "learning_rate": 3.65693858465205e-05, "loss": 0.1082, "step": 11380 }, { "epoch": 70.69158878504673, "grad_norm": 2.28125, "learning_rate": 3.6567201725257664e-05, "loss": 0.1645, "step": 11381 }, { "epoch": 70.69781931464175, "grad_norm": 3.15625, "learning_rate": 3.6565017491651844e-05, "loss": 0.1653, "step": 11382 }, { "epoch": 70.70404984423676, "grad_norm": 3.09375, "learning_rate": 3.6562833145724265e-05, "loss": 0.1565, "step": 11383 }, { "epoch": 70.71028037383178, "grad_norm": 4.0625, "learning_rate": 3.656064868749614e-05, "loss": 0.1462, "step": 11384 }, { "epoch": 70.7165109034268, "grad_norm": 3.3125, "learning_rate": 3.655846411698869e-05, "loss": 0.215, "step": 11385 }, { "epoch": 70.72274143302181, "grad_norm": 2.65625, "learning_rate": 3.6556279434223116e-05, "loss": 0.192, "step": 11386 }, { "epoch": 70.72897196261682, "grad_norm": 3.4375, "learning_rate": 3.655409463922065e-05, "loss": 0.185, "step": 11387 }, { "epoch": 70.73520249221184, "grad_norm": 2.1875, "learning_rate": 3.655190973200251e-05, "loss": 0.1146, "step": 11388 }, { "epoch": 70.74143302180686, "grad_norm": 2.3125, "learning_rate": 3.6549724712589905e-05, "loss": 0.1555, "step": 11389 }, { "epoch": 70.74766355140187, "grad_norm": 3.625, "learning_rate": 3.654753958100408e-05, "loss": 0.3005, "step": 11390 }, { "epoch": 70.75389408099689, "grad_norm": 2.015625, "learning_rate": 3.654535433726623e-05, "loss": 0.1066, "step": 11391 }, { "epoch": 70.7601246105919, "grad_norm": 1.6953125, "learning_rate": 3.654316898139759e-05, "loss": 0.1243, "step": 11392 }, { "epoch": 70.76635514018692, "grad_norm": 2.765625, "learning_rate": 3.6540983513419393e-05, "loss": 0.1275, "step": 11393 }, { "epoch": 70.77258566978193, "grad_norm": 2.609375, "learning_rate": 3.653879793335285e-05, "loss": 0.1261, "step": 11394 }, { "epoch": 70.77881619937695, "grad_norm": 3.796875, "learning_rate": 3.65366122412192e-05, "loss": 0.2154, "step": 11395 }, { "epoch": 70.78504672897196, "grad_norm": 3.46875, "learning_rate": 3.6534426437039664e-05, "loss": 0.1172, "step": 11396 }, { "epoch": 70.79127725856698, "grad_norm": 3.796875, "learning_rate": 3.653224052083547e-05, "loss": 0.2277, "step": 11397 }, { "epoch": 70.797507788162, "grad_norm": 2.828125, "learning_rate": 3.6530054492627856e-05, "loss": 0.1791, "step": 11398 }, { "epoch": 70.80373831775701, "grad_norm": 2.171875, "learning_rate": 3.652786835243804e-05, "loss": 0.1704, "step": 11399 }, { "epoch": 70.80996884735202, "grad_norm": 4.625, "learning_rate": 3.652568210028727e-05, "loss": 0.1989, "step": 11400 }, { "epoch": 70.81619937694704, "grad_norm": 4.25, "learning_rate": 3.652349573619678e-05, "loss": 0.2102, "step": 11401 }, { "epoch": 70.82242990654206, "grad_norm": 3.359375, "learning_rate": 3.6521309260187776e-05, "loss": 0.1257, "step": 11402 }, { "epoch": 70.82866043613707, "grad_norm": 3.046875, "learning_rate": 3.651912267228153e-05, "loss": 0.1743, "step": 11403 }, { "epoch": 70.83489096573209, "grad_norm": 3.328125, "learning_rate": 3.651693597249925e-05, "loss": 0.2128, "step": 11404 }, { "epoch": 70.8411214953271, "grad_norm": 3.703125, "learning_rate": 3.6514749160862194e-05, "loss": 0.119, "step": 11405 }, { "epoch": 70.84735202492212, "grad_norm": 2.171875, "learning_rate": 3.651256223739158e-05, "loss": 0.1485, "step": 11406 }, { "epoch": 70.85358255451713, "grad_norm": 2.96875, "learning_rate": 3.6510375202108675e-05, "loss": 0.2981, "step": 11407 }, { "epoch": 70.85981308411215, "grad_norm": 3.390625, "learning_rate": 3.6508188055034696e-05, "loss": 0.251, "step": 11408 }, { "epoch": 70.86604361370716, "grad_norm": 3.1875, "learning_rate": 3.65060007961909e-05, "loss": 0.2634, "step": 11409 }, { "epoch": 70.87227414330218, "grad_norm": 2.828125, "learning_rate": 3.650381342559852e-05, "loss": 0.2557, "step": 11410 }, { "epoch": 70.8785046728972, "grad_norm": 2.21875, "learning_rate": 3.6501625943278805e-05, "loss": 0.2505, "step": 11411 }, { "epoch": 70.88473520249221, "grad_norm": 2.484375, "learning_rate": 3.6499438349253004e-05, "loss": 0.2187, "step": 11412 }, { "epoch": 70.89096573208722, "grad_norm": 1.7578125, "learning_rate": 3.6497250643542345e-05, "loss": 0.1305, "step": 11413 }, { "epoch": 70.89719626168224, "grad_norm": 2.796875, "learning_rate": 3.649506282616811e-05, "loss": 0.2302, "step": 11414 }, { "epoch": 70.90342679127725, "grad_norm": 2.90625, "learning_rate": 3.6492874897151514e-05, "loss": 0.292, "step": 11415 }, { "epoch": 70.90965732087227, "grad_norm": 1.1875, "learning_rate": 3.649068685651382e-05, "loss": 0.1098, "step": 11416 }, { "epoch": 70.91588785046729, "grad_norm": 2.609375, "learning_rate": 3.648849870427629e-05, "loss": 0.234, "step": 11417 }, { "epoch": 70.9221183800623, "grad_norm": 2.234375, "learning_rate": 3.648631044046015e-05, "loss": 0.1417, "step": 11418 }, { "epoch": 70.92834890965732, "grad_norm": 2.0625, "learning_rate": 3.648412206508667e-05, "loss": 0.1905, "step": 11419 }, { "epoch": 70.93457943925233, "grad_norm": 2.171875, "learning_rate": 3.6481933578177105e-05, "loss": 0.1939, "step": 11420 }, { "epoch": 70.94080996884735, "grad_norm": 1.265625, "learning_rate": 3.6479744979752706e-05, "loss": 0.11, "step": 11421 }, { "epoch": 70.94704049844236, "grad_norm": 2.46875, "learning_rate": 3.647755626983472e-05, "loss": 0.1874, "step": 11422 }, { "epoch": 70.95327102803738, "grad_norm": 1.8359375, "learning_rate": 3.6475367448444426e-05, "loss": 0.1854, "step": 11423 }, { "epoch": 70.9595015576324, "grad_norm": 1.8671875, "learning_rate": 3.6473178515603065e-05, "loss": 0.1409, "step": 11424 }, { "epoch": 70.96573208722741, "grad_norm": 1.640625, "learning_rate": 3.64709894713319e-05, "loss": 0.1552, "step": 11425 }, { "epoch": 70.97196261682242, "grad_norm": 2.15625, "learning_rate": 3.6468800315652193e-05, "loss": 0.0981, "step": 11426 }, { "epoch": 70.97819314641744, "grad_norm": 2.484375, "learning_rate": 3.6466611048585204e-05, "loss": 0.1936, "step": 11427 }, { "epoch": 70.98442367601245, "grad_norm": 1.40625, "learning_rate": 3.646442167015219e-05, "loss": 0.1173, "step": 11428 }, { "epoch": 70.99065420560747, "grad_norm": 1.984375, "learning_rate": 3.6462232180374425e-05, "loss": 0.1249, "step": 11429 }, { "epoch": 70.99688473520249, "grad_norm": 2.359375, "learning_rate": 3.6460042579273176e-05, "loss": 0.1981, "step": 11430 }, { "epoch": 71.0, "grad_norm": 1.0546875, "learning_rate": 3.64578528668697e-05, "loss": 0.0636, "step": 11431 }, { "epoch": 71.00623052959502, "grad_norm": 4.28125, "learning_rate": 3.645566304318526e-05, "loss": 0.1995, "step": 11432 }, { "epoch": 71.01246105919003, "grad_norm": 2.609375, "learning_rate": 3.645347310824114e-05, "loss": 0.1392, "step": 11433 }, { "epoch": 71.01869158878505, "grad_norm": 3.421875, "learning_rate": 3.6451283062058595e-05, "loss": 0.1553, "step": 11434 }, { "epoch": 71.02492211838006, "grad_norm": 3.21875, "learning_rate": 3.64490929046589e-05, "loss": 0.1099, "step": 11435 }, { "epoch": 71.03115264797508, "grad_norm": 2.21875, "learning_rate": 3.644690263606333e-05, "loss": 0.1273, "step": 11436 }, { "epoch": 71.03738317757009, "grad_norm": 3.0625, "learning_rate": 3.644471225629315e-05, "loss": 0.1261, "step": 11437 }, { "epoch": 71.04361370716511, "grad_norm": 2.53125, "learning_rate": 3.644252176536963e-05, "loss": 0.1182, "step": 11438 }, { "epoch": 71.04984423676012, "grad_norm": 5.65625, "learning_rate": 3.644033116331406e-05, "loss": 0.3512, "step": 11439 }, { "epoch": 71.05607476635514, "grad_norm": 3.8125, "learning_rate": 3.6438140450147707e-05, "loss": 0.1984, "step": 11440 }, { "epoch": 71.06230529595015, "grad_norm": 4.15625, "learning_rate": 3.643594962589184e-05, "loss": 0.2149, "step": 11441 }, { "epoch": 71.06853582554517, "grad_norm": 3.3125, "learning_rate": 3.6433758690567756e-05, "loss": 0.2541, "step": 11442 }, { "epoch": 71.07476635514018, "grad_norm": 2.671875, "learning_rate": 3.643156764419671e-05, "loss": 0.2096, "step": 11443 }, { "epoch": 71.0809968847352, "grad_norm": 2.8125, "learning_rate": 3.6429376486799994e-05, "loss": 0.2404, "step": 11444 }, { "epoch": 71.08722741433021, "grad_norm": 3.171875, "learning_rate": 3.6427185218398895e-05, "loss": 0.2696, "step": 11445 }, { "epoch": 71.09345794392523, "grad_norm": 3.28125, "learning_rate": 3.6424993839014695e-05, "loss": 0.1388, "step": 11446 }, { "epoch": 71.09968847352025, "grad_norm": 2.890625, "learning_rate": 3.642280234866866e-05, "loss": 0.1154, "step": 11447 }, { "epoch": 71.10591900311526, "grad_norm": 1.78125, "learning_rate": 3.6420610747382085e-05, "loss": 0.1368, "step": 11448 }, { "epoch": 71.11214953271028, "grad_norm": 3.296875, "learning_rate": 3.6418419035176264e-05, "loss": 0.3373, "step": 11449 }, { "epoch": 71.11838006230529, "grad_norm": 2.34375, "learning_rate": 3.6416227212072465e-05, "loss": 0.1116, "step": 11450 }, { "epoch": 71.1246105919003, "grad_norm": 2.765625, "learning_rate": 3.6414035278092e-05, "loss": 0.1631, "step": 11451 }, { "epoch": 71.13084112149532, "grad_norm": 2.3125, "learning_rate": 3.641184323325613e-05, "loss": 0.1496, "step": 11452 }, { "epoch": 71.13707165109034, "grad_norm": 3.09375, "learning_rate": 3.640965107758616e-05, "loss": 0.2338, "step": 11453 }, { "epoch": 71.14330218068535, "grad_norm": 2.6875, "learning_rate": 3.640745881110337e-05, "loss": 0.2029, "step": 11454 }, { "epoch": 71.14953271028037, "grad_norm": 2.40625, "learning_rate": 3.6405266433829075e-05, "loss": 0.2485, "step": 11455 }, { "epoch": 71.15576323987538, "grad_norm": 2.90625, "learning_rate": 3.6403073945784544e-05, "loss": 0.1292, "step": 11456 }, { "epoch": 71.1619937694704, "grad_norm": 2.125, "learning_rate": 3.6400881346991086e-05, "loss": 0.1138, "step": 11457 }, { "epoch": 71.16822429906541, "grad_norm": 2.296875, "learning_rate": 3.6398688637469984e-05, "loss": 0.2513, "step": 11458 }, { "epoch": 71.17445482866043, "grad_norm": 2.6875, "learning_rate": 3.639649581724254e-05, "loss": 0.2557, "step": 11459 }, { "epoch": 71.18068535825545, "grad_norm": 3.859375, "learning_rate": 3.639430288633005e-05, "loss": 0.2597, "step": 11460 }, { "epoch": 71.18691588785046, "grad_norm": 4.03125, "learning_rate": 3.639210984475382e-05, "loss": 0.371, "step": 11461 }, { "epoch": 71.19314641744548, "grad_norm": 1.78125, "learning_rate": 3.638991669253513e-05, "loss": 0.1156, "step": 11462 }, { "epoch": 71.19937694704049, "grad_norm": 2.359375, "learning_rate": 3.6387723429695295e-05, "loss": 0.1593, "step": 11463 }, { "epoch": 71.20560747663552, "grad_norm": 1.2890625, "learning_rate": 3.638553005625562e-05, "loss": 0.1189, "step": 11464 }, { "epoch": 71.21183800623054, "grad_norm": 1.90625, "learning_rate": 3.63833365722374e-05, "loss": 0.1519, "step": 11465 }, { "epoch": 71.21806853582555, "grad_norm": 2.046875, "learning_rate": 3.638114297766193e-05, "loss": 0.1742, "step": 11466 }, { "epoch": 71.22429906542057, "grad_norm": 1.7890625, "learning_rate": 3.6378949272550536e-05, "loss": 0.1323, "step": 11467 }, { "epoch": 71.23052959501558, "grad_norm": 1.5390625, "learning_rate": 3.637675545692451e-05, "loss": 0.1487, "step": 11468 }, { "epoch": 71.2367601246106, "grad_norm": 1.2265625, "learning_rate": 3.637456153080515e-05, "loss": 0.1211, "step": 11469 }, { "epoch": 71.24299065420561, "grad_norm": 3.0, "learning_rate": 3.637236749421379e-05, "loss": 0.2078, "step": 11470 }, { "epoch": 71.24922118380063, "grad_norm": 2.9375, "learning_rate": 3.637017334717171e-05, "loss": 0.2823, "step": 11471 }, { "epoch": 71.25545171339564, "grad_norm": 1.6953125, "learning_rate": 3.6367979089700244e-05, "loss": 0.1287, "step": 11472 }, { "epoch": 71.26168224299066, "grad_norm": 2.953125, "learning_rate": 3.6365784721820694e-05, "loss": 0.2646, "step": 11473 }, { "epoch": 71.26791277258567, "grad_norm": 2.65625, "learning_rate": 3.636359024355436e-05, "loss": 0.1635, "step": 11474 }, { "epoch": 71.27414330218069, "grad_norm": 2.796875, "learning_rate": 3.6361395654922575e-05, "loss": 0.2456, "step": 11475 }, { "epoch": 71.2803738317757, "grad_norm": 2.96875, "learning_rate": 3.6359200955946635e-05, "loss": 0.2517, "step": 11476 }, { "epoch": 71.28660436137072, "grad_norm": 2.6875, "learning_rate": 3.635700614664788e-05, "loss": 0.2482, "step": 11477 }, { "epoch": 71.29283489096574, "grad_norm": 2.625, "learning_rate": 3.63548112270476e-05, "loss": 0.1176, "step": 11478 }, { "epoch": 71.29906542056075, "grad_norm": 1.59375, "learning_rate": 3.6352616197167124e-05, "loss": 0.1507, "step": 11479 }, { "epoch": 71.30529595015577, "grad_norm": 1.9375, "learning_rate": 3.6350421057027775e-05, "loss": 0.162, "step": 11480 }, { "epoch": 71.31152647975078, "grad_norm": 2.828125, "learning_rate": 3.634822580665086e-05, "loss": 0.1881, "step": 11481 }, { "epoch": 71.3177570093458, "grad_norm": 3.453125, "learning_rate": 3.634603044605772e-05, "loss": 0.119, "step": 11482 }, { "epoch": 71.32398753894081, "grad_norm": 3.34375, "learning_rate": 3.6343834975269645e-05, "loss": 0.3343, "step": 11483 }, { "epoch": 71.33021806853583, "grad_norm": 3.296875, "learning_rate": 3.6341639394308e-05, "loss": 0.1502, "step": 11484 }, { "epoch": 71.33644859813084, "grad_norm": 3.0625, "learning_rate": 3.6339443703194065e-05, "loss": 0.2893, "step": 11485 }, { "epoch": 71.34267912772586, "grad_norm": 3.9375, "learning_rate": 3.63372479019492e-05, "loss": 0.4323, "step": 11486 }, { "epoch": 71.34890965732087, "grad_norm": 2.5625, "learning_rate": 3.633505199059471e-05, "loss": 0.1695, "step": 11487 }, { "epoch": 71.35514018691589, "grad_norm": 3.078125, "learning_rate": 3.633285596915193e-05, "loss": 0.1322, "step": 11488 }, { "epoch": 71.3613707165109, "grad_norm": 2.0625, "learning_rate": 3.6330659837642187e-05, "loss": 0.1291, "step": 11489 }, { "epoch": 71.36760124610592, "grad_norm": 4.1875, "learning_rate": 3.632846359608682e-05, "loss": 0.2537, "step": 11490 }, { "epoch": 71.37383177570094, "grad_norm": 2.09375, "learning_rate": 3.632626724450714e-05, "loss": 0.1319, "step": 11491 }, { "epoch": 71.38006230529595, "grad_norm": 2.203125, "learning_rate": 3.632407078292449e-05, "loss": 0.1563, "step": 11492 }, { "epoch": 71.38629283489097, "grad_norm": 2.828125, "learning_rate": 3.632187421136021e-05, "loss": 0.2872, "step": 11493 }, { "epoch": 71.39252336448598, "grad_norm": 2.765625, "learning_rate": 3.631967752983561e-05, "loss": 0.1858, "step": 11494 }, { "epoch": 71.398753894081, "grad_norm": 3.75, "learning_rate": 3.631748073837204e-05, "loss": 0.1333, "step": 11495 }, { "epoch": 71.40498442367601, "grad_norm": 2.796875, "learning_rate": 3.631528383699085e-05, "loss": 0.2048, "step": 11496 }, { "epoch": 71.41121495327103, "grad_norm": 1.625, "learning_rate": 3.631308682571335e-05, "loss": 0.1491, "step": 11497 }, { "epoch": 71.41744548286604, "grad_norm": 2.359375, "learning_rate": 3.631088970456089e-05, "loss": 0.1291, "step": 11498 }, { "epoch": 71.42367601246106, "grad_norm": 2.640625, "learning_rate": 3.6308692473554814e-05, "loss": 0.1661, "step": 11499 }, { "epoch": 71.42990654205607, "grad_norm": 2.296875, "learning_rate": 3.630649513271645e-05, "loss": 0.1749, "step": 11500 }, { "epoch": 71.43613707165109, "grad_norm": 3.21875, "learning_rate": 3.6304297682067144e-05, "loss": 0.1713, "step": 11501 }, { "epoch": 71.4423676012461, "grad_norm": 2.0625, "learning_rate": 3.630210012162824e-05, "loss": 0.1651, "step": 11502 }, { "epoch": 71.44859813084112, "grad_norm": 1.15625, "learning_rate": 3.6299902451421086e-05, "loss": 0.0919, "step": 11503 }, { "epoch": 71.45482866043614, "grad_norm": 2.28125, "learning_rate": 3.629770467146701e-05, "loss": 0.1505, "step": 11504 }, { "epoch": 71.46105919003115, "grad_norm": 1.421875, "learning_rate": 3.629550678178738e-05, "loss": 0.1036, "step": 11505 }, { "epoch": 71.46728971962617, "grad_norm": 2.25, "learning_rate": 3.629330878240352e-05, "loss": 0.1451, "step": 11506 }, { "epoch": 71.47352024922118, "grad_norm": 2.296875, "learning_rate": 3.62911106733368e-05, "loss": 0.2121, "step": 11507 }, { "epoch": 71.4797507788162, "grad_norm": 2.21875, "learning_rate": 3.628891245460855e-05, "loss": 0.15, "step": 11508 }, { "epoch": 71.48598130841121, "grad_norm": 2.671875, "learning_rate": 3.628671412624012e-05, "loss": 0.263, "step": 11509 }, { "epoch": 71.49221183800623, "grad_norm": 1.59375, "learning_rate": 3.6284515688252876e-05, "loss": 0.1021, "step": 11510 }, { "epoch": 71.49844236760124, "grad_norm": 2.578125, "learning_rate": 3.6282317140668154e-05, "loss": 0.2566, "step": 11511 }, { "epoch": 71.50467289719626, "grad_norm": 2.09375, "learning_rate": 3.628011848350732e-05, "loss": 0.1315, "step": 11512 }, { "epoch": 71.51090342679127, "grad_norm": 2.140625, "learning_rate": 3.627791971679171e-05, "loss": 0.1631, "step": 11513 }, { "epoch": 71.51713395638629, "grad_norm": 2.234375, "learning_rate": 3.6275720840542686e-05, "loss": 0.1619, "step": 11514 }, { "epoch": 71.5233644859813, "grad_norm": 1.859375, "learning_rate": 3.627352185478162e-05, "loss": 0.1379, "step": 11515 }, { "epoch": 71.52959501557632, "grad_norm": 3.015625, "learning_rate": 3.627132275952985e-05, "loss": 0.1372, "step": 11516 }, { "epoch": 71.53582554517133, "grad_norm": 3.140625, "learning_rate": 3.626912355480874e-05, "loss": 0.2141, "step": 11517 }, { "epoch": 71.54205607476635, "grad_norm": 1.8203125, "learning_rate": 3.6266924240639644e-05, "loss": 0.1339, "step": 11518 }, { "epoch": 71.54828660436137, "grad_norm": 3.0625, "learning_rate": 3.626472481704394e-05, "loss": 0.1521, "step": 11519 }, { "epoch": 71.55451713395638, "grad_norm": 2.109375, "learning_rate": 3.626252528404297e-05, "loss": 0.1435, "step": 11520 }, { "epoch": 71.5607476635514, "grad_norm": 2.390625, "learning_rate": 3.62603256416581e-05, "loss": 0.1349, "step": 11521 }, { "epoch": 71.56697819314641, "grad_norm": 2.0, "learning_rate": 3.62581258899107e-05, "loss": 0.1365, "step": 11522 }, { "epoch": 71.57320872274143, "grad_norm": 3.71875, "learning_rate": 3.6255926028822137e-05, "loss": 0.1201, "step": 11523 }, { "epoch": 71.57943925233644, "grad_norm": 2.421875, "learning_rate": 3.625372605841376e-05, "loss": 0.1302, "step": 11524 }, { "epoch": 71.58566978193146, "grad_norm": 3.734375, "learning_rate": 3.6251525978706955e-05, "loss": 0.3899, "step": 11525 }, { "epoch": 71.59190031152647, "grad_norm": 2.71875, "learning_rate": 3.624932578972307e-05, "loss": 0.1811, "step": 11526 }, { "epoch": 71.59813084112149, "grad_norm": 2.46875, "learning_rate": 3.624712549148349e-05, "loss": 0.1655, "step": 11527 }, { "epoch": 71.60436137071652, "grad_norm": 2.15625, "learning_rate": 3.6244925084009586e-05, "loss": 0.1186, "step": 11528 }, { "epoch": 71.61059190031153, "grad_norm": 3.109375, "learning_rate": 3.624272456732271e-05, "loss": 0.2816, "step": 11529 }, { "epoch": 71.61682242990655, "grad_norm": 2.671875, "learning_rate": 3.624052394144425e-05, "loss": 0.151, "step": 11530 }, { "epoch": 71.62305295950156, "grad_norm": 3.234375, "learning_rate": 3.623832320639558e-05, "loss": 0.2066, "step": 11531 }, { "epoch": 71.62928348909658, "grad_norm": 3.40625, "learning_rate": 3.623612236219806e-05, "loss": 0.2082, "step": 11532 }, { "epoch": 71.6355140186916, "grad_norm": 1.7109375, "learning_rate": 3.623392140887307e-05, "loss": 0.1452, "step": 11533 }, { "epoch": 71.64174454828661, "grad_norm": 2.265625, "learning_rate": 3.6231720346442e-05, "loss": 0.1426, "step": 11534 }, { "epoch": 71.64797507788163, "grad_norm": 1.6015625, "learning_rate": 3.6229519174926217e-05, "loss": 0.1275, "step": 11535 }, { "epoch": 71.65420560747664, "grad_norm": 2.296875, "learning_rate": 3.6227317894347094e-05, "loss": 0.1945, "step": 11536 }, { "epoch": 71.66043613707166, "grad_norm": 2.015625, "learning_rate": 3.622511650472601e-05, "loss": 0.1575, "step": 11537 }, { "epoch": 71.66666666666667, "grad_norm": 2.15625, "learning_rate": 3.622291500608437e-05, "loss": 0.1292, "step": 11538 }, { "epoch": 71.67289719626169, "grad_norm": 1.171875, "learning_rate": 3.622071339844352e-05, "loss": 0.1059, "step": 11539 }, { "epoch": 71.6791277258567, "grad_norm": 3.75, "learning_rate": 3.6218511681824865e-05, "loss": 0.2518, "step": 11540 }, { "epoch": 71.68535825545172, "grad_norm": 1.3828125, "learning_rate": 3.6216309856249776e-05, "loss": 0.0987, "step": 11541 }, { "epoch": 71.69158878504673, "grad_norm": 2.984375, "learning_rate": 3.621410792173965e-05, "loss": 0.2039, "step": 11542 }, { "epoch": 71.69781931464175, "grad_norm": 2.671875, "learning_rate": 3.621190587831586e-05, "loss": 0.243, "step": 11543 }, { "epoch": 71.70404984423676, "grad_norm": 2.34375, "learning_rate": 3.620970372599981e-05, "loss": 0.1388, "step": 11544 }, { "epoch": 71.71028037383178, "grad_norm": 1.421875, "learning_rate": 3.6207501464812865e-05, "loss": 0.1243, "step": 11545 }, { "epoch": 71.7165109034268, "grad_norm": 2.796875, "learning_rate": 3.620529909477642e-05, "loss": 0.2574, "step": 11546 }, { "epoch": 71.72274143302181, "grad_norm": 2.21875, "learning_rate": 3.6203096615911884e-05, "loss": 0.1465, "step": 11547 }, { "epoch": 71.72897196261682, "grad_norm": 2.765625, "learning_rate": 3.6200894028240625e-05, "loss": 0.2785, "step": 11548 }, { "epoch": 71.73520249221184, "grad_norm": 2.390625, "learning_rate": 3.619869133178405e-05, "loss": 0.2201, "step": 11549 }, { "epoch": 71.74143302180686, "grad_norm": 2.328125, "learning_rate": 3.6196488526563546e-05, "loss": 0.2138, "step": 11550 }, { "epoch": 71.74766355140187, "grad_norm": 2.71875, "learning_rate": 3.619428561260051e-05, "loss": 0.2629, "step": 11551 }, { "epoch": 71.75389408099689, "grad_norm": 3.125, "learning_rate": 3.6192082589916326e-05, "loss": 0.2294, "step": 11552 }, { "epoch": 71.7601246105919, "grad_norm": 2.03125, "learning_rate": 3.61898794585324e-05, "loss": 0.1018, "step": 11553 }, { "epoch": 71.76635514018692, "grad_norm": 1.578125, "learning_rate": 3.618767621847013e-05, "loss": 0.1287, "step": 11554 }, { "epoch": 71.77258566978193, "grad_norm": 1.234375, "learning_rate": 3.618547286975092e-05, "loss": 0.0923, "step": 11555 }, { "epoch": 71.77881619937695, "grad_norm": 2.6875, "learning_rate": 3.6183269412396156e-05, "loss": 0.2496, "step": 11556 }, { "epoch": 71.78504672897196, "grad_norm": 2.375, "learning_rate": 3.618106584642724e-05, "loss": 0.1578, "step": 11557 }, { "epoch": 71.79127725856698, "grad_norm": 2.921875, "learning_rate": 3.6178862171865576e-05, "loss": 0.3672, "step": 11558 }, { "epoch": 71.797507788162, "grad_norm": 2.09375, "learning_rate": 3.617665838873258e-05, "loss": 0.1171, "step": 11559 }, { "epoch": 71.80373831775701, "grad_norm": 2.359375, "learning_rate": 3.617445449704964e-05, "loss": 0.1303, "step": 11560 }, { "epoch": 71.80996884735202, "grad_norm": 2.125, "learning_rate": 3.617225049683815e-05, "loss": 0.2559, "step": 11561 }, { "epoch": 71.81619937694704, "grad_norm": 3.09375, "learning_rate": 3.617004638811954e-05, "loss": 0.331, "step": 11562 }, { "epoch": 71.82242990654206, "grad_norm": 1.5703125, "learning_rate": 3.6167842170915214e-05, "loss": 0.1027, "step": 11563 }, { "epoch": 71.82866043613707, "grad_norm": 1.546875, "learning_rate": 3.616563784524656e-05, "loss": 0.1067, "step": 11564 }, { "epoch": 71.83489096573209, "grad_norm": 2.328125, "learning_rate": 3.6163433411135004e-05, "loss": 0.181, "step": 11565 }, { "epoch": 71.8411214953271, "grad_norm": 2.34375, "learning_rate": 3.6161228868601955e-05, "loss": 0.1236, "step": 11566 }, { "epoch": 71.84735202492212, "grad_norm": 3.03125, "learning_rate": 3.6159024217668815e-05, "loss": 0.1649, "step": 11567 }, { "epoch": 71.85358255451713, "grad_norm": 1.6171875, "learning_rate": 3.6156819458356996e-05, "loss": 0.1293, "step": 11568 }, { "epoch": 71.85981308411215, "grad_norm": 2.609375, "learning_rate": 3.615461459068793e-05, "loss": 0.1521, "step": 11569 }, { "epoch": 71.86604361370716, "grad_norm": 2.671875, "learning_rate": 3.615240961468301e-05, "loss": 0.1318, "step": 11570 }, { "epoch": 71.87227414330218, "grad_norm": 2.140625, "learning_rate": 3.615020453036366e-05, "loss": 0.1838, "step": 11571 }, { "epoch": 71.8785046728972, "grad_norm": 4.03125, "learning_rate": 3.614799933775129e-05, "loss": 0.2315, "step": 11572 }, { "epoch": 71.88473520249221, "grad_norm": 2.515625, "learning_rate": 3.614579403686733e-05, "loss": 0.1796, "step": 11573 }, { "epoch": 71.89096573208722, "grad_norm": 1.703125, "learning_rate": 3.614358862773318e-05, "loss": 0.1049, "step": 11574 }, { "epoch": 71.89719626168224, "grad_norm": 2.953125, "learning_rate": 3.6141383110370276e-05, "loss": 0.1687, "step": 11575 }, { "epoch": 71.90342679127725, "grad_norm": 3.0, "learning_rate": 3.613917748480004e-05, "loss": 0.1908, "step": 11576 }, { "epoch": 71.90965732087227, "grad_norm": 2.890625, "learning_rate": 3.613697175104387e-05, "loss": 0.1888, "step": 11577 }, { "epoch": 71.91588785046729, "grad_norm": 2.171875, "learning_rate": 3.6134765909123216e-05, "loss": 0.159, "step": 11578 }, { "epoch": 71.9221183800623, "grad_norm": 2.546875, "learning_rate": 3.613255995905948e-05, "loss": 0.1903, "step": 11579 }, { "epoch": 71.92834890965732, "grad_norm": 1.3125, "learning_rate": 3.61303539008741e-05, "loss": 0.1166, "step": 11580 }, { "epoch": 71.93457943925233, "grad_norm": 1.78125, "learning_rate": 3.61281477345885e-05, "loss": 0.2625, "step": 11581 }, { "epoch": 71.94080996884735, "grad_norm": 2.53125, "learning_rate": 3.612594146022411e-05, "loss": 0.1794, "step": 11582 }, { "epoch": 71.94704049844236, "grad_norm": 2.859375, "learning_rate": 3.612373507780234e-05, "loss": 0.2084, "step": 11583 }, { "epoch": 71.95327102803738, "grad_norm": 2.9375, "learning_rate": 3.612152858734464e-05, "loss": 0.2227, "step": 11584 }, { "epoch": 71.9595015576324, "grad_norm": 0.96484375, "learning_rate": 3.6119321988872425e-05, "loss": 0.1001, "step": 11585 }, { "epoch": 71.96573208722741, "grad_norm": 2.21875, "learning_rate": 3.611711528240714e-05, "loss": 0.1722, "step": 11586 }, { "epoch": 71.97196261682242, "grad_norm": 3.6875, "learning_rate": 3.61149084679702e-05, "loss": 0.384, "step": 11587 }, { "epoch": 71.97819314641744, "grad_norm": 2.875, "learning_rate": 3.6112701545583056e-05, "loss": 0.2326, "step": 11588 }, { "epoch": 71.98442367601245, "grad_norm": 3.03125, "learning_rate": 3.6110494515267126e-05, "loss": 0.1274, "step": 11589 }, { "epoch": 71.99065420560747, "grad_norm": 1.734375, "learning_rate": 3.610828737704386e-05, "loss": 0.1475, "step": 11590 }, { "epoch": 71.99688473520249, "grad_norm": 1.765625, "learning_rate": 3.610608013093468e-05, "loss": 0.1634, "step": 11591 }, { "epoch": 72.0, "grad_norm": 1.0234375, "learning_rate": 3.610387277696104e-05, "loss": 0.0573, "step": 11592 }, { "epoch": 72.00623052959502, "grad_norm": 2.578125, "learning_rate": 3.610166531514436e-05, "loss": 0.1651, "step": 11593 }, { "epoch": 72.01246105919003, "grad_norm": 3.609375, "learning_rate": 3.6099457745506085e-05, "loss": 0.2964, "step": 11594 }, { "epoch": 72.01869158878505, "grad_norm": 2.515625, "learning_rate": 3.609725006806766e-05, "loss": 0.181, "step": 11595 }, { "epoch": 72.02492211838006, "grad_norm": 1.171875, "learning_rate": 3.609504228285052e-05, "loss": 0.1263, "step": 11596 }, { "epoch": 72.03115264797508, "grad_norm": 3.1875, "learning_rate": 3.609283438987612e-05, "loss": 0.2038, "step": 11597 }, { "epoch": 72.03738317757009, "grad_norm": 2.390625, "learning_rate": 3.609062638916589e-05, "loss": 0.149, "step": 11598 }, { "epoch": 72.04361370716511, "grad_norm": 2.9375, "learning_rate": 3.608841828074128e-05, "loss": 0.2269, "step": 11599 }, { "epoch": 72.04984423676012, "grad_norm": 2.9375, "learning_rate": 3.608621006462373e-05, "loss": 0.143, "step": 11600 }, { "epoch": 72.05607476635514, "grad_norm": 1.890625, "learning_rate": 3.6084001740834706e-05, "loss": 0.099, "step": 11601 }, { "epoch": 72.06230529595015, "grad_norm": 1.3203125, "learning_rate": 3.608179330939563e-05, "loss": 0.1159, "step": 11602 }, { "epoch": 72.06853582554517, "grad_norm": 1.9765625, "learning_rate": 3.607958477032796e-05, "loss": 0.1502, "step": 11603 }, { "epoch": 72.07476635514018, "grad_norm": 2.9375, "learning_rate": 3.607737612365316e-05, "loss": 0.1332, "step": 11604 }, { "epoch": 72.0809968847352, "grad_norm": 1.0546875, "learning_rate": 3.607516736939266e-05, "loss": 0.1069, "step": 11605 }, { "epoch": 72.08722741433021, "grad_norm": 1.7734375, "learning_rate": 3.607295850756792e-05, "loss": 0.156, "step": 11606 }, { "epoch": 72.09345794392523, "grad_norm": 1.8203125, "learning_rate": 3.6070749538200385e-05, "loss": 0.2412, "step": 11607 }, { "epoch": 72.09968847352025, "grad_norm": 2.5625, "learning_rate": 3.606854046131154e-05, "loss": 0.2454, "step": 11608 }, { "epoch": 72.10591900311526, "grad_norm": 2.375, "learning_rate": 3.60663312769228e-05, "loss": 0.281, "step": 11609 }, { "epoch": 72.11214953271028, "grad_norm": 1.7109375, "learning_rate": 3.606412198505564e-05, "loss": 0.0904, "step": 11610 }, { "epoch": 72.11838006230529, "grad_norm": 2.546875, "learning_rate": 3.606191258573153e-05, "loss": 0.1847, "step": 11611 }, { "epoch": 72.1246105919003, "grad_norm": 2.140625, "learning_rate": 3.60597030789719e-05, "loss": 0.1682, "step": 11612 }, { "epoch": 72.13084112149532, "grad_norm": 2.0625, "learning_rate": 3.605749346479822e-05, "loss": 0.1561, "step": 11613 }, { "epoch": 72.13707165109034, "grad_norm": 1.6484375, "learning_rate": 3.605528374323197e-05, "loss": 0.0937, "step": 11614 }, { "epoch": 72.14330218068535, "grad_norm": 1.6875, "learning_rate": 3.6053073914294575e-05, "loss": 0.111, "step": 11615 }, { "epoch": 72.14953271028037, "grad_norm": 2.375, "learning_rate": 3.605086397800753e-05, "loss": 0.1943, "step": 11616 }, { "epoch": 72.15576323987538, "grad_norm": 2.015625, "learning_rate": 3.6048653934392284e-05, "loss": 0.1292, "step": 11617 }, { "epoch": 72.1619937694704, "grad_norm": 3.078125, "learning_rate": 3.60464437834703e-05, "loss": 0.1472, "step": 11618 }, { "epoch": 72.16822429906541, "grad_norm": 3.125, "learning_rate": 3.604423352526304e-05, "loss": 0.1444, "step": 11619 }, { "epoch": 72.17445482866043, "grad_norm": 3.109375, "learning_rate": 3.6042023159791974e-05, "loss": 0.3075, "step": 11620 }, { "epoch": 72.18068535825545, "grad_norm": 1.7578125, "learning_rate": 3.603981268707859e-05, "loss": 0.1536, "step": 11621 }, { "epoch": 72.18691588785046, "grad_norm": 1.9453125, "learning_rate": 3.603760210714432e-05, "loss": 0.1626, "step": 11622 }, { "epoch": 72.19314641744548, "grad_norm": 1.9140625, "learning_rate": 3.6035391420010665e-05, "loss": 0.1157, "step": 11623 }, { "epoch": 72.19937694704049, "grad_norm": 3.265625, "learning_rate": 3.603318062569907e-05, "loss": 0.1542, "step": 11624 }, { "epoch": 72.20560747663552, "grad_norm": 1.5546875, "learning_rate": 3.603096972423103e-05, "loss": 0.1566, "step": 11625 }, { "epoch": 72.21183800623054, "grad_norm": 1.84375, "learning_rate": 3.6028758715628005e-05, "loss": 0.1326, "step": 11626 }, { "epoch": 72.21806853582555, "grad_norm": 2.546875, "learning_rate": 3.6026547599911464e-05, "loss": 0.2957, "step": 11627 }, { "epoch": 72.22429906542057, "grad_norm": 1.8671875, "learning_rate": 3.6024336377102895e-05, "loss": 0.1566, "step": 11628 }, { "epoch": 72.23052959501558, "grad_norm": 1.8515625, "learning_rate": 3.6022125047223764e-05, "loss": 0.1394, "step": 11629 }, { "epoch": 72.2367601246106, "grad_norm": 3.109375, "learning_rate": 3.601991361029556e-05, "loss": 0.1138, "step": 11630 }, { "epoch": 72.24299065420561, "grad_norm": 2.546875, "learning_rate": 3.601770206633974e-05, "loss": 0.1619, "step": 11631 }, { "epoch": 72.24922118380063, "grad_norm": 2.75, "learning_rate": 3.60154904153778e-05, "loss": 0.1775, "step": 11632 }, { "epoch": 72.25545171339564, "grad_norm": 1.78125, "learning_rate": 3.601327865743122e-05, "loss": 0.1208, "step": 11633 }, { "epoch": 72.26168224299066, "grad_norm": 3.34375, "learning_rate": 3.601106679252147e-05, "loss": 0.2367, "step": 11634 }, { "epoch": 72.26791277258567, "grad_norm": 1.9921875, "learning_rate": 3.600885482067004e-05, "loss": 0.138, "step": 11635 }, { "epoch": 72.27414330218069, "grad_norm": 1.9140625, "learning_rate": 3.600664274189841e-05, "loss": 0.1568, "step": 11636 }, { "epoch": 72.2803738317757, "grad_norm": 2.796875, "learning_rate": 3.6004430556228063e-05, "loss": 0.1563, "step": 11637 }, { "epoch": 72.28660436137072, "grad_norm": 2.15625, "learning_rate": 3.600221826368049e-05, "loss": 0.1837, "step": 11638 }, { "epoch": 72.29283489096574, "grad_norm": 1.84375, "learning_rate": 3.600000586427718e-05, "loss": 0.1061, "step": 11639 }, { "epoch": 72.29906542056075, "grad_norm": 3.234375, "learning_rate": 3.5997793358039604e-05, "loss": 0.2319, "step": 11640 }, { "epoch": 72.30529595015577, "grad_norm": 1.1796875, "learning_rate": 3.5995580744989265e-05, "loss": 0.1148, "step": 11641 }, { "epoch": 72.31152647975078, "grad_norm": 2.453125, "learning_rate": 3.599336802514764e-05, "loss": 0.1546, "step": 11642 }, { "epoch": 72.3177570093458, "grad_norm": 2.90625, "learning_rate": 3.599115519853625e-05, "loss": 0.19, "step": 11643 }, { "epoch": 72.32398753894081, "grad_norm": 2.859375, "learning_rate": 3.598894226517654e-05, "loss": 0.169, "step": 11644 }, { "epoch": 72.33021806853583, "grad_norm": 1.4609375, "learning_rate": 3.598672922509004e-05, "loss": 0.1276, "step": 11645 }, { "epoch": 72.33644859813084, "grad_norm": 2.171875, "learning_rate": 3.5984516078298226e-05, "loss": 0.1648, "step": 11646 }, { "epoch": 72.34267912772586, "grad_norm": 3.3125, "learning_rate": 3.5982302824822586e-05, "loss": 0.1762, "step": 11647 }, { "epoch": 72.34890965732087, "grad_norm": 2.640625, "learning_rate": 3.5980089464684634e-05, "loss": 0.1524, "step": 11648 }, { "epoch": 72.35514018691589, "grad_norm": 1.65625, "learning_rate": 3.5977875997905864e-05, "loss": 0.1023, "step": 11649 }, { "epoch": 72.3613707165109, "grad_norm": 2.375, "learning_rate": 3.5975662424507764e-05, "loss": 0.2093, "step": 11650 }, { "epoch": 72.36760124610592, "grad_norm": 1.953125, "learning_rate": 3.597344874451183e-05, "loss": 0.2219, "step": 11651 }, { "epoch": 72.37383177570094, "grad_norm": 0.98828125, "learning_rate": 3.597123495793957e-05, "loss": 0.1025, "step": 11652 }, { "epoch": 72.38006230529595, "grad_norm": 3.078125, "learning_rate": 3.596902106481249e-05, "loss": 0.2244, "step": 11653 }, { "epoch": 72.38629283489097, "grad_norm": 2.953125, "learning_rate": 3.596680706515208e-05, "loss": 0.2239, "step": 11654 }, { "epoch": 72.39252336448598, "grad_norm": 1.1484375, "learning_rate": 3.5964592958979846e-05, "loss": 0.1001, "step": 11655 }, { "epoch": 72.398753894081, "grad_norm": 2.25, "learning_rate": 3.59623787463173e-05, "loss": 0.1995, "step": 11656 }, { "epoch": 72.40498442367601, "grad_norm": 2.78125, "learning_rate": 3.596016442718593e-05, "loss": 0.1748, "step": 11657 }, { "epoch": 72.41121495327103, "grad_norm": 2.328125, "learning_rate": 3.595795000160726e-05, "loss": 0.2483, "step": 11658 }, { "epoch": 72.41744548286604, "grad_norm": 1.3984375, "learning_rate": 3.595573546960279e-05, "loss": 0.1046, "step": 11659 }, { "epoch": 72.42367601246106, "grad_norm": 0.8515625, "learning_rate": 3.595352083119403e-05, "loss": 0.0798, "step": 11660 }, { "epoch": 72.42990654205607, "grad_norm": 2.140625, "learning_rate": 3.595130608640247e-05, "loss": 0.1392, "step": 11661 }, { "epoch": 72.43613707165109, "grad_norm": 1.2890625, "learning_rate": 3.594909123524965e-05, "loss": 0.0957, "step": 11662 }, { "epoch": 72.4423676012461, "grad_norm": 2.1875, "learning_rate": 3.5946876277757066e-05, "loss": 0.1678, "step": 11663 }, { "epoch": 72.44859813084112, "grad_norm": 2.359375, "learning_rate": 3.5944661213946226e-05, "loss": 0.1856, "step": 11664 }, { "epoch": 72.45482866043614, "grad_norm": 1.3359375, "learning_rate": 3.5942446043838653e-05, "loss": 0.1134, "step": 11665 }, { "epoch": 72.46105919003115, "grad_norm": 2.640625, "learning_rate": 3.5940230767455856e-05, "loss": 0.1395, "step": 11666 }, { "epoch": 72.46728971962617, "grad_norm": 1.8984375, "learning_rate": 3.593801538481935e-05, "loss": 0.1061, "step": 11667 }, { "epoch": 72.47352024922118, "grad_norm": 1.5234375, "learning_rate": 3.593579989595066e-05, "loss": 0.1391, "step": 11668 }, { "epoch": 72.4797507788162, "grad_norm": 3.0, "learning_rate": 3.593358430087128e-05, "loss": 0.1776, "step": 11669 }, { "epoch": 72.48598130841121, "grad_norm": 2.671875, "learning_rate": 3.593136859960275e-05, "loss": 0.2003, "step": 11670 }, { "epoch": 72.49221183800623, "grad_norm": 4.46875, "learning_rate": 3.592915279216659e-05, "loss": 0.3466, "step": 11671 }, { "epoch": 72.49844236760124, "grad_norm": 1.7265625, "learning_rate": 3.59269368785843e-05, "loss": 0.1, "step": 11672 }, { "epoch": 72.50467289719626, "grad_norm": 2.046875, "learning_rate": 3.592472085887742e-05, "loss": 0.1583, "step": 11673 }, { "epoch": 72.51090342679127, "grad_norm": 4.0625, "learning_rate": 3.5922504733067474e-05, "loss": 0.2344, "step": 11674 }, { "epoch": 72.51713395638629, "grad_norm": 4.34375, "learning_rate": 3.592028850117598e-05, "loss": 0.1564, "step": 11675 }, { "epoch": 72.5233644859813, "grad_norm": 5.03125, "learning_rate": 3.5918072163224456e-05, "loss": 0.3307, "step": 11676 }, { "epoch": 72.52959501557632, "grad_norm": 2.9375, "learning_rate": 3.591585571923443e-05, "loss": 0.1469, "step": 11677 }, { "epoch": 72.53582554517133, "grad_norm": 2.15625, "learning_rate": 3.591363916922744e-05, "loss": 0.1265, "step": 11678 }, { "epoch": 72.54205607476635, "grad_norm": 2.234375, "learning_rate": 3.591142251322499e-05, "loss": 0.2843, "step": 11679 }, { "epoch": 72.54828660436137, "grad_norm": 2.40625, "learning_rate": 3.590920575124863e-05, "loss": 0.1329, "step": 11680 }, { "epoch": 72.55451713395638, "grad_norm": 2.734375, "learning_rate": 3.5906988883319894e-05, "loss": 0.2125, "step": 11681 }, { "epoch": 72.5607476635514, "grad_norm": 2.484375, "learning_rate": 3.590477190946029e-05, "loss": 0.1511, "step": 11682 }, { "epoch": 72.56697819314641, "grad_norm": 3.546875, "learning_rate": 3.590255482969136e-05, "loss": 0.1692, "step": 11683 }, { "epoch": 72.57320872274143, "grad_norm": 2.828125, "learning_rate": 3.5900337644034644e-05, "loss": 0.1327, "step": 11684 }, { "epoch": 72.57943925233644, "grad_norm": 2.796875, "learning_rate": 3.589812035251167e-05, "loss": 0.2739, "step": 11685 }, { "epoch": 72.58566978193146, "grad_norm": 3.34375, "learning_rate": 3.5895902955143965e-05, "loss": 0.2339, "step": 11686 }, { "epoch": 72.59190031152647, "grad_norm": 2.53125, "learning_rate": 3.589368545195308e-05, "loss": 0.1694, "step": 11687 }, { "epoch": 72.59813084112149, "grad_norm": 2.578125, "learning_rate": 3.5891467842960545e-05, "loss": 0.2115, "step": 11688 }, { "epoch": 72.60436137071652, "grad_norm": 1.9609375, "learning_rate": 3.588925012818789e-05, "loss": 0.1563, "step": 11689 }, { "epoch": 72.61059190031153, "grad_norm": 3.859375, "learning_rate": 3.588703230765666e-05, "loss": 0.29, "step": 11690 }, { "epoch": 72.61682242990655, "grad_norm": 3.703125, "learning_rate": 3.5884814381388406e-05, "loss": 0.3233, "step": 11691 }, { "epoch": 72.62305295950156, "grad_norm": 2.265625, "learning_rate": 3.588259634940466e-05, "loss": 0.1326, "step": 11692 }, { "epoch": 72.62928348909658, "grad_norm": 3.0625, "learning_rate": 3.5880378211726954e-05, "loss": 0.1702, "step": 11693 }, { "epoch": 72.6355140186916, "grad_norm": 1.734375, "learning_rate": 3.587815996837685e-05, "loss": 0.1153, "step": 11694 }, { "epoch": 72.64174454828661, "grad_norm": 2.6875, "learning_rate": 3.587594161937587e-05, "loss": 0.1991, "step": 11695 }, { "epoch": 72.64797507788163, "grad_norm": 2.9375, "learning_rate": 3.587372316474557e-05, "loss": 0.3098, "step": 11696 }, { "epoch": 72.65420560747664, "grad_norm": 1.7734375, "learning_rate": 3.5871504604507514e-05, "loss": 0.1456, "step": 11697 }, { "epoch": 72.66043613707166, "grad_norm": 2.609375, "learning_rate": 3.586928593868321e-05, "loss": 0.123, "step": 11698 }, { "epoch": 72.66666666666667, "grad_norm": 3.21875, "learning_rate": 3.586706716729424e-05, "loss": 0.2126, "step": 11699 }, { "epoch": 72.67289719626169, "grad_norm": 2.734375, "learning_rate": 3.586484829036214e-05, "loss": 0.2062, "step": 11700 }, { "epoch": 72.6791277258567, "grad_norm": 3.0625, "learning_rate": 3.586262930790847e-05, "loss": 0.2063, "step": 11701 }, { "epoch": 72.68535825545172, "grad_norm": 3.59375, "learning_rate": 3.5860410219954765e-05, "loss": 0.285, "step": 11702 }, { "epoch": 72.69158878504673, "grad_norm": 2.8125, "learning_rate": 3.5858191026522584e-05, "loss": 0.1955, "step": 11703 }, { "epoch": 72.69781931464175, "grad_norm": 1.984375, "learning_rate": 3.585597172763348e-05, "loss": 0.1821, "step": 11704 }, { "epoch": 72.70404984423676, "grad_norm": 2.140625, "learning_rate": 3.585375232330901e-05, "loss": 0.1655, "step": 11705 }, { "epoch": 72.71028037383178, "grad_norm": 2.90625, "learning_rate": 3.585153281357073e-05, "loss": 0.2318, "step": 11706 }, { "epoch": 72.7165109034268, "grad_norm": 1.7265625, "learning_rate": 3.5849313198440193e-05, "loss": 0.1362, "step": 11707 }, { "epoch": 72.72274143302181, "grad_norm": 1.40625, "learning_rate": 3.5847093477938956e-05, "loss": 0.1058, "step": 11708 }, { "epoch": 72.72897196261682, "grad_norm": 2.5625, "learning_rate": 3.584487365208857e-05, "loss": 0.2501, "step": 11709 }, { "epoch": 72.73520249221184, "grad_norm": 2.609375, "learning_rate": 3.584265372091062e-05, "loss": 0.2066, "step": 11710 }, { "epoch": 72.74143302180686, "grad_norm": 2.953125, "learning_rate": 3.5840433684426643e-05, "loss": 0.2214, "step": 11711 }, { "epoch": 72.74766355140187, "grad_norm": 2.484375, "learning_rate": 3.5838213542658204e-05, "loss": 0.1668, "step": 11712 }, { "epoch": 72.75389408099689, "grad_norm": 2.171875, "learning_rate": 3.583599329562688e-05, "loss": 0.0993, "step": 11713 }, { "epoch": 72.7601246105919, "grad_norm": 2.3125, "learning_rate": 3.583377294335421e-05, "loss": 0.1322, "step": 11714 }, { "epoch": 72.76635514018692, "grad_norm": 3.875, "learning_rate": 3.583155248586178e-05, "loss": 0.289, "step": 11715 }, { "epoch": 72.77258566978193, "grad_norm": 2.40625, "learning_rate": 3.582933192317114e-05, "loss": 0.1882, "step": 11716 }, { "epoch": 72.77881619937695, "grad_norm": 2.09375, "learning_rate": 3.5827111255303867e-05, "loss": 0.252, "step": 11717 }, { "epoch": 72.78504672897196, "grad_norm": 2.328125, "learning_rate": 3.582489048228153e-05, "loss": 0.1777, "step": 11718 }, { "epoch": 72.79127725856698, "grad_norm": 1.6640625, "learning_rate": 3.5822669604125684e-05, "loss": 0.1238, "step": 11719 }, { "epoch": 72.797507788162, "grad_norm": 2.875, "learning_rate": 3.582044862085791e-05, "loss": 0.2153, "step": 11720 }, { "epoch": 72.80373831775701, "grad_norm": 2.921875, "learning_rate": 3.5818227532499774e-05, "loss": 0.153, "step": 11721 }, { "epoch": 72.80996884735202, "grad_norm": 1.3203125, "learning_rate": 3.581600633907286e-05, "loss": 0.1144, "step": 11722 }, { "epoch": 72.81619937694704, "grad_norm": 2.890625, "learning_rate": 3.581378504059872e-05, "loss": 0.2277, "step": 11723 }, { "epoch": 72.82242990654206, "grad_norm": 3.015625, "learning_rate": 3.581156363709894e-05, "loss": 0.1977, "step": 11724 }, { "epoch": 72.82866043613707, "grad_norm": 2.578125, "learning_rate": 3.580934212859509e-05, "loss": 0.2905, "step": 11725 }, { "epoch": 72.83489096573209, "grad_norm": 2.859375, "learning_rate": 3.580712051510876e-05, "loss": 0.2345, "step": 11726 }, { "epoch": 72.8411214953271, "grad_norm": 2.25, "learning_rate": 3.5804898796661505e-05, "loss": 0.1764, "step": 11727 }, { "epoch": 72.84735202492212, "grad_norm": 1.9296875, "learning_rate": 3.580267697327492e-05, "loss": 0.126, "step": 11728 }, { "epoch": 72.85358255451713, "grad_norm": 1.6640625, "learning_rate": 3.580045504497057e-05, "loss": 0.1248, "step": 11729 }, { "epoch": 72.85981308411215, "grad_norm": 3.453125, "learning_rate": 3.579823301177005e-05, "loss": 0.3768, "step": 11730 }, { "epoch": 72.86604361370716, "grad_norm": 2.03125, "learning_rate": 3.579601087369492e-05, "loss": 0.1771, "step": 11731 }, { "epoch": 72.87227414330218, "grad_norm": 1.7890625, "learning_rate": 3.5793788630766784e-05, "loss": 0.1677, "step": 11732 }, { "epoch": 72.8785046728972, "grad_norm": 2.0625, "learning_rate": 3.579156628300721e-05, "loss": 0.1692, "step": 11733 }, { "epoch": 72.88473520249221, "grad_norm": 2.703125, "learning_rate": 3.578934383043779e-05, "loss": 0.1293, "step": 11734 }, { "epoch": 72.89096573208722, "grad_norm": 2.046875, "learning_rate": 3.57871212730801e-05, "loss": 0.2455, "step": 11735 }, { "epoch": 72.89719626168224, "grad_norm": 1.3671875, "learning_rate": 3.578489861095574e-05, "loss": 0.1171, "step": 11736 }, { "epoch": 72.90342679127725, "grad_norm": 2.09375, "learning_rate": 3.5782675844086286e-05, "loss": 0.2404, "step": 11737 }, { "epoch": 72.90965732087227, "grad_norm": 1.671875, "learning_rate": 3.5780452972493336e-05, "loss": 0.1257, "step": 11738 }, { "epoch": 72.91588785046729, "grad_norm": 3.03125, "learning_rate": 3.5778229996198466e-05, "loss": 0.3145, "step": 11739 }, { "epoch": 72.9221183800623, "grad_norm": 2.15625, "learning_rate": 3.577600691522327e-05, "loss": 0.1339, "step": 11740 }, { "epoch": 72.92834890965732, "grad_norm": 2.78125, "learning_rate": 3.5773783729589335e-05, "loss": 0.192, "step": 11741 }, { "epoch": 72.93457943925233, "grad_norm": 1.4375, "learning_rate": 3.5771560439318265e-05, "loss": 0.141, "step": 11742 }, { "epoch": 72.94080996884735, "grad_norm": 1.75, "learning_rate": 3.576933704443165e-05, "loss": 0.132, "step": 11743 }, { "epoch": 72.94704049844236, "grad_norm": 1.9609375, "learning_rate": 3.576711354495107e-05, "loss": 0.1324, "step": 11744 }, { "epoch": 72.95327102803738, "grad_norm": 2.40625, "learning_rate": 3.576488994089815e-05, "loss": 0.1593, "step": 11745 }, { "epoch": 72.9595015576324, "grad_norm": 1.7265625, "learning_rate": 3.5762666232294445e-05, "loss": 0.1099, "step": 11746 }, { "epoch": 72.96573208722741, "grad_norm": 2.9375, "learning_rate": 3.576044241916159e-05, "loss": 0.227, "step": 11747 }, { "epoch": 72.97196261682242, "grad_norm": 3.515625, "learning_rate": 3.5758218501521156e-05, "loss": 0.18, "step": 11748 }, { "epoch": 72.97819314641744, "grad_norm": 2.96875, "learning_rate": 3.575599447939476e-05, "loss": 0.1241, "step": 11749 }, { "epoch": 72.98442367601245, "grad_norm": 2.796875, "learning_rate": 3.5753770352804e-05, "loss": 0.2131, "step": 11750 }, { "epoch": 72.99065420560747, "grad_norm": 2.515625, "learning_rate": 3.575154612177047e-05, "loss": 0.1676, "step": 11751 }, { "epoch": 72.99688473520249, "grad_norm": 1.6875, "learning_rate": 3.574932178631577e-05, "loss": 0.1262, "step": 11752 }, { "epoch": 73.0, "grad_norm": 1.2578125, "learning_rate": 3.574709734646151e-05, "loss": 0.055, "step": 11753 }, { "epoch": 73.00623052959502, "grad_norm": 3.125, "learning_rate": 3.5744872802229296e-05, "loss": 0.2333, "step": 11754 }, { "epoch": 73.01246105919003, "grad_norm": 2.359375, "learning_rate": 3.574264815364072e-05, "loss": 0.148, "step": 11755 }, { "epoch": 73.01869158878505, "grad_norm": 3.125, "learning_rate": 3.574042340071741e-05, "loss": 0.2392, "step": 11756 }, { "epoch": 73.02492211838006, "grad_norm": 4.15625, "learning_rate": 3.573819854348095e-05, "loss": 0.1934, "step": 11757 }, { "epoch": 73.03115264797508, "grad_norm": 3.578125, "learning_rate": 3.573597358195297e-05, "loss": 0.1678, "step": 11758 }, { "epoch": 73.03738317757009, "grad_norm": 3.109375, "learning_rate": 3.5733748516155055e-05, "loss": 0.1243, "step": 11759 }, { "epoch": 73.04361370716511, "grad_norm": 2.75, "learning_rate": 3.573152334610884e-05, "loss": 0.143, "step": 11760 }, { "epoch": 73.04984423676012, "grad_norm": 3.78125, "learning_rate": 3.572929807183593e-05, "loss": 0.3102, "step": 11761 }, { "epoch": 73.05607476635514, "grad_norm": 2.59375, "learning_rate": 3.572707269335792e-05, "loss": 0.2324, "step": 11762 }, { "epoch": 73.06230529595015, "grad_norm": 3.21875, "learning_rate": 3.572484721069643e-05, "loss": 0.1492, "step": 11763 }, { "epoch": 73.06853582554517, "grad_norm": 4.25, "learning_rate": 3.57226216238731e-05, "loss": 0.2197, "step": 11764 }, { "epoch": 73.07476635514018, "grad_norm": 2.875, "learning_rate": 3.572039593290951e-05, "loss": 0.1735, "step": 11765 }, { "epoch": 73.0809968847352, "grad_norm": 2.640625, "learning_rate": 3.5718170137827295e-05, "loss": 0.1565, "step": 11766 }, { "epoch": 73.08722741433021, "grad_norm": 1.9375, "learning_rate": 3.571594423864807e-05, "loss": 0.1717, "step": 11767 }, { "epoch": 73.09345794392523, "grad_norm": 2.8125, "learning_rate": 3.571371823539346e-05, "loss": 0.1336, "step": 11768 }, { "epoch": 73.09968847352025, "grad_norm": 4.28125, "learning_rate": 3.5711492128085066e-05, "loss": 0.2388, "step": 11769 }, { "epoch": 73.10591900311526, "grad_norm": 3.578125, "learning_rate": 3.570926591674453e-05, "loss": 0.2474, "step": 11770 }, { "epoch": 73.11214953271028, "grad_norm": 1.1640625, "learning_rate": 3.570703960139345e-05, "loss": 0.0976, "step": 11771 }, { "epoch": 73.11838006230529, "grad_norm": 3.1875, "learning_rate": 3.570481318205347e-05, "loss": 0.1894, "step": 11772 }, { "epoch": 73.1246105919003, "grad_norm": 2.203125, "learning_rate": 3.57025866587462e-05, "loss": 0.1716, "step": 11773 }, { "epoch": 73.13084112149532, "grad_norm": 1.828125, "learning_rate": 3.570036003149327e-05, "loss": 0.1422, "step": 11774 }, { "epoch": 73.13707165109034, "grad_norm": 2.5625, "learning_rate": 3.5698133300316304e-05, "loss": 0.1267, "step": 11775 }, { "epoch": 73.14330218068535, "grad_norm": 2.296875, "learning_rate": 3.569590646523694e-05, "loss": 0.157, "step": 11776 }, { "epoch": 73.14953271028037, "grad_norm": 2.21875, "learning_rate": 3.569367952627677e-05, "loss": 0.1204, "step": 11777 }, { "epoch": 73.15576323987538, "grad_norm": 2.53125, "learning_rate": 3.569145248345747e-05, "loss": 0.2752, "step": 11778 }, { "epoch": 73.1619937694704, "grad_norm": 2.59375, "learning_rate": 3.568922533680063e-05, "loss": 0.0903, "step": 11779 }, { "epoch": 73.16822429906541, "grad_norm": 2.609375, "learning_rate": 3.568699808632791e-05, "loss": 0.2033, "step": 11780 }, { "epoch": 73.17445482866043, "grad_norm": 1.8515625, "learning_rate": 3.5684770732060915e-05, "loss": 0.1684, "step": 11781 }, { "epoch": 73.18068535825545, "grad_norm": 1.40625, "learning_rate": 3.5682543274021305e-05, "loss": 0.1507, "step": 11782 }, { "epoch": 73.18691588785046, "grad_norm": 3.125, "learning_rate": 3.5680315712230695e-05, "loss": 0.2094, "step": 11783 }, { "epoch": 73.19314641744548, "grad_norm": 1.7421875, "learning_rate": 3.567808804671072e-05, "loss": 0.1725, "step": 11784 }, { "epoch": 73.19937694704049, "grad_norm": 2.765625, "learning_rate": 3.567586027748302e-05, "loss": 0.4228, "step": 11785 }, { "epoch": 73.20560747663552, "grad_norm": 3.03125, "learning_rate": 3.567363240456923e-05, "loss": 0.2431, "step": 11786 }, { "epoch": 73.21183800623054, "grad_norm": 2.171875, "learning_rate": 3.567140442799099e-05, "loss": 0.2126, "step": 11787 }, { "epoch": 73.21806853582555, "grad_norm": 3.328125, "learning_rate": 3.566917634776994e-05, "loss": 0.2642, "step": 11788 }, { "epoch": 73.22429906542057, "grad_norm": 1.234375, "learning_rate": 3.5666948163927716e-05, "loss": 0.1192, "step": 11789 }, { "epoch": 73.23052959501558, "grad_norm": 2.09375, "learning_rate": 3.566471987648595e-05, "loss": 0.1695, "step": 11790 }, { "epoch": 73.2367601246106, "grad_norm": 1.5546875, "learning_rate": 3.56624914854663e-05, "loss": 0.1734, "step": 11791 }, { "epoch": 73.24299065420561, "grad_norm": 2.671875, "learning_rate": 3.566026299089041e-05, "loss": 0.1784, "step": 11792 }, { "epoch": 73.24922118380063, "grad_norm": 2.078125, "learning_rate": 3.565803439277991e-05, "loss": 0.2191, "step": 11793 }, { "epoch": 73.25545171339564, "grad_norm": 3.015625, "learning_rate": 3.5655805691156434e-05, "loss": 0.3843, "step": 11794 }, { "epoch": 73.26168224299066, "grad_norm": 4.21875, "learning_rate": 3.5653576886041655e-05, "loss": 0.385, "step": 11795 }, { "epoch": 73.26791277258567, "grad_norm": 2.6875, "learning_rate": 3.5651347977457214e-05, "loss": 0.1808, "step": 11796 }, { "epoch": 73.27414330218069, "grad_norm": 2.90625, "learning_rate": 3.564911896542474e-05, "loss": 0.2381, "step": 11797 }, { "epoch": 73.2803738317757, "grad_norm": 2.515625, "learning_rate": 3.5646889849965906e-05, "loss": 0.1401, "step": 11798 }, { "epoch": 73.28660436137072, "grad_norm": 2.390625, "learning_rate": 3.564466063110234e-05, "loss": 0.1186, "step": 11799 }, { "epoch": 73.29283489096574, "grad_norm": 2.03125, "learning_rate": 3.5642431308855705e-05, "loss": 0.159, "step": 11800 }, { "epoch": 73.29906542056075, "grad_norm": 2.578125, "learning_rate": 3.5640201883247645e-05, "loss": 0.1617, "step": 11801 }, { "epoch": 73.30529595015577, "grad_norm": 2.1875, "learning_rate": 3.563797235429983e-05, "loss": 0.1209, "step": 11802 }, { "epoch": 73.31152647975078, "grad_norm": 2.328125, "learning_rate": 3.5635742722033886e-05, "loss": 0.2045, "step": 11803 }, { "epoch": 73.3177570093458, "grad_norm": 4.28125, "learning_rate": 3.5633512986471494e-05, "loss": 0.2306, "step": 11804 }, { "epoch": 73.32398753894081, "grad_norm": 3.390625, "learning_rate": 3.563128314763429e-05, "loss": 0.1655, "step": 11805 }, { "epoch": 73.33021806853583, "grad_norm": 4.875, "learning_rate": 3.562905320554395e-05, "loss": 0.2888, "step": 11806 }, { "epoch": 73.33644859813084, "grad_norm": 1.890625, "learning_rate": 3.562682316022211e-05, "loss": 0.1476, "step": 11807 }, { "epoch": 73.34267912772586, "grad_norm": 1.890625, "learning_rate": 3.562459301169044e-05, "loss": 0.1652, "step": 11808 }, { "epoch": 73.34890965732087, "grad_norm": 2.6875, "learning_rate": 3.562236275997061e-05, "loss": 0.2354, "step": 11809 }, { "epoch": 73.35514018691589, "grad_norm": 2.0625, "learning_rate": 3.562013240508425e-05, "loss": 0.1265, "step": 11810 }, { "epoch": 73.3613707165109, "grad_norm": 1.96875, "learning_rate": 3.5617901947053064e-05, "loss": 0.1241, "step": 11811 }, { "epoch": 73.36760124610592, "grad_norm": 1.9453125, "learning_rate": 3.5615671385898676e-05, "loss": 0.1525, "step": 11812 }, { "epoch": 73.37383177570094, "grad_norm": 2.578125, "learning_rate": 3.5613440721642774e-05, "loss": 0.1767, "step": 11813 }, { "epoch": 73.38006230529595, "grad_norm": 2.53125, "learning_rate": 3.5611209954307e-05, "loss": 0.1078, "step": 11814 }, { "epoch": 73.38629283489097, "grad_norm": 2.828125, "learning_rate": 3.5608979083913054e-05, "loss": 0.1478, "step": 11815 }, { "epoch": 73.39252336448598, "grad_norm": 3.0625, "learning_rate": 3.560674811048256e-05, "loss": 0.261, "step": 11816 }, { "epoch": 73.398753894081, "grad_norm": 2.84375, "learning_rate": 3.560451703403722e-05, "loss": 0.1633, "step": 11817 }, { "epoch": 73.40498442367601, "grad_norm": 3.15625, "learning_rate": 3.5602285854598695e-05, "loss": 0.2106, "step": 11818 }, { "epoch": 73.41121495327103, "grad_norm": 2.0, "learning_rate": 3.5600054572188644e-05, "loss": 0.1368, "step": 11819 }, { "epoch": 73.41744548286604, "grad_norm": 2.421875, "learning_rate": 3.559782318682875e-05, "loss": 0.1791, "step": 11820 }, { "epoch": 73.42367601246106, "grad_norm": 2.015625, "learning_rate": 3.5595591698540676e-05, "loss": 0.1498, "step": 11821 }, { "epoch": 73.42990654205607, "grad_norm": 2.609375, "learning_rate": 3.55933601073461e-05, "loss": 0.1414, "step": 11822 }, { "epoch": 73.43613707165109, "grad_norm": 1.953125, "learning_rate": 3.5591128413266686e-05, "loss": 0.145, "step": 11823 }, { "epoch": 73.4423676012461, "grad_norm": 1.109375, "learning_rate": 3.558889661632413e-05, "loss": 0.1026, "step": 11824 }, { "epoch": 73.44859813084112, "grad_norm": 2.140625, "learning_rate": 3.558666471654007e-05, "loss": 0.1473, "step": 11825 }, { "epoch": 73.45482866043614, "grad_norm": 2.15625, "learning_rate": 3.558443271393622e-05, "loss": 0.2559, "step": 11826 }, { "epoch": 73.46105919003115, "grad_norm": 2.71875, "learning_rate": 3.5582200608534246e-05, "loss": 0.2234, "step": 11827 }, { "epoch": 73.46728971962617, "grad_norm": 2.578125, "learning_rate": 3.557996840035582e-05, "loss": 0.2319, "step": 11828 }, { "epoch": 73.47352024922118, "grad_norm": 2.140625, "learning_rate": 3.557773608942262e-05, "loss": 0.1496, "step": 11829 }, { "epoch": 73.4797507788162, "grad_norm": 1.84375, "learning_rate": 3.5575503675756346e-05, "loss": 0.2514, "step": 11830 }, { "epoch": 73.48598130841121, "grad_norm": 2.8125, "learning_rate": 3.5573271159378656e-05, "loss": 0.1609, "step": 11831 }, { "epoch": 73.49221183800623, "grad_norm": 1.765625, "learning_rate": 3.5571038540311244e-05, "loss": 0.1253, "step": 11832 }, { "epoch": 73.49844236760124, "grad_norm": 2.578125, "learning_rate": 3.55688058185758e-05, "loss": 0.27, "step": 11833 }, { "epoch": 73.50467289719626, "grad_norm": 1.4140625, "learning_rate": 3.5566572994193994e-05, "loss": 0.1302, "step": 11834 }, { "epoch": 73.51090342679127, "grad_norm": 2.0625, "learning_rate": 3.5564340067187525e-05, "loss": 0.1427, "step": 11835 }, { "epoch": 73.51713395638629, "grad_norm": 2.890625, "learning_rate": 3.556210703757806e-05, "loss": 0.1593, "step": 11836 }, { "epoch": 73.5233644859813, "grad_norm": 4.0, "learning_rate": 3.555987390538732e-05, "loss": 0.2959, "step": 11837 }, { "epoch": 73.52959501557632, "grad_norm": 3.09375, "learning_rate": 3.555764067063696e-05, "loss": 0.2564, "step": 11838 }, { "epoch": 73.53582554517133, "grad_norm": 2.890625, "learning_rate": 3.555540733334869e-05, "loss": 0.2171, "step": 11839 }, { "epoch": 73.54205607476635, "grad_norm": 1.8515625, "learning_rate": 3.555317389354419e-05, "loss": 0.1024, "step": 11840 }, { "epoch": 73.54828660436137, "grad_norm": 1.5390625, "learning_rate": 3.555094035124516e-05, "loss": 0.1273, "step": 11841 }, { "epoch": 73.55451713395638, "grad_norm": 2.09375, "learning_rate": 3.5548706706473275e-05, "loss": 0.1551, "step": 11842 }, { "epoch": 73.5607476635514, "grad_norm": 2.359375, "learning_rate": 3.5546472959250263e-05, "loss": 0.1127, "step": 11843 }, { "epoch": 73.56697819314641, "grad_norm": 3.15625, "learning_rate": 3.554423910959778e-05, "loss": 0.1619, "step": 11844 }, { "epoch": 73.57320872274143, "grad_norm": 3.609375, "learning_rate": 3.554200515753755e-05, "loss": 0.1613, "step": 11845 }, { "epoch": 73.57943925233644, "grad_norm": 2.5625, "learning_rate": 3.553977110309125e-05, "loss": 0.2002, "step": 11846 }, { "epoch": 73.58566978193146, "grad_norm": 2.109375, "learning_rate": 3.553753694628059e-05, "loss": 0.1301, "step": 11847 }, { "epoch": 73.59190031152647, "grad_norm": 1.71875, "learning_rate": 3.553530268712727e-05, "loss": 0.1296, "step": 11848 }, { "epoch": 73.59813084112149, "grad_norm": 2.40625, "learning_rate": 3.5533068325652977e-05, "loss": 0.1174, "step": 11849 }, { "epoch": 73.60436137071652, "grad_norm": 3.671875, "learning_rate": 3.553083386187942e-05, "loss": 0.2855, "step": 11850 }, { "epoch": 73.61059190031153, "grad_norm": 1.2578125, "learning_rate": 3.5528599295828305e-05, "loss": 0.0973, "step": 11851 }, { "epoch": 73.61682242990655, "grad_norm": 3.5, "learning_rate": 3.552636462752132e-05, "loss": 0.1095, "step": 11852 }, { "epoch": 73.62305295950156, "grad_norm": 2.625, "learning_rate": 3.5524129856980184e-05, "loss": 0.1679, "step": 11853 }, { "epoch": 73.62928348909658, "grad_norm": 2.875, "learning_rate": 3.552189498422659e-05, "loss": 0.2552, "step": 11854 }, { "epoch": 73.6355140186916, "grad_norm": 3.015625, "learning_rate": 3.551966000928225e-05, "loss": 0.2956, "step": 11855 }, { "epoch": 73.64174454828661, "grad_norm": 1.6796875, "learning_rate": 3.551742493216887e-05, "loss": 0.1219, "step": 11856 }, { "epoch": 73.64797507788163, "grad_norm": 2.046875, "learning_rate": 3.5515189752908164e-05, "loss": 0.1469, "step": 11857 }, { "epoch": 73.65420560747664, "grad_norm": 2.15625, "learning_rate": 3.551295447152182e-05, "loss": 0.1212, "step": 11858 }, { "epoch": 73.66043613707166, "grad_norm": 1.734375, "learning_rate": 3.5510719088031576e-05, "loss": 0.155, "step": 11859 }, { "epoch": 73.66666666666667, "grad_norm": 3.171875, "learning_rate": 3.550848360245912e-05, "loss": 0.2633, "step": 11860 }, { "epoch": 73.67289719626169, "grad_norm": 3.453125, "learning_rate": 3.550624801482617e-05, "loss": 0.2364, "step": 11861 }, { "epoch": 73.6791277258567, "grad_norm": 1.796875, "learning_rate": 3.550401232515444e-05, "loss": 0.1799, "step": 11862 }, { "epoch": 73.68535825545172, "grad_norm": 2.765625, "learning_rate": 3.550177653346563e-05, "loss": 0.1969, "step": 11863 }, { "epoch": 73.69158878504673, "grad_norm": 2.84375, "learning_rate": 3.549954063978148e-05, "loss": 0.2325, "step": 11864 }, { "epoch": 73.69781931464175, "grad_norm": 2.546875, "learning_rate": 3.549730464412369e-05, "loss": 0.2118, "step": 11865 }, { "epoch": 73.70404984423676, "grad_norm": 1.4453125, "learning_rate": 3.549506854651398e-05, "loss": 0.1231, "step": 11866 }, { "epoch": 73.71028037383178, "grad_norm": 2.640625, "learning_rate": 3.549283234697407e-05, "loss": 0.1681, "step": 11867 }, { "epoch": 73.7165109034268, "grad_norm": 2.15625, "learning_rate": 3.549059604552567e-05, "loss": 0.1272, "step": 11868 }, { "epoch": 73.72274143302181, "grad_norm": 2.046875, "learning_rate": 3.54883596421905e-05, "loss": 0.1763, "step": 11869 }, { "epoch": 73.72897196261682, "grad_norm": 1.6328125, "learning_rate": 3.548612313699029e-05, "loss": 0.106, "step": 11870 }, { "epoch": 73.73520249221184, "grad_norm": 1.890625, "learning_rate": 3.5483886529946756e-05, "loss": 0.2049, "step": 11871 }, { "epoch": 73.74143302180686, "grad_norm": 1.890625, "learning_rate": 3.548164982108162e-05, "loss": 0.1165, "step": 11872 }, { "epoch": 73.74766355140187, "grad_norm": 2.859375, "learning_rate": 3.547941301041661e-05, "loss": 0.2856, "step": 11873 }, { "epoch": 73.75389408099689, "grad_norm": 2.671875, "learning_rate": 3.5477176097973435e-05, "loss": 0.198, "step": 11874 }, { "epoch": 73.7601246105919, "grad_norm": 3.296875, "learning_rate": 3.547493908377384e-05, "loss": 0.2989, "step": 11875 }, { "epoch": 73.76635514018692, "grad_norm": 3.578125, "learning_rate": 3.547270196783954e-05, "loss": 0.1797, "step": 11876 }, { "epoch": 73.77258566978193, "grad_norm": 0.80859375, "learning_rate": 3.547046475019226e-05, "loss": 0.082, "step": 11877 }, { "epoch": 73.77881619937695, "grad_norm": 1.328125, "learning_rate": 3.5468227430853745e-05, "loss": 0.1276, "step": 11878 }, { "epoch": 73.78504672897196, "grad_norm": 3.4375, "learning_rate": 3.546599000984571e-05, "loss": 0.2823, "step": 11879 }, { "epoch": 73.79127725856698, "grad_norm": 2.078125, "learning_rate": 3.5463752487189885e-05, "loss": 0.1142, "step": 11880 }, { "epoch": 73.797507788162, "grad_norm": 1.7421875, "learning_rate": 3.5461514862907996e-05, "loss": 0.1317, "step": 11881 }, { "epoch": 73.80373831775701, "grad_norm": 2.234375, "learning_rate": 3.5459277137021794e-05, "loss": 0.1642, "step": 11882 }, { "epoch": 73.80996884735202, "grad_norm": 2.921875, "learning_rate": 3.5457039309553e-05, "loss": 0.2822, "step": 11883 }, { "epoch": 73.81619937694704, "grad_norm": 2.71875, "learning_rate": 3.5454801380523345e-05, "loss": 0.1538, "step": 11884 }, { "epoch": 73.82242990654206, "grad_norm": 2.390625, "learning_rate": 3.5452563349954575e-05, "loss": 0.1431, "step": 11885 }, { "epoch": 73.82866043613707, "grad_norm": 1.9765625, "learning_rate": 3.545032521786842e-05, "loss": 0.1328, "step": 11886 }, { "epoch": 73.83489096573209, "grad_norm": 2.59375, "learning_rate": 3.544808698428661e-05, "loss": 0.2652, "step": 11887 }, { "epoch": 73.8411214953271, "grad_norm": 1.8671875, "learning_rate": 3.54458486492309e-05, "loss": 0.1899, "step": 11888 }, { "epoch": 73.84735202492212, "grad_norm": 1.84375, "learning_rate": 3.5443610212723015e-05, "loss": 0.1321, "step": 11889 }, { "epoch": 73.85358255451713, "grad_norm": 3.109375, "learning_rate": 3.5441371674784694e-05, "loss": 0.2208, "step": 11890 }, { "epoch": 73.85981308411215, "grad_norm": 3.578125, "learning_rate": 3.54391330354377e-05, "loss": 0.1153, "step": 11891 }, { "epoch": 73.86604361370716, "grad_norm": 2.484375, "learning_rate": 3.543689429470375e-05, "loss": 0.1942, "step": 11892 }, { "epoch": 73.87227414330218, "grad_norm": 1.765625, "learning_rate": 3.543465545260459e-05, "loss": 0.1574, "step": 11893 }, { "epoch": 73.8785046728972, "grad_norm": 2.328125, "learning_rate": 3.543241650916198e-05, "loss": 0.1614, "step": 11894 }, { "epoch": 73.88473520249221, "grad_norm": 2.5, "learning_rate": 3.5430177464397654e-05, "loss": 0.1397, "step": 11895 }, { "epoch": 73.89096573208722, "grad_norm": 2.359375, "learning_rate": 3.5427938318333356e-05, "loss": 0.1808, "step": 11896 }, { "epoch": 73.89719626168224, "grad_norm": 2.5625, "learning_rate": 3.542569907099085e-05, "loss": 0.2329, "step": 11897 }, { "epoch": 73.90342679127725, "grad_norm": 2.03125, "learning_rate": 3.542345972239186e-05, "loss": 0.1006, "step": 11898 }, { "epoch": 73.90965732087227, "grad_norm": 2.390625, "learning_rate": 3.542122027255813e-05, "loss": 0.1592, "step": 11899 }, { "epoch": 73.91588785046729, "grad_norm": 3.5625, "learning_rate": 3.5418980721511446e-05, "loss": 0.2463, "step": 11900 }, { "epoch": 73.9221183800623, "grad_norm": 3.40625, "learning_rate": 3.5416741069273534e-05, "loss": 0.1814, "step": 11901 }, { "epoch": 73.92834890965732, "grad_norm": 3.5, "learning_rate": 3.541450131586615e-05, "loss": 0.2289, "step": 11902 }, { "epoch": 73.93457943925233, "grad_norm": 2.59375, "learning_rate": 3.541226146131105e-05, "loss": 0.1312, "step": 11903 }, { "epoch": 73.94080996884735, "grad_norm": 2.65625, "learning_rate": 3.541002150562998e-05, "loss": 0.1973, "step": 11904 }, { "epoch": 73.94704049844236, "grad_norm": 1.7578125, "learning_rate": 3.54077814488447e-05, "loss": 0.158, "step": 11905 }, { "epoch": 73.95327102803738, "grad_norm": 3.0, "learning_rate": 3.540554129097697e-05, "loss": 0.1417, "step": 11906 }, { "epoch": 73.9595015576324, "grad_norm": 2.984375, "learning_rate": 3.540330103204854e-05, "loss": 0.2161, "step": 11907 }, { "epoch": 73.96573208722741, "grad_norm": 2.515625, "learning_rate": 3.540106067208117e-05, "loss": 0.1091, "step": 11908 }, { "epoch": 73.97196261682242, "grad_norm": 2.734375, "learning_rate": 3.539882021109662e-05, "loss": 0.1611, "step": 11909 }, { "epoch": 73.97819314641744, "grad_norm": 2.71875, "learning_rate": 3.5396579649116654e-05, "loss": 0.2148, "step": 11910 }, { "epoch": 73.98442367601245, "grad_norm": 1.84375, "learning_rate": 3.5394338986163027e-05, "loss": 0.1542, "step": 11911 }, { "epoch": 73.99065420560747, "grad_norm": 2.46875, "learning_rate": 3.53920982222575e-05, "loss": 0.1733, "step": 11912 }, { "epoch": 73.99688473520249, "grad_norm": 2.25, "learning_rate": 3.538985735742184e-05, "loss": 0.1421, "step": 11913 }, { "epoch": 74.0, "grad_norm": 2.765625, "learning_rate": 3.53876163916778e-05, "loss": 0.1521, "step": 11914 }, { "epoch": 74.00623052959502, "grad_norm": 2.46875, "learning_rate": 3.5385375325047166e-05, "loss": 0.2094, "step": 11915 }, { "epoch": 74.01246105919003, "grad_norm": 2.15625, "learning_rate": 3.538313415755168e-05, "loss": 0.1649, "step": 11916 }, { "epoch": 74.01869158878505, "grad_norm": 1.7421875, "learning_rate": 3.5380892889213125e-05, "loss": 0.1155, "step": 11917 }, { "epoch": 74.02492211838006, "grad_norm": 2.84375, "learning_rate": 3.5378651520053266e-05, "loss": 0.1444, "step": 11918 }, { "epoch": 74.03115264797508, "grad_norm": 3.125, "learning_rate": 3.5376410050093864e-05, "loss": 0.2257, "step": 11919 }, { "epoch": 74.03738317757009, "grad_norm": 1.8828125, "learning_rate": 3.53741684793567e-05, "loss": 0.1608, "step": 11920 }, { "epoch": 74.04361370716511, "grad_norm": 1.203125, "learning_rate": 3.537192680786352e-05, "loss": 0.1236, "step": 11921 }, { "epoch": 74.04984423676012, "grad_norm": 1.984375, "learning_rate": 3.536968503563613e-05, "loss": 0.1403, "step": 11922 }, { "epoch": 74.05607476635514, "grad_norm": 3.25, "learning_rate": 3.5367443162696287e-05, "loss": 0.1137, "step": 11923 }, { "epoch": 74.06230529595015, "grad_norm": 1.140625, "learning_rate": 3.536520118906576e-05, "loss": 0.1068, "step": 11924 }, { "epoch": 74.06853582554517, "grad_norm": 2.3125, "learning_rate": 3.536295911476631e-05, "loss": 0.2175, "step": 11925 }, { "epoch": 74.07476635514018, "grad_norm": 2.84375, "learning_rate": 3.536071693981976e-05, "loss": 0.1915, "step": 11926 }, { "epoch": 74.0809968847352, "grad_norm": 3.59375, "learning_rate": 3.535847466424783e-05, "loss": 0.249, "step": 11927 }, { "epoch": 74.08722741433021, "grad_norm": 1.6328125, "learning_rate": 3.535623228807234e-05, "loss": 0.0942, "step": 11928 }, { "epoch": 74.09345794392523, "grad_norm": 3.203125, "learning_rate": 3.535398981131505e-05, "loss": 0.2364, "step": 11929 }, { "epoch": 74.09968847352025, "grad_norm": 2.140625, "learning_rate": 3.5351747233997735e-05, "loss": 0.1351, "step": 11930 }, { "epoch": 74.10591900311526, "grad_norm": 3.171875, "learning_rate": 3.534950455614219e-05, "loss": 0.2414, "step": 11931 }, { "epoch": 74.11214953271028, "grad_norm": 2.453125, "learning_rate": 3.534726177777018e-05, "loss": 0.1306, "step": 11932 }, { "epoch": 74.11838006230529, "grad_norm": 2.5625, "learning_rate": 3.53450188989035e-05, "loss": 0.1918, "step": 11933 }, { "epoch": 74.1246105919003, "grad_norm": 2.578125, "learning_rate": 3.534277591956392e-05, "loss": 0.198, "step": 11934 }, { "epoch": 74.13084112149532, "grad_norm": 2.140625, "learning_rate": 3.534053283977324e-05, "loss": 0.1483, "step": 11935 }, { "epoch": 74.13707165109034, "grad_norm": 2.046875, "learning_rate": 3.533828965955324e-05, "loss": 0.1168, "step": 11936 }, { "epoch": 74.14330218068535, "grad_norm": 1.8359375, "learning_rate": 3.53360463789257e-05, "loss": 0.2086, "step": 11937 }, { "epoch": 74.14953271028037, "grad_norm": 1.8671875, "learning_rate": 3.533380299791241e-05, "loss": 0.1245, "step": 11938 }, { "epoch": 74.15576323987538, "grad_norm": 1.5703125, "learning_rate": 3.533155951653516e-05, "loss": 0.1205, "step": 11939 }, { "epoch": 74.1619937694704, "grad_norm": 1.9921875, "learning_rate": 3.5329315934815744e-05, "loss": 0.1481, "step": 11940 }, { "epoch": 74.16822429906541, "grad_norm": 2.171875, "learning_rate": 3.532707225277594e-05, "loss": 0.1827, "step": 11941 }, { "epoch": 74.17445482866043, "grad_norm": 2.453125, "learning_rate": 3.5324828470437546e-05, "loss": 0.2305, "step": 11942 }, { "epoch": 74.18068535825545, "grad_norm": 2.5, "learning_rate": 3.532258458782236e-05, "loss": 0.1301, "step": 11943 }, { "epoch": 74.18691588785046, "grad_norm": 1.5859375, "learning_rate": 3.5320340604952175e-05, "loss": 0.1258, "step": 11944 }, { "epoch": 74.19314641744548, "grad_norm": 2.8125, "learning_rate": 3.5318096521848765e-05, "loss": 0.1888, "step": 11945 }, { "epoch": 74.19937694704049, "grad_norm": 2.125, "learning_rate": 3.531585233853395e-05, "loss": 0.1287, "step": 11946 }, { "epoch": 74.20560747663552, "grad_norm": 2.734375, "learning_rate": 3.53136080550295e-05, "loss": 0.2144, "step": 11947 }, { "epoch": 74.21183800623054, "grad_norm": 2.59375, "learning_rate": 3.5311363671357244e-05, "loss": 0.1183, "step": 11948 }, { "epoch": 74.21806853582555, "grad_norm": 1.9921875, "learning_rate": 3.530911918753895e-05, "loss": 0.1356, "step": 11949 }, { "epoch": 74.22429906542057, "grad_norm": 2.578125, "learning_rate": 3.530687460359644e-05, "loss": 0.159, "step": 11950 }, { "epoch": 74.23052959501558, "grad_norm": 1.9609375, "learning_rate": 3.53046299195515e-05, "loss": 0.148, "step": 11951 }, { "epoch": 74.2367601246106, "grad_norm": 2.6875, "learning_rate": 3.530238513542593e-05, "loss": 0.2197, "step": 11952 }, { "epoch": 74.24299065420561, "grad_norm": 2.046875, "learning_rate": 3.530014025124154e-05, "loss": 0.1351, "step": 11953 }, { "epoch": 74.24922118380063, "grad_norm": 2.6875, "learning_rate": 3.529789526702012e-05, "loss": 0.1376, "step": 11954 }, { "epoch": 74.25545171339564, "grad_norm": 2.53125, "learning_rate": 3.52956501827835e-05, "loss": 0.1696, "step": 11955 }, { "epoch": 74.26168224299066, "grad_norm": 4.25, "learning_rate": 3.529340499855346e-05, "loss": 0.3152, "step": 11956 }, { "epoch": 74.26791277258567, "grad_norm": 2.6875, "learning_rate": 3.529115971435181e-05, "loss": 0.1567, "step": 11957 }, { "epoch": 74.27414330218069, "grad_norm": 1.421875, "learning_rate": 3.5288914330200363e-05, "loss": 0.1268, "step": 11958 }, { "epoch": 74.2803738317757, "grad_norm": 1.5234375, "learning_rate": 3.528666884612092e-05, "loss": 0.1392, "step": 11959 }, { "epoch": 74.28660436137072, "grad_norm": 3.03125, "learning_rate": 3.528442326213529e-05, "loss": 0.1361, "step": 11960 }, { "epoch": 74.29283489096574, "grad_norm": 2.96875, "learning_rate": 3.5282177578265296e-05, "loss": 0.2087, "step": 11961 }, { "epoch": 74.29906542056075, "grad_norm": 2.953125, "learning_rate": 3.5279931794532726e-05, "loss": 0.1513, "step": 11962 }, { "epoch": 74.30529595015577, "grad_norm": 1.890625, "learning_rate": 3.5277685910959406e-05, "loss": 0.1004, "step": 11963 }, { "epoch": 74.31152647975078, "grad_norm": 3.984375, "learning_rate": 3.527543992756716e-05, "loss": 0.2321, "step": 11964 }, { "epoch": 74.3177570093458, "grad_norm": 1.453125, "learning_rate": 3.527319384437777e-05, "loss": 0.1263, "step": 11965 }, { "epoch": 74.32398753894081, "grad_norm": 2.328125, "learning_rate": 3.527094766141308e-05, "loss": 0.1468, "step": 11966 }, { "epoch": 74.33021806853583, "grad_norm": 1.9140625, "learning_rate": 3.5268701378694884e-05, "loss": 0.1289, "step": 11967 }, { "epoch": 74.33644859813084, "grad_norm": 3.484375, "learning_rate": 3.526645499624501e-05, "loss": 0.244, "step": 11968 }, { "epoch": 74.34267912772586, "grad_norm": 2.09375, "learning_rate": 3.526420851408527e-05, "loss": 0.0991, "step": 11969 }, { "epoch": 74.34890965732087, "grad_norm": 3.734375, "learning_rate": 3.526196193223749e-05, "loss": 0.3764, "step": 11970 }, { "epoch": 74.35514018691589, "grad_norm": 1.671875, "learning_rate": 3.5259715250723485e-05, "loss": 0.1033, "step": 11971 }, { "epoch": 74.3613707165109, "grad_norm": 2.15625, "learning_rate": 3.525746846956507e-05, "loss": 0.1555, "step": 11972 }, { "epoch": 74.36760124610592, "grad_norm": 2.453125, "learning_rate": 3.525522158878407e-05, "loss": 0.2142, "step": 11973 }, { "epoch": 74.37383177570094, "grad_norm": 2.34375, "learning_rate": 3.5252974608402324e-05, "loss": 0.1218, "step": 11974 }, { "epoch": 74.38006230529595, "grad_norm": 3.09375, "learning_rate": 3.525072752844162e-05, "loss": 0.1222, "step": 11975 }, { "epoch": 74.38629283489097, "grad_norm": 2.734375, "learning_rate": 3.524848034892381e-05, "loss": 0.1302, "step": 11976 }, { "epoch": 74.39252336448598, "grad_norm": 2.125, "learning_rate": 3.52462330698707e-05, "loss": 0.1199, "step": 11977 }, { "epoch": 74.398753894081, "grad_norm": 2.359375, "learning_rate": 3.5243985691304147e-05, "loss": 0.1036, "step": 11978 }, { "epoch": 74.40498442367601, "grad_norm": 1.359375, "learning_rate": 3.524173821324594e-05, "loss": 0.1122, "step": 11979 }, { "epoch": 74.41121495327103, "grad_norm": 2.015625, "learning_rate": 3.523949063571793e-05, "loss": 0.1294, "step": 11980 }, { "epoch": 74.41744548286604, "grad_norm": 2.359375, "learning_rate": 3.523724295874194e-05, "loss": 0.1656, "step": 11981 }, { "epoch": 74.42367601246106, "grad_norm": 4.125, "learning_rate": 3.523499518233979e-05, "loss": 0.33, "step": 11982 }, { "epoch": 74.42990654205607, "grad_norm": 1.5859375, "learning_rate": 3.5232747306533344e-05, "loss": 0.1185, "step": 11983 }, { "epoch": 74.43613707165109, "grad_norm": 2.78125, "learning_rate": 3.523049933134439e-05, "loss": 0.1634, "step": 11984 }, { "epoch": 74.4423676012461, "grad_norm": 1.203125, "learning_rate": 3.5228251256794784e-05, "loss": 0.1419, "step": 11985 }, { "epoch": 74.44859813084112, "grad_norm": 2.421875, "learning_rate": 3.5226003082906365e-05, "loss": 0.1826, "step": 11986 }, { "epoch": 74.45482866043614, "grad_norm": 1.7734375, "learning_rate": 3.522375480970096e-05, "loss": 0.1202, "step": 11987 }, { "epoch": 74.46105919003115, "grad_norm": 3.125, "learning_rate": 3.522150643720039e-05, "loss": 0.2577, "step": 11988 }, { "epoch": 74.46728971962617, "grad_norm": 2.359375, "learning_rate": 3.521925796542652e-05, "loss": 0.1175, "step": 11989 }, { "epoch": 74.47352024922118, "grad_norm": 1.65625, "learning_rate": 3.5217009394401176e-05, "loss": 0.1328, "step": 11990 }, { "epoch": 74.4797507788162, "grad_norm": 1.75, "learning_rate": 3.521476072414619e-05, "loss": 0.0931, "step": 11991 }, { "epoch": 74.48598130841121, "grad_norm": 2.171875, "learning_rate": 3.52125119546834e-05, "loss": 0.1942, "step": 11992 }, { "epoch": 74.49221183800623, "grad_norm": 1.484375, "learning_rate": 3.5210263086034664e-05, "loss": 0.112, "step": 11993 }, { "epoch": 74.49844236760124, "grad_norm": 2.875, "learning_rate": 3.52080141182218e-05, "loss": 0.2489, "step": 11994 }, { "epoch": 74.50467289719626, "grad_norm": 2.96875, "learning_rate": 3.520576505126667e-05, "loss": 0.2937, "step": 11995 }, { "epoch": 74.51090342679127, "grad_norm": 2.828125, "learning_rate": 3.520351588519111e-05, "loss": 0.3379, "step": 11996 }, { "epoch": 74.51713395638629, "grad_norm": 2.046875, "learning_rate": 3.520126662001696e-05, "loss": 0.1892, "step": 11997 }, { "epoch": 74.5233644859813, "grad_norm": 1.5703125, "learning_rate": 3.519901725576607e-05, "loss": 0.1182, "step": 11998 }, { "epoch": 74.52959501557632, "grad_norm": 2.4375, "learning_rate": 3.5196767792460286e-05, "loss": 0.2021, "step": 11999 }, { "epoch": 74.53582554517133, "grad_norm": 2.421875, "learning_rate": 3.5194518230121456e-05, "loss": 0.1475, "step": 12000 }, { "epoch": 74.54205607476635, "grad_norm": 2.359375, "learning_rate": 3.519226856877143e-05, "loss": 0.1884, "step": 12001 }, { "epoch": 74.54828660436137, "grad_norm": 1.265625, "learning_rate": 3.519001880843205e-05, "loss": 0.1196, "step": 12002 }, { "epoch": 74.55451713395638, "grad_norm": 2.671875, "learning_rate": 3.518776894912517e-05, "loss": 0.1306, "step": 12003 }, { "epoch": 74.5607476635514, "grad_norm": 1.96875, "learning_rate": 3.518551899087264e-05, "loss": 0.1654, "step": 12004 }, { "epoch": 74.56697819314641, "grad_norm": 2.015625, "learning_rate": 3.5183268933696315e-05, "loss": 0.1233, "step": 12005 }, { "epoch": 74.57320872274143, "grad_norm": 1.84375, "learning_rate": 3.518101877761805e-05, "loss": 0.124, "step": 12006 }, { "epoch": 74.57943925233644, "grad_norm": 3.4375, "learning_rate": 3.5178768522659697e-05, "loss": 0.1305, "step": 12007 }, { "epoch": 74.58566978193146, "grad_norm": 1.125, "learning_rate": 3.5176518168843096e-05, "loss": 0.1055, "step": 12008 }, { "epoch": 74.59190031152647, "grad_norm": 1.265625, "learning_rate": 3.517426771619014e-05, "loss": 0.1272, "step": 12009 }, { "epoch": 74.59813084112149, "grad_norm": 2.1875, "learning_rate": 3.517201716472264e-05, "loss": 0.1929, "step": 12010 }, { "epoch": 74.60436137071652, "grad_norm": 2.15625, "learning_rate": 3.5169766514462486e-05, "loss": 0.1698, "step": 12011 }, { "epoch": 74.61059190031153, "grad_norm": 1.7265625, "learning_rate": 3.516751576543153e-05, "loss": 0.1018, "step": 12012 }, { "epoch": 74.61682242990655, "grad_norm": 2.875, "learning_rate": 3.516526491765162e-05, "loss": 0.2385, "step": 12013 }, { "epoch": 74.62305295950156, "grad_norm": 3.109375, "learning_rate": 3.516301397114463e-05, "loss": 0.2865, "step": 12014 }, { "epoch": 74.62928348909658, "grad_norm": 2.375, "learning_rate": 3.516076292593242e-05, "loss": 0.1336, "step": 12015 }, { "epoch": 74.6355140186916, "grad_norm": 2.40625, "learning_rate": 3.515851178203684e-05, "loss": 0.1483, "step": 12016 }, { "epoch": 74.64174454828661, "grad_norm": 2.21875, "learning_rate": 3.515626053947976e-05, "loss": 0.2755, "step": 12017 }, { "epoch": 74.64797507788163, "grad_norm": 2.171875, "learning_rate": 3.515400919828307e-05, "loss": 0.2086, "step": 12018 }, { "epoch": 74.65420560747664, "grad_norm": 2.90625, "learning_rate": 3.515175775846859e-05, "loss": 0.1824, "step": 12019 }, { "epoch": 74.66043613707166, "grad_norm": 3.953125, "learning_rate": 3.5149506220058214e-05, "loss": 0.2008, "step": 12020 }, { "epoch": 74.66666666666667, "grad_norm": 1.6953125, "learning_rate": 3.514725458307381e-05, "loss": 0.1273, "step": 12021 }, { "epoch": 74.67289719626169, "grad_norm": 2.96875, "learning_rate": 3.514500284753724e-05, "loss": 0.2416, "step": 12022 }, { "epoch": 74.6791277258567, "grad_norm": 3.546875, "learning_rate": 3.514275101347037e-05, "loss": 0.1922, "step": 12023 }, { "epoch": 74.68535825545172, "grad_norm": 3.53125, "learning_rate": 3.514049908089508e-05, "loss": 0.1048, "step": 12024 }, { "epoch": 74.69158878504673, "grad_norm": 2.578125, "learning_rate": 3.5138247049833236e-05, "loss": 0.186, "step": 12025 }, { "epoch": 74.69781931464175, "grad_norm": 2.71875, "learning_rate": 3.5135994920306705e-05, "loss": 0.1774, "step": 12026 }, { "epoch": 74.70404984423676, "grad_norm": 1.8203125, "learning_rate": 3.513374269233737e-05, "loss": 0.1288, "step": 12027 }, { "epoch": 74.71028037383178, "grad_norm": 2.53125, "learning_rate": 3.51314903659471e-05, "loss": 0.2005, "step": 12028 }, { "epoch": 74.7165109034268, "grad_norm": 2.484375, "learning_rate": 3.5129237941157766e-05, "loss": 0.2109, "step": 12029 }, { "epoch": 74.72274143302181, "grad_norm": 3.921875, "learning_rate": 3.5126985417991254e-05, "loss": 0.2028, "step": 12030 }, { "epoch": 74.72897196261682, "grad_norm": 4.3125, "learning_rate": 3.512473279646943e-05, "loss": 0.2903, "step": 12031 }, { "epoch": 74.73520249221184, "grad_norm": 2.703125, "learning_rate": 3.5122480076614184e-05, "loss": 0.1988, "step": 12032 }, { "epoch": 74.74143302180686, "grad_norm": 2.96875, "learning_rate": 3.512022725844738e-05, "loss": 0.1166, "step": 12033 }, { "epoch": 74.74766355140187, "grad_norm": 2.671875, "learning_rate": 3.511797434199091e-05, "loss": 0.1937, "step": 12034 }, { "epoch": 74.75389408099689, "grad_norm": 2.21875, "learning_rate": 3.5115721327266653e-05, "loss": 0.1211, "step": 12035 }, { "epoch": 74.7601246105919, "grad_norm": 3.078125, "learning_rate": 3.511346821429648e-05, "loss": 0.1909, "step": 12036 }, { "epoch": 74.76635514018692, "grad_norm": 2.46875, "learning_rate": 3.511121500310229e-05, "loss": 0.1807, "step": 12037 }, { "epoch": 74.77258566978193, "grad_norm": 2.453125, "learning_rate": 3.5108961693705956e-05, "loss": 0.1095, "step": 12038 }, { "epoch": 74.77881619937695, "grad_norm": 2.375, "learning_rate": 3.5106708286129365e-05, "loss": 0.1085, "step": 12039 }, { "epoch": 74.78504672897196, "grad_norm": 1.84375, "learning_rate": 3.510445478039441e-05, "loss": 0.1748, "step": 12040 }, { "epoch": 74.79127725856698, "grad_norm": 1.2109375, "learning_rate": 3.510220117652297e-05, "loss": 0.1108, "step": 12041 }, { "epoch": 74.797507788162, "grad_norm": 2.21875, "learning_rate": 3.509994747453692e-05, "loss": 0.1223, "step": 12042 }, { "epoch": 74.80373831775701, "grad_norm": 2.90625, "learning_rate": 3.509769367445816e-05, "loss": 0.1055, "step": 12043 }, { "epoch": 74.80996884735202, "grad_norm": 3.140625, "learning_rate": 3.50954397763086e-05, "loss": 0.2381, "step": 12044 }, { "epoch": 74.81619937694704, "grad_norm": 2.34375, "learning_rate": 3.509318578011009e-05, "loss": 0.1753, "step": 12045 }, { "epoch": 74.82242990654206, "grad_norm": 1.203125, "learning_rate": 3.5090931685884554e-05, "loss": 0.1309, "step": 12046 }, { "epoch": 74.82866043613707, "grad_norm": 2.09375, "learning_rate": 3.508867749365387e-05, "loss": 0.1303, "step": 12047 }, { "epoch": 74.83489096573209, "grad_norm": 2.53125, "learning_rate": 3.508642320343993e-05, "loss": 0.1301, "step": 12048 }, { "epoch": 74.8411214953271, "grad_norm": 1.734375, "learning_rate": 3.508416881526464e-05, "loss": 0.1408, "step": 12049 }, { "epoch": 74.84735202492212, "grad_norm": 3.21875, "learning_rate": 3.508191432914988e-05, "loss": 0.1578, "step": 12050 }, { "epoch": 74.85358255451713, "grad_norm": 2.984375, "learning_rate": 3.5079659745117554e-05, "loss": 0.3101, "step": 12051 }, { "epoch": 74.85981308411215, "grad_norm": 1.4921875, "learning_rate": 3.507740506318956e-05, "loss": 0.1128, "step": 12052 }, { "epoch": 74.86604361370716, "grad_norm": 1.5234375, "learning_rate": 3.507515028338779e-05, "loss": 0.1017, "step": 12053 }, { "epoch": 74.87227414330218, "grad_norm": 1.703125, "learning_rate": 3.507289540573415e-05, "loss": 0.1244, "step": 12054 }, { "epoch": 74.8785046728972, "grad_norm": 1.6796875, "learning_rate": 3.507064043025053e-05, "loss": 0.1356, "step": 12055 }, { "epoch": 74.88473520249221, "grad_norm": 3.03125, "learning_rate": 3.506838535695884e-05, "loss": 0.19, "step": 12056 }, { "epoch": 74.89096573208722, "grad_norm": 2.984375, "learning_rate": 3.506613018588098e-05, "loss": 0.2886, "step": 12057 }, { "epoch": 74.89719626168224, "grad_norm": 2.25, "learning_rate": 3.5063874917038855e-05, "loss": 0.1928, "step": 12058 }, { "epoch": 74.90342679127725, "grad_norm": 2.46875, "learning_rate": 3.506161955045436e-05, "loss": 0.1941, "step": 12059 }, { "epoch": 74.90965732087227, "grad_norm": 2.5, "learning_rate": 3.5059364086149404e-05, "loss": 0.2325, "step": 12060 }, { "epoch": 74.91588785046729, "grad_norm": 2.171875, "learning_rate": 3.50571085241459e-05, "loss": 0.1671, "step": 12061 }, { "epoch": 74.9221183800623, "grad_norm": 1.1171875, "learning_rate": 3.505485286446574e-05, "loss": 0.1136, "step": 12062 }, { "epoch": 74.92834890965732, "grad_norm": 2.703125, "learning_rate": 3.5052597107130846e-05, "loss": 0.095, "step": 12063 }, { "epoch": 74.93457943925233, "grad_norm": 1.140625, "learning_rate": 3.5050341252163115e-05, "loss": 0.1017, "step": 12064 }, { "epoch": 74.94080996884735, "grad_norm": 2.28125, "learning_rate": 3.5048085299584454e-05, "loss": 0.2397, "step": 12065 }, { "epoch": 74.94704049844236, "grad_norm": 2.25, "learning_rate": 3.5045829249416795e-05, "loss": 0.1516, "step": 12066 }, { "epoch": 74.95327102803738, "grad_norm": 2.96875, "learning_rate": 3.504357310168203e-05, "loss": 0.3454, "step": 12067 }, { "epoch": 74.9595015576324, "grad_norm": 2.25, "learning_rate": 3.504131685640206e-05, "loss": 0.1661, "step": 12068 }, { "epoch": 74.96573208722741, "grad_norm": 3.5625, "learning_rate": 3.503906051359883e-05, "loss": 0.1646, "step": 12069 }, { "epoch": 74.97196261682242, "grad_norm": 2.75, "learning_rate": 3.5036804073294236e-05, "loss": 0.1431, "step": 12070 }, { "epoch": 74.97819314641744, "grad_norm": 1.6171875, "learning_rate": 3.503454753551019e-05, "loss": 0.0932, "step": 12071 }, { "epoch": 74.98442367601245, "grad_norm": 2.46875, "learning_rate": 3.5032290900268615e-05, "loss": 0.1408, "step": 12072 }, { "epoch": 74.99065420560747, "grad_norm": 2.078125, "learning_rate": 3.503003416759142e-05, "loss": 0.1365, "step": 12073 }, { "epoch": 74.99688473520249, "grad_norm": 2.625, "learning_rate": 3.5027777337500536e-05, "loss": 0.2278, "step": 12074 }, { "epoch": 75.0, "grad_norm": 1.921875, "learning_rate": 3.502552041001787e-05, "loss": 0.2268, "step": 12075 }, { "epoch": 75.00623052959502, "grad_norm": 2.40625, "learning_rate": 3.502326338516534e-05, "loss": 0.1599, "step": 12076 }, { "epoch": 75.01246105919003, "grad_norm": 1.8359375, "learning_rate": 3.502100626296488e-05, "loss": 0.1239, "step": 12077 }, { "epoch": 75.01869158878505, "grad_norm": 1.984375, "learning_rate": 3.50187490434384e-05, "loss": 0.1516, "step": 12078 }, { "epoch": 75.02492211838006, "grad_norm": 1.9765625, "learning_rate": 3.5016491726607835e-05, "loss": 0.1589, "step": 12079 }, { "epoch": 75.03115264797508, "grad_norm": 3.140625, "learning_rate": 3.501423431249509e-05, "loss": 0.1504, "step": 12080 }, { "epoch": 75.03738317757009, "grad_norm": 3.5, "learning_rate": 3.5011976801122105e-05, "loss": 0.1347, "step": 12081 }, { "epoch": 75.04361370716511, "grad_norm": 2.359375, "learning_rate": 3.5009719192510805e-05, "loss": 0.1414, "step": 12082 }, { "epoch": 75.04984423676012, "grad_norm": 2.515625, "learning_rate": 3.50074614866831e-05, "loss": 0.1839, "step": 12083 }, { "epoch": 75.05607476635514, "grad_norm": 3.03125, "learning_rate": 3.500520368366093e-05, "loss": 0.3241, "step": 12084 }, { "epoch": 75.06230529595015, "grad_norm": 3.90625, "learning_rate": 3.5002945783466235e-05, "loss": 0.1974, "step": 12085 }, { "epoch": 75.06853582554517, "grad_norm": 3.59375, "learning_rate": 3.5000687786120914e-05, "loss": 0.1374, "step": 12086 }, { "epoch": 75.07476635514018, "grad_norm": 3.421875, "learning_rate": 3.499842969164692e-05, "loss": 0.1634, "step": 12087 }, { "epoch": 75.0809968847352, "grad_norm": 3.046875, "learning_rate": 3.4996171500066185e-05, "loss": 0.1731, "step": 12088 }, { "epoch": 75.08722741433021, "grad_norm": 1.3671875, "learning_rate": 3.499391321140063e-05, "loss": 0.0978, "step": 12089 }, { "epoch": 75.09345794392523, "grad_norm": 4.6875, "learning_rate": 3.499165482567219e-05, "loss": 0.2319, "step": 12090 }, { "epoch": 75.09968847352025, "grad_norm": 4.65625, "learning_rate": 3.49893963429028e-05, "loss": 0.191, "step": 12091 }, { "epoch": 75.10591900311526, "grad_norm": 4.25, "learning_rate": 3.49871377631144e-05, "loss": 0.1174, "step": 12092 }, { "epoch": 75.11214953271028, "grad_norm": 4.5, "learning_rate": 3.4984879086328914e-05, "loss": 0.1141, "step": 12093 }, { "epoch": 75.11838006230529, "grad_norm": 3.375, "learning_rate": 3.498262031256829e-05, "loss": 0.1407, "step": 12094 }, { "epoch": 75.1246105919003, "grad_norm": 1.359375, "learning_rate": 3.498036144185447e-05, "loss": 0.1154, "step": 12095 }, { "epoch": 75.13084112149532, "grad_norm": 2.875, "learning_rate": 3.497810247420938e-05, "loss": 0.1816, "step": 12096 }, { "epoch": 75.13707165109034, "grad_norm": 1.6640625, "learning_rate": 3.4975843409654965e-05, "loss": 0.1325, "step": 12097 }, { "epoch": 75.14330218068535, "grad_norm": 3.40625, "learning_rate": 3.497358424821316e-05, "loss": 0.1437, "step": 12098 }, { "epoch": 75.14953271028037, "grad_norm": 2.078125, "learning_rate": 3.497132498990592e-05, "loss": 0.1813, "step": 12099 }, { "epoch": 75.15576323987538, "grad_norm": 2.296875, "learning_rate": 3.496906563475516e-05, "loss": 0.2391, "step": 12100 }, { "epoch": 75.1619937694704, "grad_norm": 3.0625, "learning_rate": 3.496680618278286e-05, "loss": 0.2892, "step": 12101 }, { "epoch": 75.16822429906541, "grad_norm": 2.640625, "learning_rate": 3.4964546634010944e-05, "loss": 0.1809, "step": 12102 }, { "epoch": 75.17445482866043, "grad_norm": 1.96875, "learning_rate": 3.496228698846135e-05, "loss": 0.1279, "step": 12103 }, { "epoch": 75.18068535825545, "grad_norm": 2.21875, "learning_rate": 3.496002724615604e-05, "loss": 0.1717, "step": 12104 }, { "epoch": 75.18691588785046, "grad_norm": 2.75, "learning_rate": 3.4957767407116956e-05, "loss": 0.1849, "step": 12105 }, { "epoch": 75.19314641744548, "grad_norm": 2.234375, "learning_rate": 3.4955507471366036e-05, "loss": 0.2592, "step": 12106 }, { "epoch": 75.19937694704049, "grad_norm": 2.6875, "learning_rate": 3.495324743892525e-05, "loss": 0.2067, "step": 12107 }, { "epoch": 75.20560747663552, "grad_norm": 1.078125, "learning_rate": 3.495098730981652e-05, "loss": 0.1175, "step": 12108 }, { "epoch": 75.21183800623054, "grad_norm": 2.015625, "learning_rate": 3.494872708406182e-05, "loss": 0.1443, "step": 12109 }, { "epoch": 75.21806853582555, "grad_norm": 2.546875, "learning_rate": 3.49464667616831e-05, "loss": 0.18, "step": 12110 }, { "epoch": 75.22429906542057, "grad_norm": 2.21875, "learning_rate": 3.494420634270229e-05, "loss": 0.1869, "step": 12111 }, { "epoch": 75.23052959501558, "grad_norm": 2.40625, "learning_rate": 3.4941945827141376e-05, "loss": 0.1343, "step": 12112 }, { "epoch": 75.2367601246106, "grad_norm": 3.171875, "learning_rate": 3.493968521502229e-05, "loss": 0.2829, "step": 12113 }, { "epoch": 75.24299065420561, "grad_norm": 2.625, "learning_rate": 3.4937424506367e-05, "loss": 0.1584, "step": 12114 }, { "epoch": 75.24922118380063, "grad_norm": 2.359375, "learning_rate": 3.493516370119745e-05, "loss": 0.1508, "step": 12115 }, { "epoch": 75.25545171339564, "grad_norm": 3.328125, "learning_rate": 3.4932902799535606e-05, "loss": 0.2654, "step": 12116 }, { "epoch": 75.26168224299066, "grad_norm": 1.8671875, "learning_rate": 3.493064180140343e-05, "loss": 0.1066, "step": 12117 }, { "epoch": 75.26791277258567, "grad_norm": 2.390625, "learning_rate": 3.4928380706822866e-05, "loss": 0.1079, "step": 12118 }, { "epoch": 75.27414330218069, "grad_norm": 2.0625, "learning_rate": 3.492611951581589e-05, "loss": 0.1624, "step": 12119 }, { "epoch": 75.2803738317757, "grad_norm": 1.5078125, "learning_rate": 3.492385822840446e-05, "loss": 0.1092, "step": 12120 }, { "epoch": 75.28660436137072, "grad_norm": 1.4609375, "learning_rate": 3.4921596844610525e-05, "loss": 0.1369, "step": 12121 }, { "epoch": 75.29283489096574, "grad_norm": 2.609375, "learning_rate": 3.491933536445606e-05, "loss": 0.1714, "step": 12122 }, { "epoch": 75.29906542056075, "grad_norm": 3.875, "learning_rate": 3.491707378796304e-05, "loss": 0.1739, "step": 12123 }, { "epoch": 75.30529595015577, "grad_norm": 4.375, "learning_rate": 3.4914812115153403e-05, "loss": 0.2337, "step": 12124 }, { "epoch": 75.31152647975078, "grad_norm": 3.625, "learning_rate": 3.4912550346049136e-05, "loss": 0.161, "step": 12125 }, { "epoch": 75.3177570093458, "grad_norm": 3.390625, "learning_rate": 3.4910288480672194e-05, "loss": 0.1864, "step": 12126 }, { "epoch": 75.32398753894081, "grad_norm": 2.1875, "learning_rate": 3.4908026519044554e-05, "loss": 0.1373, "step": 12127 }, { "epoch": 75.33021806853583, "grad_norm": 2.078125, "learning_rate": 3.490576446118817e-05, "loss": 0.1069, "step": 12128 }, { "epoch": 75.33644859813084, "grad_norm": 2.59375, "learning_rate": 3.4903502307125034e-05, "loss": 0.1455, "step": 12129 }, { "epoch": 75.34267912772586, "grad_norm": 2.40625, "learning_rate": 3.4901240056877104e-05, "loss": 0.1267, "step": 12130 }, { "epoch": 75.34890965732087, "grad_norm": 2.375, "learning_rate": 3.489897771046634e-05, "loss": 0.1257, "step": 12131 }, { "epoch": 75.35514018691589, "grad_norm": 2.875, "learning_rate": 3.4896715267914734e-05, "loss": 0.2432, "step": 12132 }, { "epoch": 75.3613707165109, "grad_norm": 1.8515625, "learning_rate": 3.489445272924425e-05, "loss": 0.1185, "step": 12133 }, { "epoch": 75.36760124610592, "grad_norm": 1.9609375, "learning_rate": 3.489219009447686e-05, "loss": 0.1097, "step": 12134 }, { "epoch": 75.37383177570094, "grad_norm": 1.484375, "learning_rate": 3.4889927363634544e-05, "loss": 0.1219, "step": 12135 }, { "epoch": 75.38006230529595, "grad_norm": 1.625, "learning_rate": 3.488766453673928e-05, "loss": 0.1141, "step": 12136 }, { "epoch": 75.38629283489097, "grad_norm": 2.640625, "learning_rate": 3.488540161381304e-05, "loss": 0.2392, "step": 12137 }, { "epoch": 75.39252336448598, "grad_norm": 1.46875, "learning_rate": 3.48831385948778e-05, "loss": 0.1098, "step": 12138 }, { "epoch": 75.398753894081, "grad_norm": 1.9609375, "learning_rate": 3.488087547995555e-05, "loss": 0.15, "step": 12139 }, { "epoch": 75.40498442367601, "grad_norm": 1.5078125, "learning_rate": 3.487861226906826e-05, "loss": 0.1207, "step": 12140 }, { "epoch": 75.41121495327103, "grad_norm": 2.265625, "learning_rate": 3.487634896223791e-05, "loss": 0.1947, "step": 12141 }, { "epoch": 75.41744548286604, "grad_norm": 1.3125, "learning_rate": 3.487408555948648e-05, "loss": 0.1217, "step": 12142 }, { "epoch": 75.42367601246106, "grad_norm": 2.234375, "learning_rate": 3.487182206083596e-05, "loss": 0.2456, "step": 12143 }, { "epoch": 75.42990654205607, "grad_norm": 1.578125, "learning_rate": 3.4869558466308335e-05, "loss": 0.1153, "step": 12144 }, { "epoch": 75.43613707165109, "grad_norm": 3.28125, "learning_rate": 3.486729477592558e-05, "loss": 0.2344, "step": 12145 }, { "epoch": 75.4423676012461, "grad_norm": 1.1640625, "learning_rate": 3.486503098970969e-05, "loss": 0.0879, "step": 12146 }, { "epoch": 75.44859813084112, "grad_norm": 2.953125, "learning_rate": 3.4862767107682645e-05, "loss": 0.2335, "step": 12147 }, { "epoch": 75.45482866043614, "grad_norm": 1.0390625, "learning_rate": 3.486050312986643e-05, "loss": 0.0963, "step": 12148 }, { "epoch": 75.46105919003115, "grad_norm": 2.328125, "learning_rate": 3.485823905628305e-05, "loss": 0.1775, "step": 12149 }, { "epoch": 75.46728971962617, "grad_norm": 2.34375, "learning_rate": 3.4855974886954466e-05, "loss": 0.1792, "step": 12150 }, { "epoch": 75.47352024922118, "grad_norm": 2.390625, "learning_rate": 3.48537106219027e-05, "loss": 0.2712, "step": 12151 }, { "epoch": 75.4797507788162, "grad_norm": 3.25, "learning_rate": 3.485144626114971e-05, "loss": 0.2196, "step": 12152 }, { "epoch": 75.48598130841121, "grad_norm": 2.328125, "learning_rate": 3.484918180471752e-05, "loss": 0.1625, "step": 12153 }, { "epoch": 75.49221183800623, "grad_norm": 2.0625, "learning_rate": 3.484691725262809e-05, "loss": 0.1293, "step": 12154 }, { "epoch": 75.49844236760124, "grad_norm": 1.5625, "learning_rate": 3.484465260490345e-05, "loss": 0.1306, "step": 12155 }, { "epoch": 75.50467289719626, "grad_norm": 3.828125, "learning_rate": 3.484238786156556e-05, "loss": 0.2021, "step": 12156 }, { "epoch": 75.51090342679127, "grad_norm": 2.046875, "learning_rate": 3.484012302263644e-05, "loss": 0.1313, "step": 12157 }, { "epoch": 75.51713395638629, "grad_norm": 1.7421875, "learning_rate": 3.483785808813809e-05, "loss": 0.1226, "step": 12158 }, { "epoch": 75.5233644859813, "grad_norm": 2.328125, "learning_rate": 3.483559305809248e-05, "loss": 0.1819, "step": 12159 }, { "epoch": 75.52959501557632, "grad_norm": 1.9375, "learning_rate": 3.483332793252162e-05, "loss": 0.1319, "step": 12160 }, { "epoch": 75.53582554517133, "grad_norm": 1.640625, "learning_rate": 3.483106271144752e-05, "loss": 0.1655, "step": 12161 }, { "epoch": 75.54205607476635, "grad_norm": 3.90625, "learning_rate": 3.482879739489218e-05, "loss": 0.3125, "step": 12162 }, { "epoch": 75.54828660436137, "grad_norm": 1.3203125, "learning_rate": 3.4826531982877585e-05, "loss": 0.123, "step": 12163 }, { "epoch": 75.55451713395638, "grad_norm": 3.484375, "learning_rate": 3.4824266475425754e-05, "loss": 0.2243, "step": 12164 }, { "epoch": 75.5607476635514, "grad_norm": 1.4140625, "learning_rate": 3.482200087255868e-05, "loss": 0.0875, "step": 12165 }, { "epoch": 75.56697819314641, "grad_norm": 1.8359375, "learning_rate": 3.481973517429837e-05, "loss": 0.1509, "step": 12166 }, { "epoch": 75.57320872274143, "grad_norm": 2.03125, "learning_rate": 3.481746938066684e-05, "loss": 0.1722, "step": 12167 }, { "epoch": 75.57943925233644, "grad_norm": 3.640625, "learning_rate": 3.481520349168607e-05, "loss": 0.1839, "step": 12168 }, { "epoch": 75.58566978193146, "grad_norm": 3.109375, "learning_rate": 3.481293750737808e-05, "loss": 0.1804, "step": 12169 }, { "epoch": 75.59190031152647, "grad_norm": 2.578125, "learning_rate": 3.481067142776489e-05, "loss": 0.1595, "step": 12170 }, { "epoch": 75.59813084112149, "grad_norm": 2.015625, "learning_rate": 3.480840525286849e-05, "loss": 0.1222, "step": 12171 }, { "epoch": 75.60436137071652, "grad_norm": 1.2890625, "learning_rate": 3.4806138982710904e-05, "loss": 0.1163, "step": 12172 }, { "epoch": 75.61059190031153, "grad_norm": 3.453125, "learning_rate": 3.480387261731414e-05, "loss": 0.3148, "step": 12173 }, { "epoch": 75.61682242990655, "grad_norm": 2.046875, "learning_rate": 3.480160615670019e-05, "loss": 0.1448, "step": 12174 }, { "epoch": 75.62305295950156, "grad_norm": 3.796875, "learning_rate": 3.47993396008911e-05, "loss": 0.2021, "step": 12175 }, { "epoch": 75.62928348909658, "grad_norm": 2.484375, "learning_rate": 3.479707294990885e-05, "loss": 0.1243, "step": 12176 }, { "epoch": 75.6355140186916, "grad_norm": 1.7578125, "learning_rate": 3.4794806203775484e-05, "loss": 0.1564, "step": 12177 }, { "epoch": 75.64174454828661, "grad_norm": 2.1875, "learning_rate": 3.479253936251299e-05, "loss": 0.1997, "step": 12178 }, { "epoch": 75.64797507788163, "grad_norm": 2.296875, "learning_rate": 3.4790272426143395e-05, "loss": 0.1139, "step": 12179 }, { "epoch": 75.65420560747664, "grad_norm": 2.03125, "learning_rate": 3.478800539468872e-05, "loss": 0.1511, "step": 12180 }, { "epoch": 75.66043613707166, "grad_norm": 3.3125, "learning_rate": 3.478573826817099e-05, "loss": 0.192, "step": 12181 }, { "epoch": 75.66666666666667, "grad_norm": 2.75, "learning_rate": 3.478347104661219e-05, "loss": 0.3649, "step": 12182 }, { "epoch": 75.67289719626169, "grad_norm": 2.4375, "learning_rate": 3.478120373003439e-05, "loss": 0.1891, "step": 12183 }, { "epoch": 75.6791277258567, "grad_norm": 3.265625, "learning_rate": 3.4778936318459565e-05, "loss": 0.2358, "step": 12184 }, { "epoch": 75.68535825545172, "grad_norm": 3.890625, "learning_rate": 3.4776668811909765e-05, "loss": 0.2288, "step": 12185 }, { "epoch": 75.69158878504673, "grad_norm": 2.640625, "learning_rate": 3.4774401210407e-05, "loss": 0.1675, "step": 12186 }, { "epoch": 75.69781931464175, "grad_norm": 1.609375, "learning_rate": 3.4772133513973306e-05, "loss": 0.1163, "step": 12187 }, { "epoch": 75.70404984423676, "grad_norm": 1.7109375, "learning_rate": 3.476986572263069e-05, "loss": 0.1407, "step": 12188 }, { "epoch": 75.71028037383178, "grad_norm": 2.78125, "learning_rate": 3.476759783640118e-05, "loss": 0.1675, "step": 12189 }, { "epoch": 75.7165109034268, "grad_norm": 2.6875, "learning_rate": 3.4765329855306816e-05, "loss": 0.141, "step": 12190 }, { "epoch": 75.72274143302181, "grad_norm": 2.515625, "learning_rate": 3.476306177936961e-05, "loss": 0.1328, "step": 12191 }, { "epoch": 75.72897196261682, "grad_norm": 2.890625, "learning_rate": 3.4760793608611606e-05, "loss": 0.1755, "step": 12192 }, { "epoch": 75.73520249221184, "grad_norm": 2.734375, "learning_rate": 3.475852534305481e-05, "loss": 0.2781, "step": 12193 }, { "epoch": 75.74143302180686, "grad_norm": 2.84375, "learning_rate": 3.4756256982721274e-05, "loss": 0.2417, "step": 12194 }, { "epoch": 75.74766355140187, "grad_norm": 1.75, "learning_rate": 3.475398852763302e-05, "loss": 0.0957, "step": 12195 }, { "epoch": 75.75389408099689, "grad_norm": 4.375, "learning_rate": 3.4751719977812076e-05, "loss": 0.1859, "step": 12196 }, { "epoch": 75.7601246105919, "grad_norm": 3.78125, "learning_rate": 3.474945133328048e-05, "loss": 0.1819, "step": 12197 }, { "epoch": 75.76635514018692, "grad_norm": 1.5078125, "learning_rate": 3.4747182594060265e-05, "loss": 0.1407, "step": 12198 }, { "epoch": 75.77258566978193, "grad_norm": 2.046875, "learning_rate": 3.474491376017346e-05, "loss": 0.1911, "step": 12199 }, { "epoch": 75.77881619937695, "grad_norm": 2.828125, "learning_rate": 3.474264483164211e-05, "loss": 0.0991, "step": 12200 }, { "epoch": 75.78504672897196, "grad_norm": 2.21875, "learning_rate": 3.474037580848824e-05, "loss": 0.1459, "step": 12201 }, { "epoch": 75.79127725856698, "grad_norm": 1.8125, "learning_rate": 3.4738106690733896e-05, "loss": 0.1368, "step": 12202 }, { "epoch": 75.797507788162, "grad_norm": 2.921875, "learning_rate": 3.473583747840112e-05, "loss": 0.1475, "step": 12203 }, { "epoch": 75.80373831775701, "grad_norm": 1.7421875, "learning_rate": 3.4733568171511935e-05, "loss": 0.1321, "step": 12204 }, { "epoch": 75.80996884735202, "grad_norm": 2.09375, "learning_rate": 3.473129877008839e-05, "loss": 0.1658, "step": 12205 }, { "epoch": 75.81619937694704, "grad_norm": 1.8203125, "learning_rate": 3.472902927415253e-05, "loss": 0.1338, "step": 12206 }, { "epoch": 75.82242990654206, "grad_norm": 2.09375, "learning_rate": 3.472675968372639e-05, "loss": 0.2655, "step": 12207 }, { "epoch": 75.82866043613707, "grad_norm": 1.75, "learning_rate": 3.4724489998832025e-05, "loss": 0.1469, "step": 12208 }, { "epoch": 75.83489096573209, "grad_norm": 2.609375, "learning_rate": 3.472222021949146e-05, "loss": 0.2249, "step": 12209 }, { "epoch": 75.8411214953271, "grad_norm": 3.515625, "learning_rate": 3.4719950345726746e-05, "loss": 0.2285, "step": 12210 }, { "epoch": 75.84735202492212, "grad_norm": 2.890625, "learning_rate": 3.4717680377559935e-05, "loss": 0.2407, "step": 12211 }, { "epoch": 75.85358255451713, "grad_norm": 3.5, "learning_rate": 3.471541031501307e-05, "loss": 0.223, "step": 12212 }, { "epoch": 75.85981308411215, "grad_norm": 1.96875, "learning_rate": 3.47131401581082e-05, "loss": 0.2277, "step": 12213 }, { "epoch": 75.86604361370716, "grad_norm": 2.953125, "learning_rate": 3.471086990686737e-05, "loss": 0.2061, "step": 12214 }, { "epoch": 75.87227414330218, "grad_norm": 2.4375, "learning_rate": 3.4708599561312626e-05, "loss": 0.1282, "step": 12215 }, { "epoch": 75.8785046728972, "grad_norm": 1.0703125, "learning_rate": 3.470632912146603e-05, "loss": 0.1072, "step": 12216 }, { "epoch": 75.88473520249221, "grad_norm": 1.7265625, "learning_rate": 3.470405858734962e-05, "loss": 0.1339, "step": 12217 }, { "epoch": 75.89096573208722, "grad_norm": 2.03125, "learning_rate": 3.470178795898545e-05, "loss": 0.1339, "step": 12218 }, { "epoch": 75.89719626168224, "grad_norm": 1.96875, "learning_rate": 3.469951723639558e-05, "loss": 0.2027, "step": 12219 }, { "epoch": 75.90342679127725, "grad_norm": 2.140625, "learning_rate": 3.4697246419602066e-05, "loss": 0.1947, "step": 12220 }, { "epoch": 75.90965732087227, "grad_norm": 1.5390625, "learning_rate": 3.469497550862695e-05, "loss": 0.1688, "step": 12221 }, { "epoch": 75.91588785046729, "grad_norm": 1.2109375, "learning_rate": 3.46927045034923e-05, "loss": 0.1048, "step": 12222 }, { "epoch": 75.9221183800623, "grad_norm": 1.390625, "learning_rate": 3.469043340422016e-05, "loss": 0.122, "step": 12223 }, { "epoch": 75.92834890965732, "grad_norm": 2.078125, "learning_rate": 3.468816221083259e-05, "loss": 0.1673, "step": 12224 }, { "epoch": 75.93457943925233, "grad_norm": 1.5, "learning_rate": 3.4685890923351665e-05, "loss": 0.1306, "step": 12225 }, { "epoch": 75.94080996884735, "grad_norm": 2.53125, "learning_rate": 3.468361954179942e-05, "loss": 0.1431, "step": 12226 }, { "epoch": 75.94704049844236, "grad_norm": 2.203125, "learning_rate": 3.468134806619793e-05, "loss": 0.1057, "step": 12227 }, { "epoch": 75.95327102803738, "grad_norm": 2.015625, "learning_rate": 3.467907649656925e-05, "loss": 0.151, "step": 12228 }, { "epoch": 75.9595015576324, "grad_norm": 1.4140625, "learning_rate": 3.467680483293545e-05, "loss": 0.1209, "step": 12229 }, { "epoch": 75.96573208722741, "grad_norm": 2.765625, "learning_rate": 3.4674533075318575e-05, "loss": 0.2012, "step": 12230 }, { "epoch": 75.97196261682242, "grad_norm": 1.671875, "learning_rate": 3.467226122374071e-05, "loss": 0.1128, "step": 12231 }, { "epoch": 75.97819314641744, "grad_norm": 1.7265625, "learning_rate": 3.466998927822391e-05, "loss": 0.1255, "step": 12232 }, { "epoch": 75.98442367601245, "grad_norm": 2.078125, "learning_rate": 3.466771723879024e-05, "loss": 0.1565, "step": 12233 }, { "epoch": 75.99065420560747, "grad_norm": 1.4140625, "learning_rate": 3.4665445105461767e-05, "loss": 0.1422, "step": 12234 }, { "epoch": 75.99688473520249, "grad_norm": 1.25, "learning_rate": 3.4663172878260566e-05, "loss": 0.1023, "step": 12235 }, { "epoch": 76.0, "grad_norm": 1.9609375, "learning_rate": 3.466090055720869e-05, "loss": 0.0854, "step": 12236 }, { "epoch": 76.00623052959502, "grad_norm": 2.046875, "learning_rate": 3.465862814232822e-05, "loss": 0.1521, "step": 12237 }, { "epoch": 76.01246105919003, "grad_norm": 4.5, "learning_rate": 3.4656355633641224e-05, "loss": 0.2963, "step": 12238 }, { "epoch": 76.01869158878505, "grad_norm": 2.234375, "learning_rate": 3.4654083031169766e-05, "loss": 0.1153, "step": 12239 }, { "epoch": 76.02492211838006, "grad_norm": 1.609375, "learning_rate": 3.465181033493593e-05, "loss": 0.1405, "step": 12240 }, { "epoch": 76.03115264797508, "grad_norm": 2.21875, "learning_rate": 3.464953754496178e-05, "loss": 0.1399, "step": 12241 }, { "epoch": 76.03738317757009, "grad_norm": 2.375, "learning_rate": 3.464726466126939e-05, "loss": 0.1829, "step": 12242 }, { "epoch": 76.04361370716511, "grad_norm": 1.4453125, "learning_rate": 3.464499168388084e-05, "loss": 0.1089, "step": 12243 }, { "epoch": 76.04984423676012, "grad_norm": 2.25, "learning_rate": 3.46427186128182e-05, "loss": 0.1358, "step": 12244 }, { "epoch": 76.05607476635514, "grad_norm": 1.6015625, "learning_rate": 3.464044544810356e-05, "loss": 0.1367, "step": 12245 }, { "epoch": 76.06230529595015, "grad_norm": 2.8125, "learning_rate": 3.4638172189758975e-05, "loss": 0.2181, "step": 12246 }, { "epoch": 76.06853582554517, "grad_norm": 2.25, "learning_rate": 3.4635898837806535e-05, "loss": 0.1732, "step": 12247 }, { "epoch": 76.07476635514018, "grad_norm": 3.421875, "learning_rate": 3.4633625392268325e-05, "loss": 0.2259, "step": 12248 }, { "epoch": 76.0809968847352, "grad_norm": 3.28125, "learning_rate": 3.4631351853166416e-05, "loss": 0.2273, "step": 12249 }, { "epoch": 76.08722741433021, "grad_norm": 3.203125, "learning_rate": 3.462907822052289e-05, "loss": 0.2313, "step": 12250 }, { "epoch": 76.09345794392523, "grad_norm": 1.71875, "learning_rate": 3.4626804494359846e-05, "loss": 0.1111, "step": 12251 }, { "epoch": 76.09968847352025, "grad_norm": 2.515625, "learning_rate": 3.462453067469933e-05, "loss": 0.1615, "step": 12252 }, { "epoch": 76.10591900311526, "grad_norm": 2.875, "learning_rate": 3.462225676156346e-05, "loss": 0.22, "step": 12253 }, { "epoch": 76.11214953271028, "grad_norm": 1.71875, "learning_rate": 3.461998275497431e-05, "loss": 0.1181, "step": 12254 }, { "epoch": 76.11838006230529, "grad_norm": 2.171875, "learning_rate": 3.461770865495397e-05, "loss": 0.1803, "step": 12255 }, { "epoch": 76.1246105919003, "grad_norm": 2.734375, "learning_rate": 3.4615434461524516e-05, "loss": 0.1014, "step": 12256 }, { "epoch": 76.13084112149532, "grad_norm": 1.3203125, "learning_rate": 3.461316017470804e-05, "loss": 0.1043, "step": 12257 }, { "epoch": 76.13707165109034, "grad_norm": 1.2265625, "learning_rate": 3.4610885794526625e-05, "loss": 0.0939, "step": 12258 }, { "epoch": 76.14330218068535, "grad_norm": 2.5625, "learning_rate": 3.460861132100237e-05, "loss": 0.2466, "step": 12259 }, { "epoch": 76.14953271028037, "grad_norm": 1.2734375, "learning_rate": 3.460633675415736e-05, "loss": 0.1389, "step": 12260 }, { "epoch": 76.15576323987538, "grad_norm": 1.1171875, "learning_rate": 3.460406209401369e-05, "loss": 0.1122, "step": 12261 }, { "epoch": 76.1619937694704, "grad_norm": 2.71875, "learning_rate": 3.460178734059345e-05, "loss": 0.2231, "step": 12262 }, { "epoch": 76.16822429906541, "grad_norm": 3.375, "learning_rate": 3.459951249391873e-05, "loss": 0.27, "step": 12263 }, { "epoch": 76.17445482866043, "grad_norm": 2.625, "learning_rate": 3.459723755401163e-05, "loss": 0.2283, "step": 12264 }, { "epoch": 76.18068535825545, "grad_norm": 0.921875, "learning_rate": 3.4594962520894233e-05, "loss": 0.1023, "step": 12265 }, { "epoch": 76.18691588785046, "grad_norm": 2.46875, "learning_rate": 3.459268739458864e-05, "loss": 0.161, "step": 12266 }, { "epoch": 76.19314641744548, "grad_norm": 4.3125, "learning_rate": 3.459041217511696e-05, "loss": 0.3409, "step": 12267 }, { "epoch": 76.19937694704049, "grad_norm": 1.6171875, "learning_rate": 3.4588136862501273e-05, "loss": 0.1329, "step": 12268 }, { "epoch": 76.20560747663552, "grad_norm": 2.078125, "learning_rate": 3.4585861456763686e-05, "loss": 0.2233, "step": 12269 }, { "epoch": 76.21183800623054, "grad_norm": 3.3125, "learning_rate": 3.4583585957926295e-05, "loss": 0.0936, "step": 12270 }, { "epoch": 76.21806853582555, "grad_norm": 2.796875, "learning_rate": 3.4581310366011205e-05, "loss": 0.2314, "step": 12271 }, { "epoch": 76.22429906542057, "grad_norm": 2.9375, "learning_rate": 3.4579034681040505e-05, "loss": 0.1302, "step": 12272 }, { "epoch": 76.23052959501558, "grad_norm": 2.75, "learning_rate": 3.457675890303632e-05, "loss": 0.2196, "step": 12273 }, { "epoch": 76.2367601246106, "grad_norm": 2.3125, "learning_rate": 3.4574483032020725e-05, "loss": 0.1217, "step": 12274 }, { "epoch": 76.24299065420561, "grad_norm": 3.34375, "learning_rate": 3.457220706801584e-05, "loss": 0.1201, "step": 12275 }, { "epoch": 76.24922118380063, "grad_norm": 3.359375, "learning_rate": 3.456993101104377e-05, "loss": 0.2792, "step": 12276 }, { "epoch": 76.25545171339564, "grad_norm": 1.875, "learning_rate": 3.456765486112661e-05, "loss": 0.1252, "step": 12277 }, { "epoch": 76.26168224299066, "grad_norm": 3.09375, "learning_rate": 3.456537861828648e-05, "loss": 0.192, "step": 12278 }, { "epoch": 76.26791277258567, "grad_norm": 2.078125, "learning_rate": 3.4563102282545485e-05, "loss": 0.1439, "step": 12279 }, { "epoch": 76.27414330218069, "grad_norm": 3.21875, "learning_rate": 3.456082585392571e-05, "loss": 0.1928, "step": 12280 }, { "epoch": 76.2803738317757, "grad_norm": 2.203125, "learning_rate": 3.455854933244931e-05, "loss": 0.1118, "step": 12281 }, { "epoch": 76.28660436137072, "grad_norm": 3.3125, "learning_rate": 3.455627271813835e-05, "loss": 0.1967, "step": 12282 }, { "epoch": 76.29283489096574, "grad_norm": 1.6796875, "learning_rate": 3.455399601101497e-05, "loss": 0.1555, "step": 12283 }, { "epoch": 76.29906542056075, "grad_norm": 2.59375, "learning_rate": 3.4551719211101266e-05, "loss": 0.1387, "step": 12284 }, { "epoch": 76.30529595015577, "grad_norm": 4.53125, "learning_rate": 3.454944231841934e-05, "loss": 0.1845, "step": 12285 }, { "epoch": 76.31152647975078, "grad_norm": 2.359375, "learning_rate": 3.454716533299134e-05, "loss": 0.2284, "step": 12286 }, { "epoch": 76.3177570093458, "grad_norm": 4.1875, "learning_rate": 3.454488825483936e-05, "loss": 0.2902, "step": 12287 }, { "epoch": 76.32398753894081, "grad_norm": 2.421875, "learning_rate": 3.454261108398551e-05, "loss": 0.2385, "step": 12288 }, { "epoch": 76.33021806853583, "grad_norm": 1.1015625, "learning_rate": 3.454033382045192e-05, "loss": 0.1078, "step": 12289 }, { "epoch": 76.33644859813084, "grad_norm": 2.109375, "learning_rate": 3.45380564642607e-05, "loss": 0.1307, "step": 12290 }, { "epoch": 76.34267912772586, "grad_norm": 1.984375, "learning_rate": 3.4535779015433975e-05, "loss": 0.1516, "step": 12291 }, { "epoch": 76.34890965732087, "grad_norm": 1.9375, "learning_rate": 3.4533501473993844e-05, "loss": 0.1067, "step": 12292 }, { "epoch": 76.35514018691589, "grad_norm": 3.9375, "learning_rate": 3.4531223839962453e-05, "loss": 0.1498, "step": 12293 }, { "epoch": 76.3613707165109, "grad_norm": 4.4375, "learning_rate": 3.45289461133619e-05, "loss": 0.4258, "step": 12294 }, { "epoch": 76.36760124610592, "grad_norm": 2.40625, "learning_rate": 3.4526668294214335e-05, "loss": 0.1872, "step": 12295 }, { "epoch": 76.37383177570094, "grad_norm": 1.8671875, "learning_rate": 3.452439038254184e-05, "loss": 0.2428, "step": 12296 }, { "epoch": 76.38006230529595, "grad_norm": 2.953125, "learning_rate": 3.452211237836658e-05, "loss": 0.133, "step": 12297 }, { "epoch": 76.38629283489097, "grad_norm": 1.4296875, "learning_rate": 3.451983428171065e-05, "loss": 0.1452, "step": 12298 }, { "epoch": 76.39252336448598, "grad_norm": 1.46875, "learning_rate": 3.4517556092596194e-05, "loss": 0.1339, "step": 12299 }, { "epoch": 76.398753894081, "grad_norm": 1.1171875, "learning_rate": 3.4515277811045325e-05, "loss": 0.092, "step": 12300 }, { "epoch": 76.40498442367601, "grad_norm": 1.03125, "learning_rate": 3.451299943708018e-05, "loss": 0.108, "step": 12301 }, { "epoch": 76.41121495327103, "grad_norm": 2.078125, "learning_rate": 3.451072097072288e-05, "loss": 0.1522, "step": 12302 }, { "epoch": 76.41744548286604, "grad_norm": 2.34375, "learning_rate": 3.4508442411995556e-05, "loss": 0.1976, "step": 12303 }, { "epoch": 76.42367601246106, "grad_norm": 2.21875, "learning_rate": 3.450616376092034e-05, "loss": 0.1906, "step": 12304 }, { "epoch": 76.42990654205607, "grad_norm": 2.25, "learning_rate": 3.4503885017519364e-05, "loss": 0.2025, "step": 12305 }, { "epoch": 76.43613707165109, "grad_norm": 1.40625, "learning_rate": 3.450160618181476e-05, "loss": 0.1171, "step": 12306 }, { "epoch": 76.4423676012461, "grad_norm": 1.59375, "learning_rate": 3.449932725382865e-05, "loss": 0.1409, "step": 12307 }, { "epoch": 76.44859813084112, "grad_norm": 2.34375, "learning_rate": 3.449704823358318e-05, "loss": 0.1879, "step": 12308 }, { "epoch": 76.45482866043614, "grad_norm": 2.609375, "learning_rate": 3.449476912110047e-05, "loss": 0.1395, "step": 12309 }, { "epoch": 76.46105919003115, "grad_norm": 2.328125, "learning_rate": 3.449248991640267e-05, "loss": 0.1364, "step": 12310 }, { "epoch": 76.46728971962617, "grad_norm": 2.96875, "learning_rate": 3.449021061951191e-05, "loss": 0.2659, "step": 12311 }, { "epoch": 76.47352024922118, "grad_norm": 2.15625, "learning_rate": 3.448793123045033e-05, "loss": 0.1871, "step": 12312 }, { "epoch": 76.4797507788162, "grad_norm": 2.03125, "learning_rate": 3.4485651749240066e-05, "loss": 0.1393, "step": 12313 }, { "epoch": 76.48598130841121, "grad_norm": 2.203125, "learning_rate": 3.4483372175903264e-05, "loss": 0.1411, "step": 12314 }, { "epoch": 76.49221183800623, "grad_norm": 2.25, "learning_rate": 3.4481092510462036e-05, "loss": 0.1643, "step": 12315 }, { "epoch": 76.49844236760124, "grad_norm": 1.703125, "learning_rate": 3.447881275293855e-05, "loss": 0.1197, "step": 12316 }, { "epoch": 76.50467289719626, "grad_norm": 1.296875, "learning_rate": 3.447653290335495e-05, "loss": 0.1092, "step": 12317 }, { "epoch": 76.51090342679127, "grad_norm": 1.484375, "learning_rate": 3.4474252961733356e-05, "loss": 0.1136, "step": 12318 }, { "epoch": 76.51713395638629, "grad_norm": 1.8203125, "learning_rate": 3.447197292809593e-05, "loss": 0.1601, "step": 12319 }, { "epoch": 76.5233644859813, "grad_norm": 3.4375, "learning_rate": 3.4469692802464795e-05, "loss": 0.1463, "step": 12320 }, { "epoch": 76.52959501557632, "grad_norm": 2.0625, "learning_rate": 3.4467412584862136e-05, "loss": 0.128, "step": 12321 }, { "epoch": 76.53582554517133, "grad_norm": 2.046875, "learning_rate": 3.446513227531005e-05, "loss": 0.14, "step": 12322 }, { "epoch": 76.54205607476635, "grad_norm": 2.640625, "learning_rate": 3.446285187383072e-05, "loss": 0.1792, "step": 12323 }, { "epoch": 76.54828660436137, "grad_norm": 1.3671875, "learning_rate": 3.446057138044628e-05, "loss": 0.114, "step": 12324 }, { "epoch": 76.55451713395638, "grad_norm": 2.75, "learning_rate": 3.445829079517888e-05, "loss": 0.1694, "step": 12325 }, { "epoch": 76.5607476635514, "grad_norm": 1.2265625, "learning_rate": 3.445601011805066e-05, "loss": 0.0976, "step": 12326 }, { "epoch": 76.56697819314641, "grad_norm": 1.40625, "learning_rate": 3.445372934908379e-05, "loss": 0.1072, "step": 12327 }, { "epoch": 76.57320872274143, "grad_norm": 2.765625, "learning_rate": 3.4451448488300404e-05, "loss": 0.3112, "step": 12328 }, { "epoch": 76.57943925233644, "grad_norm": 1.3828125, "learning_rate": 3.444916753572266e-05, "loss": 0.124, "step": 12329 }, { "epoch": 76.58566978193146, "grad_norm": 1.2421875, "learning_rate": 3.444688649137272e-05, "loss": 0.1069, "step": 12330 }, { "epoch": 76.59190031152647, "grad_norm": 3.046875, "learning_rate": 3.444460535527273e-05, "loss": 0.2357, "step": 12331 }, { "epoch": 76.59813084112149, "grad_norm": 1.5078125, "learning_rate": 3.444232412744484e-05, "loss": 0.1063, "step": 12332 }, { "epoch": 76.60436137071652, "grad_norm": 2.328125, "learning_rate": 3.444004280791121e-05, "loss": 0.1844, "step": 12333 }, { "epoch": 76.61059190031153, "grad_norm": 1.953125, "learning_rate": 3.4437761396694e-05, "loss": 0.2156, "step": 12334 }, { "epoch": 76.61682242990655, "grad_norm": 3.0625, "learning_rate": 3.443547989381536e-05, "loss": 0.1372, "step": 12335 }, { "epoch": 76.62305295950156, "grad_norm": 1.46875, "learning_rate": 3.443319829929746e-05, "loss": 0.1145, "step": 12336 }, { "epoch": 76.62928348909658, "grad_norm": 1.7265625, "learning_rate": 3.443091661316245e-05, "loss": 0.1513, "step": 12337 }, { "epoch": 76.6355140186916, "grad_norm": 1.703125, "learning_rate": 3.4428634835432496e-05, "loss": 0.127, "step": 12338 }, { "epoch": 76.64174454828661, "grad_norm": 1.2734375, "learning_rate": 3.442635296612975e-05, "loss": 0.0985, "step": 12339 }, { "epoch": 76.64797507788163, "grad_norm": 2.71875, "learning_rate": 3.442407100527638e-05, "loss": 0.323, "step": 12340 }, { "epoch": 76.65420560747664, "grad_norm": 3.5, "learning_rate": 3.442178895289455e-05, "loss": 0.1581, "step": 12341 }, { "epoch": 76.66043613707166, "grad_norm": 2.390625, "learning_rate": 3.441950680900642e-05, "loss": 0.1281, "step": 12342 }, { "epoch": 76.66666666666667, "grad_norm": 2.703125, "learning_rate": 3.441722457363416e-05, "loss": 0.2698, "step": 12343 }, { "epoch": 76.67289719626169, "grad_norm": 1.828125, "learning_rate": 3.441494224679993e-05, "loss": 0.1547, "step": 12344 }, { "epoch": 76.6791277258567, "grad_norm": 2.3125, "learning_rate": 3.44126598285259e-05, "loss": 0.1109, "step": 12345 }, { "epoch": 76.68535825545172, "grad_norm": 2.53125, "learning_rate": 3.441037731883424e-05, "loss": 0.1342, "step": 12346 }, { "epoch": 76.69158878504673, "grad_norm": 2.015625, "learning_rate": 3.440809471774711e-05, "loss": 0.1456, "step": 12347 }, { "epoch": 76.69781931464175, "grad_norm": 2.078125, "learning_rate": 3.440581202528668e-05, "loss": 0.1303, "step": 12348 }, { "epoch": 76.70404984423676, "grad_norm": 2.59375, "learning_rate": 3.440352924147514e-05, "loss": 0.3177, "step": 12349 }, { "epoch": 76.71028037383178, "grad_norm": 4.34375, "learning_rate": 3.440124636633462e-05, "loss": 0.1756, "step": 12350 }, { "epoch": 76.7165109034268, "grad_norm": 1.6875, "learning_rate": 3.439896339988733e-05, "loss": 0.1185, "step": 12351 }, { "epoch": 76.72274143302181, "grad_norm": 3.703125, "learning_rate": 3.439668034215543e-05, "loss": 0.2428, "step": 12352 }, { "epoch": 76.72897196261682, "grad_norm": 2.84375, "learning_rate": 3.439439719316109e-05, "loss": 0.2187, "step": 12353 }, { "epoch": 76.73520249221184, "grad_norm": 3.453125, "learning_rate": 3.439211395292649e-05, "loss": 0.2566, "step": 12354 }, { "epoch": 76.74143302180686, "grad_norm": 2.53125, "learning_rate": 3.438983062147379e-05, "loss": 0.2847, "step": 12355 }, { "epoch": 76.74766355140187, "grad_norm": 2.59375, "learning_rate": 3.4387547198825186e-05, "loss": 0.1552, "step": 12356 }, { "epoch": 76.75389408099689, "grad_norm": 1.796875, "learning_rate": 3.4385263685002846e-05, "loss": 0.1189, "step": 12357 }, { "epoch": 76.7601246105919, "grad_norm": 2.5625, "learning_rate": 3.438298008002895e-05, "loss": 0.1324, "step": 12358 }, { "epoch": 76.76635514018692, "grad_norm": 3.3125, "learning_rate": 3.4380696383925676e-05, "loss": 0.2143, "step": 12359 }, { "epoch": 76.77258566978193, "grad_norm": 1.859375, "learning_rate": 3.4378412596715205e-05, "loss": 0.1355, "step": 12360 }, { "epoch": 76.77881619937695, "grad_norm": 2.328125, "learning_rate": 3.437612871841971e-05, "loss": 0.1347, "step": 12361 }, { "epoch": 76.78504672897196, "grad_norm": 4.625, "learning_rate": 3.437384474906139e-05, "loss": 0.2465, "step": 12362 }, { "epoch": 76.79127725856698, "grad_norm": 2.265625, "learning_rate": 3.43715606886624e-05, "loss": 0.1152, "step": 12363 }, { "epoch": 76.797507788162, "grad_norm": 1.8203125, "learning_rate": 3.436927653724495e-05, "loss": 0.1563, "step": 12364 }, { "epoch": 76.80373831775701, "grad_norm": 2.296875, "learning_rate": 3.436699229483121e-05, "loss": 0.2227, "step": 12365 }, { "epoch": 76.80996884735202, "grad_norm": 2.171875, "learning_rate": 3.4364707961443374e-05, "loss": 0.2255, "step": 12366 }, { "epoch": 76.81619937694704, "grad_norm": 2.515625, "learning_rate": 3.436242353710362e-05, "loss": 0.2405, "step": 12367 }, { "epoch": 76.82242990654206, "grad_norm": 1.1640625, "learning_rate": 3.4360139021834134e-05, "loss": 0.1067, "step": 12368 }, { "epoch": 76.82866043613707, "grad_norm": 2.453125, "learning_rate": 3.435785441565711e-05, "loss": 0.1687, "step": 12369 }, { "epoch": 76.83489096573209, "grad_norm": 1.734375, "learning_rate": 3.435556971859473e-05, "loss": 0.1189, "step": 12370 }, { "epoch": 76.8411214953271, "grad_norm": 2.9375, "learning_rate": 3.435328493066918e-05, "loss": 0.3521, "step": 12371 }, { "epoch": 76.84735202492212, "grad_norm": 2.046875, "learning_rate": 3.435100005190267e-05, "loss": 0.1356, "step": 12372 }, { "epoch": 76.85358255451713, "grad_norm": 2.53125, "learning_rate": 3.434871508231737e-05, "loss": 0.2017, "step": 12373 }, { "epoch": 76.85981308411215, "grad_norm": 2.90625, "learning_rate": 3.4346430021935486e-05, "loss": 0.2784, "step": 12374 }, { "epoch": 76.86604361370716, "grad_norm": 1.25, "learning_rate": 3.434414487077921e-05, "loss": 0.089, "step": 12375 }, { "epoch": 76.87227414330218, "grad_norm": 2.3125, "learning_rate": 3.434185962887071e-05, "loss": 0.1918, "step": 12376 }, { "epoch": 76.8785046728972, "grad_norm": 2.421875, "learning_rate": 3.4339574296232224e-05, "loss": 0.1416, "step": 12377 }, { "epoch": 76.88473520249221, "grad_norm": 2.453125, "learning_rate": 3.433728887288592e-05, "loss": 0.1647, "step": 12378 }, { "epoch": 76.89096573208722, "grad_norm": 4.3125, "learning_rate": 3.4335003358853996e-05, "loss": 0.1337, "step": 12379 }, { "epoch": 76.89719626168224, "grad_norm": 2.46875, "learning_rate": 3.4332717754158655e-05, "loss": 0.2506, "step": 12380 }, { "epoch": 76.90342679127725, "grad_norm": 1.671875, "learning_rate": 3.4330432058822095e-05, "loss": 0.1446, "step": 12381 }, { "epoch": 76.90965732087227, "grad_norm": 2.0625, "learning_rate": 3.4328146272866514e-05, "loss": 0.1463, "step": 12382 }, { "epoch": 76.91588785046729, "grad_norm": 2.578125, "learning_rate": 3.4325860396314105e-05, "loss": 0.2301, "step": 12383 }, { "epoch": 76.9221183800623, "grad_norm": 1.953125, "learning_rate": 3.432357442918708e-05, "loss": 0.1903, "step": 12384 }, { "epoch": 76.92834890965732, "grad_norm": 2.015625, "learning_rate": 3.4321288371507634e-05, "loss": 0.1559, "step": 12385 }, { "epoch": 76.93457943925233, "grad_norm": 3.640625, "learning_rate": 3.431900222329798e-05, "loss": 0.2928, "step": 12386 }, { "epoch": 76.94080996884735, "grad_norm": 4.15625, "learning_rate": 3.431671598458031e-05, "loss": 0.2795, "step": 12387 }, { "epoch": 76.94704049844236, "grad_norm": 3.375, "learning_rate": 3.4314429655376826e-05, "loss": 0.2356, "step": 12388 }, { "epoch": 76.95327102803738, "grad_norm": 2.59375, "learning_rate": 3.431214323570975e-05, "loss": 0.2081, "step": 12389 }, { "epoch": 76.9595015576324, "grad_norm": 3.578125, "learning_rate": 3.430985672560126e-05, "loss": 0.2625, "step": 12390 }, { "epoch": 76.96573208722741, "grad_norm": 2.75, "learning_rate": 3.4307570125073605e-05, "loss": 0.1125, "step": 12391 }, { "epoch": 76.97196261682242, "grad_norm": 1.234375, "learning_rate": 3.430528343414895e-05, "loss": 0.1128, "step": 12392 }, { "epoch": 76.97819314641744, "grad_norm": 2.625, "learning_rate": 3.430299665284953e-05, "loss": 0.2261, "step": 12393 }, { "epoch": 76.98442367601245, "grad_norm": 1.953125, "learning_rate": 3.430070978119754e-05, "loss": 0.1245, "step": 12394 }, { "epoch": 76.99065420560747, "grad_norm": 1.9921875, "learning_rate": 3.42984228192152e-05, "loss": 0.1882, "step": 12395 }, { "epoch": 76.99688473520249, "grad_norm": 2.6875, "learning_rate": 3.429613576692472e-05, "loss": 0.1521, "step": 12396 }, { "epoch": 77.0, "grad_norm": 2.1875, "learning_rate": 3.4293848624348316e-05, "loss": 0.1274, "step": 12397 }, { "epoch": 77.00623052959502, "grad_norm": 2.953125, "learning_rate": 3.4291561391508185e-05, "loss": 0.2284, "step": 12398 }, { "epoch": 77.01246105919003, "grad_norm": 2.875, "learning_rate": 3.428927406842657e-05, "loss": 0.2049, "step": 12399 }, { "epoch": 77.01869158878505, "grad_norm": 2.953125, "learning_rate": 3.428698665512565e-05, "loss": 0.1637, "step": 12400 }, { "epoch": 77.02492211838006, "grad_norm": 1.8515625, "learning_rate": 3.428469915162767e-05, "loss": 0.1486, "step": 12401 }, { "epoch": 77.03115264797508, "grad_norm": 2.84375, "learning_rate": 3.428241155795484e-05, "loss": 0.2027, "step": 12402 }, { "epoch": 77.03738317757009, "grad_norm": 3.671875, "learning_rate": 3.428012387412936e-05, "loss": 0.2405, "step": 12403 }, { "epoch": 77.04361370716511, "grad_norm": 2.578125, "learning_rate": 3.4277836100173475e-05, "loss": 0.1915, "step": 12404 }, { "epoch": 77.04984423676012, "grad_norm": 3.6875, "learning_rate": 3.4275548236109376e-05, "loss": 0.1774, "step": 12405 }, { "epoch": 77.05607476635514, "grad_norm": 3.328125, "learning_rate": 3.4273260281959304e-05, "loss": 0.2486, "step": 12406 }, { "epoch": 77.06230529595015, "grad_norm": 1.2578125, "learning_rate": 3.427097223774548e-05, "loss": 0.1254, "step": 12407 }, { "epoch": 77.06853582554517, "grad_norm": 2.9375, "learning_rate": 3.4268684103490115e-05, "loss": 0.1997, "step": 12408 }, { "epoch": 77.07476635514018, "grad_norm": 2.484375, "learning_rate": 3.426639587921544e-05, "loss": 0.1682, "step": 12409 }, { "epoch": 77.0809968847352, "grad_norm": 3.6875, "learning_rate": 3.426410756494368e-05, "loss": 0.2646, "step": 12410 }, { "epoch": 77.08722741433021, "grad_norm": 2.21875, "learning_rate": 3.426181916069705e-05, "loss": 0.1723, "step": 12411 }, { "epoch": 77.09345794392523, "grad_norm": 1.7421875, "learning_rate": 3.425953066649778e-05, "loss": 0.1137, "step": 12412 }, { "epoch": 77.09968847352025, "grad_norm": 2.265625, "learning_rate": 3.42572420823681e-05, "loss": 0.1547, "step": 12413 }, { "epoch": 77.10591900311526, "grad_norm": 3.546875, "learning_rate": 3.425495340833024e-05, "loss": 0.1977, "step": 12414 }, { "epoch": 77.11214953271028, "grad_norm": 1.484375, "learning_rate": 3.425266464440641e-05, "loss": 0.1278, "step": 12415 }, { "epoch": 77.11838006230529, "grad_norm": 3.15625, "learning_rate": 3.4250375790618864e-05, "loss": 0.2965, "step": 12416 }, { "epoch": 77.1246105919003, "grad_norm": 3.984375, "learning_rate": 3.424808684698981e-05, "loss": 0.4026, "step": 12417 }, { "epoch": 77.13084112149532, "grad_norm": 2.359375, "learning_rate": 3.424579781354148e-05, "loss": 0.107, "step": 12418 }, { "epoch": 77.13707165109034, "grad_norm": 3.84375, "learning_rate": 3.4243508690296135e-05, "loss": 0.2512, "step": 12419 }, { "epoch": 77.14330218068535, "grad_norm": 1.8203125, "learning_rate": 3.424121947727597e-05, "loss": 0.1475, "step": 12420 }, { "epoch": 77.14953271028037, "grad_norm": 2.140625, "learning_rate": 3.4238930174503245e-05, "loss": 0.2096, "step": 12421 }, { "epoch": 77.15576323987538, "grad_norm": 2.390625, "learning_rate": 3.423664078200017e-05, "loss": 0.126, "step": 12422 }, { "epoch": 77.1619937694704, "grad_norm": 2.484375, "learning_rate": 3.423435129978901e-05, "loss": 0.2226, "step": 12423 }, { "epoch": 77.16822429906541, "grad_norm": 3.59375, "learning_rate": 3.423206172789197e-05, "loss": 0.177, "step": 12424 }, { "epoch": 77.17445482866043, "grad_norm": 3.765625, "learning_rate": 3.4229772066331305e-05, "loss": 0.2879, "step": 12425 }, { "epoch": 77.18068535825545, "grad_norm": 2.515625, "learning_rate": 3.422748231512925e-05, "loss": 0.1145, "step": 12426 }, { "epoch": 77.18691588785046, "grad_norm": 3.234375, "learning_rate": 3.422519247430804e-05, "loss": 0.2393, "step": 12427 }, { "epoch": 77.19314641744548, "grad_norm": 2.828125, "learning_rate": 3.4222902543889924e-05, "loss": 0.2098, "step": 12428 }, { "epoch": 77.19937694704049, "grad_norm": 3.0, "learning_rate": 3.422061252389713e-05, "loss": 0.1666, "step": 12429 }, { "epoch": 77.20560747663552, "grad_norm": 3.265625, "learning_rate": 3.42183224143519e-05, "loss": 0.1281, "step": 12430 }, { "epoch": 77.21183800623054, "grad_norm": 2.359375, "learning_rate": 3.4216032215276486e-05, "loss": 0.1101, "step": 12431 }, { "epoch": 77.21806853582555, "grad_norm": 2.703125, "learning_rate": 3.421374192669312e-05, "loss": 0.1936, "step": 12432 }, { "epoch": 77.22429906542057, "grad_norm": 3.625, "learning_rate": 3.421145154862405e-05, "loss": 0.2992, "step": 12433 }, { "epoch": 77.23052959501558, "grad_norm": 3.296875, "learning_rate": 3.420916108109153e-05, "loss": 0.0929, "step": 12434 }, { "epoch": 77.2367601246106, "grad_norm": 1.96875, "learning_rate": 3.420687052411779e-05, "loss": 0.0929, "step": 12435 }, { "epoch": 77.24299065420561, "grad_norm": 3.828125, "learning_rate": 3.420457987772508e-05, "loss": 0.165, "step": 12436 }, { "epoch": 77.24922118380063, "grad_norm": 1.953125, "learning_rate": 3.4202289141935664e-05, "loss": 0.1524, "step": 12437 }, { "epoch": 77.25545171339564, "grad_norm": 3.09375, "learning_rate": 3.419999831677176e-05, "loss": 0.1595, "step": 12438 }, { "epoch": 77.26168224299066, "grad_norm": 3.265625, "learning_rate": 3.419770740225564e-05, "loss": 0.1186, "step": 12439 }, { "epoch": 77.26791277258567, "grad_norm": 2.015625, "learning_rate": 3.419541639840955e-05, "loss": 0.1463, "step": 12440 }, { "epoch": 77.27414330218069, "grad_norm": 1.6171875, "learning_rate": 3.419312530525573e-05, "loss": 0.1089, "step": 12441 }, { "epoch": 77.2803738317757, "grad_norm": 2.859375, "learning_rate": 3.419083412281645e-05, "loss": 0.183, "step": 12442 }, { "epoch": 77.28660436137072, "grad_norm": 3.1875, "learning_rate": 3.418854285111394e-05, "loss": 0.3166, "step": 12443 }, { "epoch": 77.29283489096574, "grad_norm": 1.6875, "learning_rate": 3.418625149017047e-05, "loss": 0.1314, "step": 12444 }, { "epoch": 77.29906542056075, "grad_norm": 3.125, "learning_rate": 3.4183960040008294e-05, "loss": 0.2668, "step": 12445 }, { "epoch": 77.30529595015577, "grad_norm": 3.015625, "learning_rate": 3.4181668500649656e-05, "loss": 0.1376, "step": 12446 }, { "epoch": 77.31152647975078, "grad_norm": 1.640625, "learning_rate": 3.417937687211682e-05, "loss": 0.1045, "step": 12447 }, { "epoch": 77.3177570093458, "grad_norm": 2.875, "learning_rate": 3.4177085154432044e-05, "loss": 0.2542, "step": 12448 }, { "epoch": 77.32398753894081, "grad_norm": 3.96875, "learning_rate": 3.417479334761758e-05, "loss": 0.1775, "step": 12449 }, { "epoch": 77.33021806853583, "grad_norm": 2.953125, "learning_rate": 3.4172501451695696e-05, "loss": 0.2464, "step": 12450 }, { "epoch": 77.33644859813084, "grad_norm": 2.46875, "learning_rate": 3.4170209466688637e-05, "loss": 0.2162, "step": 12451 }, { "epoch": 77.34267912772586, "grad_norm": 5.375, "learning_rate": 3.416791739261867e-05, "loss": 0.1853, "step": 12452 }, { "epoch": 77.34890965732087, "grad_norm": 1.8984375, "learning_rate": 3.4165625229508055e-05, "loss": 0.1131, "step": 12453 }, { "epoch": 77.35514018691589, "grad_norm": 1.4140625, "learning_rate": 3.416333297737906e-05, "loss": 0.1207, "step": 12454 }, { "epoch": 77.3613707165109, "grad_norm": 1.9296875, "learning_rate": 3.416104063625394e-05, "loss": 0.119, "step": 12455 }, { "epoch": 77.36760124610592, "grad_norm": 1.625, "learning_rate": 3.415874820615496e-05, "loss": 0.1634, "step": 12456 }, { "epoch": 77.37383177570094, "grad_norm": 2.265625, "learning_rate": 3.4156455687104386e-05, "loss": 0.1804, "step": 12457 }, { "epoch": 77.38006230529595, "grad_norm": 1.59375, "learning_rate": 3.4154163079124496e-05, "loss": 0.1463, "step": 12458 }, { "epoch": 77.38629283489097, "grad_norm": 2.421875, "learning_rate": 3.4151870382237535e-05, "loss": 0.2715, "step": 12459 }, { "epoch": 77.39252336448598, "grad_norm": 2.46875, "learning_rate": 3.414957759646578e-05, "loss": 0.1765, "step": 12460 }, { "epoch": 77.398753894081, "grad_norm": 1.859375, "learning_rate": 3.4147284721831503e-05, "loss": 0.1659, "step": 12461 }, { "epoch": 77.40498442367601, "grad_norm": 1.4140625, "learning_rate": 3.414499175835697e-05, "loss": 0.1423, "step": 12462 }, { "epoch": 77.41121495327103, "grad_norm": 2.40625, "learning_rate": 3.4142698706064434e-05, "loss": 0.1863, "step": 12463 }, { "epoch": 77.41744548286604, "grad_norm": 1.03125, "learning_rate": 3.4140405564976197e-05, "loss": 0.0944, "step": 12464 }, { "epoch": 77.42367601246106, "grad_norm": 2.0625, "learning_rate": 3.413811233511451e-05, "loss": 0.202, "step": 12465 }, { "epoch": 77.42990654205607, "grad_norm": 2.71875, "learning_rate": 3.413581901650164e-05, "loss": 0.2695, "step": 12466 }, { "epoch": 77.43613707165109, "grad_norm": 1.6171875, "learning_rate": 3.413352560915988e-05, "loss": 0.0754, "step": 12467 }, { "epoch": 77.4423676012461, "grad_norm": 2.03125, "learning_rate": 3.413123211311149e-05, "loss": 0.0826, "step": 12468 }, { "epoch": 77.44859813084112, "grad_norm": 2.078125, "learning_rate": 3.412893852837875e-05, "loss": 0.1084, "step": 12469 }, { "epoch": 77.45482866043614, "grad_norm": 3.0, "learning_rate": 3.412664485498393e-05, "loss": 0.2041, "step": 12470 }, { "epoch": 77.46105919003115, "grad_norm": 2.984375, "learning_rate": 3.412435109294932e-05, "loss": 0.1892, "step": 12471 }, { "epoch": 77.46728971962617, "grad_norm": 2.96875, "learning_rate": 3.412205724229717e-05, "loss": 0.2071, "step": 12472 }, { "epoch": 77.47352024922118, "grad_norm": 2.671875, "learning_rate": 3.41197633030498e-05, "loss": 0.311, "step": 12473 }, { "epoch": 77.4797507788162, "grad_norm": 2.53125, "learning_rate": 3.411746927522944e-05, "loss": 0.2, "step": 12474 }, { "epoch": 77.48598130841121, "grad_norm": 2.625, "learning_rate": 3.4115175158858413e-05, "loss": 0.1769, "step": 12475 }, { "epoch": 77.49221183800623, "grad_norm": 2.984375, "learning_rate": 3.411288095395898e-05, "loss": 0.2215, "step": 12476 }, { "epoch": 77.49844236760124, "grad_norm": 2.234375, "learning_rate": 3.4110586660553426e-05, "loss": 0.1219, "step": 12477 }, { "epoch": 77.50467289719626, "grad_norm": 1.9296875, "learning_rate": 3.410829227866403e-05, "loss": 0.1361, "step": 12478 }, { "epoch": 77.51090342679127, "grad_norm": 3.6875, "learning_rate": 3.4105997808313086e-05, "loss": 0.1549, "step": 12479 }, { "epoch": 77.51713395638629, "grad_norm": 2.671875, "learning_rate": 3.410370324952287e-05, "loss": 0.2409, "step": 12480 }, { "epoch": 77.5233644859813, "grad_norm": 2.109375, "learning_rate": 3.4101408602315656e-05, "loss": 0.1432, "step": 12481 }, { "epoch": 77.52959501557632, "grad_norm": 1.984375, "learning_rate": 3.409911386671375e-05, "loss": 0.1184, "step": 12482 }, { "epoch": 77.53582554517133, "grad_norm": 2.5, "learning_rate": 3.409681904273943e-05, "loss": 0.1695, "step": 12483 }, { "epoch": 77.54205607476635, "grad_norm": 2.78125, "learning_rate": 3.4094524130414996e-05, "loss": 0.2315, "step": 12484 }, { "epoch": 77.54828660436137, "grad_norm": 3.671875, "learning_rate": 3.4092229129762715e-05, "loss": 0.2299, "step": 12485 }, { "epoch": 77.55451713395638, "grad_norm": 1.890625, "learning_rate": 3.408993404080489e-05, "loss": 0.1586, "step": 12486 }, { "epoch": 77.5607476635514, "grad_norm": 2.8125, "learning_rate": 3.408763886356381e-05, "loss": 0.2757, "step": 12487 }, { "epoch": 77.56697819314641, "grad_norm": 2.921875, "learning_rate": 3.4085343598061756e-05, "loss": 0.2181, "step": 12488 }, { "epoch": 77.57320872274143, "grad_norm": 2.15625, "learning_rate": 3.408304824432104e-05, "loss": 0.1954, "step": 12489 }, { "epoch": 77.57943925233644, "grad_norm": 2.515625, "learning_rate": 3.4080752802363936e-05, "loss": 0.1468, "step": 12490 }, { "epoch": 77.58566978193146, "grad_norm": 2.0625, "learning_rate": 3.4078457272212745e-05, "loss": 0.1096, "step": 12491 }, { "epoch": 77.59190031152647, "grad_norm": 1.6796875, "learning_rate": 3.407616165388976e-05, "loss": 0.1429, "step": 12492 }, { "epoch": 77.59813084112149, "grad_norm": 2.96875, "learning_rate": 3.407386594741729e-05, "loss": 0.2024, "step": 12493 }, { "epoch": 77.60436137071652, "grad_norm": 3.84375, "learning_rate": 3.407157015281761e-05, "loss": 0.2282, "step": 12494 }, { "epoch": 77.61059190031153, "grad_norm": 1.640625, "learning_rate": 3.406927427011303e-05, "loss": 0.1242, "step": 12495 }, { "epoch": 77.61682242990655, "grad_norm": 3.6875, "learning_rate": 3.4066978299325856e-05, "loss": 0.1468, "step": 12496 }, { "epoch": 77.62305295950156, "grad_norm": 1.96875, "learning_rate": 3.406468224047836e-05, "loss": 0.142, "step": 12497 }, { "epoch": 77.62928348909658, "grad_norm": 2.546875, "learning_rate": 3.406238609359287e-05, "loss": 0.1405, "step": 12498 }, { "epoch": 77.6355140186916, "grad_norm": 1.7890625, "learning_rate": 3.406008985869167e-05, "loss": 0.1145, "step": 12499 }, { "epoch": 77.64174454828661, "grad_norm": 1.90625, "learning_rate": 3.4057793535797066e-05, "loss": 0.137, "step": 12500 }, { "epoch": 77.64797507788163, "grad_norm": 2.484375, "learning_rate": 3.4055497124931354e-05, "loss": 0.1812, "step": 12501 }, { "epoch": 77.65420560747664, "grad_norm": 3.109375, "learning_rate": 3.405320062611686e-05, "loss": 0.1855, "step": 12502 }, { "epoch": 77.66043613707166, "grad_norm": 3.71875, "learning_rate": 3.405090403937586e-05, "loss": 0.1273, "step": 12503 }, { "epoch": 77.66666666666667, "grad_norm": 1.8828125, "learning_rate": 3.404860736473067e-05, "loss": 0.0997, "step": 12504 }, { "epoch": 77.67289719626169, "grad_norm": 2.515625, "learning_rate": 3.4046310602203604e-05, "loss": 0.1788, "step": 12505 }, { "epoch": 77.6791277258567, "grad_norm": 2.296875, "learning_rate": 3.4044013751816954e-05, "loss": 0.2032, "step": 12506 }, { "epoch": 77.68535825545172, "grad_norm": 1.1484375, "learning_rate": 3.4041716813593035e-05, "loss": 0.1103, "step": 12507 }, { "epoch": 77.69158878504673, "grad_norm": 3.453125, "learning_rate": 3.403941978755416e-05, "loss": 0.3187, "step": 12508 }, { "epoch": 77.69781931464175, "grad_norm": 1.7265625, "learning_rate": 3.4037122673722634e-05, "loss": 0.1112, "step": 12509 }, { "epoch": 77.70404984423676, "grad_norm": 1.984375, "learning_rate": 3.403482547212076e-05, "loss": 0.1531, "step": 12510 }, { "epoch": 77.71028037383178, "grad_norm": 1.6328125, "learning_rate": 3.403252818277086e-05, "loss": 0.1004, "step": 12511 }, { "epoch": 77.7165109034268, "grad_norm": 2.4375, "learning_rate": 3.403023080569523e-05, "loss": 0.358, "step": 12512 }, { "epoch": 77.72274143302181, "grad_norm": 2.53125, "learning_rate": 3.402793334091621e-05, "loss": 0.1503, "step": 12513 }, { "epoch": 77.72897196261682, "grad_norm": 2.875, "learning_rate": 3.402563578845608e-05, "loss": 0.1978, "step": 12514 }, { "epoch": 77.73520249221184, "grad_norm": 2.9375, "learning_rate": 3.402333814833719e-05, "loss": 0.1453, "step": 12515 }, { "epoch": 77.74143302180686, "grad_norm": 2.0, "learning_rate": 3.402104042058182e-05, "loss": 0.1173, "step": 12516 }, { "epoch": 77.74766355140187, "grad_norm": 2.09375, "learning_rate": 3.401874260521231e-05, "loss": 0.1211, "step": 12517 }, { "epoch": 77.75389408099689, "grad_norm": 2.59375, "learning_rate": 3.401644470225096e-05, "loss": 0.2349, "step": 12518 }, { "epoch": 77.7601246105919, "grad_norm": 2.53125, "learning_rate": 3.401414671172011e-05, "loss": 0.2057, "step": 12519 }, { "epoch": 77.76635514018692, "grad_norm": 2.375, "learning_rate": 3.4011848633642054e-05, "loss": 0.1212, "step": 12520 }, { "epoch": 77.77258566978193, "grad_norm": 2.578125, "learning_rate": 3.400955046803913e-05, "loss": 0.1525, "step": 12521 }, { "epoch": 77.77881619937695, "grad_norm": 2.515625, "learning_rate": 3.4007252214933646e-05, "loss": 0.2569, "step": 12522 }, { "epoch": 77.78504672897196, "grad_norm": 2.78125, "learning_rate": 3.4004953874347925e-05, "loss": 0.1759, "step": 12523 }, { "epoch": 77.79127725856698, "grad_norm": 0.99609375, "learning_rate": 3.40026554463043e-05, "loss": 0.1009, "step": 12524 }, { "epoch": 77.797507788162, "grad_norm": 2.578125, "learning_rate": 3.4000356930825085e-05, "loss": 0.2185, "step": 12525 }, { "epoch": 77.80373831775701, "grad_norm": 2.875, "learning_rate": 3.39980583279326e-05, "loss": 0.1732, "step": 12526 }, { "epoch": 77.80996884735202, "grad_norm": 1.46875, "learning_rate": 3.399575963764917e-05, "loss": 0.1119, "step": 12527 }, { "epoch": 77.81619937694704, "grad_norm": 1.8984375, "learning_rate": 3.399346085999714e-05, "loss": 0.173, "step": 12528 }, { "epoch": 77.82242990654206, "grad_norm": 4.0625, "learning_rate": 3.399116199499881e-05, "loss": 0.2553, "step": 12529 }, { "epoch": 77.82866043613707, "grad_norm": 2.203125, "learning_rate": 3.3988863042676516e-05, "loss": 0.1324, "step": 12530 }, { "epoch": 77.83489096573209, "grad_norm": 2.359375, "learning_rate": 3.398656400305259e-05, "loss": 0.1387, "step": 12531 }, { "epoch": 77.8411214953271, "grad_norm": 2.21875, "learning_rate": 3.398426487614936e-05, "loss": 0.3028, "step": 12532 }, { "epoch": 77.84735202492212, "grad_norm": 1.6484375, "learning_rate": 3.3981965661989146e-05, "loss": 0.1429, "step": 12533 }, { "epoch": 77.85358255451713, "grad_norm": 2.484375, "learning_rate": 3.3979666360594295e-05, "loss": 0.1173, "step": 12534 }, { "epoch": 77.85981308411215, "grad_norm": 1.375, "learning_rate": 3.3977366971987124e-05, "loss": 0.1529, "step": 12535 }, { "epoch": 77.86604361370716, "grad_norm": 2.90625, "learning_rate": 3.3975067496189965e-05, "loss": 0.1929, "step": 12536 }, { "epoch": 77.87227414330218, "grad_norm": 2.125, "learning_rate": 3.397276793322517e-05, "loss": 0.1616, "step": 12537 }, { "epoch": 77.8785046728972, "grad_norm": 1.4609375, "learning_rate": 3.3970468283115045e-05, "loss": 0.1085, "step": 12538 }, { "epoch": 77.88473520249221, "grad_norm": 2.875, "learning_rate": 3.3968168545881954e-05, "loss": 0.3827, "step": 12539 }, { "epoch": 77.89096573208722, "grad_norm": 3.078125, "learning_rate": 3.396586872154821e-05, "loss": 0.1479, "step": 12540 }, { "epoch": 77.89719626168224, "grad_norm": 2.578125, "learning_rate": 3.396356881013615e-05, "loss": 0.2326, "step": 12541 }, { "epoch": 77.90342679127725, "grad_norm": 1.5234375, "learning_rate": 3.3961268811668124e-05, "loss": 0.1221, "step": 12542 }, { "epoch": 77.90965732087227, "grad_norm": 1.625, "learning_rate": 3.3958968726166466e-05, "loss": 0.1436, "step": 12543 }, { "epoch": 77.91588785046729, "grad_norm": 1.203125, "learning_rate": 3.395666855365351e-05, "loss": 0.1212, "step": 12544 }, { "epoch": 77.9221183800623, "grad_norm": 1.65625, "learning_rate": 3.39543682941516e-05, "loss": 0.1304, "step": 12545 }, { "epoch": 77.92834890965732, "grad_norm": 2.015625, "learning_rate": 3.395206794768307e-05, "loss": 0.1644, "step": 12546 }, { "epoch": 77.93457943925233, "grad_norm": 2.65625, "learning_rate": 3.394976751427028e-05, "loss": 0.2279, "step": 12547 }, { "epoch": 77.94080996884735, "grad_norm": 2.0, "learning_rate": 3.394746699393555e-05, "loss": 0.1944, "step": 12548 }, { "epoch": 77.94704049844236, "grad_norm": 2.171875, "learning_rate": 3.3945166386701225e-05, "loss": 0.2454, "step": 12549 }, { "epoch": 77.95327102803738, "grad_norm": 2.1875, "learning_rate": 3.394286569258967e-05, "loss": 0.169, "step": 12550 }, { "epoch": 77.9595015576324, "grad_norm": 1.140625, "learning_rate": 3.394056491162321e-05, "loss": 0.0921, "step": 12551 }, { "epoch": 77.96573208722741, "grad_norm": 1.7890625, "learning_rate": 3.393826404382419e-05, "loss": 0.154, "step": 12552 }, { "epoch": 77.97196261682242, "grad_norm": 1.5078125, "learning_rate": 3.393596308921497e-05, "loss": 0.1547, "step": 12553 }, { "epoch": 77.97819314641744, "grad_norm": 1.6171875, "learning_rate": 3.393366204781789e-05, "loss": 0.2178, "step": 12554 }, { "epoch": 77.98442367601245, "grad_norm": 3.671875, "learning_rate": 3.393136091965529e-05, "loss": 0.2416, "step": 12555 }, { "epoch": 77.99065420560747, "grad_norm": 1.7421875, "learning_rate": 3.392905970474955e-05, "loss": 0.124, "step": 12556 }, { "epoch": 77.99688473520249, "grad_norm": 2.0625, "learning_rate": 3.392675840312297e-05, "loss": 0.1535, "step": 12557 }, { "epoch": 78.0, "grad_norm": 1.515625, "learning_rate": 3.3924457014797936e-05, "loss": 0.1089, "step": 12558 }, { "epoch": 78.00623052959502, "grad_norm": 3.1875, "learning_rate": 3.392215553979679e-05, "loss": 0.1367, "step": 12559 }, { "epoch": 78.01246105919003, "grad_norm": 2.09375, "learning_rate": 3.39198539781419e-05, "loss": 0.1751, "step": 12560 }, { "epoch": 78.01869158878505, "grad_norm": 3.203125, "learning_rate": 3.391755232985559e-05, "loss": 0.3433, "step": 12561 }, { "epoch": 78.02492211838006, "grad_norm": 1.6640625, "learning_rate": 3.391525059496023e-05, "loss": 0.1351, "step": 12562 }, { "epoch": 78.03115264797508, "grad_norm": 1.9609375, "learning_rate": 3.3912948773478195e-05, "loss": 0.1648, "step": 12563 }, { "epoch": 78.03738317757009, "grad_norm": 1.5703125, "learning_rate": 3.3910646865431797e-05, "loss": 0.1165, "step": 12564 }, { "epoch": 78.04361370716511, "grad_norm": 2.140625, "learning_rate": 3.3908344870843425e-05, "loss": 0.1315, "step": 12565 }, { "epoch": 78.04984423676012, "grad_norm": 3.6875, "learning_rate": 3.390604278973543e-05, "loss": 0.3214, "step": 12566 }, { "epoch": 78.05607476635514, "grad_norm": 2.21875, "learning_rate": 3.3903740622130156e-05, "loss": 0.1509, "step": 12567 }, { "epoch": 78.06230529595015, "grad_norm": 2.59375, "learning_rate": 3.3901438368049976e-05, "loss": 0.1727, "step": 12568 }, { "epoch": 78.06853582554517, "grad_norm": 3.484375, "learning_rate": 3.3899136027517255e-05, "loss": 0.2109, "step": 12569 }, { "epoch": 78.07476635514018, "grad_norm": 2.296875, "learning_rate": 3.3896833600554336e-05, "loss": 0.1201, "step": 12570 }, { "epoch": 78.0809968847352, "grad_norm": 4.125, "learning_rate": 3.3894531087183594e-05, "loss": 0.254, "step": 12571 }, { "epoch": 78.08722741433021, "grad_norm": 2.234375, "learning_rate": 3.389222848742739e-05, "loss": 0.1512, "step": 12572 }, { "epoch": 78.09345794392523, "grad_norm": 2.0, "learning_rate": 3.388992580130809e-05, "loss": 0.1047, "step": 12573 }, { "epoch": 78.09968847352025, "grad_norm": 2.453125, "learning_rate": 3.3887623028848045e-05, "loss": 0.138, "step": 12574 }, { "epoch": 78.10591900311526, "grad_norm": 3.65625, "learning_rate": 3.3885320170069634e-05, "loss": 0.1465, "step": 12575 }, { "epoch": 78.11214953271028, "grad_norm": 2.890625, "learning_rate": 3.3883017224995214e-05, "loss": 0.14, "step": 12576 }, { "epoch": 78.11838006230529, "grad_norm": 2.875, "learning_rate": 3.3880714193647155e-05, "loss": 0.2974, "step": 12577 }, { "epoch": 78.1246105919003, "grad_norm": 2.03125, "learning_rate": 3.3878411076047825e-05, "loss": 0.1285, "step": 12578 }, { "epoch": 78.13084112149532, "grad_norm": 2.265625, "learning_rate": 3.38761078722196e-05, "loss": 0.2002, "step": 12579 }, { "epoch": 78.13707165109034, "grad_norm": 2.328125, "learning_rate": 3.387380458218483e-05, "loss": 0.1607, "step": 12580 }, { "epoch": 78.14330218068535, "grad_norm": 3.609375, "learning_rate": 3.3871501205965906e-05, "loss": 0.1886, "step": 12581 }, { "epoch": 78.14953271028037, "grad_norm": 1.7890625, "learning_rate": 3.386919774358518e-05, "loss": 0.1519, "step": 12582 }, { "epoch": 78.15576323987538, "grad_norm": 2.140625, "learning_rate": 3.3866894195065036e-05, "loss": 0.1998, "step": 12583 }, { "epoch": 78.1619937694704, "grad_norm": 4.0, "learning_rate": 3.386459056042784e-05, "loss": 0.3178, "step": 12584 }, { "epoch": 78.16822429906541, "grad_norm": 2.890625, "learning_rate": 3.386228683969598e-05, "loss": 0.16, "step": 12585 }, { "epoch": 78.17445482866043, "grad_norm": 4.21875, "learning_rate": 3.38599830328918e-05, "loss": 0.1683, "step": 12586 }, { "epoch": 78.18068535825545, "grad_norm": 4.3125, "learning_rate": 3.385767914003771e-05, "loss": 0.2952, "step": 12587 }, { "epoch": 78.18691588785046, "grad_norm": 3.359375, "learning_rate": 3.385537516115606e-05, "loss": 0.1745, "step": 12588 }, { "epoch": 78.19314641744548, "grad_norm": 1.8203125, "learning_rate": 3.385307109626924e-05, "loss": 0.12, "step": 12589 }, { "epoch": 78.19937694704049, "grad_norm": 3.828125, "learning_rate": 3.385076694539962e-05, "loss": 0.2726, "step": 12590 }, { "epoch": 78.20560747663552, "grad_norm": 3.84375, "learning_rate": 3.384846270856959e-05, "loss": 0.1128, "step": 12591 }, { "epoch": 78.21183800623054, "grad_norm": 2.140625, "learning_rate": 3.384615838580151e-05, "loss": 0.1645, "step": 12592 }, { "epoch": 78.21806853582555, "grad_norm": 2.96875, "learning_rate": 3.384385397711778e-05, "loss": 0.1665, "step": 12593 }, { "epoch": 78.22429906542057, "grad_norm": 1.5390625, "learning_rate": 3.384154948254078e-05, "loss": 0.138, "step": 12594 }, { "epoch": 78.23052959501558, "grad_norm": 3.671875, "learning_rate": 3.383924490209287e-05, "loss": 0.1954, "step": 12595 }, { "epoch": 78.2367601246106, "grad_norm": 3.125, "learning_rate": 3.3836940235796446e-05, "loss": 0.1601, "step": 12596 }, { "epoch": 78.24299065420561, "grad_norm": 2.96875, "learning_rate": 3.38346354836739e-05, "loss": 0.2539, "step": 12597 }, { "epoch": 78.24922118380063, "grad_norm": 1.7421875, "learning_rate": 3.383233064574761e-05, "loss": 0.1076, "step": 12598 }, { "epoch": 78.25545171339564, "grad_norm": 3.03125, "learning_rate": 3.383002572203995e-05, "loss": 0.2448, "step": 12599 }, { "epoch": 78.26168224299066, "grad_norm": 1.625, "learning_rate": 3.382772071257332e-05, "loss": 0.114, "step": 12600 }, { "epoch": 78.26791277258567, "grad_norm": 2.578125, "learning_rate": 3.3825415617370105e-05, "loss": 0.3361, "step": 12601 }, { "epoch": 78.27414330218069, "grad_norm": 2.90625, "learning_rate": 3.3823110436452685e-05, "loss": 0.2112, "step": 12602 }, { "epoch": 78.2803738317757, "grad_norm": 2.4375, "learning_rate": 3.382080516984345e-05, "loss": 0.193, "step": 12603 }, { "epoch": 78.28660436137072, "grad_norm": 1.6875, "learning_rate": 3.38184998175648e-05, "loss": 0.1678, "step": 12604 }, { "epoch": 78.29283489096574, "grad_norm": 1.828125, "learning_rate": 3.381619437963911e-05, "loss": 0.1554, "step": 12605 }, { "epoch": 78.29906542056075, "grad_norm": 2.953125, "learning_rate": 3.381388885608878e-05, "loss": 0.2483, "step": 12606 }, { "epoch": 78.30529595015577, "grad_norm": 1.921875, "learning_rate": 3.381158324693619e-05, "loss": 0.1496, "step": 12607 }, { "epoch": 78.31152647975078, "grad_norm": 2.859375, "learning_rate": 3.380927755220376e-05, "loss": 0.1433, "step": 12608 }, { "epoch": 78.3177570093458, "grad_norm": 2.4375, "learning_rate": 3.380697177191386e-05, "loss": 0.2323, "step": 12609 }, { "epoch": 78.32398753894081, "grad_norm": 2.390625, "learning_rate": 3.380466590608889e-05, "loss": 0.2012, "step": 12610 }, { "epoch": 78.33021806853583, "grad_norm": 3.0, "learning_rate": 3.380235995475124e-05, "loss": 0.1949, "step": 12611 }, { "epoch": 78.33644859813084, "grad_norm": 1.671875, "learning_rate": 3.380005391792331e-05, "loss": 0.1465, "step": 12612 }, { "epoch": 78.34267912772586, "grad_norm": 3.046875, "learning_rate": 3.37977477956275e-05, "loss": 0.1862, "step": 12613 }, { "epoch": 78.34890965732087, "grad_norm": 1.46875, "learning_rate": 3.379544158788621e-05, "loss": 0.1257, "step": 12614 }, { "epoch": 78.35514018691589, "grad_norm": 2.75, "learning_rate": 3.379313529472182e-05, "loss": 0.1799, "step": 12615 }, { "epoch": 78.3613707165109, "grad_norm": 3.03125, "learning_rate": 3.379082891615675e-05, "loss": 0.2791, "step": 12616 }, { "epoch": 78.36760124610592, "grad_norm": 1.8359375, "learning_rate": 3.37885224522134e-05, "loss": 0.1516, "step": 12617 }, { "epoch": 78.37383177570094, "grad_norm": 1.2109375, "learning_rate": 3.378621590291415e-05, "loss": 0.1319, "step": 12618 }, { "epoch": 78.38006230529595, "grad_norm": 1.375, "learning_rate": 3.378390926828142e-05, "loss": 0.1127, "step": 12619 }, { "epoch": 78.38629283489097, "grad_norm": 2.0625, "learning_rate": 3.378160254833761e-05, "loss": 0.11, "step": 12620 }, { "epoch": 78.39252336448598, "grad_norm": 2.5, "learning_rate": 3.377929574310512e-05, "loss": 0.2003, "step": 12621 }, { "epoch": 78.398753894081, "grad_norm": 1.171875, "learning_rate": 3.377698885260635e-05, "loss": 0.1061, "step": 12622 }, { "epoch": 78.40498442367601, "grad_norm": 1.3125, "learning_rate": 3.3774681876863715e-05, "loss": 0.1115, "step": 12623 }, { "epoch": 78.41121495327103, "grad_norm": 2.296875, "learning_rate": 3.377237481589961e-05, "loss": 0.1552, "step": 12624 }, { "epoch": 78.41744548286604, "grad_norm": 2.453125, "learning_rate": 3.377006766973645e-05, "loss": 0.1694, "step": 12625 }, { "epoch": 78.42367601246106, "grad_norm": 3.0, "learning_rate": 3.376776043839664e-05, "loss": 0.2352, "step": 12626 }, { "epoch": 78.42990654205607, "grad_norm": 1.8515625, "learning_rate": 3.376545312190259e-05, "loss": 0.1048, "step": 12627 }, { "epoch": 78.43613707165109, "grad_norm": 2.28125, "learning_rate": 3.376314572027671e-05, "loss": 0.1329, "step": 12628 }, { "epoch": 78.4423676012461, "grad_norm": 2.96875, "learning_rate": 3.37608382335414e-05, "loss": 0.2313, "step": 12629 }, { "epoch": 78.44859813084112, "grad_norm": 3.625, "learning_rate": 3.375853066171908e-05, "loss": 0.3246, "step": 12630 }, { "epoch": 78.45482866043614, "grad_norm": 1.84375, "learning_rate": 3.375622300483216e-05, "loss": 0.2168, "step": 12631 }, { "epoch": 78.46105919003115, "grad_norm": 2.53125, "learning_rate": 3.3753915262903045e-05, "loss": 0.1905, "step": 12632 }, { "epoch": 78.46728971962617, "grad_norm": 2.234375, "learning_rate": 3.3751607435954166e-05, "loss": 0.1514, "step": 12633 }, { "epoch": 78.47352024922118, "grad_norm": 2.71875, "learning_rate": 3.374929952400792e-05, "loss": 0.126, "step": 12634 }, { "epoch": 78.4797507788162, "grad_norm": 1.3203125, "learning_rate": 3.374699152708673e-05, "loss": 0.123, "step": 12635 }, { "epoch": 78.48598130841121, "grad_norm": 2.171875, "learning_rate": 3.374468344521301e-05, "loss": 0.1723, "step": 12636 }, { "epoch": 78.49221183800623, "grad_norm": 1.953125, "learning_rate": 3.3742375278409175e-05, "loss": 0.1362, "step": 12637 }, { "epoch": 78.49844236760124, "grad_norm": 2.53125, "learning_rate": 3.374006702669764e-05, "loss": 0.1729, "step": 12638 }, { "epoch": 78.50467289719626, "grad_norm": 1.328125, "learning_rate": 3.373775869010084e-05, "loss": 0.0993, "step": 12639 }, { "epoch": 78.51090342679127, "grad_norm": 1.125, "learning_rate": 3.373545026864117e-05, "loss": 0.0972, "step": 12640 }, { "epoch": 78.51713395638629, "grad_norm": 1.65625, "learning_rate": 3.3733141762341067e-05, "loss": 0.2191, "step": 12641 }, { "epoch": 78.5233644859813, "grad_norm": 2.171875, "learning_rate": 3.373083317122294e-05, "loss": 0.1419, "step": 12642 }, { "epoch": 78.52959501557632, "grad_norm": 2.125, "learning_rate": 3.372852449530922e-05, "loss": 0.2348, "step": 12643 }, { "epoch": 78.53582554517133, "grad_norm": 2.40625, "learning_rate": 3.372621573462232e-05, "loss": 0.1925, "step": 12644 }, { "epoch": 78.54205607476635, "grad_norm": 2.96875, "learning_rate": 3.3723906889184675e-05, "loss": 0.1786, "step": 12645 }, { "epoch": 78.54828660436137, "grad_norm": 1.8671875, "learning_rate": 3.37215979590187e-05, "loss": 0.154, "step": 12646 }, { "epoch": 78.55451713395638, "grad_norm": 2.421875, "learning_rate": 3.3719288944146814e-05, "loss": 0.2217, "step": 12647 }, { "epoch": 78.5607476635514, "grad_norm": 1.984375, "learning_rate": 3.3716979844591465e-05, "loss": 0.1532, "step": 12648 }, { "epoch": 78.56697819314641, "grad_norm": 2.078125, "learning_rate": 3.371467066037506e-05, "loss": 0.1117, "step": 12649 }, { "epoch": 78.57320872274143, "grad_norm": 1.9765625, "learning_rate": 3.371236139152003e-05, "loss": 0.1642, "step": 12650 }, { "epoch": 78.57943925233644, "grad_norm": 2.109375, "learning_rate": 3.37100520380488e-05, "loss": 0.1978, "step": 12651 }, { "epoch": 78.58566978193146, "grad_norm": 3.96875, "learning_rate": 3.370774259998381e-05, "loss": 0.3007, "step": 12652 }, { "epoch": 78.59190031152647, "grad_norm": 3.421875, "learning_rate": 3.370543307734748e-05, "loss": 0.2641, "step": 12653 }, { "epoch": 78.59813084112149, "grad_norm": 3.5625, "learning_rate": 3.370312347016225e-05, "loss": 0.2185, "step": 12654 }, { "epoch": 78.60436137071652, "grad_norm": 1.015625, "learning_rate": 3.3700813778450536e-05, "loss": 0.0792, "step": 12655 }, { "epoch": 78.61059190031153, "grad_norm": 1.2734375, "learning_rate": 3.3698504002234785e-05, "loss": 0.1214, "step": 12656 }, { "epoch": 78.61682242990655, "grad_norm": 2.09375, "learning_rate": 3.369619414153742e-05, "loss": 0.1819, "step": 12657 }, { "epoch": 78.62305295950156, "grad_norm": 4.09375, "learning_rate": 3.3693884196380885e-05, "loss": 0.1711, "step": 12658 }, { "epoch": 78.62928348909658, "grad_norm": 2.453125, "learning_rate": 3.36915741667876e-05, "loss": 0.197, "step": 12659 }, { "epoch": 78.6355140186916, "grad_norm": 2.65625, "learning_rate": 3.368926405278001e-05, "loss": 0.1497, "step": 12660 }, { "epoch": 78.64174454828661, "grad_norm": 2.015625, "learning_rate": 3.368695385438057e-05, "loss": 0.102, "step": 12661 }, { "epoch": 78.64797507788163, "grad_norm": 1.953125, "learning_rate": 3.368464357161167e-05, "loss": 0.1426, "step": 12662 }, { "epoch": 78.65420560747664, "grad_norm": 1.4921875, "learning_rate": 3.3682333204495795e-05, "loss": 0.1199, "step": 12663 }, { "epoch": 78.66043613707166, "grad_norm": 2.28125, "learning_rate": 3.3680022753055356e-05, "loss": 0.1069, "step": 12664 }, { "epoch": 78.66666666666667, "grad_norm": 2.375, "learning_rate": 3.3677712217312805e-05, "loss": 0.16, "step": 12665 }, { "epoch": 78.67289719626169, "grad_norm": 3.578125, "learning_rate": 3.367540159729057e-05, "loss": 0.207, "step": 12666 }, { "epoch": 78.6791277258567, "grad_norm": 1.0546875, "learning_rate": 3.367309089301111e-05, "loss": 0.1093, "step": 12667 }, { "epoch": 78.68535825545172, "grad_norm": 2.34375, "learning_rate": 3.367078010449686e-05, "loss": 0.1406, "step": 12668 }, { "epoch": 78.69158878504673, "grad_norm": 1.765625, "learning_rate": 3.366846923177025e-05, "loss": 0.1434, "step": 12669 }, { "epoch": 78.69781931464175, "grad_norm": 3.875, "learning_rate": 3.3666158274853746e-05, "loss": 0.1354, "step": 12670 }, { "epoch": 78.70404984423676, "grad_norm": 3.09375, "learning_rate": 3.366384723376977e-05, "loss": 0.1139, "step": 12671 }, { "epoch": 78.71028037383178, "grad_norm": 2.671875, "learning_rate": 3.366153610854078e-05, "loss": 0.1913, "step": 12672 }, { "epoch": 78.7165109034268, "grad_norm": 2.78125, "learning_rate": 3.3659224899189215e-05, "loss": 0.2284, "step": 12673 }, { "epoch": 78.72274143302181, "grad_norm": 2.953125, "learning_rate": 3.365691360573754e-05, "loss": 0.2258, "step": 12674 }, { "epoch": 78.72897196261682, "grad_norm": 3.609375, "learning_rate": 3.365460222820818e-05, "loss": 0.218, "step": 12675 }, { "epoch": 78.73520249221184, "grad_norm": 2.953125, "learning_rate": 3.36522907666236e-05, "loss": 0.2738, "step": 12676 }, { "epoch": 78.74143302180686, "grad_norm": 1.625, "learning_rate": 3.364997922100623e-05, "loss": 0.1171, "step": 12677 }, { "epoch": 78.74766355140187, "grad_norm": 2.59375, "learning_rate": 3.364766759137855e-05, "loss": 0.1687, "step": 12678 }, { "epoch": 78.75389408099689, "grad_norm": 1.609375, "learning_rate": 3.364535587776298e-05, "loss": 0.1254, "step": 12679 }, { "epoch": 78.7601246105919, "grad_norm": 1.7890625, "learning_rate": 3.3643044080182e-05, "loss": 0.185, "step": 12680 }, { "epoch": 78.76635514018692, "grad_norm": 3.875, "learning_rate": 3.364073219865803e-05, "loss": 0.1689, "step": 12681 }, { "epoch": 78.77258566978193, "grad_norm": 1.921875, "learning_rate": 3.363842023321355e-05, "loss": 0.1432, "step": 12682 }, { "epoch": 78.77881619937695, "grad_norm": 3.078125, "learning_rate": 3.363610818387101e-05, "loss": 0.1561, "step": 12683 }, { "epoch": 78.78504672897196, "grad_norm": 1.59375, "learning_rate": 3.3633796050652854e-05, "loss": 0.1169, "step": 12684 }, { "epoch": 78.79127725856698, "grad_norm": 3.015625, "learning_rate": 3.363148383358155e-05, "loss": 0.2796, "step": 12685 }, { "epoch": 78.797507788162, "grad_norm": 2.421875, "learning_rate": 3.362917153267954e-05, "loss": 0.2075, "step": 12686 }, { "epoch": 78.80373831775701, "grad_norm": 1.9140625, "learning_rate": 3.36268591479693e-05, "loss": 0.1317, "step": 12687 }, { "epoch": 78.80996884735202, "grad_norm": 1.625, "learning_rate": 3.362454667947328e-05, "loss": 0.0971, "step": 12688 }, { "epoch": 78.81619937694704, "grad_norm": 2.34375, "learning_rate": 3.362223412721394e-05, "loss": 0.1446, "step": 12689 }, { "epoch": 78.82242990654206, "grad_norm": 3.3125, "learning_rate": 3.3619921491213736e-05, "loss": 0.136, "step": 12690 }, { "epoch": 78.82866043613707, "grad_norm": 1.875, "learning_rate": 3.361760877149512e-05, "loss": 0.1366, "step": 12691 }, { "epoch": 78.83489096573209, "grad_norm": 2.921875, "learning_rate": 3.3615295968080585e-05, "loss": 0.2137, "step": 12692 }, { "epoch": 78.8411214953271, "grad_norm": 2.359375, "learning_rate": 3.3612983080992564e-05, "loss": 0.1721, "step": 12693 }, { "epoch": 78.84735202492212, "grad_norm": 2.796875, "learning_rate": 3.3610670110253526e-05, "loss": 0.2409, "step": 12694 }, { "epoch": 78.85358255451713, "grad_norm": 1.8671875, "learning_rate": 3.3608357055885944e-05, "loss": 0.1319, "step": 12695 }, { "epoch": 78.85981308411215, "grad_norm": 1.46875, "learning_rate": 3.360604391791228e-05, "loss": 0.1218, "step": 12696 }, { "epoch": 78.86604361370716, "grad_norm": 2.46875, "learning_rate": 3.360373069635498e-05, "loss": 0.1781, "step": 12697 }, { "epoch": 78.87227414330218, "grad_norm": 1.515625, "learning_rate": 3.360141739123655e-05, "loss": 0.1149, "step": 12698 }, { "epoch": 78.8785046728972, "grad_norm": 2.375, "learning_rate": 3.3599104002579425e-05, "loss": 0.1774, "step": 12699 }, { "epoch": 78.88473520249221, "grad_norm": 3.265625, "learning_rate": 3.3596790530406085e-05, "loss": 0.3647, "step": 12700 }, { "epoch": 78.89096573208722, "grad_norm": 2.109375, "learning_rate": 3.359447697473899e-05, "loss": 0.2428, "step": 12701 }, { "epoch": 78.89719626168224, "grad_norm": 1.7265625, "learning_rate": 3.3592163335600626e-05, "loss": 0.1252, "step": 12702 }, { "epoch": 78.90342679127725, "grad_norm": 2.09375, "learning_rate": 3.358984961301345e-05, "loss": 0.152, "step": 12703 }, { "epoch": 78.90965732087227, "grad_norm": 2.171875, "learning_rate": 3.358753580699994e-05, "loss": 0.1513, "step": 12704 }, { "epoch": 78.91588785046729, "grad_norm": 2.8125, "learning_rate": 3.358522191758256e-05, "loss": 0.1523, "step": 12705 }, { "epoch": 78.9221183800623, "grad_norm": 1.9921875, "learning_rate": 3.3582907944783794e-05, "loss": 0.1449, "step": 12706 }, { "epoch": 78.92834890965732, "grad_norm": 1.9765625, "learning_rate": 3.358059388862611e-05, "loss": 0.1407, "step": 12707 }, { "epoch": 78.93457943925233, "grad_norm": 2.3125, "learning_rate": 3.357827974913198e-05, "loss": 0.1446, "step": 12708 }, { "epoch": 78.94080996884735, "grad_norm": 1.8515625, "learning_rate": 3.357596552632388e-05, "loss": 0.1206, "step": 12709 }, { "epoch": 78.94704049844236, "grad_norm": 1.28125, "learning_rate": 3.357365122022429e-05, "loss": 0.0815, "step": 12710 }, { "epoch": 78.95327102803738, "grad_norm": 2.421875, "learning_rate": 3.3571336830855695e-05, "loss": 0.204, "step": 12711 }, { "epoch": 78.9595015576324, "grad_norm": 1.59375, "learning_rate": 3.3569022358240554e-05, "loss": 0.1146, "step": 12712 }, { "epoch": 78.96573208722741, "grad_norm": 2.828125, "learning_rate": 3.3566707802401356e-05, "loss": 0.2699, "step": 12713 }, { "epoch": 78.97196261682242, "grad_norm": 2.4375, "learning_rate": 3.356439316336058e-05, "loss": 0.2153, "step": 12714 }, { "epoch": 78.97819314641744, "grad_norm": 1.9453125, "learning_rate": 3.356207844114071e-05, "loss": 0.157, "step": 12715 }, { "epoch": 78.98442367601245, "grad_norm": 2.625, "learning_rate": 3.355976363576421e-05, "loss": 0.2058, "step": 12716 }, { "epoch": 78.99065420560747, "grad_norm": 3.15625, "learning_rate": 3.355744874725358e-05, "loss": 0.0926, "step": 12717 }, { "epoch": 78.99688473520249, "grad_norm": 1.703125, "learning_rate": 3.3555133775631306e-05, "loss": 0.1228, "step": 12718 }, { "epoch": 79.0, "grad_norm": 1.2734375, "learning_rate": 3.3552818720919856e-05, "loss": 0.0795, "step": 12719 }, { "epoch": 79.00623052959502, "grad_norm": 2.671875, "learning_rate": 3.355050358314172e-05, "loss": 0.2735, "step": 12720 }, { "epoch": 79.01246105919003, "grad_norm": 1.1328125, "learning_rate": 3.3548188362319386e-05, "loss": 0.0953, "step": 12721 }, { "epoch": 79.01869158878505, "grad_norm": 3.046875, "learning_rate": 3.354587305847534e-05, "loss": 0.2781, "step": 12722 }, { "epoch": 79.02492211838006, "grad_norm": 2.09375, "learning_rate": 3.354355767163206e-05, "loss": 0.116, "step": 12723 }, { "epoch": 79.03115264797508, "grad_norm": 1.96875, "learning_rate": 3.354124220181204e-05, "loss": 0.1415, "step": 12724 }, { "epoch": 79.03738317757009, "grad_norm": 2.78125, "learning_rate": 3.353892664903778e-05, "loss": 0.256, "step": 12725 }, { "epoch": 79.04361370716511, "grad_norm": 1.3828125, "learning_rate": 3.3536611013331746e-05, "loss": 0.1099, "step": 12726 }, { "epoch": 79.04984423676012, "grad_norm": 2.765625, "learning_rate": 3.353429529471644e-05, "loss": 0.1999, "step": 12727 }, { "epoch": 79.05607476635514, "grad_norm": 1.2421875, "learning_rate": 3.353197949321436e-05, "loss": 0.1146, "step": 12728 }, { "epoch": 79.06230529595015, "grad_norm": 2.4375, "learning_rate": 3.352966360884799e-05, "loss": 0.1482, "step": 12729 }, { "epoch": 79.06853582554517, "grad_norm": 1.671875, "learning_rate": 3.352734764163981e-05, "loss": 0.0956, "step": 12730 }, { "epoch": 79.07476635514018, "grad_norm": 2.09375, "learning_rate": 3.352503159161234e-05, "loss": 0.1295, "step": 12731 }, { "epoch": 79.0809968847352, "grad_norm": 1.9453125, "learning_rate": 3.352271545878805e-05, "loss": 0.1519, "step": 12732 }, { "epoch": 79.08722741433021, "grad_norm": 2.9375, "learning_rate": 3.352039924318944e-05, "loss": 0.1742, "step": 12733 }, { "epoch": 79.09345794392523, "grad_norm": 2.609375, "learning_rate": 3.351808294483902e-05, "loss": 0.175, "step": 12734 }, { "epoch": 79.09968847352025, "grad_norm": 2.59375, "learning_rate": 3.351576656375927e-05, "loss": 0.1555, "step": 12735 }, { "epoch": 79.10591900311526, "grad_norm": 2.625, "learning_rate": 3.351345009997269e-05, "loss": 0.1045, "step": 12736 }, { "epoch": 79.11214953271028, "grad_norm": 1.5546875, "learning_rate": 3.3511133553501786e-05, "loss": 0.1308, "step": 12737 }, { "epoch": 79.11838006230529, "grad_norm": 1.8984375, "learning_rate": 3.3508816924369045e-05, "loss": 0.1197, "step": 12738 }, { "epoch": 79.1246105919003, "grad_norm": 2.578125, "learning_rate": 3.350650021259698e-05, "loss": 0.1661, "step": 12739 }, { "epoch": 79.13084112149532, "grad_norm": 3.125, "learning_rate": 3.350418341820808e-05, "loss": 0.1083, "step": 12740 }, { "epoch": 79.13707165109034, "grad_norm": 1.9296875, "learning_rate": 3.350186654122486e-05, "loss": 0.1326, "step": 12741 }, { "epoch": 79.14330218068535, "grad_norm": 3.484375, "learning_rate": 3.3499549581669804e-05, "loss": 0.2278, "step": 12742 }, { "epoch": 79.14953271028037, "grad_norm": 1.1171875, "learning_rate": 3.349723253956542e-05, "loss": 0.1209, "step": 12743 }, { "epoch": 79.15576323987538, "grad_norm": 1.96875, "learning_rate": 3.3494915414934225e-05, "loss": 0.1651, "step": 12744 }, { "epoch": 79.1619937694704, "grad_norm": 2.78125, "learning_rate": 3.34925982077987e-05, "loss": 0.258, "step": 12745 }, { "epoch": 79.16822429906541, "grad_norm": 0.77734375, "learning_rate": 3.3490280918181376e-05, "loss": 0.0861, "step": 12746 }, { "epoch": 79.17445482866043, "grad_norm": 0.984375, "learning_rate": 3.348796354610475e-05, "loss": 0.1016, "step": 12747 }, { "epoch": 79.18068535825545, "grad_norm": 2.0625, "learning_rate": 3.348564609159131e-05, "loss": 0.1612, "step": 12748 }, { "epoch": 79.18691588785046, "grad_norm": 1.875, "learning_rate": 3.348332855466358e-05, "loss": 0.1738, "step": 12749 }, { "epoch": 79.19314641744548, "grad_norm": 3.671875, "learning_rate": 3.348101093534408e-05, "loss": 0.2328, "step": 12750 }, { "epoch": 79.19937694704049, "grad_norm": 1.0625, "learning_rate": 3.34786932336553e-05, "loss": 0.124, "step": 12751 }, { "epoch": 79.20560747663552, "grad_norm": 2.65625, "learning_rate": 3.3476375449619754e-05, "loss": 0.2373, "step": 12752 }, { "epoch": 79.21183800623054, "grad_norm": 2.953125, "learning_rate": 3.347405758325996e-05, "loss": 0.2411, "step": 12753 }, { "epoch": 79.21806853582555, "grad_norm": 2.640625, "learning_rate": 3.3471739634598426e-05, "loss": 0.2968, "step": 12754 }, { "epoch": 79.22429906542057, "grad_norm": 2.984375, "learning_rate": 3.346942160365766e-05, "loss": 0.2462, "step": 12755 }, { "epoch": 79.23052959501558, "grad_norm": 1.34375, "learning_rate": 3.346710349046017e-05, "loss": 0.1177, "step": 12756 }, { "epoch": 79.2367601246106, "grad_norm": 2.84375, "learning_rate": 3.34647852950285e-05, "loss": 0.2839, "step": 12757 }, { "epoch": 79.24299065420561, "grad_norm": 2.953125, "learning_rate": 3.3462467017385124e-05, "loss": 0.1606, "step": 12758 }, { "epoch": 79.24922118380063, "grad_norm": 1.4375, "learning_rate": 3.3460148657552585e-05, "loss": 0.1229, "step": 12759 }, { "epoch": 79.25545171339564, "grad_norm": 2.734375, "learning_rate": 3.345783021555339e-05, "loss": 0.2929, "step": 12760 }, { "epoch": 79.26168224299066, "grad_norm": 2.546875, "learning_rate": 3.345551169141006e-05, "loss": 0.1925, "step": 12761 }, { "epoch": 79.26791277258567, "grad_norm": 0.95703125, "learning_rate": 3.3453193085145095e-05, "loss": 0.0988, "step": 12762 }, { "epoch": 79.27414330218069, "grad_norm": 2.59375, "learning_rate": 3.345087439678105e-05, "loss": 0.1492, "step": 12763 }, { "epoch": 79.2803738317757, "grad_norm": 1.6953125, "learning_rate": 3.3448555626340404e-05, "loss": 0.1012, "step": 12764 }, { "epoch": 79.28660436137072, "grad_norm": 1.609375, "learning_rate": 3.3446236773845706e-05, "loss": 0.1397, "step": 12765 }, { "epoch": 79.29283489096574, "grad_norm": 2.953125, "learning_rate": 3.344391783931947e-05, "loss": 0.1913, "step": 12766 }, { "epoch": 79.29906542056075, "grad_norm": 2.109375, "learning_rate": 3.344159882278422e-05, "loss": 0.1648, "step": 12767 }, { "epoch": 79.30529595015577, "grad_norm": 1.484375, "learning_rate": 3.343927972426247e-05, "loss": 0.1143, "step": 12768 }, { "epoch": 79.31152647975078, "grad_norm": 2.25, "learning_rate": 3.343696054377675e-05, "loss": 0.1486, "step": 12769 }, { "epoch": 79.3177570093458, "grad_norm": 1.703125, "learning_rate": 3.343464128134958e-05, "loss": 0.1529, "step": 12770 }, { "epoch": 79.32398753894081, "grad_norm": 2.046875, "learning_rate": 3.3432321937003485e-05, "loss": 0.1, "step": 12771 }, { "epoch": 79.33021806853583, "grad_norm": 2.375, "learning_rate": 3.3430002510761e-05, "loss": 0.2809, "step": 12772 }, { "epoch": 79.33644859813084, "grad_norm": 1.7421875, "learning_rate": 3.342768300264464e-05, "loss": 0.1237, "step": 12773 }, { "epoch": 79.34267912772586, "grad_norm": 2.65625, "learning_rate": 3.342536341267694e-05, "loss": 0.1499, "step": 12774 }, { "epoch": 79.34890965732087, "grad_norm": 3.046875, "learning_rate": 3.342304374088043e-05, "loss": 0.2082, "step": 12775 }, { "epoch": 79.35514018691589, "grad_norm": 1.9765625, "learning_rate": 3.3420723987277635e-05, "loss": 0.1344, "step": 12776 }, { "epoch": 79.3613707165109, "grad_norm": 3.46875, "learning_rate": 3.341840415189108e-05, "loss": 0.193, "step": 12777 }, { "epoch": 79.36760124610592, "grad_norm": 3.375, "learning_rate": 3.3416084234743306e-05, "loss": 0.1742, "step": 12778 }, { "epoch": 79.37383177570094, "grad_norm": 1.9921875, "learning_rate": 3.3413764235856836e-05, "loss": 0.1642, "step": 12779 }, { "epoch": 79.38006230529595, "grad_norm": 1.8359375, "learning_rate": 3.3411444155254204e-05, "loss": 0.1508, "step": 12780 }, { "epoch": 79.38629283489097, "grad_norm": 2.421875, "learning_rate": 3.3409123992957955e-05, "loss": 0.1143, "step": 12781 }, { "epoch": 79.39252336448598, "grad_norm": 1.9921875, "learning_rate": 3.3406803748990606e-05, "loss": 0.1525, "step": 12782 }, { "epoch": 79.398753894081, "grad_norm": 3.234375, "learning_rate": 3.34044834233747e-05, "loss": 0.2494, "step": 12783 }, { "epoch": 79.40498442367601, "grad_norm": 1.9609375, "learning_rate": 3.340216301613277e-05, "loss": 0.1591, "step": 12784 }, { "epoch": 79.41121495327103, "grad_norm": 1.328125, "learning_rate": 3.339984252728735e-05, "loss": 0.0864, "step": 12785 }, { "epoch": 79.41744548286604, "grad_norm": 2.1875, "learning_rate": 3.3397521956860985e-05, "loss": 0.1415, "step": 12786 }, { "epoch": 79.42367601246106, "grad_norm": 1.71875, "learning_rate": 3.33952013048762e-05, "loss": 0.1506, "step": 12787 }, { "epoch": 79.42990654205607, "grad_norm": 3.09375, "learning_rate": 3.339288057135555e-05, "loss": 0.209, "step": 12788 }, { "epoch": 79.43613707165109, "grad_norm": 2.90625, "learning_rate": 3.3390559756321566e-05, "loss": 0.2762, "step": 12789 }, { "epoch": 79.4423676012461, "grad_norm": 2.125, "learning_rate": 3.338823885979679e-05, "loss": 0.163, "step": 12790 }, { "epoch": 79.44859813084112, "grad_norm": 1.828125, "learning_rate": 3.338591788180375e-05, "loss": 0.1987, "step": 12791 }, { "epoch": 79.45482866043614, "grad_norm": 1.796875, "learning_rate": 3.3383596822365016e-05, "loss": 0.1337, "step": 12792 }, { "epoch": 79.46105919003115, "grad_norm": 1.3203125, "learning_rate": 3.33812756815031e-05, "loss": 0.1186, "step": 12793 }, { "epoch": 79.46728971962617, "grad_norm": 2.828125, "learning_rate": 3.3378954459240563e-05, "loss": 0.1809, "step": 12794 }, { "epoch": 79.47352024922118, "grad_norm": 1.3828125, "learning_rate": 3.337663315559995e-05, "loss": 0.0932, "step": 12795 }, { "epoch": 79.4797507788162, "grad_norm": 1.734375, "learning_rate": 3.33743117706038e-05, "loss": 0.1166, "step": 12796 }, { "epoch": 79.48598130841121, "grad_norm": 2.09375, "learning_rate": 3.3371990304274656e-05, "loss": 0.2101, "step": 12797 }, { "epoch": 79.49221183800623, "grad_norm": 3.234375, "learning_rate": 3.3369668756635076e-05, "loss": 0.1772, "step": 12798 }, { "epoch": 79.49844236760124, "grad_norm": 3.171875, "learning_rate": 3.336734712770759e-05, "loss": 0.2308, "step": 12799 }, { "epoch": 79.50467289719626, "grad_norm": 2.9375, "learning_rate": 3.336502541751476e-05, "loss": 0.1867, "step": 12800 }, { "epoch": 79.51090342679127, "grad_norm": 2.53125, "learning_rate": 3.336270362607913e-05, "loss": 0.1459, "step": 12801 }, { "epoch": 79.51713395638629, "grad_norm": 3.125, "learning_rate": 3.336038175342325e-05, "loss": 0.272, "step": 12802 }, { "epoch": 79.5233644859813, "grad_norm": 2.46875, "learning_rate": 3.335805979956968e-05, "loss": 0.142, "step": 12803 }, { "epoch": 79.52959501557632, "grad_norm": 3.15625, "learning_rate": 3.3355737764540955e-05, "loss": 0.3099, "step": 12804 }, { "epoch": 79.53582554517133, "grad_norm": 3.1875, "learning_rate": 3.335341564835964e-05, "loss": 0.1691, "step": 12805 }, { "epoch": 79.54205607476635, "grad_norm": 2.09375, "learning_rate": 3.335109345104828e-05, "loss": 0.138, "step": 12806 }, { "epoch": 79.54828660436137, "grad_norm": 2.0625, "learning_rate": 3.334877117262942e-05, "loss": 0.1271, "step": 12807 }, { "epoch": 79.55451713395638, "grad_norm": 3.75, "learning_rate": 3.3346448813125643e-05, "loss": 0.2711, "step": 12808 }, { "epoch": 79.5607476635514, "grad_norm": 2.09375, "learning_rate": 3.3344126372559476e-05, "loss": 0.1235, "step": 12809 }, { "epoch": 79.56697819314641, "grad_norm": 2.578125, "learning_rate": 3.3341803850953485e-05, "loss": 0.2221, "step": 12810 }, { "epoch": 79.57320872274143, "grad_norm": 3.21875, "learning_rate": 3.3339481248330236e-05, "loss": 0.1358, "step": 12811 }, { "epoch": 79.57943925233644, "grad_norm": 2.28125, "learning_rate": 3.3337158564712267e-05, "loss": 0.1285, "step": 12812 }, { "epoch": 79.58566978193146, "grad_norm": 2.921875, "learning_rate": 3.3334835800122156e-05, "loss": 0.2003, "step": 12813 }, { "epoch": 79.59190031152647, "grad_norm": 2.234375, "learning_rate": 3.333251295458245e-05, "loss": 0.1544, "step": 12814 }, { "epoch": 79.59813084112149, "grad_norm": 2.296875, "learning_rate": 3.3330190028115714e-05, "loss": 0.2093, "step": 12815 }, { "epoch": 79.60436137071652, "grad_norm": 1.203125, "learning_rate": 3.332786702074451e-05, "loss": 0.0968, "step": 12816 }, { "epoch": 79.61059190031153, "grad_norm": 2.453125, "learning_rate": 3.332554393249139e-05, "loss": 0.1786, "step": 12817 }, { "epoch": 79.61682242990655, "grad_norm": 2.453125, "learning_rate": 3.3323220763378924e-05, "loss": 0.2064, "step": 12818 }, { "epoch": 79.62305295950156, "grad_norm": 2.875, "learning_rate": 3.332089751342968e-05, "loss": 0.2097, "step": 12819 }, { "epoch": 79.62928348909658, "grad_norm": 2.78125, "learning_rate": 3.331857418266622e-05, "loss": 0.2553, "step": 12820 }, { "epoch": 79.6355140186916, "grad_norm": 3.015625, "learning_rate": 3.331625077111109e-05, "loss": 0.2654, "step": 12821 }, { "epoch": 79.64174454828661, "grad_norm": 1.4296875, "learning_rate": 3.3313927278786875e-05, "loss": 0.1424, "step": 12822 }, { "epoch": 79.64797507788163, "grad_norm": 2.109375, "learning_rate": 3.331160370571614e-05, "loss": 0.1713, "step": 12823 }, { "epoch": 79.65420560747664, "grad_norm": 2.390625, "learning_rate": 3.330928005192145e-05, "loss": 0.1899, "step": 12824 }, { "epoch": 79.66043613707166, "grad_norm": 2.984375, "learning_rate": 3.330695631742537e-05, "loss": 0.2373, "step": 12825 }, { "epoch": 79.66666666666667, "grad_norm": 2.421875, "learning_rate": 3.330463250225046e-05, "loss": 0.1707, "step": 12826 }, { "epoch": 79.67289719626169, "grad_norm": 1.8828125, "learning_rate": 3.3302308606419317e-05, "loss": 0.1261, "step": 12827 }, { "epoch": 79.6791277258567, "grad_norm": 2.84375, "learning_rate": 3.3299984629954475e-05, "loss": 0.2187, "step": 12828 }, { "epoch": 79.68535825545172, "grad_norm": 2.5625, "learning_rate": 3.329766057287854e-05, "loss": 0.1892, "step": 12829 }, { "epoch": 79.69158878504673, "grad_norm": 1.125, "learning_rate": 3.329533643521405e-05, "loss": 0.1126, "step": 12830 }, { "epoch": 79.69781931464175, "grad_norm": 2.46875, "learning_rate": 3.329301221698361e-05, "loss": 0.2128, "step": 12831 }, { "epoch": 79.70404984423676, "grad_norm": 2.609375, "learning_rate": 3.3290687918209763e-05, "loss": 0.1846, "step": 12832 }, { "epoch": 79.71028037383178, "grad_norm": 2.6875, "learning_rate": 3.3288363538915114e-05, "loss": 0.2214, "step": 12833 }, { "epoch": 79.7165109034268, "grad_norm": 2.796875, "learning_rate": 3.328603907912221e-05, "loss": 0.1951, "step": 12834 }, { "epoch": 79.72274143302181, "grad_norm": 1.2734375, "learning_rate": 3.3283714538853636e-05, "loss": 0.0921, "step": 12835 }, { "epoch": 79.72897196261682, "grad_norm": 2.421875, "learning_rate": 3.328138991813198e-05, "loss": 0.2214, "step": 12836 }, { "epoch": 79.73520249221184, "grad_norm": 1.7734375, "learning_rate": 3.3279065216979804e-05, "loss": 0.1141, "step": 12837 }, { "epoch": 79.74143302180686, "grad_norm": 1.671875, "learning_rate": 3.3276740435419694e-05, "loss": 0.115, "step": 12838 }, { "epoch": 79.74766355140187, "grad_norm": 2.375, "learning_rate": 3.3274415573474227e-05, "loss": 0.2057, "step": 12839 }, { "epoch": 79.75389408099689, "grad_norm": 2.0625, "learning_rate": 3.327209063116597e-05, "loss": 0.1118, "step": 12840 }, { "epoch": 79.7601246105919, "grad_norm": 2.46875, "learning_rate": 3.326976560851753e-05, "loss": 0.1703, "step": 12841 }, { "epoch": 79.76635514018692, "grad_norm": 2.75, "learning_rate": 3.326744050555147e-05, "loss": 0.3209, "step": 12842 }, { "epoch": 79.77258566978193, "grad_norm": 2.515625, "learning_rate": 3.3265115322290374e-05, "loss": 0.1631, "step": 12843 }, { "epoch": 79.77881619937695, "grad_norm": 2.265625, "learning_rate": 3.326279005875683e-05, "loss": 0.1326, "step": 12844 }, { "epoch": 79.78504672897196, "grad_norm": 2.265625, "learning_rate": 3.32604647149734e-05, "loss": 0.1178, "step": 12845 }, { "epoch": 79.79127725856698, "grad_norm": 3.9375, "learning_rate": 3.325813929096271e-05, "loss": 0.3093, "step": 12846 }, { "epoch": 79.797507788162, "grad_norm": 2.484375, "learning_rate": 3.3255813786747305e-05, "loss": 0.2428, "step": 12847 }, { "epoch": 79.80373831775701, "grad_norm": 1.5546875, "learning_rate": 3.3253488202349784e-05, "loss": 0.1517, "step": 12848 }, { "epoch": 79.80996884735202, "grad_norm": 2.625, "learning_rate": 3.3251162537792745e-05, "loss": 0.1698, "step": 12849 }, { "epoch": 79.81619937694704, "grad_norm": 3.375, "learning_rate": 3.3248836793098766e-05, "loss": 0.126, "step": 12850 }, { "epoch": 79.82242990654206, "grad_norm": 4.125, "learning_rate": 3.324651096829043e-05, "loss": 0.1821, "step": 12851 }, { "epoch": 79.82866043613707, "grad_norm": 2.15625, "learning_rate": 3.3244185063390335e-05, "loss": 0.1877, "step": 12852 }, { "epoch": 79.83489096573209, "grad_norm": 2.515625, "learning_rate": 3.324185907842106e-05, "loss": 0.2577, "step": 12853 }, { "epoch": 79.8411214953271, "grad_norm": 1.96875, "learning_rate": 3.323953301340521e-05, "loss": 0.1695, "step": 12854 }, { "epoch": 79.84735202492212, "grad_norm": 2.625, "learning_rate": 3.323720686836537e-05, "loss": 0.1734, "step": 12855 }, { "epoch": 79.85358255451713, "grad_norm": 2.796875, "learning_rate": 3.323488064332412e-05, "loss": 0.2157, "step": 12856 }, { "epoch": 79.85981308411215, "grad_norm": 1.109375, "learning_rate": 3.323255433830407e-05, "loss": 0.1189, "step": 12857 }, { "epoch": 79.86604361370716, "grad_norm": 1.7734375, "learning_rate": 3.3230227953327796e-05, "loss": 0.1234, "step": 12858 }, { "epoch": 79.87227414330218, "grad_norm": 2.78125, "learning_rate": 3.322790148841792e-05, "loss": 0.125, "step": 12859 }, { "epoch": 79.8785046728972, "grad_norm": 1.75, "learning_rate": 3.3225574943597005e-05, "loss": 0.1643, "step": 12860 }, { "epoch": 79.88473520249221, "grad_norm": 1.1640625, "learning_rate": 3.3223248318887666e-05, "loss": 0.1075, "step": 12861 }, { "epoch": 79.89096573208722, "grad_norm": 3.09375, "learning_rate": 3.3220921614312505e-05, "loss": 0.2196, "step": 12862 }, { "epoch": 79.89719626168224, "grad_norm": 2.1875, "learning_rate": 3.32185948298941e-05, "loss": 0.1645, "step": 12863 }, { "epoch": 79.90342679127725, "grad_norm": 2.859375, "learning_rate": 3.3216267965655065e-05, "loss": 0.2557, "step": 12864 }, { "epoch": 79.90965732087227, "grad_norm": 3.578125, "learning_rate": 3.321394102161799e-05, "loss": 0.1605, "step": 12865 }, { "epoch": 79.91588785046729, "grad_norm": 1.7578125, "learning_rate": 3.321161399780548e-05, "loss": 0.1047, "step": 12866 }, { "epoch": 79.9221183800623, "grad_norm": 2.578125, "learning_rate": 3.320928689424013e-05, "loss": 0.1489, "step": 12867 }, { "epoch": 79.92834890965732, "grad_norm": 1.8515625, "learning_rate": 3.320695971094455e-05, "loss": 0.1328, "step": 12868 }, { "epoch": 79.93457943925233, "grad_norm": 2.734375, "learning_rate": 3.320463244794133e-05, "loss": 0.1185, "step": 12869 }, { "epoch": 79.94080996884735, "grad_norm": 2.140625, "learning_rate": 3.3202305105253086e-05, "loss": 0.1193, "step": 12870 }, { "epoch": 79.94704049844236, "grad_norm": 2.953125, "learning_rate": 3.319997768290241e-05, "loss": 0.213, "step": 12871 }, { "epoch": 79.95327102803738, "grad_norm": 3.359375, "learning_rate": 3.319765018091191e-05, "loss": 0.357, "step": 12872 }, { "epoch": 79.9595015576324, "grad_norm": 2.96875, "learning_rate": 3.31953225993042e-05, "loss": 0.3044, "step": 12873 }, { "epoch": 79.96573208722741, "grad_norm": 1.9375, "learning_rate": 3.319299493810187e-05, "loss": 0.1275, "step": 12874 }, { "epoch": 79.97196261682242, "grad_norm": 2.59375, "learning_rate": 3.3190667197327543e-05, "loss": 0.1728, "step": 12875 }, { "epoch": 79.97819314641744, "grad_norm": 1.03125, "learning_rate": 3.3188339377003816e-05, "loss": 0.1098, "step": 12876 }, { "epoch": 79.98442367601245, "grad_norm": 1.9921875, "learning_rate": 3.31860114771533e-05, "loss": 0.1677, "step": 12877 }, { "epoch": 79.99065420560747, "grad_norm": 1.40625, "learning_rate": 3.3183683497798606e-05, "loss": 0.1062, "step": 12878 }, { "epoch": 79.99688473520249, "grad_norm": 1.90625, "learning_rate": 3.318135543896234e-05, "loss": 0.1478, "step": 12879 }, { "epoch": 80.0, "grad_norm": 1.0234375, "learning_rate": 3.3179027300667114e-05, "loss": 0.0778, "step": 12880 }, { "epoch": 80.00623052959502, "grad_norm": 1.453125, "learning_rate": 3.3176699082935545e-05, "loss": 0.1022, "step": 12881 }, { "epoch": 80.01246105919003, "grad_norm": 2.828125, "learning_rate": 3.317437078579023e-05, "loss": 0.236, "step": 12882 }, { "epoch": 80.01869158878505, "grad_norm": 1.9375, "learning_rate": 3.31720424092538e-05, "loss": 0.1128, "step": 12883 }, { "epoch": 80.02492211838006, "grad_norm": 4.0, "learning_rate": 3.316971395334886e-05, "loss": 0.1872, "step": 12884 }, { "epoch": 80.03115264797508, "grad_norm": 2.234375, "learning_rate": 3.316738541809803e-05, "loss": 0.1456, "step": 12885 }, { "epoch": 80.03738317757009, "grad_norm": 1.6328125, "learning_rate": 3.3165056803523906e-05, "loss": 0.1126, "step": 12886 }, { "epoch": 80.04361370716511, "grad_norm": 1.515625, "learning_rate": 3.316272810964913e-05, "loss": 0.1514, "step": 12887 }, { "epoch": 80.04984423676012, "grad_norm": 3.03125, "learning_rate": 3.3160399336496306e-05, "loss": 0.2167, "step": 12888 }, { "epoch": 80.05607476635514, "grad_norm": 1.140625, "learning_rate": 3.3158070484088045e-05, "loss": 0.1046, "step": 12889 }, { "epoch": 80.06230529595015, "grad_norm": 2.671875, "learning_rate": 3.315574155244698e-05, "loss": 0.1545, "step": 12890 }, { "epoch": 80.06853582554517, "grad_norm": 2.296875, "learning_rate": 3.315341254159572e-05, "loss": 0.2165, "step": 12891 }, { "epoch": 80.07476635514018, "grad_norm": 1.8515625, "learning_rate": 3.3151083451556885e-05, "loss": 0.1378, "step": 12892 }, { "epoch": 80.0809968847352, "grad_norm": 1.578125, "learning_rate": 3.31487542823531e-05, "loss": 0.0929, "step": 12893 }, { "epoch": 80.08722741433021, "grad_norm": 3.296875, "learning_rate": 3.314642503400699e-05, "loss": 0.2045, "step": 12894 }, { "epoch": 80.09345794392523, "grad_norm": 3.25, "learning_rate": 3.3144095706541155e-05, "loss": 0.2493, "step": 12895 }, { "epoch": 80.09968847352025, "grad_norm": 2.25, "learning_rate": 3.314176629997825e-05, "loss": 0.1958, "step": 12896 }, { "epoch": 80.10591900311526, "grad_norm": 2.125, "learning_rate": 3.313943681434088e-05, "loss": 0.1493, "step": 12897 }, { "epoch": 80.11214953271028, "grad_norm": 1.21875, "learning_rate": 3.313710724965167e-05, "loss": 0.1311, "step": 12898 }, { "epoch": 80.11838006230529, "grad_norm": 1.25, "learning_rate": 3.3134777605933256e-05, "loss": 0.1161, "step": 12899 }, { "epoch": 80.1246105919003, "grad_norm": 2.453125, "learning_rate": 3.3132447883208244e-05, "loss": 0.123, "step": 12900 }, { "epoch": 80.13084112149532, "grad_norm": 3.34375, "learning_rate": 3.3130118081499287e-05, "loss": 0.178, "step": 12901 }, { "epoch": 80.13707165109034, "grad_norm": 3.8125, "learning_rate": 3.312778820082898e-05, "loss": 0.343, "step": 12902 }, { "epoch": 80.14330218068535, "grad_norm": 1.5625, "learning_rate": 3.3125458241219986e-05, "loss": 0.1185, "step": 12903 }, { "epoch": 80.14953271028037, "grad_norm": 1.8125, "learning_rate": 3.31231282026949e-05, "loss": 0.1133, "step": 12904 }, { "epoch": 80.15576323987538, "grad_norm": 3.234375, "learning_rate": 3.312079808527638e-05, "loss": 0.326, "step": 12905 }, { "epoch": 80.1619937694704, "grad_norm": 2.046875, "learning_rate": 3.3118467888987046e-05, "loss": 0.158, "step": 12906 }, { "epoch": 80.16822429906541, "grad_norm": 3.265625, "learning_rate": 3.311613761384953e-05, "loss": 0.2659, "step": 12907 }, { "epoch": 80.17445482866043, "grad_norm": 1.2890625, "learning_rate": 3.311380725988645e-05, "loss": 0.1011, "step": 12908 }, { "epoch": 80.18068535825545, "grad_norm": 2.046875, "learning_rate": 3.311147682712047e-05, "loss": 0.1248, "step": 12909 }, { "epoch": 80.18691588785046, "grad_norm": 1.71875, "learning_rate": 3.31091463155742e-05, "loss": 0.1577, "step": 12910 }, { "epoch": 80.19314641744548, "grad_norm": 1.578125, "learning_rate": 3.310681572527028e-05, "loss": 0.1288, "step": 12911 }, { "epoch": 80.19937694704049, "grad_norm": 2.265625, "learning_rate": 3.310448505623134e-05, "loss": 0.1361, "step": 12912 }, { "epoch": 80.20560747663552, "grad_norm": 3.28125, "learning_rate": 3.3102154308480025e-05, "loss": 0.1624, "step": 12913 }, { "epoch": 80.21183800623054, "grad_norm": 2.65625, "learning_rate": 3.309982348203897e-05, "loss": 0.1818, "step": 12914 }, { "epoch": 80.21806853582555, "grad_norm": 1.78125, "learning_rate": 3.30974925769308e-05, "loss": 0.1728, "step": 12915 }, { "epoch": 80.22429906542057, "grad_norm": 3.28125, "learning_rate": 3.3095161593178184e-05, "loss": 0.2325, "step": 12916 }, { "epoch": 80.23052959501558, "grad_norm": 3.015625, "learning_rate": 3.309283053080372e-05, "loss": 0.2143, "step": 12917 }, { "epoch": 80.2367601246106, "grad_norm": 1.703125, "learning_rate": 3.309049938983007e-05, "loss": 0.1423, "step": 12918 }, { "epoch": 80.24299065420561, "grad_norm": 2.21875, "learning_rate": 3.308816817027988e-05, "loss": 0.1361, "step": 12919 }, { "epoch": 80.24922118380063, "grad_norm": 2.0, "learning_rate": 3.3085836872175785e-05, "loss": 0.159, "step": 12920 }, { "epoch": 80.25545171339564, "grad_norm": 3.578125, "learning_rate": 3.308350549554042e-05, "loss": 0.2773, "step": 12921 }, { "epoch": 80.26168224299066, "grad_norm": 1.84375, "learning_rate": 3.308117404039644e-05, "loss": 0.1318, "step": 12922 }, { "epoch": 80.26791277258567, "grad_norm": 2.859375, "learning_rate": 3.3078842506766484e-05, "loss": 0.1189, "step": 12923 }, { "epoch": 80.27414330218069, "grad_norm": 2.375, "learning_rate": 3.307651089467318e-05, "loss": 0.1289, "step": 12924 }, { "epoch": 80.2803738317757, "grad_norm": 3.390625, "learning_rate": 3.307417920413921e-05, "loss": 0.118, "step": 12925 }, { "epoch": 80.28660436137072, "grad_norm": 2.171875, "learning_rate": 3.307184743518718e-05, "loss": 0.1279, "step": 12926 }, { "epoch": 80.29283489096574, "grad_norm": 3.046875, "learning_rate": 3.3069515587839754e-05, "loss": 0.1179, "step": 12927 }, { "epoch": 80.29906542056075, "grad_norm": 2.171875, "learning_rate": 3.3067183662119584e-05, "loss": 0.1445, "step": 12928 }, { "epoch": 80.30529595015577, "grad_norm": 2.28125, "learning_rate": 3.3064851658049314e-05, "loss": 0.142, "step": 12929 }, { "epoch": 80.31152647975078, "grad_norm": 2.171875, "learning_rate": 3.306251957565159e-05, "loss": 0.1174, "step": 12930 }, { "epoch": 80.3177570093458, "grad_norm": 2.9375, "learning_rate": 3.306018741494907e-05, "loss": 0.2041, "step": 12931 }, { "epoch": 80.32398753894081, "grad_norm": 2.75, "learning_rate": 3.30578551759644e-05, "loss": 0.1223, "step": 12932 }, { "epoch": 80.33021806853583, "grad_norm": 1.8671875, "learning_rate": 3.305552285872022e-05, "loss": 0.1341, "step": 12933 }, { "epoch": 80.33644859813084, "grad_norm": 3.671875, "learning_rate": 3.3053190463239205e-05, "loss": 0.2599, "step": 12934 }, { "epoch": 80.34267912772586, "grad_norm": 1.3203125, "learning_rate": 3.305085798954398e-05, "loss": 0.0935, "step": 12935 }, { "epoch": 80.34890965732087, "grad_norm": 2.265625, "learning_rate": 3.304852543765722e-05, "loss": 0.1993, "step": 12936 }, { "epoch": 80.35514018691589, "grad_norm": 1.8203125, "learning_rate": 3.304619280760157e-05, "loss": 0.1024, "step": 12937 }, { "epoch": 80.3613707165109, "grad_norm": 2.046875, "learning_rate": 3.304386009939969e-05, "loss": 0.1333, "step": 12938 }, { "epoch": 80.36760124610592, "grad_norm": 2.6875, "learning_rate": 3.304152731307423e-05, "loss": 0.2287, "step": 12939 }, { "epoch": 80.37383177570094, "grad_norm": 2.65625, "learning_rate": 3.303919444864785e-05, "loss": 0.1763, "step": 12940 }, { "epoch": 80.38006230529595, "grad_norm": 2.734375, "learning_rate": 3.303686150614321e-05, "loss": 0.1577, "step": 12941 }, { "epoch": 80.38629283489097, "grad_norm": 2.578125, "learning_rate": 3.303452848558297e-05, "loss": 0.115, "step": 12942 }, { "epoch": 80.39252336448598, "grad_norm": 3.4375, "learning_rate": 3.303219538698976e-05, "loss": 0.1539, "step": 12943 }, { "epoch": 80.398753894081, "grad_norm": 2.984375, "learning_rate": 3.3029862210386284e-05, "loss": 0.18, "step": 12944 }, { "epoch": 80.40498442367601, "grad_norm": 1.6171875, "learning_rate": 3.3027528955795174e-05, "loss": 0.152, "step": 12945 }, { "epoch": 80.41121495327103, "grad_norm": 1.078125, "learning_rate": 3.30251956232391e-05, "loss": 0.1118, "step": 12946 }, { "epoch": 80.41744548286604, "grad_norm": 1.40625, "learning_rate": 3.302286221274072e-05, "loss": 0.1203, "step": 12947 }, { "epoch": 80.42367601246106, "grad_norm": 2.5625, "learning_rate": 3.3020528724322705e-05, "loss": 0.2276, "step": 12948 }, { "epoch": 80.42990654205607, "grad_norm": 1.8984375, "learning_rate": 3.30181951580077e-05, "loss": 0.1323, "step": 12949 }, { "epoch": 80.43613707165109, "grad_norm": 1.6875, "learning_rate": 3.301586151381839e-05, "loss": 0.0974, "step": 12950 }, { "epoch": 80.4423676012461, "grad_norm": 2.6875, "learning_rate": 3.301352779177743e-05, "loss": 0.2127, "step": 12951 }, { "epoch": 80.44859813084112, "grad_norm": 1.7109375, "learning_rate": 3.301119399190748e-05, "loss": 0.1732, "step": 12952 }, { "epoch": 80.45482866043614, "grad_norm": 1.828125, "learning_rate": 3.3008860114231215e-05, "loss": 0.1711, "step": 12953 }, { "epoch": 80.46105919003115, "grad_norm": 1.546875, "learning_rate": 3.3006526158771306e-05, "loss": 0.1027, "step": 12954 }, { "epoch": 80.46728971962617, "grad_norm": 1.34375, "learning_rate": 3.300419212555041e-05, "loss": 0.1038, "step": 12955 }, { "epoch": 80.47352024922118, "grad_norm": 2.8125, "learning_rate": 3.300185801459119e-05, "loss": 0.2351, "step": 12956 }, { "epoch": 80.4797507788162, "grad_norm": 2.96875, "learning_rate": 3.2999523825916346e-05, "loss": 0.2817, "step": 12957 }, { "epoch": 80.48598130841121, "grad_norm": 3.171875, "learning_rate": 3.2997189559548515e-05, "loss": 0.2573, "step": 12958 }, { "epoch": 80.49221183800623, "grad_norm": 1.90625, "learning_rate": 3.299485521551038e-05, "loss": 0.1634, "step": 12959 }, { "epoch": 80.49844236760124, "grad_norm": 1.625, "learning_rate": 3.299252079382461e-05, "loss": 0.118, "step": 12960 }, { "epoch": 80.50467289719626, "grad_norm": 2.625, "learning_rate": 3.2990186294513894e-05, "loss": 0.1914, "step": 12961 }, { "epoch": 80.51090342679127, "grad_norm": 2.15625, "learning_rate": 3.298785171760088e-05, "loss": 0.1377, "step": 12962 }, { "epoch": 80.51713395638629, "grad_norm": 2.96875, "learning_rate": 3.298551706310825e-05, "loss": 0.2577, "step": 12963 }, { "epoch": 80.5233644859813, "grad_norm": 2.859375, "learning_rate": 3.298318233105869e-05, "loss": 0.2175, "step": 12964 }, { "epoch": 80.52959501557632, "grad_norm": 2.046875, "learning_rate": 3.298084752147487e-05, "loss": 0.1193, "step": 12965 }, { "epoch": 80.53582554517133, "grad_norm": 2.953125, "learning_rate": 3.297851263437945e-05, "loss": 0.199, "step": 12966 }, { "epoch": 80.54205607476635, "grad_norm": 1.8046875, "learning_rate": 3.297617766979514e-05, "loss": 0.1229, "step": 12967 }, { "epoch": 80.54828660436137, "grad_norm": 2.0625, "learning_rate": 3.297384262774458e-05, "loss": 0.1514, "step": 12968 }, { "epoch": 80.55451713395638, "grad_norm": 3.640625, "learning_rate": 3.297150750825048e-05, "loss": 0.1344, "step": 12969 }, { "epoch": 80.5607476635514, "grad_norm": 2.03125, "learning_rate": 3.29691723113355e-05, "loss": 0.1891, "step": 12970 }, { "epoch": 80.56697819314641, "grad_norm": 1.5078125, "learning_rate": 3.2966837037022325e-05, "loss": 0.1424, "step": 12971 }, { "epoch": 80.57320872274143, "grad_norm": 3.71875, "learning_rate": 3.2964501685333635e-05, "loss": 0.1712, "step": 12972 }, { "epoch": 80.57943925233644, "grad_norm": 3.359375, "learning_rate": 3.2962166256292113e-05, "loss": 0.3547, "step": 12973 }, { "epoch": 80.58566978193146, "grad_norm": 1.484375, "learning_rate": 3.295983074992044e-05, "loss": 0.1209, "step": 12974 }, { "epoch": 80.59190031152647, "grad_norm": 1.9453125, "learning_rate": 3.2957495166241306e-05, "loss": 0.1567, "step": 12975 }, { "epoch": 80.59813084112149, "grad_norm": 3.015625, "learning_rate": 3.2955159505277386e-05, "loss": 0.3473, "step": 12976 }, { "epoch": 80.60436137071652, "grad_norm": 2.84375, "learning_rate": 3.2952823767051363e-05, "loss": 0.3338, "step": 12977 }, { "epoch": 80.61059190031153, "grad_norm": 1.515625, "learning_rate": 3.295048795158592e-05, "loss": 0.1394, "step": 12978 }, { "epoch": 80.61682242990655, "grad_norm": 2.90625, "learning_rate": 3.294815205890376e-05, "loss": 0.2213, "step": 12979 }, { "epoch": 80.62305295950156, "grad_norm": 3.28125, "learning_rate": 3.2945816089027556e-05, "loss": 0.2119, "step": 12980 }, { "epoch": 80.62928348909658, "grad_norm": 2.828125, "learning_rate": 3.294348004197999e-05, "loss": 0.2626, "step": 12981 }, { "epoch": 80.6355140186916, "grad_norm": 1.734375, "learning_rate": 3.2941143917783766e-05, "loss": 0.146, "step": 12982 }, { "epoch": 80.64174454828661, "grad_norm": 1.28125, "learning_rate": 3.293880771646156e-05, "loss": 0.0932, "step": 12983 }, { "epoch": 80.64797507788163, "grad_norm": 1.3125, "learning_rate": 3.2936471438036075e-05, "loss": 0.1094, "step": 12984 }, { "epoch": 80.65420560747664, "grad_norm": 3.484375, "learning_rate": 3.293413508252998e-05, "loss": 0.0976, "step": 12985 }, { "epoch": 80.66043613707166, "grad_norm": 1.8671875, "learning_rate": 3.293179864996599e-05, "loss": 0.1387, "step": 12986 }, { "epoch": 80.66666666666667, "grad_norm": 1.5625, "learning_rate": 3.292946214036678e-05, "loss": 0.0813, "step": 12987 }, { "epoch": 80.67289719626169, "grad_norm": 0.8984375, "learning_rate": 3.2927125553755054e-05, "loss": 0.0922, "step": 12988 }, { "epoch": 80.6791277258567, "grad_norm": 2.625, "learning_rate": 3.29247888901535e-05, "loss": 0.1564, "step": 12989 }, { "epoch": 80.68535825545172, "grad_norm": 3.875, "learning_rate": 3.292245214958481e-05, "loss": 0.239, "step": 12990 }, { "epoch": 80.69158878504673, "grad_norm": 3.3125, "learning_rate": 3.292011533207167e-05, "loss": 0.2675, "step": 12991 }, { "epoch": 80.69781931464175, "grad_norm": 1.3125, "learning_rate": 3.29177784376368e-05, "loss": 0.0998, "step": 12992 }, { "epoch": 80.70404984423676, "grad_norm": 2.578125, "learning_rate": 3.2915441466302885e-05, "loss": 0.1259, "step": 12993 }, { "epoch": 80.71028037383178, "grad_norm": 1.9375, "learning_rate": 3.291310441809261e-05, "loss": 0.0967, "step": 12994 }, { "epoch": 80.7165109034268, "grad_norm": 2.1875, "learning_rate": 3.291076729302869e-05, "loss": 0.2181, "step": 12995 }, { "epoch": 80.72274143302181, "grad_norm": 2.234375, "learning_rate": 3.290843009113382e-05, "loss": 0.1324, "step": 12996 }, { "epoch": 80.72897196261682, "grad_norm": 2.578125, "learning_rate": 3.290609281243069e-05, "loss": 0.1126, "step": 12997 }, { "epoch": 80.73520249221184, "grad_norm": 3.484375, "learning_rate": 3.2903755456942e-05, "loss": 0.1776, "step": 12998 }, { "epoch": 80.74143302180686, "grad_norm": 2.53125, "learning_rate": 3.2901418024690474e-05, "loss": 0.1447, "step": 12999 }, { "epoch": 80.74766355140187, "grad_norm": 3.15625, "learning_rate": 3.289908051569878e-05, "loss": 0.3274, "step": 13000 }, { "epoch": 80.75389408099689, "grad_norm": 1.9921875, "learning_rate": 3.289674292998964e-05, "loss": 0.1106, "step": 13001 }, { "epoch": 80.7601246105919, "grad_norm": 2.359375, "learning_rate": 3.289440526758577e-05, "loss": 0.1378, "step": 13002 }, { "epoch": 80.76635514018692, "grad_norm": 2.515625, "learning_rate": 3.289206752850984e-05, "loss": 0.2092, "step": 13003 }, { "epoch": 80.77258566978193, "grad_norm": 3.78125, "learning_rate": 3.288972971278458e-05, "loss": 0.2464, "step": 13004 }, { "epoch": 80.77881619937695, "grad_norm": 3.875, "learning_rate": 3.288739182043269e-05, "loss": 0.1185, "step": 13005 }, { "epoch": 80.78504672897196, "grad_norm": 3.171875, "learning_rate": 3.288505385147687e-05, "loss": 0.1443, "step": 13006 }, { "epoch": 80.79127725856698, "grad_norm": 2.15625, "learning_rate": 3.2882715805939825e-05, "loss": 0.0964, "step": 13007 }, { "epoch": 80.797507788162, "grad_norm": 2.390625, "learning_rate": 3.288037768384428e-05, "loss": 0.1208, "step": 13008 }, { "epoch": 80.80373831775701, "grad_norm": 3.34375, "learning_rate": 3.287803948521292e-05, "loss": 0.1135, "step": 13009 }, { "epoch": 80.80996884735202, "grad_norm": 4.75, "learning_rate": 3.287570121006848e-05, "loss": 0.1298, "step": 13010 }, { "epoch": 80.81619937694704, "grad_norm": 3.0, "learning_rate": 3.2873362858433644e-05, "loss": 0.1059, "step": 13011 }, { "epoch": 80.82242990654206, "grad_norm": 2.9375, "learning_rate": 3.287102443033114e-05, "loss": 0.1375, "step": 13012 }, { "epoch": 80.82866043613707, "grad_norm": 3.765625, "learning_rate": 3.2868685925783665e-05, "loss": 0.2762, "step": 13013 }, { "epoch": 80.83489096573209, "grad_norm": 1.953125, "learning_rate": 3.286634734481395e-05, "loss": 0.1705, "step": 13014 }, { "epoch": 80.8411214953271, "grad_norm": 2.21875, "learning_rate": 3.2864008687444694e-05, "loss": 0.1393, "step": 13015 }, { "epoch": 80.84735202492212, "grad_norm": 2.546875, "learning_rate": 3.2861669953698606e-05, "loss": 0.1696, "step": 13016 }, { "epoch": 80.85358255451713, "grad_norm": 2.65625, "learning_rate": 3.285933114359842e-05, "loss": 0.2843, "step": 13017 }, { "epoch": 80.85981308411215, "grad_norm": 1.828125, "learning_rate": 3.285699225716683e-05, "loss": 0.1548, "step": 13018 }, { "epoch": 80.86604361370716, "grad_norm": 2.109375, "learning_rate": 3.2854653294426566e-05, "loss": 0.1847, "step": 13019 }, { "epoch": 80.87227414330218, "grad_norm": 2.234375, "learning_rate": 3.285231425540034e-05, "loss": 0.1677, "step": 13020 }, { "epoch": 80.8785046728972, "grad_norm": 1.78125, "learning_rate": 3.284997514011087e-05, "loss": 0.1007, "step": 13021 }, { "epoch": 80.88473520249221, "grad_norm": 3.421875, "learning_rate": 3.284763594858086e-05, "loss": 0.2456, "step": 13022 }, { "epoch": 80.89096573208722, "grad_norm": 2.65625, "learning_rate": 3.2845296680833046e-05, "loss": 0.1511, "step": 13023 }, { "epoch": 80.89719626168224, "grad_norm": 3.3125, "learning_rate": 3.284295733689014e-05, "loss": 0.2451, "step": 13024 }, { "epoch": 80.90342679127725, "grad_norm": 1.7109375, "learning_rate": 3.2840617916774874e-05, "loss": 0.1414, "step": 13025 }, { "epoch": 80.90965732087227, "grad_norm": 1.3515625, "learning_rate": 3.283827842050995e-05, "loss": 0.1115, "step": 13026 }, { "epoch": 80.91588785046729, "grad_norm": 1.3515625, "learning_rate": 3.283593884811811e-05, "loss": 0.1203, "step": 13027 }, { "epoch": 80.9221183800623, "grad_norm": 2.078125, "learning_rate": 3.283359919962206e-05, "loss": 0.2066, "step": 13028 }, { "epoch": 80.92834890965732, "grad_norm": 2.515625, "learning_rate": 3.283125947504453e-05, "loss": 0.2006, "step": 13029 }, { "epoch": 80.93457943925233, "grad_norm": 2.390625, "learning_rate": 3.2828919674408234e-05, "loss": 0.211, "step": 13030 }, { "epoch": 80.94080996884735, "grad_norm": 2.4375, "learning_rate": 3.282657979773592e-05, "loss": 0.282, "step": 13031 }, { "epoch": 80.94704049844236, "grad_norm": 2.4375, "learning_rate": 3.2824239845050284e-05, "loss": 0.1555, "step": 13032 }, { "epoch": 80.95327102803738, "grad_norm": 1.578125, "learning_rate": 3.282189981637407e-05, "loss": 0.0949, "step": 13033 }, { "epoch": 80.9595015576324, "grad_norm": 1.984375, "learning_rate": 3.281955971173001e-05, "loss": 0.1458, "step": 13034 }, { "epoch": 80.96573208722741, "grad_norm": 1.9375, "learning_rate": 3.281721953114081e-05, "loss": 0.1449, "step": 13035 }, { "epoch": 80.97196261682242, "grad_norm": 1.9140625, "learning_rate": 3.281487927462922e-05, "loss": 0.1682, "step": 13036 }, { "epoch": 80.97819314641744, "grad_norm": 3.296875, "learning_rate": 3.2812538942217965e-05, "loss": 0.1689, "step": 13037 }, { "epoch": 80.98442367601245, "grad_norm": 2.453125, "learning_rate": 3.281019853392977e-05, "loss": 0.0972, "step": 13038 }, { "epoch": 80.99065420560747, "grad_norm": 1.4453125, "learning_rate": 3.280785804978735e-05, "loss": 0.1275, "step": 13039 }, { "epoch": 80.99688473520249, "grad_norm": 3.34375, "learning_rate": 3.280551748981347e-05, "loss": 0.2682, "step": 13040 }, { "epoch": 81.0, "grad_norm": 0.9296875, "learning_rate": 3.280317685403083e-05, "loss": 0.0484, "step": 13041 }, { "epoch": 81.00623052959502, "grad_norm": 1.6640625, "learning_rate": 3.280083614246218e-05, "loss": 0.13, "step": 13042 }, { "epoch": 81.01246105919003, "grad_norm": 1.703125, "learning_rate": 3.279849535513025e-05, "loss": 0.1124, "step": 13043 }, { "epoch": 81.01869158878505, "grad_norm": 1.0625, "learning_rate": 3.279615449205778e-05, "loss": 0.1143, "step": 13044 }, { "epoch": 81.02492211838006, "grad_norm": 3.390625, "learning_rate": 3.2793813553267494e-05, "loss": 0.418, "step": 13045 }, { "epoch": 81.03115264797508, "grad_norm": 2.359375, "learning_rate": 3.2791472538782136e-05, "loss": 0.1815, "step": 13046 }, { "epoch": 81.03738317757009, "grad_norm": 2.46875, "learning_rate": 3.278913144862444e-05, "loss": 0.1739, "step": 13047 }, { "epoch": 81.04361370716511, "grad_norm": 1.6171875, "learning_rate": 3.2786790282817134e-05, "loss": 0.1631, "step": 13048 }, { "epoch": 81.04984423676012, "grad_norm": 1.4609375, "learning_rate": 3.278444904138297e-05, "loss": 0.1131, "step": 13049 }, { "epoch": 81.05607476635514, "grad_norm": 2.328125, "learning_rate": 3.278210772434468e-05, "loss": 0.2141, "step": 13050 }, { "epoch": 81.06230529595015, "grad_norm": 1.9296875, "learning_rate": 3.2779766331725005e-05, "loss": 0.1554, "step": 13051 }, { "epoch": 81.06853582554517, "grad_norm": 0.8984375, "learning_rate": 3.277742486354668e-05, "loss": 0.0913, "step": 13052 }, { "epoch": 81.07476635514018, "grad_norm": 1.7109375, "learning_rate": 3.277508331983246e-05, "loss": 0.1866, "step": 13053 }, { "epoch": 81.0809968847352, "grad_norm": 1.359375, "learning_rate": 3.277274170060507e-05, "loss": 0.1153, "step": 13054 }, { "epoch": 81.08722741433021, "grad_norm": 1.6484375, "learning_rate": 3.277040000588726e-05, "loss": 0.1434, "step": 13055 }, { "epoch": 81.09345794392523, "grad_norm": 2.171875, "learning_rate": 3.276805823570177e-05, "loss": 0.2301, "step": 13056 }, { "epoch": 81.09968847352025, "grad_norm": 2.09375, "learning_rate": 3.276571639007135e-05, "loss": 0.1169, "step": 13057 }, { "epoch": 81.10591900311526, "grad_norm": 3.765625, "learning_rate": 3.276337446901874e-05, "loss": 0.2633, "step": 13058 }, { "epoch": 81.11214953271028, "grad_norm": 1.2734375, "learning_rate": 3.2761032472566676e-05, "loss": 0.1166, "step": 13059 }, { "epoch": 81.11838006230529, "grad_norm": 2.984375, "learning_rate": 3.275869040073792e-05, "loss": 0.2412, "step": 13060 }, { "epoch": 81.1246105919003, "grad_norm": 2.953125, "learning_rate": 3.2756348253555215e-05, "loss": 0.1824, "step": 13061 }, { "epoch": 81.13084112149532, "grad_norm": 3.0625, "learning_rate": 3.2754006031041306e-05, "loss": 0.2629, "step": 13062 }, { "epoch": 81.13707165109034, "grad_norm": 2.25, "learning_rate": 3.275166373321893e-05, "loss": 0.1824, "step": 13063 }, { "epoch": 81.14330218068535, "grad_norm": 3.21875, "learning_rate": 3.274932136011085e-05, "loss": 0.2796, "step": 13064 }, { "epoch": 81.14953271028037, "grad_norm": 1.296875, "learning_rate": 3.274697891173982e-05, "loss": 0.1171, "step": 13065 }, { "epoch": 81.15576323987538, "grad_norm": 2.0625, "learning_rate": 3.274463638812857e-05, "loss": 0.1024, "step": 13066 }, { "epoch": 81.1619937694704, "grad_norm": 2.6875, "learning_rate": 3.274229378929988e-05, "loss": 0.1294, "step": 13067 }, { "epoch": 81.16822429906541, "grad_norm": 2.125, "learning_rate": 3.273995111527647e-05, "loss": 0.1713, "step": 13068 }, { "epoch": 81.17445482866043, "grad_norm": 2.171875, "learning_rate": 3.273760836608112e-05, "loss": 0.1389, "step": 13069 }, { "epoch": 81.18068535825545, "grad_norm": 2.109375, "learning_rate": 3.273526554173655e-05, "loss": 0.1493, "step": 13070 }, { "epoch": 81.18691588785046, "grad_norm": 3.359375, "learning_rate": 3.273292264226555e-05, "loss": 0.2064, "step": 13071 }, { "epoch": 81.19314641744548, "grad_norm": 1.9921875, "learning_rate": 3.273057966769086e-05, "loss": 0.1382, "step": 13072 }, { "epoch": 81.19937694704049, "grad_norm": 2.859375, "learning_rate": 3.272823661803523e-05, "loss": 0.4031, "step": 13073 }, { "epoch": 81.20560747663552, "grad_norm": 1.5546875, "learning_rate": 3.2725893493321427e-05, "loss": 0.1191, "step": 13074 }, { "epoch": 81.21183800623054, "grad_norm": 3.359375, "learning_rate": 3.27235502935722e-05, "loss": 0.1722, "step": 13075 }, { "epoch": 81.21806853582555, "grad_norm": 2.5, "learning_rate": 3.2721207018810305e-05, "loss": 0.169, "step": 13076 }, { "epoch": 81.22429906542057, "grad_norm": 2.0625, "learning_rate": 3.2718863669058506e-05, "loss": 0.1902, "step": 13077 }, { "epoch": 81.23052959501558, "grad_norm": 1.9609375, "learning_rate": 3.271652024433957e-05, "loss": 0.1109, "step": 13078 }, { "epoch": 81.2367601246106, "grad_norm": 2.46875, "learning_rate": 3.271417674467623e-05, "loss": 0.1909, "step": 13079 }, { "epoch": 81.24299065420561, "grad_norm": 2.703125, "learning_rate": 3.2711833170091274e-05, "loss": 0.2634, "step": 13080 }, { "epoch": 81.24922118380063, "grad_norm": 1.5234375, "learning_rate": 3.270948952060745e-05, "loss": 0.226, "step": 13081 }, { "epoch": 81.25545171339564, "grad_norm": 2.234375, "learning_rate": 3.270714579624752e-05, "loss": 0.2157, "step": 13082 }, { "epoch": 81.26168224299066, "grad_norm": 3.21875, "learning_rate": 3.270480199703425e-05, "loss": 0.2736, "step": 13083 }, { "epoch": 81.26791277258567, "grad_norm": 2.75, "learning_rate": 3.27024581229904e-05, "loss": 0.2158, "step": 13084 }, { "epoch": 81.27414330218069, "grad_norm": 2.90625, "learning_rate": 3.2700114174138746e-05, "loss": 0.3162, "step": 13085 }, { "epoch": 81.2803738317757, "grad_norm": 1.34375, "learning_rate": 3.269777015050204e-05, "loss": 0.1269, "step": 13086 }, { "epoch": 81.28660436137072, "grad_norm": 1.84375, "learning_rate": 3.269542605210305e-05, "loss": 0.1347, "step": 13087 }, { "epoch": 81.29283489096574, "grad_norm": 1.2109375, "learning_rate": 3.2693081878964546e-05, "loss": 0.1459, "step": 13088 }, { "epoch": 81.29906542056075, "grad_norm": 2.765625, "learning_rate": 3.269073763110929e-05, "loss": 0.2638, "step": 13089 }, { "epoch": 81.30529595015577, "grad_norm": 2.296875, "learning_rate": 3.268839330856005e-05, "loss": 0.2278, "step": 13090 }, { "epoch": 81.31152647975078, "grad_norm": 2.296875, "learning_rate": 3.2686048911339604e-05, "loss": 0.2169, "step": 13091 }, { "epoch": 81.3177570093458, "grad_norm": 1.8359375, "learning_rate": 3.2683704439470716e-05, "loss": 0.1358, "step": 13092 }, { "epoch": 81.32398753894081, "grad_norm": 1.7265625, "learning_rate": 3.268135989297615e-05, "loss": 0.1185, "step": 13093 }, { "epoch": 81.33021806853583, "grad_norm": 1.3515625, "learning_rate": 3.267901527187869e-05, "loss": 0.1103, "step": 13094 }, { "epoch": 81.33644859813084, "grad_norm": 2.515625, "learning_rate": 3.267667057620109e-05, "loss": 0.172, "step": 13095 }, { "epoch": 81.34267912772586, "grad_norm": 1.75, "learning_rate": 3.267432580596613e-05, "loss": 0.1224, "step": 13096 }, { "epoch": 81.34890965732087, "grad_norm": 2.53125, "learning_rate": 3.267198096119659e-05, "loss": 0.1675, "step": 13097 }, { "epoch": 81.35514018691589, "grad_norm": 4.28125, "learning_rate": 3.2669636041915224e-05, "loss": 0.1192, "step": 13098 }, { "epoch": 81.3613707165109, "grad_norm": 4.0, "learning_rate": 3.266729104814483e-05, "loss": 0.2202, "step": 13099 }, { "epoch": 81.36760124610592, "grad_norm": 3.796875, "learning_rate": 3.2664945979908174e-05, "loss": 0.2686, "step": 13100 }, { "epoch": 81.37383177570094, "grad_norm": 1.375, "learning_rate": 3.266260083722803e-05, "loss": 0.1147, "step": 13101 }, { "epoch": 81.38006230529595, "grad_norm": 2.890625, "learning_rate": 3.2660255620127166e-05, "loss": 0.2456, "step": 13102 }, { "epoch": 81.38629283489097, "grad_norm": 2.5625, "learning_rate": 3.265791032862838e-05, "loss": 0.1216, "step": 13103 }, { "epoch": 81.39252336448598, "grad_norm": 3.015625, "learning_rate": 3.2655564962754434e-05, "loss": 0.1155, "step": 13104 }, { "epoch": 81.398753894081, "grad_norm": 4.71875, "learning_rate": 3.2653219522528106e-05, "loss": 0.1319, "step": 13105 }, { "epoch": 81.40498442367601, "grad_norm": 2.65625, "learning_rate": 3.2650874007972184e-05, "loss": 0.1988, "step": 13106 }, { "epoch": 81.41121495327103, "grad_norm": 2.25, "learning_rate": 3.264852841910945e-05, "loss": 0.1589, "step": 13107 }, { "epoch": 81.41744548286604, "grad_norm": 2.53125, "learning_rate": 3.2646182755962665e-05, "loss": 0.1759, "step": 13108 }, { "epoch": 81.42367601246106, "grad_norm": 2.046875, "learning_rate": 3.264383701855464e-05, "loss": 0.1461, "step": 13109 }, { "epoch": 81.42990654205607, "grad_norm": 3.375, "learning_rate": 3.264149120690813e-05, "loss": 0.1651, "step": 13110 }, { "epoch": 81.43613707165109, "grad_norm": 2.8125, "learning_rate": 3.263914532104593e-05, "loss": 0.3089, "step": 13111 }, { "epoch": 81.4423676012461, "grad_norm": 4.375, "learning_rate": 3.2636799360990835e-05, "loss": 0.1808, "step": 13112 }, { "epoch": 81.44859813084112, "grad_norm": 4.15625, "learning_rate": 3.2634453326765604e-05, "loss": 0.1112, "step": 13113 }, { "epoch": 81.45482866043614, "grad_norm": 3.234375, "learning_rate": 3.263210721839305e-05, "loss": 0.1464, "step": 13114 }, { "epoch": 81.46105919003115, "grad_norm": 1.9453125, "learning_rate": 3.262976103589594e-05, "loss": 0.1537, "step": 13115 }, { "epoch": 81.46728971962617, "grad_norm": 2.359375, "learning_rate": 3.262741477929706e-05, "loss": 0.1724, "step": 13116 }, { "epoch": 81.47352024922118, "grad_norm": 2.234375, "learning_rate": 3.262506844861921e-05, "loss": 0.1779, "step": 13117 }, { "epoch": 81.4797507788162, "grad_norm": 2.75, "learning_rate": 3.2622722043885165e-05, "loss": 0.1357, "step": 13118 }, { "epoch": 81.48598130841121, "grad_norm": 5.09375, "learning_rate": 3.2620375565117724e-05, "loss": 0.1227, "step": 13119 }, { "epoch": 81.49221183800623, "grad_norm": 3.84375, "learning_rate": 3.261802901233967e-05, "loss": 0.1305, "step": 13120 }, { "epoch": 81.49844236760124, "grad_norm": 3.921875, "learning_rate": 3.26156823855738e-05, "loss": 0.104, "step": 13121 }, { "epoch": 81.50467289719626, "grad_norm": 3.296875, "learning_rate": 3.2613335684842886e-05, "loss": 0.189, "step": 13122 }, { "epoch": 81.51090342679127, "grad_norm": 1.40625, "learning_rate": 3.261098891016976e-05, "loss": 0.1177, "step": 13123 }, { "epoch": 81.51713395638629, "grad_norm": 2.359375, "learning_rate": 3.260864206157717e-05, "loss": 0.2354, "step": 13124 }, { "epoch": 81.5233644859813, "grad_norm": 2.3125, "learning_rate": 3.2606295139087926e-05, "loss": 0.1791, "step": 13125 }, { "epoch": 81.52959501557632, "grad_norm": 2.96875, "learning_rate": 3.260394814272483e-05, "loss": 0.1126, "step": 13126 }, { "epoch": 81.53582554517133, "grad_norm": 3.078125, "learning_rate": 3.260160107251067e-05, "loss": 0.182, "step": 13127 }, { "epoch": 81.54205607476635, "grad_norm": 2.71875, "learning_rate": 3.259925392846823e-05, "loss": 0.1393, "step": 13128 }, { "epoch": 81.54828660436137, "grad_norm": 2.203125, "learning_rate": 3.259690671062033e-05, "loss": 0.1636, "step": 13129 }, { "epoch": 81.55451713395638, "grad_norm": 2.578125, "learning_rate": 3.259455941898974e-05, "loss": 0.2369, "step": 13130 }, { "epoch": 81.5607476635514, "grad_norm": 2.375, "learning_rate": 3.259221205359928e-05, "loss": 0.1985, "step": 13131 }, { "epoch": 81.56697819314641, "grad_norm": 2.609375, "learning_rate": 3.258986461447174e-05, "loss": 0.2018, "step": 13132 }, { "epoch": 81.57320872274143, "grad_norm": 1.609375, "learning_rate": 3.258751710162992e-05, "loss": 0.1422, "step": 13133 }, { "epoch": 81.57943925233644, "grad_norm": 1.3203125, "learning_rate": 3.2585169515096615e-05, "loss": 0.0938, "step": 13134 }, { "epoch": 81.58566978193146, "grad_norm": 1.484375, "learning_rate": 3.2582821854894626e-05, "loss": 0.119, "step": 13135 }, { "epoch": 81.59190031152647, "grad_norm": 2.484375, "learning_rate": 3.258047412104675e-05, "loss": 0.2603, "step": 13136 }, { "epoch": 81.59813084112149, "grad_norm": 2.40625, "learning_rate": 3.2578126313575795e-05, "loss": 0.2257, "step": 13137 }, { "epoch": 81.60436137071652, "grad_norm": 2.375, "learning_rate": 3.2575778432504564e-05, "loss": 0.1738, "step": 13138 }, { "epoch": 81.61059190031153, "grad_norm": 2.953125, "learning_rate": 3.2573430477855864e-05, "loss": 0.1865, "step": 13139 }, { "epoch": 81.61682242990655, "grad_norm": 2.140625, "learning_rate": 3.257108244965249e-05, "loss": 0.1968, "step": 13140 }, { "epoch": 81.62305295950156, "grad_norm": 2.578125, "learning_rate": 3.256873434791725e-05, "loss": 0.1717, "step": 13141 }, { "epoch": 81.62928348909658, "grad_norm": 2.46875, "learning_rate": 3.2566386172672954e-05, "loss": 0.1015, "step": 13142 }, { "epoch": 81.6355140186916, "grad_norm": 2.140625, "learning_rate": 3.25640379239424e-05, "loss": 0.1583, "step": 13143 }, { "epoch": 81.64174454828661, "grad_norm": 3.296875, "learning_rate": 3.256168960174839e-05, "loss": 0.2369, "step": 13144 }, { "epoch": 81.64797507788163, "grad_norm": 3.765625, "learning_rate": 3.255934120611375e-05, "loss": 0.383, "step": 13145 }, { "epoch": 81.65420560747664, "grad_norm": 1.7734375, "learning_rate": 3.255699273706126e-05, "loss": 0.1593, "step": 13146 }, { "epoch": 81.66043613707166, "grad_norm": 1.4140625, "learning_rate": 3.255464419461376e-05, "loss": 0.1195, "step": 13147 }, { "epoch": 81.66666666666667, "grad_norm": 2.953125, "learning_rate": 3.255229557879405e-05, "loss": 0.1997, "step": 13148 }, { "epoch": 81.67289719626169, "grad_norm": 2.453125, "learning_rate": 3.2549946889624926e-05, "loss": 0.2279, "step": 13149 }, { "epoch": 81.6791277258567, "grad_norm": 3.96875, "learning_rate": 3.254759812712921e-05, "loss": 0.2875, "step": 13150 }, { "epoch": 81.68535825545172, "grad_norm": 3.046875, "learning_rate": 3.2545249291329713e-05, "loss": 0.3074, "step": 13151 }, { "epoch": 81.69158878504673, "grad_norm": 2.25, "learning_rate": 3.254290038224924e-05, "loss": 0.1466, "step": 13152 }, { "epoch": 81.69781931464175, "grad_norm": 1.484375, "learning_rate": 3.2540551399910625e-05, "loss": 0.1546, "step": 13153 }, { "epoch": 81.70404984423676, "grad_norm": 1.8671875, "learning_rate": 3.253820234433666e-05, "loss": 0.123, "step": 13154 }, { "epoch": 81.71028037383178, "grad_norm": 3.4375, "learning_rate": 3.2535853215550166e-05, "loss": 0.2285, "step": 13155 }, { "epoch": 81.7165109034268, "grad_norm": 1.3671875, "learning_rate": 3.253350401357396e-05, "loss": 0.1173, "step": 13156 }, { "epoch": 81.72274143302181, "grad_norm": 3.046875, "learning_rate": 3.253115473843086e-05, "loss": 0.2146, "step": 13157 }, { "epoch": 81.72897196261682, "grad_norm": 2.53125, "learning_rate": 3.252880539014368e-05, "loss": 0.1686, "step": 13158 }, { "epoch": 81.73520249221184, "grad_norm": 1.6953125, "learning_rate": 3.252645596873523e-05, "loss": 0.1323, "step": 13159 }, { "epoch": 81.74143302180686, "grad_norm": 1.171875, "learning_rate": 3.252410647422834e-05, "loss": 0.0999, "step": 13160 }, { "epoch": 81.74766355140187, "grad_norm": 1.734375, "learning_rate": 3.252175690664582e-05, "loss": 0.1522, "step": 13161 }, { "epoch": 81.75389408099689, "grad_norm": 1.296875, "learning_rate": 3.25194072660105e-05, "loss": 0.1233, "step": 13162 }, { "epoch": 81.7601246105919, "grad_norm": 1.3125, "learning_rate": 3.2517057552345194e-05, "loss": 0.1143, "step": 13163 }, { "epoch": 81.76635514018692, "grad_norm": 3.046875, "learning_rate": 3.251470776567272e-05, "loss": 0.3563, "step": 13164 }, { "epoch": 81.77258566978193, "grad_norm": 2.40625, "learning_rate": 3.25123579060159e-05, "loss": 0.1196, "step": 13165 }, { "epoch": 81.77881619937695, "grad_norm": 2.265625, "learning_rate": 3.251000797339755e-05, "loss": 0.2019, "step": 13166 }, { "epoch": 81.78504672897196, "grad_norm": 2.484375, "learning_rate": 3.250765796784052e-05, "loss": 0.1879, "step": 13167 }, { "epoch": 81.79127725856698, "grad_norm": 2.703125, "learning_rate": 3.25053078893676e-05, "loss": 0.1387, "step": 13168 }, { "epoch": 81.797507788162, "grad_norm": 3.953125, "learning_rate": 3.250295773800164e-05, "loss": 0.3548, "step": 13169 }, { "epoch": 81.80373831775701, "grad_norm": 2.515625, "learning_rate": 3.2500607513765447e-05, "loss": 0.1663, "step": 13170 }, { "epoch": 81.80996884735202, "grad_norm": 3.265625, "learning_rate": 3.249825721668185e-05, "loss": 0.318, "step": 13171 }, { "epoch": 81.81619937694704, "grad_norm": 2.53125, "learning_rate": 3.2495906846773685e-05, "loss": 0.0934, "step": 13172 }, { "epoch": 81.82242990654206, "grad_norm": 1.9375, "learning_rate": 3.249355640406378e-05, "loss": 0.1308, "step": 13173 }, { "epoch": 81.82866043613707, "grad_norm": 1.4453125, "learning_rate": 3.249120588857495e-05, "loss": 0.136, "step": 13174 }, { "epoch": 81.83489096573209, "grad_norm": 1.3203125, "learning_rate": 3.248885530033004e-05, "loss": 0.1227, "step": 13175 }, { "epoch": 81.8411214953271, "grad_norm": 1.890625, "learning_rate": 3.2486504639351865e-05, "loss": 0.2453, "step": 13176 }, { "epoch": 81.84735202492212, "grad_norm": 2.5625, "learning_rate": 3.248415390566326e-05, "loss": 0.223, "step": 13177 }, { "epoch": 81.85358255451713, "grad_norm": 2.625, "learning_rate": 3.248180309928706e-05, "loss": 0.2359, "step": 13178 }, { "epoch": 81.85981308411215, "grad_norm": 1.5, "learning_rate": 3.2479452220246085e-05, "loss": 0.1364, "step": 13179 }, { "epoch": 81.86604361370716, "grad_norm": 2.46875, "learning_rate": 3.2477101268563184e-05, "loss": 0.2004, "step": 13180 }, { "epoch": 81.87227414330218, "grad_norm": 2.609375, "learning_rate": 3.2474750244261174e-05, "loss": 0.2343, "step": 13181 }, { "epoch": 81.8785046728972, "grad_norm": 3.421875, "learning_rate": 3.24723991473629e-05, "loss": 0.2736, "step": 13182 }, { "epoch": 81.88473520249221, "grad_norm": 1.5703125, "learning_rate": 3.2470047977891184e-05, "loss": 0.1227, "step": 13183 }, { "epoch": 81.89096573208722, "grad_norm": 3.78125, "learning_rate": 3.246769673586888e-05, "loss": 0.5057, "step": 13184 }, { "epoch": 81.89719626168224, "grad_norm": 1.4609375, "learning_rate": 3.24653454213188e-05, "loss": 0.144, "step": 13185 }, { "epoch": 81.90342679127725, "grad_norm": 2.421875, "learning_rate": 3.2462994034263805e-05, "loss": 0.1772, "step": 13186 }, { "epoch": 81.90965732087227, "grad_norm": 1.4296875, "learning_rate": 3.246064257472671e-05, "loss": 0.1053, "step": 13187 }, { "epoch": 81.91588785046729, "grad_norm": 2.765625, "learning_rate": 3.245829104273037e-05, "loss": 0.2631, "step": 13188 }, { "epoch": 81.9221183800623, "grad_norm": 2.984375, "learning_rate": 3.2455939438297614e-05, "loss": 0.3428, "step": 13189 }, { "epoch": 81.92834890965732, "grad_norm": 3.625, "learning_rate": 3.245358776145128e-05, "loss": 0.2915, "step": 13190 }, { "epoch": 81.93457943925233, "grad_norm": 3.328125, "learning_rate": 3.245123601221422e-05, "loss": 0.1469, "step": 13191 }, { "epoch": 81.94080996884735, "grad_norm": 2.40625, "learning_rate": 3.244888419060926e-05, "loss": 0.2497, "step": 13192 }, { "epoch": 81.94704049844236, "grad_norm": 2.140625, "learning_rate": 3.244653229665925e-05, "loss": 0.1463, "step": 13193 }, { "epoch": 81.95327102803738, "grad_norm": 2.859375, "learning_rate": 3.244418033038703e-05, "loss": 0.3012, "step": 13194 }, { "epoch": 81.9595015576324, "grad_norm": 4.71875, "learning_rate": 3.244182829181544e-05, "loss": 0.2953, "step": 13195 }, { "epoch": 81.96573208722741, "grad_norm": 2.984375, "learning_rate": 3.243947618096733e-05, "loss": 0.2265, "step": 13196 }, { "epoch": 81.97196261682242, "grad_norm": 2.4375, "learning_rate": 3.243712399786554e-05, "loss": 0.1479, "step": 13197 }, { "epoch": 81.97819314641744, "grad_norm": 2.4375, "learning_rate": 3.243477174253291e-05, "loss": 0.2074, "step": 13198 }, { "epoch": 81.98442367601245, "grad_norm": 1.828125, "learning_rate": 3.2432419414992297e-05, "loss": 0.1127, "step": 13199 }, { "epoch": 81.99065420560747, "grad_norm": 2.234375, "learning_rate": 3.243006701526654e-05, "loss": 0.1585, "step": 13200 }, { "epoch": 81.99688473520249, "grad_norm": 1.8984375, "learning_rate": 3.242771454337848e-05, "loss": 0.13, "step": 13201 }, { "epoch": 82.0, "grad_norm": 1.1171875, "learning_rate": 3.2425361999350975e-05, "loss": 0.0585, "step": 13202 }, { "epoch": 82.00623052959502, "grad_norm": 1.9296875, "learning_rate": 3.2423009383206876e-05, "loss": 0.2608, "step": 13203 }, { "epoch": 82.01246105919003, "grad_norm": 2.5, "learning_rate": 3.242065669496902e-05, "loss": 0.1353, "step": 13204 }, { "epoch": 82.01869158878505, "grad_norm": 2.09375, "learning_rate": 3.241830393466026e-05, "loss": 0.1408, "step": 13205 }, { "epoch": 82.02492211838006, "grad_norm": 1.640625, "learning_rate": 3.241595110230346e-05, "loss": 0.1296, "step": 13206 }, { "epoch": 82.03115264797508, "grad_norm": 2.46875, "learning_rate": 3.2413598197921444e-05, "loss": 0.1796, "step": 13207 }, { "epoch": 82.03738317757009, "grad_norm": 2.609375, "learning_rate": 3.24112452215371e-05, "loss": 0.1348, "step": 13208 }, { "epoch": 82.04361370716511, "grad_norm": 1.921875, "learning_rate": 3.240889217317325e-05, "loss": 0.1636, "step": 13209 }, { "epoch": 82.04984423676012, "grad_norm": 2.625, "learning_rate": 3.240653905285276e-05, "loss": 0.1499, "step": 13210 }, { "epoch": 82.05607476635514, "grad_norm": 2.515625, "learning_rate": 3.240418586059848e-05, "loss": 0.2297, "step": 13211 }, { "epoch": 82.06230529595015, "grad_norm": 2.140625, "learning_rate": 3.240183259643327e-05, "loss": 0.1454, "step": 13212 }, { "epoch": 82.06853582554517, "grad_norm": 2.296875, "learning_rate": 3.2399479260379984e-05, "loss": 0.1714, "step": 13213 }, { "epoch": 82.07476635514018, "grad_norm": 2.8125, "learning_rate": 3.239712585246147e-05, "loss": 0.2052, "step": 13214 }, { "epoch": 82.0809968847352, "grad_norm": 3.015625, "learning_rate": 3.23947723727006e-05, "loss": 0.1395, "step": 13215 }, { "epoch": 82.08722741433021, "grad_norm": 1.6953125, "learning_rate": 3.2392418821120205e-05, "loss": 0.1398, "step": 13216 }, { "epoch": 82.09345794392523, "grad_norm": 2.578125, "learning_rate": 3.239006519774318e-05, "loss": 0.1901, "step": 13217 }, { "epoch": 82.09968847352025, "grad_norm": 2.59375, "learning_rate": 3.238771150259236e-05, "loss": 0.161, "step": 13218 }, { "epoch": 82.10591900311526, "grad_norm": 2.375, "learning_rate": 3.23853577356906e-05, "loss": 0.0892, "step": 13219 }, { "epoch": 82.11214953271028, "grad_norm": 3.4375, "learning_rate": 3.238300389706077e-05, "loss": 0.3181, "step": 13220 }, { "epoch": 82.11838006230529, "grad_norm": 4.28125, "learning_rate": 3.238064998672575e-05, "loss": 0.2249, "step": 13221 }, { "epoch": 82.1246105919003, "grad_norm": 2.71875, "learning_rate": 3.2378296004708355e-05, "loss": 0.1193, "step": 13222 }, { "epoch": 82.13084112149532, "grad_norm": 1.9140625, "learning_rate": 3.237594195103149e-05, "loss": 0.1542, "step": 13223 }, { "epoch": 82.13707165109034, "grad_norm": 2.1875, "learning_rate": 3.2373587825718e-05, "loss": 0.1111, "step": 13224 }, { "epoch": 82.14330218068535, "grad_norm": 3.296875, "learning_rate": 3.237123362879075e-05, "loss": 0.2447, "step": 13225 }, { "epoch": 82.14953271028037, "grad_norm": 2.296875, "learning_rate": 3.2368879360272606e-05, "loss": 0.1304, "step": 13226 }, { "epoch": 82.15576323987538, "grad_norm": 2.734375, "learning_rate": 3.236652502018643e-05, "loss": 0.1589, "step": 13227 }, { "epoch": 82.1619937694704, "grad_norm": 2.703125, "learning_rate": 3.23641706085551e-05, "loss": 0.1209, "step": 13228 }, { "epoch": 82.16822429906541, "grad_norm": 3.578125, "learning_rate": 3.236181612540146e-05, "loss": 0.2305, "step": 13229 }, { "epoch": 82.17445482866043, "grad_norm": 2.1875, "learning_rate": 3.23594615707484e-05, "loss": 0.1403, "step": 13230 }, { "epoch": 82.18068535825545, "grad_norm": 2.359375, "learning_rate": 3.235710694461878e-05, "loss": 0.1556, "step": 13231 }, { "epoch": 82.18691588785046, "grad_norm": 2.5, "learning_rate": 3.235475224703546e-05, "loss": 0.1878, "step": 13232 }, { "epoch": 82.19314641744548, "grad_norm": 1.8671875, "learning_rate": 3.235239747802131e-05, "loss": 0.1443, "step": 13233 }, { "epoch": 82.19937694704049, "grad_norm": 2.40625, "learning_rate": 3.235004263759922e-05, "loss": 0.2667, "step": 13234 }, { "epoch": 82.20560747663552, "grad_norm": 1.3984375, "learning_rate": 3.234768772579203e-05, "loss": 0.1019, "step": 13235 }, { "epoch": 82.21183800623054, "grad_norm": 2.046875, "learning_rate": 3.234533274262264e-05, "loss": 0.1523, "step": 13236 }, { "epoch": 82.21806853582555, "grad_norm": 2.1875, "learning_rate": 3.2342977688113915e-05, "loss": 0.2237, "step": 13237 }, { "epoch": 82.22429906542057, "grad_norm": 2.265625, "learning_rate": 3.2340622562288714e-05, "loss": 0.1305, "step": 13238 }, { "epoch": 82.23052959501558, "grad_norm": 2.328125, "learning_rate": 3.233826736516992e-05, "loss": 0.2011, "step": 13239 }, { "epoch": 82.2367601246106, "grad_norm": 2.984375, "learning_rate": 3.23359120967804e-05, "loss": 0.1996, "step": 13240 }, { "epoch": 82.24299065420561, "grad_norm": 2.46875, "learning_rate": 3.233355675714305e-05, "loss": 0.1515, "step": 13241 }, { "epoch": 82.24922118380063, "grad_norm": 1.96875, "learning_rate": 3.233120134628072e-05, "loss": 0.1496, "step": 13242 }, { "epoch": 82.25545171339564, "grad_norm": 2.125, "learning_rate": 3.23288458642163e-05, "loss": 0.2011, "step": 13243 }, { "epoch": 82.26168224299066, "grad_norm": 1.96875, "learning_rate": 3.2326490310972665e-05, "loss": 0.1467, "step": 13244 }, { "epoch": 82.26791277258567, "grad_norm": 2.921875, "learning_rate": 3.23241346865727e-05, "loss": 0.1906, "step": 13245 }, { "epoch": 82.27414330218069, "grad_norm": 2.796875, "learning_rate": 3.2321778991039254e-05, "loss": 0.1359, "step": 13246 }, { "epoch": 82.2803738317757, "grad_norm": 0.9296875, "learning_rate": 3.231942322439525e-05, "loss": 0.1055, "step": 13247 }, { "epoch": 82.28660436137072, "grad_norm": 1.796875, "learning_rate": 3.2317067386663526e-05, "loss": 0.1126, "step": 13248 }, { "epoch": 82.29283489096574, "grad_norm": 2.75, "learning_rate": 3.231471147786699e-05, "loss": 0.2353, "step": 13249 }, { "epoch": 82.29906542056075, "grad_norm": 2.828125, "learning_rate": 3.2312355498028515e-05, "loss": 0.1471, "step": 13250 }, { "epoch": 82.30529595015577, "grad_norm": 1.671875, "learning_rate": 3.230999944717098e-05, "loss": 0.131, "step": 13251 }, { "epoch": 82.31152647975078, "grad_norm": 4.34375, "learning_rate": 3.230764332531727e-05, "loss": 0.2453, "step": 13252 }, { "epoch": 82.3177570093458, "grad_norm": 2.625, "learning_rate": 3.230528713249026e-05, "loss": 0.1418, "step": 13253 }, { "epoch": 82.32398753894081, "grad_norm": 3.078125, "learning_rate": 3.230293086871285e-05, "loss": 0.1982, "step": 13254 }, { "epoch": 82.33021806853583, "grad_norm": 2.921875, "learning_rate": 3.2300574534007906e-05, "loss": 0.2914, "step": 13255 }, { "epoch": 82.33644859813084, "grad_norm": 2.390625, "learning_rate": 3.229821812839834e-05, "loss": 0.2116, "step": 13256 }, { "epoch": 82.34267912772586, "grad_norm": 3.34375, "learning_rate": 3.2295861651907e-05, "loss": 0.2696, "step": 13257 }, { "epoch": 82.34890965732087, "grad_norm": 2.40625, "learning_rate": 3.2293505104556804e-05, "loss": 0.2392, "step": 13258 }, { "epoch": 82.35514018691589, "grad_norm": 1.65625, "learning_rate": 3.2291148486370626e-05, "loss": 0.1129, "step": 13259 }, { "epoch": 82.3613707165109, "grad_norm": 2.59375, "learning_rate": 3.228879179737136e-05, "loss": 0.1353, "step": 13260 }, { "epoch": 82.36760124610592, "grad_norm": 2.25, "learning_rate": 3.228643503758189e-05, "loss": 0.1515, "step": 13261 }, { "epoch": 82.37383177570094, "grad_norm": 1.7734375, "learning_rate": 3.2284078207025096e-05, "loss": 0.2283, "step": 13262 }, { "epoch": 82.38006230529595, "grad_norm": 2.734375, "learning_rate": 3.22817213057239e-05, "loss": 0.2239, "step": 13263 }, { "epoch": 82.38629283489097, "grad_norm": 2.0625, "learning_rate": 3.227936433370115e-05, "loss": 0.1559, "step": 13264 }, { "epoch": 82.39252336448598, "grad_norm": 1.1484375, "learning_rate": 3.227700729097977e-05, "loss": 0.0987, "step": 13265 }, { "epoch": 82.398753894081, "grad_norm": 2.6875, "learning_rate": 3.227465017758264e-05, "loss": 0.1473, "step": 13266 }, { "epoch": 82.40498442367601, "grad_norm": 2.1875, "learning_rate": 3.2272292993532646e-05, "loss": 0.1829, "step": 13267 }, { "epoch": 82.41121495327103, "grad_norm": 1.015625, "learning_rate": 3.2269935738852694e-05, "loss": 0.1044, "step": 13268 }, { "epoch": 82.41744548286604, "grad_norm": 1.78125, "learning_rate": 3.2267578413565675e-05, "loss": 0.2103, "step": 13269 }, { "epoch": 82.42367601246106, "grad_norm": 4.59375, "learning_rate": 3.226522101769447e-05, "loss": 0.3018, "step": 13270 }, { "epoch": 82.42990654205607, "grad_norm": 4.28125, "learning_rate": 3.2262863551262e-05, "loss": 0.1613, "step": 13271 }, { "epoch": 82.43613707165109, "grad_norm": 2.671875, "learning_rate": 3.226050601429115e-05, "loss": 0.1191, "step": 13272 }, { "epoch": 82.4423676012461, "grad_norm": 2.40625, "learning_rate": 3.22581484068048e-05, "loss": 0.1258, "step": 13273 }, { "epoch": 82.44859813084112, "grad_norm": 1.71875, "learning_rate": 3.225579072882588e-05, "loss": 0.0986, "step": 13274 }, { "epoch": 82.45482866043614, "grad_norm": 2.34375, "learning_rate": 3.2253432980377244e-05, "loss": 0.2024, "step": 13275 }, { "epoch": 82.46105919003115, "grad_norm": 4.6875, "learning_rate": 3.225107516148184e-05, "loss": 0.2272, "step": 13276 }, { "epoch": 82.46728971962617, "grad_norm": 3.078125, "learning_rate": 3.224871727216253e-05, "loss": 0.1067, "step": 13277 }, { "epoch": 82.47352024922118, "grad_norm": 4.28125, "learning_rate": 3.2246359312442236e-05, "loss": 0.1498, "step": 13278 }, { "epoch": 82.4797507788162, "grad_norm": 3.46875, "learning_rate": 3.224400128234385e-05, "loss": 0.1177, "step": 13279 }, { "epoch": 82.48598130841121, "grad_norm": 1.5078125, "learning_rate": 3.224164318189028e-05, "loss": 0.1118, "step": 13280 }, { "epoch": 82.49221183800623, "grad_norm": 2.53125, "learning_rate": 3.223928501110441e-05, "loss": 0.2415, "step": 13281 }, { "epoch": 82.49844236760124, "grad_norm": 2.375, "learning_rate": 3.223692677000918e-05, "loss": 0.2506, "step": 13282 }, { "epoch": 82.50467289719626, "grad_norm": 4.84375, "learning_rate": 3.223456845862745e-05, "loss": 0.1748, "step": 13283 }, { "epoch": 82.51090342679127, "grad_norm": 2.453125, "learning_rate": 3.2232210076982156e-05, "loss": 0.1308, "step": 13284 }, { "epoch": 82.51713395638629, "grad_norm": 2.125, "learning_rate": 3.2229851625096194e-05, "loss": 0.1605, "step": 13285 }, { "epoch": 82.5233644859813, "grad_norm": 2.15625, "learning_rate": 3.2227493102992454e-05, "loss": 0.1498, "step": 13286 }, { "epoch": 82.52959501557632, "grad_norm": 1.65625, "learning_rate": 3.2225134510693874e-05, "loss": 0.1322, "step": 13287 }, { "epoch": 82.53582554517133, "grad_norm": 1.765625, "learning_rate": 3.222277584822333e-05, "loss": 0.2189, "step": 13288 }, { "epoch": 82.54205607476635, "grad_norm": 2.859375, "learning_rate": 3.222041711560375e-05, "loss": 0.2487, "step": 13289 }, { "epoch": 82.54828660436137, "grad_norm": 2.796875, "learning_rate": 3.221805831285803e-05, "loss": 0.1961, "step": 13290 }, { "epoch": 82.55451713395638, "grad_norm": 2.96875, "learning_rate": 3.2215699440009094e-05, "loss": 0.1839, "step": 13291 }, { "epoch": 82.5607476635514, "grad_norm": 3.3125, "learning_rate": 3.221334049707983e-05, "loss": 0.4201, "step": 13292 }, { "epoch": 82.56697819314641, "grad_norm": 3.5625, "learning_rate": 3.221098148409317e-05, "loss": 0.2648, "step": 13293 }, { "epoch": 82.57320872274143, "grad_norm": 2.125, "learning_rate": 3.220862240107201e-05, "loss": 0.1455, "step": 13294 }, { "epoch": 82.57943925233644, "grad_norm": 1.859375, "learning_rate": 3.2206263248039276e-05, "loss": 0.1345, "step": 13295 }, { "epoch": 82.58566978193146, "grad_norm": 1.9140625, "learning_rate": 3.220390402501786e-05, "loss": 0.1308, "step": 13296 }, { "epoch": 82.59190031152647, "grad_norm": 2.890625, "learning_rate": 3.22015447320307e-05, "loss": 0.162, "step": 13297 }, { "epoch": 82.59813084112149, "grad_norm": 2.46875, "learning_rate": 3.21991853691007e-05, "loss": 0.1577, "step": 13298 }, { "epoch": 82.60436137071652, "grad_norm": 2.328125, "learning_rate": 3.219682593625077e-05, "loss": 0.1057, "step": 13299 }, { "epoch": 82.61059190031153, "grad_norm": 3.71875, "learning_rate": 3.2194466433503824e-05, "loss": 0.2981, "step": 13300 }, { "epoch": 82.61682242990655, "grad_norm": 1.546875, "learning_rate": 3.219210686088278e-05, "loss": 0.1367, "step": 13301 }, { "epoch": 82.62305295950156, "grad_norm": 1.5, "learning_rate": 3.2189747218410566e-05, "loss": 0.1154, "step": 13302 }, { "epoch": 82.62928348909658, "grad_norm": 3.0625, "learning_rate": 3.218738750611008e-05, "loss": 0.2614, "step": 13303 }, { "epoch": 82.6355140186916, "grad_norm": 1.40625, "learning_rate": 3.2185027724004256e-05, "loss": 0.1336, "step": 13304 }, { "epoch": 82.64174454828661, "grad_norm": 2.421875, "learning_rate": 3.2182667872116e-05, "loss": 0.2247, "step": 13305 }, { "epoch": 82.64797507788163, "grad_norm": 1.9609375, "learning_rate": 3.218030795046824e-05, "loss": 0.1514, "step": 13306 }, { "epoch": 82.65420560747664, "grad_norm": 3.03125, "learning_rate": 3.21779479590839e-05, "loss": 0.1963, "step": 13307 }, { "epoch": 82.66043613707166, "grad_norm": 2.46875, "learning_rate": 3.217558789798589e-05, "loss": 0.1672, "step": 13308 }, { "epoch": 82.66666666666667, "grad_norm": 1.328125, "learning_rate": 3.217322776719714e-05, "loss": 0.1029, "step": 13309 }, { "epoch": 82.67289719626169, "grad_norm": 3.0, "learning_rate": 3.217086756674056e-05, "loss": 0.2629, "step": 13310 }, { "epoch": 82.6791277258567, "grad_norm": 2.0625, "learning_rate": 3.216850729663909e-05, "loss": 0.1734, "step": 13311 }, { "epoch": 82.68535825545172, "grad_norm": 2.515625, "learning_rate": 3.2166146956915634e-05, "loss": 0.1436, "step": 13312 }, { "epoch": 82.69158878504673, "grad_norm": 1.6796875, "learning_rate": 3.216378654759313e-05, "loss": 0.1123, "step": 13313 }, { "epoch": 82.69781931464175, "grad_norm": 2.46875, "learning_rate": 3.21614260686945e-05, "loss": 0.1897, "step": 13314 }, { "epoch": 82.70404984423676, "grad_norm": 2.15625, "learning_rate": 3.215906552024267e-05, "loss": 0.1713, "step": 13315 }, { "epoch": 82.71028037383178, "grad_norm": 2.25, "learning_rate": 3.215670490226056e-05, "loss": 0.1997, "step": 13316 }, { "epoch": 82.7165109034268, "grad_norm": 2.625, "learning_rate": 3.215434421477112e-05, "loss": 0.151, "step": 13317 }, { "epoch": 82.72274143302181, "grad_norm": 1.6015625, "learning_rate": 3.215198345779723e-05, "loss": 0.151, "step": 13318 }, { "epoch": 82.72897196261682, "grad_norm": 2.953125, "learning_rate": 3.214962263136187e-05, "loss": 0.2684, "step": 13319 }, { "epoch": 82.73520249221184, "grad_norm": 1.6796875, "learning_rate": 3.2147261735487934e-05, "loss": 0.1643, "step": 13320 }, { "epoch": 82.74143302180686, "grad_norm": 2.46875, "learning_rate": 3.214490077019837e-05, "loss": 0.1934, "step": 13321 }, { "epoch": 82.74766355140187, "grad_norm": 2.640625, "learning_rate": 3.214253973551609e-05, "loss": 0.1254, "step": 13322 }, { "epoch": 82.75389408099689, "grad_norm": 2.328125, "learning_rate": 3.2140178631464046e-05, "loss": 0.1912, "step": 13323 }, { "epoch": 82.7601246105919, "grad_norm": 1.609375, "learning_rate": 3.213781745806516e-05, "loss": 0.1423, "step": 13324 }, { "epoch": 82.76635514018692, "grad_norm": 2.59375, "learning_rate": 3.2135456215342356e-05, "loss": 0.117, "step": 13325 }, { "epoch": 82.77258566978193, "grad_norm": 1.859375, "learning_rate": 3.213309490331859e-05, "loss": 0.096, "step": 13326 }, { "epoch": 82.77881619937695, "grad_norm": 1.640625, "learning_rate": 3.213073352201677e-05, "loss": 0.1241, "step": 13327 }, { "epoch": 82.78504672897196, "grad_norm": 2.4375, "learning_rate": 3.212837207145984e-05, "loss": 0.1752, "step": 13328 }, { "epoch": 82.79127725856698, "grad_norm": 2.515625, "learning_rate": 3.2126010551670733e-05, "loss": 0.1074, "step": 13329 }, { "epoch": 82.797507788162, "grad_norm": 2.46875, "learning_rate": 3.2123648962672396e-05, "loss": 0.1949, "step": 13330 }, { "epoch": 82.80373831775701, "grad_norm": 1.578125, "learning_rate": 3.212128730448774e-05, "loss": 0.1285, "step": 13331 }, { "epoch": 82.80996884735202, "grad_norm": 2.953125, "learning_rate": 3.211892557713973e-05, "loss": 0.274, "step": 13332 }, { "epoch": 82.81619937694704, "grad_norm": 2.71875, "learning_rate": 3.2116563780651296e-05, "loss": 0.1616, "step": 13333 }, { "epoch": 82.82242990654206, "grad_norm": 1.890625, "learning_rate": 3.211420191504537e-05, "loss": 0.1133, "step": 13334 }, { "epoch": 82.82866043613707, "grad_norm": 2.515625, "learning_rate": 3.2111839980344894e-05, "loss": 0.117, "step": 13335 }, { "epoch": 82.83489096573209, "grad_norm": 2.25, "learning_rate": 3.21094779765728e-05, "loss": 0.2008, "step": 13336 }, { "epoch": 82.8411214953271, "grad_norm": 1.5625, "learning_rate": 3.210711590375204e-05, "loss": 0.0988, "step": 13337 }, { "epoch": 82.84735202492212, "grad_norm": 1.5625, "learning_rate": 3.210475376190554e-05, "loss": 0.1298, "step": 13338 }, { "epoch": 82.85358255451713, "grad_norm": 3.125, "learning_rate": 3.2102391551056266e-05, "loss": 0.2734, "step": 13339 }, { "epoch": 82.85981308411215, "grad_norm": 1.0859375, "learning_rate": 3.210002927122713e-05, "loss": 0.1092, "step": 13340 }, { "epoch": 82.86604361370716, "grad_norm": 2.296875, "learning_rate": 3.20976669224411e-05, "loss": 0.247, "step": 13341 }, { "epoch": 82.87227414330218, "grad_norm": 1.203125, "learning_rate": 3.209530450472111e-05, "loss": 0.1003, "step": 13342 }, { "epoch": 82.8785046728972, "grad_norm": 2.390625, "learning_rate": 3.2092942018090104e-05, "loss": 0.1507, "step": 13343 }, { "epoch": 82.88473520249221, "grad_norm": 2.3125, "learning_rate": 3.2090579462571025e-05, "loss": 0.1765, "step": 13344 }, { "epoch": 82.89096573208722, "grad_norm": 2.3125, "learning_rate": 3.2088216838186826e-05, "loss": 0.1989, "step": 13345 }, { "epoch": 82.89719626168224, "grad_norm": 1.1484375, "learning_rate": 3.2085854144960444e-05, "loss": 0.1131, "step": 13346 }, { "epoch": 82.90342679127725, "grad_norm": 2.46875, "learning_rate": 3.2083491382914824e-05, "loss": 0.2117, "step": 13347 }, { "epoch": 82.90965732087227, "grad_norm": 2.546875, "learning_rate": 3.208112855207293e-05, "loss": 0.1104, "step": 13348 }, { "epoch": 82.91588785046729, "grad_norm": 2.078125, "learning_rate": 3.20787656524577e-05, "loss": 0.2555, "step": 13349 }, { "epoch": 82.9221183800623, "grad_norm": 1.7734375, "learning_rate": 3.2076402684092074e-05, "loss": 0.1531, "step": 13350 }, { "epoch": 82.92834890965732, "grad_norm": 2.21875, "learning_rate": 3.2074039646999013e-05, "loss": 0.184, "step": 13351 }, { "epoch": 82.93457943925233, "grad_norm": 2.828125, "learning_rate": 3.207167654120147e-05, "loss": 0.257, "step": 13352 }, { "epoch": 82.94080996884735, "grad_norm": 3.078125, "learning_rate": 3.206931336672239e-05, "loss": 0.202, "step": 13353 }, { "epoch": 82.94704049844236, "grad_norm": 1.78125, "learning_rate": 3.206695012358472e-05, "loss": 0.1278, "step": 13354 }, { "epoch": 82.95327102803738, "grad_norm": 2.0625, "learning_rate": 3.2064586811811426e-05, "loss": 0.1635, "step": 13355 }, { "epoch": 82.9595015576324, "grad_norm": 1.90625, "learning_rate": 3.2062223431425446e-05, "loss": 0.1532, "step": 13356 }, { "epoch": 82.96573208722741, "grad_norm": 2.46875, "learning_rate": 3.205985998244975e-05, "loss": 0.2121, "step": 13357 }, { "epoch": 82.97196261682242, "grad_norm": 2.140625, "learning_rate": 3.205749646490727e-05, "loss": 0.1457, "step": 13358 }, { "epoch": 82.97819314641744, "grad_norm": 1.7421875, "learning_rate": 3.205513287882098e-05, "loss": 0.1414, "step": 13359 }, { "epoch": 82.98442367601245, "grad_norm": 2.5, "learning_rate": 3.2052769224213823e-05, "loss": 0.1763, "step": 13360 }, { "epoch": 82.99065420560747, "grad_norm": 1.953125, "learning_rate": 3.205040550110877e-05, "loss": 0.1364, "step": 13361 }, { "epoch": 82.99688473520249, "grad_norm": 3.109375, "learning_rate": 3.2048041709528764e-05, "loss": 0.1439, "step": 13362 }, { "epoch": 83.0, "grad_norm": 1.2265625, "learning_rate": 3.204567784949677e-05, "loss": 0.0607, "step": 13363 }, { "epoch": 83.00623052959502, "grad_norm": 2.578125, "learning_rate": 3.2043313921035743e-05, "loss": 0.1538, "step": 13364 }, { "epoch": 83.01246105919003, "grad_norm": 1.234375, "learning_rate": 3.2040949924168654e-05, "loss": 0.0924, "step": 13365 }, { "epoch": 83.01869158878505, "grad_norm": 1.2890625, "learning_rate": 3.203858585891843e-05, "loss": 0.1015, "step": 13366 }, { "epoch": 83.02492211838006, "grad_norm": 0.85546875, "learning_rate": 3.2036221725308066e-05, "loss": 0.0891, "step": 13367 }, { "epoch": 83.03115264797508, "grad_norm": 2.140625, "learning_rate": 3.203385752336051e-05, "loss": 0.1997, "step": 13368 }, { "epoch": 83.03738317757009, "grad_norm": 2.0, "learning_rate": 3.2031493253098724e-05, "loss": 0.2343, "step": 13369 }, { "epoch": 83.04361370716511, "grad_norm": 2.4375, "learning_rate": 3.202912891454567e-05, "loss": 0.1652, "step": 13370 }, { "epoch": 83.04984423676012, "grad_norm": 3.28125, "learning_rate": 3.202676450772431e-05, "loss": 0.3732, "step": 13371 }, { "epoch": 83.05607476635514, "grad_norm": 2.4375, "learning_rate": 3.202440003265762e-05, "loss": 0.1448, "step": 13372 }, { "epoch": 83.06230529595015, "grad_norm": 2.71875, "learning_rate": 3.202203548936853e-05, "loss": 0.1597, "step": 13373 }, { "epoch": 83.06853582554517, "grad_norm": 2.109375, "learning_rate": 3.201967087788005e-05, "loss": 0.1397, "step": 13374 }, { "epoch": 83.07476635514018, "grad_norm": 1.65625, "learning_rate": 3.20173061982151e-05, "loss": 0.1034, "step": 13375 }, { "epoch": 83.0809968847352, "grad_norm": 2.5, "learning_rate": 3.201494145039669e-05, "loss": 0.2466, "step": 13376 }, { "epoch": 83.08722741433021, "grad_norm": 1.515625, "learning_rate": 3.2012576634447756e-05, "loss": 0.1138, "step": 13377 }, { "epoch": 83.09345794392523, "grad_norm": 2.828125, "learning_rate": 3.2010211750391276e-05, "loss": 0.144, "step": 13378 }, { "epoch": 83.09968847352025, "grad_norm": 2.921875, "learning_rate": 3.200784679825022e-05, "loss": 0.2375, "step": 13379 }, { "epoch": 83.10591900311526, "grad_norm": 1.59375, "learning_rate": 3.2005481778047553e-05, "loss": 0.1391, "step": 13380 }, { "epoch": 83.11214953271028, "grad_norm": 2.296875, "learning_rate": 3.200311668980625e-05, "loss": 0.1657, "step": 13381 }, { "epoch": 83.11838006230529, "grad_norm": 2.71875, "learning_rate": 3.200075153354928e-05, "loss": 0.1695, "step": 13382 }, { "epoch": 83.1246105919003, "grad_norm": 1.890625, "learning_rate": 3.199838630929961e-05, "loss": 0.126, "step": 13383 }, { "epoch": 83.13084112149532, "grad_norm": 1.8359375, "learning_rate": 3.199602101708021e-05, "loss": 0.1091, "step": 13384 }, { "epoch": 83.13707165109034, "grad_norm": 1.4609375, "learning_rate": 3.199365565691406e-05, "loss": 0.1182, "step": 13385 }, { "epoch": 83.14330218068535, "grad_norm": 3.71875, "learning_rate": 3.1991290228824113e-05, "loss": 0.3673, "step": 13386 }, { "epoch": 83.14953271028037, "grad_norm": 4.125, "learning_rate": 3.1988924732833384e-05, "loss": 0.1779, "step": 13387 }, { "epoch": 83.15576323987538, "grad_norm": 2.90625, "learning_rate": 3.1986559168964795e-05, "loss": 0.1849, "step": 13388 }, { "epoch": 83.1619937694704, "grad_norm": 2.984375, "learning_rate": 3.198419353724136e-05, "loss": 0.2735, "step": 13389 }, { "epoch": 83.16822429906541, "grad_norm": 2.171875, "learning_rate": 3.198182783768604e-05, "loss": 0.1988, "step": 13390 }, { "epoch": 83.17445482866043, "grad_norm": 2.28125, "learning_rate": 3.197946207032181e-05, "loss": 0.2074, "step": 13391 }, { "epoch": 83.18068535825545, "grad_norm": 2.875, "learning_rate": 3.197709623517165e-05, "loss": 0.2816, "step": 13392 }, { "epoch": 83.18691588785046, "grad_norm": 2.78125, "learning_rate": 3.1974730332258546e-05, "loss": 0.2912, "step": 13393 }, { "epoch": 83.19314641744548, "grad_norm": 1.953125, "learning_rate": 3.1972364361605454e-05, "loss": 0.1755, "step": 13394 }, { "epoch": 83.19937694704049, "grad_norm": 1.1015625, "learning_rate": 3.196999832323537e-05, "loss": 0.1097, "step": 13395 }, { "epoch": 83.20560747663552, "grad_norm": 2.265625, "learning_rate": 3.1967632217171275e-05, "loss": 0.1943, "step": 13396 }, { "epoch": 83.21183800623054, "grad_norm": 4.28125, "learning_rate": 3.196526604343614e-05, "loss": 0.2451, "step": 13397 }, { "epoch": 83.21806853582555, "grad_norm": 2.765625, "learning_rate": 3.1962899802052946e-05, "loss": 0.1348, "step": 13398 }, { "epoch": 83.22429906542057, "grad_norm": 1.8671875, "learning_rate": 3.196053349304467e-05, "loss": 0.1081, "step": 13399 }, { "epoch": 83.23052959501558, "grad_norm": 1.3046875, "learning_rate": 3.195816711643431e-05, "loss": 0.1076, "step": 13400 }, { "epoch": 83.2367601246106, "grad_norm": 2.0625, "learning_rate": 3.195580067224484e-05, "loss": 0.1317, "step": 13401 }, { "epoch": 83.24299065420561, "grad_norm": 1.640625, "learning_rate": 3.195343416049924e-05, "loss": 0.1142, "step": 13402 }, { "epoch": 83.24922118380063, "grad_norm": 2.390625, "learning_rate": 3.1951067581220506e-05, "loss": 0.1645, "step": 13403 }, { "epoch": 83.25545171339564, "grad_norm": 1.8125, "learning_rate": 3.194870093443161e-05, "loss": 0.13, "step": 13404 }, { "epoch": 83.26168224299066, "grad_norm": 2.109375, "learning_rate": 3.194633422015554e-05, "loss": 0.1612, "step": 13405 }, { "epoch": 83.26791277258567, "grad_norm": 1.890625, "learning_rate": 3.1943967438415276e-05, "loss": 0.1442, "step": 13406 }, { "epoch": 83.27414330218069, "grad_norm": 1.703125, "learning_rate": 3.194160058923382e-05, "loss": 0.1101, "step": 13407 }, { "epoch": 83.2803738317757, "grad_norm": 2.296875, "learning_rate": 3.193923367263415e-05, "loss": 0.2041, "step": 13408 }, { "epoch": 83.28660436137072, "grad_norm": 2.109375, "learning_rate": 3.1936866688639266e-05, "loss": 0.123, "step": 13409 }, { "epoch": 83.29283489096574, "grad_norm": 1.3515625, "learning_rate": 3.193449963727213e-05, "loss": 0.1138, "step": 13410 }, { "epoch": 83.29906542056075, "grad_norm": 2.3125, "learning_rate": 3.193213251855576e-05, "loss": 0.2056, "step": 13411 }, { "epoch": 83.30529595015577, "grad_norm": 1.8203125, "learning_rate": 3.192976533251312e-05, "loss": 0.1547, "step": 13412 }, { "epoch": 83.31152647975078, "grad_norm": 2.34375, "learning_rate": 3.1927398079167226e-05, "loss": 0.1328, "step": 13413 }, { "epoch": 83.3177570093458, "grad_norm": 4.0625, "learning_rate": 3.192503075854105e-05, "loss": 0.231, "step": 13414 }, { "epoch": 83.32398753894081, "grad_norm": 3.703125, "learning_rate": 3.192266337065759e-05, "loss": 0.1715, "step": 13415 }, { "epoch": 83.33021806853583, "grad_norm": 4.21875, "learning_rate": 3.192029591553985e-05, "loss": 0.4395, "step": 13416 }, { "epoch": 83.33644859813084, "grad_norm": 4.25, "learning_rate": 3.19179283932108e-05, "loss": 0.1926, "step": 13417 }, { "epoch": 83.34267912772586, "grad_norm": 3.390625, "learning_rate": 3.191556080369345e-05, "loss": 0.2343, "step": 13418 }, { "epoch": 83.34890965732087, "grad_norm": 1.9921875, "learning_rate": 3.191319314701079e-05, "loss": 0.1043, "step": 13419 }, { "epoch": 83.35514018691589, "grad_norm": 3.640625, "learning_rate": 3.191082542318583e-05, "loss": 0.2328, "step": 13420 }, { "epoch": 83.3613707165109, "grad_norm": 2.296875, "learning_rate": 3.190845763224153e-05, "loss": 0.1793, "step": 13421 }, { "epoch": 83.36760124610592, "grad_norm": 1.1640625, "learning_rate": 3.190608977420093e-05, "loss": 0.1158, "step": 13422 }, { "epoch": 83.37383177570094, "grad_norm": 1.8828125, "learning_rate": 3.190372184908699e-05, "loss": 0.1748, "step": 13423 }, { "epoch": 83.38006230529595, "grad_norm": 1.2890625, "learning_rate": 3.1901353856922725e-05, "loss": 0.1245, "step": 13424 }, { "epoch": 83.38629283489097, "grad_norm": 1.6484375, "learning_rate": 3.1898985797731143e-05, "loss": 0.1362, "step": 13425 }, { "epoch": 83.39252336448598, "grad_norm": 1.3046875, "learning_rate": 3.1896617671535224e-05, "loss": 0.1215, "step": 13426 }, { "epoch": 83.398753894081, "grad_norm": 3.71875, "learning_rate": 3.1894249478357965e-05, "loss": 0.2276, "step": 13427 }, { "epoch": 83.40498442367601, "grad_norm": 3.21875, "learning_rate": 3.18918812182224e-05, "loss": 0.2214, "step": 13428 }, { "epoch": 83.41121495327103, "grad_norm": 2.28125, "learning_rate": 3.1889512891151485e-05, "loss": 0.1599, "step": 13429 }, { "epoch": 83.41744548286604, "grad_norm": 1.8671875, "learning_rate": 3.1887144497168254e-05, "loss": 0.1499, "step": 13430 }, { "epoch": 83.42367601246106, "grad_norm": 1.7890625, "learning_rate": 3.18847760362957e-05, "loss": 0.1676, "step": 13431 }, { "epoch": 83.42990654205607, "grad_norm": 2.25, "learning_rate": 3.1882407508556824e-05, "loss": 0.2189, "step": 13432 }, { "epoch": 83.43613707165109, "grad_norm": 2.609375, "learning_rate": 3.188003891397463e-05, "loss": 0.2042, "step": 13433 }, { "epoch": 83.4423676012461, "grad_norm": 2.296875, "learning_rate": 3.187767025257212e-05, "loss": 0.2179, "step": 13434 }, { "epoch": 83.44859813084112, "grad_norm": 2.5, "learning_rate": 3.1875301524372306e-05, "loss": 0.1655, "step": 13435 }, { "epoch": 83.45482866043614, "grad_norm": 1.359375, "learning_rate": 3.187293272939819e-05, "loss": 0.1226, "step": 13436 }, { "epoch": 83.46105919003115, "grad_norm": 2.046875, "learning_rate": 3.1870563867672773e-05, "loss": 0.1308, "step": 13437 }, { "epoch": 83.46728971962617, "grad_norm": 1.3203125, "learning_rate": 3.1868194939219064e-05, "loss": 0.1146, "step": 13438 }, { "epoch": 83.47352024922118, "grad_norm": 2.140625, "learning_rate": 3.186582594406008e-05, "loss": 0.1795, "step": 13439 }, { "epoch": 83.4797507788162, "grad_norm": 1.765625, "learning_rate": 3.1863456882218815e-05, "loss": 0.1312, "step": 13440 }, { "epoch": 83.48598130841121, "grad_norm": 3.234375, "learning_rate": 3.186108775371829e-05, "loss": 0.2296, "step": 13441 }, { "epoch": 83.49221183800623, "grad_norm": 1.734375, "learning_rate": 3.1858718558581504e-05, "loss": 0.1505, "step": 13442 }, { "epoch": 83.49844236760124, "grad_norm": 1.3125, "learning_rate": 3.1856349296831484e-05, "loss": 0.094, "step": 13443 }, { "epoch": 83.50467289719626, "grad_norm": 2.609375, "learning_rate": 3.1853979968491224e-05, "loss": 0.2338, "step": 13444 }, { "epoch": 83.51090342679127, "grad_norm": 2.046875, "learning_rate": 3.1851610573583735e-05, "loss": 0.1413, "step": 13445 }, { "epoch": 83.51713395638629, "grad_norm": 2.546875, "learning_rate": 3.1849241112132036e-05, "loss": 0.1493, "step": 13446 }, { "epoch": 83.5233644859813, "grad_norm": 2.109375, "learning_rate": 3.184687158415914e-05, "loss": 0.1673, "step": 13447 }, { "epoch": 83.52959501557632, "grad_norm": 1.5078125, "learning_rate": 3.184450198968806e-05, "loss": 0.1161, "step": 13448 }, { "epoch": 83.53582554517133, "grad_norm": 2.390625, "learning_rate": 3.18421323287418e-05, "loss": 0.1286, "step": 13449 }, { "epoch": 83.54205607476635, "grad_norm": 2.578125, "learning_rate": 3.183976260134339e-05, "loss": 0.1346, "step": 13450 }, { "epoch": 83.54828660436137, "grad_norm": 2.25, "learning_rate": 3.1837392807515844e-05, "loss": 0.2012, "step": 13451 }, { "epoch": 83.55451713395638, "grad_norm": 1.796875, "learning_rate": 3.1835022947282166e-05, "loss": 0.1417, "step": 13452 }, { "epoch": 83.5607476635514, "grad_norm": 2.125, "learning_rate": 3.1832653020665384e-05, "loss": 0.1137, "step": 13453 }, { "epoch": 83.56697819314641, "grad_norm": 1.9609375, "learning_rate": 3.1830283027688503e-05, "loss": 0.1368, "step": 13454 }, { "epoch": 83.57320872274143, "grad_norm": 2.78125, "learning_rate": 3.182791296837455e-05, "loss": 0.2264, "step": 13455 }, { "epoch": 83.57943925233644, "grad_norm": 1.9609375, "learning_rate": 3.182554284274654e-05, "loss": 0.1863, "step": 13456 }, { "epoch": 83.58566978193146, "grad_norm": 3.296875, "learning_rate": 3.1823172650827496e-05, "loss": 0.1773, "step": 13457 }, { "epoch": 83.59190031152647, "grad_norm": 2.703125, "learning_rate": 3.182080239264043e-05, "loss": 0.1848, "step": 13458 }, { "epoch": 83.59813084112149, "grad_norm": 2.5, "learning_rate": 3.181843206820837e-05, "loss": 0.1398, "step": 13459 }, { "epoch": 83.60436137071652, "grad_norm": 3.203125, "learning_rate": 3.181606167755434e-05, "loss": 0.2755, "step": 13460 }, { "epoch": 83.61059190031153, "grad_norm": 2.203125, "learning_rate": 3.181369122070136e-05, "loss": 0.2155, "step": 13461 }, { "epoch": 83.61682242990655, "grad_norm": 3.546875, "learning_rate": 3.181132069767243e-05, "loss": 0.2592, "step": 13462 }, { "epoch": 83.62305295950156, "grad_norm": 1.609375, "learning_rate": 3.180895010849061e-05, "loss": 0.1318, "step": 13463 }, { "epoch": 83.62928348909658, "grad_norm": 2.578125, "learning_rate": 3.18065794531789e-05, "loss": 0.1916, "step": 13464 }, { "epoch": 83.6355140186916, "grad_norm": 1.53125, "learning_rate": 3.1804208731760326e-05, "loss": 0.1649, "step": 13465 }, { "epoch": 83.64174454828661, "grad_norm": 1.6875, "learning_rate": 3.180183794425792e-05, "loss": 0.1151, "step": 13466 }, { "epoch": 83.64797507788163, "grad_norm": 2.265625, "learning_rate": 3.179946709069471e-05, "loss": 0.2081, "step": 13467 }, { "epoch": 83.65420560747664, "grad_norm": 3.984375, "learning_rate": 3.179709617109371e-05, "loss": 0.2136, "step": 13468 }, { "epoch": 83.66043613707166, "grad_norm": 4.5625, "learning_rate": 3.1794725185477946e-05, "loss": 0.217, "step": 13469 }, { "epoch": 83.66666666666667, "grad_norm": 3.390625, "learning_rate": 3.1792354133870466e-05, "loss": 0.1818, "step": 13470 }, { "epoch": 83.67289719626169, "grad_norm": 1.78125, "learning_rate": 3.178998301629428e-05, "loss": 0.1023, "step": 13471 }, { "epoch": 83.6791277258567, "grad_norm": 1.4609375, "learning_rate": 3.178761183277242e-05, "loss": 0.1014, "step": 13472 }, { "epoch": 83.68535825545172, "grad_norm": 2.5625, "learning_rate": 3.178524058332791e-05, "loss": 0.174, "step": 13473 }, { "epoch": 83.69158878504673, "grad_norm": 3.53125, "learning_rate": 3.178286926798381e-05, "loss": 0.1426, "step": 13474 }, { "epoch": 83.69781931464175, "grad_norm": 1.6953125, "learning_rate": 3.17804978867631e-05, "loss": 0.1174, "step": 13475 }, { "epoch": 83.70404984423676, "grad_norm": 3.453125, "learning_rate": 3.177812643968886e-05, "loss": 0.1158, "step": 13476 }, { "epoch": 83.71028037383178, "grad_norm": 2.890625, "learning_rate": 3.177575492678408e-05, "loss": 0.2574, "step": 13477 }, { "epoch": 83.7165109034268, "grad_norm": 3.828125, "learning_rate": 3.1773383348071835e-05, "loss": 0.2289, "step": 13478 }, { "epoch": 83.72274143302181, "grad_norm": 1.6484375, "learning_rate": 3.177101170357513e-05, "loss": 0.1308, "step": 13479 }, { "epoch": 83.72897196261682, "grad_norm": 2.734375, "learning_rate": 3.176863999331701e-05, "loss": 0.1435, "step": 13480 }, { "epoch": 83.73520249221184, "grad_norm": 3.171875, "learning_rate": 3.176626821732049e-05, "loss": 0.2606, "step": 13481 }, { "epoch": 83.74143302180686, "grad_norm": 2.953125, "learning_rate": 3.1763896375608634e-05, "loss": 0.2818, "step": 13482 }, { "epoch": 83.74766355140187, "grad_norm": 3.078125, "learning_rate": 3.1761524468204465e-05, "loss": 0.2935, "step": 13483 }, { "epoch": 83.75389408099689, "grad_norm": 3.453125, "learning_rate": 3.175915249513101e-05, "loss": 0.1087, "step": 13484 }, { "epoch": 83.7601246105919, "grad_norm": 2.828125, "learning_rate": 3.175678045641132e-05, "loss": 0.1179, "step": 13485 }, { "epoch": 83.76635514018692, "grad_norm": 1.8515625, "learning_rate": 3.175440835206843e-05, "loss": 0.1406, "step": 13486 }, { "epoch": 83.77258566978193, "grad_norm": 2.75, "learning_rate": 3.1752036182125374e-05, "loss": 0.1748, "step": 13487 }, { "epoch": 83.77881619937695, "grad_norm": 2.0, "learning_rate": 3.1749663946605184e-05, "loss": 0.169, "step": 13488 }, { "epoch": 83.78504672897196, "grad_norm": 1.4375, "learning_rate": 3.1747291645530926e-05, "loss": 0.1195, "step": 13489 }, { "epoch": 83.79127725856698, "grad_norm": 2.9375, "learning_rate": 3.1744919278925605e-05, "loss": 0.1105, "step": 13490 }, { "epoch": 83.797507788162, "grad_norm": 2.5625, "learning_rate": 3.1742546846812294e-05, "loss": 0.1291, "step": 13491 }, { "epoch": 83.80373831775701, "grad_norm": 2.59375, "learning_rate": 3.174017434921401e-05, "loss": 0.1437, "step": 13492 }, { "epoch": 83.80996884735202, "grad_norm": 3.96875, "learning_rate": 3.173780178615381e-05, "loss": 0.167, "step": 13493 }, { "epoch": 83.81619937694704, "grad_norm": 2.0625, "learning_rate": 3.173542915765473e-05, "loss": 0.2189, "step": 13494 }, { "epoch": 83.82242990654206, "grad_norm": 3.25, "learning_rate": 3.173305646373982e-05, "loss": 0.2825, "step": 13495 }, { "epoch": 83.82866043613707, "grad_norm": 2.890625, "learning_rate": 3.1730683704432116e-05, "loss": 0.1211, "step": 13496 }, { "epoch": 83.83489096573209, "grad_norm": 2.796875, "learning_rate": 3.172831087975466e-05, "loss": 0.1431, "step": 13497 }, { "epoch": 83.8411214953271, "grad_norm": 3.3125, "learning_rate": 3.172593798973052e-05, "loss": 0.2641, "step": 13498 }, { "epoch": 83.84735202492212, "grad_norm": 1.96875, "learning_rate": 3.172356503438272e-05, "loss": 0.1057, "step": 13499 }, { "epoch": 83.85358255451713, "grad_norm": 3.15625, "learning_rate": 3.172119201373431e-05, "loss": 0.1983, "step": 13500 }, { "epoch": 83.85981308411215, "grad_norm": 3.15625, "learning_rate": 3.1718818927808335e-05, "loss": 0.1218, "step": 13501 }, { "epoch": 83.86604361370716, "grad_norm": 4.4375, "learning_rate": 3.171644577662785e-05, "loss": 0.387, "step": 13502 }, { "epoch": 83.87227414330218, "grad_norm": 3.15625, "learning_rate": 3.171407256021591e-05, "loss": 0.127, "step": 13503 }, { "epoch": 83.8785046728972, "grad_norm": 2.5625, "learning_rate": 3.171169927859554e-05, "loss": 0.1205, "step": 13504 }, { "epoch": 83.88473520249221, "grad_norm": 2.890625, "learning_rate": 3.170932593178982e-05, "loss": 0.1989, "step": 13505 }, { "epoch": 83.89096573208722, "grad_norm": 1.7265625, "learning_rate": 3.170695251982177e-05, "loss": 0.1429, "step": 13506 }, { "epoch": 83.89719626168224, "grad_norm": 3.046875, "learning_rate": 3.1704579042714466e-05, "loss": 0.2692, "step": 13507 }, { "epoch": 83.90342679127725, "grad_norm": 2.234375, "learning_rate": 3.170220550049095e-05, "loss": 0.2566, "step": 13508 }, { "epoch": 83.90965732087227, "grad_norm": 2.5625, "learning_rate": 3.169983189317428e-05, "loss": 0.2255, "step": 13509 }, { "epoch": 83.91588785046729, "grad_norm": 2.828125, "learning_rate": 3.1697458220787484e-05, "loss": 0.2053, "step": 13510 }, { "epoch": 83.9221183800623, "grad_norm": 3.5, "learning_rate": 3.1695084483353656e-05, "loss": 0.172, "step": 13511 }, { "epoch": 83.92834890965732, "grad_norm": 3.09375, "learning_rate": 3.169271068089581e-05, "loss": 0.3556, "step": 13512 }, { "epoch": 83.93457943925233, "grad_norm": 2.0625, "learning_rate": 3.169033681343704e-05, "loss": 0.1994, "step": 13513 }, { "epoch": 83.94080996884735, "grad_norm": 1.9921875, "learning_rate": 3.1687962881000366e-05, "loss": 0.1875, "step": 13514 }, { "epoch": 83.94704049844236, "grad_norm": 1.59375, "learning_rate": 3.168558888360887e-05, "loss": 0.1466, "step": 13515 }, { "epoch": 83.95327102803738, "grad_norm": 2.40625, "learning_rate": 3.168321482128559e-05, "loss": 0.2367, "step": 13516 }, { "epoch": 83.9595015576324, "grad_norm": 2.265625, "learning_rate": 3.1680840694053606e-05, "loss": 0.1188, "step": 13517 }, { "epoch": 83.96573208722741, "grad_norm": 1.359375, "learning_rate": 3.167846650193594e-05, "loss": 0.1141, "step": 13518 }, { "epoch": 83.97196261682242, "grad_norm": 1.484375, "learning_rate": 3.1676092244955685e-05, "loss": 0.1142, "step": 13519 }, { "epoch": 83.97819314641744, "grad_norm": 3.140625, "learning_rate": 3.1673717923135884e-05, "loss": 0.2272, "step": 13520 }, { "epoch": 83.98442367601245, "grad_norm": 1.9609375, "learning_rate": 3.167134353649961e-05, "loss": 0.1765, "step": 13521 }, { "epoch": 83.99065420560747, "grad_norm": 1.40625, "learning_rate": 3.1668969085069904e-05, "loss": 0.0965, "step": 13522 }, { "epoch": 83.99688473520249, "grad_norm": 1.9765625, "learning_rate": 3.166659456886983e-05, "loss": 0.1549, "step": 13523 }, { "epoch": 84.0, "grad_norm": 2.578125, "learning_rate": 3.166421998792248e-05, "loss": 0.2513, "step": 13524 }, { "epoch": 84.00623052959502, "grad_norm": 1.75, "learning_rate": 3.166184534225087e-05, "loss": 0.1305, "step": 13525 }, { "epoch": 84.01246105919003, "grad_norm": 3.15625, "learning_rate": 3.1659470631878104e-05, "loss": 0.2045, "step": 13526 }, { "epoch": 84.01869158878505, "grad_norm": 2.359375, "learning_rate": 3.1657095856827215e-05, "loss": 0.2094, "step": 13527 }, { "epoch": 84.02492211838006, "grad_norm": 2.34375, "learning_rate": 3.165472101712129e-05, "loss": 0.1285, "step": 13528 }, { "epoch": 84.03115264797508, "grad_norm": 1.34375, "learning_rate": 3.165234611278338e-05, "loss": 0.1233, "step": 13529 }, { "epoch": 84.03738317757009, "grad_norm": 3.453125, "learning_rate": 3.164997114383656e-05, "loss": 0.1642, "step": 13530 }, { "epoch": 84.04361370716511, "grad_norm": 1.59375, "learning_rate": 3.1647596110303884e-05, "loss": 0.1177, "step": 13531 }, { "epoch": 84.04984423676012, "grad_norm": 1.7578125, "learning_rate": 3.1645221012208425e-05, "loss": 0.1026, "step": 13532 }, { "epoch": 84.05607476635514, "grad_norm": 2.265625, "learning_rate": 3.164284584957326e-05, "loss": 0.1422, "step": 13533 }, { "epoch": 84.06230529595015, "grad_norm": 2.59375, "learning_rate": 3.1640470622421434e-05, "loss": 0.2529, "step": 13534 }, { "epoch": 84.06853582554517, "grad_norm": 2.75, "learning_rate": 3.163809533077604e-05, "loss": 0.1848, "step": 13535 }, { "epoch": 84.07476635514018, "grad_norm": 1.84375, "learning_rate": 3.1635719974660136e-05, "loss": 0.1651, "step": 13536 }, { "epoch": 84.0809968847352, "grad_norm": 2.140625, "learning_rate": 3.1633344554096795e-05, "loss": 0.1433, "step": 13537 }, { "epoch": 84.08722741433021, "grad_norm": 1.359375, "learning_rate": 3.163096906910908e-05, "loss": 0.1181, "step": 13538 }, { "epoch": 84.09345794392523, "grad_norm": 2.78125, "learning_rate": 3.162859351972007e-05, "loss": 0.1778, "step": 13539 }, { "epoch": 84.09968847352025, "grad_norm": 1.9296875, "learning_rate": 3.1626217905952836e-05, "loss": 0.1589, "step": 13540 }, { "epoch": 84.10591900311526, "grad_norm": 1.6953125, "learning_rate": 3.1623842227830445e-05, "loss": 0.1189, "step": 13541 }, { "epoch": 84.11214953271028, "grad_norm": 2.375, "learning_rate": 3.162146648537598e-05, "loss": 0.1515, "step": 13542 }, { "epoch": 84.11838006230529, "grad_norm": 2.375, "learning_rate": 3.161909067861251e-05, "loss": 0.1068, "step": 13543 }, { "epoch": 84.1246105919003, "grad_norm": 2.3125, "learning_rate": 3.1616714807563106e-05, "loss": 0.16, "step": 13544 }, { "epoch": 84.13084112149532, "grad_norm": 1.375, "learning_rate": 3.1614338872250836e-05, "loss": 0.1285, "step": 13545 }, { "epoch": 84.13707165109034, "grad_norm": 1.8671875, "learning_rate": 3.16119628726988e-05, "loss": 0.1219, "step": 13546 }, { "epoch": 84.14330218068535, "grad_norm": 2.0, "learning_rate": 3.160958680893005e-05, "loss": 0.1193, "step": 13547 }, { "epoch": 84.14953271028037, "grad_norm": 3.140625, "learning_rate": 3.160721068096768e-05, "loss": 0.19, "step": 13548 }, { "epoch": 84.15576323987538, "grad_norm": 2.328125, "learning_rate": 3.160483448883475e-05, "loss": 0.137, "step": 13549 }, { "epoch": 84.1619937694704, "grad_norm": 1.9296875, "learning_rate": 3.1602458232554345e-05, "loss": 0.1393, "step": 13550 }, { "epoch": 84.16822429906541, "grad_norm": 2.09375, "learning_rate": 3.160008191214955e-05, "loss": 0.1056, "step": 13551 }, { "epoch": 84.17445482866043, "grad_norm": 2.96875, "learning_rate": 3.1597705527643444e-05, "loss": 0.2468, "step": 13552 }, { "epoch": 84.18068535825545, "grad_norm": 2.84375, "learning_rate": 3.15953290790591e-05, "loss": 0.1427, "step": 13553 }, { "epoch": 84.18691588785046, "grad_norm": 2.53125, "learning_rate": 3.15929525664196e-05, "loss": 0.1016, "step": 13554 }, { "epoch": 84.19314641744548, "grad_norm": 1.375, "learning_rate": 3.159057598974803e-05, "loss": 0.1374, "step": 13555 }, { "epoch": 84.19937694704049, "grad_norm": 1.2890625, "learning_rate": 3.158819934906746e-05, "loss": 0.1083, "step": 13556 }, { "epoch": 84.20560747663552, "grad_norm": 1.6171875, "learning_rate": 3.1585822644400994e-05, "loss": 0.1322, "step": 13557 }, { "epoch": 84.21183800623054, "grad_norm": 2.984375, "learning_rate": 3.158344587577169e-05, "loss": 0.2251, "step": 13558 }, { "epoch": 84.21806853582555, "grad_norm": 2.046875, "learning_rate": 3.1581069043202655e-05, "loss": 0.1422, "step": 13559 }, { "epoch": 84.22429906542057, "grad_norm": 2.78125, "learning_rate": 3.1578692146716956e-05, "loss": 0.2479, "step": 13560 }, { "epoch": 84.23052959501558, "grad_norm": 2.875, "learning_rate": 3.157631518633768e-05, "loss": 0.2162, "step": 13561 }, { "epoch": 84.2367601246106, "grad_norm": 2.546875, "learning_rate": 3.157393816208792e-05, "loss": 0.2212, "step": 13562 }, { "epoch": 84.24299065420561, "grad_norm": 1.7265625, "learning_rate": 3.157156107399076e-05, "loss": 0.1059, "step": 13563 }, { "epoch": 84.24922118380063, "grad_norm": 1.6015625, "learning_rate": 3.156918392206929e-05, "loss": 0.1391, "step": 13564 }, { "epoch": 84.25545171339564, "grad_norm": 2.28125, "learning_rate": 3.156680670634658e-05, "loss": 0.1141, "step": 13565 }, { "epoch": 84.26168224299066, "grad_norm": 2.890625, "learning_rate": 3.1564429426845745e-05, "loss": 0.1417, "step": 13566 }, { "epoch": 84.26791277258567, "grad_norm": 2.75, "learning_rate": 3.156205208358984e-05, "loss": 0.2096, "step": 13567 }, { "epoch": 84.27414330218069, "grad_norm": 2.109375, "learning_rate": 3.1559674676601994e-05, "loss": 0.127, "step": 13568 }, { "epoch": 84.2803738317757, "grad_norm": 2.4375, "learning_rate": 3.155729720590526e-05, "loss": 0.197, "step": 13569 }, { "epoch": 84.28660436137072, "grad_norm": 2.03125, "learning_rate": 3.155491967152275e-05, "loss": 0.1384, "step": 13570 }, { "epoch": 84.29283489096574, "grad_norm": 1.3359375, "learning_rate": 3.1552542073477555e-05, "loss": 0.0856, "step": 13571 }, { "epoch": 84.29906542056075, "grad_norm": 3.5625, "learning_rate": 3.155016441179276e-05, "loss": 0.1166, "step": 13572 }, { "epoch": 84.30529595015577, "grad_norm": 2.4375, "learning_rate": 3.154778668649146e-05, "loss": 0.1919, "step": 13573 }, { "epoch": 84.31152647975078, "grad_norm": 2.171875, "learning_rate": 3.154540889759674e-05, "loss": 0.1616, "step": 13574 }, { "epoch": 84.3177570093458, "grad_norm": 1.59375, "learning_rate": 3.15430310451317e-05, "loss": 0.107, "step": 13575 }, { "epoch": 84.32398753894081, "grad_norm": 1.1015625, "learning_rate": 3.1540653129119445e-05, "loss": 0.0812, "step": 13576 }, { "epoch": 84.33021806853583, "grad_norm": 2.484375, "learning_rate": 3.153827514958305e-05, "loss": 0.195, "step": 13577 }, { "epoch": 84.33644859813084, "grad_norm": 2.296875, "learning_rate": 3.1535897106545634e-05, "loss": 0.1198, "step": 13578 }, { "epoch": 84.34267912772586, "grad_norm": 2.375, "learning_rate": 3.153351900003027e-05, "loss": 0.2027, "step": 13579 }, { "epoch": 84.34890965732087, "grad_norm": 2.390625, "learning_rate": 3.153114083006004e-05, "loss": 0.1472, "step": 13580 }, { "epoch": 84.35514018691589, "grad_norm": 1.875, "learning_rate": 3.1528762596658104e-05, "loss": 0.2278, "step": 13581 }, { "epoch": 84.3613707165109, "grad_norm": 2.296875, "learning_rate": 3.1526384299847495e-05, "loss": 0.1883, "step": 13582 }, { "epoch": 84.36760124610592, "grad_norm": 2.453125, "learning_rate": 3.1524005939651343e-05, "loss": 0.1352, "step": 13583 }, { "epoch": 84.37383177570094, "grad_norm": 4.21875, "learning_rate": 3.1521627516092743e-05, "loss": 0.1887, "step": 13584 }, { "epoch": 84.38006230529595, "grad_norm": 2.6875, "learning_rate": 3.1519249029194795e-05, "loss": 0.2139, "step": 13585 }, { "epoch": 84.38629283489097, "grad_norm": 1.8125, "learning_rate": 3.151687047898059e-05, "loss": 0.1328, "step": 13586 }, { "epoch": 84.39252336448598, "grad_norm": 1.9453125, "learning_rate": 3.151449186547325e-05, "loss": 0.1287, "step": 13587 }, { "epoch": 84.398753894081, "grad_norm": 1.46875, "learning_rate": 3.151211318869585e-05, "loss": 0.1399, "step": 13588 }, { "epoch": 84.40498442367601, "grad_norm": 2.265625, "learning_rate": 3.150973444867151e-05, "loss": 0.1713, "step": 13589 }, { "epoch": 84.41121495327103, "grad_norm": 2.453125, "learning_rate": 3.150735564542332e-05, "loss": 0.1424, "step": 13590 }, { "epoch": 84.41744548286604, "grad_norm": 3.5, "learning_rate": 3.1504976778974404e-05, "loss": 0.1151, "step": 13591 }, { "epoch": 84.42367601246106, "grad_norm": 2.828125, "learning_rate": 3.1502597849347845e-05, "loss": 0.1867, "step": 13592 }, { "epoch": 84.42990654205607, "grad_norm": 1.84375, "learning_rate": 3.150021885656676e-05, "loss": 0.1301, "step": 13593 }, { "epoch": 84.43613707165109, "grad_norm": 3.8125, "learning_rate": 3.149783980065425e-05, "loss": 0.2608, "step": 13594 }, { "epoch": 84.4423676012461, "grad_norm": 2.09375, "learning_rate": 3.1495460681633414e-05, "loss": 0.1295, "step": 13595 }, { "epoch": 84.44859813084112, "grad_norm": 1.6796875, "learning_rate": 3.1493081499527374e-05, "loss": 0.1311, "step": 13596 }, { "epoch": 84.45482866043614, "grad_norm": 4.28125, "learning_rate": 3.149070225435923e-05, "loss": 0.3023, "step": 13597 }, { "epoch": 84.46105919003115, "grad_norm": 4.0625, "learning_rate": 3.148832294615208e-05, "loss": 0.1402, "step": 13598 }, { "epoch": 84.46728971962617, "grad_norm": 5.46875, "learning_rate": 3.148594357492904e-05, "loss": 0.3142, "step": 13599 }, { "epoch": 84.47352024922118, "grad_norm": 3.65625, "learning_rate": 3.1483564140713226e-05, "loss": 0.2051, "step": 13600 }, { "epoch": 84.4797507788162, "grad_norm": 3.328125, "learning_rate": 3.148118464352774e-05, "loss": 0.1589, "step": 13601 }, { "epoch": 84.48598130841121, "grad_norm": 1.7578125, "learning_rate": 3.147880508339569e-05, "loss": 0.1449, "step": 13602 }, { "epoch": 84.49221183800623, "grad_norm": 2.0625, "learning_rate": 3.14764254603402e-05, "loss": 0.1078, "step": 13603 }, { "epoch": 84.49844236760124, "grad_norm": 2.890625, "learning_rate": 3.1474045774384355e-05, "loss": 0.1261, "step": 13604 }, { "epoch": 84.50467289719626, "grad_norm": 1.4375, "learning_rate": 3.14716660255513e-05, "loss": 0.1356, "step": 13605 }, { "epoch": 84.51090342679127, "grad_norm": 3.921875, "learning_rate": 3.1469286213864105e-05, "loss": 0.1831, "step": 13606 }, { "epoch": 84.51713395638629, "grad_norm": 2.453125, "learning_rate": 3.1466906339345934e-05, "loss": 0.1179, "step": 13607 }, { "epoch": 84.5233644859813, "grad_norm": 3.734375, "learning_rate": 3.146452640201987e-05, "loss": 0.3221, "step": 13608 }, { "epoch": 84.52959501557632, "grad_norm": 2.625, "learning_rate": 3.1462146401909026e-05, "loss": 0.1711, "step": 13609 }, { "epoch": 84.53582554517133, "grad_norm": 3.234375, "learning_rate": 3.145976633903653e-05, "loss": 0.2381, "step": 13610 }, { "epoch": 84.54205607476635, "grad_norm": 4.4375, "learning_rate": 3.145738621342549e-05, "loss": 0.2837, "step": 13611 }, { "epoch": 84.54828660436137, "grad_norm": 3.59375, "learning_rate": 3.145500602509903e-05, "loss": 0.1413, "step": 13612 }, { "epoch": 84.55451713395638, "grad_norm": 1.1953125, "learning_rate": 3.1452625774080255e-05, "loss": 0.1185, "step": 13613 }, { "epoch": 84.5607476635514, "grad_norm": 2.375, "learning_rate": 3.14502454603923e-05, "loss": 0.1844, "step": 13614 }, { "epoch": 84.56697819314641, "grad_norm": 2.875, "learning_rate": 3.1447865084058254e-05, "loss": 0.1317, "step": 13615 }, { "epoch": 84.57320872274143, "grad_norm": 2.296875, "learning_rate": 3.144548464510127e-05, "loss": 0.192, "step": 13616 }, { "epoch": 84.57943925233644, "grad_norm": 2.046875, "learning_rate": 3.144310414354444e-05, "loss": 0.1627, "step": 13617 }, { "epoch": 84.58566978193146, "grad_norm": 1.796875, "learning_rate": 3.14407235794109e-05, "loss": 0.1327, "step": 13618 }, { "epoch": 84.59190031152647, "grad_norm": 1.3515625, "learning_rate": 3.143834295272376e-05, "loss": 0.121, "step": 13619 }, { "epoch": 84.59813084112149, "grad_norm": 3.0, "learning_rate": 3.143596226350615e-05, "loss": 0.2254, "step": 13620 }, { "epoch": 84.60436137071652, "grad_norm": 2.53125, "learning_rate": 3.143358151178118e-05, "loss": 0.1914, "step": 13621 }, { "epoch": 84.61059190031153, "grad_norm": 2.140625, "learning_rate": 3.1431200697571996e-05, "loss": 0.1398, "step": 13622 }, { "epoch": 84.61682242990655, "grad_norm": 1.96875, "learning_rate": 3.142881982090169e-05, "loss": 0.1262, "step": 13623 }, { "epoch": 84.62305295950156, "grad_norm": 1.953125, "learning_rate": 3.142643888179341e-05, "loss": 0.1256, "step": 13624 }, { "epoch": 84.62928348909658, "grad_norm": 2.8125, "learning_rate": 3.142405788027026e-05, "loss": 0.3117, "step": 13625 }, { "epoch": 84.6355140186916, "grad_norm": 3.03125, "learning_rate": 3.142167681635539e-05, "loss": 0.3, "step": 13626 }, { "epoch": 84.64174454828661, "grad_norm": 2.703125, "learning_rate": 3.1419295690071907e-05, "loss": 0.1759, "step": 13627 }, { "epoch": 84.64797507788163, "grad_norm": 4.875, "learning_rate": 3.1416914501442935e-05, "loss": 0.1684, "step": 13628 }, { "epoch": 84.65420560747664, "grad_norm": 1.734375, "learning_rate": 3.141453325049161e-05, "loss": 0.111, "step": 13629 }, { "epoch": 84.66043613707166, "grad_norm": 2.078125, "learning_rate": 3.141215193724106e-05, "loss": 0.1306, "step": 13630 }, { "epoch": 84.66666666666667, "grad_norm": 3.1875, "learning_rate": 3.14097705617144e-05, "loss": 0.3026, "step": 13631 }, { "epoch": 84.67289719626169, "grad_norm": 3.09375, "learning_rate": 3.1407389123934774e-05, "loss": 0.3332, "step": 13632 }, { "epoch": 84.6791277258567, "grad_norm": 3.171875, "learning_rate": 3.14050076239253e-05, "loss": 0.1836, "step": 13633 }, { "epoch": 84.68535825545172, "grad_norm": 3.28125, "learning_rate": 3.140262606170911e-05, "loss": 0.2488, "step": 13634 }, { "epoch": 84.69158878504673, "grad_norm": 3.15625, "learning_rate": 3.1400244437309346e-05, "loss": 0.2293, "step": 13635 }, { "epoch": 84.69781931464175, "grad_norm": 2.3125, "learning_rate": 3.1397862750749115e-05, "loss": 0.2101, "step": 13636 }, { "epoch": 84.70404984423676, "grad_norm": 2.046875, "learning_rate": 3.139548100205158e-05, "loss": 0.1466, "step": 13637 }, { "epoch": 84.71028037383178, "grad_norm": 2.09375, "learning_rate": 3.139309919123984e-05, "loss": 0.2287, "step": 13638 }, { "epoch": 84.7165109034268, "grad_norm": 2.671875, "learning_rate": 3.139071731833705e-05, "loss": 0.206, "step": 13639 }, { "epoch": 84.72274143302181, "grad_norm": 2.265625, "learning_rate": 3.138833538336633e-05, "loss": 0.2793, "step": 13640 }, { "epoch": 84.72897196261682, "grad_norm": 3.125, "learning_rate": 3.138595338635082e-05, "loss": 0.2356, "step": 13641 }, { "epoch": 84.73520249221184, "grad_norm": 3.234375, "learning_rate": 3.138357132731366e-05, "loss": 0.2662, "step": 13642 }, { "epoch": 84.74143302180686, "grad_norm": 3.109375, "learning_rate": 3.1381189206277975e-05, "loss": 0.1213, "step": 13643 }, { "epoch": 84.74766355140187, "grad_norm": 2.4375, "learning_rate": 3.13788070232669e-05, "loss": 0.2336, "step": 13644 }, { "epoch": 84.75389408099689, "grad_norm": 1.7734375, "learning_rate": 3.137642477830359e-05, "loss": 0.126, "step": 13645 }, { "epoch": 84.7601246105919, "grad_norm": 1.453125, "learning_rate": 3.1374042471411155e-05, "loss": 0.1425, "step": 13646 }, { "epoch": 84.76635514018692, "grad_norm": 1.953125, "learning_rate": 3.1371660102612757e-05, "loss": 0.1538, "step": 13647 }, { "epoch": 84.77258566978193, "grad_norm": 1.6875, "learning_rate": 3.136927767193151e-05, "loss": 0.1563, "step": 13648 }, { "epoch": 84.77881619937695, "grad_norm": 2.0, "learning_rate": 3.136689517939058e-05, "loss": 0.1187, "step": 13649 }, { "epoch": 84.78504672897196, "grad_norm": 2.421875, "learning_rate": 3.136451262501307e-05, "loss": 0.1946, "step": 13650 }, { "epoch": 84.79127725856698, "grad_norm": 2.09375, "learning_rate": 3.136213000882216e-05, "loss": 0.1395, "step": 13651 }, { "epoch": 84.797507788162, "grad_norm": 1.671875, "learning_rate": 3.135974733084096e-05, "loss": 0.1588, "step": 13652 }, { "epoch": 84.80373831775701, "grad_norm": 2.015625, "learning_rate": 3.135736459109263e-05, "loss": 0.1408, "step": 13653 }, { "epoch": 84.80996884735202, "grad_norm": 2.234375, "learning_rate": 3.13549817896003e-05, "loss": 0.1355, "step": 13654 }, { "epoch": 84.81619937694704, "grad_norm": 0.97265625, "learning_rate": 3.1352598926387124e-05, "loss": 0.0863, "step": 13655 }, { "epoch": 84.82242990654206, "grad_norm": 1.75, "learning_rate": 3.135021600147622e-05, "loss": 0.1324, "step": 13656 }, { "epoch": 84.82866043613707, "grad_norm": 3.4375, "learning_rate": 3.134783301489077e-05, "loss": 0.2675, "step": 13657 }, { "epoch": 84.83489096573209, "grad_norm": 2.515625, "learning_rate": 3.134544996665388e-05, "loss": 0.1157, "step": 13658 }, { "epoch": 84.8411214953271, "grad_norm": 1.7265625, "learning_rate": 3.134306685678872e-05, "loss": 0.1176, "step": 13659 }, { "epoch": 84.84735202492212, "grad_norm": 2.171875, "learning_rate": 3.134068368531843e-05, "loss": 0.1523, "step": 13660 }, { "epoch": 84.85358255451713, "grad_norm": 1.4765625, "learning_rate": 3.1338300452266145e-05, "loss": 0.1243, "step": 13661 }, { "epoch": 84.85981308411215, "grad_norm": 2.65625, "learning_rate": 3.133591715765502e-05, "loss": 0.1898, "step": 13662 }, { "epoch": 84.86604361370716, "grad_norm": 2.578125, "learning_rate": 3.1333533801508204e-05, "loss": 0.2425, "step": 13663 }, { "epoch": 84.87227414330218, "grad_norm": 3.25, "learning_rate": 3.133115038384885e-05, "loss": 0.2442, "step": 13664 }, { "epoch": 84.8785046728972, "grad_norm": 3.921875, "learning_rate": 3.132876690470008e-05, "loss": 0.116, "step": 13665 }, { "epoch": 84.88473520249221, "grad_norm": 1.609375, "learning_rate": 3.132638336408507e-05, "loss": 0.1125, "step": 13666 }, { "epoch": 84.89096573208722, "grad_norm": 2.859375, "learning_rate": 3.132399976202696e-05, "loss": 0.2328, "step": 13667 }, { "epoch": 84.89719626168224, "grad_norm": 1.28125, "learning_rate": 3.13216160985489e-05, "loss": 0.0943, "step": 13668 }, { "epoch": 84.90342679127725, "grad_norm": 1.71875, "learning_rate": 3.131923237367404e-05, "loss": 0.1247, "step": 13669 }, { "epoch": 84.90965732087227, "grad_norm": 1.859375, "learning_rate": 3.131684858742554e-05, "loss": 0.1375, "step": 13670 }, { "epoch": 84.91588785046729, "grad_norm": 1.8203125, "learning_rate": 3.131446473982653e-05, "loss": 0.1134, "step": 13671 }, { "epoch": 84.9221183800623, "grad_norm": 3.0, "learning_rate": 3.131208083090018e-05, "loss": 0.1795, "step": 13672 }, { "epoch": 84.92834890965732, "grad_norm": 2.359375, "learning_rate": 3.130969686066965e-05, "loss": 0.1401, "step": 13673 }, { "epoch": 84.93457943925233, "grad_norm": 2.078125, "learning_rate": 3.130731282915807e-05, "loss": 0.1157, "step": 13674 }, { "epoch": 84.94080996884735, "grad_norm": 1.6328125, "learning_rate": 3.1304928736388614e-05, "loss": 0.1116, "step": 13675 }, { "epoch": 84.94704049844236, "grad_norm": 2.65625, "learning_rate": 3.1302544582384416e-05, "loss": 0.1173, "step": 13676 }, { "epoch": 84.95327102803738, "grad_norm": 1.46875, "learning_rate": 3.130016036716867e-05, "loss": 0.1087, "step": 13677 }, { "epoch": 84.9595015576324, "grad_norm": 1.515625, "learning_rate": 3.1297776090764484e-05, "loss": 0.1352, "step": 13678 }, { "epoch": 84.96573208722741, "grad_norm": 2.859375, "learning_rate": 3.1295391753195047e-05, "loss": 0.2335, "step": 13679 }, { "epoch": 84.97196261682242, "grad_norm": 1.90625, "learning_rate": 3.129300735448351e-05, "loss": 0.1067, "step": 13680 }, { "epoch": 84.97819314641744, "grad_norm": 2.015625, "learning_rate": 3.1290622894653026e-05, "loss": 0.1543, "step": 13681 }, { "epoch": 84.98442367601245, "grad_norm": 2.171875, "learning_rate": 3.128823837372675e-05, "loss": 0.1791, "step": 13682 }, { "epoch": 84.99065420560747, "grad_norm": 1.4140625, "learning_rate": 3.128585379172785e-05, "loss": 0.1253, "step": 13683 }, { "epoch": 84.99688473520249, "grad_norm": 2.296875, "learning_rate": 3.1283469148679485e-05, "loss": 0.1483, "step": 13684 }, { "epoch": 85.0, "grad_norm": 1.5859375, "learning_rate": 3.12810844446048e-05, "loss": 0.0976, "step": 13685 }, { "epoch": 85.00623052959502, "grad_norm": 3.5, "learning_rate": 3.127869967952698e-05, "loss": 0.1338, "step": 13686 }, { "epoch": 85.01246105919003, "grad_norm": 2.359375, "learning_rate": 3.127631485346917e-05, "loss": 0.1076, "step": 13687 }, { "epoch": 85.01869158878505, "grad_norm": 1.1640625, "learning_rate": 3.1273929966454535e-05, "loss": 0.096, "step": 13688 }, { "epoch": 85.02492211838006, "grad_norm": 1.5078125, "learning_rate": 3.127154501850624e-05, "loss": 0.1215, "step": 13689 }, { "epoch": 85.03115264797508, "grad_norm": 2.984375, "learning_rate": 3.126916000964744e-05, "loss": 0.1073, "step": 13690 }, { "epoch": 85.03738317757009, "grad_norm": 2.859375, "learning_rate": 3.126677493990131e-05, "loss": 0.1197, "step": 13691 }, { "epoch": 85.04361370716511, "grad_norm": 2.609375, "learning_rate": 3.126438980929101e-05, "loss": 0.1739, "step": 13692 }, { "epoch": 85.04984423676012, "grad_norm": 1.4453125, "learning_rate": 3.1262004617839693e-05, "loss": 0.1156, "step": 13693 }, { "epoch": 85.05607476635514, "grad_norm": 1.7578125, "learning_rate": 3.125961936557055e-05, "loss": 0.1395, "step": 13694 }, { "epoch": 85.06230529595015, "grad_norm": 3.140625, "learning_rate": 3.125723405250672e-05, "loss": 0.2452, "step": 13695 }, { "epoch": 85.06853582554517, "grad_norm": 3.53125, "learning_rate": 3.1254848678671396e-05, "loss": 0.2397, "step": 13696 }, { "epoch": 85.07476635514018, "grad_norm": 2.65625, "learning_rate": 3.125246324408772e-05, "loss": 0.1043, "step": 13697 }, { "epoch": 85.0809968847352, "grad_norm": 3.421875, "learning_rate": 3.125007774877887e-05, "loss": 0.2802, "step": 13698 }, { "epoch": 85.08722741433021, "grad_norm": 2.96875, "learning_rate": 3.124769219276803e-05, "loss": 0.1726, "step": 13699 }, { "epoch": 85.09345794392523, "grad_norm": 2.0625, "learning_rate": 3.1245306576078336e-05, "loss": 0.1641, "step": 13700 }, { "epoch": 85.09968847352025, "grad_norm": 1.9609375, "learning_rate": 3.1242920898732986e-05, "loss": 0.1248, "step": 13701 }, { "epoch": 85.10591900311526, "grad_norm": 2.421875, "learning_rate": 3.124053516075514e-05, "loss": 0.1329, "step": 13702 }, { "epoch": 85.11214953271028, "grad_norm": 1.640625, "learning_rate": 3.123814936216797e-05, "loss": 0.1169, "step": 13703 }, { "epoch": 85.11838006230529, "grad_norm": 1.9921875, "learning_rate": 3.123576350299464e-05, "loss": 0.1026, "step": 13704 }, { "epoch": 85.1246105919003, "grad_norm": 2.5, "learning_rate": 3.123337758325834e-05, "loss": 0.1517, "step": 13705 }, { "epoch": 85.13084112149532, "grad_norm": 1.484375, "learning_rate": 3.123099160298222e-05, "loss": 0.1063, "step": 13706 }, { "epoch": 85.13707165109034, "grad_norm": 1.109375, "learning_rate": 3.122860556218947e-05, "loss": 0.1058, "step": 13707 }, { "epoch": 85.14330218068535, "grad_norm": 1.3203125, "learning_rate": 3.1226219460903256e-05, "loss": 0.0939, "step": 13708 }, { "epoch": 85.14953271028037, "grad_norm": 1.1953125, "learning_rate": 3.1223833299146756e-05, "loss": 0.1274, "step": 13709 }, { "epoch": 85.15576323987538, "grad_norm": 2.609375, "learning_rate": 3.122144707694314e-05, "loss": 0.1648, "step": 13710 }, { "epoch": 85.1619937694704, "grad_norm": 3.03125, "learning_rate": 3.121906079431558e-05, "loss": 0.2304, "step": 13711 }, { "epoch": 85.16822429906541, "grad_norm": 2.4375, "learning_rate": 3.121667445128727e-05, "loss": 0.1306, "step": 13712 }, { "epoch": 85.17445482866043, "grad_norm": 3.09375, "learning_rate": 3.121428804788137e-05, "loss": 0.1023, "step": 13713 }, { "epoch": 85.18068535825545, "grad_norm": 3.53125, "learning_rate": 3.1211901584121064e-05, "loss": 0.1128, "step": 13714 }, { "epoch": 85.18691588785046, "grad_norm": 2.734375, "learning_rate": 3.120951506002952e-05, "loss": 0.339, "step": 13715 }, { "epoch": 85.19314641744548, "grad_norm": 3.484375, "learning_rate": 3.120712847562993e-05, "loss": 0.1829, "step": 13716 }, { "epoch": 85.19937694704049, "grad_norm": 5.1875, "learning_rate": 3.120474183094547e-05, "loss": 0.1365, "step": 13717 }, { "epoch": 85.20560747663552, "grad_norm": 3.140625, "learning_rate": 3.1202355125999324e-05, "loss": 0.2344, "step": 13718 }, { "epoch": 85.21183800623054, "grad_norm": 4.28125, "learning_rate": 3.119996836081465e-05, "loss": 0.2404, "step": 13719 }, { "epoch": 85.21806853582555, "grad_norm": 2.078125, "learning_rate": 3.119758153541465e-05, "loss": 0.098, "step": 13720 }, { "epoch": 85.22429906542057, "grad_norm": 3.359375, "learning_rate": 3.11951946498225e-05, "loss": 0.1756, "step": 13721 }, { "epoch": 85.23052959501558, "grad_norm": 3.28125, "learning_rate": 3.1192807704061385e-05, "loss": 0.2025, "step": 13722 }, { "epoch": 85.2367601246106, "grad_norm": 2.625, "learning_rate": 3.119042069815448e-05, "loss": 0.1486, "step": 13723 }, { "epoch": 85.24299065420561, "grad_norm": 3.828125, "learning_rate": 3.118803363212497e-05, "loss": 0.1318, "step": 13724 }, { "epoch": 85.24922118380063, "grad_norm": 3.3125, "learning_rate": 3.1185646505996043e-05, "loss": 0.2235, "step": 13725 }, { "epoch": 85.25545171339564, "grad_norm": 2.328125, "learning_rate": 3.118325931979088e-05, "loss": 0.1858, "step": 13726 }, { "epoch": 85.26168224299066, "grad_norm": 1.34375, "learning_rate": 3.118087207353266e-05, "loss": 0.0919, "step": 13727 }, { "epoch": 85.26791277258567, "grad_norm": 1.234375, "learning_rate": 3.117848476724458e-05, "loss": 0.0859, "step": 13728 }, { "epoch": 85.27414330218069, "grad_norm": 3.46875, "learning_rate": 3.117609740094982e-05, "loss": 0.1494, "step": 13729 }, { "epoch": 85.2803738317757, "grad_norm": 1.4375, "learning_rate": 3.117370997467157e-05, "loss": 0.1269, "step": 13730 }, { "epoch": 85.28660436137072, "grad_norm": 3.203125, "learning_rate": 3.117132248843302e-05, "loss": 0.3183, "step": 13731 }, { "epoch": 85.29283489096574, "grad_norm": 2.203125, "learning_rate": 3.116893494225734e-05, "loss": 0.1545, "step": 13732 }, { "epoch": 85.29906542056075, "grad_norm": 2.078125, "learning_rate": 3.116654733616774e-05, "loss": 0.1461, "step": 13733 }, { "epoch": 85.30529595015577, "grad_norm": 2.4375, "learning_rate": 3.11641596701874e-05, "loss": 0.1892, "step": 13734 }, { "epoch": 85.31152647975078, "grad_norm": 2.75, "learning_rate": 3.11617719443395e-05, "loss": 0.2092, "step": 13735 }, { "epoch": 85.3177570093458, "grad_norm": 2.328125, "learning_rate": 3.1159384158647246e-05, "loss": 0.1909, "step": 13736 }, { "epoch": 85.32398753894081, "grad_norm": 3.28125, "learning_rate": 3.1156996313133826e-05, "loss": 0.0896, "step": 13737 }, { "epoch": 85.33021806853583, "grad_norm": 1.5703125, "learning_rate": 3.1154608407822413e-05, "loss": 0.1486, "step": 13738 }, { "epoch": 85.33644859813084, "grad_norm": 0.8671875, "learning_rate": 3.1152220442736225e-05, "loss": 0.0992, "step": 13739 }, { "epoch": 85.34267912772586, "grad_norm": 2.1875, "learning_rate": 3.114983241789844e-05, "loss": 0.1539, "step": 13740 }, { "epoch": 85.34890965732087, "grad_norm": 1.78125, "learning_rate": 3.1147444333332245e-05, "loss": 0.1599, "step": 13741 }, { "epoch": 85.35514018691589, "grad_norm": 1.1796875, "learning_rate": 3.114505618906086e-05, "loss": 0.0966, "step": 13742 }, { "epoch": 85.3613707165109, "grad_norm": 1.9375, "learning_rate": 3.114266798510744e-05, "loss": 0.1532, "step": 13743 }, { "epoch": 85.36760124610592, "grad_norm": 3.015625, "learning_rate": 3.114027972149521e-05, "loss": 0.2508, "step": 13744 }, { "epoch": 85.37383177570094, "grad_norm": 2.125, "learning_rate": 3.113789139824737e-05, "loss": 0.1899, "step": 13745 }, { "epoch": 85.38006230529595, "grad_norm": 3.453125, "learning_rate": 3.113550301538708e-05, "loss": 0.2849, "step": 13746 }, { "epoch": 85.38629283489097, "grad_norm": 2.59375, "learning_rate": 3.113311457293757e-05, "loss": 0.2061, "step": 13747 }, { "epoch": 85.39252336448598, "grad_norm": 1.2890625, "learning_rate": 3.1130726070922016e-05, "loss": 0.1067, "step": 13748 }, { "epoch": 85.398753894081, "grad_norm": 2.890625, "learning_rate": 3.112833750936363e-05, "loss": 0.1419, "step": 13749 }, { "epoch": 85.40498442367601, "grad_norm": 2.75, "learning_rate": 3.1125948888285604e-05, "loss": 0.1949, "step": 13750 }, { "epoch": 85.41121495327103, "grad_norm": 2.046875, "learning_rate": 3.1123560207711146e-05, "loss": 0.1639, "step": 13751 }, { "epoch": 85.41744548286604, "grad_norm": 2.453125, "learning_rate": 3.1121171467663435e-05, "loss": 0.1643, "step": 13752 }, { "epoch": 85.42367601246106, "grad_norm": 1.234375, "learning_rate": 3.111878266816569e-05, "loss": 0.1264, "step": 13753 }, { "epoch": 85.42990654205607, "grad_norm": 1.4140625, "learning_rate": 3.11163938092411e-05, "loss": 0.1358, "step": 13754 }, { "epoch": 85.43613707165109, "grad_norm": 2.78125, "learning_rate": 3.111400489091288e-05, "loss": 0.2656, "step": 13755 }, { "epoch": 85.4423676012461, "grad_norm": 2.421875, "learning_rate": 3.1111615913204215e-05, "loss": 0.1937, "step": 13756 }, { "epoch": 85.44859813084112, "grad_norm": 3.15625, "learning_rate": 3.1109226876138315e-05, "loss": 0.1777, "step": 13757 }, { "epoch": 85.45482866043614, "grad_norm": 3.390625, "learning_rate": 3.110683777973838e-05, "loss": 0.2585, "step": 13758 }, { "epoch": 85.46105919003115, "grad_norm": 3.296875, "learning_rate": 3.110444862402763e-05, "loss": 0.2638, "step": 13759 }, { "epoch": 85.46728971962617, "grad_norm": 2.328125, "learning_rate": 3.110205940902924e-05, "loss": 0.182, "step": 13760 }, { "epoch": 85.47352024922118, "grad_norm": 1.3515625, "learning_rate": 3.1099670134766426e-05, "loss": 0.0984, "step": 13761 }, { "epoch": 85.4797507788162, "grad_norm": 1.8125, "learning_rate": 3.1097280801262414e-05, "loss": 0.1712, "step": 13762 }, { "epoch": 85.48598130841121, "grad_norm": 2.375, "learning_rate": 3.109489140854037e-05, "loss": 0.2639, "step": 13763 }, { "epoch": 85.49221183800623, "grad_norm": 1.796875, "learning_rate": 3.109250195662354e-05, "loss": 0.1393, "step": 13764 }, { "epoch": 85.49844236760124, "grad_norm": 5.0625, "learning_rate": 3.1090112445535104e-05, "loss": 0.2282, "step": 13765 }, { "epoch": 85.50467289719626, "grad_norm": 4.40625, "learning_rate": 3.1087722875298285e-05, "loss": 0.2672, "step": 13766 }, { "epoch": 85.51090342679127, "grad_norm": 2.125, "learning_rate": 3.1085333245936276e-05, "loss": 0.1439, "step": 13767 }, { "epoch": 85.51713395638629, "grad_norm": 3.328125, "learning_rate": 3.10829435574723e-05, "loss": 0.219, "step": 13768 }, { "epoch": 85.5233644859813, "grad_norm": 3.25, "learning_rate": 3.108055380992956e-05, "loss": 0.1271, "step": 13769 }, { "epoch": 85.52959501557632, "grad_norm": 4.03125, "learning_rate": 3.1078164003331264e-05, "loss": 0.1493, "step": 13770 }, { "epoch": 85.53582554517133, "grad_norm": 3.078125, "learning_rate": 3.107577413770062e-05, "loss": 0.1889, "step": 13771 }, { "epoch": 85.54205607476635, "grad_norm": 3.140625, "learning_rate": 3.1073384213060854e-05, "loss": 0.1377, "step": 13772 }, { "epoch": 85.54828660436137, "grad_norm": 2.546875, "learning_rate": 3.107099422943516e-05, "loss": 0.1492, "step": 13773 }, { "epoch": 85.55451713395638, "grad_norm": 2.453125, "learning_rate": 3.1068604186846754e-05, "loss": 0.1868, "step": 13774 }, { "epoch": 85.5607476635514, "grad_norm": 2.65625, "learning_rate": 3.106621408531886e-05, "loss": 0.2315, "step": 13775 }, { "epoch": 85.56697819314641, "grad_norm": 3.171875, "learning_rate": 3.1063823924874674e-05, "loss": 0.121, "step": 13776 }, { "epoch": 85.57320872274143, "grad_norm": 3.65625, "learning_rate": 3.106143370553742e-05, "loss": 0.2913, "step": 13777 }, { "epoch": 85.57943925233644, "grad_norm": 3.953125, "learning_rate": 3.105904342733032e-05, "loss": 0.4088, "step": 13778 }, { "epoch": 85.58566978193146, "grad_norm": 2.359375, "learning_rate": 3.105665309027656e-05, "loss": 0.1201, "step": 13779 }, { "epoch": 85.59190031152647, "grad_norm": 2.59375, "learning_rate": 3.105426269439938e-05, "loss": 0.1396, "step": 13780 }, { "epoch": 85.59813084112149, "grad_norm": 1.515625, "learning_rate": 3.1051872239721996e-05, "loss": 0.1159, "step": 13781 }, { "epoch": 85.60436137071652, "grad_norm": 1.3125, "learning_rate": 3.104948172626762e-05, "loss": 0.1232, "step": 13782 }, { "epoch": 85.61059190031153, "grad_norm": 3.359375, "learning_rate": 3.104709115405947e-05, "loss": 0.2111, "step": 13783 }, { "epoch": 85.61682242990655, "grad_norm": 4.46875, "learning_rate": 3.1044700523120756e-05, "loss": 0.1108, "step": 13784 }, { "epoch": 85.62305295950156, "grad_norm": 3.078125, "learning_rate": 3.1042309833474705e-05, "loss": 0.1391, "step": 13785 }, { "epoch": 85.62928348909658, "grad_norm": 3.0625, "learning_rate": 3.103991908514453e-05, "loss": 0.2468, "step": 13786 }, { "epoch": 85.6355140186916, "grad_norm": 1.8125, "learning_rate": 3.103752827815345e-05, "loss": 0.1414, "step": 13787 }, { "epoch": 85.64174454828661, "grad_norm": 2.625, "learning_rate": 3.10351374125247e-05, "loss": 0.2067, "step": 13788 }, { "epoch": 85.64797507788163, "grad_norm": 3.296875, "learning_rate": 3.1032746488281485e-05, "loss": 0.1225, "step": 13789 }, { "epoch": 85.65420560747664, "grad_norm": 2.609375, "learning_rate": 3.103035550544702e-05, "loss": 0.159, "step": 13790 }, { "epoch": 85.66043613707166, "grad_norm": 2.078125, "learning_rate": 3.102796446404455e-05, "loss": 0.1465, "step": 13791 }, { "epoch": 85.66666666666667, "grad_norm": 3.609375, "learning_rate": 3.1025573364097285e-05, "loss": 0.3258, "step": 13792 }, { "epoch": 85.67289719626169, "grad_norm": 2.734375, "learning_rate": 3.102318220562844e-05, "loss": 0.2775, "step": 13793 }, { "epoch": 85.6791277258567, "grad_norm": 2.25, "learning_rate": 3.1020790988661246e-05, "loss": 0.2802, "step": 13794 }, { "epoch": 85.68535825545172, "grad_norm": 2.125, "learning_rate": 3.1018399713218924e-05, "loss": 0.1582, "step": 13795 }, { "epoch": 85.69158878504673, "grad_norm": 2.84375, "learning_rate": 3.10160083793247e-05, "loss": 0.2208, "step": 13796 }, { "epoch": 85.69781931464175, "grad_norm": 1.7109375, "learning_rate": 3.1013616987001816e-05, "loss": 0.1213, "step": 13797 }, { "epoch": 85.70404984423676, "grad_norm": 3.0625, "learning_rate": 3.1011225536273466e-05, "loss": 0.273, "step": 13798 }, { "epoch": 85.71028037383178, "grad_norm": 1.8359375, "learning_rate": 3.100883402716289e-05, "loss": 0.1384, "step": 13799 }, { "epoch": 85.7165109034268, "grad_norm": 1.4921875, "learning_rate": 3.100644245969332e-05, "loss": 0.1151, "step": 13800 }, { "epoch": 85.72274143302181, "grad_norm": 2.15625, "learning_rate": 3.1004050833887985e-05, "loss": 0.157, "step": 13801 }, { "epoch": 85.72897196261682, "grad_norm": 3.34375, "learning_rate": 3.100165914977011e-05, "loss": 0.3649, "step": 13802 }, { "epoch": 85.73520249221184, "grad_norm": 1.640625, "learning_rate": 3.0999267407362915e-05, "loss": 0.1291, "step": 13803 }, { "epoch": 85.74143302180686, "grad_norm": 3.1875, "learning_rate": 3.099687560668963e-05, "loss": 0.2904, "step": 13804 }, { "epoch": 85.74766355140187, "grad_norm": 3.125, "learning_rate": 3.099448374777351e-05, "loss": 0.2366, "step": 13805 }, { "epoch": 85.75389408099689, "grad_norm": 3.65625, "learning_rate": 3.099209183063776e-05, "loss": 0.3131, "step": 13806 }, { "epoch": 85.7601246105919, "grad_norm": 1.2421875, "learning_rate": 3.09896998553056e-05, "loss": 0.1243, "step": 13807 }, { "epoch": 85.76635514018692, "grad_norm": 2.59375, "learning_rate": 3.0987307821800296e-05, "loss": 0.1883, "step": 13808 }, { "epoch": 85.77258566978193, "grad_norm": 3.390625, "learning_rate": 3.0984915730145046e-05, "loss": 0.2472, "step": 13809 }, { "epoch": 85.77881619937695, "grad_norm": 1.9921875, "learning_rate": 3.098252358036311e-05, "loss": 0.1174, "step": 13810 }, { "epoch": 85.78504672897196, "grad_norm": 2.53125, "learning_rate": 3.09801313724777e-05, "loss": 0.1481, "step": 13811 }, { "epoch": 85.79127725856698, "grad_norm": 1.5, "learning_rate": 3.097773910651207e-05, "loss": 0.1159, "step": 13812 }, { "epoch": 85.797507788162, "grad_norm": 2.078125, "learning_rate": 3.097534678248944e-05, "loss": 0.2004, "step": 13813 }, { "epoch": 85.80373831775701, "grad_norm": 3.109375, "learning_rate": 3.097295440043304e-05, "loss": 0.1479, "step": 13814 }, { "epoch": 85.80996884735202, "grad_norm": 1.7578125, "learning_rate": 3.097056196036611e-05, "loss": 0.1767, "step": 13815 }, { "epoch": 85.81619937694704, "grad_norm": 2.34375, "learning_rate": 3.09681694623119e-05, "loss": 0.2919, "step": 13816 }, { "epoch": 85.82242990654206, "grad_norm": 2.90625, "learning_rate": 3.096577690629363e-05, "loss": 0.2199, "step": 13817 }, { "epoch": 85.82866043613707, "grad_norm": 2.421875, "learning_rate": 3.096338429233454e-05, "loss": 0.1701, "step": 13818 }, { "epoch": 85.83489096573209, "grad_norm": 1.7734375, "learning_rate": 3.0960991620457874e-05, "loss": 0.1305, "step": 13819 }, { "epoch": 85.8411214953271, "grad_norm": 1.828125, "learning_rate": 3.095859889068687e-05, "loss": 0.1119, "step": 13820 }, { "epoch": 85.84735202492212, "grad_norm": 2.125, "learning_rate": 3.095620610304475e-05, "loss": 0.1155, "step": 13821 }, { "epoch": 85.85358255451713, "grad_norm": 1.5703125, "learning_rate": 3.095381325755478e-05, "loss": 0.1711, "step": 13822 }, { "epoch": 85.85981308411215, "grad_norm": 1.4296875, "learning_rate": 3.095142035424018e-05, "loss": 0.1136, "step": 13823 }, { "epoch": 85.86604361370716, "grad_norm": 2.9375, "learning_rate": 3.0949027393124185e-05, "loss": 0.1528, "step": 13824 }, { "epoch": 85.87227414330218, "grad_norm": 2.40625, "learning_rate": 3.094663437423006e-05, "loss": 0.2258, "step": 13825 }, { "epoch": 85.8785046728972, "grad_norm": 1.484375, "learning_rate": 3.094424129758103e-05, "loss": 0.1344, "step": 13826 }, { "epoch": 85.88473520249221, "grad_norm": 1.140625, "learning_rate": 3.094184816320034e-05, "loss": 0.1245, "step": 13827 }, { "epoch": 85.89096573208722, "grad_norm": 3.203125, "learning_rate": 3.0939454971111224e-05, "loss": 0.2033, "step": 13828 }, { "epoch": 85.89719626168224, "grad_norm": 2.15625, "learning_rate": 3.093706172133695e-05, "loss": 0.1843, "step": 13829 }, { "epoch": 85.90342679127725, "grad_norm": 2.609375, "learning_rate": 3.093466841390074e-05, "loss": 0.1787, "step": 13830 }, { "epoch": 85.90965732087227, "grad_norm": 2.125, "learning_rate": 3.093227504882584e-05, "loss": 0.176, "step": 13831 }, { "epoch": 85.91588785046729, "grad_norm": 2.6875, "learning_rate": 3.09298816261355e-05, "loss": 0.1214, "step": 13832 }, { "epoch": 85.9221183800623, "grad_norm": 3.109375, "learning_rate": 3.092748814585297e-05, "loss": 0.1965, "step": 13833 }, { "epoch": 85.92834890965732, "grad_norm": 1.3828125, "learning_rate": 3.092509460800148e-05, "loss": 0.1005, "step": 13834 }, { "epoch": 85.93457943925233, "grad_norm": 3.03125, "learning_rate": 3.092270101260429e-05, "loss": 0.2112, "step": 13835 }, { "epoch": 85.94080996884735, "grad_norm": 2.15625, "learning_rate": 3.0920307359684655e-05, "loss": 0.1289, "step": 13836 }, { "epoch": 85.94704049844236, "grad_norm": 2.25, "learning_rate": 3.09179136492658e-05, "loss": 0.1196, "step": 13837 }, { "epoch": 85.95327102803738, "grad_norm": 1.671875, "learning_rate": 3.091551988137098e-05, "loss": 0.1328, "step": 13838 }, { "epoch": 85.9595015576324, "grad_norm": 2.59375, "learning_rate": 3.091312605602346e-05, "loss": 0.1636, "step": 13839 }, { "epoch": 85.96573208722741, "grad_norm": 2.453125, "learning_rate": 3.091073217324648e-05, "loss": 0.1978, "step": 13840 }, { "epoch": 85.97196261682242, "grad_norm": 2.9375, "learning_rate": 3.090833823306328e-05, "loss": 0.2488, "step": 13841 }, { "epoch": 85.97819314641744, "grad_norm": 1.9140625, "learning_rate": 3.090594423549711e-05, "loss": 0.1617, "step": 13842 }, { "epoch": 85.98442367601245, "grad_norm": 2.53125, "learning_rate": 3.0903550180571246e-05, "loss": 0.2306, "step": 13843 }, { "epoch": 85.99065420560747, "grad_norm": 2.421875, "learning_rate": 3.0901156068308905e-05, "loss": 0.2598, "step": 13844 }, { "epoch": 85.99688473520249, "grad_norm": 2.484375, "learning_rate": 3.0898761898733375e-05, "loss": 0.1794, "step": 13845 }, { "epoch": 86.0, "grad_norm": 0.72265625, "learning_rate": 3.0896367671867876e-05, "loss": 0.0515, "step": 13846 }, { "epoch": 86.00623052959502, "grad_norm": 3.15625, "learning_rate": 3.0893973387735687e-05, "loss": 0.2479, "step": 13847 }, { "epoch": 86.01246105919003, "grad_norm": 2.921875, "learning_rate": 3.089157904636004e-05, "loss": 0.2867, "step": 13848 }, { "epoch": 86.01869158878505, "grad_norm": 2.375, "learning_rate": 3.088918464776421e-05, "loss": 0.2089, "step": 13849 }, { "epoch": 86.02492211838006, "grad_norm": 2.34375, "learning_rate": 3.088679019197143e-05, "loss": 0.1266, "step": 13850 }, { "epoch": 86.03115264797508, "grad_norm": 2.84375, "learning_rate": 3.088439567900497e-05, "loss": 0.1512, "step": 13851 }, { "epoch": 86.03738317757009, "grad_norm": 3.1875, "learning_rate": 3.088200110888809e-05, "loss": 0.2383, "step": 13852 }, { "epoch": 86.04361370716511, "grad_norm": 3.109375, "learning_rate": 3.087960648164403e-05, "loss": 0.1318, "step": 13853 }, { "epoch": 86.04984423676012, "grad_norm": 2.78125, "learning_rate": 3.087721179729606e-05, "loss": 0.0941, "step": 13854 }, { "epoch": 86.05607476635514, "grad_norm": 2.296875, "learning_rate": 3.087481705586744e-05, "loss": 0.1229, "step": 13855 }, { "epoch": 86.06230529595015, "grad_norm": 1.6875, "learning_rate": 3.0872422257381416e-05, "loss": 0.1419, "step": 13856 }, { "epoch": 86.06853582554517, "grad_norm": 2.796875, "learning_rate": 3.087002740186126e-05, "loss": 0.2578, "step": 13857 }, { "epoch": 86.07476635514018, "grad_norm": 2.859375, "learning_rate": 3.086763248933022e-05, "loss": 0.1424, "step": 13858 }, { "epoch": 86.0809968847352, "grad_norm": 4.4375, "learning_rate": 3.0865237519811555e-05, "loss": 0.1842, "step": 13859 }, { "epoch": 86.08722741433021, "grad_norm": 1.8125, "learning_rate": 3.0862842493328536e-05, "loss": 0.1501, "step": 13860 }, { "epoch": 86.09345794392523, "grad_norm": 1.6796875, "learning_rate": 3.086044740990442e-05, "loss": 0.1085, "step": 13861 }, { "epoch": 86.09968847352025, "grad_norm": 2.90625, "learning_rate": 3.0858052269562466e-05, "loss": 0.2218, "step": 13862 }, { "epoch": 86.10591900311526, "grad_norm": 2.9375, "learning_rate": 3.085565707232593e-05, "loss": 0.1538, "step": 13863 }, { "epoch": 86.11214953271028, "grad_norm": 2.21875, "learning_rate": 3.08532618182181e-05, "loss": 0.1843, "step": 13864 }, { "epoch": 86.11838006230529, "grad_norm": 3.0, "learning_rate": 3.085086650726221e-05, "loss": 0.2024, "step": 13865 }, { "epoch": 86.1246105919003, "grad_norm": 2.390625, "learning_rate": 3.084847113948153e-05, "loss": 0.1485, "step": 13866 }, { "epoch": 86.13084112149532, "grad_norm": 1.640625, "learning_rate": 3.084607571489934e-05, "loss": 0.158, "step": 13867 }, { "epoch": 86.13707165109034, "grad_norm": 2.640625, "learning_rate": 3.0843680233538885e-05, "loss": 0.318, "step": 13868 }, { "epoch": 86.14330218068535, "grad_norm": 1.7265625, "learning_rate": 3.084128469542344e-05, "loss": 0.128, "step": 13869 }, { "epoch": 86.14953271028037, "grad_norm": 1.9609375, "learning_rate": 3.083888910057627e-05, "loss": 0.1282, "step": 13870 }, { "epoch": 86.15576323987538, "grad_norm": 2.015625, "learning_rate": 3.083649344902065e-05, "loss": 0.1213, "step": 13871 }, { "epoch": 86.1619937694704, "grad_norm": 2.671875, "learning_rate": 3.0834097740779835e-05, "loss": 0.2337, "step": 13872 }, { "epoch": 86.16822429906541, "grad_norm": 2.453125, "learning_rate": 3.0831701975877095e-05, "loss": 0.3376, "step": 13873 }, { "epoch": 86.17445482866043, "grad_norm": 1.4921875, "learning_rate": 3.0829306154335694e-05, "loss": 0.1204, "step": 13874 }, { "epoch": 86.18068535825545, "grad_norm": 2.25, "learning_rate": 3.0826910276178916e-05, "loss": 0.2799, "step": 13875 }, { "epoch": 86.18691588785046, "grad_norm": 2.890625, "learning_rate": 3.082451434143001e-05, "loss": 0.1803, "step": 13876 }, { "epoch": 86.19314641744548, "grad_norm": 2.109375, "learning_rate": 3.082211835011227e-05, "loss": 0.2441, "step": 13877 }, { "epoch": 86.19937694704049, "grad_norm": 2.1875, "learning_rate": 3.081972230224894e-05, "loss": 0.1752, "step": 13878 }, { "epoch": 86.20560747663552, "grad_norm": 2.046875, "learning_rate": 3.081732619786332e-05, "loss": 0.1686, "step": 13879 }, { "epoch": 86.21183800623054, "grad_norm": 2.640625, "learning_rate": 3.081493003697866e-05, "loss": 0.2107, "step": 13880 }, { "epoch": 86.21806853582555, "grad_norm": 2.421875, "learning_rate": 3.0812533819618225e-05, "loss": 0.1721, "step": 13881 }, { "epoch": 86.22429906542057, "grad_norm": 3.5625, "learning_rate": 3.08101375458053e-05, "loss": 0.2317, "step": 13882 }, { "epoch": 86.23052959501558, "grad_norm": 2.890625, "learning_rate": 3.0807741215563175e-05, "loss": 0.2212, "step": 13883 }, { "epoch": 86.2367601246106, "grad_norm": 0.80859375, "learning_rate": 3.08053448289151e-05, "loss": 0.085, "step": 13884 }, { "epoch": 86.24299065420561, "grad_norm": 2.03125, "learning_rate": 3.080294838588434e-05, "loss": 0.1291, "step": 13885 }, { "epoch": 86.24922118380063, "grad_norm": 2.03125, "learning_rate": 3.08005518864942e-05, "loss": 0.1837, "step": 13886 }, { "epoch": 86.25545171339564, "grad_norm": 3.0, "learning_rate": 3.0798155330767934e-05, "loss": 0.1908, "step": 13887 }, { "epoch": 86.26168224299066, "grad_norm": 1.4140625, "learning_rate": 3.079575871872883e-05, "loss": 0.1163, "step": 13888 }, { "epoch": 86.26791277258567, "grad_norm": 2.78125, "learning_rate": 3.079336205040016e-05, "loss": 0.1848, "step": 13889 }, { "epoch": 86.27414330218069, "grad_norm": 2.5625, "learning_rate": 3.07909653258052e-05, "loss": 0.2379, "step": 13890 }, { "epoch": 86.2803738317757, "grad_norm": 2.4375, "learning_rate": 3.078856854496722e-05, "loss": 0.2146, "step": 13891 }, { "epoch": 86.28660436137072, "grad_norm": 2.03125, "learning_rate": 3.078617170790951e-05, "loss": 0.1402, "step": 13892 }, { "epoch": 86.29283489096574, "grad_norm": 3.046875, "learning_rate": 3.078377481465534e-05, "loss": 0.2285, "step": 13893 }, { "epoch": 86.29906542056075, "grad_norm": 1.7265625, "learning_rate": 3.0781377865227994e-05, "loss": 0.1306, "step": 13894 }, { "epoch": 86.30529595015577, "grad_norm": 1.7578125, "learning_rate": 3.077898085965075e-05, "loss": 0.1205, "step": 13895 }, { "epoch": 86.31152647975078, "grad_norm": 2.375, "learning_rate": 3.077658379794689e-05, "loss": 0.1197, "step": 13896 }, { "epoch": 86.3177570093458, "grad_norm": 1.7578125, "learning_rate": 3.07741866801397e-05, "loss": 0.154, "step": 13897 }, { "epoch": 86.32398753894081, "grad_norm": 1.84375, "learning_rate": 3.077178950625244e-05, "loss": 0.1497, "step": 13898 }, { "epoch": 86.33021806853583, "grad_norm": 1.5703125, "learning_rate": 3.0769392276308424e-05, "loss": 0.1357, "step": 13899 }, { "epoch": 86.33644859813084, "grad_norm": 2.796875, "learning_rate": 3.07669949903309e-05, "loss": 0.21, "step": 13900 }, { "epoch": 86.34267912772586, "grad_norm": 1.6875, "learning_rate": 3.076459764834318e-05, "loss": 0.1237, "step": 13901 }, { "epoch": 86.34890965732087, "grad_norm": 2.03125, "learning_rate": 3.0762200250368535e-05, "loss": 0.2113, "step": 13902 }, { "epoch": 86.35514018691589, "grad_norm": 2.03125, "learning_rate": 3.0759802796430246e-05, "loss": 0.1077, "step": 13903 }, { "epoch": 86.3613707165109, "grad_norm": 2.828125, "learning_rate": 3.07574052865516e-05, "loss": 0.2177, "step": 13904 }, { "epoch": 86.36760124610592, "grad_norm": 1.1953125, "learning_rate": 3.075500772075588e-05, "loss": 0.0883, "step": 13905 }, { "epoch": 86.37383177570094, "grad_norm": 0.90234375, "learning_rate": 3.075261009906638e-05, "loss": 0.0921, "step": 13906 }, { "epoch": 86.38006230529595, "grad_norm": 1.2734375, "learning_rate": 3.075021242150637e-05, "loss": 0.118, "step": 13907 }, { "epoch": 86.38629283489097, "grad_norm": 2.34375, "learning_rate": 3.0747814688099164e-05, "loss": 0.191, "step": 13908 }, { "epoch": 86.39252336448598, "grad_norm": 2.484375, "learning_rate": 3.0745416898868025e-05, "loss": 0.1696, "step": 13909 }, { "epoch": 86.398753894081, "grad_norm": 2.5625, "learning_rate": 3.074301905383625e-05, "loss": 0.1616, "step": 13910 }, { "epoch": 86.40498442367601, "grad_norm": 2.90625, "learning_rate": 3.074062115302711e-05, "loss": 0.1848, "step": 13911 }, { "epoch": 86.41121495327103, "grad_norm": 2.828125, "learning_rate": 3.073822319646394e-05, "loss": 0.1804, "step": 13912 }, { "epoch": 86.41744548286604, "grad_norm": 1.9765625, "learning_rate": 3.0735825184169974e-05, "loss": 0.1264, "step": 13913 }, { "epoch": 86.42367601246106, "grad_norm": 2.40625, "learning_rate": 3.073342711616854e-05, "loss": 0.1856, "step": 13914 }, { "epoch": 86.42990654205607, "grad_norm": 1.078125, "learning_rate": 3.0731028992482915e-05, "loss": 0.1195, "step": 13915 }, { "epoch": 86.43613707165109, "grad_norm": 1.296875, "learning_rate": 3.072863081313639e-05, "loss": 0.1108, "step": 13916 }, { "epoch": 86.4423676012461, "grad_norm": 2.375, "learning_rate": 3.072623257815226e-05, "loss": 0.1849, "step": 13917 }, { "epoch": 86.44859813084112, "grad_norm": 2.109375, "learning_rate": 3.072383428755381e-05, "loss": 0.1297, "step": 13918 }, { "epoch": 86.45482866043614, "grad_norm": 1.96875, "learning_rate": 3.072143594136433e-05, "loss": 0.1547, "step": 13919 }, { "epoch": 86.46105919003115, "grad_norm": 2.8125, "learning_rate": 3.071903753960713e-05, "loss": 0.182, "step": 13920 }, { "epoch": 86.46728971962617, "grad_norm": 1.75, "learning_rate": 3.071663908230549e-05, "loss": 0.1438, "step": 13921 }, { "epoch": 86.47352024922118, "grad_norm": 1.9453125, "learning_rate": 3.0714240569482715e-05, "loss": 0.106, "step": 13922 }, { "epoch": 86.4797507788162, "grad_norm": 1.65625, "learning_rate": 3.071184200116209e-05, "loss": 0.1359, "step": 13923 }, { "epoch": 86.48598130841121, "grad_norm": 2.578125, "learning_rate": 3.0709443377366905e-05, "loss": 0.2456, "step": 13924 }, { "epoch": 86.49221183800623, "grad_norm": 2.5, "learning_rate": 3.070704469812048e-05, "loss": 0.1902, "step": 13925 }, { "epoch": 86.49844236760124, "grad_norm": 2.75, "learning_rate": 3.0704645963446086e-05, "loss": 0.2707, "step": 13926 }, { "epoch": 86.50467289719626, "grad_norm": 2.09375, "learning_rate": 3.070224717336703e-05, "loss": 0.1401, "step": 13927 }, { "epoch": 86.51090342679127, "grad_norm": 2.390625, "learning_rate": 3.0699848327906625e-05, "loss": 0.1959, "step": 13928 }, { "epoch": 86.51713395638629, "grad_norm": 3.03125, "learning_rate": 3.069744942708813e-05, "loss": 0.1965, "step": 13929 }, { "epoch": 86.5233644859813, "grad_norm": 2.546875, "learning_rate": 3.069505047093488e-05, "loss": 0.2033, "step": 13930 }, { "epoch": 86.52959501557632, "grad_norm": 2.5625, "learning_rate": 3.069265145947016e-05, "loss": 0.153, "step": 13931 }, { "epoch": 86.53582554517133, "grad_norm": 2.125, "learning_rate": 3.0690252392717273e-05, "loss": 0.1806, "step": 13932 }, { "epoch": 86.54205607476635, "grad_norm": 1.5390625, "learning_rate": 3.068785327069951e-05, "loss": 0.1248, "step": 13933 }, { "epoch": 86.54828660436137, "grad_norm": 1.5546875, "learning_rate": 3.068545409344019e-05, "loss": 0.1208, "step": 13934 }, { "epoch": 86.55451713395638, "grad_norm": 2.0625, "learning_rate": 3.068305486096259e-05, "loss": 0.1737, "step": 13935 }, { "epoch": 86.5607476635514, "grad_norm": 2.796875, "learning_rate": 3.068065557329004e-05, "loss": 0.2294, "step": 13936 }, { "epoch": 86.56697819314641, "grad_norm": 3.265625, "learning_rate": 3.067825623044582e-05, "loss": 0.2673, "step": 13937 }, { "epoch": 86.57320872274143, "grad_norm": 4.03125, "learning_rate": 3.067585683245324e-05, "loss": 0.2803, "step": 13938 }, { "epoch": 86.57943925233644, "grad_norm": 1.46875, "learning_rate": 3.067345737933561e-05, "loss": 0.145, "step": 13939 }, { "epoch": 86.58566978193146, "grad_norm": 2.21875, "learning_rate": 3.0671057871116216e-05, "loss": 0.0943, "step": 13940 }, { "epoch": 86.59190031152647, "grad_norm": 3.34375, "learning_rate": 3.0668658307818385e-05, "loss": 0.2314, "step": 13941 }, { "epoch": 86.59813084112149, "grad_norm": 2.71875, "learning_rate": 3.06662586894654e-05, "loss": 0.223, "step": 13942 }, { "epoch": 86.60436137071652, "grad_norm": 2.25, "learning_rate": 3.0663859016080586e-05, "loss": 0.1375, "step": 13943 }, { "epoch": 86.61059190031153, "grad_norm": 3.25, "learning_rate": 3.0661459287687245e-05, "loss": 0.1333, "step": 13944 }, { "epoch": 86.61682242990655, "grad_norm": 2.8125, "learning_rate": 3.065905950430867e-05, "loss": 0.1398, "step": 13945 }, { "epoch": 86.62305295950156, "grad_norm": 1.265625, "learning_rate": 3.065665966596818e-05, "loss": 0.1045, "step": 13946 }, { "epoch": 86.62928348909658, "grad_norm": 3.40625, "learning_rate": 3.0654259772689086e-05, "loss": 0.2474, "step": 13947 }, { "epoch": 86.6355140186916, "grad_norm": 2.671875, "learning_rate": 3.0651859824494686e-05, "loss": 0.3056, "step": 13948 }, { "epoch": 86.64174454828661, "grad_norm": 1.921875, "learning_rate": 3.0649459821408296e-05, "loss": 0.1362, "step": 13949 }, { "epoch": 86.64797507788163, "grad_norm": 2.0, "learning_rate": 3.0647059763453226e-05, "loss": 0.1447, "step": 13950 }, { "epoch": 86.65420560747664, "grad_norm": 2.453125, "learning_rate": 3.064465965065278e-05, "loss": 0.1731, "step": 13951 }, { "epoch": 86.66043613707166, "grad_norm": 1.1484375, "learning_rate": 3.0642259483030266e-05, "loss": 0.117, "step": 13952 }, { "epoch": 86.66666666666667, "grad_norm": 1.3828125, "learning_rate": 3.063985926060901e-05, "loss": 0.1572, "step": 13953 }, { "epoch": 86.67289719626169, "grad_norm": 1.6640625, "learning_rate": 3.063745898341231e-05, "loss": 0.1553, "step": 13954 }, { "epoch": 86.6791277258567, "grad_norm": 2.921875, "learning_rate": 3.0635058651463474e-05, "loss": 0.1937, "step": 13955 }, { "epoch": 86.68535825545172, "grad_norm": 2.125, "learning_rate": 3.063265826478583e-05, "loss": 0.1759, "step": 13956 }, { "epoch": 86.69158878504673, "grad_norm": 2.1875, "learning_rate": 3.063025782340267e-05, "loss": 0.1313, "step": 13957 }, { "epoch": 86.69781931464175, "grad_norm": 2.171875, "learning_rate": 3.062785732733734e-05, "loss": 0.1642, "step": 13958 }, { "epoch": 86.70404984423676, "grad_norm": 2.296875, "learning_rate": 3.062545677661312e-05, "loss": 0.1812, "step": 13959 }, { "epoch": 86.71028037383178, "grad_norm": 2.53125, "learning_rate": 3.062305617125335e-05, "loss": 0.1948, "step": 13960 }, { "epoch": 86.7165109034268, "grad_norm": 3.296875, "learning_rate": 3.0620655511281325e-05, "loss": 0.1904, "step": 13961 }, { "epoch": 86.72274143302181, "grad_norm": 2.265625, "learning_rate": 3.0618254796720375e-05, "loss": 0.192, "step": 13962 }, { "epoch": 86.72897196261682, "grad_norm": 3.3125, "learning_rate": 3.061585402759381e-05, "loss": 0.2827, "step": 13963 }, { "epoch": 86.73520249221184, "grad_norm": 1.515625, "learning_rate": 3.061345320392495e-05, "loss": 0.1175, "step": 13964 }, { "epoch": 86.74143302180686, "grad_norm": 2.90625, "learning_rate": 3.061105232573711e-05, "loss": 0.1373, "step": 13965 }, { "epoch": 86.74766355140187, "grad_norm": 1.578125, "learning_rate": 3.0608651393053615e-05, "loss": 0.1117, "step": 13966 }, { "epoch": 86.75389408099689, "grad_norm": 2.9375, "learning_rate": 3.060625040589776e-05, "loss": 0.1854, "step": 13967 }, { "epoch": 86.7601246105919, "grad_norm": 1.28125, "learning_rate": 3.0603849364292886e-05, "loss": 0.0932, "step": 13968 }, { "epoch": 86.76635514018692, "grad_norm": 1.65625, "learning_rate": 3.060144826826232e-05, "loss": 0.1153, "step": 13969 }, { "epoch": 86.77258566978193, "grad_norm": 1.6015625, "learning_rate": 3.0599047117829356e-05, "loss": 0.1105, "step": 13970 }, { "epoch": 86.77881619937695, "grad_norm": 2.015625, "learning_rate": 3.059664591301733e-05, "loss": 0.0924, "step": 13971 }, { "epoch": 86.78504672897196, "grad_norm": 3.171875, "learning_rate": 3.059424465384955e-05, "loss": 0.18, "step": 13972 }, { "epoch": 86.79127725856698, "grad_norm": 2.515625, "learning_rate": 3.059184334034937e-05, "loss": 0.2026, "step": 13973 }, { "epoch": 86.797507788162, "grad_norm": 1.234375, "learning_rate": 3.058944197254007e-05, "loss": 0.0802, "step": 13974 }, { "epoch": 86.80373831775701, "grad_norm": 2.21875, "learning_rate": 3.0587040550445e-05, "loss": 0.106, "step": 13975 }, { "epoch": 86.80996884735202, "grad_norm": 2.34375, "learning_rate": 3.058463907408748e-05, "loss": 0.1551, "step": 13976 }, { "epoch": 86.81619937694704, "grad_norm": 1.96875, "learning_rate": 3.0582237543490825e-05, "loss": 0.1431, "step": 13977 }, { "epoch": 86.82242990654206, "grad_norm": 3.328125, "learning_rate": 3.057983595867836e-05, "loss": 0.1764, "step": 13978 }, { "epoch": 86.82866043613707, "grad_norm": 1.65625, "learning_rate": 3.0577434319673415e-05, "loss": 0.1263, "step": 13979 }, { "epoch": 86.83489096573209, "grad_norm": 2.71875, "learning_rate": 3.057503262649931e-05, "loss": 0.2422, "step": 13980 }, { "epoch": 86.8411214953271, "grad_norm": 2.96875, "learning_rate": 3.0572630879179374e-05, "loss": 0.1396, "step": 13981 }, { "epoch": 86.84735202492212, "grad_norm": 2.109375, "learning_rate": 3.057022907773694e-05, "loss": 0.2586, "step": 13982 }, { "epoch": 86.85358255451713, "grad_norm": 3.25, "learning_rate": 3.0567827222195325e-05, "loss": 0.2808, "step": 13983 }, { "epoch": 86.85981308411215, "grad_norm": 2.125, "learning_rate": 3.056542531257786e-05, "loss": 0.2472, "step": 13984 }, { "epoch": 86.86604361370716, "grad_norm": 2.46875, "learning_rate": 3.056302334890786e-05, "loss": 0.1835, "step": 13985 }, { "epoch": 86.87227414330218, "grad_norm": 2.609375, "learning_rate": 3.056062133120868e-05, "loss": 0.1316, "step": 13986 }, { "epoch": 86.8785046728972, "grad_norm": 2.734375, "learning_rate": 3.055821925950363e-05, "loss": 0.1858, "step": 13987 }, { "epoch": 86.88473520249221, "grad_norm": 1.796875, "learning_rate": 3.0555817133816037e-05, "loss": 0.1221, "step": 13988 }, { "epoch": 86.89096573208722, "grad_norm": 1.953125, "learning_rate": 3.055341495416925e-05, "loss": 0.1359, "step": 13989 }, { "epoch": 86.89719626168224, "grad_norm": 2.421875, "learning_rate": 3.055101272058657e-05, "loss": 0.1765, "step": 13990 }, { "epoch": 86.90342679127725, "grad_norm": 1.796875, "learning_rate": 3.0548610433091356e-05, "loss": 0.1304, "step": 13991 }, { "epoch": 86.90965732087227, "grad_norm": 2.375, "learning_rate": 3.0546208091706927e-05, "loss": 0.2077, "step": 13992 }, { "epoch": 86.91588785046729, "grad_norm": 2.140625, "learning_rate": 3.0543805696456616e-05, "loss": 0.135, "step": 13993 }, { "epoch": 86.9221183800623, "grad_norm": 3.171875, "learning_rate": 3.0541403247363755e-05, "loss": 0.1328, "step": 13994 }, { "epoch": 86.92834890965732, "grad_norm": 2.40625, "learning_rate": 3.053900074445168e-05, "loss": 0.1939, "step": 13995 }, { "epoch": 86.93457943925233, "grad_norm": 1.6171875, "learning_rate": 3.053659818774372e-05, "loss": 0.122, "step": 13996 }, { "epoch": 86.94080996884735, "grad_norm": 3.125, "learning_rate": 3.053419557726321e-05, "loss": 0.2001, "step": 13997 }, { "epoch": 86.94704049844236, "grad_norm": 3.671875, "learning_rate": 3.053179291303349e-05, "loss": 0.2337, "step": 13998 }, { "epoch": 86.95327102803738, "grad_norm": 1.4453125, "learning_rate": 3.052939019507789e-05, "loss": 0.1092, "step": 13999 }, { "epoch": 86.9595015576324, "grad_norm": 2.828125, "learning_rate": 3.052698742341975e-05, "loss": 0.1898, "step": 14000 }, { "epoch": 86.96573208722741, "grad_norm": 2.28125, "learning_rate": 3.05245845980824e-05, "loss": 0.177, "step": 14001 }, { "epoch": 86.97196261682242, "grad_norm": 2.859375, "learning_rate": 3.052218171908918e-05, "loss": 0.2132, "step": 14002 }, { "epoch": 86.97819314641744, "grad_norm": 2.0625, "learning_rate": 3.051977878646343e-05, "loss": 0.1754, "step": 14003 }, { "epoch": 86.98442367601245, "grad_norm": 2.84375, "learning_rate": 3.0517375800228487e-05, "loss": 0.2103, "step": 14004 }, { "epoch": 86.99065420560747, "grad_norm": 2.015625, "learning_rate": 3.051497276040768e-05, "loss": 0.151, "step": 14005 }, { "epoch": 86.99688473520249, "grad_norm": 1.3359375, "learning_rate": 3.051256966702436e-05, "loss": 0.1066, "step": 14006 }, { "epoch": 87.0, "grad_norm": 2.390625, "learning_rate": 3.051016652010185e-05, "loss": 0.1488, "step": 14007 }, { "epoch": 87.00623052959502, "grad_norm": 2.328125, "learning_rate": 3.0507763319663517e-05, "loss": 0.1874, "step": 14008 }, { "epoch": 87.01246105919003, "grad_norm": 2.671875, "learning_rate": 3.050536006573268e-05, "loss": 0.1904, "step": 14009 }, { "epoch": 87.01869158878505, "grad_norm": 1.59375, "learning_rate": 3.0502956758332686e-05, "loss": 0.12, "step": 14010 }, { "epoch": 87.02492211838006, "grad_norm": 2.984375, "learning_rate": 3.0500553397486875e-05, "loss": 0.2048, "step": 14011 }, { "epoch": 87.03115264797508, "grad_norm": 3.359375, "learning_rate": 3.0498149983218592e-05, "loss": 0.2485, "step": 14012 }, { "epoch": 87.03738317757009, "grad_norm": 3.703125, "learning_rate": 3.0495746515551177e-05, "loss": 0.2222, "step": 14013 }, { "epoch": 87.04361370716511, "grad_norm": 3.859375, "learning_rate": 3.0493342994507973e-05, "loss": 0.1311, "step": 14014 }, { "epoch": 87.04984423676012, "grad_norm": 1.6796875, "learning_rate": 3.0490939420112326e-05, "loss": 0.1263, "step": 14015 }, { "epoch": 87.05607476635514, "grad_norm": 1.875, "learning_rate": 3.048853579238757e-05, "loss": 0.1639, "step": 14016 }, { "epoch": 87.06230529595015, "grad_norm": 1.9453125, "learning_rate": 3.0486132111357064e-05, "loss": 0.112, "step": 14017 }, { "epoch": 87.06853582554517, "grad_norm": 3.84375, "learning_rate": 3.048372837704414e-05, "loss": 0.204, "step": 14018 }, { "epoch": 87.07476635514018, "grad_norm": 4.46875, "learning_rate": 3.0481324589472154e-05, "loss": 0.2229, "step": 14019 }, { "epoch": 87.0809968847352, "grad_norm": 1.859375, "learning_rate": 3.0478920748664448e-05, "loss": 0.1625, "step": 14020 }, { "epoch": 87.08722741433021, "grad_norm": 1.2421875, "learning_rate": 3.047651685464437e-05, "loss": 0.1175, "step": 14021 }, { "epoch": 87.09345794392523, "grad_norm": 3.078125, "learning_rate": 3.047411290743526e-05, "loss": 0.3447, "step": 14022 }, { "epoch": 87.09968847352025, "grad_norm": 1.4296875, "learning_rate": 3.0471708907060482e-05, "loss": 0.1258, "step": 14023 }, { "epoch": 87.10591900311526, "grad_norm": 1.8046875, "learning_rate": 3.046930485354337e-05, "loss": 0.1245, "step": 14024 }, { "epoch": 87.11214953271028, "grad_norm": 1.8125, "learning_rate": 3.046690074690726e-05, "loss": 0.1246, "step": 14025 }, { "epoch": 87.11838006230529, "grad_norm": 3.03125, "learning_rate": 3.0464496587175533e-05, "loss": 0.2316, "step": 14026 }, { "epoch": 87.1246105919003, "grad_norm": 1.390625, "learning_rate": 3.046209237437152e-05, "loss": 0.1263, "step": 14027 }, { "epoch": 87.13084112149532, "grad_norm": 2.453125, "learning_rate": 3.0459688108518574e-05, "loss": 0.1386, "step": 14028 }, { "epoch": 87.13707165109034, "grad_norm": 2.015625, "learning_rate": 3.045728378964004e-05, "loss": 0.1728, "step": 14029 }, { "epoch": 87.14330218068535, "grad_norm": 2.34375, "learning_rate": 3.0454879417759287e-05, "loss": 0.1231, "step": 14030 }, { "epoch": 87.14953271028037, "grad_norm": 2.515625, "learning_rate": 3.0452474992899643e-05, "loss": 0.1308, "step": 14031 }, { "epoch": 87.15576323987538, "grad_norm": 1.0, "learning_rate": 3.045007051508448e-05, "loss": 0.0919, "step": 14032 }, { "epoch": 87.1619937694704, "grad_norm": 2.171875, "learning_rate": 3.0447665984337138e-05, "loss": 0.1319, "step": 14033 }, { "epoch": 87.16822429906541, "grad_norm": 2.46875, "learning_rate": 3.0445261400680975e-05, "loss": 0.194, "step": 14034 }, { "epoch": 87.17445482866043, "grad_norm": 2.796875, "learning_rate": 3.0442856764139348e-05, "loss": 0.1688, "step": 14035 }, { "epoch": 87.18068535825545, "grad_norm": 1.4765625, "learning_rate": 3.0440452074735604e-05, "loss": 0.1135, "step": 14036 }, { "epoch": 87.18691588785046, "grad_norm": 2.21875, "learning_rate": 3.0438047332493108e-05, "loss": 0.1736, "step": 14037 }, { "epoch": 87.19314641744548, "grad_norm": 2.203125, "learning_rate": 3.04356425374352e-05, "loss": 0.1786, "step": 14038 }, { "epoch": 87.19937694704049, "grad_norm": 2.625, "learning_rate": 3.043323768958526e-05, "loss": 0.1472, "step": 14039 }, { "epoch": 87.20560747663552, "grad_norm": 2.703125, "learning_rate": 3.0430832788966614e-05, "loss": 0.255, "step": 14040 }, { "epoch": 87.21183800623054, "grad_norm": 2.484375, "learning_rate": 3.042842783560264e-05, "loss": 0.2173, "step": 14041 }, { "epoch": 87.21806853582555, "grad_norm": 2.546875, "learning_rate": 3.042602282951669e-05, "loss": 0.1133, "step": 14042 }, { "epoch": 87.22429906542057, "grad_norm": 2.671875, "learning_rate": 3.0423617770732122e-05, "loss": 0.309, "step": 14043 }, { "epoch": 87.23052959501558, "grad_norm": 1.96875, "learning_rate": 3.042121265927229e-05, "loss": 0.1476, "step": 14044 }, { "epoch": 87.2367601246106, "grad_norm": 2.296875, "learning_rate": 3.0418807495160563e-05, "loss": 0.1545, "step": 14045 }, { "epoch": 87.24299065420561, "grad_norm": 1.5390625, "learning_rate": 3.0416402278420293e-05, "loss": 0.1646, "step": 14046 }, { "epoch": 87.24922118380063, "grad_norm": 1.03125, "learning_rate": 3.041399700907484e-05, "loss": 0.1032, "step": 14047 }, { "epoch": 87.25545171339564, "grad_norm": 1.671875, "learning_rate": 3.0411591687147568e-05, "loss": 0.1502, "step": 14048 }, { "epoch": 87.26168224299066, "grad_norm": 1.5625, "learning_rate": 3.040918631266183e-05, "loss": 0.1278, "step": 14049 }, { "epoch": 87.26791277258567, "grad_norm": 2.796875, "learning_rate": 3.0406780885641e-05, "loss": 0.2554, "step": 14050 }, { "epoch": 87.27414330218069, "grad_norm": 2.90625, "learning_rate": 3.0404375406108427e-05, "loss": 0.1793, "step": 14051 }, { "epoch": 87.2803738317757, "grad_norm": 3.71875, "learning_rate": 3.0401969874087488e-05, "loss": 0.1632, "step": 14052 }, { "epoch": 87.28660436137072, "grad_norm": 1.8359375, "learning_rate": 3.0399564289601528e-05, "loss": 0.134, "step": 14053 }, { "epoch": 87.29283489096574, "grad_norm": 3.21875, "learning_rate": 3.039715865267393e-05, "loss": 0.3595, "step": 14054 }, { "epoch": 87.29906542056075, "grad_norm": 2.515625, "learning_rate": 3.0394752963328045e-05, "loss": 0.1495, "step": 14055 }, { "epoch": 87.30529595015577, "grad_norm": 2.328125, "learning_rate": 3.0392347221587237e-05, "loss": 0.1669, "step": 14056 }, { "epoch": 87.31152647975078, "grad_norm": 1.828125, "learning_rate": 3.0389941427474873e-05, "loss": 0.1169, "step": 14057 }, { "epoch": 87.3177570093458, "grad_norm": 2.953125, "learning_rate": 3.038753558101433e-05, "loss": 0.1661, "step": 14058 }, { "epoch": 87.32398753894081, "grad_norm": 2.390625, "learning_rate": 3.0385129682228953e-05, "loss": 0.1962, "step": 14059 }, { "epoch": 87.33021806853583, "grad_norm": 2.453125, "learning_rate": 3.0382723731142126e-05, "loss": 0.174, "step": 14060 }, { "epoch": 87.33644859813084, "grad_norm": 3.515625, "learning_rate": 3.0380317727777214e-05, "loss": 0.2001, "step": 14061 }, { "epoch": 87.34267912772586, "grad_norm": 1.6796875, "learning_rate": 3.0377911672157577e-05, "loss": 0.1657, "step": 14062 }, { "epoch": 87.34890965732087, "grad_norm": 2.25, "learning_rate": 3.0375505564306588e-05, "loss": 0.212, "step": 14063 }, { "epoch": 87.35514018691589, "grad_norm": 2.515625, "learning_rate": 3.0373099404247605e-05, "loss": 0.1213, "step": 14064 }, { "epoch": 87.3613707165109, "grad_norm": 1.5859375, "learning_rate": 3.0370693192004024e-05, "loss": 0.1237, "step": 14065 }, { "epoch": 87.36760124610592, "grad_norm": 1.875, "learning_rate": 3.0368286927599188e-05, "loss": 0.1432, "step": 14066 }, { "epoch": 87.37383177570094, "grad_norm": 2.234375, "learning_rate": 3.0365880611056475e-05, "loss": 0.2538, "step": 14067 }, { "epoch": 87.38006230529595, "grad_norm": 1.265625, "learning_rate": 3.0363474242399258e-05, "loss": 0.1272, "step": 14068 }, { "epoch": 87.38629283489097, "grad_norm": 1.3359375, "learning_rate": 3.0361067821650906e-05, "loss": 0.0985, "step": 14069 }, { "epoch": 87.39252336448598, "grad_norm": 1.0859375, "learning_rate": 3.035866134883479e-05, "loss": 0.1263, "step": 14070 }, { "epoch": 87.398753894081, "grad_norm": 2.59375, "learning_rate": 3.0356254823974295e-05, "loss": 0.1838, "step": 14071 }, { "epoch": 87.40498442367601, "grad_norm": 1.921875, "learning_rate": 3.0353848247092774e-05, "loss": 0.1505, "step": 14072 }, { "epoch": 87.41121495327103, "grad_norm": 1.9765625, "learning_rate": 3.0351441618213604e-05, "loss": 0.1527, "step": 14073 }, { "epoch": 87.41744548286604, "grad_norm": 2.484375, "learning_rate": 3.034903493736017e-05, "loss": 0.1656, "step": 14074 }, { "epoch": 87.42367601246106, "grad_norm": 0.91015625, "learning_rate": 3.034662820455584e-05, "loss": 0.0994, "step": 14075 }, { "epoch": 87.42990654205607, "grad_norm": 2.203125, "learning_rate": 3.034422141982398e-05, "loss": 0.143, "step": 14076 }, { "epoch": 87.43613707165109, "grad_norm": 2.53125, "learning_rate": 3.0341814583187978e-05, "loss": 0.1761, "step": 14077 }, { "epoch": 87.4423676012461, "grad_norm": 1.7109375, "learning_rate": 3.0339407694671214e-05, "loss": 0.1589, "step": 14078 }, { "epoch": 87.44859813084112, "grad_norm": 3.1875, "learning_rate": 3.033700075429704e-05, "loss": 0.2862, "step": 14079 }, { "epoch": 87.45482866043614, "grad_norm": 1.8671875, "learning_rate": 3.0334593762088855e-05, "loss": 0.1168, "step": 14080 }, { "epoch": 87.46105919003115, "grad_norm": 0.97265625, "learning_rate": 3.033218671807003e-05, "loss": 0.0915, "step": 14081 }, { "epoch": 87.46728971962617, "grad_norm": 2.9375, "learning_rate": 3.0329779622263944e-05, "loss": 0.1245, "step": 14082 }, { "epoch": 87.47352024922118, "grad_norm": 2.484375, "learning_rate": 3.0327372474693964e-05, "loss": 0.1075, "step": 14083 }, { "epoch": 87.4797507788162, "grad_norm": 1.59375, "learning_rate": 3.032496527538349e-05, "loss": 0.0838, "step": 14084 }, { "epoch": 87.48598130841121, "grad_norm": 2.25, "learning_rate": 3.032255802435588e-05, "loss": 0.149, "step": 14085 }, { "epoch": 87.49221183800623, "grad_norm": 2.234375, "learning_rate": 3.0320150721634515e-05, "loss": 0.1689, "step": 14086 }, { "epoch": 87.49844236760124, "grad_norm": 2.0, "learning_rate": 3.03177433672428e-05, "loss": 0.1923, "step": 14087 }, { "epoch": 87.50467289719626, "grad_norm": 3.171875, "learning_rate": 3.0315335961204088e-05, "loss": 0.2628, "step": 14088 }, { "epoch": 87.51090342679127, "grad_norm": 2.3125, "learning_rate": 3.031292850354177e-05, "loss": 0.1683, "step": 14089 }, { "epoch": 87.51713395638629, "grad_norm": 2.34375, "learning_rate": 3.0310520994279236e-05, "loss": 0.2008, "step": 14090 }, { "epoch": 87.5233644859813, "grad_norm": 3.0, "learning_rate": 3.0308113433439855e-05, "loss": 0.2086, "step": 14091 }, { "epoch": 87.52959501557632, "grad_norm": 2.171875, "learning_rate": 3.0305705821047014e-05, "loss": 0.1968, "step": 14092 }, { "epoch": 87.53582554517133, "grad_norm": 2.59375, "learning_rate": 3.0303298157124106e-05, "loss": 0.1384, "step": 14093 }, { "epoch": 87.54205607476635, "grad_norm": 2.421875, "learning_rate": 3.0300890441694494e-05, "loss": 0.1932, "step": 14094 }, { "epoch": 87.54828660436137, "grad_norm": 2.59375, "learning_rate": 3.029848267478158e-05, "loss": 0.2233, "step": 14095 }, { "epoch": 87.55451713395638, "grad_norm": 2.46875, "learning_rate": 3.0296074856408747e-05, "loss": 0.1587, "step": 14096 }, { "epoch": 87.5607476635514, "grad_norm": 1.1640625, "learning_rate": 3.0293666986599374e-05, "loss": 0.1017, "step": 14097 }, { "epoch": 87.56697819314641, "grad_norm": 2.59375, "learning_rate": 3.029125906537685e-05, "loss": 0.2076, "step": 14098 }, { "epoch": 87.57320872274143, "grad_norm": 1.1015625, "learning_rate": 3.0288851092764548e-05, "loss": 0.1289, "step": 14099 }, { "epoch": 87.57943925233644, "grad_norm": 2.625, "learning_rate": 3.0286443068785885e-05, "loss": 0.2335, "step": 14100 }, { "epoch": 87.58566978193146, "grad_norm": 2.921875, "learning_rate": 3.0284034993464215e-05, "loss": 0.292, "step": 14101 }, { "epoch": 87.59190031152647, "grad_norm": 2.296875, "learning_rate": 3.0281626866822953e-05, "loss": 0.1138, "step": 14102 }, { "epoch": 87.59813084112149, "grad_norm": 1.296875, "learning_rate": 3.0279218688885468e-05, "loss": 0.1122, "step": 14103 }, { "epoch": 87.60436137071652, "grad_norm": 2.359375, "learning_rate": 3.027681045967516e-05, "loss": 0.2277, "step": 14104 }, { "epoch": 87.61059190031153, "grad_norm": 1.90625, "learning_rate": 3.0274402179215405e-05, "loss": 0.1425, "step": 14105 }, { "epoch": 87.61682242990655, "grad_norm": 2.21875, "learning_rate": 3.027199384752962e-05, "loss": 0.1797, "step": 14106 }, { "epoch": 87.62305295950156, "grad_norm": 2.8125, "learning_rate": 3.0269585464641158e-05, "loss": 0.1383, "step": 14107 }, { "epoch": 87.62928348909658, "grad_norm": 1.9609375, "learning_rate": 3.0267177030573435e-05, "loss": 0.1573, "step": 14108 }, { "epoch": 87.6355140186916, "grad_norm": 3.46875, "learning_rate": 3.026476854534984e-05, "loss": 0.1688, "step": 14109 }, { "epoch": 87.64174454828661, "grad_norm": 2.15625, "learning_rate": 3.0262360008993756e-05, "loss": 0.1323, "step": 14110 }, { "epoch": 87.64797507788163, "grad_norm": 2.0, "learning_rate": 3.0259951421528583e-05, "loss": 0.1316, "step": 14111 }, { "epoch": 87.65420560747664, "grad_norm": 1.71875, "learning_rate": 3.0257542782977705e-05, "loss": 0.1414, "step": 14112 }, { "epoch": 87.66043613707166, "grad_norm": 4.46875, "learning_rate": 3.025513409336453e-05, "loss": 0.1406, "step": 14113 }, { "epoch": 87.66666666666667, "grad_norm": 3.21875, "learning_rate": 3.025272535271243e-05, "loss": 0.1919, "step": 14114 }, { "epoch": 87.67289719626169, "grad_norm": 2.578125, "learning_rate": 3.0250316561044823e-05, "loss": 0.1601, "step": 14115 }, { "epoch": 87.6791277258567, "grad_norm": 2.125, "learning_rate": 3.0247907718385083e-05, "loss": 0.1686, "step": 14116 }, { "epoch": 87.68535825545172, "grad_norm": 2.0625, "learning_rate": 3.024549882475662e-05, "loss": 0.2041, "step": 14117 }, { "epoch": 87.69158878504673, "grad_norm": 2.84375, "learning_rate": 3.0243089880182823e-05, "loss": 0.2658, "step": 14118 }, { "epoch": 87.69781931464175, "grad_norm": 3.59375, "learning_rate": 3.024068088468709e-05, "loss": 0.4023, "step": 14119 }, { "epoch": 87.70404984423676, "grad_norm": 2.03125, "learning_rate": 3.0238271838292813e-05, "loss": 0.163, "step": 14120 }, { "epoch": 87.71028037383178, "grad_norm": 1.765625, "learning_rate": 3.023586274102339e-05, "loss": 0.1188, "step": 14121 }, { "epoch": 87.7165109034268, "grad_norm": 2.125, "learning_rate": 3.0233453592902232e-05, "loss": 0.2295, "step": 14122 }, { "epoch": 87.72274143302181, "grad_norm": 1.7265625, "learning_rate": 3.0231044393952712e-05, "loss": 0.1257, "step": 14123 }, { "epoch": 87.72897196261682, "grad_norm": 1.59375, "learning_rate": 3.022863514419825e-05, "loss": 0.1216, "step": 14124 }, { "epoch": 87.73520249221184, "grad_norm": 1.7421875, "learning_rate": 3.022622584366224e-05, "loss": 0.1647, "step": 14125 }, { "epoch": 87.74143302180686, "grad_norm": 3.046875, "learning_rate": 3.0223816492368086e-05, "loss": 0.1035, "step": 14126 }, { "epoch": 87.74766355140187, "grad_norm": 3.640625, "learning_rate": 3.022140709033917e-05, "loss": 0.2418, "step": 14127 }, { "epoch": 87.75389408099689, "grad_norm": 2.296875, "learning_rate": 3.0218997637598916e-05, "loss": 0.136, "step": 14128 }, { "epoch": 87.7601246105919, "grad_norm": 2.296875, "learning_rate": 3.0216588134170705e-05, "loss": 0.1492, "step": 14129 }, { "epoch": 87.76635514018692, "grad_norm": 2.40625, "learning_rate": 3.021417858007795e-05, "loss": 0.3027, "step": 14130 }, { "epoch": 87.77258566978193, "grad_norm": 2.109375, "learning_rate": 3.0211768975344053e-05, "loss": 0.1666, "step": 14131 }, { "epoch": 87.77881619937695, "grad_norm": 1.3359375, "learning_rate": 3.020935931999241e-05, "loss": 0.1359, "step": 14132 }, { "epoch": 87.78504672897196, "grad_norm": 2.125, "learning_rate": 3.0206949614046433e-05, "loss": 0.1186, "step": 14133 }, { "epoch": 87.79127725856698, "grad_norm": 2.3125, "learning_rate": 3.0204539857529507e-05, "loss": 0.2741, "step": 14134 }, { "epoch": 87.797507788162, "grad_norm": 2.578125, "learning_rate": 3.0202130050465066e-05, "loss": 0.1264, "step": 14135 }, { "epoch": 87.80373831775701, "grad_norm": 1.6640625, "learning_rate": 3.019972019287649e-05, "loss": 0.1396, "step": 14136 }, { "epoch": 87.80996884735202, "grad_norm": 2.15625, "learning_rate": 3.0197310284787195e-05, "loss": 0.2361, "step": 14137 }, { "epoch": 87.81619937694704, "grad_norm": 2.1875, "learning_rate": 3.0194900326220582e-05, "loss": 0.1661, "step": 14138 }, { "epoch": 87.82242990654206, "grad_norm": 2.421875, "learning_rate": 3.0192490317200058e-05, "loss": 0.1739, "step": 14139 }, { "epoch": 87.82866043613707, "grad_norm": 2.875, "learning_rate": 3.0190080257749025e-05, "loss": 0.2369, "step": 14140 }, { "epoch": 87.83489096573209, "grad_norm": 2.59375, "learning_rate": 3.0187670147890906e-05, "loss": 0.1874, "step": 14141 }, { "epoch": 87.8411214953271, "grad_norm": 2.4375, "learning_rate": 3.018525998764909e-05, "loss": 0.1729, "step": 14142 }, { "epoch": 87.84735202492212, "grad_norm": 3.265625, "learning_rate": 3.0182849777046995e-05, "loss": 0.2204, "step": 14143 }, { "epoch": 87.85358255451713, "grad_norm": 3.453125, "learning_rate": 3.0180439516108027e-05, "loss": 0.2273, "step": 14144 }, { "epoch": 87.85981308411215, "grad_norm": 1.7734375, "learning_rate": 3.0178029204855597e-05, "loss": 0.12, "step": 14145 }, { "epoch": 87.86604361370716, "grad_norm": 2.515625, "learning_rate": 3.017561884331311e-05, "loss": 0.2276, "step": 14146 }, { "epoch": 87.87227414330218, "grad_norm": 2.046875, "learning_rate": 3.0173208431503974e-05, "loss": 0.1706, "step": 14147 }, { "epoch": 87.8785046728972, "grad_norm": 1.671875, "learning_rate": 3.0170797969451615e-05, "loss": 0.1347, "step": 14148 }, { "epoch": 87.88473520249221, "grad_norm": 0.80078125, "learning_rate": 3.016838745717942e-05, "loss": 0.0918, "step": 14149 }, { "epoch": 87.89096573208722, "grad_norm": 2.125, "learning_rate": 3.016597689471082e-05, "loss": 0.1623, "step": 14150 }, { "epoch": 87.89719626168224, "grad_norm": 2.734375, "learning_rate": 3.0163566282069216e-05, "loss": 0.1989, "step": 14151 }, { "epoch": 87.90342679127725, "grad_norm": 3.59375, "learning_rate": 3.0161155619278024e-05, "loss": 0.2243, "step": 14152 }, { "epoch": 87.90965732087227, "grad_norm": 3.09375, "learning_rate": 3.015874490636066e-05, "loss": 0.1042, "step": 14153 }, { "epoch": 87.91588785046729, "grad_norm": 2.53125, "learning_rate": 3.0156334143340537e-05, "loss": 0.1572, "step": 14154 }, { "epoch": 87.9221183800623, "grad_norm": 1.09375, "learning_rate": 3.015392333024106e-05, "loss": 0.1093, "step": 14155 }, { "epoch": 87.92834890965732, "grad_norm": 1.84375, "learning_rate": 3.0151512467085653e-05, "loss": 0.1472, "step": 14156 }, { "epoch": 87.93457943925233, "grad_norm": 3.203125, "learning_rate": 3.0149101553897723e-05, "loss": 0.2233, "step": 14157 }, { "epoch": 87.94080996884735, "grad_norm": 3.1875, "learning_rate": 3.0146690590700692e-05, "loss": 0.2195, "step": 14158 }, { "epoch": 87.94704049844236, "grad_norm": 2.90625, "learning_rate": 3.0144279577517975e-05, "loss": 0.2291, "step": 14159 }, { "epoch": 87.95327102803738, "grad_norm": 2.59375, "learning_rate": 3.0141868514372984e-05, "loss": 0.157, "step": 14160 }, { "epoch": 87.9595015576324, "grad_norm": 1.3203125, "learning_rate": 3.013945740128914e-05, "loss": 0.1128, "step": 14161 }, { "epoch": 87.96573208722741, "grad_norm": 2.046875, "learning_rate": 3.0137046238289852e-05, "loss": 0.1089, "step": 14162 }, { "epoch": 87.97196261682242, "grad_norm": 2.546875, "learning_rate": 3.013463502539855e-05, "loss": 0.1747, "step": 14163 }, { "epoch": 87.97819314641744, "grad_norm": 2.359375, "learning_rate": 3.0132223762638646e-05, "loss": 0.2508, "step": 14164 }, { "epoch": 87.98442367601245, "grad_norm": 2.59375, "learning_rate": 3.0129812450033562e-05, "loss": 0.2307, "step": 14165 }, { "epoch": 87.99065420560747, "grad_norm": 2.296875, "learning_rate": 3.0127401087606712e-05, "loss": 0.1401, "step": 14166 }, { "epoch": 87.99688473520249, "grad_norm": 2.578125, "learning_rate": 3.0124989675381514e-05, "loss": 0.2443, "step": 14167 }, { "epoch": 88.0, "grad_norm": 1.671875, "learning_rate": 3.0122578213381398e-05, "loss": 0.0986, "step": 14168 }, { "epoch": 88.00623052959502, "grad_norm": 3.484375, "learning_rate": 3.012016670162977e-05, "loss": 0.2524, "step": 14169 }, { "epoch": 88.01246105919003, "grad_norm": 1.671875, "learning_rate": 3.0117755140150066e-05, "loss": 0.1178, "step": 14170 }, { "epoch": 88.01869158878505, "grad_norm": 3.0625, "learning_rate": 3.0115343528965696e-05, "loss": 0.2335, "step": 14171 }, { "epoch": 88.02492211838006, "grad_norm": 2.96875, "learning_rate": 3.0112931868100092e-05, "loss": 0.1406, "step": 14172 }, { "epoch": 88.03115264797508, "grad_norm": 4.21875, "learning_rate": 3.011052015757667e-05, "loss": 0.2604, "step": 14173 }, { "epoch": 88.03738317757009, "grad_norm": 2.4375, "learning_rate": 3.0108108397418856e-05, "loss": 0.184, "step": 14174 }, { "epoch": 88.04361370716511, "grad_norm": 4.625, "learning_rate": 3.0105696587650067e-05, "loss": 0.2044, "step": 14175 }, { "epoch": 88.04984423676012, "grad_norm": 4.84375, "learning_rate": 3.010328472829374e-05, "loss": 0.1353, "step": 14176 }, { "epoch": 88.05607476635514, "grad_norm": 2.96875, "learning_rate": 3.010087281937328e-05, "loss": 0.1354, "step": 14177 }, { "epoch": 88.06230529595015, "grad_norm": 2.46875, "learning_rate": 3.009846086091213e-05, "loss": 0.103, "step": 14178 }, { "epoch": 88.06853582554517, "grad_norm": 2.203125, "learning_rate": 3.009604885293371e-05, "loss": 0.1266, "step": 14179 }, { "epoch": 88.07476635514018, "grad_norm": 2.09375, "learning_rate": 3.009363679546145e-05, "loss": 0.2627, "step": 14180 }, { "epoch": 88.0809968847352, "grad_norm": 1.7734375, "learning_rate": 3.0091224688518764e-05, "loss": 0.1354, "step": 14181 }, { "epoch": 88.08722741433021, "grad_norm": 3.078125, "learning_rate": 3.0088812532129084e-05, "loss": 0.1164, "step": 14182 }, { "epoch": 88.09345794392523, "grad_norm": 1.5703125, "learning_rate": 3.008640032631585e-05, "loss": 0.1667, "step": 14183 }, { "epoch": 88.09968847352025, "grad_norm": 2.65625, "learning_rate": 3.008398807110247e-05, "loss": 0.1429, "step": 14184 }, { "epoch": 88.10591900311526, "grad_norm": 1.8125, "learning_rate": 3.008157576651238e-05, "loss": 0.1555, "step": 14185 }, { "epoch": 88.11214953271028, "grad_norm": 1.9140625, "learning_rate": 3.0079163412569017e-05, "loss": 0.1601, "step": 14186 }, { "epoch": 88.11838006230529, "grad_norm": 1.7265625, "learning_rate": 3.0076751009295806e-05, "loss": 0.144, "step": 14187 }, { "epoch": 88.1246105919003, "grad_norm": 1.84375, "learning_rate": 3.0074338556716164e-05, "loss": 0.1312, "step": 14188 }, { "epoch": 88.13084112149532, "grad_norm": 1.1796875, "learning_rate": 3.0071926054853546e-05, "loss": 0.1047, "step": 14189 }, { "epoch": 88.13707165109034, "grad_norm": 1.6015625, "learning_rate": 3.006951350373135e-05, "loss": 0.1052, "step": 14190 }, { "epoch": 88.14330218068535, "grad_norm": 2.84375, "learning_rate": 3.0067100903373034e-05, "loss": 0.2372, "step": 14191 }, { "epoch": 88.14953271028037, "grad_norm": 2.375, "learning_rate": 3.0064688253802026e-05, "loss": 0.1046, "step": 14192 }, { "epoch": 88.15576323987538, "grad_norm": 1.9375, "learning_rate": 3.006227555504175e-05, "loss": 0.1625, "step": 14193 }, { "epoch": 88.1619937694704, "grad_norm": 1.7265625, "learning_rate": 3.005986280711564e-05, "loss": 0.1194, "step": 14194 }, { "epoch": 88.16822429906541, "grad_norm": 2.109375, "learning_rate": 3.0057450010047133e-05, "loss": 0.1791, "step": 14195 }, { "epoch": 88.17445482866043, "grad_norm": 2.078125, "learning_rate": 3.0055037163859663e-05, "loss": 0.1767, "step": 14196 }, { "epoch": 88.18068535825545, "grad_norm": 1.7265625, "learning_rate": 3.0052624268576658e-05, "loss": 0.1215, "step": 14197 }, { "epoch": 88.18691588785046, "grad_norm": 3.34375, "learning_rate": 3.0050211324221555e-05, "loss": 0.1456, "step": 14198 }, { "epoch": 88.19314641744548, "grad_norm": 2.390625, "learning_rate": 3.004779833081779e-05, "loss": 0.1687, "step": 14199 }, { "epoch": 88.19937694704049, "grad_norm": 2.4375, "learning_rate": 3.00453852883888e-05, "loss": 0.2152, "step": 14200 }, { "epoch": 88.20560747663552, "grad_norm": 2.65625, "learning_rate": 3.004297219695802e-05, "loss": 0.1632, "step": 14201 }, { "epoch": 88.21183800623054, "grad_norm": 2.328125, "learning_rate": 3.0040559056548888e-05, "loss": 0.1674, "step": 14202 }, { "epoch": 88.21806853582555, "grad_norm": 1.5546875, "learning_rate": 3.003814586718483e-05, "loss": 0.1272, "step": 14203 }, { "epoch": 88.22429906542057, "grad_norm": 1.8671875, "learning_rate": 3.00357326288893e-05, "loss": 0.1136, "step": 14204 }, { "epoch": 88.23052959501558, "grad_norm": 2.046875, "learning_rate": 3.003331934168573e-05, "loss": 0.1283, "step": 14205 }, { "epoch": 88.2367601246106, "grad_norm": 2.734375, "learning_rate": 3.0030906005597553e-05, "loss": 0.1606, "step": 14206 }, { "epoch": 88.24299065420561, "grad_norm": 3.5625, "learning_rate": 3.0028492620648214e-05, "loss": 0.2788, "step": 14207 }, { "epoch": 88.24922118380063, "grad_norm": 2.859375, "learning_rate": 3.0026079186861146e-05, "loss": 0.2301, "step": 14208 }, { "epoch": 88.25545171339564, "grad_norm": 1.6875, "learning_rate": 3.0023665704259797e-05, "loss": 0.1338, "step": 14209 }, { "epoch": 88.26168224299066, "grad_norm": 2.59375, "learning_rate": 3.0021252172867592e-05, "loss": 0.1439, "step": 14210 }, { "epoch": 88.26791277258567, "grad_norm": 4.0625, "learning_rate": 3.0018838592707993e-05, "loss": 0.2739, "step": 14211 }, { "epoch": 88.27414330218069, "grad_norm": 1.78125, "learning_rate": 3.0016424963804423e-05, "loss": 0.1616, "step": 14212 }, { "epoch": 88.2803738317757, "grad_norm": 2.21875, "learning_rate": 3.001401128618033e-05, "loss": 0.1527, "step": 14213 }, { "epoch": 88.28660436137072, "grad_norm": 2.125, "learning_rate": 3.0011597559859167e-05, "loss": 0.1457, "step": 14214 }, { "epoch": 88.29283489096574, "grad_norm": 2.546875, "learning_rate": 3.0009183784864365e-05, "loss": 0.2019, "step": 14215 }, { "epoch": 88.29906542056075, "grad_norm": 1.4296875, "learning_rate": 3.0006769961219362e-05, "loss": 0.1041, "step": 14216 }, { "epoch": 88.30529595015577, "grad_norm": 2.25, "learning_rate": 3.0004356088947604e-05, "loss": 0.1438, "step": 14217 }, { "epoch": 88.31152647975078, "grad_norm": 1.4765625, "learning_rate": 3.0001942168072556e-05, "loss": 0.1313, "step": 14218 }, { "epoch": 88.3177570093458, "grad_norm": 1.6015625, "learning_rate": 2.999952819861763e-05, "loss": 0.1469, "step": 14219 }, { "epoch": 88.32398753894081, "grad_norm": 2.875, "learning_rate": 2.9997114180606296e-05, "loss": 0.2241, "step": 14220 }, { "epoch": 88.33021806853583, "grad_norm": 1.6171875, "learning_rate": 2.9994700114061985e-05, "loss": 0.0971, "step": 14221 }, { "epoch": 88.33644859813084, "grad_norm": 2.609375, "learning_rate": 2.999228599900815e-05, "loss": 0.1239, "step": 14222 }, { "epoch": 88.34267912772586, "grad_norm": 1.59375, "learning_rate": 2.998987183546823e-05, "loss": 0.1186, "step": 14223 }, { "epoch": 88.34890965732087, "grad_norm": 1.3671875, "learning_rate": 2.998745762346569e-05, "loss": 0.1241, "step": 14224 }, { "epoch": 88.35514018691589, "grad_norm": 3.203125, "learning_rate": 2.9985043363023958e-05, "loss": 0.2643, "step": 14225 }, { "epoch": 88.3613707165109, "grad_norm": 3.25, "learning_rate": 2.9982629054166482e-05, "loss": 0.2175, "step": 14226 }, { "epoch": 88.36760124610592, "grad_norm": 1.9140625, "learning_rate": 2.9980214696916726e-05, "loss": 0.1022, "step": 14227 }, { "epoch": 88.37383177570094, "grad_norm": 3.109375, "learning_rate": 2.997780029129813e-05, "loss": 0.1601, "step": 14228 }, { "epoch": 88.38006230529595, "grad_norm": 3.515625, "learning_rate": 2.9975385837334135e-05, "loss": 0.2655, "step": 14229 }, { "epoch": 88.38629283489097, "grad_norm": 1.65625, "learning_rate": 2.9972971335048205e-05, "loss": 0.1085, "step": 14230 }, { "epoch": 88.39252336448598, "grad_norm": 2.5, "learning_rate": 2.9970556784463776e-05, "loss": 0.1472, "step": 14231 }, { "epoch": 88.398753894081, "grad_norm": 1.59375, "learning_rate": 2.996814218560431e-05, "loss": 0.1399, "step": 14232 }, { "epoch": 88.40498442367601, "grad_norm": 2.453125, "learning_rate": 2.9965727538493255e-05, "loss": 0.1652, "step": 14233 }, { "epoch": 88.41121495327103, "grad_norm": 2.828125, "learning_rate": 2.9963312843154056e-05, "loss": 0.2431, "step": 14234 }, { "epoch": 88.41744548286604, "grad_norm": 1.640625, "learning_rate": 2.996089809961018e-05, "loss": 0.1215, "step": 14235 }, { "epoch": 88.42367601246106, "grad_norm": 1.546875, "learning_rate": 2.995848330788506e-05, "loss": 0.1204, "step": 14236 }, { "epoch": 88.42990654205607, "grad_norm": 2.765625, "learning_rate": 2.9956068468002168e-05, "loss": 0.2139, "step": 14237 }, { "epoch": 88.43613707165109, "grad_norm": 2.9375, "learning_rate": 2.9953653579984942e-05, "loss": 0.3266, "step": 14238 }, { "epoch": 88.4423676012461, "grad_norm": 2.109375, "learning_rate": 2.9951238643856844e-05, "loss": 0.1574, "step": 14239 }, { "epoch": 88.44859813084112, "grad_norm": 1.6875, "learning_rate": 2.9948823659641327e-05, "loss": 0.1533, "step": 14240 }, { "epoch": 88.45482866043614, "grad_norm": 2.046875, "learning_rate": 2.9946408627361844e-05, "loss": 0.1582, "step": 14241 }, { "epoch": 88.46105919003115, "grad_norm": 2.578125, "learning_rate": 2.9943993547041854e-05, "loss": 0.0888, "step": 14242 }, { "epoch": 88.46728971962617, "grad_norm": 2.453125, "learning_rate": 2.9941578418704813e-05, "loss": 0.1686, "step": 14243 }, { "epoch": 88.47352024922118, "grad_norm": 2.984375, "learning_rate": 2.993916324237417e-05, "loss": 0.1619, "step": 14244 }, { "epoch": 88.4797507788162, "grad_norm": 2.046875, "learning_rate": 2.9936748018073378e-05, "loss": 0.1181, "step": 14245 }, { "epoch": 88.48598130841121, "grad_norm": 1.53125, "learning_rate": 2.993433274582592e-05, "loss": 0.1259, "step": 14246 }, { "epoch": 88.49221183800623, "grad_norm": 2.5625, "learning_rate": 2.9931917425655225e-05, "loss": 0.2002, "step": 14247 }, { "epoch": 88.49844236760124, "grad_norm": 1.921875, "learning_rate": 2.9929502057584763e-05, "loss": 0.1509, "step": 14248 }, { "epoch": 88.50467289719626, "grad_norm": 2.828125, "learning_rate": 2.9927086641637997e-05, "loss": 0.2661, "step": 14249 }, { "epoch": 88.51090342679127, "grad_norm": 2.34375, "learning_rate": 2.9924671177838374e-05, "loss": 0.2028, "step": 14250 }, { "epoch": 88.51713395638629, "grad_norm": 2.9375, "learning_rate": 2.992225566620936e-05, "loss": 0.2506, "step": 14251 }, { "epoch": 88.5233644859813, "grad_norm": 1.734375, "learning_rate": 2.9919840106774423e-05, "loss": 0.127, "step": 14252 }, { "epoch": 88.52959501557632, "grad_norm": 1.7109375, "learning_rate": 2.9917424499557005e-05, "loss": 0.1025, "step": 14253 }, { "epoch": 88.53582554517133, "grad_norm": 2.203125, "learning_rate": 2.9915008844580583e-05, "loss": 0.195, "step": 14254 }, { "epoch": 88.54205607476635, "grad_norm": 1.84375, "learning_rate": 2.9912593141868616e-05, "loss": 0.0996, "step": 14255 }, { "epoch": 88.54828660436137, "grad_norm": 2.265625, "learning_rate": 2.991017739144456e-05, "loss": 0.1848, "step": 14256 }, { "epoch": 88.55451713395638, "grad_norm": 3.0625, "learning_rate": 2.990776159333188e-05, "loss": 0.1363, "step": 14257 }, { "epoch": 88.5607476635514, "grad_norm": 3.203125, "learning_rate": 2.990534574755403e-05, "loss": 0.1684, "step": 14258 }, { "epoch": 88.56697819314641, "grad_norm": 3.5625, "learning_rate": 2.99029298541345e-05, "loss": 0.2617, "step": 14259 }, { "epoch": 88.57320872274143, "grad_norm": 2.953125, "learning_rate": 2.990051391309671e-05, "loss": 0.1533, "step": 14260 }, { "epoch": 88.57943925233644, "grad_norm": 2.5625, "learning_rate": 2.989809792446417e-05, "loss": 0.1707, "step": 14261 }, { "epoch": 88.58566978193146, "grad_norm": 3.0625, "learning_rate": 2.989568188826031e-05, "loss": 0.1328, "step": 14262 }, { "epoch": 88.59190031152647, "grad_norm": 2.203125, "learning_rate": 2.989326580450862e-05, "loss": 0.1545, "step": 14263 }, { "epoch": 88.59813084112149, "grad_norm": 2.21875, "learning_rate": 2.9890849673232553e-05, "loss": 0.148, "step": 14264 }, { "epoch": 88.60436137071652, "grad_norm": 3.296875, "learning_rate": 2.988843349445557e-05, "loss": 0.1482, "step": 14265 }, { "epoch": 88.61059190031153, "grad_norm": 3.34375, "learning_rate": 2.988601726820115e-05, "loss": 0.2739, "step": 14266 }, { "epoch": 88.61682242990655, "grad_norm": 3.484375, "learning_rate": 2.9883600994492743e-05, "loss": 0.1085, "step": 14267 }, { "epoch": 88.62305295950156, "grad_norm": 4.375, "learning_rate": 2.9881184673353836e-05, "loss": 0.2312, "step": 14268 }, { "epoch": 88.62928348909658, "grad_norm": 2.109375, "learning_rate": 2.987876830480789e-05, "loss": 0.1145, "step": 14269 }, { "epoch": 88.6355140186916, "grad_norm": 3.03125, "learning_rate": 2.9876351888878367e-05, "loss": 0.2243, "step": 14270 }, { "epoch": 88.64174454828661, "grad_norm": 3.109375, "learning_rate": 2.9873935425588734e-05, "loss": 0.1125, "step": 14271 }, { "epoch": 88.64797507788163, "grad_norm": 2.625, "learning_rate": 2.9871518914962475e-05, "loss": 0.242, "step": 14272 }, { "epoch": 88.65420560747664, "grad_norm": 3.078125, "learning_rate": 2.986910235702305e-05, "loss": 0.1601, "step": 14273 }, { "epoch": 88.66043613707166, "grad_norm": 3.28125, "learning_rate": 2.986668575179392e-05, "loss": 0.1489, "step": 14274 }, { "epoch": 88.66666666666667, "grad_norm": 2.8125, "learning_rate": 2.9864269099298574e-05, "loss": 0.1186, "step": 14275 }, { "epoch": 88.67289719626169, "grad_norm": 2.34375, "learning_rate": 2.9861852399560476e-05, "loss": 0.142, "step": 14276 }, { "epoch": 88.6791277258567, "grad_norm": 2.140625, "learning_rate": 2.985943565260309e-05, "loss": 0.092, "step": 14277 }, { "epoch": 88.68535825545172, "grad_norm": 4.4375, "learning_rate": 2.98570188584499e-05, "loss": 0.2763, "step": 14278 }, { "epoch": 88.69158878504673, "grad_norm": 2.5625, "learning_rate": 2.9854602017124366e-05, "loss": 0.1764, "step": 14279 }, { "epoch": 88.69781931464175, "grad_norm": 3.1875, "learning_rate": 2.9852185128649964e-05, "loss": 0.2409, "step": 14280 }, { "epoch": 88.70404984423676, "grad_norm": 2.859375, "learning_rate": 2.9849768193050175e-05, "loss": 0.1999, "step": 14281 }, { "epoch": 88.71028037383178, "grad_norm": 3.171875, "learning_rate": 2.9847351210348467e-05, "loss": 0.2787, "step": 14282 }, { "epoch": 88.7165109034268, "grad_norm": 1.609375, "learning_rate": 2.984493418056831e-05, "loss": 0.1343, "step": 14283 }, { "epoch": 88.72274143302181, "grad_norm": 3.953125, "learning_rate": 2.9842517103733192e-05, "loss": 0.3218, "step": 14284 }, { "epoch": 88.72897196261682, "grad_norm": 1.484375, "learning_rate": 2.984009997986658e-05, "loss": 0.129, "step": 14285 }, { "epoch": 88.73520249221184, "grad_norm": 2.46875, "learning_rate": 2.983768280899194e-05, "loss": 0.1178, "step": 14286 }, { "epoch": 88.74143302180686, "grad_norm": 1.875, "learning_rate": 2.9835265591132767e-05, "loss": 0.1642, "step": 14287 }, { "epoch": 88.74766355140187, "grad_norm": 1.8984375, "learning_rate": 2.9832848326312522e-05, "loss": 0.1038, "step": 14288 }, { "epoch": 88.75389408099689, "grad_norm": 1.7734375, "learning_rate": 2.9830431014554696e-05, "loss": 0.1393, "step": 14289 }, { "epoch": 88.7601246105919, "grad_norm": 1.53125, "learning_rate": 2.9828013655882753e-05, "loss": 0.0868, "step": 14290 }, { "epoch": 88.76635514018692, "grad_norm": 1.703125, "learning_rate": 2.9825596250320174e-05, "loss": 0.1416, "step": 14291 }, { "epoch": 88.77258566978193, "grad_norm": 1.96875, "learning_rate": 2.9823178797890443e-05, "loss": 0.2621, "step": 14292 }, { "epoch": 88.77881619937695, "grad_norm": 1.7265625, "learning_rate": 2.9820761298617028e-05, "loss": 0.1328, "step": 14293 }, { "epoch": 88.78504672897196, "grad_norm": 3.453125, "learning_rate": 2.981834375252343e-05, "loss": 0.1294, "step": 14294 }, { "epoch": 88.79127725856698, "grad_norm": 1.65625, "learning_rate": 2.98159261596331e-05, "loss": 0.0916, "step": 14295 }, { "epoch": 88.797507788162, "grad_norm": 1.5078125, "learning_rate": 2.9813508519969542e-05, "loss": 0.1291, "step": 14296 }, { "epoch": 88.80373831775701, "grad_norm": 2.140625, "learning_rate": 2.9811090833556222e-05, "loss": 0.2155, "step": 14297 }, { "epoch": 88.80996884735202, "grad_norm": 2.15625, "learning_rate": 2.9808673100416628e-05, "loss": 0.194, "step": 14298 }, { "epoch": 88.81619937694704, "grad_norm": 3.3125, "learning_rate": 2.980625532057424e-05, "loss": 0.1792, "step": 14299 }, { "epoch": 88.82242990654206, "grad_norm": 1.4921875, "learning_rate": 2.9803837494052545e-05, "loss": 0.1431, "step": 14300 }, { "epoch": 88.82866043613707, "grad_norm": 3.203125, "learning_rate": 2.980141962087501e-05, "loss": 0.3333, "step": 14301 }, { "epoch": 88.83489096573209, "grad_norm": 2.28125, "learning_rate": 2.979900170106513e-05, "loss": 0.1345, "step": 14302 }, { "epoch": 88.8411214953271, "grad_norm": 2.5625, "learning_rate": 2.979658373464639e-05, "loss": 0.2678, "step": 14303 }, { "epoch": 88.84735202492212, "grad_norm": 1.9375, "learning_rate": 2.979416572164227e-05, "loss": 0.1333, "step": 14304 }, { "epoch": 88.85358255451713, "grad_norm": 1.8125, "learning_rate": 2.9791747662076252e-05, "loss": 0.1292, "step": 14305 }, { "epoch": 88.85981308411215, "grad_norm": 1.984375, "learning_rate": 2.978932955597182e-05, "loss": 0.1706, "step": 14306 }, { "epoch": 88.86604361370716, "grad_norm": 2.640625, "learning_rate": 2.978691140335247e-05, "loss": 0.1974, "step": 14307 }, { "epoch": 88.87227414330218, "grad_norm": 2.609375, "learning_rate": 2.978449320424167e-05, "loss": 0.2381, "step": 14308 }, { "epoch": 88.8785046728972, "grad_norm": 2.671875, "learning_rate": 2.978207495866292e-05, "loss": 0.2889, "step": 14309 }, { "epoch": 88.88473520249221, "grad_norm": 2.6875, "learning_rate": 2.97796566666397e-05, "loss": 0.2028, "step": 14310 }, { "epoch": 88.89096573208722, "grad_norm": 1.265625, "learning_rate": 2.9777238328195506e-05, "loss": 0.1275, "step": 14311 }, { "epoch": 88.89719626168224, "grad_norm": 1.8046875, "learning_rate": 2.9774819943353814e-05, "loss": 0.1359, "step": 14312 }, { "epoch": 88.90342679127725, "grad_norm": 3.875, "learning_rate": 2.9772401512138115e-05, "loss": 0.2171, "step": 14313 }, { "epoch": 88.90965732087227, "grad_norm": 0.98046875, "learning_rate": 2.9769983034571902e-05, "loss": 0.0841, "step": 14314 }, { "epoch": 88.91588785046729, "grad_norm": 2.53125, "learning_rate": 2.9767564510678652e-05, "loss": 0.1161, "step": 14315 }, { "epoch": 88.9221183800623, "grad_norm": 2.71875, "learning_rate": 2.9765145940481875e-05, "loss": 0.2908, "step": 14316 }, { "epoch": 88.92834890965732, "grad_norm": 1.9140625, "learning_rate": 2.9762727324005037e-05, "loss": 0.1008, "step": 14317 }, { "epoch": 88.93457943925233, "grad_norm": 2.265625, "learning_rate": 2.976030866127164e-05, "loss": 0.1493, "step": 14318 }, { "epoch": 88.94080996884735, "grad_norm": 3.03125, "learning_rate": 2.975788995230518e-05, "loss": 0.2191, "step": 14319 }, { "epoch": 88.94704049844236, "grad_norm": 3.359375, "learning_rate": 2.975547119712914e-05, "loss": 0.2423, "step": 14320 }, { "epoch": 88.95327102803738, "grad_norm": 4.21875, "learning_rate": 2.9753052395767007e-05, "loss": 0.2172, "step": 14321 }, { "epoch": 88.9595015576324, "grad_norm": 2.84375, "learning_rate": 2.9750633548242285e-05, "loss": 0.1176, "step": 14322 }, { "epoch": 88.96573208722741, "grad_norm": 2.859375, "learning_rate": 2.9748214654578462e-05, "loss": 0.2291, "step": 14323 }, { "epoch": 88.97196261682242, "grad_norm": 2.125, "learning_rate": 2.9745795714799024e-05, "loss": 0.1546, "step": 14324 }, { "epoch": 88.97819314641744, "grad_norm": 3.046875, "learning_rate": 2.9743376728927476e-05, "loss": 0.1953, "step": 14325 }, { "epoch": 88.98442367601245, "grad_norm": 3.109375, "learning_rate": 2.97409576969873e-05, "loss": 0.2671, "step": 14326 }, { "epoch": 88.99065420560747, "grad_norm": 3.40625, "learning_rate": 2.9738538619002e-05, "loss": 0.2877, "step": 14327 }, { "epoch": 88.99688473520249, "grad_norm": 3.40625, "learning_rate": 2.9736119494995053e-05, "loss": 0.1643, "step": 14328 }, { "epoch": 89.0, "grad_norm": 0.9609375, "learning_rate": 2.9733700324989984e-05, "loss": 0.0403, "step": 14329 }, { "epoch": 89.00623052959502, "grad_norm": 1.1328125, "learning_rate": 2.9731281109010256e-05, "loss": 0.1174, "step": 14330 }, { "epoch": 89.01246105919003, "grad_norm": 2.828125, "learning_rate": 2.9728861847079388e-05, "loss": 0.2679, "step": 14331 }, { "epoch": 89.01869158878505, "grad_norm": 2.90625, "learning_rate": 2.9726442539220868e-05, "loss": 0.2558, "step": 14332 }, { "epoch": 89.02492211838006, "grad_norm": 2.046875, "learning_rate": 2.9724023185458194e-05, "loss": 0.1317, "step": 14333 }, { "epoch": 89.03115264797508, "grad_norm": 3.359375, "learning_rate": 2.9721603785814856e-05, "loss": 0.1549, "step": 14334 }, { "epoch": 89.03738317757009, "grad_norm": 2.296875, "learning_rate": 2.971918434031437e-05, "loss": 0.1694, "step": 14335 }, { "epoch": 89.04361370716511, "grad_norm": 1.984375, "learning_rate": 2.971676484898021e-05, "loss": 0.1796, "step": 14336 }, { "epoch": 89.04984423676012, "grad_norm": 2.0625, "learning_rate": 2.9714345311835895e-05, "loss": 0.1288, "step": 14337 }, { "epoch": 89.05607476635514, "grad_norm": 2.296875, "learning_rate": 2.971192572890491e-05, "loss": 0.1758, "step": 14338 }, { "epoch": 89.06230529595015, "grad_norm": 2.25, "learning_rate": 2.9709506100210764e-05, "loss": 0.1645, "step": 14339 }, { "epoch": 89.06853582554517, "grad_norm": 2.3125, "learning_rate": 2.9707086425776958e-05, "loss": 0.1154, "step": 14340 }, { "epoch": 89.07476635514018, "grad_norm": 2.046875, "learning_rate": 2.9704666705626977e-05, "loss": 0.1494, "step": 14341 }, { "epoch": 89.0809968847352, "grad_norm": 1.734375, "learning_rate": 2.9702246939784347e-05, "loss": 0.1516, "step": 14342 }, { "epoch": 89.08722741433021, "grad_norm": 1.9140625, "learning_rate": 2.969982712827254e-05, "loss": 0.1822, "step": 14343 }, { "epoch": 89.09345794392523, "grad_norm": 1.5078125, "learning_rate": 2.969740727111508e-05, "loss": 0.1071, "step": 14344 }, { "epoch": 89.09968847352025, "grad_norm": 3.484375, "learning_rate": 2.969498736833546e-05, "loss": 0.2371, "step": 14345 }, { "epoch": 89.10591900311526, "grad_norm": 1.7265625, "learning_rate": 2.9692567419957184e-05, "loss": 0.1518, "step": 14346 }, { "epoch": 89.11214953271028, "grad_norm": 1.3046875, "learning_rate": 2.969014742600375e-05, "loss": 0.1232, "step": 14347 }, { "epoch": 89.11838006230529, "grad_norm": 2.1875, "learning_rate": 2.9687727386498682e-05, "loss": 0.2609, "step": 14348 }, { "epoch": 89.1246105919003, "grad_norm": 2.90625, "learning_rate": 2.9685307301465454e-05, "loss": 0.2144, "step": 14349 }, { "epoch": 89.13084112149532, "grad_norm": 2.140625, "learning_rate": 2.9682887170927587e-05, "loss": 0.1691, "step": 14350 }, { "epoch": 89.13707165109034, "grad_norm": 1.4296875, "learning_rate": 2.968046699490859e-05, "loss": 0.1264, "step": 14351 }, { "epoch": 89.14330218068535, "grad_norm": 2.1875, "learning_rate": 2.9678046773431963e-05, "loss": 0.1759, "step": 14352 }, { "epoch": 89.14953271028037, "grad_norm": 1.265625, "learning_rate": 2.9675626506521205e-05, "loss": 0.1254, "step": 14353 }, { "epoch": 89.15576323987538, "grad_norm": 3.328125, "learning_rate": 2.9673206194199825e-05, "loss": 0.3192, "step": 14354 }, { "epoch": 89.1619937694704, "grad_norm": 1.3046875, "learning_rate": 2.9670785836491337e-05, "loss": 0.1174, "step": 14355 }, { "epoch": 89.16822429906541, "grad_norm": 1.671875, "learning_rate": 2.9668365433419237e-05, "loss": 0.1146, "step": 14356 }, { "epoch": 89.17445482866043, "grad_norm": 3.5, "learning_rate": 2.966594498500705e-05, "loss": 0.2619, "step": 14357 }, { "epoch": 89.18068535825545, "grad_norm": 3.140625, "learning_rate": 2.9663524491278266e-05, "loss": 0.1445, "step": 14358 }, { "epoch": 89.18691588785046, "grad_norm": 1.578125, "learning_rate": 2.9661103952256396e-05, "loss": 0.1218, "step": 14359 }, { "epoch": 89.19314641744548, "grad_norm": 2.296875, "learning_rate": 2.965868336796496e-05, "loss": 0.196, "step": 14360 }, { "epoch": 89.19937694704049, "grad_norm": 1.625, "learning_rate": 2.965626273842746e-05, "loss": 0.1352, "step": 14361 }, { "epoch": 89.20560747663552, "grad_norm": 2.59375, "learning_rate": 2.9653842063667403e-05, "loss": 0.203, "step": 14362 }, { "epoch": 89.21183800623054, "grad_norm": 2.984375, "learning_rate": 2.9651421343708297e-05, "loss": 0.1784, "step": 14363 }, { "epoch": 89.21806853582555, "grad_norm": 2.40625, "learning_rate": 2.9649000578573664e-05, "loss": 0.2233, "step": 14364 }, { "epoch": 89.22429906542057, "grad_norm": 1.046875, "learning_rate": 2.9646579768287002e-05, "loss": 0.1114, "step": 14365 }, { "epoch": 89.23052959501558, "grad_norm": 2.625, "learning_rate": 2.9644158912871834e-05, "loss": 0.1852, "step": 14366 }, { "epoch": 89.2367601246106, "grad_norm": 2.296875, "learning_rate": 2.9641738012351667e-05, "loss": 0.1945, "step": 14367 }, { "epoch": 89.24299065420561, "grad_norm": 1.59375, "learning_rate": 2.963931706675001e-05, "loss": 0.1708, "step": 14368 }, { "epoch": 89.24922118380063, "grad_norm": 1.703125, "learning_rate": 2.963689607609037e-05, "loss": 0.1832, "step": 14369 }, { "epoch": 89.25545171339564, "grad_norm": 1.7109375, "learning_rate": 2.9634475040396286e-05, "loss": 0.1255, "step": 14370 }, { "epoch": 89.26168224299066, "grad_norm": 1.1484375, "learning_rate": 2.9632053959691235e-05, "loss": 0.1018, "step": 14371 }, { "epoch": 89.26791277258567, "grad_norm": 3.515625, "learning_rate": 2.9629632833998766e-05, "loss": 0.2878, "step": 14372 }, { "epoch": 89.27414330218069, "grad_norm": 3.078125, "learning_rate": 2.962721166334237e-05, "loss": 0.2685, "step": 14373 }, { "epoch": 89.2803738317757, "grad_norm": 2.625, "learning_rate": 2.962479044774557e-05, "loss": 0.1237, "step": 14374 }, { "epoch": 89.28660436137072, "grad_norm": 2.9375, "learning_rate": 2.962236918723188e-05, "loss": 0.2335, "step": 14375 }, { "epoch": 89.29283489096574, "grad_norm": 1.4375, "learning_rate": 2.9619947881824818e-05, "loss": 0.1162, "step": 14376 }, { "epoch": 89.29906542056075, "grad_norm": 1.265625, "learning_rate": 2.9617526531547903e-05, "loss": 0.1103, "step": 14377 }, { "epoch": 89.30529595015577, "grad_norm": 2.453125, "learning_rate": 2.961510513642464e-05, "loss": 0.1989, "step": 14378 }, { "epoch": 89.31152647975078, "grad_norm": 1.984375, "learning_rate": 2.961268369647855e-05, "loss": 0.1735, "step": 14379 }, { "epoch": 89.3177570093458, "grad_norm": 2.53125, "learning_rate": 2.9610262211733164e-05, "loss": 0.1444, "step": 14380 }, { "epoch": 89.32398753894081, "grad_norm": 2.84375, "learning_rate": 2.9607840682211985e-05, "loss": 0.1487, "step": 14381 }, { "epoch": 89.33021806853583, "grad_norm": 3.765625, "learning_rate": 2.960541910793853e-05, "loss": 0.192, "step": 14382 }, { "epoch": 89.33644859813084, "grad_norm": 3.28125, "learning_rate": 2.960299748893634e-05, "loss": 0.2725, "step": 14383 }, { "epoch": 89.34267912772586, "grad_norm": 1.7890625, "learning_rate": 2.9600575825228904e-05, "loss": 0.1513, "step": 14384 }, { "epoch": 89.34890965732087, "grad_norm": 1.4609375, "learning_rate": 2.959815411683976e-05, "loss": 0.1232, "step": 14385 }, { "epoch": 89.35514018691589, "grad_norm": 3.34375, "learning_rate": 2.9595732363792427e-05, "loss": 0.3577, "step": 14386 }, { "epoch": 89.3613707165109, "grad_norm": 2.046875, "learning_rate": 2.9593310566110422e-05, "loss": 0.1776, "step": 14387 }, { "epoch": 89.36760124610592, "grad_norm": 1.3828125, "learning_rate": 2.9590888723817262e-05, "loss": 0.1099, "step": 14388 }, { "epoch": 89.37383177570094, "grad_norm": 1.7265625, "learning_rate": 2.958846683693648e-05, "loss": 0.187, "step": 14389 }, { "epoch": 89.38006230529595, "grad_norm": 1.2265625, "learning_rate": 2.9586044905491588e-05, "loss": 0.125, "step": 14390 }, { "epoch": 89.38629283489097, "grad_norm": 2.234375, "learning_rate": 2.9583622929506105e-05, "loss": 0.1703, "step": 14391 }, { "epoch": 89.39252336448598, "grad_norm": 2.265625, "learning_rate": 2.9581200909003566e-05, "loss": 0.2048, "step": 14392 }, { "epoch": 89.398753894081, "grad_norm": 1.421875, "learning_rate": 2.9578778844007486e-05, "loss": 0.1133, "step": 14393 }, { "epoch": 89.40498442367601, "grad_norm": 2.046875, "learning_rate": 2.9576356734541395e-05, "loss": 0.1453, "step": 14394 }, { "epoch": 89.41121495327103, "grad_norm": 1.921875, "learning_rate": 2.9573934580628805e-05, "loss": 0.1056, "step": 14395 }, { "epoch": 89.41744548286604, "grad_norm": 2.75, "learning_rate": 2.9571512382293253e-05, "loss": 0.1729, "step": 14396 }, { "epoch": 89.42367601246106, "grad_norm": 1.921875, "learning_rate": 2.956909013955826e-05, "loss": 0.1198, "step": 14397 }, { "epoch": 89.42990654205607, "grad_norm": 3.171875, "learning_rate": 2.9566667852447344e-05, "loss": 0.2536, "step": 14398 }, { "epoch": 89.43613707165109, "grad_norm": 1.8125, "learning_rate": 2.9564245520984047e-05, "loss": 0.1207, "step": 14399 }, { "epoch": 89.4423676012461, "grad_norm": 1.953125, "learning_rate": 2.9561823145191874e-05, "loss": 0.1348, "step": 14400 }, { "epoch": 89.44859813084112, "grad_norm": 1.859375, "learning_rate": 2.9559400725094372e-05, "loss": 0.1434, "step": 14401 }, { "epoch": 89.45482866043614, "grad_norm": 1.4453125, "learning_rate": 2.9556978260715055e-05, "loss": 0.1254, "step": 14402 }, { "epoch": 89.46105919003115, "grad_norm": 3.359375, "learning_rate": 2.9554555752077457e-05, "loss": 0.317, "step": 14403 }, { "epoch": 89.46728971962617, "grad_norm": 2.078125, "learning_rate": 2.9552133199205095e-05, "loss": 0.1352, "step": 14404 }, { "epoch": 89.47352024922118, "grad_norm": 3.71875, "learning_rate": 2.9549710602121516e-05, "loss": 0.1969, "step": 14405 }, { "epoch": 89.4797507788162, "grad_norm": 1.46875, "learning_rate": 2.954728796085023e-05, "loss": 0.1301, "step": 14406 }, { "epoch": 89.48598130841121, "grad_norm": 1.515625, "learning_rate": 2.9544865275414775e-05, "loss": 0.1316, "step": 14407 }, { "epoch": 89.49221183800623, "grad_norm": 2.296875, "learning_rate": 2.9542442545838682e-05, "loss": 0.3538, "step": 14408 }, { "epoch": 89.49844236760124, "grad_norm": 3.8125, "learning_rate": 2.9540019772145484e-05, "loss": 0.347, "step": 14409 }, { "epoch": 89.50467289719626, "grad_norm": 2.671875, "learning_rate": 2.95375969543587e-05, "loss": 0.1257, "step": 14410 }, { "epoch": 89.51090342679127, "grad_norm": 1.9296875, "learning_rate": 2.9535174092501866e-05, "loss": 0.1643, "step": 14411 }, { "epoch": 89.51713395638629, "grad_norm": 2.765625, "learning_rate": 2.953275118659852e-05, "loss": 0.1394, "step": 14412 }, { "epoch": 89.5233644859813, "grad_norm": 2.40625, "learning_rate": 2.9530328236672184e-05, "loss": 0.155, "step": 14413 }, { "epoch": 89.52959501557632, "grad_norm": 1.734375, "learning_rate": 2.9527905242746394e-05, "loss": 0.1201, "step": 14414 }, { "epoch": 89.53582554517133, "grad_norm": 2.578125, "learning_rate": 2.952548220484469e-05, "loss": 0.1963, "step": 14415 }, { "epoch": 89.54205607476635, "grad_norm": 2.40625, "learning_rate": 2.9523059122990593e-05, "loss": 0.097, "step": 14416 }, { "epoch": 89.54828660436137, "grad_norm": 3.203125, "learning_rate": 2.952063599720764e-05, "loss": 0.1877, "step": 14417 }, { "epoch": 89.55451713395638, "grad_norm": 2.40625, "learning_rate": 2.9518212827519376e-05, "loss": 0.1516, "step": 14418 }, { "epoch": 89.5607476635514, "grad_norm": 4.15625, "learning_rate": 2.9515789613949314e-05, "loss": 0.2125, "step": 14419 }, { "epoch": 89.56697819314641, "grad_norm": 3.390625, "learning_rate": 2.9513366356521004e-05, "loss": 0.279, "step": 14420 }, { "epoch": 89.57320872274143, "grad_norm": 3.25, "learning_rate": 2.9510943055257976e-05, "loss": 0.2677, "step": 14421 }, { "epoch": 89.57943925233644, "grad_norm": 2.546875, "learning_rate": 2.9508519710183772e-05, "loss": 0.2491, "step": 14422 }, { "epoch": 89.58566978193146, "grad_norm": 1.125, "learning_rate": 2.9506096321321926e-05, "loss": 0.1164, "step": 14423 }, { "epoch": 89.59190031152647, "grad_norm": 1.984375, "learning_rate": 2.9503672888695967e-05, "loss": 0.1257, "step": 14424 }, { "epoch": 89.59813084112149, "grad_norm": 2.84375, "learning_rate": 2.950124941232944e-05, "loss": 0.1133, "step": 14425 }, { "epoch": 89.60436137071652, "grad_norm": 2.765625, "learning_rate": 2.9498825892245873e-05, "loss": 0.2145, "step": 14426 }, { "epoch": 89.61059190031153, "grad_norm": 1.2734375, "learning_rate": 2.9496402328468813e-05, "loss": 0.1259, "step": 14427 }, { "epoch": 89.61682242990655, "grad_norm": 1.296875, "learning_rate": 2.9493978721021793e-05, "loss": 0.106, "step": 14428 }, { "epoch": 89.62305295950156, "grad_norm": 1.2265625, "learning_rate": 2.9491555069928358e-05, "loss": 0.1245, "step": 14429 }, { "epoch": 89.62928348909658, "grad_norm": 3.625, "learning_rate": 2.948913137521203e-05, "loss": 0.2902, "step": 14430 }, { "epoch": 89.6355140186916, "grad_norm": 1.4296875, "learning_rate": 2.948670763689638e-05, "loss": 0.1157, "step": 14431 }, { "epoch": 89.64174454828661, "grad_norm": 3.265625, "learning_rate": 2.948428385500491e-05, "loss": 0.1917, "step": 14432 }, { "epoch": 89.64797507788163, "grad_norm": 2.265625, "learning_rate": 2.9481860029561193e-05, "loss": 0.1036, "step": 14433 }, { "epoch": 89.65420560747664, "grad_norm": 3.0625, "learning_rate": 2.9479436160588746e-05, "loss": 0.2264, "step": 14434 }, { "epoch": 89.66043613707166, "grad_norm": 3.078125, "learning_rate": 2.947701224811113e-05, "loss": 0.2195, "step": 14435 }, { "epoch": 89.66666666666667, "grad_norm": 2.078125, "learning_rate": 2.9474588292151866e-05, "loss": 0.1971, "step": 14436 }, { "epoch": 89.67289719626169, "grad_norm": 2.484375, "learning_rate": 2.9472164292734512e-05, "loss": 0.1781, "step": 14437 }, { "epoch": 89.6791277258567, "grad_norm": 2.9375, "learning_rate": 2.9469740249882603e-05, "loss": 0.2539, "step": 14438 }, { "epoch": 89.68535825545172, "grad_norm": 2.390625, "learning_rate": 2.946731616361968e-05, "loss": 0.1598, "step": 14439 }, { "epoch": 89.69158878504673, "grad_norm": 2.265625, "learning_rate": 2.9464892033969292e-05, "loss": 0.1467, "step": 14440 }, { "epoch": 89.69781931464175, "grad_norm": 2.828125, "learning_rate": 2.9462467860954978e-05, "loss": 0.2108, "step": 14441 }, { "epoch": 89.70404984423676, "grad_norm": 1.671875, "learning_rate": 2.946004364460029e-05, "loss": 0.1372, "step": 14442 }, { "epoch": 89.71028037383178, "grad_norm": 1.4453125, "learning_rate": 2.9457619384928752e-05, "loss": 0.1309, "step": 14443 }, { "epoch": 89.7165109034268, "grad_norm": 2.171875, "learning_rate": 2.9455195081963945e-05, "loss": 0.1854, "step": 14444 }, { "epoch": 89.72274143302181, "grad_norm": 3.171875, "learning_rate": 2.9452770735729374e-05, "loss": 0.1208, "step": 14445 }, { "epoch": 89.72897196261682, "grad_norm": 2.0, "learning_rate": 2.9450346346248614e-05, "loss": 0.1218, "step": 14446 }, { "epoch": 89.73520249221184, "grad_norm": 1.2265625, "learning_rate": 2.9447921913545197e-05, "loss": 0.0903, "step": 14447 }, { "epoch": 89.74143302180686, "grad_norm": 2.15625, "learning_rate": 2.944549743764268e-05, "loss": 0.2046, "step": 14448 }, { "epoch": 89.74766355140187, "grad_norm": 1.7578125, "learning_rate": 2.9443072918564594e-05, "loss": 0.1049, "step": 14449 }, { "epoch": 89.75389408099689, "grad_norm": 1.7421875, "learning_rate": 2.9440648356334498e-05, "loss": 0.1395, "step": 14450 }, { "epoch": 89.7601246105919, "grad_norm": 3.984375, "learning_rate": 2.943822375097594e-05, "loss": 0.1925, "step": 14451 }, { "epoch": 89.76635514018692, "grad_norm": 2.53125, "learning_rate": 2.943579910251246e-05, "loss": 0.1268, "step": 14452 }, { "epoch": 89.77258566978193, "grad_norm": 2.96875, "learning_rate": 2.9433374410967624e-05, "loss": 0.1824, "step": 14453 }, { "epoch": 89.77881619937695, "grad_norm": 2.4375, "learning_rate": 2.9430949676364953e-05, "loss": 0.2031, "step": 14454 }, { "epoch": 89.78504672897196, "grad_norm": 1.78125, "learning_rate": 2.9428524898728026e-05, "loss": 0.1064, "step": 14455 }, { "epoch": 89.79127725856698, "grad_norm": 1.8359375, "learning_rate": 2.9426100078080377e-05, "loss": 0.1338, "step": 14456 }, { "epoch": 89.797507788162, "grad_norm": 3.125, "learning_rate": 2.9423675214445563e-05, "loss": 0.1283, "step": 14457 }, { "epoch": 89.80373831775701, "grad_norm": 2.078125, "learning_rate": 2.9421250307847127e-05, "loss": 0.148, "step": 14458 }, { "epoch": 89.80996884735202, "grad_norm": 1.6953125, "learning_rate": 2.941882535830862e-05, "loss": 0.1554, "step": 14459 }, { "epoch": 89.81619937694704, "grad_norm": 2.21875, "learning_rate": 2.9416400365853603e-05, "loss": 0.1284, "step": 14460 }, { "epoch": 89.82242990654206, "grad_norm": 2.09375, "learning_rate": 2.9413975330505617e-05, "loss": 0.1929, "step": 14461 }, { "epoch": 89.82866043613707, "grad_norm": 1.6171875, "learning_rate": 2.941155025228823e-05, "loss": 0.1366, "step": 14462 }, { "epoch": 89.83489096573209, "grad_norm": 2.203125, "learning_rate": 2.9409125131224974e-05, "loss": 0.109, "step": 14463 }, { "epoch": 89.8411214953271, "grad_norm": 2.578125, "learning_rate": 2.9406699967339423e-05, "loss": 0.1027, "step": 14464 }, { "epoch": 89.84735202492212, "grad_norm": 2.609375, "learning_rate": 2.940427476065511e-05, "loss": 0.1581, "step": 14465 }, { "epoch": 89.85358255451713, "grad_norm": 2.09375, "learning_rate": 2.9401849511195616e-05, "loss": 0.1242, "step": 14466 }, { "epoch": 89.85981308411215, "grad_norm": 2.578125, "learning_rate": 2.9399424218984465e-05, "loss": 0.24, "step": 14467 }, { "epoch": 89.86604361370716, "grad_norm": 3.15625, "learning_rate": 2.9396998884045235e-05, "loss": 0.2273, "step": 14468 }, { "epoch": 89.87227414330218, "grad_norm": 2.234375, "learning_rate": 2.9394573506401467e-05, "loss": 0.1858, "step": 14469 }, { "epoch": 89.8785046728972, "grad_norm": 3.015625, "learning_rate": 2.9392148086076732e-05, "loss": 0.2849, "step": 14470 }, { "epoch": 89.88473520249221, "grad_norm": 1.6875, "learning_rate": 2.9389722623094567e-05, "loss": 0.1491, "step": 14471 }, { "epoch": 89.89096573208722, "grad_norm": 3.5625, "learning_rate": 2.9387297117478548e-05, "loss": 0.3916, "step": 14472 }, { "epoch": 89.89719626168224, "grad_norm": 1.8359375, "learning_rate": 2.9384871569252214e-05, "loss": 0.1548, "step": 14473 }, { "epoch": 89.90342679127725, "grad_norm": 2.609375, "learning_rate": 2.9382445978439126e-05, "loss": 0.1779, "step": 14474 }, { "epoch": 89.90965732087227, "grad_norm": 1.6171875, "learning_rate": 2.9380020345062864e-05, "loss": 0.1498, "step": 14475 }, { "epoch": 89.91588785046729, "grad_norm": 1.40625, "learning_rate": 2.9377594669146953e-05, "loss": 0.1147, "step": 14476 }, { "epoch": 89.9221183800623, "grad_norm": 3.375, "learning_rate": 2.937516895071497e-05, "loss": 0.2675, "step": 14477 }, { "epoch": 89.92834890965732, "grad_norm": 2.109375, "learning_rate": 2.9372743189790476e-05, "loss": 0.1781, "step": 14478 }, { "epoch": 89.93457943925233, "grad_norm": 2.296875, "learning_rate": 2.937031738639703e-05, "loss": 0.1824, "step": 14479 }, { "epoch": 89.94080996884735, "grad_norm": 1.0546875, "learning_rate": 2.9367891540558178e-05, "loss": 0.1041, "step": 14480 }, { "epoch": 89.94704049844236, "grad_norm": 2.515625, "learning_rate": 2.9365465652297497e-05, "loss": 0.228, "step": 14481 }, { "epoch": 89.95327102803738, "grad_norm": 1.703125, "learning_rate": 2.9363039721638542e-05, "loss": 0.1103, "step": 14482 }, { "epoch": 89.9595015576324, "grad_norm": 3.578125, "learning_rate": 2.9360613748604866e-05, "loss": 0.2997, "step": 14483 }, { "epoch": 89.96573208722741, "grad_norm": 2.40625, "learning_rate": 2.9358187733220045e-05, "loss": 0.1845, "step": 14484 }, { "epoch": 89.97196261682242, "grad_norm": 1.328125, "learning_rate": 2.935576167550763e-05, "loss": 0.1065, "step": 14485 }, { "epoch": 89.97819314641744, "grad_norm": 2.09375, "learning_rate": 2.9353335575491187e-05, "loss": 0.1982, "step": 14486 }, { "epoch": 89.98442367601245, "grad_norm": 2.65625, "learning_rate": 2.9350909433194274e-05, "loss": 0.121, "step": 14487 }, { "epoch": 89.99065420560747, "grad_norm": 4.4375, "learning_rate": 2.934848324864047e-05, "loss": 0.1763, "step": 14488 }, { "epoch": 89.99688473520249, "grad_norm": 1.7578125, "learning_rate": 2.9346057021853317e-05, "loss": 0.1077, "step": 14489 }, { "epoch": 90.0, "grad_norm": 1.40625, "learning_rate": 2.9343630752856392e-05, "loss": 0.095, "step": 14490 }, { "epoch": 90.00623052959502, "grad_norm": 2.671875, "learning_rate": 2.9341204441673266e-05, "loss": 0.1292, "step": 14491 }, { "epoch": 90.01246105919003, "grad_norm": 3.578125, "learning_rate": 2.9338778088327485e-05, "loss": 0.2936, "step": 14492 }, { "epoch": 90.01869158878505, "grad_norm": 3.5625, "learning_rate": 2.9336351692842622e-05, "loss": 0.1201, "step": 14493 }, { "epoch": 90.02492211838006, "grad_norm": 2.328125, "learning_rate": 2.9333925255242244e-05, "loss": 0.1671, "step": 14494 }, { "epoch": 90.03115264797508, "grad_norm": 1.421875, "learning_rate": 2.9331498775549926e-05, "loss": 0.1073, "step": 14495 }, { "epoch": 90.03738317757009, "grad_norm": 3.90625, "learning_rate": 2.9329072253789225e-05, "loss": 0.2315, "step": 14496 }, { "epoch": 90.04361370716511, "grad_norm": 2.828125, "learning_rate": 2.9326645689983706e-05, "loss": 0.1671, "step": 14497 }, { "epoch": 90.04984423676012, "grad_norm": 3.234375, "learning_rate": 2.932421908415694e-05, "loss": 0.1257, "step": 14498 }, { "epoch": 90.05607476635514, "grad_norm": 3.546875, "learning_rate": 2.9321792436332495e-05, "loss": 0.1989, "step": 14499 }, { "epoch": 90.06230529595015, "grad_norm": 1.9609375, "learning_rate": 2.9319365746533928e-05, "loss": 0.1558, "step": 14500 }, { "epoch": 90.06853582554517, "grad_norm": 2.40625, "learning_rate": 2.9316939014784832e-05, "loss": 0.1463, "step": 14501 }, { "epoch": 90.07476635514018, "grad_norm": 1.3515625, "learning_rate": 2.931451224110875e-05, "loss": 0.1123, "step": 14502 }, { "epoch": 90.0809968847352, "grad_norm": 2.546875, "learning_rate": 2.9312085425529267e-05, "loss": 0.1406, "step": 14503 }, { "epoch": 90.08722741433021, "grad_norm": 1.71875, "learning_rate": 2.9309658568069954e-05, "loss": 0.1577, "step": 14504 }, { "epoch": 90.09345794392523, "grad_norm": 2.546875, "learning_rate": 2.930723166875437e-05, "loss": 0.1546, "step": 14505 }, { "epoch": 90.09968847352025, "grad_norm": 2.515625, "learning_rate": 2.9304804727606095e-05, "loss": 0.1061, "step": 14506 }, { "epoch": 90.10591900311526, "grad_norm": 2.671875, "learning_rate": 2.930237774464869e-05, "loss": 0.1694, "step": 14507 }, { "epoch": 90.11214953271028, "grad_norm": 2.953125, "learning_rate": 2.929995071990574e-05, "loss": 0.2151, "step": 14508 }, { "epoch": 90.11838006230529, "grad_norm": 2.0625, "learning_rate": 2.9297523653400794e-05, "loss": 0.1535, "step": 14509 }, { "epoch": 90.1246105919003, "grad_norm": 1.28125, "learning_rate": 2.9295096545157458e-05, "loss": 0.1286, "step": 14510 }, { "epoch": 90.13084112149532, "grad_norm": 2.8125, "learning_rate": 2.9292669395199268e-05, "loss": 0.1323, "step": 14511 }, { "epoch": 90.13707165109034, "grad_norm": 1.8515625, "learning_rate": 2.9290242203549828e-05, "loss": 0.1341, "step": 14512 }, { "epoch": 90.14330218068535, "grad_norm": 3.609375, "learning_rate": 2.9287814970232686e-05, "loss": 0.1394, "step": 14513 }, { "epoch": 90.14953271028037, "grad_norm": 2.90625, "learning_rate": 2.9285387695271444e-05, "loss": 0.223, "step": 14514 }, { "epoch": 90.15576323987538, "grad_norm": 4.09375, "learning_rate": 2.928296037868965e-05, "loss": 0.2255, "step": 14515 }, { "epoch": 90.1619937694704, "grad_norm": 2.984375, "learning_rate": 2.9280533020510885e-05, "loss": 0.2079, "step": 14516 }, { "epoch": 90.16822429906541, "grad_norm": 1.953125, "learning_rate": 2.9278105620758738e-05, "loss": 0.2554, "step": 14517 }, { "epoch": 90.17445482866043, "grad_norm": 1.4375, "learning_rate": 2.9275678179456766e-05, "loss": 0.1216, "step": 14518 }, { "epoch": 90.18068535825545, "grad_norm": 3.625, "learning_rate": 2.9273250696628562e-05, "loss": 0.2304, "step": 14519 }, { "epoch": 90.18691588785046, "grad_norm": 3.25, "learning_rate": 2.9270823172297683e-05, "loss": 0.1521, "step": 14520 }, { "epoch": 90.19314641744548, "grad_norm": 3.40625, "learning_rate": 2.926839560648772e-05, "loss": 0.2564, "step": 14521 }, { "epoch": 90.19937694704049, "grad_norm": 3.875, "learning_rate": 2.9265967999222244e-05, "loss": 0.2344, "step": 14522 }, { "epoch": 90.20560747663552, "grad_norm": 1.046875, "learning_rate": 2.9263540350524844e-05, "loss": 0.1207, "step": 14523 }, { "epoch": 90.21183800623054, "grad_norm": 1.3828125, "learning_rate": 2.926111266041907e-05, "loss": 0.0865, "step": 14524 }, { "epoch": 90.21806853582555, "grad_norm": 1.0546875, "learning_rate": 2.9258684928928525e-05, "loss": 0.0888, "step": 14525 }, { "epoch": 90.22429906542057, "grad_norm": 3.25, "learning_rate": 2.925625715607679e-05, "loss": 0.1843, "step": 14526 }, { "epoch": 90.23052959501558, "grad_norm": 2.875, "learning_rate": 2.9253829341887424e-05, "loss": 0.2082, "step": 14527 }, { "epoch": 90.2367601246106, "grad_norm": 2.625, "learning_rate": 2.9251401486384016e-05, "loss": 0.1936, "step": 14528 }, { "epoch": 90.24299065420561, "grad_norm": 1.875, "learning_rate": 2.9248973589590155e-05, "loss": 0.1447, "step": 14529 }, { "epoch": 90.24922118380063, "grad_norm": 2.34375, "learning_rate": 2.924654565152941e-05, "loss": 0.1508, "step": 14530 }, { "epoch": 90.25545171339564, "grad_norm": 2.90625, "learning_rate": 2.9244117672225364e-05, "loss": 0.2187, "step": 14531 }, { "epoch": 90.26168224299066, "grad_norm": 1.75, "learning_rate": 2.9241689651701598e-05, "loss": 0.1151, "step": 14532 }, { "epoch": 90.26791277258567, "grad_norm": 1.21875, "learning_rate": 2.9239261589981697e-05, "loss": 0.1034, "step": 14533 }, { "epoch": 90.27414330218069, "grad_norm": 2.078125, "learning_rate": 2.923683348708924e-05, "loss": 0.1723, "step": 14534 }, { "epoch": 90.2803738317757, "grad_norm": 3.34375, "learning_rate": 2.9234405343047804e-05, "loss": 0.1013, "step": 14535 }, { "epoch": 90.28660436137072, "grad_norm": 1.984375, "learning_rate": 2.9231977157880986e-05, "loss": 0.1196, "step": 14536 }, { "epoch": 90.29283489096574, "grad_norm": 1.0390625, "learning_rate": 2.9229548931612348e-05, "loss": 0.099, "step": 14537 }, { "epoch": 90.29906542056075, "grad_norm": 2.984375, "learning_rate": 2.9227120664265495e-05, "loss": 0.2095, "step": 14538 }, { "epoch": 90.30529595015577, "grad_norm": 2.0, "learning_rate": 2.9224692355863996e-05, "loss": 0.1846, "step": 14539 }, { "epoch": 90.31152647975078, "grad_norm": 2.203125, "learning_rate": 2.9222264006431445e-05, "loss": 0.1553, "step": 14540 }, { "epoch": 90.3177570093458, "grad_norm": 4.625, "learning_rate": 2.9219835615991416e-05, "loss": 0.2749, "step": 14541 }, { "epoch": 90.32398753894081, "grad_norm": 4.0625, "learning_rate": 2.921740718456751e-05, "loss": 0.1503, "step": 14542 }, { "epoch": 90.33021806853583, "grad_norm": 1.40625, "learning_rate": 2.921497871218329e-05, "loss": 0.1045, "step": 14543 }, { "epoch": 90.33644859813084, "grad_norm": 2.40625, "learning_rate": 2.921255019886236e-05, "loss": 0.1361, "step": 14544 }, { "epoch": 90.34267912772586, "grad_norm": 3.0625, "learning_rate": 2.9210121644628303e-05, "loss": 0.1786, "step": 14545 }, { "epoch": 90.34890965732087, "grad_norm": 3.21875, "learning_rate": 2.9207693049504697e-05, "loss": 0.1859, "step": 14546 }, { "epoch": 90.35514018691589, "grad_norm": 3.53125, "learning_rate": 2.9205264413515137e-05, "loss": 0.2132, "step": 14547 }, { "epoch": 90.3613707165109, "grad_norm": 3.59375, "learning_rate": 2.920283573668321e-05, "loss": 0.1189, "step": 14548 }, { "epoch": 90.36760124610592, "grad_norm": 3.578125, "learning_rate": 2.9200407019032507e-05, "loss": 0.1348, "step": 14549 }, { "epoch": 90.37383177570094, "grad_norm": 2.171875, "learning_rate": 2.91979782605866e-05, "loss": 0.1429, "step": 14550 }, { "epoch": 90.38006230529595, "grad_norm": 2.5, "learning_rate": 2.9195549461369098e-05, "loss": 0.2539, "step": 14551 }, { "epoch": 90.38629283489097, "grad_norm": 1.890625, "learning_rate": 2.9193120621403575e-05, "loss": 0.1499, "step": 14552 }, { "epoch": 90.39252336448598, "grad_norm": 2.578125, "learning_rate": 2.9190691740713633e-05, "loss": 0.1357, "step": 14553 }, { "epoch": 90.398753894081, "grad_norm": 3.546875, "learning_rate": 2.9188262819322853e-05, "loss": 0.2048, "step": 14554 }, { "epoch": 90.40498442367601, "grad_norm": 3.03125, "learning_rate": 2.9185833857254824e-05, "loss": 0.1289, "step": 14555 }, { "epoch": 90.41121495327103, "grad_norm": 2.625, "learning_rate": 2.9183404854533143e-05, "loss": 0.0923, "step": 14556 }, { "epoch": 90.41744548286604, "grad_norm": 3.265625, "learning_rate": 2.9180975811181392e-05, "loss": 0.3195, "step": 14557 }, { "epoch": 90.42367601246106, "grad_norm": 1.828125, "learning_rate": 2.9178546727223177e-05, "loss": 0.1136, "step": 14558 }, { "epoch": 90.42990654205607, "grad_norm": 3.890625, "learning_rate": 2.9176117602682075e-05, "loss": 0.1385, "step": 14559 }, { "epoch": 90.43613707165109, "grad_norm": 3.234375, "learning_rate": 2.917368843758168e-05, "loss": 0.1112, "step": 14560 }, { "epoch": 90.4423676012461, "grad_norm": 3.3125, "learning_rate": 2.9171259231945598e-05, "loss": 0.108, "step": 14561 }, { "epoch": 90.44859813084112, "grad_norm": 4.46875, "learning_rate": 2.916882998579741e-05, "loss": 0.2159, "step": 14562 }, { "epoch": 90.45482866043614, "grad_norm": 2.75, "learning_rate": 2.9166400699160706e-05, "loss": 0.1922, "step": 14563 }, { "epoch": 90.46105919003115, "grad_norm": 2.0625, "learning_rate": 2.916397137205909e-05, "loss": 0.1519, "step": 14564 }, { "epoch": 90.46728971962617, "grad_norm": 2.671875, "learning_rate": 2.9161542004516153e-05, "loss": 0.2619, "step": 14565 }, { "epoch": 90.47352024922118, "grad_norm": 1.6953125, "learning_rate": 2.915911259655549e-05, "loss": 0.1657, "step": 14566 }, { "epoch": 90.4797507788162, "grad_norm": 4.0625, "learning_rate": 2.915668314820069e-05, "loss": 0.1557, "step": 14567 }, { "epoch": 90.48598130841121, "grad_norm": 4.3125, "learning_rate": 2.9154253659475354e-05, "loss": 0.2722, "step": 14568 }, { "epoch": 90.49221183800623, "grad_norm": 2.40625, "learning_rate": 2.9151824130403077e-05, "loss": 0.1805, "step": 14569 }, { "epoch": 90.49844236760124, "grad_norm": 2.265625, "learning_rate": 2.9149394561007447e-05, "loss": 0.1766, "step": 14570 }, { "epoch": 90.50467289719626, "grad_norm": 2.578125, "learning_rate": 2.9146964951312077e-05, "loss": 0.1537, "step": 14571 }, { "epoch": 90.51090342679127, "grad_norm": 2.140625, "learning_rate": 2.9144535301340547e-05, "loss": 0.1498, "step": 14572 }, { "epoch": 90.51713395638629, "grad_norm": 2.234375, "learning_rate": 2.9142105611116467e-05, "loss": 0.1634, "step": 14573 }, { "epoch": 90.5233644859813, "grad_norm": 2.421875, "learning_rate": 2.913967588066343e-05, "loss": 0.2434, "step": 14574 }, { "epoch": 90.52959501557632, "grad_norm": 3.84375, "learning_rate": 2.9137246110005028e-05, "loss": 0.2211, "step": 14575 }, { "epoch": 90.53582554517133, "grad_norm": 4.03125, "learning_rate": 2.9134816299164862e-05, "loss": 0.1282, "step": 14576 }, { "epoch": 90.54205607476635, "grad_norm": 3.875, "learning_rate": 2.913238644816655e-05, "loss": 0.3374, "step": 14577 }, { "epoch": 90.54828660436137, "grad_norm": 1.921875, "learning_rate": 2.9129956557033654e-05, "loss": 0.1078, "step": 14578 }, { "epoch": 90.55451713395638, "grad_norm": 1.1796875, "learning_rate": 2.9127526625789808e-05, "loss": 0.1174, "step": 14579 }, { "epoch": 90.5607476635514, "grad_norm": 2.1875, "learning_rate": 2.9125096654458593e-05, "loss": 0.1737, "step": 14580 }, { "epoch": 90.56697819314641, "grad_norm": 1.6171875, "learning_rate": 2.9122666643063617e-05, "loss": 0.1069, "step": 14581 }, { "epoch": 90.57320872274143, "grad_norm": 2.953125, "learning_rate": 2.912023659162848e-05, "loss": 0.1597, "step": 14582 }, { "epoch": 90.57943925233644, "grad_norm": 2.625, "learning_rate": 2.9117806500176774e-05, "loss": 0.141, "step": 14583 }, { "epoch": 90.58566978193146, "grad_norm": 3.125, "learning_rate": 2.911537636873212e-05, "loss": 0.2879, "step": 14584 }, { "epoch": 90.59190031152647, "grad_norm": 1.984375, "learning_rate": 2.9112946197318097e-05, "loss": 0.1725, "step": 14585 }, { "epoch": 90.59813084112149, "grad_norm": 2.296875, "learning_rate": 2.9110515985958324e-05, "loss": 0.1161, "step": 14586 }, { "epoch": 90.60436137071652, "grad_norm": 2.21875, "learning_rate": 2.9108085734676397e-05, "loss": 0.1461, "step": 14587 }, { "epoch": 90.61059190031153, "grad_norm": 3.015625, "learning_rate": 2.9105655443495923e-05, "loss": 0.189, "step": 14588 }, { "epoch": 90.61682242990655, "grad_norm": 3.0, "learning_rate": 2.9103225112440498e-05, "loss": 0.1202, "step": 14589 }, { "epoch": 90.62305295950156, "grad_norm": 3.1875, "learning_rate": 2.910079474153373e-05, "loss": 0.1205, "step": 14590 }, { "epoch": 90.62928348909658, "grad_norm": 1.1796875, "learning_rate": 2.9098364330799228e-05, "loss": 0.123, "step": 14591 }, { "epoch": 90.6355140186916, "grad_norm": 1.8671875, "learning_rate": 2.909593388026059e-05, "loss": 0.1428, "step": 14592 }, { "epoch": 90.64174454828661, "grad_norm": 3.34375, "learning_rate": 2.9093503389941423e-05, "loss": 0.2433, "step": 14593 }, { "epoch": 90.64797507788163, "grad_norm": 2.59375, "learning_rate": 2.9091072859865337e-05, "loss": 0.1635, "step": 14594 }, { "epoch": 90.65420560747664, "grad_norm": 3.421875, "learning_rate": 2.908864229005593e-05, "loss": 0.2397, "step": 14595 }, { "epoch": 90.66043613707166, "grad_norm": 1.875, "learning_rate": 2.9086211680536813e-05, "loss": 0.1263, "step": 14596 }, { "epoch": 90.66666666666667, "grad_norm": 3.0, "learning_rate": 2.9083781031331592e-05, "loss": 0.0907, "step": 14597 }, { "epoch": 90.67289719626169, "grad_norm": 4.25, "learning_rate": 2.9081350342463866e-05, "loss": 0.1736, "step": 14598 }, { "epoch": 90.6791277258567, "grad_norm": 3.015625, "learning_rate": 2.9078919613957255e-05, "loss": 0.1622, "step": 14599 }, { "epoch": 90.68535825545172, "grad_norm": 1.53125, "learning_rate": 2.9076488845835364e-05, "loss": 0.1498, "step": 14600 }, { "epoch": 90.69158878504673, "grad_norm": 1.203125, "learning_rate": 2.9074058038121794e-05, "loss": 0.0829, "step": 14601 }, { "epoch": 90.69781931464175, "grad_norm": 2.84375, "learning_rate": 2.907162719084016e-05, "loss": 0.1854, "step": 14602 }, { "epoch": 90.70404984423676, "grad_norm": 2.90625, "learning_rate": 2.9069196304014072e-05, "loss": 0.2379, "step": 14603 }, { "epoch": 90.71028037383178, "grad_norm": 2.9375, "learning_rate": 2.906676537766713e-05, "loss": 0.2061, "step": 14604 }, { "epoch": 90.7165109034268, "grad_norm": 1.984375, "learning_rate": 2.9064334411822952e-05, "loss": 0.1184, "step": 14605 }, { "epoch": 90.72274143302181, "grad_norm": 1.5078125, "learning_rate": 2.9061903406505154e-05, "loss": 0.1195, "step": 14606 }, { "epoch": 90.72897196261682, "grad_norm": 2.25, "learning_rate": 2.905947236173732e-05, "loss": 0.1189, "step": 14607 }, { "epoch": 90.73520249221184, "grad_norm": 1.4375, "learning_rate": 2.9057041277543097e-05, "loss": 0.1103, "step": 14608 }, { "epoch": 90.74143302180686, "grad_norm": 2.15625, "learning_rate": 2.9054610153946065e-05, "loss": 0.1566, "step": 14609 }, { "epoch": 90.74766355140187, "grad_norm": 1.15625, "learning_rate": 2.905217899096986e-05, "loss": 0.0982, "step": 14610 }, { "epoch": 90.75389408099689, "grad_norm": 1.3203125, "learning_rate": 2.904974778863807e-05, "loss": 0.1186, "step": 14611 }, { "epoch": 90.7601246105919, "grad_norm": 3.609375, "learning_rate": 2.9047316546974334e-05, "loss": 0.2078, "step": 14612 }, { "epoch": 90.76635514018692, "grad_norm": 2.359375, "learning_rate": 2.9044885266002236e-05, "loss": 0.1795, "step": 14613 }, { "epoch": 90.77258566978193, "grad_norm": 2.6875, "learning_rate": 2.904245394574541e-05, "loss": 0.1686, "step": 14614 }, { "epoch": 90.77881619937695, "grad_norm": 1.90625, "learning_rate": 2.9040022586227468e-05, "loss": 0.1879, "step": 14615 }, { "epoch": 90.78504672897196, "grad_norm": 2.734375, "learning_rate": 2.903759118747202e-05, "loss": 0.1754, "step": 14616 }, { "epoch": 90.79127725856698, "grad_norm": 2.203125, "learning_rate": 2.903515974950267e-05, "loss": 0.2097, "step": 14617 }, { "epoch": 90.797507788162, "grad_norm": 2.421875, "learning_rate": 2.9032728272343045e-05, "loss": 0.1727, "step": 14618 }, { "epoch": 90.80373831775701, "grad_norm": 1.515625, "learning_rate": 2.9030296756016768e-05, "loss": 0.1134, "step": 14619 }, { "epoch": 90.80996884735202, "grad_norm": 1.21875, "learning_rate": 2.902786520054743e-05, "loss": 0.126, "step": 14620 }, { "epoch": 90.81619937694704, "grad_norm": 3.25, "learning_rate": 2.902543360595866e-05, "loss": 0.296, "step": 14621 }, { "epoch": 90.82242990654206, "grad_norm": 2.75, "learning_rate": 2.902300197227408e-05, "loss": 0.252, "step": 14622 }, { "epoch": 90.82866043613707, "grad_norm": 1.9296875, "learning_rate": 2.9020570299517297e-05, "loss": 0.1548, "step": 14623 }, { "epoch": 90.83489096573209, "grad_norm": 3.125, "learning_rate": 2.901813858771193e-05, "loss": 0.2567, "step": 14624 }, { "epoch": 90.8411214953271, "grad_norm": 1.5078125, "learning_rate": 2.9015706836881607e-05, "loss": 0.117, "step": 14625 }, { "epoch": 90.84735202492212, "grad_norm": 2.53125, "learning_rate": 2.9013275047049926e-05, "loss": 0.25, "step": 14626 }, { "epoch": 90.85358255451713, "grad_norm": 1.9453125, "learning_rate": 2.9010843218240523e-05, "loss": 0.1643, "step": 14627 }, { "epoch": 90.85981308411215, "grad_norm": 1.8359375, "learning_rate": 2.9008411350477004e-05, "loss": 0.1372, "step": 14628 }, { "epoch": 90.86604361370716, "grad_norm": 3.15625, "learning_rate": 2.9005979443782993e-05, "loss": 0.3841, "step": 14629 }, { "epoch": 90.87227414330218, "grad_norm": 3.046875, "learning_rate": 2.9003547498182116e-05, "loss": 0.269, "step": 14630 }, { "epoch": 90.8785046728972, "grad_norm": 2.6875, "learning_rate": 2.9001115513697978e-05, "loss": 0.1869, "step": 14631 }, { "epoch": 90.88473520249221, "grad_norm": 2.1875, "learning_rate": 2.899868349035421e-05, "loss": 0.1633, "step": 14632 }, { "epoch": 90.89096573208722, "grad_norm": 2.671875, "learning_rate": 2.899625142817442e-05, "loss": 0.1847, "step": 14633 }, { "epoch": 90.89719626168224, "grad_norm": 2.203125, "learning_rate": 2.8993819327182246e-05, "loss": 0.193, "step": 14634 }, { "epoch": 90.90342679127725, "grad_norm": 2.171875, "learning_rate": 2.89913871874013e-05, "loss": 0.1837, "step": 14635 }, { "epoch": 90.90965732087227, "grad_norm": 1.6796875, "learning_rate": 2.89889550088552e-05, "loss": 0.1138, "step": 14636 }, { "epoch": 90.91588785046729, "grad_norm": 2.53125, "learning_rate": 2.898652279156757e-05, "loss": 0.1003, "step": 14637 }, { "epoch": 90.9221183800623, "grad_norm": 0.80078125, "learning_rate": 2.8984090535562048e-05, "loss": 0.0825, "step": 14638 }, { "epoch": 90.92834890965732, "grad_norm": 2.390625, "learning_rate": 2.8981658240862224e-05, "loss": 0.2814, "step": 14639 }, { "epoch": 90.93457943925233, "grad_norm": 2.234375, "learning_rate": 2.8979225907491747e-05, "loss": 0.1749, "step": 14640 }, { "epoch": 90.94080996884735, "grad_norm": 2.671875, "learning_rate": 2.8976793535474233e-05, "loss": 0.134, "step": 14641 }, { "epoch": 90.94704049844236, "grad_norm": 1.671875, "learning_rate": 2.8974361124833312e-05, "loss": 0.1698, "step": 14642 }, { "epoch": 90.95327102803738, "grad_norm": 2.171875, "learning_rate": 2.897192867559259e-05, "loss": 0.2366, "step": 14643 }, { "epoch": 90.9595015576324, "grad_norm": 1.8984375, "learning_rate": 2.896949618777571e-05, "loss": 0.1433, "step": 14644 }, { "epoch": 90.96573208722741, "grad_norm": 1.46875, "learning_rate": 2.8967063661406285e-05, "loss": 0.1139, "step": 14645 }, { "epoch": 90.97196261682242, "grad_norm": 1.453125, "learning_rate": 2.8964631096507944e-05, "loss": 0.1131, "step": 14646 }, { "epoch": 90.97819314641744, "grad_norm": 2.046875, "learning_rate": 2.896219849310432e-05, "loss": 0.1625, "step": 14647 }, { "epoch": 90.98442367601245, "grad_norm": 2.65625, "learning_rate": 2.8959765851219022e-05, "loss": 0.1703, "step": 14648 }, { "epoch": 90.99065420560747, "grad_norm": 3.40625, "learning_rate": 2.8957333170875695e-05, "loss": 0.142, "step": 14649 }, { "epoch": 90.99688473520249, "grad_norm": 3.109375, "learning_rate": 2.8954900452097953e-05, "loss": 0.2733, "step": 14650 }, { "epoch": 91.0, "grad_norm": 1.7421875, "learning_rate": 2.8952467694909436e-05, "loss": 0.1621, "step": 14651 }, { "epoch": 91.00623052959502, "grad_norm": 1.625, "learning_rate": 2.895003489933375e-05, "loss": 0.1407, "step": 14652 }, { "epoch": 91.01246105919003, "grad_norm": 1.8203125, "learning_rate": 2.894760206539454e-05, "loss": 0.1389, "step": 14653 }, { "epoch": 91.01869158878505, "grad_norm": 1.8515625, "learning_rate": 2.8945169193115435e-05, "loss": 0.1497, "step": 14654 }, { "epoch": 91.02492211838006, "grad_norm": 2.1875, "learning_rate": 2.894273628252005e-05, "loss": 0.2359, "step": 14655 }, { "epoch": 91.03115264797508, "grad_norm": 2.34375, "learning_rate": 2.8940303333632024e-05, "loss": 0.1163, "step": 14656 }, { "epoch": 91.03738317757009, "grad_norm": 3.640625, "learning_rate": 2.893787034647499e-05, "loss": 0.26, "step": 14657 }, { "epoch": 91.04361370716511, "grad_norm": 2.296875, "learning_rate": 2.8935437321072567e-05, "loss": 0.131, "step": 14658 }, { "epoch": 91.04984423676012, "grad_norm": 1.6015625, "learning_rate": 2.8933004257448387e-05, "loss": 0.1368, "step": 14659 }, { "epoch": 91.05607476635514, "grad_norm": 2.0625, "learning_rate": 2.8930571155626096e-05, "loss": 0.1085, "step": 14660 }, { "epoch": 91.06230529595015, "grad_norm": 2.53125, "learning_rate": 2.89281380156293e-05, "loss": 0.2644, "step": 14661 }, { "epoch": 91.06853582554517, "grad_norm": 1.75, "learning_rate": 2.8925704837481642e-05, "loss": 0.1473, "step": 14662 }, { "epoch": 91.07476635514018, "grad_norm": 1.921875, "learning_rate": 2.892327162120676e-05, "loss": 0.1247, "step": 14663 }, { "epoch": 91.0809968847352, "grad_norm": 2.3125, "learning_rate": 2.892083836682828e-05, "loss": 0.202, "step": 14664 }, { "epoch": 91.08722741433021, "grad_norm": 1.4609375, "learning_rate": 2.891840507436983e-05, "loss": 0.1114, "step": 14665 }, { "epoch": 91.09345794392523, "grad_norm": 1.171875, "learning_rate": 2.8915971743855054e-05, "loss": 0.1147, "step": 14666 }, { "epoch": 91.09968847352025, "grad_norm": 1.921875, "learning_rate": 2.891353837530757e-05, "loss": 0.1588, "step": 14667 }, { "epoch": 91.10591900311526, "grad_norm": 1.375, "learning_rate": 2.891110496875102e-05, "loss": 0.092, "step": 14668 }, { "epoch": 91.11214953271028, "grad_norm": 1.1796875, "learning_rate": 2.8908671524209037e-05, "loss": 0.1297, "step": 14669 }, { "epoch": 91.11838006230529, "grad_norm": 2.78125, "learning_rate": 2.890623804170525e-05, "loss": 0.1817, "step": 14670 }, { "epoch": 91.1246105919003, "grad_norm": 2.828125, "learning_rate": 2.890380452126331e-05, "loss": 0.257, "step": 14671 }, { "epoch": 91.13084112149532, "grad_norm": 2.265625, "learning_rate": 2.8901370962906826e-05, "loss": 0.1261, "step": 14672 }, { "epoch": 91.13707165109034, "grad_norm": 1.9140625, "learning_rate": 2.889893736665946e-05, "loss": 0.0936, "step": 14673 }, { "epoch": 91.14330218068535, "grad_norm": 3.234375, "learning_rate": 2.8896503732544832e-05, "loss": 0.2308, "step": 14674 }, { "epoch": 91.14953271028037, "grad_norm": 2.625, "learning_rate": 2.8894070060586576e-05, "loss": 0.2132, "step": 14675 }, { "epoch": 91.15576323987538, "grad_norm": 2.984375, "learning_rate": 2.889163635080834e-05, "loss": 0.1291, "step": 14676 }, { "epoch": 91.1619937694704, "grad_norm": 1.75, "learning_rate": 2.888920260323375e-05, "loss": 0.1556, "step": 14677 }, { "epoch": 91.16822429906541, "grad_norm": 2.140625, "learning_rate": 2.888676881788645e-05, "loss": 0.1488, "step": 14678 }, { "epoch": 91.17445482866043, "grad_norm": 2.484375, "learning_rate": 2.888433499479007e-05, "loss": 0.1781, "step": 14679 }, { "epoch": 91.18068535825545, "grad_norm": 2.375, "learning_rate": 2.8881901133968254e-05, "loss": 0.208, "step": 14680 }, { "epoch": 91.18691588785046, "grad_norm": 2.75, "learning_rate": 2.8879467235444636e-05, "loss": 0.2095, "step": 14681 }, { "epoch": 91.19314641744548, "grad_norm": 2.265625, "learning_rate": 2.887703329924286e-05, "loss": 0.2177, "step": 14682 }, { "epoch": 91.19937694704049, "grad_norm": 1.75, "learning_rate": 2.887459932538656e-05, "loss": 0.1341, "step": 14683 }, { "epoch": 91.20560747663552, "grad_norm": 2.125, "learning_rate": 2.8872165313899374e-05, "loss": 0.1461, "step": 14684 }, { "epoch": 91.21183800623054, "grad_norm": 1.875, "learning_rate": 2.886973126480495e-05, "loss": 0.1269, "step": 14685 }, { "epoch": 91.21806853582555, "grad_norm": 3.171875, "learning_rate": 2.8867297178126918e-05, "loss": 0.2639, "step": 14686 }, { "epoch": 91.22429906542057, "grad_norm": 2.515625, "learning_rate": 2.8864863053888925e-05, "loss": 0.2173, "step": 14687 }, { "epoch": 91.23052959501558, "grad_norm": 2.859375, "learning_rate": 2.8862428892114605e-05, "loss": 0.1956, "step": 14688 }, { "epoch": 91.2367601246106, "grad_norm": 1.875, "learning_rate": 2.885999469282761e-05, "loss": 0.1683, "step": 14689 }, { "epoch": 91.24299065420561, "grad_norm": 1.8515625, "learning_rate": 2.8857560456051575e-05, "loss": 0.1076, "step": 14690 }, { "epoch": 91.24922118380063, "grad_norm": 2.203125, "learning_rate": 2.885512618181014e-05, "loss": 0.1988, "step": 14691 }, { "epoch": 91.25545171339564, "grad_norm": 2.5, "learning_rate": 2.8852691870126946e-05, "loss": 0.1694, "step": 14692 }, { "epoch": 91.26168224299066, "grad_norm": 2.21875, "learning_rate": 2.8850257521025642e-05, "loss": 0.155, "step": 14693 }, { "epoch": 91.26791277258567, "grad_norm": 1.9609375, "learning_rate": 2.884782313452986e-05, "loss": 0.146, "step": 14694 }, { "epoch": 91.27414330218069, "grad_norm": 3.25, "learning_rate": 2.8845388710663262e-05, "loss": 0.1946, "step": 14695 }, { "epoch": 91.2803738317757, "grad_norm": 2.53125, "learning_rate": 2.884295424944947e-05, "loss": 0.331, "step": 14696 }, { "epoch": 91.28660436137072, "grad_norm": 1.109375, "learning_rate": 2.884051975091214e-05, "loss": 0.1016, "step": 14697 }, { "epoch": 91.29283489096574, "grad_norm": 2.671875, "learning_rate": 2.8838085215074923e-05, "loss": 0.2091, "step": 14698 }, { "epoch": 91.29906542056075, "grad_norm": 1.546875, "learning_rate": 2.883565064196145e-05, "loss": 0.0944, "step": 14699 }, { "epoch": 91.30529595015577, "grad_norm": 2.03125, "learning_rate": 2.8833216031595373e-05, "loss": 0.2725, "step": 14700 }, { "epoch": 91.31152647975078, "grad_norm": 1.9453125, "learning_rate": 2.8830781384000333e-05, "loss": 0.1141, "step": 14701 }, { "epoch": 91.3177570093458, "grad_norm": 2.671875, "learning_rate": 2.8828346699199978e-05, "loss": 0.2797, "step": 14702 }, { "epoch": 91.32398753894081, "grad_norm": 1.7890625, "learning_rate": 2.8825911977217952e-05, "loss": 0.1522, "step": 14703 }, { "epoch": 91.33021806853583, "grad_norm": 1.921875, "learning_rate": 2.8823477218077905e-05, "loss": 0.1468, "step": 14704 }, { "epoch": 91.33644859813084, "grad_norm": 1.75, "learning_rate": 2.882104242180349e-05, "loss": 0.1443, "step": 14705 }, { "epoch": 91.34267912772586, "grad_norm": 2.265625, "learning_rate": 2.881860758841834e-05, "loss": 0.1712, "step": 14706 }, { "epoch": 91.34890965732087, "grad_norm": 1.671875, "learning_rate": 2.881617271794611e-05, "loss": 0.1059, "step": 14707 }, { "epoch": 91.35514018691589, "grad_norm": 2.46875, "learning_rate": 2.8813737810410457e-05, "loss": 0.143, "step": 14708 }, { "epoch": 91.3613707165109, "grad_norm": 1.15625, "learning_rate": 2.8811302865835004e-05, "loss": 0.0904, "step": 14709 }, { "epoch": 91.36760124610592, "grad_norm": 2.859375, "learning_rate": 2.8808867884243427e-05, "loss": 0.2804, "step": 14710 }, { "epoch": 91.37383177570094, "grad_norm": 2.109375, "learning_rate": 2.880643286565936e-05, "loss": 0.1998, "step": 14711 }, { "epoch": 91.38006230529595, "grad_norm": 2.5625, "learning_rate": 2.880399781010646e-05, "loss": 0.149, "step": 14712 }, { "epoch": 91.38629283489097, "grad_norm": 2.59375, "learning_rate": 2.880156271760837e-05, "loss": 0.2374, "step": 14713 }, { "epoch": 91.39252336448598, "grad_norm": 2.125, "learning_rate": 2.8799127588188746e-05, "loss": 0.0921, "step": 14714 }, { "epoch": 91.398753894081, "grad_norm": 1.828125, "learning_rate": 2.8796692421871234e-05, "loss": 0.1286, "step": 14715 }, { "epoch": 91.40498442367601, "grad_norm": 1.8046875, "learning_rate": 2.8794257218679477e-05, "loss": 0.1026, "step": 14716 }, { "epoch": 91.41121495327103, "grad_norm": 2.828125, "learning_rate": 2.8791821978637153e-05, "loss": 0.1956, "step": 14717 }, { "epoch": 91.41744548286604, "grad_norm": 2.078125, "learning_rate": 2.878938670176788e-05, "loss": 0.2149, "step": 14718 }, { "epoch": 91.42367601246106, "grad_norm": 2.65625, "learning_rate": 2.8786951388095335e-05, "loss": 0.2215, "step": 14719 }, { "epoch": 91.42990654205607, "grad_norm": 2.375, "learning_rate": 2.878451603764316e-05, "loss": 0.1972, "step": 14720 }, { "epoch": 91.43613707165109, "grad_norm": 2.75, "learning_rate": 2.8782080650435006e-05, "loss": 0.2418, "step": 14721 }, { "epoch": 91.4423676012461, "grad_norm": 2.140625, "learning_rate": 2.8779645226494524e-05, "loss": 0.1644, "step": 14722 }, { "epoch": 91.44859813084112, "grad_norm": 1.8203125, "learning_rate": 2.8777209765845382e-05, "loss": 0.1081, "step": 14723 }, { "epoch": 91.45482866043614, "grad_norm": 2.390625, "learning_rate": 2.8774774268511216e-05, "loss": 0.0932, "step": 14724 }, { "epoch": 91.46105919003115, "grad_norm": 2.6875, "learning_rate": 2.8772338734515692e-05, "loss": 0.2051, "step": 14725 }, { "epoch": 91.46728971962617, "grad_norm": 2.5625, "learning_rate": 2.8769903163882463e-05, "loss": 0.1874, "step": 14726 }, { "epoch": 91.47352024922118, "grad_norm": 2.9375, "learning_rate": 2.8767467556635174e-05, "loss": 0.2828, "step": 14727 }, { "epoch": 91.4797507788162, "grad_norm": 2.203125, "learning_rate": 2.8765031912797498e-05, "loss": 0.1453, "step": 14728 }, { "epoch": 91.48598130841121, "grad_norm": 3.453125, "learning_rate": 2.8762596232393064e-05, "loss": 0.2762, "step": 14729 }, { "epoch": 91.49221183800623, "grad_norm": 1.421875, "learning_rate": 2.8760160515445556e-05, "loss": 0.1165, "step": 14730 }, { "epoch": 91.49844236760124, "grad_norm": 1.7578125, "learning_rate": 2.875772476197861e-05, "loss": 0.1712, "step": 14731 }, { "epoch": 91.50467289719626, "grad_norm": 1.875, "learning_rate": 2.8755288972015894e-05, "loss": 0.119, "step": 14732 }, { "epoch": 91.51090342679127, "grad_norm": 1.4921875, "learning_rate": 2.875285314558106e-05, "loss": 0.1394, "step": 14733 }, { "epoch": 91.51713395638629, "grad_norm": 1.1484375, "learning_rate": 2.8750417282697763e-05, "loss": 0.099, "step": 14734 }, { "epoch": 91.5233644859813, "grad_norm": 3.203125, "learning_rate": 2.874798138338966e-05, "loss": 0.2467, "step": 14735 }, { "epoch": 91.52959501557632, "grad_norm": 2.1875, "learning_rate": 2.8745545447680427e-05, "loss": 0.1484, "step": 14736 }, { "epoch": 91.53582554517133, "grad_norm": 1.65625, "learning_rate": 2.8743109475593693e-05, "loss": 0.1296, "step": 14737 }, { "epoch": 91.54205607476635, "grad_norm": 1.7578125, "learning_rate": 2.8740673467153138e-05, "loss": 0.1008, "step": 14738 }, { "epoch": 91.54828660436137, "grad_norm": 1.6875, "learning_rate": 2.873823742238242e-05, "loss": 0.1234, "step": 14739 }, { "epoch": 91.55451713395638, "grad_norm": 1.2265625, "learning_rate": 2.8735801341305187e-05, "loss": 0.1135, "step": 14740 }, { "epoch": 91.5607476635514, "grad_norm": 1.90625, "learning_rate": 2.873336522394511e-05, "loss": 0.1128, "step": 14741 }, { "epoch": 91.56697819314641, "grad_norm": 2.859375, "learning_rate": 2.873092907032583e-05, "loss": 0.3101, "step": 14742 }, { "epoch": 91.57320872274143, "grad_norm": 2.953125, "learning_rate": 2.8728492880471036e-05, "loss": 0.2601, "step": 14743 }, { "epoch": 91.57943925233644, "grad_norm": 2.46875, "learning_rate": 2.872605665440436e-05, "loss": 0.1378, "step": 14744 }, { "epoch": 91.58566978193146, "grad_norm": 1.8359375, "learning_rate": 2.8723620392149487e-05, "loss": 0.1001, "step": 14745 }, { "epoch": 91.59190031152647, "grad_norm": 2.078125, "learning_rate": 2.872118409373007e-05, "loss": 0.1159, "step": 14746 }, { "epoch": 91.59813084112149, "grad_norm": 1.2890625, "learning_rate": 2.871874775916976e-05, "loss": 0.1138, "step": 14747 }, { "epoch": 91.60436137071652, "grad_norm": 2.59375, "learning_rate": 2.871631138849224e-05, "loss": 0.1697, "step": 14748 }, { "epoch": 91.61059190031153, "grad_norm": 1.6796875, "learning_rate": 2.871387498172115e-05, "loss": 0.117, "step": 14749 }, { "epoch": 91.61682242990655, "grad_norm": 2.59375, "learning_rate": 2.871143853888017e-05, "loss": 0.1811, "step": 14750 }, { "epoch": 91.62305295950156, "grad_norm": 2.1875, "learning_rate": 2.870900205999295e-05, "loss": 0.2581, "step": 14751 }, { "epoch": 91.62928348909658, "grad_norm": 2.125, "learning_rate": 2.8706565545083176e-05, "loss": 0.13, "step": 14752 }, { "epoch": 91.6355140186916, "grad_norm": 1.6953125, "learning_rate": 2.8704128994174478e-05, "loss": 0.0927, "step": 14753 }, { "epoch": 91.64174454828661, "grad_norm": 1.8984375, "learning_rate": 2.870169240729055e-05, "loss": 0.1666, "step": 14754 }, { "epoch": 91.64797507788163, "grad_norm": 1.421875, "learning_rate": 2.8699255784455043e-05, "loss": 0.114, "step": 14755 }, { "epoch": 91.65420560747664, "grad_norm": 1.703125, "learning_rate": 2.8696819125691627e-05, "loss": 0.1042, "step": 14756 }, { "epoch": 91.66043613707166, "grad_norm": 1.71875, "learning_rate": 2.869438243102396e-05, "loss": 0.1319, "step": 14757 }, { "epoch": 91.66666666666667, "grad_norm": 1.578125, "learning_rate": 2.8691945700475707e-05, "loss": 0.1284, "step": 14758 }, { "epoch": 91.67289719626169, "grad_norm": 2.9375, "learning_rate": 2.8689508934070552e-05, "loss": 0.2944, "step": 14759 }, { "epoch": 91.6791277258567, "grad_norm": 1.4921875, "learning_rate": 2.8687072131832142e-05, "loss": 0.1389, "step": 14760 }, { "epoch": 91.68535825545172, "grad_norm": 1.140625, "learning_rate": 2.868463529378415e-05, "loss": 0.1165, "step": 14761 }, { "epoch": 91.69158878504673, "grad_norm": 1.6796875, "learning_rate": 2.8682198419950247e-05, "loss": 0.1034, "step": 14762 }, { "epoch": 91.69781931464175, "grad_norm": 1.6640625, "learning_rate": 2.8679761510354098e-05, "loss": 0.1133, "step": 14763 }, { "epoch": 91.70404984423676, "grad_norm": 2.15625, "learning_rate": 2.8677324565019358e-05, "loss": 0.2042, "step": 14764 }, { "epoch": 91.71028037383178, "grad_norm": 2.65625, "learning_rate": 2.867488758396972e-05, "loss": 0.1861, "step": 14765 }, { "epoch": 91.7165109034268, "grad_norm": 1.5390625, "learning_rate": 2.867245056722883e-05, "loss": 0.1338, "step": 14766 }, { "epoch": 91.72274143302181, "grad_norm": 2.203125, "learning_rate": 2.8670013514820375e-05, "loss": 0.1189, "step": 14767 }, { "epoch": 91.72897196261682, "grad_norm": 3.34375, "learning_rate": 2.866757642676801e-05, "loss": 0.3182, "step": 14768 }, { "epoch": 91.73520249221184, "grad_norm": 1.5703125, "learning_rate": 2.8665139303095413e-05, "loss": 0.1468, "step": 14769 }, { "epoch": 91.74143302180686, "grad_norm": 2.0, "learning_rate": 2.8662702143826243e-05, "loss": 0.1429, "step": 14770 }, { "epoch": 91.74766355140187, "grad_norm": 1.5625, "learning_rate": 2.866026494898419e-05, "loss": 0.0882, "step": 14771 }, { "epoch": 91.75389408099689, "grad_norm": 2.53125, "learning_rate": 2.8657827718592895e-05, "loss": 0.1949, "step": 14772 }, { "epoch": 91.7601246105919, "grad_norm": 1.640625, "learning_rate": 2.8655390452676057e-05, "loss": 0.0989, "step": 14773 }, { "epoch": 91.76635514018692, "grad_norm": 1.2421875, "learning_rate": 2.8652953151257333e-05, "loss": 0.1066, "step": 14774 }, { "epoch": 91.77258566978193, "grad_norm": 2.203125, "learning_rate": 2.86505158143604e-05, "loss": 0.1104, "step": 14775 }, { "epoch": 91.77881619937695, "grad_norm": 1.953125, "learning_rate": 2.8648078442008928e-05, "loss": 0.1433, "step": 14776 }, { "epoch": 91.78504672897196, "grad_norm": 1.3984375, "learning_rate": 2.8645641034226583e-05, "loss": 0.1394, "step": 14777 }, { "epoch": 91.79127725856698, "grad_norm": 1.203125, "learning_rate": 2.864320359103706e-05, "loss": 0.1126, "step": 14778 }, { "epoch": 91.797507788162, "grad_norm": 2.671875, "learning_rate": 2.8640766112463996e-05, "loss": 0.232, "step": 14779 }, { "epoch": 91.80373831775701, "grad_norm": 2.328125, "learning_rate": 2.8638328598531095e-05, "loss": 0.1358, "step": 14780 }, { "epoch": 91.80996884735202, "grad_norm": 2.25, "learning_rate": 2.8635891049262013e-05, "loss": 0.1743, "step": 14781 }, { "epoch": 91.81619937694704, "grad_norm": 4.0625, "learning_rate": 2.8633453464680436e-05, "loss": 0.5072, "step": 14782 }, { "epoch": 91.82242990654206, "grad_norm": 2.296875, "learning_rate": 2.8631015844810026e-05, "loss": 0.1577, "step": 14783 }, { "epoch": 91.82866043613707, "grad_norm": 1.640625, "learning_rate": 2.862857818967447e-05, "loss": 0.1277, "step": 14784 }, { "epoch": 91.83489096573209, "grad_norm": 1.375, "learning_rate": 2.862614049929743e-05, "loss": 0.1416, "step": 14785 }, { "epoch": 91.8411214953271, "grad_norm": 2.671875, "learning_rate": 2.86237027737026e-05, "loss": 0.1819, "step": 14786 }, { "epoch": 91.84735202492212, "grad_norm": 1.8203125, "learning_rate": 2.8621265012913638e-05, "loss": 0.1724, "step": 14787 }, { "epoch": 91.85358255451713, "grad_norm": 1.6171875, "learning_rate": 2.861882721695423e-05, "loss": 0.1327, "step": 14788 }, { "epoch": 91.85981308411215, "grad_norm": 2.3125, "learning_rate": 2.8616389385848047e-05, "loss": 0.1921, "step": 14789 }, { "epoch": 91.86604361370716, "grad_norm": 2.3125, "learning_rate": 2.8613951519618765e-05, "loss": 0.2016, "step": 14790 }, { "epoch": 91.87227414330218, "grad_norm": 2.71875, "learning_rate": 2.8611513618290064e-05, "loss": 0.191, "step": 14791 }, { "epoch": 91.8785046728972, "grad_norm": 2.75, "learning_rate": 2.8609075681885617e-05, "loss": 0.2271, "step": 14792 }, { "epoch": 91.88473520249221, "grad_norm": 2.53125, "learning_rate": 2.8606637710429107e-05, "loss": 0.253, "step": 14793 }, { "epoch": 91.89096573208722, "grad_norm": 2.828125, "learning_rate": 2.8604199703944217e-05, "loss": 0.2049, "step": 14794 }, { "epoch": 91.89719626168224, "grad_norm": 2.625, "learning_rate": 2.8601761662454617e-05, "loss": 0.2329, "step": 14795 }, { "epoch": 91.90342679127725, "grad_norm": 4.15625, "learning_rate": 2.8599323585983985e-05, "loss": 0.2139, "step": 14796 }, { "epoch": 91.90965732087227, "grad_norm": 2.65625, "learning_rate": 2.8596885474556005e-05, "loss": 0.141, "step": 14797 }, { "epoch": 91.91588785046729, "grad_norm": 2.0, "learning_rate": 2.859444732819435e-05, "loss": 0.1368, "step": 14798 }, { "epoch": 91.9221183800623, "grad_norm": 2.609375, "learning_rate": 2.8592009146922704e-05, "loss": 0.15, "step": 14799 }, { "epoch": 91.92834890965732, "grad_norm": 1.609375, "learning_rate": 2.858957093076476e-05, "loss": 0.1441, "step": 14800 }, { "epoch": 91.93457943925233, "grad_norm": 3.84375, "learning_rate": 2.858713267974417e-05, "loss": 0.1297, "step": 14801 }, { "epoch": 91.94080996884735, "grad_norm": 2.1875, "learning_rate": 2.8584694393884637e-05, "loss": 0.1746, "step": 14802 }, { "epoch": 91.94704049844236, "grad_norm": 1.65625, "learning_rate": 2.8582256073209836e-05, "loss": 0.1228, "step": 14803 }, { "epoch": 91.95327102803738, "grad_norm": 1.6015625, "learning_rate": 2.8579817717743444e-05, "loss": 0.0931, "step": 14804 }, { "epoch": 91.9595015576324, "grad_norm": 3.359375, "learning_rate": 2.8577379327509146e-05, "loss": 0.2222, "step": 14805 }, { "epoch": 91.96573208722741, "grad_norm": 3.0625, "learning_rate": 2.857494090253064e-05, "loss": 0.1421, "step": 14806 }, { "epoch": 91.97196261682242, "grad_norm": 3.65625, "learning_rate": 2.8572502442831573e-05, "loss": 0.2849, "step": 14807 }, { "epoch": 91.97819314641744, "grad_norm": 1.546875, "learning_rate": 2.857006394843566e-05, "loss": 0.1149, "step": 14808 }, { "epoch": 91.98442367601245, "grad_norm": 2.75, "learning_rate": 2.8567625419366567e-05, "loss": 0.1176, "step": 14809 }, { "epoch": 91.99065420560747, "grad_norm": 2.734375, "learning_rate": 2.8565186855647984e-05, "loss": 0.3225, "step": 14810 }, { "epoch": 91.99688473520249, "grad_norm": 2.21875, "learning_rate": 2.8562748257303596e-05, "loss": 0.1392, "step": 14811 }, { "epoch": 92.0, "grad_norm": 2.828125, "learning_rate": 2.856030962435708e-05, "loss": 0.1057, "step": 14812 }, { "epoch": 92.00623052959502, "grad_norm": 2.828125, "learning_rate": 2.8557870956832132e-05, "loss": 0.1123, "step": 14813 }, { "epoch": 92.01246105919003, "grad_norm": 2.265625, "learning_rate": 2.8555432254752425e-05, "loss": 0.1181, "step": 14814 }, { "epoch": 92.01869158878505, "grad_norm": 3.234375, "learning_rate": 2.855299351814165e-05, "loss": 0.3431, "step": 14815 }, { "epoch": 92.02492211838006, "grad_norm": 2.109375, "learning_rate": 2.8550554747023494e-05, "loss": 0.1258, "step": 14816 }, { "epoch": 92.03115264797508, "grad_norm": 2.546875, "learning_rate": 2.8548115941421636e-05, "loss": 0.1704, "step": 14817 }, { "epoch": 92.03738317757009, "grad_norm": 2.953125, "learning_rate": 2.8545677101359768e-05, "loss": 0.2312, "step": 14818 }, { "epoch": 92.04361370716511, "grad_norm": 1.7265625, "learning_rate": 2.8543238226861585e-05, "loss": 0.1174, "step": 14819 }, { "epoch": 92.04984423676012, "grad_norm": 1.96875, "learning_rate": 2.854079931795075e-05, "loss": 0.1461, "step": 14820 }, { "epoch": 92.05607476635514, "grad_norm": 2.5625, "learning_rate": 2.853836037465097e-05, "loss": 0.187, "step": 14821 }, { "epoch": 92.06230529595015, "grad_norm": 1.484375, "learning_rate": 2.853592139698593e-05, "loss": 0.129, "step": 14822 }, { "epoch": 92.06853582554517, "grad_norm": 1.7421875, "learning_rate": 2.8533482384979308e-05, "loss": 0.1369, "step": 14823 }, { "epoch": 92.07476635514018, "grad_norm": 2.453125, "learning_rate": 2.8531043338654807e-05, "loss": 0.1724, "step": 14824 }, { "epoch": 92.0809968847352, "grad_norm": 1.484375, "learning_rate": 2.8528604258036096e-05, "loss": 0.1149, "step": 14825 }, { "epoch": 92.08722741433021, "grad_norm": 1.8828125, "learning_rate": 2.8526165143146887e-05, "loss": 0.1322, "step": 14826 }, { "epoch": 92.09345794392523, "grad_norm": 1.296875, "learning_rate": 2.8523725994010847e-05, "loss": 0.1101, "step": 14827 }, { "epoch": 92.09968847352025, "grad_norm": 2.875, "learning_rate": 2.8521286810651682e-05, "loss": 0.2346, "step": 14828 }, { "epoch": 92.10591900311526, "grad_norm": 1.7421875, "learning_rate": 2.851884759309308e-05, "loss": 0.0966, "step": 14829 }, { "epoch": 92.11214953271028, "grad_norm": 3.015625, "learning_rate": 2.851640834135872e-05, "loss": 0.1282, "step": 14830 }, { "epoch": 92.11838006230529, "grad_norm": 2.6875, "learning_rate": 2.8513969055472296e-05, "loss": 0.2917, "step": 14831 }, { "epoch": 92.1246105919003, "grad_norm": 2.9375, "learning_rate": 2.851152973545751e-05, "loss": 0.2873, "step": 14832 }, { "epoch": 92.13084112149532, "grad_norm": 2.171875, "learning_rate": 2.8509090381338038e-05, "loss": 0.156, "step": 14833 }, { "epoch": 92.13707165109034, "grad_norm": 1.546875, "learning_rate": 2.8506650993137585e-05, "loss": 0.136, "step": 14834 }, { "epoch": 92.14330218068535, "grad_norm": 3.28125, "learning_rate": 2.8504211570879834e-05, "loss": 0.3025, "step": 14835 }, { "epoch": 92.14953271028037, "grad_norm": 1.734375, "learning_rate": 2.8501772114588476e-05, "loss": 0.1052, "step": 14836 }, { "epoch": 92.15576323987538, "grad_norm": 1.8046875, "learning_rate": 2.849933262428721e-05, "loss": 0.1486, "step": 14837 }, { "epoch": 92.1619937694704, "grad_norm": 2.5, "learning_rate": 2.8496893099999733e-05, "loss": 0.1658, "step": 14838 }, { "epoch": 92.16822429906541, "grad_norm": 2.4375, "learning_rate": 2.8494453541749728e-05, "loss": 0.2383, "step": 14839 }, { "epoch": 92.17445482866043, "grad_norm": 1.8359375, "learning_rate": 2.849201394956088e-05, "loss": 0.1355, "step": 14840 }, { "epoch": 92.18068535825545, "grad_norm": 2.453125, "learning_rate": 2.8489574323456913e-05, "loss": 0.184, "step": 14841 }, { "epoch": 92.18691588785046, "grad_norm": 2.015625, "learning_rate": 2.8487134663461483e-05, "loss": 0.1461, "step": 14842 }, { "epoch": 92.19314641744548, "grad_norm": 2.125, "learning_rate": 2.848469496959832e-05, "loss": 0.1728, "step": 14843 }, { "epoch": 92.19937694704049, "grad_norm": 1.875, "learning_rate": 2.84822552418911e-05, "loss": 0.1267, "step": 14844 }, { "epoch": 92.20560747663552, "grad_norm": 1.5234375, "learning_rate": 2.8479815480363513e-05, "loss": 0.1051, "step": 14845 }, { "epoch": 92.21183800623054, "grad_norm": 1.609375, "learning_rate": 2.847737568503927e-05, "loss": 0.1351, "step": 14846 }, { "epoch": 92.21806853582555, "grad_norm": 1.3984375, "learning_rate": 2.8474935855942053e-05, "loss": 0.109, "step": 14847 }, { "epoch": 92.22429906542057, "grad_norm": 2.40625, "learning_rate": 2.847249599309557e-05, "loss": 0.1314, "step": 14848 }, { "epoch": 92.23052959501558, "grad_norm": 1.9765625, "learning_rate": 2.8470056096523507e-05, "loss": 0.1202, "step": 14849 }, { "epoch": 92.2367601246106, "grad_norm": 0.94921875, "learning_rate": 2.846761616624957e-05, "loss": 0.0899, "step": 14850 }, { "epoch": 92.24299065420561, "grad_norm": 1.1875, "learning_rate": 2.846517620229745e-05, "loss": 0.0913, "step": 14851 }, { "epoch": 92.24922118380063, "grad_norm": 1.5234375, "learning_rate": 2.8462736204690848e-05, "loss": 0.1171, "step": 14852 }, { "epoch": 92.25545171339564, "grad_norm": 2.046875, "learning_rate": 2.8460296173453454e-05, "loss": 0.1203, "step": 14853 }, { "epoch": 92.26168224299066, "grad_norm": 1.8203125, "learning_rate": 2.8457856108608984e-05, "loss": 0.1291, "step": 14854 }, { "epoch": 92.26791277258567, "grad_norm": 1.953125, "learning_rate": 2.8455416010181108e-05, "loss": 0.1692, "step": 14855 }, { "epoch": 92.27414330218069, "grad_norm": 2.109375, "learning_rate": 2.845297587819355e-05, "loss": 0.1797, "step": 14856 }, { "epoch": 92.2803738317757, "grad_norm": 2.859375, "learning_rate": 2.8450535712670002e-05, "loss": 0.2873, "step": 14857 }, { "epoch": 92.28660436137072, "grad_norm": 1.921875, "learning_rate": 2.8448095513634155e-05, "loss": 0.1433, "step": 14858 }, { "epoch": 92.29283489096574, "grad_norm": 3.046875, "learning_rate": 2.8445655281109718e-05, "loss": 0.1199, "step": 14859 }, { "epoch": 92.29906542056075, "grad_norm": 1.6015625, "learning_rate": 2.8443215015120385e-05, "loss": 0.1326, "step": 14860 }, { "epoch": 92.30529595015577, "grad_norm": 4.09375, "learning_rate": 2.8440774715689873e-05, "loss": 0.1562, "step": 14861 }, { "epoch": 92.31152647975078, "grad_norm": 2.0625, "learning_rate": 2.843833438284185e-05, "loss": 0.127, "step": 14862 }, { "epoch": 92.3177570093458, "grad_norm": 2.34375, "learning_rate": 2.843589401660005e-05, "loss": 0.1064, "step": 14863 }, { "epoch": 92.32398753894081, "grad_norm": 3.359375, "learning_rate": 2.8433453616988153e-05, "loss": 0.1958, "step": 14864 }, { "epoch": 92.33021806853583, "grad_norm": 2.953125, "learning_rate": 2.8431013184029876e-05, "loss": 0.1379, "step": 14865 }, { "epoch": 92.33644859813084, "grad_norm": 2.484375, "learning_rate": 2.8428572717748903e-05, "loss": 0.1794, "step": 14866 }, { "epoch": 92.34267912772586, "grad_norm": 1.859375, "learning_rate": 2.8426132218168955e-05, "loss": 0.1131, "step": 14867 }, { "epoch": 92.34890965732087, "grad_norm": 2.578125, "learning_rate": 2.842369168531372e-05, "loss": 0.2497, "step": 14868 }, { "epoch": 92.35514018691589, "grad_norm": 2.875, "learning_rate": 2.842125111920691e-05, "loss": 0.1338, "step": 14869 }, { "epoch": 92.3613707165109, "grad_norm": 2.59375, "learning_rate": 2.8418810519872225e-05, "loss": 0.1633, "step": 14870 }, { "epoch": 92.36760124610592, "grad_norm": 2.796875, "learning_rate": 2.8416369887333372e-05, "loss": 0.2362, "step": 14871 }, { "epoch": 92.37383177570094, "grad_norm": 4.25, "learning_rate": 2.8413929221614043e-05, "loss": 0.27, "step": 14872 }, { "epoch": 92.38006230529595, "grad_norm": 1.3984375, "learning_rate": 2.841148852273796e-05, "loss": 0.0946, "step": 14873 }, { "epoch": 92.38629283489097, "grad_norm": 1.609375, "learning_rate": 2.8409047790728814e-05, "loss": 0.1314, "step": 14874 }, { "epoch": 92.39252336448598, "grad_norm": 2.484375, "learning_rate": 2.840660702561031e-05, "loss": 0.249, "step": 14875 }, { "epoch": 92.398753894081, "grad_norm": 1.9140625, "learning_rate": 2.840416622740617e-05, "loss": 0.0961, "step": 14876 }, { "epoch": 92.40498442367601, "grad_norm": 2.859375, "learning_rate": 2.840172539614008e-05, "loss": 0.2727, "step": 14877 }, { "epoch": 92.41121495327103, "grad_norm": 3.078125, "learning_rate": 2.8399284531835752e-05, "loss": 0.2064, "step": 14878 }, { "epoch": 92.41744548286604, "grad_norm": 2.140625, "learning_rate": 2.839684363451689e-05, "loss": 0.1827, "step": 14879 }, { "epoch": 92.42367601246106, "grad_norm": 2.21875, "learning_rate": 2.8394402704207213e-05, "loss": 0.1737, "step": 14880 }, { "epoch": 92.42990654205607, "grad_norm": 2.0625, "learning_rate": 2.8391961740930407e-05, "loss": 0.1583, "step": 14881 }, { "epoch": 92.43613707165109, "grad_norm": 2.984375, "learning_rate": 2.8389520744710196e-05, "loss": 0.1882, "step": 14882 }, { "epoch": 92.4423676012461, "grad_norm": 3.359375, "learning_rate": 2.8387079715570287e-05, "loss": 0.3267, "step": 14883 }, { "epoch": 92.44859813084112, "grad_norm": 1.59375, "learning_rate": 2.838463865353438e-05, "loss": 0.13, "step": 14884 }, { "epoch": 92.45482866043614, "grad_norm": 1.6328125, "learning_rate": 2.8382197558626183e-05, "loss": 0.1529, "step": 14885 }, { "epoch": 92.46105919003115, "grad_norm": 1.484375, "learning_rate": 2.837975643086941e-05, "loss": 0.134, "step": 14886 }, { "epoch": 92.46728971962617, "grad_norm": 1.546875, "learning_rate": 2.8377315270287764e-05, "loss": 0.1129, "step": 14887 }, { "epoch": 92.47352024922118, "grad_norm": 2.921875, "learning_rate": 2.8374874076904954e-05, "loss": 0.1388, "step": 14888 }, { "epoch": 92.4797507788162, "grad_norm": 2.5625, "learning_rate": 2.83724328507447e-05, "loss": 0.1833, "step": 14889 }, { "epoch": 92.48598130841121, "grad_norm": 3.078125, "learning_rate": 2.8369991591830695e-05, "loss": 0.2818, "step": 14890 }, { "epoch": 92.49221183800623, "grad_norm": 2.09375, "learning_rate": 2.836755030018667e-05, "loss": 0.1501, "step": 14891 }, { "epoch": 92.49844236760124, "grad_norm": 2.703125, "learning_rate": 2.8365108975836313e-05, "loss": 0.103, "step": 14892 }, { "epoch": 92.50467289719626, "grad_norm": 1.7421875, "learning_rate": 2.836266761880335e-05, "loss": 0.1161, "step": 14893 }, { "epoch": 92.51090342679127, "grad_norm": 1.328125, "learning_rate": 2.8360226229111487e-05, "loss": 0.1275, "step": 14894 }, { "epoch": 92.51713395638629, "grad_norm": 2.5, "learning_rate": 2.8357784806784433e-05, "loss": 0.2068, "step": 14895 }, { "epoch": 92.5233644859813, "grad_norm": 1.7109375, "learning_rate": 2.835534335184591e-05, "loss": 0.1254, "step": 14896 }, { "epoch": 92.52959501557632, "grad_norm": 1.6953125, "learning_rate": 2.835290186431961e-05, "loss": 0.1137, "step": 14897 }, { "epoch": 92.53582554517133, "grad_norm": 1.65625, "learning_rate": 2.8350460344229256e-05, "loss": 0.1143, "step": 14898 }, { "epoch": 92.54205607476635, "grad_norm": 3.78125, "learning_rate": 2.8348018791598567e-05, "loss": 0.3477, "step": 14899 }, { "epoch": 92.54828660436137, "grad_norm": 2.3125, "learning_rate": 2.8345577206451252e-05, "loss": 0.1956, "step": 14900 }, { "epoch": 92.55451713395638, "grad_norm": 2.546875, "learning_rate": 2.8343135588811015e-05, "loss": 0.2235, "step": 14901 }, { "epoch": 92.5607476635514, "grad_norm": 3.25, "learning_rate": 2.834069393870159e-05, "loss": 0.2847, "step": 14902 }, { "epoch": 92.56697819314641, "grad_norm": 1.4609375, "learning_rate": 2.8338252256146668e-05, "loss": 0.0893, "step": 14903 }, { "epoch": 92.57320872274143, "grad_norm": 2.625, "learning_rate": 2.833581054116997e-05, "loss": 0.1569, "step": 14904 }, { "epoch": 92.57943925233644, "grad_norm": 2.578125, "learning_rate": 2.8333368793795224e-05, "loss": 0.233, "step": 14905 }, { "epoch": 92.58566978193146, "grad_norm": 1.609375, "learning_rate": 2.8330927014046128e-05, "loss": 0.1194, "step": 14906 }, { "epoch": 92.59190031152647, "grad_norm": 2.359375, "learning_rate": 2.8328485201946404e-05, "loss": 0.2344, "step": 14907 }, { "epoch": 92.59813084112149, "grad_norm": 3.0625, "learning_rate": 2.8326043357519766e-05, "loss": 0.1622, "step": 14908 }, { "epoch": 92.60436137071652, "grad_norm": 2.109375, "learning_rate": 2.8323601480789928e-05, "loss": 0.157, "step": 14909 }, { "epoch": 92.61059190031153, "grad_norm": 2.609375, "learning_rate": 2.8321159571780603e-05, "loss": 0.3077, "step": 14910 }, { "epoch": 92.61682242990655, "grad_norm": 1.6015625, "learning_rate": 2.8318717630515524e-05, "loss": 0.1501, "step": 14911 }, { "epoch": 92.62305295950156, "grad_norm": 2.359375, "learning_rate": 2.831627565701839e-05, "loss": 0.1317, "step": 14912 }, { "epoch": 92.62928348909658, "grad_norm": 1.1875, "learning_rate": 2.8313833651312927e-05, "loss": 0.1001, "step": 14913 }, { "epoch": 92.6355140186916, "grad_norm": 1.9921875, "learning_rate": 2.831139161342284e-05, "loss": 0.1038, "step": 14914 }, { "epoch": 92.64174454828661, "grad_norm": 2.90625, "learning_rate": 2.8308949543371866e-05, "loss": 0.2581, "step": 14915 }, { "epoch": 92.64797507788163, "grad_norm": 1.96875, "learning_rate": 2.8306507441183704e-05, "loss": 0.1368, "step": 14916 }, { "epoch": 92.65420560747664, "grad_norm": 2.0625, "learning_rate": 2.830406530688209e-05, "loss": 0.1559, "step": 14917 }, { "epoch": 92.66043613707166, "grad_norm": 1.2421875, "learning_rate": 2.830162314049073e-05, "loss": 0.0967, "step": 14918 }, { "epoch": 92.66666666666667, "grad_norm": 1.1796875, "learning_rate": 2.8299180942033344e-05, "loss": 0.126, "step": 14919 }, { "epoch": 92.67289719626169, "grad_norm": 1.0703125, "learning_rate": 2.8296738711533656e-05, "loss": 0.1247, "step": 14920 }, { "epoch": 92.6791277258567, "grad_norm": 2.453125, "learning_rate": 2.8294296449015377e-05, "loss": 0.231, "step": 14921 }, { "epoch": 92.68535825545172, "grad_norm": 2.109375, "learning_rate": 2.829185415450224e-05, "loss": 0.1673, "step": 14922 }, { "epoch": 92.69158878504673, "grad_norm": 1.34375, "learning_rate": 2.828941182801795e-05, "loss": 0.119, "step": 14923 }, { "epoch": 92.69781931464175, "grad_norm": 1.734375, "learning_rate": 2.8286969469586244e-05, "loss": 0.1386, "step": 14924 }, { "epoch": 92.70404984423676, "grad_norm": 1.609375, "learning_rate": 2.8284527079230822e-05, "loss": 0.0999, "step": 14925 }, { "epoch": 92.71028037383178, "grad_norm": 3.515625, "learning_rate": 2.8282084656975428e-05, "loss": 0.2204, "step": 14926 }, { "epoch": 92.7165109034268, "grad_norm": 1.6640625, "learning_rate": 2.8279642202843765e-05, "loss": 0.1039, "step": 14927 }, { "epoch": 92.72274143302181, "grad_norm": 2.578125, "learning_rate": 2.827719971685956e-05, "loss": 0.2667, "step": 14928 }, { "epoch": 92.72897196261682, "grad_norm": 2.25, "learning_rate": 2.8274757199046537e-05, "loss": 0.2359, "step": 14929 }, { "epoch": 92.73520249221184, "grad_norm": 2.15625, "learning_rate": 2.8272314649428416e-05, "loss": 0.1274, "step": 14930 }, { "epoch": 92.74143302180686, "grad_norm": 2.75, "learning_rate": 2.8269872068028925e-05, "loss": 0.1629, "step": 14931 }, { "epoch": 92.74766355140187, "grad_norm": 1.5546875, "learning_rate": 2.8267429454871786e-05, "loss": 0.1146, "step": 14932 }, { "epoch": 92.75389408099689, "grad_norm": 0.953125, "learning_rate": 2.8264986809980714e-05, "loss": 0.0985, "step": 14933 }, { "epoch": 92.7601246105919, "grad_norm": 3.0, "learning_rate": 2.8262544133379442e-05, "loss": 0.2368, "step": 14934 }, { "epoch": 92.76635514018692, "grad_norm": 2.953125, "learning_rate": 2.8260101425091684e-05, "loss": 0.2193, "step": 14935 }, { "epoch": 92.77258566978193, "grad_norm": 2.34375, "learning_rate": 2.8257658685141164e-05, "loss": 0.1388, "step": 14936 }, { "epoch": 92.77881619937695, "grad_norm": 2.3125, "learning_rate": 2.8255215913551625e-05, "loss": 0.1618, "step": 14937 }, { "epoch": 92.78504672897196, "grad_norm": 2.1875, "learning_rate": 2.825277311034677e-05, "loss": 0.1516, "step": 14938 }, { "epoch": 92.79127725856698, "grad_norm": 2.765625, "learning_rate": 2.8250330275550336e-05, "loss": 0.1308, "step": 14939 }, { "epoch": 92.797507788162, "grad_norm": 0.98828125, "learning_rate": 2.824788740918604e-05, "loss": 0.1047, "step": 14940 }, { "epoch": 92.80373831775701, "grad_norm": 1.2421875, "learning_rate": 2.8245444511277618e-05, "loss": 0.1072, "step": 14941 }, { "epoch": 92.80996884735202, "grad_norm": 2.25, "learning_rate": 2.8243001581848788e-05, "loss": 0.1567, "step": 14942 }, { "epoch": 92.81619937694704, "grad_norm": 2.65625, "learning_rate": 2.824055862092328e-05, "loss": 0.2014, "step": 14943 }, { "epoch": 92.82242990654206, "grad_norm": 1.875, "learning_rate": 2.8238115628524814e-05, "loss": 0.0926, "step": 14944 }, { "epoch": 92.82866043613707, "grad_norm": 1.609375, "learning_rate": 2.823567260467712e-05, "loss": 0.1061, "step": 14945 }, { "epoch": 92.83489096573209, "grad_norm": 2.328125, "learning_rate": 2.823322954940394e-05, "loss": 0.1672, "step": 14946 }, { "epoch": 92.8411214953271, "grad_norm": 0.6796875, "learning_rate": 2.8230786462728975e-05, "loss": 0.0722, "step": 14947 }, { "epoch": 92.84735202492212, "grad_norm": 2.015625, "learning_rate": 2.822834334467597e-05, "loss": 0.1253, "step": 14948 }, { "epoch": 92.85358255451713, "grad_norm": 3.296875, "learning_rate": 2.8225900195268647e-05, "loss": 0.155, "step": 14949 }, { "epoch": 92.85981308411215, "grad_norm": 2.15625, "learning_rate": 2.822345701453074e-05, "loss": 0.1607, "step": 14950 }, { "epoch": 92.86604361370716, "grad_norm": 2.6875, "learning_rate": 2.8221013802485975e-05, "loss": 0.1595, "step": 14951 }, { "epoch": 92.87227414330218, "grad_norm": 2.484375, "learning_rate": 2.8218570559158076e-05, "loss": 0.1122, "step": 14952 }, { "epoch": 92.8785046728972, "grad_norm": 2.25, "learning_rate": 2.821612728457078e-05, "loss": 0.1531, "step": 14953 }, { "epoch": 92.88473520249221, "grad_norm": 2.328125, "learning_rate": 2.821368397874781e-05, "loss": 0.1942, "step": 14954 }, { "epoch": 92.89096573208722, "grad_norm": 2.1875, "learning_rate": 2.8211240641712904e-05, "loss": 0.1121, "step": 14955 }, { "epoch": 92.89719626168224, "grad_norm": 0.92578125, "learning_rate": 2.8208797273489783e-05, "loss": 0.0913, "step": 14956 }, { "epoch": 92.90342679127725, "grad_norm": 3.53125, "learning_rate": 2.8206353874102182e-05, "loss": 0.3532, "step": 14957 }, { "epoch": 92.90965732087227, "grad_norm": 2.21875, "learning_rate": 2.8203910443573828e-05, "loss": 0.1921, "step": 14958 }, { "epoch": 92.91588785046729, "grad_norm": 1.7578125, "learning_rate": 2.8201466981928465e-05, "loss": 0.1211, "step": 14959 }, { "epoch": 92.9221183800623, "grad_norm": 1.8203125, "learning_rate": 2.8199023489189803e-05, "loss": 0.1363, "step": 14960 }, { "epoch": 92.92834890965732, "grad_norm": 3.3125, "learning_rate": 2.8196579965381593e-05, "loss": 0.3335, "step": 14961 }, { "epoch": 92.93457943925233, "grad_norm": 3.171875, "learning_rate": 2.8194136410527556e-05, "loss": 0.3651, "step": 14962 }, { "epoch": 92.94080996884735, "grad_norm": 3.65625, "learning_rate": 2.8191692824651428e-05, "loss": 0.204, "step": 14963 }, { "epoch": 92.94704049844236, "grad_norm": 2.640625, "learning_rate": 2.8189249207776936e-05, "loss": 0.1199, "step": 14964 }, { "epoch": 92.95327102803738, "grad_norm": 2.859375, "learning_rate": 2.8186805559927826e-05, "loss": 0.1945, "step": 14965 }, { "epoch": 92.9595015576324, "grad_norm": 1.2890625, "learning_rate": 2.8184361881127823e-05, "loss": 0.1046, "step": 14966 }, { "epoch": 92.96573208722741, "grad_norm": 3.359375, "learning_rate": 2.8181918171400655e-05, "loss": 0.1516, "step": 14967 }, { "epoch": 92.97196261682242, "grad_norm": 5.1875, "learning_rate": 2.817947443077007e-05, "loss": 0.3958, "step": 14968 }, { "epoch": 92.97819314641744, "grad_norm": 5.125, "learning_rate": 2.817703065925979e-05, "loss": 0.0993, "step": 14969 }, { "epoch": 92.98442367601245, "grad_norm": 2.34375, "learning_rate": 2.8174586856893554e-05, "loss": 0.1639, "step": 14970 }, { "epoch": 92.99065420560747, "grad_norm": 2.359375, "learning_rate": 2.817214302369509e-05, "loss": 0.1631, "step": 14971 }, { "epoch": 92.99688473520249, "grad_norm": 2.125, "learning_rate": 2.816969915968815e-05, "loss": 0.1689, "step": 14972 }, { "epoch": 93.0, "grad_norm": 1.6875, "learning_rate": 2.8167255264896446e-05, "loss": 0.1031, "step": 14973 }, { "epoch": 93.00623052959502, "grad_norm": 1.7265625, "learning_rate": 2.8164811339343732e-05, "loss": 0.1368, "step": 14974 }, { "epoch": 93.01246105919003, "grad_norm": 1.6953125, "learning_rate": 2.816236738305374e-05, "loss": 0.1345, "step": 14975 }, { "epoch": 93.01869158878505, "grad_norm": 2.03125, "learning_rate": 2.8159923396050205e-05, "loss": 0.2255, "step": 14976 }, { "epoch": 93.02492211838006, "grad_norm": 1.796875, "learning_rate": 2.815747937835685e-05, "loss": 0.1097, "step": 14977 }, { "epoch": 93.03115264797508, "grad_norm": 2.375, "learning_rate": 2.8155035329997436e-05, "loss": 0.1608, "step": 14978 }, { "epoch": 93.03738317757009, "grad_norm": 2.828125, "learning_rate": 2.815259125099568e-05, "loss": 0.2188, "step": 14979 }, { "epoch": 93.04361370716511, "grad_norm": 1.7109375, "learning_rate": 2.8150147141375337e-05, "loss": 0.0945, "step": 14980 }, { "epoch": 93.04984423676012, "grad_norm": 2.125, "learning_rate": 2.814770300116013e-05, "loss": 0.0953, "step": 14981 }, { "epoch": 93.05607476635514, "grad_norm": 2.125, "learning_rate": 2.8145258830373806e-05, "loss": 0.1481, "step": 14982 }, { "epoch": 93.06230529595015, "grad_norm": 2.703125, "learning_rate": 2.8142814629040093e-05, "loss": 0.2117, "step": 14983 }, { "epoch": 93.06853582554517, "grad_norm": 2.09375, "learning_rate": 2.8140370397182735e-05, "loss": 0.1423, "step": 14984 }, { "epoch": 93.07476635514018, "grad_norm": 2.078125, "learning_rate": 2.813792613482548e-05, "loss": 0.1781, "step": 14985 }, { "epoch": 93.0809968847352, "grad_norm": 2.4375, "learning_rate": 2.813548184199205e-05, "loss": 0.2089, "step": 14986 }, { "epoch": 93.08722741433021, "grad_norm": 2.96875, "learning_rate": 2.81330375187062e-05, "loss": 0.1482, "step": 14987 }, { "epoch": 93.09345794392523, "grad_norm": 2.65625, "learning_rate": 2.8130593164991663e-05, "loss": 0.151, "step": 14988 }, { "epoch": 93.09968847352025, "grad_norm": 1.78125, "learning_rate": 2.8128148780872183e-05, "loss": 0.1529, "step": 14989 }, { "epoch": 93.10591900311526, "grad_norm": 1.953125, "learning_rate": 2.8125704366371492e-05, "loss": 0.1523, "step": 14990 }, { "epoch": 93.11214953271028, "grad_norm": 1.125, "learning_rate": 2.8123259921513334e-05, "loss": 0.103, "step": 14991 }, { "epoch": 93.11838006230529, "grad_norm": 2.46875, "learning_rate": 2.8120815446321458e-05, "loss": 0.2287, "step": 14992 }, { "epoch": 93.1246105919003, "grad_norm": 1.390625, "learning_rate": 2.811837094081959e-05, "loss": 0.1269, "step": 14993 }, { "epoch": 93.13084112149532, "grad_norm": 2.015625, "learning_rate": 2.8115926405031488e-05, "loss": 0.2604, "step": 14994 }, { "epoch": 93.13707165109034, "grad_norm": 2.109375, "learning_rate": 2.8113481838980875e-05, "loss": 0.1361, "step": 14995 }, { "epoch": 93.14330218068535, "grad_norm": 2.265625, "learning_rate": 2.811103724269151e-05, "loss": 0.164, "step": 14996 }, { "epoch": 93.14953271028037, "grad_norm": 2.328125, "learning_rate": 2.8108592616187133e-05, "loss": 0.1828, "step": 14997 }, { "epoch": 93.15576323987538, "grad_norm": 2.046875, "learning_rate": 2.810614795949148e-05, "loss": 0.1337, "step": 14998 }, { "epoch": 93.1619937694704, "grad_norm": 2.0625, "learning_rate": 2.8103703272628296e-05, "loss": 0.1527, "step": 14999 }, { "epoch": 93.16822429906541, "grad_norm": 2.0625, "learning_rate": 2.8101258555621324e-05, "loss": 0.1762, "step": 15000 }, { "epoch": 93.17445482866043, "grad_norm": 2.140625, "learning_rate": 2.8098813808494312e-05, "loss": 0.1817, "step": 15001 }, { "epoch": 93.18068535825545, "grad_norm": 2.40625, "learning_rate": 2.8096369031271002e-05, "loss": 0.2605, "step": 15002 }, { "epoch": 93.18691588785046, "grad_norm": 3.703125, "learning_rate": 2.8093924223975136e-05, "loss": 0.1775, "step": 15003 }, { "epoch": 93.19314641744548, "grad_norm": 1.1875, "learning_rate": 2.809147938663046e-05, "loss": 0.1119, "step": 15004 }, { "epoch": 93.19937694704049, "grad_norm": 1.3671875, "learning_rate": 2.8089034519260722e-05, "loss": 0.0862, "step": 15005 }, { "epoch": 93.20560747663552, "grad_norm": 2.171875, "learning_rate": 2.8086589621889648e-05, "loss": 0.1685, "step": 15006 }, { "epoch": 93.21183800623054, "grad_norm": 1.84375, "learning_rate": 2.8084144694541014e-05, "loss": 0.1551, "step": 15007 }, { "epoch": 93.21806853582555, "grad_norm": 1.3515625, "learning_rate": 2.8081699737238544e-05, "loss": 0.1115, "step": 15008 }, { "epoch": 93.22429906542057, "grad_norm": 2.703125, "learning_rate": 2.807925475000599e-05, "loss": 0.2062, "step": 15009 }, { "epoch": 93.23052959501558, "grad_norm": 3.5625, "learning_rate": 2.8076809732867106e-05, "loss": 0.2673, "step": 15010 }, { "epoch": 93.2367601246106, "grad_norm": 3.03125, "learning_rate": 2.807436468584562e-05, "loss": 0.317, "step": 15011 }, { "epoch": 93.24299065420561, "grad_norm": 2.03125, "learning_rate": 2.807191960896529e-05, "loss": 0.1682, "step": 15012 }, { "epoch": 93.24922118380063, "grad_norm": 2.328125, "learning_rate": 2.8069474502249875e-05, "loss": 0.1918, "step": 15013 }, { "epoch": 93.25545171339564, "grad_norm": 2.1875, "learning_rate": 2.8067029365723097e-05, "loss": 0.1298, "step": 15014 }, { "epoch": 93.26168224299066, "grad_norm": 3.0, "learning_rate": 2.8064584199408722e-05, "loss": 0.1751, "step": 15015 }, { "epoch": 93.26791277258567, "grad_norm": 3.359375, "learning_rate": 2.806213900333049e-05, "loss": 0.2285, "step": 15016 }, { "epoch": 93.27414330218069, "grad_norm": 3.0, "learning_rate": 2.8059693777512158e-05, "loss": 0.2384, "step": 15017 }, { "epoch": 93.2803738317757, "grad_norm": 2.546875, "learning_rate": 2.8057248521977465e-05, "loss": 0.1988, "step": 15018 }, { "epoch": 93.28660436137072, "grad_norm": 1.6640625, "learning_rate": 2.805480323675016e-05, "loss": 0.1365, "step": 15019 }, { "epoch": 93.29283489096574, "grad_norm": 1.7421875, "learning_rate": 2.8052357921854e-05, "loss": 0.1222, "step": 15020 }, { "epoch": 93.29906542056075, "grad_norm": 1.703125, "learning_rate": 2.8049912577312727e-05, "loss": 0.1152, "step": 15021 }, { "epoch": 93.30529595015577, "grad_norm": 2.34375, "learning_rate": 2.804746720315009e-05, "loss": 0.1545, "step": 15022 }, { "epoch": 93.31152647975078, "grad_norm": 3.015625, "learning_rate": 2.804502179938985e-05, "loss": 0.1901, "step": 15023 }, { "epoch": 93.3177570093458, "grad_norm": 2.328125, "learning_rate": 2.804257636605575e-05, "loss": 0.2426, "step": 15024 }, { "epoch": 93.32398753894081, "grad_norm": 2.125, "learning_rate": 2.804013090317153e-05, "loss": 0.1371, "step": 15025 }, { "epoch": 93.33021806853583, "grad_norm": 2.03125, "learning_rate": 2.803768541076096e-05, "loss": 0.1472, "step": 15026 }, { "epoch": 93.33644859813084, "grad_norm": 1.9140625, "learning_rate": 2.8035239888847774e-05, "loss": 0.1637, "step": 15027 }, { "epoch": 93.34267912772586, "grad_norm": 1.59375, "learning_rate": 2.803279433745574e-05, "loss": 0.1281, "step": 15028 }, { "epoch": 93.34890965732087, "grad_norm": 1.703125, "learning_rate": 2.80303487566086e-05, "loss": 0.1351, "step": 15029 }, { "epoch": 93.35514018691589, "grad_norm": 2.4375, "learning_rate": 2.8027903146330103e-05, "loss": 0.1855, "step": 15030 }, { "epoch": 93.3613707165109, "grad_norm": 2.578125, "learning_rate": 2.802545750664401e-05, "loss": 0.1403, "step": 15031 }, { "epoch": 93.36760124610592, "grad_norm": 2.5, "learning_rate": 2.8023011837574065e-05, "loss": 0.1714, "step": 15032 }, { "epoch": 93.37383177570094, "grad_norm": 2.28125, "learning_rate": 2.8020566139144028e-05, "loss": 0.1706, "step": 15033 }, { "epoch": 93.38006230529595, "grad_norm": 2.515625, "learning_rate": 2.8018120411377648e-05, "loss": 0.2228, "step": 15034 }, { "epoch": 93.38629283489097, "grad_norm": 3.0, "learning_rate": 2.8015674654298673e-05, "loss": 0.2462, "step": 15035 }, { "epoch": 93.39252336448598, "grad_norm": 2.953125, "learning_rate": 2.8013228867930868e-05, "loss": 0.181, "step": 15036 }, { "epoch": 93.398753894081, "grad_norm": 2.625, "learning_rate": 2.8010783052297986e-05, "loss": 0.2048, "step": 15037 }, { "epoch": 93.40498442367601, "grad_norm": 3.046875, "learning_rate": 2.8008337207423775e-05, "loss": 0.2063, "step": 15038 }, { "epoch": 93.41121495327103, "grad_norm": 1.703125, "learning_rate": 2.800589133333199e-05, "loss": 0.1282, "step": 15039 }, { "epoch": 93.41744548286604, "grad_norm": 3.09375, "learning_rate": 2.8003445430046386e-05, "loss": 0.1335, "step": 15040 }, { "epoch": 93.42367601246106, "grad_norm": 3.375, "learning_rate": 2.800099949759072e-05, "loss": 0.2564, "step": 15041 }, { "epoch": 93.42990654205607, "grad_norm": 2.0625, "learning_rate": 2.7998553535988752e-05, "loss": 0.131, "step": 15042 }, { "epoch": 93.43613707165109, "grad_norm": 1.796875, "learning_rate": 2.7996107545264223e-05, "loss": 0.1423, "step": 15043 }, { "epoch": 93.4423676012461, "grad_norm": 2.375, "learning_rate": 2.7993661525440906e-05, "loss": 0.1288, "step": 15044 }, { "epoch": 93.44859813084112, "grad_norm": 2.515625, "learning_rate": 2.7991215476542547e-05, "loss": 0.1929, "step": 15045 }, { "epoch": 93.45482866043614, "grad_norm": 1.6015625, "learning_rate": 2.7988769398592906e-05, "loss": 0.1151, "step": 15046 }, { "epoch": 93.46105919003115, "grad_norm": 2.546875, "learning_rate": 2.7986323291615734e-05, "loss": 0.2029, "step": 15047 }, { "epoch": 93.46728971962617, "grad_norm": 2.8125, "learning_rate": 2.7983877155634808e-05, "loss": 0.1312, "step": 15048 }, { "epoch": 93.47352024922118, "grad_norm": 1.59375, "learning_rate": 2.7981430990673856e-05, "loss": 0.1455, "step": 15049 }, { "epoch": 93.4797507788162, "grad_norm": 3.453125, "learning_rate": 2.7978984796756657e-05, "loss": 0.2893, "step": 15050 }, { "epoch": 93.48598130841121, "grad_norm": 2.578125, "learning_rate": 2.7976538573906963e-05, "loss": 0.1687, "step": 15051 }, { "epoch": 93.49221183800623, "grad_norm": 1.6875, "learning_rate": 2.7974092322148533e-05, "loss": 0.1317, "step": 15052 }, { "epoch": 93.49844236760124, "grad_norm": 2.21875, "learning_rate": 2.7971646041505117e-05, "loss": 0.1247, "step": 15053 }, { "epoch": 93.50467289719626, "grad_norm": 1.3515625, "learning_rate": 2.796919973200048e-05, "loss": 0.106, "step": 15054 }, { "epoch": 93.51090342679127, "grad_norm": 1.5390625, "learning_rate": 2.796675339365839e-05, "loss": 0.102, "step": 15055 }, { "epoch": 93.51713395638629, "grad_norm": 2.890625, "learning_rate": 2.796430702650259e-05, "loss": 0.268, "step": 15056 }, { "epoch": 93.5233644859813, "grad_norm": 1.265625, "learning_rate": 2.7961860630556853e-05, "loss": 0.0784, "step": 15057 }, { "epoch": 93.52959501557632, "grad_norm": 3.203125, "learning_rate": 2.795941420584493e-05, "loss": 0.2462, "step": 15058 }, { "epoch": 93.53582554517133, "grad_norm": 1.3984375, "learning_rate": 2.7956967752390588e-05, "loss": 0.1106, "step": 15059 }, { "epoch": 93.54205607476635, "grad_norm": 1.34375, "learning_rate": 2.7954521270217577e-05, "loss": 0.086, "step": 15060 }, { "epoch": 93.54828660436137, "grad_norm": 0.9921875, "learning_rate": 2.7952074759349677e-05, "loss": 0.105, "step": 15061 }, { "epoch": 93.55451713395638, "grad_norm": 1.8203125, "learning_rate": 2.7949628219810624e-05, "loss": 0.1429, "step": 15062 }, { "epoch": 93.5607476635514, "grad_norm": 1.828125, "learning_rate": 2.79471816516242e-05, "loss": 0.1066, "step": 15063 }, { "epoch": 93.56697819314641, "grad_norm": 2.5, "learning_rate": 2.7944735054814158e-05, "loss": 0.1106, "step": 15064 }, { "epoch": 93.57320872274143, "grad_norm": 3.390625, "learning_rate": 2.7942288429404256e-05, "loss": 0.1663, "step": 15065 }, { "epoch": 93.57943925233644, "grad_norm": 4.53125, "learning_rate": 2.793984177541827e-05, "loss": 0.1787, "step": 15066 }, { "epoch": 93.58566978193146, "grad_norm": 2.390625, "learning_rate": 2.7937395092879942e-05, "loss": 0.2212, "step": 15067 }, { "epoch": 93.59190031152647, "grad_norm": 2.515625, "learning_rate": 2.7934948381813054e-05, "loss": 0.1371, "step": 15068 }, { "epoch": 93.59813084112149, "grad_norm": 2.765625, "learning_rate": 2.7932501642241353e-05, "loss": 0.2761, "step": 15069 }, { "epoch": 93.60436137071652, "grad_norm": 2.609375, "learning_rate": 2.7930054874188616e-05, "loss": 0.2038, "step": 15070 }, { "epoch": 93.61059190031153, "grad_norm": 2.3125, "learning_rate": 2.7927608077678595e-05, "loss": 0.1625, "step": 15071 }, { "epoch": 93.61682242990655, "grad_norm": 2.203125, "learning_rate": 2.7925161252735067e-05, "loss": 0.2997, "step": 15072 }, { "epoch": 93.62305295950156, "grad_norm": 2.828125, "learning_rate": 2.792271439938178e-05, "loss": 0.1784, "step": 15073 }, { "epoch": 93.62928348909658, "grad_norm": 4.40625, "learning_rate": 2.7920267517642513e-05, "loss": 0.1368, "step": 15074 }, { "epoch": 93.6355140186916, "grad_norm": 1.7109375, "learning_rate": 2.791782060754102e-05, "loss": 0.1241, "step": 15075 }, { "epoch": 93.64174454828661, "grad_norm": 1.84375, "learning_rate": 2.791537366910107e-05, "loss": 0.1382, "step": 15076 }, { "epoch": 93.64797507788163, "grad_norm": 0.8984375, "learning_rate": 2.791292670234643e-05, "loss": 0.1034, "step": 15077 }, { "epoch": 93.65420560747664, "grad_norm": 1.1796875, "learning_rate": 2.7910479707300862e-05, "loss": 0.1029, "step": 15078 }, { "epoch": 93.66043613707166, "grad_norm": 3.03125, "learning_rate": 2.790803268398814e-05, "loss": 0.2653, "step": 15079 }, { "epoch": 93.66666666666667, "grad_norm": 1.796875, "learning_rate": 2.7905585632432012e-05, "loss": 0.183, "step": 15080 }, { "epoch": 93.67289719626169, "grad_norm": 1.625, "learning_rate": 2.7903138552656267e-05, "loss": 0.1581, "step": 15081 }, { "epoch": 93.6791277258567, "grad_norm": 2.953125, "learning_rate": 2.7900691444684644e-05, "loss": 0.2497, "step": 15082 }, { "epoch": 93.68535825545172, "grad_norm": 3.078125, "learning_rate": 2.7898244308540943e-05, "loss": 0.2216, "step": 15083 }, { "epoch": 93.69158878504673, "grad_norm": 2.25, "learning_rate": 2.7895797144248904e-05, "loss": 0.2016, "step": 15084 }, { "epoch": 93.69781931464175, "grad_norm": 1.1796875, "learning_rate": 2.789334995183231e-05, "loss": 0.1203, "step": 15085 }, { "epoch": 93.70404984423676, "grad_norm": 2.234375, "learning_rate": 2.789090273131492e-05, "loss": 0.1791, "step": 15086 }, { "epoch": 93.71028037383178, "grad_norm": 1.53125, "learning_rate": 2.7888455482720505e-05, "loss": 0.108, "step": 15087 }, { "epoch": 93.7165109034268, "grad_norm": 1.9140625, "learning_rate": 2.7886008206072832e-05, "loss": 0.1346, "step": 15088 }, { "epoch": 93.72274143302181, "grad_norm": 1.5234375, "learning_rate": 2.7883560901395667e-05, "loss": 0.0994, "step": 15089 }, { "epoch": 93.72897196261682, "grad_norm": 2.0, "learning_rate": 2.7881113568712797e-05, "loss": 0.1168, "step": 15090 }, { "epoch": 93.73520249221184, "grad_norm": 2.921875, "learning_rate": 2.787866620804796e-05, "loss": 0.1923, "step": 15091 }, { "epoch": 93.74143302180686, "grad_norm": 2.109375, "learning_rate": 2.7876218819424947e-05, "loss": 0.1435, "step": 15092 }, { "epoch": 93.74766355140187, "grad_norm": 2.0625, "learning_rate": 2.7873771402867523e-05, "loss": 0.1468, "step": 15093 }, { "epoch": 93.75389408099689, "grad_norm": 1.578125, "learning_rate": 2.7871323958399455e-05, "loss": 0.1237, "step": 15094 }, { "epoch": 93.7601246105919, "grad_norm": 2.546875, "learning_rate": 2.7868876486044508e-05, "loss": 0.1988, "step": 15095 }, { "epoch": 93.76635514018692, "grad_norm": 2.375, "learning_rate": 2.7866428985826476e-05, "loss": 0.1969, "step": 15096 }, { "epoch": 93.77258566978193, "grad_norm": 3.625, "learning_rate": 2.7863981457769094e-05, "loss": 0.3258, "step": 15097 }, { "epoch": 93.77881619937695, "grad_norm": 1.890625, "learning_rate": 2.7861533901896163e-05, "loss": 0.1109, "step": 15098 }, { "epoch": 93.78504672897196, "grad_norm": 3.46875, "learning_rate": 2.785908631823144e-05, "loss": 0.262, "step": 15099 }, { "epoch": 93.79127725856698, "grad_norm": 2.71875, "learning_rate": 2.7856638706798705e-05, "loss": 0.2797, "step": 15100 }, { "epoch": 93.797507788162, "grad_norm": 2.046875, "learning_rate": 2.7854191067621717e-05, "loss": 0.139, "step": 15101 }, { "epoch": 93.80373831775701, "grad_norm": 2.671875, "learning_rate": 2.7851743400724258e-05, "loss": 0.3255, "step": 15102 }, { "epoch": 93.80996884735202, "grad_norm": 3.171875, "learning_rate": 2.7849295706130095e-05, "loss": 0.2237, "step": 15103 }, { "epoch": 93.81619937694704, "grad_norm": 2.109375, "learning_rate": 2.7846847983862996e-05, "loss": 0.1855, "step": 15104 }, { "epoch": 93.82242990654206, "grad_norm": 2.28125, "learning_rate": 2.784440023394675e-05, "loss": 0.1067, "step": 15105 }, { "epoch": 93.82866043613707, "grad_norm": 2.1875, "learning_rate": 2.7841952456405117e-05, "loss": 0.1823, "step": 15106 }, { "epoch": 93.83489096573209, "grad_norm": 2.421875, "learning_rate": 2.7839504651261872e-05, "loss": 0.2307, "step": 15107 }, { "epoch": 93.8411214953271, "grad_norm": 2.328125, "learning_rate": 2.783705681854079e-05, "loss": 0.2108, "step": 15108 }, { "epoch": 93.84735202492212, "grad_norm": 1.5234375, "learning_rate": 2.783460895826565e-05, "loss": 0.1056, "step": 15109 }, { "epoch": 93.85358255451713, "grad_norm": 2.375, "learning_rate": 2.7832161070460216e-05, "loss": 0.1591, "step": 15110 }, { "epoch": 93.85981308411215, "grad_norm": 1.6171875, "learning_rate": 2.7829713155148275e-05, "loss": 0.1492, "step": 15111 }, { "epoch": 93.86604361370716, "grad_norm": 3.015625, "learning_rate": 2.7827265212353587e-05, "loss": 0.2168, "step": 15112 }, { "epoch": 93.87227414330218, "grad_norm": 1.796875, "learning_rate": 2.7824817242099938e-05, "loss": 0.1534, "step": 15113 }, { "epoch": 93.8785046728972, "grad_norm": 2.625, "learning_rate": 2.7822369244411094e-05, "loss": 0.2106, "step": 15114 }, { "epoch": 93.88473520249221, "grad_norm": 1.640625, "learning_rate": 2.7819921219310842e-05, "loss": 0.1328, "step": 15115 }, { "epoch": 93.89096573208722, "grad_norm": 2.140625, "learning_rate": 2.7817473166822948e-05, "loss": 0.1726, "step": 15116 }, { "epoch": 93.89719626168224, "grad_norm": 1.9296875, "learning_rate": 2.781502508697119e-05, "loss": 0.1546, "step": 15117 }, { "epoch": 93.90342679127725, "grad_norm": 2.21875, "learning_rate": 2.7812576979779354e-05, "loss": 0.1881, "step": 15118 }, { "epoch": 93.90965732087227, "grad_norm": 2.765625, "learning_rate": 2.78101288452712e-05, "loss": 0.2339, "step": 15119 }, { "epoch": 93.91588785046729, "grad_norm": 1.3828125, "learning_rate": 2.780768068347051e-05, "loss": 0.0877, "step": 15120 }, { "epoch": 93.9221183800623, "grad_norm": 1.71875, "learning_rate": 2.7805232494401073e-05, "loss": 0.0942, "step": 15121 }, { "epoch": 93.92834890965732, "grad_norm": 1.8515625, "learning_rate": 2.7802784278086653e-05, "loss": 0.1447, "step": 15122 }, { "epoch": 93.93457943925233, "grad_norm": 2.421875, "learning_rate": 2.7800336034551027e-05, "loss": 0.1852, "step": 15123 }, { "epoch": 93.94080996884735, "grad_norm": 2.65625, "learning_rate": 2.7797887763817982e-05, "loss": 0.1515, "step": 15124 }, { "epoch": 93.94704049844236, "grad_norm": 1.4453125, "learning_rate": 2.7795439465911294e-05, "loss": 0.1444, "step": 15125 }, { "epoch": 93.95327102803738, "grad_norm": 2.953125, "learning_rate": 2.779299114085474e-05, "loss": 0.1675, "step": 15126 }, { "epoch": 93.9595015576324, "grad_norm": 3.03125, "learning_rate": 2.7790542788672092e-05, "loss": 0.195, "step": 15127 }, { "epoch": 93.96573208722741, "grad_norm": 1.78125, "learning_rate": 2.778809440938714e-05, "loss": 0.1238, "step": 15128 }, { "epoch": 93.97196261682242, "grad_norm": 1.796875, "learning_rate": 2.7785646003023656e-05, "loss": 0.161, "step": 15129 }, { "epoch": 93.97819314641744, "grad_norm": 3.28125, "learning_rate": 2.778319756960542e-05, "loss": 0.1947, "step": 15130 }, { "epoch": 93.98442367601245, "grad_norm": 4.21875, "learning_rate": 2.7780749109156218e-05, "loss": 0.2259, "step": 15131 }, { "epoch": 93.99065420560747, "grad_norm": 2.671875, "learning_rate": 2.7778300621699816e-05, "loss": 0.1889, "step": 15132 }, { "epoch": 93.99688473520249, "grad_norm": 2.421875, "learning_rate": 2.7775852107260004e-05, "loss": 0.1732, "step": 15133 }, { "epoch": 94.0, "grad_norm": 1.1875, "learning_rate": 2.7773403565860566e-05, "loss": 0.0587, "step": 15134 }, { "epoch": 94.00623052959502, "grad_norm": 1.265625, "learning_rate": 2.7770954997525277e-05, "loss": 0.1322, "step": 15135 }, { "epoch": 94.01246105919003, "grad_norm": 2.609375, "learning_rate": 2.7768506402277922e-05, "loss": 0.1227, "step": 15136 }, { "epoch": 94.01869158878505, "grad_norm": 1.7109375, "learning_rate": 2.776605778014228e-05, "loss": 0.1213, "step": 15137 }, { "epoch": 94.02492211838006, "grad_norm": 1.8828125, "learning_rate": 2.7763609131142128e-05, "loss": 0.1339, "step": 15138 }, { "epoch": 94.03115264797508, "grad_norm": 1.28125, "learning_rate": 2.7761160455301244e-05, "loss": 0.1077, "step": 15139 }, { "epoch": 94.03738317757009, "grad_norm": 1.328125, "learning_rate": 2.775871175264343e-05, "loss": 0.1419, "step": 15140 }, { "epoch": 94.04361370716511, "grad_norm": 2.5, "learning_rate": 2.7756263023192453e-05, "loss": 0.1534, "step": 15141 }, { "epoch": 94.04984423676012, "grad_norm": 1.78125, "learning_rate": 2.7753814266972093e-05, "loss": 0.1495, "step": 15142 }, { "epoch": 94.05607476635514, "grad_norm": 1.2421875, "learning_rate": 2.7751365484006143e-05, "loss": 0.1133, "step": 15143 }, { "epoch": 94.06230529595015, "grad_norm": 2.90625, "learning_rate": 2.7748916674318388e-05, "loss": 0.2081, "step": 15144 }, { "epoch": 94.06853582554517, "grad_norm": 1.9296875, "learning_rate": 2.7746467837932587e-05, "loss": 0.1197, "step": 15145 }, { "epoch": 94.07476635514018, "grad_norm": 2.515625, "learning_rate": 2.7744018974872553e-05, "loss": 0.2158, "step": 15146 }, { "epoch": 94.0809968847352, "grad_norm": 2.515625, "learning_rate": 2.774157008516206e-05, "loss": 0.2134, "step": 15147 }, { "epoch": 94.08722741433021, "grad_norm": 1.7265625, "learning_rate": 2.773912116882489e-05, "loss": 0.1387, "step": 15148 }, { "epoch": 94.09345794392523, "grad_norm": 1.796875, "learning_rate": 2.7736672225884824e-05, "loss": 0.111, "step": 15149 }, { "epoch": 94.09968847352025, "grad_norm": 1.5078125, "learning_rate": 2.7734223256365654e-05, "loss": 0.1073, "step": 15150 }, { "epoch": 94.10591900311526, "grad_norm": 2.796875, "learning_rate": 2.773177426029116e-05, "loss": 0.1773, "step": 15151 }, { "epoch": 94.11214953271028, "grad_norm": 2.796875, "learning_rate": 2.7729325237685123e-05, "loss": 0.2331, "step": 15152 }, { "epoch": 94.11838006230529, "grad_norm": 2.1875, "learning_rate": 2.7726876188571345e-05, "loss": 0.1811, "step": 15153 }, { "epoch": 94.1246105919003, "grad_norm": 2.5625, "learning_rate": 2.7724427112973587e-05, "loss": 0.1143, "step": 15154 }, { "epoch": 94.13084112149532, "grad_norm": 2.1875, "learning_rate": 2.7721978010915656e-05, "loss": 0.1423, "step": 15155 }, { "epoch": 94.13707165109034, "grad_norm": 2.171875, "learning_rate": 2.7719528882421324e-05, "loss": 0.1192, "step": 15156 }, { "epoch": 94.14330218068535, "grad_norm": 1.3984375, "learning_rate": 2.7717079727514395e-05, "loss": 0.1284, "step": 15157 }, { "epoch": 94.14953271028037, "grad_norm": 1.4296875, "learning_rate": 2.7714630546218635e-05, "loss": 0.1103, "step": 15158 }, { "epoch": 94.15576323987538, "grad_norm": 3.203125, "learning_rate": 2.7712181338557843e-05, "loss": 0.1835, "step": 15159 }, { "epoch": 94.1619937694704, "grad_norm": 2.75, "learning_rate": 2.7709732104555806e-05, "loss": 0.1836, "step": 15160 }, { "epoch": 94.16822429906541, "grad_norm": 3.390625, "learning_rate": 2.770728284423631e-05, "loss": 0.1574, "step": 15161 }, { "epoch": 94.17445482866043, "grad_norm": 2.734375, "learning_rate": 2.7704833557623137e-05, "loss": 0.1076, "step": 15162 }, { "epoch": 94.18068535825545, "grad_norm": 2.84375, "learning_rate": 2.7702384244740088e-05, "loss": 0.2364, "step": 15163 }, { "epoch": 94.18691588785046, "grad_norm": 1.4765625, "learning_rate": 2.7699934905610936e-05, "loss": 0.1053, "step": 15164 }, { "epoch": 94.19314641744548, "grad_norm": 2.78125, "learning_rate": 2.769748554025947e-05, "loss": 0.2672, "step": 15165 }, { "epoch": 94.19937694704049, "grad_norm": 3.3125, "learning_rate": 2.7695036148709502e-05, "loss": 0.2375, "step": 15166 }, { "epoch": 94.20560747663552, "grad_norm": 2.328125, "learning_rate": 2.7692586730984793e-05, "loss": 0.1534, "step": 15167 }, { "epoch": 94.21183800623054, "grad_norm": 3.796875, "learning_rate": 2.769013728710914e-05, "loss": 0.1157, "step": 15168 }, { "epoch": 94.21806853582555, "grad_norm": 1.5390625, "learning_rate": 2.7687687817106345e-05, "loss": 0.1138, "step": 15169 }, { "epoch": 94.22429906542057, "grad_norm": 2.09375, "learning_rate": 2.768523832100019e-05, "loss": 0.207, "step": 15170 }, { "epoch": 94.23052959501558, "grad_norm": 0.9609375, "learning_rate": 2.768278879881445e-05, "loss": 0.1161, "step": 15171 }, { "epoch": 94.2367601246106, "grad_norm": 1.390625, "learning_rate": 2.7680339250572946e-05, "loss": 0.1253, "step": 15172 }, { "epoch": 94.24299065420561, "grad_norm": 2.078125, "learning_rate": 2.767788967629944e-05, "loss": 0.1406, "step": 15173 }, { "epoch": 94.24922118380063, "grad_norm": 2.171875, "learning_rate": 2.7675440076017733e-05, "loss": 0.147, "step": 15174 }, { "epoch": 94.25545171339564, "grad_norm": 1.234375, "learning_rate": 2.7672990449751622e-05, "loss": 0.1082, "step": 15175 }, { "epoch": 94.26168224299066, "grad_norm": 1.40625, "learning_rate": 2.7670540797524896e-05, "loss": 0.0922, "step": 15176 }, { "epoch": 94.26791277258567, "grad_norm": 1.578125, "learning_rate": 2.7668091119361344e-05, "loss": 0.1204, "step": 15177 }, { "epoch": 94.27414330218069, "grad_norm": 1.96875, "learning_rate": 2.766564141528475e-05, "loss": 0.1344, "step": 15178 }, { "epoch": 94.2803738317757, "grad_norm": 1.9609375, "learning_rate": 2.7663191685318923e-05, "loss": 0.2187, "step": 15179 }, { "epoch": 94.28660436137072, "grad_norm": 1.96875, "learning_rate": 2.7660741929487637e-05, "loss": 0.1438, "step": 15180 }, { "epoch": 94.29283489096574, "grad_norm": 1.390625, "learning_rate": 2.76582921478147e-05, "loss": 0.161, "step": 15181 }, { "epoch": 94.29906542056075, "grad_norm": 2.203125, "learning_rate": 2.7655842340323895e-05, "loss": 0.2253, "step": 15182 }, { "epoch": 94.30529595015577, "grad_norm": 4.9375, "learning_rate": 2.765339250703902e-05, "loss": 0.2413, "step": 15183 }, { "epoch": 94.31152647975078, "grad_norm": 2.734375, "learning_rate": 2.765094264798387e-05, "loss": 0.1561, "step": 15184 }, { "epoch": 94.3177570093458, "grad_norm": 4.3125, "learning_rate": 2.764849276318223e-05, "loss": 0.2191, "step": 15185 }, { "epoch": 94.32398753894081, "grad_norm": 1.515625, "learning_rate": 2.76460428526579e-05, "loss": 0.119, "step": 15186 }, { "epoch": 94.33021806853583, "grad_norm": 2.546875, "learning_rate": 2.764359291643467e-05, "loss": 0.3061, "step": 15187 }, { "epoch": 94.33644859813084, "grad_norm": 2.140625, "learning_rate": 2.7641142954536346e-05, "loss": 0.0999, "step": 15188 }, { "epoch": 94.34267912772586, "grad_norm": 2.265625, "learning_rate": 2.7638692966986702e-05, "loss": 0.151, "step": 15189 }, { "epoch": 94.34890965732087, "grad_norm": 4.1875, "learning_rate": 2.7636242953809554e-05, "loss": 0.3052, "step": 15190 }, { "epoch": 94.35514018691589, "grad_norm": 3.40625, "learning_rate": 2.7633792915028677e-05, "loss": 0.2401, "step": 15191 }, { "epoch": 94.3613707165109, "grad_norm": 3.90625, "learning_rate": 2.763134285066789e-05, "loss": 0.1556, "step": 15192 }, { "epoch": 94.36760124610592, "grad_norm": 2.046875, "learning_rate": 2.7628892760750963e-05, "loss": 0.1037, "step": 15193 }, { "epoch": 94.37383177570094, "grad_norm": 1.84375, "learning_rate": 2.762644264530171e-05, "loss": 0.1491, "step": 15194 }, { "epoch": 94.38006230529595, "grad_norm": 2.34375, "learning_rate": 2.762399250434392e-05, "loss": 0.1572, "step": 15195 }, { "epoch": 94.38629283489097, "grad_norm": 1.9453125, "learning_rate": 2.7621542337901397e-05, "loss": 0.1254, "step": 15196 }, { "epoch": 94.39252336448598, "grad_norm": 3.84375, "learning_rate": 2.7619092145997922e-05, "loss": 0.1662, "step": 15197 }, { "epoch": 94.398753894081, "grad_norm": 3.625, "learning_rate": 2.7616641928657306e-05, "loss": 0.1561, "step": 15198 }, { "epoch": 94.40498442367601, "grad_norm": 2.9375, "learning_rate": 2.7614191685903336e-05, "loss": 0.2455, "step": 15199 }, { "epoch": 94.41121495327103, "grad_norm": 4.21875, "learning_rate": 2.761174141775981e-05, "loss": 0.2723, "step": 15200 }, { "epoch": 94.41744548286604, "grad_norm": 2.078125, "learning_rate": 2.760929112425054e-05, "loss": 0.1814, "step": 15201 }, { "epoch": 94.42367601246106, "grad_norm": 1.78125, "learning_rate": 2.7606840805399303e-05, "loss": 0.1031, "step": 15202 }, { "epoch": 94.42990654205607, "grad_norm": 2.96875, "learning_rate": 2.7604390461229913e-05, "loss": 0.2236, "step": 15203 }, { "epoch": 94.43613707165109, "grad_norm": 2.53125, "learning_rate": 2.7601940091766164e-05, "loss": 0.2236, "step": 15204 }, { "epoch": 94.4423676012461, "grad_norm": 3.125, "learning_rate": 2.759948969703185e-05, "loss": 0.3305, "step": 15205 }, { "epoch": 94.44859813084112, "grad_norm": 2.5, "learning_rate": 2.759703927705077e-05, "loss": 0.246, "step": 15206 }, { "epoch": 94.45482866043614, "grad_norm": 2.734375, "learning_rate": 2.759458883184673e-05, "loss": 0.1412, "step": 15207 }, { "epoch": 94.46105919003115, "grad_norm": 1.7578125, "learning_rate": 2.759213836144352e-05, "loss": 0.1016, "step": 15208 }, { "epoch": 94.46728971962617, "grad_norm": 3.265625, "learning_rate": 2.7589687865864945e-05, "loss": 0.113, "step": 15209 }, { "epoch": 94.47352024922118, "grad_norm": 1.4609375, "learning_rate": 2.758723734513481e-05, "loss": 0.1089, "step": 15210 }, { "epoch": 94.4797507788162, "grad_norm": 2.765625, "learning_rate": 2.75847867992769e-05, "loss": 0.3032, "step": 15211 }, { "epoch": 94.48598130841121, "grad_norm": 2.046875, "learning_rate": 2.758233622831503e-05, "loss": 0.1549, "step": 15212 }, { "epoch": 94.49221183800623, "grad_norm": 1.609375, "learning_rate": 2.757988563227299e-05, "loss": 0.1168, "step": 15213 }, { "epoch": 94.49844236760124, "grad_norm": 3.140625, "learning_rate": 2.757743501117459e-05, "loss": 0.268, "step": 15214 }, { "epoch": 94.50467289719626, "grad_norm": 2.421875, "learning_rate": 2.757498436504362e-05, "loss": 0.2163, "step": 15215 }, { "epoch": 94.51090342679127, "grad_norm": 1.9296875, "learning_rate": 2.757253369390389e-05, "loss": 0.142, "step": 15216 }, { "epoch": 94.51713395638629, "grad_norm": 1.0390625, "learning_rate": 2.75700829977792e-05, "loss": 0.1071, "step": 15217 }, { "epoch": 94.5233644859813, "grad_norm": 1.234375, "learning_rate": 2.756763227669335e-05, "loss": 0.107, "step": 15218 }, { "epoch": 94.52959501557632, "grad_norm": 2.671875, "learning_rate": 2.7565181530670136e-05, "loss": 0.1563, "step": 15219 }, { "epoch": 94.53582554517133, "grad_norm": 2.109375, "learning_rate": 2.7562730759733373e-05, "loss": 0.2196, "step": 15220 }, { "epoch": 94.54205607476635, "grad_norm": 1.328125, "learning_rate": 2.7560279963906855e-05, "loss": 0.1117, "step": 15221 }, { "epoch": 94.54828660436137, "grad_norm": 2.03125, "learning_rate": 2.7557829143214388e-05, "loss": 0.1517, "step": 15222 }, { "epoch": 94.55451713395638, "grad_norm": 1.6015625, "learning_rate": 2.755537829767977e-05, "loss": 0.1279, "step": 15223 }, { "epoch": 94.5607476635514, "grad_norm": 2.796875, "learning_rate": 2.7552927427326806e-05, "loss": 0.1788, "step": 15224 }, { "epoch": 94.56697819314641, "grad_norm": 2.765625, "learning_rate": 2.7550476532179302e-05, "loss": 0.1469, "step": 15225 }, { "epoch": 94.57320872274143, "grad_norm": 2.640625, "learning_rate": 2.754802561226106e-05, "loss": 0.2588, "step": 15226 }, { "epoch": 94.57943925233644, "grad_norm": 2.625, "learning_rate": 2.754557466759589e-05, "loss": 0.2086, "step": 15227 }, { "epoch": 94.58566978193146, "grad_norm": 2.421875, "learning_rate": 2.7543123698207578e-05, "loss": 0.1133, "step": 15228 }, { "epoch": 94.59190031152647, "grad_norm": 2.484375, "learning_rate": 2.754067270411995e-05, "loss": 0.2169, "step": 15229 }, { "epoch": 94.59813084112149, "grad_norm": 2.875, "learning_rate": 2.7538221685356798e-05, "loss": 0.2043, "step": 15230 }, { "epoch": 94.60436137071652, "grad_norm": 2.953125, "learning_rate": 2.7535770641941933e-05, "loss": 0.1516, "step": 15231 }, { "epoch": 94.61059190031153, "grad_norm": 3.40625, "learning_rate": 2.7533319573899152e-05, "loss": 0.2516, "step": 15232 }, { "epoch": 94.61682242990655, "grad_norm": 2.015625, "learning_rate": 2.7530868481252266e-05, "loss": 0.1687, "step": 15233 }, { "epoch": 94.62305295950156, "grad_norm": 2.703125, "learning_rate": 2.7528417364025076e-05, "loss": 0.1785, "step": 15234 }, { "epoch": 94.62928348909658, "grad_norm": 3.4375, "learning_rate": 2.7525966222241395e-05, "loss": 0.204, "step": 15235 }, { "epoch": 94.6355140186916, "grad_norm": 2.203125, "learning_rate": 2.752351505592503e-05, "loss": 0.192, "step": 15236 }, { "epoch": 94.64174454828661, "grad_norm": 1.8046875, "learning_rate": 2.7521063865099772e-05, "loss": 0.1587, "step": 15237 }, { "epoch": 94.64797507788163, "grad_norm": 2.296875, "learning_rate": 2.7518612649789444e-05, "loss": 0.1924, "step": 15238 }, { "epoch": 94.65420560747664, "grad_norm": 2.390625, "learning_rate": 2.7516161410017844e-05, "loss": 0.1541, "step": 15239 }, { "epoch": 94.66043613707166, "grad_norm": 3.25, "learning_rate": 2.7513710145808785e-05, "loss": 0.3306, "step": 15240 }, { "epoch": 94.66666666666667, "grad_norm": 3.328125, "learning_rate": 2.751125885718606e-05, "loss": 0.1937, "step": 15241 }, { "epoch": 94.67289719626169, "grad_norm": 1.9609375, "learning_rate": 2.7508807544173504e-05, "loss": 0.1654, "step": 15242 }, { "epoch": 94.6791277258567, "grad_norm": 2.140625, "learning_rate": 2.7506356206794892e-05, "loss": 0.228, "step": 15243 }, { "epoch": 94.68535825545172, "grad_norm": 2.171875, "learning_rate": 2.750390484507406e-05, "loss": 0.1822, "step": 15244 }, { "epoch": 94.69158878504673, "grad_norm": 2.171875, "learning_rate": 2.7501453459034793e-05, "loss": 0.1423, "step": 15245 }, { "epoch": 94.69781931464175, "grad_norm": 2.390625, "learning_rate": 2.7499002048700916e-05, "loss": 0.2, "step": 15246 }, { "epoch": 94.70404984423676, "grad_norm": 1.375, "learning_rate": 2.7496550614096234e-05, "loss": 0.1148, "step": 15247 }, { "epoch": 94.71028037383178, "grad_norm": 1.28125, "learning_rate": 2.7494099155244546e-05, "loss": 0.1069, "step": 15248 }, { "epoch": 94.7165109034268, "grad_norm": 1.390625, "learning_rate": 2.7491647672169678e-05, "loss": 0.1239, "step": 15249 }, { "epoch": 94.72274143302181, "grad_norm": 1.90625, "learning_rate": 2.748919616489542e-05, "loss": 0.154, "step": 15250 }, { "epoch": 94.72897196261682, "grad_norm": 2.625, "learning_rate": 2.7486744633445593e-05, "loss": 0.1372, "step": 15251 }, { "epoch": 94.73520249221184, "grad_norm": 2.78125, "learning_rate": 2.748429307784401e-05, "loss": 0.1144, "step": 15252 }, { "epoch": 94.74143302180686, "grad_norm": 1.9921875, "learning_rate": 2.7481841498114476e-05, "loss": 0.1165, "step": 15253 }, { "epoch": 94.74766355140187, "grad_norm": 2.640625, "learning_rate": 2.7479389894280794e-05, "loss": 0.1664, "step": 15254 }, { "epoch": 94.75389408099689, "grad_norm": 2.921875, "learning_rate": 2.747693826636679e-05, "loss": 0.1855, "step": 15255 }, { "epoch": 94.7601246105919, "grad_norm": 2.0, "learning_rate": 2.7474486614396254e-05, "loss": 0.1375, "step": 15256 }, { "epoch": 94.76635514018692, "grad_norm": 2.15625, "learning_rate": 2.747203493839302e-05, "loss": 0.1593, "step": 15257 }, { "epoch": 94.77258566978193, "grad_norm": 1.515625, "learning_rate": 2.746958323838089e-05, "loss": 0.1248, "step": 15258 }, { "epoch": 94.77881619937695, "grad_norm": 2.203125, "learning_rate": 2.746713151438367e-05, "loss": 0.1759, "step": 15259 }, { "epoch": 94.78504672897196, "grad_norm": 3.234375, "learning_rate": 2.7464679766425176e-05, "loss": 0.2503, "step": 15260 }, { "epoch": 94.79127725856698, "grad_norm": 3.390625, "learning_rate": 2.7462227994529217e-05, "loss": 0.2058, "step": 15261 }, { "epoch": 94.797507788162, "grad_norm": 2.125, "learning_rate": 2.7459776198719617e-05, "loss": 0.1112, "step": 15262 }, { "epoch": 94.80373831775701, "grad_norm": 2.078125, "learning_rate": 2.7457324379020167e-05, "loss": 0.1224, "step": 15263 }, { "epoch": 94.80996884735202, "grad_norm": 2.1875, "learning_rate": 2.7454872535454697e-05, "loss": 0.1052, "step": 15264 }, { "epoch": 94.81619937694704, "grad_norm": 2.671875, "learning_rate": 2.7452420668047015e-05, "loss": 0.1237, "step": 15265 }, { "epoch": 94.82242990654206, "grad_norm": 1.9921875, "learning_rate": 2.7449968776820935e-05, "loss": 0.093, "step": 15266 }, { "epoch": 94.82866043613707, "grad_norm": 3.0, "learning_rate": 2.744751686180026e-05, "loss": 0.3121, "step": 15267 }, { "epoch": 94.83489096573209, "grad_norm": 2.25, "learning_rate": 2.7445064923008822e-05, "loss": 0.1123, "step": 15268 }, { "epoch": 94.8411214953271, "grad_norm": 3.75, "learning_rate": 2.7442612960470415e-05, "loss": 0.3644, "step": 15269 }, { "epoch": 94.84735202492212, "grad_norm": 1.9140625, "learning_rate": 2.744016097420887e-05, "loss": 0.1168, "step": 15270 }, { "epoch": 94.85358255451713, "grad_norm": 1.625, "learning_rate": 2.7437708964247987e-05, "loss": 0.1164, "step": 15271 }, { "epoch": 94.85981308411215, "grad_norm": 2.625, "learning_rate": 2.7435256930611597e-05, "loss": 0.1332, "step": 15272 }, { "epoch": 94.86604361370716, "grad_norm": 2.296875, "learning_rate": 2.74328048733235e-05, "loss": 0.2194, "step": 15273 }, { "epoch": 94.87227414330218, "grad_norm": 1.7890625, "learning_rate": 2.7430352792407514e-05, "loss": 0.0933, "step": 15274 }, { "epoch": 94.8785046728972, "grad_norm": 3.59375, "learning_rate": 2.742790068788746e-05, "loss": 0.1444, "step": 15275 }, { "epoch": 94.88473520249221, "grad_norm": 3.890625, "learning_rate": 2.7425448559787142e-05, "loss": 0.1629, "step": 15276 }, { "epoch": 94.89096573208722, "grad_norm": 3.734375, "learning_rate": 2.7422996408130392e-05, "loss": 0.1535, "step": 15277 }, { "epoch": 94.89719626168224, "grad_norm": 2.78125, "learning_rate": 2.742054423294101e-05, "loss": 0.1511, "step": 15278 }, { "epoch": 94.90342679127725, "grad_norm": 1.953125, "learning_rate": 2.741809203424282e-05, "loss": 0.1433, "step": 15279 }, { "epoch": 94.90965732087227, "grad_norm": 2.78125, "learning_rate": 2.7415639812059633e-05, "loss": 0.2127, "step": 15280 }, { "epoch": 94.91588785046729, "grad_norm": 2.0625, "learning_rate": 2.7413187566415277e-05, "loss": 0.1488, "step": 15281 }, { "epoch": 94.9221183800623, "grad_norm": 1.765625, "learning_rate": 2.7410735297333556e-05, "loss": 0.1453, "step": 15282 }, { "epoch": 94.92834890965732, "grad_norm": 3.015625, "learning_rate": 2.7408283004838293e-05, "loss": 0.1473, "step": 15283 }, { "epoch": 94.93457943925233, "grad_norm": 2.78125, "learning_rate": 2.740583068895331e-05, "loss": 0.1064, "step": 15284 }, { "epoch": 94.94080996884735, "grad_norm": 3.921875, "learning_rate": 2.7403378349702407e-05, "loss": 0.2477, "step": 15285 }, { "epoch": 94.94704049844236, "grad_norm": 1.6171875, "learning_rate": 2.7400925987109417e-05, "loss": 0.1351, "step": 15286 }, { "epoch": 94.95327102803738, "grad_norm": 0.9296875, "learning_rate": 2.739847360119816e-05, "loss": 0.089, "step": 15287 }, { "epoch": 94.9595015576324, "grad_norm": 2.234375, "learning_rate": 2.7396021191992443e-05, "loss": 0.1216, "step": 15288 }, { "epoch": 94.96573208722741, "grad_norm": 1.5703125, "learning_rate": 2.7393568759516087e-05, "loss": 0.1251, "step": 15289 }, { "epoch": 94.97196261682242, "grad_norm": 2.84375, "learning_rate": 2.739111630379292e-05, "loss": 0.2113, "step": 15290 }, { "epoch": 94.97819314641744, "grad_norm": 2.859375, "learning_rate": 2.7388663824846745e-05, "loss": 0.1451, "step": 15291 }, { "epoch": 94.98442367601245, "grad_norm": 1.875, "learning_rate": 2.7386211322701395e-05, "loss": 0.1693, "step": 15292 }, { "epoch": 94.99065420560747, "grad_norm": 3.09375, "learning_rate": 2.738375879738068e-05, "loss": 0.2695, "step": 15293 }, { "epoch": 94.99688473520249, "grad_norm": 2.3125, "learning_rate": 2.738130624890843e-05, "loss": 0.1762, "step": 15294 }, { "epoch": 95.0, "grad_norm": 0.99609375, "learning_rate": 2.7378853677308454e-05, "loss": 0.0585, "step": 15295 }, { "epoch": 95.00623052959502, "grad_norm": 1.703125, "learning_rate": 2.7376401082604564e-05, "loss": 0.1354, "step": 15296 }, { "epoch": 95.01246105919003, "grad_norm": 3.453125, "learning_rate": 2.737394846482061e-05, "loss": 0.3118, "step": 15297 }, { "epoch": 95.01869158878505, "grad_norm": 2.75, "learning_rate": 2.737149582398038e-05, "loss": 0.1382, "step": 15298 }, { "epoch": 95.02492211838006, "grad_norm": 1.5546875, "learning_rate": 2.7369043160107715e-05, "loss": 0.1046, "step": 15299 }, { "epoch": 95.03115264797508, "grad_norm": 2.75, "learning_rate": 2.7366590473226434e-05, "loss": 0.1644, "step": 15300 }, { "epoch": 95.03738317757009, "grad_norm": 2.125, "learning_rate": 2.736413776336035e-05, "loss": 0.187, "step": 15301 }, { "epoch": 95.04361370716511, "grad_norm": 2.40625, "learning_rate": 2.736168503053328e-05, "loss": 0.2679, "step": 15302 }, { "epoch": 95.04984423676012, "grad_norm": 1.296875, "learning_rate": 2.7359232274769063e-05, "loss": 0.1016, "step": 15303 }, { "epoch": 95.05607476635514, "grad_norm": 2.9375, "learning_rate": 2.73567794960915e-05, "loss": 0.2781, "step": 15304 }, { "epoch": 95.06230529595015, "grad_norm": 1.3984375, "learning_rate": 2.735432669452443e-05, "loss": 0.133, "step": 15305 }, { "epoch": 95.06853582554517, "grad_norm": 1.9375, "learning_rate": 2.7351873870091672e-05, "loss": 0.1163, "step": 15306 }, { "epoch": 95.07476635514018, "grad_norm": 1.7890625, "learning_rate": 2.7349421022817038e-05, "loss": 0.1027, "step": 15307 }, { "epoch": 95.0809968847352, "grad_norm": 2.46875, "learning_rate": 2.734696815272436e-05, "loss": 0.2598, "step": 15308 }, { "epoch": 95.08722741433021, "grad_norm": 2.5, "learning_rate": 2.734451525983746e-05, "loss": 0.1574, "step": 15309 }, { "epoch": 95.09345794392523, "grad_norm": 2.078125, "learning_rate": 2.734206234418016e-05, "loss": 0.1687, "step": 15310 }, { "epoch": 95.09968847352025, "grad_norm": 1.6796875, "learning_rate": 2.7339609405776273e-05, "loss": 0.1122, "step": 15311 }, { "epoch": 95.10591900311526, "grad_norm": 2.46875, "learning_rate": 2.7337156444649647e-05, "loss": 0.2395, "step": 15312 }, { "epoch": 95.11214953271028, "grad_norm": 2.671875, "learning_rate": 2.7334703460824073e-05, "loss": 0.166, "step": 15313 }, { "epoch": 95.11838006230529, "grad_norm": 2.078125, "learning_rate": 2.7332250454323403e-05, "loss": 0.1161, "step": 15314 }, { "epoch": 95.1246105919003, "grad_norm": 2.71875, "learning_rate": 2.7329797425171443e-05, "loss": 0.2708, "step": 15315 }, { "epoch": 95.13084112149532, "grad_norm": 2.453125, "learning_rate": 2.7327344373392037e-05, "loss": 0.1617, "step": 15316 }, { "epoch": 95.13707165109034, "grad_norm": 1.8203125, "learning_rate": 2.7324891299008985e-05, "loss": 0.1048, "step": 15317 }, { "epoch": 95.14330218068535, "grad_norm": 3.015625, "learning_rate": 2.7322438202046135e-05, "loss": 0.3016, "step": 15318 }, { "epoch": 95.14953271028037, "grad_norm": 1.4609375, "learning_rate": 2.7319985082527292e-05, "loss": 0.1342, "step": 15319 }, { "epoch": 95.15576323987538, "grad_norm": 0.96875, "learning_rate": 2.73175319404763e-05, "loss": 0.1044, "step": 15320 }, { "epoch": 95.1619937694704, "grad_norm": 1.90625, "learning_rate": 2.7315078775916968e-05, "loss": 0.1519, "step": 15321 }, { "epoch": 95.16822429906541, "grad_norm": 2.703125, "learning_rate": 2.7312625588873126e-05, "loss": 0.1962, "step": 15322 }, { "epoch": 95.17445482866043, "grad_norm": 1.2890625, "learning_rate": 2.7310172379368603e-05, "loss": 0.1291, "step": 15323 }, { "epoch": 95.18068535825545, "grad_norm": 1.5234375, "learning_rate": 2.7307719147427224e-05, "loss": 0.1293, "step": 15324 }, { "epoch": 95.18691588785046, "grad_norm": 1.8359375, "learning_rate": 2.7305265893072822e-05, "loss": 0.1273, "step": 15325 }, { "epoch": 95.19314641744548, "grad_norm": 2.21875, "learning_rate": 2.730281261632921e-05, "loss": 0.1587, "step": 15326 }, { "epoch": 95.19937694704049, "grad_norm": 2.203125, "learning_rate": 2.730035931722022e-05, "loss": 0.1831, "step": 15327 }, { "epoch": 95.20560747663552, "grad_norm": 1.7578125, "learning_rate": 2.7297905995769685e-05, "loss": 0.1475, "step": 15328 }, { "epoch": 95.21183800623054, "grad_norm": 1.6640625, "learning_rate": 2.729545265200143e-05, "loss": 0.2164, "step": 15329 }, { "epoch": 95.21806853582555, "grad_norm": 2.546875, "learning_rate": 2.7292999285939273e-05, "loss": 0.1157, "step": 15330 }, { "epoch": 95.22429906542057, "grad_norm": 2.5, "learning_rate": 2.729054589760705e-05, "loss": 0.1825, "step": 15331 }, { "epoch": 95.23052959501558, "grad_norm": 2.4375, "learning_rate": 2.7288092487028595e-05, "loss": 0.1983, "step": 15332 }, { "epoch": 95.2367601246106, "grad_norm": 3.15625, "learning_rate": 2.728563905422772e-05, "loss": 0.2999, "step": 15333 }, { "epoch": 95.24299065420561, "grad_norm": 2.078125, "learning_rate": 2.728318559922826e-05, "loss": 0.1498, "step": 15334 }, { "epoch": 95.24922118380063, "grad_norm": 2.78125, "learning_rate": 2.7280732122054055e-05, "loss": 0.2553, "step": 15335 }, { "epoch": 95.25545171339564, "grad_norm": 2.5, "learning_rate": 2.7278278622728915e-05, "loss": 0.1956, "step": 15336 }, { "epoch": 95.26168224299066, "grad_norm": 1.203125, "learning_rate": 2.7275825101276676e-05, "loss": 0.0835, "step": 15337 }, { "epoch": 95.26791277258567, "grad_norm": 2.40625, "learning_rate": 2.7273371557721176e-05, "loss": 0.1707, "step": 15338 }, { "epoch": 95.27414330218069, "grad_norm": 2.65625, "learning_rate": 2.727091799208623e-05, "loss": 0.1907, "step": 15339 }, { "epoch": 95.2803738317757, "grad_norm": 1.6953125, "learning_rate": 2.7268464404395684e-05, "loss": 0.1256, "step": 15340 }, { "epoch": 95.28660436137072, "grad_norm": 2.203125, "learning_rate": 2.7266010794673352e-05, "loss": 0.1554, "step": 15341 }, { "epoch": 95.29283489096574, "grad_norm": 3.390625, "learning_rate": 2.726355716294307e-05, "loss": 0.2406, "step": 15342 }, { "epoch": 95.29906542056075, "grad_norm": 1.6328125, "learning_rate": 2.726110350922867e-05, "loss": 0.1345, "step": 15343 }, { "epoch": 95.30529595015577, "grad_norm": 1.1640625, "learning_rate": 2.7258649833553984e-05, "loss": 0.0972, "step": 15344 }, { "epoch": 95.31152647975078, "grad_norm": 3.046875, "learning_rate": 2.7256196135942836e-05, "loss": 0.2515, "step": 15345 }, { "epoch": 95.3177570093458, "grad_norm": 2.40625, "learning_rate": 2.7253742416419054e-05, "loss": 0.1846, "step": 15346 }, { "epoch": 95.32398753894081, "grad_norm": 1.6015625, "learning_rate": 2.7251288675006487e-05, "loss": 0.1614, "step": 15347 }, { "epoch": 95.33021806853583, "grad_norm": 2.671875, "learning_rate": 2.7248834911728943e-05, "loss": 0.2165, "step": 15348 }, { "epoch": 95.33644859813084, "grad_norm": 1.921875, "learning_rate": 2.7246381126610272e-05, "loss": 0.1449, "step": 15349 }, { "epoch": 95.34267912772586, "grad_norm": 1.859375, "learning_rate": 2.724392731967429e-05, "loss": 0.176, "step": 15350 }, { "epoch": 95.34890965732087, "grad_norm": 2.90625, "learning_rate": 2.7241473490944853e-05, "loss": 0.2188, "step": 15351 }, { "epoch": 95.35514018691589, "grad_norm": 1.125, "learning_rate": 2.7239019640445762e-05, "loss": 0.104, "step": 15352 }, { "epoch": 95.3613707165109, "grad_norm": 1.7421875, "learning_rate": 2.7236565768200872e-05, "loss": 0.1303, "step": 15353 }, { "epoch": 95.36760124610592, "grad_norm": 2.1875, "learning_rate": 2.723411187423401e-05, "loss": 0.095, "step": 15354 }, { "epoch": 95.37383177570094, "grad_norm": 2.703125, "learning_rate": 2.7231657958569e-05, "loss": 0.2245, "step": 15355 }, { "epoch": 95.38006230529595, "grad_norm": 2.65625, "learning_rate": 2.7229204021229686e-05, "loss": 0.165, "step": 15356 }, { "epoch": 95.38629283489097, "grad_norm": 2.46875, "learning_rate": 2.7226750062239898e-05, "loss": 0.2127, "step": 15357 }, { "epoch": 95.39252336448598, "grad_norm": 2.65625, "learning_rate": 2.7224296081623468e-05, "loss": 0.2007, "step": 15358 }, { "epoch": 95.398753894081, "grad_norm": 2.375, "learning_rate": 2.7221842079404225e-05, "loss": 0.1849, "step": 15359 }, { "epoch": 95.40498442367601, "grad_norm": 2.03125, "learning_rate": 2.7219388055606015e-05, "loss": 0.1283, "step": 15360 }, { "epoch": 95.41121495327103, "grad_norm": 2.265625, "learning_rate": 2.721693401025266e-05, "loss": 0.1703, "step": 15361 }, { "epoch": 95.41744548286604, "grad_norm": 1.7578125, "learning_rate": 2.7214479943367998e-05, "loss": 0.1749, "step": 15362 }, { "epoch": 95.42367601246106, "grad_norm": 1.9765625, "learning_rate": 2.7212025854975865e-05, "loss": 0.1062, "step": 15363 }, { "epoch": 95.42990654205607, "grad_norm": 2.359375, "learning_rate": 2.72095717451001e-05, "loss": 0.1792, "step": 15364 }, { "epoch": 95.43613707165109, "grad_norm": 1.34375, "learning_rate": 2.720711761376452e-05, "loss": 0.1175, "step": 15365 }, { "epoch": 95.4423676012461, "grad_norm": 2.453125, "learning_rate": 2.7204663460992985e-05, "loss": 0.1941, "step": 15366 }, { "epoch": 95.44859813084112, "grad_norm": 2.1875, "learning_rate": 2.720220928680931e-05, "loss": 0.1684, "step": 15367 }, { "epoch": 95.45482866043614, "grad_norm": 2.59375, "learning_rate": 2.7199755091237344e-05, "loss": 0.2022, "step": 15368 }, { "epoch": 95.46105919003115, "grad_norm": 2.671875, "learning_rate": 2.7197300874300914e-05, "loss": 0.2275, "step": 15369 }, { "epoch": 95.46728971962617, "grad_norm": 1.4296875, "learning_rate": 2.7194846636023862e-05, "loss": 0.1255, "step": 15370 }, { "epoch": 95.47352024922118, "grad_norm": 3.0, "learning_rate": 2.7192392376430016e-05, "loss": 0.126, "step": 15371 }, { "epoch": 95.4797507788162, "grad_norm": 2.671875, "learning_rate": 2.7189938095543215e-05, "loss": 0.222, "step": 15372 }, { "epoch": 95.48598130841121, "grad_norm": 2.140625, "learning_rate": 2.718748379338731e-05, "loss": 0.2118, "step": 15373 }, { "epoch": 95.49221183800623, "grad_norm": 2.359375, "learning_rate": 2.718502946998611e-05, "loss": 0.1963, "step": 15374 }, { "epoch": 95.49844236760124, "grad_norm": 2.5625, "learning_rate": 2.7182575125363473e-05, "loss": 0.2197, "step": 15375 }, { "epoch": 95.50467289719626, "grad_norm": 1.609375, "learning_rate": 2.718012075954323e-05, "loss": 0.1116, "step": 15376 }, { "epoch": 95.51090342679127, "grad_norm": 1.15625, "learning_rate": 2.717766637254922e-05, "loss": 0.1077, "step": 15377 }, { "epoch": 95.51713395638629, "grad_norm": 1.8046875, "learning_rate": 2.717521196440528e-05, "loss": 0.1625, "step": 15378 }, { "epoch": 95.5233644859813, "grad_norm": 1.9765625, "learning_rate": 2.7172757535135247e-05, "loss": 0.1728, "step": 15379 }, { "epoch": 95.52959501557632, "grad_norm": 1.8046875, "learning_rate": 2.7170303084762954e-05, "loss": 0.125, "step": 15380 }, { "epoch": 95.53582554517133, "grad_norm": 3.15625, "learning_rate": 2.7167848613312243e-05, "loss": 0.2428, "step": 15381 }, { "epoch": 95.54205607476635, "grad_norm": 2.09375, "learning_rate": 2.716539412080696e-05, "loss": 0.1188, "step": 15382 }, { "epoch": 95.54828660436137, "grad_norm": 1.4765625, "learning_rate": 2.7162939607270928e-05, "loss": 0.1328, "step": 15383 }, { "epoch": 95.55451713395638, "grad_norm": 2.28125, "learning_rate": 2.7160485072727998e-05, "loss": 0.1631, "step": 15384 }, { "epoch": 95.5607476635514, "grad_norm": 3.296875, "learning_rate": 2.7158030517202e-05, "loss": 0.1279, "step": 15385 }, { "epoch": 95.56697819314641, "grad_norm": 2.96875, "learning_rate": 2.7155575940716786e-05, "loss": 0.1577, "step": 15386 }, { "epoch": 95.57320872274143, "grad_norm": 1.9453125, "learning_rate": 2.7153121343296186e-05, "loss": 0.138, "step": 15387 }, { "epoch": 95.57943925233644, "grad_norm": 2.765625, "learning_rate": 2.7150666724964035e-05, "loss": 0.1942, "step": 15388 }, { "epoch": 95.58566978193146, "grad_norm": 3.984375, "learning_rate": 2.7148212085744185e-05, "loss": 0.1214, "step": 15389 }, { "epoch": 95.59190031152647, "grad_norm": 4.09375, "learning_rate": 2.7145757425660472e-05, "loss": 0.1344, "step": 15390 }, { "epoch": 95.59813084112149, "grad_norm": 3.078125, "learning_rate": 2.714330274473673e-05, "loss": 0.1945, "step": 15391 }, { "epoch": 95.60436137071652, "grad_norm": 2.96875, "learning_rate": 2.7140848042996803e-05, "loss": 0.098, "step": 15392 }, { "epoch": 95.61059190031153, "grad_norm": 2.359375, "learning_rate": 2.7138393320464533e-05, "loss": 0.1768, "step": 15393 }, { "epoch": 95.61682242990655, "grad_norm": 2.65625, "learning_rate": 2.7135938577163754e-05, "loss": 0.1522, "step": 15394 }, { "epoch": 95.62305295950156, "grad_norm": 3.71875, "learning_rate": 2.7133483813118328e-05, "loss": 0.1394, "step": 15395 }, { "epoch": 95.62928348909658, "grad_norm": 4.28125, "learning_rate": 2.713102902835206e-05, "loss": 0.2736, "step": 15396 }, { "epoch": 95.6355140186916, "grad_norm": 2.6875, "learning_rate": 2.712857422288883e-05, "loss": 0.1918, "step": 15397 }, { "epoch": 95.64174454828661, "grad_norm": 2.171875, "learning_rate": 2.712611939675245e-05, "loss": 0.1559, "step": 15398 }, { "epoch": 95.64797507788163, "grad_norm": 3.078125, "learning_rate": 2.712366454996678e-05, "loss": 0.1168, "step": 15399 }, { "epoch": 95.65420560747664, "grad_norm": 4.125, "learning_rate": 2.7121209682555647e-05, "loss": 0.247, "step": 15400 }, { "epoch": 95.66043613707166, "grad_norm": 2.015625, "learning_rate": 2.711875479454291e-05, "loss": 0.1731, "step": 15401 }, { "epoch": 95.66666666666667, "grad_norm": 1.4609375, "learning_rate": 2.71162998859524e-05, "loss": 0.131, "step": 15402 }, { "epoch": 95.67289719626169, "grad_norm": 1.8125, "learning_rate": 2.7113844956807965e-05, "loss": 0.1539, "step": 15403 }, { "epoch": 95.6791277258567, "grad_norm": 0.890625, "learning_rate": 2.7111390007133446e-05, "loss": 0.0973, "step": 15404 }, { "epoch": 95.68535825545172, "grad_norm": 2.796875, "learning_rate": 2.7108935036952683e-05, "loss": 0.1562, "step": 15405 }, { "epoch": 95.69158878504673, "grad_norm": 4.03125, "learning_rate": 2.7106480046289525e-05, "loss": 0.2244, "step": 15406 }, { "epoch": 95.69781931464175, "grad_norm": 3.90625, "learning_rate": 2.71040250351678e-05, "loss": 0.1295, "step": 15407 }, { "epoch": 95.70404984423676, "grad_norm": 3.578125, "learning_rate": 2.7101570003611375e-05, "loss": 0.1812, "step": 15408 }, { "epoch": 95.71028037383178, "grad_norm": 2.46875, "learning_rate": 2.7099114951644072e-05, "loss": 0.1798, "step": 15409 }, { "epoch": 95.7165109034268, "grad_norm": 2.3125, "learning_rate": 2.7096659879289753e-05, "loss": 0.1916, "step": 15410 }, { "epoch": 95.72274143302181, "grad_norm": 2.1875, "learning_rate": 2.7094204786572254e-05, "loss": 0.2104, "step": 15411 }, { "epoch": 95.72897196261682, "grad_norm": 3.6875, "learning_rate": 2.709174967351542e-05, "loss": 0.2555, "step": 15412 }, { "epoch": 95.73520249221184, "grad_norm": 1.03125, "learning_rate": 2.7089294540143083e-05, "loss": 0.0728, "step": 15413 }, { "epoch": 95.74143302180686, "grad_norm": 1.71875, "learning_rate": 2.7086839386479117e-05, "loss": 0.1121, "step": 15414 }, { "epoch": 95.74766355140187, "grad_norm": 2.328125, "learning_rate": 2.7084384212547337e-05, "loss": 0.1783, "step": 15415 }, { "epoch": 95.75389408099689, "grad_norm": 1.171875, "learning_rate": 2.708192901837161e-05, "loss": 0.1229, "step": 15416 }, { "epoch": 95.7601246105919, "grad_norm": 2.25, "learning_rate": 2.707947380397577e-05, "loss": 0.1524, "step": 15417 }, { "epoch": 95.76635514018692, "grad_norm": 3.140625, "learning_rate": 2.707701856938366e-05, "loss": 0.1842, "step": 15418 }, { "epoch": 95.77258566978193, "grad_norm": 1.2109375, "learning_rate": 2.7074563314619135e-05, "loss": 0.1355, "step": 15419 }, { "epoch": 95.77881619937695, "grad_norm": 1.7890625, "learning_rate": 2.7072108039706028e-05, "loss": 0.1117, "step": 15420 }, { "epoch": 95.78504672897196, "grad_norm": 1.1953125, "learning_rate": 2.7069652744668207e-05, "loss": 0.1108, "step": 15421 }, { "epoch": 95.79127725856698, "grad_norm": 2.5, "learning_rate": 2.7067197429529495e-05, "loss": 0.1867, "step": 15422 }, { "epoch": 95.797507788162, "grad_norm": 2.203125, "learning_rate": 2.706474209431375e-05, "loss": 0.1799, "step": 15423 }, { "epoch": 95.80373831775701, "grad_norm": 2.203125, "learning_rate": 2.7062286739044824e-05, "loss": 0.1606, "step": 15424 }, { "epoch": 95.80996884735202, "grad_norm": 1.2421875, "learning_rate": 2.7059831363746545e-05, "loss": 0.1115, "step": 15425 }, { "epoch": 95.81619937694704, "grad_norm": 2.015625, "learning_rate": 2.7057375968442782e-05, "loss": 0.1525, "step": 15426 }, { "epoch": 95.82242990654206, "grad_norm": 1.671875, "learning_rate": 2.705492055315737e-05, "loss": 0.1433, "step": 15427 }, { "epoch": 95.82866043613707, "grad_norm": 1.0703125, "learning_rate": 2.7052465117914162e-05, "loss": 0.1038, "step": 15428 }, { "epoch": 95.83489096573209, "grad_norm": 2.03125, "learning_rate": 2.705000966273699e-05, "loss": 0.1448, "step": 15429 }, { "epoch": 95.8411214953271, "grad_norm": 1.75, "learning_rate": 2.7047554187649727e-05, "loss": 0.165, "step": 15430 }, { "epoch": 95.84735202492212, "grad_norm": 2.0, "learning_rate": 2.7045098692676203e-05, "loss": 0.1686, "step": 15431 }, { "epoch": 95.85358255451713, "grad_norm": 3.296875, "learning_rate": 2.704264317784027e-05, "loss": 0.2208, "step": 15432 }, { "epoch": 95.85981308411215, "grad_norm": 2.484375, "learning_rate": 2.704018764316578e-05, "loss": 0.1684, "step": 15433 }, { "epoch": 95.86604361370716, "grad_norm": 1.5, "learning_rate": 2.7037732088676582e-05, "loss": 0.1169, "step": 15434 }, { "epoch": 95.87227414330218, "grad_norm": 1.53125, "learning_rate": 2.703527651439652e-05, "loss": 0.1265, "step": 15435 }, { "epoch": 95.8785046728972, "grad_norm": 3.15625, "learning_rate": 2.7032820920349444e-05, "loss": 0.2028, "step": 15436 }, { "epoch": 95.88473520249221, "grad_norm": 2.9375, "learning_rate": 2.7030365306559207e-05, "loss": 0.1809, "step": 15437 }, { "epoch": 95.89096573208722, "grad_norm": 1.8125, "learning_rate": 2.7027909673049663e-05, "loss": 0.1612, "step": 15438 }, { "epoch": 95.89719626168224, "grad_norm": 2.75, "learning_rate": 2.702545401984465e-05, "loss": 0.179, "step": 15439 }, { "epoch": 95.90342679127725, "grad_norm": 2.5625, "learning_rate": 2.702299834696802e-05, "loss": 0.1839, "step": 15440 }, { "epoch": 95.90965732087227, "grad_norm": 2.015625, "learning_rate": 2.7020542654443627e-05, "loss": 0.1669, "step": 15441 }, { "epoch": 95.91588785046729, "grad_norm": 2.59375, "learning_rate": 2.7018086942295317e-05, "loss": 0.1928, "step": 15442 }, { "epoch": 95.9221183800623, "grad_norm": 2.265625, "learning_rate": 2.701563121054695e-05, "loss": 0.1648, "step": 15443 }, { "epoch": 95.92834890965732, "grad_norm": 3.0, "learning_rate": 2.701317545922236e-05, "loss": 0.1332, "step": 15444 }, { "epoch": 95.93457943925233, "grad_norm": 1.6484375, "learning_rate": 2.7010719688345415e-05, "loss": 0.1163, "step": 15445 }, { "epoch": 95.94080996884735, "grad_norm": 2.0625, "learning_rate": 2.7008263897939955e-05, "loss": 0.247, "step": 15446 }, { "epoch": 95.94704049844236, "grad_norm": 1.9921875, "learning_rate": 2.7005808088029844e-05, "loss": 0.1598, "step": 15447 }, { "epoch": 95.95327102803738, "grad_norm": 1.234375, "learning_rate": 2.700335225863891e-05, "loss": 0.1128, "step": 15448 }, { "epoch": 95.9595015576324, "grad_norm": 0.98828125, "learning_rate": 2.7000896409791033e-05, "loss": 0.1123, "step": 15449 }, { "epoch": 95.96573208722741, "grad_norm": 3.265625, "learning_rate": 2.6998440541510035e-05, "loss": 0.3369, "step": 15450 }, { "epoch": 95.97196261682242, "grad_norm": 2.375, "learning_rate": 2.699598465381979e-05, "loss": 0.0953, "step": 15451 }, { "epoch": 95.97819314641744, "grad_norm": 3.15625, "learning_rate": 2.6993528746744146e-05, "loss": 0.2223, "step": 15452 }, { "epoch": 95.98442367601245, "grad_norm": 3.21875, "learning_rate": 2.6991072820306947e-05, "loss": 0.1706, "step": 15453 }, { "epoch": 95.99065420560747, "grad_norm": 1.75, "learning_rate": 2.6988616874532053e-05, "loss": 0.1459, "step": 15454 }, { "epoch": 95.99688473520249, "grad_norm": 2.25, "learning_rate": 2.698616090944331e-05, "loss": 0.1647, "step": 15455 }, { "epoch": 96.0, "grad_norm": 1.2421875, "learning_rate": 2.6983704925064585e-05, "loss": 0.0695, "step": 15456 }, { "epoch": 96.00623052959502, "grad_norm": 3.40625, "learning_rate": 2.698124892141971e-05, "loss": 0.2059, "step": 15457 }, { "epoch": 96.01246105919003, "grad_norm": 3.375, "learning_rate": 2.6978792898532555e-05, "loss": 0.1258, "step": 15458 }, { "epoch": 96.01869158878505, "grad_norm": 3.359375, "learning_rate": 2.697633685642697e-05, "loss": 0.1363, "step": 15459 }, { "epoch": 96.02492211838006, "grad_norm": 1.4921875, "learning_rate": 2.6973880795126803e-05, "loss": 0.1047, "step": 15460 }, { "epoch": 96.03115264797508, "grad_norm": 1.703125, "learning_rate": 2.6971424714655914e-05, "loss": 0.2074, "step": 15461 }, { "epoch": 96.03738317757009, "grad_norm": 2.578125, "learning_rate": 2.696896861503816e-05, "loss": 0.2303, "step": 15462 }, { "epoch": 96.04361370716511, "grad_norm": 2.96875, "learning_rate": 2.6966512496297376e-05, "loss": 0.2313, "step": 15463 }, { "epoch": 96.04984423676012, "grad_norm": 2.734375, "learning_rate": 2.6964056358457436e-05, "loss": 0.1292, "step": 15464 }, { "epoch": 96.05607476635514, "grad_norm": 3.203125, "learning_rate": 2.6961600201542185e-05, "loss": 0.2221, "step": 15465 }, { "epoch": 96.06230529595015, "grad_norm": 3.125, "learning_rate": 2.6959144025575483e-05, "loss": 0.1675, "step": 15466 }, { "epoch": 96.06853582554517, "grad_norm": 2.609375, "learning_rate": 2.6956687830581186e-05, "loss": 0.172, "step": 15467 }, { "epoch": 96.07476635514018, "grad_norm": 2.78125, "learning_rate": 2.695423161658314e-05, "loss": 0.3281, "step": 15468 }, { "epoch": 96.0809968847352, "grad_norm": 2.625, "learning_rate": 2.6951775383605216e-05, "loss": 0.114, "step": 15469 }, { "epoch": 96.08722741433021, "grad_norm": 1.2578125, "learning_rate": 2.6949319131671243e-05, "loss": 0.1052, "step": 15470 }, { "epoch": 96.09345794392523, "grad_norm": 1.203125, "learning_rate": 2.6946862860805104e-05, "loss": 0.1078, "step": 15471 }, { "epoch": 96.09968847352025, "grad_norm": 1.3984375, "learning_rate": 2.6944406571030638e-05, "loss": 0.1325, "step": 15472 }, { "epoch": 96.10591900311526, "grad_norm": 1.5390625, "learning_rate": 2.6941950262371712e-05, "loss": 0.1039, "step": 15473 }, { "epoch": 96.11214953271028, "grad_norm": 2.53125, "learning_rate": 2.6939493934852177e-05, "loss": 0.0917, "step": 15474 }, { "epoch": 96.11838006230529, "grad_norm": 2.8125, "learning_rate": 2.693703758849589e-05, "loss": 0.2727, "step": 15475 }, { "epoch": 96.1246105919003, "grad_norm": 2.34375, "learning_rate": 2.6934581223326704e-05, "loss": 0.1807, "step": 15476 }, { "epoch": 96.13084112149532, "grad_norm": 1.7578125, "learning_rate": 2.6932124839368473e-05, "loss": 0.2347, "step": 15477 }, { "epoch": 96.13707165109034, "grad_norm": 2.640625, "learning_rate": 2.692966843664507e-05, "loss": 0.1208, "step": 15478 }, { "epoch": 96.14330218068535, "grad_norm": 2.625, "learning_rate": 2.692721201518033e-05, "loss": 0.1808, "step": 15479 }, { "epoch": 96.14953271028037, "grad_norm": 2.453125, "learning_rate": 2.692475557499813e-05, "loss": 0.1512, "step": 15480 }, { "epoch": 96.15576323987538, "grad_norm": 2.15625, "learning_rate": 2.6922299116122317e-05, "loss": 0.1104, "step": 15481 }, { "epoch": 96.1619937694704, "grad_norm": 3.375, "learning_rate": 2.6919842638576754e-05, "loss": 0.1455, "step": 15482 }, { "epoch": 96.16822429906541, "grad_norm": 2.265625, "learning_rate": 2.6917386142385286e-05, "loss": 0.1751, "step": 15483 }, { "epoch": 96.17445482866043, "grad_norm": 3.09375, "learning_rate": 2.6914929627571794e-05, "loss": 0.1937, "step": 15484 }, { "epoch": 96.18068535825545, "grad_norm": 1.8046875, "learning_rate": 2.6912473094160106e-05, "loss": 0.1297, "step": 15485 }, { "epoch": 96.18691588785046, "grad_norm": 3.859375, "learning_rate": 2.691001654217411e-05, "loss": 0.1246, "step": 15486 }, { "epoch": 96.19314641744548, "grad_norm": 3.453125, "learning_rate": 2.6907559971637648e-05, "loss": 0.2148, "step": 15487 }, { "epoch": 96.19937694704049, "grad_norm": 1.8125, "learning_rate": 2.690510338257458e-05, "loss": 0.158, "step": 15488 }, { "epoch": 96.20560747663552, "grad_norm": 2.0, "learning_rate": 2.690264677500877e-05, "loss": 0.1661, "step": 15489 }, { "epoch": 96.21183800623054, "grad_norm": 2.65625, "learning_rate": 2.6900190148964072e-05, "loss": 0.2048, "step": 15490 }, { "epoch": 96.21806853582555, "grad_norm": 2.484375, "learning_rate": 2.6897733504464355e-05, "loss": 0.1821, "step": 15491 }, { "epoch": 96.22429906542057, "grad_norm": 2.015625, "learning_rate": 2.689527684153346e-05, "loss": 0.204, "step": 15492 }, { "epoch": 96.23052959501558, "grad_norm": 1.7265625, "learning_rate": 2.6892820160195265e-05, "loss": 0.1243, "step": 15493 }, { "epoch": 96.2367601246106, "grad_norm": 3.03125, "learning_rate": 2.6890363460473617e-05, "loss": 0.4208, "step": 15494 }, { "epoch": 96.24299065420561, "grad_norm": 2.390625, "learning_rate": 2.6887906742392388e-05, "loss": 0.1235, "step": 15495 }, { "epoch": 96.24922118380063, "grad_norm": 1.3671875, "learning_rate": 2.688545000597542e-05, "loss": 0.1033, "step": 15496 }, { "epoch": 96.25545171339564, "grad_norm": 2.453125, "learning_rate": 2.6882993251246595e-05, "loss": 0.1506, "step": 15497 }, { "epoch": 96.26168224299066, "grad_norm": 2.90625, "learning_rate": 2.6880536478229757e-05, "loss": 0.1582, "step": 15498 }, { "epoch": 96.26791277258567, "grad_norm": 2.859375, "learning_rate": 2.6878079686948777e-05, "loss": 0.2451, "step": 15499 }, { "epoch": 96.27414330218069, "grad_norm": 2.78125, "learning_rate": 2.687562287742751e-05, "loss": 0.2092, "step": 15500 }, { "epoch": 96.2803738317757, "grad_norm": 2.25, "learning_rate": 2.6873166049689818e-05, "loss": 0.2491, "step": 15501 }, { "epoch": 96.28660436137072, "grad_norm": 0.88671875, "learning_rate": 2.6870709203759564e-05, "loss": 0.0891, "step": 15502 }, { "epoch": 96.29283489096574, "grad_norm": 2.15625, "learning_rate": 2.686825233966061e-05, "loss": 0.2, "step": 15503 }, { "epoch": 96.29906542056075, "grad_norm": 1.8125, "learning_rate": 2.6865795457416815e-05, "loss": 0.1532, "step": 15504 }, { "epoch": 96.30529595015577, "grad_norm": 2.84375, "learning_rate": 2.6863338557052036e-05, "loss": 0.2562, "step": 15505 }, { "epoch": 96.31152647975078, "grad_norm": 3.5, "learning_rate": 2.6860881638590147e-05, "loss": 0.32, "step": 15506 }, { "epoch": 96.3177570093458, "grad_norm": 2.5, "learning_rate": 2.6858424702055e-05, "loss": 0.1095, "step": 15507 }, { "epoch": 96.32398753894081, "grad_norm": 1.0859375, "learning_rate": 2.6855967747470463e-05, "loss": 0.0877, "step": 15508 }, { "epoch": 96.33021806853583, "grad_norm": 2.6875, "learning_rate": 2.685351077486039e-05, "loss": 0.1334, "step": 15509 }, { "epoch": 96.33644859813084, "grad_norm": 2.109375, "learning_rate": 2.6851053784248666e-05, "loss": 0.1086, "step": 15510 }, { "epoch": 96.34267912772586, "grad_norm": 2.796875, "learning_rate": 2.6848596775659117e-05, "loss": 0.2016, "step": 15511 }, { "epoch": 96.34890965732087, "grad_norm": 2.25, "learning_rate": 2.684613974911564e-05, "loss": 0.1043, "step": 15512 }, { "epoch": 96.35514018691589, "grad_norm": 1.484375, "learning_rate": 2.684368270464208e-05, "loss": 0.1388, "step": 15513 }, { "epoch": 96.3613707165109, "grad_norm": 2.015625, "learning_rate": 2.6841225642262307e-05, "loss": 0.1131, "step": 15514 }, { "epoch": 96.36760124610592, "grad_norm": 2.5625, "learning_rate": 2.6838768562000184e-05, "loss": 0.1253, "step": 15515 }, { "epoch": 96.37383177570094, "grad_norm": 2.875, "learning_rate": 2.683631146387957e-05, "loss": 0.2641, "step": 15516 }, { "epoch": 96.38006230529595, "grad_norm": 2.265625, "learning_rate": 2.6833854347924338e-05, "loss": 0.1259, "step": 15517 }, { "epoch": 96.38629283489097, "grad_norm": 1.3046875, "learning_rate": 2.6831397214158332e-05, "loss": 0.1222, "step": 15518 }, { "epoch": 96.39252336448598, "grad_norm": 2.53125, "learning_rate": 2.6828940062605444e-05, "loss": 0.2676, "step": 15519 }, { "epoch": 96.398753894081, "grad_norm": 1.5703125, "learning_rate": 2.6826482893289517e-05, "loss": 0.1375, "step": 15520 }, { "epoch": 96.40498442367601, "grad_norm": 1.9609375, "learning_rate": 2.6824025706234422e-05, "loss": 0.1483, "step": 15521 }, { "epoch": 96.41121495327103, "grad_norm": 2.921875, "learning_rate": 2.6821568501464033e-05, "loss": 0.1217, "step": 15522 }, { "epoch": 96.41744548286604, "grad_norm": 3.953125, "learning_rate": 2.6819111279002203e-05, "loss": 0.1736, "step": 15523 }, { "epoch": 96.42367601246106, "grad_norm": 1.875, "learning_rate": 2.6816654038872803e-05, "loss": 0.1428, "step": 15524 }, { "epoch": 96.42990654205607, "grad_norm": 2.59375, "learning_rate": 2.6814196781099686e-05, "loss": 0.2122, "step": 15525 }, { "epoch": 96.43613707165109, "grad_norm": 1.5625, "learning_rate": 2.681173950570674e-05, "loss": 0.1097, "step": 15526 }, { "epoch": 96.4423676012461, "grad_norm": 2.5625, "learning_rate": 2.6809282212717808e-05, "loss": 0.2, "step": 15527 }, { "epoch": 96.44859813084112, "grad_norm": 1.2578125, "learning_rate": 2.680682490215677e-05, "loss": 0.1167, "step": 15528 }, { "epoch": 96.45482866043614, "grad_norm": 3.03125, "learning_rate": 2.6804367574047483e-05, "loss": 0.2668, "step": 15529 }, { "epoch": 96.46105919003115, "grad_norm": 1.0703125, "learning_rate": 2.6801910228413824e-05, "loss": 0.1137, "step": 15530 }, { "epoch": 96.46728971962617, "grad_norm": 1.4375, "learning_rate": 2.6799452865279638e-05, "loss": 0.1301, "step": 15531 }, { "epoch": 96.47352024922118, "grad_norm": 2.625, "learning_rate": 2.679699548466883e-05, "loss": 0.1729, "step": 15532 }, { "epoch": 96.4797507788162, "grad_norm": 3.515625, "learning_rate": 2.6794538086605226e-05, "loss": 0.3054, "step": 15533 }, { "epoch": 96.48598130841121, "grad_norm": 1.5390625, "learning_rate": 2.6792080671112714e-05, "loss": 0.1238, "step": 15534 }, { "epoch": 96.49221183800623, "grad_norm": 2.5, "learning_rate": 2.6789623238215156e-05, "loss": 0.161, "step": 15535 }, { "epoch": 96.49844236760124, "grad_norm": 1.296875, "learning_rate": 2.6787165787936418e-05, "loss": 0.107, "step": 15536 }, { "epoch": 96.50467289719626, "grad_norm": 1.9140625, "learning_rate": 2.6784708320300372e-05, "loss": 0.1222, "step": 15537 }, { "epoch": 96.51090342679127, "grad_norm": 2.609375, "learning_rate": 2.678225083533088e-05, "loss": 0.1781, "step": 15538 }, { "epoch": 96.51713395638629, "grad_norm": 3.203125, "learning_rate": 2.677979333305181e-05, "loss": 0.2084, "step": 15539 }, { "epoch": 96.5233644859813, "grad_norm": 0.94921875, "learning_rate": 2.6777335813487025e-05, "loss": 0.0991, "step": 15540 }, { "epoch": 96.52959501557632, "grad_norm": 1.359375, "learning_rate": 2.677487827666041e-05, "loss": 0.0842, "step": 15541 }, { "epoch": 96.53582554517133, "grad_norm": 1.96875, "learning_rate": 2.677242072259582e-05, "loss": 0.2039, "step": 15542 }, { "epoch": 96.54205607476635, "grad_norm": 3.453125, "learning_rate": 2.676996315131713e-05, "loss": 0.2916, "step": 15543 }, { "epoch": 96.54828660436137, "grad_norm": 2.640625, "learning_rate": 2.6767505562848194e-05, "loss": 0.2238, "step": 15544 }, { "epoch": 96.55451713395638, "grad_norm": 2.265625, "learning_rate": 2.6765047957212896e-05, "loss": 0.1321, "step": 15545 }, { "epoch": 96.5607476635514, "grad_norm": 1.671875, "learning_rate": 2.67625903344351e-05, "loss": 0.1398, "step": 15546 }, { "epoch": 96.56697819314641, "grad_norm": 0.9296875, "learning_rate": 2.676013269453867e-05, "loss": 0.1077, "step": 15547 }, { "epoch": 96.57320872274143, "grad_norm": 2.140625, "learning_rate": 2.6757675037547484e-05, "loss": 0.2315, "step": 15548 }, { "epoch": 96.57943925233644, "grad_norm": 1.875, "learning_rate": 2.6755217363485406e-05, "loss": 0.1747, "step": 15549 }, { "epoch": 96.58566978193146, "grad_norm": 1.6015625, "learning_rate": 2.675275967237631e-05, "loss": 0.1367, "step": 15550 }, { "epoch": 96.59190031152647, "grad_norm": 1.578125, "learning_rate": 2.6750301964244055e-05, "loss": 0.154, "step": 15551 }, { "epoch": 96.59813084112149, "grad_norm": 2.609375, "learning_rate": 2.6747844239112524e-05, "loss": 0.2448, "step": 15552 }, { "epoch": 96.60436137071652, "grad_norm": 2.4375, "learning_rate": 2.674538649700557e-05, "loss": 0.122, "step": 15553 }, { "epoch": 96.61059190031153, "grad_norm": 1.640625, "learning_rate": 2.674292873794709e-05, "loss": 0.1523, "step": 15554 }, { "epoch": 96.61682242990655, "grad_norm": 1.0546875, "learning_rate": 2.6740470961960916e-05, "loss": 0.0988, "step": 15555 }, { "epoch": 96.62305295950156, "grad_norm": 1.6171875, "learning_rate": 2.6738013169070954e-05, "loss": 0.1321, "step": 15556 }, { "epoch": 96.62928348909658, "grad_norm": 0.984375, "learning_rate": 2.6735555359301058e-05, "loss": 0.1134, "step": 15557 }, { "epoch": 96.6355140186916, "grad_norm": 1.40625, "learning_rate": 2.6733097532675105e-05, "loss": 0.128, "step": 15558 }, { "epoch": 96.64174454828661, "grad_norm": 2.375, "learning_rate": 2.673063968921695e-05, "loss": 0.1547, "step": 15559 }, { "epoch": 96.64797507788163, "grad_norm": 2.515625, "learning_rate": 2.672818182895049e-05, "loss": 0.1212, "step": 15560 }, { "epoch": 96.65420560747664, "grad_norm": 2.078125, "learning_rate": 2.6725723951899577e-05, "loss": 0.1097, "step": 15561 }, { "epoch": 96.66043613707166, "grad_norm": 2.734375, "learning_rate": 2.6723266058088092e-05, "loss": 0.1169, "step": 15562 }, { "epoch": 96.66666666666667, "grad_norm": 3.25, "learning_rate": 2.67208081475399e-05, "loss": 0.3239, "step": 15563 }, { "epoch": 96.67289719626169, "grad_norm": 1.296875, "learning_rate": 2.6718350220278875e-05, "loss": 0.1298, "step": 15564 }, { "epoch": 96.6791277258567, "grad_norm": 2.9375, "learning_rate": 2.671589227632889e-05, "loss": 0.2244, "step": 15565 }, { "epoch": 96.68535825545172, "grad_norm": 3.046875, "learning_rate": 2.6713434315713808e-05, "loss": 0.1786, "step": 15566 }, { "epoch": 96.69158878504673, "grad_norm": 1.9609375, "learning_rate": 2.6710976338457523e-05, "loss": 0.1638, "step": 15567 }, { "epoch": 96.69781931464175, "grad_norm": 2.515625, "learning_rate": 2.670851834458388e-05, "loss": 0.2131, "step": 15568 }, { "epoch": 96.70404984423676, "grad_norm": 1.21875, "learning_rate": 2.6706060334116777e-05, "loss": 0.1049, "step": 15569 }, { "epoch": 96.71028037383178, "grad_norm": 2.203125, "learning_rate": 2.6703602307080074e-05, "loss": 0.1512, "step": 15570 }, { "epoch": 96.7165109034268, "grad_norm": 2.046875, "learning_rate": 2.670114426349764e-05, "loss": 0.148, "step": 15571 }, { "epoch": 96.72274143302181, "grad_norm": 1.8046875, "learning_rate": 2.6698686203393354e-05, "loss": 0.1632, "step": 15572 }, { "epoch": 96.72897196261682, "grad_norm": 2.078125, "learning_rate": 2.6696228126791088e-05, "loss": 0.1089, "step": 15573 }, { "epoch": 96.73520249221184, "grad_norm": 1.171875, "learning_rate": 2.6693770033714716e-05, "loss": 0.097, "step": 15574 }, { "epoch": 96.74143302180686, "grad_norm": 1.9453125, "learning_rate": 2.6691311924188108e-05, "loss": 0.1633, "step": 15575 }, { "epoch": 96.74766355140187, "grad_norm": 2.53125, "learning_rate": 2.668885379823514e-05, "loss": 0.1112, "step": 15576 }, { "epoch": 96.75389408099689, "grad_norm": 1.4375, "learning_rate": 2.6686395655879694e-05, "loss": 0.1151, "step": 15577 }, { "epoch": 96.7601246105919, "grad_norm": 1.3359375, "learning_rate": 2.6683937497145634e-05, "loss": 0.0943, "step": 15578 }, { "epoch": 96.76635514018692, "grad_norm": 3.625, "learning_rate": 2.668147932205683e-05, "loss": 0.205, "step": 15579 }, { "epoch": 96.77258566978193, "grad_norm": 2.265625, "learning_rate": 2.667902113063717e-05, "loss": 0.1796, "step": 15580 }, { "epoch": 96.77881619937695, "grad_norm": 1.96875, "learning_rate": 2.6676562922910515e-05, "loss": 0.087, "step": 15581 }, { "epoch": 96.78504672897196, "grad_norm": 2.203125, "learning_rate": 2.667410469890075e-05, "loss": 0.2153, "step": 15582 }, { "epoch": 96.79127725856698, "grad_norm": 1.015625, "learning_rate": 2.6671646458631745e-05, "loss": 0.1008, "step": 15583 }, { "epoch": 96.797507788162, "grad_norm": 1.4921875, "learning_rate": 2.666918820212737e-05, "loss": 0.1144, "step": 15584 }, { "epoch": 96.80373831775701, "grad_norm": 1.6796875, "learning_rate": 2.6666729929411515e-05, "loss": 0.1343, "step": 15585 }, { "epoch": 96.80996884735202, "grad_norm": 2.0625, "learning_rate": 2.666427164050804e-05, "loss": 0.1306, "step": 15586 }, { "epoch": 96.81619937694704, "grad_norm": 1.359375, "learning_rate": 2.6661813335440827e-05, "loss": 0.1114, "step": 15587 }, { "epoch": 96.82242990654206, "grad_norm": 2.921875, "learning_rate": 2.665935501423375e-05, "loss": 0.2554, "step": 15588 }, { "epoch": 96.82866043613707, "grad_norm": 2.421875, "learning_rate": 2.665689667691069e-05, "loss": 0.2001, "step": 15589 }, { "epoch": 96.83489096573209, "grad_norm": 3.34375, "learning_rate": 2.665443832349551e-05, "loss": 0.3765, "step": 15590 }, { "epoch": 96.8411214953271, "grad_norm": 1.9765625, "learning_rate": 2.66519799540121e-05, "loss": 0.1576, "step": 15591 }, { "epoch": 96.84735202492212, "grad_norm": 1.8984375, "learning_rate": 2.6649521568484327e-05, "loss": 0.1691, "step": 15592 }, { "epoch": 96.85358255451713, "grad_norm": 2.015625, "learning_rate": 2.6647063166936075e-05, "loss": 0.11, "step": 15593 }, { "epoch": 96.85981308411215, "grad_norm": 2.515625, "learning_rate": 2.664460474939121e-05, "loss": 0.1648, "step": 15594 }, { "epoch": 96.86604361370716, "grad_norm": 1.4921875, "learning_rate": 2.6642146315873622e-05, "loss": 0.1027, "step": 15595 }, { "epoch": 96.87227414330218, "grad_norm": 2.234375, "learning_rate": 2.6639687866407175e-05, "loss": 0.1442, "step": 15596 }, { "epoch": 96.8785046728972, "grad_norm": 1.84375, "learning_rate": 2.6637229401015757e-05, "loss": 0.1978, "step": 15597 }, { "epoch": 96.88473520249221, "grad_norm": 2.28125, "learning_rate": 2.663477091972324e-05, "loss": 0.2043, "step": 15598 }, { "epoch": 96.89096573208722, "grad_norm": 1.703125, "learning_rate": 2.6632312422553497e-05, "loss": 0.146, "step": 15599 }, { "epoch": 96.89719626168224, "grad_norm": 2.953125, "learning_rate": 2.662985390953041e-05, "loss": 0.176, "step": 15600 }, { "epoch": 96.90342679127725, "grad_norm": 2.3125, "learning_rate": 2.6627395380677855e-05, "loss": 0.1853, "step": 15601 }, { "epoch": 96.90965732087227, "grad_norm": 2.234375, "learning_rate": 2.662493683601972e-05, "loss": 0.2337, "step": 15602 }, { "epoch": 96.91588785046729, "grad_norm": 2.34375, "learning_rate": 2.662247827557986e-05, "loss": 0.229, "step": 15603 }, { "epoch": 96.9221183800623, "grad_norm": 1.2890625, "learning_rate": 2.6620019699382177e-05, "loss": 0.1054, "step": 15604 }, { "epoch": 96.92834890965732, "grad_norm": 1.8125, "learning_rate": 2.6617561107450535e-05, "loss": 0.1593, "step": 15605 }, { "epoch": 96.93457943925233, "grad_norm": 2.53125, "learning_rate": 2.6615102499808813e-05, "loss": 0.1026, "step": 15606 }, { "epoch": 96.94080996884735, "grad_norm": 1.5234375, "learning_rate": 2.6612643876480892e-05, "loss": 0.161, "step": 15607 }, { "epoch": 96.94704049844236, "grad_norm": 3.21875, "learning_rate": 2.6610185237490664e-05, "loss": 0.2916, "step": 15608 }, { "epoch": 96.95327102803738, "grad_norm": 2.203125, "learning_rate": 2.660772658286198e-05, "loss": 0.2159, "step": 15609 }, { "epoch": 96.9595015576324, "grad_norm": 1.2109375, "learning_rate": 2.6605267912618743e-05, "loss": 0.0967, "step": 15610 }, { "epoch": 96.96573208722741, "grad_norm": 0.92578125, "learning_rate": 2.6602809226784818e-05, "loss": 0.0989, "step": 15611 }, { "epoch": 96.97196261682242, "grad_norm": 1.9296875, "learning_rate": 2.6600350525384095e-05, "loss": 0.13, "step": 15612 }, { "epoch": 96.97819314641744, "grad_norm": 2.328125, "learning_rate": 2.6597891808440446e-05, "loss": 0.2832, "step": 15613 }, { "epoch": 96.98442367601245, "grad_norm": 4.0, "learning_rate": 2.6595433075977744e-05, "loss": 0.3025, "step": 15614 }, { "epoch": 96.99065420560747, "grad_norm": 1.984375, "learning_rate": 2.659297432801989e-05, "loss": 0.1233, "step": 15615 }, { "epoch": 96.99688473520249, "grad_norm": 2.0, "learning_rate": 2.659051556459074e-05, "loss": 0.1207, "step": 15616 }, { "epoch": 97.0, "grad_norm": 2.03125, "learning_rate": 2.658805678571419e-05, "loss": 0.0576, "step": 15617 }, { "epoch": 97.00623052959502, "grad_norm": 2.359375, "learning_rate": 2.6585597991414114e-05, "loss": 0.1465, "step": 15618 }, { "epoch": 97.01246105919003, "grad_norm": 1.78125, "learning_rate": 2.6583139181714395e-05, "loss": 0.1281, "step": 15619 }, { "epoch": 97.01869158878505, "grad_norm": 2.328125, "learning_rate": 2.6580680356638916e-05, "loss": 0.1793, "step": 15620 }, { "epoch": 97.02492211838006, "grad_norm": 1.21875, "learning_rate": 2.6578221516211544e-05, "loss": 0.102, "step": 15621 }, { "epoch": 97.03115264797508, "grad_norm": 2.21875, "learning_rate": 2.6575762660456177e-05, "loss": 0.1579, "step": 15622 }, { "epoch": 97.03738317757009, "grad_norm": 2.6875, "learning_rate": 2.6573303789396682e-05, "loss": 0.1893, "step": 15623 }, { "epoch": 97.04361370716511, "grad_norm": 2.015625, "learning_rate": 2.6570844903056953e-05, "loss": 0.1443, "step": 15624 }, { "epoch": 97.04984423676012, "grad_norm": 1.9140625, "learning_rate": 2.6568386001460854e-05, "loss": 0.175, "step": 15625 }, { "epoch": 97.05607476635514, "grad_norm": 1.4765625, "learning_rate": 2.656592708463228e-05, "loss": 0.1547, "step": 15626 }, { "epoch": 97.06230529595015, "grad_norm": 2.953125, "learning_rate": 2.656346815259511e-05, "loss": 0.1715, "step": 15627 }, { "epoch": 97.06853582554517, "grad_norm": 1.5, "learning_rate": 2.656100920537323e-05, "loss": 0.1438, "step": 15628 }, { "epoch": 97.07476635514018, "grad_norm": 1.6328125, "learning_rate": 2.655855024299051e-05, "loss": 0.1603, "step": 15629 }, { "epoch": 97.0809968847352, "grad_norm": 2.28125, "learning_rate": 2.6556091265470846e-05, "loss": 0.1156, "step": 15630 }, { "epoch": 97.08722741433021, "grad_norm": 1.703125, "learning_rate": 2.6553632272838108e-05, "loss": 0.1233, "step": 15631 }, { "epoch": 97.09345794392523, "grad_norm": 3.34375, "learning_rate": 2.655117326511618e-05, "loss": 0.2317, "step": 15632 }, { "epoch": 97.09968847352025, "grad_norm": 2.234375, "learning_rate": 2.6548714242328954e-05, "loss": 0.1142, "step": 15633 }, { "epoch": 97.10591900311526, "grad_norm": 3.296875, "learning_rate": 2.6546255204500297e-05, "loss": 0.4162, "step": 15634 }, { "epoch": 97.11214953271028, "grad_norm": 2.375, "learning_rate": 2.654379615165411e-05, "loss": 0.1481, "step": 15635 }, { "epoch": 97.11838006230529, "grad_norm": 2.65625, "learning_rate": 2.6541337083814256e-05, "loss": 0.1413, "step": 15636 }, { "epoch": 97.1246105919003, "grad_norm": 3.390625, "learning_rate": 2.6538878001004643e-05, "loss": 0.2453, "step": 15637 }, { "epoch": 97.13084112149532, "grad_norm": 2.5, "learning_rate": 2.6536418903249123e-05, "loss": 0.2147, "step": 15638 }, { "epoch": 97.13707165109034, "grad_norm": 1.5703125, "learning_rate": 2.65339597905716e-05, "loss": 0.1083, "step": 15639 }, { "epoch": 97.14330218068535, "grad_norm": 1.453125, "learning_rate": 2.6531500662995956e-05, "loss": 0.1198, "step": 15640 }, { "epoch": 97.14953271028037, "grad_norm": 2.328125, "learning_rate": 2.652904152054607e-05, "loss": 0.1215, "step": 15641 }, { "epoch": 97.15576323987538, "grad_norm": 1.3828125, "learning_rate": 2.6526582363245822e-05, "loss": 0.1301, "step": 15642 }, { "epoch": 97.1619937694704, "grad_norm": 2.3125, "learning_rate": 2.6524123191119116e-05, "loss": 0.2314, "step": 15643 }, { "epoch": 97.16822429906541, "grad_norm": 2.328125, "learning_rate": 2.6521664004189807e-05, "loss": 0.0903, "step": 15644 }, { "epoch": 97.17445482866043, "grad_norm": 2.046875, "learning_rate": 2.6519204802481795e-05, "loss": 0.1367, "step": 15645 }, { "epoch": 97.18068535825545, "grad_norm": 1.84375, "learning_rate": 2.6516745586018965e-05, "loss": 0.1155, "step": 15646 }, { "epoch": 97.18691588785046, "grad_norm": 1.7421875, "learning_rate": 2.65142863548252e-05, "loss": 0.1357, "step": 15647 }, { "epoch": 97.19314641744548, "grad_norm": 2.203125, "learning_rate": 2.6511827108924388e-05, "loss": 0.16, "step": 15648 }, { "epoch": 97.19937694704049, "grad_norm": 2.359375, "learning_rate": 2.6509367848340394e-05, "loss": 0.1296, "step": 15649 }, { "epoch": 97.20560747663552, "grad_norm": 3.53125, "learning_rate": 2.6506908573097133e-05, "loss": 0.1283, "step": 15650 }, { "epoch": 97.21183800623054, "grad_norm": 2.765625, "learning_rate": 2.6504449283218463e-05, "loss": 0.21, "step": 15651 }, { "epoch": 97.21806853582555, "grad_norm": 1.234375, "learning_rate": 2.6501989978728286e-05, "loss": 0.1078, "step": 15652 }, { "epoch": 97.22429906542057, "grad_norm": 2.359375, "learning_rate": 2.6499530659650484e-05, "loss": 0.2288, "step": 15653 }, { "epoch": 97.23052959501558, "grad_norm": 1.6328125, "learning_rate": 2.649707132600894e-05, "loss": 0.15, "step": 15654 }, { "epoch": 97.2367601246106, "grad_norm": 1.875, "learning_rate": 2.6494611977827532e-05, "loss": 0.1307, "step": 15655 }, { "epoch": 97.24299065420561, "grad_norm": 2.546875, "learning_rate": 2.6492152615130165e-05, "loss": 0.1405, "step": 15656 }, { "epoch": 97.24922118380063, "grad_norm": 2.96875, "learning_rate": 2.6489693237940705e-05, "loss": 0.2768, "step": 15657 }, { "epoch": 97.25545171339564, "grad_norm": 4.375, "learning_rate": 2.648723384628305e-05, "loss": 0.0994, "step": 15658 }, { "epoch": 97.26168224299066, "grad_norm": 1.5078125, "learning_rate": 2.648477444018108e-05, "loss": 0.119, "step": 15659 }, { "epoch": 97.26791277258567, "grad_norm": 1.7265625, "learning_rate": 2.648231501965869e-05, "loss": 0.121, "step": 15660 }, { "epoch": 97.27414330218069, "grad_norm": 2.3125, "learning_rate": 2.647985558473976e-05, "loss": 0.2475, "step": 15661 }, { "epoch": 97.2803738317757, "grad_norm": 2.8125, "learning_rate": 2.647739613544817e-05, "loss": 0.1576, "step": 15662 }, { "epoch": 97.28660436137072, "grad_norm": 1.5, "learning_rate": 2.6474936671807822e-05, "loss": 0.1165, "step": 15663 }, { "epoch": 97.29283489096574, "grad_norm": 3.046875, "learning_rate": 2.6472477193842583e-05, "loss": 0.2339, "step": 15664 }, { "epoch": 97.29906542056075, "grad_norm": 3.5, "learning_rate": 2.6470017701576356e-05, "loss": 0.3859, "step": 15665 }, { "epoch": 97.30529595015577, "grad_norm": 2.34375, "learning_rate": 2.6467558195033028e-05, "loss": 0.1404, "step": 15666 }, { "epoch": 97.31152647975078, "grad_norm": 2.5, "learning_rate": 2.6465098674236477e-05, "loss": 0.1679, "step": 15667 }, { "epoch": 97.3177570093458, "grad_norm": 2.609375, "learning_rate": 2.64626391392106e-05, "loss": 0.18, "step": 15668 }, { "epoch": 97.32398753894081, "grad_norm": 1.8671875, "learning_rate": 2.646017958997927e-05, "loss": 0.1243, "step": 15669 }, { "epoch": 97.33021806853583, "grad_norm": 1.6953125, "learning_rate": 2.645772002656639e-05, "loss": 0.1273, "step": 15670 }, { "epoch": 97.33644859813084, "grad_norm": 1.6953125, "learning_rate": 2.645526044899584e-05, "loss": 0.1167, "step": 15671 }, { "epoch": 97.34267912772586, "grad_norm": 2.03125, "learning_rate": 2.6452800857291514e-05, "loss": 0.1421, "step": 15672 }, { "epoch": 97.34890965732087, "grad_norm": 2.703125, "learning_rate": 2.645034125147729e-05, "loss": 0.2732, "step": 15673 }, { "epoch": 97.35514018691589, "grad_norm": 2.265625, "learning_rate": 2.644788163157706e-05, "loss": 0.1704, "step": 15674 }, { "epoch": 97.3613707165109, "grad_norm": 2.140625, "learning_rate": 2.644542199761472e-05, "loss": 0.2329, "step": 15675 }, { "epoch": 97.36760124610592, "grad_norm": 2.0625, "learning_rate": 2.6442962349614154e-05, "loss": 0.1135, "step": 15676 }, { "epoch": 97.37383177570094, "grad_norm": 3.828125, "learning_rate": 2.644050268759924e-05, "loss": 0.1709, "step": 15677 }, { "epoch": 97.38006230529595, "grad_norm": 3.1875, "learning_rate": 2.643804301159389e-05, "loss": 0.3793, "step": 15678 }, { "epoch": 97.38629283489097, "grad_norm": 3.53125, "learning_rate": 2.6435583321621966e-05, "loss": 0.2781, "step": 15679 }, { "epoch": 97.39252336448598, "grad_norm": 1.21875, "learning_rate": 2.643312361770737e-05, "loss": 0.0966, "step": 15680 }, { "epoch": 97.398753894081, "grad_norm": 1.8359375, "learning_rate": 2.6430663899873996e-05, "loss": 0.1386, "step": 15681 }, { "epoch": 97.40498442367601, "grad_norm": 1.828125, "learning_rate": 2.642820416814573e-05, "loss": 0.1176, "step": 15682 }, { "epoch": 97.41121495327103, "grad_norm": 1.890625, "learning_rate": 2.6425744422546455e-05, "loss": 0.1046, "step": 15683 }, { "epoch": 97.41744548286604, "grad_norm": 1.328125, "learning_rate": 2.6423284663100066e-05, "loss": 0.0867, "step": 15684 }, { "epoch": 97.42367601246106, "grad_norm": 1.671875, "learning_rate": 2.6420824889830458e-05, "loss": 0.1438, "step": 15685 }, { "epoch": 97.42990654205607, "grad_norm": 2.703125, "learning_rate": 2.6418365102761504e-05, "loss": 0.1739, "step": 15686 }, { "epoch": 97.43613707165109, "grad_norm": 4.3125, "learning_rate": 2.6415905301917114e-05, "loss": 0.2915, "step": 15687 }, { "epoch": 97.4423676012461, "grad_norm": 2.890625, "learning_rate": 2.6413445487321165e-05, "loss": 0.2637, "step": 15688 }, { "epoch": 97.44859813084112, "grad_norm": 2.25, "learning_rate": 2.641098565899755e-05, "loss": 0.1342, "step": 15689 }, { "epoch": 97.45482866043614, "grad_norm": 1.9453125, "learning_rate": 2.640852581697016e-05, "loss": 0.1678, "step": 15690 }, { "epoch": 97.46105919003115, "grad_norm": 2.171875, "learning_rate": 2.6406065961262894e-05, "loss": 0.1821, "step": 15691 }, { "epoch": 97.46728971962617, "grad_norm": 2.8125, "learning_rate": 2.6403606091899618e-05, "loss": 0.2789, "step": 15692 }, { "epoch": 97.47352024922118, "grad_norm": 1.5, "learning_rate": 2.6401146208904247e-05, "loss": 0.1044, "step": 15693 }, { "epoch": 97.4797507788162, "grad_norm": 1.5703125, "learning_rate": 2.639868631230067e-05, "loss": 0.1507, "step": 15694 }, { "epoch": 97.48598130841121, "grad_norm": 2.0625, "learning_rate": 2.639622640211277e-05, "loss": 0.1687, "step": 15695 }, { "epoch": 97.49221183800623, "grad_norm": 3.296875, "learning_rate": 2.6393766478364434e-05, "loss": 0.2785, "step": 15696 }, { "epoch": 97.49844236760124, "grad_norm": 2.46875, "learning_rate": 2.6391306541079557e-05, "loss": 0.2602, "step": 15697 }, { "epoch": 97.50467289719626, "grad_norm": 1.25, "learning_rate": 2.6388846590282044e-05, "loss": 0.1054, "step": 15698 }, { "epoch": 97.51090342679127, "grad_norm": 3.328125, "learning_rate": 2.638638662599577e-05, "loss": 0.352, "step": 15699 }, { "epoch": 97.51713395638629, "grad_norm": 2.140625, "learning_rate": 2.638392664824463e-05, "loss": 0.1885, "step": 15700 }, { "epoch": 97.5233644859813, "grad_norm": 2.25, "learning_rate": 2.6381466657052522e-05, "loss": 0.2311, "step": 15701 }, { "epoch": 97.52959501557632, "grad_norm": 2.9375, "learning_rate": 2.637900665244333e-05, "loss": 0.3889, "step": 15702 }, { "epoch": 97.53582554517133, "grad_norm": 2.515625, "learning_rate": 2.6376546634440945e-05, "loss": 0.159, "step": 15703 }, { "epoch": 97.54205607476635, "grad_norm": 2.359375, "learning_rate": 2.6374086603069275e-05, "loss": 0.1128, "step": 15704 }, { "epoch": 97.54828660436137, "grad_norm": 2.84375, "learning_rate": 2.637162655835219e-05, "loss": 0.1841, "step": 15705 }, { "epoch": 97.55451713395638, "grad_norm": 2.0, "learning_rate": 2.6369166500313602e-05, "loss": 0.1568, "step": 15706 }, { "epoch": 97.5607476635514, "grad_norm": 1.84375, "learning_rate": 2.6366706428977396e-05, "loss": 0.137, "step": 15707 }, { "epoch": 97.56697819314641, "grad_norm": 2.234375, "learning_rate": 2.6364246344367462e-05, "loss": 0.1341, "step": 15708 }, { "epoch": 97.57320872274143, "grad_norm": 2.0, "learning_rate": 2.6361786246507693e-05, "loss": 0.1392, "step": 15709 }, { "epoch": 97.57943925233644, "grad_norm": 2.390625, "learning_rate": 2.6359326135421986e-05, "loss": 0.1419, "step": 15710 }, { "epoch": 97.58566978193146, "grad_norm": 2.046875, "learning_rate": 2.635686601113423e-05, "loss": 0.1305, "step": 15711 }, { "epoch": 97.59190031152647, "grad_norm": 1.5703125, "learning_rate": 2.6354405873668315e-05, "loss": 0.1338, "step": 15712 }, { "epoch": 97.59813084112149, "grad_norm": 2.453125, "learning_rate": 2.6351945723048155e-05, "loss": 0.1735, "step": 15713 }, { "epoch": 97.60436137071652, "grad_norm": 2.59375, "learning_rate": 2.6349485559297608e-05, "loss": 0.1682, "step": 15714 }, { "epoch": 97.61059190031153, "grad_norm": 4.625, "learning_rate": 2.63470253824406e-05, "loss": 0.1392, "step": 15715 }, { "epoch": 97.61682242990655, "grad_norm": 2.609375, "learning_rate": 2.634456519250101e-05, "loss": 0.2384, "step": 15716 }, { "epoch": 97.62305295950156, "grad_norm": 2.140625, "learning_rate": 2.6342104989502738e-05, "loss": 0.1884, "step": 15717 }, { "epoch": 97.62928348909658, "grad_norm": 3.953125, "learning_rate": 2.633964477346967e-05, "loss": 0.1352, "step": 15718 }, { "epoch": 97.6355140186916, "grad_norm": 1.65625, "learning_rate": 2.6337184544425696e-05, "loss": 0.1484, "step": 15719 }, { "epoch": 97.64174454828661, "grad_norm": 1.7734375, "learning_rate": 2.6334724302394732e-05, "loss": 0.1469, "step": 15720 }, { "epoch": 97.64797507788163, "grad_norm": 1.3515625, "learning_rate": 2.6332264047400645e-05, "loss": 0.1197, "step": 15721 }, { "epoch": 97.65420560747664, "grad_norm": 2.046875, "learning_rate": 2.6329803779467353e-05, "loss": 0.231, "step": 15722 }, { "epoch": 97.66043613707166, "grad_norm": 2.75, "learning_rate": 2.632734349861874e-05, "loss": 0.2049, "step": 15723 }, { "epoch": 97.66666666666667, "grad_norm": 2.671875, "learning_rate": 2.6324883204878702e-05, "loss": 0.129, "step": 15724 }, { "epoch": 97.67289719626169, "grad_norm": 2.140625, "learning_rate": 2.6322422898271126e-05, "loss": 0.1785, "step": 15725 }, { "epoch": 97.6791277258567, "grad_norm": 1.703125, "learning_rate": 2.631996257881993e-05, "loss": 0.1723, "step": 15726 }, { "epoch": 97.68535825545172, "grad_norm": 1.265625, "learning_rate": 2.6317502246548976e-05, "loss": 0.0977, "step": 15727 }, { "epoch": 97.69158878504673, "grad_norm": 3.46875, "learning_rate": 2.6315041901482184e-05, "loss": 0.5277, "step": 15728 }, { "epoch": 97.69781931464175, "grad_norm": 2.125, "learning_rate": 2.6312581543643443e-05, "loss": 0.1607, "step": 15729 }, { "epoch": 97.70404984423676, "grad_norm": 1.3125, "learning_rate": 2.6310121173056652e-05, "loss": 0.1379, "step": 15730 }, { "epoch": 97.71028037383178, "grad_norm": 1.59375, "learning_rate": 2.6307660789745696e-05, "loss": 0.1261, "step": 15731 }, { "epoch": 97.7165109034268, "grad_norm": 5.15625, "learning_rate": 2.6305200393734473e-05, "loss": 0.4062, "step": 15732 }, { "epoch": 97.72274143302181, "grad_norm": 1.4453125, "learning_rate": 2.6302739985046898e-05, "loss": 0.1325, "step": 15733 }, { "epoch": 97.72897196261682, "grad_norm": 2.953125, "learning_rate": 2.6300279563706842e-05, "loss": 0.144, "step": 15734 }, { "epoch": 97.73520249221184, "grad_norm": 3.109375, "learning_rate": 2.629781912973821e-05, "loss": 0.2107, "step": 15735 }, { "epoch": 97.74143302180686, "grad_norm": 2.390625, "learning_rate": 2.62953586831649e-05, "loss": 0.1463, "step": 15736 }, { "epoch": 97.74766355140187, "grad_norm": 1.8515625, "learning_rate": 2.629289822401081e-05, "loss": 0.1003, "step": 15737 }, { "epoch": 97.75389408099689, "grad_norm": 3.59375, "learning_rate": 2.629043775229983e-05, "loss": 0.1329, "step": 15738 }, { "epoch": 97.7601246105919, "grad_norm": 2.59375, "learning_rate": 2.628797726805587e-05, "loss": 0.1312, "step": 15739 }, { "epoch": 97.76635514018692, "grad_norm": 1.7890625, "learning_rate": 2.6285516771302803e-05, "loss": 0.1172, "step": 15740 }, { "epoch": 97.77258566978193, "grad_norm": 3.578125, "learning_rate": 2.628305626206455e-05, "loss": 0.1896, "step": 15741 }, { "epoch": 97.77881619937695, "grad_norm": 2.984375, "learning_rate": 2.6280595740364994e-05, "loss": 0.1582, "step": 15742 }, { "epoch": 97.78504672897196, "grad_norm": 3.234375, "learning_rate": 2.6278135206228037e-05, "loss": 0.2095, "step": 15743 }, { "epoch": 97.79127725856698, "grad_norm": 2.40625, "learning_rate": 2.627567465967758e-05, "loss": 0.1085, "step": 15744 }, { "epoch": 97.797507788162, "grad_norm": 2.578125, "learning_rate": 2.6273214100737516e-05, "loss": 0.1443, "step": 15745 }, { "epoch": 97.80373831775701, "grad_norm": 1.9609375, "learning_rate": 2.6270753529431734e-05, "loss": 0.0997, "step": 15746 }, { "epoch": 97.80996884735202, "grad_norm": 2.0, "learning_rate": 2.626829294578414e-05, "loss": 0.1345, "step": 15747 }, { "epoch": 97.81619937694704, "grad_norm": 3.25, "learning_rate": 2.6265832349818643e-05, "loss": 0.2175, "step": 15748 }, { "epoch": 97.82242990654206, "grad_norm": 1.3828125, "learning_rate": 2.626337174155911e-05, "loss": 0.1045, "step": 15749 }, { "epoch": 97.82866043613707, "grad_norm": 2.484375, "learning_rate": 2.6260911121029473e-05, "loss": 0.2012, "step": 15750 }, { "epoch": 97.83489096573209, "grad_norm": 2.296875, "learning_rate": 2.6258450488253605e-05, "loss": 0.1587, "step": 15751 }, { "epoch": 97.8411214953271, "grad_norm": 2.59375, "learning_rate": 2.6255989843255428e-05, "loss": 0.1093, "step": 15752 }, { "epoch": 97.84735202492212, "grad_norm": 3.28125, "learning_rate": 2.6253529186058818e-05, "loss": 0.1604, "step": 15753 }, { "epoch": 97.85358255451713, "grad_norm": 2.515625, "learning_rate": 2.625106851668768e-05, "loss": 0.1383, "step": 15754 }, { "epoch": 97.85981308411215, "grad_norm": 1.4453125, "learning_rate": 2.624860783516592e-05, "loss": 0.1065, "step": 15755 }, { "epoch": 97.86604361370716, "grad_norm": 1.078125, "learning_rate": 2.624614714151743e-05, "loss": 0.0975, "step": 15756 }, { "epoch": 97.87227414330218, "grad_norm": 1.5, "learning_rate": 2.6243686435766114e-05, "loss": 0.1145, "step": 15757 }, { "epoch": 97.8785046728972, "grad_norm": 2.6875, "learning_rate": 2.6241225717935857e-05, "loss": 0.2183, "step": 15758 }, { "epoch": 97.88473520249221, "grad_norm": 1.3046875, "learning_rate": 2.6238764988050575e-05, "loss": 0.1095, "step": 15759 }, { "epoch": 97.89096573208722, "grad_norm": 4.03125, "learning_rate": 2.6236304246134153e-05, "loss": 0.3269, "step": 15760 }, { "epoch": 97.89719626168224, "grad_norm": 1.9140625, "learning_rate": 2.6233843492210507e-05, "loss": 0.1548, "step": 15761 }, { "epoch": 97.90342679127725, "grad_norm": 1.765625, "learning_rate": 2.6231382726303517e-05, "loss": 0.1036, "step": 15762 }, { "epoch": 97.90965732087227, "grad_norm": 2.25, "learning_rate": 2.6228921948437095e-05, "loss": 0.1373, "step": 15763 }, { "epoch": 97.91588785046729, "grad_norm": 1.984375, "learning_rate": 2.622646115863514e-05, "loss": 0.2094, "step": 15764 }, { "epoch": 97.9221183800623, "grad_norm": 1.59375, "learning_rate": 2.6224000356921546e-05, "loss": 0.1085, "step": 15765 }, { "epoch": 97.92834890965732, "grad_norm": 2.625, "learning_rate": 2.6221539543320217e-05, "loss": 0.1711, "step": 15766 }, { "epoch": 97.93457943925233, "grad_norm": 2.578125, "learning_rate": 2.6219078717855043e-05, "loss": 0.178, "step": 15767 }, { "epoch": 97.94080996884735, "grad_norm": 1.515625, "learning_rate": 2.621661788054995e-05, "loss": 0.1395, "step": 15768 }, { "epoch": 97.94704049844236, "grad_norm": 2.03125, "learning_rate": 2.621415703142881e-05, "loss": 0.2036, "step": 15769 }, { "epoch": 97.95327102803738, "grad_norm": 1.4140625, "learning_rate": 2.6211696170515536e-05, "loss": 0.1581, "step": 15770 }, { "epoch": 97.9595015576324, "grad_norm": 1.3203125, "learning_rate": 2.620923529783403e-05, "loss": 0.117, "step": 15771 }, { "epoch": 97.96573208722741, "grad_norm": 2.578125, "learning_rate": 2.6206774413408186e-05, "loss": 0.2143, "step": 15772 }, { "epoch": 97.97196261682242, "grad_norm": 2.15625, "learning_rate": 2.6204313517261904e-05, "loss": 0.1194, "step": 15773 }, { "epoch": 97.97819314641744, "grad_norm": 1.953125, "learning_rate": 2.6201852609419093e-05, "loss": 0.1122, "step": 15774 }, { "epoch": 97.98442367601245, "grad_norm": 2.265625, "learning_rate": 2.6199391689903647e-05, "loss": 0.1968, "step": 15775 }, { "epoch": 97.99065420560747, "grad_norm": 1.71875, "learning_rate": 2.6196930758739468e-05, "loss": 0.1541, "step": 15776 }, { "epoch": 97.99688473520249, "grad_norm": 2.984375, "learning_rate": 2.6194469815950463e-05, "loss": 0.2687, "step": 15777 }, { "epoch": 98.0, "grad_norm": 2.265625, "learning_rate": 2.6192008861560526e-05, "loss": 0.112, "step": 15778 }, { "epoch": 98.00623052959502, "grad_norm": 1.0859375, "learning_rate": 2.6189547895593562e-05, "loss": 0.1256, "step": 15779 }, { "epoch": 98.01246105919003, "grad_norm": 1.859375, "learning_rate": 2.6187086918073472e-05, "loss": 0.1064, "step": 15780 }, { "epoch": 98.01869158878505, "grad_norm": 2.71875, "learning_rate": 2.6184625929024153e-05, "loss": 0.1848, "step": 15781 }, { "epoch": 98.02492211838006, "grad_norm": 1.8515625, "learning_rate": 2.6182164928469503e-05, "loss": 0.1321, "step": 15782 }, { "epoch": 98.03115264797508, "grad_norm": 1.765625, "learning_rate": 2.6179703916433444e-05, "loss": 0.1008, "step": 15783 }, { "epoch": 98.03738317757009, "grad_norm": 1.59375, "learning_rate": 2.6177242892939862e-05, "loss": 0.1115, "step": 15784 }, { "epoch": 98.04361370716511, "grad_norm": 1.3984375, "learning_rate": 2.617478185801266e-05, "loss": 0.1199, "step": 15785 }, { "epoch": 98.04984423676012, "grad_norm": 3.921875, "learning_rate": 2.6172320811675737e-05, "loss": 0.1879, "step": 15786 }, { "epoch": 98.05607476635514, "grad_norm": 4.1875, "learning_rate": 2.6169859753953008e-05, "loss": 0.2454, "step": 15787 }, { "epoch": 98.06230529595015, "grad_norm": 2.78125, "learning_rate": 2.616739868486836e-05, "loss": 0.2223, "step": 15788 }, { "epoch": 98.06853582554517, "grad_norm": 3.046875, "learning_rate": 2.6164937604445705e-05, "loss": 0.1291, "step": 15789 }, { "epoch": 98.07476635514018, "grad_norm": 1.9140625, "learning_rate": 2.6162476512708945e-05, "loss": 0.1502, "step": 15790 }, { "epoch": 98.0809968847352, "grad_norm": 1.8046875, "learning_rate": 2.616001540968198e-05, "loss": 0.1616, "step": 15791 }, { "epoch": 98.08722741433021, "grad_norm": 3.21875, "learning_rate": 2.615755429538872e-05, "loss": 0.2185, "step": 15792 }, { "epoch": 98.09345794392523, "grad_norm": 1.9453125, "learning_rate": 2.615509316985305e-05, "loss": 0.2202, "step": 15793 }, { "epoch": 98.09968847352025, "grad_norm": 2.453125, "learning_rate": 2.6152632033098895e-05, "loss": 0.1776, "step": 15794 }, { "epoch": 98.10591900311526, "grad_norm": 2.59375, "learning_rate": 2.615017088515013e-05, "loss": 0.2503, "step": 15795 }, { "epoch": 98.11214953271028, "grad_norm": 2.125, "learning_rate": 2.61477097260307e-05, "loss": 0.2319, "step": 15796 }, { "epoch": 98.11838006230529, "grad_norm": 3.09375, "learning_rate": 2.6145248555764463e-05, "loss": 0.2431, "step": 15797 }, { "epoch": 98.1246105919003, "grad_norm": 3.328125, "learning_rate": 2.614278737437535e-05, "loss": 0.1301, "step": 15798 }, { "epoch": 98.13084112149532, "grad_norm": 1.1953125, "learning_rate": 2.614032618188726e-05, "loss": 0.102, "step": 15799 }, { "epoch": 98.13707165109034, "grad_norm": 1.71875, "learning_rate": 2.6137864978324095e-05, "loss": 0.1188, "step": 15800 }, { "epoch": 98.14330218068535, "grad_norm": 1.1640625, "learning_rate": 2.613540376370975e-05, "loss": 0.1136, "step": 15801 }, { "epoch": 98.14953271028037, "grad_norm": 1.3046875, "learning_rate": 2.6132942538068146e-05, "loss": 0.1071, "step": 15802 }, { "epoch": 98.15576323987538, "grad_norm": 3.5, "learning_rate": 2.613048130142317e-05, "loss": 0.2665, "step": 15803 }, { "epoch": 98.1619937694704, "grad_norm": 1.9609375, "learning_rate": 2.612802005379874e-05, "loss": 0.1938, "step": 15804 }, { "epoch": 98.16822429906541, "grad_norm": 2.1875, "learning_rate": 2.612555879521875e-05, "loss": 0.1846, "step": 15805 }, { "epoch": 98.17445482866043, "grad_norm": 1.8203125, "learning_rate": 2.6123097525707108e-05, "loss": 0.1559, "step": 15806 }, { "epoch": 98.18068535825545, "grad_norm": 2.15625, "learning_rate": 2.612063624528772e-05, "loss": 0.144, "step": 15807 }, { "epoch": 98.18691588785046, "grad_norm": 2.609375, "learning_rate": 2.6118174953984486e-05, "loss": 0.1817, "step": 15808 }, { "epoch": 98.19314641744548, "grad_norm": 2.328125, "learning_rate": 2.6115713651821323e-05, "loss": 0.1661, "step": 15809 }, { "epoch": 98.19937694704049, "grad_norm": 2.203125, "learning_rate": 2.6113252338822115e-05, "loss": 0.1729, "step": 15810 }, { "epoch": 98.20560747663552, "grad_norm": 3.40625, "learning_rate": 2.611079101501078e-05, "loss": 0.3649, "step": 15811 }, { "epoch": 98.21183800623054, "grad_norm": 1.375, "learning_rate": 2.610832968041122e-05, "loss": 0.1251, "step": 15812 }, { "epoch": 98.21806853582555, "grad_norm": 3.421875, "learning_rate": 2.6105868335047344e-05, "loss": 0.2917, "step": 15813 }, { "epoch": 98.22429906542057, "grad_norm": 2.859375, "learning_rate": 2.6103406978943053e-05, "loss": 0.2312, "step": 15814 }, { "epoch": 98.23052959501558, "grad_norm": 2.390625, "learning_rate": 2.6100945612122253e-05, "loss": 0.1927, "step": 15815 }, { "epoch": 98.2367601246106, "grad_norm": 1.234375, "learning_rate": 2.609848423460885e-05, "loss": 0.0991, "step": 15816 }, { "epoch": 98.24299065420561, "grad_norm": 1.546875, "learning_rate": 2.609602284642674e-05, "loss": 0.141, "step": 15817 }, { "epoch": 98.24922118380063, "grad_norm": 1.203125, "learning_rate": 2.6093561447599847e-05, "loss": 0.0865, "step": 15818 }, { "epoch": 98.25545171339564, "grad_norm": 1.9296875, "learning_rate": 2.609110003815206e-05, "loss": 0.1448, "step": 15819 }, { "epoch": 98.26168224299066, "grad_norm": 2.921875, "learning_rate": 2.6088638618107298e-05, "loss": 0.1981, "step": 15820 }, { "epoch": 98.26791277258567, "grad_norm": 1.90625, "learning_rate": 2.6086177187489453e-05, "loss": 0.1154, "step": 15821 }, { "epoch": 98.27414330218069, "grad_norm": 3.0625, "learning_rate": 2.6083715746322444e-05, "loss": 0.1279, "step": 15822 }, { "epoch": 98.2803738317757, "grad_norm": 3.984375, "learning_rate": 2.6081254294630163e-05, "loss": 0.1992, "step": 15823 }, { "epoch": 98.28660436137072, "grad_norm": 2.21875, "learning_rate": 2.6078792832436533e-05, "loss": 0.1578, "step": 15824 }, { "epoch": 98.29283489096574, "grad_norm": 2.59375, "learning_rate": 2.6076331359765448e-05, "loss": 0.1925, "step": 15825 }, { "epoch": 98.29906542056075, "grad_norm": 1.328125, "learning_rate": 2.6073869876640818e-05, "loss": 0.1111, "step": 15826 }, { "epoch": 98.30529595015577, "grad_norm": 2.953125, "learning_rate": 2.607140838308655e-05, "loss": 0.1737, "step": 15827 }, { "epoch": 98.31152647975078, "grad_norm": 2.796875, "learning_rate": 2.6068946879126547e-05, "loss": 0.2394, "step": 15828 }, { "epoch": 98.3177570093458, "grad_norm": 2.84375, "learning_rate": 2.6066485364784728e-05, "loss": 0.157, "step": 15829 }, { "epoch": 98.32398753894081, "grad_norm": 2.65625, "learning_rate": 2.6064023840084973e-05, "loss": 0.1805, "step": 15830 }, { "epoch": 98.33021806853583, "grad_norm": 2.796875, "learning_rate": 2.606156230505122e-05, "loss": 0.255, "step": 15831 }, { "epoch": 98.33644859813084, "grad_norm": 2.3125, "learning_rate": 2.605910075970735e-05, "loss": 0.1756, "step": 15832 }, { "epoch": 98.34267912772586, "grad_norm": 2.859375, "learning_rate": 2.6056639204077292e-05, "loss": 0.2522, "step": 15833 }, { "epoch": 98.34890965732087, "grad_norm": 3.21875, "learning_rate": 2.605417763818494e-05, "loss": 0.1152, "step": 15834 }, { "epoch": 98.35514018691589, "grad_norm": 2.5625, "learning_rate": 2.60517160620542e-05, "loss": 0.2227, "step": 15835 }, { "epoch": 98.3613707165109, "grad_norm": 1.3046875, "learning_rate": 2.6049254475708983e-05, "loss": 0.0932, "step": 15836 }, { "epoch": 98.36760124610592, "grad_norm": 2.40625, "learning_rate": 2.60467928791732e-05, "loss": 0.1075, "step": 15837 }, { "epoch": 98.37383177570094, "grad_norm": 2.5625, "learning_rate": 2.6044331272470755e-05, "loss": 0.1985, "step": 15838 }, { "epoch": 98.38006230529595, "grad_norm": 1.609375, "learning_rate": 2.6041869655625562e-05, "loss": 0.108, "step": 15839 }, { "epoch": 98.38629283489097, "grad_norm": 4.28125, "learning_rate": 2.6039408028661517e-05, "loss": 0.1354, "step": 15840 }, { "epoch": 98.39252336448598, "grad_norm": 3.328125, "learning_rate": 2.603694639160253e-05, "loss": 0.1831, "step": 15841 }, { "epoch": 98.398753894081, "grad_norm": 3.234375, "learning_rate": 2.603448474447252e-05, "loss": 0.1834, "step": 15842 }, { "epoch": 98.40498442367601, "grad_norm": 2.921875, "learning_rate": 2.6032023087295375e-05, "loss": 0.1514, "step": 15843 }, { "epoch": 98.41121495327103, "grad_norm": 1.8984375, "learning_rate": 2.6029561420095032e-05, "loss": 0.1448, "step": 15844 }, { "epoch": 98.41744548286604, "grad_norm": 1.984375, "learning_rate": 2.6027099742895367e-05, "loss": 0.0931, "step": 15845 }, { "epoch": 98.42367601246106, "grad_norm": 1.46875, "learning_rate": 2.6024638055720313e-05, "loss": 0.1387, "step": 15846 }, { "epoch": 98.42990654205607, "grad_norm": 1.3671875, "learning_rate": 2.602217635859377e-05, "loss": 0.0906, "step": 15847 }, { "epoch": 98.43613707165109, "grad_norm": 1.8359375, "learning_rate": 2.6019714651539646e-05, "loss": 0.138, "step": 15848 }, { "epoch": 98.4423676012461, "grad_norm": 2.984375, "learning_rate": 2.6017252934581837e-05, "loss": 0.1805, "step": 15849 }, { "epoch": 98.44859813084112, "grad_norm": 3.109375, "learning_rate": 2.6014791207744282e-05, "loss": 0.184, "step": 15850 }, { "epoch": 98.45482866043614, "grad_norm": 2.28125, "learning_rate": 2.6012329471050862e-05, "loss": 0.1744, "step": 15851 }, { "epoch": 98.46105919003115, "grad_norm": 2.203125, "learning_rate": 2.6009867724525494e-05, "loss": 0.1558, "step": 15852 }, { "epoch": 98.46728971962617, "grad_norm": 2.9375, "learning_rate": 2.6007405968192094e-05, "loss": 0.3639, "step": 15853 }, { "epoch": 98.47352024922118, "grad_norm": 2.90625, "learning_rate": 2.6004944202074566e-05, "loss": 0.2476, "step": 15854 }, { "epoch": 98.4797507788162, "grad_norm": 1.8984375, "learning_rate": 2.6002482426196816e-05, "loss": 0.151, "step": 15855 }, { "epoch": 98.48598130841121, "grad_norm": 2.703125, "learning_rate": 2.6000020640582757e-05, "loss": 0.2287, "step": 15856 }, { "epoch": 98.49221183800623, "grad_norm": 4.1875, "learning_rate": 2.5997558845256304e-05, "loss": 0.2127, "step": 15857 }, { "epoch": 98.49844236760124, "grad_norm": 1.609375, "learning_rate": 2.599509704024135e-05, "loss": 0.1452, "step": 15858 }, { "epoch": 98.50467289719626, "grad_norm": 3.359375, "learning_rate": 2.599263522556182e-05, "loss": 0.2929, "step": 15859 }, { "epoch": 98.51090342679127, "grad_norm": 2.21875, "learning_rate": 2.5990173401241618e-05, "loss": 0.1556, "step": 15860 }, { "epoch": 98.51713395638629, "grad_norm": 2.28125, "learning_rate": 2.5987711567304652e-05, "loss": 0.1679, "step": 15861 }, { "epoch": 98.5233644859813, "grad_norm": 2.125, "learning_rate": 2.598524972377484e-05, "loss": 0.1528, "step": 15862 }, { "epoch": 98.52959501557632, "grad_norm": 1.9765625, "learning_rate": 2.598278787067608e-05, "loss": 0.1774, "step": 15863 }, { "epoch": 98.53582554517133, "grad_norm": 1.7734375, "learning_rate": 2.5980326008032292e-05, "loss": 0.1552, "step": 15864 }, { "epoch": 98.54205607476635, "grad_norm": 2.390625, "learning_rate": 2.597786413586737e-05, "loss": 0.1981, "step": 15865 }, { "epoch": 98.54828660436137, "grad_norm": 1.5390625, "learning_rate": 2.597540225420525e-05, "loss": 0.1302, "step": 15866 }, { "epoch": 98.55451713395638, "grad_norm": 2.53125, "learning_rate": 2.597294036306982e-05, "loss": 0.2067, "step": 15867 }, { "epoch": 98.5607476635514, "grad_norm": 2.84375, "learning_rate": 2.5970478462485005e-05, "loss": 0.2974, "step": 15868 }, { "epoch": 98.56697819314641, "grad_norm": 1.7890625, "learning_rate": 2.5968016552474705e-05, "loss": 0.1756, "step": 15869 }, { "epoch": 98.57320872274143, "grad_norm": 2.296875, "learning_rate": 2.596555463306283e-05, "loss": 0.2031, "step": 15870 }, { "epoch": 98.57943925233644, "grad_norm": 2.3125, "learning_rate": 2.59630927042733e-05, "loss": 0.1898, "step": 15871 }, { "epoch": 98.58566978193146, "grad_norm": 2.953125, "learning_rate": 2.5960630766130024e-05, "loss": 0.2697, "step": 15872 }, { "epoch": 98.59190031152647, "grad_norm": 2.796875, "learning_rate": 2.5958168818656907e-05, "loss": 0.1982, "step": 15873 }, { "epoch": 98.59813084112149, "grad_norm": 2.25, "learning_rate": 2.5955706861877866e-05, "loss": 0.1393, "step": 15874 }, { "epoch": 98.60436137071652, "grad_norm": 2.984375, "learning_rate": 2.5953244895816808e-05, "loss": 0.2703, "step": 15875 }, { "epoch": 98.61059190031153, "grad_norm": 1.84375, "learning_rate": 2.5950782920497646e-05, "loss": 0.1643, "step": 15876 }, { "epoch": 98.61682242990655, "grad_norm": 2.09375, "learning_rate": 2.5948320935944286e-05, "loss": 0.1249, "step": 15877 }, { "epoch": 98.62305295950156, "grad_norm": 1.0546875, "learning_rate": 2.5945858942180646e-05, "loss": 0.1006, "step": 15878 }, { "epoch": 98.62928348909658, "grad_norm": 1.9375, "learning_rate": 2.5943396939230636e-05, "loss": 0.1155, "step": 15879 }, { "epoch": 98.6355140186916, "grad_norm": 3.15625, "learning_rate": 2.594093492711816e-05, "loss": 0.1277, "step": 15880 }, { "epoch": 98.64174454828661, "grad_norm": 2.453125, "learning_rate": 2.5938472905867143e-05, "loss": 0.1671, "step": 15881 }, { "epoch": 98.64797507788163, "grad_norm": 1.8203125, "learning_rate": 2.593601087550149e-05, "loss": 0.1879, "step": 15882 }, { "epoch": 98.65420560747664, "grad_norm": 1.0390625, "learning_rate": 2.5933548836045107e-05, "loss": 0.1193, "step": 15883 }, { "epoch": 98.66043613707166, "grad_norm": 1.671875, "learning_rate": 2.5931086787521908e-05, "loss": 0.1284, "step": 15884 }, { "epoch": 98.66666666666667, "grad_norm": 3.015625, "learning_rate": 2.592862472995582e-05, "loss": 0.1668, "step": 15885 }, { "epoch": 98.67289719626169, "grad_norm": 2.203125, "learning_rate": 2.5926162663370734e-05, "loss": 0.1905, "step": 15886 }, { "epoch": 98.6791277258567, "grad_norm": 3.125, "learning_rate": 2.5923700587790572e-05, "loss": 0.1823, "step": 15887 }, { "epoch": 98.68535825545172, "grad_norm": 1.703125, "learning_rate": 2.5921238503239248e-05, "loss": 0.1361, "step": 15888 }, { "epoch": 98.69158878504673, "grad_norm": 2.296875, "learning_rate": 2.5918776409740676e-05, "loss": 0.1306, "step": 15889 }, { "epoch": 98.69781931464175, "grad_norm": 2.234375, "learning_rate": 2.5916314307318755e-05, "loss": 0.1152, "step": 15890 }, { "epoch": 98.70404984423676, "grad_norm": 2.140625, "learning_rate": 2.5913852195997408e-05, "loss": 0.145, "step": 15891 }, { "epoch": 98.71028037383178, "grad_norm": 2.03125, "learning_rate": 2.591139007580055e-05, "loss": 0.1249, "step": 15892 }, { "epoch": 98.7165109034268, "grad_norm": 1.9609375, "learning_rate": 2.5908927946752082e-05, "loss": 0.1292, "step": 15893 }, { "epoch": 98.72274143302181, "grad_norm": 2.09375, "learning_rate": 2.590646580887593e-05, "loss": 0.1644, "step": 15894 }, { "epoch": 98.72897196261682, "grad_norm": 2.53125, "learning_rate": 2.5904003662196002e-05, "loss": 0.2331, "step": 15895 }, { "epoch": 98.73520249221184, "grad_norm": 2.3125, "learning_rate": 2.590154150673621e-05, "loss": 0.2181, "step": 15896 }, { "epoch": 98.74143302180686, "grad_norm": 2.3125, "learning_rate": 2.589907934252046e-05, "loss": 0.1102, "step": 15897 }, { "epoch": 98.74766355140187, "grad_norm": 2.4375, "learning_rate": 2.5896617169572678e-05, "loss": 0.1121, "step": 15898 }, { "epoch": 98.75389408099689, "grad_norm": 3.125, "learning_rate": 2.5894154987916764e-05, "loss": 0.1886, "step": 15899 }, { "epoch": 98.7601246105919, "grad_norm": 2.953125, "learning_rate": 2.5891692797576645e-05, "loss": 0.1716, "step": 15900 }, { "epoch": 98.76635514018692, "grad_norm": 2.5625, "learning_rate": 2.5889230598576223e-05, "loss": 0.28, "step": 15901 }, { "epoch": 98.77258566978193, "grad_norm": 1.1953125, "learning_rate": 2.588676839093942e-05, "loss": 0.1146, "step": 15902 }, { "epoch": 98.77881619937695, "grad_norm": 1.8125, "learning_rate": 2.5884306174690142e-05, "loss": 0.1412, "step": 15903 }, { "epoch": 98.78504672897196, "grad_norm": 2.296875, "learning_rate": 2.588184394985231e-05, "loss": 0.1224, "step": 15904 }, { "epoch": 98.79127725856698, "grad_norm": 2.90625, "learning_rate": 2.587938171644983e-05, "loss": 0.2177, "step": 15905 }, { "epoch": 98.797507788162, "grad_norm": 2.25, "learning_rate": 2.5876919474506615e-05, "loss": 0.2139, "step": 15906 }, { "epoch": 98.80373831775701, "grad_norm": 3.859375, "learning_rate": 2.5874457224046588e-05, "loss": 0.0976, "step": 15907 }, { "epoch": 98.80996884735202, "grad_norm": 3.578125, "learning_rate": 2.5871994965093656e-05, "loss": 0.3622, "step": 15908 }, { "epoch": 98.81619937694704, "grad_norm": 3.328125, "learning_rate": 2.5869532697671732e-05, "loss": 0.2802, "step": 15909 }, { "epoch": 98.82242990654206, "grad_norm": 1.765625, "learning_rate": 2.5867070421804738e-05, "loss": 0.122, "step": 15910 }, { "epoch": 98.82866043613707, "grad_norm": 1.5078125, "learning_rate": 2.5864608137516583e-05, "loss": 0.1288, "step": 15911 }, { "epoch": 98.83489096573209, "grad_norm": 2.65625, "learning_rate": 2.5862145844831177e-05, "loss": 0.1461, "step": 15912 }, { "epoch": 98.8411214953271, "grad_norm": 2.703125, "learning_rate": 2.5859683543772434e-05, "loss": 0.201, "step": 15913 }, { "epoch": 98.84735202492212, "grad_norm": 1.578125, "learning_rate": 2.5857221234364287e-05, "loss": 0.1212, "step": 15914 }, { "epoch": 98.85358255451713, "grad_norm": 3.21875, "learning_rate": 2.5854758916630623e-05, "loss": 0.2548, "step": 15915 }, { "epoch": 98.85981308411215, "grad_norm": 2.234375, "learning_rate": 2.585229659059537e-05, "loss": 0.1234, "step": 15916 }, { "epoch": 98.86604361370716, "grad_norm": 2.28125, "learning_rate": 2.5849834256282447e-05, "loss": 0.2688, "step": 15917 }, { "epoch": 98.87227414330218, "grad_norm": 2.53125, "learning_rate": 2.584737191371576e-05, "loss": 0.2144, "step": 15918 }, { "epoch": 98.8785046728972, "grad_norm": 2.265625, "learning_rate": 2.584490956291923e-05, "loss": 0.1325, "step": 15919 }, { "epoch": 98.88473520249221, "grad_norm": 2.1875, "learning_rate": 2.5842447203916775e-05, "loss": 0.1492, "step": 15920 }, { "epoch": 98.89096573208722, "grad_norm": 2.265625, "learning_rate": 2.5839984836732294e-05, "loss": 0.1138, "step": 15921 }, { "epoch": 98.89719626168224, "grad_norm": 3.0, "learning_rate": 2.5837522461389717e-05, "loss": 0.1703, "step": 15922 }, { "epoch": 98.90342679127725, "grad_norm": 2.46875, "learning_rate": 2.5835060077912953e-05, "loss": 0.1703, "step": 15923 }, { "epoch": 98.90965732087227, "grad_norm": 2.28125, "learning_rate": 2.583259768632592e-05, "loss": 0.1446, "step": 15924 }, { "epoch": 98.91588785046729, "grad_norm": 2.265625, "learning_rate": 2.5830135286652535e-05, "loss": 0.1735, "step": 15925 }, { "epoch": 98.9221183800623, "grad_norm": 2.46875, "learning_rate": 2.5827672878916704e-05, "loss": 0.2348, "step": 15926 }, { "epoch": 98.92834890965732, "grad_norm": 2.6875, "learning_rate": 2.5825210463142357e-05, "loss": 0.224, "step": 15927 }, { "epoch": 98.93457943925233, "grad_norm": 1.671875, "learning_rate": 2.582274803935339e-05, "loss": 0.1338, "step": 15928 }, { "epoch": 98.94080996884735, "grad_norm": 2.828125, "learning_rate": 2.5820285607573734e-05, "loss": 0.1994, "step": 15929 }, { "epoch": 98.94704049844236, "grad_norm": 2.8125, "learning_rate": 2.5817823167827305e-05, "loss": 0.3056, "step": 15930 }, { "epoch": 98.95327102803738, "grad_norm": 2.609375, "learning_rate": 2.581536072013801e-05, "loss": 0.2285, "step": 15931 }, { "epoch": 98.9595015576324, "grad_norm": 2.125, "learning_rate": 2.5812898264529768e-05, "loss": 0.1119, "step": 15932 }, { "epoch": 98.96573208722741, "grad_norm": 2.71875, "learning_rate": 2.58104358010265e-05, "loss": 0.2168, "step": 15933 }, { "epoch": 98.97196261682242, "grad_norm": 1.5546875, "learning_rate": 2.5807973329652108e-05, "loss": 0.1161, "step": 15934 }, { "epoch": 98.97819314641744, "grad_norm": 2.453125, "learning_rate": 2.5805510850430525e-05, "loss": 0.2173, "step": 15935 }, { "epoch": 98.98442367601245, "grad_norm": 2.84375, "learning_rate": 2.5803048363385664e-05, "loss": 0.2139, "step": 15936 }, { "epoch": 98.99065420560747, "grad_norm": 3.0, "learning_rate": 2.580058586854143e-05, "loss": 0.3051, "step": 15937 }, { "epoch": 98.99688473520249, "grad_norm": 2.65625, "learning_rate": 2.5798123365921746e-05, "loss": 0.2166, "step": 15938 }, { "epoch": 99.0, "grad_norm": 1.7265625, "learning_rate": 2.579566085555053e-05, "loss": 0.0652, "step": 15939 }, { "epoch": 99.00623052959502, "grad_norm": 3.21875, "learning_rate": 2.5793198337451696e-05, "loss": 0.3836, "step": 15940 }, { "epoch": 99.01246105919003, "grad_norm": 2.09375, "learning_rate": 2.5790735811649158e-05, "loss": 0.1172, "step": 15941 }, { "epoch": 99.01869158878505, "grad_norm": 1.359375, "learning_rate": 2.578827327816684e-05, "loss": 0.1241, "step": 15942 }, { "epoch": 99.02492211838006, "grad_norm": 2.3125, "learning_rate": 2.5785810737028653e-05, "loss": 0.2634, "step": 15943 }, { "epoch": 99.03115264797508, "grad_norm": 1.3984375, "learning_rate": 2.5783348188258516e-05, "loss": 0.098, "step": 15944 }, { "epoch": 99.03738317757009, "grad_norm": 1.125, "learning_rate": 2.5780885631880343e-05, "loss": 0.1135, "step": 15945 }, { "epoch": 99.04361370716511, "grad_norm": 2.484375, "learning_rate": 2.5778423067918056e-05, "loss": 0.2664, "step": 15946 }, { "epoch": 99.04984423676012, "grad_norm": 1.5546875, "learning_rate": 2.5775960496395564e-05, "loss": 0.1246, "step": 15947 }, { "epoch": 99.05607476635514, "grad_norm": 1.9375, "learning_rate": 2.5773497917336792e-05, "loss": 0.1529, "step": 15948 }, { "epoch": 99.06230529595015, "grad_norm": 2.296875, "learning_rate": 2.577103533076565e-05, "loss": 0.118, "step": 15949 }, { "epoch": 99.06853582554517, "grad_norm": 1.671875, "learning_rate": 2.5768572736706065e-05, "loss": 0.1299, "step": 15950 }, { "epoch": 99.07476635514018, "grad_norm": 3.96875, "learning_rate": 2.5766110135181943e-05, "loss": 0.3661, "step": 15951 }, { "epoch": 99.0809968847352, "grad_norm": 2.953125, "learning_rate": 2.576364752621721e-05, "loss": 0.1886, "step": 15952 }, { "epoch": 99.08722741433021, "grad_norm": 1.7109375, "learning_rate": 2.5761184909835778e-05, "loss": 0.1538, "step": 15953 }, { "epoch": 99.09345794392523, "grad_norm": 2.5, "learning_rate": 2.575872228606156e-05, "loss": 0.2156, "step": 15954 }, { "epoch": 99.09968847352025, "grad_norm": 3.4375, "learning_rate": 2.575625965491849e-05, "loss": 0.2734, "step": 15955 }, { "epoch": 99.10591900311526, "grad_norm": 2.25, "learning_rate": 2.5753797016430463e-05, "loss": 0.1061, "step": 15956 }, { "epoch": 99.11214953271028, "grad_norm": 1.4609375, "learning_rate": 2.5751334370621416e-05, "loss": 0.0998, "step": 15957 }, { "epoch": 99.11838006230529, "grad_norm": 2.375, "learning_rate": 2.574887171751526e-05, "loss": 0.1854, "step": 15958 }, { "epoch": 99.1246105919003, "grad_norm": 2.421875, "learning_rate": 2.5746409057135912e-05, "loss": 0.1233, "step": 15959 }, { "epoch": 99.13084112149532, "grad_norm": 2.8125, "learning_rate": 2.574394638950729e-05, "loss": 0.1682, "step": 15960 }, { "epoch": 99.13707165109034, "grad_norm": 2.609375, "learning_rate": 2.574148371465331e-05, "loss": 0.184, "step": 15961 }, { "epoch": 99.14330218068535, "grad_norm": 3.296875, "learning_rate": 2.5739021032597893e-05, "loss": 0.3147, "step": 15962 }, { "epoch": 99.14953271028037, "grad_norm": 2.015625, "learning_rate": 2.5736558343364953e-05, "loss": 0.2409, "step": 15963 }, { "epoch": 99.15576323987538, "grad_norm": 1.9765625, "learning_rate": 2.5734095646978413e-05, "loss": 0.1866, "step": 15964 }, { "epoch": 99.1619937694704, "grad_norm": 1.34375, "learning_rate": 2.573163294346219e-05, "loss": 0.1135, "step": 15965 }, { "epoch": 99.16822429906541, "grad_norm": 4.25, "learning_rate": 2.5729170232840205e-05, "loss": 0.3505, "step": 15966 }, { "epoch": 99.17445482866043, "grad_norm": 2.359375, "learning_rate": 2.5726707515136366e-05, "loss": 0.1757, "step": 15967 }, { "epoch": 99.18068535825545, "grad_norm": 3.765625, "learning_rate": 2.572424479037461e-05, "loss": 0.218, "step": 15968 }, { "epoch": 99.18691588785046, "grad_norm": 2.46875, "learning_rate": 2.572178205857883e-05, "loss": 0.1881, "step": 15969 }, { "epoch": 99.19314641744548, "grad_norm": 1.390625, "learning_rate": 2.5719319319772967e-05, "loss": 0.1201, "step": 15970 }, { "epoch": 99.19937694704049, "grad_norm": 2.296875, "learning_rate": 2.5716856573980924e-05, "loss": 0.1871, "step": 15971 }, { "epoch": 99.20560747663552, "grad_norm": 1.734375, "learning_rate": 2.5714393821226635e-05, "loss": 0.1045, "step": 15972 }, { "epoch": 99.21183800623054, "grad_norm": 3.125, "learning_rate": 2.5711931061534004e-05, "loss": 0.2282, "step": 15973 }, { "epoch": 99.21806853582555, "grad_norm": 3.171875, "learning_rate": 2.5709468294926963e-05, "loss": 0.2407, "step": 15974 }, { "epoch": 99.22429906542057, "grad_norm": 2.140625, "learning_rate": 2.5707005521429417e-05, "loss": 0.2467, "step": 15975 }, { "epoch": 99.23052959501558, "grad_norm": 1.8203125, "learning_rate": 2.570454274106529e-05, "loss": 0.1315, "step": 15976 }, { "epoch": 99.2367601246106, "grad_norm": 2.953125, "learning_rate": 2.570207995385851e-05, "loss": 0.1002, "step": 15977 }, { "epoch": 99.24299065420561, "grad_norm": 2.984375, "learning_rate": 2.569961715983299e-05, "loss": 0.183, "step": 15978 }, { "epoch": 99.24922118380063, "grad_norm": 2.125, "learning_rate": 2.5697154359012644e-05, "loss": 0.167, "step": 15979 }, { "epoch": 99.25545171339564, "grad_norm": 2.140625, "learning_rate": 2.569469155142139e-05, "loss": 0.2111, "step": 15980 }, { "epoch": 99.26168224299066, "grad_norm": 1.1484375, "learning_rate": 2.569222873708317e-05, "loss": 0.0986, "step": 15981 }, { "epoch": 99.26791277258567, "grad_norm": 2.390625, "learning_rate": 2.568976591602187e-05, "loss": 0.1801, "step": 15982 }, { "epoch": 99.27414330218069, "grad_norm": 3.65625, "learning_rate": 2.568730308826143e-05, "loss": 0.2492, "step": 15983 }, { "epoch": 99.2803738317757, "grad_norm": 1.9296875, "learning_rate": 2.568484025382577e-05, "loss": 0.1408, "step": 15984 }, { "epoch": 99.28660436137072, "grad_norm": 2.21875, "learning_rate": 2.56823774127388e-05, "loss": 0.1221, "step": 15985 }, { "epoch": 99.29283489096574, "grad_norm": 3.515625, "learning_rate": 2.5679914565024443e-05, "loss": 0.2801, "step": 15986 }, { "epoch": 99.29906542056075, "grad_norm": 1.7421875, "learning_rate": 2.5677451710706625e-05, "loss": 0.0968, "step": 15987 }, { "epoch": 99.30529595015577, "grad_norm": 2.515625, "learning_rate": 2.5674988849809255e-05, "loss": 0.1825, "step": 15988 }, { "epoch": 99.31152647975078, "grad_norm": 2.515625, "learning_rate": 2.567252598235626e-05, "loss": 0.1989, "step": 15989 }, { "epoch": 99.3177570093458, "grad_norm": 3.8125, "learning_rate": 2.5670063108371563e-05, "loss": 0.2525, "step": 15990 }, { "epoch": 99.32398753894081, "grad_norm": 2.671875, "learning_rate": 2.5667600227879074e-05, "loss": 0.2132, "step": 15991 }, { "epoch": 99.33021806853583, "grad_norm": 3.4375, "learning_rate": 2.5665137340902718e-05, "loss": 0.1168, "step": 15992 }, { "epoch": 99.33644859813084, "grad_norm": 3.421875, "learning_rate": 2.566267444746641e-05, "loss": 0.1583, "step": 15993 }, { "epoch": 99.34267912772586, "grad_norm": 2.890625, "learning_rate": 2.5660211547594087e-05, "loss": 0.2115, "step": 15994 }, { "epoch": 99.34890965732087, "grad_norm": 2.921875, "learning_rate": 2.565774864130965e-05, "loss": 0.1903, "step": 15995 }, { "epoch": 99.35514018691589, "grad_norm": 1.25, "learning_rate": 2.565528572863703e-05, "loss": 0.094, "step": 15996 }, { "epoch": 99.3613707165109, "grad_norm": 1.7109375, "learning_rate": 2.5652822809600142e-05, "loss": 0.1268, "step": 15997 }, { "epoch": 99.36760124610592, "grad_norm": 2.5625, "learning_rate": 2.5650359884222913e-05, "loss": 0.2277, "step": 15998 }, { "epoch": 99.37383177570094, "grad_norm": 2.515625, "learning_rate": 2.5647896952529255e-05, "loss": 0.1746, "step": 15999 }, { "epoch": 99.38006230529595, "grad_norm": 2.0625, "learning_rate": 2.564543401454309e-05, "loss": 0.158, "step": 16000 }, { "epoch": 99.38629283489097, "grad_norm": 2.71875, "learning_rate": 2.5642971070288345e-05, "loss": 0.1116, "step": 16001 }, { "epoch": 99.39252336448598, "grad_norm": 2.015625, "learning_rate": 2.5640508119788932e-05, "loss": 0.1533, "step": 16002 }, { "epoch": 99.398753894081, "grad_norm": 2.28125, "learning_rate": 2.5638045163068785e-05, "loss": 0.238, "step": 16003 }, { "epoch": 99.40498442367601, "grad_norm": 3.359375, "learning_rate": 2.56355822001518e-05, "loss": 0.2234, "step": 16004 }, { "epoch": 99.41121495327103, "grad_norm": 2.734375, "learning_rate": 2.5633119231061923e-05, "loss": 0.1794, "step": 16005 }, { "epoch": 99.41744548286604, "grad_norm": 2.296875, "learning_rate": 2.563065625582307e-05, "loss": 0.1842, "step": 16006 }, { "epoch": 99.42367601246106, "grad_norm": 2.03125, "learning_rate": 2.5628193274459157e-05, "loss": 0.1626, "step": 16007 }, { "epoch": 99.42990654205607, "grad_norm": 2.5625, "learning_rate": 2.5625730286994098e-05, "loss": 0.3734, "step": 16008 }, { "epoch": 99.43613707165109, "grad_norm": 1.640625, "learning_rate": 2.5623267293451826e-05, "loss": 0.1531, "step": 16009 }, { "epoch": 99.4423676012461, "grad_norm": 2.625, "learning_rate": 2.562080429385626e-05, "loss": 0.2175, "step": 16010 }, { "epoch": 99.44859813084112, "grad_norm": 1.9296875, "learning_rate": 2.5618341288231306e-05, "loss": 0.1114, "step": 16011 }, { "epoch": 99.45482866043614, "grad_norm": 2.296875, "learning_rate": 2.561587827660091e-05, "loss": 0.157, "step": 16012 }, { "epoch": 99.46105919003115, "grad_norm": 1.234375, "learning_rate": 2.5613415258988976e-05, "loss": 0.1215, "step": 16013 }, { "epoch": 99.46728971962617, "grad_norm": 1.2265625, "learning_rate": 2.5610952235419433e-05, "loss": 0.1184, "step": 16014 }, { "epoch": 99.47352024922118, "grad_norm": 2.890625, "learning_rate": 2.5608489205916197e-05, "loss": 0.1424, "step": 16015 }, { "epoch": 99.4797507788162, "grad_norm": 1.6953125, "learning_rate": 2.56060261705032e-05, "loss": 0.1497, "step": 16016 }, { "epoch": 99.48598130841121, "grad_norm": 2.484375, "learning_rate": 2.5603563129204345e-05, "loss": 0.1859, "step": 16017 }, { "epoch": 99.49221183800623, "grad_norm": 1.8515625, "learning_rate": 2.5601100082043565e-05, "loss": 0.1585, "step": 16018 }, { "epoch": 99.49844236760124, "grad_norm": 2.796875, "learning_rate": 2.559863702904479e-05, "loss": 0.1456, "step": 16019 }, { "epoch": 99.50467289719626, "grad_norm": 2.8125, "learning_rate": 2.5596173970231928e-05, "loss": 0.1227, "step": 16020 }, { "epoch": 99.51090342679127, "grad_norm": 2.453125, "learning_rate": 2.5593710905628904e-05, "loss": 0.2241, "step": 16021 }, { "epoch": 99.51713395638629, "grad_norm": 2.984375, "learning_rate": 2.5591247835259642e-05, "loss": 0.3092, "step": 16022 }, { "epoch": 99.5233644859813, "grad_norm": 1.359375, "learning_rate": 2.5588784759148066e-05, "loss": 0.1177, "step": 16023 }, { "epoch": 99.52959501557632, "grad_norm": 2.453125, "learning_rate": 2.5586321677318085e-05, "loss": 0.2312, "step": 16024 }, { "epoch": 99.53582554517133, "grad_norm": 2.1875, "learning_rate": 2.558385858979364e-05, "loss": 0.1906, "step": 16025 }, { "epoch": 99.54205607476635, "grad_norm": 2.640625, "learning_rate": 2.5581395496598637e-05, "loss": 0.1306, "step": 16026 }, { "epoch": 99.54828660436137, "grad_norm": 2.109375, "learning_rate": 2.557893239775701e-05, "loss": 0.1678, "step": 16027 }, { "epoch": 99.55451713395638, "grad_norm": 1.7109375, "learning_rate": 2.557646929329267e-05, "loss": 0.2398, "step": 16028 }, { "epoch": 99.5607476635514, "grad_norm": 1.5234375, "learning_rate": 2.5574006183229553e-05, "loss": 0.1549, "step": 16029 }, { "epoch": 99.56697819314641, "grad_norm": 2.21875, "learning_rate": 2.5571543067591564e-05, "loss": 0.1755, "step": 16030 }, { "epoch": 99.57320872274143, "grad_norm": 2.109375, "learning_rate": 2.556907994640264e-05, "loss": 0.1445, "step": 16031 }, { "epoch": 99.57943925233644, "grad_norm": 2.328125, "learning_rate": 2.55666168196867e-05, "loss": 0.2027, "step": 16032 }, { "epoch": 99.58566978193146, "grad_norm": 2.75, "learning_rate": 2.5564153687467664e-05, "loss": 0.2389, "step": 16033 }, { "epoch": 99.59190031152647, "grad_norm": 1.6796875, "learning_rate": 2.556169054976945e-05, "loss": 0.1099, "step": 16034 }, { "epoch": 99.59813084112149, "grad_norm": 1.6171875, "learning_rate": 2.5559227406615993e-05, "loss": 0.1181, "step": 16035 }, { "epoch": 99.60436137071652, "grad_norm": 2.828125, "learning_rate": 2.55567642580312e-05, "loss": 0.1566, "step": 16036 }, { "epoch": 99.61059190031153, "grad_norm": 2.171875, "learning_rate": 2.5554301104038997e-05, "loss": 0.1182, "step": 16037 }, { "epoch": 99.61682242990655, "grad_norm": 1.765625, "learning_rate": 2.5551837944663326e-05, "loss": 0.1627, "step": 16038 }, { "epoch": 99.62305295950156, "grad_norm": 1.6796875, "learning_rate": 2.5549374779928082e-05, "loss": 0.1612, "step": 16039 }, { "epoch": 99.62928348909658, "grad_norm": 2.46875, "learning_rate": 2.5546911609857204e-05, "loss": 0.1702, "step": 16040 }, { "epoch": 99.6355140186916, "grad_norm": 2.21875, "learning_rate": 2.554444843447461e-05, "loss": 0.1547, "step": 16041 }, { "epoch": 99.64174454828661, "grad_norm": 2.421875, "learning_rate": 2.554198525380423e-05, "loss": 0.2413, "step": 16042 }, { "epoch": 99.64797507788163, "grad_norm": 2.890625, "learning_rate": 2.553952206786997e-05, "loss": 0.2822, "step": 16043 }, { "epoch": 99.65420560747664, "grad_norm": 3.71875, "learning_rate": 2.5537058876695775e-05, "loss": 0.291, "step": 16044 }, { "epoch": 99.66043613707166, "grad_norm": 1.3203125, "learning_rate": 2.5534595680305544e-05, "loss": 0.1071, "step": 16045 }, { "epoch": 99.66666666666667, "grad_norm": 1.265625, "learning_rate": 2.5532132478723225e-05, "loss": 0.1106, "step": 16046 }, { "epoch": 99.67289719626169, "grad_norm": 2.90625, "learning_rate": 2.5529669271972724e-05, "loss": 0.1735, "step": 16047 }, { "epoch": 99.6791277258567, "grad_norm": 1.1484375, "learning_rate": 2.5527206060077968e-05, "loss": 0.1202, "step": 16048 }, { "epoch": 99.68535825545172, "grad_norm": 1.140625, "learning_rate": 2.5524742843062888e-05, "loss": 0.0998, "step": 16049 }, { "epoch": 99.69158878504673, "grad_norm": 2.734375, "learning_rate": 2.552227962095139e-05, "loss": 0.2996, "step": 16050 }, { "epoch": 99.69781931464175, "grad_norm": 2.71875, "learning_rate": 2.551981639376742e-05, "loss": 0.2154, "step": 16051 }, { "epoch": 99.70404984423676, "grad_norm": 2.25, "learning_rate": 2.5517353161534874e-05, "loss": 0.1141, "step": 16052 }, { "epoch": 99.71028037383178, "grad_norm": 1.71875, "learning_rate": 2.5514889924277702e-05, "loss": 0.1277, "step": 16053 }, { "epoch": 99.7165109034268, "grad_norm": 1.2734375, "learning_rate": 2.5512426682019814e-05, "loss": 0.1161, "step": 16054 }, { "epoch": 99.72274143302181, "grad_norm": 3.328125, "learning_rate": 2.550996343478514e-05, "loss": 0.2394, "step": 16055 }, { "epoch": 99.72897196261682, "grad_norm": 2.96875, "learning_rate": 2.5507500182597592e-05, "loss": 0.2178, "step": 16056 }, { "epoch": 99.73520249221184, "grad_norm": 2.90625, "learning_rate": 2.5505036925481103e-05, "loss": 0.1934, "step": 16057 }, { "epoch": 99.74143302180686, "grad_norm": 3.859375, "learning_rate": 2.5502573663459594e-05, "loss": 0.2243, "step": 16058 }, { "epoch": 99.74766355140187, "grad_norm": 3.09375, "learning_rate": 2.5500110396556988e-05, "loss": 0.166, "step": 16059 }, { "epoch": 99.75389408099689, "grad_norm": 2.234375, "learning_rate": 2.549764712479722e-05, "loss": 0.1279, "step": 16060 }, { "epoch": 99.7601246105919, "grad_norm": 2.671875, "learning_rate": 2.549518384820419e-05, "loss": 0.2496, "step": 16061 }, { "epoch": 99.76635514018692, "grad_norm": 1.4921875, "learning_rate": 2.5492720566801843e-05, "loss": 0.0782, "step": 16062 }, { "epoch": 99.77258566978193, "grad_norm": 2.5, "learning_rate": 2.549025728061408e-05, "loss": 0.0993, "step": 16063 }, { "epoch": 99.77881619937695, "grad_norm": 2.8125, "learning_rate": 2.5487793989664867e-05, "loss": 0.2671, "step": 16064 }, { "epoch": 99.78504672897196, "grad_norm": 2.171875, "learning_rate": 2.548533069397808e-05, "loss": 0.1543, "step": 16065 }, { "epoch": 99.79127725856698, "grad_norm": 2.78125, "learning_rate": 2.5482867393577677e-05, "loss": 0.1981, "step": 16066 }, { "epoch": 99.797507788162, "grad_norm": 3.265625, "learning_rate": 2.5480404088487565e-05, "loss": 0.2274, "step": 16067 }, { "epoch": 99.80373831775701, "grad_norm": 2.171875, "learning_rate": 2.5477940778731668e-05, "loss": 0.1892, "step": 16068 }, { "epoch": 99.80996884735202, "grad_norm": 2.5625, "learning_rate": 2.5475477464333918e-05, "loss": 0.1888, "step": 16069 }, { "epoch": 99.81619937694704, "grad_norm": 2.9375, "learning_rate": 2.5473014145318243e-05, "loss": 0.2683, "step": 16070 }, { "epoch": 99.82242990654206, "grad_norm": 1.5859375, "learning_rate": 2.547055082170855e-05, "loss": 0.1256, "step": 16071 }, { "epoch": 99.82866043613707, "grad_norm": 4.34375, "learning_rate": 2.546808749352877e-05, "loss": 0.1912, "step": 16072 }, { "epoch": 99.83489096573209, "grad_norm": 3.796875, "learning_rate": 2.5465624160802847e-05, "loss": 0.1593, "step": 16073 }, { "epoch": 99.8411214953271, "grad_norm": 3.609375, "learning_rate": 2.5463160823554672e-05, "loss": 0.2273, "step": 16074 }, { "epoch": 99.84735202492212, "grad_norm": 2.171875, "learning_rate": 2.5460697481808194e-05, "loss": 0.2192, "step": 16075 }, { "epoch": 99.85358255451713, "grad_norm": 1.3046875, "learning_rate": 2.5458234135587335e-05, "loss": 0.1168, "step": 16076 }, { "epoch": 99.85981308411215, "grad_norm": 2.578125, "learning_rate": 2.5455770784916005e-05, "loss": 0.1398, "step": 16077 }, { "epoch": 99.86604361370716, "grad_norm": 2.796875, "learning_rate": 2.545330742981814e-05, "loss": 0.2551, "step": 16078 }, { "epoch": 99.87227414330218, "grad_norm": 2.78125, "learning_rate": 2.545084407031767e-05, "loss": 0.1886, "step": 16079 }, { "epoch": 99.8785046728972, "grad_norm": 3.046875, "learning_rate": 2.5448380706438502e-05, "loss": 0.239, "step": 16080 }, { "epoch": 99.88473520249221, "grad_norm": 5.25, "learning_rate": 2.544591733820458e-05, "loss": 0.2399, "step": 16081 }, { "epoch": 99.89096573208722, "grad_norm": 2.125, "learning_rate": 2.544345396563982e-05, "loss": 0.1554, "step": 16082 }, { "epoch": 99.89719626168224, "grad_norm": 1.6171875, "learning_rate": 2.5440990588768138e-05, "loss": 0.1221, "step": 16083 }, { "epoch": 99.90342679127725, "grad_norm": 2.890625, "learning_rate": 2.5438527207613472e-05, "loss": 0.2013, "step": 16084 }, { "epoch": 99.90965732087227, "grad_norm": 3.96875, "learning_rate": 2.543606382219974e-05, "loss": 0.2936, "step": 16085 }, { "epoch": 99.91588785046729, "grad_norm": 1.4609375, "learning_rate": 2.5433600432550876e-05, "loss": 0.107, "step": 16086 }, { "epoch": 99.9221183800623, "grad_norm": 1.2109375, "learning_rate": 2.543113703869079e-05, "loss": 0.1243, "step": 16087 }, { "epoch": 99.92834890965732, "grad_norm": 1.3203125, "learning_rate": 2.5428673640643414e-05, "loss": 0.0923, "step": 16088 }, { "epoch": 99.93457943925233, "grad_norm": 1.25, "learning_rate": 2.5426210238432678e-05, "loss": 0.1087, "step": 16089 }, { "epoch": 99.94080996884735, "grad_norm": 2.265625, "learning_rate": 2.5423746832082505e-05, "loss": 0.2344, "step": 16090 }, { "epoch": 99.94704049844236, "grad_norm": 3.203125, "learning_rate": 2.542128342161681e-05, "loss": 0.3002, "step": 16091 }, { "epoch": 99.95327102803738, "grad_norm": 2.15625, "learning_rate": 2.541882000705954e-05, "loss": 0.1461, "step": 16092 }, { "epoch": 99.9595015576324, "grad_norm": 1.78125, "learning_rate": 2.5416356588434594e-05, "loss": 0.126, "step": 16093 }, { "epoch": 99.96573208722741, "grad_norm": 1.9453125, "learning_rate": 2.541389316576591e-05, "loss": 0.2037, "step": 16094 }, { "epoch": 99.97196261682242, "grad_norm": 2.078125, "learning_rate": 2.541142973907742e-05, "loss": 0.1743, "step": 16095 }, { "epoch": 99.97819314641744, "grad_norm": 1.8671875, "learning_rate": 2.5408966308393038e-05, "loss": 0.1189, "step": 16096 }, { "epoch": 99.98442367601245, "grad_norm": 2.109375, "learning_rate": 2.540650287373669e-05, "loss": 0.1803, "step": 16097 }, { "epoch": 99.99065420560747, "grad_norm": 1.5625, "learning_rate": 2.540403943513231e-05, "loss": 0.1497, "step": 16098 }, { "epoch": 99.99688473520249, "grad_norm": 1.59375, "learning_rate": 2.5401575992603822e-05, "loss": 0.1163, "step": 16099 }, { "epoch": 100.0, "grad_norm": 2.21875, "learning_rate": 2.539911254617514e-05, "loss": 0.1699, "step": 16100 }, { "epoch": 100.00623052959502, "grad_norm": 1.9140625, "learning_rate": 2.5396649095870202e-05, "loss": 0.1605, "step": 16101 }, { "epoch": 100.01246105919003, "grad_norm": 2.859375, "learning_rate": 2.5394185641712924e-05, "loss": 0.2566, "step": 16102 }, { "epoch": 100.01869158878505, "grad_norm": 1.0703125, "learning_rate": 2.539172218372724e-05, "loss": 0.1132, "step": 16103 }, { "epoch": 100.02492211838006, "grad_norm": 1.2265625, "learning_rate": 2.538925872193707e-05, "loss": 0.1216, "step": 16104 }, { "epoch": 100.03115264797508, "grad_norm": 2.765625, "learning_rate": 2.5386795256366348e-05, "loss": 0.2422, "step": 16105 }, { "epoch": 100.03738317757009, "grad_norm": 2.09375, "learning_rate": 2.5384331787038984e-05, "loss": 0.1659, "step": 16106 }, { "epoch": 100.04361370716511, "grad_norm": 2.84375, "learning_rate": 2.5381868313978912e-05, "loss": 0.284, "step": 16107 }, { "epoch": 100.04984423676012, "grad_norm": 2.21875, "learning_rate": 2.537940483721007e-05, "loss": 0.1487, "step": 16108 }, { "epoch": 100.05607476635514, "grad_norm": 2.234375, "learning_rate": 2.5376941356756362e-05, "loss": 0.1196, "step": 16109 }, { "epoch": 100.06230529595015, "grad_norm": 2.046875, "learning_rate": 2.5374477872641728e-05, "loss": 0.1084, "step": 16110 }, { "epoch": 100.06853582554517, "grad_norm": 1.53125, "learning_rate": 2.5372014384890086e-05, "loss": 0.109, "step": 16111 }, { "epoch": 100.07476635514018, "grad_norm": 1.7421875, "learning_rate": 2.5369550893525375e-05, "loss": 0.1259, "step": 16112 }, { "epoch": 100.0809968847352, "grad_norm": 3.078125, "learning_rate": 2.53670873985715e-05, "loss": 0.3018, "step": 16113 }, { "epoch": 100.08722741433021, "grad_norm": 1.75, "learning_rate": 2.5364623900052403e-05, "loss": 0.1379, "step": 16114 }, { "epoch": 100.09345794392523, "grad_norm": 2.578125, "learning_rate": 2.5362160397992013e-05, "loss": 0.2376, "step": 16115 }, { "epoch": 100.09968847352025, "grad_norm": 2.3125, "learning_rate": 2.535969689241424e-05, "loss": 0.1857, "step": 16116 }, { "epoch": 100.10591900311526, "grad_norm": 1.4140625, "learning_rate": 2.5357233383343028e-05, "loss": 0.1467, "step": 16117 }, { "epoch": 100.11214953271028, "grad_norm": 2.671875, "learning_rate": 2.5354769870802286e-05, "loss": 0.2263, "step": 16118 }, { "epoch": 100.11838006230529, "grad_norm": 1.5, "learning_rate": 2.5352306354815953e-05, "loss": 0.1262, "step": 16119 }, { "epoch": 100.1246105919003, "grad_norm": 1.25, "learning_rate": 2.5349842835407946e-05, "loss": 0.1186, "step": 16120 }, { "epoch": 100.13084112149532, "grad_norm": 2.09375, "learning_rate": 2.5347379312602198e-05, "loss": 0.147, "step": 16121 }, { "epoch": 100.13707165109034, "grad_norm": 1.9609375, "learning_rate": 2.534491578642263e-05, "loss": 0.169, "step": 16122 }, { "epoch": 100.14330218068535, "grad_norm": 2.390625, "learning_rate": 2.5342452256893172e-05, "loss": 0.1061, "step": 16123 }, { "epoch": 100.14953271028037, "grad_norm": 1.0390625, "learning_rate": 2.533998872403775e-05, "loss": 0.093, "step": 16124 }, { "epoch": 100.15576323987538, "grad_norm": 2.40625, "learning_rate": 2.533752518788029e-05, "loss": 0.1884, "step": 16125 }, { "epoch": 100.1619937694704, "grad_norm": 1.015625, "learning_rate": 2.5335061648444715e-05, "loss": 0.1199, "step": 16126 }, { "epoch": 100.16822429906541, "grad_norm": 1.7578125, "learning_rate": 2.5332598105754964e-05, "loss": 0.1042, "step": 16127 }, { "epoch": 100.17445482866043, "grad_norm": 1.9375, "learning_rate": 2.533013455983494e-05, "loss": 0.1302, "step": 16128 }, { "epoch": 100.18068535825545, "grad_norm": 2.609375, "learning_rate": 2.5327671010708592e-05, "loss": 0.1207, "step": 16129 }, { "epoch": 100.18691588785046, "grad_norm": 2.34375, "learning_rate": 2.5325207458399837e-05, "loss": 0.199, "step": 16130 }, { "epoch": 100.19314641744548, "grad_norm": 1.9765625, "learning_rate": 2.53227439029326e-05, "loss": 0.1593, "step": 16131 }, { "epoch": 100.19937694704049, "grad_norm": 2.46875, "learning_rate": 2.532028034433081e-05, "loss": 0.2138, "step": 16132 }, { "epoch": 100.20560747663552, "grad_norm": 2.359375, "learning_rate": 2.5317816782618393e-05, "loss": 0.1602, "step": 16133 }, { "epoch": 100.21183800623054, "grad_norm": 1.7734375, "learning_rate": 2.5315353217819282e-05, "loss": 0.1424, "step": 16134 }, { "epoch": 100.21806853582555, "grad_norm": 2.140625, "learning_rate": 2.5312889649957382e-05, "loss": 0.217, "step": 16135 }, { "epoch": 100.22429906542057, "grad_norm": 1.4453125, "learning_rate": 2.531042607905665e-05, "loss": 0.1365, "step": 16136 }, { "epoch": 100.23052959501558, "grad_norm": 2.640625, "learning_rate": 2.530796250514099e-05, "loss": 0.1751, "step": 16137 }, { "epoch": 100.2367601246106, "grad_norm": 1.8671875, "learning_rate": 2.5305498928234345e-05, "loss": 0.1406, "step": 16138 }, { "epoch": 100.24299065420561, "grad_norm": 1.4453125, "learning_rate": 2.530303534836062e-05, "loss": 0.1141, "step": 16139 }, { "epoch": 100.24922118380063, "grad_norm": 1.375, "learning_rate": 2.5300571765543767e-05, "loss": 0.1017, "step": 16140 }, { "epoch": 100.25545171339564, "grad_norm": 1.4140625, "learning_rate": 2.5298108179807696e-05, "loss": 0.1264, "step": 16141 }, { "epoch": 100.26168224299066, "grad_norm": 2.28125, "learning_rate": 2.5295644591176338e-05, "loss": 0.1124, "step": 16142 }, { "epoch": 100.26791277258567, "grad_norm": 1.4140625, "learning_rate": 2.5293180999673633e-05, "loss": 0.1096, "step": 16143 }, { "epoch": 100.27414330218069, "grad_norm": 2.0, "learning_rate": 2.5290717405323478e-05, "loss": 0.1226, "step": 16144 }, { "epoch": 100.2803738317757, "grad_norm": 1.359375, "learning_rate": 2.5288253808149827e-05, "loss": 0.14, "step": 16145 }, { "epoch": 100.28660436137072, "grad_norm": 1.1796875, "learning_rate": 2.5285790208176592e-05, "loss": 0.1197, "step": 16146 }, { "epoch": 100.29283489096574, "grad_norm": 2.46875, "learning_rate": 2.528332660542771e-05, "loss": 0.1345, "step": 16147 }, { "epoch": 100.29906542056075, "grad_norm": 2.03125, "learning_rate": 2.5280862999927097e-05, "loss": 0.1866, "step": 16148 }, { "epoch": 100.30529595015577, "grad_norm": 3.046875, "learning_rate": 2.5278399391698693e-05, "loss": 0.282, "step": 16149 }, { "epoch": 100.31152647975078, "grad_norm": 1.796875, "learning_rate": 2.5275935780766414e-05, "loss": 0.1169, "step": 16150 }, { "epoch": 100.3177570093458, "grad_norm": 2.265625, "learning_rate": 2.5273472167154195e-05, "loss": 0.1283, "step": 16151 }, { "epoch": 100.32398753894081, "grad_norm": 2.84375, "learning_rate": 2.527100855088596e-05, "loss": 0.3077, "step": 16152 }, { "epoch": 100.33021806853583, "grad_norm": 2.484375, "learning_rate": 2.5268544931985633e-05, "loss": 0.2047, "step": 16153 }, { "epoch": 100.33644859813084, "grad_norm": 1.1796875, "learning_rate": 2.5266081310477145e-05, "loss": 0.099, "step": 16154 }, { "epoch": 100.34267912772586, "grad_norm": 2.34375, "learning_rate": 2.526361768638442e-05, "loss": 0.2171, "step": 16155 }, { "epoch": 100.34890965732087, "grad_norm": 1.2578125, "learning_rate": 2.5261154059731397e-05, "loss": 0.1033, "step": 16156 }, { "epoch": 100.35514018691589, "grad_norm": 1.3984375, "learning_rate": 2.525869043054198e-05, "loss": 0.1208, "step": 16157 }, { "epoch": 100.3613707165109, "grad_norm": 2.421875, "learning_rate": 2.5256226798840115e-05, "loss": 0.166, "step": 16158 }, { "epoch": 100.36760124610592, "grad_norm": 2.265625, "learning_rate": 2.525376316464973e-05, "loss": 0.1863, "step": 16159 }, { "epoch": 100.37383177570094, "grad_norm": 2.71875, "learning_rate": 2.5251299527994742e-05, "loss": 0.2418, "step": 16160 }, { "epoch": 100.38006230529595, "grad_norm": 1.2734375, "learning_rate": 2.5248835888899076e-05, "loss": 0.1131, "step": 16161 }, { "epoch": 100.38629283489097, "grad_norm": 2.109375, "learning_rate": 2.5246372247386678e-05, "loss": 0.1031, "step": 16162 }, { "epoch": 100.39252336448598, "grad_norm": 1.4921875, "learning_rate": 2.5243908603481453e-05, "loss": 0.1363, "step": 16163 }, { "epoch": 100.398753894081, "grad_norm": 3.078125, "learning_rate": 2.5241444957207345e-05, "loss": 0.1831, "step": 16164 }, { "epoch": 100.40498442367601, "grad_norm": 1.84375, "learning_rate": 2.5238981308588277e-05, "loss": 0.1567, "step": 16165 }, { "epoch": 100.41121495327103, "grad_norm": 2.53125, "learning_rate": 2.5236517657648172e-05, "loss": 0.1747, "step": 16166 }, { "epoch": 100.41744548286604, "grad_norm": 3.6875, "learning_rate": 2.5234054004410962e-05, "loss": 0.3604, "step": 16167 }, { "epoch": 100.42367601246106, "grad_norm": 2.34375, "learning_rate": 2.5231590348900562e-05, "loss": 0.1499, "step": 16168 }, { "epoch": 100.42990654205607, "grad_norm": 1.75, "learning_rate": 2.522912669114093e-05, "loss": 0.1371, "step": 16169 }, { "epoch": 100.43613707165109, "grad_norm": 2.25, "learning_rate": 2.5226663031155954e-05, "loss": 0.1696, "step": 16170 }, { "epoch": 100.4423676012461, "grad_norm": 2.671875, "learning_rate": 2.5224199368969594e-05, "loss": 0.1825, "step": 16171 }, { "epoch": 100.44859813084112, "grad_norm": 1.7578125, "learning_rate": 2.522173570460576e-05, "loss": 0.1609, "step": 16172 }, { "epoch": 100.45482866043614, "grad_norm": 2.875, "learning_rate": 2.5219272038088386e-05, "loss": 0.2371, "step": 16173 }, { "epoch": 100.46105919003115, "grad_norm": 1.40625, "learning_rate": 2.5216808369441396e-05, "loss": 0.1186, "step": 16174 }, { "epoch": 100.46728971962617, "grad_norm": 3.90625, "learning_rate": 2.5214344698688724e-05, "loss": 0.4477, "step": 16175 }, { "epoch": 100.47352024922118, "grad_norm": 2.703125, "learning_rate": 2.5211881025854288e-05, "loss": 0.1377, "step": 16176 }, { "epoch": 100.4797507788162, "grad_norm": 1.2578125, "learning_rate": 2.520941735096203e-05, "loss": 0.1204, "step": 16177 }, { "epoch": 100.48598130841121, "grad_norm": 2.046875, "learning_rate": 2.520695367403586e-05, "loss": 0.1655, "step": 16178 }, { "epoch": 100.49221183800623, "grad_norm": 2.78125, "learning_rate": 2.5204489995099716e-05, "loss": 0.2409, "step": 16179 }, { "epoch": 100.49844236760124, "grad_norm": 2.90625, "learning_rate": 2.520202631417753e-05, "loss": 0.1708, "step": 16180 }, { "epoch": 100.50467289719626, "grad_norm": 1.765625, "learning_rate": 2.5199562631293216e-05, "loss": 0.0947, "step": 16181 }, { "epoch": 100.51090342679127, "grad_norm": 2.71875, "learning_rate": 2.5197098946470716e-05, "loss": 0.2617, "step": 16182 }, { "epoch": 100.51713395638629, "grad_norm": 1.03125, "learning_rate": 2.5194635259733946e-05, "loss": 0.0999, "step": 16183 }, { "epoch": 100.5233644859813, "grad_norm": 1.5, "learning_rate": 2.519217157110684e-05, "loss": 0.1274, "step": 16184 }, { "epoch": 100.52959501557632, "grad_norm": 2.109375, "learning_rate": 2.5189707880613334e-05, "loss": 0.1557, "step": 16185 }, { "epoch": 100.53582554517133, "grad_norm": 0.7578125, "learning_rate": 2.5187244188277344e-05, "loss": 0.0948, "step": 16186 }, { "epoch": 100.54205607476635, "grad_norm": 1.875, "learning_rate": 2.518478049412279e-05, "loss": 0.0987, "step": 16187 }, { "epoch": 100.54828660436137, "grad_norm": 1.921875, "learning_rate": 2.518231679817363e-05, "loss": 0.1371, "step": 16188 }, { "epoch": 100.55451713395638, "grad_norm": 3.09375, "learning_rate": 2.5179853100453755e-05, "loss": 0.1364, "step": 16189 }, { "epoch": 100.5607476635514, "grad_norm": 2.015625, "learning_rate": 2.5177389400987115e-05, "loss": 0.1156, "step": 16190 }, { "epoch": 100.56697819314641, "grad_norm": 1.2265625, "learning_rate": 2.5174925699797642e-05, "loss": 0.1012, "step": 16191 }, { "epoch": 100.57320872274143, "grad_norm": 3.109375, "learning_rate": 2.5172461996909247e-05, "loss": 0.1611, "step": 16192 }, { "epoch": 100.57943925233644, "grad_norm": 2.546875, "learning_rate": 2.516999829234587e-05, "loss": 0.1716, "step": 16193 }, { "epoch": 100.58566978193146, "grad_norm": 3.546875, "learning_rate": 2.5167534586131437e-05, "loss": 0.2423, "step": 16194 }, { "epoch": 100.59190031152647, "grad_norm": 1.96875, "learning_rate": 2.516507087828987e-05, "loss": 0.1363, "step": 16195 }, { "epoch": 100.59813084112149, "grad_norm": 2.796875, "learning_rate": 2.5162607168845103e-05, "loss": 0.1413, "step": 16196 }, { "epoch": 100.60436137071652, "grad_norm": 2.515625, "learning_rate": 2.5160143457821068e-05, "loss": 0.1596, "step": 16197 }, { "epoch": 100.61059190031153, "grad_norm": 2.625, "learning_rate": 2.515767974524168e-05, "loss": 0.2148, "step": 16198 }, { "epoch": 100.61682242990655, "grad_norm": 1.328125, "learning_rate": 2.515521603113088e-05, "loss": 0.1114, "step": 16199 }, { "epoch": 100.62305295950156, "grad_norm": 1.8671875, "learning_rate": 2.515275231551259e-05, "loss": 0.1219, "step": 16200 }, { "epoch": 100.62928348909658, "grad_norm": 2.375, "learning_rate": 2.5150288598410738e-05, "loss": 0.241, "step": 16201 }, { "epoch": 100.6355140186916, "grad_norm": 2.375, "learning_rate": 2.514782487984926e-05, "loss": 0.1807, "step": 16202 }, { "epoch": 100.64174454828661, "grad_norm": 1.8515625, "learning_rate": 2.514536115985206e-05, "loss": 0.1507, "step": 16203 }, { "epoch": 100.64797507788163, "grad_norm": 3.109375, "learning_rate": 2.5142897438443103e-05, "loss": 0.3063, "step": 16204 }, { "epoch": 100.65420560747664, "grad_norm": 2.203125, "learning_rate": 2.5140433715646283e-05, "loss": 0.137, "step": 16205 }, { "epoch": 100.66043613707166, "grad_norm": 2.1875, "learning_rate": 2.513796999148555e-05, "loss": 0.1162, "step": 16206 }, { "epoch": 100.66666666666667, "grad_norm": 2.109375, "learning_rate": 2.5135506265984822e-05, "loss": 0.1364, "step": 16207 }, { "epoch": 100.67289719626169, "grad_norm": 1.734375, "learning_rate": 2.5133042539168034e-05, "loss": 0.1038, "step": 16208 }, { "epoch": 100.6791277258567, "grad_norm": 2.03125, "learning_rate": 2.51305788110591e-05, "loss": 0.1283, "step": 16209 }, { "epoch": 100.68535825545172, "grad_norm": 2.046875, "learning_rate": 2.5128115081681974e-05, "loss": 0.1648, "step": 16210 }, { "epoch": 100.69158878504673, "grad_norm": 2.90625, "learning_rate": 2.5125651351060552e-05, "loss": 0.1716, "step": 16211 }, { "epoch": 100.69781931464175, "grad_norm": 2.65625, "learning_rate": 2.5123187619218792e-05, "loss": 0.1337, "step": 16212 }, { "epoch": 100.70404984423676, "grad_norm": 1.640625, "learning_rate": 2.512072388618061e-05, "loss": 0.1042, "step": 16213 }, { "epoch": 100.71028037383178, "grad_norm": 2.640625, "learning_rate": 2.5118260151969925e-05, "loss": 0.2142, "step": 16214 }, { "epoch": 100.7165109034268, "grad_norm": 2.453125, "learning_rate": 2.5115796416610675e-05, "loss": 0.1102, "step": 16215 }, { "epoch": 100.72274143302181, "grad_norm": 2.109375, "learning_rate": 2.5113332680126795e-05, "loss": 0.1344, "step": 16216 }, { "epoch": 100.72897196261682, "grad_norm": 3.171875, "learning_rate": 2.51108689425422e-05, "loss": 0.2002, "step": 16217 }, { "epoch": 100.73520249221184, "grad_norm": 2.625, "learning_rate": 2.5108405203880812e-05, "loss": 0.2136, "step": 16218 }, { "epoch": 100.74143302180686, "grad_norm": 2.890625, "learning_rate": 2.510594146416659e-05, "loss": 0.2426, "step": 16219 }, { "epoch": 100.74766355140187, "grad_norm": 2.3125, "learning_rate": 2.5103477723423435e-05, "loss": 0.0999, "step": 16220 }, { "epoch": 100.75389408099689, "grad_norm": 1.7734375, "learning_rate": 2.510101398167528e-05, "loss": 0.1007, "step": 16221 }, { "epoch": 100.7601246105919, "grad_norm": 2.21875, "learning_rate": 2.509855023894606e-05, "loss": 0.0989, "step": 16222 }, { "epoch": 100.76635514018692, "grad_norm": 1.671875, "learning_rate": 2.5096086495259707e-05, "loss": 0.1372, "step": 16223 }, { "epoch": 100.77258566978193, "grad_norm": 2.734375, "learning_rate": 2.5093622750640134e-05, "loss": 0.2541, "step": 16224 }, { "epoch": 100.77881619937695, "grad_norm": 2.84375, "learning_rate": 2.509115900511128e-05, "loss": 0.1339, "step": 16225 }, { "epoch": 100.78504672897196, "grad_norm": 2.359375, "learning_rate": 2.5088695258697075e-05, "loss": 0.1485, "step": 16226 }, { "epoch": 100.79127725856698, "grad_norm": 1.75, "learning_rate": 2.5086231511421443e-05, "loss": 0.1643, "step": 16227 }, { "epoch": 100.797507788162, "grad_norm": 2.03125, "learning_rate": 2.5083767763308313e-05, "loss": 0.1438, "step": 16228 }, { "epoch": 100.80373831775701, "grad_norm": 1.71875, "learning_rate": 2.5081304014381614e-05, "loss": 0.1474, "step": 16229 }, { "epoch": 100.80996884735202, "grad_norm": 1.53125, "learning_rate": 2.507884026466527e-05, "loss": 0.1517, "step": 16230 }, { "epoch": 100.81619937694704, "grad_norm": 1.75, "learning_rate": 2.5076376514183213e-05, "loss": 0.1324, "step": 16231 }, { "epoch": 100.82242990654206, "grad_norm": 2.234375, "learning_rate": 2.5073912762959384e-05, "loss": 0.2222, "step": 16232 }, { "epoch": 100.82866043613707, "grad_norm": 2.03125, "learning_rate": 2.507144901101769e-05, "loss": 0.1218, "step": 16233 }, { "epoch": 100.83489096573209, "grad_norm": 1.453125, "learning_rate": 2.5068985258382067e-05, "loss": 0.1493, "step": 16234 }, { "epoch": 100.8411214953271, "grad_norm": 1.5078125, "learning_rate": 2.506652150507645e-05, "loss": 0.1266, "step": 16235 }, { "epoch": 100.84735202492212, "grad_norm": 1.7734375, "learning_rate": 2.5064057751124764e-05, "loss": 0.146, "step": 16236 }, { "epoch": 100.85358255451713, "grad_norm": 2.671875, "learning_rate": 2.5061593996550923e-05, "loss": 0.2266, "step": 16237 }, { "epoch": 100.85981308411215, "grad_norm": 1.4921875, "learning_rate": 2.5059130241378885e-05, "loss": 0.1403, "step": 16238 }, { "epoch": 100.86604361370716, "grad_norm": 3.6875, "learning_rate": 2.505666648563256e-05, "loss": 0.209, "step": 16239 }, { "epoch": 100.87227414330218, "grad_norm": 2.34375, "learning_rate": 2.5054202729335873e-05, "loss": 0.1424, "step": 16240 }, { "epoch": 100.8785046728972, "grad_norm": 1.65625, "learning_rate": 2.505173897251276e-05, "loss": 0.1072, "step": 16241 }, { "epoch": 100.88473520249221, "grad_norm": 1.6015625, "learning_rate": 2.504927521518715e-05, "loss": 0.098, "step": 16242 }, { "epoch": 100.89096573208722, "grad_norm": 2.1875, "learning_rate": 2.5046811457382967e-05, "loss": 0.0943, "step": 16243 }, { "epoch": 100.89719626168224, "grad_norm": 1.9921875, "learning_rate": 2.5044347699124144e-05, "loss": 0.1436, "step": 16244 }, { "epoch": 100.90342679127725, "grad_norm": 1.3203125, "learning_rate": 2.5041883940434608e-05, "loss": 0.0861, "step": 16245 }, { "epoch": 100.90965732087227, "grad_norm": 2.984375, "learning_rate": 2.503942018133828e-05, "loss": 0.1207, "step": 16246 }, { "epoch": 100.91588785046729, "grad_norm": 2.65625, "learning_rate": 2.5036956421859102e-05, "loss": 0.175, "step": 16247 }, { "epoch": 100.9221183800623, "grad_norm": 0.96875, "learning_rate": 2.5034492662020992e-05, "loss": 0.0887, "step": 16248 }, { "epoch": 100.92834890965732, "grad_norm": 2.15625, "learning_rate": 2.503202890184789e-05, "loss": 0.1812, "step": 16249 }, { "epoch": 100.93457943925233, "grad_norm": 2.09375, "learning_rate": 2.502956514136371e-05, "loss": 0.2292, "step": 16250 }, { "epoch": 100.94080996884735, "grad_norm": 2.265625, "learning_rate": 2.5027101380592387e-05, "loss": 0.1308, "step": 16251 }, { "epoch": 100.94704049844236, "grad_norm": 2.1875, "learning_rate": 2.5024637619557857e-05, "loss": 0.1193, "step": 16252 }, { "epoch": 100.95327102803738, "grad_norm": 2.453125, "learning_rate": 2.5022173858284027e-05, "loss": 0.1787, "step": 16253 }, { "epoch": 100.9595015576324, "grad_norm": 2.15625, "learning_rate": 2.501971009679485e-05, "loss": 0.1173, "step": 16254 }, { "epoch": 100.96573208722741, "grad_norm": 2.390625, "learning_rate": 2.5017246335114248e-05, "loss": 0.2307, "step": 16255 }, { "epoch": 100.97196261682242, "grad_norm": 1.8671875, "learning_rate": 2.501478257326614e-05, "loss": 0.1353, "step": 16256 }, { "epoch": 100.97819314641744, "grad_norm": 1.1875, "learning_rate": 2.5012318811274456e-05, "loss": 0.1124, "step": 16257 }, { "epoch": 100.98442367601245, "grad_norm": 2.203125, "learning_rate": 2.5009855049163144e-05, "loss": 0.2628, "step": 16258 }, { "epoch": 100.99065420560747, "grad_norm": 1.6953125, "learning_rate": 2.5007391286956107e-05, "loss": 0.0957, "step": 16259 }, { "epoch": 100.99688473520249, "grad_norm": 2.484375, "learning_rate": 2.5004927524677285e-05, "loss": 0.1904, "step": 16260 }, { "epoch": 101.0, "grad_norm": 0.9609375, "learning_rate": 2.500246376235061e-05, "loss": 0.0622, "step": 16261 }, { "epoch": 101.00623052959502, "grad_norm": 2.671875, "learning_rate": 2.5e-05, "loss": 0.1926, "step": 16262 }, { "epoch": 101.01246105919003, "grad_norm": 2.734375, "learning_rate": 2.4997536237649396e-05, "loss": 0.2586, "step": 16263 }, { "epoch": 101.01869158878505, "grad_norm": 1.5546875, "learning_rate": 2.4995072475322714e-05, "loss": 0.1127, "step": 16264 }, { "epoch": 101.02492211838006, "grad_norm": 2.796875, "learning_rate": 2.4992608713043896e-05, "loss": 0.2081, "step": 16265 }, { "epoch": 101.03115264797508, "grad_norm": 1.3984375, "learning_rate": 2.499014495083687e-05, "loss": 0.14, "step": 16266 }, { "epoch": 101.03738317757009, "grad_norm": 1.1875, "learning_rate": 2.4987681188725547e-05, "loss": 0.1065, "step": 16267 }, { "epoch": 101.04361370716511, "grad_norm": 1.3984375, "learning_rate": 2.4985217426733866e-05, "loss": 0.1159, "step": 16268 }, { "epoch": 101.04984423676012, "grad_norm": 2.296875, "learning_rate": 2.498275366488576e-05, "loss": 0.1768, "step": 16269 }, { "epoch": 101.05607476635514, "grad_norm": 2.265625, "learning_rate": 2.4980289903205152e-05, "loss": 0.147, "step": 16270 }, { "epoch": 101.06230529595015, "grad_norm": 1.15625, "learning_rate": 2.4977826141715972e-05, "loss": 0.1109, "step": 16271 }, { "epoch": 101.06853582554517, "grad_norm": 2.765625, "learning_rate": 2.497536238044216e-05, "loss": 0.184, "step": 16272 }, { "epoch": 101.07476635514018, "grad_norm": 2.953125, "learning_rate": 2.4972898619407622e-05, "loss": 0.2337, "step": 16273 }, { "epoch": 101.0809968847352, "grad_norm": 2.953125, "learning_rate": 2.4970434858636296e-05, "loss": 0.192, "step": 16274 }, { "epoch": 101.08722741433021, "grad_norm": 2.125, "learning_rate": 2.4967971098152117e-05, "loss": 0.1364, "step": 16275 }, { "epoch": 101.09345794392523, "grad_norm": 1.75, "learning_rate": 2.496550733797901e-05, "loss": 0.1273, "step": 16276 }, { "epoch": 101.09968847352025, "grad_norm": 1.1171875, "learning_rate": 2.49630435781409e-05, "loss": 0.1082, "step": 16277 }, { "epoch": 101.10591900311526, "grad_norm": 2.625, "learning_rate": 2.4960579818661726e-05, "loss": 0.2134, "step": 16278 }, { "epoch": 101.11214953271028, "grad_norm": 1.921875, "learning_rate": 2.4958116059565405e-05, "loss": 0.1573, "step": 16279 }, { "epoch": 101.11838006230529, "grad_norm": 2.140625, "learning_rate": 2.4955652300875865e-05, "loss": 0.1501, "step": 16280 }, { "epoch": 101.1246105919003, "grad_norm": 1.4140625, "learning_rate": 2.495318854261704e-05, "loss": 0.1059, "step": 16281 }, { "epoch": 101.13084112149532, "grad_norm": 1.671875, "learning_rate": 2.4950724784812855e-05, "loss": 0.1169, "step": 16282 }, { "epoch": 101.13707165109034, "grad_norm": 2.65625, "learning_rate": 2.494826102748724e-05, "loss": 0.3052, "step": 16283 }, { "epoch": 101.14330218068535, "grad_norm": 3.625, "learning_rate": 2.4945797270664126e-05, "loss": 0.252, "step": 16284 }, { "epoch": 101.14953271028037, "grad_norm": 3.34375, "learning_rate": 2.4943333514367448e-05, "loss": 0.1789, "step": 16285 }, { "epoch": 101.15576323987538, "grad_norm": 3.359375, "learning_rate": 2.494086975862112e-05, "loss": 0.2779, "step": 16286 }, { "epoch": 101.1619937694704, "grad_norm": 2.625, "learning_rate": 2.493840600344908e-05, "loss": 0.2674, "step": 16287 }, { "epoch": 101.16822429906541, "grad_norm": 1.8359375, "learning_rate": 2.4935942248875242e-05, "loss": 0.1011, "step": 16288 }, { "epoch": 101.17445482866043, "grad_norm": 2.453125, "learning_rate": 2.4933478494923554e-05, "loss": 0.2058, "step": 16289 }, { "epoch": 101.18068535825545, "grad_norm": 0.91015625, "learning_rate": 2.4931014741617932e-05, "loss": 0.0922, "step": 16290 }, { "epoch": 101.18691588785046, "grad_norm": 1.25, "learning_rate": 2.4928550988982324e-05, "loss": 0.1121, "step": 16291 }, { "epoch": 101.19314641744548, "grad_norm": 2.09375, "learning_rate": 2.4926087237040628e-05, "loss": 0.1598, "step": 16292 }, { "epoch": 101.19937694704049, "grad_norm": 1.8125, "learning_rate": 2.492362348581679e-05, "loss": 0.1666, "step": 16293 }, { "epoch": 101.20560747663552, "grad_norm": 1.296875, "learning_rate": 2.4921159735334732e-05, "loss": 0.0936, "step": 16294 }, { "epoch": 101.21183800623054, "grad_norm": 1.4375, "learning_rate": 2.4918695985618392e-05, "loss": 0.1028, "step": 16295 }, { "epoch": 101.21806853582555, "grad_norm": 3.59375, "learning_rate": 2.4916232236691693e-05, "loss": 0.3308, "step": 16296 }, { "epoch": 101.22429906542057, "grad_norm": 2.234375, "learning_rate": 2.4913768488578556e-05, "loss": 0.1408, "step": 16297 }, { "epoch": 101.23052959501558, "grad_norm": 1.9765625, "learning_rate": 2.4911304741302928e-05, "loss": 0.143, "step": 16298 }, { "epoch": 101.2367601246106, "grad_norm": 1.421875, "learning_rate": 2.4908840994888725e-05, "loss": 0.105, "step": 16299 }, { "epoch": 101.24299065420561, "grad_norm": 3.546875, "learning_rate": 2.4906377249359872e-05, "loss": 0.183, "step": 16300 }, { "epoch": 101.24922118380063, "grad_norm": 2.171875, "learning_rate": 2.49039135047403e-05, "loss": 0.1637, "step": 16301 }, { "epoch": 101.25545171339564, "grad_norm": 1.1484375, "learning_rate": 2.4901449761053942e-05, "loss": 0.1265, "step": 16302 }, { "epoch": 101.26168224299066, "grad_norm": 2.3125, "learning_rate": 2.4898986018324716e-05, "loss": 0.1082, "step": 16303 }, { "epoch": 101.26791277258567, "grad_norm": 2.25, "learning_rate": 2.4896522276576577e-05, "loss": 0.1593, "step": 16304 }, { "epoch": 101.27414330218069, "grad_norm": 2.203125, "learning_rate": 2.4894058535833416e-05, "loss": 0.1665, "step": 16305 }, { "epoch": 101.2803738317757, "grad_norm": 0.97265625, "learning_rate": 2.489159479611919e-05, "loss": 0.1027, "step": 16306 }, { "epoch": 101.28660436137072, "grad_norm": 2.03125, "learning_rate": 2.4889131057457806e-05, "loss": 0.1485, "step": 16307 }, { "epoch": 101.29283489096574, "grad_norm": 1.90625, "learning_rate": 2.4886667319873214e-05, "loss": 0.1284, "step": 16308 }, { "epoch": 101.29906542056075, "grad_norm": 2.453125, "learning_rate": 2.4884203583389327e-05, "loss": 0.1838, "step": 16309 }, { "epoch": 101.30529595015577, "grad_norm": 1.4140625, "learning_rate": 2.4881739848030087e-05, "loss": 0.1031, "step": 16310 }, { "epoch": 101.31152647975078, "grad_norm": 1.8046875, "learning_rate": 2.48792761138194e-05, "loss": 0.1134, "step": 16311 }, { "epoch": 101.3177570093458, "grad_norm": 1.859375, "learning_rate": 2.4876812380781213e-05, "loss": 0.1247, "step": 16312 }, { "epoch": 101.32398753894081, "grad_norm": 2.546875, "learning_rate": 2.487434864893945e-05, "loss": 0.1331, "step": 16313 }, { "epoch": 101.33021806853583, "grad_norm": 2.515625, "learning_rate": 2.487188491831803e-05, "loss": 0.1903, "step": 16314 }, { "epoch": 101.33644859813084, "grad_norm": 1.703125, "learning_rate": 2.4869421188940897e-05, "loss": 0.1324, "step": 16315 }, { "epoch": 101.34267912772586, "grad_norm": 1.3984375, "learning_rate": 2.4866957460831965e-05, "loss": 0.1381, "step": 16316 }, { "epoch": 101.34890965732087, "grad_norm": 1.796875, "learning_rate": 2.4864493734015187e-05, "loss": 0.0976, "step": 16317 }, { "epoch": 101.35514018691589, "grad_norm": 3.015625, "learning_rate": 2.4862030008514455e-05, "loss": 0.1166, "step": 16318 }, { "epoch": 101.3613707165109, "grad_norm": 3.75, "learning_rate": 2.4859566284353723e-05, "loss": 0.1762, "step": 16319 }, { "epoch": 101.36760124610592, "grad_norm": 1.3359375, "learning_rate": 2.4857102561556903e-05, "loss": 0.1043, "step": 16320 }, { "epoch": 101.37383177570094, "grad_norm": 2.6875, "learning_rate": 2.485463884014794e-05, "loss": 0.3619, "step": 16321 }, { "epoch": 101.38006230529595, "grad_norm": 2.546875, "learning_rate": 2.485217512015075e-05, "loss": 0.2126, "step": 16322 }, { "epoch": 101.38629283489097, "grad_norm": 2.515625, "learning_rate": 2.484971140158927e-05, "loss": 0.1922, "step": 16323 }, { "epoch": 101.39252336448598, "grad_norm": 3.390625, "learning_rate": 2.4847247684487416e-05, "loss": 0.3925, "step": 16324 }, { "epoch": 101.398753894081, "grad_norm": 2.65625, "learning_rate": 2.4844783968869126e-05, "loss": 0.1842, "step": 16325 }, { "epoch": 101.40498442367601, "grad_norm": 3.359375, "learning_rate": 2.4842320254758324e-05, "loss": 0.2718, "step": 16326 }, { "epoch": 101.41121495327103, "grad_norm": 1.2265625, "learning_rate": 2.4839856542178934e-05, "loss": 0.1041, "step": 16327 }, { "epoch": 101.41744548286604, "grad_norm": 3.3125, "learning_rate": 2.48373928311549e-05, "loss": 0.1896, "step": 16328 }, { "epoch": 101.42367601246106, "grad_norm": 2.09375, "learning_rate": 2.4834929121710138e-05, "loss": 0.2481, "step": 16329 }, { "epoch": 101.42990654205607, "grad_norm": 1.5078125, "learning_rate": 2.4832465413868576e-05, "loss": 0.1308, "step": 16330 }, { "epoch": 101.43613707165109, "grad_norm": 2.234375, "learning_rate": 2.4830001707654134e-05, "loss": 0.1476, "step": 16331 }, { "epoch": 101.4423676012461, "grad_norm": 2.390625, "learning_rate": 2.482753800309076e-05, "loss": 0.222, "step": 16332 }, { "epoch": 101.44859813084112, "grad_norm": 2.078125, "learning_rate": 2.482507430020236e-05, "loss": 0.108, "step": 16333 }, { "epoch": 101.45482866043614, "grad_norm": 4.71875, "learning_rate": 2.4822610599012887e-05, "loss": 0.2888, "step": 16334 }, { "epoch": 101.46105919003115, "grad_norm": 2.578125, "learning_rate": 2.4820146899546248e-05, "loss": 0.2203, "step": 16335 }, { "epoch": 101.46728971962617, "grad_norm": 2.40625, "learning_rate": 2.4817683201826387e-05, "loss": 0.1898, "step": 16336 }, { "epoch": 101.47352024922118, "grad_norm": 2.125, "learning_rate": 2.4815219505877215e-05, "loss": 0.1767, "step": 16337 }, { "epoch": 101.4797507788162, "grad_norm": 1.421875, "learning_rate": 2.4812755811722665e-05, "loss": 0.1094, "step": 16338 }, { "epoch": 101.48598130841121, "grad_norm": 1.1953125, "learning_rate": 2.4810292119386675e-05, "loss": 0.0962, "step": 16339 }, { "epoch": 101.49221183800623, "grad_norm": 1.671875, "learning_rate": 2.4807828428893157e-05, "loss": 0.1376, "step": 16340 }, { "epoch": 101.49844236760124, "grad_norm": 2.40625, "learning_rate": 2.4805364740266053e-05, "loss": 0.18, "step": 16341 }, { "epoch": 101.50467289719626, "grad_norm": 2.515625, "learning_rate": 2.4802901053529293e-05, "loss": 0.2163, "step": 16342 }, { "epoch": 101.51090342679127, "grad_norm": 2.640625, "learning_rate": 2.4800437368706793e-05, "loss": 0.2139, "step": 16343 }, { "epoch": 101.51713395638629, "grad_norm": 1.984375, "learning_rate": 2.4797973685822477e-05, "loss": 0.1219, "step": 16344 }, { "epoch": 101.5233644859813, "grad_norm": 2.171875, "learning_rate": 2.479551000490029e-05, "loss": 0.1345, "step": 16345 }, { "epoch": 101.52959501557632, "grad_norm": 1.4921875, "learning_rate": 2.4793046325964144e-05, "loss": 0.0844, "step": 16346 }, { "epoch": 101.53582554517133, "grad_norm": 3.109375, "learning_rate": 2.4790582649037973e-05, "loss": 0.2086, "step": 16347 }, { "epoch": 101.54205607476635, "grad_norm": 3.09375, "learning_rate": 2.4788118974145715e-05, "loss": 0.2077, "step": 16348 }, { "epoch": 101.54828660436137, "grad_norm": 4.6875, "learning_rate": 2.4785655301311282e-05, "loss": 0.3038, "step": 16349 }, { "epoch": 101.55451713395638, "grad_norm": 1.3359375, "learning_rate": 2.478319163055861e-05, "loss": 0.1199, "step": 16350 }, { "epoch": 101.5607476635514, "grad_norm": 3.171875, "learning_rate": 2.4780727961911617e-05, "loss": 0.2597, "step": 16351 }, { "epoch": 101.56697819314641, "grad_norm": 2.5, "learning_rate": 2.4778264295394244e-05, "loss": 0.1462, "step": 16352 }, { "epoch": 101.57320872274143, "grad_norm": 1.2578125, "learning_rate": 2.477580063103041e-05, "loss": 0.1076, "step": 16353 }, { "epoch": 101.57943925233644, "grad_norm": 2.859375, "learning_rate": 2.4773336968844045e-05, "loss": 0.2084, "step": 16354 }, { "epoch": 101.58566978193146, "grad_norm": 1.890625, "learning_rate": 2.477087330885908e-05, "loss": 0.1054, "step": 16355 }, { "epoch": 101.59190031152647, "grad_norm": 2.53125, "learning_rate": 2.476840965109944e-05, "loss": 0.2806, "step": 16356 }, { "epoch": 101.59813084112149, "grad_norm": 2.578125, "learning_rate": 2.4765945995589044e-05, "loss": 0.1765, "step": 16357 }, { "epoch": 101.60436137071652, "grad_norm": 2.0625, "learning_rate": 2.4763482342351834e-05, "loss": 0.147, "step": 16358 }, { "epoch": 101.61059190031153, "grad_norm": 1.875, "learning_rate": 2.476101869141173e-05, "loss": 0.1645, "step": 16359 }, { "epoch": 101.61682242990655, "grad_norm": 2.671875, "learning_rate": 2.4758555042792654e-05, "loss": 0.2644, "step": 16360 }, { "epoch": 101.62305295950156, "grad_norm": 2.015625, "learning_rate": 2.475609139651855e-05, "loss": 0.1078, "step": 16361 }, { "epoch": 101.62928348909658, "grad_norm": 2.8125, "learning_rate": 2.475362775261333e-05, "loss": 0.1725, "step": 16362 }, { "epoch": 101.6355140186916, "grad_norm": 2.828125, "learning_rate": 2.4751164111100927e-05, "loss": 0.2407, "step": 16363 }, { "epoch": 101.64174454828661, "grad_norm": 2.96875, "learning_rate": 2.4748700472005267e-05, "loss": 0.1787, "step": 16364 }, { "epoch": 101.64797507788163, "grad_norm": 1.1171875, "learning_rate": 2.4746236835350275e-05, "loss": 0.1066, "step": 16365 }, { "epoch": 101.65420560747664, "grad_norm": 1.4296875, "learning_rate": 2.474377320115988e-05, "loss": 0.0976, "step": 16366 }, { "epoch": 101.66043613707166, "grad_norm": 1.703125, "learning_rate": 2.474130956945802e-05, "loss": 0.1162, "step": 16367 }, { "epoch": 101.66666666666667, "grad_norm": 2.5, "learning_rate": 2.4738845940268616e-05, "loss": 0.2254, "step": 16368 }, { "epoch": 101.67289719626169, "grad_norm": 1.828125, "learning_rate": 2.4736382313615585e-05, "loss": 0.1472, "step": 16369 }, { "epoch": 101.6791277258567, "grad_norm": 3.25, "learning_rate": 2.4733918689522857e-05, "loss": 0.1027, "step": 16370 }, { "epoch": 101.68535825545172, "grad_norm": 2.171875, "learning_rate": 2.473145506801437e-05, "loss": 0.1393, "step": 16371 }, { "epoch": 101.69158878504673, "grad_norm": 1.6953125, "learning_rate": 2.4728991449114045e-05, "loss": 0.1214, "step": 16372 }, { "epoch": 101.69781931464175, "grad_norm": 2.453125, "learning_rate": 2.4726527832845804e-05, "loss": 0.2281, "step": 16373 }, { "epoch": 101.70404984423676, "grad_norm": 2.171875, "learning_rate": 2.4724064219233588e-05, "loss": 0.1201, "step": 16374 }, { "epoch": 101.71028037383178, "grad_norm": 2.203125, "learning_rate": 2.4721600608301316e-05, "loss": 0.1669, "step": 16375 }, { "epoch": 101.7165109034268, "grad_norm": 1.828125, "learning_rate": 2.471913700007291e-05, "loss": 0.1319, "step": 16376 }, { "epoch": 101.72274143302181, "grad_norm": 2.015625, "learning_rate": 2.4716673394572297e-05, "loss": 0.1451, "step": 16377 }, { "epoch": 101.72897196261682, "grad_norm": 2.515625, "learning_rate": 2.471420979182341e-05, "loss": 0.1054, "step": 16378 }, { "epoch": 101.73520249221184, "grad_norm": 1.65625, "learning_rate": 2.4711746191850173e-05, "loss": 0.0969, "step": 16379 }, { "epoch": 101.74143302180686, "grad_norm": 2.34375, "learning_rate": 2.4709282594676535e-05, "loss": 0.1592, "step": 16380 }, { "epoch": 101.74766355140187, "grad_norm": 4.71875, "learning_rate": 2.470681900032638e-05, "loss": 0.3827, "step": 16381 }, { "epoch": 101.75389408099689, "grad_norm": 2.546875, "learning_rate": 2.4704355408823664e-05, "loss": 0.1542, "step": 16382 }, { "epoch": 101.7601246105919, "grad_norm": 2.484375, "learning_rate": 2.470189182019231e-05, "loss": 0.2106, "step": 16383 }, { "epoch": 101.76635514018692, "grad_norm": 1.703125, "learning_rate": 2.4699428234456236e-05, "loss": 0.1199, "step": 16384 }, { "epoch": 101.77258566978193, "grad_norm": 1.9453125, "learning_rate": 2.469696465163938e-05, "loss": 0.1327, "step": 16385 }, { "epoch": 101.77881619937695, "grad_norm": 2.703125, "learning_rate": 2.4694501071765657e-05, "loss": 0.1142, "step": 16386 }, { "epoch": 101.78504672897196, "grad_norm": 2.484375, "learning_rate": 2.4692037494859018e-05, "loss": 0.1395, "step": 16387 }, { "epoch": 101.79127725856698, "grad_norm": 3.21875, "learning_rate": 2.4689573920943356e-05, "loss": 0.1818, "step": 16388 }, { "epoch": 101.797507788162, "grad_norm": 1.9296875, "learning_rate": 2.4687110350042624e-05, "loss": 0.1107, "step": 16389 }, { "epoch": 101.80373831775701, "grad_norm": 3.28125, "learning_rate": 2.4684646782180727e-05, "loss": 0.1643, "step": 16390 }, { "epoch": 101.80996884735202, "grad_norm": 3.40625, "learning_rate": 2.468218321738161e-05, "loss": 0.3196, "step": 16391 }, { "epoch": 101.81619937694704, "grad_norm": 2.984375, "learning_rate": 2.4679719655669196e-05, "loss": 0.2087, "step": 16392 }, { "epoch": 101.82242990654206, "grad_norm": 1.546875, "learning_rate": 2.4677256097067412e-05, "loss": 0.118, "step": 16393 }, { "epoch": 101.82866043613707, "grad_norm": 2.15625, "learning_rate": 2.467479254160017e-05, "loss": 0.1785, "step": 16394 }, { "epoch": 101.83489096573209, "grad_norm": 2.28125, "learning_rate": 2.4672328989291413e-05, "loss": 0.1086, "step": 16395 }, { "epoch": 101.8411214953271, "grad_norm": 3.28125, "learning_rate": 2.4669865440165065e-05, "loss": 0.3098, "step": 16396 }, { "epoch": 101.84735202492212, "grad_norm": 2.578125, "learning_rate": 2.4667401894245042e-05, "loss": 0.0886, "step": 16397 }, { "epoch": 101.85358255451713, "grad_norm": 3.890625, "learning_rate": 2.4664938351555288e-05, "loss": 0.3014, "step": 16398 }, { "epoch": 101.85981308411215, "grad_norm": 1.671875, "learning_rate": 2.4662474812119718e-05, "loss": 0.1343, "step": 16399 }, { "epoch": 101.86604361370716, "grad_norm": 1.46875, "learning_rate": 2.4660011275962258e-05, "loss": 0.1366, "step": 16400 }, { "epoch": 101.87227414330218, "grad_norm": 2.171875, "learning_rate": 2.465754774310683e-05, "loss": 0.2395, "step": 16401 }, { "epoch": 101.8785046728972, "grad_norm": 1.6796875, "learning_rate": 2.4655084213577376e-05, "loss": 0.1236, "step": 16402 }, { "epoch": 101.88473520249221, "grad_norm": 2.5625, "learning_rate": 2.4652620687397808e-05, "loss": 0.1885, "step": 16403 }, { "epoch": 101.89096573208722, "grad_norm": 1.3671875, "learning_rate": 2.4650157164592056e-05, "loss": 0.1077, "step": 16404 }, { "epoch": 101.89719626168224, "grad_norm": 1.6953125, "learning_rate": 2.4647693645184053e-05, "loss": 0.1287, "step": 16405 }, { "epoch": 101.90342679127725, "grad_norm": 2.5625, "learning_rate": 2.4645230129197723e-05, "loss": 0.3021, "step": 16406 }, { "epoch": 101.90965732087227, "grad_norm": 2.90625, "learning_rate": 2.4642766616656978e-05, "loss": 0.1003, "step": 16407 }, { "epoch": 101.91588785046729, "grad_norm": 2.296875, "learning_rate": 2.4640303107585765e-05, "loss": 0.1132, "step": 16408 }, { "epoch": 101.9221183800623, "grad_norm": 2.5, "learning_rate": 2.4637839602007996e-05, "loss": 0.1825, "step": 16409 }, { "epoch": 101.92834890965732, "grad_norm": 2.125, "learning_rate": 2.4635376099947593e-05, "loss": 0.1489, "step": 16410 }, { "epoch": 101.93457943925233, "grad_norm": 2.3125, "learning_rate": 2.4632912601428502e-05, "loss": 0.1622, "step": 16411 }, { "epoch": 101.94080996884735, "grad_norm": 1.875, "learning_rate": 2.4630449106474638e-05, "loss": 0.1719, "step": 16412 }, { "epoch": 101.94704049844236, "grad_norm": 2.390625, "learning_rate": 2.4627985615109923e-05, "loss": 0.1904, "step": 16413 }, { "epoch": 101.95327102803738, "grad_norm": 1.8671875, "learning_rate": 2.462552212735828e-05, "loss": 0.1551, "step": 16414 }, { "epoch": 101.9595015576324, "grad_norm": 1.765625, "learning_rate": 2.4623058643243647e-05, "loss": 0.167, "step": 16415 }, { "epoch": 101.96573208722741, "grad_norm": 1.484375, "learning_rate": 2.4620595162789936e-05, "loss": 0.1292, "step": 16416 }, { "epoch": 101.97196261682242, "grad_norm": 2.765625, "learning_rate": 2.4618131686021087e-05, "loss": 0.2449, "step": 16417 }, { "epoch": 101.97819314641744, "grad_norm": 2.265625, "learning_rate": 2.4615668212961028e-05, "loss": 0.1967, "step": 16418 }, { "epoch": 101.98442367601245, "grad_norm": 0.92578125, "learning_rate": 2.4613204743633668e-05, "loss": 0.0916, "step": 16419 }, { "epoch": 101.99065420560747, "grad_norm": 3.328125, "learning_rate": 2.4610741278062933e-05, "loss": 0.3475, "step": 16420 }, { "epoch": 101.99688473520249, "grad_norm": 2.15625, "learning_rate": 2.4608277816272766e-05, "loss": 0.1006, "step": 16421 }, { "epoch": 102.0, "grad_norm": 1.5703125, "learning_rate": 2.460581435828708e-05, "loss": 0.0827, "step": 16422 }, { "epoch": 102.00623052959502, "grad_norm": 1.9375, "learning_rate": 2.46033509041298e-05, "loss": 0.1531, "step": 16423 }, { "epoch": 102.01246105919003, "grad_norm": 2.046875, "learning_rate": 2.4600887453824862e-05, "loss": 0.1255, "step": 16424 }, { "epoch": 102.01869158878505, "grad_norm": 2.734375, "learning_rate": 2.4598424007396187e-05, "loss": 0.2201, "step": 16425 }, { "epoch": 102.02492211838006, "grad_norm": 2.015625, "learning_rate": 2.4595960564867696e-05, "loss": 0.1446, "step": 16426 }, { "epoch": 102.03115264797508, "grad_norm": 2.34375, "learning_rate": 2.459349712626331e-05, "loss": 0.1638, "step": 16427 }, { "epoch": 102.03738317757009, "grad_norm": 2.59375, "learning_rate": 2.4591033691606968e-05, "loss": 0.2297, "step": 16428 }, { "epoch": 102.04361370716511, "grad_norm": 1.8125, "learning_rate": 2.4588570260922585e-05, "loss": 0.1492, "step": 16429 }, { "epoch": 102.04984423676012, "grad_norm": 1.375, "learning_rate": 2.4586106834234088e-05, "loss": 0.129, "step": 16430 }, { "epoch": 102.05607476635514, "grad_norm": 1.5703125, "learning_rate": 2.4583643411565412e-05, "loss": 0.1246, "step": 16431 }, { "epoch": 102.06230529595015, "grad_norm": 3.5, "learning_rate": 2.458117999294047e-05, "loss": 0.2448, "step": 16432 }, { "epoch": 102.06853582554517, "grad_norm": 0.9375, "learning_rate": 2.4578716578383196e-05, "loss": 0.0898, "step": 16433 }, { "epoch": 102.07476635514018, "grad_norm": 1.4140625, "learning_rate": 2.45762531679175e-05, "loss": 0.1061, "step": 16434 }, { "epoch": 102.0809968847352, "grad_norm": 2.53125, "learning_rate": 2.4573789761567325e-05, "loss": 0.1857, "step": 16435 }, { "epoch": 102.08722741433021, "grad_norm": 2.796875, "learning_rate": 2.4571326359356585e-05, "loss": 0.1819, "step": 16436 }, { "epoch": 102.09345794392523, "grad_norm": 2.40625, "learning_rate": 2.4568862961309212e-05, "loss": 0.2172, "step": 16437 }, { "epoch": 102.09968847352025, "grad_norm": 3.015625, "learning_rate": 2.4566399567449136e-05, "loss": 0.317, "step": 16438 }, { "epoch": 102.10591900311526, "grad_norm": 1.5859375, "learning_rate": 2.4563936177800268e-05, "loss": 0.142, "step": 16439 }, { "epoch": 102.11214953271028, "grad_norm": 2.828125, "learning_rate": 2.456147279238653e-05, "loss": 0.2182, "step": 16440 }, { "epoch": 102.11838006230529, "grad_norm": 3.046875, "learning_rate": 2.4559009411231868e-05, "loss": 0.2234, "step": 16441 }, { "epoch": 102.1246105919003, "grad_norm": 1.6328125, "learning_rate": 2.455654603436019e-05, "loss": 0.106, "step": 16442 }, { "epoch": 102.13084112149532, "grad_norm": 2.0625, "learning_rate": 2.455408266179542e-05, "loss": 0.1225, "step": 16443 }, { "epoch": 102.13707165109034, "grad_norm": 1.9453125, "learning_rate": 2.45516192935615e-05, "loss": 0.1688, "step": 16444 }, { "epoch": 102.14330218068535, "grad_norm": 3.296875, "learning_rate": 2.454915592968234e-05, "loss": 0.0998, "step": 16445 }, { "epoch": 102.14953271028037, "grad_norm": 2.015625, "learning_rate": 2.4546692570181863e-05, "loss": 0.1097, "step": 16446 }, { "epoch": 102.15576323987538, "grad_norm": 1.2421875, "learning_rate": 2.4544229215083997e-05, "loss": 0.1125, "step": 16447 }, { "epoch": 102.1619937694704, "grad_norm": 2.4375, "learning_rate": 2.454176586441267e-05, "loss": 0.1862, "step": 16448 }, { "epoch": 102.16822429906541, "grad_norm": 2.890625, "learning_rate": 2.4539302518191802e-05, "loss": 0.2557, "step": 16449 }, { "epoch": 102.17445482866043, "grad_norm": 2.390625, "learning_rate": 2.4536839176445337e-05, "loss": 0.1613, "step": 16450 }, { "epoch": 102.18068535825545, "grad_norm": 2.015625, "learning_rate": 2.4534375839197166e-05, "loss": 0.1505, "step": 16451 }, { "epoch": 102.18691588785046, "grad_norm": 2.171875, "learning_rate": 2.4531912506471232e-05, "loss": 0.1664, "step": 16452 }, { "epoch": 102.19314641744548, "grad_norm": 1.390625, "learning_rate": 2.4529449178291457e-05, "loss": 0.1166, "step": 16453 }, { "epoch": 102.19937694704049, "grad_norm": 2.28125, "learning_rate": 2.4526985854681766e-05, "loss": 0.1525, "step": 16454 }, { "epoch": 102.20560747663552, "grad_norm": 2.265625, "learning_rate": 2.4524522535666088e-05, "loss": 0.1768, "step": 16455 }, { "epoch": 102.21183800623054, "grad_norm": 1.9296875, "learning_rate": 2.452205922126833e-05, "loss": 0.1555, "step": 16456 }, { "epoch": 102.21806853582555, "grad_norm": 2.6875, "learning_rate": 2.4519595911512444e-05, "loss": 0.2155, "step": 16457 }, { "epoch": 102.22429906542057, "grad_norm": 1.59375, "learning_rate": 2.4517132606422333e-05, "loss": 0.1184, "step": 16458 }, { "epoch": 102.23052959501558, "grad_norm": 2.453125, "learning_rate": 2.4514669306021924e-05, "loss": 0.1886, "step": 16459 }, { "epoch": 102.2367601246106, "grad_norm": 1.9453125, "learning_rate": 2.451220601033514e-05, "loss": 0.1112, "step": 16460 }, { "epoch": 102.24299065420561, "grad_norm": 2.640625, "learning_rate": 2.4509742719385914e-05, "loss": 0.257, "step": 16461 }, { "epoch": 102.24922118380063, "grad_norm": 1.84375, "learning_rate": 2.450727943319816e-05, "loss": 0.1597, "step": 16462 }, { "epoch": 102.25545171339564, "grad_norm": 1.34375, "learning_rate": 2.4504816151795822e-05, "loss": 0.1005, "step": 16463 }, { "epoch": 102.26168224299066, "grad_norm": 1.2421875, "learning_rate": 2.4502352875202793e-05, "loss": 0.1077, "step": 16464 }, { "epoch": 102.26791277258567, "grad_norm": 1.6484375, "learning_rate": 2.4499889603443018e-05, "loss": 0.1475, "step": 16465 }, { "epoch": 102.27414330218069, "grad_norm": 2.65625, "learning_rate": 2.4497426336540408e-05, "loss": 0.26, "step": 16466 }, { "epoch": 102.2803738317757, "grad_norm": 1.8203125, "learning_rate": 2.44949630745189e-05, "loss": 0.1129, "step": 16467 }, { "epoch": 102.28660436137072, "grad_norm": 2.03125, "learning_rate": 2.4492499817402414e-05, "loss": 0.1423, "step": 16468 }, { "epoch": 102.29283489096574, "grad_norm": 1.984375, "learning_rate": 2.4490036565214873e-05, "loss": 0.1498, "step": 16469 }, { "epoch": 102.29906542056075, "grad_norm": 1.7265625, "learning_rate": 2.448757331798019e-05, "loss": 0.1213, "step": 16470 }, { "epoch": 102.30529595015577, "grad_norm": 2.5625, "learning_rate": 2.4485110075722304e-05, "loss": 0.1718, "step": 16471 }, { "epoch": 102.31152647975078, "grad_norm": 2.171875, "learning_rate": 2.448264683846513e-05, "loss": 0.2441, "step": 16472 }, { "epoch": 102.3177570093458, "grad_norm": 3.234375, "learning_rate": 2.4480183606232587e-05, "loss": 0.183, "step": 16473 }, { "epoch": 102.32398753894081, "grad_norm": 1.1484375, "learning_rate": 2.4477720379048612e-05, "loss": 0.1119, "step": 16474 }, { "epoch": 102.33021806853583, "grad_norm": 3.828125, "learning_rate": 2.4475257156937114e-05, "loss": 0.2115, "step": 16475 }, { "epoch": 102.33644859813084, "grad_norm": 2.078125, "learning_rate": 2.447279393992204e-05, "loss": 0.2189, "step": 16476 }, { "epoch": 102.34267912772586, "grad_norm": 1.9140625, "learning_rate": 2.4470330728027282e-05, "loss": 0.1239, "step": 16477 }, { "epoch": 102.34890965732087, "grad_norm": 2.8125, "learning_rate": 2.4467867521276784e-05, "loss": 0.2182, "step": 16478 }, { "epoch": 102.35514018691589, "grad_norm": 3.0, "learning_rate": 2.4465404319694458e-05, "loss": 0.2273, "step": 16479 }, { "epoch": 102.3613707165109, "grad_norm": 3.28125, "learning_rate": 2.446294112330423e-05, "loss": 0.2964, "step": 16480 }, { "epoch": 102.36760124610592, "grad_norm": 2.140625, "learning_rate": 2.4460477932130032e-05, "loss": 0.1863, "step": 16481 }, { "epoch": 102.37383177570094, "grad_norm": 2.28125, "learning_rate": 2.4458014746195786e-05, "loss": 0.1185, "step": 16482 }, { "epoch": 102.38006230529595, "grad_norm": 1.5078125, "learning_rate": 2.44555515655254e-05, "loss": 0.1241, "step": 16483 }, { "epoch": 102.38629283489097, "grad_norm": 1.859375, "learning_rate": 2.4453088390142805e-05, "loss": 0.1382, "step": 16484 }, { "epoch": 102.39252336448598, "grad_norm": 2.484375, "learning_rate": 2.4450625220071927e-05, "loss": 0.181, "step": 16485 }, { "epoch": 102.398753894081, "grad_norm": 1.484375, "learning_rate": 2.444816205533668e-05, "loss": 0.1209, "step": 16486 }, { "epoch": 102.40498442367601, "grad_norm": 3.015625, "learning_rate": 2.4445698895961e-05, "loss": 0.1523, "step": 16487 }, { "epoch": 102.41121495327103, "grad_norm": 2.875, "learning_rate": 2.4443235741968813e-05, "loss": 0.1367, "step": 16488 }, { "epoch": 102.41744548286604, "grad_norm": 3.203125, "learning_rate": 2.444077259338402e-05, "loss": 0.2609, "step": 16489 }, { "epoch": 102.42367601246106, "grad_norm": 2.96875, "learning_rate": 2.4438309450230555e-05, "loss": 0.1516, "step": 16490 }, { "epoch": 102.42990654205607, "grad_norm": 2.984375, "learning_rate": 2.4435846312532342e-05, "loss": 0.1373, "step": 16491 }, { "epoch": 102.43613707165109, "grad_norm": 1.2734375, "learning_rate": 2.4433383180313303e-05, "loss": 0.1299, "step": 16492 }, { "epoch": 102.4423676012461, "grad_norm": 1.1484375, "learning_rate": 2.4430920053597356e-05, "loss": 0.0996, "step": 16493 }, { "epoch": 102.44859813084112, "grad_norm": 2.578125, "learning_rate": 2.442845693240843e-05, "loss": 0.0925, "step": 16494 }, { "epoch": 102.45482866043614, "grad_norm": 1.8515625, "learning_rate": 2.4425993816770456e-05, "loss": 0.135, "step": 16495 }, { "epoch": 102.46105919003115, "grad_norm": 1.921875, "learning_rate": 2.4423530706707336e-05, "loss": 0.1639, "step": 16496 }, { "epoch": 102.46728971962617, "grad_norm": 2.234375, "learning_rate": 2.4421067602242996e-05, "loss": 0.1599, "step": 16497 }, { "epoch": 102.47352024922118, "grad_norm": 2.90625, "learning_rate": 2.441860450340137e-05, "loss": 0.2591, "step": 16498 }, { "epoch": 102.4797507788162, "grad_norm": 4.28125, "learning_rate": 2.441614141020636e-05, "loss": 0.2768, "step": 16499 }, { "epoch": 102.48598130841121, "grad_norm": 3.1875, "learning_rate": 2.4413678322681917e-05, "loss": 0.2782, "step": 16500 }, { "epoch": 102.49221183800623, "grad_norm": 1.71875, "learning_rate": 2.4411215240851947e-05, "loss": 0.1099, "step": 16501 }, { "epoch": 102.49844236760124, "grad_norm": 1.9453125, "learning_rate": 2.4408752164740367e-05, "loss": 0.1159, "step": 16502 }, { "epoch": 102.50467289719626, "grad_norm": 2.03125, "learning_rate": 2.4406289094371102e-05, "loss": 0.2518, "step": 16503 }, { "epoch": 102.51090342679127, "grad_norm": 2.265625, "learning_rate": 2.4403826029768078e-05, "loss": 0.2173, "step": 16504 }, { "epoch": 102.51713395638629, "grad_norm": 2.421875, "learning_rate": 2.4401362970955215e-05, "loss": 0.1416, "step": 16505 }, { "epoch": 102.5233644859813, "grad_norm": 2.8125, "learning_rate": 2.439889991795643e-05, "loss": 0.1049, "step": 16506 }, { "epoch": 102.52959501557632, "grad_norm": 2.375, "learning_rate": 2.4396436870795654e-05, "loss": 0.1861, "step": 16507 }, { "epoch": 102.53582554517133, "grad_norm": 2.203125, "learning_rate": 2.439397382949681e-05, "loss": 0.1699, "step": 16508 }, { "epoch": 102.54205607476635, "grad_norm": 1.5546875, "learning_rate": 2.439151079408381e-05, "loss": 0.1495, "step": 16509 }, { "epoch": 102.54828660436137, "grad_norm": 3.03125, "learning_rate": 2.438904776458057e-05, "loss": 0.2994, "step": 16510 }, { "epoch": 102.55451713395638, "grad_norm": 3.21875, "learning_rate": 2.438658474101103e-05, "loss": 0.2301, "step": 16511 }, { "epoch": 102.5607476635514, "grad_norm": 1.671875, "learning_rate": 2.438412172339909e-05, "loss": 0.1531, "step": 16512 }, { "epoch": 102.56697819314641, "grad_norm": 1.625, "learning_rate": 2.4381658711768693e-05, "loss": 0.1441, "step": 16513 }, { "epoch": 102.57320872274143, "grad_norm": 1.953125, "learning_rate": 2.4379195706143753e-05, "loss": 0.1176, "step": 16514 }, { "epoch": 102.57943925233644, "grad_norm": 2.421875, "learning_rate": 2.4376732706548183e-05, "loss": 0.1736, "step": 16515 }, { "epoch": 102.58566978193146, "grad_norm": 1.234375, "learning_rate": 2.4374269713005907e-05, "loss": 0.104, "step": 16516 }, { "epoch": 102.59190031152647, "grad_norm": 2.140625, "learning_rate": 2.4371806725540852e-05, "loss": 0.1392, "step": 16517 }, { "epoch": 102.59813084112149, "grad_norm": 2.46875, "learning_rate": 2.4369343744176935e-05, "loss": 0.1532, "step": 16518 }, { "epoch": 102.60436137071652, "grad_norm": 3.28125, "learning_rate": 2.4366880768938072e-05, "loss": 0.2189, "step": 16519 }, { "epoch": 102.61059190031153, "grad_norm": 2.265625, "learning_rate": 2.43644177998482e-05, "loss": 0.1884, "step": 16520 }, { "epoch": 102.61682242990655, "grad_norm": 1.7734375, "learning_rate": 2.4361954836931228e-05, "loss": 0.1033, "step": 16521 }, { "epoch": 102.62305295950156, "grad_norm": 1.5625, "learning_rate": 2.4359491880211073e-05, "loss": 0.0959, "step": 16522 }, { "epoch": 102.62928348909658, "grad_norm": 2.484375, "learning_rate": 2.4357028929711658e-05, "loss": 0.2017, "step": 16523 }, { "epoch": 102.6355140186916, "grad_norm": 2.890625, "learning_rate": 2.4354565985456914e-05, "loss": 0.2831, "step": 16524 }, { "epoch": 102.64174454828661, "grad_norm": 1.3828125, "learning_rate": 2.4352103047470754e-05, "loss": 0.1284, "step": 16525 }, { "epoch": 102.64797507788163, "grad_norm": 2.15625, "learning_rate": 2.434964011577709e-05, "loss": 0.1677, "step": 16526 }, { "epoch": 102.65420560747664, "grad_norm": 2.453125, "learning_rate": 2.4347177190399864e-05, "loss": 0.1926, "step": 16527 }, { "epoch": 102.66043613707166, "grad_norm": 2.015625, "learning_rate": 2.4344714271362975e-05, "loss": 0.1562, "step": 16528 }, { "epoch": 102.66666666666667, "grad_norm": 2.4375, "learning_rate": 2.4342251358690357e-05, "loss": 0.2578, "step": 16529 }, { "epoch": 102.67289719626169, "grad_norm": 1.265625, "learning_rate": 2.433978845240592e-05, "loss": 0.0863, "step": 16530 }, { "epoch": 102.6791277258567, "grad_norm": 1.1953125, "learning_rate": 2.4337325552533588e-05, "loss": 0.1318, "step": 16531 }, { "epoch": 102.68535825545172, "grad_norm": 2.359375, "learning_rate": 2.433486265909728e-05, "loss": 0.168, "step": 16532 }, { "epoch": 102.69158878504673, "grad_norm": 1.609375, "learning_rate": 2.4332399772120938e-05, "loss": 0.1304, "step": 16533 }, { "epoch": 102.69781931464175, "grad_norm": 2.828125, "learning_rate": 2.4329936891628447e-05, "loss": 0.2274, "step": 16534 }, { "epoch": 102.70404984423676, "grad_norm": 2.796875, "learning_rate": 2.4327474017643746e-05, "loss": 0.1865, "step": 16535 }, { "epoch": 102.71028037383178, "grad_norm": 2.171875, "learning_rate": 2.4325011150190748e-05, "loss": 0.0969, "step": 16536 }, { "epoch": 102.7165109034268, "grad_norm": 0.96875, "learning_rate": 2.432254828929338e-05, "loss": 0.088, "step": 16537 }, { "epoch": 102.72274143302181, "grad_norm": 2.265625, "learning_rate": 2.432008543497556e-05, "loss": 0.1412, "step": 16538 }, { "epoch": 102.72897196261682, "grad_norm": 1.4140625, "learning_rate": 2.4317622587261213e-05, "loss": 0.1129, "step": 16539 }, { "epoch": 102.73520249221184, "grad_norm": 3.125, "learning_rate": 2.431515974617424e-05, "loss": 0.1702, "step": 16540 }, { "epoch": 102.74143302180686, "grad_norm": 2.859375, "learning_rate": 2.4312696911738574e-05, "loss": 0.1957, "step": 16541 }, { "epoch": 102.74766355140187, "grad_norm": 2.0, "learning_rate": 2.4310234083978136e-05, "loss": 0.1351, "step": 16542 }, { "epoch": 102.75389408099689, "grad_norm": 1.453125, "learning_rate": 2.4307771262916837e-05, "loss": 0.1347, "step": 16543 }, { "epoch": 102.7601246105919, "grad_norm": 2.15625, "learning_rate": 2.4305308448578608e-05, "loss": 0.1976, "step": 16544 }, { "epoch": 102.76635514018692, "grad_norm": 1.34375, "learning_rate": 2.430284564098736e-05, "loss": 0.112, "step": 16545 }, { "epoch": 102.77258566978193, "grad_norm": 1.765625, "learning_rate": 2.4300382840167023e-05, "loss": 0.1608, "step": 16546 }, { "epoch": 102.77881619937695, "grad_norm": 2.78125, "learning_rate": 2.4297920046141494e-05, "loss": 0.1819, "step": 16547 }, { "epoch": 102.78504672897196, "grad_norm": 2.109375, "learning_rate": 2.4295457258934713e-05, "loss": 0.1621, "step": 16548 }, { "epoch": 102.79127725856698, "grad_norm": 2.375, "learning_rate": 2.4292994478570586e-05, "loss": 0.1854, "step": 16549 }, { "epoch": 102.797507788162, "grad_norm": 1.9296875, "learning_rate": 2.4290531705073043e-05, "loss": 0.2035, "step": 16550 }, { "epoch": 102.80373831775701, "grad_norm": 1.546875, "learning_rate": 2.4288068938465998e-05, "loss": 0.1191, "step": 16551 }, { "epoch": 102.80996884735202, "grad_norm": 2.390625, "learning_rate": 2.4285606178773378e-05, "loss": 0.1568, "step": 16552 }, { "epoch": 102.81619937694704, "grad_norm": 2.640625, "learning_rate": 2.428314342601908e-05, "loss": 0.2081, "step": 16553 }, { "epoch": 102.82242990654206, "grad_norm": 3.078125, "learning_rate": 2.4280680680227042e-05, "loss": 0.3008, "step": 16554 }, { "epoch": 102.82866043613707, "grad_norm": 2.734375, "learning_rate": 2.4278217941421177e-05, "loss": 0.1974, "step": 16555 }, { "epoch": 102.83489096573209, "grad_norm": 1.2890625, "learning_rate": 2.4275755209625397e-05, "loss": 0.1033, "step": 16556 }, { "epoch": 102.8411214953271, "grad_norm": 1.5546875, "learning_rate": 2.4273292484863636e-05, "loss": 0.1219, "step": 16557 }, { "epoch": 102.84735202492212, "grad_norm": 0.9296875, "learning_rate": 2.4270829767159804e-05, "loss": 0.1004, "step": 16558 }, { "epoch": 102.85358255451713, "grad_norm": 1.296875, "learning_rate": 2.426836705653782e-05, "loss": 0.1216, "step": 16559 }, { "epoch": 102.85981308411215, "grad_norm": 1.890625, "learning_rate": 2.426590435302159e-05, "loss": 0.1391, "step": 16560 }, { "epoch": 102.86604361370716, "grad_norm": 1.3203125, "learning_rate": 2.4263441656635053e-05, "loss": 0.1, "step": 16561 }, { "epoch": 102.87227414330218, "grad_norm": 2.46875, "learning_rate": 2.426097896740211e-05, "loss": 0.2408, "step": 16562 }, { "epoch": 102.8785046728972, "grad_norm": 1.9765625, "learning_rate": 2.4258516285346693e-05, "loss": 0.1665, "step": 16563 }, { "epoch": 102.88473520249221, "grad_norm": 2.109375, "learning_rate": 2.4256053610492716e-05, "loss": 0.12, "step": 16564 }, { "epoch": 102.89096573208722, "grad_norm": 2.046875, "learning_rate": 2.42535909428641e-05, "loss": 0.1471, "step": 16565 }, { "epoch": 102.89719626168224, "grad_norm": 1.59375, "learning_rate": 2.4251128282484745e-05, "loss": 0.128, "step": 16566 }, { "epoch": 102.90342679127725, "grad_norm": 2.15625, "learning_rate": 2.424866562937859e-05, "loss": 0.1163, "step": 16567 }, { "epoch": 102.90965732087227, "grad_norm": 2.640625, "learning_rate": 2.424620298356954e-05, "loss": 0.185, "step": 16568 }, { "epoch": 102.91588785046729, "grad_norm": 2.734375, "learning_rate": 2.4243740345081517e-05, "loss": 0.1228, "step": 16569 }, { "epoch": 102.9221183800623, "grad_norm": 2.9375, "learning_rate": 2.424127771393844e-05, "loss": 0.1535, "step": 16570 }, { "epoch": 102.92834890965732, "grad_norm": 1.6796875, "learning_rate": 2.4238815090164235e-05, "loss": 0.1485, "step": 16571 }, { "epoch": 102.93457943925233, "grad_norm": 2.53125, "learning_rate": 2.4236352473782803e-05, "loss": 0.2502, "step": 16572 }, { "epoch": 102.94080996884735, "grad_norm": 2.671875, "learning_rate": 2.4233889864818063e-05, "loss": 0.1915, "step": 16573 }, { "epoch": 102.94704049844236, "grad_norm": 1.9140625, "learning_rate": 2.4231427263293944e-05, "loss": 0.1652, "step": 16574 }, { "epoch": 102.95327102803738, "grad_norm": 2.703125, "learning_rate": 2.4228964669234355e-05, "loss": 0.3236, "step": 16575 }, { "epoch": 102.9595015576324, "grad_norm": 3.1875, "learning_rate": 2.422650208266321e-05, "loss": 0.3176, "step": 16576 }, { "epoch": 102.96573208722741, "grad_norm": 2.9375, "learning_rate": 2.4224039503604435e-05, "loss": 0.2496, "step": 16577 }, { "epoch": 102.97196261682242, "grad_norm": 1.5078125, "learning_rate": 2.4221576932081953e-05, "loss": 0.1099, "step": 16578 }, { "epoch": 102.97819314641744, "grad_norm": 3.046875, "learning_rate": 2.4219114368119667e-05, "loss": 0.1513, "step": 16579 }, { "epoch": 102.98442367601245, "grad_norm": 1.390625, "learning_rate": 2.421665181174149e-05, "loss": 0.1179, "step": 16580 }, { "epoch": 102.99065420560747, "grad_norm": 3.65625, "learning_rate": 2.4214189262971352e-05, "loss": 0.3483, "step": 16581 }, { "epoch": 102.99688473520249, "grad_norm": 3.171875, "learning_rate": 2.421172672183316e-05, "loss": 0.316, "step": 16582 }, { "epoch": 103.0, "grad_norm": 0.8046875, "learning_rate": 2.4209264188350844e-05, "loss": 0.0415, "step": 16583 }, { "epoch": 103.00623052959502, "grad_norm": 2.75, "learning_rate": 2.4206801662548314e-05, "loss": 0.1798, "step": 16584 }, { "epoch": 103.01246105919003, "grad_norm": 1.984375, "learning_rate": 2.420433914444948e-05, "loss": 0.1351, "step": 16585 }, { "epoch": 103.01869158878505, "grad_norm": 2.625, "learning_rate": 2.420187663407826e-05, "loss": 0.1282, "step": 16586 }, { "epoch": 103.02492211838006, "grad_norm": 4.625, "learning_rate": 2.4199414131458577e-05, "loss": 0.2225, "step": 16587 }, { "epoch": 103.03115264797508, "grad_norm": 2.0, "learning_rate": 2.4196951636614345e-05, "loss": 0.1203, "step": 16588 }, { "epoch": 103.03738317757009, "grad_norm": 2.625, "learning_rate": 2.4194489149569474e-05, "loss": 0.2511, "step": 16589 }, { "epoch": 103.04361370716511, "grad_norm": 2.328125, "learning_rate": 2.4192026670347894e-05, "loss": 0.2689, "step": 16590 }, { "epoch": 103.04984423676012, "grad_norm": 3.046875, "learning_rate": 2.418956419897351e-05, "loss": 0.2423, "step": 16591 }, { "epoch": 103.05607476635514, "grad_norm": 3.109375, "learning_rate": 2.418710173547024e-05, "loss": 0.2163, "step": 16592 }, { "epoch": 103.06230529595015, "grad_norm": 2.625, "learning_rate": 2.4184639279861995e-05, "loss": 0.1395, "step": 16593 }, { "epoch": 103.06853582554517, "grad_norm": 2.234375, "learning_rate": 2.41821768321727e-05, "loss": 0.1131, "step": 16594 }, { "epoch": 103.07476635514018, "grad_norm": 3.015625, "learning_rate": 2.417971439242626e-05, "loss": 0.1613, "step": 16595 }, { "epoch": 103.0809968847352, "grad_norm": 1.015625, "learning_rate": 2.4177251960646607e-05, "loss": 0.1065, "step": 16596 }, { "epoch": 103.08722741433021, "grad_norm": 2.234375, "learning_rate": 2.4174789536857652e-05, "loss": 0.1732, "step": 16597 }, { "epoch": 103.09345794392523, "grad_norm": 2.5625, "learning_rate": 2.41723271210833e-05, "loss": 0.1853, "step": 16598 }, { "epoch": 103.09968847352025, "grad_norm": 2.84375, "learning_rate": 2.416986471334747e-05, "loss": 0.3064, "step": 16599 }, { "epoch": 103.10591900311526, "grad_norm": 1.6640625, "learning_rate": 2.4167402313674082e-05, "loss": 0.1181, "step": 16600 }, { "epoch": 103.11214953271028, "grad_norm": 1.15625, "learning_rate": 2.416493992208705e-05, "loss": 0.1193, "step": 16601 }, { "epoch": 103.11838006230529, "grad_norm": 2.0625, "learning_rate": 2.4162477538610285e-05, "loss": 0.1671, "step": 16602 }, { "epoch": 103.1246105919003, "grad_norm": 1.34375, "learning_rate": 2.416001516326771e-05, "loss": 0.1268, "step": 16603 }, { "epoch": 103.13084112149532, "grad_norm": 1.46875, "learning_rate": 2.4157552796083234e-05, "loss": 0.1245, "step": 16604 }, { "epoch": 103.13707165109034, "grad_norm": 2.03125, "learning_rate": 2.4155090437080776e-05, "loss": 0.1566, "step": 16605 }, { "epoch": 103.14330218068535, "grad_norm": 1.984375, "learning_rate": 2.4152628086284244e-05, "loss": 0.1545, "step": 16606 }, { "epoch": 103.14953271028037, "grad_norm": 2.046875, "learning_rate": 2.4150165743717556e-05, "loss": 0.1572, "step": 16607 }, { "epoch": 103.15576323987538, "grad_norm": 2.046875, "learning_rate": 2.4147703409404628e-05, "loss": 0.1653, "step": 16608 }, { "epoch": 103.1619937694704, "grad_norm": 2.5, "learning_rate": 2.414524108336939e-05, "loss": 0.2215, "step": 16609 }, { "epoch": 103.16822429906541, "grad_norm": 1.265625, "learning_rate": 2.4142778765635726e-05, "loss": 0.107, "step": 16610 }, { "epoch": 103.17445482866043, "grad_norm": 2.0625, "learning_rate": 2.4140316456227572e-05, "loss": 0.2078, "step": 16611 }, { "epoch": 103.18068535825545, "grad_norm": 2.71875, "learning_rate": 2.4137854155168825e-05, "loss": 0.3104, "step": 16612 }, { "epoch": 103.18691588785046, "grad_norm": 1.8359375, "learning_rate": 2.4135391862483423e-05, "loss": 0.1542, "step": 16613 }, { "epoch": 103.19314641744548, "grad_norm": 0.98046875, "learning_rate": 2.4132929578195264e-05, "loss": 0.092, "step": 16614 }, { "epoch": 103.19937694704049, "grad_norm": 1.2890625, "learning_rate": 2.4130467302328267e-05, "loss": 0.0963, "step": 16615 }, { "epoch": 103.20560747663552, "grad_norm": 2.234375, "learning_rate": 2.412800503490635e-05, "loss": 0.1955, "step": 16616 }, { "epoch": 103.21183800623054, "grad_norm": 1.84375, "learning_rate": 2.4125542775953418e-05, "loss": 0.1029, "step": 16617 }, { "epoch": 103.21806853582555, "grad_norm": 2.59375, "learning_rate": 2.412308052549339e-05, "loss": 0.26, "step": 16618 }, { "epoch": 103.22429906542057, "grad_norm": 4.1875, "learning_rate": 2.4120618283550174e-05, "loss": 0.3272, "step": 16619 }, { "epoch": 103.23052959501558, "grad_norm": 3.140625, "learning_rate": 2.4118156050147693e-05, "loss": 0.2628, "step": 16620 }, { "epoch": 103.2367601246106, "grad_norm": 1.796875, "learning_rate": 2.411569382530986e-05, "loss": 0.1264, "step": 16621 }, { "epoch": 103.24299065420561, "grad_norm": 2.015625, "learning_rate": 2.4113231609060592e-05, "loss": 0.1732, "step": 16622 }, { "epoch": 103.24922118380063, "grad_norm": 2.515625, "learning_rate": 2.411076940142378e-05, "loss": 0.1261, "step": 16623 }, { "epoch": 103.25545171339564, "grad_norm": 1.640625, "learning_rate": 2.410830720242336e-05, "loss": 0.1251, "step": 16624 }, { "epoch": 103.26168224299066, "grad_norm": 1.375, "learning_rate": 2.410584501208324e-05, "loss": 0.0845, "step": 16625 }, { "epoch": 103.26791277258567, "grad_norm": 1.6875, "learning_rate": 2.4103382830427325e-05, "loss": 0.1255, "step": 16626 }, { "epoch": 103.27414330218069, "grad_norm": 2.296875, "learning_rate": 2.4100920657479543e-05, "loss": 0.1305, "step": 16627 }, { "epoch": 103.2803738317757, "grad_norm": 1.359375, "learning_rate": 2.4098458493263794e-05, "loss": 0.119, "step": 16628 }, { "epoch": 103.28660436137072, "grad_norm": 2.53125, "learning_rate": 2.4095996337804007e-05, "loss": 0.2344, "step": 16629 }, { "epoch": 103.29283489096574, "grad_norm": 2.046875, "learning_rate": 2.4093534191124072e-05, "loss": 0.1174, "step": 16630 }, { "epoch": 103.29906542056075, "grad_norm": 2.828125, "learning_rate": 2.409107205324792e-05, "loss": 0.184, "step": 16631 }, { "epoch": 103.30529595015577, "grad_norm": 2.34375, "learning_rate": 2.4088609924199454e-05, "loss": 0.1721, "step": 16632 }, { "epoch": 103.31152647975078, "grad_norm": 2.390625, "learning_rate": 2.4086147804002594e-05, "loss": 0.118, "step": 16633 }, { "epoch": 103.3177570093458, "grad_norm": 1.828125, "learning_rate": 2.4083685692681247e-05, "loss": 0.1219, "step": 16634 }, { "epoch": 103.32398753894081, "grad_norm": 1.6953125, "learning_rate": 2.408122359025934e-05, "loss": 0.1792, "step": 16635 }, { "epoch": 103.33021806853583, "grad_norm": 2.484375, "learning_rate": 2.4078761496760755e-05, "loss": 0.2243, "step": 16636 }, { "epoch": 103.33644859813084, "grad_norm": 2.15625, "learning_rate": 2.407629941220943e-05, "loss": 0.152, "step": 16637 }, { "epoch": 103.34267912772586, "grad_norm": 2.03125, "learning_rate": 2.407383733662927e-05, "loss": 0.1973, "step": 16638 }, { "epoch": 103.34890965732087, "grad_norm": 1.46875, "learning_rate": 2.4071375270044183e-05, "loss": 0.1032, "step": 16639 }, { "epoch": 103.35514018691589, "grad_norm": 1.9375, "learning_rate": 2.4068913212478088e-05, "loss": 0.1279, "step": 16640 }, { "epoch": 103.3613707165109, "grad_norm": 2.0625, "learning_rate": 2.4066451163954902e-05, "loss": 0.1948, "step": 16641 }, { "epoch": 103.36760124610592, "grad_norm": 2.546875, "learning_rate": 2.4063989124498524e-05, "loss": 0.2561, "step": 16642 }, { "epoch": 103.37383177570094, "grad_norm": 1.890625, "learning_rate": 2.4061527094132862e-05, "loss": 0.1775, "step": 16643 }, { "epoch": 103.38006230529595, "grad_norm": 1.3046875, "learning_rate": 2.4059065072881844e-05, "loss": 0.1145, "step": 16644 }, { "epoch": 103.38629283489097, "grad_norm": 1.3359375, "learning_rate": 2.4056603060769367e-05, "loss": 0.1003, "step": 16645 }, { "epoch": 103.39252336448598, "grad_norm": 1.34375, "learning_rate": 2.4054141057819356e-05, "loss": 0.1155, "step": 16646 }, { "epoch": 103.398753894081, "grad_norm": 1.5703125, "learning_rate": 2.405167906405572e-05, "loss": 0.1494, "step": 16647 }, { "epoch": 103.40498442367601, "grad_norm": 1.5234375, "learning_rate": 2.4049217079502367e-05, "loss": 0.1318, "step": 16648 }, { "epoch": 103.41121495327103, "grad_norm": 2.0, "learning_rate": 2.4046755104183198e-05, "loss": 0.1398, "step": 16649 }, { "epoch": 103.41744548286604, "grad_norm": 1.2109375, "learning_rate": 2.404429313812214e-05, "loss": 0.1227, "step": 16650 }, { "epoch": 103.42367601246106, "grad_norm": 2.234375, "learning_rate": 2.4041831181343095e-05, "loss": 0.0945, "step": 16651 }, { "epoch": 103.42990654205607, "grad_norm": 2.59375, "learning_rate": 2.4039369233869975e-05, "loss": 0.1822, "step": 16652 }, { "epoch": 103.43613707165109, "grad_norm": 2.515625, "learning_rate": 2.40369072957267e-05, "loss": 0.1161, "step": 16653 }, { "epoch": 103.4423676012461, "grad_norm": 1.625, "learning_rate": 2.4034445366937174e-05, "loss": 0.1521, "step": 16654 }, { "epoch": 103.44859813084112, "grad_norm": 1.84375, "learning_rate": 2.4031983447525307e-05, "loss": 0.175, "step": 16655 }, { "epoch": 103.45482866043614, "grad_norm": 2.65625, "learning_rate": 2.4029521537515004e-05, "loss": 0.2225, "step": 16656 }, { "epoch": 103.46105919003115, "grad_norm": 2.703125, "learning_rate": 2.4027059636930186e-05, "loss": 0.223, "step": 16657 }, { "epoch": 103.46728971962617, "grad_norm": 1.4921875, "learning_rate": 2.402459774579475e-05, "loss": 0.1222, "step": 16658 }, { "epoch": 103.47352024922118, "grad_norm": 2.5, "learning_rate": 2.402213586413263e-05, "loss": 0.2426, "step": 16659 }, { "epoch": 103.4797507788162, "grad_norm": 2.40625, "learning_rate": 2.4019673991967724e-05, "loss": 0.1062, "step": 16660 }, { "epoch": 103.48598130841121, "grad_norm": 1.7421875, "learning_rate": 2.4017212129323934e-05, "loss": 0.1107, "step": 16661 }, { "epoch": 103.49221183800623, "grad_norm": 2.265625, "learning_rate": 2.4014750276225168e-05, "loss": 0.1706, "step": 16662 }, { "epoch": 103.49844236760124, "grad_norm": 1.9609375, "learning_rate": 2.401228843269535e-05, "loss": 0.1742, "step": 16663 }, { "epoch": 103.50467289719626, "grad_norm": 2.3125, "learning_rate": 2.4009826598758388e-05, "loss": 0.2951, "step": 16664 }, { "epoch": 103.51090342679127, "grad_norm": 2.265625, "learning_rate": 2.400736477443818e-05, "loss": 0.1545, "step": 16665 }, { "epoch": 103.51713395638629, "grad_norm": 2.03125, "learning_rate": 2.4004902959758653e-05, "loss": 0.1765, "step": 16666 }, { "epoch": 103.5233644859813, "grad_norm": 2.59375, "learning_rate": 2.400244115474371e-05, "loss": 0.1219, "step": 16667 }, { "epoch": 103.52959501557632, "grad_norm": 1.4765625, "learning_rate": 2.399997935941725e-05, "loss": 0.0946, "step": 16668 }, { "epoch": 103.53582554517133, "grad_norm": 2.453125, "learning_rate": 2.3997517573803187e-05, "loss": 0.1507, "step": 16669 }, { "epoch": 103.54205607476635, "grad_norm": 3.546875, "learning_rate": 2.399505579792544e-05, "loss": 0.2228, "step": 16670 }, { "epoch": 103.54828660436137, "grad_norm": 2.796875, "learning_rate": 2.399259403180791e-05, "loss": 0.212, "step": 16671 }, { "epoch": 103.55451713395638, "grad_norm": 3.6875, "learning_rate": 2.39901322754745e-05, "loss": 0.1487, "step": 16672 }, { "epoch": 103.5607476635514, "grad_norm": 1.9375, "learning_rate": 2.3987670528949147e-05, "loss": 0.1307, "step": 16673 }, { "epoch": 103.56697819314641, "grad_norm": 1.15625, "learning_rate": 2.3985208792255727e-05, "loss": 0.1083, "step": 16674 }, { "epoch": 103.57320872274143, "grad_norm": 4.21875, "learning_rate": 2.3982747065418166e-05, "loss": 0.3166, "step": 16675 }, { "epoch": 103.57943925233644, "grad_norm": 2.453125, "learning_rate": 2.3980285348460363e-05, "loss": 0.1842, "step": 16676 }, { "epoch": 103.58566978193146, "grad_norm": 2.65625, "learning_rate": 2.3977823641406235e-05, "loss": 0.2308, "step": 16677 }, { "epoch": 103.59190031152647, "grad_norm": 2.921875, "learning_rate": 2.3975361944279686e-05, "loss": 0.1612, "step": 16678 }, { "epoch": 103.59813084112149, "grad_norm": 1.5, "learning_rate": 2.3972900257104642e-05, "loss": 0.1114, "step": 16679 }, { "epoch": 103.60436137071652, "grad_norm": 2.96875, "learning_rate": 2.397043857990498e-05, "loss": 0.2403, "step": 16680 }, { "epoch": 103.61059190031153, "grad_norm": 2.4375, "learning_rate": 2.3967976912704627e-05, "loss": 0.1194, "step": 16681 }, { "epoch": 103.61682242990655, "grad_norm": 2.8125, "learning_rate": 2.396551525552749e-05, "loss": 0.1564, "step": 16682 }, { "epoch": 103.62305295950156, "grad_norm": 3.34375, "learning_rate": 2.3963053608397473e-05, "loss": 0.2249, "step": 16683 }, { "epoch": 103.62928348909658, "grad_norm": 2.71875, "learning_rate": 2.3960591971338493e-05, "loss": 0.1939, "step": 16684 }, { "epoch": 103.6355140186916, "grad_norm": 2.359375, "learning_rate": 2.395813034437444e-05, "loss": 0.1273, "step": 16685 }, { "epoch": 103.64174454828661, "grad_norm": 2.328125, "learning_rate": 2.3955668727529247e-05, "loss": 0.1024, "step": 16686 }, { "epoch": 103.64797507788163, "grad_norm": 1.5234375, "learning_rate": 2.3953207120826802e-05, "loss": 0.1312, "step": 16687 }, { "epoch": 103.65420560747664, "grad_norm": 3.9375, "learning_rate": 2.395074552429102e-05, "loss": 0.3032, "step": 16688 }, { "epoch": 103.66043613707166, "grad_norm": 2.53125, "learning_rate": 2.3948283937945802e-05, "loss": 0.1208, "step": 16689 }, { "epoch": 103.66666666666667, "grad_norm": 2.578125, "learning_rate": 2.3945822361815067e-05, "loss": 0.1983, "step": 16690 }, { "epoch": 103.67289719626169, "grad_norm": 2.734375, "learning_rate": 2.394336079592271e-05, "loss": 0.1729, "step": 16691 }, { "epoch": 103.6791277258567, "grad_norm": 1.734375, "learning_rate": 2.394089924029266e-05, "loss": 0.1324, "step": 16692 }, { "epoch": 103.68535825545172, "grad_norm": 2.0625, "learning_rate": 2.393843769494879e-05, "loss": 0.1136, "step": 16693 }, { "epoch": 103.69158878504673, "grad_norm": 2.265625, "learning_rate": 2.3935976159915033e-05, "loss": 0.1508, "step": 16694 }, { "epoch": 103.69781931464175, "grad_norm": 3.140625, "learning_rate": 2.393351463521528e-05, "loss": 0.3107, "step": 16695 }, { "epoch": 103.70404984423676, "grad_norm": 1.328125, "learning_rate": 2.3931053120873455e-05, "loss": 0.1186, "step": 16696 }, { "epoch": 103.71028037383178, "grad_norm": 1.46875, "learning_rate": 2.3928591616913454e-05, "loss": 0.1474, "step": 16697 }, { "epoch": 103.7165109034268, "grad_norm": 2.59375, "learning_rate": 2.392613012335918e-05, "loss": 0.1695, "step": 16698 }, { "epoch": 103.72274143302181, "grad_norm": 0.828125, "learning_rate": 2.3923668640234558e-05, "loss": 0.0916, "step": 16699 }, { "epoch": 103.72897196261682, "grad_norm": 2.296875, "learning_rate": 2.3921207167563473e-05, "loss": 0.2145, "step": 16700 }, { "epoch": 103.73520249221184, "grad_norm": 2.546875, "learning_rate": 2.391874570536984e-05, "loss": 0.1717, "step": 16701 }, { "epoch": 103.74143302180686, "grad_norm": 1.765625, "learning_rate": 2.3916284253677558e-05, "loss": 0.1509, "step": 16702 }, { "epoch": 103.74766355140187, "grad_norm": 2.84375, "learning_rate": 2.391382281251055e-05, "loss": 0.3064, "step": 16703 }, { "epoch": 103.75389408099689, "grad_norm": 3.265625, "learning_rate": 2.3911361381892705e-05, "loss": 0.2216, "step": 16704 }, { "epoch": 103.7601246105919, "grad_norm": 2.53125, "learning_rate": 2.390889996184795e-05, "loss": 0.2589, "step": 16705 }, { "epoch": 103.76635514018692, "grad_norm": 1.625, "learning_rate": 2.3906438552400163e-05, "loss": 0.1471, "step": 16706 }, { "epoch": 103.77258566978193, "grad_norm": 2.34375, "learning_rate": 2.3903977153573264e-05, "loss": 0.1402, "step": 16707 }, { "epoch": 103.77881619937695, "grad_norm": 1.3359375, "learning_rate": 2.3901515765391157e-05, "loss": 0.0989, "step": 16708 }, { "epoch": 103.78504672897196, "grad_norm": 2.09375, "learning_rate": 2.3899054387877753e-05, "loss": 0.1797, "step": 16709 }, { "epoch": 103.79127725856698, "grad_norm": 3.5, "learning_rate": 2.3896593021056953e-05, "loss": 0.2566, "step": 16710 }, { "epoch": 103.797507788162, "grad_norm": 1.984375, "learning_rate": 2.3894131664952668e-05, "loss": 0.1531, "step": 16711 }, { "epoch": 103.80373831775701, "grad_norm": 1.5703125, "learning_rate": 2.3891670319588785e-05, "loss": 0.085, "step": 16712 }, { "epoch": 103.80996884735202, "grad_norm": 2.578125, "learning_rate": 2.3889208984989227e-05, "loss": 0.0981, "step": 16713 }, { "epoch": 103.81619937694704, "grad_norm": 2.671875, "learning_rate": 2.3886747661177894e-05, "loss": 0.1899, "step": 16714 }, { "epoch": 103.82242990654206, "grad_norm": 2.8125, "learning_rate": 2.3884286348178686e-05, "loss": 0.2096, "step": 16715 }, { "epoch": 103.82866043613707, "grad_norm": 2.265625, "learning_rate": 2.3881825046015516e-05, "loss": 0.1834, "step": 16716 }, { "epoch": 103.83489096573209, "grad_norm": 2.28125, "learning_rate": 2.3879363754712284e-05, "loss": 0.2124, "step": 16717 }, { "epoch": 103.8411214953271, "grad_norm": 2.015625, "learning_rate": 2.38769024742929e-05, "loss": 0.1651, "step": 16718 }, { "epoch": 103.84735202492212, "grad_norm": 2.515625, "learning_rate": 2.3874441204781255e-05, "loss": 0.1943, "step": 16719 }, { "epoch": 103.85358255451713, "grad_norm": 1.375, "learning_rate": 2.3871979946201268e-05, "loss": 0.1028, "step": 16720 }, { "epoch": 103.85981308411215, "grad_norm": 2.265625, "learning_rate": 2.3869518698576835e-05, "loss": 0.2764, "step": 16721 }, { "epoch": 103.86604361370716, "grad_norm": 1.4375, "learning_rate": 2.3867057461931857e-05, "loss": 0.1544, "step": 16722 }, { "epoch": 103.87227414330218, "grad_norm": 1.1171875, "learning_rate": 2.386459623629025e-05, "loss": 0.094, "step": 16723 }, { "epoch": 103.8785046728972, "grad_norm": 1.9140625, "learning_rate": 2.3862135021675918e-05, "loss": 0.1397, "step": 16724 }, { "epoch": 103.88473520249221, "grad_norm": 2.625, "learning_rate": 2.385967381811275e-05, "loss": 0.1679, "step": 16725 }, { "epoch": 103.89096573208722, "grad_norm": 2.515625, "learning_rate": 2.3857212625624652e-05, "loss": 0.1751, "step": 16726 }, { "epoch": 103.89719626168224, "grad_norm": 2.296875, "learning_rate": 2.3854751444235543e-05, "loss": 0.2061, "step": 16727 }, { "epoch": 103.90342679127725, "grad_norm": 1.75, "learning_rate": 2.3852290273969308e-05, "loss": 0.1403, "step": 16728 }, { "epoch": 103.90965732087227, "grad_norm": 2.03125, "learning_rate": 2.3849829114849864e-05, "loss": 0.1662, "step": 16729 }, { "epoch": 103.91588785046729, "grad_norm": 2.9375, "learning_rate": 2.3847367966901117e-05, "loss": 0.0878, "step": 16730 }, { "epoch": 103.9221183800623, "grad_norm": 1.8984375, "learning_rate": 2.3844906830146958e-05, "loss": 0.1544, "step": 16731 }, { "epoch": 103.92834890965732, "grad_norm": 2.28125, "learning_rate": 2.384244570461129e-05, "loss": 0.1421, "step": 16732 }, { "epoch": 103.93457943925233, "grad_norm": 3.015625, "learning_rate": 2.3839984590318025e-05, "loss": 0.1755, "step": 16733 }, { "epoch": 103.94080996884735, "grad_norm": 2.6875, "learning_rate": 2.3837523487291058e-05, "loss": 0.3134, "step": 16734 }, { "epoch": 103.94704049844236, "grad_norm": 2.171875, "learning_rate": 2.383506239555429e-05, "loss": 0.208, "step": 16735 }, { "epoch": 103.95327102803738, "grad_norm": 1.7890625, "learning_rate": 2.3832601315131637e-05, "loss": 0.1519, "step": 16736 }, { "epoch": 103.9595015576324, "grad_norm": 3.0625, "learning_rate": 2.3830140246047e-05, "loss": 0.2331, "step": 16737 }, { "epoch": 103.96573208722741, "grad_norm": 2.84375, "learning_rate": 2.382767918832427e-05, "loss": 0.2172, "step": 16738 }, { "epoch": 103.97196261682242, "grad_norm": 2.375, "learning_rate": 2.3825218141987346e-05, "loss": 0.1203, "step": 16739 }, { "epoch": 103.97819314641744, "grad_norm": 2.765625, "learning_rate": 2.3822757107060144e-05, "loss": 0.1877, "step": 16740 }, { "epoch": 103.98442367601245, "grad_norm": 1.6484375, "learning_rate": 2.3820296083566558e-05, "loss": 0.1482, "step": 16741 }, { "epoch": 103.99065420560747, "grad_norm": 2.75, "learning_rate": 2.3817835071530492e-05, "loss": 0.2296, "step": 16742 }, { "epoch": 103.99688473520249, "grad_norm": 3.328125, "learning_rate": 2.3815374070975857e-05, "loss": 0.2553, "step": 16743 }, { "epoch": 104.0, "grad_norm": 0.5546875, "learning_rate": 2.381291308192654e-05, "loss": 0.0405, "step": 16744 }, { "epoch": 104.00623052959502, "grad_norm": 3.171875, "learning_rate": 2.3810452104406444e-05, "loss": 0.2706, "step": 16745 }, { "epoch": 104.01246105919003, "grad_norm": 1.6328125, "learning_rate": 2.3807991138439477e-05, "loss": 0.1212, "step": 16746 }, { "epoch": 104.01869158878505, "grad_norm": 2.09375, "learning_rate": 2.3805530184049543e-05, "loss": 0.1586, "step": 16747 }, { "epoch": 104.02492211838006, "grad_norm": 1.46875, "learning_rate": 2.380306924126053e-05, "loss": 0.1189, "step": 16748 }, { "epoch": 104.03115264797508, "grad_norm": 1.3203125, "learning_rate": 2.3800608310096362e-05, "loss": 0.1185, "step": 16749 }, { "epoch": 104.03738317757009, "grad_norm": 1.546875, "learning_rate": 2.3798147390580916e-05, "loss": 0.1306, "step": 16750 }, { "epoch": 104.04361370716511, "grad_norm": 1.4296875, "learning_rate": 2.3795686482738105e-05, "loss": 0.1158, "step": 16751 }, { "epoch": 104.04984423676012, "grad_norm": 1.3984375, "learning_rate": 2.3793225586591823e-05, "loss": 0.1325, "step": 16752 }, { "epoch": 104.05607476635514, "grad_norm": 0.91015625, "learning_rate": 2.379076470216598e-05, "loss": 0.0909, "step": 16753 }, { "epoch": 104.06230529595015, "grad_norm": 1.84375, "learning_rate": 2.3788303829484466e-05, "loss": 0.1214, "step": 16754 }, { "epoch": 104.06853582554517, "grad_norm": 1.5546875, "learning_rate": 2.378584296857119e-05, "loss": 0.1511, "step": 16755 }, { "epoch": 104.07476635514018, "grad_norm": 3.09375, "learning_rate": 2.378338211945006e-05, "loss": 0.1976, "step": 16756 }, { "epoch": 104.0809968847352, "grad_norm": 3.515625, "learning_rate": 2.378092128214496e-05, "loss": 0.3756, "step": 16757 }, { "epoch": 104.08722741433021, "grad_norm": 1.078125, "learning_rate": 2.377846045667979e-05, "loss": 0.1046, "step": 16758 }, { "epoch": 104.09345794392523, "grad_norm": 2.28125, "learning_rate": 2.377599964307846e-05, "loss": 0.2374, "step": 16759 }, { "epoch": 104.09968847352025, "grad_norm": 2.34375, "learning_rate": 2.3773538841364866e-05, "loss": 0.1568, "step": 16760 }, { "epoch": 104.10591900311526, "grad_norm": 1.3359375, "learning_rate": 2.3771078051562907e-05, "loss": 0.0875, "step": 16761 }, { "epoch": 104.11214953271028, "grad_norm": 1.9921875, "learning_rate": 2.376861727369649e-05, "loss": 0.1411, "step": 16762 }, { "epoch": 104.11838006230529, "grad_norm": 2.484375, "learning_rate": 2.3766156507789505e-05, "loss": 0.1865, "step": 16763 }, { "epoch": 104.1246105919003, "grad_norm": 3.140625, "learning_rate": 2.3763695753865852e-05, "loss": 0.1901, "step": 16764 }, { "epoch": 104.13084112149532, "grad_norm": 2.28125, "learning_rate": 2.376123501194943e-05, "loss": 0.1835, "step": 16765 }, { "epoch": 104.13707165109034, "grad_norm": 2.578125, "learning_rate": 2.3758774282064146e-05, "loss": 0.2163, "step": 16766 }, { "epoch": 104.14330218068535, "grad_norm": 1.046875, "learning_rate": 2.3756313564233895e-05, "loss": 0.1156, "step": 16767 }, { "epoch": 104.14953271028037, "grad_norm": 1.2890625, "learning_rate": 2.375385285848257e-05, "loss": 0.1096, "step": 16768 }, { "epoch": 104.15576323987538, "grad_norm": 2.25, "learning_rate": 2.3751392164834084e-05, "loss": 0.1518, "step": 16769 }, { "epoch": 104.1619937694704, "grad_norm": 3.796875, "learning_rate": 2.374893148331232e-05, "loss": 0.3055, "step": 16770 }, { "epoch": 104.16822429906541, "grad_norm": 2.046875, "learning_rate": 2.3746470813941188e-05, "loss": 0.1347, "step": 16771 }, { "epoch": 104.17445482866043, "grad_norm": 2.3125, "learning_rate": 2.3744010156744574e-05, "loss": 0.1518, "step": 16772 }, { "epoch": 104.18068535825545, "grad_norm": 1.46875, "learning_rate": 2.374154951174639e-05, "loss": 0.0984, "step": 16773 }, { "epoch": 104.18691588785046, "grad_norm": 1.7109375, "learning_rate": 2.3739088878970526e-05, "loss": 0.1336, "step": 16774 }, { "epoch": 104.19314641744548, "grad_norm": 1.90625, "learning_rate": 2.3736628258440895e-05, "loss": 0.1133, "step": 16775 }, { "epoch": 104.19937694704049, "grad_norm": 3.078125, "learning_rate": 2.3734167650181372e-05, "loss": 0.1424, "step": 16776 }, { "epoch": 104.20560747663552, "grad_norm": 2.46875, "learning_rate": 2.3731707054215866e-05, "loss": 0.1742, "step": 16777 }, { "epoch": 104.21183800623054, "grad_norm": 2.671875, "learning_rate": 2.372924647056827e-05, "loss": 0.1501, "step": 16778 }, { "epoch": 104.21806853582555, "grad_norm": 2.5, "learning_rate": 2.3726785899262493e-05, "loss": 0.1221, "step": 16779 }, { "epoch": 104.22429906542057, "grad_norm": 3.0, "learning_rate": 2.3724325340322427e-05, "loss": 0.166, "step": 16780 }, { "epoch": 104.23052959501558, "grad_norm": 2.609375, "learning_rate": 2.3721864793771972e-05, "loss": 0.2145, "step": 16781 }, { "epoch": 104.2367601246106, "grad_norm": 1.3515625, "learning_rate": 2.3719404259635012e-05, "loss": 0.1223, "step": 16782 }, { "epoch": 104.24299065420561, "grad_norm": 2.71875, "learning_rate": 2.3716943737935457e-05, "loss": 0.1265, "step": 16783 }, { "epoch": 104.24922118380063, "grad_norm": 1.390625, "learning_rate": 2.37144832286972e-05, "loss": 0.1338, "step": 16784 }, { "epoch": 104.25545171339564, "grad_norm": 2.203125, "learning_rate": 2.3712022731944137e-05, "loss": 0.1787, "step": 16785 }, { "epoch": 104.26168224299066, "grad_norm": 1.21875, "learning_rate": 2.370956224770017e-05, "loss": 0.1077, "step": 16786 }, { "epoch": 104.26791277258567, "grad_norm": 1.703125, "learning_rate": 2.370710177598919e-05, "loss": 0.1587, "step": 16787 }, { "epoch": 104.27414330218069, "grad_norm": 2.5, "learning_rate": 2.3704641316835112e-05, "loss": 0.1829, "step": 16788 }, { "epoch": 104.2803738317757, "grad_norm": 1.859375, "learning_rate": 2.3702180870261796e-05, "loss": 0.1362, "step": 16789 }, { "epoch": 104.28660436137072, "grad_norm": 1.625, "learning_rate": 2.3699720436293167e-05, "loss": 0.0889, "step": 16790 }, { "epoch": 104.29283489096574, "grad_norm": 2.03125, "learning_rate": 2.3697260014953108e-05, "loss": 0.1979, "step": 16791 }, { "epoch": 104.29906542056075, "grad_norm": 1.71875, "learning_rate": 2.3694799606265522e-05, "loss": 0.1614, "step": 16792 }, { "epoch": 104.30529595015577, "grad_norm": 1.3203125, "learning_rate": 2.3692339210254306e-05, "loss": 0.0915, "step": 16793 }, { "epoch": 104.31152647975078, "grad_norm": 1.6875, "learning_rate": 2.368987882694336e-05, "loss": 0.1322, "step": 16794 }, { "epoch": 104.3177570093458, "grad_norm": 2.359375, "learning_rate": 2.368741845635656e-05, "loss": 0.2003, "step": 16795 }, { "epoch": 104.32398753894081, "grad_norm": 1.0546875, "learning_rate": 2.368495809851782e-05, "loss": 0.1155, "step": 16796 }, { "epoch": 104.33021806853583, "grad_norm": 3.203125, "learning_rate": 2.3682497753451027e-05, "loss": 0.176, "step": 16797 }, { "epoch": 104.33644859813084, "grad_norm": 1.5546875, "learning_rate": 2.3680037421180077e-05, "loss": 0.1241, "step": 16798 }, { "epoch": 104.34267912772586, "grad_norm": 2.734375, "learning_rate": 2.3677577101728873e-05, "loss": 0.2559, "step": 16799 }, { "epoch": 104.34890965732087, "grad_norm": 1.9140625, "learning_rate": 2.367511679512131e-05, "loss": 0.1483, "step": 16800 }, { "epoch": 104.35514018691589, "grad_norm": 2.359375, "learning_rate": 2.3672656501381272e-05, "loss": 0.2335, "step": 16801 }, { "epoch": 104.3613707165109, "grad_norm": 1.7734375, "learning_rate": 2.367019622053265e-05, "loss": 0.1612, "step": 16802 }, { "epoch": 104.36760124610592, "grad_norm": 1.84375, "learning_rate": 2.3667735952599358e-05, "loss": 0.1598, "step": 16803 }, { "epoch": 104.37383177570094, "grad_norm": 2.015625, "learning_rate": 2.3665275697605273e-05, "loss": 0.1566, "step": 16804 }, { "epoch": 104.38006230529595, "grad_norm": 3.015625, "learning_rate": 2.3662815455574307e-05, "loss": 0.3504, "step": 16805 }, { "epoch": 104.38629283489097, "grad_norm": 2.828125, "learning_rate": 2.366035522653034e-05, "loss": 0.2935, "step": 16806 }, { "epoch": 104.39252336448598, "grad_norm": 1.3203125, "learning_rate": 2.3657895010497278e-05, "loss": 0.1173, "step": 16807 }, { "epoch": 104.398753894081, "grad_norm": 2.265625, "learning_rate": 2.3655434807498996e-05, "loss": 0.1451, "step": 16808 }, { "epoch": 104.40498442367601, "grad_norm": 2.046875, "learning_rate": 2.3652974617559407e-05, "loss": 0.167, "step": 16809 }, { "epoch": 104.41121495327103, "grad_norm": 1.890625, "learning_rate": 2.3650514440702394e-05, "loss": 0.1215, "step": 16810 }, { "epoch": 104.41744548286604, "grad_norm": 1.703125, "learning_rate": 2.3648054276951854e-05, "loss": 0.1608, "step": 16811 }, { "epoch": 104.42367601246106, "grad_norm": 1.6953125, "learning_rate": 2.3645594126331684e-05, "loss": 0.1212, "step": 16812 }, { "epoch": 104.42990654205607, "grad_norm": 1.7734375, "learning_rate": 2.3643133988865783e-05, "loss": 0.1329, "step": 16813 }, { "epoch": 104.43613707165109, "grad_norm": 1.390625, "learning_rate": 2.3640673864578023e-05, "loss": 0.0887, "step": 16814 }, { "epoch": 104.4423676012461, "grad_norm": 2.296875, "learning_rate": 2.363821375349231e-05, "loss": 0.2619, "step": 16815 }, { "epoch": 104.44859813084112, "grad_norm": 2.0625, "learning_rate": 2.3635753655632544e-05, "loss": 0.2136, "step": 16816 }, { "epoch": 104.45482866043614, "grad_norm": 1.953125, "learning_rate": 2.363329357102261e-05, "loss": 0.1513, "step": 16817 }, { "epoch": 104.46105919003115, "grad_norm": 1.59375, "learning_rate": 2.3630833499686397e-05, "loss": 0.1359, "step": 16818 }, { "epoch": 104.46728971962617, "grad_norm": 3.34375, "learning_rate": 2.3628373441647812e-05, "loss": 0.2499, "step": 16819 }, { "epoch": 104.47352024922118, "grad_norm": 3.765625, "learning_rate": 2.3625913396930734e-05, "loss": 0.1254, "step": 16820 }, { "epoch": 104.4797507788162, "grad_norm": 3.25, "learning_rate": 2.362345336555906e-05, "loss": 0.1542, "step": 16821 }, { "epoch": 104.48598130841121, "grad_norm": 2.875, "learning_rate": 2.3620993347556674e-05, "loss": 0.1255, "step": 16822 }, { "epoch": 104.49221183800623, "grad_norm": 1.2421875, "learning_rate": 2.3618533342947487e-05, "loss": 0.1145, "step": 16823 }, { "epoch": 104.49844236760124, "grad_norm": 1.1953125, "learning_rate": 2.361607335175537e-05, "loss": 0.1197, "step": 16824 }, { "epoch": 104.50467289719626, "grad_norm": 3.03125, "learning_rate": 2.3613613374004233e-05, "loss": 0.2738, "step": 16825 }, { "epoch": 104.51090342679127, "grad_norm": 3.078125, "learning_rate": 2.3611153409717965e-05, "loss": 0.2011, "step": 16826 }, { "epoch": 104.51713395638629, "grad_norm": 2.9375, "learning_rate": 2.3608693458920446e-05, "loss": 0.1864, "step": 16827 }, { "epoch": 104.5233644859813, "grad_norm": 3.296875, "learning_rate": 2.3606233521635572e-05, "loss": 0.173, "step": 16828 }, { "epoch": 104.52959501557632, "grad_norm": 2.0625, "learning_rate": 2.3603773597887237e-05, "loss": 0.1991, "step": 16829 }, { "epoch": 104.53582554517133, "grad_norm": 1.703125, "learning_rate": 2.3601313687699337e-05, "loss": 0.1519, "step": 16830 }, { "epoch": 104.54205607476635, "grad_norm": 4.09375, "learning_rate": 2.359885379109575e-05, "loss": 0.2353, "step": 16831 }, { "epoch": 104.54828660436137, "grad_norm": 1.296875, "learning_rate": 2.3596393908100388e-05, "loss": 0.1088, "step": 16832 }, { "epoch": 104.55451713395638, "grad_norm": 1.1796875, "learning_rate": 2.359393403873712e-05, "loss": 0.0939, "step": 16833 }, { "epoch": 104.5607476635514, "grad_norm": 4.03125, "learning_rate": 2.3591474183029845e-05, "loss": 0.3786, "step": 16834 }, { "epoch": 104.56697819314641, "grad_norm": 3.671875, "learning_rate": 2.3589014341002456e-05, "loss": 0.3424, "step": 16835 }, { "epoch": 104.57320872274143, "grad_norm": 1.71875, "learning_rate": 2.358655451267884e-05, "loss": 0.1186, "step": 16836 }, { "epoch": 104.57943925233644, "grad_norm": 2.296875, "learning_rate": 2.3584094698082888e-05, "loss": 0.1427, "step": 16837 }, { "epoch": 104.58566978193146, "grad_norm": 1.9140625, "learning_rate": 2.3581634897238495e-05, "loss": 0.1736, "step": 16838 }, { "epoch": 104.59190031152647, "grad_norm": 1.1953125, "learning_rate": 2.357917511016955e-05, "loss": 0.0913, "step": 16839 }, { "epoch": 104.59813084112149, "grad_norm": 1.953125, "learning_rate": 2.357671533689994e-05, "loss": 0.1193, "step": 16840 }, { "epoch": 104.60436137071652, "grad_norm": 1.1171875, "learning_rate": 2.3574255577453547e-05, "loss": 0.0892, "step": 16841 }, { "epoch": 104.61059190031153, "grad_norm": 1.7578125, "learning_rate": 2.3571795831854276e-05, "loss": 0.1133, "step": 16842 }, { "epoch": 104.61682242990655, "grad_norm": 1.1171875, "learning_rate": 2.3569336100126006e-05, "loss": 0.1005, "step": 16843 }, { "epoch": 104.62305295950156, "grad_norm": 1.9140625, "learning_rate": 2.356687638229263e-05, "loss": 0.1184, "step": 16844 }, { "epoch": 104.62928348909658, "grad_norm": 1.375, "learning_rate": 2.3564416678378043e-05, "loss": 0.1231, "step": 16845 }, { "epoch": 104.6355140186916, "grad_norm": 1.6328125, "learning_rate": 2.3561956988406126e-05, "loss": 0.1151, "step": 16846 }, { "epoch": 104.64174454828661, "grad_norm": 1.8828125, "learning_rate": 2.3559497312400766e-05, "loss": 0.0881, "step": 16847 }, { "epoch": 104.64797507788163, "grad_norm": 1.4375, "learning_rate": 2.3557037650385856e-05, "loss": 0.1207, "step": 16848 }, { "epoch": 104.65420560747664, "grad_norm": 1.0546875, "learning_rate": 2.3554578002385285e-05, "loss": 0.1016, "step": 16849 }, { "epoch": 104.66043613707166, "grad_norm": 2.046875, "learning_rate": 2.3552118368422937e-05, "loss": 0.2376, "step": 16850 }, { "epoch": 104.66666666666667, "grad_norm": 2.1875, "learning_rate": 2.3549658748522725e-05, "loss": 0.136, "step": 16851 }, { "epoch": 104.67289719626169, "grad_norm": 1.265625, "learning_rate": 2.3547199142708498e-05, "loss": 0.096, "step": 16852 }, { "epoch": 104.6791277258567, "grad_norm": 1.3828125, "learning_rate": 2.354473955100417e-05, "loss": 0.1175, "step": 16853 }, { "epoch": 104.68535825545172, "grad_norm": 1.6328125, "learning_rate": 2.3542279973433613e-05, "loss": 0.1109, "step": 16854 }, { "epoch": 104.69158878504673, "grad_norm": 2.328125, "learning_rate": 2.3539820410020734e-05, "loss": 0.2082, "step": 16855 }, { "epoch": 104.69781931464175, "grad_norm": 2.40625, "learning_rate": 2.353736086078941e-05, "loss": 0.1392, "step": 16856 }, { "epoch": 104.70404984423676, "grad_norm": 2.328125, "learning_rate": 2.353490132576352e-05, "loss": 0.1547, "step": 16857 }, { "epoch": 104.71028037383178, "grad_norm": 2.265625, "learning_rate": 2.3532441804966978e-05, "loss": 0.2131, "step": 16858 }, { "epoch": 104.7165109034268, "grad_norm": 1.4296875, "learning_rate": 2.3529982298423647e-05, "loss": 0.1298, "step": 16859 }, { "epoch": 104.72274143302181, "grad_norm": 2.78125, "learning_rate": 2.3527522806157422e-05, "loss": 0.2496, "step": 16860 }, { "epoch": 104.72897196261682, "grad_norm": 2.59375, "learning_rate": 2.3525063328192187e-05, "loss": 0.0959, "step": 16861 }, { "epoch": 104.73520249221184, "grad_norm": 1.8984375, "learning_rate": 2.3522603864551834e-05, "loss": 0.1681, "step": 16862 }, { "epoch": 104.74143302180686, "grad_norm": 2.40625, "learning_rate": 2.352014441526025e-05, "loss": 0.1742, "step": 16863 }, { "epoch": 104.74766355140187, "grad_norm": 3.46875, "learning_rate": 2.351768498034132e-05, "loss": 0.2217, "step": 16864 }, { "epoch": 104.75389408099689, "grad_norm": 1.671875, "learning_rate": 2.3515225559818925e-05, "loss": 0.1104, "step": 16865 }, { "epoch": 104.7601246105919, "grad_norm": 2.546875, "learning_rate": 2.3512766153716957e-05, "loss": 0.1585, "step": 16866 }, { "epoch": 104.76635514018692, "grad_norm": 2.25, "learning_rate": 2.35103067620593e-05, "loss": 0.1538, "step": 16867 }, { "epoch": 104.77258566978193, "grad_norm": 2.921875, "learning_rate": 2.350784738486984e-05, "loss": 0.1454, "step": 16868 }, { "epoch": 104.77881619937695, "grad_norm": 1.53125, "learning_rate": 2.350538802217247e-05, "loss": 0.1383, "step": 16869 }, { "epoch": 104.78504672897196, "grad_norm": 4.15625, "learning_rate": 2.3502928673991073e-05, "loss": 0.2741, "step": 16870 }, { "epoch": 104.79127725856698, "grad_norm": 1.84375, "learning_rate": 2.350046934034953e-05, "loss": 0.1599, "step": 16871 }, { "epoch": 104.797507788162, "grad_norm": 2.34375, "learning_rate": 2.349801002127172e-05, "loss": 0.1789, "step": 16872 }, { "epoch": 104.80373831775701, "grad_norm": 2.296875, "learning_rate": 2.3495550716781546e-05, "loss": 0.1494, "step": 16873 }, { "epoch": 104.80996884735202, "grad_norm": 2.421875, "learning_rate": 2.3493091426902873e-05, "loss": 0.1533, "step": 16874 }, { "epoch": 104.81619937694704, "grad_norm": 2.328125, "learning_rate": 2.3490632151659605e-05, "loss": 0.2276, "step": 16875 }, { "epoch": 104.82242990654206, "grad_norm": 1.8203125, "learning_rate": 2.348817289107562e-05, "loss": 0.1441, "step": 16876 }, { "epoch": 104.82866043613707, "grad_norm": 0.9453125, "learning_rate": 2.3485713645174808e-05, "loss": 0.1075, "step": 16877 }, { "epoch": 104.83489096573209, "grad_norm": 2.0625, "learning_rate": 2.3483254413981038e-05, "loss": 0.1988, "step": 16878 }, { "epoch": 104.8411214953271, "grad_norm": 2.96875, "learning_rate": 2.3480795197518207e-05, "loss": 0.2439, "step": 16879 }, { "epoch": 104.84735202492212, "grad_norm": 1.3984375, "learning_rate": 2.34783359958102e-05, "loss": 0.0967, "step": 16880 }, { "epoch": 104.85358255451713, "grad_norm": 1.703125, "learning_rate": 2.347587680888089e-05, "loss": 0.1178, "step": 16881 }, { "epoch": 104.85981308411215, "grad_norm": 1.6015625, "learning_rate": 2.3473417636754173e-05, "loss": 0.1088, "step": 16882 }, { "epoch": 104.86604361370716, "grad_norm": 2.21875, "learning_rate": 2.3470958479453938e-05, "loss": 0.1822, "step": 16883 }, { "epoch": 104.87227414330218, "grad_norm": 2.0, "learning_rate": 2.3468499337004053e-05, "loss": 0.2421, "step": 16884 }, { "epoch": 104.8785046728972, "grad_norm": 1.46875, "learning_rate": 2.3466040209428406e-05, "loss": 0.1216, "step": 16885 }, { "epoch": 104.88473520249221, "grad_norm": 3.25, "learning_rate": 2.3463581096750883e-05, "loss": 0.2402, "step": 16886 }, { "epoch": 104.89096573208722, "grad_norm": 2.359375, "learning_rate": 2.3461121998995366e-05, "loss": 0.1603, "step": 16887 }, { "epoch": 104.89719626168224, "grad_norm": 2.140625, "learning_rate": 2.3458662916185746e-05, "loss": 0.1684, "step": 16888 }, { "epoch": 104.90342679127725, "grad_norm": 1.28125, "learning_rate": 2.34562038483459e-05, "loss": 0.0919, "step": 16889 }, { "epoch": 104.90965732087227, "grad_norm": 2.703125, "learning_rate": 2.345374479549971e-05, "loss": 0.1766, "step": 16890 }, { "epoch": 104.91588785046729, "grad_norm": 1.8828125, "learning_rate": 2.3451285757671055e-05, "loss": 0.098, "step": 16891 }, { "epoch": 104.9221183800623, "grad_norm": 2.40625, "learning_rate": 2.3448826734883826e-05, "loss": 0.0981, "step": 16892 }, { "epoch": 104.92834890965732, "grad_norm": 2.359375, "learning_rate": 2.34463677271619e-05, "loss": 0.1773, "step": 16893 }, { "epoch": 104.93457943925233, "grad_norm": 1.9765625, "learning_rate": 2.3443908734529157e-05, "loss": 0.1426, "step": 16894 }, { "epoch": 104.94080996884735, "grad_norm": 1.28125, "learning_rate": 2.344144975700949e-05, "loss": 0.1238, "step": 16895 }, { "epoch": 104.94704049844236, "grad_norm": 2.65625, "learning_rate": 2.3438990794626778e-05, "loss": 0.1969, "step": 16896 }, { "epoch": 104.95327102803738, "grad_norm": 2.765625, "learning_rate": 2.3436531847404896e-05, "loss": 0.1989, "step": 16897 }, { "epoch": 104.9595015576324, "grad_norm": 2.0625, "learning_rate": 2.3434072915367723e-05, "loss": 0.106, "step": 16898 }, { "epoch": 104.96573208722741, "grad_norm": 2.515625, "learning_rate": 2.3431613998539152e-05, "loss": 0.0963, "step": 16899 }, { "epoch": 104.97196261682242, "grad_norm": 2.328125, "learning_rate": 2.3429155096943052e-05, "loss": 0.1526, "step": 16900 }, { "epoch": 104.97819314641744, "grad_norm": 2.1875, "learning_rate": 2.342669621060332e-05, "loss": 0.119, "step": 16901 }, { "epoch": 104.98442367601245, "grad_norm": 2.125, "learning_rate": 2.3424237339543836e-05, "loss": 0.1168, "step": 16902 }, { "epoch": 104.99065420560747, "grad_norm": 3.15625, "learning_rate": 2.3421778483788465e-05, "loss": 0.2176, "step": 16903 }, { "epoch": 104.99688473520249, "grad_norm": 2.0625, "learning_rate": 2.3419319643361093e-05, "loss": 0.1745, "step": 16904 }, { "epoch": 105.0, "grad_norm": 1.203125, "learning_rate": 2.341686081828561e-05, "loss": 0.0787, "step": 16905 }, { "epoch": 105.00623052959502, "grad_norm": 2.265625, "learning_rate": 2.3414402008585888e-05, "loss": 0.1296, "step": 16906 }, { "epoch": 105.01246105919003, "grad_norm": 2.6875, "learning_rate": 2.341194321428581e-05, "loss": 0.141, "step": 16907 }, { "epoch": 105.01869158878505, "grad_norm": 2.015625, "learning_rate": 2.340948443540926e-05, "loss": 0.1509, "step": 16908 }, { "epoch": 105.02492211838006, "grad_norm": 1.8828125, "learning_rate": 2.340702567198012e-05, "loss": 0.1511, "step": 16909 }, { "epoch": 105.03115264797508, "grad_norm": 1.8984375, "learning_rate": 2.3404566924022262e-05, "loss": 0.2308, "step": 16910 }, { "epoch": 105.03738317757009, "grad_norm": 1.7734375, "learning_rate": 2.3402108191559563e-05, "loss": 0.1136, "step": 16911 }, { "epoch": 105.04361370716511, "grad_norm": 1.8828125, "learning_rate": 2.339964947461591e-05, "loss": 0.1292, "step": 16912 }, { "epoch": 105.04984423676012, "grad_norm": 2.734375, "learning_rate": 2.3397190773215188e-05, "loss": 0.2084, "step": 16913 }, { "epoch": 105.05607476635514, "grad_norm": 2.609375, "learning_rate": 2.339473208738126e-05, "loss": 0.1575, "step": 16914 }, { "epoch": 105.06230529595015, "grad_norm": 1.4921875, "learning_rate": 2.3392273417138023e-05, "loss": 0.1059, "step": 16915 }, { "epoch": 105.06853582554517, "grad_norm": 3.015625, "learning_rate": 2.3389814762509348e-05, "loss": 0.2406, "step": 16916 }, { "epoch": 105.07476635514018, "grad_norm": 1.3671875, "learning_rate": 2.3387356123519114e-05, "loss": 0.097, "step": 16917 }, { "epoch": 105.0809968847352, "grad_norm": 1.0859375, "learning_rate": 2.3384897500191192e-05, "loss": 0.1038, "step": 16918 }, { "epoch": 105.08722741433021, "grad_norm": 1.015625, "learning_rate": 2.338243889254947e-05, "loss": 0.0865, "step": 16919 }, { "epoch": 105.09345794392523, "grad_norm": 1.75, "learning_rate": 2.3379980300617826e-05, "loss": 0.0952, "step": 16920 }, { "epoch": 105.09968847352025, "grad_norm": 1.109375, "learning_rate": 2.3377521724420153e-05, "loss": 0.0993, "step": 16921 }, { "epoch": 105.10591900311526, "grad_norm": 2.078125, "learning_rate": 2.337506316398029e-05, "loss": 0.1568, "step": 16922 }, { "epoch": 105.11214953271028, "grad_norm": 2.03125, "learning_rate": 2.337260461932215e-05, "loss": 0.125, "step": 16923 }, { "epoch": 105.11838006230529, "grad_norm": 2.53125, "learning_rate": 2.337014609046959e-05, "loss": 0.2558, "step": 16924 }, { "epoch": 105.1246105919003, "grad_norm": 2.21875, "learning_rate": 2.3367687577446505e-05, "loss": 0.2006, "step": 16925 }, { "epoch": 105.13084112149532, "grad_norm": 0.8984375, "learning_rate": 2.3365229080276765e-05, "loss": 0.0953, "step": 16926 }, { "epoch": 105.13707165109034, "grad_norm": 2.625, "learning_rate": 2.3362770598984242e-05, "loss": 0.0957, "step": 16927 }, { "epoch": 105.14330218068535, "grad_norm": 1.8125, "learning_rate": 2.3360312133592827e-05, "loss": 0.1139, "step": 16928 }, { "epoch": 105.14953271028037, "grad_norm": 2.125, "learning_rate": 2.3357853684126384e-05, "loss": 0.1641, "step": 16929 }, { "epoch": 105.15576323987538, "grad_norm": 2.890625, "learning_rate": 2.3355395250608795e-05, "loss": 0.1433, "step": 16930 }, { "epoch": 105.1619937694704, "grad_norm": 2.890625, "learning_rate": 2.335293683306393e-05, "loss": 0.243, "step": 16931 }, { "epoch": 105.16822429906541, "grad_norm": 4.125, "learning_rate": 2.3350478431515676e-05, "loss": 0.2194, "step": 16932 }, { "epoch": 105.17445482866043, "grad_norm": 2.390625, "learning_rate": 2.33480200459879e-05, "loss": 0.1789, "step": 16933 }, { "epoch": 105.18068535825545, "grad_norm": 2.96875, "learning_rate": 2.3345561676504502e-05, "loss": 0.2358, "step": 16934 }, { "epoch": 105.18691588785046, "grad_norm": 2.796875, "learning_rate": 2.3343103323089322e-05, "loss": 0.1467, "step": 16935 }, { "epoch": 105.19314641744548, "grad_norm": 1.4765625, "learning_rate": 2.3340644985766256e-05, "loss": 0.1338, "step": 16936 }, { "epoch": 105.19937694704049, "grad_norm": 1.96875, "learning_rate": 2.3338186664559175e-05, "loss": 0.1499, "step": 16937 }, { "epoch": 105.20560747663552, "grad_norm": 2.4375, "learning_rate": 2.3335728359491964e-05, "loss": 0.1252, "step": 16938 }, { "epoch": 105.21183800623054, "grad_norm": 1.578125, "learning_rate": 2.333327007058849e-05, "loss": 0.0988, "step": 16939 }, { "epoch": 105.21806853582555, "grad_norm": 2.515625, "learning_rate": 2.3330811797872636e-05, "loss": 0.207, "step": 16940 }, { "epoch": 105.22429906542057, "grad_norm": 1.15625, "learning_rate": 2.332835354136826e-05, "loss": 0.1175, "step": 16941 }, { "epoch": 105.23052959501558, "grad_norm": 3.65625, "learning_rate": 2.3325895301099257e-05, "loss": 0.1552, "step": 16942 }, { "epoch": 105.2367601246106, "grad_norm": 3.25, "learning_rate": 2.332343707708949e-05, "loss": 0.1858, "step": 16943 }, { "epoch": 105.24299065420561, "grad_norm": 4.59375, "learning_rate": 2.3320978869362835e-05, "loss": 0.177, "step": 16944 }, { "epoch": 105.24922118380063, "grad_norm": 2.21875, "learning_rate": 2.3318520677943174e-05, "loss": 0.0929, "step": 16945 }, { "epoch": 105.25545171339564, "grad_norm": 3.828125, "learning_rate": 2.3316062502854368e-05, "loss": 0.1828, "step": 16946 }, { "epoch": 105.26168224299066, "grad_norm": 2.578125, "learning_rate": 2.331360434412032e-05, "loss": 0.1619, "step": 16947 }, { "epoch": 105.26791277258567, "grad_norm": 2.0, "learning_rate": 2.331114620176486e-05, "loss": 0.1163, "step": 16948 }, { "epoch": 105.27414330218069, "grad_norm": 2.328125, "learning_rate": 2.3308688075811895e-05, "loss": 0.1468, "step": 16949 }, { "epoch": 105.2803738317757, "grad_norm": 4.59375, "learning_rate": 2.3306229966285286e-05, "loss": 0.2915, "step": 16950 }, { "epoch": 105.28660436137072, "grad_norm": 4.3125, "learning_rate": 2.3303771873208914e-05, "loss": 0.1481, "step": 16951 }, { "epoch": 105.29283489096574, "grad_norm": 4.9375, "learning_rate": 2.3301313796606652e-05, "loss": 0.1277, "step": 16952 }, { "epoch": 105.29906542056075, "grad_norm": 3.75, "learning_rate": 2.3298855736502373e-05, "loss": 0.2113, "step": 16953 }, { "epoch": 105.30529595015577, "grad_norm": 3.640625, "learning_rate": 2.3296397692919935e-05, "loss": 0.1023, "step": 16954 }, { "epoch": 105.31152647975078, "grad_norm": 2.765625, "learning_rate": 2.329393966588323e-05, "loss": 0.1111, "step": 16955 }, { "epoch": 105.3177570093458, "grad_norm": 3.40625, "learning_rate": 2.3291481655416124e-05, "loss": 0.2365, "step": 16956 }, { "epoch": 105.32398753894081, "grad_norm": 1.390625, "learning_rate": 2.328902366154248e-05, "loss": 0.119, "step": 16957 }, { "epoch": 105.33021806853583, "grad_norm": 3.421875, "learning_rate": 2.328656568428619e-05, "loss": 0.103, "step": 16958 }, { "epoch": 105.33644859813084, "grad_norm": 3.953125, "learning_rate": 2.328410772367112e-05, "loss": 0.1189, "step": 16959 }, { "epoch": 105.34267912772586, "grad_norm": 2.546875, "learning_rate": 2.3281649779721138e-05, "loss": 0.1395, "step": 16960 }, { "epoch": 105.34890965732087, "grad_norm": 4.15625, "learning_rate": 2.327919185246011e-05, "loss": 0.2252, "step": 16961 }, { "epoch": 105.35514018691589, "grad_norm": 2.09375, "learning_rate": 2.3276733941911917e-05, "loss": 0.1437, "step": 16962 }, { "epoch": 105.3613707165109, "grad_norm": 2.71875, "learning_rate": 2.327427604810043e-05, "loss": 0.0954, "step": 16963 }, { "epoch": 105.36760124610592, "grad_norm": 2.078125, "learning_rate": 2.327181817104951e-05, "loss": 0.2098, "step": 16964 }, { "epoch": 105.37383177570094, "grad_norm": 2.3125, "learning_rate": 2.3269360310783048e-05, "loss": 0.0934, "step": 16965 }, { "epoch": 105.38006230529595, "grad_norm": 2.859375, "learning_rate": 2.3266902467324907e-05, "loss": 0.1963, "step": 16966 }, { "epoch": 105.38629283489097, "grad_norm": 1.4765625, "learning_rate": 2.326444464069895e-05, "loss": 0.1224, "step": 16967 }, { "epoch": 105.39252336448598, "grad_norm": 1.0234375, "learning_rate": 2.326198683092905e-05, "loss": 0.0898, "step": 16968 }, { "epoch": 105.398753894081, "grad_norm": 2.15625, "learning_rate": 2.3259529038039087e-05, "loss": 0.1794, "step": 16969 }, { "epoch": 105.40498442367601, "grad_norm": 2.34375, "learning_rate": 2.325707126205292e-05, "loss": 0.2025, "step": 16970 }, { "epoch": 105.41121495327103, "grad_norm": 1.359375, "learning_rate": 2.325461350299443e-05, "loss": 0.1283, "step": 16971 }, { "epoch": 105.41744548286604, "grad_norm": 2.125, "learning_rate": 2.325215576088749e-05, "loss": 0.1757, "step": 16972 }, { "epoch": 105.42367601246106, "grad_norm": 2.3125, "learning_rate": 2.3249698035755954e-05, "loss": 0.2102, "step": 16973 }, { "epoch": 105.42990654205607, "grad_norm": 1.8671875, "learning_rate": 2.3247240327623697e-05, "loss": 0.1371, "step": 16974 }, { "epoch": 105.43613707165109, "grad_norm": 3.4375, "learning_rate": 2.3244782636514596e-05, "loss": 0.1421, "step": 16975 }, { "epoch": 105.4423676012461, "grad_norm": 1.8125, "learning_rate": 2.3242324962452518e-05, "loss": 0.1576, "step": 16976 }, { "epoch": 105.44859813084112, "grad_norm": 2.515625, "learning_rate": 2.323986730546133e-05, "loss": 0.1841, "step": 16977 }, { "epoch": 105.45482866043614, "grad_norm": 1.578125, "learning_rate": 2.3237409665564904e-05, "loss": 0.1413, "step": 16978 }, { "epoch": 105.46105919003115, "grad_norm": 3.921875, "learning_rate": 2.3234952042787113e-05, "loss": 0.2455, "step": 16979 }, { "epoch": 105.46728971962617, "grad_norm": 2.15625, "learning_rate": 2.3232494437151815e-05, "loss": 0.1655, "step": 16980 }, { "epoch": 105.47352024922118, "grad_norm": 2.453125, "learning_rate": 2.3230036848682877e-05, "loss": 0.1846, "step": 16981 }, { "epoch": 105.4797507788162, "grad_norm": 2.015625, "learning_rate": 2.3227579277404185e-05, "loss": 0.1282, "step": 16982 }, { "epoch": 105.48598130841121, "grad_norm": 1.8671875, "learning_rate": 2.3225121723339587e-05, "loss": 0.1518, "step": 16983 }, { "epoch": 105.49221183800623, "grad_norm": 1.953125, "learning_rate": 2.322266418651297e-05, "loss": 0.1318, "step": 16984 }, { "epoch": 105.49844236760124, "grad_norm": 1.6953125, "learning_rate": 2.3220206666948198e-05, "loss": 0.1531, "step": 16985 }, { "epoch": 105.50467289719626, "grad_norm": 1.6875, "learning_rate": 2.3217749164669135e-05, "loss": 0.17, "step": 16986 }, { "epoch": 105.51090342679127, "grad_norm": 1.5546875, "learning_rate": 2.3215291679699637e-05, "loss": 0.1082, "step": 16987 }, { "epoch": 105.51713395638629, "grad_norm": 1.8046875, "learning_rate": 2.3212834212063588e-05, "loss": 0.1694, "step": 16988 }, { "epoch": 105.5233644859813, "grad_norm": 3.453125, "learning_rate": 2.3210376761784853e-05, "loss": 0.1731, "step": 16989 }, { "epoch": 105.52959501557632, "grad_norm": 1.5078125, "learning_rate": 2.3207919328887288e-05, "loss": 0.1101, "step": 16990 }, { "epoch": 105.53582554517133, "grad_norm": 1.6796875, "learning_rate": 2.3205461913394784e-05, "loss": 0.1331, "step": 16991 }, { "epoch": 105.54205607476635, "grad_norm": 2.640625, "learning_rate": 2.3203004515331185e-05, "loss": 0.2009, "step": 16992 }, { "epoch": 105.54828660436137, "grad_norm": 2.9375, "learning_rate": 2.3200547134720364e-05, "loss": 0.1309, "step": 16993 }, { "epoch": 105.55451713395638, "grad_norm": 1.5, "learning_rate": 2.3198089771586185e-05, "loss": 0.1105, "step": 16994 }, { "epoch": 105.5607476635514, "grad_norm": 0.9921875, "learning_rate": 2.319563242595252e-05, "loss": 0.0922, "step": 16995 }, { "epoch": 105.56697819314641, "grad_norm": 2.75, "learning_rate": 2.3193175097843234e-05, "loss": 0.2333, "step": 16996 }, { "epoch": 105.57320872274143, "grad_norm": 3.078125, "learning_rate": 2.3190717787282194e-05, "loss": 0.1003, "step": 16997 }, { "epoch": 105.57943925233644, "grad_norm": 3.15625, "learning_rate": 2.3188260494293273e-05, "loss": 0.169, "step": 16998 }, { "epoch": 105.58566978193146, "grad_norm": 2.15625, "learning_rate": 2.318580321890032e-05, "loss": 0.1785, "step": 16999 }, { "epoch": 105.59190031152647, "grad_norm": 2.28125, "learning_rate": 2.3183345961127206e-05, "loss": 0.1794, "step": 17000 }, { "epoch": 105.59813084112149, "grad_norm": 2.5625, "learning_rate": 2.3180888720997803e-05, "loss": 0.1456, "step": 17001 }, { "epoch": 105.60436137071652, "grad_norm": 2.625, "learning_rate": 2.3178431498535972e-05, "loss": 0.1865, "step": 17002 }, { "epoch": 105.61059190031153, "grad_norm": 1.6328125, "learning_rate": 2.3175974293765573e-05, "loss": 0.1372, "step": 17003 }, { "epoch": 105.61682242990655, "grad_norm": 3.203125, "learning_rate": 2.317351710671049e-05, "loss": 0.1113, "step": 17004 }, { "epoch": 105.62305295950156, "grad_norm": 1.546875, "learning_rate": 2.3171059937394565e-05, "loss": 0.1352, "step": 17005 }, { "epoch": 105.62928348909658, "grad_norm": 2.109375, "learning_rate": 2.3168602785841674e-05, "loss": 0.1815, "step": 17006 }, { "epoch": 105.6355140186916, "grad_norm": 3.75, "learning_rate": 2.316614565207567e-05, "loss": 0.2159, "step": 17007 }, { "epoch": 105.64174454828661, "grad_norm": 3.1875, "learning_rate": 2.3163688536120434e-05, "loss": 0.2697, "step": 17008 }, { "epoch": 105.64797507788163, "grad_norm": 3.78125, "learning_rate": 2.316123143799982e-05, "loss": 0.3855, "step": 17009 }, { "epoch": 105.65420560747664, "grad_norm": 2.125, "learning_rate": 2.3158774357737702e-05, "loss": 0.1387, "step": 17010 }, { "epoch": 105.66043613707166, "grad_norm": 2.109375, "learning_rate": 2.3156317295357923e-05, "loss": 0.1664, "step": 17011 }, { "epoch": 105.66666666666667, "grad_norm": 1.9609375, "learning_rate": 2.3153860250884365e-05, "loss": 0.1426, "step": 17012 }, { "epoch": 105.67289719626169, "grad_norm": 2.03125, "learning_rate": 2.3151403224340886e-05, "loss": 0.2491, "step": 17013 }, { "epoch": 105.6791277258567, "grad_norm": 1.9609375, "learning_rate": 2.3148946215751344e-05, "loss": 0.1196, "step": 17014 }, { "epoch": 105.68535825545172, "grad_norm": 2.90625, "learning_rate": 2.314648922513961e-05, "loss": 0.2741, "step": 17015 }, { "epoch": 105.69158878504673, "grad_norm": 1.65625, "learning_rate": 2.3144032252529536e-05, "loss": 0.1241, "step": 17016 }, { "epoch": 105.69781931464175, "grad_norm": 1.9921875, "learning_rate": 2.3141575297945012e-05, "loss": 0.1256, "step": 17017 }, { "epoch": 105.70404984423676, "grad_norm": 1.8984375, "learning_rate": 2.313911836140986e-05, "loss": 0.1337, "step": 17018 }, { "epoch": 105.71028037383178, "grad_norm": 1.6796875, "learning_rate": 2.3136661442947967e-05, "loss": 0.1159, "step": 17019 }, { "epoch": 105.7165109034268, "grad_norm": 1.1953125, "learning_rate": 2.313420454258319e-05, "loss": 0.1032, "step": 17020 }, { "epoch": 105.72274143302181, "grad_norm": 2.046875, "learning_rate": 2.3131747660339394e-05, "loss": 0.1653, "step": 17021 }, { "epoch": 105.72897196261682, "grad_norm": 1.3203125, "learning_rate": 2.312929079624044e-05, "loss": 0.1183, "step": 17022 }, { "epoch": 105.73520249221184, "grad_norm": 2.6875, "learning_rate": 2.312683395031019e-05, "loss": 0.2186, "step": 17023 }, { "epoch": 105.74143302180686, "grad_norm": 3.3125, "learning_rate": 2.3124377122572497e-05, "loss": 0.4324, "step": 17024 }, { "epoch": 105.74766355140187, "grad_norm": 2.5, "learning_rate": 2.312192031305123e-05, "loss": 0.2081, "step": 17025 }, { "epoch": 105.75389408099689, "grad_norm": 2.53125, "learning_rate": 2.311946352177025e-05, "loss": 0.1944, "step": 17026 }, { "epoch": 105.7601246105919, "grad_norm": 3.03125, "learning_rate": 2.3117006748753407e-05, "loss": 0.2125, "step": 17027 }, { "epoch": 105.76635514018692, "grad_norm": 1.6328125, "learning_rate": 2.3114549994024583e-05, "loss": 0.1401, "step": 17028 }, { "epoch": 105.77258566978193, "grad_norm": 1.9375, "learning_rate": 2.3112093257607615e-05, "loss": 0.1208, "step": 17029 }, { "epoch": 105.77881619937695, "grad_norm": 3.25, "learning_rate": 2.3109636539526396e-05, "loss": 0.2314, "step": 17030 }, { "epoch": 105.78504672897196, "grad_norm": 2.203125, "learning_rate": 2.3107179839804745e-05, "loss": 0.09, "step": 17031 }, { "epoch": 105.79127725856698, "grad_norm": 1.40625, "learning_rate": 2.3104723158466544e-05, "loss": 0.1102, "step": 17032 }, { "epoch": 105.797507788162, "grad_norm": 1.734375, "learning_rate": 2.310226649553565e-05, "loss": 0.1576, "step": 17033 }, { "epoch": 105.80373831775701, "grad_norm": 1.3125, "learning_rate": 2.3099809851035927e-05, "loss": 0.124, "step": 17034 }, { "epoch": 105.80996884735202, "grad_norm": 1.6875, "learning_rate": 2.309735322499123e-05, "loss": 0.1671, "step": 17035 }, { "epoch": 105.81619937694704, "grad_norm": 1.34375, "learning_rate": 2.3094896617425426e-05, "loss": 0.1183, "step": 17036 }, { "epoch": 105.82242990654206, "grad_norm": 1.4453125, "learning_rate": 2.3092440028362355e-05, "loss": 0.1549, "step": 17037 }, { "epoch": 105.82866043613707, "grad_norm": 1.5078125, "learning_rate": 2.3089983457825897e-05, "loss": 0.1166, "step": 17038 }, { "epoch": 105.83489096573209, "grad_norm": 1.5234375, "learning_rate": 2.3087526905839897e-05, "loss": 0.093, "step": 17039 }, { "epoch": 105.8411214953271, "grad_norm": 1.6484375, "learning_rate": 2.3085070372428215e-05, "loss": 0.1247, "step": 17040 }, { "epoch": 105.84735202492212, "grad_norm": 2.75, "learning_rate": 2.3082613857614717e-05, "loss": 0.1649, "step": 17041 }, { "epoch": 105.85358255451713, "grad_norm": 3.0, "learning_rate": 2.308015736142326e-05, "loss": 0.2935, "step": 17042 }, { "epoch": 105.85981308411215, "grad_norm": 2.125, "learning_rate": 2.3077700883877692e-05, "loss": 0.1225, "step": 17043 }, { "epoch": 105.86604361370716, "grad_norm": 1.7734375, "learning_rate": 2.3075244425001874e-05, "loss": 0.1116, "step": 17044 }, { "epoch": 105.87227414330218, "grad_norm": 1.5, "learning_rate": 2.3072787984819673e-05, "loss": 0.1548, "step": 17045 }, { "epoch": 105.8785046728972, "grad_norm": 2.8125, "learning_rate": 2.3070331563354935e-05, "loss": 0.1119, "step": 17046 }, { "epoch": 105.88473520249221, "grad_norm": 1.84375, "learning_rate": 2.3067875160631526e-05, "loss": 0.146, "step": 17047 }, { "epoch": 105.89096573208722, "grad_norm": 1.859375, "learning_rate": 2.3065418776673302e-05, "loss": 0.1306, "step": 17048 }, { "epoch": 105.89719626168224, "grad_norm": 3.34375, "learning_rate": 2.3062962411504122e-05, "loss": 0.285, "step": 17049 }, { "epoch": 105.90342679127725, "grad_norm": 2.15625, "learning_rate": 2.306050606514783e-05, "loss": 0.1117, "step": 17050 }, { "epoch": 105.90965732087227, "grad_norm": 3.53125, "learning_rate": 2.305804973762829e-05, "loss": 0.2628, "step": 17051 }, { "epoch": 105.91588785046729, "grad_norm": 3.296875, "learning_rate": 2.3055593428969364e-05, "loss": 0.1643, "step": 17052 }, { "epoch": 105.9221183800623, "grad_norm": 2.703125, "learning_rate": 2.30531371391949e-05, "loss": 0.2285, "step": 17053 }, { "epoch": 105.92834890965732, "grad_norm": 2.3125, "learning_rate": 2.3050680868328756e-05, "loss": 0.0891, "step": 17054 }, { "epoch": 105.93457943925233, "grad_norm": 1.9453125, "learning_rate": 2.3048224616394796e-05, "loss": 0.1266, "step": 17055 }, { "epoch": 105.94080996884735, "grad_norm": 3.703125, "learning_rate": 2.3045768383416865e-05, "loss": 0.3064, "step": 17056 }, { "epoch": 105.94704049844236, "grad_norm": 2.5625, "learning_rate": 2.3043312169418817e-05, "loss": 0.1682, "step": 17057 }, { "epoch": 105.95327102803738, "grad_norm": 1.2265625, "learning_rate": 2.304085597442452e-05, "loss": 0.1231, "step": 17058 }, { "epoch": 105.9595015576324, "grad_norm": 1.03125, "learning_rate": 2.303839979845782e-05, "loss": 0.0896, "step": 17059 }, { "epoch": 105.96573208722741, "grad_norm": 2.609375, "learning_rate": 2.3035943641542566e-05, "loss": 0.1947, "step": 17060 }, { "epoch": 105.97196261682242, "grad_norm": 2.46875, "learning_rate": 2.303348750370263e-05, "loss": 0.2033, "step": 17061 }, { "epoch": 105.97819314641744, "grad_norm": 2.234375, "learning_rate": 2.3031031384961853e-05, "loss": 0.2198, "step": 17062 }, { "epoch": 105.98442367601245, "grad_norm": 2.546875, "learning_rate": 2.3028575285344092e-05, "loss": 0.144, "step": 17063 }, { "epoch": 105.99065420560747, "grad_norm": 2.546875, "learning_rate": 2.30261192048732e-05, "loss": 0.2318, "step": 17064 }, { "epoch": 105.99688473520249, "grad_norm": 1.625, "learning_rate": 2.3023663143573033e-05, "loss": 0.1189, "step": 17065 }, { "epoch": 106.0, "grad_norm": 0.78125, "learning_rate": 2.302120710146744e-05, "loss": 0.0523, "step": 17066 }, { "epoch": 106.00623052959502, "grad_norm": 1.65625, "learning_rate": 2.3018751078580287e-05, "loss": 0.132, "step": 17067 }, { "epoch": 106.01246105919003, "grad_norm": 3.0, "learning_rate": 2.3016295074935424e-05, "loss": 0.2638, "step": 17068 }, { "epoch": 106.01869158878505, "grad_norm": 2.390625, "learning_rate": 2.3013839090556692e-05, "loss": 0.2243, "step": 17069 }, { "epoch": 106.02492211838006, "grad_norm": 2.046875, "learning_rate": 2.301138312546795e-05, "loss": 0.1711, "step": 17070 }, { "epoch": 106.03115264797508, "grad_norm": 1.9609375, "learning_rate": 2.3008927179693056e-05, "loss": 0.1289, "step": 17071 }, { "epoch": 106.03738317757009, "grad_norm": 3.5625, "learning_rate": 2.3006471253255863e-05, "loss": 0.2944, "step": 17072 }, { "epoch": 106.04361370716511, "grad_norm": 1.65625, "learning_rate": 2.300401534618021e-05, "loss": 0.1136, "step": 17073 }, { "epoch": 106.04984423676012, "grad_norm": 2.59375, "learning_rate": 2.300155945848997e-05, "loss": 0.1979, "step": 17074 }, { "epoch": 106.05607476635514, "grad_norm": 2.484375, "learning_rate": 2.2999103590208983e-05, "loss": 0.2157, "step": 17075 }, { "epoch": 106.06230529595015, "grad_norm": 2.78125, "learning_rate": 2.2996647741361096e-05, "loss": 0.1518, "step": 17076 }, { "epoch": 106.06853582554517, "grad_norm": 3.046875, "learning_rate": 2.2994191911970165e-05, "loss": 0.2228, "step": 17077 }, { "epoch": 106.07476635514018, "grad_norm": 1.640625, "learning_rate": 2.2991736102060047e-05, "loss": 0.1171, "step": 17078 }, { "epoch": 106.0809968847352, "grad_norm": 1.2421875, "learning_rate": 2.2989280311654584e-05, "loss": 0.1228, "step": 17079 }, { "epoch": 106.08722741433021, "grad_norm": 1.1015625, "learning_rate": 2.298682454077765e-05, "loss": 0.0949, "step": 17080 }, { "epoch": 106.09345794392523, "grad_norm": 1.71875, "learning_rate": 2.298436878945306e-05, "loss": 0.1059, "step": 17081 }, { "epoch": 106.09968847352025, "grad_norm": 3.5, "learning_rate": 2.2981913057704692e-05, "loss": 0.2714, "step": 17082 }, { "epoch": 106.10591900311526, "grad_norm": 2.765625, "learning_rate": 2.297945734555638e-05, "loss": 0.2237, "step": 17083 }, { "epoch": 106.11214953271028, "grad_norm": 2.953125, "learning_rate": 2.2977001653031987e-05, "loss": 0.2109, "step": 17084 }, { "epoch": 106.11838006230529, "grad_norm": 2.3125, "learning_rate": 2.297454598015536e-05, "loss": 0.1085, "step": 17085 }, { "epoch": 106.1246105919003, "grad_norm": 1.953125, "learning_rate": 2.297209032695034e-05, "loss": 0.1617, "step": 17086 }, { "epoch": 106.13084112149532, "grad_norm": 2.75, "learning_rate": 2.2969634693440796e-05, "loss": 0.2441, "step": 17087 }, { "epoch": 106.13707165109034, "grad_norm": 1.5, "learning_rate": 2.2967179079650562e-05, "loss": 0.1341, "step": 17088 }, { "epoch": 106.14330218068535, "grad_norm": 2.03125, "learning_rate": 2.2964723485603487e-05, "loss": 0.1658, "step": 17089 }, { "epoch": 106.14953271028037, "grad_norm": 2.1875, "learning_rate": 2.296226791132342e-05, "loss": 0.1725, "step": 17090 }, { "epoch": 106.15576323987538, "grad_norm": 2.046875, "learning_rate": 2.2959812356834224e-05, "loss": 0.1549, "step": 17091 }, { "epoch": 106.1619937694704, "grad_norm": 2.25, "learning_rate": 2.2957356822159728e-05, "loss": 0.1752, "step": 17092 }, { "epoch": 106.16822429906541, "grad_norm": 2.09375, "learning_rate": 2.295490130732381e-05, "loss": 0.1731, "step": 17093 }, { "epoch": 106.17445482866043, "grad_norm": 2.203125, "learning_rate": 2.2952445812350282e-05, "loss": 0.1308, "step": 17094 }, { "epoch": 106.18068535825545, "grad_norm": 2.03125, "learning_rate": 2.2949990337263015e-05, "loss": 0.1512, "step": 17095 }, { "epoch": 106.18691588785046, "grad_norm": 1.484375, "learning_rate": 2.2947534882085847e-05, "loss": 0.1348, "step": 17096 }, { "epoch": 106.19314641744548, "grad_norm": 1.1875, "learning_rate": 2.2945079446842634e-05, "loss": 0.1104, "step": 17097 }, { "epoch": 106.19937694704049, "grad_norm": 1.34375, "learning_rate": 2.294262403155722e-05, "loss": 0.1031, "step": 17098 }, { "epoch": 106.20560747663552, "grad_norm": 2.90625, "learning_rate": 2.294016863625345e-05, "loss": 0.2772, "step": 17099 }, { "epoch": 106.21183800623054, "grad_norm": 1.453125, "learning_rate": 2.2937713260955186e-05, "loss": 0.1331, "step": 17100 }, { "epoch": 106.21806853582555, "grad_norm": 1.484375, "learning_rate": 2.293525790568625e-05, "loss": 0.1356, "step": 17101 }, { "epoch": 106.22429906542057, "grad_norm": 3.046875, "learning_rate": 2.293280257047051e-05, "loss": 0.2897, "step": 17102 }, { "epoch": 106.23052959501558, "grad_norm": 1.59375, "learning_rate": 2.2930347255331795e-05, "loss": 0.1455, "step": 17103 }, { "epoch": 106.2367601246106, "grad_norm": 1.0546875, "learning_rate": 2.292789196029397e-05, "loss": 0.0914, "step": 17104 }, { "epoch": 106.24299065420561, "grad_norm": 1.3984375, "learning_rate": 2.292543668538087e-05, "loss": 0.1046, "step": 17105 }, { "epoch": 106.24922118380063, "grad_norm": 2.640625, "learning_rate": 2.292298143061635e-05, "loss": 0.2517, "step": 17106 }, { "epoch": 106.25545171339564, "grad_norm": 2.125, "learning_rate": 2.292052619602424e-05, "loss": 0.1796, "step": 17107 }, { "epoch": 106.26168224299066, "grad_norm": 1.875, "learning_rate": 2.2918070981628398e-05, "loss": 0.1147, "step": 17108 }, { "epoch": 106.26791277258567, "grad_norm": 1.65625, "learning_rate": 2.2915615787452666e-05, "loss": 0.1112, "step": 17109 }, { "epoch": 106.27414330218069, "grad_norm": 1.7421875, "learning_rate": 2.291316061352089e-05, "loss": 0.1432, "step": 17110 }, { "epoch": 106.2803738317757, "grad_norm": 1.828125, "learning_rate": 2.2910705459856912e-05, "loss": 0.1985, "step": 17111 }, { "epoch": 106.28660436137072, "grad_norm": 1.484375, "learning_rate": 2.2908250326484593e-05, "loss": 0.1138, "step": 17112 }, { "epoch": 106.29283489096574, "grad_norm": 2.265625, "learning_rate": 2.290579521342776e-05, "loss": 0.1769, "step": 17113 }, { "epoch": 106.29906542056075, "grad_norm": 1.609375, "learning_rate": 2.290334012071025e-05, "loss": 0.1236, "step": 17114 }, { "epoch": 106.30529595015577, "grad_norm": 2.140625, "learning_rate": 2.290088504835593e-05, "loss": 0.1533, "step": 17115 }, { "epoch": 106.31152647975078, "grad_norm": 1.984375, "learning_rate": 2.289842999638863e-05, "loss": 0.1712, "step": 17116 }, { "epoch": 106.3177570093458, "grad_norm": 2.546875, "learning_rate": 2.2895974964832202e-05, "loss": 0.1775, "step": 17117 }, { "epoch": 106.32398753894081, "grad_norm": 2.28125, "learning_rate": 2.2893519953710485e-05, "loss": 0.1791, "step": 17118 }, { "epoch": 106.33021806853583, "grad_norm": 1.6484375, "learning_rate": 2.289106496304733e-05, "loss": 0.0859, "step": 17119 }, { "epoch": 106.33644859813084, "grad_norm": 1.7890625, "learning_rate": 2.288860999286656e-05, "loss": 0.1314, "step": 17120 }, { "epoch": 106.34267912772586, "grad_norm": 1.3984375, "learning_rate": 2.288615504319204e-05, "loss": 0.0803, "step": 17121 }, { "epoch": 106.34890965732087, "grad_norm": 2.3125, "learning_rate": 2.2883700114047602e-05, "loss": 0.1241, "step": 17122 }, { "epoch": 106.35514018691589, "grad_norm": 2.15625, "learning_rate": 2.2881245205457088e-05, "loss": 0.1711, "step": 17123 }, { "epoch": 106.3613707165109, "grad_norm": 3.34375, "learning_rate": 2.2878790317444348e-05, "loss": 0.1274, "step": 17124 }, { "epoch": 106.36760124610592, "grad_norm": 1.6875, "learning_rate": 2.287633545003323e-05, "loss": 0.1548, "step": 17125 }, { "epoch": 106.37383177570094, "grad_norm": 2.609375, "learning_rate": 2.2873880603247558e-05, "loss": 0.1258, "step": 17126 }, { "epoch": 106.38006230529595, "grad_norm": 2.28125, "learning_rate": 2.2871425777111178e-05, "loss": 0.1364, "step": 17127 }, { "epoch": 106.38629283489097, "grad_norm": 3.390625, "learning_rate": 2.2868970971647942e-05, "loss": 0.2454, "step": 17128 }, { "epoch": 106.39252336448598, "grad_norm": 2.296875, "learning_rate": 2.2866516186881678e-05, "loss": 0.1106, "step": 17129 }, { "epoch": 106.398753894081, "grad_norm": 2.5, "learning_rate": 2.2864061422836245e-05, "loss": 0.1385, "step": 17130 }, { "epoch": 106.40498442367601, "grad_norm": 2.390625, "learning_rate": 2.2861606679535473e-05, "loss": 0.1666, "step": 17131 }, { "epoch": 106.41121495327103, "grad_norm": 3.296875, "learning_rate": 2.2859151957003207e-05, "loss": 0.2559, "step": 17132 }, { "epoch": 106.41744548286604, "grad_norm": 1.2890625, "learning_rate": 2.2856697255263275e-05, "loss": 0.1039, "step": 17133 }, { "epoch": 106.42367601246106, "grad_norm": 3.65625, "learning_rate": 2.2854242574339533e-05, "loss": 0.3566, "step": 17134 }, { "epoch": 106.42990654205607, "grad_norm": 1.6875, "learning_rate": 2.2851787914255818e-05, "loss": 0.1096, "step": 17135 }, { "epoch": 106.43613707165109, "grad_norm": 4.09375, "learning_rate": 2.2849333275035964e-05, "loss": 0.3767, "step": 17136 }, { "epoch": 106.4423676012461, "grad_norm": 2.078125, "learning_rate": 2.2846878656703817e-05, "loss": 0.1295, "step": 17137 }, { "epoch": 106.44859813084112, "grad_norm": 1.765625, "learning_rate": 2.2844424059283223e-05, "loss": 0.1444, "step": 17138 }, { "epoch": 106.45482866043614, "grad_norm": 2.453125, "learning_rate": 2.2841969482798007e-05, "loss": 0.1295, "step": 17139 }, { "epoch": 106.46105919003115, "grad_norm": 3.265625, "learning_rate": 2.283951492727201e-05, "loss": 0.2179, "step": 17140 }, { "epoch": 106.46728971962617, "grad_norm": 1.6796875, "learning_rate": 2.2837060392729078e-05, "loss": 0.1415, "step": 17141 }, { "epoch": 106.47352024922118, "grad_norm": 2.703125, "learning_rate": 2.2834605879193047e-05, "loss": 0.2424, "step": 17142 }, { "epoch": 106.4797507788162, "grad_norm": 2.140625, "learning_rate": 2.283215138668776e-05, "loss": 0.1695, "step": 17143 }, { "epoch": 106.48598130841121, "grad_norm": 1.7421875, "learning_rate": 2.2829696915237055e-05, "loss": 0.1111, "step": 17144 }, { "epoch": 106.49221183800623, "grad_norm": 1.453125, "learning_rate": 2.282724246486477e-05, "loss": 0.1206, "step": 17145 }, { "epoch": 106.49844236760124, "grad_norm": 2.328125, "learning_rate": 2.2824788035594726e-05, "loss": 0.1671, "step": 17146 }, { "epoch": 106.50467289719626, "grad_norm": 2.9375, "learning_rate": 2.2822333627450786e-05, "loss": 0.187, "step": 17147 }, { "epoch": 106.51090342679127, "grad_norm": 2.71875, "learning_rate": 2.2819879240456774e-05, "loss": 0.2223, "step": 17148 }, { "epoch": 106.51713395638629, "grad_norm": 1.6640625, "learning_rate": 2.281742487463653e-05, "loss": 0.1232, "step": 17149 }, { "epoch": 106.5233644859813, "grad_norm": 2.65625, "learning_rate": 2.2814970530013894e-05, "loss": 0.2106, "step": 17150 }, { "epoch": 106.52959501557632, "grad_norm": 2.640625, "learning_rate": 2.2812516206612704e-05, "loss": 0.1985, "step": 17151 }, { "epoch": 106.53582554517133, "grad_norm": 2.46875, "learning_rate": 2.281006190445679e-05, "loss": 0.1755, "step": 17152 }, { "epoch": 106.54205607476635, "grad_norm": 1.6171875, "learning_rate": 2.2807607623569986e-05, "loss": 0.1073, "step": 17153 }, { "epoch": 106.54828660436137, "grad_norm": 2.4375, "learning_rate": 2.2805153363976144e-05, "loss": 0.1217, "step": 17154 }, { "epoch": 106.55451713395638, "grad_norm": 2.265625, "learning_rate": 2.2802699125699088e-05, "loss": 0.1755, "step": 17155 }, { "epoch": 106.5607476635514, "grad_norm": 2.75, "learning_rate": 2.2800244908762655e-05, "loss": 0.1241, "step": 17156 }, { "epoch": 106.56697819314641, "grad_norm": 2.484375, "learning_rate": 2.2797790713190696e-05, "loss": 0.2235, "step": 17157 }, { "epoch": 106.57320872274143, "grad_norm": 2.515625, "learning_rate": 2.2795336539007024e-05, "loss": 0.1875, "step": 17158 }, { "epoch": 106.57943925233644, "grad_norm": 2.078125, "learning_rate": 2.2792882386235485e-05, "loss": 0.1683, "step": 17159 }, { "epoch": 106.58566978193146, "grad_norm": 2.703125, "learning_rate": 2.2790428254899906e-05, "loss": 0.2028, "step": 17160 }, { "epoch": 106.59190031152647, "grad_norm": 2.03125, "learning_rate": 2.2787974145024138e-05, "loss": 0.1274, "step": 17161 }, { "epoch": 106.59813084112149, "grad_norm": 3.28125, "learning_rate": 2.2785520056632e-05, "loss": 0.1735, "step": 17162 }, { "epoch": 106.60436137071652, "grad_norm": 2.046875, "learning_rate": 2.2783065989747353e-05, "loss": 0.1438, "step": 17163 }, { "epoch": 106.61059190031153, "grad_norm": 1.3828125, "learning_rate": 2.2780611944393994e-05, "loss": 0.1194, "step": 17164 }, { "epoch": 106.61682242990655, "grad_norm": 2.84375, "learning_rate": 2.277815792059578e-05, "loss": 0.2458, "step": 17165 }, { "epoch": 106.62305295950156, "grad_norm": 2.140625, "learning_rate": 2.2775703918376538e-05, "loss": 0.1452, "step": 17166 }, { "epoch": 106.62928348909658, "grad_norm": 2.046875, "learning_rate": 2.2773249937760108e-05, "loss": 0.1546, "step": 17167 }, { "epoch": 106.6355140186916, "grad_norm": 1.2890625, "learning_rate": 2.2770795978770317e-05, "loss": 0.0908, "step": 17168 }, { "epoch": 106.64174454828661, "grad_norm": 1.7265625, "learning_rate": 2.2768342041431e-05, "loss": 0.1221, "step": 17169 }, { "epoch": 106.64797507788163, "grad_norm": 2.90625, "learning_rate": 2.2765888125766e-05, "loss": 0.1902, "step": 17170 }, { "epoch": 106.65420560747664, "grad_norm": 1.359375, "learning_rate": 2.2763434231799137e-05, "loss": 0.0909, "step": 17171 }, { "epoch": 106.66043613707166, "grad_norm": 3.046875, "learning_rate": 2.2760980359554244e-05, "loss": 0.1946, "step": 17172 }, { "epoch": 106.66666666666667, "grad_norm": 1.15625, "learning_rate": 2.2758526509055156e-05, "loss": 0.0915, "step": 17173 }, { "epoch": 106.67289719626169, "grad_norm": 2.328125, "learning_rate": 2.275607268032571e-05, "loss": 0.2123, "step": 17174 }, { "epoch": 106.6791277258567, "grad_norm": 2.28125, "learning_rate": 2.275361887338973e-05, "loss": 0.0945, "step": 17175 }, { "epoch": 106.68535825545172, "grad_norm": 2.515625, "learning_rate": 2.2751165088271066e-05, "loss": 0.1929, "step": 17176 }, { "epoch": 106.69158878504673, "grad_norm": 2.78125, "learning_rate": 2.2748711324993523e-05, "loss": 0.1226, "step": 17177 }, { "epoch": 106.69781931464175, "grad_norm": 2.03125, "learning_rate": 2.2746257583580952e-05, "loss": 0.186, "step": 17178 }, { "epoch": 106.70404984423676, "grad_norm": 2.734375, "learning_rate": 2.274380386405717e-05, "loss": 0.2556, "step": 17179 }, { "epoch": 106.71028037383178, "grad_norm": 2.515625, "learning_rate": 2.2741350166446025e-05, "loss": 0.2268, "step": 17180 }, { "epoch": 106.7165109034268, "grad_norm": 1.921875, "learning_rate": 2.2738896490771333e-05, "loss": 0.1778, "step": 17181 }, { "epoch": 106.72274143302181, "grad_norm": 1.1875, "learning_rate": 2.273644283705694e-05, "loss": 0.0974, "step": 17182 }, { "epoch": 106.72897196261682, "grad_norm": 2.0625, "learning_rate": 2.2733989205326654e-05, "loss": 0.1541, "step": 17183 }, { "epoch": 106.73520249221184, "grad_norm": 0.94921875, "learning_rate": 2.2731535595604322e-05, "loss": 0.0946, "step": 17184 }, { "epoch": 106.74143302180686, "grad_norm": 1.7265625, "learning_rate": 2.2729082007913774e-05, "loss": 0.1062, "step": 17185 }, { "epoch": 106.74766355140187, "grad_norm": 2.71875, "learning_rate": 2.2726628442278826e-05, "loss": 0.2512, "step": 17186 }, { "epoch": 106.75389408099689, "grad_norm": 2.515625, "learning_rate": 2.2724174898723323e-05, "loss": 0.2856, "step": 17187 }, { "epoch": 106.7601246105919, "grad_norm": 3.65625, "learning_rate": 2.2721721377271084e-05, "loss": 0.3754, "step": 17188 }, { "epoch": 106.76635514018692, "grad_norm": 1.859375, "learning_rate": 2.271926787794596e-05, "loss": 0.1566, "step": 17189 }, { "epoch": 106.77258566978193, "grad_norm": 2.15625, "learning_rate": 2.2716814400771742e-05, "loss": 0.2202, "step": 17190 }, { "epoch": 106.77881619937695, "grad_norm": 1.9921875, "learning_rate": 2.271436094577229e-05, "loss": 0.1193, "step": 17191 }, { "epoch": 106.78504672897196, "grad_norm": 1.890625, "learning_rate": 2.271190751297141e-05, "loss": 0.1208, "step": 17192 }, { "epoch": 106.79127725856698, "grad_norm": 1.765625, "learning_rate": 2.2709454102392952e-05, "loss": 0.1844, "step": 17193 }, { "epoch": 106.797507788162, "grad_norm": 1.796875, "learning_rate": 2.270700071406073e-05, "loss": 0.1712, "step": 17194 }, { "epoch": 106.80373831775701, "grad_norm": 1.84375, "learning_rate": 2.2704547347998585e-05, "loss": 0.1335, "step": 17195 }, { "epoch": 106.80996884735202, "grad_norm": 1.9375, "learning_rate": 2.270209400423032e-05, "loss": 0.1335, "step": 17196 }, { "epoch": 106.81619937694704, "grad_norm": 2.453125, "learning_rate": 2.2699640682779782e-05, "loss": 0.136, "step": 17197 }, { "epoch": 106.82242990654206, "grad_norm": 3.046875, "learning_rate": 2.2697187383670798e-05, "loss": 0.2626, "step": 17198 }, { "epoch": 106.82866043613707, "grad_norm": 1.28125, "learning_rate": 2.269473410692718e-05, "loss": 0.1004, "step": 17199 }, { "epoch": 106.83489096573209, "grad_norm": 1.703125, "learning_rate": 2.2692280852572775e-05, "loss": 0.1247, "step": 17200 }, { "epoch": 106.8411214953271, "grad_norm": 4.25, "learning_rate": 2.2689827620631403e-05, "loss": 0.192, "step": 17201 }, { "epoch": 106.84735202492212, "grad_norm": 2.0, "learning_rate": 2.2687374411126883e-05, "loss": 0.1151, "step": 17202 }, { "epoch": 106.85358255451713, "grad_norm": 4.5, "learning_rate": 2.2684921224083038e-05, "loss": 0.1103, "step": 17203 }, { "epoch": 106.85981308411215, "grad_norm": 1.9921875, "learning_rate": 2.268246805952371e-05, "loss": 0.2155, "step": 17204 }, { "epoch": 106.86604361370716, "grad_norm": 2.15625, "learning_rate": 2.268001491747271e-05, "loss": 0.1771, "step": 17205 }, { "epoch": 106.87227414330218, "grad_norm": 2.34375, "learning_rate": 2.2677561797953867e-05, "loss": 0.1095, "step": 17206 }, { "epoch": 106.8785046728972, "grad_norm": 2.484375, "learning_rate": 2.267510870099101e-05, "loss": 0.2343, "step": 17207 }, { "epoch": 106.88473520249221, "grad_norm": 2.6875, "learning_rate": 2.2672655626607972e-05, "loss": 0.2136, "step": 17208 }, { "epoch": 106.89096573208722, "grad_norm": 2.359375, "learning_rate": 2.267020257482856e-05, "loss": 0.1154, "step": 17209 }, { "epoch": 106.89719626168224, "grad_norm": 1.4453125, "learning_rate": 2.26677495456766e-05, "loss": 0.1226, "step": 17210 }, { "epoch": 106.90342679127725, "grad_norm": 2.046875, "learning_rate": 2.266529653917593e-05, "loss": 0.1425, "step": 17211 }, { "epoch": 106.90965732087227, "grad_norm": 1.7265625, "learning_rate": 2.266284355535036e-05, "loss": 0.0996, "step": 17212 }, { "epoch": 106.91588785046729, "grad_norm": 2.203125, "learning_rate": 2.2660390594223726e-05, "loss": 0.2076, "step": 17213 }, { "epoch": 106.9221183800623, "grad_norm": 1.765625, "learning_rate": 2.2657937655819854e-05, "loss": 0.1087, "step": 17214 }, { "epoch": 106.92834890965732, "grad_norm": 1.9921875, "learning_rate": 2.265548474016255e-05, "loss": 0.1544, "step": 17215 }, { "epoch": 106.93457943925233, "grad_norm": 1.5546875, "learning_rate": 2.2653031847275642e-05, "loss": 0.1226, "step": 17216 }, { "epoch": 106.94080996884735, "grad_norm": 2.484375, "learning_rate": 2.2650578977182967e-05, "loss": 0.1037, "step": 17217 }, { "epoch": 106.94704049844236, "grad_norm": 2.390625, "learning_rate": 2.2648126129908337e-05, "loss": 0.1895, "step": 17218 }, { "epoch": 106.95327102803738, "grad_norm": 2.859375, "learning_rate": 2.264567330547557e-05, "loss": 0.1973, "step": 17219 }, { "epoch": 106.9595015576324, "grad_norm": 2.5, "learning_rate": 2.2643220503908505e-05, "loss": 0.1122, "step": 17220 }, { "epoch": 106.96573208722741, "grad_norm": 1.171875, "learning_rate": 2.264076772523095e-05, "loss": 0.0894, "step": 17221 }, { "epoch": 106.97196261682242, "grad_norm": 3.328125, "learning_rate": 2.263831496946673e-05, "loss": 0.1099, "step": 17222 }, { "epoch": 106.97819314641744, "grad_norm": 0.875, "learning_rate": 2.263586223663966e-05, "loss": 0.0852, "step": 17223 }, { "epoch": 106.98442367601245, "grad_norm": 2.921875, "learning_rate": 2.2633409526773575e-05, "loss": 0.1983, "step": 17224 }, { "epoch": 106.99065420560747, "grad_norm": 2.90625, "learning_rate": 2.2630956839892284e-05, "loss": 0.1723, "step": 17225 }, { "epoch": 106.99688473520249, "grad_norm": 2.703125, "learning_rate": 2.262850417601962e-05, "loss": 0.203, "step": 17226 }, { "epoch": 107.0, "grad_norm": 3.3125, "learning_rate": 2.26260515351794e-05, "loss": 0.2029, "step": 17227 }, { "epoch": 107.00623052959502, "grad_norm": 2.671875, "learning_rate": 2.2623598917395438e-05, "loss": 0.2103, "step": 17228 }, { "epoch": 107.01246105919003, "grad_norm": 1.125, "learning_rate": 2.2621146322691555e-05, "loss": 0.1023, "step": 17229 }, { "epoch": 107.01869158878505, "grad_norm": 2.421875, "learning_rate": 2.261869375109158e-05, "loss": 0.1892, "step": 17230 }, { "epoch": 107.02492211838006, "grad_norm": 3.53125, "learning_rate": 2.2616241202619325e-05, "loss": 0.1401, "step": 17231 }, { "epoch": 107.03115264797508, "grad_norm": 1.4375, "learning_rate": 2.2613788677298608e-05, "loss": 0.131, "step": 17232 }, { "epoch": 107.03738317757009, "grad_norm": 1.9921875, "learning_rate": 2.261133617515326e-05, "loss": 0.1324, "step": 17233 }, { "epoch": 107.04361370716511, "grad_norm": 3.1875, "learning_rate": 2.2608883696207093e-05, "loss": 0.2144, "step": 17234 }, { "epoch": 107.04984423676012, "grad_norm": 1.3515625, "learning_rate": 2.2606431240483922e-05, "loss": 0.1229, "step": 17235 }, { "epoch": 107.05607476635514, "grad_norm": 3.21875, "learning_rate": 2.2603978808007563e-05, "loss": 0.2694, "step": 17236 }, { "epoch": 107.06230529595015, "grad_norm": 3.09375, "learning_rate": 2.260152639880185e-05, "loss": 0.3043, "step": 17237 }, { "epoch": 107.06853582554517, "grad_norm": 2.6875, "learning_rate": 2.2599074012890582e-05, "loss": 0.1372, "step": 17238 }, { "epoch": 107.07476635514018, "grad_norm": 1.8984375, "learning_rate": 2.2596621650297596e-05, "loss": 0.1543, "step": 17239 }, { "epoch": 107.0809968847352, "grad_norm": 1.8359375, "learning_rate": 2.2594169311046703e-05, "loss": 0.1203, "step": 17240 }, { "epoch": 107.08722741433021, "grad_norm": 2.890625, "learning_rate": 2.2591716995161716e-05, "loss": 0.1605, "step": 17241 }, { "epoch": 107.09345794392523, "grad_norm": 2.21875, "learning_rate": 2.2589264702666446e-05, "loss": 0.2635, "step": 17242 }, { "epoch": 107.09968847352025, "grad_norm": 2.921875, "learning_rate": 2.258681243358473e-05, "loss": 0.2132, "step": 17243 }, { "epoch": 107.10591900311526, "grad_norm": 2.265625, "learning_rate": 2.258436018794037e-05, "loss": 0.196, "step": 17244 }, { "epoch": 107.11214953271028, "grad_norm": 3.453125, "learning_rate": 2.2581907965757182e-05, "loss": 0.305, "step": 17245 }, { "epoch": 107.11838006230529, "grad_norm": 2.6875, "learning_rate": 2.2579455767059e-05, "loss": 0.1633, "step": 17246 }, { "epoch": 107.1246105919003, "grad_norm": 2.953125, "learning_rate": 2.257700359186962e-05, "loss": 0.277, "step": 17247 }, { "epoch": 107.13084112149532, "grad_norm": 1.4140625, "learning_rate": 2.2574551440212867e-05, "loss": 0.1087, "step": 17248 }, { "epoch": 107.13707165109034, "grad_norm": 2.890625, "learning_rate": 2.2572099312112546e-05, "loss": 0.1814, "step": 17249 }, { "epoch": 107.14330218068535, "grad_norm": 2.6875, "learning_rate": 2.2569647207592488e-05, "loss": 0.1017, "step": 17250 }, { "epoch": 107.14953271028037, "grad_norm": 1.9921875, "learning_rate": 2.2567195126676507e-05, "loss": 0.1861, "step": 17251 }, { "epoch": 107.15576323987538, "grad_norm": 1.3984375, "learning_rate": 2.2564743069388416e-05, "loss": 0.1421, "step": 17252 }, { "epoch": 107.1619937694704, "grad_norm": 2.171875, "learning_rate": 2.2562291035752015e-05, "loss": 0.1798, "step": 17253 }, { "epoch": 107.16822429906541, "grad_norm": 2.515625, "learning_rate": 2.2559839025791137e-05, "loss": 0.1809, "step": 17254 }, { "epoch": 107.17445482866043, "grad_norm": 1.6953125, "learning_rate": 2.2557387039529594e-05, "loss": 0.1067, "step": 17255 }, { "epoch": 107.18068535825545, "grad_norm": 1.3125, "learning_rate": 2.2554935076991184e-05, "loss": 0.126, "step": 17256 }, { "epoch": 107.18691588785046, "grad_norm": 1.78125, "learning_rate": 2.2552483138199742e-05, "loss": 0.22, "step": 17257 }, { "epoch": 107.19314641744548, "grad_norm": 2.140625, "learning_rate": 2.2550031223179067e-05, "loss": 0.1851, "step": 17258 }, { "epoch": 107.19937694704049, "grad_norm": 2.078125, "learning_rate": 2.2547579331952998e-05, "loss": 0.1673, "step": 17259 }, { "epoch": 107.20560747663552, "grad_norm": 2.90625, "learning_rate": 2.254512746454531e-05, "loss": 0.1741, "step": 17260 }, { "epoch": 107.21183800623054, "grad_norm": 3.140625, "learning_rate": 2.254267562097984e-05, "loss": 0.2735, "step": 17261 }, { "epoch": 107.21806853582555, "grad_norm": 3.53125, "learning_rate": 2.254022380128039e-05, "loss": 0.1335, "step": 17262 }, { "epoch": 107.22429906542057, "grad_norm": 2.703125, "learning_rate": 2.2537772005470782e-05, "loss": 0.156, "step": 17263 }, { "epoch": 107.23052959501558, "grad_norm": 2.5, "learning_rate": 2.2535320233574826e-05, "loss": 0.0935, "step": 17264 }, { "epoch": 107.2367601246106, "grad_norm": 2.140625, "learning_rate": 2.253286848561634e-05, "loss": 0.1956, "step": 17265 }, { "epoch": 107.24299065420561, "grad_norm": 2.34375, "learning_rate": 2.2530416761619118e-05, "loss": 0.1703, "step": 17266 }, { "epoch": 107.24922118380063, "grad_norm": 2.75, "learning_rate": 2.2527965061606982e-05, "loss": 0.1954, "step": 17267 }, { "epoch": 107.25545171339564, "grad_norm": 3.171875, "learning_rate": 2.252551338560375e-05, "loss": 0.2517, "step": 17268 }, { "epoch": 107.26168224299066, "grad_norm": 1.703125, "learning_rate": 2.2523061733633218e-05, "loss": 0.0905, "step": 17269 }, { "epoch": 107.26791277258567, "grad_norm": 2.609375, "learning_rate": 2.252061010571921e-05, "loss": 0.1245, "step": 17270 }, { "epoch": 107.27414330218069, "grad_norm": 1.6796875, "learning_rate": 2.251815850188554e-05, "loss": 0.1326, "step": 17271 }, { "epoch": 107.2803738317757, "grad_norm": 3.0625, "learning_rate": 2.2515706922156003e-05, "loss": 0.1079, "step": 17272 }, { "epoch": 107.28660436137072, "grad_norm": 3.671875, "learning_rate": 2.2513255366554413e-05, "loss": 0.3901, "step": 17273 }, { "epoch": 107.29283489096574, "grad_norm": 1.609375, "learning_rate": 2.251080383510459e-05, "loss": 0.1134, "step": 17274 }, { "epoch": 107.29906542056075, "grad_norm": 1.6953125, "learning_rate": 2.2508352327830328e-05, "loss": 0.1305, "step": 17275 }, { "epoch": 107.30529595015577, "grad_norm": 3.34375, "learning_rate": 2.2505900844755456e-05, "loss": 0.3222, "step": 17276 }, { "epoch": 107.31152647975078, "grad_norm": 3.140625, "learning_rate": 2.2503449385903772e-05, "loss": 0.1474, "step": 17277 }, { "epoch": 107.3177570093458, "grad_norm": 3.671875, "learning_rate": 2.2500997951299093e-05, "loss": 0.1796, "step": 17278 }, { "epoch": 107.32398753894081, "grad_norm": 2.734375, "learning_rate": 2.249854654096521e-05, "loss": 0.0892, "step": 17279 }, { "epoch": 107.33021806853583, "grad_norm": 3.515625, "learning_rate": 2.249609515492595e-05, "loss": 0.0902, "step": 17280 }, { "epoch": 107.33644859813084, "grad_norm": 2.78125, "learning_rate": 2.2493643793205113e-05, "loss": 0.2363, "step": 17281 }, { "epoch": 107.34267912772586, "grad_norm": 2.28125, "learning_rate": 2.2491192455826502e-05, "loss": 0.1317, "step": 17282 }, { "epoch": 107.34890965732087, "grad_norm": 3.453125, "learning_rate": 2.248874114281394e-05, "loss": 0.2224, "step": 17283 }, { "epoch": 107.35514018691589, "grad_norm": 2.78125, "learning_rate": 2.2486289854191228e-05, "loss": 0.2214, "step": 17284 }, { "epoch": 107.3613707165109, "grad_norm": 3.1875, "learning_rate": 2.248383858998217e-05, "loss": 0.2237, "step": 17285 }, { "epoch": 107.36760124610592, "grad_norm": 3.15625, "learning_rate": 2.2481387350210565e-05, "loss": 0.136, "step": 17286 }, { "epoch": 107.37383177570094, "grad_norm": 3.890625, "learning_rate": 2.2478936134900237e-05, "loss": 0.2034, "step": 17287 }, { "epoch": 107.38006230529595, "grad_norm": 4.34375, "learning_rate": 2.2476484944074978e-05, "loss": 0.2599, "step": 17288 }, { "epoch": 107.38629283489097, "grad_norm": 2.609375, "learning_rate": 2.2474033777758607e-05, "loss": 0.158, "step": 17289 }, { "epoch": 107.39252336448598, "grad_norm": 2.609375, "learning_rate": 2.2471582635974926e-05, "loss": 0.1953, "step": 17290 }, { "epoch": 107.398753894081, "grad_norm": 1.1328125, "learning_rate": 2.2469131518747747e-05, "loss": 0.0926, "step": 17291 }, { "epoch": 107.40498442367601, "grad_norm": 2.359375, "learning_rate": 2.2466680426100854e-05, "loss": 0.2083, "step": 17292 }, { "epoch": 107.41121495327103, "grad_norm": 2.078125, "learning_rate": 2.2464229358058076e-05, "loss": 0.1251, "step": 17293 }, { "epoch": 107.41744548286604, "grad_norm": 3.4375, "learning_rate": 2.2461778314643208e-05, "loss": 0.1418, "step": 17294 }, { "epoch": 107.42367601246106, "grad_norm": 2.46875, "learning_rate": 2.245932729588005e-05, "loss": 0.1012, "step": 17295 }, { "epoch": 107.42990654205607, "grad_norm": 2.515625, "learning_rate": 2.245687630179242e-05, "loss": 0.2221, "step": 17296 }, { "epoch": 107.43613707165109, "grad_norm": 2.890625, "learning_rate": 2.2454425332404122e-05, "loss": 0.1898, "step": 17297 }, { "epoch": 107.4423676012461, "grad_norm": 1.4453125, "learning_rate": 2.2451974387738947e-05, "loss": 0.1327, "step": 17298 }, { "epoch": 107.44859813084112, "grad_norm": 1.640625, "learning_rate": 2.24495234678207e-05, "loss": 0.148, "step": 17299 }, { "epoch": 107.45482866043614, "grad_norm": 2.0625, "learning_rate": 2.2447072572673196e-05, "loss": 0.1595, "step": 17300 }, { "epoch": 107.46105919003115, "grad_norm": 2.734375, "learning_rate": 2.2444621702320237e-05, "loss": 0.1817, "step": 17301 }, { "epoch": 107.46728971962617, "grad_norm": 2.28125, "learning_rate": 2.2442170856785615e-05, "loss": 0.1242, "step": 17302 }, { "epoch": 107.47352024922118, "grad_norm": 3.140625, "learning_rate": 2.2439720036093154e-05, "loss": 0.1651, "step": 17303 }, { "epoch": 107.4797507788162, "grad_norm": 2.109375, "learning_rate": 2.2437269240266633e-05, "loss": 0.1161, "step": 17304 }, { "epoch": 107.48598130841121, "grad_norm": 1.2109375, "learning_rate": 2.243481846932987e-05, "loss": 0.1054, "step": 17305 }, { "epoch": 107.49221183800623, "grad_norm": 2.671875, "learning_rate": 2.2432367723306657e-05, "loss": 0.208, "step": 17306 }, { "epoch": 107.49844236760124, "grad_norm": 2.453125, "learning_rate": 2.242991700222081e-05, "loss": 0.1266, "step": 17307 }, { "epoch": 107.50467289719626, "grad_norm": 2.75, "learning_rate": 2.242746630609611e-05, "loss": 0.1541, "step": 17308 }, { "epoch": 107.51090342679127, "grad_norm": 2.859375, "learning_rate": 2.2425015634956383e-05, "loss": 0.1451, "step": 17309 }, { "epoch": 107.51713395638629, "grad_norm": 2.890625, "learning_rate": 2.242256498882542e-05, "loss": 0.0926, "step": 17310 }, { "epoch": 107.5233644859813, "grad_norm": 3.4375, "learning_rate": 2.242011436772702e-05, "loss": 0.1993, "step": 17311 }, { "epoch": 107.52959501557632, "grad_norm": 2.0, "learning_rate": 2.2417663771684978e-05, "loss": 0.1473, "step": 17312 }, { "epoch": 107.53582554517133, "grad_norm": 2.25, "learning_rate": 2.2415213200723102e-05, "loss": 0.1254, "step": 17313 }, { "epoch": 107.54205607476635, "grad_norm": 3.515625, "learning_rate": 2.24127626548652e-05, "loss": 0.3333, "step": 17314 }, { "epoch": 107.54828660436137, "grad_norm": 3.109375, "learning_rate": 2.2410312134135054e-05, "loss": 0.2082, "step": 17315 }, { "epoch": 107.55451713395638, "grad_norm": 3.546875, "learning_rate": 2.2407861638556488e-05, "loss": 0.1784, "step": 17316 }, { "epoch": 107.5607476635514, "grad_norm": 3.875, "learning_rate": 2.240541116815328e-05, "loss": 0.2557, "step": 17317 }, { "epoch": 107.56697819314641, "grad_norm": 1.9140625, "learning_rate": 2.2402960722949237e-05, "loss": 0.1619, "step": 17318 }, { "epoch": 107.57320872274143, "grad_norm": 2.78125, "learning_rate": 2.2400510302968155e-05, "loss": 0.1716, "step": 17319 }, { "epoch": 107.57943925233644, "grad_norm": 2.921875, "learning_rate": 2.2398059908233842e-05, "loss": 0.144, "step": 17320 }, { "epoch": 107.58566978193146, "grad_norm": 2.328125, "learning_rate": 2.2395609538770086e-05, "loss": 0.1674, "step": 17321 }, { "epoch": 107.59190031152647, "grad_norm": 1.703125, "learning_rate": 2.2393159194600706e-05, "loss": 0.1067, "step": 17322 }, { "epoch": 107.59813084112149, "grad_norm": 2.015625, "learning_rate": 2.239070887574947e-05, "loss": 0.1114, "step": 17323 }, { "epoch": 107.60436137071652, "grad_norm": 2.421875, "learning_rate": 2.2388258582240194e-05, "loss": 0.1311, "step": 17324 }, { "epoch": 107.61059190031153, "grad_norm": 1.578125, "learning_rate": 2.2385808314096666e-05, "loss": 0.1164, "step": 17325 }, { "epoch": 107.61682242990655, "grad_norm": 1.375, "learning_rate": 2.2383358071342704e-05, "loss": 0.1231, "step": 17326 }, { "epoch": 107.62305295950156, "grad_norm": 2.90625, "learning_rate": 2.2380907854002084e-05, "loss": 0.2625, "step": 17327 }, { "epoch": 107.62928348909658, "grad_norm": 2.671875, "learning_rate": 2.2378457662098606e-05, "loss": 0.1089, "step": 17328 }, { "epoch": 107.6355140186916, "grad_norm": 1.375, "learning_rate": 2.2376007495656084e-05, "loss": 0.1056, "step": 17329 }, { "epoch": 107.64174454828661, "grad_norm": 1.96875, "learning_rate": 2.2373557354698294e-05, "loss": 0.1102, "step": 17330 }, { "epoch": 107.64797507788163, "grad_norm": 1.3359375, "learning_rate": 2.2371107239249043e-05, "loss": 0.0937, "step": 17331 }, { "epoch": 107.65420560747664, "grad_norm": 1.6953125, "learning_rate": 2.2368657149332116e-05, "loss": 0.1134, "step": 17332 }, { "epoch": 107.66043613707166, "grad_norm": 2.9375, "learning_rate": 2.2366207084971325e-05, "loss": 0.2296, "step": 17333 }, { "epoch": 107.66666666666667, "grad_norm": 3.21875, "learning_rate": 2.2363757046190448e-05, "loss": 0.1255, "step": 17334 }, { "epoch": 107.67289719626169, "grad_norm": 2.46875, "learning_rate": 2.2361307033013307e-05, "loss": 0.1648, "step": 17335 }, { "epoch": 107.6791277258567, "grad_norm": 2.125, "learning_rate": 2.2358857045463666e-05, "loss": 0.2401, "step": 17336 }, { "epoch": 107.68535825545172, "grad_norm": 2.359375, "learning_rate": 2.2356407083565337e-05, "loss": 0.2008, "step": 17337 }, { "epoch": 107.69158878504673, "grad_norm": 1.171875, "learning_rate": 2.23539571473421e-05, "loss": 0.1068, "step": 17338 }, { "epoch": 107.69781931464175, "grad_norm": 2.625, "learning_rate": 2.2351507236817775e-05, "loss": 0.2161, "step": 17339 }, { "epoch": 107.70404984423676, "grad_norm": 1.1875, "learning_rate": 2.2349057352016137e-05, "loss": 0.0955, "step": 17340 }, { "epoch": 107.71028037383178, "grad_norm": 3.15625, "learning_rate": 2.2346607492960988e-05, "loss": 0.2676, "step": 17341 }, { "epoch": 107.7165109034268, "grad_norm": 2.890625, "learning_rate": 2.234415765967611e-05, "loss": 0.1637, "step": 17342 }, { "epoch": 107.72274143302181, "grad_norm": 2.09375, "learning_rate": 2.2341707852185305e-05, "loss": 0.1779, "step": 17343 }, { "epoch": 107.72897196261682, "grad_norm": 1.9609375, "learning_rate": 2.233925807051237e-05, "loss": 0.1576, "step": 17344 }, { "epoch": 107.73520249221184, "grad_norm": 2.328125, "learning_rate": 2.2336808314681083e-05, "loss": 0.2163, "step": 17345 }, { "epoch": 107.74143302180686, "grad_norm": 1.4765625, "learning_rate": 2.2334358584715252e-05, "loss": 0.1083, "step": 17346 }, { "epoch": 107.74766355140187, "grad_norm": 2.1875, "learning_rate": 2.2331908880638665e-05, "loss": 0.1987, "step": 17347 }, { "epoch": 107.75389408099689, "grad_norm": 2.40625, "learning_rate": 2.2329459202475113e-05, "loss": 0.1002, "step": 17348 }, { "epoch": 107.7601246105919, "grad_norm": 1.703125, "learning_rate": 2.232700955024838e-05, "loss": 0.2227, "step": 17349 }, { "epoch": 107.76635514018692, "grad_norm": 1.1875, "learning_rate": 2.232455992398227e-05, "loss": 0.112, "step": 17350 }, { "epoch": 107.77258566978193, "grad_norm": 2.09375, "learning_rate": 2.2322110323700568e-05, "loss": 0.1475, "step": 17351 }, { "epoch": 107.77881619937695, "grad_norm": 1.4921875, "learning_rate": 2.231966074942706e-05, "loss": 0.1358, "step": 17352 }, { "epoch": 107.78504672897196, "grad_norm": 2.15625, "learning_rate": 2.231721120118555e-05, "loss": 0.1411, "step": 17353 }, { "epoch": 107.79127725856698, "grad_norm": 2.796875, "learning_rate": 2.2314761678999824e-05, "loss": 0.2276, "step": 17354 }, { "epoch": 107.797507788162, "grad_norm": 1.171875, "learning_rate": 2.2312312182893667e-05, "loss": 0.0857, "step": 17355 }, { "epoch": 107.80373831775701, "grad_norm": 1.578125, "learning_rate": 2.230986271289086e-05, "loss": 0.1398, "step": 17356 }, { "epoch": 107.80996884735202, "grad_norm": 2.765625, "learning_rate": 2.2307413269015216e-05, "loss": 0.2961, "step": 17357 }, { "epoch": 107.81619937694704, "grad_norm": 1.7265625, "learning_rate": 2.2304963851290504e-05, "loss": 0.1544, "step": 17358 }, { "epoch": 107.82242990654206, "grad_norm": 1.8203125, "learning_rate": 2.230251445974053e-05, "loss": 0.1966, "step": 17359 }, { "epoch": 107.82866043613707, "grad_norm": 1.9375, "learning_rate": 2.230006509438907e-05, "loss": 0.1535, "step": 17360 }, { "epoch": 107.83489096573209, "grad_norm": 1.3515625, "learning_rate": 2.2297615755259925e-05, "loss": 0.1139, "step": 17361 }, { "epoch": 107.8411214953271, "grad_norm": 2.421875, "learning_rate": 2.2295166442376865e-05, "loss": 0.1538, "step": 17362 }, { "epoch": 107.84735202492212, "grad_norm": 2.15625, "learning_rate": 2.2292717155763698e-05, "loss": 0.1006, "step": 17363 }, { "epoch": 107.85358255451713, "grad_norm": 2.515625, "learning_rate": 2.22902678954442e-05, "loss": 0.1748, "step": 17364 }, { "epoch": 107.85981308411215, "grad_norm": 2.078125, "learning_rate": 2.2287818661442156e-05, "loss": 0.1264, "step": 17365 }, { "epoch": 107.86604361370716, "grad_norm": 2.34375, "learning_rate": 2.2285369453781364e-05, "loss": 0.1665, "step": 17366 }, { "epoch": 107.87227414330218, "grad_norm": 1.4921875, "learning_rate": 2.2282920272485615e-05, "loss": 0.1341, "step": 17367 }, { "epoch": 107.8785046728972, "grad_norm": 1.6328125, "learning_rate": 2.228047111757868e-05, "loss": 0.1064, "step": 17368 }, { "epoch": 107.88473520249221, "grad_norm": 2.6875, "learning_rate": 2.227802198908435e-05, "loss": 0.2463, "step": 17369 }, { "epoch": 107.89096573208722, "grad_norm": 2.96875, "learning_rate": 2.227557288702642e-05, "loss": 0.1256, "step": 17370 }, { "epoch": 107.89719626168224, "grad_norm": 1.3515625, "learning_rate": 2.227312381142866e-05, "loss": 0.1412, "step": 17371 }, { "epoch": 107.90342679127725, "grad_norm": 1.78125, "learning_rate": 2.227067476231488e-05, "loss": 0.1419, "step": 17372 }, { "epoch": 107.90965732087227, "grad_norm": 3.234375, "learning_rate": 2.226822573970885e-05, "loss": 0.1752, "step": 17373 }, { "epoch": 107.91588785046729, "grad_norm": 2.890625, "learning_rate": 2.2265776743634355e-05, "loss": 0.3089, "step": 17374 }, { "epoch": 107.9221183800623, "grad_norm": 2.359375, "learning_rate": 2.226332777411518e-05, "loss": 0.228, "step": 17375 }, { "epoch": 107.92834890965732, "grad_norm": 2.59375, "learning_rate": 2.2260878831175117e-05, "loss": 0.2964, "step": 17376 }, { "epoch": 107.93457943925233, "grad_norm": 2.53125, "learning_rate": 2.2258429914837944e-05, "loss": 0.1433, "step": 17377 }, { "epoch": 107.94080996884735, "grad_norm": 2.15625, "learning_rate": 2.2255981025127442e-05, "loss": 0.1451, "step": 17378 }, { "epoch": 107.94704049844236, "grad_norm": 1.984375, "learning_rate": 2.225353216206741e-05, "loss": 0.1797, "step": 17379 }, { "epoch": 107.95327102803738, "grad_norm": 1.40625, "learning_rate": 2.2251083325681625e-05, "loss": 0.1001, "step": 17380 }, { "epoch": 107.9595015576324, "grad_norm": 1.75, "learning_rate": 2.2248634515993866e-05, "loss": 0.1353, "step": 17381 }, { "epoch": 107.96573208722741, "grad_norm": 1.046875, "learning_rate": 2.224618573302791e-05, "loss": 0.0968, "step": 17382 }, { "epoch": 107.97196261682242, "grad_norm": 0.93359375, "learning_rate": 2.2243736976807557e-05, "loss": 0.1244, "step": 17383 }, { "epoch": 107.97819314641744, "grad_norm": 2.453125, "learning_rate": 2.2241288247356572e-05, "loss": 0.1476, "step": 17384 }, { "epoch": 107.98442367601245, "grad_norm": 2.171875, "learning_rate": 2.2238839544698755e-05, "loss": 0.1943, "step": 17385 }, { "epoch": 107.99065420560747, "grad_norm": 3.265625, "learning_rate": 2.2236390868857885e-05, "loss": 0.1709, "step": 17386 }, { "epoch": 107.99688473520249, "grad_norm": 2.984375, "learning_rate": 2.2233942219857734e-05, "loss": 0.2362, "step": 17387 }, { "epoch": 108.0, "grad_norm": 1.1875, "learning_rate": 2.2231493597722084e-05, "loss": 0.0463, "step": 17388 }, { "epoch": 108.00623052959502, "grad_norm": 1.0625, "learning_rate": 2.222904500247473e-05, "loss": 0.1158, "step": 17389 }, { "epoch": 108.01246105919003, "grad_norm": 1.6953125, "learning_rate": 2.2226596434139436e-05, "loss": 0.1363, "step": 17390 }, { "epoch": 108.01869158878505, "grad_norm": 1.703125, "learning_rate": 2.2224147892739995e-05, "loss": 0.2147, "step": 17391 }, { "epoch": 108.02492211838006, "grad_norm": 2.015625, "learning_rate": 2.2221699378300193e-05, "loss": 0.0957, "step": 17392 }, { "epoch": 108.03115264797508, "grad_norm": 2.34375, "learning_rate": 2.2219250890843794e-05, "loss": 0.1709, "step": 17393 }, { "epoch": 108.03738317757009, "grad_norm": 2.53125, "learning_rate": 2.221680243039459e-05, "loss": 0.164, "step": 17394 }, { "epoch": 108.04361370716511, "grad_norm": 1.9375, "learning_rate": 2.221435399697635e-05, "loss": 0.193, "step": 17395 }, { "epoch": 108.04984423676012, "grad_norm": 1.375, "learning_rate": 2.2211905590612862e-05, "loss": 0.1014, "step": 17396 }, { "epoch": 108.05607476635514, "grad_norm": 2.484375, "learning_rate": 2.220945721132791e-05, "loss": 0.1418, "step": 17397 }, { "epoch": 108.06230529595015, "grad_norm": 2.203125, "learning_rate": 2.2207008859145262e-05, "loss": 0.142, "step": 17398 }, { "epoch": 108.06853582554517, "grad_norm": 3.703125, "learning_rate": 2.220456053408871e-05, "loss": 0.3293, "step": 17399 }, { "epoch": 108.07476635514018, "grad_norm": 2.828125, "learning_rate": 2.2202112236182023e-05, "loss": 0.2321, "step": 17400 }, { "epoch": 108.0809968847352, "grad_norm": 1.921875, "learning_rate": 2.219966396544898e-05, "loss": 0.109, "step": 17401 }, { "epoch": 108.08722741433021, "grad_norm": 2.5625, "learning_rate": 2.2197215721913353e-05, "loss": 0.162, "step": 17402 }, { "epoch": 108.09345794392523, "grad_norm": 2.171875, "learning_rate": 2.2194767505598936e-05, "loss": 0.1861, "step": 17403 }, { "epoch": 108.09968847352025, "grad_norm": 3.171875, "learning_rate": 2.2192319316529487e-05, "loss": 0.2744, "step": 17404 }, { "epoch": 108.10591900311526, "grad_norm": 2.9375, "learning_rate": 2.2189871154728813e-05, "loss": 0.2072, "step": 17405 }, { "epoch": 108.11214953271028, "grad_norm": 1.8671875, "learning_rate": 2.218742302022066e-05, "loss": 0.1363, "step": 17406 }, { "epoch": 108.11838006230529, "grad_norm": 2.671875, "learning_rate": 2.2184974913028815e-05, "loss": 0.1496, "step": 17407 }, { "epoch": 108.1246105919003, "grad_norm": 1.8046875, "learning_rate": 2.2182526833177054e-05, "loss": 0.1288, "step": 17408 }, { "epoch": 108.13084112149532, "grad_norm": 2.734375, "learning_rate": 2.218007878068916e-05, "loss": 0.2388, "step": 17409 }, { "epoch": 108.13707165109034, "grad_norm": 2.75, "learning_rate": 2.2177630755588908e-05, "loss": 0.101, "step": 17410 }, { "epoch": 108.14330218068535, "grad_norm": 3.625, "learning_rate": 2.2175182757900075e-05, "loss": 0.3496, "step": 17411 }, { "epoch": 108.14953271028037, "grad_norm": 2.265625, "learning_rate": 2.217273478764642e-05, "loss": 0.1398, "step": 17412 }, { "epoch": 108.15576323987538, "grad_norm": 2.4375, "learning_rate": 2.2170286844851734e-05, "loss": 0.2235, "step": 17413 }, { "epoch": 108.1619937694704, "grad_norm": 1.4296875, "learning_rate": 2.216783892953979e-05, "loss": 0.0946, "step": 17414 }, { "epoch": 108.16822429906541, "grad_norm": 2.34375, "learning_rate": 2.2165391041734352e-05, "loss": 0.1835, "step": 17415 }, { "epoch": 108.17445482866043, "grad_norm": 1.3359375, "learning_rate": 2.2162943181459213e-05, "loss": 0.1452, "step": 17416 }, { "epoch": 108.18068535825545, "grad_norm": 1.21875, "learning_rate": 2.2160495348738123e-05, "loss": 0.1012, "step": 17417 }, { "epoch": 108.18691588785046, "grad_norm": 2.546875, "learning_rate": 2.2158047543594895e-05, "loss": 0.2143, "step": 17418 }, { "epoch": 108.19314641744548, "grad_norm": 2.15625, "learning_rate": 2.2155599766053256e-05, "loss": 0.124, "step": 17419 }, { "epoch": 108.19937694704049, "grad_norm": 1.3984375, "learning_rate": 2.2153152016137006e-05, "loss": 0.1279, "step": 17420 }, { "epoch": 108.20560747663552, "grad_norm": 2.96875, "learning_rate": 2.215070429386991e-05, "loss": 0.2108, "step": 17421 }, { "epoch": 108.21183800623054, "grad_norm": 2.25, "learning_rate": 2.2148256599275748e-05, "loss": 0.1321, "step": 17422 }, { "epoch": 108.21806853582555, "grad_norm": 2.359375, "learning_rate": 2.214580893237829e-05, "loss": 0.1886, "step": 17423 }, { "epoch": 108.22429906542057, "grad_norm": 1.390625, "learning_rate": 2.214336129320131e-05, "loss": 0.1226, "step": 17424 }, { "epoch": 108.23052959501558, "grad_norm": 2.96875, "learning_rate": 2.2140913681768565e-05, "loss": 0.1969, "step": 17425 }, { "epoch": 108.2367601246106, "grad_norm": 2.265625, "learning_rate": 2.2138466098103842e-05, "loss": 0.188, "step": 17426 }, { "epoch": 108.24299065420561, "grad_norm": 1.7890625, "learning_rate": 2.213601854223091e-05, "loss": 0.0893, "step": 17427 }, { "epoch": 108.24922118380063, "grad_norm": 2.703125, "learning_rate": 2.2133571014173533e-05, "loss": 0.1613, "step": 17428 }, { "epoch": 108.25545171339564, "grad_norm": 2.171875, "learning_rate": 2.213112351395549e-05, "loss": 0.1084, "step": 17429 }, { "epoch": 108.26168224299066, "grad_norm": 3.0, "learning_rate": 2.2128676041600547e-05, "loss": 0.241, "step": 17430 }, { "epoch": 108.26791277258567, "grad_norm": 3.671875, "learning_rate": 2.212622859713249e-05, "loss": 0.2615, "step": 17431 }, { "epoch": 108.27414330218069, "grad_norm": 2.0, "learning_rate": 2.212378118057506e-05, "loss": 0.1299, "step": 17432 }, { "epoch": 108.2803738317757, "grad_norm": 1.5234375, "learning_rate": 2.212133379195205e-05, "loss": 0.0932, "step": 17433 }, { "epoch": 108.28660436137072, "grad_norm": 1.828125, "learning_rate": 2.2118886431287212e-05, "loss": 0.1744, "step": 17434 }, { "epoch": 108.29283489096574, "grad_norm": 1.71875, "learning_rate": 2.211643909860433e-05, "loss": 0.0987, "step": 17435 }, { "epoch": 108.29906542056075, "grad_norm": 1.78125, "learning_rate": 2.211399179392717e-05, "loss": 0.0957, "step": 17436 }, { "epoch": 108.30529595015577, "grad_norm": 2.140625, "learning_rate": 2.2111544517279508e-05, "loss": 0.136, "step": 17437 }, { "epoch": 108.31152647975078, "grad_norm": 2.734375, "learning_rate": 2.2109097268685085e-05, "loss": 0.1859, "step": 17438 }, { "epoch": 108.3177570093458, "grad_norm": 2.46875, "learning_rate": 2.2106650048167697e-05, "loss": 0.1068, "step": 17439 }, { "epoch": 108.32398753894081, "grad_norm": 2.578125, "learning_rate": 2.2104202855751098e-05, "loss": 0.2187, "step": 17440 }, { "epoch": 108.33021806853583, "grad_norm": 2.609375, "learning_rate": 2.210175569145906e-05, "loss": 0.3096, "step": 17441 }, { "epoch": 108.33644859813084, "grad_norm": 2.59375, "learning_rate": 2.209930855531535e-05, "loss": 0.2236, "step": 17442 }, { "epoch": 108.34267912772586, "grad_norm": 2.453125, "learning_rate": 2.2096861447343746e-05, "loss": 0.1722, "step": 17443 }, { "epoch": 108.34890965732087, "grad_norm": 1.9765625, "learning_rate": 2.2094414367567994e-05, "loss": 0.0845, "step": 17444 }, { "epoch": 108.35514018691589, "grad_norm": 2.15625, "learning_rate": 2.209196731601187e-05, "loss": 0.1646, "step": 17445 }, { "epoch": 108.3613707165109, "grad_norm": 2.25, "learning_rate": 2.2089520292699144e-05, "loss": 0.1457, "step": 17446 }, { "epoch": 108.36760124610592, "grad_norm": 2.359375, "learning_rate": 2.2087073297653578e-05, "loss": 0.1885, "step": 17447 }, { "epoch": 108.37383177570094, "grad_norm": 2.421875, "learning_rate": 2.208462633089893e-05, "loss": 0.1403, "step": 17448 }, { "epoch": 108.38006230529595, "grad_norm": 2.03125, "learning_rate": 2.2082179392458983e-05, "loss": 0.1873, "step": 17449 }, { "epoch": 108.38629283489097, "grad_norm": 1.21875, "learning_rate": 2.2079732482357497e-05, "loss": 0.1025, "step": 17450 }, { "epoch": 108.39252336448598, "grad_norm": 2.140625, "learning_rate": 2.2077285600618227e-05, "loss": 0.1301, "step": 17451 }, { "epoch": 108.398753894081, "grad_norm": 1.515625, "learning_rate": 2.207483874726494e-05, "loss": 0.1162, "step": 17452 }, { "epoch": 108.40498442367601, "grad_norm": 1.7421875, "learning_rate": 2.2072391922321407e-05, "loss": 0.1314, "step": 17453 }, { "epoch": 108.41121495327103, "grad_norm": 1.328125, "learning_rate": 2.2069945125811386e-05, "loss": 0.1219, "step": 17454 }, { "epoch": 108.41744548286604, "grad_norm": 1.3828125, "learning_rate": 2.206749835775865e-05, "loss": 0.1143, "step": 17455 }, { "epoch": 108.42367601246106, "grad_norm": 1.5546875, "learning_rate": 2.2065051618186958e-05, "loss": 0.1082, "step": 17456 }, { "epoch": 108.42990654205607, "grad_norm": 1.0234375, "learning_rate": 2.2062604907120067e-05, "loss": 0.1026, "step": 17457 }, { "epoch": 108.43613707165109, "grad_norm": 2.640625, "learning_rate": 2.206015822458174e-05, "loss": 0.1576, "step": 17458 }, { "epoch": 108.4423676012461, "grad_norm": 1.8125, "learning_rate": 2.2057711570595746e-05, "loss": 0.1488, "step": 17459 }, { "epoch": 108.44859813084112, "grad_norm": 2.015625, "learning_rate": 2.2055264945185848e-05, "loss": 0.1519, "step": 17460 }, { "epoch": 108.45482866043614, "grad_norm": 2.5625, "learning_rate": 2.20528183483758e-05, "loss": 0.194, "step": 17461 }, { "epoch": 108.46105919003115, "grad_norm": 2.3125, "learning_rate": 2.2050371780189378e-05, "loss": 0.1347, "step": 17462 }, { "epoch": 108.46728971962617, "grad_norm": 2.65625, "learning_rate": 2.2047925240650335e-05, "loss": 0.1998, "step": 17463 }, { "epoch": 108.47352024922118, "grad_norm": 1.765625, "learning_rate": 2.204547872978243e-05, "loss": 0.1325, "step": 17464 }, { "epoch": 108.4797507788162, "grad_norm": 1.09375, "learning_rate": 2.2043032247609418e-05, "loss": 0.0952, "step": 17465 }, { "epoch": 108.48598130841121, "grad_norm": 2.21875, "learning_rate": 2.2040585794155075e-05, "loss": 0.147, "step": 17466 }, { "epoch": 108.49221183800623, "grad_norm": 1.859375, "learning_rate": 2.203813936944315e-05, "loss": 0.2087, "step": 17467 }, { "epoch": 108.49844236760124, "grad_norm": 2.578125, "learning_rate": 2.203569297349741e-05, "loss": 0.1284, "step": 17468 }, { "epoch": 108.50467289719626, "grad_norm": 0.98046875, "learning_rate": 2.203324660634162e-05, "loss": 0.0941, "step": 17469 }, { "epoch": 108.51090342679127, "grad_norm": 0.87109375, "learning_rate": 2.2030800267999526e-05, "loss": 0.1149, "step": 17470 }, { "epoch": 108.51713395638629, "grad_norm": 1.6796875, "learning_rate": 2.202835395849489e-05, "loss": 0.1003, "step": 17471 }, { "epoch": 108.5233644859813, "grad_norm": 2.359375, "learning_rate": 2.2025907677851477e-05, "loss": 0.1721, "step": 17472 }, { "epoch": 108.52959501557632, "grad_norm": 1.625, "learning_rate": 2.2023461426093043e-05, "loss": 0.1735, "step": 17473 }, { "epoch": 108.53582554517133, "grad_norm": 1.9765625, "learning_rate": 2.2021015203243342e-05, "loss": 0.1629, "step": 17474 }, { "epoch": 108.54205607476635, "grad_norm": 1.8203125, "learning_rate": 2.201856900932615e-05, "loss": 0.1028, "step": 17475 }, { "epoch": 108.54828660436137, "grad_norm": 2.3125, "learning_rate": 2.2016122844365204e-05, "loss": 0.1942, "step": 17476 }, { "epoch": 108.55451713395638, "grad_norm": 1.4765625, "learning_rate": 2.201367670838427e-05, "loss": 0.1274, "step": 17477 }, { "epoch": 108.5607476635514, "grad_norm": 2.359375, "learning_rate": 2.2011230601407097e-05, "loss": 0.1734, "step": 17478 }, { "epoch": 108.56697819314641, "grad_norm": 2.28125, "learning_rate": 2.2008784523457455e-05, "loss": 0.1858, "step": 17479 }, { "epoch": 108.57320872274143, "grad_norm": 3.1875, "learning_rate": 2.2006338474559093e-05, "loss": 0.3053, "step": 17480 }, { "epoch": 108.57943925233644, "grad_norm": 2.671875, "learning_rate": 2.2003892454735786e-05, "loss": 0.2049, "step": 17481 }, { "epoch": 108.58566978193146, "grad_norm": 1.953125, "learning_rate": 2.200144646401126e-05, "loss": 0.0993, "step": 17482 }, { "epoch": 108.59190031152647, "grad_norm": 2.875, "learning_rate": 2.199900050240929e-05, "loss": 0.2471, "step": 17483 }, { "epoch": 108.59813084112149, "grad_norm": 2.5625, "learning_rate": 2.1996554569953616e-05, "loss": 0.1658, "step": 17484 }, { "epoch": 108.60436137071652, "grad_norm": 2.796875, "learning_rate": 2.1994108666668017e-05, "loss": 0.1763, "step": 17485 }, { "epoch": 108.61059190031153, "grad_norm": 2.5625, "learning_rate": 2.199166279257623e-05, "loss": 0.1777, "step": 17486 }, { "epoch": 108.61682242990655, "grad_norm": 4.09375, "learning_rate": 2.1989216947702017e-05, "loss": 0.255, "step": 17487 }, { "epoch": 108.62305295950156, "grad_norm": 1.7265625, "learning_rate": 2.1986771132069135e-05, "loss": 0.1245, "step": 17488 }, { "epoch": 108.62928348909658, "grad_norm": 3.5, "learning_rate": 2.1984325345701333e-05, "loss": 0.132, "step": 17489 }, { "epoch": 108.6355140186916, "grad_norm": 3.109375, "learning_rate": 2.198187958862236e-05, "loss": 0.1428, "step": 17490 }, { "epoch": 108.64174454828661, "grad_norm": 2.046875, "learning_rate": 2.1979433860855975e-05, "loss": 0.1227, "step": 17491 }, { "epoch": 108.64797507788163, "grad_norm": 2.078125, "learning_rate": 2.1976988162425937e-05, "loss": 0.1243, "step": 17492 }, { "epoch": 108.65420560747664, "grad_norm": 3.296875, "learning_rate": 2.1974542493355996e-05, "loss": 0.1351, "step": 17493 }, { "epoch": 108.66043613707166, "grad_norm": 2.828125, "learning_rate": 2.1972096853669906e-05, "loss": 0.2297, "step": 17494 }, { "epoch": 108.66666666666667, "grad_norm": 3.1875, "learning_rate": 2.1969651243391407e-05, "loss": 0.3254, "step": 17495 }, { "epoch": 108.67289719626169, "grad_norm": 2.1875, "learning_rate": 2.1967205662544264e-05, "loss": 0.1375, "step": 17496 }, { "epoch": 108.6791277258567, "grad_norm": 3.53125, "learning_rate": 2.1964760111152228e-05, "loss": 0.3389, "step": 17497 }, { "epoch": 108.68535825545172, "grad_norm": 4.90625, "learning_rate": 2.1962314589239042e-05, "loss": 0.1314, "step": 17498 }, { "epoch": 108.69158878504673, "grad_norm": 3.0, "learning_rate": 2.195986909682847e-05, "loss": 0.1577, "step": 17499 }, { "epoch": 108.69781931464175, "grad_norm": 2.25, "learning_rate": 2.1957423633944253e-05, "loss": 0.1288, "step": 17500 }, { "epoch": 108.70404984423676, "grad_norm": 1.625, "learning_rate": 2.195497820061016e-05, "loss": 0.1265, "step": 17501 }, { "epoch": 108.71028037383178, "grad_norm": 2.5, "learning_rate": 2.1952532796849912e-05, "loss": 0.1245, "step": 17502 }, { "epoch": 108.7165109034268, "grad_norm": 2.09375, "learning_rate": 2.195008742268728e-05, "loss": 0.1518, "step": 17503 }, { "epoch": 108.72274143302181, "grad_norm": 2.34375, "learning_rate": 2.1947642078146004e-05, "loss": 0.1875, "step": 17504 }, { "epoch": 108.72897196261682, "grad_norm": 3.640625, "learning_rate": 2.1945196763249842e-05, "loss": 0.1074, "step": 17505 }, { "epoch": 108.73520249221184, "grad_norm": 3.359375, "learning_rate": 2.194275147802254e-05, "loss": 0.2832, "step": 17506 }, { "epoch": 108.74143302180686, "grad_norm": 4.6875, "learning_rate": 2.194030622248785e-05, "loss": 0.1931, "step": 17507 }, { "epoch": 108.74766355140187, "grad_norm": 3.25, "learning_rate": 2.1937860996669513e-05, "loss": 0.1245, "step": 17508 }, { "epoch": 108.75389408099689, "grad_norm": 1.8046875, "learning_rate": 2.1935415800591284e-05, "loss": 0.1189, "step": 17509 }, { "epoch": 108.7601246105919, "grad_norm": 2.359375, "learning_rate": 2.193297063427691e-05, "loss": 0.1235, "step": 17510 }, { "epoch": 108.76635514018692, "grad_norm": 1.875, "learning_rate": 2.193052549775013e-05, "loss": 0.2238, "step": 17511 }, { "epoch": 108.77258566978193, "grad_norm": 2.328125, "learning_rate": 2.1928080391034708e-05, "loss": 0.1587, "step": 17512 }, { "epoch": 108.77881619937695, "grad_norm": 2.59375, "learning_rate": 2.1925635314154388e-05, "loss": 0.1213, "step": 17513 }, { "epoch": 108.78504672897196, "grad_norm": 2.609375, "learning_rate": 2.192319026713291e-05, "loss": 0.164, "step": 17514 }, { "epoch": 108.79127725856698, "grad_norm": 2.34375, "learning_rate": 2.1920745249994014e-05, "loss": 0.1455, "step": 17515 }, { "epoch": 108.797507788162, "grad_norm": 2.203125, "learning_rate": 2.191830026276146e-05, "loss": 0.1376, "step": 17516 }, { "epoch": 108.80373831775701, "grad_norm": 1.734375, "learning_rate": 2.1915855305458988e-05, "loss": 0.1095, "step": 17517 }, { "epoch": 108.80996884735202, "grad_norm": 1.6015625, "learning_rate": 2.191341037811035e-05, "loss": 0.1101, "step": 17518 }, { "epoch": 108.81619937694704, "grad_norm": 1.828125, "learning_rate": 2.1910965480739287e-05, "loss": 0.1495, "step": 17519 }, { "epoch": 108.82242990654206, "grad_norm": 2.28125, "learning_rate": 2.190852061336955e-05, "loss": 0.1885, "step": 17520 }, { "epoch": 108.82866043613707, "grad_norm": 2.5625, "learning_rate": 2.1906075776024866e-05, "loss": 0.1922, "step": 17521 }, { "epoch": 108.83489096573209, "grad_norm": 1.8046875, "learning_rate": 2.1903630968729003e-05, "loss": 0.1102, "step": 17522 }, { "epoch": 108.8411214953271, "grad_norm": 4.8125, "learning_rate": 2.1901186191505694e-05, "loss": 0.2158, "step": 17523 }, { "epoch": 108.84735202492212, "grad_norm": 1.9609375, "learning_rate": 2.1898741444378675e-05, "loss": 0.1228, "step": 17524 }, { "epoch": 108.85358255451713, "grad_norm": 2.984375, "learning_rate": 2.1896296727371706e-05, "loss": 0.2609, "step": 17525 }, { "epoch": 108.85981308411215, "grad_norm": 2.375, "learning_rate": 2.189385204050853e-05, "loss": 0.2471, "step": 17526 }, { "epoch": 108.86604361370716, "grad_norm": 1.734375, "learning_rate": 2.189140738381288e-05, "loss": 0.1013, "step": 17527 }, { "epoch": 108.87227414330218, "grad_norm": 3.140625, "learning_rate": 2.188896275730849e-05, "loss": 0.1256, "step": 17528 }, { "epoch": 108.8785046728972, "grad_norm": 2.59375, "learning_rate": 2.1886518161019127e-05, "loss": 0.1702, "step": 17529 }, { "epoch": 108.88473520249221, "grad_norm": 2.5, "learning_rate": 2.1884073594968518e-05, "loss": 0.185, "step": 17530 }, { "epoch": 108.89096573208722, "grad_norm": 1.8046875, "learning_rate": 2.188162905918041e-05, "loss": 0.1004, "step": 17531 }, { "epoch": 108.89719626168224, "grad_norm": 3.296875, "learning_rate": 2.1879184553678555e-05, "loss": 0.2401, "step": 17532 }, { "epoch": 108.90342679127725, "grad_norm": 3.53125, "learning_rate": 2.1876740078486676e-05, "loss": 0.2841, "step": 17533 }, { "epoch": 108.90965732087227, "grad_norm": 2.484375, "learning_rate": 2.1874295633628513e-05, "loss": 0.1657, "step": 17534 }, { "epoch": 108.91588785046729, "grad_norm": 2.90625, "learning_rate": 2.1871851219127822e-05, "loss": 0.2438, "step": 17535 }, { "epoch": 108.9221183800623, "grad_norm": 2.140625, "learning_rate": 2.186940683500834e-05, "loss": 0.1266, "step": 17536 }, { "epoch": 108.92834890965732, "grad_norm": 3.78125, "learning_rate": 2.1866962481293796e-05, "loss": 0.2412, "step": 17537 }, { "epoch": 108.93457943925233, "grad_norm": 1.9765625, "learning_rate": 2.1864518158007945e-05, "loss": 0.2055, "step": 17538 }, { "epoch": 108.94080996884735, "grad_norm": 1.625, "learning_rate": 2.1862073865174528e-05, "loss": 0.1301, "step": 17539 }, { "epoch": 108.94704049844236, "grad_norm": 2.71875, "learning_rate": 2.1859629602817268e-05, "loss": 0.2767, "step": 17540 }, { "epoch": 108.95327102803738, "grad_norm": 2.359375, "learning_rate": 2.1857185370959913e-05, "loss": 0.1699, "step": 17541 }, { "epoch": 108.9595015576324, "grad_norm": 1.5078125, "learning_rate": 2.1854741169626203e-05, "loss": 0.0971, "step": 17542 }, { "epoch": 108.96573208722741, "grad_norm": 3.078125, "learning_rate": 2.1852296998839876e-05, "loss": 0.1797, "step": 17543 }, { "epoch": 108.97196261682242, "grad_norm": 1.734375, "learning_rate": 2.1849852858624665e-05, "loss": 0.2004, "step": 17544 }, { "epoch": 108.97819314641744, "grad_norm": 1.578125, "learning_rate": 2.1847408749004322e-05, "loss": 0.1202, "step": 17545 }, { "epoch": 108.98442367601245, "grad_norm": 4.0625, "learning_rate": 2.184496467000257e-05, "loss": 0.271, "step": 17546 }, { "epoch": 108.99065420560747, "grad_norm": 1.2265625, "learning_rate": 2.1842520621643156e-05, "loss": 0.1154, "step": 17547 }, { "epoch": 108.99688473520249, "grad_norm": 1.2890625, "learning_rate": 2.1840076603949805e-05, "loss": 0.1142, "step": 17548 }, { "epoch": 109.0, "grad_norm": 2.015625, "learning_rate": 2.1837632616946266e-05, "loss": 0.1078, "step": 17549 }, { "epoch": 109.00623052959502, "grad_norm": 1.3828125, "learning_rate": 2.1835188660656267e-05, "loss": 0.1103, "step": 17550 }, { "epoch": 109.01246105919003, "grad_norm": 1.6796875, "learning_rate": 2.1832744735103563e-05, "loss": 0.1234, "step": 17551 }, { "epoch": 109.01869158878505, "grad_norm": 2.375, "learning_rate": 2.183030084031186e-05, "loss": 0.2064, "step": 17552 }, { "epoch": 109.02492211838006, "grad_norm": 3.359375, "learning_rate": 2.1827856976304915e-05, "loss": 0.1919, "step": 17553 }, { "epoch": 109.03115264797508, "grad_norm": 1.5859375, "learning_rate": 2.182541314310645e-05, "loss": 0.1185, "step": 17554 }, { "epoch": 109.03738317757009, "grad_norm": 2.484375, "learning_rate": 2.1822969340740212e-05, "loss": 0.1687, "step": 17555 }, { "epoch": 109.04361370716511, "grad_norm": 3.671875, "learning_rate": 2.1820525569229936e-05, "loss": 0.2956, "step": 17556 }, { "epoch": 109.04984423676012, "grad_norm": 1.9375, "learning_rate": 2.181808182859934e-05, "loss": 0.1448, "step": 17557 }, { "epoch": 109.05607476635514, "grad_norm": 2.1875, "learning_rate": 2.1815638118872183e-05, "loss": 0.1498, "step": 17558 }, { "epoch": 109.06230529595015, "grad_norm": 1.875, "learning_rate": 2.181319444007218e-05, "loss": 0.1598, "step": 17559 }, { "epoch": 109.06853582554517, "grad_norm": 2.703125, "learning_rate": 2.1810750792223066e-05, "loss": 0.1976, "step": 17560 }, { "epoch": 109.07476635514018, "grad_norm": 1.8828125, "learning_rate": 2.1808307175348578e-05, "loss": 0.1614, "step": 17561 }, { "epoch": 109.0809968847352, "grad_norm": 2.84375, "learning_rate": 2.180586358947245e-05, "loss": 0.1613, "step": 17562 }, { "epoch": 109.08722741433021, "grad_norm": 2.171875, "learning_rate": 2.180342003461841e-05, "loss": 0.1178, "step": 17563 }, { "epoch": 109.09345794392523, "grad_norm": 1.734375, "learning_rate": 2.180097651081021e-05, "loss": 0.1239, "step": 17564 }, { "epoch": 109.09968847352025, "grad_norm": 3.546875, "learning_rate": 2.1798533018071548e-05, "loss": 0.2135, "step": 17565 }, { "epoch": 109.10591900311526, "grad_norm": 2.609375, "learning_rate": 2.1796089556426178e-05, "loss": 0.3046, "step": 17566 }, { "epoch": 109.11214953271028, "grad_norm": 2.25, "learning_rate": 2.1793646125897824e-05, "loss": 0.2134, "step": 17567 }, { "epoch": 109.11838006230529, "grad_norm": 1.3359375, "learning_rate": 2.1791202726510223e-05, "loss": 0.1403, "step": 17568 }, { "epoch": 109.1246105919003, "grad_norm": 1.7890625, "learning_rate": 2.17887593582871e-05, "loss": 0.1611, "step": 17569 }, { "epoch": 109.13084112149532, "grad_norm": 2.484375, "learning_rate": 2.1786316021252188e-05, "loss": 0.1726, "step": 17570 }, { "epoch": 109.13707165109034, "grad_norm": 2.03125, "learning_rate": 2.1783872715429228e-05, "loss": 0.1764, "step": 17571 }, { "epoch": 109.14330218068535, "grad_norm": 2.53125, "learning_rate": 2.178142944084193e-05, "loss": 0.1427, "step": 17572 }, { "epoch": 109.14953271028037, "grad_norm": 1.9609375, "learning_rate": 2.1778986197514034e-05, "loss": 0.1481, "step": 17573 }, { "epoch": 109.15576323987538, "grad_norm": 4.03125, "learning_rate": 2.1776542985469263e-05, "loss": 0.4098, "step": 17574 }, { "epoch": 109.1619937694704, "grad_norm": 1.71875, "learning_rate": 2.1774099804731356e-05, "loss": 0.1344, "step": 17575 }, { "epoch": 109.16822429906541, "grad_norm": 1.265625, "learning_rate": 2.177165665532403e-05, "loss": 0.1045, "step": 17576 }, { "epoch": 109.17445482866043, "grad_norm": 2.140625, "learning_rate": 2.1769213537271037e-05, "loss": 0.1707, "step": 17577 }, { "epoch": 109.18068535825545, "grad_norm": 1.421875, "learning_rate": 2.176677045059607e-05, "loss": 0.1287, "step": 17578 }, { "epoch": 109.18691588785046, "grad_norm": 2.015625, "learning_rate": 2.1764327395322885e-05, "loss": 0.1563, "step": 17579 }, { "epoch": 109.19314641744548, "grad_norm": 3.4375, "learning_rate": 2.1761884371475188e-05, "loss": 0.1743, "step": 17580 }, { "epoch": 109.19937694704049, "grad_norm": 3.359375, "learning_rate": 2.1759441379076727e-05, "loss": 0.1241, "step": 17581 }, { "epoch": 109.20560747663552, "grad_norm": 2.1875, "learning_rate": 2.1756998418151218e-05, "loss": 0.1821, "step": 17582 }, { "epoch": 109.21183800623054, "grad_norm": 3.21875, "learning_rate": 2.175455548872239e-05, "loss": 0.2036, "step": 17583 }, { "epoch": 109.21806853582555, "grad_norm": 2.09375, "learning_rate": 2.1752112590813965e-05, "loss": 0.1497, "step": 17584 }, { "epoch": 109.22429906542057, "grad_norm": 1.7890625, "learning_rate": 2.174966972444967e-05, "loss": 0.143, "step": 17585 }, { "epoch": 109.23052959501558, "grad_norm": 3.03125, "learning_rate": 2.1747226889653235e-05, "loss": 0.1835, "step": 17586 }, { "epoch": 109.2367601246106, "grad_norm": 2.25, "learning_rate": 2.1744784086448378e-05, "loss": 0.1939, "step": 17587 }, { "epoch": 109.24299065420561, "grad_norm": 1.8671875, "learning_rate": 2.1742341314858832e-05, "loss": 0.174, "step": 17588 }, { "epoch": 109.24922118380063, "grad_norm": 1.8359375, "learning_rate": 2.1739898574908322e-05, "loss": 0.143, "step": 17589 }, { "epoch": 109.25545171339564, "grad_norm": 3.015625, "learning_rate": 2.173745586662057e-05, "loss": 0.2585, "step": 17590 }, { "epoch": 109.26168224299066, "grad_norm": 1.40625, "learning_rate": 2.1735013190019292e-05, "loss": 0.1211, "step": 17591 }, { "epoch": 109.26791277258567, "grad_norm": 3.65625, "learning_rate": 2.173257054512822e-05, "loss": 0.2298, "step": 17592 }, { "epoch": 109.27414330218069, "grad_norm": 2.0625, "learning_rate": 2.1730127931971078e-05, "loss": 0.1515, "step": 17593 }, { "epoch": 109.2803738317757, "grad_norm": 2.109375, "learning_rate": 2.172768535057158e-05, "loss": 0.1295, "step": 17594 }, { "epoch": 109.28660436137072, "grad_norm": 2.796875, "learning_rate": 2.1725242800953465e-05, "loss": 0.1373, "step": 17595 }, { "epoch": 109.29283489096574, "grad_norm": 1.703125, "learning_rate": 2.172280028314045e-05, "loss": 0.1347, "step": 17596 }, { "epoch": 109.29906542056075, "grad_norm": 2.359375, "learning_rate": 2.1720357797156248e-05, "loss": 0.189, "step": 17597 }, { "epoch": 109.30529595015577, "grad_norm": 1.953125, "learning_rate": 2.171791534302458e-05, "loss": 0.1485, "step": 17598 }, { "epoch": 109.31152647975078, "grad_norm": 1.5234375, "learning_rate": 2.171547292076918e-05, "loss": 0.1277, "step": 17599 }, { "epoch": 109.3177570093458, "grad_norm": 2.65625, "learning_rate": 2.1713030530413762e-05, "loss": 0.2385, "step": 17600 }, { "epoch": 109.32398753894081, "grad_norm": 2.4375, "learning_rate": 2.171058817198205e-05, "loss": 0.1732, "step": 17601 }, { "epoch": 109.33021806853583, "grad_norm": 1.5390625, "learning_rate": 2.170814584549777e-05, "loss": 0.1356, "step": 17602 }, { "epoch": 109.33644859813084, "grad_norm": 2.640625, "learning_rate": 2.1705703550984628e-05, "loss": 0.233, "step": 17603 }, { "epoch": 109.34267912772586, "grad_norm": 1.96875, "learning_rate": 2.170326128846635e-05, "loss": 0.224, "step": 17604 }, { "epoch": 109.34890965732087, "grad_norm": 2.296875, "learning_rate": 2.1700819057966662e-05, "loss": 0.113, "step": 17605 }, { "epoch": 109.35514018691589, "grad_norm": 2.78125, "learning_rate": 2.1698376859509278e-05, "loss": 0.1442, "step": 17606 }, { "epoch": 109.3613707165109, "grad_norm": 2.984375, "learning_rate": 2.1695934693117913e-05, "loss": 0.0988, "step": 17607 }, { "epoch": 109.36760124610592, "grad_norm": 1.8828125, "learning_rate": 2.1693492558816295e-05, "loss": 0.1674, "step": 17608 }, { "epoch": 109.37383177570094, "grad_norm": 2.625, "learning_rate": 2.1691050456628143e-05, "loss": 0.2643, "step": 17609 }, { "epoch": 109.38006230529595, "grad_norm": 2.4375, "learning_rate": 2.1688608386577166e-05, "loss": 0.1554, "step": 17610 }, { "epoch": 109.38629283489097, "grad_norm": 2.546875, "learning_rate": 2.1686166348687082e-05, "loss": 0.161, "step": 17611 }, { "epoch": 109.39252336448598, "grad_norm": 1.9140625, "learning_rate": 2.1683724342981617e-05, "loss": 0.1382, "step": 17612 }, { "epoch": 109.398753894081, "grad_norm": 2.640625, "learning_rate": 2.168128236948448e-05, "loss": 0.1886, "step": 17613 }, { "epoch": 109.40498442367601, "grad_norm": 2.625, "learning_rate": 2.1678840428219396e-05, "loss": 0.2101, "step": 17614 }, { "epoch": 109.41121495327103, "grad_norm": 3.125, "learning_rate": 2.1676398519210084e-05, "loss": 0.2353, "step": 17615 }, { "epoch": 109.41744548286604, "grad_norm": 2.703125, "learning_rate": 2.1673956642480246e-05, "loss": 0.2087, "step": 17616 }, { "epoch": 109.42367601246106, "grad_norm": 2.359375, "learning_rate": 2.1671514798053605e-05, "loss": 0.15, "step": 17617 }, { "epoch": 109.42990654205607, "grad_norm": 1.6796875, "learning_rate": 2.1669072985953878e-05, "loss": 0.1651, "step": 17618 }, { "epoch": 109.43613707165109, "grad_norm": 1.0546875, "learning_rate": 2.1666631206204786e-05, "loss": 0.1282, "step": 17619 }, { "epoch": 109.4423676012461, "grad_norm": 4.03125, "learning_rate": 2.1664189458830024e-05, "loss": 0.3533, "step": 17620 }, { "epoch": 109.44859813084112, "grad_norm": 2.5, "learning_rate": 2.1661747743853335e-05, "loss": 0.2143, "step": 17621 }, { "epoch": 109.45482866043614, "grad_norm": 1.6875, "learning_rate": 2.165930606129842e-05, "loss": 0.1282, "step": 17622 }, { "epoch": 109.46105919003115, "grad_norm": 2.75, "learning_rate": 2.1656864411188987e-05, "loss": 0.2266, "step": 17623 }, { "epoch": 109.46728971962617, "grad_norm": 3.96875, "learning_rate": 2.165442279354875e-05, "loss": 0.5725, "step": 17624 }, { "epoch": 109.47352024922118, "grad_norm": 2.015625, "learning_rate": 2.1651981208401435e-05, "loss": 0.1493, "step": 17625 }, { "epoch": 109.4797507788162, "grad_norm": 0.97265625, "learning_rate": 2.164953965577074e-05, "loss": 0.1054, "step": 17626 }, { "epoch": 109.48598130841121, "grad_norm": 1.4921875, "learning_rate": 2.1647098135680393e-05, "loss": 0.1438, "step": 17627 }, { "epoch": 109.49221183800623, "grad_norm": 1.4609375, "learning_rate": 2.1644656648154103e-05, "loss": 0.1105, "step": 17628 }, { "epoch": 109.49844236760124, "grad_norm": 2.109375, "learning_rate": 2.1642215193215576e-05, "loss": 0.1171, "step": 17629 }, { "epoch": 109.50467289719626, "grad_norm": 1.4375, "learning_rate": 2.1639773770888516e-05, "loss": 0.1323, "step": 17630 }, { "epoch": 109.51090342679127, "grad_norm": 2.34375, "learning_rate": 2.1637332381196652e-05, "loss": 0.1573, "step": 17631 }, { "epoch": 109.51713395638629, "grad_norm": 1.6640625, "learning_rate": 2.163489102416369e-05, "loss": 0.1356, "step": 17632 }, { "epoch": 109.5233644859813, "grad_norm": 2.109375, "learning_rate": 2.1632449699813333e-05, "loss": 0.1774, "step": 17633 }, { "epoch": 109.52959501557632, "grad_norm": 1.1640625, "learning_rate": 2.1630008408169307e-05, "loss": 0.0982, "step": 17634 }, { "epoch": 109.53582554517133, "grad_norm": 2.75, "learning_rate": 2.162756714925531e-05, "loss": 0.2912, "step": 17635 }, { "epoch": 109.54205607476635, "grad_norm": 1.2578125, "learning_rate": 2.162512592309505e-05, "loss": 0.1376, "step": 17636 }, { "epoch": 109.54828660436137, "grad_norm": 2.8125, "learning_rate": 2.1622684729712242e-05, "loss": 0.1829, "step": 17637 }, { "epoch": 109.55451713395638, "grad_norm": 2.890625, "learning_rate": 2.1620243569130596e-05, "loss": 0.2032, "step": 17638 }, { "epoch": 109.5607476635514, "grad_norm": 1.78125, "learning_rate": 2.1617802441373826e-05, "loss": 0.1143, "step": 17639 }, { "epoch": 109.56697819314641, "grad_norm": 1.9296875, "learning_rate": 2.1615361346465623e-05, "loss": 0.1063, "step": 17640 }, { "epoch": 109.57320872274143, "grad_norm": 1.6796875, "learning_rate": 2.1612920284429722e-05, "loss": 0.1156, "step": 17641 }, { "epoch": 109.57943925233644, "grad_norm": 2.578125, "learning_rate": 2.161047925528981e-05, "loss": 0.2273, "step": 17642 }, { "epoch": 109.58566978193146, "grad_norm": 2.3125, "learning_rate": 2.16080382590696e-05, "loss": 0.195, "step": 17643 }, { "epoch": 109.59190031152647, "grad_norm": 2.4375, "learning_rate": 2.1605597295792793e-05, "loss": 0.1178, "step": 17644 }, { "epoch": 109.59813084112149, "grad_norm": 2.515625, "learning_rate": 2.1603156365483112e-05, "loss": 0.2561, "step": 17645 }, { "epoch": 109.60436137071652, "grad_norm": 2.4375, "learning_rate": 2.160071546816425e-05, "loss": 0.1776, "step": 17646 }, { "epoch": 109.61059190031153, "grad_norm": 1.7890625, "learning_rate": 2.1598274603859934e-05, "loss": 0.1463, "step": 17647 }, { "epoch": 109.61682242990655, "grad_norm": 1.3203125, "learning_rate": 2.159583377259384e-05, "loss": 0.1093, "step": 17648 }, { "epoch": 109.62305295950156, "grad_norm": 2.90625, "learning_rate": 2.159339297438969e-05, "loss": 0.1763, "step": 17649 }, { "epoch": 109.62928348909658, "grad_norm": 2.25, "learning_rate": 2.1590952209271188e-05, "loss": 0.1289, "step": 17650 }, { "epoch": 109.6355140186916, "grad_norm": 1.6328125, "learning_rate": 2.1588511477262048e-05, "loss": 0.1249, "step": 17651 }, { "epoch": 109.64174454828661, "grad_norm": 2.921875, "learning_rate": 2.158607077838596e-05, "loss": 0.1473, "step": 17652 }, { "epoch": 109.64797507788163, "grad_norm": 2.625, "learning_rate": 2.158363011266664e-05, "loss": 0.3043, "step": 17653 }, { "epoch": 109.65420560747664, "grad_norm": 2.34375, "learning_rate": 2.158118948012778e-05, "loss": 0.1863, "step": 17654 }, { "epoch": 109.66043613707166, "grad_norm": 3.875, "learning_rate": 2.1578748880793094e-05, "loss": 0.3789, "step": 17655 }, { "epoch": 109.66666666666667, "grad_norm": 2.796875, "learning_rate": 2.1576308314686286e-05, "loss": 0.2383, "step": 17656 }, { "epoch": 109.67289719626169, "grad_norm": 1.5546875, "learning_rate": 2.157386778183105e-05, "loss": 0.1609, "step": 17657 }, { "epoch": 109.6791277258567, "grad_norm": 3.265625, "learning_rate": 2.15714272822511e-05, "loss": 0.3076, "step": 17658 }, { "epoch": 109.68535825545172, "grad_norm": 2.59375, "learning_rate": 2.1568986815970126e-05, "loss": 0.2414, "step": 17659 }, { "epoch": 109.69158878504673, "grad_norm": 1.53125, "learning_rate": 2.1566546383011856e-05, "loss": 0.0926, "step": 17660 }, { "epoch": 109.69781931464175, "grad_norm": 3.046875, "learning_rate": 2.1564105983399956e-05, "loss": 0.1818, "step": 17661 }, { "epoch": 109.70404984423676, "grad_norm": 1.890625, "learning_rate": 2.156166561715815e-05, "loss": 0.14, "step": 17662 }, { "epoch": 109.71028037383178, "grad_norm": 1.46875, "learning_rate": 2.1559225284310136e-05, "loss": 0.1298, "step": 17663 }, { "epoch": 109.7165109034268, "grad_norm": 1.5703125, "learning_rate": 2.1556784984879614e-05, "loss": 0.1017, "step": 17664 }, { "epoch": 109.72274143302181, "grad_norm": 2.40625, "learning_rate": 2.1554344718890284e-05, "loss": 0.1478, "step": 17665 }, { "epoch": 109.72897196261682, "grad_norm": 3.59375, "learning_rate": 2.1551904486365855e-05, "loss": 0.1207, "step": 17666 }, { "epoch": 109.73520249221184, "grad_norm": 3.3125, "learning_rate": 2.1549464287330007e-05, "loss": 0.2756, "step": 17667 }, { "epoch": 109.74143302180686, "grad_norm": 4.03125, "learning_rate": 2.1547024121806456e-05, "loss": 0.1484, "step": 17668 }, { "epoch": 109.74766355140187, "grad_norm": 3.53125, "learning_rate": 2.1544583989818898e-05, "loss": 0.1915, "step": 17669 }, { "epoch": 109.75389408099689, "grad_norm": 2.265625, "learning_rate": 2.1542143891391025e-05, "loss": 0.11, "step": 17670 }, { "epoch": 109.7601246105919, "grad_norm": 3.25, "learning_rate": 2.153970382654655e-05, "loss": 0.2654, "step": 17671 }, { "epoch": 109.76635514018692, "grad_norm": 1.1796875, "learning_rate": 2.1537263795309165e-05, "loss": 0.0803, "step": 17672 }, { "epoch": 109.77258566978193, "grad_norm": 1.6953125, "learning_rate": 2.153482379770256e-05, "loss": 0.1321, "step": 17673 }, { "epoch": 109.77881619937695, "grad_norm": 2.1875, "learning_rate": 2.1532383833750435e-05, "loss": 0.1422, "step": 17674 }, { "epoch": 109.78504672897196, "grad_norm": 2.265625, "learning_rate": 2.1529943903476495e-05, "loss": 0.1097, "step": 17675 }, { "epoch": 109.79127725856698, "grad_norm": 2.40625, "learning_rate": 2.152750400690443e-05, "loss": 0.1128, "step": 17676 }, { "epoch": 109.797507788162, "grad_norm": 2.734375, "learning_rate": 2.152506414405795e-05, "loss": 0.1036, "step": 17677 }, { "epoch": 109.80373831775701, "grad_norm": 3.21875, "learning_rate": 2.1522624314960736e-05, "loss": 0.2895, "step": 17678 }, { "epoch": 109.80996884735202, "grad_norm": 2.359375, "learning_rate": 2.1520184519636496e-05, "loss": 0.1101, "step": 17679 }, { "epoch": 109.81619937694704, "grad_norm": 1.6875, "learning_rate": 2.151774475810891e-05, "loss": 0.1276, "step": 17680 }, { "epoch": 109.82242990654206, "grad_norm": 1.625, "learning_rate": 2.151530503040169e-05, "loss": 0.128, "step": 17681 }, { "epoch": 109.82866043613707, "grad_norm": 1.625, "learning_rate": 2.151286533653852e-05, "loss": 0.1277, "step": 17682 }, { "epoch": 109.83489096573209, "grad_norm": 3.375, "learning_rate": 2.1510425676543096e-05, "loss": 0.1282, "step": 17683 }, { "epoch": 109.8411214953271, "grad_norm": 2.640625, "learning_rate": 2.1507986050439118e-05, "loss": 0.1439, "step": 17684 }, { "epoch": 109.84735202492212, "grad_norm": 1.1640625, "learning_rate": 2.1505546458250288e-05, "loss": 0.1029, "step": 17685 }, { "epoch": 109.85358255451713, "grad_norm": 2.953125, "learning_rate": 2.150310690000028e-05, "loss": 0.299, "step": 17686 }, { "epoch": 109.85981308411215, "grad_norm": 1.5078125, "learning_rate": 2.1500667375712792e-05, "loss": 0.097, "step": 17687 }, { "epoch": 109.86604361370716, "grad_norm": 3.109375, "learning_rate": 2.1498227885411526e-05, "loss": 0.291, "step": 17688 }, { "epoch": 109.87227414330218, "grad_norm": 2.015625, "learning_rate": 2.149578842912017e-05, "loss": 0.1373, "step": 17689 }, { "epoch": 109.8785046728972, "grad_norm": 1.78125, "learning_rate": 2.1493349006862417e-05, "loss": 0.1365, "step": 17690 }, { "epoch": 109.88473520249221, "grad_norm": 1.3671875, "learning_rate": 2.1490909618661964e-05, "loss": 0.1168, "step": 17691 }, { "epoch": 109.89096573208722, "grad_norm": 3.078125, "learning_rate": 2.14884702645425e-05, "loss": 0.363, "step": 17692 }, { "epoch": 109.89719626168224, "grad_norm": 2.359375, "learning_rate": 2.1486030944527713e-05, "loss": 0.1451, "step": 17693 }, { "epoch": 109.90342679127725, "grad_norm": 4.3125, "learning_rate": 2.1483591658641287e-05, "loss": 0.4661, "step": 17694 }, { "epoch": 109.90965732087227, "grad_norm": 2.171875, "learning_rate": 2.148115240690693e-05, "loss": 0.1549, "step": 17695 }, { "epoch": 109.91588785046729, "grad_norm": 2.484375, "learning_rate": 2.1478713189348314e-05, "loss": 0.1425, "step": 17696 }, { "epoch": 109.9221183800623, "grad_norm": 0.91015625, "learning_rate": 2.1476274005989148e-05, "loss": 0.0806, "step": 17697 }, { "epoch": 109.92834890965732, "grad_norm": 2.859375, "learning_rate": 2.1473834856853123e-05, "loss": 0.2895, "step": 17698 }, { "epoch": 109.93457943925233, "grad_norm": 1.34375, "learning_rate": 2.147139574196391e-05, "loss": 0.1139, "step": 17699 }, { "epoch": 109.94080996884735, "grad_norm": 2.75, "learning_rate": 2.14689566613452e-05, "loss": 0.2879, "step": 17700 }, { "epoch": 109.94704049844236, "grad_norm": 2.65625, "learning_rate": 2.1466517615020694e-05, "loss": 0.2423, "step": 17701 }, { "epoch": 109.95327102803738, "grad_norm": 2.140625, "learning_rate": 2.1464078603014075e-05, "loss": 0.1133, "step": 17702 }, { "epoch": 109.9595015576324, "grad_norm": 3.265625, "learning_rate": 2.1461639625349027e-05, "loss": 0.1301, "step": 17703 }, { "epoch": 109.96573208722741, "grad_norm": 2.421875, "learning_rate": 2.1459200682049255e-05, "loss": 0.1293, "step": 17704 }, { "epoch": 109.97196261682242, "grad_norm": 2.734375, "learning_rate": 2.1456761773138427e-05, "loss": 0.1729, "step": 17705 }, { "epoch": 109.97819314641744, "grad_norm": 2.515625, "learning_rate": 2.1454322898640234e-05, "loss": 0.2131, "step": 17706 }, { "epoch": 109.98442367601245, "grad_norm": 1.5234375, "learning_rate": 2.1451884058578366e-05, "loss": 0.1324, "step": 17707 }, { "epoch": 109.99065420560747, "grad_norm": 2.828125, "learning_rate": 2.144944525297651e-05, "loss": 0.1178, "step": 17708 }, { "epoch": 109.99688473520249, "grad_norm": 2.546875, "learning_rate": 2.144700648185835e-05, "loss": 0.2071, "step": 17709 }, { "epoch": 110.0, "grad_norm": 2.46875, "learning_rate": 2.1444567745247577e-05, "loss": 0.1241, "step": 17710 }, { "epoch": 110.00623052959502, "grad_norm": 2.453125, "learning_rate": 2.1442129043167874e-05, "loss": 0.2234, "step": 17711 }, { "epoch": 110.01246105919003, "grad_norm": 1.7890625, "learning_rate": 2.1439690375642927e-05, "loss": 0.1131, "step": 17712 }, { "epoch": 110.01869158878505, "grad_norm": 2.140625, "learning_rate": 2.1437251742696406e-05, "loss": 0.1136, "step": 17713 }, { "epoch": 110.02492211838006, "grad_norm": 2.125, "learning_rate": 2.1434813144352018e-05, "loss": 0.1856, "step": 17714 }, { "epoch": 110.03115264797508, "grad_norm": 3.15625, "learning_rate": 2.143237458063344e-05, "loss": 0.2767, "step": 17715 }, { "epoch": 110.03738317757009, "grad_norm": 2.96875, "learning_rate": 2.142993605156434e-05, "loss": 0.2112, "step": 17716 }, { "epoch": 110.04361370716511, "grad_norm": 2.609375, "learning_rate": 2.1427497557168433e-05, "loss": 0.2167, "step": 17717 }, { "epoch": 110.04984423676012, "grad_norm": 3.46875, "learning_rate": 2.1425059097469376e-05, "loss": 0.1727, "step": 17718 }, { "epoch": 110.05607476635514, "grad_norm": 1.9765625, "learning_rate": 2.1422620672490857e-05, "loss": 0.1086, "step": 17719 }, { "epoch": 110.06230529595015, "grad_norm": 1.390625, "learning_rate": 2.1420182282256558e-05, "loss": 0.1128, "step": 17720 }, { "epoch": 110.06853582554517, "grad_norm": 1.2734375, "learning_rate": 2.1417743926790167e-05, "loss": 0.09, "step": 17721 }, { "epoch": 110.07476635514018, "grad_norm": 1.0546875, "learning_rate": 2.1415305606115362e-05, "loss": 0.1097, "step": 17722 }, { "epoch": 110.0809968847352, "grad_norm": 1.890625, "learning_rate": 2.141286732025584e-05, "loss": 0.159, "step": 17723 }, { "epoch": 110.08722741433021, "grad_norm": 1.8984375, "learning_rate": 2.1410429069235254e-05, "loss": 0.0904, "step": 17724 }, { "epoch": 110.09345794392523, "grad_norm": 2.140625, "learning_rate": 2.1407990853077298e-05, "loss": 0.1285, "step": 17725 }, { "epoch": 110.09968847352025, "grad_norm": 1.8046875, "learning_rate": 2.140555267180565e-05, "loss": 0.1403, "step": 17726 }, { "epoch": 110.10591900311526, "grad_norm": 1.78125, "learning_rate": 2.1403114525444e-05, "loss": 0.1624, "step": 17727 }, { "epoch": 110.11214953271028, "grad_norm": 0.9765625, "learning_rate": 2.1400676414016018e-05, "loss": 0.0956, "step": 17728 }, { "epoch": 110.11838006230529, "grad_norm": 3.078125, "learning_rate": 2.1398238337545386e-05, "loss": 0.2315, "step": 17729 }, { "epoch": 110.1246105919003, "grad_norm": 2.765625, "learning_rate": 2.139580029605579e-05, "loss": 0.1751, "step": 17730 }, { "epoch": 110.13084112149532, "grad_norm": 2.890625, "learning_rate": 2.1393362289570895e-05, "loss": 0.2032, "step": 17731 }, { "epoch": 110.13707165109034, "grad_norm": 2.4375, "learning_rate": 2.139092431811439e-05, "loss": 0.1841, "step": 17732 }, { "epoch": 110.14330218068535, "grad_norm": 2.453125, "learning_rate": 2.138848638170994e-05, "loss": 0.1504, "step": 17733 }, { "epoch": 110.14953271028037, "grad_norm": 1.96875, "learning_rate": 2.138604848038124e-05, "loss": 0.1392, "step": 17734 }, { "epoch": 110.15576323987538, "grad_norm": 1.5390625, "learning_rate": 2.1383610614151962e-05, "loss": 0.1189, "step": 17735 }, { "epoch": 110.1619937694704, "grad_norm": 1.828125, "learning_rate": 2.1381172783045783e-05, "loss": 0.1154, "step": 17736 }, { "epoch": 110.16822429906541, "grad_norm": 1.65625, "learning_rate": 2.1378734987086364e-05, "loss": 0.103, "step": 17737 }, { "epoch": 110.17445482866043, "grad_norm": 1.921875, "learning_rate": 2.1376297226297407e-05, "loss": 0.1455, "step": 17738 }, { "epoch": 110.18068535825545, "grad_norm": 1.40625, "learning_rate": 2.137385950070257e-05, "loss": 0.1316, "step": 17739 }, { "epoch": 110.18691588785046, "grad_norm": 2.484375, "learning_rate": 2.1371421810325532e-05, "loss": 0.2328, "step": 17740 }, { "epoch": 110.19314641744548, "grad_norm": 2.546875, "learning_rate": 2.1368984155189976e-05, "loss": 0.0923, "step": 17741 }, { "epoch": 110.19937694704049, "grad_norm": 1.4296875, "learning_rate": 2.1366546535319576e-05, "loss": 0.1323, "step": 17742 }, { "epoch": 110.20560747663552, "grad_norm": 3.234375, "learning_rate": 2.1364108950737996e-05, "loss": 0.216, "step": 17743 }, { "epoch": 110.21183800623054, "grad_norm": 3.515625, "learning_rate": 2.1361671401468914e-05, "loss": 0.1645, "step": 17744 }, { "epoch": 110.21806853582555, "grad_norm": 3.734375, "learning_rate": 2.135923388753601e-05, "loss": 0.2969, "step": 17745 }, { "epoch": 110.22429906542057, "grad_norm": 3.34375, "learning_rate": 2.1356796408962948e-05, "loss": 0.3155, "step": 17746 }, { "epoch": 110.23052959501558, "grad_norm": 1.890625, "learning_rate": 2.1354358965773412e-05, "loss": 0.1172, "step": 17747 }, { "epoch": 110.2367601246106, "grad_norm": 2.171875, "learning_rate": 2.1351921557991074e-05, "loss": 0.1312, "step": 17748 }, { "epoch": 110.24299065420561, "grad_norm": 2.03125, "learning_rate": 2.1349484185639606e-05, "loss": 0.1737, "step": 17749 }, { "epoch": 110.24922118380063, "grad_norm": 2.796875, "learning_rate": 2.134704684874267e-05, "loss": 0.2675, "step": 17750 }, { "epoch": 110.25545171339564, "grad_norm": 1.4140625, "learning_rate": 2.134460954732395e-05, "loss": 0.0818, "step": 17751 }, { "epoch": 110.26168224299066, "grad_norm": 1.46875, "learning_rate": 2.1342172281407107e-05, "loss": 0.1303, "step": 17752 }, { "epoch": 110.26791277258567, "grad_norm": 2.28125, "learning_rate": 2.133973505101582e-05, "loss": 0.2822, "step": 17753 }, { "epoch": 110.27414330218069, "grad_norm": 2.828125, "learning_rate": 2.133729785617376e-05, "loss": 0.3018, "step": 17754 }, { "epoch": 110.2803738317757, "grad_norm": 1.796875, "learning_rate": 2.13348606969046e-05, "loss": 0.1116, "step": 17755 }, { "epoch": 110.28660436137072, "grad_norm": 1.1484375, "learning_rate": 2.1332423573232e-05, "loss": 0.1029, "step": 17756 }, { "epoch": 110.29283489096574, "grad_norm": 2.09375, "learning_rate": 2.132998648517963e-05, "loss": 0.1351, "step": 17757 }, { "epoch": 110.29906542056075, "grad_norm": 2.78125, "learning_rate": 2.1327549432771172e-05, "loss": 0.2228, "step": 17758 }, { "epoch": 110.30529595015577, "grad_norm": 1.234375, "learning_rate": 2.132511241603028e-05, "loss": 0.1185, "step": 17759 }, { "epoch": 110.31152647975078, "grad_norm": 3.09375, "learning_rate": 2.132267543498064e-05, "loss": 0.2079, "step": 17760 }, { "epoch": 110.3177570093458, "grad_norm": 1.59375, "learning_rate": 2.132023848964591e-05, "loss": 0.1455, "step": 17761 }, { "epoch": 110.32398753894081, "grad_norm": 2.578125, "learning_rate": 2.1317801580049762e-05, "loss": 0.3128, "step": 17762 }, { "epoch": 110.33021806853583, "grad_norm": 2.265625, "learning_rate": 2.1315364706215855e-05, "loss": 0.1746, "step": 17763 }, { "epoch": 110.33644859813084, "grad_norm": 1.9921875, "learning_rate": 2.1312927868167864e-05, "loss": 0.1574, "step": 17764 }, { "epoch": 110.34267912772586, "grad_norm": 2.3125, "learning_rate": 2.1310491065929454e-05, "loss": 0.2516, "step": 17765 }, { "epoch": 110.34890965732087, "grad_norm": 3.59375, "learning_rate": 2.130805429952429e-05, "loss": 0.3417, "step": 17766 }, { "epoch": 110.35514018691589, "grad_norm": 2.59375, "learning_rate": 2.1305617568976044e-05, "loss": 0.2333, "step": 17767 }, { "epoch": 110.3613707165109, "grad_norm": 2.390625, "learning_rate": 2.1303180874308386e-05, "loss": 0.1397, "step": 17768 }, { "epoch": 110.36760124610592, "grad_norm": 1.328125, "learning_rate": 2.1300744215544967e-05, "loss": 0.0988, "step": 17769 }, { "epoch": 110.37383177570094, "grad_norm": 1.75, "learning_rate": 2.1298307592709453e-05, "loss": 0.1437, "step": 17770 }, { "epoch": 110.38006230529595, "grad_norm": 1.6328125, "learning_rate": 2.1295871005825525e-05, "loss": 0.1497, "step": 17771 }, { "epoch": 110.38629283489097, "grad_norm": 1.984375, "learning_rate": 2.129343445491683e-05, "loss": 0.1182, "step": 17772 }, { "epoch": 110.39252336448598, "grad_norm": 2.765625, "learning_rate": 2.1290997940007046e-05, "loss": 0.15, "step": 17773 }, { "epoch": 110.398753894081, "grad_norm": 2.53125, "learning_rate": 2.128856146111984e-05, "loss": 0.1918, "step": 17774 }, { "epoch": 110.40498442367601, "grad_norm": 3.4375, "learning_rate": 2.1286125018278857e-05, "loss": 0.1539, "step": 17775 }, { "epoch": 110.41121495327103, "grad_norm": 3.453125, "learning_rate": 2.1283688611507767e-05, "loss": 0.1965, "step": 17776 }, { "epoch": 110.41744548286604, "grad_norm": 2.109375, "learning_rate": 2.128125224083024e-05, "loss": 0.1197, "step": 17777 }, { "epoch": 110.42367601246106, "grad_norm": 1.390625, "learning_rate": 2.127881590626994e-05, "loss": 0.1129, "step": 17778 }, { "epoch": 110.42990654205607, "grad_norm": 1.859375, "learning_rate": 2.1276379607850512e-05, "loss": 0.1548, "step": 17779 }, { "epoch": 110.43613707165109, "grad_norm": 1.375, "learning_rate": 2.1273943345595637e-05, "loss": 0.1169, "step": 17780 }, { "epoch": 110.4423676012461, "grad_norm": 2.15625, "learning_rate": 2.1271507119528977e-05, "loss": 0.1738, "step": 17781 }, { "epoch": 110.44859813084112, "grad_norm": 3.28125, "learning_rate": 2.1269070929674175e-05, "loss": 0.2772, "step": 17782 }, { "epoch": 110.45482866043614, "grad_norm": 2.8125, "learning_rate": 2.12666347760549e-05, "loss": 0.1788, "step": 17783 }, { "epoch": 110.46105919003115, "grad_norm": 3.125, "learning_rate": 2.126419865869482e-05, "loss": 0.1564, "step": 17784 }, { "epoch": 110.46728971962617, "grad_norm": 2.578125, "learning_rate": 2.1261762577617587e-05, "loss": 0.1354, "step": 17785 }, { "epoch": 110.47352024922118, "grad_norm": 1.171875, "learning_rate": 2.125932653284686e-05, "loss": 0.0977, "step": 17786 }, { "epoch": 110.4797507788162, "grad_norm": 1.625, "learning_rate": 2.125689052440631e-05, "loss": 0.1104, "step": 17787 }, { "epoch": 110.48598130841121, "grad_norm": 1.8203125, "learning_rate": 2.1254454552319582e-05, "loss": 0.1377, "step": 17788 }, { "epoch": 110.49221183800623, "grad_norm": 1.4609375, "learning_rate": 2.125201861661034e-05, "loss": 0.1301, "step": 17789 }, { "epoch": 110.49844236760124, "grad_norm": 2.484375, "learning_rate": 2.124958271730224e-05, "loss": 0.1443, "step": 17790 }, { "epoch": 110.50467289719626, "grad_norm": 2.09375, "learning_rate": 2.1247146854418947e-05, "loss": 0.2165, "step": 17791 }, { "epoch": 110.51090342679127, "grad_norm": 1.1328125, "learning_rate": 2.124471102798411e-05, "loss": 0.106, "step": 17792 }, { "epoch": 110.51713395638629, "grad_norm": 1.7890625, "learning_rate": 2.1242275238021405e-05, "loss": 0.1321, "step": 17793 }, { "epoch": 110.5233644859813, "grad_norm": 1.7265625, "learning_rate": 2.1239839484554453e-05, "loss": 0.1454, "step": 17794 }, { "epoch": 110.52959501557632, "grad_norm": 1.109375, "learning_rate": 2.123740376760694e-05, "loss": 0.1087, "step": 17795 }, { "epoch": 110.53582554517133, "grad_norm": 2.671875, "learning_rate": 2.123496808720251e-05, "loss": 0.2123, "step": 17796 }, { "epoch": 110.54205607476635, "grad_norm": 2.359375, "learning_rate": 2.123253244336483e-05, "loss": 0.0916, "step": 17797 }, { "epoch": 110.54828660436137, "grad_norm": 3.171875, "learning_rate": 2.1230096836117543e-05, "loss": 0.1984, "step": 17798 }, { "epoch": 110.55451713395638, "grad_norm": 1.984375, "learning_rate": 2.1227661265484307e-05, "loss": 0.1146, "step": 17799 }, { "epoch": 110.5607476635514, "grad_norm": 1.3046875, "learning_rate": 2.1225225731488786e-05, "loss": 0.1141, "step": 17800 }, { "epoch": 110.56697819314641, "grad_norm": 2.4375, "learning_rate": 2.1222790234154627e-05, "loss": 0.2167, "step": 17801 }, { "epoch": 110.57320872274143, "grad_norm": 2.25, "learning_rate": 2.1220354773505478e-05, "loss": 0.1526, "step": 17802 }, { "epoch": 110.57943925233644, "grad_norm": 1.234375, "learning_rate": 2.1217919349565e-05, "loss": 0.122, "step": 17803 }, { "epoch": 110.58566978193146, "grad_norm": 3.453125, "learning_rate": 2.1215483962356846e-05, "loss": 0.2279, "step": 17804 }, { "epoch": 110.59190031152647, "grad_norm": 3.109375, "learning_rate": 2.1213048611904664e-05, "loss": 0.2157, "step": 17805 }, { "epoch": 110.59813084112149, "grad_norm": 1.6171875, "learning_rate": 2.121061329823213e-05, "loss": 0.115, "step": 17806 }, { "epoch": 110.60436137071652, "grad_norm": 1.453125, "learning_rate": 2.120817802136286e-05, "loss": 0.0774, "step": 17807 }, { "epoch": 110.61059190031153, "grad_norm": 3.390625, "learning_rate": 2.1205742781320525e-05, "loss": 0.2838, "step": 17808 }, { "epoch": 110.61682242990655, "grad_norm": 1.171875, "learning_rate": 2.1203307578128772e-05, "loss": 0.0946, "step": 17809 }, { "epoch": 110.62305295950156, "grad_norm": 2.234375, "learning_rate": 2.120087241181126e-05, "loss": 0.2426, "step": 17810 }, { "epoch": 110.62928348909658, "grad_norm": 1.6640625, "learning_rate": 2.1198437282391635e-05, "loss": 0.1427, "step": 17811 }, { "epoch": 110.6355140186916, "grad_norm": 3.09375, "learning_rate": 2.119600218989355e-05, "loss": 0.2876, "step": 17812 }, { "epoch": 110.64174454828661, "grad_norm": 2.890625, "learning_rate": 2.119356713434064e-05, "loss": 0.2714, "step": 17813 }, { "epoch": 110.64797507788163, "grad_norm": 1.828125, "learning_rate": 2.1191132115756578e-05, "loss": 0.1161, "step": 17814 }, { "epoch": 110.65420560747664, "grad_norm": 3.40625, "learning_rate": 2.1188697134165e-05, "loss": 0.1407, "step": 17815 }, { "epoch": 110.66043613707166, "grad_norm": 1.9140625, "learning_rate": 2.1186262189589552e-05, "loss": 0.1284, "step": 17816 }, { "epoch": 110.66666666666667, "grad_norm": 2.421875, "learning_rate": 2.1183827282053893e-05, "loss": 0.144, "step": 17817 }, { "epoch": 110.67289719626169, "grad_norm": 1.625, "learning_rate": 2.118139241158166e-05, "loss": 0.1231, "step": 17818 }, { "epoch": 110.6791277258567, "grad_norm": 2.015625, "learning_rate": 2.1178957578196522e-05, "loss": 0.2035, "step": 17819 }, { "epoch": 110.68535825545172, "grad_norm": 1.78125, "learning_rate": 2.1176522781922097e-05, "loss": 0.1368, "step": 17820 }, { "epoch": 110.69158878504673, "grad_norm": 2.875, "learning_rate": 2.1174088022782054e-05, "loss": 0.1962, "step": 17821 }, { "epoch": 110.69781931464175, "grad_norm": 2.75, "learning_rate": 2.1171653300800025e-05, "loss": 0.1038, "step": 17822 }, { "epoch": 110.70404984423676, "grad_norm": 2.015625, "learning_rate": 2.1169218615999676e-05, "loss": 0.2258, "step": 17823 }, { "epoch": 110.71028037383178, "grad_norm": 1.875, "learning_rate": 2.1166783968404636e-05, "loss": 0.1887, "step": 17824 }, { "epoch": 110.7165109034268, "grad_norm": 1.5625, "learning_rate": 2.116434935803856e-05, "loss": 0.1029, "step": 17825 }, { "epoch": 110.72274143302181, "grad_norm": 2.234375, "learning_rate": 2.1161914784925083e-05, "loss": 0.1986, "step": 17826 }, { "epoch": 110.72897196261682, "grad_norm": 1.515625, "learning_rate": 2.115948024908786e-05, "loss": 0.1094, "step": 17827 }, { "epoch": 110.73520249221184, "grad_norm": 2.296875, "learning_rate": 2.1157045750550532e-05, "loss": 0.1446, "step": 17828 }, { "epoch": 110.74143302180686, "grad_norm": 2.625, "learning_rate": 2.1154611289336744e-05, "loss": 0.2023, "step": 17829 }, { "epoch": 110.74766355140187, "grad_norm": 1.234375, "learning_rate": 2.115217686547014e-05, "loss": 0.1087, "step": 17830 }, { "epoch": 110.75389408099689, "grad_norm": 2.875, "learning_rate": 2.1149742478974363e-05, "loss": 0.2185, "step": 17831 }, { "epoch": 110.7601246105919, "grad_norm": 2.15625, "learning_rate": 2.1147308129873063e-05, "loss": 0.1663, "step": 17832 }, { "epoch": 110.76635514018692, "grad_norm": 1.1953125, "learning_rate": 2.1144873818189867e-05, "loss": 0.0937, "step": 17833 }, { "epoch": 110.77258566978193, "grad_norm": 2.21875, "learning_rate": 2.1142439543948434e-05, "loss": 0.1193, "step": 17834 }, { "epoch": 110.77881619937695, "grad_norm": 2.203125, "learning_rate": 2.1140005307172393e-05, "loss": 0.1979, "step": 17835 }, { "epoch": 110.78504672897196, "grad_norm": 2.5625, "learning_rate": 2.1137571107885394e-05, "loss": 0.1284, "step": 17836 }, { "epoch": 110.79127725856698, "grad_norm": 2.140625, "learning_rate": 2.1135136946111078e-05, "loss": 0.0938, "step": 17837 }, { "epoch": 110.797507788162, "grad_norm": 2.015625, "learning_rate": 2.113270282187309e-05, "loss": 0.149, "step": 17838 }, { "epoch": 110.80373831775701, "grad_norm": 1.8203125, "learning_rate": 2.113026873519506e-05, "loss": 0.1117, "step": 17839 }, { "epoch": 110.80996884735202, "grad_norm": 3.15625, "learning_rate": 2.112783468610063e-05, "loss": 0.2306, "step": 17840 }, { "epoch": 110.81619937694704, "grad_norm": 1.9296875, "learning_rate": 2.1125400674613447e-05, "loss": 0.1401, "step": 17841 }, { "epoch": 110.82242990654206, "grad_norm": 1.4296875, "learning_rate": 2.1122966700757145e-05, "loss": 0.1119, "step": 17842 }, { "epoch": 110.82866043613707, "grad_norm": 2.421875, "learning_rate": 2.1120532764555367e-05, "loss": 0.1528, "step": 17843 }, { "epoch": 110.83489096573209, "grad_norm": 1.78125, "learning_rate": 2.1118098866031755e-05, "loss": 0.1343, "step": 17844 }, { "epoch": 110.8411214953271, "grad_norm": 2.34375, "learning_rate": 2.111566500520994e-05, "loss": 0.1924, "step": 17845 }, { "epoch": 110.84735202492212, "grad_norm": 2.046875, "learning_rate": 2.111323118211356e-05, "loss": 0.2044, "step": 17846 }, { "epoch": 110.85358255451713, "grad_norm": 1.9921875, "learning_rate": 2.1110797396766255e-05, "loss": 0.2005, "step": 17847 }, { "epoch": 110.85981308411215, "grad_norm": 2.75, "learning_rate": 2.1108363649191663e-05, "loss": 0.1849, "step": 17848 }, { "epoch": 110.86604361370716, "grad_norm": 2.40625, "learning_rate": 2.110592993941342e-05, "loss": 0.2281, "step": 17849 }, { "epoch": 110.87227414330218, "grad_norm": 2.109375, "learning_rate": 2.110349626745517e-05, "loss": 0.2302, "step": 17850 }, { "epoch": 110.8785046728972, "grad_norm": 2.28125, "learning_rate": 2.1101062633340544e-05, "loss": 0.2115, "step": 17851 }, { "epoch": 110.88473520249221, "grad_norm": 1.734375, "learning_rate": 2.1098629037093176e-05, "loss": 0.1524, "step": 17852 }, { "epoch": 110.89096573208722, "grad_norm": 3.34375, "learning_rate": 2.1096195478736696e-05, "loss": 0.2042, "step": 17853 }, { "epoch": 110.89719626168224, "grad_norm": 3.3125, "learning_rate": 2.109376195829475e-05, "loss": 0.2037, "step": 17854 }, { "epoch": 110.90342679127725, "grad_norm": 1.9375, "learning_rate": 2.1091328475790965e-05, "loss": 0.2475, "step": 17855 }, { "epoch": 110.90965732087227, "grad_norm": 2.578125, "learning_rate": 2.1088895031248984e-05, "loss": 0.193, "step": 17856 }, { "epoch": 110.91588785046729, "grad_norm": 2.75, "learning_rate": 2.1086461624692442e-05, "loss": 0.1067, "step": 17857 }, { "epoch": 110.9221183800623, "grad_norm": 2.671875, "learning_rate": 2.1084028256144962e-05, "loss": 0.2022, "step": 17858 }, { "epoch": 110.92834890965732, "grad_norm": 1.5234375, "learning_rate": 2.1081594925630173e-05, "loss": 0.1281, "step": 17859 }, { "epoch": 110.93457943925233, "grad_norm": 1.8203125, "learning_rate": 2.1079161633171727e-05, "loss": 0.1534, "step": 17860 }, { "epoch": 110.94080996884735, "grad_norm": 2.6875, "learning_rate": 2.1076728378793247e-05, "loss": 0.185, "step": 17861 }, { "epoch": 110.94704049844236, "grad_norm": 1.9453125, "learning_rate": 2.1074295162518357e-05, "loss": 0.1455, "step": 17862 }, { "epoch": 110.95327102803738, "grad_norm": 1.125, "learning_rate": 2.1071861984370707e-05, "loss": 0.0992, "step": 17863 }, { "epoch": 110.9595015576324, "grad_norm": 1.484375, "learning_rate": 2.1069428844373916e-05, "loss": 0.0988, "step": 17864 }, { "epoch": 110.96573208722741, "grad_norm": 1.2109375, "learning_rate": 2.106699574255162e-05, "loss": 0.0841, "step": 17865 }, { "epoch": 110.97196261682242, "grad_norm": 1.7890625, "learning_rate": 2.106456267892744e-05, "loss": 0.1589, "step": 17866 }, { "epoch": 110.97819314641744, "grad_norm": 2.296875, "learning_rate": 2.1062129653525017e-05, "loss": 0.1915, "step": 17867 }, { "epoch": 110.98442367601245, "grad_norm": 2.265625, "learning_rate": 2.1059696666367972e-05, "loss": 0.205, "step": 17868 }, { "epoch": 110.99065420560747, "grad_norm": 1.2734375, "learning_rate": 2.1057263717479954e-05, "loss": 0.1136, "step": 17869 }, { "epoch": 110.99688473520249, "grad_norm": 1.7734375, "learning_rate": 2.1054830806884577e-05, "loss": 0.1479, "step": 17870 }, { "epoch": 111.0, "grad_norm": 0.92578125, "learning_rate": 2.1052397934605468e-05, "loss": 0.0405, "step": 17871 }, { "epoch": 111.00623052959502, "grad_norm": 2.421875, "learning_rate": 2.104996510066625e-05, "loss": 0.1427, "step": 17872 }, { "epoch": 111.01246105919003, "grad_norm": 1.984375, "learning_rate": 2.1047532305090574e-05, "loss": 0.1285, "step": 17873 }, { "epoch": 111.01869158878505, "grad_norm": 2.015625, "learning_rate": 2.104509954790205e-05, "loss": 0.1251, "step": 17874 }, { "epoch": 111.02492211838006, "grad_norm": 2.9375, "learning_rate": 2.1042666829124304e-05, "loss": 0.2471, "step": 17875 }, { "epoch": 111.03115264797508, "grad_norm": 2.3125, "learning_rate": 2.104023414878098e-05, "loss": 0.1891, "step": 17876 }, { "epoch": 111.03738317757009, "grad_norm": 1.8046875, "learning_rate": 2.103780150689569e-05, "loss": 0.1635, "step": 17877 }, { "epoch": 111.04361370716511, "grad_norm": 1.03125, "learning_rate": 2.103536890349206e-05, "loss": 0.1089, "step": 17878 }, { "epoch": 111.04984423676012, "grad_norm": 1.4609375, "learning_rate": 2.1032936338593718e-05, "loss": 0.0942, "step": 17879 }, { "epoch": 111.05607476635514, "grad_norm": 2.796875, "learning_rate": 2.1030503812224294e-05, "loss": 0.2126, "step": 17880 }, { "epoch": 111.06230529595015, "grad_norm": 1.9453125, "learning_rate": 2.1028071324407415e-05, "loss": 0.1465, "step": 17881 }, { "epoch": 111.06853582554517, "grad_norm": 2.140625, "learning_rate": 2.1025638875166704e-05, "loss": 0.1558, "step": 17882 }, { "epoch": 111.07476635514018, "grad_norm": 2.03125, "learning_rate": 2.102320646452577e-05, "loss": 0.158, "step": 17883 }, { "epoch": 111.0809968847352, "grad_norm": 2.0, "learning_rate": 2.1020774092508255e-05, "loss": 0.1556, "step": 17884 }, { "epoch": 111.08722741433021, "grad_norm": 3.03125, "learning_rate": 2.101834175913778e-05, "loss": 0.2398, "step": 17885 }, { "epoch": 111.09345794392523, "grad_norm": 2.109375, "learning_rate": 2.1015909464437958e-05, "loss": 0.2253, "step": 17886 }, { "epoch": 111.09968847352025, "grad_norm": 2.75, "learning_rate": 2.1013477208432428e-05, "loss": 0.1396, "step": 17887 }, { "epoch": 111.10591900311526, "grad_norm": 2.609375, "learning_rate": 2.1011044991144795e-05, "loss": 0.1197, "step": 17888 }, { "epoch": 111.11214953271028, "grad_norm": 1.46875, "learning_rate": 2.100861281259871e-05, "loss": 0.1407, "step": 17889 }, { "epoch": 111.11838006230529, "grad_norm": 2.65625, "learning_rate": 2.1006180672817756e-05, "loss": 0.251, "step": 17890 }, { "epoch": 111.1246105919003, "grad_norm": 2.296875, "learning_rate": 2.100374857182558e-05, "loss": 0.1727, "step": 17891 }, { "epoch": 111.13084112149532, "grad_norm": 2.34375, "learning_rate": 2.1001316509645798e-05, "loss": 0.1154, "step": 17892 }, { "epoch": 111.13707165109034, "grad_norm": 1.9765625, "learning_rate": 2.0998884486302024e-05, "loss": 0.1596, "step": 17893 }, { "epoch": 111.14330218068535, "grad_norm": 2.21875, "learning_rate": 2.099645250181789e-05, "loss": 0.22, "step": 17894 }, { "epoch": 111.14953271028037, "grad_norm": 1.765625, "learning_rate": 2.0994020556217013e-05, "loss": 0.1316, "step": 17895 }, { "epoch": 111.15576323987538, "grad_norm": 1.5859375, "learning_rate": 2.0991588649523002e-05, "loss": 0.1127, "step": 17896 }, { "epoch": 111.1619937694704, "grad_norm": 1.59375, "learning_rate": 2.0989156781759482e-05, "loss": 0.1338, "step": 17897 }, { "epoch": 111.16822429906541, "grad_norm": 1.3359375, "learning_rate": 2.098672495295008e-05, "loss": 0.1078, "step": 17898 }, { "epoch": 111.17445482866043, "grad_norm": 2.84375, "learning_rate": 2.09842931631184e-05, "loss": 0.2019, "step": 17899 }, { "epoch": 111.18068535825545, "grad_norm": 1.171875, "learning_rate": 2.098186141228807e-05, "loss": 0.1254, "step": 17900 }, { "epoch": 111.18691588785046, "grad_norm": 0.8984375, "learning_rate": 2.09794297004827e-05, "loss": 0.1118, "step": 17901 }, { "epoch": 111.19314641744548, "grad_norm": 2.515625, "learning_rate": 2.097699802772593e-05, "loss": 0.2269, "step": 17902 }, { "epoch": 111.19937694704049, "grad_norm": 2.703125, "learning_rate": 2.0974566394041344e-05, "loss": 0.1739, "step": 17903 }, { "epoch": 111.20560747663552, "grad_norm": 0.72265625, "learning_rate": 2.097213479945258e-05, "loss": 0.0837, "step": 17904 }, { "epoch": 111.21183800623054, "grad_norm": 3.234375, "learning_rate": 2.096970324398324e-05, "loss": 0.1251, "step": 17905 }, { "epoch": 111.21806853582555, "grad_norm": 1.7421875, "learning_rate": 2.0967271727656958e-05, "loss": 0.2288, "step": 17906 }, { "epoch": 111.22429906542057, "grad_norm": 3.46875, "learning_rate": 2.096484025049733e-05, "loss": 0.1853, "step": 17907 }, { "epoch": 111.23052959501558, "grad_norm": 1.7421875, "learning_rate": 2.0962408812527996e-05, "loss": 0.1594, "step": 17908 }, { "epoch": 111.2367601246106, "grad_norm": 1.796875, "learning_rate": 2.0959977413772535e-05, "loss": 0.1532, "step": 17909 }, { "epoch": 111.24299065420561, "grad_norm": 2.015625, "learning_rate": 2.095754605425459e-05, "loss": 0.1513, "step": 17910 }, { "epoch": 111.24922118380063, "grad_norm": 1.390625, "learning_rate": 2.0955114733997766e-05, "loss": 0.1092, "step": 17911 }, { "epoch": 111.25545171339564, "grad_norm": 1.5859375, "learning_rate": 2.0952683453025672e-05, "loss": 0.118, "step": 17912 }, { "epoch": 111.26168224299066, "grad_norm": 1.8046875, "learning_rate": 2.0950252211361927e-05, "loss": 0.1134, "step": 17913 }, { "epoch": 111.26791277258567, "grad_norm": 2.546875, "learning_rate": 2.094782100903015e-05, "loss": 0.1294, "step": 17914 }, { "epoch": 111.27414330218069, "grad_norm": 2.15625, "learning_rate": 2.094538984605394e-05, "loss": 0.148, "step": 17915 }, { "epoch": 111.2803738317757, "grad_norm": 1.1328125, "learning_rate": 2.0942958722456912e-05, "loss": 0.0895, "step": 17916 }, { "epoch": 111.28660436137072, "grad_norm": 3.109375, "learning_rate": 2.094052763826268e-05, "loss": 0.2258, "step": 17917 }, { "epoch": 111.29283489096574, "grad_norm": 2.578125, "learning_rate": 2.0938096593494855e-05, "loss": 0.1195, "step": 17918 }, { "epoch": 111.29906542056075, "grad_norm": 2.671875, "learning_rate": 2.093566558817705e-05, "loss": 0.2652, "step": 17919 }, { "epoch": 111.30529595015577, "grad_norm": 1.9765625, "learning_rate": 2.093323462233287e-05, "loss": 0.158, "step": 17920 }, { "epoch": 111.31152647975078, "grad_norm": 1.828125, "learning_rate": 2.093080369598594e-05, "loss": 0.1252, "step": 17921 }, { "epoch": 111.3177570093458, "grad_norm": 1.7734375, "learning_rate": 2.0928372809159842e-05, "loss": 0.1105, "step": 17922 }, { "epoch": 111.32398753894081, "grad_norm": 2.59375, "learning_rate": 2.092594196187821e-05, "loss": 0.2095, "step": 17923 }, { "epoch": 111.33021806853583, "grad_norm": 2.6875, "learning_rate": 2.0923511154164642e-05, "loss": 0.1328, "step": 17924 }, { "epoch": 111.33644859813084, "grad_norm": 2.203125, "learning_rate": 2.0921080386042747e-05, "loss": 0.1539, "step": 17925 }, { "epoch": 111.34267912772586, "grad_norm": 4.0625, "learning_rate": 2.0918649657536136e-05, "loss": 0.1136, "step": 17926 }, { "epoch": 111.34890965732087, "grad_norm": 2.25, "learning_rate": 2.091621896866842e-05, "loss": 0.2293, "step": 17927 }, { "epoch": 111.35514018691589, "grad_norm": 2.25, "learning_rate": 2.09137883194632e-05, "loss": 0.1557, "step": 17928 }, { "epoch": 111.3613707165109, "grad_norm": 1.859375, "learning_rate": 2.0911357709944077e-05, "loss": 0.2001, "step": 17929 }, { "epoch": 111.36760124610592, "grad_norm": 1.390625, "learning_rate": 2.0908927140134672e-05, "loss": 0.1159, "step": 17930 }, { "epoch": 111.37383177570094, "grad_norm": 2.171875, "learning_rate": 2.0906496610058583e-05, "loss": 0.1548, "step": 17931 }, { "epoch": 111.38006230529595, "grad_norm": 1.9609375, "learning_rate": 2.090406611973941e-05, "loss": 0.2132, "step": 17932 }, { "epoch": 111.38629283489097, "grad_norm": 1.046875, "learning_rate": 2.0901635669200778e-05, "loss": 0.0865, "step": 17933 }, { "epoch": 111.39252336448598, "grad_norm": 1.890625, "learning_rate": 2.0899205258466275e-05, "loss": 0.1281, "step": 17934 }, { "epoch": 111.398753894081, "grad_norm": 2.546875, "learning_rate": 2.089677488755951e-05, "loss": 0.1505, "step": 17935 }, { "epoch": 111.40498442367601, "grad_norm": 2.40625, "learning_rate": 2.0894344556504083e-05, "loss": 0.162, "step": 17936 }, { "epoch": 111.41121495327103, "grad_norm": 1.8828125, "learning_rate": 2.0891914265323605e-05, "loss": 0.0858, "step": 17937 }, { "epoch": 111.41744548286604, "grad_norm": 1.8515625, "learning_rate": 2.0889484014041675e-05, "loss": 0.152, "step": 17938 }, { "epoch": 111.42367601246106, "grad_norm": 3.0625, "learning_rate": 2.0887053802681902e-05, "loss": 0.2246, "step": 17939 }, { "epoch": 111.42990654205607, "grad_norm": 2.703125, "learning_rate": 2.0884623631267892e-05, "loss": 0.3049, "step": 17940 }, { "epoch": 111.43613707165109, "grad_norm": 1.5390625, "learning_rate": 2.088219349982323e-05, "loss": 0.1958, "step": 17941 }, { "epoch": 111.4423676012461, "grad_norm": 1.484375, "learning_rate": 2.0879763408371526e-05, "loss": 0.1135, "step": 17942 }, { "epoch": 111.44859813084112, "grad_norm": 2.265625, "learning_rate": 2.0877333356936385e-05, "loss": 0.2067, "step": 17943 }, { "epoch": 111.45482866043614, "grad_norm": 1.1171875, "learning_rate": 2.087490334554141e-05, "loss": 0.0846, "step": 17944 }, { "epoch": 111.46105919003115, "grad_norm": 3.546875, "learning_rate": 2.087247337421019e-05, "loss": 0.1707, "step": 17945 }, { "epoch": 111.46728971962617, "grad_norm": 2.734375, "learning_rate": 2.087004344296635e-05, "loss": 0.2154, "step": 17946 }, { "epoch": 111.47352024922118, "grad_norm": 2.28125, "learning_rate": 2.0867613551833463e-05, "loss": 0.1526, "step": 17947 }, { "epoch": 111.4797507788162, "grad_norm": 1.8125, "learning_rate": 2.0865183700835143e-05, "loss": 0.1534, "step": 17948 }, { "epoch": 111.48598130841121, "grad_norm": 1.84375, "learning_rate": 2.0862753889994978e-05, "loss": 0.1227, "step": 17949 }, { "epoch": 111.49221183800623, "grad_norm": 0.97265625, "learning_rate": 2.0860324119336578e-05, "loss": 0.1069, "step": 17950 }, { "epoch": 111.49844236760124, "grad_norm": 1.875, "learning_rate": 2.0857894388883535e-05, "loss": 0.1552, "step": 17951 }, { "epoch": 111.50467289719626, "grad_norm": 2.203125, "learning_rate": 2.085546469865945e-05, "loss": 0.2963, "step": 17952 }, { "epoch": 111.51090342679127, "grad_norm": 1.2578125, "learning_rate": 2.0853035048687932e-05, "loss": 0.0831, "step": 17953 }, { "epoch": 111.51713395638629, "grad_norm": 2.25, "learning_rate": 2.085060543899256e-05, "loss": 0.1448, "step": 17954 }, { "epoch": 111.5233644859813, "grad_norm": 2.3125, "learning_rate": 2.0848175869596932e-05, "loss": 0.1757, "step": 17955 }, { "epoch": 111.52959501557632, "grad_norm": 2.78125, "learning_rate": 2.084574634052465e-05, "loss": 0.261, "step": 17956 }, { "epoch": 111.53582554517133, "grad_norm": 1.8828125, "learning_rate": 2.0843316851799315e-05, "loss": 0.1281, "step": 17957 }, { "epoch": 111.54205607476635, "grad_norm": 1.9453125, "learning_rate": 2.084088740344451e-05, "loss": 0.1183, "step": 17958 }, { "epoch": 111.54828660436137, "grad_norm": 2.359375, "learning_rate": 2.083845799548385e-05, "loss": 0.1927, "step": 17959 }, { "epoch": 111.55451713395638, "grad_norm": 1.8671875, "learning_rate": 2.0836028627940915e-05, "loss": 0.1636, "step": 17960 }, { "epoch": 111.5607476635514, "grad_norm": 2.59375, "learning_rate": 2.08335993008393e-05, "loss": 0.1213, "step": 17961 }, { "epoch": 111.56697819314641, "grad_norm": 1.75, "learning_rate": 2.0831170014202594e-05, "loss": 0.1303, "step": 17962 }, { "epoch": 111.57320872274143, "grad_norm": 2.421875, "learning_rate": 2.0828740768054405e-05, "loss": 0.1357, "step": 17963 }, { "epoch": 111.57943925233644, "grad_norm": 2.453125, "learning_rate": 2.0826311562418314e-05, "loss": 0.148, "step": 17964 }, { "epoch": 111.58566978193146, "grad_norm": 2.609375, "learning_rate": 2.082388239731794e-05, "loss": 0.2229, "step": 17965 }, { "epoch": 111.59190031152647, "grad_norm": 1.6953125, "learning_rate": 2.0821453272776832e-05, "loss": 0.146, "step": 17966 }, { "epoch": 111.59813084112149, "grad_norm": 1.1484375, "learning_rate": 2.0819024188818613e-05, "loss": 0.1195, "step": 17967 }, { "epoch": 111.60436137071652, "grad_norm": 1.015625, "learning_rate": 2.0816595145466863e-05, "loss": 0.12, "step": 17968 }, { "epoch": 111.61059190031153, "grad_norm": 2.875, "learning_rate": 2.0814166142745182e-05, "loss": 0.2543, "step": 17969 }, { "epoch": 111.61682242990655, "grad_norm": 1.9140625, "learning_rate": 2.0811737180677153e-05, "loss": 0.1515, "step": 17970 }, { "epoch": 111.62305295950156, "grad_norm": 1.6953125, "learning_rate": 2.080930825928637e-05, "loss": 0.1078, "step": 17971 }, { "epoch": 111.62928348909658, "grad_norm": 2.484375, "learning_rate": 2.0806879378596427e-05, "loss": 0.2021, "step": 17972 }, { "epoch": 111.6355140186916, "grad_norm": 1.9140625, "learning_rate": 2.080445053863091e-05, "loss": 0.2364, "step": 17973 }, { "epoch": 111.64174454828661, "grad_norm": 2.0625, "learning_rate": 2.0802021739413406e-05, "loss": 0.1636, "step": 17974 }, { "epoch": 111.64797507788163, "grad_norm": 3.484375, "learning_rate": 2.07995929809675e-05, "loss": 0.318, "step": 17975 }, { "epoch": 111.65420560747664, "grad_norm": 1.6484375, "learning_rate": 2.0797164263316793e-05, "loss": 0.1236, "step": 17976 }, { "epoch": 111.66043613707166, "grad_norm": 1.3671875, "learning_rate": 2.0794735586484866e-05, "loss": 0.1373, "step": 17977 }, { "epoch": 111.66666666666667, "grad_norm": 2.53125, "learning_rate": 2.0792306950495312e-05, "loss": 0.2278, "step": 17978 }, { "epoch": 111.67289719626169, "grad_norm": 1.1796875, "learning_rate": 2.0789878355371706e-05, "loss": 0.1288, "step": 17979 }, { "epoch": 111.6791277258567, "grad_norm": 1.6328125, "learning_rate": 2.0787449801137647e-05, "loss": 0.1077, "step": 17980 }, { "epoch": 111.68535825545172, "grad_norm": 1.7421875, "learning_rate": 2.0785021287816716e-05, "loss": 0.1707, "step": 17981 }, { "epoch": 111.69158878504673, "grad_norm": 1.0234375, "learning_rate": 2.0782592815432497e-05, "loss": 0.1073, "step": 17982 }, { "epoch": 111.69781931464175, "grad_norm": 1.3203125, "learning_rate": 2.0780164384008586e-05, "loss": 0.1127, "step": 17983 }, { "epoch": 111.70404984423676, "grad_norm": 2.046875, "learning_rate": 2.0777735993568565e-05, "loss": 0.1037, "step": 17984 }, { "epoch": 111.71028037383178, "grad_norm": 1.171875, "learning_rate": 2.0775307644136013e-05, "loss": 0.1076, "step": 17985 }, { "epoch": 111.7165109034268, "grad_norm": 1.9453125, "learning_rate": 2.077287933573451e-05, "loss": 0.1547, "step": 17986 }, { "epoch": 111.72274143302181, "grad_norm": 2.796875, "learning_rate": 2.0770451068387654e-05, "loss": 0.1849, "step": 17987 }, { "epoch": 111.72897196261682, "grad_norm": 1.8984375, "learning_rate": 2.076802284211902e-05, "loss": 0.0831, "step": 17988 }, { "epoch": 111.73520249221184, "grad_norm": 2.015625, "learning_rate": 2.0765594656952198e-05, "loss": 0.1382, "step": 17989 }, { "epoch": 111.74143302180686, "grad_norm": 1.46875, "learning_rate": 2.0763166512910764e-05, "loss": 0.1388, "step": 17990 }, { "epoch": 111.74766355140187, "grad_norm": 1.6953125, "learning_rate": 2.0760738410018312e-05, "loss": 0.129, "step": 17991 }, { "epoch": 111.75389408099689, "grad_norm": 2.765625, "learning_rate": 2.0758310348298404e-05, "loss": 0.1498, "step": 17992 }, { "epoch": 111.7601246105919, "grad_norm": 1.9921875, "learning_rate": 2.0755882327774642e-05, "loss": 0.1475, "step": 17993 }, { "epoch": 111.76635514018692, "grad_norm": 2.703125, "learning_rate": 2.0753454348470596e-05, "loss": 0.2372, "step": 17994 }, { "epoch": 111.77258566978193, "grad_norm": 1.6484375, "learning_rate": 2.0751026410409848e-05, "loss": 0.1094, "step": 17995 }, { "epoch": 111.77881619937695, "grad_norm": 2.75, "learning_rate": 2.0748598513615983e-05, "loss": 0.2074, "step": 17996 }, { "epoch": 111.78504672897196, "grad_norm": 2.453125, "learning_rate": 2.0746170658112585e-05, "loss": 0.1001, "step": 17997 }, { "epoch": 111.79127725856698, "grad_norm": 1.828125, "learning_rate": 2.0743742843923226e-05, "loss": 0.1494, "step": 17998 }, { "epoch": 111.797507788162, "grad_norm": 1.796875, "learning_rate": 2.0741315071071477e-05, "loss": 0.187, "step": 17999 }, { "epoch": 111.80373831775701, "grad_norm": 2.5625, "learning_rate": 2.0738887339580933e-05, "loss": 0.0979, "step": 18000 }, { "epoch": 111.80996884735202, "grad_norm": 2.9375, "learning_rate": 2.0736459649475165e-05, "loss": 0.2135, "step": 18001 }, { "epoch": 111.81619937694704, "grad_norm": 2.265625, "learning_rate": 2.0734032000777755e-05, "loss": 0.1568, "step": 18002 }, { "epoch": 111.82242990654206, "grad_norm": 2.328125, "learning_rate": 2.0731604393512287e-05, "loss": 0.1724, "step": 18003 }, { "epoch": 111.82866043613707, "grad_norm": 2.4375, "learning_rate": 2.0729176827702326e-05, "loss": 0.1239, "step": 18004 }, { "epoch": 111.83489096573209, "grad_norm": 1.1875, "learning_rate": 2.0726749303371447e-05, "loss": 0.1071, "step": 18005 }, { "epoch": 111.8411214953271, "grad_norm": 1.3828125, "learning_rate": 2.0724321820543237e-05, "loss": 0.1044, "step": 18006 }, { "epoch": 111.84735202492212, "grad_norm": 2.046875, "learning_rate": 2.0721894379241268e-05, "loss": 0.3004, "step": 18007 }, { "epoch": 111.85358255451713, "grad_norm": 2.515625, "learning_rate": 2.071946697948911e-05, "loss": 0.2048, "step": 18008 }, { "epoch": 111.85981308411215, "grad_norm": 1.3671875, "learning_rate": 2.071703962131035e-05, "loss": 0.1252, "step": 18009 }, { "epoch": 111.86604361370716, "grad_norm": 1.96875, "learning_rate": 2.0714612304728566e-05, "loss": 0.158, "step": 18010 }, { "epoch": 111.87227414330218, "grad_norm": 2.953125, "learning_rate": 2.0712185029767316e-05, "loss": 0.276, "step": 18011 }, { "epoch": 111.8785046728972, "grad_norm": 2.4375, "learning_rate": 2.0709757796450178e-05, "loss": 0.155, "step": 18012 }, { "epoch": 111.88473520249221, "grad_norm": 2.890625, "learning_rate": 2.0707330604800735e-05, "loss": 0.2237, "step": 18013 }, { "epoch": 111.89096573208722, "grad_norm": 1.0234375, "learning_rate": 2.0704903454842548e-05, "loss": 0.0951, "step": 18014 }, { "epoch": 111.89719626168224, "grad_norm": 2.1875, "learning_rate": 2.0702476346599202e-05, "loss": 0.1222, "step": 18015 }, { "epoch": 111.90342679127725, "grad_norm": 1.6015625, "learning_rate": 2.0700049280094276e-05, "loss": 0.1047, "step": 18016 }, { "epoch": 111.90965732087227, "grad_norm": 1.203125, "learning_rate": 2.069762225535132e-05, "loss": 0.0972, "step": 18017 }, { "epoch": 111.91588785046729, "grad_norm": 1.7578125, "learning_rate": 2.0695195272393915e-05, "loss": 0.1152, "step": 18018 }, { "epoch": 111.9221183800623, "grad_norm": 2.078125, "learning_rate": 2.0692768331245636e-05, "loss": 0.2268, "step": 18019 }, { "epoch": 111.92834890965732, "grad_norm": 2.28125, "learning_rate": 2.069034143193005e-05, "loss": 0.146, "step": 18020 }, { "epoch": 111.93457943925233, "grad_norm": 3.53125, "learning_rate": 2.068791457447073e-05, "loss": 0.2022, "step": 18021 }, { "epoch": 111.94080996884735, "grad_norm": 3.28125, "learning_rate": 2.068548775889125e-05, "loss": 0.1713, "step": 18022 }, { "epoch": 111.94704049844236, "grad_norm": 2.609375, "learning_rate": 2.0683060985215177e-05, "loss": 0.2907, "step": 18023 }, { "epoch": 111.95327102803738, "grad_norm": 2.5625, "learning_rate": 2.0680634253466074e-05, "loss": 0.1591, "step": 18024 }, { "epoch": 111.9595015576324, "grad_norm": 2.78125, "learning_rate": 2.067820756366751e-05, "loss": 0.2406, "step": 18025 }, { "epoch": 111.96573208722741, "grad_norm": 1.65625, "learning_rate": 2.0675780915843066e-05, "loss": 0.1343, "step": 18026 }, { "epoch": 111.97196261682242, "grad_norm": 1.390625, "learning_rate": 2.06733543100163e-05, "loss": 0.1134, "step": 18027 }, { "epoch": 111.97819314641744, "grad_norm": 2.5, "learning_rate": 2.0670927746210778e-05, "loss": 0.2037, "step": 18028 }, { "epoch": 111.98442367601245, "grad_norm": 2.6875, "learning_rate": 2.066850122445008e-05, "loss": 0.2131, "step": 18029 }, { "epoch": 111.99065420560747, "grad_norm": 3.84375, "learning_rate": 2.0666074744757758e-05, "loss": 0.3079, "step": 18030 }, { "epoch": 111.99688473520249, "grad_norm": 0.94140625, "learning_rate": 2.0663648307157384e-05, "loss": 0.0859, "step": 18031 }, { "epoch": 112.0, "grad_norm": 1.03125, "learning_rate": 2.066122191167252e-05, "loss": 0.0539, "step": 18032 }, { "epoch": 112.00623052959502, "grad_norm": 4.84375, "learning_rate": 2.0658795558326743e-05, "loss": 0.1377, "step": 18033 }, { "epoch": 112.01246105919003, "grad_norm": 2.90625, "learning_rate": 2.0656369247143607e-05, "loss": 0.1797, "step": 18034 }, { "epoch": 112.01869158878505, "grad_norm": 1.515625, "learning_rate": 2.0653942978146692e-05, "loss": 0.1151, "step": 18035 }, { "epoch": 112.02492211838006, "grad_norm": 2.0625, "learning_rate": 2.065151675135954e-05, "loss": 0.1652, "step": 18036 }, { "epoch": 112.03115264797508, "grad_norm": 2.40625, "learning_rate": 2.064909056680573e-05, "loss": 0.2142, "step": 18037 }, { "epoch": 112.03738317757009, "grad_norm": 3.265625, "learning_rate": 2.064666442450882e-05, "loss": 0.3187, "step": 18038 }, { "epoch": 112.04361370716511, "grad_norm": 3.453125, "learning_rate": 2.0644238324492378e-05, "loss": 0.2838, "step": 18039 }, { "epoch": 112.04984423676012, "grad_norm": 3.5625, "learning_rate": 2.064181226677996e-05, "loss": 0.1977, "step": 18040 }, { "epoch": 112.05607476635514, "grad_norm": 2.515625, "learning_rate": 2.0639386251395133e-05, "loss": 0.0998, "step": 18041 }, { "epoch": 112.06230529595015, "grad_norm": 1.8203125, "learning_rate": 2.0636960278361467e-05, "loss": 0.1142, "step": 18042 }, { "epoch": 112.06853582554517, "grad_norm": 1.609375, "learning_rate": 2.0634534347702512e-05, "loss": 0.1127, "step": 18043 }, { "epoch": 112.07476635514018, "grad_norm": 2.25, "learning_rate": 2.0632108459441828e-05, "loss": 0.2156, "step": 18044 }, { "epoch": 112.0809968847352, "grad_norm": 1.765625, "learning_rate": 2.0629682613602977e-05, "loss": 0.1575, "step": 18045 }, { "epoch": 112.08722741433021, "grad_norm": 2.875, "learning_rate": 2.0627256810209526e-05, "loss": 0.3386, "step": 18046 }, { "epoch": 112.09345794392523, "grad_norm": 1.7890625, "learning_rate": 2.0624831049285026e-05, "loss": 0.1355, "step": 18047 }, { "epoch": 112.09968847352025, "grad_norm": 2.15625, "learning_rate": 2.062240533085306e-05, "loss": 0.243, "step": 18048 }, { "epoch": 112.10591900311526, "grad_norm": 1.6796875, "learning_rate": 2.061997965493715e-05, "loss": 0.1101, "step": 18049 }, { "epoch": 112.11214953271028, "grad_norm": 2.109375, "learning_rate": 2.0617554021560876e-05, "loss": 0.1442, "step": 18050 }, { "epoch": 112.11838006230529, "grad_norm": 2.25, "learning_rate": 2.0615128430747792e-05, "loss": 0.1593, "step": 18051 }, { "epoch": 112.1246105919003, "grad_norm": 1.40625, "learning_rate": 2.061270288252146e-05, "loss": 0.1125, "step": 18052 }, { "epoch": 112.13084112149532, "grad_norm": 2.203125, "learning_rate": 2.061027737690544e-05, "loss": 0.0862, "step": 18053 }, { "epoch": 112.13707165109034, "grad_norm": 1.84375, "learning_rate": 2.060785191392328e-05, "loss": 0.2401, "step": 18054 }, { "epoch": 112.14330218068535, "grad_norm": 2.171875, "learning_rate": 2.060542649359854e-05, "loss": 0.1917, "step": 18055 }, { "epoch": 112.14953271028037, "grad_norm": 1.7734375, "learning_rate": 2.0603001115954774e-05, "loss": 0.1598, "step": 18056 }, { "epoch": 112.15576323987538, "grad_norm": 1.3203125, "learning_rate": 2.060057578101554e-05, "loss": 0.086, "step": 18057 }, { "epoch": 112.1619937694704, "grad_norm": 1.5078125, "learning_rate": 2.059815048880439e-05, "loss": 0.1148, "step": 18058 }, { "epoch": 112.16822429906541, "grad_norm": 2.28125, "learning_rate": 2.059572523934489e-05, "loss": 0.1205, "step": 18059 }, { "epoch": 112.17445482866043, "grad_norm": 3.0, "learning_rate": 2.059330003266058e-05, "loss": 0.3081, "step": 18060 }, { "epoch": 112.18068535825545, "grad_norm": 0.9296875, "learning_rate": 2.0590874868775035e-05, "loss": 0.1025, "step": 18061 }, { "epoch": 112.18691588785046, "grad_norm": 1.8203125, "learning_rate": 2.058844974771178e-05, "loss": 0.137, "step": 18062 }, { "epoch": 112.19314641744548, "grad_norm": 1.2421875, "learning_rate": 2.058602466949439e-05, "loss": 0.1337, "step": 18063 }, { "epoch": 112.19937694704049, "grad_norm": 2.9375, "learning_rate": 2.0583599634146403e-05, "loss": 0.1089, "step": 18064 }, { "epoch": 112.20560747663552, "grad_norm": 2.8125, "learning_rate": 2.0581174641691382e-05, "loss": 0.1716, "step": 18065 }, { "epoch": 112.21183800623054, "grad_norm": 2.484375, "learning_rate": 2.0578749692152882e-05, "loss": 0.2344, "step": 18066 }, { "epoch": 112.21806853582555, "grad_norm": 3.21875, "learning_rate": 2.0576324785554453e-05, "loss": 0.2154, "step": 18067 }, { "epoch": 112.22429906542057, "grad_norm": 2.0, "learning_rate": 2.057389992191963e-05, "loss": 0.0992, "step": 18068 }, { "epoch": 112.23052959501558, "grad_norm": 2.25, "learning_rate": 2.0571475101271977e-05, "loss": 0.1077, "step": 18069 }, { "epoch": 112.2367601246106, "grad_norm": 2.09375, "learning_rate": 2.056905032363505e-05, "loss": 0.2024, "step": 18070 }, { "epoch": 112.24299065420561, "grad_norm": 0.87890625, "learning_rate": 2.0566625589032382e-05, "loss": 0.088, "step": 18071 }, { "epoch": 112.24922118380063, "grad_norm": 1.7578125, "learning_rate": 2.056420089748754e-05, "loss": 0.1209, "step": 18072 }, { "epoch": 112.25545171339564, "grad_norm": 1.5546875, "learning_rate": 2.056177624902407e-05, "loss": 0.1323, "step": 18073 }, { "epoch": 112.26168224299066, "grad_norm": 4.09375, "learning_rate": 2.055935164366551e-05, "loss": 0.2895, "step": 18074 }, { "epoch": 112.26791277258567, "grad_norm": 1.265625, "learning_rate": 2.055692708143541e-05, "loss": 0.0926, "step": 18075 }, { "epoch": 112.27414330218069, "grad_norm": 2.078125, "learning_rate": 2.055450256235733e-05, "loss": 0.1844, "step": 18076 }, { "epoch": 112.2803738317757, "grad_norm": 3.578125, "learning_rate": 2.0552078086454805e-05, "loss": 0.1809, "step": 18077 }, { "epoch": 112.28660436137072, "grad_norm": 2.078125, "learning_rate": 2.0549653653751388e-05, "loss": 0.1522, "step": 18078 }, { "epoch": 112.29283489096574, "grad_norm": 1.2578125, "learning_rate": 2.0547229264270625e-05, "loss": 0.1025, "step": 18079 }, { "epoch": 112.29906542056075, "grad_norm": 1.9375, "learning_rate": 2.0544804918036068e-05, "loss": 0.0985, "step": 18080 }, { "epoch": 112.30529595015577, "grad_norm": 1.9375, "learning_rate": 2.054238061507125e-05, "loss": 0.1024, "step": 18081 }, { "epoch": 112.31152647975078, "grad_norm": 2.28125, "learning_rate": 2.053995635539972e-05, "loss": 0.1743, "step": 18082 }, { "epoch": 112.3177570093458, "grad_norm": 2.796875, "learning_rate": 2.0537532139045028e-05, "loss": 0.2377, "step": 18083 }, { "epoch": 112.32398753894081, "grad_norm": 2.375, "learning_rate": 2.053510796603071e-05, "loss": 0.0898, "step": 18084 }, { "epoch": 112.33021806853583, "grad_norm": 2.1875, "learning_rate": 2.0532683836380323e-05, "loss": 0.1428, "step": 18085 }, { "epoch": 112.33644859813084, "grad_norm": 1.953125, "learning_rate": 2.053025975011741e-05, "loss": 0.1299, "step": 18086 }, { "epoch": 112.34267912772586, "grad_norm": 1.671875, "learning_rate": 2.0527835707265497e-05, "loss": 0.1103, "step": 18087 }, { "epoch": 112.34890965732087, "grad_norm": 1.8984375, "learning_rate": 2.052541170784814e-05, "loss": 0.1146, "step": 18088 }, { "epoch": 112.35514018691589, "grad_norm": 1.34375, "learning_rate": 2.0522987751888878e-05, "loss": 0.1126, "step": 18089 }, { "epoch": 112.3613707165109, "grad_norm": 2.234375, "learning_rate": 2.0520563839411257e-05, "loss": 0.1394, "step": 18090 }, { "epoch": 112.36760124610592, "grad_norm": 1.84375, "learning_rate": 2.051813997043881e-05, "loss": 0.1982, "step": 18091 }, { "epoch": 112.37383177570094, "grad_norm": 1.375, "learning_rate": 2.0515716144995088e-05, "loss": 0.1008, "step": 18092 }, { "epoch": 112.38006230529595, "grad_norm": 1.796875, "learning_rate": 2.051329236310363e-05, "loss": 0.1102, "step": 18093 }, { "epoch": 112.38629283489097, "grad_norm": 1.484375, "learning_rate": 2.051086862478797e-05, "loss": 0.1227, "step": 18094 }, { "epoch": 112.39252336448598, "grad_norm": 1.828125, "learning_rate": 2.0508444930071648e-05, "loss": 0.2304, "step": 18095 }, { "epoch": 112.398753894081, "grad_norm": 2.59375, "learning_rate": 2.050602127897821e-05, "loss": 0.1619, "step": 18096 }, { "epoch": 112.40498442367601, "grad_norm": 1.8203125, "learning_rate": 2.050359767153119e-05, "loss": 0.101, "step": 18097 }, { "epoch": 112.41121495327103, "grad_norm": 2.359375, "learning_rate": 2.050117410775413e-05, "loss": 0.1837, "step": 18098 }, { "epoch": 112.41744548286604, "grad_norm": 2.609375, "learning_rate": 2.0498750587670572e-05, "loss": 0.1868, "step": 18099 }, { "epoch": 112.42367601246106, "grad_norm": 2.515625, "learning_rate": 2.0496327111304042e-05, "loss": 0.1464, "step": 18100 }, { "epoch": 112.42990654205607, "grad_norm": 2.6875, "learning_rate": 2.0493903678678077e-05, "loss": 0.1635, "step": 18101 }, { "epoch": 112.43613707165109, "grad_norm": 1.5625, "learning_rate": 2.049148028981623e-05, "loss": 0.1158, "step": 18102 }, { "epoch": 112.4423676012461, "grad_norm": 2.125, "learning_rate": 2.0489056944742026e-05, "loss": 0.1645, "step": 18103 }, { "epoch": 112.44859813084112, "grad_norm": 3.046875, "learning_rate": 2.0486633643479e-05, "loss": 0.2655, "step": 18104 }, { "epoch": 112.45482866043614, "grad_norm": 2.796875, "learning_rate": 2.0484210386050695e-05, "loss": 0.2785, "step": 18105 }, { "epoch": 112.46105919003115, "grad_norm": 2.578125, "learning_rate": 2.0481787172480637e-05, "loss": 0.2432, "step": 18106 }, { "epoch": 112.46728971962617, "grad_norm": 2.84375, "learning_rate": 2.0479364002792367e-05, "loss": 0.141, "step": 18107 }, { "epoch": 112.47352024922118, "grad_norm": 1.1640625, "learning_rate": 2.0476940877009413e-05, "loss": 0.095, "step": 18108 }, { "epoch": 112.4797507788162, "grad_norm": 2.609375, "learning_rate": 2.047451779515532e-05, "loss": 0.1525, "step": 18109 }, { "epoch": 112.48598130841121, "grad_norm": 2.046875, "learning_rate": 2.0472094757253605e-05, "loss": 0.0886, "step": 18110 }, { "epoch": 112.49221183800623, "grad_norm": 1.1953125, "learning_rate": 2.0469671763327815e-05, "loss": 0.1005, "step": 18111 }, { "epoch": 112.49844236760124, "grad_norm": 1.4375, "learning_rate": 2.046724881340149e-05, "loss": 0.1368, "step": 18112 }, { "epoch": 112.50467289719626, "grad_norm": 2.34375, "learning_rate": 2.0464825907498143e-05, "loss": 0.184, "step": 18113 }, { "epoch": 112.51090342679127, "grad_norm": 1.5390625, "learning_rate": 2.0462403045641304e-05, "loss": 0.1143, "step": 18114 }, { "epoch": 112.51713395638629, "grad_norm": 1.3671875, "learning_rate": 2.0459980227854525e-05, "loss": 0.12, "step": 18115 }, { "epoch": 112.5233644859813, "grad_norm": 2.25, "learning_rate": 2.045755745416132e-05, "loss": 0.118, "step": 18116 }, { "epoch": 112.52959501557632, "grad_norm": 1.9140625, "learning_rate": 2.0455134724585224e-05, "loss": 0.1141, "step": 18117 }, { "epoch": 112.53582554517133, "grad_norm": 1.71875, "learning_rate": 2.0452712039149776e-05, "loss": 0.1215, "step": 18118 }, { "epoch": 112.54205607476635, "grad_norm": 2.625, "learning_rate": 2.0450289397878494e-05, "loss": 0.1959, "step": 18119 }, { "epoch": 112.54828660436137, "grad_norm": 1.5, "learning_rate": 2.044786680079491e-05, "loss": 0.0966, "step": 18120 }, { "epoch": 112.55451713395638, "grad_norm": 3.328125, "learning_rate": 2.0445444247922552e-05, "loss": 0.2776, "step": 18121 }, { "epoch": 112.5607476635514, "grad_norm": 1.828125, "learning_rate": 2.044302173928495e-05, "loss": 0.1382, "step": 18122 }, { "epoch": 112.56697819314641, "grad_norm": 2.15625, "learning_rate": 2.0440599274905633e-05, "loss": 0.1585, "step": 18123 }, { "epoch": 112.57320872274143, "grad_norm": 2.125, "learning_rate": 2.043817685480813e-05, "loss": 0.1586, "step": 18124 }, { "epoch": 112.57943925233644, "grad_norm": 1.53125, "learning_rate": 2.0435754479015962e-05, "loss": 0.1262, "step": 18125 }, { "epoch": 112.58566978193146, "grad_norm": 1.203125, "learning_rate": 2.043333214755266e-05, "loss": 0.1245, "step": 18126 }, { "epoch": 112.59190031152647, "grad_norm": 2.5, "learning_rate": 2.043090986044175e-05, "loss": 0.1161, "step": 18127 }, { "epoch": 112.59813084112149, "grad_norm": 2.515625, "learning_rate": 2.042848761770675e-05, "loss": 0.2017, "step": 18128 }, { "epoch": 112.60436137071652, "grad_norm": 1.4765625, "learning_rate": 2.0426065419371198e-05, "loss": 0.1252, "step": 18129 }, { "epoch": 112.61059190031153, "grad_norm": 2.453125, "learning_rate": 2.0423643265458607e-05, "loss": 0.1048, "step": 18130 }, { "epoch": 112.61682242990655, "grad_norm": 1.484375, "learning_rate": 2.0421221155992526e-05, "loss": 0.1305, "step": 18131 }, { "epoch": 112.62305295950156, "grad_norm": 3.703125, "learning_rate": 2.041879909099644e-05, "loss": 0.1917, "step": 18132 }, { "epoch": 112.62928348909658, "grad_norm": 2.078125, "learning_rate": 2.0416377070493904e-05, "loss": 0.1832, "step": 18133 }, { "epoch": 112.6355140186916, "grad_norm": 1.984375, "learning_rate": 2.0413955094508418e-05, "loss": 0.158, "step": 18134 }, { "epoch": 112.64174454828661, "grad_norm": 2.15625, "learning_rate": 2.0411533163063525e-05, "loss": 0.1268, "step": 18135 }, { "epoch": 112.64797507788163, "grad_norm": 2.046875, "learning_rate": 2.040911127618274e-05, "loss": 0.1729, "step": 18136 }, { "epoch": 112.65420560747664, "grad_norm": 1.9609375, "learning_rate": 2.040668943388959e-05, "loss": 0.1326, "step": 18137 }, { "epoch": 112.66043613707166, "grad_norm": 1.796875, "learning_rate": 2.040426763620758e-05, "loss": 0.139, "step": 18138 }, { "epoch": 112.66666666666667, "grad_norm": 2.625, "learning_rate": 2.0401845883160246e-05, "loss": 0.1879, "step": 18139 }, { "epoch": 112.67289719626169, "grad_norm": 2.234375, "learning_rate": 2.0399424174771102e-05, "loss": 0.1743, "step": 18140 }, { "epoch": 112.6791277258567, "grad_norm": 4.3125, "learning_rate": 2.0397002511063668e-05, "loss": 0.1806, "step": 18141 }, { "epoch": 112.68535825545172, "grad_norm": 3.484375, "learning_rate": 2.0394580892061468e-05, "loss": 0.1848, "step": 18142 }, { "epoch": 112.69158878504673, "grad_norm": 2.078125, "learning_rate": 2.0392159317788028e-05, "loss": 0.1139, "step": 18143 }, { "epoch": 112.69781931464175, "grad_norm": 1.578125, "learning_rate": 2.038973778826685e-05, "loss": 0.1288, "step": 18144 }, { "epoch": 112.70404984423676, "grad_norm": 1.0, "learning_rate": 2.0387316303521454e-05, "loss": 0.1113, "step": 18145 }, { "epoch": 112.71028037383178, "grad_norm": 1.859375, "learning_rate": 2.038489486357537e-05, "loss": 0.1502, "step": 18146 }, { "epoch": 112.7165109034268, "grad_norm": 2.453125, "learning_rate": 2.0382473468452103e-05, "loss": 0.2328, "step": 18147 }, { "epoch": 112.72274143302181, "grad_norm": 1.9609375, "learning_rate": 2.0380052118175184e-05, "loss": 0.2186, "step": 18148 }, { "epoch": 112.72897196261682, "grad_norm": 3.234375, "learning_rate": 2.037763081276812e-05, "loss": 0.1787, "step": 18149 }, { "epoch": 112.73520249221184, "grad_norm": 1.1171875, "learning_rate": 2.037520955225444e-05, "loss": 0.0955, "step": 18150 }, { "epoch": 112.74143302180686, "grad_norm": 2.65625, "learning_rate": 2.0372788336657636e-05, "loss": 0.1242, "step": 18151 }, { "epoch": 112.74766355140187, "grad_norm": 2.421875, "learning_rate": 2.0370367166001243e-05, "loss": 0.2018, "step": 18152 }, { "epoch": 112.75389408099689, "grad_norm": 2.25, "learning_rate": 2.0367946040308768e-05, "loss": 0.1907, "step": 18153 }, { "epoch": 112.7601246105919, "grad_norm": 2.78125, "learning_rate": 2.036552495960372e-05, "loss": 0.2208, "step": 18154 }, { "epoch": 112.76635514018692, "grad_norm": 0.8828125, "learning_rate": 2.0363103923909627e-05, "loss": 0.0902, "step": 18155 }, { "epoch": 112.77258566978193, "grad_norm": 2.84375, "learning_rate": 2.0360682933250002e-05, "loss": 0.1394, "step": 18156 }, { "epoch": 112.77881619937695, "grad_norm": 1.640625, "learning_rate": 2.0358261987648346e-05, "loss": 0.1169, "step": 18157 }, { "epoch": 112.78504672897196, "grad_norm": 2.171875, "learning_rate": 2.035584108712817e-05, "loss": 0.1095, "step": 18158 }, { "epoch": 112.79127725856698, "grad_norm": 1.6015625, "learning_rate": 2.0353420231713004e-05, "loss": 0.1718, "step": 18159 }, { "epoch": 112.797507788162, "grad_norm": 2.140625, "learning_rate": 2.0350999421426338e-05, "loss": 0.1729, "step": 18160 }, { "epoch": 112.80373831775701, "grad_norm": 1.7421875, "learning_rate": 2.0348578656291702e-05, "loss": 0.2395, "step": 18161 }, { "epoch": 112.80996884735202, "grad_norm": 1.671875, "learning_rate": 2.0346157936332603e-05, "loss": 0.2167, "step": 18162 }, { "epoch": 112.81619937694704, "grad_norm": 2.8125, "learning_rate": 2.034373726157255e-05, "loss": 0.1217, "step": 18163 }, { "epoch": 112.82242990654206, "grad_norm": 3.6875, "learning_rate": 2.034131663203504e-05, "loss": 0.2674, "step": 18164 }, { "epoch": 112.82866043613707, "grad_norm": 1.6015625, "learning_rate": 2.0338896047743607e-05, "loss": 0.1188, "step": 18165 }, { "epoch": 112.83489096573209, "grad_norm": 2.390625, "learning_rate": 2.0336475508721743e-05, "loss": 0.2234, "step": 18166 }, { "epoch": 112.8411214953271, "grad_norm": 2.265625, "learning_rate": 2.0334055014992953e-05, "loss": 0.1657, "step": 18167 }, { "epoch": 112.84735202492212, "grad_norm": 2.1875, "learning_rate": 2.033163456658076e-05, "loss": 0.177, "step": 18168 }, { "epoch": 112.85358255451713, "grad_norm": 1.390625, "learning_rate": 2.0329214163508672e-05, "loss": 0.1018, "step": 18169 }, { "epoch": 112.85981308411215, "grad_norm": 1.890625, "learning_rate": 2.0326793805800184e-05, "loss": 0.1329, "step": 18170 }, { "epoch": 112.86604361370716, "grad_norm": 2.90625, "learning_rate": 2.0324373493478804e-05, "loss": 0.2062, "step": 18171 }, { "epoch": 112.87227414330218, "grad_norm": 1.1015625, "learning_rate": 2.0321953226568046e-05, "loss": 0.1045, "step": 18172 }, { "epoch": 112.8785046728972, "grad_norm": 2.171875, "learning_rate": 2.0319533005091415e-05, "loss": 0.1054, "step": 18173 }, { "epoch": 112.88473520249221, "grad_norm": 0.671875, "learning_rate": 2.0317112829072408e-05, "loss": 0.0895, "step": 18174 }, { "epoch": 112.89096573208722, "grad_norm": 1.828125, "learning_rate": 2.031469269853455e-05, "loss": 0.1282, "step": 18175 }, { "epoch": 112.89719626168224, "grad_norm": 1.921875, "learning_rate": 2.031227261350133e-05, "loss": 0.0887, "step": 18176 }, { "epoch": 112.90342679127725, "grad_norm": 1.6953125, "learning_rate": 2.0309852573996253e-05, "loss": 0.1199, "step": 18177 }, { "epoch": 112.90965732087227, "grad_norm": 2.15625, "learning_rate": 2.030743258004282e-05, "loss": 0.1426, "step": 18178 }, { "epoch": 112.91588785046729, "grad_norm": 2.21875, "learning_rate": 2.0305012631664543e-05, "loss": 0.18, "step": 18179 }, { "epoch": 112.9221183800623, "grad_norm": 1.6328125, "learning_rate": 2.0302592728884918e-05, "loss": 0.103, "step": 18180 }, { "epoch": 112.92834890965732, "grad_norm": 1.890625, "learning_rate": 2.030017287172746e-05, "loss": 0.1095, "step": 18181 }, { "epoch": 112.93457943925233, "grad_norm": 1.578125, "learning_rate": 2.0297753060215666e-05, "loss": 0.1476, "step": 18182 }, { "epoch": 112.94080996884735, "grad_norm": 2.859375, "learning_rate": 2.029533329437303e-05, "loss": 0.3279, "step": 18183 }, { "epoch": 112.94704049844236, "grad_norm": 2.0625, "learning_rate": 2.0292913574223048e-05, "loss": 0.1804, "step": 18184 }, { "epoch": 112.95327102803738, "grad_norm": 2.65625, "learning_rate": 2.0290493899789238e-05, "loss": 0.1737, "step": 18185 }, { "epoch": 112.9595015576324, "grad_norm": 1.1484375, "learning_rate": 2.0288074271095092e-05, "loss": 0.0857, "step": 18186 }, { "epoch": 112.96573208722741, "grad_norm": 2.109375, "learning_rate": 2.0285654688164107e-05, "loss": 0.1624, "step": 18187 }, { "epoch": 112.97196261682242, "grad_norm": 2.328125, "learning_rate": 2.0283235151019793e-05, "loss": 0.1458, "step": 18188 }, { "epoch": 112.97819314641744, "grad_norm": 1.9296875, "learning_rate": 2.0280815659685643e-05, "loss": 0.1287, "step": 18189 }, { "epoch": 112.98442367601245, "grad_norm": 2.53125, "learning_rate": 2.027839621418515e-05, "loss": 0.1351, "step": 18190 }, { "epoch": 112.99065420560747, "grad_norm": 2.53125, "learning_rate": 2.0275976814541812e-05, "loss": 0.1831, "step": 18191 }, { "epoch": 112.99688473520249, "grad_norm": 2.96875, "learning_rate": 2.0273557460779134e-05, "loss": 0.2144, "step": 18192 }, { "epoch": 113.0, "grad_norm": 1.5078125, "learning_rate": 2.027113815292061e-05, "loss": 0.0612, "step": 18193 }, { "epoch": 113.00623052959502, "grad_norm": 1.03125, "learning_rate": 2.0268718890989753e-05, "loss": 0.1086, "step": 18194 }, { "epoch": 113.01246105919003, "grad_norm": 3.0625, "learning_rate": 2.026629967501003e-05, "loss": 0.2593, "step": 18195 }, { "epoch": 113.01869158878505, "grad_norm": 2.859375, "learning_rate": 2.026388050500495e-05, "loss": 0.1703, "step": 18196 }, { "epoch": 113.02492211838006, "grad_norm": 2.390625, "learning_rate": 2.026146138099801e-05, "loss": 0.2275, "step": 18197 }, { "epoch": 113.03115264797508, "grad_norm": 1.375, "learning_rate": 2.0259042303012703e-05, "loss": 0.113, "step": 18198 }, { "epoch": 113.03738317757009, "grad_norm": 2.953125, "learning_rate": 2.025662327107253e-05, "loss": 0.3338, "step": 18199 }, { "epoch": 113.04361370716511, "grad_norm": 2.109375, "learning_rate": 2.025420428520097e-05, "loss": 0.1683, "step": 18200 }, { "epoch": 113.04984423676012, "grad_norm": 1.078125, "learning_rate": 2.0251785345421544e-05, "loss": 0.1217, "step": 18201 }, { "epoch": 113.05607476635514, "grad_norm": 1.0234375, "learning_rate": 2.024936645175772e-05, "loss": 0.0989, "step": 18202 }, { "epoch": 113.06230529595015, "grad_norm": 1.3984375, "learning_rate": 2.0246947604232995e-05, "loss": 0.111, "step": 18203 }, { "epoch": 113.06853582554517, "grad_norm": 1.375, "learning_rate": 2.0244528802870864e-05, "loss": 0.0807, "step": 18204 }, { "epoch": 113.07476635514018, "grad_norm": 1.3515625, "learning_rate": 2.0242110047694824e-05, "loss": 0.1099, "step": 18205 }, { "epoch": 113.0809968847352, "grad_norm": 1.875, "learning_rate": 2.0239691338728354e-05, "loss": 0.1449, "step": 18206 }, { "epoch": 113.08722741433021, "grad_norm": 1.1484375, "learning_rate": 2.0237272675994976e-05, "loss": 0.1101, "step": 18207 }, { "epoch": 113.09345794392523, "grad_norm": 1.796875, "learning_rate": 2.0234854059518138e-05, "loss": 0.1063, "step": 18208 }, { "epoch": 113.09968847352025, "grad_norm": 1.96875, "learning_rate": 2.0232435489321354e-05, "loss": 0.1213, "step": 18209 }, { "epoch": 113.10591900311526, "grad_norm": 1.765625, "learning_rate": 2.0230016965428107e-05, "loss": 0.1433, "step": 18210 }, { "epoch": 113.11214953271028, "grad_norm": 2.28125, "learning_rate": 2.022759848786189e-05, "loss": 0.2036, "step": 18211 }, { "epoch": 113.11838006230529, "grad_norm": 2.0, "learning_rate": 2.022518005664619e-05, "loss": 0.1825, "step": 18212 }, { "epoch": 113.1246105919003, "grad_norm": 1.5390625, "learning_rate": 2.0222761671804506e-05, "loss": 0.1117, "step": 18213 }, { "epoch": 113.13084112149532, "grad_norm": 1.734375, "learning_rate": 2.0220343333360304e-05, "loss": 0.0993, "step": 18214 }, { "epoch": 113.13707165109034, "grad_norm": 2.671875, "learning_rate": 2.0217925041337088e-05, "loss": 0.2349, "step": 18215 }, { "epoch": 113.14330218068535, "grad_norm": 2.015625, "learning_rate": 2.021550679575834e-05, "loss": 0.17, "step": 18216 }, { "epoch": 113.14953271028037, "grad_norm": 1.796875, "learning_rate": 2.0213088596647537e-05, "loss": 0.1062, "step": 18217 }, { "epoch": 113.15576323987538, "grad_norm": 2.328125, "learning_rate": 2.0210670444028184e-05, "loss": 0.1329, "step": 18218 }, { "epoch": 113.1619937694704, "grad_norm": 2.96875, "learning_rate": 2.0208252337923757e-05, "loss": 0.1867, "step": 18219 }, { "epoch": 113.16822429906541, "grad_norm": 3.25, "learning_rate": 2.0205834278357742e-05, "loss": 0.2073, "step": 18220 }, { "epoch": 113.17445482866043, "grad_norm": 2.578125, "learning_rate": 2.0203416265353617e-05, "loss": 0.1757, "step": 18221 }, { "epoch": 113.18068535825545, "grad_norm": 2.875, "learning_rate": 2.0200998298934874e-05, "loss": 0.1252, "step": 18222 }, { "epoch": 113.18691588785046, "grad_norm": 1.6953125, "learning_rate": 2.0198580379124997e-05, "loss": 0.091, "step": 18223 }, { "epoch": 113.19314641744548, "grad_norm": 1.625, "learning_rate": 2.0196162505947464e-05, "loss": 0.0934, "step": 18224 }, { "epoch": 113.19937694704049, "grad_norm": 1.7578125, "learning_rate": 2.0193744679425764e-05, "loss": 0.1498, "step": 18225 }, { "epoch": 113.20560747663552, "grad_norm": 1.921875, "learning_rate": 2.0191326899583378e-05, "loss": 0.1545, "step": 18226 }, { "epoch": 113.21183800623054, "grad_norm": 2.640625, "learning_rate": 2.0188909166443787e-05, "loss": 0.1438, "step": 18227 }, { "epoch": 113.21806853582555, "grad_norm": 2.3125, "learning_rate": 2.0186491480030464e-05, "loss": 0.1588, "step": 18228 }, { "epoch": 113.22429906542057, "grad_norm": 2.015625, "learning_rate": 2.01840738403669e-05, "loss": 0.2251, "step": 18229 }, { "epoch": 113.23052959501558, "grad_norm": 2.828125, "learning_rate": 2.0181656247476573e-05, "loss": 0.2318, "step": 18230 }, { "epoch": 113.2367601246106, "grad_norm": 3.390625, "learning_rate": 2.0179238701382968e-05, "loss": 0.2116, "step": 18231 }, { "epoch": 113.24299065420561, "grad_norm": 1.171875, "learning_rate": 2.017682120210956e-05, "loss": 0.1228, "step": 18232 }, { "epoch": 113.24922118380063, "grad_norm": 2.796875, "learning_rate": 2.0174403749679835e-05, "loss": 0.2886, "step": 18233 }, { "epoch": 113.25545171339564, "grad_norm": 2.21875, "learning_rate": 2.0171986344117256e-05, "loss": 0.1794, "step": 18234 }, { "epoch": 113.26168224299066, "grad_norm": 1.453125, "learning_rate": 2.016956898544531e-05, "loss": 0.1328, "step": 18235 }, { "epoch": 113.26791277258567, "grad_norm": 1.4921875, "learning_rate": 2.016715167368748e-05, "loss": 0.1154, "step": 18236 }, { "epoch": 113.27414330218069, "grad_norm": 1.4296875, "learning_rate": 2.0164734408867232e-05, "loss": 0.0937, "step": 18237 }, { "epoch": 113.2803738317757, "grad_norm": 1.0546875, "learning_rate": 2.016231719100806e-05, "loss": 0.0939, "step": 18238 }, { "epoch": 113.28660436137072, "grad_norm": 2.484375, "learning_rate": 2.015990002013343e-05, "loss": 0.1789, "step": 18239 }, { "epoch": 113.29283489096574, "grad_norm": 0.91796875, "learning_rate": 2.0157482896266817e-05, "loss": 0.1052, "step": 18240 }, { "epoch": 113.29906542056075, "grad_norm": 1.7734375, "learning_rate": 2.015506581943169e-05, "loss": 0.1207, "step": 18241 }, { "epoch": 113.30529595015577, "grad_norm": 2.703125, "learning_rate": 2.015264878965154e-05, "loss": 0.2081, "step": 18242 }, { "epoch": 113.31152647975078, "grad_norm": 2.578125, "learning_rate": 2.0150231806949827e-05, "loss": 0.165, "step": 18243 }, { "epoch": 113.3177570093458, "grad_norm": 1.453125, "learning_rate": 2.014781487135004e-05, "loss": 0.1134, "step": 18244 }, { "epoch": 113.32398753894081, "grad_norm": 3.21875, "learning_rate": 2.0145397982875647e-05, "loss": 0.2875, "step": 18245 }, { "epoch": 113.33021806853583, "grad_norm": 1.2578125, "learning_rate": 2.0142981141550113e-05, "loss": 0.1175, "step": 18246 }, { "epoch": 113.33644859813084, "grad_norm": 1.28125, "learning_rate": 2.0140564347396916e-05, "loss": 0.1254, "step": 18247 }, { "epoch": 113.34267912772586, "grad_norm": 2.1875, "learning_rate": 2.013814760043953e-05, "loss": 0.1455, "step": 18248 }, { "epoch": 113.34890965732087, "grad_norm": 1.546875, "learning_rate": 2.013573090070143e-05, "loss": 0.1732, "step": 18249 }, { "epoch": 113.35514018691589, "grad_norm": 2.65625, "learning_rate": 2.0133314248206076e-05, "loss": 0.31, "step": 18250 }, { "epoch": 113.3613707165109, "grad_norm": 1.8828125, "learning_rate": 2.0130897642976953e-05, "loss": 0.1528, "step": 18251 }, { "epoch": 113.36760124610592, "grad_norm": 0.98828125, "learning_rate": 2.012848108503753e-05, "loss": 0.0876, "step": 18252 }, { "epoch": 113.37383177570094, "grad_norm": 1.875, "learning_rate": 2.012606457441127e-05, "loss": 0.1437, "step": 18253 }, { "epoch": 113.38006230529595, "grad_norm": 3.171875, "learning_rate": 2.012364811112164e-05, "loss": 0.3214, "step": 18254 }, { "epoch": 113.38629283489097, "grad_norm": 3.109375, "learning_rate": 2.0121231695192118e-05, "loss": 0.1545, "step": 18255 }, { "epoch": 113.39252336448598, "grad_norm": 0.8984375, "learning_rate": 2.0118815326646163e-05, "loss": 0.105, "step": 18256 }, { "epoch": 113.398753894081, "grad_norm": 3.25, "learning_rate": 2.0116399005507252e-05, "loss": 0.2247, "step": 18257 }, { "epoch": 113.40498442367601, "grad_norm": 1.7890625, "learning_rate": 2.011398273179886e-05, "loss": 0.1321, "step": 18258 }, { "epoch": 113.41121495327103, "grad_norm": 2.328125, "learning_rate": 2.0111566505544438e-05, "loss": 0.227, "step": 18259 }, { "epoch": 113.41744548286604, "grad_norm": 1.703125, "learning_rate": 2.0109150326767453e-05, "loss": 0.1326, "step": 18260 }, { "epoch": 113.42367601246106, "grad_norm": 0.80859375, "learning_rate": 2.0106734195491384e-05, "loss": 0.0753, "step": 18261 }, { "epoch": 113.42990654205607, "grad_norm": 1.1171875, "learning_rate": 2.010431811173969e-05, "loss": 0.1273, "step": 18262 }, { "epoch": 113.43613707165109, "grad_norm": 2.328125, "learning_rate": 2.0101902075535834e-05, "loss": 0.1924, "step": 18263 }, { "epoch": 113.4423676012461, "grad_norm": 2.453125, "learning_rate": 2.0099486086903295e-05, "loss": 0.2109, "step": 18264 }, { "epoch": 113.44859813084112, "grad_norm": 2.171875, "learning_rate": 2.0097070145865517e-05, "loss": 0.12, "step": 18265 }, { "epoch": 113.45482866043614, "grad_norm": 2.015625, "learning_rate": 2.0094654252445973e-05, "loss": 0.1184, "step": 18266 }, { "epoch": 113.46105919003115, "grad_norm": 1.5625, "learning_rate": 2.0092238406668125e-05, "loss": 0.137, "step": 18267 }, { "epoch": 113.46728971962617, "grad_norm": 2.34375, "learning_rate": 2.0089822608555448e-05, "loss": 0.171, "step": 18268 }, { "epoch": 113.47352024922118, "grad_norm": 1.546875, "learning_rate": 2.008740685813139e-05, "loss": 0.1106, "step": 18269 }, { "epoch": 113.4797507788162, "grad_norm": 2.609375, "learning_rate": 2.0084991155419413e-05, "loss": 0.2085, "step": 18270 }, { "epoch": 113.48598130841121, "grad_norm": 2.359375, "learning_rate": 2.0082575500442998e-05, "loss": 0.2422, "step": 18271 }, { "epoch": 113.49221183800623, "grad_norm": 2.75, "learning_rate": 2.0080159893225587e-05, "loss": 0.2233, "step": 18272 }, { "epoch": 113.49844236760124, "grad_norm": 2.78125, "learning_rate": 2.0077744333790642e-05, "loss": 0.2633, "step": 18273 }, { "epoch": 113.50467289719626, "grad_norm": 3.5625, "learning_rate": 2.0075328822161628e-05, "loss": 0.3576, "step": 18274 }, { "epoch": 113.51090342679127, "grad_norm": 3.265625, "learning_rate": 2.0072913358362013e-05, "loss": 0.17, "step": 18275 }, { "epoch": 113.51713395638629, "grad_norm": 2.6875, "learning_rate": 2.0070497942415232e-05, "loss": 0.1355, "step": 18276 }, { "epoch": 113.5233644859813, "grad_norm": 2.4375, "learning_rate": 2.0068082574344787e-05, "loss": 0.1215, "step": 18277 }, { "epoch": 113.52959501557632, "grad_norm": 1.859375, "learning_rate": 2.006566725417409e-05, "loss": 0.1726, "step": 18278 }, { "epoch": 113.53582554517133, "grad_norm": 3.34375, "learning_rate": 2.0063251981926624e-05, "loss": 0.355, "step": 18279 }, { "epoch": 113.54205607476635, "grad_norm": 2.328125, "learning_rate": 2.0060836757625836e-05, "loss": 0.1334, "step": 18280 }, { "epoch": 113.54828660436137, "grad_norm": 2.890625, "learning_rate": 2.0058421581295196e-05, "loss": 0.1546, "step": 18281 }, { "epoch": 113.55451713395638, "grad_norm": 2.625, "learning_rate": 2.0056006452958152e-05, "loss": 0.2211, "step": 18282 }, { "epoch": 113.5607476635514, "grad_norm": 3.296875, "learning_rate": 2.005359137263815e-05, "loss": 0.1627, "step": 18283 }, { "epoch": 113.56697819314641, "grad_norm": 2.0625, "learning_rate": 2.005117634035868e-05, "loss": 0.1983, "step": 18284 }, { "epoch": 113.57320872274143, "grad_norm": 3.46875, "learning_rate": 2.0048761356143165e-05, "loss": 0.2165, "step": 18285 }, { "epoch": 113.57943925233644, "grad_norm": 1.8671875, "learning_rate": 2.0046346420015067e-05, "loss": 0.1656, "step": 18286 }, { "epoch": 113.58566978193146, "grad_norm": 2.125, "learning_rate": 2.0043931531997838e-05, "loss": 0.1422, "step": 18287 }, { "epoch": 113.59190031152647, "grad_norm": 1.6328125, "learning_rate": 2.0041516692114943e-05, "loss": 0.1056, "step": 18288 }, { "epoch": 113.59813084112149, "grad_norm": 1.0703125, "learning_rate": 2.003910190038982e-05, "loss": 0.0899, "step": 18289 }, { "epoch": 113.60436137071652, "grad_norm": 2.109375, "learning_rate": 2.003668715684595e-05, "loss": 0.0901, "step": 18290 }, { "epoch": 113.61059190031153, "grad_norm": 2.53125, "learning_rate": 2.0034272461506754e-05, "loss": 0.1358, "step": 18291 }, { "epoch": 113.61682242990655, "grad_norm": 1.03125, "learning_rate": 2.0031857814395697e-05, "loss": 0.1005, "step": 18292 }, { "epoch": 113.62305295950156, "grad_norm": 2.734375, "learning_rate": 2.0029443215536227e-05, "loss": 0.2807, "step": 18293 }, { "epoch": 113.62928348909658, "grad_norm": 2.09375, "learning_rate": 2.0027028664951804e-05, "loss": 0.1658, "step": 18294 }, { "epoch": 113.6355140186916, "grad_norm": 2.875, "learning_rate": 2.002461416266587e-05, "loss": 0.2426, "step": 18295 }, { "epoch": 113.64174454828661, "grad_norm": 2.296875, "learning_rate": 2.0022199708701883e-05, "loss": 0.1676, "step": 18296 }, { "epoch": 113.64797507788163, "grad_norm": 1.015625, "learning_rate": 2.0019785303083276e-05, "loss": 0.1022, "step": 18297 }, { "epoch": 113.65420560747664, "grad_norm": 1.1953125, "learning_rate": 2.001737094583352e-05, "loss": 0.0894, "step": 18298 }, { "epoch": 113.66043613707166, "grad_norm": 2.015625, "learning_rate": 2.001495663697605e-05, "loss": 0.1682, "step": 18299 }, { "epoch": 113.66666666666667, "grad_norm": 2.359375, "learning_rate": 2.0012542376534314e-05, "loss": 0.2226, "step": 18300 }, { "epoch": 113.67289719626169, "grad_norm": 1.4765625, "learning_rate": 2.0010128164531766e-05, "loss": 0.1167, "step": 18301 }, { "epoch": 113.6791277258567, "grad_norm": 1.015625, "learning_rate": 2.000771400099185e-05, "loss": 0.1069, "step": 18302 }, { "epoch": 113.68535825545172, "grad_norm": 1.7109375, "learning_rate": 2.0005299885938025e-05, "loss": 0.1077, "step": 18303 }, { "epoch": 113.69158878504673, "grad_norm": 1.75, "learning_rate": 2.000288581939371e-05, "loss": 0.1546, "step": 18304 }, { "epoch": 113.69781931464175, "grad_norm": 1.7265625, "learning_rate": 2.0000471801382374e-05, "loss": 0.1083, "step": 18305 }, { "epoch": 113.70404984423676, "grad_norm": 1.3515625, "learning_rate": 1.999805783192745e-05, "loss": 0.116, "step": 18306 }, { "epoch": 113.71028037383178, "grad_norm": 0.89453125, "learning_rate": 1.9995643911052392e-05, "loss": 0.1074, "step": 18307 }, { "epoch": 113.7165109034268, "grad_norm": 2.9375, "learning_rate": 1.999323003878064e-05, "loss": 0.2172, "step": 18308 }, { "epoch": 113.72274143302181, "grad_norm": 2.0, "learning_rate": 1.999081621513565e-05, "loss": 0.1525, "step": 18309 }, { "epoch": 113.72897196261682, "grad_norm": 1.6484375, "learning_rate": 1.998840244014084e-05, "loss": 0.1389, "step": 18310 }, { "epoch": 113.73520249221184, "grad_norm": 1.609375, "learning_rate": 1.998598871381967e-05, "loss": 0.1099, "step": 18311 }, { "epoch": 113.74143302180686, "grad_norm": 1.5234375, "learning_rate": 1.9983575036195583e-05, "loss": 0.1144, "step": 18312 }, { "epoch": 113.74766355140187, "grad_norm": 1.6953125, "learning_rate": 1.998116140729201e-05, "loss": 0.1305, "step": 18313 }, { "epoch": 113.75389408099689, "grad_norm": 1.703125, "learning_rate": 1.9978747827132407e-05, "loss": 0.1228, "step": 18314 }, { "epoch": 113.7601246105919, "grad_norm": 1.7109375, "learning_rate": 1.997633429574022e-05, "loss": 0.1421, "step": 18315 }, { "epoch": 113.76635514018692, "grad_norm": 1.1015625, "learning_rate": 1.9973920813138863e-05, "loss": 0.121, "step": 18316 }, { "epoch": 113.77258566978193, "grad_norm": 1.1640625, "learning_rate": 1.9971507379351792e-05, "loss": 0.1232, "step": 18317 }, { "epoch": 113.77881619937695, "grad_norm": 1.9140625, "learning_rate": 1.9969093994402452e-05, "loss": 0.1562, "step": 18318 }, { "epoch": 113.78504672897196, "grad_norm": 3.984375, "learning_rate": 1.9966680658314273e-05, "loss": 0.2334, "step": 18319 }, { "epoch": 113.79127725856698, "grad_norm": 1.3125, "learning_rate": 1.99642673711107e-05, "loss": 0.0912, "step": 18320 }, { "epoch": 113.797507788162, "grad_norm": 2.921875, "learning_rate": 1.9961854132815166e-05, "loss": 0.287, "step": 18321 }, { "epoch": 113.80373831775701, "grad_norm": 1.453125, "learning_rate": 1.995944094345112e-05, "loss": 0.111, "step": 18322 }, { "epoch": 113.80996884735202, "grad_norm": 2.859375, "learning_rate": 1.995702780304199e-05, "loss": 0.2315, "step": 18323 }, { "epoch": 113.81619937694704, "grad_norm": 1.71875, "learning_rate": 1.9954614711611202e-05, "loss": 0.104, "step": 18324 }, { "epoch": 113.82242990654206, "grad_norm": 2.78125, "learning_rate": 1.9952201669182213e-05, "loss": 0.2637, "step": 18325 }, { "epoch": 113.82866043613707, "grad_norm": 1.2734375, "learning_rate": 1.9949788675778448e-05, "loss": 0.0876, "step": 18326 }, { "epoch": 113.83489096573209, "grad_norm": 2.546875, "learning_rate": 1.9947375731423344e-05, "loss": 0.1688, "step": 18327 }, { "epoch": 113.8411214953271, "grad_norm": 1.9609375, "learning_rate": 1.9944962836140346e-05, "loss": 0.1411, "step": 18328 }, { "epoch": 113.84735202492212, "grad_norm": 2.0625, "learning_rate": 1.9942549989952876e-05, "loss": 0.1519, "step": 18329 }, { "epoch": 113.85358255451713, "grad_norm": 2.265625, "learning_rate": 1.994013719288436e-05, "loss": 0.1199, "step": 18330 }, { "epoch": 113.85981308411215, "grad_norm": 1.4765625, "learning_rate": 1.9937724444958255e-05, "loss": 0.1246, "step": 18331 }, { "epoch": 113.86604361370716, "grad_norm": 2.53125, "learning_rate": 1.993531174619798e-05, "loss": 0.1967, "step": 18332 }, { "epoch": 113.87227414330218, "grad_norm": 3.609375, "learning_rate": 1.993289909662696e-05, "loss": 0.1734, "step": 18333 }, { "epoch": 113.8785046728972, "grad_norm": 4.03125, "learning_rate": 1.9930486496268652e-05, "loss": 0.2403, "step": 18334 }, { "epoch": 113.88473520249221, "grad_norm": 2.484375, "learning_rate": 1.992807394514647e-05, "loss": 0.1871, "step": 18335 }, { "epoch": 113.89096573208722, "grad_norm": 2.109375, "learning_rate": 1.9925661443283842e-05, "loss": 0.2058, "step": 18336 }, { "epoch": 113.89719626168224, "grad_norm": 1.7578125, "learning_rate": 1.9923248990704203e-05, "loss": 0.1346, "step": 18337 }, { "epoch": 113.90342679127725, "grad_norm": 2.578125, "learning_rate": 1.9920836587430985e-05, "loss": 0.0956, "step": 18338 }, { "epoch": 113.90965732087227, "grad_norm": 2.21875, "learning_rate": 1.9918424233487615e-05, "loss": 0.1355, "step": 18339 }, { "epoch": 113.91588785046729, "grad_norm": 3.703125, "learning_rate": 1.9916011928897534e-05, "loss": 0.1046, "step": 18340 }, { "epoch": 113.9221183800623, "grad_norm": 2.953125, "learning_rate": 1.991359967368416e-05, "loss": 0.202, "step": 18341 }, { "epoch": 113.92834890965732, "grad_norm": 1.5703125, "learning_rate": 1.991118746787092e-05, "loss": 0.1312, "step": 18342 }, { "epoch": 113.93457943925233, "grad_norm": 2.65625, "learning_rate": 1.9908775311481238e-05, "loss": 0.1575, "step": 18343 }, { "epoch": 113.94080996884735, "grad_norm": 1.8203125, "learning_rate": 1.9906363204538557e-05, "loss": 0.0952, "step": 18344 }, { "epoch": 113.94704049844236, "grad_norm": 2.625, "learning_rate": 1.9903951147066292e-05, "loss": 0.2802, "step": 18345 }, { "epoch": 113.95327102803738, "grad_norm": 3.40625, "learning_rate": 1.9901539139087864e-05, "loss": 0.3986, "step": 18346 }, { "epoch": 113.9595015576324, "grad_norm": 1.4765625, "learning_rate": 1.989912718062672e-05, "loss": 0.1504, "step": 18347 }, { "epoch": 113.96573208722741, "grad_norm": 3.21875, "learning_rate": 1.9896715271706272e-05, "loss": 0.2422, "step": 18348 }, { "epoch": 113.97196261682242, "grad_norm": 1.6171875, "learning_rate": 1.9894303412349942e-05, "loss": 0.1498, "step": 18349 }, { "epoch": 113.97819314641744, "grad_norm": 2.890625, "learning_rate": 1.989189160258115e-05, "loss": 0.1399, "step": 18350 }, { "epoch": 113.98442367601245, "grad_norm": 3.09375, "learning_rate": 1.9889479842423335e-05, "loss": 0.2326, "step": 18351 }, { "epoch": 113.99065420560747, "grad_norm": 2.9375, "learning_rate": 1.988706813189991e-05, "loss": 0.2372, "step": 18352 }, { "epoch": 113.99688473520249, "grad_norm": 0.953125, "learning_rate": 1.9884656471034306e-05, "loss": 0.1157, "step": 18353 }, { "epoch": 114.0, "grad_norm": 2.15625, "learning_rate": 1.9882244859849943e-05, "loss": 0.0945, "step": 18354 }, { "epoch": 114.00623052959502, "grad_norm": 2.1875, "learning_rate": 1.9879833298370238e-05, "loss": 0.2235, "step": 18355 }, { "epoch": 114.01246105919003, "grad_norm": 2.875, "learning_rate": 1.987742178661861e-05, "loss": 0.2499, "step": 18356 }, { "epoch": 114.01869158878505, "grad_norm": 2.875, "learning_rate": 1.9875010324618488e-05, "loss": 0.1752, "step": 18357 }, { "epoch": 114.02492211838006, "grad_norm": 2.59375, "learning_rate": 1.9872598912393294e-05, "loss": 0.2774, "step": 18358 }, { "epoch": 114.03115264797508, "grad_norm": 2.859375, "learning_rate": 1.987018754996644e-05, "loss": 0.2486, "step": 18359 }, { "epoch": 114.03738317757009, "grad_norm": 1.9296875, "learning_rate": 1.9867776237361364e-05, "loss": 0.1455, "step": 18360 }, { "epoch": 114.04361370716511, "grad_norm": 1.640625, "learning_rate": 1.986536497460145e-05, "loss": 0.09, "step": 18361 }, { "epoch": 114.04984423676012, "grad_norm": 2.71875, "learning_rate": 1.986295376171015e-05, "loss": 0.1539, "step": 18362 }, { "epoch": 114.05607476635514, "grad_norm": 1.09375, "learning_rate": 1.9860542598710865e-05, "loss": 0.114, "step": 18363 }, { "epoch": 114.06230529595015, "grad_norm": 2.609375, "learning_rate": 1.985813148562702e-05, "loss": 0.1394, "step": 18364 }, { "epoch": 114.06853582554517, "grad_norm": 3.515625, "learning_rate": 1.985572042248203e-05, "loss": 0.2279, "step": 18365 }, { "epoch": 114.07476635514018, "grad_norm": 1.3984375, "learning_rate": 1.985330940929932e-05, "loss": 0.105, "step": 18366 }, { "epoch": 114.0809968847352, "grad_norm": 2.90625, "learning_rate": 1.9850898446102283e-05, "loss": 0.163, "step": 18367 }, { "epoch": 114.08722741433021, "grad_norm": 3.1875, "learning_rate": 1.9848487532914356e-05, "loss": 0.2487, "step": 18368 }, { "epoch": 114.09345794392523, "grad_norm": 2.6875, "learning_rate": 1.9846076669758947e-05, "loss": 0.2354, "step": 18369 }, { "epoch": 114.09968847352025, "grad_norm": 2.0625, "learning_rate": 1.984366585665947e-05, "loss": 0.1326, "step": 18370 }, { "epoch": 114.10591900311526, "grad_norm": 2.234375, "learning_rate": 1.9841255093639343e-05, "loss": 0.2026, "step": 18371 }, { "epoch": 114.11214953271028, "grad_norm": 2.46875, "learning_rate": 1.9838844380721975e-05, "loss": 0.1786, "step": 18372 }, { "epoch": 114.11838006230529, "grad_norm": 1.2578125, "learning_rate": 1.9836433717930793e-05, "loss": 0.0966, "step": 18373 }, { "epoch": 114.1246105919003, "grad_norm": 1.9140625, "learning_rate": 1.9834023105289187e-05, "loss": 0.0939, "step": 18374 }, { "epoch": 114.13084112149532, "grad_norm": 3.3125, "learning_rate": 1.9831612542820588e-05, "loss": 0.2655, "step": 18375 }, { "epoch": 114.13707165109034, "grad_norm": 2.1875, "learning_rate": 1.9829202030548394e-05, "loss": 0.1596, "step": 18376 }, { "epoch": 114.14330218068535, "grad_norm": 2.015625, "learning_rate": 1.982679156849603e-05, "loss": 0.0965, "step": 18377 }, { "epoch": 114.14953271028037, "grad_norm": 1.484375, "learning_rate": 1.9824381156686893e-05, "loss": 0.1117, "step": 18378 }, { "epoch": 114.15576323987538, "grad_norm": 2.4375, "learning_rate": 1.9821970795144416e-05, "loss": 0.2647, "step": 18379 }, { "epoch": 114.1619937694704, "grad_norm": 2.203125, "learning_rate": 1.981956048389198e-05, "loss": 0.2239, "step": 18380 }, { "epoch": 114.16822429906541, "grad_norm": 1.59375, "learning_rate": 1.981715022295301e-05, "loss": 0.1202, "step": 18381 }, { "epoch": 114.17445482866043, "grad_norm": 2.234375, "learning_rate": 1.9814740012350917e-05, "loss": 0.1151, "step": 18382 }, { "epoch": 114.18068535825545, "grad_norm": 1.3984375, "learning_rate": 1.9812329852109096e-05, "loss": 0.1083, "step": 18383 }, { "epoch": 114.18691588785046, "grad_norm": 2.203125, "learning_rate": 1.9809919742250974e-05, "loss": 0.1836, "step": 18384 }, { "epoch": 114.19314641744548, "grad_norm": 1.8125, "learning_rate": 1.980750968279995e-05, "loss": 0.1477, "step": 18385 }, { "epoch": 114.19937694704049, "grad_norm": 2.53125, "learning_rate": 1.980509967377943e-05, "loss": 0.2004, "step": 18386 }, { "epoch": 114.20560747663552, "grad_norm": 1.515625, "learning_rate": 1.9802689715212814e-05, "loss": 0.1297, "step": 18387 }, { "epoch": 114.21183800623054, "grad_norm": 1.34375, "learning_rate": 1.9800279807123516e-05, "loss": 0.1344, "step": 18388 }, { "epoch": 114.21806853582555, "grad_norm": 2.71875, "learning_rate": 1.979786994953494e-05, "loss": 0.3126, "step": 18389 }, { "epoch": 114.22429906542057, "grad_norm": 3.1875, "learning_rate": 1.979546014247049e-05, "loss": 0.2182, "step": 18390 }, { "epoch": 114.23052959501558, "grad_norm": 2.703125, "learning_rate": 1.9793050385953576e-05, "loss": 0.2544, "step": 18391 }, { "epoch": 114.2367601246106, "grad_norm": 3.109375, "learning_rate": 1.97906406800076e-05, "loss": 0.1542, "step": 18392 }, { "epoch": 114.24299065420561, "grad_norm": 1.265625, "learning_rate": 1.9788231024655953e-05, "loss": 0.1141, "step": 18393 }, { "epoch": 114.24922118380063, "grad_norm": 2.28125, "learning_rate": 1.9785821419922056e-05, "loss": 0.2, "step": 18394 }, { "epoch": 114.25545171339564, "grad_norm": 3.0, "learning_rate": 1.97834118658293e-05, "loss": 0.2165, "step": 18395 }, { "epoch": 114.26168224299066, "grad_norm": 2.875, "learning_rate": 1.978100236240109e-05, "loss": 0.2948, "step": 18396 }, { "epoch": 114.26791277258567, "grad_norm": 2.890625, "learning_rate": 1.977859290966083e-05, "loss": 0.2031, "step": 18397 }, { "epoch": 114.27414330218069, "grad_norm": 3.328125, "learning_rate": 1.9776183507631926e-05, "loss": 0.2325, "step": 18398 }, { "epoch": 114.2803738317757, "grad_norm": 3.59375, "learning_rate": 1.9773774156337765e-05, "loss": 0.3077, "step": 18399 }, { "epoch": 114.28660436137072, "grad_norm": 2.875, "learning_rate": 1.977136485580175e-05, "loss": 0.1499, "step": 18400 }, { "epoch": 114.29283489096574, "grad_norm": 3.078125, "learning_rate": 1.976895560604729e-05, "loss": 0.2798, "step": 18401 }, { "epoch": 114.29906542056075, "grad_norm": 2.046875, "learning_rate": 1.9766546407097774e-05, "loss": 0.1883, "step": 18402 }, { "epoch": 114.30529595015577, "grad_norm": 2.28125, "learning_rate": 1.976413725897661e-05, "loss": 0.2128, "step": 18403 }, { "epoch": 114.31152647975078, "grad_norm": 2.03125, "learning_rate": 1.97617281617072e-05, "loss": 0.1261, "step": 18404 }, { "epoch": 114.3177570093458, "grad_norm": 1.46875, "learning_rate": 1.9759319115312925e-05, "loss": 0.0905, "step": 18405 }, { "epoch": 114.32398753894081, "grad_norm": 2.0, "learning_rate": 1.9756910119817186e-05, "loss": 0.1722, "step": 18406 }, { "epoch": 114.33021806853583, "grad_norm": 1.6796875, "learning_rate": 1.9754501175243387e-05, "loss": 0.0948, "step": 18407 }, { "epoch": 114.33644859813084, "grad_norm": 1.5390625, "learning_rate": 1.975209228161492e-05, "loss": 0.1232, "step": 18408 }, { "epoch": 114.34267912772586, "grad_norm": 2.109375, "learning_rate": 1.974968343895518e-05, "loss": 0.1615, "step": 18409 }, { "epoch": 114.34890965732087, "grad_norm": 1.390625, "learning_rate": 1.974727464728757e-05, "loss": 0.1114, "step": 18410 }, { "epoch": 114.35514018691589, "grad_norm": 3.171875, "learning_rate": 1.9744865906635483e-05, "loss": 0.24, "step": 18411 }, { "epoch": 114.3613707165109, "grad_norm": 2.78125, "learning_rate": 1.97424572170223e-05, "loss": 0.3513, "step": 18412 }, { "epoch": 114.36760124610592, "grad_norm": 1.828125, "learning_rate": 1.974004857847142e-05, "loss": 0.1716, "step": 18413 }, { "epoch": 114.37383177570094, "grad_norm": 2.71875, "learning_rate": 1.9737639991006246e-05, "loss": 0.2229, "step": 18414 }, { "epoch": 114.38006230529595, "grad_norm": 1.2421875, "learning_rate": 1.9735231454650168e-05, "loss": 0.1009, "step": 18415 }, { "epoch": 114.38629283489097, "grad_norm": 1.6640625, "learning_rate": 1.973282296942656e-05, "loss": 0.1328, "step": 18416 }, { "epoch": 114.39252336448598, "grad_norm": 3.265625, "learning_rate": 1.9730414535358848e-05, "loss": 0.3082, "step": 18417 }, { "epoch": 114.398753894081, "grad_norm": 1.1171875, "learning_rate": 1.9728006152470394e-05, "loss": 0.0853, "step": 18418 }, { "epoch": 114.40498442367601, "grad_norm": 2.6875, "learning_rate": 1.9725597820784597e-05, "loss": 0.2392, "step": 18419 }, { "epoch": 114.41121495327103, "grad_norm": 1.453125, "learning_rate": 1.9723189540324846e-05, "loss": 0.1313, "step": 18420 }, { "epoch": 114.41744548286604, "grad_norm": 3.5, "learning_rate": 1.9720781311114535e-05, "loss": 0.2189, "step": 18421 }, { "epoch": 114.42367601246106, "grad_norm": 3.0625, "learning_rate": 1.971837313317705e-05, "loss": 0.2053, "step": 18422 }, { "epoch": 114.42990654205607, "grad_norm": 2.859375, "learning_rate": 1.9715965006535784e-05, "loss": 0.199, "step": 18423 }, { "epoch": 114.43613707165109, "grad_norm": 1.5, "learning_rate": 1.9713556931214128e-05, "loss": 0.1196, "step": 18424 }, { "epoch": 114.4423676012461, "grad_norm": 0.98046875, "learning_rate": 1.9711148907235455e-05, "loss": 0.0831, "step": 18425 }, { "epoch": 114.44859813084112, "grad_norm": 1.2265625, "learning_rate": 1.970874093462316e-05, "loss": 0.1149, "step": 18426 }, { "epoch": 114.45482866043614, "grad_norm": 1.546875, "learning_rate": 1.9706333013400635e-05, "loss": 0.1321, "step": 18427 }, { "epoch": 114.46105919003115, "grad_norm": 2.28125, "learning_rate": 1.9703925143591263e-05, "loss": 0.1457, "step": 18428 }, { "epoch": 114.46728971962617, "grad_norm": 2.6875, "learning_rate": 1.970151732521842e-05, "loss": 0.151, "step": 18429 }, { "epoch": 114.47352024922118, "grad_norm": 1.3515625, "learning_rate": 1.969910955830551e-05, "loss": 0.1176, "step": 18430 }, { "epoch": 114.4797507788162, "grad_norm": 1.4921875, "learning_rate": 1.9696701842875906e-05, "loss": 0.1008, "step": 18431 }, { "epoch": 114.48598130841121, "grad_norm": 3.65625, "learning_rate": 1.9694294178952992e-05, "loss": 0.1829, "step": 18432 }, { "epoch": 114.49221183800623, "grad_norm": 2.578125, "learning_rate": 1.969188656656015e-05, "loss": 0.2472, "step": 18433 }, { "epoch": 114.49844236760124, "grad_norm": 3.171875, "learning_rate": 1.968947900572077e-05, "loss": 0.1714, "step": 18434 }, { "epoch": 114.50467289719626, "grad_norm": 1.7890625, "learning_rate": 1.9687071496458225e-05, "loss": 0.178, "step": 18435 }, { "epoch": 114.51090342679127, "grad_norm": 1.671875, "learning_rate": 1.9684664038795925e-05, "loss": 0.1081, "step": 18436 }, { "epoch": 114.51713395638629, "grad_norm": 1.1328125, "learning_rate": 1.968225663275721e-05, "loss": 0.1166, "step": 18437 }, { "epoch": 114.5233644859813, "grad_norm": 1.8125, "learning_rate": 1.9679849278365487e-05, "loss": 0.1408, "step": 18438 }, { "epoch": 114.52959501557632, "grad_norm": 3.453125, "learning_rate": 1.9677441975644127e-05, "loss": 0.165, "step": 18439 }, { "epoch": 114.53582554517133, "grad_norm": 2.921875, "learning_rate": 1.967503472461652e-05, "loss": 0.1547, "step": 18440 }, { "epoch": 114.54205607476635, "grad_norm": 2.6875, "learning_rate": 1.9672627525306038e-05, "loss": 0.1385, "step": 18441 }, { "epoch": 114.54828660436137, "grad_norm": 3.96875, "learning_rate": 1.967022037773606e-05, "loss": 0.2976, "step": 18442 }, { "epoch": 114.55451713395638, "grad_norm": 2.71875, "learning_rate": 1.9667813281929974e-05, "loss": 0.2038, "step": 18443 }, { "epoch": 114.5607476635514, "grad_norm": 2.5, "learning_rate": 1.966540623791115e-05, "loss": 0.1489, "step": 18444 }, { "epoch": 114.56697819314641, "grad_norm": 2.703125, "learning_rate": 1.966299924570297e-05, "loss": 0.1472, "step": 18445 }, { "epoch": 114.57320872274143, "grad_norm": 1.4453125, "learning_rate": 1.9660592305328795e-05, "loss": 0.1192, "step": 18446 }, { "epoch": 114.57943925233644, "grad_norm": 2.46875, "learning_rate": 1.9658185416812025e-05, "loss": 0.256, "step": 18447 }, { "epoch": 114.58566978193146, "grad_norm": 2.28125, "learning_rate": 1.965577858017602e-05, "loss": 0.1517, "step": 18448 }, { "epoch": 114.59190031152647, "grad_norm": 2.078125, "learning_rate": 1.9653371795444176e-05, "loss": 0.231, "step": 18449 }, { "epoch": 114.59813084112149, "grad_norm": 1.875, "learning_rate": 1.9650965062639838e-05, "loss": 0.143, "step": 18450 }, { "epoch": 114.60436137071652, "grad_norm": 2.0625, "learning_rate": 1.96485583817864e-05, "loss": 0.1475, "step": 18451 }, { "epoch": 114.61059190031153, "grad_norm": 1.7265625, "learning_rate": 1.9646151752907232e-05, "loss": 0.1456, "step": 18452 }, { "epoch": 114.61682242990655, "grad_norm": 1.4921875, "learning_rate": 1.9643745176025714e-05, "loss": 0.1063, "step": 18453 }, { "epoch": 114.62305295950156, "grad_norm": 1.9453125, "learning_rate": 1.964133865116521e-05, "loss": 0.1777, "step": 18454 }, { "epoch": 114.62928348909658, "grad_norm": 2.875, "learning_rate": 1.9638932178349103e-05, "loss": 0.199, "step": 18455 }, { "epoch": 114.6355140186916, "grad_norm": 0.890625, "learning_rate": 1.9636525757600755e-05, "loss": 0.0899, "step": 18456 }, { "epoch": 114.64174454828661, "grad_norm": 2.5625, "learning_rate": 1.963411938894353e-05, "loss": 0.1238, "step": 18457 }, { "epoch": 114.64797507788163, "grad_norm": 0.9375, "learning_rate": 1.963171307240082e-05, "loss": 0.0912, "step": 18458 }, { "epoch": 114.65420560747664, "grad_norm": 1.484375, "learning_rate": 1.962930680799598e-05, "loss": 0.1274, "step": 18459 }, { "epoch": 114.66043613707166, "grad_norm": 1.65625, "learning_rate": 1.962690059575239e-05, "loss": 0.1443, "step": 18460 }, { "epoch": 114.66666666666667, "grad_norm": 2.46875, "learning_rate": 1.9624494435693418e-05, "loss": 0.1939, "step": 18461 }, { "epoch": 114.67289719626169, "grad_norm": 1.234375, "learning_rate": 1.9622088327842432e-05, "loss": 0.1243, "step": 18462 }, { "epoch": 114.6791277258567, "grad_norm": 2.015625, "learning_rate": 1.961968227222279e-05, "loss": 0.195, "step": 18463 }, { "epoch": 114.68535825545172, "grad_norm": 1.5078125, "learning_rate": 1.9617276268857877e-05, "loss": 0.1297, "step": 18464 }, { "epoch": 114.69158878504673, "grad_norm": 3.34375, "learning_rate": 1.961487031777105e-05, "loss": 0.2797, "step": 18465 }, { "epoch": 114.69781931464175, "grad_norm": 3.5, "learning_rate": 1.9612464418985678e-05, "loss": 0.3022, "step": 18466 }, { "epoch": 114.70404984423676, "grad_norm": 2.015625, "learning_rate": 1.9610058572525126e-05, "loss": 0.163, "step": 18467 }, { "epoch": 114.71028037383178, "grad_norm": 1.5234375, "learning_rate": 1.9607652778412776e-05, "loss": 0.1151, "step": 18468 }, { "epoch": 114.7165109034268, "grad_norm": 2.25, "learning_rate": 1.960524703667197e-05, "loss": 0.1605, "step": 18469 }, { "epoch": 114.72274143302181, "grad_norm": 2.859375, "learning_rate": 1.9602841347326077e-05, "loss": 0.2368, "step": 18470 }, { "epoch": 114.72897196261682, "grad_norm": 2.109375, "learning_rate": 1.9600435710398478e-05, "loss": 0.1584, "step": 18471 }, { "epoch": 114.73520249221184, "grad_norm": 1.96875, "learning_rate": 1.9598030125912518e-05, "loss": 0.1034, "step": 18472 }, { "epoch": 114.74143302180686, "grad_norm": 1.578125, "learning_rate": 1.9595624593891572e-05, "loss": 0.1422, "step": 18473 }, { "epoch": 114.74766355140187, "grad_norm": 1.859375, "learning_rate": 1.959321911435901e-05, "loss": 0.0947, "step": 18474 }, { "epoch": 114.75389408099689, "grad_norm": 1.0859375, "learning_rate": 1.9590813687338178e-05, "loss": 0.099, "step": 18475 }, { "epoch": 114.7601246105919, "grad_norm": 1.8125, "learning_rate": 1.958840831285244e-05, "loss": 0.1162, "step": 18476 }, { "epoch": 114.76635514018692, "grad_norm": 3.046875, "learning_rate": 1.9586002990925167e-05, "loss": 0.3163, "step": 18477 }, { "epoch": 114.77258566978193, "grad_norm": 2.53125, "learning_rate": 1.9583597721579716e-05, "loss": 0.2066, "step": 18478 }, { "epoch": 114.77881619937695, "grad_norm": 3.296875, "learning_rate": 1.958119250483944e-05, "loss": 0.3158, "step": 18479 }, { "epoch": 114.78504672897196, "grad_norm": 1.8515625, "learning_rate": 1.957878734072771e-05, "loss": 0.1592, "step": 18480 }, { "epoch": 114.79127725856698, "grad_norm": 1.5234375, "learning_rate": 1.957638222926789e-05, "loss": 0.1128, "step": 18481 }, { "epoch": 114.797507788162, "grad_norm": 2.171875, "learning_rate": 1.957397717048332e-05, "loss": 0.1464, "step": 18482 }, { "epoch": 114.80373831775701, "grad_norm": 3.328125, "learning_rate": 1.9571572164397366e-05, "loss": 0.3202, "step": 18483 }, { "epoch": 114.80996884735202, "grad_norm": 2.046875, "learning_rate": 1.9569167211033392e-05, "loss": 0.1225, "step": 18484 }, { "epoch": 114.81619937694704, "grad_norm": 1.9140625, "learning_rate": 1.956676231041475e-05, "loss": 0.119, "step": 18485 }, { "epoch": 114.82242990654206, "grad_norm": 2.0625, "learning_rate": 1.95643574625648e-05, "loss": 0.1471, "step": 18486 }, { "epoch": 114.82866043613707, "grad_norm": 2.65625, "learning_rate": 1.95619526675069e-05, "loss": 0.206, "step": 18487 }, { "epoch": 114.83489096573209, "grad_norm": 2.546875, "learning_rate": 1.9559547925264405e-05, "loss": 0.0921, "step": 18488 }, { "epoch": 114.8411214953271, "grad_norm": 2.53125, "learning_rate": 1.9557143235860658e-05, "loss": 0.2342, "step": 18489 }, { "epoch": 114.84735202492212, "grad_norm": 3.625, "learning_rate": 1.9554738599319027e-05, "loss": 0.1879, "step": 18490 }, { "epoch": 114.85358255451713, "grad_norm": 3.78125, "learning_rate": 1.9552334015662868e-05, "loss": 0.2246, "step": 18491 }, { "epoch": 114.85981308411215, "grad_norm": 1.6640625, "learning_rate": 1.9549929484915524e-05, "loss": 0.1188, "step": 18492 }, { "epoch": 114.86604361370716, "grad_norm": 3.109375, "learning_rate": 1.954752500710036e-05, "loss": 0.2269, "step": 18493 }, { "epoch": 114.87227414330218, "grad_norm": 1.640625, "learning_rate": 1.9545120582240726e-05, "loss": 0.1343, "step": 18494 }, { "epoch": 114.8785046728972, "grad_norm": 1.4609375, "learning_rate": 1.9542716210359964e-05, "loss": 0.123, "step": 18495 }, { "epoch": 114.88473520249221, "grad_norm": 3.546875, "learning_rate": 1.954031189148143e-05, "loss": 0.2717, "step": 18496 }, { "epoch": 114.89096573208722, "grad_norm": 3.4375, "learning_rate": 1.9537907625628486e-05, "loss": 0.1015, "step": 18497 }, { "epoch": 114.89719626168224, "grad_norm": 1.96875, "learning_rate": 1.953550341282447e-05, "loss": 0.1183, "step": 18498 }, { "epoch": 114.90342679127725, "grad_norm": 2.5625, "learning_rate": 1.9533099253092738e-05, "loss": 0.21, "step": 18499 }, { "epoch": 114.90965732087227, "grad_norm": 2.9375, "learning_rate": 1.9530695146456647e-05, "loss": 0.13, "step": 18500 }, { "epoch": 114.91588785046729, "grad_norm": 1.03125, "learning_rate": 1.952829109293953e-05, "loss": 0.0959, "step": 18501 }, { "epoch": 114.9221183800623, "grad_norm": 1.4375, "learning_rate": 1.9525887092564746e-05, "loss": 0.1384, "step": 18502 }, { "epoch": 114.92834890965732, "grad_norm": 1.96875, "learning_rate": 1.9523483145355635e-05, "loss": 0.1264, "step": 18503 }, { "epoch": 114.93457943925233, "grad_norm": 2.859375, "learning_rate": 1.9521079251335554e-05, "loss": 0.2646, "step": 18504 }, { "epoch": 114.94080996884735, "grad_norm": 2.703125, "learning_rate": 1.9518675410527845e-05, "loss": 0.1555, "step": 18505 }, { "epoch": 114.94704049844236, "grad_norm": 3.046875, "learning_rate": 1.951627162295587e-05, "loss": 0.2557, "step": 18506 }, { "epoch": 114.95327102803738, "grad_norm": 1.1328125, "learning_rate": 1.9513867888642945e-05, "loss": 0.0975, "step": 18507 }, { "epoch": 114.9595015576324, "grad_norm": 3.359375, "learning_rate": 1.951146420761244e-05, "loss": 0.1294, "step": 18508 }, { "epoch": 114.96573208722741, "grad_norm": 3.1875, "learning_rate": 1.9509060579887683e-05, "loss": 0.3174, "step": 18509 }, { "epoch": 114.97196261682242, "grad_norm": 1.703125, "learning_rate": 1.9506657005492036e-05, "loss": 0.1203, "step": 18510 }, { "epoch": 114.97819314641744, "grad_norm": 1.09375, "learning_rate": 1.950425348444883e-05, "loss": 0.1042, "step": 18511 }, { "epoch": 114.98442367601245, "grad_norm": 1.890625, "learning_rate": 1.950185001678141e-05, "loss": 0.1476, "step": 18512 }, { "epoch": 114.99065420560747, "grad_norm": 1.1328125, "learning_rate": 1.949944660251313e-05, "loss": 0.1147, "step": 18513 }, { "epoch": 114.99688473520249, "grad_norm": 1.484375, "learning_rate": 1.949704324166732e-05, "loss": 0.1385, "step": 18514 }, { "epoch": 115.0, "grad_norm": 2.234375, "learning_rate": 1.9494639934267327e-05, "loss": 0.1236, "step": 18515 }, { "epoch": 115.00623052959502, "grad_norm": 3.28125, "learning_rate": 1.9492236680336485e-05, "loss": 0.2171, "step": 18516 }, { "epoch": 115.01246105919003, "grad_norm": 2.609375, "learning_rate": 1.948983347989815e-05, "loss": 0.1871, "step": 18517 }, { "epoch": 115.01869158878505, "grad_norm": 2.734375, "learning_rate": 1.9487430332975643e-05, "loss": 0.2277, "step": 18518 }, { "epoch": 115.02492211838006, "grad_norm": 1.7109375, "learning_rate": 1.9485027239592334e-05, "loss": 0.1695, "step": 18519 }, { "epoch": 115.03115264797508, "grad_norm": 1.2578125, "learning_rate": 1.9482624199771526e-05, "loss": 0.1032, "step": 18520 }, { "epoch": 115.03738317757009, "grad_norm": 2.328125, "learning_rate": 1.948022121353658e-05, "loss": 0.2183, "step": 18521 }, { "epoch": 115.04361370716511, "grad_norm": 2.765625, "learning_rate": 1.9477818280910824e-05, "loss": 0.2301, "step": 18522 }, { "epoch": 115.04984423676012, "grad_norm": 1.0703125, "learning_rate": 1.9475415401917605e-05, "loss": 0.0905, "step": 18523 }, { "epoch": 115.05607476635514, "grad_norm": 1.828125, "learning_rate": 1.9473012576580258e-05, "loss": 0.1154, "step": 18524 }, { "epoch": 115.06230529595015, "grad_norm": 2.5625, "learning_rate": 1.947060980492212e-05, "loss": 0.2324, "step": 18525 }, { "epoch": 115.06853582554517, "grad_norm": 1.53125, "learning_rate": 1.9468207086966516e-05, "loss": 0.1141, "step": 18526 }, { "epoch": 115.07476635514018, "grad_norm": 3.046875, "learning_rate": 1.9465804422736794e-05, "loss": 0.2087, "step": 18527 }, { "epoch": 115.0809968847352, "grad_norm": 4.0, "learning_rate": 1.946340181225629e-05, "loss": 0.2379, "step": 18528 }, { "epoch": 115.08722741433021, "grad_norm": 1.4375, "learning_rate": 1.9460999255548325e-05, "loss": 0.1007, "step": 18529 }, { "epoch": 115.09345794392523, "grad_norm": 1.109375, "learning_rate": 1.9458596752636248e-05, "loss": 0.1069, "step": 18530 }, { "epoch": 115.09968847352025, "grad_norm": 1.6640625, "learning_rate": 1.9456194303543383e-05, "loss": 0.1065, "step": 18531 }, { "epoch": 115.10591900311526, "grad_norm": 1.0390625, "learning_rate": 1.9453791908293083e-05, "loss": 0.097, "step": 18532 }, { "epoch": 115.11214953271028, "grad_norm": 3.28125, "learning_rate": 1.945138956690865e-05, "loss": 0.1563, "step": 18533 }, { "epoch": 115.11838006230529, "grad_norm": 2.234375, "learning_rate": 1.9448987279413436e-05, "loss": 0.1669, "step": 18534 }, { "epoch": 115.1246105919003, "grad_norm": 2.015625, "learning_rate": 1.944658504583076e-05, "loss": 0.1341, "step": 18535 }, { "epoch": 115.13084112149532, "grad_norm": 1.4296875, "learning_rate": 1.9444182866183966e-05, "loss": 0.1214, "step": 18536 }, { "epoch": 115.13707165109034, "grad_norm": 3.28125, "learning_rate": 1.9441780740496378e-05, "loss": 0.2632, "step": 18537 }, { "epoch": 115.14330218068535, "grad_norm": 1.390625, "learning_rate": 1.9439378668791332e-05, "loss": 0.1251, "step": 18538 }, { "epoch": 115.14953271028037, "grad_norm": 3.375, "learning_rate": 1.9436976651092144e-05, "loss": 0.1856, "step": 18539 }, { "epoch": 115.15576323987538, "grad_norm": 1.6953125, "learning_rate": 1.9434574687422152e-05, "loss": 0.1556, "step": 18540 }, { "epoch": 115.1619937694704, "grad_norm": 0.96484375, "learning_rate": 1.9432172777804687e-05, "loss": 0.0992, "step": 18541 }, { "epoch": 115.16822429906541, "grad_norm": 1.421875, "learning_rate": 1.9429770922263062e-05, "loss": 0.122, "step": 18542 }, { "epoch": 115.17445482866043, "grad_norm": 2.640625, "learning_rate": 1.9427369120820625e-05, "loss": 0.3733, "step": 18543 }, { "epoch": 115.18068535825545, "grad_norm": 1.7734375, "learning_rate": 1.9424967373500687e-05, "loss": 0.1511, "step": 18544 }, { "epoch": 115.18691588785046, "grad_norm": 2.34375, "learning_rate": 1.9422565680326597e-05, "loss": 0.1926, "step": 18545 }, { "epoch": 115.19314641744548, "grad_norm": 2.609375, "learning_rate": 1.9420164041321648e-05, "loss": 0.2273, "step": 18546 }, { "epoch": 115.19937694704049, "grad_norm": 2.28125, "learning_rate": 1.9417762456509188e-05, "loss": 0.2017, "step": 18547 }, { "epoch": 115.20560747663552, "grad_norm": 2.75, "learning_rate": 1.9415360925912522e-05, "loss": 0.1833, "step": 18548 }, { "epoch": 115.21183800623054, "grad_norm": 2.46875, "learning_rate": 1.9412959449555e-05, "loss": 0.1179, "step": 18549 }, { "epoch": 115.21806853582555, "grad_norm": 1.28125, "learning_rate": 1.941055802745993e-05, "loss": 0.1126, "step": 18550 }, { "epoch": 115.22429906542057, "grad_norm": 2.46875, "learning_rate": 1.9408156659650646e-05, "loss": 0.2337, "step": 18551 }, { "epoch": 115.23052959501558, "grad_norm": 1.390625, "learning_rate": 1.9405755346150453e-05, "loss": 0.1026, "step": 18552 }, { "epoch": 115.2367601246106, "grad_norm": 1.4921875, "learning_rate": 1.940335408698268e-05, "loss": 0.1048, "step": 18553 }, { "epoch": 115.24299065420561, "grad_norm": 1.0234375, "learning_rate": 1.9400952882170654e-05, "loss": 0.1084, "step": 18554 }, { "epoch": 115.24922118380063, "grad_norm": 1.5703125, "learning_rate": 1.939855173173769e-05, "loss": 0.1318, "step": 18555 }, { "epoch": 115.25545171339564, "grad_norm": 1.53125, "learning_rate": 1.939615063570711e-05, "loss": 0.1496, "step": 18556 }, { "epoch": 115.26168224299066, "grad_norm": 3.453125, "learning_rate": 1.939374959410225e-05, "loss": 0.3077, "step": 18557 }, { "epoch": 115.26791277258567, "grad_norm": 1.8671875, "learning_rate": 1.93913486069464e-05, "loss": 0.166, "step": 18558 }, { "epoch": 115.27414330218069, "grad_norm": 1.90625, "learning_rate": 1.9388947674262894e-05, "loss": 0.1539, "step": 18559 }, { "epoch": 115.2803738317757, "grad_norm": 3.328125, "learning_rate": 1.9386546796075056e-05, "loss": 0.256, "step": 18560 }, { "epoch": 115.28660436137072, "grad_norm": 1.671875, "learning_rate": 1.9384145972406193e-05, "loss": 0.145, "step": 18561 }, { "epoch": 115.29283489096574, "grad_norm": 2.828125, "learning_rate": 1.9381745203279624e-05, "loss": 0.2084, "step": 18562 }, { "epoch": 115.29906542056075, "grad_norm": 2.265625, "learning_rate": 1.9379344488718678e-05, "loss": 0.1348, "step": 18563 }, { "epoch": 115.30529595015577, "grad_norm": 2.234375, "learning_rate": 1.937694382874666e-05, "loss": 0.1081, "step": 18564 }, { "epoch": 115.31152647975078, "grad_norm": 1.8984375, "learning_rate": 1.9374543223386888e-05, "loss": 0.1788, "step": 18565 }, { "epoch": 115.3177570093458, "grad_norm": 2.265625, "learning_rate": 1.9372142672662665e-05, "loss": 0.1729, "step": 18566 }, { "epoch": 115.32398753894081, "grad_norm": 2.53125, "learning_rate": 1.9369742176597327e-05, "loss": 0.2177, "step": 18567 }, { "epoch": 115.33021806853583, "grad_norm": 1.578125, "learning_rate": 1.9367341735214172e-05, "loss": 0.1123, "step": 18568 }, { "epoch": 115.33644859813084, "grad_norm": 2.8125, "learning_rate": 1.9364941348536528e-05, "loss": 0.1998, "step": 18569 }, { "epoch": 115.34267912772586, "grad_norm": 2.0625, "learning_rate": 1.9362541016587706e-05, "loss": 0.1344, "step": 18570 }, { "epoch": 115.34890965732087, "grad_norm": 1.234375, "learning_rate": 1.9360140739391e-05, "loss": 0.133, "step": 18571 }, { "epoch": 115.35514018691589, "grad_norm": 1.3515625, "learning_rate": 1.9357740516969736e-05, "loss": 0.0995, "step": 18572 }, { "epoch": 115.3613707165109, "grad_norm": 2.0625, "learning_rate": 1.9355340349347226e-05, "loss": 0.1088, "step": 18573 }, { "epoch": 115.36760124610592, "grad_norm": 1.328125, "learning_rate": 1.935294023654678e-05, "loss": 0.1174, "step": 18574 }, { "epoch": 115.37383177570094, "grad_norm": 2.609375, "learning_rate": 1.9350540178591703e-05, "loss": 0.2356, "step": 18575 }, { "epoch": 115.38006230529595, "grad_norm": 0.7890625, "learning_rate": 1.934814017550532e-05, "loss": 0.1017, "step": 18576 }, { "epoch": 115.38629283489097, "grad_norm": 2.171875, "learning_rate": 1.934574022731092e-05, "loss": 0.1147, "step": 18577 }, { "epoch": 115.39252336448598, "grad_norm": 3.828125, "learning_rate": 1.9343340334031825e-05, "loss": 0.3521, "step": 18578 }, { "epoch": 115.398753894081, "grad_norm": 1.859375, "learning_rate": 1.9340940495691334e-05, "loss": 0.1583, "step": 18579 }, { "epoch": 115.40498442367601, "grad_norm": 3.015625, "learning_rate": 1.9338540712312764e-05, "loss": 0.3308, "step": 18580 }, { "epoch": 115.41121495327103, "grad_norm": 2.625, "learning_rate": 1.9336140983919413e-05, "loss": 0.2236, "step": 18581 }, { "epoch": 115.41744548286604, "grad_norm": 2.234375, "learning_rate": 1.9333741310534597e-05, "loss": 0.2031, "step": 18582 }, { "epoch": 115.42367601246106, "grad_norm": 1.5, "learning_rate": 1.9331341692181628e-05, "loss": 0.1291, "step": 18583 }, { "epoch": 115.42990654205607, "grad_norm": 1.2578125, "learning_rate": 1.9328942128883793e-05, "loss": 0.0859, "step": 18584 }, { "epoch": 115.43613707165109, "grad_norm": 2.109375, "learning_rate": 1.9326542620664402e-05, "loss": 0.1682, "step": 18585 }, { "epoch": 115.4423676012461, "grad_norm": 2.03125, "learning_rate": 1.9324143167546764e-05, "loss": 0.1937, "step": 18586 }, { "epoch": 115.44859813084112, "grad_norm": 3.046875, "learning_rate": 1.9321743769554186e-05, "loss": 0.1251, "step": 18587 }, { "epoch": 115.45482866043614, "grad_norm": 1.65625, "learning_rate": 1.931934442670996e-05, "loss": 0.1341, "step": 18588 }, { "epoch": 115.46105919003115, "grad_norm": 1.25, "learning_rate": 1.931694513903741e-05, "loss": 0.1128, "step": 18589 }, { "epoch": 115.46728971962617, "grad_norm": 1.6171875, "learning_rate": 1.931454590655982e-05, "loss": 0.2014, "step": 18590 }, { "epoch": 115.47352024922118, "grad_norm": 2.15625, "learning_rate": 1.9312146729300494e-05, "loss": 0.1132, "step": 18591 }, { "epoch": 115.4797507788162, "grad_norm": 2.65625, "learning_rate": 1.930974760728273e-05, "loss": 0.2118, "step": 18592 }, { "epoch": 115.48598130841121, "grad_norm": 2.546875, "learning_rate": 1.9307348540529842e-05, "loss": 0.1962, "step": 18593 }, { "epoch": 115.49221183800623, "grad_norm": 1.5859375, "learning_rate": 1.9304949529065117e-05, "loss": 0.1028, "step": 18594 }, { "epoch": 115.49844236760124, "grad_norm": 2.328125, "learning_rate": 1.9302550572911876e-05, "loss": 0.1825, "step": 18595 }, { "epoch": 115.50467289719626, "grad_norm": 1.78125, "learning_rate": 1.9300151672093387e-05, "loss": 0.1989, "step": 18596 }, { "epoch": 115.51090342679127, "grad_norm": 1.8828125, "learning_rate": 1.9297752826632974e-05, "loss": 0.1488, "step": 18597 }, { "epoch": 115.51713395638629, "grad_norm": 0.9140625, "learning_rate": 1.929535403655392e-05, "loss": 0.099, "step": 18598 }, { "epoch": 115.5233644859813, "grad_norm": 1.3984375, "learning_rate": 1.9292955301879524e-05, "loss": 0.1266, "step": 18599 }, { "epoch": 115.52959501557632, "grad_norm": 1.1484375, "learning_rate": 1.929055662263309e-05, "loss": 0.1129, "step": 18600 }, { "epoch": 115.53582554517133, "grad_norm": 1.9375, "learning_rate": 1.928815799883791e-05, "loss": 0.1419, "step": 18601 }, { "epoch": 115.54205607476635, "grad_norm": 1.25, "learning_rate": 1.9285759430517298e-05, "loss": 0.1173, "step": 18602 }, { "epoch": 115.54828660436137, "grad_norm": 2.0, "learning_rate": 1.9283360917694512e-05, "loss": 0.1683, "step": 18603 }, { "epoch": 115.55451713395638, "grad_norm": 2.375, "learning_rate": 1.9280962460392876e-05, "loss": 0.1545, "step": 18604 }, { "epoch": 115.5607476635514, "grad_norm": 1.8046875, "learning_rate": 1.9278564058635666e-05, "loss": 0.155, "step": 18605 }, { "epoch": 115.56697819314641, "grad_norm": 2.15625, "learning_rate": 1.92761657124462e-05, "loss": 0.1728, "step": 18606 }, { "epoch": 115.57320872274143, "grad_norm": 2.625, "learning_rate": 1.927376742184775e-05, "loss": 0.2045, "step": 18607 }, { "epoch": 115.57943925233644, "grad_norm": 2.296875, "learning_rate": 1.927136918686362e-05, "loss": 0.1785, "step": 18608 }, { "epoch": 115.58566978193146, "grad_norm": 1.109375, "learning_rate": 1.926897100751709e-05, "loss": 0.0988, "step": 18609 }, { "epoch": 115.59190031152647, "grad_norm": 2.28125, "learning_rate": 1.9266572883831463e-05, "loss": 0.1451, "step": 18610 }, { "epoch": 115.59813084112149, "grad_norm": 1.703125, "learning_rate": 1.926417481583003e-05, "loss": 0.1819, "step": 18611 }, { "epoch": 115.60436137071652, "grad_norm": 2.1875, "learning_rate": 1.9261776803536067e-05, "loss": 0.1833, "step": 18612 }, { "epoch": 115.61059190031153, "grad_norm": 1.2109375, "learning_rate": 1.925937884697288e-05, "loss": 0.1075, "step": 18613 }, { "epoch": 115.61682242990655, "grad_norm": 2.109375, "learning_rate": 1.925698094616375e-05, "loss": 0.2072, "step": 18614 }, { "epoch": 115.62305295950156, "grad_norm": 2.28125, "learning_rate": 1.9254583101131987e-05, "loss": 0.1701, "step": 18615 }, { "epoch": 115.62928348909658, "grad_norm": 1.03125, "learning_rate": 1.9252185311900846e-05, "loss": 0.1028, "step": 18616 }, { "epoch": 115.6355140186916, "grad_norm": 3.21875, "learning_rate": 1.924978757849363e-05, "loss": 0.166, "step": 18617 }, { "epoch": 115.64174454828661, "grad_norm": 2.078125, "learning_rate": 1.9247389900933623e-05, "loss": 0.1569, "step": 18618 }, { "epoch": 115.64797507788163, "grad_norm": 1.9765625, "learning_rate": 1.9244992279244124e-05, "loss": 0.1452, "step": 18619 }, { "epoch": 115.65420560747664, "grad_norm": 0.92578125, "learning_rate": 1.924259471344841e-05, "loss": 0.1132, "step": 18620 }, { "epoch": 115.66043613707166, "grad_norm": 1.84375, "learning_rate": 1.9240197203569767e-05, "loss": 0.1227, "step": 18621 }, { "epoch": 115.66666666666667, "grad_norm": 1.90625, "learning_rate": 1.9237799749631475e-05, "loss": 0.2171, "step": 18622 }, { "epoch": 115.67289719626169, "grad_norm": 2.078125, "learning_rate": 1.9235402351656825e-05, "loss": 0.1049, "step": 18623 }, { "epoch": 115.6791277258567, "grad_norm": 3.1875, "learning_rate": 1.9233005009669102e-05, "loss": 0.2659, "step": 18624 }, { "epoch": 115.68535825545172, "grad_norm": 1.0625, "learning_rate": 1.923060772369158e-05, "loss": 0.0932, "step": 18625 }, { "epoch": 115.69158878504673, "grad_norm": 1.234375, "learning_rate": 1.9228210493747557e-05, "loss": 0.096, "step": 18626 }, { "epoch": 115.69781931464175, "grad_norm": 2.90625, "learning_rate": 1.9225813319860316e-05, "loss": 0.1768, "step": 18627 }, { "epoch": 115.70404984423676, "grad_norm": 1.7265625, "learning_rate": 1.922341620205312e-05, "loss": 0.1465, "step": 18628 }, { "epoch": 115.71028037383178, "grad_norm": 1.328125, "learning_rate": 1.9221019140349253e-05, "loss": 0.1045, "step": 18629 }, { "epoch": 115.7165109034268, "grad_norm": 1.921875, "learning_rate": 1.921862213477201e-05, "loss": 0.1726, "step": 18630 }, { "epoch": 115.72274143302181, "grad_norm": 1.5625, "learning_rate": 1.9216225185344662e-05, "loss": 0.0883, "step": 18631 }, { "epoch": 115.72897196261682, "grad_norm": 1.09375, "learning_rate": 1.9213828292090498e-05, "loss": 0.1065, "step": 18632 }, { "epoch": 115.73520249221184, "grad_norm": 1.5546875, "learning_rate": 1.9211431455032786e-05, "loss": 0.1387, "step": 18633 }, { "epoch": 115.74143302180686, "grad_norm": 2.84375, "learning_rate": 1.9209034674194815e-05, "loss": 0.2624, "step": 18634 }, { "epoch": 115.74766355140187, "grad_norm": 2.046875, "learning_rate": 1.9206637949599846e-05, "loss": 0.1194, "step": 18635 }, { "epoch": 115.75389408099689, "grad_norm": 2.65625, "learning_rate": 1.9204241281271175e-05, "loss": 0.1745, "step": 18636 }, { "epoch": 115.7601246105919, "grad_norm": 1.1640625, "learning_rate": 1.9201844669232065e-05, "loss": 0.0974, "step": 18637 }, { "epoch": 115.76635514018692, "grad_norm": 2.625, "learning_rate": 1.9199448113505802e-05, "loss": 0.099, "step": 18638 }, { "epoch": 115.77258566978193, "grad_norm": 2.421875, "learning_rate": 1.9197051614115657e-05, "loss": 0.1405, "step": 18639 }, { "epoch": 115.77881619937695, "grad_norm": 1.3515625, "learning_rate": 1.9194655171084915e-05, "loss": 0.1082, "step": 18640 }, { "epoch": 115.78504672897196, "grad_norm": 2.8125, "learning_rate": 1.919225878443684e-05, "loss": 0.1554, "step": 18641 }, { "epoch": 115.79127725856698, "grad_norm": 0.97265625, "learning_rate": 1.91898624541947e-05, "loss": 0.0911, "step": 18642 }, { "epoch": 115.797507788162, "grad_norm": 2.375, "learning_rate": 1.918746618038178e-05, "loss": 0.169, "step": 18643 }, { "epoch": 115.80373831775701, "grad_norm": 3.40625, "learning_rate": 1.918506996302135e-05, "loss": 0.1239, "step": 18644 }, { "epoch": 115.80996884735202, "grad_norm": 2.6875, "learning_rate": 1.9182673802136685e-05, "loss": 0.3077, "step": 18645 }, { "epoch": 115.81619937694704, "grad_norm": 3.0, "learning_rate": 1.9180277697751058e-05, "loss": 0.191, "step": 18646 }, { "epoch": 115.82242990654206, "grad_norm": 2.96875, "learning_rate": 1.917788164988774e-05, "loss": 0.243, "step": 18647 }, { "epoch": 115.82866043613707, "grad_norm": 2.640625, "learning_rate": 1.9175485658569992e-05, "loss": 0.2064, "step": 18648 }, { "epoch": 115.83489096573209, "grad_norm": 2.9375, "learning_rate": 1.917308972382109e-05, "loss": 0.2377, "step": 18649 }, { "epoch": 115.8411214953271, "grad_norm": 2.59375, "learning_rate": 1.9170693845664308e-05, "loss": 0.2977, "step": 18650 }, { "epoch": 115.84735202492212, "grad_norm": 2.625, "learning_rate": 1.9168298024122907e-05, "loss": 0.1484, "step": 18651 }, { "epoch": 115.85358255451713, "grad_norm": 1.6875, "learning_rate": 1.916590225922017e-05, "loss": 0.1426, "step": 18652 }, { "epoch": 115.85981308411215, "grad_norm": 3.078125, "learning_rate": 1.916350655097936e-05, "loss": 0.1106, "step": 18653 }, { "epoch": 115.86604361370716, "grad_norm": 3.515625, "learning_rate": 1.9161110899423733e-05, "loss": 0.225, "step": 18654 }, { "epoch": 115.87227414330218, "grad_norm": 2.578125, "learning_rate": 1.9158715304576562e-05, "loss": 0.1053, "step": 18655 }, { "epoch": 115.8785046728972, "grad_norm": 2.8125, "learning_rate": 1.9156319766461124e-05, "loss": 0.1838, "step": 18656 }, { "epoch": 115.88473520249221, "grad_norm": 2.078125, "learning_rate": 1.915392428510067e-05, "loss": 0.1726, "step": 18657 }, { "epoch": 115.89096573208722, "grad_norm": 0.78125, "learning_rate": 1.9151528860518472e-05, "loss": 0.0941, "step": 18658 }, { "epoch": 115.89719626168224, "grad_norm": 1.234375, "learning_rate": 1.91491334927378e-05, "loss": 0.0958, "step": 18659 }, { "epoch": 115.90342679127725, "grad_norm": 1.9921875, "learning_rate": 1.9146738181781915e-05, "loss": 0.143, "step": 18660 }, { "epoch": 115.90965732087227, "grad_norm": 1.7578125, "learning_rate": 1.9144342927674076e-05, "loss": 0.1687, "step": 18661 }, { "epoch": 115.91588785046729, "grad_norm": 1.4609375, "learning_rate": 1.914194773043754e-05, "loss": 0.109, "step": 18662 }, { "epoch": 115.9221183800623, "grad_norm": 1.8984375, "learning_rate": 1.9139552590095587e-05, "loss": 0.1402, "step": 18663 }, { "epoch": 115.92834890965732, "grad_norm": 3.59375, "learning_rate": 1.9137157506671463e-05, "loss": 0.2365, "step": 18664 }, { "epoch": 115.93457943925233, "grad_norm": 1.09375, "learning_rate": 1.9134762480188458e-05, "loss": 0.117, "step": 18665 }, { "epoch": 115.94080996884735, "grad_norm": 2.453125, "learning_rate": 1.913236751066979e-05, "loss": 0.1338, "step": 18666 }, { "epoch": 115.94704049844236, "grad_norm": 2.171875, "learning_rate": 1.912997259813875e-05, "loss": 0.2263, "step": 18667 }, { "epoch": 115.95327102803738, "grad_norm": 1.140625, "learning_rate": 1.9127577742618586e-05, "loss": 0.1082, "step": 18668 }, { "epoch": 115.9595015576324, "grad_norm": 3.015625, "learning_rate": 1.9125182944132564e-05, "loss": 0.3283, "step": 18669 }, { "epoch": 115.96573208722741, "grad_norm": 2.71875, "learning_rate": 1.912278820270394e-05, "loss": 0.322, "step": 18670 }, { "epoch": 115.97196261682242, "grad_norm": 2.5625, "learning_rate": 1.9120393518355966e-05, "loss": 0.133, "step": 18671 }, { "epoch": 115.97819314641744, "grad_norm": 1.3125, "learning_rate": 1.911799889111192e-05, "loss": 0.102, "step": 18672 }, { "epoch": 115.98442367601245, "grad_norm": 1.0703125, "learning_rate": 1.9115604320995036e-05, "loss": 0.1267, "step": 18673 }, { "epoch": 115.99065420560747, "grad_norm": 1.1640625, "learning_rate": 1.9113209808028577e-05, "loss": 0.125, "step": 18674 }, { "epoch": 115.99688473520249, "grad_norm": 2.3125, "learning_rate": 1.91108153522358e-05, "loss": 0.1619, "step": 18675 }, { "epoch": 116.0, "grad_norm": 0.83203125, "learning_rate": 1.9108420953639965e-05, "loss": 0.0546, "step": 18676 }, { "epoch": 116.00623052959502, "grad_norm": 2.09375, "learning_rate": 1.9106026612264316e-05, "loss": 0.1392, "step": 18677 }, { "epoch": 116.01246105919003, "grad_norm": 1.7109375, "learning_rate": 1.9103632328132133e-05, "loss": 0.203, "step": 18678 }, { "epoch": 116.01869158878505, "grad_norm": 2.625, "learning_rate": 1.9101238101266634e-05, "loss": 0.1984, "step": 18679 }, { "epoch": 116.02492211838006, "grad_norm": 1.7109375, "learning_rate": 1.9098843931691097e-05, "loss": 0.1267, "step": 18680 }, { "epoch": 116.03115264797508, "grad_norm": 1.9453125, "learning_rate": 1.909644981942876e-05, "loss": 0.1569, "step": 18681 }, { "epoch": 116.03738317757009, "grad_norm": 1.3359375, "learning_rate": 1.909405576450289e-05, "loss": 0.0979, "step": 18682 }, { "epoch": 116.04361370716511, "grad_norm": 2.09375, "learning_rate": 1.909166176693673e-05, "loss": 0.24, "step": 18683 }, { "epoch": 116.04984423676012, "grad_norm": 2.625, "learning_rate": 1.9089267826753525e-05, "loss": 0.2543, "step": 18684 }, { "epoch": 116.05607476635514, "grad_norm": 1.4765625, "learning_rate": 1.9086873943976547e-05, "loss": 0.0902, "step": 18685 }, { "epoch": 116.06230529595015, "grad_norm": 1.4921875, "learning_rate": 1.9084480118629023e-05, "loss": 0.1045, "step": 18686 }, { "epoch": 116.06853582554517, "grad_norm": 1.4765625, "learning_rate": 1.908208635073421e-05, "loss": 0.1246, "step": 18687 }, { "epoch": 116.07476635514018, "grad_norm": 1.3203125, "learning_rate": 1.9079692640315354e-05, "loss": 0.116, "step": 18688 }, { "epoch": 116.0809968847352, "grad_norm": 3.09375, "learning_rate": 1.907729898739571e-05, "loss": 0.2682, "step": 18689 }, { "epoch": 116.08722741433021, "grad_norm": 2.078125, "learning_rate": 1.907490539199852e-05, "loss": 0.1869, "step": 18690 }, { "epoch": 116.09345794392523, "grad_norm": 2.15625, "learning_rate": 1.9072511854147046e-05, "loss": 0.1306, "step": 18691 }, { "epoch": 116.09968847352025, "grad_norm": 1.2890625, "learning_rate": 1.9070118373864508e-05, "loss": 0.1331, "step": 18692 }, { "epoch": 116.10591900311526, "grad_norm": 1.3515625, "learning_rate": 1.9067724951174167e-05, "loss": 0.1107, "step": 18693 }, { "epoch": 116.11214953271028, "grad_norm": 2.4375, "learning_rate": 1.906533158609927e-05, "loss": 0.1958, "step": 18694 }, { "epoch": 116.11838006230529, "grad_norm": 2.125, "learning_rate": 1.9062938278663053e-05, "loss": 0.1537, "step": 18695 }, { "epoch": 116.1246105919003, "grad_norm": 2.078125, "learning_rate": 1.9060545028888775e-05, "loss": 0.2194, "step": 18696 }, { "epoch": 116.13084112149532, "grad_norm": 2.734375, "learning_rate": 1.905815183679967e-05, "loss": 0.2913, "step": 18697 }, { "epoch": 116.13707165109034, "grad_norm": 1.1875, "learning_rate": 1.905575870241898e-05, "loss": 0.0978, "step": 18698 }, { "epoch": 116.14330218068535, "grad_norm": 1.3203125, "learning_rate": 1.9053365625769948e-05, "loss": 0.1093, "step": 18699 }, { "epoch": 116.14953271028037, "grad_norm": 1.90625, "learning_rate": 1.9050972606875818e-05, "loss": 0.1483, "step": 18700 }, { "epoch": 116.15576323987538, "grad_norm": 2.0, "learning_rate": 1.9048579645759828e-05, "loss": 0.102, "step": 18701 }, { "epoch": 116.1619937694704, "grad_norm": 2.484375, "learning_rate": 1.9046186742445228e-05, "loss": 0.2443, "step": 18702 }, { "epoch": 116.16822429906541, "grad_norm": 2.234375, "learning_rate": 1.904379389695525e-05, "loss": 0.1735, "step": 18703 }, { "epoch": 116.17445482866043, "grad_norm": 2.125, "learning_rate": 1.9041401109313142e-05, "loss": 0.1464, "step": 18704 }, { "epoch": 116.18068535825545, "grad_norm": 1.984375, "learning_rate": 1.903900837954213e-05, "loss": 0.2148, "step": 18705 }, { "epoch": 116.18691588785046, "grad_norm": 1.28125, "learning_rate": 1.903661570766546e-05, "loss": 0.1279, "step": 18706 }, { "epoch": 116.19314641744548, "grad_norm": 2.03125, "learning_rate": 1.903422309370638e-05, "loss": 0.1346, "step": 18707 }, { "epoch": 116.19937694704049, "grad_norm": 1.6640625, "learning_rate": 1.9031830537688104e-05, "loss": 0.1001, "step": 18708 }, { "epoch": 116.20560747663552, "grad_norm": 2.984375, "learning_rate": 1.902943803963389e-05, "loss": 0.3163, "step": 18709 }, { "epoch": 116.21183800623054, "grad_norm": 3.390625, "learning_rate": 1.9027045599566972e-05, "loss": 0.144, "step": 18710 }, { "epoch": 116.21806853582555, "grad_norm": 1.7734375, "learning_rate": 1.9024653217510576e-05, "loss": 0.1518, "step": 18711 }, { "epoch": 116.22429906542057, "grad_norm": 2.171875, "learning_rate": 1.902226089348794e-05, "loss": 0.1971, "step": 18712 }, { "epoch": 116.23052959501558, "grad_norm": 2.46875, "learning_rate": 1.9019868627522302e-05, "loss": 0.2337, "step": 18713 }, { "epoch": 116.2367601246106, "grad_norm": 2.96875, "learning_rate": 1.9017476419636893e-05, "loss": 0.2203, "step": 18714 }, { "epoch": 116.24299065420561, "grad_norm": 2.078125, "learning_rate": 1.9015084269854956e-05, "loss": 0.1641, "step": 18715 }, { "epoch": 116.24922118380063, "grad_norm": 2.15625, "learning_rate": 1.9012692178199716e-05, "loss": 0.1719, "step": 18716 }, { "epoch": 116.25545171339564, "grad_norm": 2.71875, "learning_rate": 1.9010300144694407e-05, "loss": 0.2992, "step": 18717 }, { "epoch": 116.26168224299066, "grad_norm": 2.84375, "learning_rate": 1.9007908169362256e-05, "loss": 0.2551, "step": 18718 }, { "epoch": 116.26791277258567, "grad_norm": 2.265625, "learning_rate": 1.90055162522265e-05, "loss": 0.1489, "step": 18719 }, { "epoch": 116.27414330218069, "grad_norm": 2.546875, "learning_rate": 1.9003124393310367e-05, "loss": 0.16, "step": 18720 }, { "epoch": 116.2803738317757, "grad_norm": 1.546875, "learning_rate": 1.9000732592637085e-05, "loss": 0.0968, "step": 18721 }, { "epoch": 116.28660436137072, "grad_norm": 2.109375, "learning_rate": 1.8998340850229897e-05, "loss": 0.1698, "step": 18722 }, { "epoch": 116.29283489096574, "grad_norm": 2.453125, "learning_rate": 1.899594916611202e-05, "loss": 0.1939, "step": 18723 }, { "epoch": 116.29906542056075, "grad_norm": 2.734375, "learning_rate": 1.8993557540306684e-05, "loss": 0.2449, "step": 18724 }, { "epoch": 116.30529595015577, "grad_norm": 1.578125, "learning_rate": 1.8991165972837113e-05, "loss": 0.1256, "step": 18725 }, { "epoch": 116.31152647975078, "grad_norm": 3.4375, "learning_rate": 1.8988774463726543e-05, "loss": 0.2389, "step": 18726 }, { "epoch": 116.3177570093458, "grad_norm": 2.953125, "learning_rate": 1.8986383012998193e-05, "loss": 0.2909, "step": 18727 }, { "epoch": 116.32398753894081, "grad_norm": 2.46875, "learning_rate": 1.89839916206753e-05, "loss": 0.1872, "step": 18728 }, { "epoch": 116.33021806853583, "grad_norm": 1.765625, "learning_rate": 1.8981600286781085e-05, "loss": 0.131, "step": 18729 }, { "epoch": 116.33644859813084, "grad_norm": 1.5390625, "learning_rate": 1.8979209011338767e-05, "loss": 0.1264, "step": 18730 }, { "epoch": 116.34267912772586, "grad_norm": 1.6875, "learning_rate": 1.8976817794371568e-05, "loss": 0.1268, "step": 18731 }, { "epoch": 116.34890965732087, "grad_norm": 2.890625, "learning_rate": 1.8974426635902725e-05, "loss": 0.2352, "step": 18732 }, { "epoch": 116.35514018691589, "grad_norm": 2.3125, "learning_rate": 1.8972035535955455e-05, "loss": 0.18, "step": 18733 }, { "epoch": 116.3613707165109, "grad_norm": 1.8125, "learning_rate": 1.8969644494552972e-05, "loss": 0.2407, "step": 18734 }, { "epoch": 116.36760124610592, "grad_norm": 1.390625, "learning_rate": 1.8967253511718525e-05, "loss": 0.1293, "step": 18735 }, { "epoch": 116.37383177570094, "grad_norm": 2.46875, "learning_rate": 1.8964862587475308e-05, "loss": 0.1319, "step": 18736 }, { "epoch": 116.38006230529595, "grad_norm": 1.59375, "learning_rate": 1.8962471721846554e-05, "loss": 0.1179, "step": 18737 }, { "epoch": 116.38629283489097, "grad_norm": 2.640625, "learning_rate": 1.8960080914855475e-05, "loss": 0.2036, "step": 18738 }, { "epoch": 116.39252336448598, "grad_norm": 2.328125, "learning_rate": 1.89576901665253e-05, "loss": 0.1601, "step": 18739 }, { "epoch": 116.398753894081, "grad_norm": 2.703125, "learning_rate": 1.895529947687925e-05, "loss": 0.2342, "step": 18740 }, { "epoch": 116.40498442367601, "grad_norm": 2.203125, "learning_rate": 1.8952908845940534e-05, "loss": 0.1772, "step": 18741 }, { "epoch": 116.41121495327103, "grad_norm": 1.1640625, "learning_rate": 1.895051827373239e-05, "loss": 0.1161, "step": 18742 }, { "epoch": 116.41744548286604, "grad_norm": 1.734375, "learning_rate": 1.894812776027801e-05, "loss": 0.165, "step": 18743 }, { "epoch": 116.42367601246106, "grad_norm": 1.5234375, "learning_rate": 1.8945737305600626e-05, "loss": 0.0919, "step": 18744 }, { "epoch": 116.42990654205607, "grad_norm": 3.015625, "learning_rate": 1.8943346909723443e-05, "loss": 0.2132, "step": 18745 }, { "epoch": 116.43613707165109, "grad_norm": 1.796875, "learning_rate": 1.8940956572669692e-05, "loss": 0.1424, "step": 18746 }, { "epoch": 116.4423676012461, "grad_norm": 1.2265625, "learning_rate": 1.893856629446258e-05, "loss": 0.114, "step": 18747 }, { "epoch": 116.44859813084112, "grad_norm": 2.15625, "learning_rate": 1.893617607512534e-05, "loss": 0.166, "step": 18748 }, { "epoch": 116.45482866043614, "grad_norm": 1.3984375, "learning_rate": 1.893378591468115e-05, "loss": 0.1331, "step": 18749 }, { "epoch": 116.46105919003115, "grad_norm": 1.765625, "learning_rate": 1.893139581315325e-05, "loss": 0.133, "step": 18750 }, { "epoch": 116.46728971962617, "grad_norm": 1.8984375, "learning_rate": 1.8929005770564844e-05, "loss": 0.1504, "step": 18751 }, { "epoch": 116.47352024922118, "grad_norm": 1.3359375, "learning_rate": 1.892661578693915e-05, "loss": 0.1273, "step": 18752 }, { "epoch": 116.4797507788162, "grad_norm": 1.953125, "learning_rate": 1.892422586229938e-05, "loss": 0.1259, "step": 18753 }, { "epoch": 116.48598130841121, "grad_norm": 1.890625, "learning_rate": 1.8921835996668735e-05, "loss": 0.1539, "step": 18754 }, { "epoch": 116.49221183800623, "grad_norm": 2.53125, "learning_rate": 1.8919446190070446e-05, "loss": 0.1881, "step": 18755 }, { "epoch": 116.49844236760124, "grad_norm": 1.1328125, "learning_rate": 1.891705644252771e-05, "loss": 0.1124, "step": 18756 }, { "epoch": 116.50467289719626, "grad_norm": 2.46875, "learning_rate": 1.891466675406373e-05, "loss": 0.1973, "step": 18757 }, { "epoch": 116.51090342679127, "grad_norm": 2.109375, "learning_rate": 1.891227712470172e-05, "loss": 0.1007, "step": 18758 }, { "epoch": 116.51713395638629, "grad_norm": 1.8984375, "learning_rate": 1.89098875544649e-05, "loss": 0.1396, "step": 18759 }, { "epoch": 116.5233644859813, "grad_norm": 2.0, "learning_rate": 1.8907498043376462e-05, "loss": 0.1579, "step": 18760 }, { "epoch": 116.52959501557632, "grad_norm": 2.8125, "learning_rate": 1.8905108591459636e-05, "loss": 0.2591, "step": 18761 }, { "epoch": 116.53582554517133, "grad_norm": 2.90625, "learning_rate": 1.8902719198737602e-05, "loss": 0.2527, "step": 18762 }, { "epoch": 116.54205607476635, "grad_norm": 2.796875, "learning_rate": 1.890032986523358e-05, "loss": 0.1792, "step": 18763 }, { "epoch": 116.54828660436137, "grad_norm": 1.359375, "learning_rate": 1.8897940590970768e-05, "loss": 0.1299, "step": 18764 }, { "epoch": 116.55451713395638, "grad_norm": 2.0, "learning_rate": 1.889555137597238e-05, "loss": 0.1815, "step": 18765 }, { "epoch": 116.5607476635514, "grad_norm": 2.484375, "learning_rate": 1.889316222026162e-05, "loss": 0.2243, "step": 18766 }, { "epoch": 116.56697819314641, "grad_norm": 1.7734375, "learning_rate": 1.8890773123861694e-05, "loss": 0.135, "step": 18767 }, { "epoch": 116.57320872274143, "grad_norm": 2.46875, "learning_rate": 1.888838408679579e-05, "loss": 0.1328, "step": 18768 }, { "epoch": 116.57943925233644, "grad_norm": 1.796875, "learning_rate": 1.888599510908713e-05, "loss": 0.1038, "step": 18769 }, { "epoch": 116.58566978193146, "grad_norm": 2.140625, "learning_rate": 1.8883606190758904e-05, "loss": 0.149, "step": 18770 }, { "epoch": 116.59190031152647, "grad_norm": 1.8828125, "learning_rate": 1.888121733183431e-05, "loss": 0.1399, "step": 18771 }, { "epoch": 116.59813084112149, "grad_norm": 1.5078125, "learning_rate": 1.8878828532336568e-05, "loss": 0.1314, "step": 18772 }, { "epoch": 116.60436137071652, "grad_norm": 1.4140625, "learning_rate": 1.8876439792288857e-05, "loss": 0.1361, "step": 18773 }, { "epoch": 116.61059190031153, "grad_norm": 1.859375, "learning_rate": 1.88740511117144e-05, "loss": 0.2399, "step": 18774 }, { "epoch": 116.61682242990655, "grad_norm": 1.53125, "learning_rate": 1.8871662490636374e-05, "loss": 0.1379, "step": 18775 }, { "epoch": 116.62305295950156, "grad_norm": 1.9609375, "learning_rate": 1.886927392907799e-05, "loss": 0.1673, "step": 18776 }, { "epoch": 116.62928348909658, "grad_norm": 3.5625, "learning_rate": 1.8866885427062438e-05, "loss": 0.2663, "step": 18777 }, { "epoch": 116.6355140186916, "grad_norm": 2.171875, "learning_rate": 1.8864496984612922e-05, "loss": 0.1551, "step": 18778 }, { "epoch": 116.64174454828661, "grad_norm": 3.578125, "learning_rate": 1.886210860175264e-05, "loss": 0.2591, "step": 18779 }, { "epoch": 116.64797507788163, "grad_norm": 2.296875, "learning_rate": 1.8859720278504795e-05, "loss": 0.1245, "step": 18780 }, { "epoch": 116.65420560747664, "grad_norm": 3.375, "learning_rate": 1.8857332014892558e-05, "loss": 0.3059, "step": 18781 }, { "epoch": 116.66043613707166, "grad_norm": 1.4765625, "learning_rate": 1.885494381093915e-05, "loss": 0.1309, "step": 18782 }, { "epoch": 116.66666666666667, "grad_norm": 2.234375, "learning_rate": 1.8852555666667754e-05, "loss": 0.2796, "step": 18783 }, { "epoch": 116.67289719626169, "grad_norm": 2.5, "learning_rate": 1.8850167582101564e-05, "loss": 0.1995, "step": 18784 }, { "epoch": 116.6791277258567, "grad_norm": 1.5, "learning_rate": 1.8847779557263778e-05, "loss": 0.1273, "step": 18785 }, { "epoch": 116.68535825545172, "grad_norm": 2.09375, "learning_rate": 1.8845391592177592e-05, "loss": 0.1812, "step": 18786 }, { "epoch": 116.69158878504673, "grad_norm": 1.75, "learning_rate": 1.8843003686866187e-05, "loss": 0.1281, "step": 18787 }, { "epoch": 116.69781931464175, "grad_norm": 2.1875, "learning_rate": 1.8840615841352756e-05, "loss": 0.1387, "step": 18788 }, { "epoch": 116.70404984423676, "grad_norm": 1.796875, "learning_rate": 1.8838228055660502e-05, "loss": 0.2276, "step": 18789 }, { "epoch": 116.71028037383178, "grad_norm": 2.59375, "learning_rate": 1.883584032981261e-05, "loss": 0.1486, "step": 18790 }, { "epoch": 116.7165109034268, "grad_norm": 3.5, "learning_rate": 1.8833452663832264e-05, "loss": 0.2017, "step": 18791 }, { "epoch": 116.72274143302181, "grad_norm": 1.9765625, "learning_rate": 1.8831065057742657e-05, "loss": 0.2442, "step": 18792 }, { "epoch": 116.72897196261682, "grad_norm": 2.125, "learning_rate": 1.882867751156699e-05, "loss": 0.1912, "step": 18793 }, { "epoch": 116.73520249221184, "grad_norm": 1.2578125, "learning_rate": 1.8826290025328436e-05, "loss": 0.1036, "step": 18794 }, { "epoch": 116.74143302180686, "grad_norm": 2.328125, "learning_rate": 1.882390259905018e-05, "loss": 0.1393, "step": 18795 }, { "epoch": 116.74766355140187, "grad_norm": 1.9375, "learning_rate": 1.8821515232755425e-05, "loss": 0.1389, "step": 18796 }, { "epoch": 116.75389408099689, "grad_norm": 1.9375, "learning_rate": 1.881912792646734e-05, "loss": 0.1, "step": 18797 }, { "epoch": 116.7601246105919, "grad_norm": 2.046875, "learning_rate": 1.8816740680209126e-05, "loss": 0.1541, "step": 18798 }, { "epoch": 116.76635514018692, "grad_norm": 1.9453125, "learning_rate": 1.881435349400397e-05, "loss": 0.1567, "step": 18799 }, { "epoch": 116.77258566978193, "grad_norm": 2.03125, "learning_rate": 1.881196636787504e-05, "loss": 0.1452, "step": 18800 }, { "epoch": 116.77881619937695, "grad_norm": 0.96875, "learning_rate": 1.880957930184553e-05, "loss": 0.0869, "step": 18801 }, { "epoch": 116.78504672897196, "grad_norm": 1.8046875, "learning_rate": 1.8807192295938624e-05, "loss": 0.1992, "step": 18802 }, { "epoch": 116.79127725856698, "grad_norm": 3.671875, "learning_rate": 1.8804805350177505e-05, "loss": 0.2135, "step": 18803 }, { "epoch": 116.797507788162, "grad_norm": 3.0, "learning_rate": 1.880241846458535e-05, "loss": 0.2985, "step": 18804 }, { "epoch": 116.80373831775701, "grad_norm": 1.453125, "learning_rate": 1.8800031639185358e-05, "loss": 0.1225, "step": 18805 }, { "epoch": 116.80996884735202, "grad_norm": 2.015625, "learning_rate": 1.8797644874000692e-05, "loss": 0.1563, "step": 18806 }, { "epoch": 116.81619937694704, "grad_norm": 2.078125, "learning_rate": 1.8795258169054537e-05, "loss": 0.113, "step": 18807 }, { "epoch": 116.82242990654206, "grad_norm": 1.796875, "learning_rate": 1.879287152437007e-05, "loss": 0.1631, "step": 18808 }, { "epoch": 116.82866043613707, "grad_norm": 1.890625, "learning_rate": 1.879048493997048e-05, "loss": 0.1621, "step": 18809 }, { "epoch": 116.83489096573209, "grad_norm": 2.828125, "learning_rate": 1.8788098415878942e-05, "loss": 0.2796, "step": 18810 }, { "epoch": 116.8411214953271, "grad_norm": 1.4453125, "learning_rate": 1.8785711952118635e-05, "loss": 0.1362, "step": 18811 }, { "epoch": 116.84735202492212, "grad_norm": 1.875, "learning_rate": 1.878332554871274e-05, "loss": 0.2012, "step": 18812 }, { "epoch": 116.85358255451713, "grad_norm": 2.515625, "learning_rate": 1.8780939205684426e-05, "loss": 0.1601, "step": 18813 }, { "epoch": 116.85981308411215, "grad_norm": 2.21875, "learning_rate": 1.877855292305687e-05, "loss": 0.2475, "step": 18814 }, { "epoch": 116.86604361370716, "grad_norm": 2.0, "learning_rate": 1.8776166700853253e-05, "loss": 0.1877, "step": 18815 }, { "epoch": 116.87227414330218, "grad_norm": 2.671875, "learning_rate": 1.8773780539096754e-05, "loss": 0.2209, "step": 18816 }, { "epoch": 116.8785046728972, "grad_norm": 1.8828125, "learning_rate": 1.8771394437810532e-05, "loss": 0.1528, "step": 18817 }, { "epoch": 116.88473520249221, "grad_norm": 2.5, "learning_rate": 1.8769008397017788e-05, "loss": 0.1703, "step": 18818 }, { "epoch": 116.89096573208722, "grad_norm": 1.5703125, "learning_rate": 1.876662241674167e-05, "loss": 0.1034, "step": 18819 }, { "epoch": 116.89719626168224, "grad_norm": 2.015625, "learning_rate": 1.8764236497005365e-05, "loss": 0.1261, "step": 18820 }, { "epoch": 116.90342679127725, "grad_norm": 1.578125, "learning_rate": 1.8761850637832036e-05, "loss": 0.0949, "step": 18821 }, { "epoch": 116.90965732087227, "grad_norm": 2.09375, "learning_rate": 1.8759464839244863e-05, "loss": 0.1544, "step": 18822 }, { "epoch": 116.91588785046729, "grad_norm": 3.0, "learning_rate": 1.875707910126701e-05, "loss": 0.2868, "step": 18823 }, { "epoch": 116.9221183800623, "grad_norm": 2.078125, "learning_rate": 1.8754693423921663e-05, "loss": 0.1793, "step": 18824 }, { "epoch": 116.92834890965732, "grad_norm": 2.734375, "learning_rate": 1.8752307807231983e-05, "loss": 0.212, "step": 18825 }, { "epoch": 116.93457943925233, "grad_norm": 2.90625, "learning_rate": 1.8749922251221134e-05, "loss": 0.2179, "step": 18826 }, { "epoch": 116.94080996884735, "grad_norm": 3.046875, "learning_rate": 1.874753675591228e-05, "loss": 0.1268, "step": 18827 }, { "epoch": 116.94704049844236, "grad_norm": 4.03125, "learning_rate": 1.8745151321328614e-05, "loss": 0.1719, "step": 18828 }, { "epoch": 116.95327102803738, "grad_norm": 2.921875, "learning_rate": 1.8742765947493278e-05, "loss": 0.3103, "step": 18829 }, { "epoch": 116.9595015576324, "grad_norm": 2.765625, "learning_rate": 1.8740380634429452e-05, "loss": 0.1199, "step": 18830 }, { "epoch": 116.96573208722741, "grad_norm": 1.1484375, "learning_rate": 1.873799538216031e-05, "loss": 0.1048, "step": 18831 }, { "epoch": 116.97196261682242, "grad_norm": 2.546875, "learning_rate": 1.8735610190709002e-05, "loss": 0.2228, "step": 18832 }, { "epoch": 116.97819314641744, "grad_norm": 2.03125, "learning_rate": 1.8733225060098696e-05, "loss": 0.1681, "step": 18833 }, { "epoch": 116.98442367601245, "grad_norm": 1.9609375, "learning_rate": 1.8730839990352562e-05, "loss": 0.1374, "step": 18834 }, { "epoch": 116.99065420560747, "grad_norm": 1.9375, "learning_rate": 1.8728454981493766e-05, "loss": 0.1504, "step": 18835 }, { "epoch": 116.99688473520249, "grad_norm": 1.65625, "learning_rate": 1.8726070033545467e-05, "loss": 0.1179, "step": 18836 }, { "epoch": 117.0, "grad_norm": 0.79296875, "learning_rate": 1.872368514653084e-05, "loss": 0.0441, "step": 18837 }, { "epoch": 117.00623052959502, "grad_norm": 2.484375, "learning_rate": 1.872130032047302e-05, "loss": 0.2149, "step": 18838 }, { "epoch": 117.01246105919003, "grad_norm": 2.1875, "learning_rate": 1.87189155553952e-05, "loss": 0.2555, "step": 18839 }, { "epoch": 117.01869158878505, "grad_norm": 0.93359375, "learning_rate": 1.8716530851320525e-05, "loss": 0.0813, "step": 18840 }, { "epoch": 117.02492211838006, "grad_norm": 1.5390625, "learning_rate": 1.8714146208272147e-05, "loss": 0.148, "step": 18841 }, { "epoch": 117.03115264797508, "grad_norm": 1.5390625, "learning_rate": 1.871176162627325e-05, "loss": 0.0998, "step": 18842 }, { "epoch": 117.03738317757009, "grad_norm": 1.9765625, "learning_rate": 1.8709377105346973e-05, "loss": 0.1916, "step": 18843 }, { "epoch": 117.04361370716511, "grad_norm": 1.6953125, "learning_rate": 1.8706992645516498e-05, "loss": 0.1399, "step": 18844 }, { "epoch": 117.04984423676012, "grad_norm": 2.34375, "learning_rate": 1.8704608246804956e-05, "loss": 0.1915, "step": 18845 }, { "epoch": 117.05607476635514, "grad_norm": 1.4375, "learning_rate": 1.8702223909235518e-05, "loss": 0.1495, "step": 18846 }, { "epoch": 117.06230529595015, "grad_norm": 1.828125, "learning_rate": 1.869983963283134e-05, "loss": 0.1513, "step": 18847 }, { "epoch": 117.06853582554517, "grad_norm": 1.625, "learning_rate": 1.8697455417615577e-05, "loss": 0.1434, "step": 18848 }, { "epoch": 117.07476635514018, "grad_norm": 0.953125, "learning_rate": 1.8695071263611396e-05, "loss": 0.0791, "step": 18849 }, { "epoch": 117.0809968847352, "grad_norm": 1.3046875, "learning_rate": 1.869268717084194e-05, "loss": 0.1179, "step": 18850 }, { "epoch": 117.08722741433021, "grad_norm": 3.53125, "learning_rate": 1.869030313933036e-05, "loss": 0.2052, "step": 18851 }, { "epoch": 117.09345794392523, "grad_norm": 2.71875, "learning_rate": 1.8687919169099824e-05, "loss": 0.1783, "step": 18852 }, { "epoch": 117.09968847352025, "grad_norm": 2.359375, "learning_rate": 1.8685535260173476e-05, "loss": 0.2141, "step": 18853 }, { "epoch": 117.10591900311526, "grad_norm": 3.203125, "learning_rate": 1.868315141257447e-05, "loss": 0.1873, "step": 18854 }, { "epoch": 117.11214953271028, "grad_norm": 2.125, "learning_rate": 1.8680767626325962e-05, "loss": 0.1832, "step": 18855 }, { "epoch": 117.11838006230529, "grad_norm": 1.296875, "learning_rate": 1.8678383901451108e-05, "loss": 0.0906, "step": 18856 }, { "epoch": 117.1246105919003, "grad_norm": 2.578125, "learning_rate": 1.8676000237973048e-05, "loss": 0.1922, "step": 18857 }, { "epoch": 117.13084112149532, "grad_norm": 1.96875, "learning_rate": 1.8673616635914933e-05, "loss": 0.1566, "step": 18858 }, { "epoch": 117.13707165109034, "grad_norm": 1.890625, "learning_rate": 1.8671233095299923e-05, "loss": 0.1542, "step": 18859 }, { "epoch": 117.14330218068535, "grad_norm": 1.6015625, "learning_rate": 1.8668849616151157e-05, "loss": 0.1214, "step": 18860 }, { "epoch": 117.14953271028037, "grad_norm": 1.28125, "learning_rate": 1.8666466198491795e-05, "loss": 0.13, "step": 18861 }, { "epoch": 117.15576323987538, "grad_norm": 1.5859375, "learning_rate": 1.866408284234498e-05, "loss": 0.1274, "step": 18862 }, { "epoch": 117.1619937694704, "grad_norm": 3.03125, "learning_rate": 1.8661699547733864e-05, "loss": 0.2727, "step": 18863 }, { "epoch": 117.16822429906541, "grad_norm": 1.9375, "learning_rate": 1.8659316314681573e-05, "loss": 0.1846, "step": 18864 }, { "epoch": 117.17445482866043, "grad_norm": 2.234375, "learning_rate": 1.865693314321128e-05, "loss": 0.2104, "step": 18865 }, { "epoch": 117.18068535825545, "grad_norm": 1.28125, "learning_rate": 1.865455003334612e-05, "loss": 0.1458, "step": 18866 }, { "epoch": 117.18691588785046, "grad_norm": 3.203125, "learning_rate": 1.8652166985109236e-05, "loss": 0.2296, "step": 18867 }, { "epoch": 117.19314641744548, "grad_norm": 1.0703125, "learning_rate": 1.8649783998523776e-05, "loss": 0.092, "step": 18868 }, { "epoch": 117.19937694704049, "grad_norm": 1.296875, "learning_rate": 1.8647401073612892e-05, "loss": 0.1306, "step": 18869 }, { "epoch": 117.20560747663552, "grad_norm": 2.90625, "learning_rate": 1.864501821039971e-05, "loss": 0.2116, "step": 18870 }, { "epoch": 117.21183800623054, "grad_norm": 1.1015625, "learning_rate": 1.8642635408907377e-05, "loss": 0.0765, "step": 18871 }, { "epoch": 117.21806853582555, "grad_norm": 2.40625, "learning_rate": 1.864025266915904e-05, "loss": 0.206, "step": 18872 }, { "epoch": 117.22429906542057, "grad_norm": 2.125, "learning_rate": 1.863786999117784e-05, "loss": 0.1635, "step": 18873 }, { "epoch": 117.23052959501558, "grad_norm": 2.375, "learning_rate": 1.8635487374986926e-05, "loss": 0.2126, "step": 18874 }, { "epoch": 117.2367601246106, "grad_norm": 2.28125, "learning_rate": 1.8633104820609433e-05, "loss": 0.1828, "step": 18875 }, { "epoch": 117.24299065420561, "grad_norm": 1.5546875, "learning_rate": 1.86307223280685e-05, "loss": 0.1262, "step": 18876 }, { "epoch": 117.24922118380063, "grad_norm": 2.09375, "learning_rate": 1.862833989738725e-05, "loss": 0.0949, "step": 18877 }, { "epoch": 117.25545171339564, "grad_norm": 2.4375, "learning_rate": 1.8625957528588847e-05, "loss": 0.1785, "step": 18878 }, { "epoch": 117.26168224299066, "grad_norm": 1.7421875, "learning_rate": 1.8623575221696417e-05, "loss": 0.1683, "step": 18879 }, { "epoch": 117.26791277258567, "grad_norm": 3.265625, "learning_rate": 1.8621192976733094e-05, "loss": 0.2271, "step": 18880 }, { "epoch": 117.27414330218069, "grad_norm": 1.875, "learning_rate": 1.8618810793722028e-05, "loss": 0.1547, "step": 18881 }, { "epoch": 117.2803738317757, "grad_norm": 2.21875, "learning_rate": 1.8616428672686347e-05, "loss": 0.1166, "step": 18882 }, { "epoch": 117.28660436137072, "grad_norm": 2.515625, "learning_rate": 1.8614046613649188e-05, "loss": 0.1705, "step": 18883 }, { "epoch": 117.29283489096574, "grad_norm": 2.765625, "learning_rate": 1.8611664616633675e-05, "loss": 0.1459, "step": 18884 }, { "epoch": 117.29906542056075, "grad_norm": 3.484375, "learning_rate": 1.860928268166296e-05, "loss": 0.3204, "step": 18885 }, { "epoch": 117.30529595015577, "grad_norm": 3.125, "learning_rate": 1.8606900808760166e-05, "loss": 0.2965, "step": 18886 }, { "epoch": 117.31152647975078, "grad_norm": 1.21875, "learning_rate": 1.8604518997948425e-05, "loss": 0.0993, "step": 18887 }, { "epoch": 117.3177570093458, "grad_norm": 2.390625, "learning_rate": 1.8602137249250887e-05, "loss": 0.2127, "step": 18888 }, { "epoch": 117.32398753894081, "grad_norm": 2.09375, "learning_rate": 1.8599755562690663e-05, "loss": 0.1955, "step": 18889 }, { "epoch": 117.33021806853583, "grad_norm": 1.765625, "learning_rate": 1.8597373938290893e-05, "loss": 0.1409, "step": 18890 }, { "epoch": 117.33644859813084, "grad_norm": 1.2734375, "learning_rate": 1.8594992376074703e-05, "loss": 0.1236, "step": 18891 }, { "epoch": 117.34267912772586, "grad_norm": 1.765625, "learning_rate": 1.859261087606523e-05, "loss": 0.1445, "step": 18892 }, { "epoch": 117.34890965732087, "grad_norm": 2.40625, "learning_rate": 1.85902294382856e-05, "loss": 0.171, "step": 18893 }, { "epoch": 117.35514018691589, "grad_norm": 1.4921875, "learning_rate": 1.8587848062758944e-05, "loss": 0.103, "step": 18894 }, { "epoch": 117.3613707165109, "grad_norm": 2.578125, "learning_rate": 1.8585466749508397e-05, "loss": 0.2541, "step": 18895 }, { "epoch": 117.36760124610592, "grad_norm": 1.015625, "learning_rate": 1.8583085498557075e-05, "loss": 0.0769, "step": 18896 }, { "epoch": 117.37383177570094, "grad_norm": 2.765625, "learning_rate": 1.85807043099281e-05, "loss": 0.203, "step": 18897 }, { "epoch": 117.38006230529595, "grad_norm": 1.40625, "learning_rate": 1.8578323183644618e-05, "loss": 0.1047, "step": 18898 }, { "epoch": 117.38629283489097, "grad_norm": 3.15625, "learning_rate": 1.8575942119729738e-05, "loss": 0.2231, "step": 18899 }, { "epoch": 117.39252336448598, "grad_norm": 1.984375, "learning_rate": 1.8573561118206597e-05, "loss": 0.1518, "step": 18900 }, { "epoch": 117.398753894081, "grad_norm": 2.828125, "learning_rate": 1.8571180179098318e-05, "loss": 0.2033, "step": 18901 }, { "epoch": 117.40498442367601, "grad_norm": 2.71875, "learning_rate": 1.856879930242802e-05, "loss": 0.2084, "step": 18902 }, { "epoch": 117.41121495327103, "grad_norm": 1.328125, "learning_rate": 1.8566418488218825e-05, "loss": 0.087, "step": 18903 }, { "epoch": 117.41744548286604, "grad_norm": 1.1640625, "learning_rate": 1.8564037736493855e-05, "loss": 0.1135, "step": 18904 }, { "epoch": 117.42367601246106, "grad_norm": 1.0390625, "learning_rate": 1.8561657047276247e-05, "loss": 0.0846, "step": 18905 }, { "epoch": 117.42990654205607, "grad_norm": 3.3125, "learning_rate": 1.8559276420589105e-05, "loss": 0.2345, "step": 18906 }, { "epoch": 117.43613707165109, "grad_norm": 2.375, "learning_rate": 1.8556895856455574e-05, "loss": 0.1728, "step": 18907 }, { "epoch": 117.4423676012461, "grad_norm": 2.5625, "learning_rate": 1.8554515354898742e-05, "loss": 0.2096, "step": 18908 }, { "epoch": 117.44859813084112, "grad_norm": 1.53125, "learning_rate": 1.855213491594175e-05, "loss": 0.1304, "step": 18909 }, { "epoch": 117.45482866043614, "grad_norm": 3.359375, "learning_rate": 1.854975453960771e-05, "loss": 0.1524, "step": 18910 }, { "epoch": 117.46105919003115, "grad_norm": 1.7109375, "learning_rate": 1.8547374225919744e-05, "loss": 0.1243, "step": 18911 }, { "epoch": 117.46728971962617, "grad_norm": 1.2890625, "learning_rate": 1.8544993974900974e-05, "loss": 0.0953, "step": 18912 }, { "epoch": 117.47352024922118, "grad_norm": 1.4453125, "learning_rate": 1.8542613786574503e-05, "loss": 0.1021, "step": 18913 }, { "epoch": 117.4797507788162, "grad_norm": 2.796875, "learning_rate": 1.8540233660963474e-05, "loss": 0.2555, "step": 18914 }, { "epoch": 117.48598130841121, "grad_norm": 3.0625, "learning_rate": 1.8537853598090976e-05, "loss": 0.1914, "step": 18915 }, { "epoch": 117.49221183800623, "grad_norm": 3.15625, "learning_rate": 1.853547359798014e-05, "loss": 0.183, "step": 18916 }, { "epoch": 117.49844236760124, "grad_norm": 3.3125, "learning_rate": 1.853309366065407e-05, "loss": 0.224, "step": 18917 }, { "epoch": 117.50467289719626, "grad_norm": 2.328125, "learning_rate": 1.853071378613589e-05, "loss": 0.2454, "step": 18918 }, { "epoch": 117.51090342679127, "grad_norm": 2.109375, "learning_rate": 1.852833397444871e-05, "loss": 0.1278, "step": 18919 }, { "epoch": 117.51713395638629, "grad_norm": 1.1875, "learning_rate": 1.8525954225615654e-05, "loss": 0.113, "step": 18920 }, { "epoch": 117.5233644859813, "grad_norm": 1.28125, "learning_rate": 1.8523574539659818e-05, "loss": 0.1024, "step": 18921 }, { "epoch": 117.52959501557632, "grad_norm": 1.8515625, "learning_rate": 1.852119491660432e-05, "loss": 0.1009, "step": 18922 }, { "epoch": 117.53582554517133, "grad_norm": 1.3515625, "learning_rate": 1.8518815356472265e-05, "loss": 0.1342, "step": 18923 }, { "epoch": 117.54205607476635, "grad_norm": 2.28125, "learning_rate": 1.8516435859286777e-05, "loss": 0.1833, "step": 18924 }, { "epoch": 117.54828660436137, "grad_norm": 2.1875, "learning_rate": 1.851405642507096e-05, "loss": 0.1506, "step": 18925 }, { "epoch": 117.55451713395638, "grad_norm": 3.375, "learning_rate": 1.8511677053847932e-05, "loss": 0.0971, "step": 18926 }, { "epoch": 117.5607476635514, "grad_norm": 2.5, "learning_rate": 1.850929774564078e-05, "loss": 0.1429, "step": 18927 }, { "epoch": 117.56697819314641, "grad_norm": 2.75, "learning_rate": 1.8506918500472632e-05, "loss": 0.1889, "step": 18928 }, { "epoch": 117.57320872274143, "grad_norm": 2.703125, "learning_rate": 1.850453931836659e-05, "loss": 0.2108, "step": 18929 }, { "epoch": 117.57943925233644, "grad_norm": 2.765625, "learning_rate": 1.8502160199345753e-05, "loss": 0.1626, "step": 18930 }, { "epoch": 117.58566978193146, "grad_norm": 1.1328125, "learning_rate": 1.8499781143433242e-05, "loss": 0.1083, "step": 18931 }, { "epoch": 117.59190031152647, "grad_norm": 1.53125, "learning_rate": 1.8497402150652154e-05, "loss": 0.1182, "step": 18932 }, { "epoch": 117.59813084112149, "grad_norm": 2.96875, "learning_rate": 1.8495023221025605e-05, "loss": 0.3088, "step": 18933 }, { "epoch": 117.60436137071652, "grad_norm": 2.625, "learning_rate": 1.8492644354576676e-05, "loss": 0.3478, "step": 18934 }, { "epoch": 117.61059190031153, "grad_norm": 2.546875, "learning_rate": 1.8490265551328496e-05, "loss": 0.3013, "step": 18935 }, { "epoch": 117.61682242990655, "grad_norm": 2.453125, "learning_rate": 1.8487886811304157e-05, "loss": 0.2578, "step": 18936 }, { "epoch": 117.62305295950156, "grad_norm": 1.859375, "learning_rate": 1.8485508134526755e-05, "loss": 0.1227, "step": 18937 }, { "epoch": 117.62928348909658, "grad_norm": 2.53125, "learning_rate": 1.8483129521019407e-05, "loss": 0.3023, "step": 18938 }, { "epoch": 117.6355140186916, "grad_norm": 3.1875, "learning_rate": 1.8480750970805214e-05, "loss": 0.2332, "step": 18939 }, { "epoch": 117.64174454828661, "grad_norm": 0.796875, "learning_rate": 1.8478372483907262e-05, "loss": 0.0938, "step": 18940 }, { "epoch": 117.64797507788163, "grad_norm": 1.3828125, "learning_rate": 1.8475994060348662e-05, "loss": 0.1148, "step": 18941 }, { "epoch": 117.65420560747664, "grad_norm": 4.0, "learning_rate": 1.847361570015251e-05, "loss": 0.3491, "step": 18942 }, { "epoch": 117.66043613707166, "grad_norm": 1.671875, "learning_rate": 1.8471237403341905e-05, "loss": 0.1406, "step": 18943 }, { "epoch": 117.66666666666667, "grad_norm": 1.765625, "learning_rate": 1.846885916993995e-05, "loss": 0.1755, "step": 18944 }, { "epoch": 117.67289719626169, "grad_norm": 2.484375, "learning_rate": 1.8466480999969742e-05, "loss": 0.1126, "step": 18945 }, { "epoch": 117.6791277258567, "grad_norm": 2.625, "learning_rate": 1.846410289345438e-05, "loss": 0.1579, "step": 18946 }, { "epoch": 117.68535825545172, "grad_norm": 4.09375, "learning_rate": 1.846172485041695e-05, "loss": 0.1189, "step": 18947 }, { "epoch": 117.69158878504673, "grad_norm": 3.265625, "learning_rate": 1.845934687088056e-05, "loss": 0.1126, "step": 18948 }, { "epoch": 117.69781931464175, "grad_norm": 3.421875, "learning_rate": 1.84569689548683e-05, "loss": 0.2241, "step": 18949 }, { "epoch": 117.70404984423676, "grad_norm": 0.9140625, "learning_rate": 1.845459110240326e-05, "loss": 0.0821, "step": 18950 }, { "epoch": 117.71028037383178, "grad_norm": 2.59375, "learning_rate": 1.8452213313508545e-05, "loss": 0.2114, "step": 18951 }, { "epoch": 117.7165109034268, "grad_norm": 1.15625, "learning_rate": 1.8449835588207248e-05, "loss": 0.0816, "step": 18952 }, { "epoch": 117.72274143302181, "grad_norm": 3.0, "learning_rate": 1.8447457926522454e-05, "loss": 0.1666, "step": 18953 }, { "epoch": 117.72897196261682, "grad_norm": 2.328125, "learning_rate": 1.844508032847725e-05, "loss": 0.1976, "step": 18954 }, { "epoch": 117.73520249221184, "grad_norm": 2.25, "learning_rate": 1.844270279409474e-05, "loss": 0.1163, "step": 18955 }, { "epoch": 117.74143302180686, "grad_norm": 3.0625, "learning_rate": 1.844032532339801e-05, "loss": 0.2176, "step": 18956 }, { "epoch": 117.74766355140187, "grad_norm": 2.9375, "learning_rate": 1.8437947916410157e-05, "loss": 0.1308, "step": 18957 }, { "epoch": 117.75389408099689, "grad_norm": 1.8984375, "learning_rate": 1.843557057315427e-05, "loss": 0.103, "step": 18958 }, { "epoch": 117.7601246105919, "grad_norm": 2.6875, "learning_rate": 1.8433193293653427e-05, "loss": 0.0949, "step": 18959 }, { "epoch": 117.76635514018692, "grad_norm": 1.4375, "learning_rate": 1.8430816077930718e-05, "loss": 0.0833, "step": 18960 }, { "epoch": 117.77258566978193, "grad_norm": 2.578125, "learning_rate": 1.8428438926009244e-05, "loss": 0.2857, "step": 18961 }, { "epoch": 117.77881619937695, "grad_norm": 2.515625, "learning_rate": 1.8426061837912083e-05, "loss": 0.1773, "step": 18962 }, { "epoch": 117.78504672897196, "grad_norm": 2.3125, "learning_rate": 1.8423684813662316e-05, "loss": 0.213, "step": 18963 }, { "epoch": 117.79127725856698, "grad_norm": 1.7265625, "learning_rate": 1.842130785328305e-05, "loss": 0.128, "step": 18964 }, { "epoch": 117.797507788162, "grad_norm": 3.0625, "learning_rate": 1.8418930956797354e-05, "loss": 0.2208, "step": 18965 }, { "epoch": 117.80373831775701, "grad_norm": 2.78125, "learning_rate": 1.8416554124228312e-05, "loss": 0.1088, "step": 18966 }, { "epoch": 117.80996884735202, "grad_norm": 2.734375, "learning_rate": 1.841417735559901e-05, "loss": 0.3022, "step": 18967 }, { "epoch": 117.81619937694704, "grad_norm": 2.25, "learning_rate": 1.8411800650932544e-05, "loss": 0.1037, "step": 18968 }, { "epoch": 117.82242990654206, "grad_norm": 2.5625, "learning_rate": 1.8409424010251974e-05, "loss": 0.1759, "step": 18969 }, { "epoch": 117.82866043613707, "grad_norm": 2.71875, "learning_rate": 1.84070474335804e-05, "loss": 0.1419, "step": 18970 }, { "epoch": 117.83489096573209, "grad_norm": 2.34375, "learning_rate": 1.840467092094091e-05, "loss": 0.1523, "step": 18971 }, { "epoch": 117.8411214953271, "grad_norm": 1.8203125, "learning_rate": 1.8402294472356565e-05, "loss": 0.1572, "step": 18972 }, { "epoch": 117.84735202492212, "grad_norm": 2.578125, "learning_rate": 1.839991808785045e-05, "loss": 0.1898, "step": 18973 }, { "epoch": 117.85358255451713, "grad_norm": 1.3359375, "learning_rate": 1.8397541767445657e-05, "loss": 0.1046, "step": 18974 }, { "epoch": 117.85981308411215, "grad_norm": 1.3984375, "learning_rate": 1.8395165511165258e-05, "loss": 0.1223, "step": 18975 }, { "epoch": 117.86604361370716, "grad_norm": 2.53125, "learning_rate": 1.8392789319032328e-05, "loss": 0.1706, "step": 18976 }, { "epoch": 117.87227414330218, "grad_norm": 4.3125, "learning_rate": 1.8390413191069957e-05, "loss": 0.3548, "step": 18977 }, { "epoch": 117.8785046728972, "grad_norm": 2.3125, "learning_rate": 1.8388037127301207e-05, "loss": 0.1652, "step": 18978 }, { "epoch": 117.88473520249221, "grad_norm": 2.65625, "learning_rate": 1.8385661127749166e-05, "loss": 0.2057, "step": 18979 }, { "epoch": 117.89096573208722, "grad_norm": 3.875, "learning_rate": 1.83832851924369e-05, "loss": 0.2519, "step": 18980 }, { "epoch": 117.89719626168224, "grad_norm": 2.1875, "learning_rate": 1.8380909321387495e-05, "loss": 0.1787, "step": 18981 }, { "epoch": 117.90342679127725, "grad_norm": 1.84375, "learning_rate": 1.8378533514624026e-05, "loss": 0.086, "step": 18982 }, { "epoch": 117.90965732087227, "grad_norm": 1.9609375, "learning_rate": 1.837615777216955e-05, "loss": 0.143, "step": 18983 }, { "epoch": 117.91588785046729, "grad_norm": 1.3125, "learning_rate": 1.837378209404717e-05, "loss": 0.1212, "step": 18984 }, { "epoch": 117.9221183800623, "grad_norm": 1.21875, "learning_rate": 1.8371406480279934e-05, "loss": 0.1061, "step": 18985 }, { "epoch": 117.92834890965732, "grad_norm": 2.515625, "learning_rate": 1.836903093089093e-05, "loss": 0.2794, "step": 18986 }, { "epoch": 117.93457943925233, "grad_norm": 2.15625, "learning_rate": 1.8366655445903214e-05, "loss": 0.1613, "step": 18987 }, { "epoch": 117.94080996884735, "grad_norm": 1.2265625, "learning_rate": 1.836428002533987e-05, "loss": 0.1036, "step": 18988 }, { "epoch": 117.94704049844236, "grad_norm": 1.375, "learning_rate": 1.836190466922396e-05, "loss": 0.145, "step": 18989 }, { "epoch": 117.95327102803738, "grad_norm": 2.421875, "learning_rate": 1.8359529377578575e-05, "loss": 0.1828, "step": 18990 }, { "epoch": 117.9595015576324, "grad_norm": 1.9140625, "learning_rate": 1.835715415042675e-05, "loss": 0.1539, "step": 18991 }, { "epoch": 117.96573208722741, "grad_norm": 2.515625, "learning_rate": 1.8354778987791584e-05, "loss": 0.1446, "step": 18992 }, { "epoch": 117.97196261682242, "grad_norm": 2.28125, "learning_rate": 1.8352403889696122e-05, "loss": 0.141, "step": 18993 }, { "epoch": 117.97819314641744, "grad_norm": 1.6796875, "learning_rate": 1.835002885616345e-05, "loss": 0.1138, "step": 18994 }, { "epoch": 117.98442367601245, "grad_norm": 3.375, "learning_rate": 1.8347653887216623e-05, "loss": 0.3313, "step": 18995 }, { "epoch": 117.99065420560747, "grad_norm": 3.25, "learning_rate": 1.8345278982878718e-05, "loss": 0.2587, "step": 18996 }, { "epoch": 117.99688473520249, "grad_norm": 2.0, "learning_rate": 1.8342904143172784e-05, "loss": 0.1623, "step": 18997 }, { "epoch": 118.0, "grad_norm": 1.1328125, "learning_rate": 1.8340529368121902e-05, "loss": 0.0622, "step": 18998 }, { "epoch": 118.00623052959502, "grad_norm": 2.328125, "learning_rate": 1.8338154657749128e-05, "loss": 0.1838, "step": 18999 }, { "epoch": 118.01246105919003, "grad_norm": 1.6640625, "learning_rate": 1.8335780012077524e-05, "loss": 0.1421, "step": 19000 }, { "epoch": 118.01869158878505, "grad_norm": 1.6796875, "learning_rate": 1.833340543113016e-05, "loss": 0.1626, "step": 19001 }, { "epoch": 118.02492211838006, "grad_norm": 1.75, "learning_rate": 1.8331030914930095e-05, "loss": 0.1425, "step": 19002 }, { "epoch": 118.03115264797508, "grad_norm": 2.640625, "learning_rate": 1.8328656463500406e-05, "loss": 0.212, "step": 19003 }, { "epoch": 118.03738317757009, "grad_norm": 2.71875, "learning_rate": 1.8326282076864115e-05, "loss": 0.1902, "step": 19004 }, { "epoch": 118.04361370716511, "grad_norm": 2.078125, "learning_rate": 1.8323907755044317e-05, "loss": 0.1149, "step": 19005 }, { "epoch": 118.04984423676012, "grad_norm": 1.6640625, "learning_rate": 1.8321533498064058e-05, "loss": 0.1229, "step": 19006 }, { "epoch": 118.05607476635514, "grad_norm": 1.875, "learning_rate": 1.8319159305946403e-05, "loss": 0.0928, "step": 19007 }, { "epoch": 118.06230529595015, "grad_norm": 1.296875, "learning_rate": 1.8316785178714414e-05, "loss": 0.1288, "step": 19008 }, { "epoch": 118.06853582554517, "grad_norm": 3.5, "learning_rate": 1.8314411116391138e-05, "loss": 0.2572, "step": 19009 }, { "epoch": 118.07476635514018, "grad_norm": 1.859375, "learning_rate": 1.8312037118999636e-05, "loss": 0.148, "step": 19010 }, { "epoch": 118.0809968847352, "grad_norm": 3.625, "learning_rate": 1.830966318656297e-05, "loss": 0.3846, "step": 19011 }, { "epoch": 118.08722741433021, "grad_norm": 1.84375, "learning_rate": 1.830728931910419e-05, "loss": 0.1431, "step": 19012 }, { "epoch": 118.09345794392523, "grad_norm": 1.4140625, "learning_rate": 1.830491551664635e-05, "loss": 0.1215, "step": 19013 }, { "epoch": 118.09968847352025, "grad_norm": 1.4140625, "learning_rate": 1.8302541779212515e-05, "loss": 0.1065, "step": 19014 }, { "epoch": 118.10591900311526, "grad_norm": 1.6953125, "learning_rate": 1.8300168106825726e-05, "loss": 0.1344, "step": 19015 }, { "epoch": 118.11214953271028, "grad_norm": 2.65625, "learning_rate": 1.829779449950906e-05, "loss": 0.1093, "step": 19016 }, { "epoch": 118.11838006230529, "grad_norm": 2.21875, "learning_rate": 1.829542095728554e-05, "loss": 0.202, "step": 19017 }, { "epoch": 118.1246105919003, "grad_norm": 2.03125, "learning_rate": 1.8293047480178232e-05, "loss": 0.1162, "step": 19018 }, { "epoch": 118.13084112149532, "grad_norm": 1.9296875, "learning_rate": 1.8290674068210187e-05, "loss": 0.1695, "step": 19019 }, { "epoch": 118.13707165109034, "grad_norm": 2.484375, "learning_rate": 1.8288300721404462e-05, "loss": 0.1681, "step": 19020 }, { "epoch": 118.14330218068535, "grad_norm": 2.21875, "learning_rate": 1.82859274397841e-05, "loss": 0.1439, "step": 19021 }, { "epoch": 118.14953271028037, "grad_norm": 2.6875, "learning_rate": 1.8283554223372156e-05, "loss": 0.151, "step": 19022 }, { "epoch": 118.15576323987538, "grad_norm": 1.9453125, "learning_rate": 1.8281181072191668e-05, "loss": 0.2344, "step": 19023 }, { "epoch": 118.1619937694704, "grad_norm": 1.8359375, "learning_rate": 1.82788079862657e-05, "loss": 0.1621, "step": 19024 }, { "epoch": 118.16822429906541, "grad_norm": 2.75, "learning_rate": 1.827643496561729e-05, "loss": 0.2405, "step": 19025 }, { "epoch": 118.17445482866043, "grad_norm": 2.421875, "learning_rate": 1.8274062010269483e-05, "loss": 0.1509, "step": 19026 }, { "epoch": 118.18068535825545, "grad_norm": 2.4375, "learning_rate": 1.8271689120245332e-05, "loss": 0.167, "step": 19027 }, { "epoch": 118.18691588785046, "grad_norm": 2.5625, "learning_rate": 1.826931629556789e-05, "loss": 0.171, "step": 19028 }, { "epoch": 118.19314641744548, "grad_norm": 1.8671875, "learning_rate": 1.8266943536260188e-05, "loss": 0.1346, "step": 19029 }, { "epoch": 118.19937694704049, "grad_norm": 1.9375, "learning_rate": 1.8264570842345273e-05, "loss": 0.1662, "step": 19030 }, { "epoch": 118.20560747663552, "grad_norm": 2.921875, "learning_rate": 1.8262198213846194e-05, "loss": 0.2466, "step": 19031 }, { "epoch": 118.21183800623054, "grad_norm": 3.671875, "learning_rate": 1.8259825650785994e-05, "loss": 0.2386, "step": 19032 }, { "epoch": 118.21806853582555, "grad_norm": 3.46875, "learning_rate": 1.825745315318771e-05, "loss": 0.275, "step": 19033 }, { "epoch": 118.22429906542057, "grad_norm": 3.046875, "learning_rate": 1.825508072107439e-05, "loss": 0.2507, "step": 19034 }, { "epoch": 118.23052959501558, "grad_norm": 1.625, "learning_rate": 1.8252708354469086e-05, "loss": 0.1085, "step": 19035 }, { "epoch": 118.2367601246106, "grad_norm": 1.6171875, "learning_rate": 1.825033605339482e-05, "loss": 0.1645, "step": 19036 }, { "epoch": 118.24299065420561, "grad_norm": 1.6640625, "learning_rate": 1.8247963817874632e-05, "loss": 0.1354, "step": 19037 }, { "epoch": 118.24922118380063, "grad_norm": 2.828125, "learning_rate": 1.8245591647931577e-05, "loss": 0.3089, "step": 19038 }, { "epoch": 118.25545171339564, "grad_norm": 1.9453125, "learning_rate": 1.824321954358868e-05, "loss": 0.126, "step": 19039 }, { "epoch": 118.26168224299066, "grad_norm": 2.921875, "learning_rate": 1.824084750486899e-05, "loss": 0.2275, "step": 19040 }, { "epoch": 118.26791277258567, "grad_norm": 1.796875, "learning_rate": 1.8238475531795545e-05, "loss": 0.0725, "step": 19041 }, { "epoch": 118.27414330218069, "grad_norm": 2.546875, "learning_rate": 1.8236103624391375e-05, "loss": 0.2404, "step": 19042 }, { "epoch": 118.2803738317757, "grad_norm": 2.453125, "learning_rate": 1.823373178267951e-05, "loss": 0.1627, "step": 19043 }, { "epoch": 118.28660436137072, "grad_norm": 2.109375, "learning_rate": 1.8231360006683002e-05, "loss": 0.1636, "step": 19044 }, { "epoch": 118.29283489096574, "grad_norm": 1.3359375, "learning_rate": 1.8228988296424877e-05, "loss": 0.1257, "step": 19045 }, { "epoch": 118.29906542056075, "grad_norm": 1.1171875, "learning_rate": 1.8226616651928164e-05, "loss": 0.0975, "step": 19046 }, { "epoch": 118.30529595015577, "grad_norm": 2.453125, "learning_rate": 1.822424507321592e-05, "loss": 0.1604, "step": 19047 }, { "epoch": 118.31152647975078, "grad_norm": 1.90625, "learning_rate": 1.8221873560311155e-05, "loss": 0.1057, "step": 19048 }, { "epoch": 118.3177570093458, "grad_norm": 1.7890625, "learning_rate": 1.8219502113236903e-05, "loss": 0.1639, "step": 19049 }, { "epoch": 118.32398753894081, "grad_norm": 0.8828125, "learning_rate": 1.82171307320162e-05, "loss": 0.0934, "step": 19050 }, { "epoch": 118.33021806853583, "grad_norm": 1.7578125, "learning_rate": 1.821475941667209e-05, "loss": 0.1276, "step": 19051 }, { "epoch": 118.33644859813084, "grad_norm": 2.453125, "learning_rate": 1.8212388167227583e-05, "loss": 0.1994, "step": 19052 }, { "epoch": 118.34267912772586, "grad_norm": 1.140625, "learning_rate": 1.8210016983705725e-05, "loss": 0.1054, "step": 19053 }, { "epoch": 118.34890965732087, "grad_norm": 2.625, "learning_rate": 1.8207645866129544e-05, "loss": 0.2146, "step": 19054 }, { "epoch": 118.35514018691589, "grad_norm": 1.7578125, "learning_rate": 1.820527481452206e-05, "loss": 0.1519, "step": 19055 }, { "epoch": 118.3613707165109, "grad_norm": 1.671875, "learning_rate": 1.82029038289063e-05, "loss": 0.2269, "step": 19056 }, { "epoch": 118.36760124610592, "grad_norm": 1.8125, "learning_rate": 1.82005329093053e-05, "loss": 0.1314, "step": 19057 }, { "epoch": 118.37383177570094, "grad_norm": 2.890625, "learning_rate": 1.8198162055742084e-05, "loss": 0.2092, "step": 19058 }, { "epoch": 118.38006230529595, "grad_norm": 3.578125, "learning_rate": 1.8195791268239673e-05, "loss": 0.3828, "step": 19059 }, { "epoch": 118.38629283489097, "grad_norm": 1.890625, "learning_rate": 1.819342054682111e-05, "loss": 0.1799, "step": 19060 }, { "epoch": 118.39252336448598, "grad_norm": 3.0, "learning_rate": 1.8191049891509397e-05, "loss": 0.1592, "step": 19061 }, { "epoch": 118.398753894081, "grad_norm": 2.03125, "learning_rate": 1.818867930232757e-05, "loss": 0.1515, "step": 19062 }, { "epoch": 118.40498442367601, "grad_norm": 2.953125, "learning_rate": 1.818630877929865e-05, "loss": 0.2182, "step": 19063 }, { "epoch": 118.41121495327103, "grad_norm": 1.7109375, "learning_rate": 1.8183938322445664e-05, "loss": 0.2455, "step": 19064 }, { "epoch": 118.41744548286604, "grad_norm": 2.140625, "learning_rate": 1.8181567931791626e-05, "loss": 0.0873, "step": 19065 }, { "epoch": 118.42367601246106, "grad_norm": 1.28125, "learning_rate": 1.817919760735958e-05, "loss": 0.1116, "step": 19066 }, { "epoch": 118.42990654205607, "grad_norm": 1.0546875, "learning_rate": 1.8176827349172516e-05, "loss": 0.0928, "step": 19067 }, { "epoch": 118.43613707165109, "grad_norm": 2.6875, "learning_rate": 1.8174457157253472e-05, "loss": 0.1972, "step": 19068 }, { "epoch": 118.4423676012461, "grad_norm": 2.484375, "learning_rate": 1.8172087031625455e-05, "loss": 0.1831, "step": 19069 }, { "epoch": 118.44859813084112, "grad_norm": 4.15625, "learning_rate": 1.8169716972311506e-05, "loss": 0.2582, "step": 19070 }, { "epoch": 118.45482866043614, "grad_norm": 1.8125, "learning_rate": 1.8167346979334625e-05, "loss": 0.1326, "step": 19071 }, { "epoch": 118.46105919003115, "grad_norm": 2.046875, "learning_rate": 1.8164977052717837e-05, "loss": 0.1656, "step": 19072 }, { "epoch": 118.46728971962617, "grad_norm": 1.8125, "learning_rate": 1.8162607192484162e-05, "loss": 0.1371, "step": 19073 }, { "epoch": 118.47352024922118, "grad_norm": 1.6328125, "learning_rate": 1.8160237398656613e-05, "loss": 0.1101, "step": 19074 }, { "epoch": 118.4797507788162, "grad_norm": 2.984375, "learning_rate": 1.8157867671258202e-05, "loss": 0.1046, "step": 19075 }, { "epoch": 118.48598130841121, "grad_norm": 1.3515625, "learning_rate": 1.8155498010311946e-05, "loss": 0.1387, "step": 19076 }, { "epoch": 118.49221183800623, "grad_norm": 1.4296875, "learning_rate": 1.8153128415840865e-05, "loss": 0.1277, "step": 19077 }, { "epoch": 118.49844236760124, "grad_norm": 1.9375, "learning_rate": 1.815075888786797e-05, "loss": 0.1363, "step": 19078 }, { "epoch": 118.50467289719626, "grad_norm": 2.65625, "learning_rate": 1.8148389426416278e-05, "loss": 0.1955, "step": 19079 }, { "epoch": 118.51090342679127, "grad_norm": 3.21875, "learning_rate": 1.8146020031508785e-05, "loss": 0.343, "step": 19080 }, { "epoch": 118.51713395638629, "grad_norm": 2.296875, "learning_rate": 1.8143650703168525e-05, "loss": 0.1634, "step": 19081 }, { "epoch": 118.5233644859813, "grad_norm": 1.1953125, "learning_rate": 1.8141281441418495e-05, "loss": 0.0998, "step": 19082 }, { "epoch": 118.52959501557632, "grad_norm": 1.296875, "learning_rate": 1.8138912246281707e-05, "loss": 0.1071, "step": 19083 }, { "epoch": 118.53582554517133, "grad_norm": 2.484375, "learning_rate": 1.8136543117781184e-05, "loss": 0.1427, "step": 19084 }, { "epoch": 118.54205607476635, "grad_norm": 1.671875, "learning_rate": 1.8134174055939918e-05, "loss": 0.1289, "step": 19085 }, { "epoch": 118.54828660436137, "grad_norm": 2.984375, "learning_rate": 1.813180506078094e-05, "loss": 0.2555, "step": 19086 }, { "epoch": 118.55451713395638, "grad_norm": 2.265625, "learning_rate": 1.8129436132327232e-05, "loss": 0.2193, "step": 19087 }, { "epoch": 118.5607476635514, "grad_norm": 2.65625, "learning_rate": 1.812706727060182e-05, "loss": 0.2222, "step": 19088 }, { "epoch": 118.56697819314641, "grad_norm": 1.7421875, "learning_rate": 1.8124698475627696e-05, "loss": 0.1435, "step": 19089 }, { "epoch": 118.57320872274143, "grad_norm": 1.46875, "learning_rate": 1.812232974742788e-05, "loss": 0.1319, "step": 19090 }, { "epoch": 118.57943925233644, "grad_norm": 2.078125, "learning_rate": 1.8119961086025374e-05, "loss": 0.1382, "step": 19091 }, { "epoch": 118.58566978193146, "grad_norm": 1.15625, "learning_rate": 1.8117592491443185e-05, "loss": 0.1305, "step": 19092 }, { "epoch": 118.59190031152647, "grad_norm": 1.171875, "learning_rate": 1.8115223963704304e-05, "loss": 0.1014, "step": 19093 }, { "epoch": 118.59813084112149, "grad_norm": 1.453125, "learning_rate": 1.811285550283175e-05, "loss": 0.1463, "step": 19094 }, { "epoch": 118.60436137071652, "grad_norm": 2.03125, "learning_rate": 1.8110487108848518e-05, "loss": 0.2065, "step": 19095 }, { "epoch": 118.61059190031153, "grad_norm": 1.7578125, "learning_rate": 1.810811878177761e-05, "loss": 0.2039, "step": 19096 }, { "epoch": 118.61682242990655, "grad_norm": 2.1875, "learning_rate": 1.8105750521642034e-05, "loss": 0.1226, "step": 19097 }, { "epoch": 118.62305295950156, "grad_norm": 2.65625, "learning_rate": 1.8103382328464792e-05, "loss": 0.1657, "step": 19098 }, { "epoch": 118.62928348909658, "grad_norm": 3.0625, "learning_rate": 1.8101014202268872e-05, "loss": 0.2932, "step": 19099 }, { "epoch": 118.6355140186916, "grad_norm": 1.5234375, "learning_rate": 1.809864614307728e-05, "loss": 0.1269, "step": 19100 }, { "epoch": 118.64174454828661, "grad_norm": 1.8828125, "learning_rate": 1.809627815091302e-05, "loss": 0.1242, "step": 19101 }, { "epoch": 118.64797507788163, "grad_norm": 2.046875, "learning_rate": 1.809391022579908e-05, "loss": 0.1987, "step": 19102 }, { "epoch": 118.65420560747664, "grad_norm": 1.625, "learning_rate": 1.809154236775847e-05, "loss": 0.1211, "step": 19103 }, { "epoch": 118.66043613707166, "grad_norm": 2.265625, "learning_rate": 1.8089174576814182e-05, "loss": 0.1671, "step": 19104 }, { "epoch": 118.66666666666667, "grad_norm": 2.421875, "learning_rate": 1.8086806852989216e-05, "loss": 0.1844, "step": 19105 }, { "epoch": 118.67289719626169, "grad_norm": 1.828125, "learning_rate": 1.8084439196306556e-05, "loss": 0.1543, "step": 19106 }, { "epoch": 118.6791277258567, "grad_norm": 1.8984375, "learning_rate": 1.8082071606789207e-05, "loss": 0.1426, "step": 19107 }, { "epoch": 118.68535825545172, "grad_norm": 3.28125, "learning_rate": 1.8079704084460162e-05, "loss": 0.2747, "step": 19108 }, { "epoch": 118.69158878504673, "grad_norm": 2.046875, "learning_rate": 1.8077336629342413e-05, "loss": 0.1398, "step": 19109 }, { "epoch": 118.69781931464175, "grad_norm": 2.203125, "learning_rate": 1.8074969241458957e-05, "loss": 0.1578, "step": 19110 }, { "epoch": 118.70404984423676, "grad_norm": 1.921875, "learning_rate": 1.8072601920832786e-05, "loss": 0.1638, "step": 19111 }, { "epoch": 118.71028037383178, "grad_norm": 3.734375, "learning_rate": 1.807023466748689e-05, "loss": 0.2966, "step": 19112 }, { "epoch": 118.7165109034268, "grad_norm": 1.40625, "learning_rate": 1.806786748144425e-05, "loss": 0.1048, "step": 19113 }, { "epoch": 118.72274143302181, "grad_norm": 2.3125, "learning_rate": 1.806550036272788e-05, "loss": 0.163, "step": 19114 }, { "epoch": 118.72897196261682, "grad_norm": 2.09375, "learning_rate": 1.8063133311360743e-05, "loss": 0.177, "step": 19115 }, { "epoch": 118.73520249221184, "grad_norm": 1.921875, "learning_rate": 1.8060766327365846e-05, "loss": 0.1617, "step": 19116 }, { "epoch": 118.74143302180686, "grad_norm": 3.5625, "learning_rate": 1.8058399410766185e-05, "loss": 0.2542, "step": 19117 }, { "epoch": 118.74766355140187, "grad_norm": 1.9140625, "learning_rate": 1.805603256158473e-05, "loss": 0.1401, "step": 19118 }, { "epoch": 118.75389408099689, "grad_norm": 3.84375, "learning_rate": 1.8053665779844468e-05, "loss": 0.1885, "step": 19119 }, { "epoch": 118.7601246105919, "grad_norm": 1.6328125, "learning_rate": 1.80512990655684e-05, "loss": 0.1377, "step": 19120 }, { "epoch": 118.76635514018692, "grad_norm": 2.078125, "learning_rate": 1.8048932418779503e-05, "loss": 0.1506, "step": 19121 }, { "epoch": 118.77258566978193, "grad_norm": 1.890625, "learning_rate": 1.8046565839500756e-05, "loss": 0.2086, "step": 19122 }, { "epoch": 118.77881619937695, "grad_norm": 2.03125, "learning_rate": 1.8044199327755164e-05, "loss": 0.2492, "step": 19123 }, { "epoch": 118.78504672897196, "grad_norm": 1.9921875, "learning_rate": 1.8041832883565697e-05, "loss": 0.1452, "step": 19124 }, { "epoch": 118.79127725856698, "grad_norm": 2.890625, "learning_rate": 1.8039466506955334e-05, "loss": 0.188, "step": 19125 }, { "epoch": 118.797507788162, "grad_norm": 2.4375, "learning_rate": 1.8037100197947064e-05, "loss": 0.2608, "step": 19126 }, { "epoch": 118.80373831775701, "grad_norm": 2.65625, "learning_rate": 1.803473395656387e-05, "loss": 0.1793, "step": 19127 }, { "epoch": 118.80996884735202, "grad_norm": 2.078125, "learning_rate": 1.8032367782828734e-05, "loss": 0.2025, "step": 19128 }, { "epoch": 118.81619937694704, "grad_norm": 3.109375, "learning_rate": 1.803000167676463e-05, "loss": 0.3096, "step": 19129 }, { "epoch": 118.82242990654206, "grad_norm": 1.796875, "learning_rate": 1.8027635638394552e-05, "loss": 0.1232, "step": 19130 }, { "epoch": 118.82866043613707, "grad_norm": 2.046875, "learning_rate": 1.8025269667741467e-05, "loss": 0.1624, "step": 19131 }, { "epoch": 118.83489096573209, "grad_norm": 1.4921875, "learning_rate": 1.8022903764828354e-05, "loss": 0.0965, "step": 19132 }, { "epoch": 118.8411214953271, "grad_norm": 1.765625, "learning_rate": 1.802053792967819e-05, "loss": 0.108, "step": 19133 }, { "epoch": 118.84735202492212, "grad_norm": 2.5625, "learning_rate": 1.8018172162313962e-05, "loss": 0.1922, "step": 19134 }, { "epoch": 118.85358255451713, "grad_norm": 3.0625, "learning_rate": 1.801580646275864e-05, "loss": 0.1638, "step": 19135 }, { "epoch": 118.85981308411215, "grad_norm": 1.875, "learning_rate": 1.8013440831035214e-05, "loss": 0.2212, "step": 19136 }, { "epoch": 118.86604361370716, "grad_norm": 2.28125, "learning_rate": 1.8011075267166632e-05, "loss": 0.098, "step": 19137 }, { "epoch": 118.87227414330218, "grad_norm": 2.015625, "learning_rate": 1.8008709771175892e-05, "loss": 0.1519, "step": 19138 }, { "epoch": 118.8785046728972, "grad_norm": 1.671875, "learning_rate": 1.8006344343085953e-05, "loss": 0.159, "step": 19139 }, { "epoch": 118.88473520249221, "grad_norm": 3.53125, "learning_rate": 1.80039789829198e-05, "loss": 0.3048, "step": 19140 }, { "epoch": 118.89096573208722, "grad_norm": 0.98828125, "learning_rate": 1.80016136907004e-05, "loss": 0.0951, "step": 19141 }, { "epoch": 118.89719626168224, "grad_norm": 1.28125, "learning_rate": 1.7999248466450724e-05, "loss": 0.1102, "step": 19142 }, { "epoch": 118.90342679127725, "grad_norm": 3.453125, "learning_rate": 1.7996883310193756e-05, "loss": 0.2495, "step": 19143 }, { "epoch": 118.90965732087227, "grad_norm": 1.7734375, "learning_rate": 1.7994518221952452e-05, "loss": 0.1266, "step": 19144 }, { "epoch": 118.91588785046729, "grad_norm": 2.921875, "learning_rate": 1.799215320174979e-05, "loss": 0.2354, "step": 19145 }, { "epoch": 118.9221183800623, "grad_norm": 2.234375, "learning_rate": 1.798978824960873e-05, "loss": 0.1704, "step": 19146 }, { "epoch": 118.92834890965732, "grad_norm": 3.265625, "learning_rate": 1.7987423365552254e-05, "loss": 0.2596, "step": 19147 }, { "epoch": 118.93457943925233, "grad_norm": 1.9140625, "learning_rate": 1.7985058549603315e-05, "loss": 0.1486, "step": 19148 }, { "epoch": 118.94080996884735, "grad_norm": 1.5078125, "learning_rate": 1.798269380178491e-05, "loss": 0.114, "step": 19149 }, { "epoch": 118.94704049844236, "grad_norm": 1.8359375, "learning_rate": 1.7980329122119966e-05, "loss": 0.1115, "step": 19150 }, { "epoch": 118.95327102803738, "grad_norm": 2.6875, "learning_rate": 1.7977964510631476e-05, "loss": 0.3106, "step": 19151 }, { "epoch": 118.9595015576324, "grad_norm": 1.875, "learning_rate": 1.7975599967342395e-05, "loss": 0.1574, "step": 19152 }, { "epoch": 118.96573208722741, "grad_norm": 1.9296875, "learning_rate": 1.7973235492275693e-05, "loss": 0.2044, "step": 19153 }, { "epoch": 118.97196261682242, "grad_norm": 1.6953125, "learning_rate": 1.7970871085454333e-05, "loss": 0.141, "step": 19154 }, { "epoch": 118.97819314641744, "grad_norm": 1.6640625, "learning_rate": 1.796850674690127e-05, "loss": 0.1139, "step": 19155 }, { "epoch": 118.98442367601245, "grad_norm": 2.171875, "learning_rate": 1.7966142476639492e-05, "loss": 0.1697, "step": 19156 }, { "epoch": 118.99065420560747, "grad_norm": 2.390625, "learning_rate": 1.7963778274691937e-05, "loss": 0.1191, "step": 19157 }, { "epoch": 118.99688473520249, "grad_norm": 3.328125, "learning_rate": 1.7961414141081573e-05, "loss": 0.2769, "step": 19158 }, { "epoch": 119.0, "grad_norm": 2.140625, "learning_rate": 1.795905007583136e-05, "loss": 0.1661, "step": 19159 }, { "epoch": 119.00623052959502, "grad_norm": 1.0625, "learning_rate": 1.795668607896426e-05, "loss": 0.1, "step": 19160 }, { "epoch": 119.01246105919003, "grad_norm": 1.9140625, "learning_rate": 1.795432215050323e-05, "loss": 0.1538, "step": 19161 }, { "epoch": 119.01869158878505, "grad_norm": 2.328125, "learning_rate": 1.7951958290471242e-05, "loss": 0.1134, "step": 19162 }, { "epoch": 119.02492211838006, "grad_norm": 2.328125, "learning_rate": 1.7949594498891237e-05, "loss": 0.1857, "step": 19163 }, { "epoch": 119.03115264797508, "grad_norm": 1.8046875, "learning_rate": 1.7947230775786182e-05, "loss": 0.1714, "step": 19164 }, { "epoch": 119.03738317757009, "grad_norm": 2.796875, "learning_rate": 1.7944867121179025e-05, "loss": 0.2105, "step": 19165 }, { "epoch": 119.04361370716511, "grad_norm": 1.5859375, "learning_rate": 1.7942503535092734e-05, "loss": 0.1392, "step": 19166 }, { "epoch": 119.04984423676012, "grad_norm": 1.1796875, "learning_rate": 1.7940140017550262e-05, "loss": 0.1047, "step": 19167 }, { "epoch": 119.05607476635514, "grad_norm": 3.25, "learning_rate": 1.793777656857456e-05, "loss": 0.2299, "step": 19168 }, { "epoch": 119.06230529595015, "grad_norm": 2.75, "learning_rate": 1.793541318818858e-05, "loss": 0.2671, "step": 19169 }, { "epoch": 119.06853582554517, "grad_norm": 1.328125, "learning_rate": 1.793304987641528e-05, "loss": 0.1263, "step": 19170 }, { "epoch": 119.07476635514018, "grad_norm": 2.0, "learning_rate": 1.7930686633277616e-05, "loss": 0.1384, "step": 19171 }, { "epoch": 119.0809968847352, "grad_norm": 2.28125, "learning_rate": 1.792832345879853e-05, "loss": 0.2272, "step": 19172 }, { "epoch": 119.08722741433021, "grad_norm": 3.171875, "learning_rate": 1.7925960353000986e-05, "loss": 0.2544, "step": 19173 }, { "epoch": 119.09345794392523, "grad_norm": 2.28125, "learning_rate": 1.7923597315907928e-05, "loss": 0.3053, "step": 19174 }, { "epoch": 119.09968847352025, "grad_norm": 2.296875, "learning_rate": 1.792123434754231e-05, "loss": 0.2116, "step": 19175 }, { "epoch": 119.10591900311526, "grad_norm": 2.6875, "learning_rate": 1.7918871447927075e-05, "loss": 0.2477, "step": 19176 }, { "epoch": 119.11214953271028, "grad_norm": 1.953125, "learning_rate": 1.7916508617085175e-05, "loss": 0.1074, "step": 19177 }, { "epoch": 119.11838006230529, "grad_norm": 2.328125, "learning_rate": 1.7914145855039565e-05, "loss": 0.148, "step": 19178 }, { "epoch": 119.1246105919003, "grad_norm": 2.09375, "learning_rate": 1.7911783161813177e-05, "loss": 0.22, "step": 19179 }, { "epoch": 119.13084112149532, "grad_norm": 1.4453125, "learning_rate": 1.7909420537428974e-05, "loss": 0.1316, "step": 19180 }, { "epoch": 119.13707165109034, "grad_norm": 2.5, "learning_rate": 1.7907057981909902e-05, "loss": 0.2022, "step": 19181 }, { "epoch": 119.14330218068535, "grad_norm": 1.3203125, "learning_rate": 1.79046954952789e-05, "loss": 0.098, "step": 19182 }, { "epoch": 119.14953271028037, "grad_norm": 1.9609375, "learning_rate": 1.7902333077558904e-05, "loss": 0.1505, "step": 19183 }, { "epoch": 119.15576323987538, "grad_norm": 1.765625, "learning_rate": 1.789997072877287e-05, "loss": 0.1334, "step": 19184 }, { "epoch": 119.1619937694704, "grad_norm": 1.6171875, "learning_rate": 1.7897608448943737e-05, "loss": 0.113, "step": 19185 }, { "epoch": 119.16822429906541, "grad_norm": 1.4453125, "learning_rate": 1.7895246238094456e-05, "loss": 0.1089, "step": 19186 }, { "epoch": 119.17445482866043, "grad_norm": 1.6171875, "learning_rate": 1.789288409624797e-05, "loss": 0.1239, "step": 19187 }, { "epoch": 119.18068535825545, "grad_norm": 2.03125, "learning_rate": 1.7890522023427208e-05, "loss": 0.1587, "step": 19188 }, { "epoch": 119.18691588785046, "grad_norm": 2.65625, "learning_rate": 1.7888160019655115e-05, "loss": 0.2136, "step": 19189 }, { "epoch": 119.19314641744548, "grad_norm": 1.6640625, "learning_rate": 1.7885798084954635e-05, "loss": 0.1392, "step": 19190 }, { "epoch": 119.19937694704049, "grad_norm": 3.390625, "learning_rate": 1.7883436219348706e-05, "loss": 0.2879, "step": 19191 }, { "epoch": 119.20560747663552, "grad_norm": 1.6015625, "learning_rate": 1.7881074422860263e-05, "loss": 0.2192, "step": 19192 }, { "epoch": 119.21183800623054, "grad_norm": 2.0, "learning_rate": 1.7878712695512256e-05, "loss": 0.196, "step": 19193 }, { "epoch": 119.21806853582555, "grad_norm": 1.8046875, "learning_rate": 1.7876351037327617e-05, "loss": 0.1302, "step": 19194 }, { "epoch": 119.22429906542057, "grad_norm": 2.265625, "learning_rate": 1.7873989448329276e-05, "loss": 0.2486, "step": 19195 }, { "epoch": 119.23052959501558, "grad_norm": 2.15625, "learning_rate": 1.787162792854017e-05, "loss": 0.1688, "step": 19196 }, { "epoch": 119.2367601246106, "grad_norm": 1.453125, "learning_rate": 1.786926647798324e-05, "loss": 0.1091, "step": 19197 }, { "epoch": 119.24299065420561, "grad_norm": 2.703125, "learning_rate": 1.7866905096681415e-05, "loss": 0.2284, "step": 19198 }, { "epoch": 119.24922118380063, "grad_norm": 1.8046875, "learning_rate": 1.7864543784657643e-05, "loss": 0.1546, "step": 19199 }, { "epoch": 119.25545171339564, "grad_norm": 2.484375, "learning_rate": 1.7862182541934848e-05, "loss": 0.1902, "step": 19200 }, { "epoch": 119.26168224299066, "grad_norm": 1.1640625, "learning_rate": 1.785982136853596e-05, "loss": 0.1092, "step": 19201 }, { "epoch": 119.26791277258567, "grad_norm": 1.5625, "learning_rate": 1.785746026448391e-05, "loss": 0.1282, "step": 19202 }, { "epoch": 119.27414330218069, "grad_norm": 1.625, "learning_rate": 1.7855099229801643e-05, "loss": 0.1121, "step": 19203 }, { "epoch": 119.2803738317757, "grad_norm": 2.734375, "learning_rate": 1.7852738264512072e-05, "loss": 0.2218, "step": 19204 }, { "epoch": 119.28660436137072, "grad_norm": 2.140625, "learning_rate": 1.7850377368638135e-05, "loss": 0.1377, "step": 19205 }, { "epoch": 119.29283489096574, "grad_norm": 2.65625, "learning_rate": 1.7848016542202767e-05, "loss": 0.1841, "step": 19206 }, { "epoch": 119.29906542056075, "grad_norm": 3.5625, "learning_rate": 1.7845655785228898e-05, "loss": 0.4005, "step": 19207 }, { "epoch": 119.30529595015577, "grad_norm": 1.5390625, "learning_rate": 1.7843295097739442e-05, "loss": 0.1275, "step": 19208 }, { "epoch": 119.31152647975078, "grad_norm": 1.8515625, "learning_rate": 1.784093447975733e-05, "loss": 0.1269, "step": 19209 }, { "epoch": 119.3177570093458, "grad_norm": 1.6953125, "learning_rate": 1.7838573931305504e-05, "loss": 0.1313, "step": 19210 }, { "epoch": 119.32398753894081, "grad_norm": 2.171875, "learning_rate": 1.783621345240687e-05, "loss": 0.103, "step": 19211 }, { "epoch": 119.33021806853583, "grad_norm": 2.171875, "learning_rate": 1.7833853043084365e-05, "loss": 0.154, "step": 19212 }, { "epoch": 119.33644859813084, "grad_norm": 2.46875, "learning_rate": 1.783149270336092e-05, "loss": 0.2583, "step": 19213 }, { "epoch": 119.34267912772586, "grad_norm": 2.84375, "learning_rate": 1.782913243325945e-05, "loss": 0.2, "step": 19214 }, { "epoch": 119.34890965732087, "grad_norm": 2.703125, "learning_rate": 1.782677223280287e-05, "loss": 0.202, "step": 19215 }, { "epoch": 119.35514018691589, "grad_norm": 2.515625, "learning_rate": 1.7824412102014115e-05, "loss": 0.2802, "step": 19216 }, { "epoch": 119.3613707165109, "grad_norm": 1.8046875, "learning_rate": 1.7822052040916105e-05, "loss": 0.1557, "step": 19217 }, { "epoch": 119.36760124610592, "grad_norm": 2.796875, "learning_rate": 1.7819692049531756e-05, "loss": 0.2146, "step": 19218 }, { "epoch": 119.37383177570094, "grad_norm": 1.96875, "learning_rate": 1.7817332127884008e-05, "loss": 0.1211, "step": 19219 }, { "epoch": 119.38006230529595, "grad_norm": 1.53125, "learning_rate": 1.7814972275995753e-05, "loss": 0.1112, "step": 19220 }, { "epoch": 119.38629283489097, "grad_norm": 2.21875, "learning_rate": 1.7812612493889928e-05, "loss": 0.1518, "step": 19221 }, { "epoch": 119.39252336448598, "grad_norm": 3.4375, "learning_rate": 1.7810252781589444e-05, "loss": 0.179, "step": 19222 }, { "epoch": 119.398753894081, "grad_norm": 2.59375, "learning_rate": 1.780789313911722e-05, "loss": 0.153, "step": 19223 }, { "epoch": 119.40498442367601, "grad_norm": 1.875, "learning_rate": 1.7805533566496182e-05, "loss": 0.1723, "step": 19224 }, { "epoch": 119.41121495327103, "grad_norm": 3.71875, "learning_rate": 1.7803174063749233e-05, "loss": 0.2473, "step": 19225 }, { "epoch": 119.41744548286604, "grad_norm": 2.453125, "learning_rate": 1.7800814630899307e-05, "loss": 0.1953, "step": 19226 }, { "epoch": 119.42367601246106, "grad_norm": 3.265625, "learning_rate": 1.7798455267969304e-05, "loss": 0.2254, "step": 19227 }, { "epoch": 119.42990654205607, "grad_norm": 2.8125, "learning_rate": 1.779609597498214e-05, "loss": 0.1784, "step": 19228 }, { "epoch": 119.43613707165109, "grad_norm": 1.5078125, "learning_rate": 1.779373675196073e-05, "loss": 0.1228, "step": 19229 }, { "epoch": 119.4423676012461, "grad_norm": 1.2890625, "learning_rate": 1.7791377598927996e-05, "loss": 0.1335, "step": 19230 }, { "epoch": 119.44859813084112, "grad_norm": 1.6171875, "learning_rate": 1.778901851590683e-05, "loss": 0.149, "step": 19231 }, { "epoch": 119.45482866043614, "grad_norm": 1.7109375, "learning_rate": 1.778665950292018e-05, "loss": 0.1443, "step": 19232 }, { "epoch": 119.46105919003115, "grad_norm": 1.578125, "learning_rate": 1.778430055999092e-05, "loss": 0.115, "step": 19233 }, { "epoch": 119.46728971962617, "grad_norm": 1.4296875, "learning_rate": 1.778194168714198e-05, "loss": 0.1068, "step": 19234 }, { "epoch": 119.47352024922118, "grad_norm": 1.8203125, "learning_rate": 1.7779582884396257e-05, "loss": 0.1034, "step": 19235 }, { "epoch": 119.4797507788162, "grad_norm": 1.3671875, "learning_rate": 1.777722415177667e-05, "loss": 0.0932, "step": 19236 }, { "epoch": 119.48598130841121, "grad_norm": 1.296875, "learning_rate": 1.7774865489306135e-05, "loss": 0.1089, "step": 19237 }, { "epoch": 119.49221183800623, "grad_norm": 1.875, "learning_rate": 1.7772506897007552e-05, "loss": 0.1315, "step": 19238 }, { "epoch": 119.49844236760124, "grad_norm": 0.86328125, "learning_rate": 1.7770148374903815e-05, "loss": 0.0978, "step": 19239 }, { "epoch": 119.50467289719626, "grad_norm": 1.171875, "learning_rate": 1.776778992301785e-05, "loss": 0.0904, "step": 19240 }, { "epoch": 119.51090342679127, "grad_norm": 3.234375, "learning_rate": 1.7765431541372552e-05, "loss": 0.3524, "step": 19241 }, { "epoch": 119.51713395638629, "grad_norm": 3.078125, "learning_rate": 1.7763073229990827e-05, "loss": 0.1255, "step": 19242 }, { "epoch": 119.5233644859813, "grad_norm": 2.625, "learning_rate": 1.7760714988895583e-05, "loss": 0.105, "step": 19243 }, { "epoch": 119.52959501557632, "grad_norm": 1.3125, "learning_rate": 1.775835681810972e-05, "loss": 0.1014, "step": 19244 }, { "epoch": 119.53582554517133, "grad_norm": 2.375, "learning_rate": 1.7755998717656158e-05, "loss": 0.2004, "step": 19245 }, { "epoch": 119.54205607476635, "grad_norm": 3.421875, "learning_rate": 1.7753640687557767e-05, "loss": 0.3627, "step": 19246 }, { "epoch": 119.54828660436137, "grad_norm": 1.1953125, "learning_rate": 1.7751282727837473e-05, "loss": 0.1245, "step": 19247 }, { "epoch": 119.55451713395638, "grad_norm": 1.4140625, "learning_rate": 1.7748924838518167e-05, "loss": 0.1094, "step": 19248 }, { "epoch": 119.5607476635514, "grad_norm": 1.4375, "learning_rate": 1.7746567019622755e-05, "loss": 0.1226, "step": 19249 }, { "epoch": 119.56697819314641, "grad_norm": 1.0546875, "learning_rate": 1.7744209271174135e-05, "loss": 0.0945, "step": 19250 }, { "epoch": 119.57320872274143, "grad_norm": 1.296875, "learning_rate": 1.7741851593195208e-05, "loss": 0.0981, "step": 19251 }, { "epoch": 119.57943925233644, "grad_norm": 1.9765625, "learning_rate": 1.773949398570886e-05, "loss": 0.1545, "step": 19252 }, { "epoch": 119.58566978193146, "grad_norm": 1.9453125, "learning_rate": 1.7737136448738003e-05, "loss": 0.1558, "step": 19253 }, { "epoch": 119.59190031152647, "grad_norm": 2.671875, "learning_rate": 1.773477898230553e-05, "loss": 0.2687, "step": 19254 }, { "epoch": 119.59813084112149, "grad_norm": 3.09375, "learning_rate": 1.7732421586434327e-05, "loss": 0.2649, "step": 19255 }, { "epoch": 119.60436137071652, "grad_norm": 2.28125, "learning_rate": 1.7730064261147308e-05, "loss": 0.2187, "step": 19256 }, { "epoch": 119.61059190031153, "grad_norm": 2.859375, "learning_rate": 1.7727707006467363e-05, "loss": 0.1865, "step": 19257 }, { "epoch": 119.61682242990655, "grad_norm": 1.984375, "learning_rate": 1.7725349822417376e-05, "loss": 0.1233, "step": 19258 }, { "epoch": 119.62305295950156, "grad_norm": 2.0, "learning_rate": 1.7722992709020236e-05, "loss": 0.2462, "step": 19259 }, { "epoch": 119.62928348909658, "grad_norm": 1.59375, "learning_rate": 1.7720635666298854e-05, "loss": 0.1247, "step": 19260 }, { "epoch": 119.6355140186916, "grad_norm": 1.8515625, "learning_rate": 1.7718278694276108e-05, "loss": 0.0989, "step": 19261 }, { "epoch": 119.64174454828661, "grad_norm": 1.5859375, "learning_rate": 1.77159217929749e-05, "loss": 0.1288, "step": 19262 }, { "epoch": 119.64797507788163, "grad_norm": 3.28125, "learning_rate": 1.7713564962418117e-05, "loss": 0.2326, "step": 19263 }, { "epoch": 119.65420560747664, "grad_norm": 1.640625, "learning_rate": 1.771120820262865e-05, "loss": 0.1219, "step": 19264 }, { "epoch": 119.66043613707166, "grad_norm": 3.09375, "learning_rate": 1.7708851513629377e-05, "loss": 0.2448, "step": 19265 }, { "epoch": 119.66666666666667, "grad_norm": 2.734375, "learning_rate": 1.7706494895443202e-05, "loss": 0.3384, "step": 19266 }, { "epoch": 119.67289719626169, "grad_norm": 2.609375, "learning_rate": 1.7704138348093004e-05, "loss": 0.1863, "step": 19267 }, { "epoch": 119.6791277258567, "grad_norm": 1.9375, "learning_rate": 1.770178187160167e-05, "loss": 0.0889, "step": 19268 }, { "epoch": 119.68535825545172, "grad_norm": 1.765625, "learning_rate": 1.7699425465992093e-05, "loss": 0.1177, "step": 19269 }, { "epoch": 119.69158878504673, "grad_norm": 2.515625, "learning_rate": 1.769706913128716e-05, "loss": 0.2817, "step": 19270 }, { "epoch": 119.69781931464175, "grad_norm": 1.921875, "learning_rate": 1.769471286750975e-05, "loss": 0.1736, "step": 19271 }, { "epoch": 119.70404984423676, "grad_norm": 2.65625, "learning_rate": 1.769235667468274e-05, "loss": 0.157, "step": 19272 }, { "epoch": 119.71028037383178, "grad_norm": 3.09375, "learning_rate": 1.7690000552829028e-05, "loss": 0.3854, "step": 19273 }, { "epoch": 119.7165109034268, "grad_norm": 1.0703125, "learning_rate": 1.768764450197149e-05, "loss": 0.1091, "step": 19274 }, { "epoch": 119.72274143302181, "grad_norm": 1.2890625, "learning_rate": 1.7685288522133008e-05, "loss": 0.0996, "step": 19275 }, { "epoch": 119.72897196261682, "grad_norm": 1.5078125, "learning_rate": 1.7682932613336473e-05, "loss": 0.13, "step": 19276 }, { "epoch": 119.73520249221184, "grad_norm": 2.71875, "learning_rate": 1.7680576775604762e-05, "loss": 0.1554, "step": 19277 }, { "epoch": 119.74143302180686, "grad_norm": 1.6015625, "learning_rate": 1.767822100896075e-05, "loss": 0.1419, "step": 19278 }, { "epoch": 119.74766355140187, "grad_norm": 4.1875, "learning_rate": 1.767586531342731e-05, "loss": 0.3729, "step": 19279 }, { "epoch": 119.75389408099689, "grad_norm": 2.5625, "learning_rate": 1.7673509689027337e-05, "loss": 0.2061, "step": 19280 }, { "epoch": 119.7601246105919, "grad_norm": 1.0078125, "learning_rate": 1.7671154135783697e-05, "loss": 0.1029, "step": 19281 }, { "epoch": 119.76635514018692, "grad_norm": 3.203125, "learning_rate": 1.766879865371928e-05, "loss": 0.2245, "step": 19282 }, { "epoch": 119.77258566978193, "grad_norm": 2.828125, "learning_rate": 1.766644324285696e-05, "loss": 0.2756, "step": 19283 }, { "epoch": 119.77881619937695, "grad_norm": 2.296875, "learning_rate": 1.76640879032196e-05, "loss": 0.2077, "step": 19284 }, { "epoch": 119.78504672897196, "grad_norm": 3.5625, "learning_rate": 1.7661732634830087e-05, "loss": 0.2417, "step": 19285 }, { "epoch": 119.79127725856698, "grad_norm": 2.234375, "learning_rate": 1.7659377437711295e-05, "loss": 0.1479, "step": 19286 }, { "epoch": 119.797507788162, "grad_norm": 2.59375, "learning_rate": 1.7657022311886097e-05, "loss": 0.1719, "step": 19287 }, { "epoch": 119.80373831775701, "grad_norm": 1.5546875, "learning_rate": 1.7654667257377356e-05, "loss": 0.1495, "step": 19288 }, { "epoch": 119.80996884735202, "grad_norm": 2.21875, "learning_rate": 1.765231227420797e-05, "loss": 0.1785, "step": 19289 }, { "epoch": 119.81619937694704, "grad_norm": 1.375, "learning_rate": 1.7649957362400795e-05, "loss": 0.1243, "step": 19290 }, { "epoch": 119.82242990654206, "grad_norm": 2.234375, "learning_rate": 1.7647602521978696e-05, "loss": 0.1804, "step": 19291 }, { "epoch": 119.82866043613707, "grad_norm": 1.6328125, "learning_rate": 1.7645247752964548e-05, "loss": 0.0985, "step": 19292 }, { "epoch": 119.83489096573209, "grad_norm": 3.0625, "learning_rate": 1.764289305538123e-05, "loss": 0.3437, "step": 19293 }, { "epoch": 119.8411214953271, "grad_norm": 2.59375, "learning_rate": 1.76405384292516e-05, "loss": 0.1413, "step": 19294 }, { "epoch": 119.84735202492212, "grad_norm": 2.28125, "learning_rate": 1.7638183874598537e-05, "loss": 0.2202, "step": 19295 }, { "epoch": 119.85358255451713, "grad_norm": 2.375, "learning_rate": 1.7635829391444907e-05, "loss": 0.2295, "step": 19296 }, { "epoch": 119.85981308411215, "grad_norm": 2.859375, "learning_rate": 1.7633474979813573e-05, "loss": 0.185, "step": 19297 }, { "epoch": 119.86604361370716, "grad_norm": 2.421875, "learning_rate": 1.7631120639727393e-05, "loss": 0.2237, "step": 19298 }, { "epoch": 119.87227414330218, "grad_norm": 2.6875, "learning_rate": 1.762876637120925e-05, "loss": 0.2321, "step": 19299 }, { "epoch": 119.8785046728972, "grad_norm": 2.359375, "learning_rate": 1.7626412174282003e-05, "loss": 0.2287, "step": 19300 }, { "epoch": 119.88473520249221, "grad_norm": 2.078125, "learning_rate": 1.7624058048968507e-05, "loss": 0.1652, "step": 19301 }, { "epoch": 119.89096573208722, "grad_norm": 2.6875, "learning_rate": 1.7621703995291647e-05, "loss": 0.2959, "step": 19302 }, { "epoch": 119.89719626168224, "grad_norm": 2.65625, "learning_rate": 1.7619350013274266e-05, "loss": 0.1951, "step": 19303 }, { "epoch": 119.90342679127725, "grad_norm": 1.40625, "learning_rate": 1.7616996102939233e-05, "loss": 0.1243, "step": 19304 }, { "epoch": 119.90965732087227, "grad_norm": 2.6875, "learning_rate": 1.76146422643094e-05, "loss": 0.1734, "step": 19305 }, { "epoch": 119.91588785046729, "grad_norm": 2.796875, "learning_rate": 1.7612288497407648e-05, "loss": 0.172, "step": 19306 }, { "epoch": 119.9221183800623, "grad_norm": 2.1875, "learning_rate": 1.7609934802256822e-05, "loss": 0.1592, "step": 19307 }, { "epoch": 119.92834890965732, "grad_norm": 1.796875, "learning_rate": 1.76075811788798e-05, "loss": 0.1763, "step": 19308 }, { "epoch": 119.93457943925233, "grad_norm": 2.515625, "learning_rate": 1.7605227627299413e-05, "loss": 0.2221, "step": 19309 }, { "epoch": 119.94080996884735, "grad_norm": 1.6796875, "learning_rate": 1.760287414753854e-05, "loss": 0.1236, "step": 19310 }, { "epoch": 119.94704049844236, "grad_norm": 1.2421875, "learning_rate": 1.760052073962002e-05, "loss": 0.0887, "step": 19311 }, { "epoch": 119.95327102803738, "grad_norm": 1.3515625, "learning_rate": 1.7598167403566734e-05, "loss": 0.1138, "step": 19312 }, { "epoch": 119.9595015576324, "grad_norm": 2.734375, "learning_rate": 1.759581413940152e-05, "loss": 0.2556, "step": 19313 }, { "epoch": 119.96573208722741, "grad_norm": 1.4609375, "learning_rate": 1.7593460947147238e-05, "loss": 0.1328, "step": 19314 }, { "epoch": 119.97196261682242, "grad_norm": 1.4296875, "learning_rate": 1.7591107826826757e-05, "loss": 0.1194, "step": 19315 }, { "epoch": 119.97819314641744, "grad_norm": 3.84375, "learning_rate": 1.7588754778462907e-05, "loss": 0.2559, "step": 19316 }, { "epoch": 119.98442367601245, "grad_norm": 2.828125, "learning_rate": 1.7586401802078555e-05, "loss": 0.2533, "step": 19317 }, { "epoch": 119.99065420560747, "grad_norm": 2.25, "learning_rate": 1.7584048897696544e-05, "loss": 0.1351, "step": 19318 }, { "epoch": 119.99688473520249, "grad_norm": 1.609375, "learning_rate": 1.758169606533974e-05, "loss": 0.1458, "step": 19319 }, { "epoch": 120.0, "grad_norm": 1.1328125, "learning_rate": 1.7579343305030985e-05, "loss": 0.0674, "step": 19320 }, { "epoch": 120.00623052959502, "grad_norm": 1.375, "learning_rate": 1.7576990616793137e-05, "loss": 0.1036, "step": 19321 }, { "epoch": 120.01246105919003, "grad_norm": 1.625, "learning_rate": 1.757463800064903e-05, "loss": 0.1021, "step": 19322 }, { "epoch": 120.01869158878505, "grad_norm": 2.359375, "learning_rate": 1.7572285456621523e-05, "loss": 0.2357, "step": 19323 }, { "epoch": 120.02492211838006, "grad_norm": 2.046875, "learning_rate": 1.7569932984733474e-05, "loss": 0.1615, "step": 19324 }, { "epoch": 120.03115264797508, "grad_norm": 1.9296875, "learning_rate": 1.756758058500771e-05, "loss": 0.1493, "step": 19325 }, { "epoch": 120.03738317757009, "grad_norm": 2.25, "learning_rate": 1.756522825746709e-05, "loss": 0.1839, "step": 19326 }, { "epoch": 120.04361370716511, "grad_norm": 2.875, "learning_rate": 1.756287600213447e-05, "loss": 0.2698, "step": 19327 }, { "epoch": 120.04984423676012, "grad_norm": 1.5078125, "learning_rate": 1.7560523819032683e-05, "loss": 0.096, "step": 19328 }, { "epoch": 120.05607476635514, "grad_norm": 1.7109375, "learning_rate": 1.7558171708184564e-05, "loss": 0.1313, "step": 19329 }, { "epoch": 120.06230529595015, "grad_norm": 1.7109375, "learning_rate": 1.7555819669612976e-05, "loss": 0.1607, "step": 19330 }, { "epoch": 120.06853582554517, "grad_norm": 2.46875, "learning_rate": 1.7553467703340754e-05, "loss": 0.1319, "step": 19331 }, { "epoch": 120.07476635514018, "grad_norm": 1.3046875, "learning_rate": 1.7551115809390745e-05, "loss": 0.1313, "step": 19332 }, { "epoch": 120.0809968847352, "grad_norm": 2.125, "learning_rate": 1.7548763987785783e-05, "loss": 0.1545, "step": 19333 }, { "epoch": 120.08722741433021, "grad_norm": 2.234375, "learning_rate": 1.7546412238548725e-05, "loss": 0.1592, "step": 19334 }, { "epoch": 120.09345794392523, "grad_norm": 3.078125, "learning_rate": 1.7544060561702392e-05, "loss": 0.1212, "step": 19335 }, { "epoch": 120.09968847352025, "grad_norm": 2.046875, "learning_rate": 1.7541708957269637e-05, "loss": 0.1398, "step": 19336 }, { "epoch": 120.10591900311526, "grad_norm": 1.8671875, "learning_rate": 1.7539357425273294e-05, "loss": 0.1324, "step": 19337 }, { "epoch": 120.11214953271028, "grad_norm": 2.5, "learning_rate": 1.7537005965736204e-05, "loss": 0.2578, "step": 19338 }, { "epoch": 120.11838006230529, "grad_norm": 1.296875, "learning_rate": 1.7534654578681204e-05, "loss": 0.1154, "step": 19339 }, { "epoch": 120.1246105919003, "grad_norm": 1.15625, "learning_rate": 1.7532303264131134e-05, "loss": 0.1058, "step": 19340 }, { "epoch": 120.13084112149532, "grad_norm": 2.46875, "learning_rate": 1.7529952022108826e-05, "loss": 0.2013, "step": 19341 }, { "epoch": 120.13707165109034, "grad_norm": 2.453125, "learning_rate": 1.7527600852637112e-05, "loss": 0.173, "step": 19342 }, { "epoch": 120.14330218068535, "grad_norm": 2.265625, "learning_rate": 1.7525249755738832e-05, "loss": 0.1783, "step": 19343 }, { "epoch": 120.14953271028037, "grad_norm": 3.046875, "learning_rate": 1.7522898731436822e-05, "loss": 0.2407, "step": 19344 }, { "epoch": 120.15576323987538, "grad_norm": 2.71875, "learning_rate": 1.7520547779753914e-05, "loss": 0.2175, "step": 19345 }, { "epoch": 120.1619937694704, "grad_norm": 1.4921875, "learning_rate": 1.7518196900712946e-05, "loss": 0.1424, "step": 19346 }, { "epoch": 120.16822429906541, "grad_norm": 2.484375, "learning_rate": 1.7515846094336748e-05, "loss": 0.2531, "step": 19347 }, { "epoch": 120.17445482866043, "grad_norm": 2.046875, "learning_rate": 1.751349536064814e-05, "loss": 0.1376, "step": 19348 }, { "epoch": 120.18068535825545, "grad_norm": 2.703125, "learning_rate": 1.7511144699669966e-05, "loss": 0.1736, "step": 19349 }, { "epoch": 120.18691588785046, "grad_norm": 1.734375, "learning_rate": 1.750879411142505e-05, "loss": 0.1528, "step": 19350 }, { "epoch": 120.19314641744548, "grad_norm": 1.8203125, "learning_rate": 1.7506443595936217e-05, "loss": 0.1495, "step": 19351 }, { "epoch": 120.19937694704049, "grad_norm": 1.9296875, "learning_rate": 1.750409315322631e-05, "loss": 0.1573, "step": 19352 }, { "epoch": 120.20560747663552, "grad_norm": 1.03125, "learning_rate": 1.7501742783318154e-05, "loss": 0.097, "step": 19353 }, { "epoch": 120.21183800623054, "grad_norm": 3.375, "learning_rate": 1.7499392486234566e-05, "loss": 0.2157, "step": 19354 }, { "epoch": 120.21806853582555, "grad_norm": 3.453125, "learning_rate": 1.749704226199837e-05, "loss": 0.2406, "step": 19355 }, { "epoch": 120.22429906542057, "grad_norm": 1.125, "learning_rate": 1.7494692110632405e-05, "loss": 0.1128, "step": 19356 }, { "epoch": 120.23052959501558, "grad_norm": 0.97265625, "learning_rate": 1.7492342032159486e-05, "loss": 0.0908, "step": 19357 }, { "epoch": 120.2367601246106, "grad_norm": 1.21875, "learning_rate": 1.7489992026602447e-05, "loss": 0.0972, "step": 19358 }, { "epoch": 120.24299065420561, "grad_norm": 1.2109375, "learning_rate": 1.7487642093984113e-05, "loss": 0.1214, "step": 19359 }, { "epoch": 120.24922118380063, "grad_norm": 2.625, "learning_rate": 1.7485292234327292e-05, "loss": 0.271, "step": 19360 }, { "epoch": 120.25545171339564, "grad_norm": 2.21875, "learning_rate": 1.7482942447654815e-05, "loss": 0.1165, "step": 19361 }, { "epoch": 120.26168224299066, "grad_norm": 1.40625, "learning_rate": 1.7480592733989504e-05, "loss": 0.1449, "step": 19362 }, { "epoch": 120.26791277258567, "grad_norm": 2.46875, "learning_rate": 1.7478243093354184e-05, "loss": 0.1924, "step": 19363 }, { "epoch": 120.27414330218069, "grad_norm": 2.578125, "learning_rate": 1.747589352577166e-05, "loss": 0.159, "step": 19364 }, { "epoch": 120.2803738317757, "grad_norm": 3.109375, "learning_rate": 1.747354403126477e-05, "loss": 0.2694, "step": 19365 }, { "epoch": 120.28660436137072, "grad_norm": 1.3828125, "learning_rate": 1.7471194609856332e-05, "loss": 0.1135, "step": 19366 }, { "epoch": 120.29283489096574, "grad_norm": 4.3125, "learning_rate": 1.746884526156915e-05, "loss": 0.1082, "step": 19367 }, { "epoch": 120.29906542056075, "grad_norm": 1.6953125, "learning_rate": 1.746649598642604e-05, "loss": 0.1386, "step": 19368 }, { "epoch": 120.30529595015577, "grad_norm": 1.9765625, "learning_rate": 1.7464146784449837e-05, "loss": 0.1738, "step": 19369 }, { "epoch": 120.31152647975078, "grad_norm": 1.234375, "learning_rate": 1.7461797655663344e-05, "loss": 0.1201, "step": 19370 }, { "epoch": 120.3177570093458, "grad_norm": 2.78125, "learning_rate": 1.7459448600089377e-05, "loss": 0.3483, "step": 19371 }, { "epoch": 120.32398753894081, "grad_norm": 1.78125, "learning_rate": 1.745709961775076e-05, "loss": 0.168, "step": 19372 }, { "epoch": 120.33021806853583, "grad_norm": 3.0625, "learning_rate": 1.7454750708670296e-05, "loss": 0.138, "step": 19373 }, { "epoch": 120.33644859813084, "grad_norm": 3.359375, "learning_rate": 1.7452401872870798e-05, "loss": 0.2289, "step": 19374 }, { "epoch": 120.34267912772586, "grad_norm": 2.109375, "learning_rate": 1.745005311037508e-05, "loss": 0.1217, "step": 19375 }, { "epoch": 120.34890965732087, "grad_norm": 3.25, "learning_rate": 1.744770442120596e-05, "loss": 0.1809, "step": 19376 }, { "epoch": 120.35514018691589, "grad_norm": 2.3125, "learning_rate": 1.744535580538624e-05, "loss": 0.1222, "step": 19377 }, { "epoch": 120.3613707165109, "grad_norm": 1.0390625, "learning_rate": 1.7443007262938747e-05, "loss": 0.0939, "step": 19378 }, { "epoch": 120.36760124610592, "grad_norm": 2.4375, "learning_rate": 1.7440658793886264e-05, "loss": 0.2624, "step": 19379 }, { "epoch": 120.37383177570094, "grad_norm": 1.4921875, "learning_rate": 1.7438310398251618e-05, "loss": 0.1146, "step": 19380 }, { "epoch": 120.38006230529595, "grad_norm": 2.875, "learning_rate": 1.743596207605761e-05, "loss": 0.1879, "step": 19381 }, { "epoch": 120.38629283489097, "grad_norm": 1.9765625, "learning_rate": 1.7433613827327055e-05, "loss": 0.15, "step": 19382 }, { "epoch": 120.39252336448598, "grad_norm": 1.671875, "learning_rate": 1.7431265652082752e-05, "loss": 0.1516, "step": 19383 }, { "epoch": 120.398753894081, "grad_norm": 2.6875, "learning_rate": 1.7428917550347506e-05, "loss": 0.1888, "step": 19384 }, { "epoch": 120.40498442367601, "grad_norm": 3.28125, "learning_rate": 1.742656952214414e-05, "loss": 0.2255, "step": 19385 }, { "epoch": 120.41121495327103, "grad_norm": 2.046875, "learning_rate": 1.7424221567495435e-05, "loss": 0.1588, "step": 19386 }, { "epoch": 120.41744548286604, "grad_norm": 2.40625, "learning_rate": 1.742187368642421e-05, "loss": 0.1979, "step": 19387 }, { "epoch": 120.42367601246106, "grad_norm": 1.46875, "learning_rate": 1.741952587895325e-05, "loss": 0.0963, "step": 19388 }, { "epoch": 120.42990654205607, "grad_norm": 1.3515625, "learning_rate": 1.7417178145105383e-05, "loss": 0.1143, "step": 19389 }, { "epoch": 120.43613707165109, "grad_norm": 1.171875, "learning_rate": 1.7414830484903387e-05, "loss": 0.0995, "step": 19390 }, { "epoch": 120.4423676012461, "grad_norm": 3.140625, "learning_rate": 1.741248289837009e-05, "loss": 0.2796, "step": 19391 }, { "epoch": 120.44859813084112, "grad_norm": 2.84375, "learning_rate": 1.7410135385528266e-05, "loss": 0.2585, "step": 19392 }, { "epoch": 120.45482866043614, "grad_norm": 1.8984375, "learning_rate": 1.740778794640072e-05, "loss": 0.1127, "step": 19393 }, { "epoch": 120.46105919003115, "grad_norm": 1.2265625, "learning_rate": 1.7405440581010257e-05, "loss": 0.1128, "step": 19394 }, { "epoch": 120.46728971962617, "grad_norm": 2.25, "learning_rate": 1.7403093289379676e-05, "loss": 0.0949, "step": 19395 }, { "epoch": 120.47352024922118, "grad_norm": 2.203125, "learning_rate": 1.740074607153177e-05, "loss": 0.129, "step": 19396 }, { "epoch": 120.4797507788162, "grad_norm": 2.359375, "learning_rate": 1.7398398927489345e-05, "loss": 0.2049, "step": 19397 }, { "epoch": 120.48598130841121, "grad_norm": 4.0, "learning_rate": 1.739605185727518e-05, "loss": 0.3183, "step": 19398 }, { "epoch": 120.49221183800623, "grad_norm": 1.8203125, "learning_rate": 1.739370486091208e-05, "loss": 0.1055, "step": 19399 }, { "epoch": 120.49844236760124, "grad_norm": 2.046875, "learning_rate": 1.7391357938422842e-05, "loss": 0.1569, "step": 19400 }, { "epoch": 120.50467289719626, "grad_norm": 1.90625, "learning_rate": 1.7389011089830252e-05, "loss": 0.1058, "step": 19401 }, { "epoch": 120.51090342679127, "grad_norm": 3.484375, "learning_rate": 1.738666431515711e-05, "loss": 0.2366, "step": 19402 }, { "epoch": 120.51713395638629, "grad_norm": 2.828125, "learning_rate": 1.7384317614426203e-05, "loss": 0.1451, "step": 19403 }, { "epoch": 120.5233644859813, "grad_norm": 3.46875, "learning_rate": 1.738197098766034e-05, "loss": 0.0943, "step": 19404 }, { "epoch": 120.52959501557632, "grad_norm": 1.546875, "learning_rate": 1.737962443488228e-05, "loss": 0.1003, "step": 19405 }, { "epoch": 120.53582554517133, "grad_norm": 3.796875, "learning_rate": 1.737727795611484e-05, "loss": 0.2018, "step": 19406 }, { "epoch": 120.54205607476635, "grad_norm": 3.703125, "learning_rate": 1.7374931551380796e-05, "loss": 0.1897, "step": 19407 }, { "epoch": 120.54828660436137, "grad_norm": 2.515625, "learning_rate": 1.737258522070294e-05, "loss": 0.1851, "step": 19408 }, { "epoch": 120.55451713395638, "grad_norm": 2.453125, "learning_rate": 1.7370238964104067e-05, "loss": 0.2271, "step": 19409 }, { "epoch": 120.5607476635514, "grad_norm": 2.203125, "learning_rate": 1.7367892781606962e-05, "loss": 0.1421, "step": 19410 }, { "epoch": 120.56697819314641, "grad_norm": 2.375, "learning_rate": 1.7365546673234395e-05, "loss": 0.2807, "step": 19411 }, { "epoch": 120.57320872274143, "grad_norm": 2.203125, "learning_rate": 1.7363200639009174e-05, "loss": 0.114, "step": 19412 }, { "epoch": 120.57943925233644, "grad_norm": 1.5859375, "learning_rate": 1.736085467895407e-05, "loss": 0.0991, "step": 19413 }, { "epoch": 120.58566978193146, "grad_norm": 1.2890625, "learning_rate": 1.735850879309187e-05, "loss": 0.1026, "step": 19414 }, { "epoch": 120.59190031152647, "grad_norm": 2.796875, "learning_rate": 1.7356162981445368e-05, "loss": 0.2456, "step": 19415 }, { "epoch": 120.59813084112149, "grad_norm": 3.234375, "learning_rate": 1.7353817244037334e-05, "loss": 0.1672, "step": 19416 }, { "epoch": 120.60436137071652, "grad_norm": 1.7890625, "learning_rate": 1.7351471580890566e-05, "loss": 0.1248, "step": 19417 }, { "epoch": 120.61059190031153, "grad_norm": 2.5, "learning_rate": 1.734912599202782e-05, "loss": 0.2276, "step": 19418 }, { "epoch": 120.61682242990655, "grad_norm": 3.015625, "learning_rate": 1.73467804774719e-05, "loss": 0.2275, "step": 19419 }, { "epoch": 120.62305295950156, "grad_norm": 2.640625, "learning_rate": 1.7344435037245575e-05, "loss": 0.2162, "step": 19420 }, { "epoch": 120.62928348909658, "grad_norm": 2.5625, "learning_rate": 1.7342089671371624e-05, "loss": 0.121, "step": 19421 }, { "epoch": 120.6355140186916, "grad_norm": 1.375, "learning_rate": 1.7339744379872836e-05, "loss": 0.1263, "step": 19422 }, { "epoch": 120.64174454828661, "grad_norm": 1.21875, "learning_rate": 1.7337399162771982e-05, "loss": 0.0981, "step": 19423 }, { "epoch": 120.64797507788163, "grad_norm": 1.8359375, "learning_rate": 1.733505402009184e-05, "loss": 0.1841, "step": 19424 }, { "epoch": 120.65420560747664, "grad_norm": 2.640625, "learning_rate": 1.733270895185517e-05, "loss": 0.2186, "step": 19425 }, { "epoch": 120.66043613707166, "grad_norm": 2.75, "learning_rate": 1.733036395808478e-05, "loss": 0.1745, "step": 19426 }, { "epoch": 120.66666666666667, "grad_norm": 1.6640625, "learning_rate": 1.7328019038803416e-05, "loss": 0.1503, "step": 19427 }, { "epoch": 120.67289719626169, "grad_norm": 2.453125, "learning_rate": 1.7325674194033874e-05, "loss": 0.1169, "step": 19428 }, { "epoch": 120.6791277258567, "grad_norm": 2.390625, "learning_rate": 1.7323329423798922e-05, "loss": 0.178, "step": 19429 }, { "epoch": 120.68535825545172, "grad_norm": 2.0625, "learning_rate": 1.7320984728121324e-05, "loss": 0.1431, "step": 19430 }, { "epoch": 120.69158878504673, "grad_norm": 1.09375, "learning_rate": 1.731864010702385e-05, "loss": 0.105, "step": 19431 }, { "epoch": 120.69781931464175, "grad_norm": 1.3984375, "learning_rate": 1.731629556052929e-05, "loss": 0.1157, "step": 19432 }, { "epoch": 120.70404984423676, "grad_norm": 2.625, "learning_rate": 1.7313951088660395e-05, "loss": 0.2544, "step": 19433 }, { "epoch": 120.71028037383178, "grad_norm": 1.5546875, "learning_rate": 1.7311606691439942e-05, "loss": 0.178, "step": 19434 }, { "epoch": 120.7165109034268, "grad_norm": 1.6953125, "learning_rate": 1.7309262368890712e-05, "loss": 0.1208, "step": 19435 }, { "epoch": 120.72274143302181, "grad_norm": 2.09375, "learning_rate": 1.730691812103546e-05, "loss": 0.1799, "step": 19436 }, { "epoch": 120.72897196261682, "grad_norm": 2.3125, "learning_rate": 1.7304573947896956e-05, "loss": 0.1831, "step": 19437 }, { "epoch": 120.73520249221184, "grad_norm": 3.546875, "learning_rate": 1.7302229849497963e-05, "loss": 0.2756, "step": 19438 }, { "epoch": 120.74143302180686, "grad_norm": 2.484375, "learning_rate": 1.7299885825861256e-05, "loss": 0.1486, "step": 19439 }, { "epoch": 120.74766355140187, "grad_norm": 1.390625, "learning_rate": 1.7297541877009593e-05, "loss": 0.1182, "step": 19440 }, { "epoch": 120.75389408099689, "grad_norm": 1.0390625, "learning_rate": 1.729519800296575e-05, "loss": 0.0917, "step": 19441 }, { "epoch": 120.7601246105919, "grad_norm": 3.75, "learning_rate": 1.7292854203752487e-05, "loss": 0.3167, "step": 19442 }, { "epoch": 120.76635514018692, "grad_norm": 1.984375, "learning_rate": 1.729051047939256e-05, "loss": 0.1906, "step": 19443 }, { "epoch": 120.77258566978193, "grad_norm": 1.390625, "learning_rate": 1.728816682990873e-05, "loss": 0.1071, "step": 19444 }, { "epoch": 120.77881619937695, "grad_norm": 2.203125, "learning_rate": 1.7285823255323773e-05, "loss": 0.1891, "step": 19445 }, { "epoch": 120.78504672897196, "grad_norm": 1.9140625, "learning_rate": 1.728347975566044e-05, "loss": 0.1012, "step": 19446 }, { "epoch": 120.79127725856698, "grad_norm": 1.7890625, "learning_rate": 1.7281136330941493e-05, "loss": 0.1247, "step": 19447 }, { "epoch": 120.797507788162, "grad_norm": 2.5, "learning_rate": 1.72787929811897e-05, "loss": 0.2103, "step": 19448 }, { "epoch": 120.80373831775701, "grad_norm": 2.203125, "learning_rate": 1.727644970642781e-05, "loss": 0.2058, "step": 19449 }, { "epoch": 120.80996884735202, "grad_norm": 2.375, "learning_rate": 1.727410650667858e-05, "loss": 0.1671, "step": 19450 }, { "epoch": 120.81619937694704, "grad_norm": 2.765625, "learning_rate": 1.727176338196477e-05, "loss": 0.1275, "step": 19451 }, { "epoch": 120.82242990654206, "grad_norm": 1.5, "learning_rate": 1.7269420332309144e-05, "loss": 0.1219, "step": 19452 }, { "epoch": 120.82866043613707, "grad_norm": 1.8828125, "learning_rate": 1.7267077357734445e-05, "loss": 0.1169, "step": 19453 }, { "epoch": 120.83489096573209, "grad_norm": 1.2421875, "learning_rate": 1.7264734458263446e-05, "loss": 0.1257, "step": 19454 }, { "epoch": 120.8411214953271, "grad_norm": 2.953125, "learning_rate": 1.7262391633918897e-05, "loss": 0.262, "step": 19455 }, { "epoch": 120.84735202492212, "grad_norm": 1.3359375, "learning_rate": 1.726004888472354e-05, "loss": 0.1051, "step": 19456 }, { "epoch": 120.85358255451713, "grad_norm": 1.375, "learning_rate": 1.7257706210700132e-05, "loss": 0.1351, "step": 19457 }, { "epoch": 120.85981308411215, "grad_norm": 2.671875, "learning_rate": 1.7255363611871427e-05, "loss": 0.1959, "step": 19458 }, { "epoch": 120.86604361370716, "grad_norm": 3.71875, "learning_rate": 1.725302108826019e-05, "loss": 0.1409, "step": 19459 }, { "epoch": 120.87227414330218, "grad_norm": 2.640625, "learning_rate": 1.7250678639889147e-05, "loss": 0.1621, "step": 19460 }, { "epoch": 120.8785046728972, "grad_norm": 3.546875, "learning_rate": 1.7248336266781078e-05, "loss": 0.2281, "step": 19461 }, { "epoch": 120.88473520249221, "grad_norm": 3.171875, "learning_rate": 1.7245993968958706e-05, "loss": 0.1568, "step": 19462 }, { "epoch": 120.89096573208722, "grad_norm": 3.734375, "learning_rate": 1.7243651746444795e-05, "loss": 0.3202, "step": 19463 }, { "epoch": 120.89719626168224, "grad_norm": 1.8203125, "learning_rate": 1.7241309599262083e-05, "loss": 0.1282, "step": 19464 }, { "epoch": 120.90342679127725, "grad_norm": 2.296875, "learning_rate": 1.7238967527433324e-05, "loss": 0.1818, "step": 19465 }, { "epoch": 120.90965732087227, "grad_norm": 1.8671875, "learning_rate": 1.723662553098127e-05, "loss": 0.133, "step": 19466 }, { "epoch": 120.91588785046729, "grad_norm": 4.4375, "learning_rate": 1.7234283609928665e-05, "loss": 0.2336, "step": 19467 }, { "epoch": 120.9221183800623, "grad_norm": 2.765625, "learning_rate": 1.7231941764298237e-05, "loss": 0.1408, "step": 19468 }, { "epoch": 120.92834890965732, "grad_norm": 1.84375, "learning_rate": 1.722959999411275e-05, "loss": 0.1328, "step": 19469 }, { "epoch": 120.93457943925233, "grad_norm": 2.0625, "learning_rate": 1.7227258299394937e-05, "loss": 0.1923, "step": 19470 }, { "epoch": 120.94080996884735, "grad_norm": 2.53125, "learning_rate": 1.7224916680167544e-05, "loss": 0.0912, "step": 19471 }, { "epoch": 120.94704049844236, "grad_norm": 2.234375, "learning_rate": 1.7222575136453316e-05, "loss": 0.1974, "step": 19472 }, { "epoch": 120.95327102803738, "grad_norm": 1.03125, "learning_rate": 1.7220233668274994e-05, "loss": 0.1084, "step": 19473 }, { "epoch": 120.9595015576324, "grad_norm": 2.125, "learning_rate": 1.7217892275655327e-05, "loss": 0.1469, "step": 19474 }, { "epoch": 120.96573208722741, "grad_norm": 2.828125, "learning_rate": 1.7215550958617034e-05, "loss": 0.3209, "step": 19475 }, { "epoch": 120.97196261682242, "grad_norm": 1.8671875, "learning_rate": 1.7213209717182872e-05, "loss": 0.1479, "step": 19476 }, { "epoch": 120.97819314641744, "grad_norm": 1.5, "learning_rate": 1.721086855137557e-05, "loss": 0.1142, "step": 19477 }, { "epoch": 120.98442367601245, "grad_norm": 2.875, "learning_rate": 1.7208527461217873e-05, "loss": 0.3228, "step": 19478 }, { "epoch": 120.99065420560747, "grad_norm": 1.984375, "learning_rate": 1.720618644673251e-05, "loss": 0.1674, "step": 19479 }, { "epoch": 120.99688473520249, "grad_norm": 2.4375, "learning_rate": 1.7203845507942232e-05, "loss": 0.1666, "step": 19480 }, { "epoch": 121.0, "grad_norm": 2.25, "learning_rate": 1.7201504644869753e-05, "loss": 0.1046, "step": 19481 }, { "epoch": 121.00623052959502, "grad_norm": 4.78125, "learning_rate": 1.7199163857537825e-05, "loss": 0.39, "step": 19482 }, { "epoch": 121.01246105919003, "grad_norm": 2.1875, "learning_rate": 1.719682314596918e-05, "loss": 0.0922, "step": 19483 }, { "epoch": 121.01869158878505, "grad_norm": 2.953125, "learning_rate": 1.719448251018654e-05, "loss": 0.1638, "step": 19484 }, { "epoch": 121.02492211838006, "grad_norm": 1.859375, "learning_rate": 1.7192141950212653e-05, "loss": 0.1322, "step": 19485 }, { "epoch": 121.03115264797508, "grad_norm": 1.9140625, "learning_rate": 1.718980146607024e-05, "loss": 0.1011, "step": 19486 }, { "epoch": 121.03738317757009, "grad_norm": 1.609375, "learning_rate": 1.7187461057782048e-05, "loss": 0.1346, "step": 19487 }, { "epoch": 121.04361370716511, "grad_norm": 3.15625, "learning_rate": 1.7185120725370783e-05, "loss": 0.1624, "step": 19488 }, { "epoch": 121.04984423676012, "grad_norm": 2.28125, "learning_rate": 1.718278046885919e-05, "loss": 0.1557, "step": 19489 }, { "epoch": 121.05607476635514, "grad_norm": 1.71875, "learning_rate": 1.7180440288269993e-05, "loss": 0.1174, "step": 19490 }, { "epoch": 121.06230529595015, "grad_norm": 1.1953125, "learning_rate": 1.717810018362593e-05, "loss": 0.0894, "step": 19491 }, { "epoch": 121.06853582554517, "grad_norm": 1.796875, "learning_rate": 1.717576015494972e-05, "loss": 0.1213, "step": 19492 }, { "epoch": 121.07476635514018, "grad_norm": 1.875, "learning_rate": 1.7173420202264094e-05, "loss": 0.1503, "step": 19493 }, { "epoch": 121.0809968847352, "grad_norm": 1.4453125, "learning_rate": 1.7171080325591768e-05, "loss": 0.1065, "step": 19494 }, { "epoch": 121.08722741433021, "grad_norm": 1.484375, "learning_rate": 1.716874052495548e-05, "loss": 0.1037, "step": 19495 }, { "epoch": 121.09345794392523, "grad_norm": 4.46875, "learning_rate": 1.7166400800377948e-05, "loss": 0.1823, "step": 19496 }, { "epoch": 121.09968847352025, "grad_norm": 2.859375, "learning_rate": 1.7164061151881895e-05, "loss": 0.3495, "step": 19497 }, { "epoch": 121.10591900311526, "grad_norm": 0.96875, "learning_rate": 1.716172157949005e-05, "loss": 0.0998, "step": 19498 }, { "epoch": 121.11214953271028, "grad_norm": 1.21875, "learning_rate": 1.7159382083225135e-05, "loss": 0.111, "step": 19499 }, { "epoch": 121.11838006230529, "grad_norm": 2.171875, "learning_rate": 1.7157042663109863e-05, "loss": 0.2232, "step": 19500 }, { "epoch": 121.1246105919003, "grad_norm": 3.0625, "learning_rate": 1.715470331916696e-05, "loss": 0.2299, "step": 19501 }, { "epoch": 121.13084112149532, "grad_norm": 1.21875, "learning_rate": 1.715236405141915e-05, "loss": 0.0971, "step": 19502 }, { "epoch": 121.13707165109034, "grad_norm": 1.625, "learning_rate": 1.715002485988914e-05, "loss": 0.1302, "step": 19503 }, { "epoch": 121.14330218068535, "grad_norm": 2.21875, "learning_rate": 1.7147685744599668e-05, "loss": 0.2288, "step": 19504 }, { "epoch": 121.14953271028037, "grad_norm": 1.640625, "learning_rate": 1.714534670557344e-05, "loss": 0.1195, "step": 19505 }, { "epoch": 121.15576323987538, "grad_norm": 2.125, "learning_rate": 1.7143007742833178e-05, "loss": 0.1523, "step": 19506 }, { "epoch": 121.1619937694704, "grad_norm": 1.7578125, "learning_rate": 1.7140668856401588e-05, "loss": 0.1387, "step": 19507 }, { "epoch": 121.16822429906541, "grad_norm": 1.90625, "learning_rate": 1.7138330046301393e-05, "loss": 0.1662, "step": 19508 }, { "epoch": 121.17445482866043, "grad_norm": 0.984375, "learning_rate": 1.7135991312555312e-05, "loss": 0.082, "step": 19509 }, { "epoch": 121.18068535825545, "grad_norm": 1.6640625, "learning_rate": 1.7133652655186053e-05, "loss": 0.1469, "step": 19510 }, { "epoch": 121.18691588785046, "grad_norm": 3.59375, "learning_rate": 1.713131407421633e-05, "loss": 0.2958, "step": 19511 }, { "epoch": 121.19314641744548, "grad_norm": 1.5234375, "learning_rate": 1.712897556966887e-05, "loss": 0.1078, "step": 19512 }, { "epoch": 121.19937694704049, "grad_norm": 1.7890625, "learning_rate": 1.7126637141566366e-05, "loss": 0.0961, "step": 19513 }, { "epoch": 121.20560747663552, "grad_norm": 2.640625, "learning_rate": 1.7124298789931528e-05, "loss": 0.2317, "step": 19514 }, { "epoch": 121.21183800623054, "grad_norm": 1.53125, "learning_rate": 1.7121960514787083e-05, "loss": 0.1363, "step": 19515 }, { "epoch": 121.21806853582555, "grad_norm": 2.53125, "learning_rate": 1.711962231615573e-05, "loss": 0.349, "step": 19516 }, { "epoch": 121.22429906542057, "grad_norm": 2.546875, "learning_rate": 1.7117284194060174e-05, "loss": 0.1941, "step": 19517 }, { "epoch": 121.23052959501558, "grad_norm": 3.0625, "learning_rate": 1.711494614852314e-05, "loss": 0.218, "step": 19518 }, { "epoch": 121.2367601246106, "grad_norm": 2.46875, "learning_rate": 1.7112608179567324e-05, "loss": 0.1962, "step": 19519 }, { "epoch": 121.24299065420561, "grad_norm": 3.15625, "learning_rate": 1.711027028721543e-05, "loss": 0.2401, "step": 19520 }, { "epoch": 121.24922118380063, "grad_norm": 2.4375, "learning_rate": 1.7107932471490163e-05, "loss": 0.2563, "step": 19521 }, { "epoch": 121.25545171339564, "grad_norm": 1.890625, "learning_rate": 1.7105594732414242e-05, "loss": 0.107, "step": 19522 }, { "epoch": 121.26168224299066, "grad_norm": 2.359375, "learning_rate": 1.7103257070010355e-05, "loss": 0.1938, "step": 19523 }, { "epoch": 121.26791277258567, "grad_norm": 3.25, "learning_rate": 1.710091948430122e-05, "loss": 0.2096, "step": 19524 }, { "epoch": 121.27414330218069, "grad_norm": 2.25, "learning_rate": 1.709858197530954e-05, "loss": 0.1286, "step": 19525 }, { "epoch": 121.2803738317757, "grad_norm": 1.828125, "learning_rate": 1.7096244543058004e-05, "loss": 0.1539, "step": 19526 }, { "epoch": 121.28660436137072, "grad_norm": 3.671875, "learning_rate": 1.7093907187569315e-05, "loss": 0.2183, "step": 19527 }, { "epoch": 121.29283489096574, "grad_norm": 2.421875, "learning_rate": 1.709156990886619e-05, "loss": 0.2484, "step": 19528 }, { "epoch": 121.29906542056075, "grad_norm": 3.140625, "learning_rate": 1.7089232706971313e-05, "loss": 0.2669, "step": 19529 }, { "epoch": 121.30529595015577, "grad_norm": 1.890625, "learning_rate": 1.7086895581907386e-05, "loss": 0.1622, "step": 19530 }, { "epoch": 121.31152647975078, "grad_norm": 1.765625, "learning_rate": 1.7084558533697124e-05, "loss": 0.1219, "step": 19531 }, { "epoch": 121.3177570093458, "grad_norm": 1.2734375, "learning_rate": 1.7082221562363203e-05, "loss": 0.113, "step": 19532 }, { "epoch": 121.32398753894081, "grad_norm": 2.546875, "learning_rate": 1.707988466792833e-05, "loss": 0.1887, "step": 19533 }, { "epoch": 121.33021806853583, "grad_norm": 1.421875, "learning_rate": 1.70775478504152e-05, "loss": 0.1554, "step": 19534 }, { "epoch": 121.33644859813084, "grad_norm": 2.15625, "learning_rate": 1.7075211109846508e-05, "loss": 0.1359, "step": 19535 }, { "epoch": 121.34267912772586, "grad_norm": 1.4765625, "learning_rate": 1.7072874446244945e-05, "loss": 0.1217, "step": 19536 }, { "epoch": 121.34890965732087, "grad_norm": 3.3125, "learning_rate": 1.7070537859633216e-05, "loss": 0.1862, "step": 19537 }, { "epoch": 121.35514018691589, "grad_norm": 3.0, "learning_rate": 1.7068201350034017e-05, "loss": 0.194, "step": 19538 }, { "epoch": 121.3613707165109, "grad_norm": 1.875, "learning_rate": 1.7065864917470022e-05, "loss": 0.1493, "step": 19539 }, { "epoch": 121.36760124610592, "grad_norm": 1.671875, "learning_rate": 1.706352856196393e-05, "loss": 0.1185, "step": 19540 }, { "epoch": 121.37383177570094, "grad_norm": 2.59375, "learning_rate": 1.706119228353844e-05, "loss": 0.0953, "step": 19541 }, { "epoch": 121.38006230529595, "grad_norm": 1.1328125, "learning_rate": 1.705885608221624e-05, "loss": 0.104, "step": 19542 }, { "epoch": 121.38629283489097, "grad_norm": 1.6640625, "learning_rate": 1.7056519958020007e-05, "loss": 0.087, "step": 19543 }, { "epoch": 121.39252336448598, "grad_norm": 1.8984375, "learning_rate": 1.7054183910972456e-05, "loss": 0.1658, "step": 19544 }, { "epoch": 121.398753894081, "grad_norm": 2.15625, "learning_rate": 1.705184794109625e-05, "loss": 0.1586, "step": 19545 }, { "epoch": 121.40498442367601, "grad_norm": 3.359375, "learning_rate": 1.7049512048414085e-05, "loss": 0.2157, "step": 19546 }, { "epoch": 121.41121495327103, "grad_norm": 3.640625, "learning_rate": 1.7047176232948646e-05, "loss": 0.2889, "step": 19547 }, { "epoch": 121.41744548286604, "grad_norm": 1.734375, "learning_rate": 1.7044840494722626e-05, "loss": 0.1411, "step": 19548 }, { "epoch": 121.42367601246106, "grad_norm": 3.109375, "learning_rate": 1.7042504833758697e-05, "loss": 0.2808, "step": 19549 }, { "epoch": 121.42990654205607, "grad_norm": 3.71875, "learning_rate": 1.7040169250079568e-05, "loss": 0.2674, "step": 19550 }, { "epoch": 121.43613707165109, "grad_norm": 1.5703125, "learning_rate": 1.7037833743707892e-05, "loss": 0.145, "step": 19551 }, { "epoch": 121.4423676012461, "grad_norm": 1.296875, "learning_rate": 1.7035498314666375e-05, "loss": 0.0967, "step": 19552 }, { "epoch": 121.44859813084112, "grad_norm": 1.2734375, "learning_rate": 1.7033162962977678e-05, "loss": 0.1272, "step": 19553 }, { "epoch": 121.45482866043614, "grad_norm": 2.609375, "learning_rate": 1.7030827688664505e-05, "loss": 0.3137, "step": 19554 }, { "epoch": 121.46105919003115, "grad_norm": 2.984375, "learning_rate": 1.7028492491749527e-05, "loss": 0.2155, "step": 19555 }, { "epoch": 121.46728971962617, "grad_norm": 1.859375, "learning_rate": 1.7026157372255414e-05, "loss": 0.1214, "step": 19556 }, { "epoch": 121.47352024922118, "grad_norm": 1.0703125, "learning_rate": 1.702382233020487e-05, "loss": 0.1044, "step": 19557 }, { "epoch": 121.4797507788162, "grad_norm": 1.53125, "learning_rate": 1.702148736562055e-05, "loss": 0.1239, "step": 19558 }, { "epoch": 121.48598130841121, "grad_norm": 2.140625, "learning_rate": 1.701915247852514e-05, "loss": 0.1462, "step": 19559 }, { "epoch": 121.49221183800623, "grad_norm": 2.90625, "learning_rate": 1.7016817668941308e-05, "loss": 0.1587, "step": 19560 }, { "epoch": 121.49844236760124, "grad_norm": 0.87890625, "learning_rate": 1.7014482936891747e-05, "loss": 0.0868, "step": 19561 }, { "epoch": 121.50467289719626, "grad_norm": 3.5, "learning_rate": 1.7012148282399125e-05, "loss": 0.172, "step": 19562 }, { "epoch": 121.51090342679127, "grad_norm": 3.78125, "learning_rate": 1.7009813705486122e-05, "loss": 0.1858, "step": 19563 }, { "epoch": 121.51713395638629, "grad_norm": 1.6953125, "learning_rate": 1.700747920617539e-05, "loss": 0.1407, "step": 19564 }, { "epoch": 121.5233644859813, "grad_norm": 1.7578125, "learning_rate": 1.7005144784489628e-05, "loss": 0.1232, "step": 19565 }, { "epoch": 121.52959501557632, "grad_norm": 2.3125, "learning_rate": 1.7002810440451495e-05, "loss": 0.1311, "step": 19566 }, { "epoch": 121.53582554517133, "grad_norm": 2.203125, "learning_rate": 1.7000476174083663e-05, "loss": 0.1777, "step": 19567 }, { "epoch": 121.54205607476635, "grad_norm": 1.8984375, "learning_rate": 1.6998141985408804e-05, "loss": 0.1457, "step": 19568 }, { "epoch": 121.54828660436137, "grad_norm": 2.4375, "learning_rate": 1.6995807874449605e-05, "loss": 0.157, "step": 19569 }, { "epoch": 121.55451713395638, "grad_norm": 1.6875, "learning_rate": 1.6993473841228707e-05, "loss": 0.1646, "step": 19570 }, { "epoch": 121.5607476635514, "grad_norm": 2.03125, "learning_rate": 1.6991139885768788e-05, "loss": 0.1787, "step": 19571 }, { "epoch": 121.56697819314641, "grad_norm": 2.59375, "learning_rate": 1.6988806008092524e-05, "loss": 0.2004, "step": 19572 }, { "epoch": 121.57320872274143, "grad_norm": 2.09375, "learning_rate": 1.6986472208222576e-05, "loss": 0.1856, "step": 19573 }, { "epoch": 121.57943925233644, "grad_norm": 2.0, "learning_rate": 1.698413848618161e-05, "loss": 0.1844, "step": 19574 }, { "epoch": 121.58566978193146, "grad_norm": 2.5, "learning_rate": 1.69818048419923e-05, "loss": 0.2204, "step": 19575 }, { "epoch": 121.59190031152647, "grad_norm": 1.8828125, "learning_rate": 1.6979471275677304e-05, "loss": 0.0924, "step": 19576 }, { "epoch": 121.59813084112149, "grad_norm": 3.671875, "learning_rate": 1.6977137787259284e-05, "loss": 0.1623, "step": 19577 }, { "epoch": 121.60436137071652, "grad_norm": 1.359375, "learning_rate": 1.6974804376760905e-05, "loss": 0.1216, "step": 19578 }, { "epoch": 121.61059190031153, "grad_norm": 2.828125, "learning_rate": 1.6972471044204828e-05, "loss": 0.2082, "step": 19579 }, { "epoch": 121.61682242990655, "grad_norm": 0.9296875, "learning_rate": 1.6970137789613715e-05, "loss": 0.1024, "step": 19580 }, { "epoch": 121.62305295950156, "grad_norm": 2.203125, "learning_rate": 1.6967804613010234e-05, "loss": 0.2114, "step": 19581 }, { "epoch": 121.62928348909658, "grad_norm": 2.625, "learning_rate": 1.6965471514417047e-05, "loss": 0.1906, "step": 19582 }, { "epoch": 121.6355140186916, "grad_norm": 1.3359375, "learning_rate": 1.69631384938568e-05, "loss": 0.1134, "step": 19583 }, { "epoch": 121.64174454828661, "grad_norm": 2.5, "learning_rate": 1.6960805551352152e-05, "loss": 0.2126, "step": 19584 }, { "epoch": 121.64797507788163, "grad_norm": 3.21875, "learning_rate": 1.6958472686925774e-05, "loss": 0.2532, "step": 19585 }, { "epoch": 121.65420560747664, "grad_norm": 2.078125, "learning_rate": 1.695613990060031e-05, "loss": 0.1627, "step": 19586 }, { "epoch": 121.66043613707166, "grad_norm": 3.46875, "learning_rate": 1.695380719239843e-05, "loss": 0.1544, "step": 19587 }, { "epoch": 121.66666666666667, "grad_norm": 2.9375, "learning_rate": 1.6951474562342786e-05, "loss": 0.1934, "step": 19588 }, { "epoch": 121.67289719626169, "grad_norm": 1.7734375, "learning_rate": 1.6949142010456027e-05, "loss": 0.1158, "step": 19589 }, { "epoch": 121.6791277258567, "grad_norm": 1.4453125, "learning_rate": 1.6946809536760804e-05, "loss": 0.1242, "step": 19590 }, { "epoch": 121.68535825545172, "grad_norm": 2.546875, "learning_rate": 1.6944477141279784e-05, "loss": 0.2302, "step": 19591 }, { "epoch": 121.69158878504673, "grad_norm": 2.3125, "learning_rate": 1.6942144824035612e-05, "loss": 0.163, "step": 19592 }, { "epoch": 121.69781931464175, "grad_norm": 2.28125, "learning_rate": 1.6939812585050928e-05, "loss": 0.1658, "step": 19593 }, { "epoch": 121.70404984423676, "grad_norm": 3.203125, "learning_rate": 1.6937480424348406e-05, "loss": 0.1659, "step": 19594 }, { "epoch": 121.71028037383178, "grad_norm": 2.453125, "learning_rate": 1.6935148341950692e-05, "loss": 0.1267, "step": 19595 }, { "epoch": 121.7165109034268, "grad_norm": 1.671875, "learning_rate": 1.6932816337880425e-05, "loss": 0.1364, "step": 19596 }, { "epoch": 121.72274143302181, "grad_norm": 1.59375, "learning_rate": 1.693048441216025e-05, "loss": 0.1676, "step": 19597 }, { "epoch": 121.72897196261682, "grad_norm": 1.7578125, "learning_rate": 1.692815256481283e-05, "loss": 0.1073, "step": 19598 }, { "epoch": 121.73520249221184, "grad_norm": 1.9765625, "learning_rate": 1.69258207958608e-05, "loss": 0.1488, "step": 19599 }, { "epoch": 121.74143302180686, "grad_norm": 1.5078125, "learning_rate": 1.6923489105326817e-05, "loss": 0.1406, "step": 19600 }, { "epoch": 121.74766355140187, "grad_norm": 2.15625, "learning_rate": 1.6921157493233532e-05, "loss": 0.1564, "step": 19601 }, { "epoch": 121.75389408099689, "grad_norm": 1.9296875, "learning_rate": 1.691882595960357e-05, "loss": 0.1983, "step": 19602 }, { "epoch": 121.7601246105919, "grad_norm": 1.265625, "learning_rate": 1.6916494504459582e-05, "loss": 0.1034, "step": 19603 }, { "epoch": 121.76635514018692, "grad_norm": 1.7265625, "learning_rate": 1.6914163127824224e-05, "loss": 0.1283, "step": 19604 }, { "epoch": 121.77258566978193, "grad_norm": 3.484375, "learning_rate": 1.6911831829720123e-05, "loss": 0.3279, "step": 19605 }, { "epoch": 121.77881619937695, "grad_norm": 1.8671875, "learning_rate": 1.6909500610169926e-05, "loss": 0.146, "step": 19606 }, { "epoch": 121.78504672897196, "grad_norm": 3.140625, "learning_rate": 1.690716946919628e-05, "loss": 0.3171, "step": 19607 }, { "epoch": 121.79127725856698, "grad_norm": 2.359375, "learning_rate": 1.6904838406821832e-05, "loss": 0.1811, "step": 19608 }, { "epoch": 121.797507788162, "grad_norm": 1.421875, "learning_rate": 1.69025074230692e-05, "loss": 0.1077, "step": 19609 }, { "epoch": 121.80373831775701, "grad_norm": 1.796875, "learning_rate": 1.6900176517961036e-05, "loss": 0.1401, "step": 19610 }, { "epoch": 121.80996884735202, "grad_norm": 2.34375, "learning_rate": 1.689784569151998e-05, "loss": 0.1561, "step": 19611 }, { "epoch": 121.81619937694704, "grad_norm": 1.71875, "learning_rate": 1.6895514943768663e-05, "loss": 0.1017, "step": 19612 }, { "epoch": 121.82242990654206, "grad_norm": 2.703125, "learning_rate": 1.6893184274729725e-05, "loss": 0.16, "step": 19613 }, { "epoch": 121.82866043613707, "grad_norm": 1.796875, "learning_rate": 1.6890853684425806e-05, "loss": 0.1695, "step": 19614 }, { "epoch": 121.83489096573209, "grad_norm": 2.609375, "learning_rate": 1.6888523172879535e-05, "loss": 0.2648, "step": 19615 }, { "epoch": 121.8411214953271, "grad_norm": 0.8515625, "learning_rate": 1.688619274011355e-05, "loss": 0.0939, "step": 19616 }, { "epoch": 121.84735202492212, "grad_norm": 1.84375, "learning_rate": 1.6883862386150475e-05, "loss": 0.1032, "step": 19617 }, { "epoch": 121.85358255451713, "grad_norm": 1.75, "learning_rate": 1.688153211101296e-05, "loss": 0.0929, "step": 19618 }, { "epoch": 121.85981308411215, "grad_norm": 1.4140625, "learning_rate": 1.6879201914723618e-05, "loss": 0.1216, "step": 19619 }, { "epoch": 121.86604361370716, "grad_norm": 1.890625, "learning_rate": 1.6876871797305108e-05, "loss": 0.0965, "step": 19620 }, { "epoch": 121.87227414330218, "grad_norm": 2.78125, "learning_rate": 1.6874541758780027e-05, "loss": 0.2334, "step": 19621 }, { "epoch": 121.8785046728972, "grad_norm": 1.203125, "learning_rate": 1.6872211799171026e-05, "loss": 0.1199, "step": 19622 }, { "epoch": 121.88473520249221, "grad_norm": 2.46875, "learning_rate": 1.6869881918500723e-05, "loss": 0.1963, "step": 19623 }, { "epoch": 121.89096573208722, "grad_norm": 2.203125, "learning_rate": 1.686755211679176e-05, "loss": 0.0971, "step": 19624 }, { "epoch": 121.89719626168224, "grad_norm": 2.125, "learning_rate": 1.6865222394066753e-05, "loss": 0.1494, "step": 19625 }, { "epoch": 121.90342679127725, "grad_norm": 1.9453125, "learning_rate": 1.6862892750348326e-05, "loss": 0.1389, "step": 19626 }, { "epoch": 121.90965732087227, "grad_norm": 2.3125, "learning_rate": 1.6860563185659127e-05, "loss": 0.1341, "step": 19627 }, { "epoch": 121.91588785046729, "grad_norm": 1.40625, "learning_rate": 1.6858233700021754e-05, "loss": 0.111, "step": 19628 }, { "epoch": 121.9221183800623, "grad_norm": 2.5625, "learning_rate": 1.6855904293458847e-05, "loss": 0.1347, "step": 19629 }, { "epoch": 121.92834890965732, "grad_norm": 2.625, "learning_rate": 1.685357496599302e-05, "loss": 0.1447, "step": 19630 }, { "epoch": 121.93457943925233, "grad_norm": 1.7109375, "learning_rate": 1.6851245717646905e-05, "loss": 0.1746, "step": 19631 }, { "epoch": 121.94080996884735, "grad_norm": 1.4296875, "learning_rate": 1.6848916548443118e-05, "loss": 0.1244, "step": 19632 }, { "epoch": 121.94704049844236, "grad_norm": 1.390625, "learning_rate": 1.6846587458404294e-05, "loss": 0.1211, "step": 19633 }, { "epoch": 121.95327102803738, "grad_norm": 2.421875, "learning_rate": 1.6844258447553028e-05, "loss": 0.31, "step": 19634 }, { "epoch": 121.9595015576324, "grad_norm": 3.8125, "learning_rate": 1.6841929515911964e-05, "loss": 0.4962, "step": 19635 }, { "epoch": 121.96573208722741, "grad_norm": 2.4375, "learning_rate": 1.6839600663503703e-05, "loss": 0.1731, "step": 19636 }, { "epoch": 121.97196261682242, "grad_norm": 2.828125, "learning_rate": 1.6837271890350876e-05, "loss": 0.2055, "step": 19637 }, { "epoch": 121.97819314641744, "grad_norm": 2.421875, "learning_rate": 1.6834943196476093e-05, "loss": 0.1963, "step": 19638 }, { "epoch": 121.98442367601245, "grad_norm": 1.3203125, "learning_rate": 1.6832614581901986e-05, "loss": 0.1105, "step": 19639 }, { "epoch": 121.99065420560747, "grad_norm": 2.09375, "learning_rate": 1.6830286046651145e-05, "loss": 0.1893, "step": 19640 }, { "epoch": 121.99688473520249, "grad_norm": 1.40625, "learning_rate": 1.6827957590746203e-05, "loss": 0.0909, "step": 19641 }, { "epoch": 122.0, "grad_norm": 0.765625, "learning_rate": 1.682562921420977e-05, "loss": 0.0421, "step": 19642 }, { "epoch": 122.00623052959502, "grad_norm": 3.1875, "learning_rate": 1.682330091706446e-05, "loss": 0.1793, "step": 19643 }, { "epoch": 122.01246105919003, "grad_norm": 1.4140625, "learning_rate": 1.682097269933289e-05, "loss": 0.0976, "step": 19644 }, { "epoch": 122.01869158878505, "grad_norm": 2.46875, "learning_rate": 1.681864456103766e-05, "loss": 0.132, "step": 19645 }, { "epoch": 122.02492211838006, "grad_norm": 2.53125, "learning_rate": 1.6816316502201403e-05, "loss": 0.1638, "step": 19646 }, { "epoch": 122.03115264797508, "grad_norm": 2.125, "learning_rate": 1.6813988522846702e-05, "loss": 0.1476, "step": 19647 }, { "epoch": 122.03738317757009, "grad_norm": 1.40625, "learning_rate": 1.681166062299619e-05, "loss": 0.1353, "step": 19648 }, { "epoch": 122.04361370716511, "grad_norm": 2.734375, "learning_rate": 1.6809332802672456e-05, "loss": 0.2211, "step": 19649 }, { "epoch": 122.04984423676012, "grad_norm": 1.5546875, "learning_rate": 1.680700506189813e-05, "loss": 0.1243, "step": 19650 }, { "epoch": 122.05607476635514, "grad_norm": 1.609375, "learning_rate": 1.6804677400695806e-05, "loss": 0.1088, "step": 19651 }, { "epoch": 122.06230529595015, "grad_norm": 2.375, "learning_rate": 1.6802349819088098e-05, "loss": 0.2125, "step": 19652 }, { "epoch": 122.06853582554517, "grad_norm": 2.34375, "learning_rate": 1.6800022317097597e-05, "loss": 0.1453, "step": 19653 }, { "epoch": 122.07476635514018, "grad_norm": 2.40625, "learning_rate": 1.6797694894746923e-05, "loss": 0.126, "step": 19654 }, { "epoch": 122.0809968847352, "grad_norm": 2.296875, "learning_rate": 1.679536755205868e-05, "loss": 0.1395, "step": 19655 }, { "epoch": 122.08722741433021, "grad_norm": 2.640625, "learning_rate": 1.6793040289055458e-05, "loss": 0.2033, "step": 19656 }, { "epoch": 122.09345794392523, "grad_norm": 1.34375, "learning_rate": 1.6790713105759874e-05, "loss": 0.1354, "step": 19657 }, { "epoch": 122.09968847352025, "grad_norm": 1.953125, "learning_rate": 1.678838600219453e-05, "loss": 0.1366, "step": 19658 }, { "epoch": 122.10591900311526, "grad_norm": 1.765625, "learning_rate": 1.6786058978382018e-05, "loss": 0.1247, "step": 19659 }, { "epoch": 122.11214953271028, "grad_norm": 2.421875, "learning_rate": 1.678373203434494e-05, "loss": 0.1865, "step": 19660 }, { "epoch": 122.11838006230529, "grad_norm": 1.234375, "learning_rate": 1.6781405170105902e-05, "loss": 0.1079, "step": 19661 }, { "epoch": 122.1246105919003, "grad_norm": 2.78125, "learning_rate": 1.67790783856875e-05, "loss": 0.2603, "step": 19662 }, { "epoch": 122.13084112149532, "grad_norm": 3.125, "learning_rate": 1.677675168111233e-05, "loss": 0.2805, "step": 19663 }, { "epoch": 122.13707165109034, "grad_norm": 3.0, "learning_rate": 1.6774425056402994e-05, "loss": 0.1834, "step": 19664 }, { "epoch": 122.14330218068535, "grad_norm": 1.5390625, "learning_rate": 1.6772098511582092e-05, "loss": 0.1258, "step": 19665 }, { "epoch": 122.14953271028037, "grad_norm": 2.03125, "learning_rate": 1.676977204667221e-05, "loss": 0.1492, "step": 19666 }, { "epoch": 122.15576323987538, "grad_norm": 1.71875, "learning_rate": 1.676744566169594e-05, "loss": 0.0997, "step": 19667 }, { "epoch": 122.1619937694704, "grad_norm": 2.1875, "learning_rate": 1.676511935667589e-05, "loss": 0.1658, "step": 19668 }, { "epoch": 122.16822429906541, "grad_norm": 2.09375, "learning_rate": 1.676279313163464e-05, "loss": 0.1669, "step": 19669 }, { "epoch": 122.17445482866043, "grad_norm": 1.0546875, "learning_rate": 1.6760466986594795e-05, "loss": 0.0911, "step": 19670 }, { "epoch": 122.18068535825545, "grad_norm": 2.078125, "learning_rate": 1.6758140921578946e-05, "loss": 0.1594, "step": 19671 }, { "epoch": 122.18691588785046, "grad_norm": 2.53125, "learning_rate": 1.6755814936609675e-05, "loss": 0.1761, "step": 19672 }, { "epoch": 122.19314641744548, "grad_norm": 2.125, "learning_rate": 1.6753489031709574e-05, "loss": 0.0939, "step": 19673 }, { "epoch": 122.19937694704049, "grad_norm": 3.796875, "learning_rate": 1.6751163206901243e-05, "loss": 0.2276, "step": 19674 }, { "epoch": 122.20560747663552, "grad_norm": 2.09375, "learning_rate": 1.674883746220726e-05, "loss": 0.1523, "step": 19675 }, { "epoch": 122.21183800623054, "grad_norm": 2.640625, "learning_rate": 1.6746511797650215e-05, "loss": 0.1878, "step": 19676 }, { "epoch": 122.21806853582555, "grad_norm": 2.875, "learning_rate": 1.67441862132527e-05, "loss": 0.1166, "step": 19677 }, { "epoch": 122.22429906542057, "grad_norm": 1.7421875, "learning_rate": 1.6741860709037304e-05, "loss": 0.0921, "step": 19678 }, { "epoch": 122.23052959501558, "grad_norm": 2.53125, "learning_rate": 1.6739535285026604e-05, "loss": 0.199, "step": 19679 }, { "epoch": 122.2367601246106, "grad_norm": 2.484375, "learning_rate": 1.673720994124318e-05, "loss": 0.203, "step": 19680 }, { "epoch": 122.24299065420561, "grad_norm": 2.34375, "learning_rate": 1.6734884677709632e-05, "loss": 0.185, "step": 19681 }, { "epoch": 122.24922118380063, "grad_norm": 1.4140625, "learning_rate": 1.673255949444853e-05, "loss": 0.0951, "step": 19682 }, { "epoch": 122.25545171339564, "grad_norm": 3.640625, "learning_rate": 1.673023439148247e-05, "loss": 0.2066, "step": 19683 }, { "epoch": 122.26168224299066, "grad_norm": 2.15625, "learning_rate": 1.6727909368834033e-05, "loss": 0.1193, "step": 19684 }, { "epoch": 122.26791277258567, "grad_norm": 2.09375, "learning_rate": 1.6725584426525786e-05, "loss": 0.096, "step": 19685 }, { "epoch": 122.27414330218069, "grad_norm": 2.609375, "learning_rate": 1.672325956458031e-05, "loss": 0.1073, "step": 19686 }, { "epoch": 122.2803738317757, "grad_norm": 1.078125, "learning_rate": 1.6720934783020205e-05, "loss": 0.1098, "step": 19687 }, { "epoch": 122.28660436137072, "grad_norm": 2.421875, "learning_rate": 1.6718610081868028e-05, "loss": 0.2414, "step": 19688 }, { "epoch": 122.29283489096574, "grad_norm": 1.875, "learning_rate": 1.6716285461146363e-05, "loss": 0.0981, "step": 19689 }, { "epoch": 122.29906542056075, "grad_norm": 1.3515625, "learning_rate": 1.6713960920877798e-05, "loss": 0.123, "step": 19690 }, { "epoch": 122.30529595015577, "grad_norm": 1.2734375, "learning_rate": 1.67116364610849e-05, "loss": 0.1021, "step": 19691 }, { "epoch": 122.31152647975078, "grad_norm": 2.03125, "learning_rate": 1.670931208179024e-05, "loss": 0.1649, "step": 19692 }, { "epoch": 122.3177570093458, "grad_norm": 1.2578125, "learning_rate": 1.6706987783016398e-05, "loss": 0.1252, "step": 19693 }, { "epoch": 122.32398753894081, "grad_norm": 1.4609375, "learning_rate": 1.6704663564785948e-05, "loss": 0.131, "step": 19694 }, { "epoch": 122.33021806853583, "grad_norm": 1.96875, "learning_rate": 1.6702339427121467e-05, "loss": 0.1487, "step": 19695 }, { "epoch": 122.33644859813084, "grad_norm": 1.546875, "learning_rate": 1.670001537004552e-05, "loss": 0.1032, "step": 19696 }, { "epoch": 122.34267912772586, "grad_norm": 1.984375, "learning_rate": 1.6697691393580696e-05, "loss": 0.2129, "step": 19697 }, { "epoch": 122.34890965732087, "grad_norm": 1.890625, "learning_rate": 1.6695367497749542e-05, "loss": 0.1674, "step": 19698 }, { "epoch": 122.35514018691589, "grad_norm": 2.015625, "learning_rate": 1.6693043682574635e-05, "loss": 0.1165, "step": 19699 }, { "epoch": 122.3613707165109, "grad_norm": 2.4375, "learning_rate": 1.6690719948078555e-05, "loss": 0.1894, "step": 19700 }, { "epoch": 122.36760124610592, "grad_norm": 3.28125, "learning_rate": 1.6688396294283864e-05, "loss": 0.28, "step": 19701 }, { "epoch": 122.37383177570094, "grad_norm": 1.6015625, "learning_rate": 1.668607272121312e-05, "loss": 0.1199, "step": 19702 }, { "epoch": 122.38006230529595, "grad_norm": 3.0, "learning_rate": 1.6683749228888917e-05, "loss": 0.3911, "step": 19703 }, { "epoch": 122.38629283489097, "grad_norm": 2.390625, "learning_rate": 1.6681425817333795e-05, "loss": 0.226, "step": 19704 }, { "epoch": 122.39252336448598, "grad_norm": 3.546875, "learning_rate": 1.6679102486570328e-05, "loss": 0.2936, "step": 19705 }, { "epoch": 122.398753894081, "grad_norm": 2.671875, "learning_rate": 1.6676779236621075e-05, "loss": 0.2169, "step": 19706 }, { "epoch": 122.40498442367601, "grad_norm": 0.6796875, "learning_rate": 1.6674456067508615e-05, "loss": 0.0853, "step": 19707 }, { "epoch": 122.41121495327103, "grad_norm": 2.125, "learning_rate": 1.66721329792555e-05, "loss": 0.2316, "step": 19708 }, { "epoch": 122.41744548286604, "grad_norm": 1.1328125, "learning_rate": 1.6669809971884298e-05, "loss": 0.1159, "step": 19709 }, { "epoch": 122.42367601246106, "grad_norm": 2.359375, "learning_rate": 1.6667487045417558e-05, "loss": 0.2469, "step": 19710 }, { "epoch": 122.42990654205607, "grad_norm": 2.140625, "learning_rate": 1.6665164199877853e-05, "loss": 0.159, "step": 19711 }, { "epoch": 122.43613707165109, "grad_norm": 2.109375, "learning_rate": 1.666284143528774e-05, "loss": 0.1686, "step": 19712 }, { "epoch": 122.4423676012461, "grad_norm": 1.6328125, "learning_rate": 1.666051875166977e-05, "loss": 0.172, "step": 19713 }, { "epoch": 122.44859813084112, "grad_norm": 1.75, "learning_rate": 1.6658196149046518e-05, "loss": 0.136, "step": 19714 }, { "epoch": 122.45482866043614, "grad_norm": 2.140625, "learning_rate": 1.6655873627440527e-05, "loss": 0.1858, "step": 19715 }, { "epoch": 122.46105919003115, "grad_norm": 1.9140625, "learning_rate": 1.6653551186874372e-05, "loss": 0.1307, "step": 19716 }, { "epoch": 122.46728971962617, "grad_norm": 1.5390625, "learning_rate": 1.6651228827370582e-05, "loss": 0.1202, "step": 19717 }, { "epoch": 122.47352024922118, "grad_norm": 3.453125, "learning_rate": 1.6648906548951732e-05, "loss": 0.2468, "step": 19718 }, { "epoch": 122.4797507788162, "grad_norm": 1.2109375, "learning_rate": 1.6646584351640365e-05, "loss": 0.1293, "step": 19719 }, { "epoch": 122.48598130841121, "grad_norm": 3.359375, "learning_rate": 1.6644262235459047e-05, "loss": 0.2294, "step": 19720 }, { "epoch": 122.49221183800623, "grad_norm": 1.40625, "learning_rate": 1.6641940200430323e-05, "loss": 0.0993, "step": 19721 }, { "epoch": 122.49844236760124, "grad_norm": 2.015625, "learning_rate": 1.6639618246576754e-05, "loss": 0.2098, "step": 19722 }, { "epoch": 122.50467289719626, "grad_norm": 0.7421875, "learning_rate": 1.6637296373920875e-05, "loss": 0.0853, "step": 19723 }, { "epoch": 122.51090342679127, "grad_norm": 3.578125, "learning_rate": 1.6634974582485247e-05, "loss": 0.2345, "step": 19724 }, { "epoch": 122.51713395638629, "grad_norm": 2.21875, "learning_rate": 1.6632652872292418e-05, "loss": 0.2029, "step": 19725 }, { "epoch": 122.5233644859813, "grad_norm": 0.7578125, "learning_rate": 1.6630331243364933e-05, "loss": 0.0787, "step": 19726 }, { "epoch": 122.52959501557632, "grad_norm": 2.546875, "learning_rate": 1.6628009695725346e-05, "loss": 0.2174, "step": 19727 }, { "epoch": 122.53582554517133, "grad_norm": 1.359375, "learning_rate": 1.662568822939621e-05, "loss": 0.1052, "step": 19728 }, { "epoch": 122.54205607476635, "grad_norm": 2.84375, "learning_rate": 1.6623366844400058e-05, "loss": 0.2196, "step": 19729 }, { "epoch": 122.54828660436137, "grad_norm": 2.109375, "learning_rate": 1.662104554075944e-05, "loss": 0.2021, "step": 19730 }, { "epoch": 122.55451713395638, "grad_norm": 1.609375, "learning_rate": 1.6618724318496904e-05, "loss": 0.1352, "step": 19731 }, { "epoch": 122.5607476635514, "grad_norm": 2.328125, "learning_rate": 1.661640317763499e-05, "loss": 0.1408, "step": 19732 }, { "epoch": 122.56697819314641, "grad_norm": 2.625, "learning_rate": 1.6614082118196246e-05, "loss": 0.2796, "step": 19733 }, { "epoch": 122.57320872274143, "grad_norm": 3.0, "learning_rate": 1.6611761140203217e-05, "loss": 0.2493, "step": 19734 }, { "epoch": 122.57943925233644, "grad_norm": 2.234375, "learning_rate": 1.6609440243678443e-05, "loss": 0.2332, "step": 19735 }, { "epoch": 122.58566978193146, "grad_norm": 2.28125, "learning_rate": 1.6607119428644453e-05, "loss": 0.1972, "step": 19736 }, { "epoch": 122.59190031152647, "grad_norm": 1.796875, "learning_rate": 1.66047986951238e-05, "loss": 0.1203, "step": 19737 }, { "epoch": 122.59813084112149, "grad_norm": 2.453125, "learning_rate": 1.6602478043139024e-05, "loss": 0.1703, "step": 19738 }, { "epoch": 122.60436137071652, "grad_norm": 2.203125, "learning_rate": 1.6600157472712652e-05, "loss": 0.1488, "step": 19739 }, { "epoch": 122.61059190031153, "grad_norm": 1.515625, "learning_rate": 1.6597836983867238e-05, "loss": 0.1572, "step": 19740 }, { "epoch": 122.61682242990655, "grad_norm": 2.53125, "learning_rate": 1.6595516576625313e-05, "loss": 0.2247, "step": 19741 }, { "epoch": 122.62305295950156, "grad_norm": 1.765625, "learning_rate": 1.6593196251009406e-05, "loss": 0.1264, "step": 19742 }, { "epoch": 122.62928348909658, "grad_norm": 3.03125, "learning_rate": 1.6590876007042054e-05, "loss": 0.1692, "step": 19743 }, { "epoch": 122.6355140186916, "grad_norm": 3.765625, "learning_rate": 1.65885558447458e-05, "loss": 0.2845, "step": 19744 }, { "epoch": 122.64174454828661, "grad_norm": 2.0, "learning_rate": 1.6586235764143167e-05, "loss": 0.2083, "step": 19745 }, { "epoch": 122.64797507788163, "grad_norm": 3.03125, "learning_rate": 1.6583915765256697e-05, "loss": 0.1479, "step": 19746 }, { "epoch": 122.65420560747664, "grad_norm": 1.734375, "learning_rate": 1.6581595848108923e-05, "loss": 0.107, "step": 19747 }, { "epoch": 122.66043613707166, "grad_norm": 2.1875, "learning_rate": 1.6579276012722377e-05, "loss": 0.1414, "step": 19748 }, { "epoch": 122.66666666666667, "grad_norm": 2.265625, "learning_rate": 1.6576956259119576e-05, "loss": 0.1908, "step": 19749 }, { "epoch": 122.67289719626169, "grad_norm": 3.21875, "learning_rate": 1.657463658732306e-05, "loss": 0.2748, "step": 19750 }, { "epoch": 122.6791277258567, "grad_norm": 2.46875, "learning_rate": 1.6572316997355364e-05, "loss": 0.1555, "step": 19751 }, { "epoch": 122.68535825545172, "grad_norm": 1.8203125, "learning_rate": 1.6569997489239004e-05, "loss": 0.1826, "step": 19752 }, { "epoch": 122.69158878504673, "grad_norm": 3.796875, "learning_rate": 1.6567678062996517e-05, "loss": 0.1788, "step": 19753 }, { "epoch": 122.69781931464175, "grad_norm": 3.859375, "learning_rate": 1.656535871865043e-05, "loss": 0.3272, "step": 19754 }, { "epoch": 122.70404984423676, "grad_norm": 3.015625, "learning_rate": 1.6563039456223262e-05, "loss": 0.1523, "step": 19755 }, { "epoch": 122.71028037383178, "grad_norm": 3.75, "learning_rate": 1.6560720275737536e-05, "loss": 0.1524, "step": 19756 }, { "epoch": 122.7165109034268, "grad_norm": 3.421875, "learning_rate": 1.6558401177215786e-05, "loss": 0.2005, "step": 19757 }, { "epoch": 122.72274143302181, "grad_norm": 1.65625, "learning_rate": 1.655608216068053e-05, "loss": 0.1436, "step": 19758 }, { "epoch": 122.72897196261682, "grad_norm": 3.03125, "learning_rate": 1.655376322615429e-05, "loss": 0.3319, "step": 19759 }, { "epoch": 122.73520249221184, "grad_norm": 2.4375, "learning_rate": 1.65514443736596e-05, "loss": 0.1803, "step": 19760 }, { "epoch": 122.74143302180686, "grad_norm": 3.15625, "learning_rate": 1.654912560321896e-05, "loss": 0.2203, "step": 19761 }, { "epoch": 122.74766355140187, "grad_norm": 2.953125, "learning_rate": 1.6546806914854908e-05, "loss": 0.1786, "step": 19762 }, { "epoch": 122.75389408099689, "grad_norm": 2.734375, "learning_rate": 1.654448830858995e-05, "loss": 0.2032, "step": 19763 }, { "epoch": 122.7601246105919, "grad_norm": 2.5, "learning_rate": 1.654216978444662e-05, "loss": 0.1931, "step": 19764 }, { "epoch": 122.76635514018692, "grad_norm": 1.9765625, "learning_rate": 1.6539851342447414e-05, "loss": 0.236, "step": 19765 }, { "epoch": 122.77258566978193, "grad_norm": 1.4453125, "learning_rate": 1.6537532982614878e-05, "loss": 0.0981, "step": 19766 }, { "epoch": 122.77881619937695, "grad_norm": 1.109375, "learning_rate": 1.6535214704971514e-05, "loss": 0.0938, "step": 19767 }, { "epoch": 122.78504672897196, "grad_norm": 2.625, "learning_rate": 1.653289650953983e-05, "loss": 0.2033, "step": 19768 }, { "epoch": 122.79127725856698, "grad_norm": 1.3203125, "learning_rate": 1.6530578396342346e-05, "loss": 0.1445, "step": 19769 }, { "epoch": 122.797507788162, "grad_norm": 2.125, "learning_rate": 1.652826036540158e-05, "loss": 0.1705, "step": 19770 }, { "epoch": 122.80373831775701, "grad_norm": 1.4765625, "learning_rate": 1.6525942416740043e-05, "loss": 0.1183, "step": 19771 }, { "epoch": 122.80996884735202, "grad_norm": 1.8828125, "learning_rate": 1.652362455038024e-05, "loss": 0.1295, "step": 19772 }, { "epoch": 122.81619937694704, "grad_norm": 2.296875, "learning_rate": 1.6521306766344708e-05, "loss": 0.1677, "step": 19773 }, { "epoch": 122.82242990654206, "grad_norm": 1.640625, "learning_rate": 1.6518989064655928e-05, "loss": 0.1468, "step": 19774 }, { "epoch": 122.82866043613707, "grad_norm": 1.4140625, "learning_rate": 1.651667144533642e-05, "loss": 0.0942, "step": 19775 }, { "epoch": 122.83489096573209, "grad_norm": 2.328125, "learning_rate": 1.651435390840869e-05, "loss": 0.2219, "step": 19776 }, { "epoch": 122.8411214953271, "grad_norm": 2.28125, "learning_rate": 1.6512036453895262e-05, "loss": 0.2125, "step": 19777 }, { "epoch": 122.84735202492212, "grad_norm": 1.296875, "learning_rate": 1.6509719081818627e-05, "loss": 0.1045, "step": 19778 }, { "epoch": 122.85358255451713, "grad_norm": 2.171875, "learning_rate": 1.6507401792201306e-05, "loss": 0.2105, "step": 19779 }, { "epoch": 122.85981308411215, "grad_norm": 3.53125, "learning_rate": 1.6505084585065787e-05, "loss": 0.259, "step": 19780 }, { "epoch": 122.86604361370716, "grad_norm": 2.25, "learning_rate": 1.6502767460434588e-05, "loss": 0.191, "step": 19781 }, { "epoch": 122.87227414330218, "grad_norm": 1.515625, "learning_rate": 1.6500450418330205e-05, "loss": 0.1129, "step": 19782 }, { "epoch": 122.8785046728972, "grad_norm": 1.984375, "learning_rate": 1.649813345877515e-05, "loss": 0.1164, "step": 19783 }, { "epoch": 122.88473520249221, "grad_norm": 2.8125, "learning_rate": 1.6495816581791925e-05, "loss": 0.191, "step": 19784 }, { "epoch": 122.89096573208722, "grad_norm": 1.0859375, "learning_rate": 1.649349978740302e-05, "loss": 0.1027, "step": 19785 }, { "epoch": 122.89719626168224, "grad_norm": 2.203125, "learning_rate": 1.6491183075630958e-05, "loss": 0.2455, "step": 19786 }, { "epoch": 122.90342679127725, "grad_norm": 1.9453125, "learning_rate": 1.6488866446498223e-05, "loss": 0.1263, "step": 19787 }, { "epoch": 122.90965732087227, "grad_norm": 2.3125, "learning_rate": 1.648654990002732e-05, "loss": 0.1052, "step": 19788 }, { "epoch": 122.91588785046729, "grad_norm": 3.28125, "learning_rate": 1.6484233436240735e-05, "loss": 0.2185, "step": 19789 }, { "epoch": 122.9221183800623, "grad_norm": 1.6796875, "learning_rate": 1.648191705516099e-05, "loss": 0.1248, "step": 19790 }, { "epoch": 122.92834890965732, "grad_norm": 1.8125, "learning_rate": 1.647960075681056e-05, "loss": 0.1506, "step": 19791 }, { "epoch": 122.93457943925233, "grad_norm": 1.9375, "learning_rate": 1.647728454121196e-05, "loss": 0.111, "step": 19792 }, { "epoch": 122.94080996884735, "grad_norm": 1.046875, "learning_rate": 1.647496840838767e-05, "loss": 0.1138, "step": 19793 }, { "epoch": 122.94704049844236, "grad_norm": 1.84375, "learning_rate": 1.6472652358360193e-05, "loss": 0.1642, "step": 19794 }, { "epoch": 122.95327102803738, "grad_norm": 1.1875, "learning_rate": 1.6470336391152018e-05, "loss": 0.0916, "step": 19795 }, { "epoch": 122.9595015576324, "grad_norm": 1.59375, "learning_rate": 1.6468020506785644e-05, "loss": 0.159, "step": 19796 }, { "epoch": 122.96573208722741, "grad_norm": 2.6875, "learning_rate": 1.6465704705283557e-05, "loss": 0.2544, "step": 19797 }, { "epoch": 122.97196261682242, "grad_norm": 1.515625, "learning_rate": 1.646338898666826e-05, "loss": 0.1021, "step": 19798 }, { "epoch": 122.97819314641744, "grad_norm": 1.140625, "learning_rate": 1.6461073350962223e-05, "loss": 0.1077, "step": 19799 }, { "epoch": 122.98442367601245, "grad_norm": 2.90625, "learning_rate": 1.645875779818796e-05, "loss": 0.2064, "step": 19800 }, { "epoch": 122.99065420560747, "grad_norm": 2.46875, "learning_rate": 1.6456442328367943e-05, "loss": 0.2816, "step": 19801 }, { "epoch": 122.99688473520249, "grad_norm": 2.296875, "learning_rate": 1.6454126941524663e-05, "loss": 0.2226, "step": 19802 }, { "epoch": 123.0, "grad_norm": 1.1015625, "learning_rate": 1.6451811637680617e-05, "loss": 0.0597, "step": 19803 }, { "epoch": 123.00623052959502, "grad_norm": 2.203125, "learning_rate": 1.6449496416858284e-05, "loss": 0.1362, "step": 19804 }, { "epoch": 123.01246105919003, "grad_norm": 2.53125, "learning_rate": 1.6447181279080153e-05, "loss": 0.3459, "step": 19805 }, { "epoch": 123.01869158878505, "grad_norm": 1.296875, "learning_rate": 1.64448662243687e-05, "loss": 0.1059, "step": 19806 }, { "epoch": 123.02492211838006, "grad_norm": 2.609375, "learning_rate": 1.6442551252746422e-05, "loss": 0.2176, "step": 19807 }, { "epoch": 123.03115264797508, "grad_norm": 1.0390625, "learning_rate": 1.6440236364235796e-05, "loss": 0.0859, "step": 19808 }, { "epoch": 123.03738317757009, "grad_norm": 1.1796875, "learning_rate": 1.64379215588593e-05, "loss": 0.1092, "step": 19809 }, { "epoch": 123.04361370716511, "grad_norm": 3.125, "learning_rate": 1.6435606836639425e-05, "loss": 0.2413, "step": 19810 }, { "epoch": 123.04984423676012, "grad_norm": 1.5703125, "learning_rate": 1.6433292197598653e-05, "loss": 0.1048, "step": 19811 }, { "epoch": 123.05607476635514, "grad_norm": 2.015625, "learning_rate": 1.643097764175946e-05, "loss": 0.1509, "step": 19812 }, { "epoch": 123.06230529595015, "grad_norm": 1.8125, "learning_rate": 1.6428663169144314e-05, "loss": 0.1242, "step": 19813 }, { "epoch": 123.06853582554517, "grad_norm": 1.5234375, "learning_rate": 1.6426348779775713e-05, "loss": 0.1352, "step": 19814 }, { "epoch": 123.07476635514018, "grad_norm": 0.91796875, "learning_rate": 1.642403447367612e-05, "loss": 0.1042, "step": 19815 }, { "epoch": 123.0809968847352, "grad_norm": 1.6484375, "learning_rate": 1.6421720250868025e-05, "loss": 0.117, "step": 19816 }, { "epoch": 123.08722741433021, "grad_norm": 3.140625, "learning_rate": 1.6419406111373896e-05, "loss": 0.2531, "step": 19817 }, { "epoch": 123.09345794392523, "grad_norm": 1.7421875, "learning_rate": 1.6417092055216215e-05, "loss": 0.1132, "step": 19818 }, { "epoch": 123.09968847352025, "grad_norm": 3.234375, "learning_rate": 1.6414778082417445e-05, "loss": 0.2369, "step": 19819 }, { "epoch": 123.10591900311526, "grad_norm": 1.75, "learning_rate": 1.641246419300007e-05, "loss": 0.0891, "step": 19820 }, { "epoch": 123.11214953271028, "grad_norm": 2.671875, "learning_rate": 1.6410150386986556e-05, "loss": 0.2357, "step": 19821 }, { "epoch": 123.11838006230529, "grad_norm": 2.78125, "learning_rate": 1.640783666439938e-05, "loss": 0.2511, "step": 19822 }, { "epoch": 123.1246105919003, "grad_norm": 1.203125, "learning_rate": 1.640552302526101e-05, "loss": 0.0998, "step": 19823 }, { "epoch": 123.13084112149532, "grad_norm": 3.171875, "learning_rate": 1.6403209469593928e-05, "loss": 0.2572, "step": 19824 }, { "epoch": 123.13707165109034, "grad_norm": 1.1171875, "learning_rate": 1.6400895997420585e-05, "loss": 0.113, "step": 19825 }, { "epoch": 123.14330218068535, "grad_norm": 2.5625, "learning_rate": 1.6398582608763457e-05, "loss": 0.1673, "step": 19826 }, { "epoch": 123.14953271028037, "grad_norm": 2.390625, "learning_rate": 1.6396269303645017e-05, "loss": 0.1633, "step": 19827 }, { "epoch": 123.15576323987538, "grad_norm": 1.53125, "learning_rate": 1.6393956082087724e-05, "loss": 0.1348, "step": 19828 }, { "epoch": 123.1619937694704, "grad_norm": 2.28125, "learning_rate": 1.639164294411406e-05, "loss": 0.1768, "step": 19829 }, { "epoch": 123.16822429906541, "grad_norm": 2.09375, "learning_rate": 1.638932988974648e-05, "loss": 0.161, "step": 19830 }, { "epoch": 123.17445482866043, "grad_norm": 2.71875, "learning_rate": 1.638701691900745e-05, "loss": 0.2432, "step": 19831 }, { "epoch": 123.18068535825545, "grad_norm": 2.171875, "learning_rate": 1.6384704031919424e-05, "loss": 0.2546, "step": 19832 }, { "epoch": 123.18691588785046, "grad_norm": 1.40625, "learning_rate": 1.6382391228504877e-05, "loss": 0.1084, "step": 19833 }, { "epoch": 123.19314641744548, "grad_norm": 1.953125, "learning_rate": 1.6380078508786273e-05, "loss": 0.1541, "step": 19834 }, { "epoch": 123.19937694704049, "grad_norm": 3.671875, "learning_rate": 1.6377765872786068e-05, "loss": 0.4282, "step": 19835 }, { "epoch": 123.20560747663552, "grad_norm": 2.421875, "learning_rate": 1.6375453320526724e-05, "loss": 0.2388, "step": 19836 }, { "epoch": 123.21183800623054, "grad_norm": 2.15625, "learning_rate": 1.6373140852030706e-05, "loss": 0.0981, "step": 19837 }, { "epoch": 123.21806853582555, "grad_norm": 2.25, "learning_rate": 1.6370828467320467e-05, "loss": 0.1286, "step": 19838 }, { "epoch": 123.22429906542057, "grad_norm": 2.9375, "learning_rate": 1.636851616641846e-05, "loss": 0.2812, "step": 19839 }, { "epoch": 123.23052959501558, "grad_norm": 3.046875, "learning_rate": 1.6366203949347152e-05, "loss": 0.2952, "step": 19840 }, { "epoch": 123.2367601246106, "grad_norm": 1.7265625, "learning_rate": 1.6363891816128994e-05, "loss": 0.147, "step": 19841 }, { "epoch": 123.24299065420561, "grad_norm": 2.5625, "learning_rate": 1.636157976678645e-05, "loss": 0.2393, "step": 19842 }, { "epoch": 123.24922118380063, "grad_norm": 2.171875, "learning_rate": 1.6359267801341975e-05, "loss": 0.161, "step": 19843 }, { "epoch": 123.25545171339564, "grad_norm": 1.265625, "learning_rate": 1.6356955919818017e-05, "loss": 0.1114, "step": 19844 }, { "epoch": 123.26168224299066, "grad_norm": 1.1015625, "learning_rate": 1.6354644122237023e-05, "loss": 0.0834, "step": 19845 }, { "epoch": 123.26791277258567, "grad_norm": 2.265625, "learning_rate": 1.6352332408621457e-05, "loss": 0.1119, "step": 19846 }, { "epoch": 123.27414330218069, "grad_norm": 1.9375, "learning_rate": 1.6350020778993768e-05, "loss": 0.0963, "step": 19847 }, { "epoch": 123.2803738317757, "grad_norm": 2.734375, "learning_rate": 1.6347709233376402e-05, "loss": 0.2052, "step": 19848 }, { "epoch": 123.28660436137072, "grad_norm": 2.171875, "learning_rate": 1.6345397771791827e-05, "loss": 0.1521, "step": 19849 }, { "epoch": 123.29283489096574, "grad_norm": 2.46875, "learning_rate": 1.6343086394262467e-05, "loss": 0.187, "step": 19850 }, { "epoch": 123.29906542056075, "grad_norm": 2.28125, "learning_rate": 1.6340775100810784e-05, "loss": 0.2379, "step": 19851 }, { "epoch": 123.30529595015577, "grad_norm": 3.09375, "learning_rate": 1.633846389145922e-05, "loss": 0.3143, "step": 19852 }, { "epoch": 123.31152647975078, "grad_norm": 3.078125, "learning_rate": 1.6336152766230232e-05, "loss": 0.2226, "step": 19853 }, { "epoch": 123.3177570093458, "grad_norm": 2.59375, "learning_rate": 1.6333841725146266e-05, "loss": 0.1953, "step": 19854 }, { "epoch": 123.32398753894081, "grad_norm": 2.515625, "learning_rate": 1.633153076822975e-05, "loss": 0.1777, "step": 19855 }, { "epoch": 123.33021806853583, "grad_norm": 2.15625, "learning_rate": 1.6329219895503152e-05, "loss": 0.1848, "step": 19856 }, { "epoch": 123.33644859813084, "grad_norm": 1.515625, "learning_rate": 1.6326909106988893e-05, "loss": 0.1084, "step": 19857 }, { "epoch": 123.34267912772586, "grad_norm": 1.890625, "learning_rate": 1.6324598402709432e-05, "loss": 0.1296, "step": 19858 }, { "epoch": 123.34890965732087, "grad_norm": 2.359375, "learning_rate": 1.6322287782687197e-05, "loss": 0.1608, "step": 19859 }, { "epoch": 123.35514018691589, "grad_norm": 2.15625, "learning_rate": 1.6319977246944646e-05, "loss": 0.2252, "step": 19860 }, { "epoch": 123.3613707165109, "grad_norm": 2.078125, "learning_rate": 1.6317666795504204e-05, "loss": 0.1604, "step": 19861 }, { "epoch": 123.36760124610592, "grad_norm": 1.03125, "learning_rate": 1.6315356428388335e-05, "loss": 0.0928, "step": 19862 }, { "epoch": 123.37383177570094, "grad_norm": 1.5859375, "learning_rate": 1.631304614561944e-05, "loss": 0.1397, "step": 19863 }, { "epoch": 123.38006230529595, "grad_norm": 2.125, "learning_rate": 1.631073594721999e-05, "loss": 0.195, "step": 19864 }, { "epoch": 123.38629283489097, "grad_norm": 1.671875, "learning_rate": 1.63084258332124e-05, "loss": 0.107, "step": 19865 }, { "epoch": 123.39252336448598, "grad_norm": 1.125, "learning_rate": 1.630611580361912e-05, "loss": 0.1105, "step": 19866 }, { "epoch": 123.398753894081, "grad_norm": 2.171875, "learning_rate": 1.6303805858462585e-05, "loss": 0.1652, "step": 19867 }, { "epoch": 123.40498442367601, "grad_norm": 2.203125, "learning_rate": 1.630149599776522e-05, "loss": 0.178, "step": 19868 }, { "epoch": 123.41121495327103, "grad_norm": 2.09375, "learning_rate": 1.6299186221549473e-05, "loss": 0.1743, "step": 19869 }, { "epoch": 123.41744548286604, "grad_norm": 1.2890625, "learning_rate": 1.629687652983776e-05, "loss": 0.1089, "step": 19870 }, { "epoch": 123.42367601246106, "grad_norm": 2.375, "learning_rate": 1.629456692265253e-05, "loss": 0.1527, "step": 19871 }, { "epoch": 123.42990654205607, "grad_norm": 1.6953125, "learning_rate": 1.6292257400016197e-05, "loss": 0.1421, "step": 19872 }, { "epoch": 123.43613707165109, "grad_norm": 1.203125, "learning_rate": 1.6289947961951205e-05, "loss": 0.1169, "step": 19873 }, { "epoch": 123.4423676012461, "grad_norm": 2.9375, "learning_rate": 1.6287638608479975e-05, "loss": 0.3268, "step": 19874 }, { "epoch": 123.44859813084112, "grad_norm": 1.5859375, "learning_rate": 1.6285329339624956e-05, "loss": 0.0953, "step": 19875 }, { "epoch": 123.45482866043614, "grad_norm": 1.8046875, "learning_rate": 1.6283020155408544e-05, "loss": 0.1808, "step": 19876 }, { "epoch": 123.46105919003115, "grad_norm": 1.359375, "learning_rate": 1.6280711055853188e-05, "loss": 0.0944, "step": 19877 }, { "epoch": 123.46728971962617, "grad_norm": 1.6640625, "learning_rate": 1.6278402040981306e-05, "loss": 0.155, "step": 19878 }, { "epoch": 123.47352024922118, "grad_norm": 2.5, "learning_rate": 1.6276093110815334e-05, "loss": 0.1264, "step": 19879 }, { "epoch": 123.4797507788162, "grad_norm": 2.890625, "learning_rate": 1.6273784265377685e-05, "loss": 0.3025, "step": 19880 }, { "epoch": 123.48598130841121, "grad_norm": 2.078125, "learning_rate": 1.6271475504690792e-05, "loss": 0.2492, "step": 19881 }, { "epoch": 123.49221183800623, "grad_norm": 0.74609375, "learning_rate": 1.6269166828777067e-05, "loss": 0.0774, "step": 19882 }, { "epoch": 123.49844236760124, "grad_norm": 3.078125, "learning_rate": 1.6266858237658943e-05, "loss": 0.3929, "step": 19883 }, { "epoch": 123.50467289719626, "grad_norm": 1.890625, "learning_rate": 1.6264549731358835e-05, "loss": 0.2459, "step": 19884 }, { "epoch": 123.51090342679127, "grad_norm": 2.375, "learning_rate": 1.626224130989916e-05, "loss": 0.134, "step": 19885 }, { "epoch": 123.51713395638629, "grad_norm": 2.3125, "learning_rate": 1.6259932973302355e-05, "loss": 0.143, "step": 19886 }, { "epoch": 123.5233644859813, "grad_norm": 2.765625, "learning_rate": 1.625762472159082e-05, "loss": 0.1909, "step": 19887 }, { "epoch": 123.52959501557632, "grad_norm": 2.578125, "learning_rate": 1.6255316554786998e-05, "loss": 0.2062, "step": 19888 }, { "epoch": 123.53582554517133, "grad_norm": 1.1875, "learning_rate": 1.6253008472913274e-05, "loss": 0.0983, "step": 19889 }, { "epoch": 123.54205607476635, "grad_norm": 2.96875, "learning_rate": 1.6250700475992083e-05, "loss": 0.1121, "step": 19890 }, { "epoch": 123.54828660436137, "grad_norm": 1.125, "learning_rate": 1.6248392564045836e-05, "loss": 0.119, "step": 19891 }, { "epoch": 123.55451713395638, "grad_norm": 2.125, "learning_rate": 1.6246084737096954e-05, "loss": 0.0982, "step": 19892 }, { "epoch": 123.5607476635514, "grad_norm": 3.03125, "learning_rate": 1.624377699516785e-05, "loss": 0.2535, "step": 19893 }, { "epoch": 123.56697819314641, "grad_norm": 1.7578125, "learning_rate": 1.624146933828093e-05, "loss": 0.1745, "step": 19894 }, { "epoch": 123.57320872274143, "grad_norm": 2.5625, "learning_rate": 1.6239161766458605e-05, "loss": 0.1648, "step": 19895 }, { "epoch": 123.57943925233644, "grad_norm": 1.171875, "learning_rate": 1.62368542797233e-05, "loss": 0.0992, "step": 19896 }, { "epoch": 123.58566978193146, "grad_norm": 1.9375, "learning_rate": 1.6234546878097417e-05, "loss": 0.1481, "step": 19897 }, { "epoch": 123.59190031152647, "grad_norm": 1.3359375, "learning_rate": 1.623223956160336e-05, "loss": 0.1189, "step": 19898 }, { "epoch": 123.59813084112149, "grad_norm": 1.5625, "learning_rate": 1.6229932330263553e-05, "loss": 0.1389, "step": 19899 }, { "epoch": 123.60436137071652, "grad_norm": 1.609375, "learning_rate": 1.6227625184100397e-05, "loss": 0.1098, "step": 19900 }, { "epoch": 123.61059190031153, "grad_norm": 2.515625, "learning_rate": 1.6225318123136294e-05, "loss": 0.1971, "step": 19901 }, { "epoch": 123.61682242990655, "grad_norm": 2.6875, "learning_rate": 1.6223011147393654e-05, "loss": 0.2592, "step": 19902 }, { "epoch": 123.62305295950156, "grad_norm": 2.515625, "learning_rate": 1.6220704256894888e-05, "loss": 0.1231, "step": 19903 }, { "epoch": 123.62928348909658, "grad_norm": 1.390625, "learning_rate": 1.6218397451662398e-05, "loss": 0.1065, "step": 19904 }, { "epoch": 123.6355140186916, "grad_norm": 1.9765625, "learning_rate": 1.621609073171858e-05, "loss": 0.1615, "step": 19905 }, { "epoch": 123.64174454828661, "grad_norm": 2.484375, "learning_rate": 1.6213784097085853e-05, "loss": 0.174, "step": 19906 }, { "epoch": 123.64797507788163, "grad_norm": 0.92578125, "learning_rate": 1.6211477547786615e-05, "loss": 0.1064, "step": 19907 }, { "epoch": 123.65420560747664, "grad_norm": 1.4765625, "learning_rate": 1.6209171083843256e-05, "loss": 0.1307, "step": 19908 }, { "epoch": 123.66043613707166, "grad_norm": 2.3125, "learning_rate": 1.6206864705278183e-05, "loss": 0.0988, "step": 19909 }, { "epoch": 123.66666666666667, "grad_norm": 1.1171875, "learning_rate": 1.6204558412113803e-05, "loss": 0.1334, "step": 19910 }, { "epoch": 123.67289719626169, "grad_norm": 1.2578125, "learning_rate": 1.62022522043725e-05, "loss": 0.1195, "step": 19911 }, { "epoch": 123.6791277258567, "grad_norm": 2.234375, "learning_rate": 1.619994608207669e-05, "loss": 0.2186, "step": 19912 }, { "epoch": 123.68535825545172, "grad_norm": 2.8125, "learning_rate": 1.619764004524877e-05, "loss": 0.242, "step": 19913 }, { "epoch": 123.69158878504673, "grad_norm": 1.734375, "learning_rate": 1.6195334093911126e-05, "loss": 0.1001, "step": 19914 }, { "epoch": 123.69781931464175, "grad_norm": 1.484375, "learning_rate": 1.6193028228086144e-05, "loss": 0.1494, "step": 19915 }, { "epoch": 123.70404984423676, "grad_norm": 1.2734375, "learning_rate": 1.619072244779624e-05, "loss": 0.1089, "step": 19916 }, { "epoch": 123.71028037383178, "grad_norm": 1.3359375, "learning_rate": 1.6188416753063806e-05, "loss": 0.1141, "step": 19917 }, { "epoch": 123.7165109034268, "grad_norm": 1.3671875, "learning_rate": 1.618611114391122e-05, "loss": 0.115, "step": 19918 }, { "epoch": 123.72274143302181, "grad_norm": 1.296875, "learning_rate": 1.61838056203609e-05, "loss": 0.1309, "step": 19919 }, { "epoch": 123.72897196261682, "grad_norm": 2.109375, "learning_rate": 1.618150018243521e-05, "loss": 0.1608, "step": 19920 }, { "epoch": 123.73520249221184, "grad_norm": 1.28125, "learning_rate": 1.6179194830156557e-05, "loss": 0.1359, "step": 19921 }, { "epoch": 123.74143302180686, "grad_norm": 1.5625, "learning_rate": 1.617688956354732e-05, "loss": 0.1358, "step": 19922 }, { "epoch": 123.74766355140187, "grad_norm": 3.25, "learning_rate": 1.6174584382629904e-05, "loss": 0.2981, "step": 19923 }, { "epoch": 123.75389408099689, "grad_norm": 3.09375, "learning_rate": 1.6172279287426682e-05, "loss": 0.3313, "step": 19924 }, { "epoch": 123.7601246105919, "grad_norm": 2.59375, "learning_rate": 1.616997427796005e-05, "loss": 0.226, "step": 19925 }, { "epoch": 123.76635514018692, "grad_norm": 2.828125, "learning_rate": 1.61676693542524e-05, "loss": 0.2448, "step": 19926 }, { "epoch": 123.77258566978193, "grad_norm": 2.5625, "learning_rate": 1.616536451632611e-05, "loss": 0.1792, "step": 19927 }, { "epoch": 123.77881619937695, "grad_norm": 1.3125, "learning_rate": 1.6163059764203553e-05, "loss": 0.1298, "step": 19928 }, { "epoch": 123.78504672897196, "grad_norm": 3.015625, "learning_rate": 1.616075509790714e-05, "loss": 0.2661, "step": 19929 }, { "epoch": 123.79127725856698, "grad_norm": 2.984375, "learning_rate": 1.6158450517459235e-05, "loss": 0.1893, "step": 19930 }, { "epoch": 123.797507788162, "grad_norm": 1.8828125, "learning_rate": 1.6156146022882217e-05, "loss": 0.1077, "step": 19931 }, { "epoch": 123.80373831775701, "grad_norm": 3.28125, "learning_rate": 1.615384161419849e-05, "loss": 0.2617, "step": 19932 }, { "epoch": 123.80996884735202, "grad_norm": 1.8046875, "learning_rate": 1.6151537291430417e-05, "loss": 0.1229, "step": 19933 }, { "epoch": 123.81619937694704, "grad_norm": 2.984375, "learning_rate": 1.614923305460039e-05, "loss": 0.2666, "step": 19934 }, { "epoch": 123.82242990654206, "grad_norm": 1.5078125, "learning_rate": 1.6146928903730764e-05, "loss": 0.1031, "step": 19935 }, { "epoch": 123.82866043613707, "grad_norm": 1.2421875, "learning_rate": 1.6144624838843948e-05, "loss": 0.1125, "step": 19936 }, { "epoch": 123.83489096573209, "grad_norm": 1.8671875, "learning_rate": 1.6142320859962296e-05, "loss": 0.1746, "step": 19937 }, { "epoch": 123.8411214953271, "grad_norm": 1.7421875, "learning_rate": 1.6140016967108197e-05, "loss": 0.1784, "step": 19938 }, { "epoch": 123.84735202492212, "grad_norm": 2.90625, "learning_rate": 1.6137713160304037e-05, "loss": 0.2737, "step": 19939 }, { "epoch": 123.85358255451713, "grad_norm": 1.53125, "learning_rate": 1.6135409439572165e-05, "loss": 0.1098, "step": 19940 }, { "epoch": 123.85981308411215, "grad_norm": 3.28125, "learning_rate": 1.6133105804934966e-05, "loss": 0.2379, "step": 19941 }, { "epoch": 123.86604361370716, "grad_norm": 1.984375, "learning_rate": 1.6130802256414824e-05, "loss": 0.1617, "step": 19942 }, { "epoch": 123.87227414330218, "grad_norm": 2.546875, "learning_rate": 1.6128498794034104e-05, "loss": 0.1368, "step": 19943 }, { "epoch": 123.8785046728972, "grad_norm": 2.359375, "learning_rate": 1.6126195417815167e-05, "loss": 0.1924, "step": 19944 }, { "epoch": 123.88473520249221, "grad_norm": 2.9375, "learning_rate": 1.612389212778041e-05, "loss": 0.285, "step": 19945 }, { "epoch": 123.89096573208722, "grad_norm": 1.328125, "learning_rate": 1.6121588923952177e-05, "loss": 0.121, "step": 19946 }, { "epoch": 123.89719626168224, "grad_norm": 1.15625, "learning_rate": 1.611928580635285e-05, "loss": 0.1052, "step": 19947 }, { "epoch": 123.90342679127725, "grad_norm": 3.4375, "learning_rate": 1.6116982775004788e-05, "loss": 0.4311, "step": 19948 }, { "epoch": 123.90965732087227, "grad_norm": 1.78125, "learning_rate": 1.611467982993037e-05, "loss": 0.134, "step": 19949 }, { "epoch": 123.91588785046729, "grad_norm": 2.5, "learning_rate": 1.6112376971151957e-05, "loss": 0.1211, "step": 19950 }, { "epoch": 123.9221183800623, "grad_norm": 1.1640625, "learning_rate": 1.6110074198691923e-05, "loss": 0.1046, "step": 19951 }, { "epoch": 123.92834890965732, "grad_norm": 1.46875, "learning_rate": 1.6107771512572612e-05, "loss": 0.1277, "step": 19952 }, { "epoch": 123.93457943925233, "grad_norm": 2.703125, "learning_rate": 1.610546891281641e-05, "loss": 0.2349, "step": 19953 }, { "epoch": 123.94080996884735, "grad_norm": 3.640625, "learning_rate": 1.610316639944567e-05, "loss": 0.289, "step": 19954 }, { "epoch": 123.94704049844236, "grad_norm": 1.90625, "learning_rate": 1.610086397248275e-05, "loss": 0.1376, "step": 19955 }, { "epoch": 123.95327102803738, "grad_norm": 1.3125, "learning_rate": 1.6098561631950027e-05, "loss": 0.1308, "step": 19956 }, { "epoch": 123.9595015576324, "grad_norm": 0.90625, "learning_rate": 1.6096259377869843e-05, "loss": 0.0976, "step": 19957 }, { "epoch": 123.96573208722741, "grad_norm": 2.21875, "learning_rate": 1.6093957210264587e-05, "loss": 0.1941, "step": 19958 }, { "epoch": 123.97196261682242, "grad_norm": 2.515625, "learning_rate": 1.609165512915658e-05, "loss": 0.2032, "step": 19959 }, { "epoch": 123.97819314641744, "grad_norm": 3.0, "learning_rate": 1.608935313456821e-05, "loss": 0.2945, "step": 19960 }, { "epoch": 123.98442367601245, "grad_norm": 3.875, "learning_rate": 1.6087051226521814e-05, "loss": 0.3843, "step": 19961 }, { "epoch": 123.99065420560747, "grad_norm": 2.96875, "learning_rate": 1.608474940503976e-05, "loss": 0.3469, "step": 19962 }, { "epoch": 123.99688473520249, "grad_norm": 1.0703125, "learning_rate": 1.6082447670144406e-05, "loss": 0.0821, "step": 19963 }, { "epoch": 124.0, "grad_norm": 1.109375, "learning_rate": 1.608014602185811e-05, "loss": 0.0642, "step": 19964 }, { "epoch": 124.00623052959502, "grad_norm": 1.46875, "learning_rate": 1.6077844460203206e-05, "loss": 0.1262, "step": 19965 }, { "epoch": 124.01246105919003, "grad_norm": 2.421875, "learning_rate": 1.6075542985202067e-05, "loss": 0.1956, "step": 19966 }, { "epoch": 124.01869158878505, "grad_norm": 1.5859375, "learning_rate": 1.6073241596877036e-05, "loss": 0.1322, "step": 19967 }, { "epoch": 124.02492211838006, "grad_norm": 1.65625, "learning_rate": 1.6070940295250462e-05, "loss": 0.0979, "step": 19968 }, { "epoch": 124.03115264797508, "grad_norm": 2.375, "learning_rate": 1.6068639080344708e-05, "loss": 0.2076, "step": 19969 }, { "epoch": 124.03738317757009, "grad_norm": 1.1328125, "learning_rate": 1.606633795218212e-05, "loss": 0.0928, "step": 19970 }, { "epoch": 124.04361370716511, "grad_norm": 2.421875, "learning_rate": 1.606403691078504e-05, "loss": 0.155, "step": 19971 }, { "epoch": 124.04984423676012, "grad_norm": 1.90625, "learning_rate": 1.6061735956175814e-05, "loss": 0.2446, "step": 19972 }, { "epoch": 124.05607476635514, "grad_norm": 3.640625, "learning_rate": 1.6059435088376802e-05, "loss": 0.2791, "step": 19973 }, { "epoch": 124.06230529595015, "grad_norm": 1.4921875, "learning_rate": 1.6057134307410336e-05, "loss": 0.1271, "step": 19974 }, { "epoch": 124.06853582554517, "grad_norm": 2.375, "learning_rate": 1.6054833613298774e-05, "loss": 0.107, "step": 19975 }, { "epoch": 124.07476635514018, "grad_norm": 1.2734375, "learning_rate": 1.6052533006064457e-05, "loss": 0.1047, "step": 19976 }, { "epoch": 124.0809968847352, "grad_norm": 1.8046875, "learning_rate": 1.6050232485729734e-05, "loss": 0.1407, "step": 19977 }, { "epoch": 124.08722741433021, "grad_norm": 2.1875, "learning_rate": 1.6047932052316926e-05, "loss": 0.1217, "step": 19978 }, { "epoch": 124.09345794392523, "grad_norm": 2.21875, "learning_rate": 1.6045631705848404e-05, "loss": 0.2099, "step": 19979 }, { "epoch": 124.09968847352025, "grad_norm": 1.9765625, "learning_rate": 1.6043331446346493e-05, "loss": 0.1083, "step": 19980 }, { "epoch": 124.10591900311526, "grad_norm": 2.78125, "learning_rate": 1.6041031273833533e-05, "loss": 0.2531, "step": 19981 }, { "epoch": 124.11214953271028, "grad_norm": 1.75, "learning_rate": 1.6038731188331878e-05, "loss": 0.1289, "step": 19982 }, { "epoch": 124.11838006230529, "grad_norm": 2.25, "learning_rate": 1.6036431189863854e-05, "loss": 0.2228, "step": 19983 }, { "epoch": 124.1246105919003, "grad_norm": 1.0703125, "learning_rate": 1.6034131278451803e-05, "loss": 0.0928, "step": 19984 }, { "epoch": 124.13084112149532, "grad_norm": 2.453125, "learning_rate": 1.6031831454118056e-05, "loss": 0.1659, "step": 19985 }, { "epoch": 124.13707165109034, "grad_norm": 1.84375, "learning_rate": 1.6029531716884954e-05, "loss": 0.1268, "step": 19986 }, { "epoch": 124.14330218068535, "grad_norm": 1.53125, "learning_rate": 1.6027232066774835e-05, "loss": 0.143, "step": 19987 }, { "epoch": 124.14953271028037, "grad_norm": 2.390625, "learning_rate": 1.602493250381003e-05, "loss": 0.1454, "step": 19988 }, { "epoch": 124.15576323987538, "grad_norm": 1.75, "learning_rate": 1.6022633028012886e-05, "loss": 0.1184, "step": 19989 }, { "epoch": 124.1619937694704, "grad_norm": 2.859375, "learning_rate": 1.6020333639405717e-05, "loss": 0.2477, "step": 19990 }, { "epoch": 124.16822429906541, "grad_norm": 1.0625, "learning_rate": 1.6018034338010857e-05, "loss": 0.0921, "step": 19991 }, { "epoch": 124.17445482866043, "grad_norm": 3.46875, "learning_rate": 1.601573512385065e-05, "loss": 0.1822, "step": 19992 }, { "epoch": 124.18068535825545, "grad_norm": 1.921875, "learning_rate": 1.601343599694742e-05, "loss": 0.132, "step": 19993 }, { "epoch": 124.18691588785046, "grad_norm": 1.40625, "learning_rate": 1.6011136957323486e-05, "loss": 0.1302, "step": 19994 }, { "epoch": 124.19314641744548, "grad_norm": 2.171875, "learning_rate": 1.6008838005001193e-05, "loss": 0.1003, "step": 19995 }, { "epoch": 124.19937694704049, "grad_norm": 3.09375, "learning_rate": 1.6006539140002867e-05, "loss": 0.2689, "step": 19996 }, { "epoch": 124.20560747663552, "grad_norm": 3.234375, "learning_rate": 1.600424036235083e-05, "loss": 0.2081, "step": 19997 }, { "epoch": 124.21183800623054, "grad_norm": 1.7109375, "learning_rate": 1.60019416720674e-05, "loss": 0.1337, "step": 19998 }, { "epoch": 124.21806853582555, "grad_norm": 2.859375, "learning_rate": 1.599964306917492e-05, "loss": 0.2636, "step": 19999 }, { "epoch": 124.22429906542057, "grad_norm": 2.40625, "learning_rate": 1.5997344553695704e-05, "loss": 0.2163, "step": 20000 }, { "epoch": 124.23052959501558, "grad_norm": 1.7265625, "learning_rate": 1.599504612565207e-05, "loss": 0.152, "step": 20001 }, { "epoch": 124.2367601246106, "grad_norm": 2.1875, "learning_rate": 1.599274778506636e-05, "loss": 0.2645, "step": 20002 }, { "epoch": 124.24299065420561, "grad_norm": 2.234375, "learning_rate": 1.599044953196088e-05, "loss": 0.1711, "step": 20003 }, { "epoch": 124.24922118380063, "grad_norm": 3.875, "learning_rate": 1.5988151366357955e-05, "loss": 0.2524, "step": 20004 }, { "epoch": 124.25545171339564, "grad_norm": 4.21875, "learning_rate": 1.59858532882799e-05, "loss": 0.1097, "step": 20005 }, { "epoch": 124.26168224299066, "grad_norm": 1.296875, "learning_rate": 1.5983555297749043e-05, "loss": 0.1187, "step": 20006 }, { "epoch": 124.26791277258567, "grad_norm": 1.8671875, "learning_rate": 1.5981257394787692e-05, "loss": 0.1464, "step": 20007 }, { "epoch": 124.27414330218069, "grad_norm": 1.3125, "learning_rate": 1.597895957941818e-05, "loss": 0.1157, "step": 20008 }, { "epoch": 124.2803738317757, "grad_norm": 1.4609375, "learning_rate": 1.597666185166282e-05, "loss": 0.134, "step": 20009 }, { "epoch": 124.28660436137072, "grad_norm": 2.140625, "learning_rate": 1.5974364211543923e-05, "loss": 0.1591, "step": 20010 }, { "epoch": 124.29283489096574, "grad_norm": 0.94140625, "learning_rate": 1.5972066659083796e-05, "loss": 0.112, "step": 20011 }, { "epoch": 124.29906542056075, "grad_norm": 2.171875, "learning_rate": 1.5969769194304767e-05, "loss": 0.1039, "step": 20012 }, { "epoch": 124.30529595015577, "grad_norm": 1.640625, "learning_rate": 1.5967471817229148e-05, "loss": 0.1154, "step": 20013 }, { "epoch": 124.31152647975078, "grad_norm": 3.484375, "learning_rate": 1.596517452787924e-05, "loss": 0.2879, "step": 20014 }, { "epoch": 124.3177570093458, "grad_norm": 2.421875, "learning_rate": 1.5962877326277376e-05, "loss": 0.179, "step": 20015 }, { "epoch": 124.32398753894081, "grad_norm": 1.1875, "learning_rate": 1.5960580212445848e-05, "loss": 0.0979, "step": 20016 }, { "epoch": 124.33021806853583, "grad_norm": 3.03125, "learning_rate": 1.595828318640697e-05, "loss": 0.2666, "step": 20017 }, { "epoch": 124.33644859813084, "grad_norm": 1.859375, "learning_rate": 1.595598624818305e-05, "loss": 0.1087, "step": 20018 }, { "epoch": 124.34267912772586, "grad_norm": 2.46875, "learning_rate": 1.5953689397796402e-05, "loss": 0.2686, "step": 20019 }, { "epoch": 124.34890965732087, "grad_norm": 2.53125, "learning_rate": 1.5951392635269326e-05, "loss": 0.1774, "step": 20020 }, { "epoch": 124.35514018691589, "grad_norm": 3.0, "learning_rate": 1.5949095960624154e-05, "loss": 0.2029, "step": 20021 }, { "epoch": 124.3613707165109, "grad_norm": 2.359375, "learning_rate": 1.5946799373883147e-05, "loss": 0.1805, "step": 20022 }, { "epoch": 124.36760124610592, "grad_norm": 1.875, "learning_rate": 1.5944502875068645e-05, "loss": 0.1586, "step": 20023 }, { "epoch": 124.37383177570094, "grad_norm": 2.09375, "learning_rate": 1.5942206464202937e-05, "loss": 0.1043, "step": 20024 }, { "epoch": 124.38006230529595, "grad_norm": 3.015625, "learning_rate": 1.5939910141308335e-05, "loss": 0.3354, "step": 20025 }, { "epoch": 124.38629283489097, "grad_norm": 1.7265625, "learning_rate": 1.5937613906407135e-05, "loss": 0.144, "step": 20026 }, { "epoch": 124.39252336448598, "grad_norm": 2.546875, "learning_rate": 1.5935317759521636e-05, "loss": 0.1117, "step": 20027 }, { "epoch": 124.398753894081, "grad_norm": 3.03125, "learning_rate": 1.5933021700674157e-05, "loss": 0.1682, "step": 20028 }, { "epoch": 124.40498442367601, "grad_norm": 2.578125, "learning_rate": 1.593072572988697e-05, "loss": 0.1069, "step": 20029 }, { "epoch": 124.41121495327103, "grad_norm": 2.59375, "learning_rate": 1.5928429847182395e-05, "loss": 0.2434, "step": 20030 }, { "epoch": 124.41744548286604, "grad_norm": 2.828125, "learning_rate": 1.5926134052582714e-05, "loss": 0.2303, "step": 20031 }, { "epoch": 124.42367601246106, "grad_norm": 1.1875, "learning_rate": 1.5923838346110238e-05, "loss": 0.106, "step": 20032 }, { "epoch": 124.42990654205607, "grad_norm": 1.3515625, "learning_rate": 1.5921542727787254e-05, "loss": 0.1103, "step": 20033 }, { "epoch": 124.43613707165109, "grad_norm": 2.125, "learning_rate": 1.5919247197636077e-05, "loss": 0.1529, "step": 20034 }, { "epoch": 124.4423676012461, "grad_norm": 2.0625, "learning_rate": 1.5916951755678973e-05, "loss": 0.1991, "step": 20035 }, { "epoch": 124.44859813084112, "grad_norm": 1.46875, "learning_rate": 1.5914656401938253e-05, "loss": 0.1355, "step": 20036 }, { "epoch": 124.45482866043614, "grad_norm": 2.78125, "learning_rate": 1.5912361136436198e-05, "loss": 0.2248, "step": 20037 }, { "epoch": 124.46105919003115, "grad_norm": 2.234375, "learning_rate": 1.591006595919512e-05, "loss": 0.2021, "step": 20038 }, { "epoch": 124.46728971962617, "grad_norm": 2.359375, "learning_rate": 1.5907770870237295e-05, "loss": 0.1978, "step": 20039 }, { "epoch": 124.47352024922118, "grad_norm": 2.515625, "learning_rate": 1.5905475869585017e-05, "loss": 0.0979, "step": 20040 }, { "epoch": 124.4797507788162, "grad_norm": 2.15625, "learning_rate": 1.5903180957260573e-05, "loss": 0.0992, "step": 20041 }, { "epoch": 124.48598130841121, "grad_norm": 2.765625, "learning_rate": 1.5900886133286252e-05, "loss": 0.2413, "step": 20042 }, { "epoch": 124.49221183800623, "grad_norm": 1.9375, "learning_rate": 1.5898591397684346e-05, "loss": 0.1997, "step": 20043 }, { "epoch": 124.49844236760124, "grad_norm": 1.484375, "learning_rate": 1.589629675047714e-05, "loss": 0.0935, "step": 20044 }, { "epoch": 124.50467289719626, "grad_norm": 2.90625, "learning_rate": 1.589400219168692e-05, "loss": 0.3313, "step": 20045 }, { "epoch": 124.51090342679127, "grad_norm": 1.0859375, "learning_rate": 1.589170772133597e-05, "loss": 0.1061, "step": 20046 }, { "epoch": 124.51713395638629, "grad_norm": 1.2578125, "learning_rate": 1.5889413339446586e-05, "loss": 0.1211, "step": 20047 }, { "epoch": 124.5233644859813, "grad_norm": 2.171875, "learning_rate": 1.5887119046041022e-05, "loss": 0.1521, "step": 20048 }, { "epoch": 124.52959501557632, "grad_norm": 2.03125, "learning_rate": 1.5884824841141592e-05, "loss": 0.2334, "step": 20049 }, { "epoch": 124.53582554517133, "grad_norm": 2.171875, "learning_rate": 1.5882530724770563e-05, "loss": 0.2081, "step": 20050 }, { "epoch": 124.54205607476635, "grad_norm": 2.078125, "learning_rate": 1.588023669695021e-05, "loss": 0.2389, "step": 20051 }, { "epoch": 124.54828660436137, "grad_norm": 1.3828125, "learning_rate": 1.587794275770283e-05, "loss": 0.1042, "step": 20052 }, { "epoch": 124.55451713395638, "grad_norm": 1.8984375, "learning_rate": 1.5875648907050696e-05, "loss": 0.1946, "step": 20053 }, { "epoch": 124.5607476635514, "grad_norm": 1.6171875, "learning_rate": 1.587335514501608e-05, "loss": 0.1468, "step": 20054 }, { "epoch": 124.56697819314641, "grad_norm": 1.8828125, "learning_rate": 1.5871061471621258e-05, "loss": 0.1684, "step": 20055 }, { "epoch": 124.57320872274143, "grad_norm": 1.234375, "learning_rate": 1.586876788688852e-05, "loss": 0.1331, "step": 20056 }, { "epoch": 124.57943925233644, "grad_norm": 1.921875, "learning_rate": 1.5866474390840125e-05, "loss": 0.1643, "step": 20057 }, { "epoch": 124.58566978193146, "grad_norm": 2.484375, "learning_rate": 1.5864180983498363e-05, "loss": 0.229, "step": 20058 }, { "epoch": 124.59190031152647, "grad_norm": 2.546875, "learning_rate": 1.5861887664885504e-05, "loss": 0.2504, "step": 20059 }, { "epoch": 124.59813084112149, "grad_norm": 1.5, "learning_rate": 1.5859594435023816e-05, "loss": 0.1403, "step": 20060 }, { "epoch": 124.60436137071652, "grad_norm": 3.890625, "learning_rate": 1.585730129393557e-05, "loss": 0.2697, "step": 20061 }, { "epoch": 124.61059190031153, "grad_norm": 2.28125, "learning_rate": 1.5855008241643045e-05, "loss": 0.1769, "step": 20062 }, { "epoch": 124.61682242990655, "grad_norm": 1.3203125, "learning_rate": 1.5852715278168506e-05, "loss": 0.1218, "step": 20063 }, { "epoch": 124.62305295950156, "grad_norm": 1.6875, "learning_rate": 1.585042240353422e-05, "loss": 0.1495, "step": 20064 }, { "epoch": 124.62928348909658, "grad_norm": 2.15625, "learning_rate": 1.5848129617762468e-05, "loss": 0.1381, "step": 20065 }, { "epoch": 124.6355140186916, "grad_norm": 2.3125, "learning_rate": 1.5845836920875513e-05, "loss": 0.1348, "step": 20066 }, { "epoch": 124.64174454828661, "grad_norm": 3.125, "learning_rate": 1.5843544312895616e-05, "loss": 0.3304, "step": 20067 }, { "epoch": 124.64797507788163, "grad_norm": 2.109375, "learning_rate": 1.584125179384504e-05, "loss": 0.1405, "step": 20068 }, { "epoch": 124.65420560747664, "grad_norm": 1.8359375, "learning_rate": 1.5838959363746068e-05, "loss": 0.1202, "step": 20069 }, { "epoch": 124.66043613707166, "grad_norm": 3.1875, "learning_rate": 1.5836667022620943e-05, "loss": 0.2353, "step": 20070 }, { "epoch": 124.66666666666667, "grad_norm": 2.5, "learning_rate": 1.583437477049195e-05, "loss": 0.1238, "step": 20071 }, { "epoch": 124.67289719626169, "grad_norm": 1.3125, "learning_rate": 1.5832082607381344e-05, "loss": 0.128, "step": 20072 }, { "epoch": 124.6791277258567, "grad_norm": 1.515625, "learning_rate": 1.582979053331138e-05, "loss": 0.1047, "step": 20073 }, { "epoch": 124.68535825545172, "grad_norm": 1.7421875, "learning_rate": 1.5827498548304313e-05, "loss": 0.1547, "step": 20074 }, { "epoch": 124.69158878504673, "grad_norm": 2.734375, "learning_rate": 1.5825206652382423e-05, "loss": 0.0877, "step": 20075 }, { "epoch": 124.69781931464175, "grad_norm": 2.375, "learning_rate": 1.582291484556796e-05, "loss": 0.2627, "step": 20076 }, { "epoch": 124.70404984423676, "grad_norm": 2.3125, "learning_rate": 1.5820623127883176e-05, "loss": 0.19, "step": 20077 }, { "epoch": 124.71028037383178, "grad_norm": 2.40625, "learning_rate": 1.5818331499350343e-05, "loss": 0.1723, "step": 20078 }, { "epoch": 124.7165109034268, "grad_norm": 1.2578125, "learning_rate": 1.5816039959991712e-05, "loss": 0.1225, "step": 20079 }, { "epoch": 124.72274143302181, "grad_norm": 1.9375, "learning_rate": 1.5813748509829536e-05, "loss": 0.1765, "step": 20080 }, { "epoch": 124.72897196261682, "grad_norm": 3.140625, "learning_rate": 1.581145714888606e-05, "loss": 0.1785, "step": 20081 }, { "epoch": 124.73520249221184, "grad_norm": 1.140625, "learning_rate": 1.5809165877183558e-05, "loss": 0.0993, "step": 20082 }, { "epoch": 124.74143302180686, "grad_norm": 2.796875, "learning_rate": 1.580687469474427e-05, "loss": 0.2091, "step": 20083 }, { "epoch": 124.74766355140187, "grad_norm": 2.015625, "learning_rate": 1.5804583601590454e-05, "loss": 0.1597, "step": 20084 }, { "epoch": 124.75389408099689, "grad_norm": 1.9765625, "learning_rate": 1.5802292597744365e-05, "loss": 0.1453, "step": 20085 }, { "epoch": 124.7601246105919, "grad_norm": 2.453125, "learning_rate": 1.5800001683228248e-05, "loss": 0.2497, "step": 20086 }, { "epoch": 124.76635514018692, "grad_norm": 2.015625, "learning_rate": 1.5797710858064345e-05, "loss": 0.1776, "step": 20087 }, { "epoch": 124.77258566978193, "grad_norm": 2.09375, "learning_rate": 1.5795420122274924e-05, "loss": 0.1607, "step": 20088 }, { "epoch": 124.77881619937695, "grad_norm": 1.125, "learning_rate": 1.5793129475882218e-05, "loss": 0.0947, "step": 20089 }, { "epoch": 124.78504672897196, "grad_norm": 2.640625, "learning_rate": 1.5790838918908475e-05, "loss": 0.2245, "step": 20090 }, { "epoch": 124.79127725856698, "grad_norm": 3.25, "learning_rate": 1.578854845137595e-05, "loss": 0.2341, "step": 20091 }, { "epoch": 124.797507788162, "grad_norm": 2.0625, "learning_rate": 1.5786258073306887e-05, "loss": 0.1185, "step": 20092 }, { "epoch": 124.80373831775701, "grad_norm": 2.28125, "learning_rate": 1.5783967784723526e-05, "loss": 0.1994, "step": 20093 }, { "epoch": 124.80996884735202, "grad_norm": 1.5625, "learning_rate": 1.5781677585648103e-05, "loss": 0.146, "step": 20094 }, { "epoch": 124.81619937694704, "grad_norm": 3.796875, "learning_rate": 1.577938747610288e-05, "loss": 0.3429, "step": 20095 }, { "epoch": 124.82242990654206, "grad_norm": 2.921875, "learning_rate": 1.5777097456110085e-05, "loss": 0.2185, "step": 20096 }, { "epoch": 124.82866043613707, "grad_norm": 1.984375, "learning_rate": 1.577480752569196e-05, "loss": 0.1247, "step": 20097 }, { "epoch": 124.83489096573209, "grad_norm": 1.828125, "learning_rate": 1.5772517684870754e-05, "loss": 0.1435, "step": 20098 }, { "epoch": 124.8411214953271, "grad_norm": 1.0703125, "learning_rate": 1.57702279336687e-05, "loss": 0.1211, "step": 20099 }, { "epoch": 124.84735202492212, "grad_norm": 2.609375, "learning_rate": 1.576793827210804e-05, "loss": 0.1942, "step": 20100 }, { "epoch": 124.85358255451713, "grad_norm": 1.96875, "learning_rate": 1.5765648700211e-05, "loss": 0.1714, "step": 20101 }, { "epoch": 124.85981308411215, "grad_norm": 0.84765625, "learning_rate": 1.576335921799983e-05, "loss": 0.094, "step": 20102 }, { "epoch": 124.86604361370716, "grad_norm": 1.390625, "learning_rate": 1.576106982549676e-05, "loss": 0.1015, "step": 20103 }, { "epoch": 124.87227414330218, "grad_norm": 2.1875, "learning_rate": 1.5758780522724036e-05, "loss": 0.1948, "step": 20104 }, { "epoch": 124.8785046728972, "grad_norm": 1.9453125, "learning_rate": 1.5756491309703875e-05, "loss": 0.1513, "step": 20105 }, { "epoch": 124.88473520249221, "grad_norm": 2.109375, "learning_rate": 1.575420218645852e-05, "loss": 0.2292, "step": 20106 }, { "epoch": 124.89096573208722, "grad_norm": 2.6875, "learning_rate": 1.5751913153010196e-05, "loss": 0.2587, "step": 20107 }, { "epoch": 124.89719626168224, "grad_norm": 2.5, "learning_rate": 1.5749624209381145e-05, "loss": 0.1539, "step": 20108 }, { "epoch": 124.90342679127725, "grad_norm": 2.421875, "learning_rate": 1.5747335355593593e-05, "loss": 0.2124, "step": 20109 }, { "epoch": 124.90965732087227, "grad_norm": 2.421875, "learning_rate": 1.5745046591669778e-05, "loss": 0.2266, "step": 20110 }, { "epoch": 124.91588785046729, "grad_norm": 1.96875, "learning_rate": 1.5742757917631907e-05, "loss": 0.1078, "step": 20111 }, { "epoch": 124.9221183800623, "grad_norm": 1.3359375, "learning_rate": 1.5740469333502226e-05, "loss": 0.1117, "step": 20112 }, { "epoch": 124.92834890965732, "grad_norm": 1.0703125, "learning_rate": 1.5738180839302958e-05, "loss": 0.0982, "step": 20113 }, { "epoch": 124.93457943925233, "grad_norm": 2.109375, "learning_rate": 1.5735892435056326e-05, "loss": 0.1958, "step": 20114 }, { "epoch": 124.94080996884735, "grad_norm": 1.859375, "learning_rate": 1.5733604120784563e-05, "loss": 0.1119, "step": 20115 }, { "epoch": 124.94704049844236, "grad_norm": 2.421875, "learning_rate": 1.573131589650988e-05, "loss": 0.1759, "step": 20116 }, { "epoch": 124.95327102803738, "grad_norm": 1.09375, "learning_rate": 1.572902776225453e-05, "loss": 0.1047, "step": 20117 }, { "epoch": 124.9595015576324, "grad_norm": 1.4609375, "learning_rate": 1.57267397180407e-05, "loss": 0.1266, "step": 20118 }, { "epoch": 124.96573208722741, "grad_norm": 2.0625, "learning_rate": 1.572445176389063e-05, "loss": 0.1411, "step": 20119 }, { "epoch": 124.97196261682242, "grad_norm": 2.28125, "learning_rate": 1.5722163899826538e-05, "loss": 0.1439, "step": 20120 }, { "epoch": 124.97819314641744, "grad_norm": 2.0625, "learning_rate": 1.5719876125870644e-05, "loss": 0.1746, "step": 20121 }, { "epoch": 124.98442367601245, "grad_norm": 1.953125, "learning_rate": 1.571758844204517e-05, "loss": 0.2169, "step": 20122 }, { "epoch": 124.99065420560747, "grad_norm": 1.5, "learning_rate": 1.571530084837234e-05, "loss": 0.1307, "step": 20123 }, { "epoch": 124.99688473520249, "grad_norm": 2.21875, "learning_rate": 1.5713013344874353e-05, "loss": 0.1579, "step": 20124 }, { "epoch": 125.0, "grad_norm": 1.3203125, "learning_rate": 1.571072593157344e-05, "loss": 0.0951, "step": 20125 }, { "epoch": 125.00623052959502, "grad_norm": 1.3671875, "learning_rate": 1.5708438608491814e-05, "loss": 0.1398, "step": 20126 }, { "epoch": 125.01246105919003, "grad_norm": 2.203125, "learning_rate": 1.570615137565169e-05, "loss": 0.2569, "step": 20127 }, { "epoch": 125.01869158878505, "grad_norm": 1.234375, "learning_rate": 1.570386423307528e-05, "loss": 0.0969, "step": 20128 }, { "epoch": 125.02492211838006, "grad_norm": 2.0625, "learning_rate": 1.5701577180784806e-05, "loss": 0.1297, "step": 20129 }, { "epoch": 125.03115264797508, "grad_norm": 3.28125, "learning_rate": 1.569929021880247e-05, "loss": 0.3064, "step": 20130 }, { "epoch": 125.03738317757009, "grad_norm": 2.984375, "learning_rate": 1.569700334715048e-05, "loss": 0.1814, "step": 20131 }, { "epoch": 125.04361370716511, "grad_norm": 2.046875, "learning_rate": 1.5694716565851058e-05, "loss": 0.1708, "step": 20132 }, { "epoch": 125.04984423676012, "grad_norm": 2.609375, "learning_rate": 1.56924298749264e-05, "loss": 0.3062, "step": 20133 }, { "epoch": 125.05607476635514, "grad_norm": 2.609375, "learning_rate": 1.5690143274398734e-05, "loss": 0.2621, "step": 20134 }, { "epoch": 125.06230529595015, "grad_norm": 3.109375, "learning_rate": 1.568785676429026e-05, "loss": 0.1134, "step": 20135 }, { "epoch": 125.06853582554517, "grad_norm": 1.5078125, "learning_rate": 1.568557034462318e-05, "loss": 0.1347, "step": 20136 }, { "epoch": 125.07476635514018, "grad_norm": 2.0, "learning_rate": 1.56832840154197e-05, "loss": 0.2096, "step": 20137 }, { "epoch": 125.0809968847352, "grad_norm": 2.90625, "learning_rate": 1.568099777670203e-05, "loss": 0.299, "step": 20138 }, { "epoch": 125.08722741433021, "grad_norm": 2.859375, "learning_rate": 1.5678711628492368e-05, "loss": 0.2512, "step": 20139 }, { "epoch": 125.09345794392523, "grad_norm": 2.0, "learning_rate": 1.567642557081292e-05, "loss": 0.171, "step": 20140 }, { "epoch": 125.09968847352025, "grad_norm": 2.1875, "learning_rate": 1.5674139603685897e-05, "loss": 0.133, "step": 20141 }, { "epoch": 125.10591900311526, "grad_norm": 4.46875, "learning_rate": 1.56718537271335e-05, "loss": 0.3835, "step": 20142 }, { "epoch": 125.11214953271028, "grad_norm": 2.46875, "learning_rate": 1.5669567941177915e-05, "loss": 0.0855, "step": 20143 }, { "epoch": 125.11838006230529, "grad_norm": 1.9765625, "learning_rate": 1.5667282245841354e-05, "loss": 0.1814, "step": 20144 }, { "epoch": 125.1246105919003, "grad_norm": 1.84375, "learning_rate": 1.566499664114601e-05, "loss": 0.1645, "step": 20145 }, { "epoch": 125.13084112149532, "grad_norm": 1.6796875, "learning_rate": 1.5662711127114087e-05, "loss": 0.1427, "step": 20146 }, { "epoch": 125.13707165109034, "grad_norm": 1.6875, "learning_rate": 1.566042570376778e-05, "loss": 0.1155, "step": 20147 }, { "epoch": 125.14330218068535, "grad_norm": 1.6640625, "learning_rate": 1.5658140371129286e-05, "loss": 0.1354, "step": 20148 }, { "epoch": 125.14953271028037, "grad_norm": 1.46875, "learning_rate": 1.5655855129220803e-05, "loss": 0.1438, "step": 20149 }, { "epoch": 125.15576323987538, "grad_norm": 2.078125, "learning_rate": 1.5653569978064523e-05, "loss": 0.1872, "step": 20150 }, { "epoch": 125.1619937694704, "grad_norm": 1.890625, "learning_rate": 1.565128491768263e-05, "loss": 0.1701, "step": 20151 }, { "epoch": 125.16822429906541, "grad_norm": 3.234375, "learning_rate": 1.5648999948097337e-05, "loss": 0.1943, "step": 20152 }, { "epoch": 125.17445482866043, "grad_norm": 1.9921875, "learning_rate": 1.5646715069330817e-05, "loss": 0.1763, "step": 20153 }, { "epoch": 125.18068535825545, "grad_norm": 3.078125, "learning_rate": 1.5644430281405274e-05, "loss": 0.3076, "step": 20154 }, { "epoch": 125.18691588785046, "grad_norm": 2.125, "learning_rate": 1.5642145584342904e-05, "loss": 0.1613, "step": 20155 }, { "epoch": 125.19314641744548, "grad_norm": 1.5078125, "learning_rate": 1.563986097816588e-05, "loss": 0.1158, "step": 20156 }, { "epoch": 125.19937694704049, "grad_norm": 1.7734375, "learning_rate": 1.5637576462896386e-05, "loss": 0.1706, "step": 20157 }, { "epoch": 125.20560747663552, "grad_norm": 2.46875, "learning_rate": 1.5635292038556632e-05, "loss": 0.1561, "step": 20158 }, { "epoch": 125.21183800623054, "grad_norm": 1.6484375, "learning_rate": 1.5633007705168796e-05, "loss": 0.1351, "step": 20159 }, { "epoch": 125.21806853582555, "grad_norm": 2.25, "learning_rate": 1.5630723462755047e-05, "loss": 0.1228, "step": 20160 }, { "epoch": 125.22429906542057, "grad_norm": 1.5859375, "learning_rate": 1.5628439311337603e-05, "loss": 0.1124, "step": 20161 }, { "epoch": 125.23052959501558, "grad_norm": 1.96875, "learning_rate": 1.5626155250938623e-05, "loss": 0.0925, "step": 20162 }, { "epoch": 125.2367601246106, "grad_norm": 1.421875, "learning_rate": 1.5623871281580296e-05, "loss": 0.148, "step": 20163 }, { "epoch": 125.24299065420561, "grad_norm": 2.453125, "learning_rate": 1.56215874032848e-05, "loss": 0.273, "step": 20164 }, { "epoch": 125.24922118380063, "grad_norm": 2.25, "learning_rate": 1.561930361607433e-05, "loss": 0.1231, "step": 20165 }, { "epoch": 125.25545171339564, "grad_norm": 1.6796875, "learning_rate": 1.5617019919971048e-05, "loss": 0.1595, "step": 20166 }, { "epoch": 125.26168224299066, "grad_norm": 2.546875, "learning_rate": 1.5614736314997153e-05, "loss": 0.2005, "step": 20167 }, { "epoch": 125.26791277258567, "grad_norm": 2.75, "learning_rate": 1.561245280117482e-05, "loss": 0.2587, "step": 20168 }, { "epoch": 125.27414330218069, "grad_norm": 1.7734375, "learning_rate": 1.5610169378526217e-05, "loss": 0.1992, "step": 20169 }, { "epoch": 125.2803738317757, "grad_norm": 2.34375, "learning_rate": 1.560788604707352e-05, "loss": 0.1981, "step": 20170 }, { "epoch": 125.28660436137072, "grad_norm": 2.71875, "learning_rate": 1.5605602806838914e-05, "loss": 0.2742, "step": 20171 }, { "epoch": 125.29283489096574, "grad_norm": 1.96875, "learning_rate": 1.5603319657844576e-05, "loss": 0.1154, "step": 20172 }, { "epoch": 125.29906542056075, "grad_norm": 3.296875, "learning_rate": 1.560103660011267e-05, "loss": 0.2199, "step": 20173 }, { "epoch": 125.30529595015577, "grad_norm": 2.265625, "learning_rate": 1.559875363366538e-05, "loss": 0.2204, "step": 20174 }, { "epoch": 125.31152647975078, "grad_norm": 2.078125, "learning_rate": 1.5596470758524876e-05, "loss": 0.1136, "step": 20175 }, { "epoch": 125.3177570093458, "grad_norm": 2.3125, "learning_rate": 1.5594187974713327e-05, "loss": 0.1635, "step": 20176 }, { "epoch": 125.32398753894081, "grad_norm": 1.4296875, "learning_rate": 1.5591905282252895e-05, "loss": 0.1037, "step": 20177 }, { "epoch": 125.33021806853583, "grad_norm": 2.546875, "learning_rate": 1.5589622681165767e-05, "loss": 0.2059, "step": 20178 }, { "epoch": 125.33644859813084, "grad_norm": 1.796875, "learning_rate": 1.5587340171474097e-05, "loss": 0.1455, "step": 20179 }, { "epoch": 125.34267912772586, "grad_norm": 2.4375, "learning_rate": 1.558505775320008e-05, "loss": 0.2902, "step": 20180 }, { "epoch": 125.34890965732087, "grad_norm": 3.296875, "learning_rate": 1.558277542636585e-05, "loss": 0.1617, "step": 20181 }, { "epoch": 125.35514018691589, "grad_norm": 1.390625, "learning_rate": 1.5580493190993585e-05, "loss": 0.1164, "step": 20182 }, { "epoch": 125.3613707165109, "grad_norm": 1.53125, "learning_rate": 1.5578211047105453e-05, "loss": 0.143, "step": 20183 }, { "epoch": 125.36760124610592, "grad_norm": 2.125, "learning_rate": 1.5575928994723625e-05, "loss": 0.1353, "step": 20184 }, { "epoch": 125.37383177570094, "grad_norm": 2.1875, "learning_rate": 1.557364703387026e-05, "loss": 0.1627, "step": 20185 }, { "epoch": 125.38006230529595, "grad_norm": 3.234375, "learning_rate": 1.557136516456751e-05, "loss": 0.2211, "step": 20186 }, { "epoch": 125.38629283489097, "grad_norm": 0.8125, "learning_rate": 1.5569083386837558e-05, "loss": 0.0807, "step": 20187 }, { "epoch": 125.39252336448598, "grad_norm": 1.8515625, "learning_rate": 1.5566801700702545e-05, "loss": 0.1207, "step": 20188 }, { "epoch": 125.398753894081, "grad_norm": 1.265625, "learning_rate": 1.5564520106184644e-05, "loss": 0.1002, "step": 20189 }, { "epoch": 125.40498442367601, "grad_norm": 1.7734375, "learning_rate": 1.5562238603306002e-05, "loss": 0.1365, "step": 20190 }, { "epoch": 125.41121495327103, "grad_norm": 1.4375, "learning_rate": 1.5559957192088793e-05, "loss": 0.1439, "step": 20191 }, { "epoch": 125.41744548286604, "grad_norm": 1.890625, "learning_rate": 1.5557675872555165e-05, "loss": 0.1469, "step": 20192 }, { "epoch": 125.42367601246106, "grad_norm": 2.453125, "learning_rate": 1.555539464472728e-05, "loss": 0.3075, "step": 20193 }, { "epoch": 125.42990654205607, "grad_norm": 2.75, "learning_rate": 1.555311350862728e-05, "loss": 0.3476, "step": 20194 }, { "epoch": 125.43613707165109, "grad_norm": 3.46875, "learning_rate": 1.555083246427734e-05, "loss": 0.1427, "step": 20195 }, { "epoch": 125.4423676012461, "grad_norm": 1.7734375, "learning_rate": 1.5548551511699598e-05, "loss": 0.1057, "step": 20196 }, { "epoch": 125.44859813084112, "grad_norm": 1.765625, "learning_rate": 1.5546270650916212e-05, "loss": 0.1377, "step": 20197 }, { "epoch": 125.45482866043614, "grad_norm": 1.75, "learning_rate": 1.5543989881949338e-05, "loss": 0.1569, "step": 20198 }, { "epoch": 125.46105919003115, "grad_norm": 1.4453125, "learning_rate": 1.5541709204821124e-05, "loss": 0.1406, "step": 20199 }, { "epoch": 125.46728971962617, "grad_norm": 3.03125, "learning_rate": 1.553942861955373e-05, "loss": 0.145, "step": 20200 }, { "epoch": 125.47352024922118, "grad_norm": 1.171875, "learning_rate": 1.5537148126169283e-05, "loss": 0.1183, "step": 20201 }, { "epoch": 125.4797507788162, "grad_norm": 1.953125, "learning_rate": 1.5534867724689954e-05, "loss": 0.1464, "step": 20202 }, { "epoch": 125.48598130841121, "grad_norm": 1.484375, "learning_rate": 1.5532587415137873e-05, "loss": 0.1471, "step": 20203 }, { "epoch": 125.49221183800623, "grad_norm": 2.28125, "learning_rate": 1.55303071975352e-05, "loss": 0.2358, "step": 20204 }, { "epoch": 125.49844236760124, "grad_norm": 1.515625, "learning_rate": 1.5528027071904076e-05, "loss": 0.1233, "step": 20205 }, { "epoch": 125.50467289719626, "grad_norm": 1.8046875, "learning_rate": 1.5525747038266656e-05, "loss": 0.1458, "step": 20206 }, { "epoch": 125.51090342679127, "grad_norm": 2.109375, "learning_rate": 1.552346709664506e-05, "loss": 0.1845, "step": 20207 }, { "epoch": 125.51713395638629, "grad_norm": 2.484375, "learning_rate": 1.5521187247061453e-05, "loss": 0.1624, "step": 20208 }, { "epoch": 125.5233644859813, "grad_norm": 2.359375, "learning_rate": 1.551890748953797e-05, "loss": 0.2666, "step": 20209 }, { "epoch": 125.52959501557632, "grad_norm": 1.46875, "learning_rate": 1.5516627824096746e-05, "loss": 0.1213, "step": 20210 }, { "epoch": 125.53582554517133, "grad_norm": 1.5390625, "learning_rate": 1.5514348250759936e-05, "loss": 0.0924, "step": 20211 }, { "epoch": 125.54205607476635, "grad_norm": 1.84375, "learning_rate": 1.5512068769549676e-05, "loss": 0.111, "step": 20212 }, { "epoch": 125.54828660436137, "grad_norm": 1.7109375, "learning_rate": 1.5509789380488093e-05, "loss": 0.11, "step": 20213 }, { "epoch": 125.55451713395638, "grad_norm": 2.03125, "learning_rate": 1.550751008359733e-05, "loss": 0.137, "step": 20214 }, { "epoch": 125.5607476635514, "grad_norm": 2.734375, "learning_rate": 1.5505230878899534e-05, "loss": 0.224, "step": 20215 }, { "epoch": 125.56697819314641, "grad_norm": 2.890625, "learning_rate": 1.5502951766416823e-05, "loss": 0.1816, "step": 20216 }, { "epoch": 125.57320872274143, "grad_norm": 2.625, "learning_rate": 1.5500672746171353e-05, "loss": 0.2085, "step": 20217 }, { "epoch": 125.57943925233644, "grad_norm": 2.734375, "learning_rate": 1.5498393818185246e-05, "loss": 0.1932, "step": 20218 }, { "epoch": 125.58566978193146, "grad_norm": 1.875, "learning_rate": 1.549611498248064e-05, "loss": 0.1542, "step": 20219 }, { "epoch": 125.59190031152647, "grad_norm": 1.2421875, "learning_rate": 1.5493836239079663e-05, "loss": 0.1148, "step": 20220 }, { "epoch": 125.59813084112149, "grad_norm": 2.3125, "learning_rate": 1.5491557588004446e-05, "loss": 0.1742, "step": 20221 }, { "epoch": 125.60436137071652, "grad_norm": 1.1171875, "learning_rate": 1.5489279029277125e-05, "loss": 0.1065, "step": 20222 }, { "epoch": 125.61059190031153, "grad_norm": 2.34375, "learning_rate": 1.5487000562919824e-05, "loss": 0.1632, "step": 20223 }, { "epoch": 125.61682242990655, "grad_norm": 2.234375, "learning_rate": 1.5484722188954674e-05, "loss": 0.1609, "step": 20224 }, { "epoch": 125.62305295950156, "grad_norm": 2.296875, "learning_rate": 1.5482443907403815e-05, "loss": 0.195, "step": 20225 }, { "epoch": 125.62928348909658, "grad_norm": 2.484375, "learning_rate": 1.548016571828936e-05, "loss": 0.1554, "step": 20226 }, { "epoch": 125.6355140186916, "grad_norm": 1.796875, "learning_rate": 1.547788762163343e-05, "loss": 0.1363, "step": 20227 }, { "epoch": 125.64174454828661, "grad_norm": 2.578125, "learning_rate": 1.547560961745816e-05, "loss": 0.2354, "step": 20228 }, { "epoch": 125.64797507788163, "grad_norm": 1.578125, "learning_rate": 1.5473331705785678e-05, "loss": 0.1274, "step": 20229 }, { "epoch": 125.65420560747664, "grad_norm": 1.0859375, "learning_rate": 1.5471053886638097e-05, "loss": 0.1045, "step": 20230 }, { "epoch": 125.66043613707166, "grad_norm": 1.9921875, "learning_rate": 1.5468776160037556e-05, "loss": 0.1536, "step": 20231 }, { "epoch": 125.66666666666667, "grad_norm": 2.0, "learning_rate": 1.5466498526006162e-05, "loss": 0.1972, "step": 20232 }, { "epoch": 125.67289719626169, "grad_norm": 3.03125, "learning_rate": 1.5464220984566037e-05, "loss": 0.2401, "step": 20233 }, { "epoch": 125.6791277258567, "grad_norm": 1.359375, "learning_rate": 1.5461943535739302e-05, "loss": 0.1138, "step": 20234 }, { "epoch": 125.68535825545172, "grad_norm": 1.671875, "learning_rate": 1.5459666179548082e-05, "loss": 0.1509, "step": 20235 }, { "epoch": 125.69158878504673, "grad_norm": 1.484375, "learning_rate": 1.5457388916014486e-05, "loss": 0.1022, "step": 20236 }, { "epoch": 125.69781931464175, "grad_norm": 1.21875, "learning_rate": 1.545511174516064e-05, "loss": 0.1313, "step": 20237 }, { "epoch": 125.70404984423676, "grad_norm": 2.734375, "learning_rate": 1.5452834667008663e-05, "loss": 0.205, "step": 20238 }, { "epoch": 125.71028037383178, "grad_norm": 1.7109375, "learning_rate": 1.5450557681580663e-05, "loss": 0.1555, "step": 20239 }, { "epoch": 125.7165109034268, "grad_norm": 2.0625, "learning_rate": 1.5448280788898746e-05, "loss": 0.1144, "step": 20240 }, { "epoch": 125.72274143302181, "grad_norm": 1.1640625, "learning_rate": 1.5446003988985043e-05, "loss": 0.1145, "step": 20241 }, { "epoch": 125.72897196261682, "grad_norm": 0.84375, "learning_rate": 1.5443727281861655e-05, "loss": 0.0942, "step": 20242 }, { "epoch": 125.73520249221184, "grad_norm": 2.125, "learning_rate": 1.5441450667550695e-05, "loss": 0.1421, "step": 20243 }, { "epoch": 125.74143302180686, "grad_norm": 3.015625, "learning_rate": 1.5439174146074286e-05, "loss": 0.2218, "step": 20244 }, { "epoch": 125.74766355140187, "grad_norm": 1.8203125, "learning_rate": 1.5436897717454524e-05, "loss": 0.1569, "step": 20245 }, { "epoch": 125.75389408099689, "grad_norm": 2.171875, "learning_rate": 1.5434621381713528e-05, "loss": 0.1802, "step": 20246 }, { "epoch": 125.7601246105919, "grad_norm": 1.8984375, "learning_rate": 1.5432345138873387e-05, "loss": 0.1521, "step": 20247 }, { "epoch": 125.76635514018692, "grad_norm": 2.046875, "learning_rate": 1.5430068988956237e-05, "loss": 0.1696, "step": 20248 }, { "epoch": 125.77258566978193, "grad_norm": 2.625, "learning_rate": 1.5427792931984158e-05, "loss": 0.2203, "step": 20249 }, { "epoch": 125.77881619937695, "grad_norm": 2.484375, "learning_rate": 1.5425516967979288e-05, "loss": 0.1559, "step": 20250 }, { "epoch": 125.78504672897196, "grad_norm": 1.859375, "learning_rate": 1.542324109696369e-05, "loss": 0.1809, "step": 20251 }, { "epoch": 125.79127725856698, "grad_norm": 2.546875, "learning_rate": 1.54209653189595e-05, "loss": 0.165, "step": 20252 }, { "epoch": 125.797507788162, "grad_norm": 1.546875, "learning_rate": 1.54186896339888e-05, "loss": 0.1374, "step": 20253 }, { "epoch": 125.80373831775701, "grad_norm": 1.3046875, "learning_rate": 1.5416414042073707e-05, "loss": 0.1065, "step": 20254 }, { "epoch": 125.80996884735202, "grad_norm": 2.109375, "learning_rate": 1.5414138543236316e-05, "loss": 0.1541, "step": 20255 }, { "epoch": 125.81619937694704, "grad_norm": 1.3515625, "learning_rate": 1.541186313749873e-05, "loss": 0.1278, "step": 20256 }, { "epoch": 125.82242990654206, "grad_norm": 2.265625, "learning_rate": 1.540958782488305e-05, "loss": 0.2222, "step": 20257 }, { "epoch": 125.82866043613707, "grad_norm": 1.7421875, "learning_rate": 1.540731260541136e-05, "loss": 0.1279, "step": 20258 }, { "epoch": 125.83489096573209, "grad_norm": 1.3515625, "learning_rate": 1.5405037479105776e-05, "loss": 0.135, "step": 20259 }, { "epoch": 125.8411214953271, "grad_norm": 1.2109375, "learning_rate": 1.5402762445988377e-05, "loss": 0.119, "step": 20260 }, { "epoch": 125.84735202492212, "grad_norm": 1.9609375, "learning_rate": 1.5400487506081275e-05, "loss": 0.2134, "step": 20261 }, { "epoch": 125.85358255451713, "grad_norm": 2.71875, "learning_rate": 1.539821265940655e-05, "loss": 0.2689, "step": 20262 }, { "epoch": 125.85981308411215, "grad_norm": 1.40625, "learning_rate": 1.539593790598632e-05, "loss": 0.1024, "step": 20263 }, { "epoch": 125.86604361370716, "grad_norm": 1.953125, "learning_rate": 1.5393663245842643e-05, "loss": 0.1427, "step": 20264 }, { "epoch": 125.87227414330218, "grad_norm": 1.796875, "learning_rate": 1.5391388678997637e-05, "loss": 0.0947, "step": 20265 }, { "epoch": 125.8785046728972, "grad_norm": 1.609375, "learning_rate": 1.5389114205473377e-05, "loss": 0.1179, "step": 20266 }, { "epoch": 125.88473520249221, "grad_norm": 2.859375, "learning_rate": 1.5386839825291966e-05, "loss": 0.2414, "step": 20267 }, { "epoch": 125.89096573208722, "grad_norm": 1.5859375, "learning_rate": 1.5384565538475493e-05, "loss": 0.1378, "step": 20268 }, { "epoch": 125.89719626168224, "grad_norm": 2.921875, "learning_rate": 1.538229134504603e-05, "loss": 0.1483, "step": 20269 }, { "epoch": 125.90342679127725, "grad_norm": 2.125, "learning_rate": 1.538001724502569e-05, "loss": 0.2315, "step": 20270 }, { "epoch": 125.90965732087227, "grad_norm": 1.09375, "learning_rate": 1.537774323843654e-05, "loss": 0.0978, "step": 20271 }, { "epoch": 125.91588785046729, "grad_norm": 1.984375, "learning_rate": 1.537546932530067e-05, "loss": 0.2433, "step": 20272 }, { "epoch": 125.9221183800623, "grad_norm": 3.1875, "learning_rate": 1.537319550564016e-05, "loss": 0.2866, "step": 20273 }, { "epoch": 125.92834890965732, "grad_norm": 3.578125, "learning_rate": 1.537092177947711e-05, "loss": 0.1881, "step": 20274 }, { "epoch": 125.93457943925233, "grad_norm": 2.1875, "learning_rate": 1.536864814683358e-05, "loss": 0.2076, "step": 20275 }, { "epoch": 125.94080996884735, "grad_norm": 2.84375, "learning_rate": 1.5366374607731687e-05, "loss": 0.2479, "step": 20276 }, { "epoch": 125.94704049844236, "grad_norm": 1.4296875, "learning_rate": 1.5364101162193467e-05, "loss": 0.1004, "step": 20277 }, { "epoch": 125.95327102803738, "grad_norm": 1.96875, "learning_rate": 1.5361827810241034e-05, "loss": 0.1606, "step": 20278 }, { "epoch": 125.9595015576324, "grad_norm": 1.9140625, "learning_rate": 1.5359554551896448e-05, "loss": 0.1496, "step": 20279 }, { "epoch": 125.96573208722741, "grad_norm": 1.125, "learning_rate": 1.53572813871818e-05, "loss": 0.1019, "step": 20280 }, { "epoch": 125.97196261682242, "grad_norm": 1.4453125, "learning_rate": 1.5355008316119164e-05, "loss": 0.1385, "step": 20281 }, { "epoch": 125.97819314641744, "grad_norm": 2.828125, "learning_rate": 1.5352735338730616e-05, "loss": 0.1954, "step": 20282 }, { "epoch": 125.98442367601245, "grad_norm": 2.546875, "learning_rate": 1.535046245503823e-05, "loss": 0.2034, "step": 20283 }, { "epoch": 125.99065420560747, "grad_norm": 2.40625, "learning_rate": 1.5348189665064076e-05, "loss": 0.2003, "step": 20284 }, { "epoch": 125.99688473520249, "grad_norm": 1.90625, "learning_rate": 1.534591696883024e-05, "loss": 0.1223, "step": 20285 }, { "epoch": 126.0, "grad_norm": 0.8671875, "learning_rate": 1.5343644366358782e-05, "loss": 0.0549, "step": 20286 }, { "epoch": 126.00623052959502, "grad_norm": 2.140625, "learning_rate": 1.5341371857671782e-05, "loss": 0.1703, "step": 20287 }, { "epoch": 126.01246105919003, "grad_norm": 1.9140625, "learning_rate": 1.5339099442791316e-05, "loss": 0.1453, "step": 20288 }, { "epoch": 126.01869158878505, "grad_norm": 1.9140625, "learning_rate": 1.5336827121739447e-05, "loss": 0.1037, "step": 20289 }, { "epoch": 126.02492211838006, "grad_norm": 1.203125, "learning_rate": 1.5334554894538232e-05, "loss": 0.1079, "step": 20290 }, { "epoch": 126.03115264797508, "grad_norm": 1.921875, "learning_rate": 1.5332282761209766e-05, "loss": 0.2099, "step": 20291 }, { "epoch": 126.03738317757009, "grad_norm": 2.28125, "learning_rate": 1.5330010721776095e-05, "loss": 0.1993, "step": 20292 }, { "epoch": 126.04361370716511, "grad_norm": 2.640625, "learning_rate": 1.532773877625929e-05, "loss": 0.173, "step": 20293 }, { "epoch": 126.04984423676012, "grad_norm": 2.015625, "learning_rate": 1.5325466924681424e-05, "loss": 0.146, "step": 20294 }, { "epoch": 126.05607476635514, "grad_norm": 2.75, "learning_rate": 1.5323195167064563e-05, "loss": 0.097, "step": 20295 }, { "epoch": 126.06230529595015, "grad_norm": 0.796875, "learning_rate": 1.5320923503430763e-05, "loss": 0.1046, "step": 20296 }, { "epoch": 126.06853582554517, "grad_norm": 2.65625, "learning_rate": 1.5318651933802076e-05, "loss": 0.2199, "step": 20297 }, { "epoch": 126.07476635514018, "grad_norm": 1.2265625, "learning_rate": 1.5316380458200586e-05, "loss": 0.1247, "step": 20298 }, { "epoch": 126.0809968847352, "grad_norm": 2.296875, "learning_rate": 1.531410907664834e-05, "loss": 0.2144, "step": 20299 }, { "epoch": 126.08722741433021, "grad_norm": 2.9375, "learning_rate": 1.531183778916741e-05, "loss": 0.2692, "step": 20300 }, { "epoch": 126.09345794392523, "grad_norm": 2.53125, "learning_rate": 1.530956659577985e-05, "loss": 0.2547, "step": 20301 }, { "epoch": 126.09968847352025, "grad_norm": 1.78125, "learning_rate": 1.5307295496507713e-05, "loss": 0.1513, "step": 20302 }, { "epoch": 126.10591900311526, "grad_norm": 2.5, "learning_rate": 1.530502449137305e-05, "loss": 0.1986, "step": 20303 }, { "epoch": 126.11214953271028, "grad_norm": 1.921875, "learning_rate": 1.530275358039794e-05, "loss": 0.1308, "step": 20304 }, { "epoch": 126.11838006230529, "grad_norm": 1.625, "learning_rate": 1.5300482763604418e-05, "loss": 0.166, "step": 20305 }, { "epoch": 126.1246105919003, "grad_norm": 1.96875, "learning_rate": 1.5298212041014543e-05, "loss": 0.1615, "step": 20306 }, { "epoch": 126.13084112149532, "grad_norm": 2.140625, "learning_rate": 1.5295941412650382e-05, "loss": 0.1273, "step": 20307 }, { "epoch": 126.13707165109034, "grad_norm": 1.21875, "learning_rate": 1.5293670878533977e-05, "loss": 0.1042, "step": 20308 }, { "epoch": 126.14330218068535, "grad_norm": 3.71875, "learning_rate": 1.529140043868738e-05, "loss": 0.2706, "step": 20309 }, { "epoch": 126.14953271028037, "grad_norm": 1.6796875, "learning_rate": 1.5289130093132632e-05, "loss": 0.1028, "step": 20310 }, { "epoch": 126.15576323987538, "grad_norm": 1.8515625, "learning_rate": 1.5286859841891805e-05, "loss": 0.1279, "step": 20311 }, { "epoch": 126.1619937694704, "grad_norm": 2.296875, "learning_rate": 1.5284589684986928e-05, "loss": 0.1638, "step": 20312 }, { "epoch": 126.16822429906541, "grad_norm": 1.96875, "learning_rate": 1.5282319622440067e-05, "loss": 0.1663, "step": 20313 }, { "epoch": 126.17445482866043, "grad_norm": 2.921875, "learning_rate": 1.528004965427326e-05, "loss": 0.1681, "step": 20314 }, { "epoch": 126.18068535825545, "grad_norm": 1.78125, "learning_rate": 1.5277779780508552e-05, "loss": 0.1428, "step": 20315 }, { "epoch": 126.18691588785046, "grad_norm": 3.171875, "learning_rate": 1.5275510001167988e-05, "loss": 0.2605, "step": 20316 }, { "epoch": 126.19314641744548, "grad_norm": 2.671875, "learning_rate": 1.5273240316273613e-05, "loss": 0.2134, "step": 20317 }, { "epoch": 126.19937694704049, "grad_norm": 1.515625, "learning_rate": 1.5270970725847477e-05, "loss": 0.1452, "step": 20318 }, { "epoch": 126.20560747663552, "grad_norm": 1.6796875, "learning_rate": 1.5268701229911607e-05, "loss": 0.1574, "step": 20319 }, { "epoch": 126.21183800623054, "grad_norm": 1.484375, "learning_rate": 1.5266431828488077e-05, "loss": 0.1172, "step": 20320 }, { "epoch": 126.21806853582555, "grad_norm": 2.28125, "learning_rate": 1.5264162521598892e-05, "loss": 0.1557, "step": 20321 }, { "epoch": 126.22429906542057, "grad_norm": 1.8203125, "learning_rate": 1.5261893309266113e-05, "loss": 0.2383, "step": 20322 }, { "epoch": 126.23052959501558, "grad_norm": 1.5703125, "learning_rate": 1.5259624191511762e-05, "loss": 0.115, "step": 20323 }, { "epoch": 126.2367601246106, "grad_norm": 2.625, "learning_rate": 1.5257355168357896e-05, "loss": 0.2206, "step": 20324 }, { "epoch": 126.24299065420561, "grad_norm": 3.09375, "learning_rate": 1.5255086239826538e-05, "loss": 0.2204, "step": 20325 }, { "epoch": 126.24922118380063, "grad_norm": 1.28125, "learning_rate": 1.525281740593974e-05, "loss": 0.1075, "step": 20326 }, { "epoch": 126.25545171339564, "grad_norm": 1.2265625, "learning_rate": 1.5250548666719527e-05, "loss": 0.0992, "step": 20327 }, { "epoch": 126.26168224299066, "grad_norm": 2.5625, "learning_rate": 1.5248280022187933e-05, "loss": 0.1205, "step": 20328 }, { "epoch": 126.26791277258567, "grad_norm": 1.7578125, "learning_rate": 1.5246011472366984e-05, "loss": 0.1072, "step": 20329 }, { "epoch": 126.27414330218069, "grad_norm": 1.6484375, "learning_rate": 1.524374301727873e-05, "loss": 0.1023, "step": 20330 }, { "epoch": 126.2803738317757, "grad_norm": 2.078125, "learning_rate": 1.5241474656945193e-05, "loss": 0.1279, "step": 20331 }, { "epoch": 126.28660436137072, "grad_norm": 2.171875, "learning_rate": 1.52392063913884e-05, "loss": 0.1762, "step": 20332 }, { "epoch": 126.29283489096574, "grad_norm": 1.6640625, "learning_rate": 1.5236938220630392e-05, "loss": 0.1165, "step": 20333 }, { "epoch": 126.29906542056075, "grad_norm": 2.640625, "learning_rate": 1.5234670144693192e-05, "loss": 0.0987, "step": 20334 }, { "epoch": 126.30529595015577, "grad_norm": 2.46875, "learning_rate": 1.5232402163598824e-05, "loss": 0.2359, "step": 20335 }, { "epoch": 126.31152647975078, "grad_norm": 2.234375, "learning_rate": 1.5230134277369317e-05, "loss": 0.2617, "step": 20336 }, { "epoch": 126.3177570093458, "grad_norm": 0.91015625, "learning_rate": 1.5227866486026701e-05, "loss": 0.0947, "step": 20337 }, { "epoch": 126.32398753894081, "grad_norm": 1.296875, "learning_rate": 1.5225598789592999e-05, "loss": 0.1068, "step": 20338 }, { "epoch": 126.33021806853583, "grad_norm": 2.8125, "learning_rate": 1.5223331188090234e-05, "loss": 0.169, "step": 20339 }, { "epoch": 126.33644859813084, "grad_norm": 1.9765625, "learning_rate": 1.5221063681540437e-05, "loss": 0.1048, "step": 20340 }, { "epoch": 126.34267912772586, "grad_norm": 2.453125, "learning_rate": 1.521879626996562e-05, "loss": 0.1725, "step": 20341 }, { "epoch": 126.34890965732087, "grad_norm": 2.609375, "learning_rate": 1.521652895338781e-05, "loss": 0.2849, "step": 20342 }, { "epoch": 126.35514018691589, "grad_norm": 1.3828125, "learning_rate": 1.5214261731829022e-05, "loss": 0.1161, "step": 20343 }, { "epoch": 126.3613707165109, "grad_norm": 1.828125, "learning_rate": 1.5211994605311284e-05, "loss": 0.1167, "step": 20344 }, { "epoch": 126.36760124610592, "grad_norm": 2.90625, "learning_rate": 1.5209727573856602e-05, "loss": 0.2067, "step": 20345 }, { "epoch": 126.37383177570094, "grad_norm": 1.2265625, "learning_rate": 1.5207460637487025e-05, "loss": 0.1142, "step": 20346 }, { "epoch": 126.38006230529595, "grad_norm": 2.15625, "learning_rate": 1.5205193796224529e-05, "loss": 0.0961, "step": 20347 }, { "epoch": 126.38629283489097, "grad_norm": 2.46875, "learning_rate": 1.5202927050091154e-05, "loss": 0.1748, "step": 20348 }, { "epoch": 126.39252336448598, "grad_norm": 2.796875, "learning_rate": 1.5200660399108906e-05, "loss": 0.2457, "step": 20349 }, { "epoch": 126.398753894081, "grad_norm": 2.625, "learning_rate": 1.5198393843299808e-05, "loss": 0.1961, "step": 20350 }, { "epoch": 126.40498442367601, "grad_norm": 3.015625, "learning_rate": 1.5196127382685866e-05, "loss": 0.1608, "step": 20351 }, { "epoch": 126.41121495327103, "grad_norm": 1.609375, "learning_rate": 1.5193861017289102e-05, "loss": 0.1563, "step": 20352 }, { "epoch": 126.41744548286604, "grad_norm": 1.2265625, "learning_rate": 1.519159474713151e-05, "loss": 0.1136, "step": 20353 }, { "epoch": 126.42367601246106, "grad_norm": 3.234375, "learning_rate": 1.5189328572235115e-05, "loss": 0.2133, "step": 20354 }, { "epoch": 126.42990654205607, "grad_norm": 2.234375, "learning_rate": 1.5187062492621923e-05, "loss": 0.1889, "step": 20355 }, { "epoch": 126.43613707165109, "grad_norm": 2.28125, "learning_rate": 1.5184796508313934e-05, "loss": 0.2312, "step": 20356 }, { "epoch": 126.4423676012461, "grad_norm": 1.4609375, "learning_rate": 1.5182530619333169e-05, "loss": 0.1029, "step": 20357 }, { "epoch": 126.44859813084112, "grad_norm": 1.859375, "learning_rate": 1.5180264825701623e-05, "loss": 0.1689, "step": 20358 }, { "epoch": 126.45482866043614, "grad_norm": 1.5546875, "learning_rate": 1.5177999127441328e-05, "loss": 0.1361, "step": 20359 }, { "epoch": 126.46105919003115, "grad_norm": 2.53125, "learning_rate": 1.5175733524574248e-05, "loss": 0.2126, "step": 20360 }, { "epoch": 126.46728971962617, "grad_norm": 3.0, "learning_rate": 1.5173468017122416e-05, "loss": 0.2036, "step": 20361 }, { "epoch": 126.47352024922118, "grad_norm": 1.5, "learning_rate": 1.517120260510782e-05, "loss": 0.1217, "step": 20362 }, { "epoch": 126.4797507788162, "grad_norm": 1.4375, "learning_rate": 1.5168937288552476e-05, "loss": 0.1489, "step": 20363 }, { "epoch": 126.48598130841121, "grad_norm": 2.203125, "learning_rate": 1.516667206747838e-05, "loss": 0.1854, "step": 20364 }, { "epoch": 126.49221183800623, "grad_norm": 2.171875, "learning_rate": 1.5164406941907534e-05, "loss": 0.1517, "step": 20365 }, { "epoch": 126.49844236760124, "grad_norm": 2.5625, "learning_rate": 1.516214191186192e-05, "loss": 0.1911, "step": 20366 }, { "epoch": 126.50467289719626, "grad_norm": 1.875, "learning_rate": 1.515987697736356e-05, "loss": 0.1271, "step": 20367 }, { "epoch": 126.51090342679127, "grad_norm": 2.109375, "learning_rate": 1.515761213843444e-05, "loss": 0.1538, "step": 20368 }, { "epoch": 126.51713395638629, "grad_norm": 2.46875, "learning_rate": 1.5155347395096554e-05, "loss": 0.1748, "step": 20369 }, { "epoch": 126.5233644859813, "grad_norm": 2.359375, "learning_rate": 1.5153082747371904e-05, "loss": 0.1638, "step": 20370 }, { "epoch": 126.52959501557632, "grad_norm": 2.375, "learning_rate": 1.5150818195282492e-05, "loss": 0.1684, "step": 20371 }, { "epoch": 126.53582554517133, "grad_norm": 0.90625, "learning_rate": 1.5148553738850296e-05, "loss": 0.0879, "step": 20372 }, { "epoch": 126.54205607476635, "grad_norm": 1.78125, "learning_rate": 1.5146289378097311e-05, "loss": 0.1408, "step": 20373 }, { "epoch": 126.54828660436137, "grad_norm": 2.390625, "learning_rate": 1.5144025113045537e-05, "loss": 0.0826, "step": 20374 }, { "epoch": 126.55451713395638, "grad_norm": 2.046875, "learning_rate": 1.5141760943716957e-05, "loss": 0.1222, "step": 20375 }, { "epoch": 126.5607476635514, "grad_norm": 1.8359375, "learning_rate": 1.5139496870133569e-05, "loss": 0.137, "step": 20376 }, { "epoch": 126.56697819314641, "grad_norm": 3.125, "learning_rate": 1.513723289231736e-05, "loss": 0.2484, "step": 20377 }, { "epoch": 126.57320872274143, "grad_norm": 2.5625, "learning_rate": 1.5134969010290322e-05, "loss": 0.1665, "step": 20378 }, { "epoch": 126.57943925233644, "grad_norm": 2.546875, "learning_rate": 1.5132705224074425e-05, "loss": 0.1837, "step": 20379 }, { "epoch": 126.58566978193146, "grad_norm": 2.640625, "learning_rate": 1.5130441533691674e-05, "loss": 0.1942, "step": 20380 }, { "epoch": 126.59190031152647, "grad_norm": 1.765625, "learning_rate": 1.5128177939164046e-05, "loss": 0.1601, "step": 20381 }, { "epoch": 126.59813084112149, "grad_norm": 2.453125, "learning_rate": 1.5125914440513522e-05, "loss": 0.1668, "step": 20382 }, { "epoch": 126.60436137071652, "grad_norm": 2.515625, "learning_rate": 1.5123651037762099e-05, "loss": 0.1704, "step": 20383 }, { "epoch": 126.61059190031153, "grad_norm": 2.6875, "learning_rate": 1.5121387730931754e-05, "loss": 0.2254, "step": 20384 }, { "epoch": 126.61682242990655, "grad_norm": 1.5078125, "learning_rate": 1.5119124520044461e-05, "loss": 0.1381, "step": 20385 }, { "epoch": 126.62305295950156, "grad_norm": 1.7578125, "learning_rate": 1.5116861405122202e-05, "loss": 0.1055, "step": 20386 }, { "epoch": 126.62928348909658, "grad_norm": 2.140625, "learning_rate": 1.5114598386186965e-05, "loss": 0.1421, "step": 20387 }, { "epoch": 126.6355140186916, "grad_norm": 2.375, "learning_rate": 1.5112335463260723e-05, "loss": 0.1276, "step": 20388 }, { "epoch": 126.64174454828661, "grad_norm": 1.7578125, "learning_rate": 1.5110072636365451e-05, "loss": 0.0925, "step": 20389 }, { "epoch": 126.64797507788163, "grad_norm": 2.859375, "learning_rate": 1.5107809905523143e-05, "loss": 0.2174, "step": 20390 }, { "epoch": 126.65420560747664, "grad_norm": 2.046875, "learning_rate": 1.5105547270755757e-05, "loss": 0.2831, "step": 20391 }, { "epoch": 126.66043613707166, "grad_norm": 2.5625, "learning_rate": 1.5103284732085272e-05, "loss": 0.2082, "step": 20392 }, { "epoch": 126.66666666666667, "grad_norm": 2.390625, "learning_rate": 1.5101022289533661e-05, "loss": 0.2092, "step": 20393 }, { "epoch": 126.67289719626169, "grad_norm": 2.4375, "learning_rate": 1.5098759943122904e-05, "loss": 0.1897, "step": 20394 }, { "epoch": 126.6791277258567, "grad_norm": 2.671875, "learning_rate": 1.5096497692874964e-05, "loss": 0.2181, "step": 20395 }, { "epoch": 126.68535825545172, "grad_norm": 1.4140625, "learning_rate": 1.5094235538811825e-05, "loss": 0.1339, "step": 20396 }, { "epoch": 126.69158878504673, "grad_norm": 2.8125, "learning_rate": 1.5091973480955451e-05, "loss": 0.3798, "step": 20397 }, { "epoch": 126.69781931464175, "grad_norm": 1.3203125, "learning_rate": 1.5089711519327812e-05, "loss": 0.124, "step": 20398 }, { "epoch": 126.70404984423676, "grad_norm": 1.9140625, "learning_rate": 1.5087449653950868e-05, "loss": 0.0962, "step": 20399 }, { "epoch": 126.71028037383178, "grad_norm": 1.171875, "learning_rate": 1.5085187884846599e-05, "loss": 0.1013, "step": 20400 }, { "epoch": 126.7165109034268, "grad_norm": 1.1640625, "learning_rate": 1.5082926212036969e-05, "loss": 0.1139, "step": 20401 }, { "epoch": 126.72274143302181, "grad_norm": 1.484375, "learning_rate": 1.5080664635543934e-05, "loss": 0.1259, "step": 20402 }, { "epoch": 126.72897196261682, "grad_norm": 1.28125, "learning_rate": 1.5078403155389481e-05, "loss": 0.1138, "step": 20403 }, { "epoch": 126.73520249221184, "grad_norm": 1.984375, "learning_rate": 1.5076141771595553e-05, "loss": 0.2018, "step": 20404 }, { "epoch": 126.74143302180686, "grad_norm": 1.53125, "learning_rate": 1.507388048418412e-05, "loss": 0.1567, "step": 20405 }, { "epoch": 126.74766355140187, "grad_norm": 1.671875, "learning_rate": 1.5071619293177136e-05, "loss": 0.1226, "step": 20406 }, { "epoch": 126.75389408099689, "grad_norm": 2.59375, "learning_rate": 1.5069358198596578e-05, "loss": 0.2469, "step": 20407 }, { "epoch": 126.7601246105919, "grad_norm": 2.5625, "learning_rate": 1.5067097200464391e-05, "loss": 0.1285, "step": 20408 }, { "epoch": 126.76635514018692, "grad_norm": 2.3125, "learning_rate": 1.5064836298802551e-05, "loss": 0.1175, "step": 20409 }, { "epoch": 126.77258566978193, "grad_norm": 2.203125, "learning_rate": 1.5062575493633008e-05, "loss": 0.1847, "step": 20410 }, { "epoch": 126.77881619937695, "grad_norm": 1.953125, "learning_rate": 1.5060314784977713e-05, "loss": 0.171, "step": 20411 }, { "epoch": 126.78504672897196, "grad_norm": 1.875, "learning_rate": 1.5058054172858626e-05, "loss": 0.1388, "step": 20412 }, { "epoch": 126.79127725856698, "grad_norm": 1.8359375, "learning_rate": 1.5055793657297706e-05, "loss": 0.173, "step": 20413 }, { "epoch": 126.797507788162, "grad_norm": 1.578125, "learning_rate": 1.5053533238316909e-05, "loss": 0.1433, "step": 20414 }, { "epoch": 126.80373831775701, "grad_norm": 1.609375, "learning_rate": 1.5051272915938177e-05, "loss": 0.099, "step": 20415 }, { "epoch": 126.80996884735202, "grad_norm": 1.46875, "learning_rate": 1.5049012690183485e-05, "loss": 0.1208, "step": 20416 }, { "epoch": 126.81619937694704, "grad_norm": 1.2421875, "learning_rate": 1.5046752561074762e-05, "loss": 0.1269, "step": 20417 }, { "epoch": 126.82242990654206, "grad_norm": 2.78125, "learning_rate": 1.5044492528633968e-05, "loss": 0.2226, "step": 20418 }, { "epoch": 126.82866043613707, "grad_norm": 1.625, "learning_rate": 1.5042232592883052e-05, "loss": 0.1516, "step": 20419 }, { "epoch": 126.83489096573209, "grad_norm": 3.21875, "learning_rate": 1.5039972753843965e-05, "loss": 0.297, "step": 20420 }, { "epoch": 126.8411214953271, "grad_norm": 3.3125, "learning_rate": 1.5037713011538651e-05, "loss": 0.1408, "step": 20421 }, { "epoch": 126.84735202492212, "grad_norm": 2.140625, "learning_rate": 1.5035453365989072e-05, "loss": 0.2076, "step": 20422 }, { "epoch": 126.85358255451713, "grad_norm": 3.71875, "learning_rate": 1.5033193817217145e-05, "loss": 0.2025, "step": 20423 }, { "epoch": 126.85981308411215, "grad_norm": 2.875, "learning_rate": 1.5030934365244842e-05, "loss": 0.2763, "step": 20424 }, { "epoch": 126.86604361370716, "grad_norm": 1.9609375, "learning_rate": 1.502867501009409e-05, "loss": 0.1731, "step": 20425 }, { "epoch": 126.87227414330218, "grad_norm": 1.3046875, "learning_rate": 1.5026415751786844e-05, "loss": 0.1263, "step": 20426 }, { "epoch": 126.8785046728972, "grad_norm": 1.8671875, "learning_rate": 1.502415659034504e-05, "loss": 0.1174, "step": 20427 }, { "epoch": 126.88473520249221, "grad_norm": 2.03125, "learning_rate": 1.5021897525790621e-05, "loss": 0.1363, "step": 20428 }, { "epoch": 126.89096573208722, "grad_norm": 1.3984375, "learning_rate": 1.5019638558145535e-05, "loss": 0.1146, "step": 20429 }, { "epoch": 126.89719626168224, "grad_norm": 2.875, "learning_rate": 1.501737968743171e-05, "loss": 0.1971, "step": 20430 }, { "epoch": 126.90342679127725, "grad_norm": 1.34375, "learning_rate": 1.5015120913671089e-05, "loss": 0.1287, "step": 20431 }, { "epoch": 126.90965732087227, "grad_norm": 2.515625, "learning_rate": 1.5012862236885606e-05, "loss": 0.1724, "step": 20432 }, { "epoch": 126.91588785046729, "grad_norm": 1.828125, "learning_rate": 1.5010603657097203e-05, "loss": 0.132, "step": 20433 }, { "epoch": 126.9221183800623, "grad_norm": 2.765625, "learning_rate": 1.5008345174327815e-05, "loss": 0.2218, "step": 20434 }, { "epoch": 126.92834890965732, "grad_norm": 2.78125, "learning_rate": 1.5006086788599384e-05, "loss": 0.1914, "step": 20435 }, { "epoch": 126.93457943925233, "grad_norm": 4.28125, "learning_rate": 1.500382849993382e-05, "loss": 0.2382, "step": 20436 }, { "epoch": 126.94080996884735, "grad_norm": 1.46875, "learning_rate": 1.500157030835308e-05, "loss": 0.1389, "step": 20437 }, { "epoch": 126.94704049844236, "grad_norm": 2.953125, "learning_rate": 1.499931221387909e-05, "loss": 0.2417, "step": 20438 }, { "epoch": 126.95327102803738, "grad_norm": 2.515625, "learning_rate": 1.4997054216533773e-05, "loss": 0.2621, "step": 20439 }, { "epoch": 126.9595015576324, "grad_norm": 4.0, "learning_rate": 1.4994796316339069e-05, "loss": 0.1193, "step": 20440 }, { "epoch": 126.96573208722741, "grad_norm": 0.92578125, "learning_rate": 1.4992538513316906e-05, "loss": 0.0925, "step": 20441 }, { "epoch": 126.97196261682242, "grad_norm": 2.375, "learning_rate": 1.499028080748921e-05, "loss": 0.1392, "step": 20442 }, { "epoch": 126.97819314641744, "grad_norm": 2.203125, "learning_rate": 1.4988023198877899e-05, "loss": 0.1793, "step": 20443 }, { "epoch": 126.98442367601245, "grad_norm": 2.4375, "learning_rate": 1.4985765687504912e-05, "loss": 0.1825, "step": 20444 }, { "epoch": 126.99065420560747, "grad_norm": 3.109375, "learning_rate": 1.4983508273392166e-05, "loss": 0.1941, "step": 20445 }, { "epoch": 126.99688473520249, "grad_norm": 1.6640625, "learning_rate": 1.4981250956561599e-05, "loss": 0.0904, "step": 20446 }, { "epoch": 127.0, "grad_norm": 1.296875, "learning_rate": 1.4978993737035119e-05, "loss": 0.0938, "step": 20447 }, { "epoch": 127.00623052959502, "grad_norm": 2.78125, "learning_rate": 1.4976736614834664e-05, "loss": 0.2397, "step": 20448 }, { "epoch": 127.01246105919003, "grad_norm": 2.4375, "learning_rate": 1.4974479589982137e-05, "loss": 0.1076, "step": 20449 }, { "epoch": 127.01869158878505, "grad_norm": 1.8359375, "learning_rate": 1.4972222662499471e-05, "loss": 0.134, "step": 20450 }, { "epoch": 127.02492211838006, "grad_norm": 3.671875, "learning_rate": 1.4969965832408583e-05, "loss": 0.2213, "step": 20451 }, { "epoch": 127.03115264797508, "grad_norm": 2.671875, "learning_rate": 1.496770909973139e-05, "loss": 0.2683, "step": 20452 }, { "epoch": 127.03738317757009, "grad_norm": 2.21875, "learning_rate": 1.4965452464489815e-05, "loss": 0.1457, "step": 20453 }, { "epoch": 127.04361370716511, "grad_norm": 2.765625, "learning_rate": 1.4963195926705775e-05, "loss": 0.2978, "step": 20454 }, { "epoch": 127.04984423676012, "grad_norm": 2.125, "learning_rate": 1.496093948640118e-05, "loss": 0.2329, "step": 20455 }, { "epoch": 127.05607476635514, "grad_norm": 1.453125, "learning_rate": 1.4958683143597941e-05, "loss": 0.1267, "step": 20456 }, { "epoch": 127.06230529595015, "grad_norm": 2.015625, "learning_rate": 1.4956426898317983e-05, "loss": 0.1041, "step": 20457 }, { "epoch": 127.06853582554517, "grad_norm": 1.9609375, "learning_rate": 1.4954170750583207e-05, "loss": 0.1244, "step": 20458 }, { "epoch": 127.07476635514018, "grad_norm": 2.21875, "learning_rate": 1.4951914700415542e-05, "loss": 0.1863, "step": 20459 }, { "epoch": 127.0809968847352, "grad_norm": 1.96875, "learning_rate": 1.4949658747836895e-05, "loss": 0.1396, "step": 20460 }, { "epoch": 127.08722741433021, "grad_norm": 3.953125, "learning_rate": 1.4947402892869167e-05, "loss": 0.2479, "step": 20461 }, { "epoch": 127.09345794392523, "grad_norm": 1.875, "learning_rate": 1.4945147135534266e-05, "loss": 0.122, "step": 20462 }, { "epoch": 127.09968847352025, "grad_norm": 2.4375, "learning_rate": 1.4942891475854107e-05, "loss": 0.1051, "step": 20463 }, { "epoch": 127.10591900311526, "grad_norm": 2.90625, "learning_rate": 1.4940635913850599e-05, "loss": 0.2583, "step": 20464 }, { "epoch": 127.11214953271028, "grad_norm": 1.8515625, "learning_rate": 1.4938380449545641e-05, "loss": 0.1347, "step": 20465 }, { "epoch": 127.11838006230529, "grad_norm": 3.734375, "learning_rate": 1.493612508296115e-05, "loss": 0.4005, "step": 20466 }, { "epoch": 127.1246105919003, "grad_norm": 2.546875, "learning_rate": 1.4933869814119025e-05, "loss": 0.1003, "step": 20467 }, { "epoch": 127.13084112149532, "grad_norm": 2.515625, "learning_rate": 1.4931614643041167e-05, "loss": 0.1663, "step": 20468 }, { "epoch": 127.13707165109034, "grad_norm": 1.8984375, "learning_rate": 1.4929359569749474e-05, "loss": 0.176, "step": 20469 }, { "epoch": 127.14330218068535, "grad_norm": 2.9375, "learning_rate": 1.4927104594265858e-05, "loss": 0.1659, "step": 20470 }, { "epoch": 127.14953271028037, "grad_norm": 2.453125, "learning_rate": 1.492484971661221e-05, "loss": 0.1776, "step": 20471 }, { "epoch": 127.15576323987538, "grad_norm": 2.703125, "learning_rate": 1.4922594936810444e-05, "loss": 0.2119, "step": 20472 }, { "epoch": 127.1619937694704, "grad_norm": 2.4375, "learning_rate": 1.4920340254882454e-05, "loss": 0.1024, "step": 20473 }, { "epoch": 127.16822429906541, "grad_norm": 1.6640625, "learning_rate": 1.4918085670850129e-05, "loss": 0.1407, "step": 20474 }, { "epoch": 127.17445482866043, "grad_norm": 1.40625, "learning_rate": 1.4915831184735369e-05, "loss": 0.1146, "step": 20475 }, { "epoch": 127.18068535825545, "grad_norm": 1.6953125, "learning_rate": 1.4913576796560072e-05, "loss": 0.1417, "step": 20476 }, { "epoch": 127.18691588785046, "grad_norm": 1.703125, "learning_rate": 1.4911322506346137e-05, "loss": 0.1115, "step": 20477 }, { "epoch": 127.19314641744548, "grad_norm": 2.3125, "learning_rate": 1.4909068314115448e-05, "loss": 0.1387, "step": 20478 }, { "epoch": 127.19937694704049, "grad_norm": 1.6875, "learning_rate": 1.4906814219889908e-05, "loss": 0.1352, "step": 20479 }, { "epoch": 127.20560747663552, "grad_norm": 2.0, "learning_rate": 1.4904560223691417e-05, "loss": 0.1351, "step": 20480 }, { "epoch": 127.21183800623054, "grad_norm": 3.125, "learning_rate": 1.4902306325541842e-05, "loss": 0.3446, "step": 20481 }, { "epoch": 127.21806853582555, "grad_norm": 2.109375, "learning_rate": 1.4900052525463088e-05, "loss": 0.1693, "step": 20482 }, { "epoch": 127.22429906542057, "grad_norm": 2.15625, "learning_rate": 1.4897798823477043e-05, "loss": 0.1656, "step": 20483 }, { "epoch": 127.23052959501558, "grad_norm": 1.484375, "learning_rate": 1.48955452196056e-05, "loss": 0.1155, "step": 20484 }, { "epoch": 127.2367601246106, "grad_norm": 3.546875, "learning_rate": 1.4893291713870632e-05, "loss": 0.1882, "step": 20485 }, { "epoch": 127.24299065420561, "grad_norm": 1.8359375, "learning_rate": 1.4891038306294047e-05, "loss": 0.1052, "step": 20486 }, { "epoch": 127.24922118380063, "grad_norm": 0.796875, "learning_rate": 1.4888784996897714e-05, "loss": 0.0938, "step": 20487 }, { "epoch": 127.25545171339564, "grad_norm": 2.25, "learning_rate": 1.4886531785703522e-05, "loss": 0.1594, "step": 20488 }, { "epoch": 127.26168224299066, "grad_norm": 1.703125, "learning_rate": 1.4884278672733354e-05, "loss": 0.0862, "step": 20489 }, { "epoch": 127.26791277258567, "grad_norm": 1.4765625, "learning_rate": 1.4882025658009094e-05, "loss": 0.1303, "step": 20490 }, { "epoch": 127.27414330218069, "grad_norm": 3.421875, "learning_rate": 1.487977274155262e-05, "loss": 0.1039, "step": 20491 }, { "epoch": 127.2803738317757, "grad_norm": 2.34375, "learning_rate": 1.487751992338583e-05, "loss": 0.1261, "step": 20492 }, { "epoch": 127.28660436137072, "grad_norm": 2.84375, "learning_rate": 1.4875267203530573e-05, "loss": 0.2648, "step": 20493 }, { "epoch": 127.29283489096574, "grad_norm": 2.609375, "learning_rate": 1.4873014582008754e-05, "loss": 0.1272, "step": 20494 }, { "epoch": 127.29906542056075, "grad_norm": 3.5, "learning_rate": 1.4870762058842236e-05, "loss": 0.3013, "step": 20495 }, { "epoch": 127.30529595015577, "grad_norm": 2.359375, "learning_rate": 1.4868509634052904e-05, "loss": 0.1346, "step": 20496 }, { "epoch": 127.31152647975078, "grad_norm": 2.421875, "learning_rate": 1.4866257307662634e-05, "loss": 0.1637, "step": 20497 }, { "epoch": 127.3177570093458, "grad_norm": 0.98046875, "learning_rate": 1.4864005079693294e-05, "loss": 0.0847, "step": 20498 }, { "epoch": 127.32398753894081, "grad_norm": 1.953125, "learning_rate": 1.4861752950166772e-05, "loss": 0.1249, "step": 20499 }, { "epoch": 127.33021806853583, "grad_norm": 2.453125, "learning_rate": 1.4859500919104926e-05, "loss": 0.1299, "step": 20500 }, { "epoch": 127.33644859813084, "grad_norm": 2.703125, "learning_rate": 1.4857248986529636e-05, "loss": 0.2354, "step": 20501 }, { "epoch": 127.34267912772586, "grad_norm": 0.9609375, "learning_rate": 1.4854997152462764e-05, "loss": 0.0983, "step": 20502 }, { "epoch": 127.34890965732087, "grad_norm": 2.546875, "learning_rate": 1.4852745416926195e-05, "loss": 0.2486, "step": 20503 }, { "epoch": 127.35514018691589, "grad_norm": 2.65625, "learning_rate": 1.4850493779941782e-05, "loss": 0.2902, "step": 20504 }, { "epoch": 127.3613707165109, "grad_norm": 1.953125, "learning_rate": 1.4848242241531424e-05, "loss": 0.2166, "step": 20505 }, { "epoch": 127.36760124610592, "grad_norm": 1.578125, "learning_rate": 1.4845990801716946e-05, "loss": 0.1308, "step": 20506 }, { "epoch": 127.37383177570094, "grad_norm": 3.140625, "learning_rate": 1.484373946052024e-05, "loss": 0.1842, "step": 20507 }, { "epoch": 127.38006230529595, "grad_norm": 1.8203125, "learning_rate": 1.4841488217963165e-05, "loss": 0.2126, "step": 20508 }, { "epoch": 127.38629283489097, "grad_norm": 2.34375, "learning_rate": 1.4839237074067591e-05, "loss": 0.0935, "step": 20509 }, { "epoch": 127.39252336448598, "grad_norm": 2.359375, "learning_rate": 1.4836986028855377e-05, "loss": 0.1836, "step": 20510 }, { "epoch": 127.398753894081, "grad_norm": 2.171875, "learning_rate": 1.4834735082348388e-05, "loss": 0.1453, "step": 20511 }, { "epoch": 127.40498442367601, "grad_norm": 2.25, "learning_rate": 1.4832484234568479e-05, "loss": 0.2604, "step": 20512 }, { "epoch": 127.41121495327103, "grad_norm": 2.03125, "learning_rate": 1.483023348553752e-05, "loss": 0.1866, "step": 20513 }, { "epoch": 127.41744548286604, "grad_norm": 3.21875, "learning_rate": 1.4827982835277365e-05, "loss": 0.2559, "step": 20514 }, { "epoch": 127.42367601246106, "grad_norm": 1.296875, "learning_rate": 1.4825732283809868e-05, "loss": 0.0886, "step": 20515 }, { "epoch": 127.42990654205607, "grad_norm": 2.421875, "learning_rate": 1.48234818311569e-05, "loss": 0.1947, "step": 20516 }, { "epoch": 127.43613707165109, "grad_norm": 2.4375, "learning_rate": 1.4821231477340302e-05, "loss": 0.19, "step": 20517 }, { "epoch": 127.4423676012461, "grad_norm": 2.578125, "learning_rate": 1.4818981222381959e-05, "loss": 0.1933, "step": 20518 }, { "epoch": 127.44859813084112, "grad_norm": 1.34375, "learning_rate": 1.4816731066303686e-05, "loss": 0.1341, "step": 20519 }, { "epoch": 127.45482866043614, "grad_norm": 1.5078125, "learning_rate": 1.4814481009127363e-05, "loss": 0.1195, "step": 20520 }, { "epoch": 127.46105919003115, "grad_norm": 2.0625, "learning_rate": 1.481223105087483e-05, "loss": 0.1394, "step": 20521 }, { "epoch": 127.46728971962617, "grad_norm": 2.109375, "learning_rate": 1.4809981191567956e-05, "loss": 0.1616, "step": 20522 }, { "epoch": 127.47352024922118, "grad_norm": 1.6953125, "learning_rate": 1.4807731431228577e-05, "loss": 0.1433, "step": 20523 }, { "epoch": 127.4797507788162, "grad_norm": 1.9140625, "learning_rate": 1.4805481769878555e-05, "loss": 0.1371, "step": 20524 }, { "epoch": 127.48598130841121, "grad_norm": 1.921875, "learning_rate": 1.480323220753972e-05, "loss": 0.1496, "step": 20525 }, { "epoch": 127.49221183800623, "grad_norm": 2.203125, "learning_rate": 1.4800982744233937e-05, "loss": 0.2051, "step": 20526 }, { "epoch": 127.49844236760124, "grad_norm": 2.953125, "learning_rate": 1.4798733379983049e-05, "loss": 0.1256, "step": 20527 }, { "epoch": 127.50467289719626, "grad_norm": 1.484375, "learning_rate": 1.4796484114808895e-05, "loss": 0.1252, "step": 20528 }, { "epoch": 127.51090342679127, "grad_norm": 0.984375, "learning_rate": 1.4794234948733331e-05, "loss": 0.0782, "step": 20529 }, { "epoch": 127.51713395638629, "grad_norm": 2.890625, "learning_rate": 1.4791985881778204e-05, "loss": 0.1601, "step": 20530 }, { "epoch": 127.5233644859813, "grad_norm": 2.796875, "learning_rate": 1.4789736913965349e-05, "loss": 0.235, "step": 20531 }, { "epoch": 127.52959501557632, "grad_norm": 1.59375, "learning_rate": 1.4787488045316602e-05, "loss": 0.1701, "step": 20532 }, { "epoch": 127.53582554517133, "grad_norm": 1.28125, "learning_rate": 1.4785239275853818e-05, "loss": 0.1053, "step": 20533 }, { "epoch": 127.54205607476635, "grad_norm": 1.859375, "learning_rate": 1.4782990605598835e-05, "loss": 0.1661, "step": 20534 }, { "epoch": 127.54828660436137, "grad_norm": 2.6875, "learning_rate": 1.478074203457348e-05, "loss": 0.2282, "step": 20535 }, { "epoch": 127.55451713395638, "grad_norm": 0.96484375, "learning_rate": 1.4778493562799605e-05, "loss": 0.105, "step": 20536 }, { "epoch": 127.5607476635514, "grad_norm": 2.765625, "learning_rate": 1.4776245190299054e-05, "loss": 0.277, "step": 20537 }, { "epoch": 127.56697819314641, "grad_norm": 1.3671875, "learning_rate": 1.4773996917093646e-05, "loss": 0.1119, "step": 20538 }, { "epoch": 127.57320872274143, "grad_norm": 1.4296875, "learning_rate": 1.4771748743205218e-05, "loss": 0.1352, "step": 20539 }, { "epoch": 127.57943925233644, "grad_norm": 2.328125, "learning_rate": 1.4769500668655616e-05, "loss": 0.1802, "step": 20540 }, { "epoch": 127.58566978193146, "grad_norm": 1.5390625, "learning_rate": 1.4767252693466665e-05, "loss": 0.1374, "step": 20541 }, { "epoch": 127.59190031152647, "grad_norm": 1.8515625, "learning_rate": 1.4765004817660206e-05, "loss": 0.1346, "step": 20542 }, { "epoch": 127.59813084112149, "grad_norm": 1.9765625, "learning_rate": 1.4762757041258071e-05, "loss": 0.1885, "step": 20543 }, { "epoch": 127.60436137071652, "grad_norm": 1.53125, "learning_rate": 1.4760509364282082e-05, "loss": 0.1393, "step": 20544 }, { "epoch": 127.61059190031153, "grad_norm": 2.8125, "learning_rate": 1.4758261786754063e-05, "loss": 0.1929, "step": 20545 }, { "epoch": 127.61682242990655, "grad_norm": 1.859375, "learning_rate": 1.4756014308695864e-05, "loss": 0.1316, "step": 20546 }, { "epoch": 127.62305295950156, "grad_norm": 1.6015625, "learning_rate": 1.4753766930129298e-05, "loss": 0.1509, "step": 20547 }, { "epoch": 127.62928348909658, "grad_norm": 2.203125, "learning_rate": 1.4751519651076193e-05, "loss": 0.2457, "step": 20548 }, { "epoch": 127.6355140186916, "grad_norm": 3.09375, "learning_rate": 1.4749272471558381e-05, "loss": 0.2262, "step": 20549 }, { "epoch": 127.64174454828661, "grad_norm": 1.9453125, "learning_rate": 1.474702539159769e-05, "loss": 0.1761, "step": 20550 }, { "epoch": 127.64797507788163, "grad_norm": 1.015625, "learning_rate": 1.4744778411215932e-05, "loss": 0.1185, "step": 20551 }, { "epoch": 127.65420560747664, "grad_norm": 2.234375, "learning_rate": 1.4742531530434933e-05, "loss": 0.1786, "step": 20552 }, { "epoch": 127.66043613707166, "grad_norm": 1.3828125, "learning_rate": 1.474028474927652e-05, "loss": 0.1152, "step": 20553 }, { "epoch": 127.66666666666667, "grad_norm": 1.28125, "learning_rate": 1.4738038067762511e-05, "loss": 0.0982, "step": 20554 }, { "epoch": 127.67289719626169, "grad_norm": 2.203125, "learning_rate": 1.473579148591473e-05, "loss": 0.1188, "step": 20555 }, { "epoch": 127.6791277258567, "grad_norm": 1.7734375, "learning_rate": 1.4733545003755e-05, "loss": 0.1704, "step": 20556 }, { "epoch": 127.68535825545172, "grad_norm": 1.5078125, "learning_rate": 1.4731298621305128e-05, "loss": 0.1408, "step": 20557 }, { "epoch": 127.69158878504673, "grad_norm": 2.140625, "learning_rate": 1.472905233858693e-05, "loss": 0.1154, "step": 20558 }, { "epoch": 127.69781931464175, "grad_norm": 1.3671875, "learning_rate": 1.4726806155622235e-05, "loss": 0.1086, "step": 20559 }, { "epoch": 127.70404984423676, "grad_norm": 2.265625, "learning_rate": 1.4724560072432852e-05, "loss": 0.2142, "step": 20560 }, { "epoch": 127.71028037383178, "grad_norm": 1.6171875, "learning_rate": 1.472231408904059e-05, "loss": 0.1243, "step": 20561 }, { "epoch": 127.7165109034268, "grad_norm": 1.9375, "learning_rate": 1.472006820546728e-05, "loss": 0.1416, "step": 20562 }, { "epoch": 127.72274143302181, "grad_norm": 3.171875, "learning_rate": 1.4717822421734718e-05, "loss": 0.2857, "step": 20563 }, { "epoch": 127.72897196261682, "grad_norm": 2.28125, "learning_rate": 1.4715576737864716e-05, "loss": 0.2047, "step": 20564 }, { "epoch": 127.73520249221184, "grad_norm": 1.640625, "learning_rate": 1.4713331153879086e-05, "loss": 0.1351, "step": 20565 }, { "epoch": 127.74143302180686, "grad_norm": 1.40625, "learning_rate": 1.4711085669799646e-05, "loss": 0.1183, "step": 20566 }, { "epoch": 127.74766355140187, "grad_norm": 2.53125, "learning_rate": 1.4708840285648192e-05, "loss": 0.1419, "step": 20567 }, { "epoch": 127.75389408099689, "grad_norm": 1.3359375, "learning_rate": 1.4706595001446543e-05, "loss": 0.1284, "step": 20568 }, { "epoch": 127.7601246105919, "grad_norm": 2.484375, "learning_rate": 1.4704349817216506e-05, "loss": 0.1113, "step": 20569 }, { "epoch": 127.76635514018692, "grad_norm": 2.140625, "learning_rate": 1.470210473297988e-05, "loss": 0.1289, "step": 20570 }, { "epoch": 127.77258566978193, "grad_norm": 1.703125, "learning_rate": 1.4699859748758466e-05, "loss": 0.1391, "step": 20571 }, { "epoch": 127.77881619937695, "grad_norm": 2.34375, "learning_rate": 1.4697614864574073e-05, "loss": 0.097, "step": 20572 }, { "epoch": 127.78504672897196, "grad_norm": 3.421875, "learning_rate": 1.4695370080448505e-05, "loss": 0.0799, "step": 20573 }, { "epoch": 127.79127725856698, "grad_norm": 3.140625, "learning_rate": 1.4693125396403562e-05, "loss": 0.2195, "step": 20574 }, { "epoch": 127.797507788162, "grad_norm": 2.25, "learning_rate": 1.4690880812461057e-05, "loss": 0.1338, "step": 20575 }, { "epoch": 127.80373831775701, "grad_norm": 1.5703125, "learning_rate": 1.4688636328642763e-05, "loss": 0.12, "step": 20576 }, { "epoch": 127.80996884735202, "grad_norm": 1.234375, "learning_rate": 1.46863919449705e-05, "loss": 0.1313, "step": 20577 }, { "epoch": 127.81619937694704, "grad_norm": 2.6875, "learning_rate": 1.4684147661466058e-05, "loss": 0.2984, "step": 20578 }, { "epoch": 127.82242990654206, "grad_norm": 2.734375, "learning_rate": 1.4681903478151238e-05, "loss": 0.2288, "step": 20579 }, { "epoch": 127.82866043613707, "grad_norm": 2.546875, "learning_rate": 1.4679659395047834e-05, "loss": 0.2975, "step": 20580 }, { "epoch": 127.83489096573209, "grad_norm": 1.59375, "learning_rate": 1.4677415412177647e-05, "loss": 0.1256, "step": 20581 }, { "epoch": 127.8411214953271, "grad_norm": 1.265625, "learning_rate": 1.4675171529562455e-05, "loss": 0.1252, "step": 20582 }, { "epoch": 127.84735202492212, "grad_norm": 2.421875, "learning_rate": 1.4672927747224064e-05, "loss": 0.1656, "step": 20583 }, { "epoch": 127.85358255451713, "grad_norm": 2.296875, "learning_rate": 1.4670684065184265e-05, "loss": 0.1427, "step": 20584 }, { "epoch": 127.85981308411215, "grad_norm": 2.453125, "learning_rate": 1.466844048346484e-05, "loss": 0.294, "step": 20585 }, { "epoch": 127.86604361370716, "grad_norm": 1.828125, "learning_rate": 1.4666197002087594e-05, "loss": 0.1074, "step": 20586 }, { "epoch": 127.87227414330218, "grad_norm": 3.59375, "learning_rate": 1.4663953621074303e-05, "loss": 0.1755, "step": 20587 }, { "epoch": 127.8785046728972, "grad_norm": 2.140625, "learning_rate": 1.4661710340446775e-05, "loss": 0.1831, "step": 20588 }, { "epoch": 127.88473520249221, "grad_norm": 1.4296875, "learning_rate": 1.4659467160226767e-05, "loss": 0.1079, "step": 20589 }, { "epoch": 127.89096573208722, "grad_norm": 2.859375, "learning_rate": 1.4657224080436088e-05, "loss": 0.1648, "step": 20590 }, { "epoch": 127.89719626168224, "grad_norm": 1.4140625, "learning_rate": 1.465498110109651e-05, "loss": 0.1226, "step": 20591 }, { "epoch": 127.90342679127725, "grad_norm": 1.9921875, "learning_rate": 1.4652738222229827e-05, "loss": 0.1489, "step": 20592 }, { "epoch": 127.90965732087227, "grad_norm": 2.140625, "learning_rate": 1.465049544385782e-05, "loss": 0.1207, "step": 20593 }, { "epoch": 127.91588785046729, "grad_norm": 1.6484375, "learning_rate": 1.4648252766002274e-05, "loss": 0.1047, "step": 20594 }, { "epoch": 127.9221183800623, "grad_norm": 2.078125, "learning_rate": 1.4646010188684958e-05, "loss": 0.1605, "step": 20595 }, { "epoch": 127.92834890965732, "grad_norm": 2.5, "learning_rate": 1.4643767711927663e-05, "loss": 0.2616, "step": 20596 }, { "epoch": 127.93457943925233, "grad_norm": 3.421875, "learning_rate": 1.4641525335752172e-05, "loss": 0.2006, "step": 20597 }, { "epoch": 127.94080996884735, "grad_norm": 3.15625, "learning_rate": 1.4639283060180248e-05, "loss": 0.21, "step": 20598 }, { "epoch": 127.94704049844236, "grad_norm": 1.25, "learning_rate": 1.4637040885233683e-05, "loss": 0.1028, "step": 20599 }, { "epoch": 127.95327102803738, "grad_norm": 1.3125, "learning_rate": 1.4634798810934245e-05, "loss": 0.125, "step": 20600 }, { "epoch": 127.9595015576324, "grad_norm": 3.015625, "learning_rate": 1.463255683730373e-05, "loss": 0.2451, "step": 20601 }, { "epoch": 127.96573208722741, "grad_norm": 2.75, "learning_rate": 1.4630314964363876e-05, "loss": 0.1, "step": 20602 }, { "epoch": 127.97196261682242, "grad_norm": 1.3515625, "learning_rate": 1.4628073192136482e-05, "loss": 0.1136, "step": 20603 }, { "epoch": 127.97819314641744, "grad_norm": 2.578125, "learning_rate": 1.462583152064331e-05, "loss": 0.2158, "step": 20604 }, { "epoch": 127.98442367601245, "grad_norm": 1.9609375, "learning_rate": 1.462358994990614e-05, "loss": 0.1956, "step": 20605 }, { "epoch": 127.99065420560747, "grad_norm": 3.359375, "learning_rate": 1.462134847994674e-05, "loss": 0.3077, "step": 20606 }, { "epoch": 127.99688473520249, "grad_norm": 2.5, "learning_rate": 1.4619107110786884e-05, "loss": 0.2257, "step": 20607 }, { "epoch": 128.0, "grad_norm": 2.25, "learning_rate": 1.4616865842448324e-05, "loss": 0.2439, "step": 20608 }, { "epoch": 128.00623052959503, "grad_norm": 2.625, "learning_rate": 1.4614624674952842e-05, "loss": 0.2217, "step": 20609 }, { "epoch": 128.01246105919003, "grad_norm": 2.375, "learning_rate": 1.4612383608322203e-05, "loss": 0.2452, "step": 20610 }, { "epoch": 128.01869158878506, "grad_norm": 1.9765625, "learning_rate": 1.4610142642578165e-05, "loss": 0.1396, "step": 20611 }, { "epoch": 128.02492211838006, "grad_norm": 2.09375, "learning_rate": 1.4607901777742505e-05, "loss": 0.1571, "step": 20612 }, { "epoch": 128.0311526479751, "grad_norm": 3.578125, "learning_rate": 1.4605661013836983e-05, "loss": 0.2485, "step": 20613 }, { "epoch": 128.0373831775701, "grad_norm": 1.6484375, "learning_rate": 1.4603420350883357e-05, "loss": 0.1231, "step": 20614 }, { "epoch": 128.04361370716512, "grad_norm": 1.5234375, "learning_rate": 1.460117978890338e-05, "loss": 0.1191, "step": 20615 }, { "epoch": 128.04984423676012, "grad_norm": 1.5, "learning_rate": 1.4598939327918834e-05, "loss": 0.1468, "step": 20616 }, { "epoch": 128.05607476635515, "grad_norm": 2.609375, "learning_rate": 1.459669896795146e-05, "loss": 0.2006, "step": 20617 }, { "epoch": 128.06230529595015, "grad_norm": 1.9453125, "learning_rate": 1.4594458709023034e-05, "loss": 0.1923, "step": 20618 }, { "epoch": 128.06853582554518, "grad_norm": 3.375, "learning_rate": 1.4592218551155299e-05, "loss": 0.2603, "step": 20619 }, { "epoch": 128.07476635514018, "grad_norm": 1.65625, "learning_rate": 1.4589978494370027e-05, "loss": 0.1253, "step": 20620 }, { "epoch": 128.0809968847352, "grad_norm": 3.546875, "learning_rate": 1.4587738538688963e-05, "loss": 0.2205, "step": 20621 }, { "epoch": 128.08722741433021, "grad_norm": 2.328125, "learning_rate": 1.4585498684133855e-05, "loss": 0.1621, "step": 20622 }, { "epoch": 128.09345794392524, "grad_norm": 1.2109375, "learning_rate": 1.4583258930726468e-05, "loss": 0.1142, "step": 20623 }, { "epoch": 128.09968847352025, "grad_norm": 3.296875, "learning_rate": 1.4581019278488561e-05, "loss": 0.3777, "step": 20624 }, { "epoch": 128.10591900311528, "grad_norm": 2.5, "learning_rate": 1.4578779727441866e-05, "loss": 0.2565, "step": 20625 }, { "epoch": 128.11214953271028, "grad_norm": 2.359375, "learning_rate": 1.4576540277608155e-05, "loss": 0.2042, "step": 20626 }, { "epoch": 128.1183800623053, "grad_norm": 1.1015625, "learning_rate": 1.457430092900916e-05, "loss": 0.0874, "step": 20627 }, { "epoch": 128.1246105919003, "grad_norm": 1.078125, "learning_rate": 1.457206168166665e-05, "loss": 0.113, "step": 20628 }, { "epoch": 128.13084112149534, "grad_norm": 1.9921875, "learning_rate": 1.456982253560235e-05, "loss": 0.2458, "step": 20629 }, { "epoch": 128.13707165109034, "grad_norm": 3.09375, "learning_rate": 1.4567583490838016e-05, "loss": 0.2557, "step": 20630 }, { "epoch": 128.14330218068537, "grad_norm": 2.890625, "learning_rate": 1.456534454739541e-05, "loss": 0.1886, "step": 20631 }, { "epoch": 128.14953271028037, "grad_norm": 3.078125, "learning_rate": 1.4563105705296263e-05, "loss": 0.2548, "step": 20632 }, { "epoch": 128.1557632398754, "grad_norm": 1.2890625, "learning_rate": 1.456086696456231e-05, "loss": 0.1315, "step": 20633 }, { "epoch": 128.1619937694704, "grad_norm": 2.203125, "learning_rate": 1.4558628325215301e-05, "loss": 0.2401, "step": 20634 }, { "epoch": 128.16822429906543, "grad_norm": 1.0625, "learning_rate": 1.4556389787276992e-05, "loss": 0.123, "step": 20635 }, { "epoch": 128.17445482866043, "grad_norm": 2.34375, "learning_rate": 1.4554151350769104e-05, "loss": 0.1182, "step": 20636 }, { "epoch": 128.18068535825546, "grad_norm": 1.609375, "learning_rate": 1.4551913015713387e-05, "loss": 0.0948, "step": 20637 }, { "epoch": 128.18691588785046, "grad_norm": 3.171875, "learning_rate": 1.4549674782131578e-05, "loss": 0.211, "step": 20638 }, { "epoch": 128.1931464174455, "grad_norm": 2.09375, "learning_rate": 1.4547436650045438e-05, "loss": 0.1647, "step": 20639 }, { "epoch": 128.1993769470405, "grad_norm": 1.6640625, "learning_rate": 1.454519861947666e-05, "loss": 0.108, "step": 20640 }, { "epoch": 128.20560747663552, "grad_norm": 1.84375, "learning_rate": 1.4542960690447005e-05, "loss": 0.1477, "step": 20641 }, { "epoch": 128.21183800623052, "grad_norm": 3.171875, "learning_rate": 1.4540722862978207e-05, "loss": 0.3542, "step": 20642 }, { "epoch": 128.21806853582555, "grad_norm": 1.125, "learning_rate": 1.4538485137092007e-05, "loss": 0.0973, "step": 20643 }, { "epoch": 128.22429906542055, "grad_norm": 2.15625, "learning_rate": 1.453624751281012e-05, "loss": 0.1841, "step": 20644 }, { "epoch": 128.23052959501558, "grad_norm": 1.4140625, "learning_rate": 1.4534009990154302e-05, "loss": 0.1332, "step": 20645 }, { "epoch": 128.23676012461058, "grad_norm": 2.171875, "learning_rate": 1.4531772569146268e-05, "loss": 0.2257, "step": 20646 }, { "epoch": 128.2429906542056, "grad_norm": 1.6328125, "learning_rate": 1.452953524980774e-05, "loss": 0.0821, "step": 20647 }, { "epoch": 128.2492211838006, "grad_norm": 2.71875, "learning_rate": 1.4527298032160464e-05, "loss": 0.2169, "step": 20648 }, { "epoch": 128.25545171339564, "grad_norm": 3.125, "learning_rate": 1.452506091622616e-05, "loss": 0.309, "step": 20649 }, { "epoch": 128.26168224299064, "grad_norm": 2.25, "learning_rate": 1.4522823902026567e-05, "loss": 0.1482, "step": 20650 }, { "epoch": 128.26791277258567, "grad_norm": 1.34375, "learning_rate": 1.4520586989583406e-05, "loss": 0.1357, "step": 20651 }, { "epoch": 128.27414330218068, "grad_norm": 2.640625, "learning_rate": 1.4518350178918388e-05, "loss": 0.1971, "step": 20652 }, { "epoch": 128.2803738317757, "grad_norm": 1.3203125, "learning_rate": 1.4516113470053243e-05, "loss": 0.116, "step": 20653 }, { "epoch": 128.2866043613707, "grad_norm": 1.078125, "learning_rate": 1.4513876863009718e-05, "loss": 0.114, "step": 20654 }, { "epoch": 128.29283489096574, "grad_norm": 1.546875, "learning_rate": 1.4511640357809503e-05, "loss": 0.1359, "step": 20655 }, { "epoch": 128.29906542056074, "grad_norm": 2.25, "learning_rate": 1.4509403954474332e-05, "loss": 0.2209, "step": 20656 }, { "epoch": 128.30529595015577, "grad_norm": 2.015625, "learning_rate": 1.4507167653025938e-05, "loss": 0.1313, "step": 20657 }, { "epoch": 128.31152647975077, "grad_norm": 1.5625, "learning_rate": 1.4504931453486031e-05, "loss": 0.1211, "step": 20658 }, { "epoch": 128.3177570093458, "grad_norm": 2.40625, "learning_rate": 1.4502695355876316e-05, "loss": 0.201, "step": 20659 }, { "epoch": 128.3239875389408, "grad_norm": 2.078125, "learning_rate": 1.4500459360218522e-05, "loss": 0.1598, "step": 20660 }, { "epoch": 128.33021806853583, "grad_norm": 1.140625, "learning_rate": 1.4498223466534374e-05, "loss": 0.1133, "step": 20661 }, { "epoch": 128.33644859813083, "grad_norm": 1.8046875, "learning_rate": 1.4495987674845573e-05, "loss": 0.1077, "step": 20662 }, { "epoch": 128.34267912772586, "grad_norm": 1.390625, "learning_rate": 1.4493751985173837e-05, "loss": 0.1246, "step": 20663 }, { "epoch": 128.34890965732086, "grad_norm": 2.25, "learning_rate": 1.449151639754089e-05, "loss": 0.1536, "step": 20664 }, { "epoch": 128.3551401869159, "grad_norm": 1.5078125, "learning_rate": 1.4489280911968438e-05, "loss": 0.1427, "step": 20665 }, { "epoch": 128.3613707165109, "grad_norm": 2.125, "learning_rate": 1.4487045528478183e-05, "loss": 0.1462, "step": 20666 }, { "epoch": 128.36760124610592, "grad_norm": 1.3671875, "learning_rate": 1.448481024709184e-05, "loss": 0.0734, "step": 20667 }, { "epoch": 128.37383177570092, "grad_norm": 2.9375, "learning_rate": 1.4482575067831133e-05, "loss": 0.2284, "step": 20668 }, { "epoch": 128.38006230529595, "grad_norm": 1.375, "learning_rate": 1.448033999071775e-05, "loss": 0.1273, "step": 20669 }, { "epoch": 128.38629283489095, "grad_norm": 2.203125, "learning_rate": 1.4478105015773408e-05, "loss": 0.1821, "step": 20670 }, { "epoch": 128.39252336448598, "grad_norm": 1.4140625, "learning_rate": 1.4475870143019824e-05, "loss": 0.1217, "step": 20671 }, { "epoch": 128.39875389408098, "grad_norm": 0.87109375, "learning_rate": 1.4473635372478692e-05, "loss": 0.084, "step": 20672 }, { "epoch": 128.404984423676, "grad_norm": 2.25, "learning_rate": 1.4471400704171706e-05, "loss": 0.2322, "step": 20673 }, { "epoch": 128.41121495327104, "grad_norm": 1.390625, "learning_rate": 1.4469166138120583e-05, "loss": 0.1073, "step": 20674 }, { "epoch": 128.41744548286604, "grad_norm": 2.9375, "learning_rate": 1.4466931674347022e-05, "loss": 0.2934, "step": 20675 }, { "epoch": 128.42367601246107, "grad_norm": 2.375, "learning_rate": 1.4464697312872738e-05, "loss": 0.2267, "step": 20676 }, { "epoch": 128.42990654205607, "grad_norm": 2.8125, "learning_rate": 1.4462463053719415e-05, "loss": 0.2521, "step": 20677 }, { "epoch": 128.4361370716511, "grad_norm": 2.953125, "learning_rate": 1.4460228896908754e-05, "loss": 0.2502, "step": 20678 }, { "epoch": 128.4423676012461, "grad_norm": 2.640625, "learning_rate": 1.4457994842462452e-05, "loss": 0.1789, "step": 20679 }, { "epoch": 128.44859813084113, "grad_norm": 1.5078125, "learning_rate": 1.4455760890402225e-05, "loss": 0.1423, "step": 20680 }, { "epoch": 128.45482866043614, "grad_norm": 1.8515625, "learning_rate": 1.4453527040749742e-05, "loss": 0.1355, "step": 20681 }, { "epoch": 128.46105919003116, "grad_norm": 2.15625, "learning_rate": 1.4451293293526716e-05, "loss": 0.1535, "step": 20682 }, { "epoch": 128.46728971962617, "grad_norm": 1.6640625, "learning_rate": 1.4449059648754848e-05, "loss": 0.1021, "step": 20683 }, { "epoch": 128.4735202492212, "grad_norm": 1.2421875, "learning_rate": 1.4446826106455819e-05, "loss": 0.1303, "step": 20684 }, { "epoch": 128.4797507788162, "grad_norm": 2.375, "learning_rate": 1.4444592666651317e-05, "loss": 0.1657, "step": 20685 }, { "epoch": 128.48598130841123, "grad_norm": 0.97265625, "learning_rate": 1.4442359329363036e-05, "loss": 0.0864, "step": 20686 }, { "epoch": 128.49221183800623, "grad_norm": 2.34375, "learning_rate": 1.4440126094612688e-05, "loss": 0.1214, "step": 20687 }, { "epoch": 128.49844236760126, "grad_norm": 3.03125, "learning_rate": 1.4437892962421934e-05, "loss": 0.3251, "step": 20688 }, { "epoch": 128.50467289719626, "grad_norm": 2.984375, "learning_rate": 1.4435659932812473e-05, "loss": 0.2251, "step": 20689 }, { "epoch": 128.5109034267913, "grad_norm": 2.546875, "learning_rate": 1.4433427005806008e-05, "loss": 0.2653, "step": 20690 }, { "epoch": 128.5171339563863, "grad_norm": 2.0625, "learning_rate": 1.443119418142421e-05, "loss": 0.1572, "step": 20691 }, { "epoch": 128.52336448598132, "grad_norm": 1.265625, "learning_rate": 1.4428961459688755e-05, "loss": 0.1051, "step": 20692 }, { "epoch": 128.52959501557632, "grad_norm": 2.59375, "learning_rate": 1.4426728840621345e-05, "loss": 0.1789, "step": 20693 }, { "epoch": 128.53582554517135, "grad_norm": 0.9921875, "learning_rate": 1.4424496324243663e-05, "loss": 0.1093, "step": 20694 }, { "epoch": 128.54205607476635, "grad_norm": 1.34375, "learning_rate": 1.4422263910577377e-05, "loss": 0.1126, "step": 20695 }, { "epoch": 128.54828660436138, "grad_norm": 2.640625, "learning_rate": 1.442003159964419e-05, "loss": 0.2091, "step": 20696 }, { "epoch": 128.55451713395638, "grad_norm": 2.734375, "learning_rate": 1.4417799391465762e-05, "loss": 0.2292, "step": 20697 }, { "epoch": 128.5607476635514, "grad_norm": 2.21875, "learning_rate": 1.4415567286063788e-05, "loss": 0.2251, "step": 20698 }, { "epoch": 128.5669781931464, "grad_norm": 2.046875, "learning_rate": 1.4413335283459934e-05, "loss": 0.1452, "step": 20699 }, { "epoch": 128.57320872274144, "grad_norm": 2.171875, "learning_rate": 1.4411103383675884e-05, "loss": 0.149, "step": 20700 }, { "epoch": 128.57943925233644, "grad_norm": 2.515625, "learning_rate": 1.4408871586733318e-05, "loss": 0.2183, "step": 20701 }, { "epoch": 128.58566978193147, "grad_norm": 1.984375, "learning_rate": 1.4406639892653917e-05, "loss": 0.1283, "step": 20702 }, { "epoch": 128.59190031152647, "grad_norm": 1.8359375, "learning_rate": 1.4404408301459332e-05, "loss": 0.1081, "step": 20703 }, { "epoch": 128.5981308411215, "grad_norm": 1.4921875, "learning_rate": 1.4402176813171254e-05, "loss": 0.0909, "step": 20704 }, { "epoch": 128.6043613707165, "grad_norm": 1.1640625, "learning_rate": 1.4399945427811362e-05, "loss": 0.1215, "step": 20705 }, { "epoch": 128.61059190031153, "grad_norm": 1.484375, "learning_rate": 1.439771414540131e-05, "loss": 0.1342, "step": 20706 }, { "epoch": 128.61682242990653, "grad_norm": 2.234375, "learning_rate": 1.4395482965962775e-05, "loss": 0.091, "step": 20707 }, { "epoch": 128.62305295950156, "grad_norm": 1.984375, "learning_rate": 1.439325188951743e-05, "loss": 0.1298, "step": 20708 }, { "epoch": 128.62928348909657, "grad_norm": 1.6171875, "learning_rate": 1.4391020916086967e-05, "loss": 0.1524, "step": 20709 }, { "epoch": 128.6355140186916, "grad_norm": 1.5, "learning_rate": 1.4388790045693004e-05, "loss": 0.1664, "step": 20710 }, { "epoch": 128.6417445482866, "grad_norm": 2.390625, "learning_rate": 1.4386559278357237e-05, "loss": 0.1153, "step": 20711 }, { "epoch": 128.64797507788163, "grad_norm": 1.4609375, "learning_rate": 1.4384328614101325e-05, "loss": 0.1105, "step": 20712 }, { "epoch": 128.65420560747663, "grad_norm": 1.90625, "learning_rate": 1.4382098052946946e-05, "loss": 0.0974, "step": 20713 }, { "epoch": 128.66043613707166, "grad_norm": 1.6328125, "learning_rate": 1.4379867594915746e-05, "loss": 0.1172, "step": 20714 }, { "epoch": 128.66666666666666, "grad_norm": 2.84375, "learning_rate": 1.4377637240029402e-05, "loss": 0.2768, "step": 20715 }, { "epoch": 128.6728971962617, "grad_norm": 2.0, "learning_rate": 1.4375406988309558e-05, "loss": 0.125, "step": 20716 }, { "epoch": 128.6791277258567, "grad_norm": 1.4296875, "learning_rate": 1.4373176839777896e-05, "loss": 0.1428, "step": 20717 }, { "epoch": 128.68535825545172, "grad_norm": 3.140625, "learning_rate": 1.4370946794456058e-05, "loss": 0.2302, "step": 20718 }, { "epoch": 128.69158878504672, "grad_norm": 2.125, "learning_rate": 1.4368716852365705e-05, "loss": 0.2269, "step": 20719 }, { "epoch": 128.69781931464175, "grad_norm": 2.375, "learning_rate": 1.4366487013528512e-05, "loss": 0.1879, "step": 20720 }, { "epoch": 128.70404984423675, "grad_norm": 2.546875, "learning_rate": 1.436425727796612e-05, "loss": 0.14, "step": 20721 }, { "epoch": 128.71028037383178, "grad_norm": 3.171875, "learning_rate": 1.4362027645700183e-05, "loss": 0.1754, "step": 20722 }, { "epoch": 128.71651090342678, "grad_norm": 1.6015625, "learning_rate": 1.4359798116752354e-05, "loss": 0.1253, "step": 20723 }, { "epoch": 128.7227414330218, "grad_norm": 2.265625, "learning_rate": 1.4357568691144304e-05, "loss": 0.1338, "step": 20724 }, { "epoch": 128.7289719626168, "grad_norm": 1.5546875, "learning_rate": 1.4355339368897663e-05, "loss": 0.1278, "step": 20725 }, { "epoch": 128.73520249221184, "grad_norm": 1.9609375, "learning_rate": 1.43531101500341e-05, "loss": 0.1401, "step": 20726 }, { "epoch": 128.74143302180684, "grad_norm": 1.9453125, "learning_rate": 1.4350881034575265e-05, "loss": 0.1363, "step": 20727 }, { "epoch": 128.74766355140187, "grad_norm": 1.1015625, "learning_rate": 1.43486520225428e-05, "loss": 0.1109, "step": 20728 }, { "epoch": 128.75389408099687, "grad_norm": 3.046875, "learning_rate": 1.4346423113958349e-05, "loss": 0.2078, "step": 20729 }, { "epoch": 128.7601246105919, "grad_norm": 1.4609375, "learning_rate": 1.4344194308843565e-05, "loss": 0.1435, "step": 20730 }, { "epoch": 128.7663551401869, "grad_norm": 0.90625, "learning_rate": 1.4341965607220107e-05, "loss": 0.0971, "step": 20731 }, { "epoch": 128.77258566978193, "grad_norm": 1.890625, "learning_rate": 1.43397370091096e-05, "loss": 0.1504, "step": 20732 }, { "epoch": 128.77881619937693, "grad_norm": 1.7265625, "learning_rate": 1.4337508514533692e-05, "loss": 0.1245, "step": 20733 }, { "epoch": 128.78504672897196, "grad_norm": 3.140625, "learning_rate": 1.433528012351405e-05, "loss": 0.1773, "step": 20734 }, { "epoch": 128.79127725856696, "grad_norm": 2.46875, "learning_rate": 1.4333051836072298e-05, "loss": 0.1125, "step": 20735 }, { "epoch": 128.797507788162, "grad_norm": 2.578125, "learning_rate": 1.4330823652230067e-05, "loss": 0.271, "step": 20736 }, { "epoch": 128.80373831775702, "grad_norm": 2.40625, "learning_rate": 1.432859557200901e-05, "loss": 0.1149, "step": 20737 }, { "epoch": 128.80996884735202, "grad_norm": 1.5, "learning_rate": 1.4326367595430767e-05, "loss": 0.1375, "step": 20738 }, { "epoch": 128.81619937694705, "grad_norm": 3.03125, "learning_rate": 1.4324139722516982e-05, "loss": 0.1929, "step": 20739 }, { "epoch": 128.82242990654206, "grad_norm": 2.203125, "learning_rate": 1.432191195328928e-05, "loss": 0.1299, "step": 20740 }, { "epoch": 128.82866043613708, "grad_norm": 2.46875, "learning_rate": 1.4319684287769314e-05, "loss": 0.1656, "step": 20741 }, { "epoch": 128.8348909657321, "grad_norm": 2.1875, "learning_rate": 1.4317456725978706e-05, "loss": 0.09, "step": 20742 }, { "epoch": 128.84112149532712, "grad_norm": 2.875, "learning_rate": 1.4315229267939084e-05, "loss": 0.1982, "step": 20743 }, { "epoch": 128.84735202492212, "grad_norm": 2.078125, "learning_rate": 1.4313001913672091e-05, "loss": 0.1665, "step": 20744 }, { "epoch": 128.85358255451715, "grad_norm": 1.859375, "learning_rate": 1.4310774663199363e-05, "loss": 0.1422, "step": 20745 }, { "epoch": 128.85981308411215, "grad_norm": 1.890625, "learning_rate": 1.4308547516542537e-05, "loss": 0.1441, "step": 20746 }, { "epoch": 128.86604361370718, "grad_norm": 2.28125, "learning_rate": 1.4306320473723234e-05, "loss": 0.2478, "step": 20747 }, { "epoch": 128.87227414330218, "grad_norm": 2.21875, "learning_rate": 1.4304093534763074e-05, "loss": 0.1768, "step": 20748 }, { "epoch": 128.8785046728972, "grad_norm": 2.3125, "learning_rate": 1.4301866699683697e-05, "loss": 0.2135, "step": 20749 }, { "epoch": 128.8847352024922, "grad_norm": 2.046875, "learning_rate": 1.4299639968506736e-05, "loss": 0.1424, "step": 20750 }, { "epoch": 128.89096573208724, "grad_norm": 2.921875, "learning_rate": 1.4297413341253801e-05, "loss": 0.2471, "step": 20751 }, { "epoch": 128.89719626168224, "grad_norm": 2.609375, "learning_rate": 1.4295186817946532e-05, "loss": 0.183, "step": 20752 }, { "epoch": 128.90342679127727, "grad_norm": 1.578125, "learning_rate": 1.4292960398606553e-05, "loss": 0.1264, "step": 20753 }, { "epoch": 128.90965732087227, "grad_norm": 1.984375, "learning_rate": 1.4290734083255487e-05, "loss": 0.1647, "step": 20754 }, { "epoch": 128.9158878504673, "grad_norm": 1.34375, "learning_rate": 1.4288507871914938e-05, "loss": 0.1054, "step": 20755 }, { "epoch": 128.9221183800623, "grad_norm": 3.0625, "learning_rate": 1.4286281764606545e-05, "loss": 0.295, "step": 20756 }, { "epoch": 128.92834890965733, "grad_norm": 1.1875, "learning_rate": 1.4284055761351934e-05, "loss": 0.1065, "step": 20757 }, { "epoch": 128.93457943925233, "grad_norm": 0.88671875, "learning_rate": 1.4281829862172708e-05, "loss": 0.0931, "step": 20758 }, { "epoch": 128.94080996884736, "grad_norm": 1.484375, "learning_rate": 1.4279604067090489e-05, "loss": 0.124, "step": 20759 }, { "epoch": 128.94704049844236, "grad_norm": 1.5703125, "learning_rate": 1.427737837612691e-05, "loss": 0.1335, "step": 20760 }, { "epoch": 128.9532710280374, "grad_norm": 1.0703125, "learning_rate": 1.4275152789303575e-05, "loss": 0.105, "step": 20761 }, { "epoch": 128.9595015576324, "grad_norm": 2.328125, "learning_rate": 1.4272927306642089e-05, "loss": 0.1841, "step": 20762 }, { "epoch": 128.96573208722742, "grad_norm": 2.078125, "learning_rate": 1.4270701928164079e-05, "loss": 0.1513, "step": 20763 }, { "epoch": 128.97196261682242, "grad_norm": 1.3046875, "learning_rate": 1.4268476653891167e-05, "loss": 0.1296, "step": 20764 }, { "epoch": 128.97819314641745, "grad_norm": 3.5625, "learning_rate": 1.4266251483844945e-05, "loss": 0.2464, "step": 20765 }, { "epoch": 128.98442367601245, "grad_norm": 2.015625, "learning_rate": 1.4264026418047043e-05, "loss": 0.1691, "step": 20766 }, { "epoch": 128.99065420560748, "grad_norm": 0.91015625, "learning_rate": 1.4261801456519053e-05, "loss": 0.0967, "step": 20767 }, { "epoch": 128.99688473520249, "grad_norm": 2.90625, "learning_rate": 1.4259576599282604e-05, "loss": 0.1012, "step": 20768 }, { "epoch": 129.0, "grad_norm": 1.3125, "learning_rate": 1.4257351846359284e-05, "loss": 0.0781, "step": 20769 }, { "epoch": 129.00623052959503, "grad_norm": 1.9140625, "learning_rate": 1.425512719777071e-05, "loss": 0.1913, "step": 20770 }, { "epoch": 129.01246105919003, "grad_norm": 1.6875, "learning_rate": 1.4252902653538487e-05, "loss": 0.1533, "step": 20771 }, { "epoch": 129.01869158878506, "grad_norm": 1.609375, "learning_rate": 1.4250678213684243e-05, "loss": 0.1301, "step": 20772 }, { "epoch": 129.02492211838006, "grad_norm": 2.5, "learning_rate": 1.4248453878229542e-05, "loss": 0.1764, "step": 20773 }, { "epoch": 129.0311526479751, "grad_norm": 2.09375, "learning_rate": 1.4246229647196005e-05, "loss": 0.1366, "step": 20774 }, { "epoch": 129.0373831775701, "grad_norm": 1.453125, "learning_rate": 1.4244005520605236e-05, "loss": 0.1218, "step": 20775 }, { "epoch": 129.04361370716512, "grad_norm": 1.9921875, "learning_rate": 1.4241781498478845e-05, "loss": 0.1722, "step": 20776 }, { "epoch": 129.04984423676012, "grad_norm": 2.078125, "learning_rate": 1.4239557580838414e-05, "loss": 0.2094, "step": 20777 }, { "epoch": 129.05607476635515, "grad_norm": 2.34375, "learning_rate": 1.4237333767705546e-05, "loss": 0.1743, "step": 20778 }, { "epoch": 129.06230529595015, "grad_norm": 1.046875, "learning_rate": 1.4235110059101859e-05, "loss": 0.1029, "step": 20779 }, { "epoch": 129.06853582554518, "grad_norm": 1.5546875, "learning_rate": 1.4232886455048932e-05, "loss": 0.1442, "step": 20780 }, { "epoch": 129.07476635514018, "grad_norm": 3.0625, "learning_rate": 1.4230662955568358e-05, "loss": 0.1119, "step": 20781 }, { "epoch": 129.0809968847352, "grad_norm": 2.0625, "learning_rate": 1.4228439560681734e-05, "loss": 0.1076, "step": 20782 }, { "epoch": 129.08722741433021, "grad_norm": 1.8828125, "learning_rate": 1.4226216270410669e-05, "loss": 0.1049, "step": 20783 }, { "epoch": 129.09345794392524, "grad_norm": 1.0546875, "learning_rate": 1.4223993084776738e-05, "loss": 0.1034, "step": 20784 }, { "epoch": 129.09968847352025, "grad_norm": 1.7421875, "learning_rate": 1.422177000380155e-05, "loss": 0.1215, "step": 20785 }, { "epoch": 129.10591900311528, "grad_norm": 2.34375, "learning_rate": 1.4219547027506672e-05, "loss": 0.2234, "step": 20786 }, { "epoch": 129.11214953271028, "grad_norm": 2.875, "learning_rate": 1.421732415591372e-05, "loss": 0.242, "step": 20787 }, { "epoch": 129.1183800623053, "grad_norm": 2.0, "learning_rate": 1.421510138904426e-05, "loss": 0.1843, "step": 20788 }, { "epoch": 129.1246105919003, "grad_norm": 1.1796875, "learning_rate": 1.4212878726919895e-05, "loss": 0.1201, "step": 20789 }, { "epoch": 129.13084112149534, "grad_norm": 2.640625, "learning_rate": 1.4210656169562214e-05, "loss": 0.2903, "step": 20790 }, { "epoch": 129.13707165109034, "grad_norm": 2.59375, "learning_rate": 1.4208433716992803e-05, "loss": 0.1808, "step": 20791 }, { "epoch": 129.14330218068537, "grad_norm": 1.96875, "learning_rate": 1.4206211369233225e-05, "loss": 0.0955, "step": 20792 }, { "epoch": 129.14953271028037, "grad_norm": 3.734375, "learning_rate": 1.4203989126305084e-05, "loss": 0.2866, "step": 20793 }, { "epoch": 129.1557632398754, "grad_norm": 2.96875, "learning_rate": 1.4201766988229964e-05, "loss": 0.16, "step": 20794 }, { "epoch": 129.1619937694704, "grad_norm": 2.578125, "learning_rate": 1.4199544955029437e-05, "loss": 0.1001, "step": 20795 }, { "epoch": 129.16822429906543, "grad_norm": 1.1015625, "learning_rate": 1.4197323026725084e-05, "loss": 0.103, "step": 20796 }, { "epoch": 129.17445482866043, "grad_norm": 1.9296875, "learning_rate": 1.41951012033385e-05, "loss": 0.1718, "step": 20797 }, { "epoch": 129.18068535825546, "grad_norm": 2.5625, "learning_rate": 1.4192879484891253e-05, "loss": 0.1889, "step": 20798 }, { "epoch": 129.18691588785046, "grad_norm": 1.1171875, "learning_rate": 1.4190657871404911e-05, "loss": 0.0895, "step": 20799 }, { "epoch": 129.1931464174455, "grad_norm": 1.703125, "learning_rate": 1.4188436362901061e-05, "loss": 0.1611, "step": 20800 }, { "epoch": 129.1993769470405, "grad_norm": 2.640625, "learning_rate": 1.418621495940129e-05, "loss": 0.1661, "step": 20801 }, { "epoch": 129.20560747663552, "grad_norm": 2.09375, "learning_rate": 1.4183993660927147e-05, "loss": 0.1707, "step": 20802 }, { "epoch": 129.21183800623052, "grad_norm": 1.9453125, "learning_rate": 1.418177246750022e-05, "loss": 0.1459, "step": 20803 }, { "epoch": 129.21806853582555, "grad_norm": 2.140625, "learning_rate": 1.4179551379142095e-05, "loss": 0.1815, "step": 20804 }, { "epoch": 129.22429906542055, "grad_norm": 2.890625, "learning_rate": 1.4177330395874327e-05, "loss": 0.3077, "step": 20805 }, { "epoch": 129.23052959501558, "grad_norm": 2.421875, "learning_rate": 1.4175109517718483e-05, "loss": 0.2247, "step": 20806 }, { "epoch": 129.23676012461058, "grad_norm": 2.203125, "learning_rate": 1.4172888744696136e-05, "loss": 0.1833, "step": 20807 }, { "epoch": 129.2429906542056, "grad_norm": 2.421875, "learning_rate": 1.4170668076828859e-05, "loss": 0.1812, "step": 20808 }, { "epoch": 129.2492211838006, "grad_norm": 1.484375, "learning_rate": 1.4168447514138228e-05, "loss": 0.1279, "step": 20809 }, { "epoch": 129.25545171339564, "grad_norm": 2.265625, "learning_rate": 1.416622705664579e-05, "loss": 0.1341, "step": 20810 }, { "epoch": 129.26168224299064, "grad_norm": 1.03125, "learning_rate": 1.4164006704373129e-05, "loss": 0.1088, "step": 20811 }, { "epoch": 129.26791277258567, "grad_norm": 2.109375, "learning_rate": 1.4161786457341791e-05, "loss": 0.1586, "step": 20812 }, { "epoch": 129.27414330218068, "grad_norm": 2.296875, "learning_rate": 1.4159566315573364e-05, "loss": 0.2161, "step": 20813 }, { "epoch": 129.2803738317757, "grad_norm": 1.421875, "learning_rate": 1.4157346279089382e-05, "loss": 0.135, "step": 20814 }, { "epoch": 129.2866043613707, "grad_norm": 1.15625, "learning_rate": 1.4155126347911418e-05, "loss": 0.1137, "step": 20815 }, { "epoch": 129.29283489096574, "grad_norm": 2.28125, "learning_rate": 1.4152906522061048e-05, "loss": 0.1865, "step": 20816 }, { "epoch": 129.29906542056074, "grad_norm": 1.1015625, "learning_rate": 1.415068680155982e-05, "loss": 0.1027, "step": 20817 }, { "epoch": 129.30529595015577, "grad_norm": 2.84375, "learning_rate": 1.4148467186429277e-05, "loss": 0.256, "step": 20818 }, { "epoch": 129.31152647975077, "grad_norm": 2.5, "learning_rate": 1.4146247676690993e-05, "loss": 0.1865, "step": 20819 }, { "epoch": 129.3177570093458, "grad_norm": 1.3671875, "learning_rate": 1.4144028272366528e-05, "loss": 0.1264, "step": 20820 }, { "epoch": 129.3239875389408, "grad_norm": 3.9375, "learning_rate": 1.4141808973477422e-05, "loss": 0.3497, "step": 20821 }, { "epoch": 129.33021806853583, "grad_norm": 2.015625, "learning_rate": 1.4139589780045237e-05, "loss": 0.1394, "step": 20822 }, { "epoch": 129.33644859813083, "grad_norm": 2.453125, "learning_rate": 1.413737069209154e-05, "loss": 0.1179, "step": 20823 }, { "epoch": 129.34267912772586, "grad_norm": 1.6875, "learning_rate": 1.4135151709637866e-05, "loss": 0.1697, "step": 20824 }, { "epoch": 129.34890965732086, "grad_norm": 3.96875, "learning_rate": 1.4132932832705763e-05, "loss": 0.3998, "step": 20825 }, { "epoch": 129.3551401869159, "grad_norm": 2.578125, "learning_rate": 1.4130714061316788e-05, "loss": 0.2396, "step": 20826 }, { "epoch": 129.3613707165109, "grad_norm": 1.375, "learning_rate": 1.4128495395492502e-05, "loss": 0.1079, "step": 20827 }, { "epoch": 129.36760124610592, "grad_norm": 1.640625, "learning_rate": 1.4126276835254429e-05, "loss": 0.0895, "step": 20828 }, { "epoch": 129.37383177570092, "grad_norm": 2.390625, "learning_rate": 1.412405838062413e-05, "loss": 0.2158, "step": 20829 }, { "epoch": 129.38006230529595, "grad_norm": 2.453125, "learning_rate": 1.4121840031623162e-05, "loss": 0.1749, "step": 20830 }, { "epoch": 129.38629283489095, "grad_norm": 0.9375, "learning_rate": 1.4119621788273057e-05, "loss": 0.1019, "step": 20831 }, { "epoch": 129.39252336448598, "grad_norm": 1.7421875, "learning_rate": 1.411740365059535e-05, "loss": 0.1152, "step": 20832 }, { "epoch": 129.39875389408098, "grad_norm": 2.015625, "learning_rate": 1.4115185618611593e-05, "loss": 0.1179, "step": 20833 }, { "epoch": 129.404984423676, "grad_norm": 2.578125, "learning_rate": 1.4112967692343332e-05, "loss": 0.1675, "step": 20834 }, { "epoch": 129.41121495327104, "grad_norm": 1.1953125, "learning_rate": 1.4110749871812113e-05, "loss": 0.1112, "step": 20835 }, { "epoch": 129.41744548286604, "grad_norm": 3.0625, "learning_rate": 1.4108532157039468e-05, "loss": 0.1988, "step": 20836 }, { "epoch": 129.42367601246107, "grad_norm": 2.125, "learning_rate": 1.4106314548046923e-05, "loss": 0.1014, "step": 20837 }, { "epoch": 129.42990654205607, "grad_norm": 2.890625, "learning_rate": 1.4104097044856044e-05, "loss": 0.1603, "step": 20838 }, { "epoch": 129.4361370716511, "grad_norm": 1.9140625, "learning_rate": 1.4101879647488342e-05, "loss": 0.1718, "step": 20839 }, { "epoch": 129.4423676012461, "grad_norm": 1.8984375, "learning_rate": 1.409966235596536e-05, "loss": 0.1288, "step": 20840 }, { "epoch": 129.44859813084113, "grad_norm": 1.375, "learning_rate": 1.4097445170308637e-05, "loss": 0.1066, "step": 20841 }, { "epoch": 129.45482866043614, "grad_norm": 1.9296875, "learning_rate": 1.4095228090539725e-05, "loss": 0.1317, "step": 20842 }, { "epoch": 129.46105919003116, "grad_norm": 1.6796875, "learning_rate": 1.4093011116680122e-05, "loss": 0.1523, "step": 20843 }, { "epoch": 129.46728971962617, "grad_norm": 1.359375, "learning_rate": 1.4090794248751372e-05, "loss": 0.1585, "step": 20844 }, { "epoch": 129.4735202492212, "grad_norm": 2.578125, "learning_rate": 1.408857748677501e-05, "loss": 0.1281, "step": 20845 }, { "epoch": 129.4797507788162, "grad_norm": 2.125, "learning_rate": 1.4086360830772571e-05, "loss": 0.1313, "step": 20846 }, { "epoch": 129.48598130841123, "grad_norm": 2.546875, "learning_rate": 1.408414428076557e-05, "loss": 0.11, "step": 20847 }, { "epoch": 129.49221183800623, "grad_norm": 1.2421875, "learning_rate": 1.4081927836775543e-05, "loss": 0.1397, "step": 20848 }, { "epoch": 129.49844236760126, "grad_norm": 2.453125, "learning_rate": 1.4079711498824027e-05, "loss": 0.2501, "step": 20849 }, { "epoch": 129.50467289719626, "grad_norm": 2.28125, "learning_rate": 1.4077495266932533e-05, "loss": 0.257, "step": 20850 }, { "epoch": 129.5109034267913, "grad_norm": 2.125, "learning_rate": 1.4075279141122577e-05, "loss": 0.1693, "step": 20851 }, { "epoch": 129.5171339563863, "grad_norm": 1.6015625, "learning_rate": 1.4073063121415698e-05, "loss": 0.155, "step": 20852 }, { "epoch": 129.52336448598132, "grad_norm": 2.59375, "learning_rate": 1.4070847207833421e-05, "loss": 0.4272, "step": 20853 }, { "epoch": 129.52959501557632, "grad_norm": 0.984375, "learning_rate": 1.406863140039725e-05, "loss": 0.0939, "step": 20854 }, { "epoch": 129.53582554517135, "grad_norm": 2.046875, "learning_rate": 1.4066415699128727e-05, "loss": 0.1874, "step": 20855 }, { "epoch": 129.54205607476635, "grad_norm": 1.640625, "learning_rate": 1.4064200104049353e-05, "loss": 0.1137, "step": 20856 }, { "epoch": 129.54828660436138, "grad_norm": 1.5546875, "learning_rate": 1.406198461518066e-05, "loss": 0.1164, "step": 20857 }, { "epoch": 129.55451713395638, "grad_norm": 1.796875, "learning_rate": 1.4059769232544152e-05, "loss": 0.157, "step": 20858 }, { "epoch": 129.5607476635514, "grad_norm": 2.40625, "learning_rate": 1.4057553956161346e-05, "loss": 0.1903, "step": 20859 }, { "epoch": 129.5669781931464, "grad_norm": 1.515625, "learning_rate": 1.405533878605378e-05, "loss": 0.1292, "step": 20860 }, { "epoch": 129.57320872274144, "grad_norm": 2.046875, "learning_rate": 1.405312372224294e-05, "loss": 0.1699, "step": 20861 }, { "epoch": 129.57943925233644, "grad_norm": 1.3203125, "learning_rate": 1.4050908764750359e-05, "loss": 0.1003, "step": 20862 }, { "epoch": 129.58566978193147, "grad_norm": 1.25, "learning_rate": 1.404869391359753e-05, "loss": 0.1071, "step": 20863 }, { "epoch": 129.59190031152647, "grad_norm": 1.1484375, "learning_rate": 1.4046479168805987e-05, "loss": 0.1048, "step": 20864 }, { "epoch": 129.5981308411215, "grad_norm": 2.15625, "learning_rate": 1.4044264530397217e-05, "loss": 0.1524, "step": 20865 }, { "epoch": 129.6043613707165, "grad_norm": 1.015625, "learning_rate": 1.404204999839274e-05, "loss": 0.1044, "step": 20866 }, { "epoch": 129.61059190031153, "grad_norm": 2.859375, "learning_rate": 1.4039835572814064e-05, "loss": 0.3548, "step": 20867 }, { "epoch": 129.61682242990653, "grad_norm": 2.34375, "learning_rate": 1.4037621253682715e-05, "loss": 0.1544, "step": 20868 }, { "epoch": 129.62305295950156, "grad_norm": 2.359375, "learning_rate": 1.403540704102016e-05, "loss": 0.1848, "step": 20869 }, { "epoch": 129.62928348909657, "grad_norm": 1.0390625, "learning_rate": 1.4033192934847922e-05, "loss": 0.1148, "step": 20870 }, { "epoch": 129.6355140186916, "grad_norm": 2.390625, "learning_rate": 1.4030978935187511e-05, "loss": 0.1958, "step": 20871 }, { "epoch": 129.6417445482866, "grad_norm": 0.90234375, "learning_rate": 1.4028765042060432e-05, "loss": 0.101, "step": 20872 }, { "epoch": 129.64797507788163, "grad_norm": 3.015625, "learning_rate": 1.402655125548817e-05, "loss": 0.1993, "step": 20873 }, { "epoch": 129.65420560747663, "grad_norm": 1.1875, "learning_rate": 1.4024337575492247e-05, "loss": 0.1019, "step": 20874 }, { "epoch": 129.66043613707166, "grad_norm": 2.84375, "learning_rate": 1.402212400209414e-05, "loss": 0.2097, "step": 20875 }, { "epoch": 129.66666666666666, "grad_norm": 1.765625, "learning_rate": 1.4019910535315373e-05, "loss": 0.1352, "step": 20876 }, { "epoch": 129.6728971962617, "grad_norm": 3.828125, "learning_rate": 1.4017697175177413e-05, "loss": 0.3208, "step": 20877 }, { "epoch": 129.6791277258567, "grad_norm": 1.6875, "learning_rate": 1.4015483921701778e-05, "loss": 0.1305, "step": 20878 }, { "epoch": 129.68535825545172, "grad_norm": 2.171875, "learning_rate": 1.401327077490997e-05, "loss": 0.1619, "step": 20879 }, { "epoch": 129.69158878504672, "grad_norm": 1.375, "learning_rate": 1.4011057734823458e-05, "loss": 0.0973, "step": 20880 }, { "epoch": 129.69781931464175, "grad_norm": 1.90625, "learning_rate": 1.4008844801463763e-05, "loss": 0.151, "step": 20881 }, { "epoch": 129.70404984423675, "grad_norm": 1.3203125, "learning_rate": 1.4006631974852352e-05, "loss": 0.1222, "step": 20882 }, { "epoch": 129.71028037383178, "grad_norm": 1.3046875, "learning_rate": 1.4004419255010742e-05, "loss": 0.0988, "step": 20883 }, { "epoch": 129.71651090342678, "grad_norm": 1.890625, "learning_rate": 1.40022066419604e-05, "loss": 0.1389, "step": 20884 }, { "epoch": 129.7227414330218, "grad_norm": 1.6015625, "learning_rate": 1.3999994135722822e-05, "loss": 0.1288, "step": 20885 }, { "epoch": 129.7289719626168, "grad_norm": 1.859375, "learning_rate": 1.3997781736319515e-05, "loss": 0.1682, "step": 20886 }, { "epoch": 129.73520249221184, "grad_norm": 1.796875, "learning_rate": 1.3995569443771947e-05, "loss": 0.152, "step": 20887 }, { "epoch": 129.74143302180684, "grad_norm": 2.15625, "learning_rate": 1.3993357258101598e-05, "loss": 0.1639, "step": 20888 }, { "epoch": 129.74766355140187, "grad_norm": 3.546875, "learning_rate": 1.3991145179329967e-05, "loss": 0.2108, "step": 20889 }, { "epoch": 129.75389408099687, "grad_norm": 1.9140625, "learning_rate": 1.3988933207478541e-05, "loss": 0.1094, "step": 20890 }, { "epoch": 129.7601246105919, "grad_norm": 1.6171875, "learning_rate": 1.3986721342568788e-05, "loss": 0.1302, "step": 20891 }, { "epoch": 129.7663551401869, "grad_norm": 1.8046875, "learning_rate": 1.3984509584622202e-05, "loss": 0.2295, "step": 20892 }, { "epoch": 129.77258566978193, "grad_norm": 1.921875, "learning_rate": 1.3982297933660266e-05, "loss": 0.2135, "step": 20893 }, { "epoch": 129.77881619937693, "grad_norm": 2.078125, "learning_rate": 1.3980086389704455e-05, "loss": 0.1616, "step": 20894 }, { "epoch": 129.78504672897196, "grad_norm": 2.421875, "learning_rate": 1.3977874952776243e-05, "loss": 0.1686, "step": 20895 }, { "epoch": 129.79127725856696, "grad_norm": 1.46875, "learning_rate": 1.3975663622897106e-05, "loss": 0.1303, "step": 20896 }, { "epoch": 129.797507788162, "grad_norm": 1.1015625, "learning_rate": 1.3973452400088543e-05, "loss": 0.0959, "step": 20897 }, { "epoch": 129.80373831775702, "grad_norm": 2.34375, "learning_rate": 1.3971241284372003e-05, "loss": 0.1858, "step": 20898 }, { "epoch": 129.80996884735202, "grad_norm": 2.65625, "learning_rate": 1.396903027576897e-05, "loss": 0.1858, "step": 20899 }, { "epoch": 129.81619937694705, "grad_norm": 1.7578125, "learning_rate": 1.3966819374300932e-05, "loss": 0.1053, "step": 20900 }, { "epoch": 129.82242990654206, "grad_norm": 1.96875, "learning_rate": 1.3964608579989351e-05, "loss": 0.1496, "step": 20901 }, { "epoch": 129.82866043613708, "grad_norm": 2.828125, "learning_rate": 1.3962397892855685e-05, "loss": 0.2005, "step": 20902 }, { "epoch": 129.8348909657321, "grad_norm": 2.671875, "learning_rate": 1.396018731292142e-05, "loss": 0.2384, "step": 20903 }, { "epoch": 129.84112149532712, "grad_norm": 1.71875, "learning_rate": 1.395797684020802e-05, "loss": 0.1015, "step": 20904 }, { "epoch": 129.84735202492212, "grad_norm": 3.6875, "learning_rate": 1.3955766474736964e-05, "loss": 0.1404, "step": 20905 }, { "epoch": 129.85358255451715, "grad_norm": 1.359375, "learning_rate": 1.3953556216529717e-05, "loss": 0.1153, "step": 20906 }, { "epoch": 129.85981308411215, "grad_norm": 3.203125, "learning_rate": 1.3951346065607725e-05, "loss": 0.2447, "step": 20907 }, { "epoch": 129.86604361370718, "grad_norm": 1.9375, "learning_rate": 1.3949136021992474e-05, "loss": 0.1106, "step": 20908 }, { "epoch": 129.87227414330218, "grad_norm": 1.40625, "learning_rate": 1.3946926085705431e-05, "loss": 0.1148, "step": 20909 }, { "epoch": 129.8785046728972, "grad_norm": 2.359375, "learning_rate": 1.3944716256768037e-05, "loss": 0.1748, "step": 20910 }, { "epoch": 129.8847352024922, "grad_norm": 1.1953125, "learning_rate": 1.3942506535201771e-05, "loss": 0.0973, "step": 20911 }, { "epoch": 129.89096573208724, "grad_norm": 2.1875, "learning_rate": 1.3940296921028106e-05, "loss": 0.1337, "step": 20912 }, { "epoch": 129.89719626168224, "grad_norm": 1.890625, "learning_rate": 1.3938087414268486e-05, "loss": 0.1903, "step": 20913 }, { "epoch": 129.90342679127727, "grad_norm": 1.7421875, "learning_rate": 1.393587801494436e-05, "loss": 0.1555, "step": 20914 }, { "epoch": 129.90965732087227, "grad_norm": 1.734375, "learning_rate": 1.3933668723077198e-05, "loss": 0.1548, "step": 20915 }, { "epoch": 129.9158878504673, "grad_norm": 2.515625, "learning_rate": 1.3931459538688472e-05, "loss": 0.2245, "step": 20916 }, { "epoch": 129.9221183800623, "grad_norm": 1.8359375, "learning_rate": 1.3929250461799609e-05, "loss": 0.1379, "step": 20917 }, { "epoch": 129.92834890965733, "grad_norm": 1.515625, "learning_rate": 1.392704149243208e-05, "loss": 0.1028, "step": 20918 }, { "epoch": 129.93457943925233, "grad_norm": 1.171875, "learning_rate": 1.3924832630607349e-05, "loss": 0.1007, "step": 20919 }, { "epoch": 129.94080996884736, "grad_norm": 2.625, "learning_rate": 1.3922623876346858e-05, "loss": 0.2143, "step": 20920 }, { "epoch": 129.94704049844236, "grad_norm": 2.71875, "learning_rate": 1.3920415229672046e-05, "loss": 0.1824, "step": 20921 }, { "epoch": 129.9532710280374, "grad_norm": 1.3828125, "learning_rate": 1.3918206690604374e-05, "loss": 0.1095, "step": 20922 }, { "epoch": 129.9595015576324, "grad_norm": 1.46875, "learning_rate": 1.3915998259165309e-05, "loss": 0.1064, "step": 20923 }, { "epoch": 129.96573208722742, "grad_norm": 4.21875, "learning_rate": 1.391378993537627e-05, "loss": 0.3128, "step": 20924 }, { "epoch": 129.97196261682242, "grad_norm": 3.0, "learning_rate": 1.3911581719258732e-05, "loss": 0.2563, "step": 20925 }, { "epoch": 129.97819314641745, "grad_norm": 2.453125, "learning_rate": 1.3909373610834114e-05, "loss": 0.2114, "step": 20926 }, { "epoch": 129.98442367601245, "grad_norm": 2.015625, "learning_rate": 1.3907165610123892e-05, "loss": 0.2713, "step": 20927 }, { "epoch": 129.99065420560748, "grad_norm": 3.296875, "learning_rate": 1.3904957717149483e-05, "loss": 0.2216, "step": 20928 }, { "epoch": 129.99688473520249, "grad_norm": 0.9765625, "learning_rate": 1.3902749931932344e-05, "loss": 0.1111, "step": 20929 }, { "epoch": 130.0, "grad_norm": 2.25, "learning_rate": 1.3900542254493915e-05, "loss": 0.1537, "step": 20930 }, { "epoch": 130.00623052959503, "grad_norm": 1.4375, "learning_rate": 1.3898334684855647e-05, "loss": 0.1353, "step": 20931 }, { "epoch": 130.01246105919003, "grad_norm": 2.375, "learning_rate": 1.3896127223038974e-05, "loss": 0.1946, "step": 20932 }, { "epoch": 130.01869158878506, "grad_norm": 2.421875, "learning_rate": 1.3893919869065319e-05, "loss": 0.1336, "step": 20933 }, { "epoch": 130.02492211838006, "grad_norm": 1.9375, "learning_rate": 1.3891712622956148e-05, "loss": 0.1631, "step": 20934 }, { "epoch": 130.0311526479751, "grad_norm": 2.1875, "learning_rate": 1.3889505484732873e-05, "loss": 0.1416, "step": 20935 }, { "epoch": 130.0373831775701, "grad_norm": 2.234375, "learning_rate": 1.3887298454416945e-05, "loss": 0.1542, "step": 20936 }, { "epoch": 130.04361370716512, "grad_norm": 2.140625, "learning_rate": 1.3885091532029792e-05, "loss": 0.186, "step": 20937 }, { "epoch": 130.04984423676012, "grad_norm": 2.296875, "learning_rate": 1.3882884717592876e-05, "loss": 0.1919, "step": 20938 }, { "epoch": 130.05607476635515, "grad_norm": 2.6875, "learning_rate": 1.3880678011127579e-05, "loss": 0.1952, "step": 20939 }, { "epoch": 130.06230529595015, "grad_norm": 2.203125, "learning_rate": 1.3878471412655363e-05, "loss": 0.2102, "step": 20940 }, { "epoch": 130.06853582554518, "grad_norm": 1.6328125, "learning_rate": 1.3876264922197658e-05, "loss": 0.1481, "step": 20941 }, { "epoch": 130.07476635514018, "grad_norm": 1.328125, "learning_rate": 1.3874058539775903e-05, "loss": 0.1071, "step": 20942 }, { "epoch": 130.0809968847352, "grad_norm": 2.578125, "learning_rate": 1.38718522654115e-05, "loss": 0.2684, "step": 20943 }, { "epoch": 130.08722741433021, "grad_norm": 2.46875, "learning_rate": 1.3869646099125905e-05, "loss": 0.165, "step": 20944 }, { "epoch": 130.09345794392524, "grad_norm": 1.734375, "learning_rate": 1.386744004094052e-05, "loss": 0.12, "step": 20945 }, { "epoch": 130.09968847352025, "grad_norm": 3.4375, "learning_rate": 1.38652340908768e-05, "loss": 0.3044, "step": 20946 }, { "epoch": 130.10591900311528, "grad_norm": 1.8671875, "learning_rate": 1.3863028248956133e-05, "loss": 0.1563, "step": 20947 }, { "epoch": 130.11214953271028, "grad_norm": 1.734375, "learning_rate": 1.3860822515199967e-05, "loss": 0.1095, "step": 20948 }, { "epoch": 130.1183800623053, "grad_norm": 1.5078125, "learning_rate": 1.3858616889629728e-05, "loss": 0.1034, "step": 20949 }, { "epoch": 130.1246105919003, "grad_norm": 2.21875, "learning_rate": 1.385641137226682e-05, "loss": 0.1555, "step": 20950 }, { "epoch": 130.13084112149534, "grad_norm": 1.3203125, "learning_rate": 1.3854205963132683e-05, "loss": 0.1311, "step": 20951 }, { "epoch": 130.13707165109034, "grad_norm": 2.046875, "learning_rate": 1.3852000662248712e-05, "loss": 0.1411, "step": 20952 }, { "epoch": 130.14330218068537, "grad_norm": 2.09375, "learning_rate": 1.384979546963635e-05, "loss": 0.1645, "step": 20953 }, { "epoch": 130.14953271028037, "grad_norm": 2.234375, "learning_rate": 1.3847590385316997e-05, "loss": 0.1594, "step": 20954 }, { "epoch": 130.1557632398754, "grad_norm": 2.28125, "learning_rate": 1.3845385409312074e-05, "loss": 0.1529, "step": 20955 }, { "epoch": 130.1619937694704, "grad_norm": 2.984375, "learning_rate": 1.3843180541643006e-05, "loss": 0.205, "step": 20956 }, { "epoch": 130.16822429906543, "grad_norm": 2.96875, "learning_rate": 1.38409757823312e-05, "loss": 0.2441, "step": 20957 }, { "epoch": 130.17445482866043, "grad_norm": 2.375, "learning_rate": 1.3838771131398054e-05, "loss": 0.2347, "step": 20958 }, { "epoch": 130.18068535825546, "grad_norm": 2.1875, "learning_rate": 1.3836566588864999e-05, "loss": 0.1158, "step": 20959 }, { "epoch": 130.18691588785046, "grad_norm": 3.671875, "learning_rate": 1.383436215475345e-05, "loss": 0.1836, "step": 20960 }, { "epoch": 130.1931464174455, "grad_norm": 2.5, "learning_rate": 1.3832157829084797e-05, "loss": 0.1368, "step": 20961 }, { "epoch": 130.1993769470405, "grad_norm": 2.25, "learning_rate": 1.3829953611880458e-05, "loss": 0.0932, "step": 20962 }, { "epoch": 130.20560747663552, "grad_norm": 3.234375, "learning_rate": 1.3827749503161852e-05, "loss": 0.2906, "step": 20963 }, { "epoch": 130.21183800623052, "grad_norm": 1.4296875, "learning_rate": 1.382554550295038e-05, "loss": 0.1068, "step": 20964 }, { "epoch": 130.21806853582555, "grad_norm": 2.09375, "learning_rate": 1.3823341611267433e-05, "loss": 0.1762, "step": 20965 }, { "epoch": 130.22429906542055, "grad_norm": 2.578125, "learning_rate": 1.3821137828134423e-05, "loss": 0.1661, "step": 20966 }, { "epoch": 130.23052959501558, "grad_norm": 2.96875, "learning_rate": 1.381893415357276e-05, "loss": 0.1943, "step": 20967 }, { "epoch": 130.23676012461058, "grad_norm": 1.3203125, "learning_rate": 1.3816730587603848e-05, "loss": 0.0873, "step": 20968 }, { "epoch": 130.2429906542056, "grad_norm": 1.8671875, "learning_rate": 1.3814527130249083e-05, "loss": 0.1497, "step": 20969 }, { "epoch": 130.2492211838006, "grad_norm": 1.796875, "learning_rate": 1.3812323781529871e-05, "loss": 0.1352, "step": 20970 }, { "epoch": 130.25545171339564, "grad_norm": 1.2734375, "learning_rate": 1.3810120541467597e-05, "loss": 0.0906, "step": 20971 }, { "epoch": 130.26168224299064, "grad_norm": 1.8359375, "learning_rate": 1.3807917410083682e-05, "loss": 0.1317, "step": 20972 }, { "epoch": 130.26791277258567, "grad_norm": 2.046875, "learning_rate": 1.3805714387399496e-05, "loss": 0.1292, "step": 20973 }, { "epoch": 130.27414330218068, "grad_norm": 1.4921875, "learning_rate": 1.3803511473436453e-05, "loss": 0.1084, "step": 20974 }, { "epoch": 130.2803738317757, "grad_norm": 1.015625, "learning_rate": 1.380130866821595e-05, "loss": 0.0854, "step": 20975 }, { "epoch": 130.2866043613707, "grad_norm": 1.6171875, "learning_rate": 1.3799105971759382e-05, "loss": 0.1225, "step": 20976 }, { "epoch": 130.29283489096574, "grad_norm": 2.96875, "learning_rate": 1.3796903384088123e-05, "loss": 0.2235, "step": 20977 }, { "epoch": 130.29906542056074, "grad_norm": 1.875, "learning_rate": 1.3794700905223579e-05, "loss": 0.1479, "step": 20978 }, { "epoch": 130.30529595015577, "grad_norm": 2.5, "learning_rate": 1.3792498535187148e-05, "loss": 0.1754, "step": 20979 }, { "epoch": 130.31152647975077, "grad_norm": 3.0, "learning_rate": 1.3790296274000203e-05, "loss": 0.1886, "step": 20980 }, { "epoch": 130.3177570093458, "grad_norm": 2.3125, "learning_rate": 1.378809412168414e-05, "loss": 0.1735, "step": 20981 }, { "epoch": 130.3239875389408, "grad_norm": 2.25, "learning_rate": 1.378589207826036e-05, "loss": 0.2222, "step": 20982 }, { "epoch": 130.33021806853583, "grad_norm": 2.203125, "learning_rate": 1.3783690143750233e-05, "loss": 0.2078, "step": 20983 }, { "epoch": 130.33644859813083, "grad_norm": 3.03125, "learning_rate": 1.3781488318175146e-05, "loss": 0.1441, "step": 20984 }, { "epoch": 130.34267912772586, "grad_norm": 2.609375, "learning_rate": 1.3779286601556482e-05, "loss": 0.1998, "step": 20985 }, { "epoch": 130.34890965732086, "grad_norm": 1.65625, "learning_rate": 1.377708499391564e-05, "loss": 0.12, "step": 20986 }, { "epoch": 130.3551401869159, "grad_norm": 2.046875, "learning_rate": 1.3774883495273985e-05, "loss": 0.1531, "step": 20987 }, { "epoch": 130.3613707165109, "grad_norm": 2.375, "learning_rate": 1.3772682105652903e-05, "loss": 0.1785, "step": 20988 }, { "epoch": 130.36760124610592, "grad_norm": 1.6015625, "learning_rate": 1.377048082507379e-05, "loss": 0.1159, "step": 20989 }, { "epoch": 130.37383177570092, "grad_norm": 1.9609375, "learning_rate": 1.3768279653558009e-05, "loss": 0.1497, "step": 20990 }, { "epoch": 130.38006230529595, "grad_norm": 1.328125, "learning_rate": 1.376607859112693e-05, "loss": 0.1319, "step": 20991 }, { "epoch": 130.38629283489095, "grad_norm": 3.015625, "learning_rate": 1.3763877637801943e-05, "loss": 0.2615, "step": 20992 }, { "epoch": 130.39252336448598, "grad_norm": 1.84375, "learning_rate": 1.3761676793604433e-05, "loss": 0.1267, "step": 20993 }, { "epoch": 130.39875389408098, "grad_norm": 1.671875, "learning_rate": 1.3759476058555754e-05, "loss": 0.0944, "step": 20994 }, { "epoch": 130.404984423676, "grad_norm": 2.859375, "learning_rate": 1.3757275432677299e-05, "loss": 0.3696, "step": 20995 }, { "epoch": 130.41121495327104, "grad_norm": 2.34375, "learning_rate": 1.3755074915990423e-05, "loss": 0.177, "step": 20996 }, { "epoch": 130.41744548286604, "grad_norm": 2.703125, "learning_rate": 1.3752874508516517e-05, "loss": 0.149, "step": 20997 }, { "epoch": 130.42367601246107, "grad_norm": 1.828125, "learning_rate": 1.3750674210276935e-05, "loss": 0.1583, "step": 20998 }, { "epoch": 130.42990654205607, "grad_norm": 1.84375, "learning_rate": 1.374847402129305e-05, "loss": 0.1424, "step": 20999 }, { "epoch": 130.4361370716511, "grad_norm": 1.859375, "learning_rate": 1.3746273941586236e-05, "loss": 0.1458, "step": 21000 }, { "epoch": 130.4423676012461, "grad_norm": 2.21875, "learning_rate": 1.3744073971177871e-05, "loss": 0.1574, "step": 21001 }, { "epoch": 130.44859813084113, "grad_norm": 2.46875, "learning_rate": 1.3741874110089306e-05, "loss": 0.1321, "step": 21002 }, { "epoch": 130.45482866043614, "grad_norm": 1.3359375, "learning_rate": 1.3739674358341903e-05, "loss": 0.1012, "step": 21003 }, { "epoch": 130.46105919003116, "grad_norm": 2.4375, "learning_rate": 1.3737474715957032e-05, "loss": 0.1347, "step": 21004 }, { "epoch": 130.46728971962617, "grad_norm": 3.46875, "learning_rate": 1.373527518295607e-05, "loss": 0.1388, "step": 21005 }, { "epoch": 130.4735202492212, "grad_norm": 2.015625, "learning_rate": 1.3733075759360354e-05, "loss": 0.1738, "step": 21006 }, { "epoch": 130.4797507788162, "grad_norm": 2.3125, "learning_rate": 1.373087644519126e-05, "loss": 0.2132, "step": 21007 }, { "epoch": 130.48598130841123, "grad_norm": 1.6640625, "learning_rate": 1.3728677240470157e-05, "loss": 0.1426, "step": 21008 }, { "epoch": 130.49221183800623, "grad_norm": 0.96875, "learning_rate": 1.3726478145218391e-05, "loss": 0.1024, "step": 21009 }, { "epoch": 130.49844236760126, "grad_norm": 1.390625, "learning_rate": 1.3724279159457314e-05, "loss": 0.1146, "step": 21010 }, { "epoch": 130.50467289719626, "grad_norm": 1.578125, "learning_rate": 1.3722080283208294e-05, "loss": 0.1138, "step": 21011 }, { "epoch": 130.5109034267913, "grad_norm": 3.078125, "learning_rate": 1.3719881516492694e-05, "loss": 0.3279, "step": 21012 }, { "epoch": 130.5171339563863, "grad_norm": 2.3125, "learning_rate": 1.3717682859331849e-05, "loss": 0.1348, "step": 21013 }, { "epoch": 130.52336448598132, "grad_norm": 1.3671875, "learning_rate": 1.3715484311747134e-05, "loss": 0.0983, "step": 21014 }, { "epoch": 130.52959501557632, "grad_norm": 3.359375, "learning_rate": 1.371328587375988e-05, "loss": 0.2962, "step": 21015 }, { "epoch": 130.53582554517135, "grad_norm": 3.8125, "learning_rate": 1.371108754539146e-05, "loss": 0.2132, "step": 21016 }, { "epoch": 130.54205607476635, "grad_norm": 2.625, "learning_rate": 1.3708889326663205e-05, "loss": 0.2289, "step": 21017 }, { "epoch": 130.54828660436138, "grad_norm": 1.984375, "learning_rate": 1.3706691217596473e-05, "loss": 0.1284, "step": 21018 }, { "epoch": 130.55451713395638, "grad_norm": 3.484375, "learning_rate": 1.3704493218212624e-05, "loss": 0.1931, "step": 21019 }, { "epoch": 130.5607476635514, "grad_norm": 2.609375, "learning_rate": 1.3702295328532985e-05, "loss": 0.1561, "step": 21020 }, { "epoch": 130.5669781931464, "grad_norm": 1.1015625, "learning_rate": 1.3700097548578921e-05, "loss": 0.0967, "step": 21021 }, { "epoch": 130.57320872274144, "grad_norm": 2.703125, "learning_rate": 1.3697899878371762e-05, "loss": 0.1973, "step": 21022 }, { "epoch": 130.57943925233644, "grad_norm": 1.3046875, "learning_rate": 1.3695702317932862e-05, "loss": 0.107, "step": 21023 }, { "epoch": 130.58566978193147, "grad_norm": 2.234375, "learning_rate": 1.3693504867283558e-05, "loss": 0.1154, "step": 21024 }, { "epoch": 130.59190031152647, "grad_norm": 2.921875, "learning_rate": 1.3691307526445191e-05, "loss": 0.1429, "step": 21025 }, { "epoch": 130.5981308411215, "grad_norm": 2.140625, "learning_rate": 1.3689110295439107e-05, "loss": 0.1156, "step": 21026 }, { "epoch": 130.6043613707165, "grad_norm": 2.015625, "learning_rate": 1.3686913174286665e-05, "loss": 0.2087, "step": 21027 }, { "epoch": 130.61059190031153, "grad_norm": 2.453125, "learning_rate": 1.368471616300916e-05, "loss": 0.176, "step": 21028 }, { "epoch": 130.61682242990653, "grad_norm": 1.6640625, "learning_rate": 1.3682519261627957e-05, "loss": 0.1222, "step": 21029 }, { "epoch": 130.62305295950156, "grad_norm": 2.671875, "learning_rate": 1.3680322470164398e-05, "loss": 0.2207, "step": 21030 }, { "epoch": 130.62928348909657, "grad_norm": 1.4609375, "learning_rate": 1.36781257886398e-05, "loss": 0.141, "step": 21031 }, { "epoch": 130.6355140186916, "grad_norm": 1.671875, "learning_rate": 1.3675929217075508e-05, "loss": 0.1316, "step": 21032 }, { "epoch": 130.6417445482866, "grad_norm": 1.6875, "learning_rate": 1.3673732755492865e-05, "loss": 0.1081, "step": 21033 }, { "epoch": 130.64797507788163, "grad_norm": 1.3515625, "learning_rate": 1.3671536403913193e-05, "loss": 0.0977, "step": 21034 }, { "epoch": 130.65420560747663, "grad_norm": 2.453125, "learning_rate": 1.3669340162357816e-05, "loss": 0.2065, "step": 21035 }, { "epoch": 130.66043613707166, "grad_norm": 2.625, "learning_rate": 1.3667144030848073e-05, "loss": 0.2151, "step": 21036 }, { "epoch": 130.66666666666666, "grad_norm": 0.96875, "learning_rate": 1.3664948009405288e-05, "loss": 0.0985, "step": 21037 }, { "epoch": 130.6728971962617, "grad_norm": 1.28125, "learning_rate": 1.3662752098050807e-05, "loss": 0.1286, "step": 21038 }, { "epoch": 130.6791277258567, "grad_norm": 1.8203125, "learning_rate": 1.3660556296805932e-05, "loss": 0.1578, "step": 21039 }, { "epoch": 130.68535825545172, "grad_norm": 2.09375, "learning_rate": 1.3658360605692014e-05, "loss": 0.2397, "step": 21040 }, { "epoch": 130.69158878504672, "grad_norm": 2.4375, "learning_rate": 1.3656165024730352e-05, "loss": 0.1698, "step": 21041 }, { "epoch": 130.69781931464175, "grad_norm": 1.3046875, "learning_rate": 1.3653969553942295e-05, "loss": 0.0818, "step": 21042 }, { "epoch": 130.70404984423675, "grad_norm": 1.9765625, "learning_rate": 1.3651774193349142e-05, "loss": 0.1247, "step": 21043 }, { "epoch": 130.71028037383178, "grad_norm": 1.6640625, "learning_rate": 1.364957894297223e-05, "loss": 0.1985, "step": 21044 }, { "epoch": 130.71651090342678, "grad_norm": 2.234375, "learning_rate": 1.364738380283288e-05, "loss": 0.1206, "step": 21045 }, { "epoch": 130.7227414330218, "grad_norm": 2.71875, "learning_rate": 1.3645188772952411e-05, "loss": 0.3019, "step": 21046 }, { "epoch": 130.7289719626168, "grad_norm": 1.4765625, "learning_rate": 1.3642993853352132e-05, "loss": 0.0972, "step": 21047 }, { "epoch": 130.73520249221184, "grad_norm": 3.453125, "learning_rate": 1.3640799044053362e-05, "loss": 0.1688, "step": 21048 }, { "epoch": 130.74143302180684, "grad_norm": 2.40625, "learning_rate": 1.3638604345077433e-05, "loss": 0.2285, "step": 21049 }, { "epoch": 130.74766355140187, "grad_norm": 1.5703125, "learning_rate": 1.363640975644564e-05, "loss": 0.1253, "step": 21050 }, { "epoch": 130.75389408099687, "grad_norm": 2.3125, "learning_rate": 1.3634215278179308e-05, "loss": 0.1365, "step": 21051 }, { "epoch": 130.7601246105919, "grad_norm": 0.78515625, "learning_rate": 1.3632020910299758e-05, "loss": 0.0908, "step": 21052 }, { "epoch": 130.7663551401869, "grad_norm": 2.171875, "learning_rate": 1.3629826652828295e-05, "loss": 0.1309, "step": 21053 }, { "epoch": 130.77258566978193, "grad_norm": 1.59375, "learning_rate": 1.362763250578622e-05, "loss": 0.1678, "step": 21054 }, { "epoch": 130.77881619937693, "grad_norm": 3.28125, "learning_rate": 1.3625438469194845e-05, "loss": 0.2527, "step": 21055 }, { "epoch": 130.78504672897196, "grad_norm": 2.109375, "learning_rate": 1.3623244543075503e-05, "loss": 0.2368, "step": 21056 }, { "epoch": 130.79127725856696, "grad_norm": 1.6171875, "learning_rate": 1.362105072744947e-05, "loss": 0.1292, "step": 21057 }, { "epoch": 130.797507788162, "grad_norm": 1.2734375, "learning_rate": 1.3618857022338066e-05, "loss": 0.1369, "step": 21058 }, { "epoch": 130.80373831775702, "grad_norm": 1.96875, "learning_rate": 1.3616663427762611e-05, "loss": 0.2253, "step": 21059 }, { "epoch": 130.80996884735202, "grad_norm": 2.625, "learning_rate": 1.3614469943744393e-05, "loss": 0.1813, "step": 21060 }, { "epoch": 130.81619937694705, "grad_norm": 3.25, "learning_rate": 1.3612276570304711e-05, "loss": 0.1557, "step": 21061 }, { "epoch": 130.82242990654206, "grad_norm": 1.96875, "learning_rate": 1.3610083307464871e-05, "loss": 0.1807, "step": 21062 }, { "epoch": 130.82866043613708, "grad_norm": 1.84375, "learning_rate": 1.3607890155246184e-05, "loss": 0.0926, "step": 21063 }, { "epoch": 130.8348909657321, "grad_norm": 1.453125, "learning_rate": 1.3605697113669952e-05, "loss": 0.1184, "step": 21064 }, { "epoch": 130.84112149532712, "grad_norm": 1.5625, "learning_rate": 1.360350418275747e-05, "loss": 0.1131, "step": 21065 }, { "epoch": 130.84735202492212, "grad_norm": 1.7109375, "learning_rate": 1.3601311362530023e-05, "loss": 0.0858, "step": 21066 }, { "epoch": 130.85358255451715, "grad_norm": 2.25, "learning_rate": 1.3599118653008916e-05, "loss": 0.2009, "step": 21067 }, { "epoch": 130.85981308411215, "grad_norm": 1.75, "learning_rate": 1.3596926054215458e-05, "loss": 0.1673, "step": 21068 }, { "epoch": 130.86604361370718, "grad_norm": 1.609375, "learning_rate": 1.3594733566170926e-05, "loss": 0.114, "step": 21069 }, { "epoch": 130.87227414330218, "grad_norm": 1.0390625, "learning_rate": 1.359254118889662e-05, "loss": 0.0845, "step": 21070 }, { "epoch": 130.8785046728972, "grad_norm": 2.046875, "learning_rate": 1.3590348922413845e-05, "loss": 0.1663, "step": 21071 }, { "epoch": 130.8847352024922, "grad_norm": 2.1875, "learning_rate": 1.3588156766743881e-05, "loss": 0.1473, "step": 21072 }, { "epoch": 130.89096573208724, "grad_norm": 1.484375, "learning_rate": 1.3585964721908012e-05, "loss": 0.1198, "step": 21073 }, { "epoch": 130.89719626168224, "grad_norm": 3.125, "learning_rate": 1.3583772787927532e-05, "loss": 0.2116, "step": 21074 }, { "epoch": 130.90342679127727, "grad_norm": 2.640625, "learning_rate": 1.3581580964823742e-05, "loss": 0.2113, "step": 21075 }, { "epoch": 130.90965732087227, "grad_norm": 1.6484375, "learning_rate": 1.3579389252617913e-05, "loss": 0.1329, "step": 21076 }, { "epoch": 130.9158878504673, "grad_norm": 3.453125, "learning_rate": 1.3577197651331341e-05, "loss": 0.3301, "step": 21077 }, { "epoch": 130.9221183800623, "grad_norm": 2.375, "learning_rate": 1.3575006160985314e-05, "loss": 0.1644, "step": 21078 }, { "epoch": 130.92834890965733, "grad_norm": 2.609375, "learning_rate": 1.3572814781601112e-05, "loss": 0.2146, "step": 21079 }, { "epoch": 130.93457943925233, "grad_norm": 1.9375, "learning_rate": 1.3570623513200008e-05, "loss": 0.1611, "step": 21080 }, { "epoch": 130.94080996884736, "grad_norm": 1.5390625, "learning_rate": 1.3568432355803292e-05, "loss": 0.1366, "step": 21081 }, { "epoch": 130.94704049844236, "grad_norm": 1.9296875, "learning_rate": 1.356624130943226e-05, "loss": 0.1569, "step": 21082 }, { "epoch": 130.9532710280374, "grad_norm": 1.4375, "learning_rate": 1.3564050374108162e-05, "loss": 0.1407, "step": 21083 }, { "epoch": 130.9595015576324, "grad_norm": 1.703125, "learning_rate": 1.3561859549852308e-05, "loss": 0.1485, "step": 21084 }, { "epoch": 130.96573208722742, "grad_norm": 1.828125, "learning_rate": 1.3559668836685946e-05, "loss": 0.2306, "step": 21085 }, { "epoch": 130.97196261682242, "grad_norm": 2.3125, "learning_rate": 1.355747823463038e-05, "loss": 0.2608, "step": 21086 }, { "epoch": 130.97819314641745, "grad_norm": 1.59375, "learning_rate": 1.3555287743706862e-05, "loss": 0.1524, "step": 21087 }, { "epoch": 130.98442367601245, "grad_norm": 2.1875, "learning_rate": 1.3553097363936678e-05, "loss": 0.1769, "step": 21088 }, { "epoch": 130.99065420560748, "grad_norm": 2.640625, "learning_rate": 1.3550907095341109e-05, "loss": 0.3324, "step": 21089 }, { "epoch": 130.99688473520249, "grad_norm": 3.625, "learning_rate": 1.3548716937941409e-05, "loss": 0.3647, "step": 21090 }, { "epoch": 131.0, "grad_norm": 1.3359375, "learning_rate": 1.3546526891758871e-05, "loss": 0.0486, "step": 21091 }, { "epoch": 131.00623052959503, "grad_norm": 2.328125, "learning_rate": 1.354433695681474e-05, "loss": 0.1763, "step": 21092 }, { "epoch": 131.01246105919003, "grad_norm": 2.0, "learning_rate": 1.3542147133130312e-05, "loss": 0.1413, "step": 21093 }, { "epoch": 131.01869158878506, "grad_norm": 1.4296875, "learning_rate": 1.3539957420726828e-05, "loss": 0.1309, "step": 21094 }, { "epoch": 131.02492211838006, "grad_norm": 1.984375, "learning_rate": 1.3537767819625574e-05, "loss": 0.2237, "step": 21095 }, { "epoch": 131.0311526479751, "grad_norm": 1.28125, "learning_rate": 1.3535578329847806e-05, "loss": 0.1242, "step": 21096 }, { "epoch": 131.0373831775701, "grad_norm": 1.9453125, "learning_rate": 1.3533388951414813e-05, "loss": 0.1369, "step": 21097 }, { "epoch": 131.04361370716512, "grad_norm": 1.8984375, "learning_rate": 1.3531199684347817e-05, "loss": 0.204, "step": 21098 }, { "epoch": 131.04984423676012, "grad_norm": 1.0703125, "learning_rate": 1.3529010528668107e-05, "loss": 0.1151, "step": 21099 }, { "epoch": 131.05607476635515, "grad_norm": 1.703125, "learning_rate": 1.3526821484396938e-05, "loss": 0.1441, "step": 21100 }, { "epoch": 131.06230529595015, "grad_norm": 1.796875, "learning_rate": 1.352463255155558e-05, "loss": 0.1546, "step": 21101 }, { "epoch": 131.06853582554518, "grad_norm": 1.34375, "learning_rate": 1.3522443730165274e-05, "loss": 0.1354, "step": 21102 }, { "epoch": 131.07476635514018, "grad_norm": 2.0625, "learning_rate": 1.3520255020247303e-05, "loss": 0.1702, "step": 21103 }, { "epoch": 131.0809968847352, "grad_norm": 2.1875, "learning_rate": 1.3518066421822898e-05, "loss": 0.1916, "step": 21104 }, { "epoch": 131.08722741433021, "grad_norm": 2.828125, "learning_rate": 1.3515877934913335e-05, "loss": 0.2416, "step": 21105 }, { "epoch": 131.09345794392524, "grad_norm": 3.015625, "learning_rate": 1.3513689559539855e-05, "loss": 0.2697, "step": 21106 }, { "epoch": 131.09968847352025, "grad_norm": 1.546875, "learning_rate": 1.3511501295723717e-05, "loss": 0.1147, "step": 21107 }, { "epoch": 131.10591900311528, "grad_norm": 1.9453125, "learning_rate": 1.350931314348618e-05, "loss": 0.1188, "step": 21108 }, { "epoch": 131.11214953271028, "grad_norm": 3.0625, "learning_rate": 1.3507125102848487e-05, "loss": 0.3581, "step": 21109 }, { "epoch": 131.1183800623053, "grad_norm": 1.6328125, "learning_rate": 1.3504937173831899e-05, "loss": 0.1175, "step": 21110 }, { "epoch": 131.1246105919003, "grad_norm": 2.640625, "learning_rate": 1.350274935645765e-05, "loss": 0.2287, "step": 21111 }, { "epoch": 131.13084112149534, "grad_norm": 1.9453125, "learning_rate": 1.3500561650747006e-05, "loss": 0.1655, "step": 21112 }, { "epoch": 131.13707165109034, "grad_norm": 1.1640625, "learning_rate": 1.3498374056721197e-05, "loss": 0.1181, "step": 21113 }, { "epoch": 131.14330218068537, "grad_norm": 1.625, "learning_rate": 1.3496186574401481e-05, "loss": 0.1574, "step": 21114 }, { "epoch": 131.14953271028037, "grad_norm": 2.359375, "learning_rate": 1.3493999203809108e-05, "loss": 0.2094, "step": 21115 }, { "epoch": 131.1557632398754, "grad_norm": 3.40625, "learning_rate": 1.3491811944965316e-05, "loss": 0.2522, "step": 21116 }, { "epoch": 131.1619937694704, "grad_norm": 1.8203125, "learning_rate": 1.3489624797891334e-05, "loss": 0.1288, "step": 21117 }, { "epoch": 131.16822429906543, "grad_norm": 1.625, "learning_rate": 1.3487437762608418e-05, "loss": 0.0923, "step": 21118 }, { "epoch": 131.17445482866043, "grad_norm": 3.109375, "learning_rate": 1.348525083913782e-05, "loss": 0.2513, "step": 21119 }, { "epoch": 131.18068535825546, "grad_norm": 1.59375, "learning_rate": 1.3483064027500755e-05, "loss": 0.1379, "step": 21120 }, { "epoch": 131.18691588785046, "grad_norm": 1.78125, "learning_rate": 1.3480877327718477e-05, "loss": 0.101, "step": 21121 }, { "epoch": 131.1931464174455, "grad_norm": 3.203125, "learning_rate": 1.3478690739812228e-05, "loss": 0.2325, "step": 21122 }, { "epoch": 131.1993769470405, "grad_norm": 2.0625, "learning_rate": 1.347650426380324e-05, "loss": 0.2382, "step": 21123 }, { "epoch": 131.20560747663552, "grad_norm": 2.796875, "learning_rate": 1.3474317899712736e-05, "loss": 0.2249, "step": 21124 }, { "epoch": 131.21183800623052, "grad_norm": 2.3125, "learning_rate": 1.347213164756196e-05, "loss": 0.1866, "step": 21125 }, { "epoch": 131.21806853582555, "grad_norm": 2.453125, "learning_rate": 1.3469945507372154e-05, "loss": 0.1528, "step": 21126 }, { "epoch": 131.22429906542055, "grad_norm": 2.3125, "learning_rate": 1.3467759479164532e-05, "loss": 0.1399, "step": 21127 }, { "epoch": 131.23052959501558, "grad_norm": 2.015625, "learning_rate": 1.3465573562960338e-05, "loss": 0.1323, "step": 21128 }, { "epoch": 131.23676012461058, "grad_norm": 2.6875, "learning_rate": 1.3463387758780806e-05, "loss": 0.2526, "step": 21129 }, { "epoch": 131.2429906542056, "grad_norm": 2.078125, "learning_rate": 1.346120206664716e-05, "loss": 0.1254, "step": 21130 }, { "epoch": 131.2492211838006, "grad_norm": 1.9453125, "learning_rate": 1.3459016486580617e-05, "loss": 0.1036, "step": 21131 }, { "epoch": 131.25545171339564, "grad_norm": 1.8515625, "learning_rate": 1.3456831018602412e-05, "loss": 0.162, "step": 21132 }, { "epoch": 131.26168224299064, "grad_norm": 1.8671875, "learning_rate": 1.3454645662733772e-05, "loss": 0.153, "step": 21133 }, { "epoch": 131.26791277258567, "grad_norm": 2.203125, "learning_rate": 1.3452460418995933e-05, "loss": 0.1676, "step": 21134 }, { "epoch": 131.27414330218068, "grad_norm": 2.03125, "learning_rate": 1.3450275287410102e-05, "loss": 0.0975, "step": 21135 }, { "epoch": 131.2803738317757, "grad_norm": 3.078125, "learning_rate": 1.3448090267997498e-05, "loss": 0.2489, "step": 21136 }, { "epoch": 131.2866043613707, "grad_norm": 2.5625, "learning_rate": 1.3445905360779354e-05, "loss": 0.1584, "step": 21137 }, { "epoch": 131.29283489096574, "grad_norm": 1.7890625, "learning_rate": 1.344372056577689e-05, "loss": 0.1335, "step": 21138 }, { "epoch": 131.29906542056074, "grad_norm": 3.515625, "learning_rate": 1.344153588301132e-05, "loss": 0.1715, "step": 21139 }, { "epoch": 131.30529595015577, "grad_norm": 2.328125, "learning_rate": 1.3439351312503856e-05, "loss": 0.1726, "step": 21140 }, { "epoch": 131.31152647975077, "grad_norm": 3.359375, "learning_rate": 1.343716685427574e-05, "loss": 0.2013, "step": 21141 }, { "epoch": 131.3177570093458, "grad_norm": 1.5, "learning_rate": 1.3434982508348165e-05, "loss": 0.1432, "step": 21142 }, { "epoch": 131.3239875389408, "grad_norm": 1.7265625, "learning_rate": 1.3432798274742345e-05, "loss": 0.1089, "step": 21143 }, { "epoch": 131.33021806853583, "grad_norm": 2.3125, "learning_rate": 1.3430614153479498e-05, "loss": 0.1625, "step": 21144 }, { "epoch": 131.33644859813083, "grad_norm": 1.7734375, "learning_rate": 1.3428430144580854e-05, "loss": 0.1386, "step": 21145 }, { "epoch": 131.34267912772586, "grad_norm": 2.203125, "learning_rate": 1.3426246248067594e-05, "loss": 0.1971, "step": 21146 }, { "epoch": 131.34890965732086, "grad_norm": 2.859375, "learning_rate": 1.3424062463960945e-05, "loss": 0.2776, "step": 21147 }, { "epoch": 131.3551401869159, "grad_norm": 1.8671875, "learning_rate": 1.3421878792282128e-05, "loss": 0.1547, "step": 21148 }, { "epoch": 131.3613707165109, "grad_norm": 1.4453125, "learning_rate": 1.3419695233052338e-05, "loss": 0.1009, "step": 21149 }, { "epoch": 131.36760124610592, "grad_norm": 1.984375, "learning_rate": 1.3417511786292772e-05, "loss": 0.1551, "step": 21150 }, { "epoch": 131.37383177570092, "grad_norm": 1.3828125, "learning_rate": 1.3415328452024648e-05, "loss": 0.1106, "step": 21151 }, { "epoch": 131.38006230529595, "grad_norm": 1.609375, "learning_rate": 1.3413145230269178e-05, "loss": 0.1092, "step": 21152 }, { "epoch": 131.38629283489095, "grad_norm": 1.7421875, "learning_rate": 1.341096212104755e-05, "loss": 0.1382, "step": 21153 }, { "epoch": 131.39252336448598, "grad_norm": 2.09375, "learning_rate": 1.3408779124380983e-05, "loss": 0.1862, "step": 21154 }, { "epoch": 131.39875389408098, "grad_norm": 1.453125, "learning_rate": 1.3406596240290659e-05, "loss": 0.1269, "step": 21155 }, { "epoch": 131.404984423676, "grad_norm": 1.5703125, "learning_rate": 1.34044134687978e-05, "loss": 0.0869, "step": 21156 }, { "epoch": 131.41121495327104, "grad_norm": 2.296875, "learning_rate": 1.3402230809923586e-05, "loss": 0.2394, "step": 21157 }, { "epoch": 131.41744548286604, "grad_norm": 2.125, "learning_rate": 1.3400048263689224e-05, "loss": 0.164, "step": 21158 }, { "epoch": 131.42367601246107, "grad_norm": 2.078125, "learning_rate": 1.3397865830115914e-05, "loss": 0.1474, "step": 21159 }, { "epoch": 131.42990654205607, "grad_norm": 1.71875, "learning_rate": 1.339568350922486e-05, "loss": 0.121, "step": 21160 }, { "epoch": 131.4361370716511, "grad_norm": 1.5390625, "learning_rate": 1.3393501301037245e-05, "loss": 0.128, "step": 21161 }, { "epoch": 131.4423676012461, "grad_norm": 1.1640625, "learning_rate": 1.3391319205574256e-05, "loss": 0.1091, "step": 21162 }, { "epoch": 131.44859813084113, "grad_norm": 2.078125, "learning_rate": 1.3389137222857096e-05, "loss": 0.1618, "step": 21163 }, { "epoch": 131.45482866043614, "grad_norm": 2.515625, "learning_rate": 1.3386955352906968e-05, "loss": 0.1898, "step": 21164 }, { "epoch": 131.46105919003116, "grad_norm": 1.609375, "learning_rate": 1.338477359574504e-05, "loss": 0.1574, "step": 21165 }, { "epoch": 131.46728971962617, "grad_norm": 2.78125, "learning_rate": 1.3382591951392512e-05, "loss": 0.2303, "step": 21166 }, { "epoch": 131.4735202492212, "grad_norm": 2.453125, "learning_rate": 1.3380410419870582e-05, "loss": 0.26, "step": 21167 }, { "epoch": 131.4797507788162, "grad_norm": 1.1953125, "learning_rate": 1.3378229001200431e-05, "loss": 0.1066, "step": 21168 }, { "epoch": 131.48598130841123, "grad_norm": 2.0625, "learning_rate": 1.3376047695403232e-05, "loss": 0.1481, "step": 21169 }, { "epoch": 131.49221183800623, "grad_norm": 1.484375, "learning_rate": 1.3373866502500181e-05, "loss": 0.1096, "step": 21170 }, { "epoch": 131.49844236760126, "grad_norm": 2.09375, "learning_rate": 1.3371685422512475e-05, "loss": 0.1204, "step": 21171 }, { "epoch": 131.50467289719626, "grad_norm": 1.453125, "learning_rate": 1.3369504455461273e-05, "loss": 0.1016, "step": 21172 }, { "epoch": 131.5109034267913, "grad_norm": 1.890625, "learning_rate": 1.336732360136778e-05, "loss": 0.155, "step": 21173 }, { "epoch": 131.5171339563863, "grad_norm": 2.390625, "learning_rate": 1.3365142860253151e-05, "loss": 0.2099, "step": 21174 }, { "epoch": 131.52336448598132, "grad_norm": 1.7109375, "learning_rate": 1.3362962232138595e-05, "loss": 0.1031, "step": 21175 }, { "epoch": 131.52959501557632, "grad_norm": 1.9921875, "learning_rate": 1.3360781717045264e-05, "loss": 0.1576, "step": 21176 }, { "epoch": 131.53582554517135, "grad_norm": 1.1796875, "learning_rate": 1.3358601314994349e-05, "loss": 0.0936, "step": 21177 }, { "epoch": 131.54205607476635, "grad_norm": 2.53125, "learning_rate": 1.3356421026007032e-05, "loss": 0.2254, "step": 21178 }, { "epoch": 131.54828660436138, "grad_norm": 2.203125, "learning_rate": 1.3354240850104476e-05, "loss": 0.1689, "step": 21179 }, { "epoch": 131.55451713395638, "grad_norm": 2.875, "learning_rate": 1.3352060787307869e-05, "loss": 0.213, "step": 21180 }, { "epoch": 131.5607476635514, "grad_norm": 1.7578125, "learning_rate": 1.3349880837638362e-05, "loss": 0.1335, "step": 21181 }, { "epoch": 131.5669781931464, "grad_norm": 2.859375, "learning_rate": 1.3347701001117157e-05, "loss": 0.2214, "step": 21182 }, { "epoch": 131.57320872274144, "grad_norm": 1.6015625, "learning_rate": 1.3345521277765394e-05, "loss": 0.1559, "step": 21183 }, { "epoch": 131.57943925233644, "grad_norm": 2.3125, "learning_rate": 1.3343341667604261e-05, "loss": 0.1628, "step": 21184 }, { "epoch": 131.58566978193147, "grad_norm": 2.890625, "learning_rate": 1.3341162170654936e-05, "loss": 0.2276, "step": 21185 }, { "epoch": 131.59190031152647, "grad_norm": 1.7578125, "learning_rate": 1.333898278693857e-05, "loss": 0.149, "step": 21186 }, { "epoch": 131.5981308411215, "grad_norm": 1.5078125, "learning_rate": 1.3336803516476326e-05, "loss": 0.1406, "step": 21187 }, { "epoch": 131.6043613707165, "grad_norm": 1.578125, "learning_rate": 1.3334624359289377e-05, "loss": 0.1, "step": 21188 }, { "epoch": 131.61059190031153, "grad_norm": 2.25, "learning_rate": 1.33324453153989e-05, "loss": 0.182, "step": 21189 }, { "epoch": 131.61682242990653, "grad_norm": 1.0546875, "learning_rate": 1.3330266384826035e-05, "loss": 0.1183, "step": 21190 }, { "epoch": 131.62305295950156, "grad_norm": 2.15625, "learning_rate": 1.3328087567591956e-05, "loss": 0.1415, "step": 21191 }, { "epoch": 131.62928348909657, "grad_norm": 2.390625, "learning_rate": 1.3325908863717823e-05, "loss": 0.2197, "step": 21192 }, { "epoch": 131.6355140186916, "grad_norm": 2.984375, "learning_rate": 1.3323730273224819e-05, "loss": 0.2289, "step": 21193 }, { "epoch": 131.6417445482866, "grad_norm": 2.0625, "learning_rate": 1.3321551796134057e-05, "loss": 0.1908, "step": 21194 }, { "epoch": 131.64797507788163, "grad_norm": 1.53125, "learning_rate": 1.3319373432466723e-05, "loss": 0.1476, "step": 21195 }, { "epoch": 131.65420560747663, "grad_norm": 3.140625, "learning_rate": 1.3317195182243967e-05, "loss": 0.2433, "step": 21196 }, { "epoch": 131.66043613707166, "grad_norm": 2.0, "learning_rate": 1.3315017045486955e-05, "loss": 0.1703, "step": 21197 }, { "epoch": 131.66666666666666, "grad_norm": 1.359375, "learning_rate": 1.3312839022216824e-05, "loss": 0.0957, "step": 21198 }, { "epoch": 131.6728971962617, "grad_norm": 2.328125, "learning_rate": 1.3310661112454748e-05, "loss": 0.2453, "step": 21199 }, { "epoch": 131.6791277258567, "grad_norm": 2.296875, "learning_rate": 1.3308483316221856e-05, "loss": 0.1904, "step": 21200 }, { "epoch": 131.68535825545172, "grad_norm": 2.09375, "learning_rate": 1.3306305633539323e-05, "loss": 0.1666, "step": 21201 }, { "epoch": 131.69158878504672, "grad_norm": 0.9609375, "learning_rate": 1.3304128064428279e-05, "loss": 0.0894, "step": 21202 }, { "epoch": 131.69781931464175, "grad_norm": 3.15625, "learning_rate": 1.3301950608909874e-05, "loss": 0.2031, "step": 21203 }, { "epoch": 131.70404984423675, "grad_norm": 3.03125, "learning_rate": 1.329977326700528e-05, "loss": 0.2411, "step": 21204 }, { "epoch": 131.71028037383178, "grad_norm": 1.953125, "learning_rate": 1.3297596038735623e-05, "loss": 0.1051, "step": 21205 }, { "epoch": 131.71651090342678, "grad_norm": 1.90625, "learning_rate": 1.3295418924122043e-05, "loss": 0.1292, "step": 21206 }, { "epoch": 131.7227414330218, "grad_norm": 2.015625, "learning_rate": 1.3293241923185697e-05, "loss": 0.1498, "step": 21207 }, { "epoch": 131.7289719626168, "grad_norm": 1.53125, "learning_rate": 1.3291065035947736e-05, "loss": 0.1362, "step": 21208 }, { "epoch": 131.73520249221184, "grad_norm": 2.171875, "learning_rate": 1.3288888262429277e-05, "loss": 0.1658, "step": 21209 }, { "epoch": 131.74143302180684, "grad_norm": 2.34375, "learning_rate": 1.328671160265148e-05, "loss": 0.1504, "step": 21210 }, { "epoch": 131.74766355140187, "grad_norm": 3.09375, "learning_rate": 1.3284535056635488e-05, "loss": 0.2895, "step": 21211 }, { "epoch": 131.75389408099687, "grad_norm": 1.4921875, "learning_rate": 1.3282358624402438e-05, "loss": 0.1165, "step": 21212 }, { "epoch": 131.7601246105919, "grad_norm": 2.1875, "learning_rate": 1.3280182305973452e-05, "loss": 0.1408, "step": 21213 }, { "epoch": 131.7663551401869, "grad_norm": 1.453125, "learning_rate": 1.3278006101369678e-05, "loss": 0.1336, "step": 21214 }, { "epoch": 131.77258566978193, "grad_norm": 1.28125, "learning_rate": 1.3275830010612262e-05, "loss": 0.086, "step": 21215 }, { "epoch": 131.77881619937693, "grad_norm": 2.046875, "learning_rate": 1.3273654033722315e-05, "loss": 0.143, "step": 21216 }, { "epoch": 131.78504672897196, "grad_norm": 2.6875, "learning_rate": 1.3271478170720986e-05, "loss": 0.1533, "step": 21217 }, { "epoch": 131.79127725856696, "grad_norm": 2.421875, "learning_rate": 1.3269302421629417e-05, "loss": 0.1979, "step": 21218 }, { "epoch": 131.797507788162, "grad_norm": 2.6875, "learning_rate": 1.3267126786468726e-05, "loss": 0.1988, "step": 21219 }, { "epoch": 131.80373831775702, "grad_norm": 1.7734375, "learning_rate": 1.326495126526004e-05, "loss": 0.0989, "step": 21220 }, { "epoch": 131.80996884735202, "grad_norm": 2.265625, "learning_rate": 1.3262775858024487e-05, "loss": 0.1511, "step": 21221 }, { "epoch": 131.81619937694705, "grad_norm": 1.5, "learning_rate": 1.3260600564783212e-05, "loss": 0.1534, "step": 21222 }, { "epoch": 131.82242990654206, "grad_norm": 1.5859375, "learning_rate": 1.3258425385557322e-05, "loss": 0.1139, "step": 21223 }, { "epoch": 131.82866043613708, "grad_norm": 2.59375, "learning_rate": 1.3256250320367963e-05, "loss": 0.1766, "step": 21224 }, { "epoch": 131.8348909657321, "grad_norm": 3.09375, "learning_rate": 1.3254075369236238e-05, "loss": 0.1835, "step": 21225 }, { "epoch": 131.84112149532712, "grad_norm": 2.0625, "learning_rate": 1.325190053218329e-05, "loss": 0.1641, "step": 21226 }, { "epoch": 131.84735202492212, "grad_norm": 1.5390625, "learning_rate": 1.3249725809230224e-05, "loss": 0.104, "step": 21227 }, { "epoch": 131.85358255451715, "grad_norm": 3.46875, "learning_rate": 1.3247551200398167e-05, "loss": 0.2508, "step": 21228 }, { "epoch": 131.85981308411215, "grad_norm": 1.2734375, "learning_rate": 1.3245376705708246e-05, "loss": 0.1082, "step": 21229 }, { "epoch": 131.86604361370718, "grad_norm": 1.34375, "learning_rate": 1.324320232518158e-05, "loss": 0.1134, "step": 21230 }, { "epoch": 131.87227414330218, "grad_norm": 1.4375, "learning_rate": 1.3241028058839288e-05, "loss": 0.1629, "step": 21231 }, { "epoch": 131.8785046728972, "grad_norm": 2.5, "learning_rate": 1.323885390670247e-05, "loss": 0.1994, "step": 21232 }, { "epoch": 131.8847352024922, "grad_norm": 2.296875, "learning_rate": 1.3236679868792257e-05, "loss": 0.1761, "step": 21233 }, { "epoch": 131.89096573208724, "grad_norm": 3.234375, "learning_rate": 1.3234505945129766e-05, "loss": 0.3571, "step": 21234 }, { "epoch": 131.89719626168224, "grad_norm": 1.09375, "learning_rate": 1.3232332135736098e-05, "loss": 0.112, "step": 21235 }, { "epoch": 131.90342679127727, "grad_norm": 1.46875, "learning_rate": 1.3230158440632373e-05, "loss": 0.1205, "step": 21236 }, { "epoch": 131.90965732087227, "grad_norm": 2.765625, "learning_rate": 1.322798485983971e-05, "loss": 0.167, "step": 21237 }, { "epoch": 131.9158878504673, "grad_norm": 1.8671875, "learning_rate": 1.3225811393379212e-05, "loss": 0.1463, "step": 21238 }, { "epoch": 131.9221183800623, "grad_norm": 2.140625, "learning_rate": 1.3223638041271979e-05, "loss": 0.1411, "step": 21239 }, { "epoch": 131.92834890965733, "grad_norm": 2.6875, "learning_rate": 1.3221464803539124e-05, "loss": 0.2631, "step": 21240 }, { "epoch": 131.93457943925233, "grad_norm": 1.7109375, "learning_rate": 1.3219291680201773e-05, "loss": 0.1311, "step": 21241 }, { "epoch": 131.94080996884736, "grad_norm": 2.390625, "learning_rate": 1.3217118671281e-05, "loss": 0.125, "step": 21242 }, { "epoch": 131.94704049844236, "grad_norm": 1.125, "learning_rate": 1.3214945776797938e-05, "loss": 0.1375, "step": 21243 }, { "epoch": 131.9532710280374, "grad_norm": 1.828125, "learning_rate": 1.321277299677367e-05, "loss": 0.1335, "step": 21244 }, { "epoch": 131.9595015576324, "grad_norm": 2.359375, "learning_rate": 1.3210600331229317e-05, "loss": 0.2491, "step": 21245 }, { "epoch": 131.96573208722742, "grad_norm": 3.03125, "learning_rate": 1.3208427780185955e-05, "loss": 0.221, "step": 21246 }, { "epoch": 131.97196261682242, "grad_norm": 1.078125, "learning_rate": 1.3206255343664705e-05, "loss": 0.1153, "step": 21247 }, { "epoch": 131.97819314641745, "grad_norm": 2.5, "learning_rate": 1.3204083021686669e-05, "loss": 0.197, "step": 21248 }, { "epoch": 131.98442367601245, "grad_norm": 3.234375, "learning_rate": 1.3201910814272927e-05, "loss": 0.2312, "step": 21249 }, { "epoch": 131.99065420560748, "grad_norm": 1.1875, "learning_rate": 1.3199738721444593e-05, "loss": 0.1222, "step": 21250 }, { "epoch": 131.99688473520249, "grad_norm": 1.578125, "learning_rate": 1.319756674322275e-05, "loss": 0.128, "step": 21251 }, { "epoch": 132.0, "grad_norm": 1.0234375, "learning_rate": 1.3195394879628504e-05, "loss": 0.0593, "step": 21252 }, { "epoch": 132.00623052959503, "grad_norm": 2.703125, "learning_rate": 1.3193223130682936e-05, "loss": 0.1923, "step": 21253 }, { "epoch": 132.01246105919003, "grad_norm": 1.5078125, "learning_rate": 1.3191051496407143e-05, "loss": 0.1592, "step": 21254 }, { "epoch": 132.01869158878506, "grad_norm": 1.6953125, "learning_rate": 1.318887997682222e-05, "loss": 0.1367, "step": 21255 }, { "epoch": 132.02492211838006, "grad_norm": 1.328125, "learning_rate": 1.3186708571949274e-05, "loss": 0.1068, "step": 21256 }, { "epoch": 132.0311526479751, "grad_norm": 2.34375, "learning_rate": 1.3184537281809356e-05, "loss": 0.2034, "step": 21257 }, { "epoch": 132.0373831775701, "grad_norm": 2.15625, "learning_rate": 1.3182366106423572e-05, "loss": 0.1275, "step": 21258 }, { "epoch": 132.04361370716512, "grad_norm": 2.75, "learning_rate": 1.3180195045813015e-05, "loss": 0.1775, "step": 21259 }, { "epoch": 132.04984423676012, "grad_norm": 2.453125, "learning_rate": 1.317802409999877e-05, "loss": 0.2463, "step": 21260 }, { "epoch": 132.05607476635515, "grad_norm": 1.7890625, "learning_rate": 1.3175853269001914e-05, "loss": 0.1395, "step": 21261 }, { "epoch": 132.06230529595015, "grad_norm": 2.0, "learning_rate": 1.3173682552843534e-05, "loss": 0.1103, "step": 21262 }, { "epoch": 132.06853582554518, "grad_norm": 1.9609375, "learning_rate": 1.3171511951544722e-05, "loss": 0.1711, "step": 21263 }, { "epoch": 132.07476635514018, "grad_norm": 2.765625, "learning_rate": 1.316934146512655e-05, "loss": 0.2447, "step": 21264 }, { "epoch": 132.0809968847352, "grad_norm": 1.34375, "learning_rate": 1.3167171093610087e-05, "loss": 0.0926, "step": 21265 }, { "epoch": 132.08722741433021, "grad_norm": 3.515625, "learning_rate": 1.3165000837016429e-05, "loss": 0.1923, "step": 21266 }, { "epoch": 132.09345794392524, "grad_norm": 1.3515625, "learning_rate": 1.316283069536665e-05, "loss": 0.1287, "step": 21267 }, { "epoch": 132.09968847352025, "grad_norm": 2.34375, "learning_rate": 1.3160660668681824e-05, "loss": 0.1663, "step": 21268 }, { "epoch": 132.10591900311528, "grad_norm": 1.484375, "learning_rate": 1.3158490756983034e-05, "loss": 0.1303, "step": 21269 }, { "epoch": 132.11214953271028, "grad_norm": 2.78125, "learning_rate": 1.315632096029134e-05, "loss": 0.2569, "step": 21270 }, { "epoch": 132.1183800623053, "grad_norm": 1.9765625, "learning_rate": 1.3154151278627835e-05, "loss": 0.095, "step": 21271 }, { "epoch": 132.1246105919003, "grad_norm": 2.125, "learning_rate": 1.3151981712013566e-05, "loss": 0.1859, "step": 21272 }, { "epoch": 132.13084112149534, "grad_norm": 1.078125, "learning_rate": 1.3149812260469625e-05, "loss": 0.111, "step": 21273 }, { "epoch": 132.13707165109034, "grad_norm": 1.3125, "learning_rate": 1.3147642924017084e-05, "loss": 0.1296, "step": 21274 }, { "epoch": 132.14330218068537, "grad_norm": 1.9453125, "learning_rate": 1.3145473702677002e-05, "loss": 0.1597, "step": 21275 }, { "epoch": 132.14953271028037, "grad_norm": 2.1875, "learning_rate": 1.3143304596470443e-05, "loss": 0.114, "step": 21276 }, { "epoch": 132.1557632398754, "grad_norm": 1.796875, "learning_rate": 1.3141135605418479e-05, "loss": 0.1153, "step": 21277 }, { "epoch": 132.1619937694704, "grad_norm": 1.7265625, "learning_rate": 1.3138966729542186e-05, "loss": 0.1578, "step": 21278 }, { "epoch": 132.16822429906543, "grad_norm": 2.125, "learning_rate": 1.3136797968862607e-05, "loss": 0.2085, "step": 21279 }, { "epoch": 132.17445482866043, "grad_norm": 3.640625, "learning_rate": 1.313462932340082e-05, "loss": 0.2236, "step": 21280 }, { "epoch": 132.18068535825546, "grad_norm": 1.59375, "learning_rate": 1.3132460793177892e-05, "loss": 0.1263, "step": 21281 }, { "epoch": 132.18691588785046, "grad_norm": 2.484375, "learning_rate": 1.3130292378214882e-05, "loss": 0.2297, "step": 21282 }, { "epoch": 132.1931464174455, "grad_norm": 2.5625, "learning_rate": 1.312812407853283e-05, "loss": 0.2159, "step": 21283 }, { "epoch": 132.1993769470405, "grad_norm": 2.1875, "learning_rate": 1.3125955894152814e-05, "loss": 0.1662, "step": 21284 }, { "epoch": 132.20560747663552, "grad_norm": 1.3125, "learning_rate": 1.3123787825095896e-05, "loss": 0.1038, "step": 21285 }, { "epoch": 132.21183800623052, "grad_norm": 1.734375, "learning_rate": 1.3121619871383115e-05, "loss": 0.1196, "step": 21286 }, { "epoch": 132.21806853582555, "grad_norm": 1.34375, "learning_rate": 1.3119452033035534e-05, "loss": 0.1076, "step": 21287 }, { "epoch": 132.22429906542055, "grad_norm": 2.1875, "learning_rate": 1.3117284310074224e-05, "loss": 0.1926, "step": 21288 }, { "epoch": 132.23052959501558, "grad_norm": 1.921875, "learning_rate": 1.3115116702520219e-05, "loss": 0.151, "step": 21289 }, { "epoch": 132.23676012461058, "grad_norm": 2.21875, "learning_rate": 1.311294921039457e-05, "loss": 0.1331, "step": 21290 }, { "epoch": 132.2429906542056, "grad_norm": 3.1875, "learning_rate": 1.3110781833718333e-05, "loss": 0.2527, "step": 21291 }, { "epoch": 132.2492211838006, "grad_norm": 2.828125, "learning_rate": 1.3108614572512563e-05, "loss": 0.2012, "step": 21292 }, { "epoch": 132.25545171339564, "grad_norm": 2.328125, "learning_rate": 1.3106447426798314e-05, "loss": 0.2377, "step": 21293 }, { "epoch": 132.26168224299064, "grad_norm": 2.03125, "learning_rate": 1.310428039659662e-05, "loss": 0.1851, "step": 21294 }, { "epoch": 132.26791277258567, "grad_norm": 2.140625, "learning_rate": 1.3102113481928526e-05, "loss": 0.1408, "step": 21295 }, { "epoch": 132.27414330218068, "grad_norm": 2.609375, "learning_rate": 1.3099946682815084e-05, "loss": 0.201, "step": 21296 }, { "epoch": 132.2803738317757, "grad_norm": 3.078125, "learning_rate": 1.309777999927735e-05, "loss": 0.1151, "step": 21297 }, { "epoch": 132.2866043613707, "grad_norm": 1.40625, "learning_rate": 1.3095613431336346e-05, "loss": 0.1423, "step": 21298 }, { "epoch": 132.29283489096574, "grad_norm": 2.875, "learning_rate": 1.3093446979013122e-05, "loss": 0.2448, "step": 21299 }, { "epoch": 132.29906542056074, "grad_norm": 2.40625, "learning_rate": 1.3091280642328733e-05, "loss": 0.2388, "step": 21300 }, { "epoch": 132.30529595015577, "grad_norm": 1.171875, "learning_rate": 1.3089114421304205e-05, "loss": 0.1097, "step": 21301 }, { "epoch": 132.31152647975077, "grad_norm": 3.484375, "learning_rate": 1.3086948315960568e-05, "loss": 0.1131, "step": 21302 }, { "epoch": 132.3177570093458, "grad_norm": 1.4453125, "learning_rate": 1.3084782326318873e-05, "loss": 0.1257, "step": 21303 }, { "epoch": 132.3239875389408, "grad_norm": 1.3203125, "learning_rate": 1.3082616452400163e-05, "loss": 0.1045, "step": 21304 }, { "epoch": 132.33021806853583, "grad_norm": 2.125, "learning_rate": 1.3080450694225455e-05, "loss": 0.1028, "step": 21305 }, { "epoch": 132.33644859813083, "grad_norm": 1.8203125, "learning_rate": 1.3078285051815793e-05, "loss": 0.1041, "step": 21306 }, { "epoch": 132.34267912772586, "grad_norm": 2.28125, "learning_rate": 1.3076119525192215e-05, "loss": 0.2568, "step": 21307 }, { "epoch": 132.34890965732086, "grad_norm": 2.71875, "learning_rate": 1.307395411437575e-05, "loss": 0.2269, "step": 21308 }, { "epoch": 132.3551401869159, "grad_norm": 2.75, "learning_rate": 1.3071788819387418e-05, "loss": 0.1246, "step": 21309 }, { "epoch": 132.3613707165109, "grad_norm": 3.25, "learning_rate": 1.3069623640248258e-05, "loss": 0.1166, "step": 21310 }, { "epoch": 132.36760124610592, "grad_norm": 1.6328125, "learning_rate": 1.3067458576979304e-05, "loss": 0.1407, "step": 21311 }, { "epoch": 132.37383177570092, "grad_norm": 2.328125, "learning_rate": 1.3065293629601571e-05, "loss": 0.1085, "step": 21312 }, { "epoch": 132.38006230529595, "grad_norm": 1.9765625, "learning_rate": 1.3063128798136104e-05, "loss": 0.1619, "step": 21313 }, { "epoch": 132.38629283489095, "grad_norm": 1.9765625, "learning_rate": 1.3060964082603902e-05, "loss": 0.1873, "step": 21314 }, { "epoch": 132.39252336448598, "grad_norm": 1.921875, "learning_rate": 1.3058799483026013e-05, "loss": 0.1636, "step": 21315 }, { "epoch": 132.39875389408098, "grad_norm": 2.0625, "learning_rate": 1.3056634999423445e-05, "loss": 0.1653, "step": 21316 }, { "epoch": 132.404984423676, "grad_norm": 2.71875, "learning_rate": 1.3054470631817223e-05, "loss": 0.1842, "step": 21317 }, { "epoch": 132.41121495327104, "grad_norm": 2.78125, "learning_rate": 1.3052306380228379e-05, "loss": 0.2835, "step": 21318 }, { "epoch": 132.41744548286604, "grad_norm": 1.5234375, "learning_rate": 1.3050142244677915e-05, "loss": 0.1246, "step": 21319 }, { "epoch": 132.42367601246107, "grad_norm": 1.171875, "learning_rate": 1.304797822518687e-05, "loss": 0.1105, "step": 21320 }, { "epoch": 132.42990654205607, "grad_norm": 2.40625, "learning_rate": 1.3045814321776234e-05, "loss": 0.1778, "step": 21321 }, { "epoch": 132.4361370716511, "grad_norm": 2.15625, "learning_rate": 1.3043650534467053e-05, "loss": 0.1647, "step": 21322 }, { "epoch": 132.4423676012461, "grad_norm": 1.5546875, "learning_rate": 1.3041486863280317e-05, "loss": 0.147, "step": 21323 }, { "epoch": 132.44859813084113, "grad_norm": 2.578125, "learning_rate": 1.303932330823705e-05, "loss": 0.2526, "step": 21324 }, { "epoch": 132.45482866043614, "grad_norm": 1.6953125, "learning_rate": 1.3037159869358268e-05, "loss": 0.1392, "step": 21325 }, { "epoch": 132.46105919003116, "grad_norm": 2.28125, "learning_rate": 1.3034996546665002e-05, "loss": 0.1702, "step": 21326 }, { "epoch": 132.46728971962617, "grad_norm": 2.75, "learning_rate": 1.3032833340178218e-05, "loss": 0.2272, "step": 21327 }, { "epoch": 132.4735202492212, "grad_norm": 0.94921875, "learning_rate": 1.303067024991895e-05, "loss": 0.0807, "step": 21328 }, { "epoch": 132.4797507788162, "grad_norm": 2.34375, "learning_rate": 1.3028507275908206e-05, "loss": 0.1726, "step": 21329 }, { "epoch": 132.48598130841123, "grad_norm": 2.609375, "learning_rate": 1.3026344418167003e-05, "loss": 0.2365, "step": 21330 }, { "epoch": 132.49221183800623, "grad_norm": 1.953125, "learning_rate": 1.3024181676716324e-05, "loss": 0.1553, "step": 21331 }, { "epoch": 132.49844236760126, "grad_norm": 3.703125, "learning_rate": 1.3022019051577188e-05, "loss": 0.2132, "step": 21332 }, { "epoch": 132.50467289719626, "grad_norm": 1.71875, "learning_rate": 1.3019856542770607e-05, "loss": 0.1366, "step": 21333 }, { "epoch": 132.5109034267913, "grad_norm": 1.703125, "learning_rate": 1.3017694150317571e-05, "loss": 0.1126, "step": 21334 }, { "epoch": 132.5171339563863, "grad_norm": 1.3828125, "learning_rate": 1.3015531874239079e-05, "loss": 0.1087, "step": 21335 }, { "epoch": 132.52336448598132, "grad_norm": 1.6171875, "learning_rate": 1.3013369714556134e-05, "loss": 0.1298, "step": 21336 }, { "epoch": 132.52959501557632, "grad_norm": 1.5, "learning_rate": 1.3011207671289753e-05, "loss": 0.1558, "step": 21337 }, { "epoch": 132.53582554517135, "grad_norm": 2.328125, "learning_rate": 1.3009045744460906e-05, "loss": 0.2027, "step": 21338 }, { "epoch": 132.54205607476635, "grad_norm": 2.6875, "learning_rate": 1.3006883934090613e-05, "loss": 0.1168, "step": 21339 }, { "epoch": 132.54828660436138, "grad_norm": 1.90625, "learning_rate": 1.3004722240199851e-05, "loss": 0.1461, "step": 21340 }, { "epoch": 132.55451713395638, "grad_norm": 1.9375, "learning_rate": 1.3002560662809638e-05, "loss": 0.1355, "step": 21341 }, { "epoch": 132.5607476635514, "grad_norm": 1.2734375, "learning_rate": 1.3000399201940941e-05, "loss": 0.1119, "step": 21342 }, { "epoch": 132.5669781931464, "grad_norm": 2.921875, "learning_rate": 1.2998237857614764e-05, "loss": 0.2401, "step": 21343 }, { "epoch": 132.57320872274144, "grad_norm": 0.88671875, "learning_rate": 1.2996076629852114e-05, "loss": 0.0937, "step": 21344 }, { "epoch": 132.57943925233644, "grad_norm": 2.84375, "learning_rate": 1.2993915518673966e-05, "loss": 0.2114, "step": 21345 }, { "epoch": 132.58566978193147, "grad_norm": 1.65625, "learning_rate": 1.29917545241013e-05, "loss": 0.1111, "step": 21346 }, { "epoch": 132.59190031152647, "grad_norm": 2.28125, "learning_rate": 1.2989593646155112e-05, "loss": 0.0884, "step": 21347 }, { "epoch": 132.5981308411215, "grad_norm": 0.84765625, "learning_rate": 1.2987432884856403e-05, "loss": 0.0815, "step": 21348 }, { "epoch": 132.6043613707165, "grad_norm": 3.0625, "learning_rate": 1.298527224022614e-05, "loss": 0.1364, "step": 21349 }, { "epoch": 132.61059190031153, "grad_norm": 2.96875, "learning_rate": 1.2983111712285312e-05, "loss": 0.204, "step": 21350 }, { "epoch": 132.61682242990653, "grad_norm": 2.109375, "learning_rate": 1.2980951301054906e-05, "loss": 0.1602, "step": 21351 }, { "epoch": 132.62305295950156, "grad_norm": 0.70703125, "learning_rate": 1.2978791006555924e-05, "loss": 0.0888, "step": 21352 }, { "epoch": 132.62928348909657, "grad_norm": 1.84375, "learning_rate": 1.2976630828809305e-05, "loss": 0.1326, "step": 21353 }, { "epoch": 132.6355140186916, "grad_norm": 1.171875, "learning_rate": 1.297447076783605e-05, "loss": 0.0971, "step": 21354 }, { "epoch": 132.6417445482866, "grad_norm": 2.453125, "learning_rate": 1.297231082365714e-05, "loss": 0.1404, "step": 21355 }, { "epoch": 132.64797507788163, "grad_norm": 2.09375, "learning_rate": 1.2970150996293564e-05, "loss": 0.1639, "step": 21356 }, { "epoch": 132.65420560747663, "grad_norm": 0.91015625, "learning_rate": 1.2967991285766274e-05, "loss": 0.0932, "step": 21357 }, { "epoch": 132.66043613707166, "grad_norm": 1.765625, "learning_rate": 1.296583169209627e-05, "loss": 0.1195, "step": 21358 }, { "epoch": 132.66666666666666, "grad_norm": 1.6328125, "learning_rate": 1.29636722153045e-05, "loss": 0.1379, "step": 21359 }, { "epoch": 132.6728971962617, "grad_norm": 2.3125, "learning_rate": 1.2961512855411964e-05, "loss": 0.202, "step": 21360 }, { "epoch": 132.6791277258567, "grad_norm": 2.40625, "learning_rate": 1.2959353612439612e-05, "loss": 0.2702, "step": 21361 }, { "epoch": 132.68535825545172, "grad_norm": 2.921875, "learning_rate": 1.2957194486408425e-05, "loss": 0.1486, "step": 21362 }, { "epoch": 132.69158878504672, "grad_norm": 2.09375, "learning_rate": 1.2955035477339381e-05, "loss": 0.1531, "step": 21363 }, { "epoch": 132.69781931464175, "grad_norm": 1.03125, "learning_rate": 1.295287658525344e-05, "loss": 0.1053, "step": 21364 }, { "epoch": 132.70404984423675, "grad_norm": 1.515625, "learning_rate": 1.2950717810171558e-05, "loss": 0.1232, "step": 21365 }, { "epoch": 132.71028037383178, "grad_norm": 1.9765625, "learning_rate": 1.294855915211472e-05, "loss": 0.1819, "step": 21366 }, { "epoch": 132.71651090342678, "grad_norm": 1.75, "learning_rate": 1.2946400611103887e-05, "loss": 0.1567, "step": 21367 }, { "epoch": 132.7227414330218, "grad_norm": 2.4375, "learning_rate": 1.2944242187160016e-05, "loss": 0.2492, "step": 21368 }, { "epoch": 132.7289719626168, "grad_norm": 2.0, "learning_rate": 1.2942083880304074e-05, "loss": 0.1015, "step": 21369 }, { "epoch": 132.73520249221184, "grad_norm": 1.7578125, "learning_rate": 1.293992569055703e-05, "loss": 0.1509, "step": 21370 }, { "epoch": 132.74143302180684, "grad_norm": 2.125, "learning_rate": 1.2937767617939839e-05, "loss": 0.1248, "step": 21371 }, { "epoch": 132.74766355140187, "grad_norm": 2.859375, "learning_rate": 1.2935609662473452e-05, "loss": 0.256, "step": 21372 }, { "epoch": 132.75389408099687, "grad_norm": 1.484375, "learning_rate": 1.2933451824178838e-05, "loss": 0.1145, "step": 21373 }, { "epoch": 132.7601246105919, "grad_norm": 2.34375, "learning_rate": 1.2931294103076957e-05, "loss": 0.1701, "step": 21374 }, { "epoch": 132.7663551401869, "grad_norm": 2.484375, "learning_rate": 1.2929136499188754e-05, "loss": 0.1957, "step": 21375 }, { "epoch": 132.77258566978193, "grad_norm": 1.640625, "learning_rate": 1.2926979012535189e-05, "loss": 0.1098, "step": 21376 }, { "epoch": 132.77881619937693, "grad_norm": 2.609375, "learning_rate": 1.2924821643137225e-05, "loss": 0.1572, "step": 21377 }, { "epoch": 132.78504672897196, "grad_norm": 1.9765625, "learning_rate": 1.292266439101581e-05, "loss": 0.179, "step": 21378 }, { "epoch": 132.79127725856696, "grad_norm": 0.8046875, "learning_rate": 1.2920507256191883e-05, "loss": 0.0932, "step": 21379 }, { "epoch": 132.797507788162, "grad_norm": 1.359375, "learning_rate": 1.29183502386864e-05, "loss": 0.1356, "step": 21380 }, { "epoch": 132.80373831775702, "grad_norm": 2.21875, "learning_rate": 1.2916193338520327e-05, "loss": 0.2434, "step": 21381 }, { "epoch": 132.80996884735202, "grad_norm": 3.0, "learning_rate": 1.2914036555714592e-05, "loss": 0.2396, "step": 21382 }, { "epoch": 132.81619937694705, "grad_norm": 1.921875, "learning_rate": 1.2911879890290158e-05, "loss": 0.1373, "step": 21383 }, { "epoch": 132.82242990654206, "grad_norm": 1.4375, "learning_rate": 1.2909723342267954e-05, "loss": 0.1267, "step": 21384 }, { "epoch": 132.82866043613708, "grad_norm": 3.484375, "learning_rate": 1.2907566911668939e-05, "loss": 0.1783, "step": 21385 }, { "epoch": 132.8348909657321, "grad_norm": 1.578125, "learning_rate": 1.2905410598514045e-05, "loss": 0.0843, "step": 21386 }, { "epoch": 132.84112149532712, "grad_norm": 3.484375, "learning_rate": 1.2903254402824223e-05, "loss": 0.3607, "step": 21387 }, { "epoch": 132.84735202492212, "grad_norm": 1.734375, "learning_rate": 1.2901098324620409e-05, "loss": 0.1164, "step": 21388 }, { "epoch": 132.85358255451715, "grad_norm": 2.5, "learning_rate": 1.2898942363923561e-05, "loss": 0.202, "step": 21389 }, { "epoch": 132.85981308411215, "grad_norm": 1.4296875, "learning_rate": 1.28967865207546e-05, "loss": 0.1447, "step": 21390 }, { "epoch": 132.86604361370718, "grad_norm": 2.5, "learning_rate": 1.2894630795134455e-05, "loss": 0.1631, "step": 21391 }, { "epoch": 132.87227414330218, "grad_norm": 1.8984375, "learning_rate": 1.289247518708408e-05, "loss": 0.2163, "step": 21392 }, { "epoch": 132.8785046728972, "grad_norm": 1.0078125, "learning_rate": 1.2890319696624414e-05, "loss": 0.1038, "step": 21393 }, { "epoch": 132.8847352024922, "grad_norm": 1.21875, "learning_rate": 1.2888164323776375e-05, "loss": 0.1105, "step": 21394 }, { "epoch": 132.89096573208724, "grad_norm": 1.1484375, "learning_rate": 1.2886009068560906e-05, "loss": 0.1054, "step": 21395 }, { "epoch": 132.89719626168224, "grad_norm": 1.6328125, "learning_rate": 1.2883853930998946e-05, "loss": 0.1535, "step": 21396 }, { "epoch": 132.90342679127727, "grad_norm": 2.5, "learning_rate": 1.2881698911111424e-05, "loss": 0.15, "step": 21397 }, { "epoch": 132.90965732087227, "grad_norm": 2.015625, "learning_rate": 1.2879544008919248e-05, "loss": 0.0911, "step": 21398 }, { "epoch": 132.9158878504673, "grad_norm": 1.6484375, "learning_rate": 1.2877389224443365e-05, "loss": 0.1279, "step": 21399 }, { "epoch": 132.9221183800623, "grad_norm": 2.375, "learning_rate": 1.2875234557704714e-05, "loss": 0.1456, "step": 21400 }, { "epoch": 132.92834890965733, "grad_norm": 1.1796875, "learning_rate": 1.2873080008724198e-05, "loss": 0.0853, "step": 21401 }, { "epoch": 132.93457943925233, "grad_norm": 2.453125, "learning_rate": 1.287092557752275e-05, "loss": 0.2256, "step": 21402 }, { "epoch": 132.94080996884736, "grad_norm": 2.453125, "learning_rate": 1.286877126412131e-05, "loss": 0.1574, "step": 21403 }, { "epoch": 132.94704049844236, "grad_norm": 1.7265625, "learning_rate": 1.2866617068540788e-05, "loss": 0.15, "step": 21404 }, { "epoch": 132.9532710280374, "grad_norm": 1.890625, "learning_rate": 1.2864462990802095e-05, "loss": 0.1073, "step": 21405 }, { "epoch": 132.9595015576324, "grad_norm": 2.359375, "learning_rate": 1.2862309030926164e-05, "loss": 0.219, "step": 21406 }, { "epoch": 132.96573208722742, "grad_norm": 2.65625, "learning_rate": 1.2860155188933926e-05, "loss": 0.1587, "step": 21407 }, { "epoch": 132.97196261682242, "grad_norm": 1.78125, "learning_rate": 1.2858001464846275e-05, "loss": 0.2307, "step": 21408 }, { "epoch": 132.97819314641745, "grad_norm": 1.6796875, "learning_rate": 1.285584785868415e-05, "loss": 0.135, "step": 21409 }, { "epoch": 132.98442367601245, "grad_norm": 3.421875, "learning_rate": 1.2853694370468449e-05, "loss": 0.205, "step": 21410 }, { "epoch": 132.99065420560748, "grad_norm": 2.625, "learning_rate": 1.2851541000220108e-05, "loss": 0.1754, "step": 21411 }, { "epoch": 132.99688473520249, "grad_norm": 1.5625, "learning_rate": 1.2849387747960019e-05, "loss": 0.1247, "step": 21412 }, { "epoch": 133.0, "grad_norm": 1.8125, "learning_rate": 1.2847234613709105e-05, "loss": 0.0517, "step": 21413 }, { "epoch": 133.00623052959503, "grad_norm": 0.77734375, "learning_rate": 1.2845081597488287e-05, "loss": 0.0847, "step": 21414 }, { "epoch": 133.01246105919003, "grad_norm": 2.296875, "learning_rate": 1.2842928699318469e-05, "loss": 0.1921, "step": 21415 }, { "epoch": 133.01869158878506, "grad_norm": 1.6953125, "learning_rate": 1.2840775919220543e-05, "loss": 0.155, "step": 21416 }, { "epoch": 133.02492211838006, "grad_norm": 1.2265625, "learning_rate": 1.2838623257215431e-05, "loss": 0.1213, "step": 21417 }, { "epoch": 133.0311526479751, "grad_norm": 2.953125, "learning_rate": 1.2836470713324056e-05, "loss": 0.254, "step": 21418 }, { "epoch": 133.0373831775701, "grad_norm": 2.28125, "learning_rate": 1.2834318287567299e-05, "loss": 0.1809, "step": 21419 }, { "epoch": 133.04361370716512, "grad_norm": 1.7734375, "learning_rate": 1.283216597996607e-05, "loss": 0.1556, "step": 21420 }, { "epoch": 133.04984423676012, "grad_norm": 1.4453125, "learning_rate": 1.2830013790541279e-05, "loss": 0.1182, "step": 21421 }, { "epoch": 133.05607476635515, "grad_norm": 1.9375, "learning_rate": 1.282786171931385e-05, "loss": 0.2371, "step": 21422 }, { "epoch": 133.06230529595015, "grad_norm": 2.890625, "learning_rate": 1.2825709766304639e-05, "loss": 0.2126, "step": 21423 }, { "epoch": 133.06853582554518, "grad_norm": 1.1640625, "learning_rate": 1.2823557931534566e-05, "loss": 0.0839, "step": 21424 }, { "epoch": 133.07476635514018, "grad_norm": 1.1953125, "learning_rate": 1.2821406215024533e-05, "loss": 0.0933, "step": 21425 }, { "epoch": 133.0809968847352, "grad_norm": 2.4375, "learning_rate": 1.281925461679545e-05, "loss": 0.1837, "step": 21426 }, { "epoch": 133.08722741433021, "grad_norm": 2.40625, "learning_rate": 1.2817103136868191e-05, "loss": 0.1496, "step": 21427 }, { "epoch": 133.09345794392524, "grad_norm": 1.875, "learning_rate": 1.281495177526367e-05, "loss": 0.1243, "step": 21428 }, { "epoch": 133.09968847352025, "grad_norm": 1.2265625, "learning_rate": 1.2812800532002763e-05, "loss": 0.1035, "step": 21429 }, { "epoch": 133.10591900311528, "grad_norm": 1.640625, "learning_rate": 1.2810649407106384e-05, "loss": 0.1345, "step": 21430 }, { "epoch": 133.11214953271028, "grad_norm": 1.171875, "learning_rate": 1.2808498400595404e-05, "loss": 0.1202, "step": 21431 }, { "epoch": 133.1183800623053, "grad_norm": 1.515625, "learning_rate": 1.2806347512490724e-05, "loss": 0.0909, "step": 21432 }, { "epoch": 133.1246105919003, "grad_norm": 1.734375, "learning_rate": 1.2804196742813245e-05, "loss": 0.1379, "step": 21433 }, { "epoch": 133.13084112149534, "grad_norm": 2.6875, "learning_rate": 1.2802046091583846e-05, "loss": 0.2312, "step": 21434 }, { "epoch": 133.13707165109034, "grad_norm": 0.92578125, "learning_rate": 1.2799895558823404e-05, "loss": 0.0995, "step": 21435 }, { "epoch": 133.14330218068537, "grad_norm": 2.421875, "learning_rate": 1.2797745144552814e-05, "loss": 0.2064, "step": 21436 }, { "epoch": 133.14953271028037, "grad_norm": 1.046875, "learning_rate": 1.2795594848792975e-05, "loss": 0.1121, "step": 21437 }, { "epoch": 133.1557632398754, "grad_norm": 2.15625, "learning_rate": 1.2793444671564745e-05, "loss": 0.1342, "step": 21438 }, { "epoch": 133.1619937694704, "grad_norm": 3.53125, "learning_rate": 1.2791294612889021e-05, "loss": 0.3367, "step": 21439 }, { "epoch": 133.16822429906543, "grad_norm": 2.21875, "learning_rate": 1.2789144672786696e-05, "loss": 0.1239, "step": 21440 }, { "epoch": 133.17445482866043, "grad_norm": 2.5, "learning_rate": 1.2786994851278636e-05, "loss": 0.1096, "step": 21441 }, { "epoch": 133.18068535825546, "grad_norm": 2.390625, "learning_rate": 1.2784845148385715e-05, "loss": 0.1585, "step": 21442 }, { "epoch": 133.18691588785046, "grad_norm": 1.1875, "learning_rate": 1.2782695564128822e-05, "loss": 0.109, "step": 21443 }, { "epoch": 133.1931464174455, "grad_norm": 2.421875, "learning_rate": 1.2780546098528837e-05, "loss": 0.2075, "step": 21444 }, { "epoch": 133.1993769470405, "grad_norm": 2.09375, "learning_rate": 1.2778396751606626e-05, "loss": 0.1663, "step": 21445 }, { "epoch": 133.20560747663552, "grad_norm": 1.8671875, "learning_rate": 1.2776247523383064e-05, "loss": 0.1758, "step": 21446 }, { "epoch": 133.21183800623052, "grad_norm": 2.453125, "learning_rate": 1.2774098413879043e-05, "loss": 0.1726, "step": 21447 }, { "epoch": 133.21806853582555, "grad_norm": 1.5703125, "learning_rate": 1.277194942311542e-05, "loss": 0.141, "step": 21448 }, { "epoch": 133.22429906542055, "grad_norm": 2.65625, "learning_rate": 1.2769800551113059e-05, "loss": 0.2208, "step": 21449 }, { "epoch": 133.23052959501558, "grad_norm": 2.046875, "learning_rate": 1.2767651797892838e-05, "loss": 0.1235, "step": 21450 }, { "epoch": 133.23676012461058, "grad_norm": 1.375, "learning_rate": 1.276550316347563e-05, "loss": 0.1104, "step": 21451 }, { "epoch": 133.2429906542056, "grad_norm": 0.97265625, "learning_rate": 1.2763354647882314e-05, "loss": 0.1101, "step": 21452 }, { "epoch": 133.2492211838006, "grad_norm": 2.890625, "learning_rate": 1.276120625113374e-05, "loss": 0.2638, "step": 21453 }, { "epoch": 133.25545171339564, "grad_norm": 2.78125, "learning_rate": 1.2759057973250765e-05, "loss": 0.1796, "step": 21454 }, { "epoch": 133.26168224299064, "grad_norm": 1.515625, "learning_rate": 1.2756909814254269e-05, "loss": 0.1442, "step": 21455 }, { "epoch": 133.26791277258567, "grad_norm": 3.078125, "learning_rate": 1.2754761774165126e-05, "loss": 0.211, "step": 21456 }, { "epoch": 133.27414330218068, "grad_norm": 1.203125, "learning_rate": 1.275261385300417e-05, "loss": 0.0929, "step": 21457 }, { "epoch": 133.2803738317757, "grad_norm": 1.9375, "learning_rate": 1.2750466050792276e-05, "loss": 0.1848, "step": 21458 }, { "epoch": 133.2866043613707, "grad_norm": 2.515625, "learning_rate": 1.2748318367550315e-05, "loss": 0.2111, "step": 21459 }, { "epoch": 133.29283489096574, "grad_norm": 2.546875, "learning_rate": 1.2746170803299135e-05, "loss": 0.2013, "step": 21460 }, { "epoch": 133.29906542056074, "grad_norm": 1.2109375, "learning_rate": 1.2744023358059583e-05, "loss": 0.1131, "step": 21461 }, { "epoch": 133.30529595015577, "grad_norm": 2.703125, "learning_rate": 1.2741876031852528e-05, "loss": 0.2111, "step": 21462 }, { "epoch": 133.31152647975077, "grad_norm": 2.078125, "learning_rate": 1.2739728824698832e-05, "loss": 0.1905, "step": 21463 }, { "epoch": 133.3177570093458, "grad_norm": 2.15625, "learning_rate": 1.273758173661933e-05, "loss": 0.1446, "step": 21464 }, { "epoch": 133.3239875389408, "grad_norm": 3.40625, "learning_rate": 1.273543476763489e-05, "loss": 0.1724, "step": 21465 }, { "epoch": 133.33021806853583, "grad_norm": 1.5703125, "learning_rate": 1.2733287917766363e-05, "loss": 0.1377, "step": 21466 }, { "epoch": 133.33644859813083, "grad_norm": 2.515625, "learning_rate": 1.2731141187034602e-05, "loss": 0.1526, "step": 21467 }, { "epoch": 133.34267912772586, "grad_norm": 0.765625, "learning_rate": 1.2728994575460435e-05, "loss": 0.0883, "step": 21468 }, { "epoch": 133.34890965732086, "grad_norm": 2.203125, "learning_rate": 1.2726848083064728e-05, "loss": 0.1415, "step": 21469 }, { "epoch": 133.3551401869159, "grad_norm": 2.3125, "learning_rate": 1.2724701709868333e-05, "loss": 0.1234, "step": 21470 }, { "epoch": 133.3613707165109, "grad_norm": 2.828125, "learning_rate": 1.2722555455892083e-05, "loss": 0.2312, "step": 21471 }, { "epoch": 133.36760124610592, "grad_norm": 1.5, "learning_rate": 1.2720409321156824e-05, "loss": 0.1206, "step": 21472 }, { "epoch": 133.37383177570092, "grad_norm": 2.59375, "learning_rate": 1.2718263305683417e-05, "loss": 0.221, "step": 21473 }, { "epoch": 133.38006230529595, "grad_norm": 1.7109375, "learning_rate": 1.271611740949269e-05, "loss": 0.1288, "step": 21474 }, { "epoch": 133.38629283489095, "grad_norm": 1.15625, "learning_rate": 1.2713971632605479e-05, "loss": 0.1145, "step": 21475 }, { "epoch": 133.39252336448598, "grad_norm": 2.359375, "learning_rate": 1.271182597504263e-05, "loss": 0.1931, "step": 21476 }, { "epoch": 133.39875389408098, "grad_norm": 1.9140625, "learning_rate": 1.2709680436824994e-05, "loss": 0.1567, "step": 21477 }, { "epoch": 133.404984423676, "grad_norm": 2.703125, "learning_rate": 1.2707535017973388e-05, "loss": 0.2417, "step": 21478 }, { "epoch": 133.41121495327104, "grad_norm": 1.1640625, "learning_rate": 1.2705389718508671e-05, "loss": 0.1081, "step": 21479 }, { "epoch": 133.41744548286604, "grad_norm": 3.875, "learning_rate": 1.2703244538451653e-05, "loss": 0.1957, "step": 21480 }, { "epoch": 133.42367601246107, "grad_norm": 1.546875, "learning_rate": 1.2701099477823195e-05, "loss": 0.1299, "step": 21481 }, { "epoch": 133.42990654205607, "grad_norm": 2.53125, "learning_rate": 1.2698954536644106e-05, "loss": 0.2524, "step": 21482 }, { "epoch": 133.4361370716511, "grad_norm": 2.625, "learning_rate": 1.2696809714935228e-05, "loss": 0.2222, "step": 21483 }, { "epoch": 133.4423676012461, "grad_norm": 0.80078125, "learning_rate": 1.2694665012717397e-05, "loss": 0.0903, "step": 21484 }, { "epoch": 133.44859813084113, "grad_norm": 2.9375, "learning_rate": 1.2692520430011457e-05, "loss": 0.2633, "step": 21485 }, { "epoch": 133.45482866043614, "grad_norm": 2.65625, "learning_rate": 1.2690375966838197e-05, "loss": 0.4246, "step": 21486 }, { "epoch": 133.46105919003116, "grad_norm": 2.140625, "learning_rate": 1.2688231623218472e-05, "loss": 0.1642, "step": 21487 }, { "epoch": 133.46728971962617, "grad_norm": 0.87890625, "learning_rate": 1.2686087399173097e-05, "loss": 0.085, "step": 21488 }, { "epoch": 133.4735202492212, "grad_norm": 1.953125, "learning_rate": 1.2683943294722917e-05, "loss": 0.1309, "step": 21489 }, { "epoch": 133.4797507788162, "grad_norm": 1.640625, "learning_rate": 1.2681799309888728e-05, "loss": 0.1287, "step": 21490 }, { "epoch": 133.48598130841123, "grad_norm": 1.453125, "learning_rate": 1.2679655444691369e-05, "loss": 0.1173, "step": 21491 }, { "epoch": 133.49221183800623, "grad_norm": 2.453125, "learning_rate": 1.2677511699151668e-05, "loss": 0.1229, "step": 21492 }, { "epoch": 133.49844236760126, "grad_norm": 2.734375, "learning_rate": 1.2675368073290434e-05, "loss": 0.1819, "step": 21493 }, { "epoch": 133.50467289719626, "grad_norm": 2.34375, "learning_rate": 1.2673224567128484e-05, "loss": 0.1741, "step": 21494 }, { "epoch": 133.5109034267913, "grad_norm": 2.453125, "learning_rate": 1.2671081180686638e-05, "loss": 0.1576, "step": 21495 }, { "epoch": 133.5171339563863, "grad_norm": 1.609375, "learning_rate": 1.2668937913985726e-05, "loss": 0.1207, "step": 21496 }, { "epoch": 133.52336448598132, "grad_norm": 1.5859375, "learning_rate": 1.2666794767046544e-05, "loss": 0.1176, "step": 21497 }, { "epoch": 133.52959501557632, "grad_norm": 3.65625, "learning_rate": 1.2664651739889927e-05, "loss": 0.1282, "step": 21498 }, { "epoch": 133.53582554517135, "grad_norm": 1.8203125, "learning_rate": 1.2662508832536668e-05, "loss": 0.1229, "step": 21499 }, { "epoch": 133.54205607476635, "grad_norm": 1.3984375, "learning_rate": 1.2660366045007599e-05, "loss": 0.1294, "step": 21500 }, { "epoch": 133.54828660436138, "grad_norm": 2.875, "learning_rate": 1.2658223377323513e-05, "loss": 0.1799, "step": 21501 }, { "epoch": 133.55451713395638, "grad_norm": 2.4375, "learning_rate": 1.265608082950523e-05, "loss": 0.1726, "step": 21502 }, { "epoch": 133.5607476635514, "grad_norm": 2.484375, "learning_rate": 1.2653938401573562e-05, "loss": 0.2218, "step": 21503 }, { "epoch": 133.5669781931464, "grad_norm": 1.390625, "learning_rate": 1.2651796093549318e-05, "loss": 0.129, "step": 21504 }, { "epoch": 133.57320872274144, "grad_norm": 1.4453125, "learning_rate": 1.2649653905453284e-05, "loss": 0.1366, "step": 21505 }, { "epoch": 133.57943925233644, "grad_norm": 1.78125, "learning_rate": 1.2647511837306284e-05, "loss": 0.1412, "step": 21506 }, { "epoch": 133.58566978193147, "grad_norm": 2.078125, "learning_rate": 1.2645369889129127e-05, "loss": 0.1204, "step": 21507 }, { "epoch": 133.59190031152647, "grad_norm": 1.1484375, "learning_rate": 1.2643228060942597e-05, "loss": 0.1121, "step": 21508 }, { "epoch": 133.5981308411215, "grad_norm": 1.6953125, "learning_rate": 1.2641086352767506e-05, "loss": 0.0919, "step": 21509 }, { "epoch": 133.6043613707165, "grad_norm": 2.21875, "learning_rate": 1.2638944764624663e-05, "loss": 0.2107, "step": 21510 }, { "epoch": 133.61059190031153, "grad_norm": 1.75, "learning_rate": 1.263680329653486e-05, "loss": 0.1369, "step": 21511 }, { "epoch": 133.61682242990653, "grad_norm": 3.171875, "learning_rate": 1.2634661948518888e-05, "loss": 0.3366, "step": 21512 }, { "epoch": 133.62305295950156, "grad_norm": 1.9921875, "learning_rate": 1.2632520720597548e-05, "loss": 0.1747, "step": 21513 }, { "epoch": 133.62928348909657, "grad_norm": 1.4609375, "learning_rate": 1.2630379612791648e-05, "loss": 0.1448, "step": 21514 }, { "epoch": 133.6355140186916, "grad_norm": 1.3828125, "learning_rate": 1.2628238625121968e-05, "loss": 0.0995, "step": 21515 }, { "epoch": 133.6417445482866, "grad_norm": 2.484375, "learning_rate": 1.2626097757609307e-05, "loss": 0.2403, "step": 21516 }, { "epoch": 133.64797507788163, "grad_norm": 1.03125, "learning_rate": 1.2623957010274463e-05, "loss": 0.1146, "step": 21517 }, { "epoch": 133.65420560747663, "grad_norm": 1.6015625, "learning_rate": 1.2621816383138227e-05, "loss": 0.1115, "step": 21518 }, { "epoch": 133.66043613707166, "grad_norm": 2.109375, "learning_rate": 1.2619675876221371e-05, "loss": 0.1749, "step": 21519 }, { "epoch": 133.66666666666666, "grad_norm": 2.140625, "learning_rate": 1.2617535489544703e-05, "loss": 0.1027, "step": 21520 }, { "epoch": 133.6728971962617, "grad_norm": 1.1484375, "learning_rate": 1.2615395223129001e-05, "loss": 0.1149, "step": 21521 }, { "epoch": 133.6791277258567, "grad_norm": 1.609375, "learning_rate": 1.2613255076995068e-05, "loss": 0.1428, "step": 21522 }, { "epoch": 133.68535825545172, "grad_norm": 1.84375, "learning_rate": 1.261111505116367e-05, "loss": 0.116, "step": 21523 }, { "epoch": 133.69158878504672, "grad_norm": 1.8046875, "learning_rate": 1.2608975145655605e-05, "loss": 0.1254, "step": 21524 }, { "epoch": 133.69781931464175, "grad_norm": 2.296875, "learning_rate": 1.2606835360491642e-05, "loss": 0.1836, "step": 21525 }, { "epoch": 133.70404984423675, "grad_norm": 1.2578125, "learning_rate": 1.260469569569258e-05, "loss": 0.1439, "step": 21526 }, { "epoch": 133.71028037383178, "grad_norm": 2.46875, "learning_rate": 1.2602556151279183e-05, "loss": 0.1248, "step": 21527 }, { "epoch": 133.71651090342678, "grad_norm": 2.296875, "learning_rate": 1.2600416727272237e-05, "loss": 0.2996, "step": 21528 }, { "epoch": 133.7227414330218, "grad_norm": 3.046875, "learning_rate": 1.2598277423692534e-05, "loss": 0.2491, "step": 21529 }, { "epoch": 133.7289719626168, "grad_norm": 1.46875, "learning_rate": 1.259613824056084e-05, "loss": 0.1223, "step": 21530 }, { "epoch": 133.73520249221184, "grad_norm": 2.359375, "learning_rate": 1.259399917789792e-05, "loss": 0.1014, "step": 21531 }, { "epoch": 133.74143302180684, "grad_norm": 2.25, "learning_rate": 1.2591860235724563e-05, "loss": 0.1362, "step": 21532 }, { "epoch": 133.74766355140187, "grad_norm": 1.3984375, "learning_rate": 1.2589721414061544e-05, "loss": 0.127, "step": 21533 }, { "epoch": 133.75389408099687, "grad_norm": 2.234375, "learning_rate": 1.2587582712929624e-05, "loss": 0.2124, "step": 21534 }, { "epoch": 133.7601246105919, "grad_norm": 2.53125, "learning_rate": 1.2585444132349583e-05, "loss": 0.1072, "step": 21535 }, { "epoch": 133.7663551401869, "grad_norm": 1.71875, "learning_rate": 1.25833056723422e-05, "loss": 0.1356, "step": 21536 }, { "epoch": 133.77258566978193, "grad_norm": 2.15625, "learning_rate": 1.2581167332928232e-05, "loss": 0.2054, "step": 21537 }, { "epoch": 133.77881619937693, "grad_norm": 2.421875, "learning_rate": 1.2579029114128444e-05, "loss": 0.1433, "step": 21538 }, { "epoch": 133.78504672897196, "grad_norm": 2.21875, "learning_rate": 1.2576891015963605e-05, "loss": 0.1051, "step": 21539 }, { "epoch": 133.79127725856696, "grad_norm": 1.7578125, "learning_rate": 1.257475303845449e-05, "loss": 0.1067, "step": 21540 }, { "epoch": 133.797507788162, "grad_norm": 2.234375, "learning_rate": 1.2572615181621853e-05, "loss": 0.1187, "step": 21541 }, { "epoch": 133.80373831775702, "grad_norm": 2.203125, "learning_rate": 1.2570477445486462e-05, "loss": 0.1542, "step": 21542 }, { "epoch": 133.80996884735202, "grad_norm": 2.1875, "learning_rate": 1.2568339830069084e-05, "loss": 0.19, "step": 21543 }, { "epoch": 133.81619937694705, "grad_norm": 2.375, "learning_rate": 1.2566202335390475e-05, "loss": 0.2215, "step": 21544 }, { "epoch": 133.82242990654206, "grad_norm": 3.921875, "learning_rate": 1.2564064961471388e-05, "loss": 0.3521, "step": 21545 }, { "epoch": 133.82866043613708, "grad_norm": 3.265625, "learning_rate": 1.2561927708332588e-05, "loss": 0.1965, "step": 21546 }, { "epoch": 133.8348909657321, "grad_norm": 1.921875, "learning_rate": 1.2559790575994829e-05, "loss": 0.1054, "step": 21547 }, { "epoch": 133.84112149532712, "grad_norm": 2.375, "learning_rate": 1.2557653564478883e-05, "loss": 0.1747, "step": 21548 }, { "epoch": 133.84735202492212, "grad_norm": 2.390625, "learning_rate": 1.2555516673805493e-05, "loss": 0.1995, "step": 21549 }, { "epoch": 133.85358255451715, "grad_norm": 1.9453125, "learning_rate": 1.25533799039954e-05, "loss": 0.2034, "step": 21550 }, { "epoch": 133.85981308411215, "grad_norm": 2.125, "learning_rate": 1.2551243255069373e-05, "loss": 0.1429, "step": 21551 }, { "epoch": 133.86604361370718, "grad_norm": 2.96875, "learning_rate": 1.2549106727048165e-05, "loss": 0.2417, "step": 21552 }, { "epoch": 133.87227414330218, "grad_norm": 1.578125, "learning_rate": 1.2546970319952516e-05, "loss": 0.0979, "step": 21553 }, { "epoch": 133.8785046728972, "grad_norm": 1.578125, "learning_rate": 1.2544834033803183e-05, "loss": 0.0873, "step": 21554 }, { "epoch": 133.8847352024922, "grad_norm": 1.3828125, "learning_rate": 1.2542697868620915e-05, "loss": 0.1253, "step": 21555 }, { "epoch": 133.89096573208724, "grad_norm": 2.359375, "learning_rate": 1.2540561824426458e-05, "loss": 0.1966, "step": 21556 }, { "epoch": 133.89719626168224, "grad_norm": 2.84375, "learning_rate": 1.2538425901240547e-05, "loss": 0.2297, "step": 21557 }, { "epoch": 133.90342679127727, "grad_norm": 1.9453125, "learning_rate": 1.253629009908393e-05, "loss": 0.1685, "step": 21558 }, { "epoch": 133.90965732087227, "grad_norm": 1.5546875, "learning_rate": 1.253415441797737e-05, "loss": 0.1357, "step": 21559 }, { "epoch": 133.9158878504673, "grad_norm": 1.828125, "learning_rate": 1.2532018857941582e-05, "loss": 0.144, "step": 21560 }, { "epoch": 133.9221183800623, "grad_norm": 1.96875, "learning_rate": 1.2529883418997318e-05, "loss": 0.1996, "step": 21561 }, { "epoch": 133.92834890965733, "grad_norm": 1.703125, "learning_rate": 1.2527748101165333e-05, "loss": 0.1034, "step": 21562 }, { "epoch": 133.93457943925233, "grad_norm": 2.296875, "learning_rate": 1.252561290446635e-05, "loss": 0.1225, "step": 21563 }, { "epoch": 133.94080996884736, "grad_norm": 2.03125, "learning_rate": 1.2523477828921093e-05, "loss": 0.1261, "step": 21564 }, { "epoch": 133.94704049844236, "grad_norm": 3.421875, "learning_rate": 1.2521342874550318e-05, "loss": 0.3124, "step": 21565 }, { "epoch": 133.9532710280374, "grad_norm": 1.4140625, "learning_rate": 1.2519208041374767e-05, "loss": 0.1061, "step": 21566 }, { "epoch": 133.9595015576324, "grad_norm": 1.84375, "learning_rate": 1.2517073329415149e-05, "loss": 0.1716, "step": 21567 }, { "epoch": 133.96573208722742, "grad_norm": 0.9765625, "learning_rate": 1.2514938738692222e-05, "loss": 0.1051, "step": 21568 }, { "epoch": 133.97196261682242, "grad_norm": 2.359375, "learning_rate": 1.2512804269226694e-05, "loss": 0.1738, "step": 21569 }, { "epoch": 133.97819314641745, "grad_norm": 1.2265625, "learning_rate": 1.251066992103932e-05, "loss": 0.1292, "step": 21570 }, { "epoch": 133.98442367601245, "grad_norm": 2.0, "learning_rate": 1.2508535694150803e-05, "loss": 0.1736, "step": 21571 }, { "epoch": 133.99065420560748, "grad_norm": 2.03125, "learning_rate": 1.2506401588581884e-05, "loss": 0.1436, "step": 21572 }, { "epoch": 133.99688473520249, "grad_norm": 1.796875, "learning_rate": 1.2504267604353303e-05, "loss": 0.1371, "step": 21573 }, { "epoch": 134.0, "grad_norm": 0.734375, "learning_rate": 1.2502133741485771e-05, "loss": 0.0516, "step": 21574 }, { "epoch": 134.00623052959503, "grad_norm": 1.671875, "learning_rate": 1.2500000000000006e-05, "loss": 0.0959, "step": 21575 }, { "epoch": 134.01246105919003, "grad_norm": 2.375, "learning_rate": 1.249786637991674e-05, "loss": 0.1874, "step": 21576 }, { "epoch": 134.01869158878506, "grad_norm": 2.625, "learning_rate": 1.2495732881256702e-05, "loss": 0.1846, "step": 21577 }, { "epoch": 134.02492211838006, "grad_norm": 2.8125, "learning_rate": 1.2493599504040598e-05, "loss": 0.1683, "step": 21578 }, { "epoch": 134.0311526479751, "grad_norm": 2.671875, "learning_rate": 1.2491466248289154e-05, "loss": 0.2033, "step": 21579 }, { "epoch": 134.0373831775701, "grad_norm": 2.296875, "learning_rate": 1.248933311402309e-05, "loss": 0.1341, "step": 21580 }, { "epoch": 134.04361370716512, "grad_norm": 3.140625, "learning_rate": 1.2487200101263146e-05, "loss": 0.3155, "step": 21581 }, { "epoch": 134.04984423676012, "grad_norm": 2.453125, "learning_rate": 1.2485067210029996e-05, "loss": 0.1139, "step": 21582 }, { "epoch": 134.05607476635515, "grad_norm": 3.296875, "learning_rate": 1.2482934440344374e-05, "loss": 0.2831, "step": 21583 }, { "epoch": 134.06230529595015, "grad_norm": 2.5, "learning_rate": 1.2480801792226992e-05, "loss": 0.1905, "step": 21584 }, { "epoch": 134.06853582554518, "grad_norm": 1.5859375, "learning_rate": 1.2478669265698579e-05, "loss": 0.1231, "step": 21585 }, { "epoch": 134.07476635514018, "grad_norm": 1.90625, "learning_rate": 1.2476536860779823e-05, "loss": 0.1434, "step": 21586 }, { "epoch": 134.0809968847352, "grad_norm": 2.546875, "learning_rate": 1.2474404577491452e-05, "loss": 0.2128, "step": 21587 }, { "epoch": 134.08722741433021, "grad_norm": 1.71875, "learning_rate": 1.247227241585416e-05, "loss": 0.1432, "step": 21588 }, { "epoch": 134.09345794392524, "grad_norm": 2.828125, "learning_rate": 1.2470140375888672e-05, "loss": 0.1947, "step": 21589 }, { "epoch": 134.09968847352025, "grad_norm": 1.375, "learning_rate": 1.2468008457615674e-05, "loss": 0.1403, "step": 21590 }, { "epoch": 134.10591900311528, "grad_norm": 2.078125, "learning_rate": 1.2465876661055884e-05, "loss": 0.1211, "step": 21591 }, { "epoch": 134.11214953271028, "grad_norm": 2.125, "learning_rate": 1.2463744986230016e-05, "loss": 0.178, "step": 21592 }, { "epoch": 134.1183800623053, "grad_norm": 1.1171875, "learning_rate": 1.2461613433158753e-05, "loss": 0.1077, "step": 21593 }, { "epoch": 134.1246105919003, "grad_norm": 1.3359375, "learning_rate": 1.2459482001862812e-05, "loss": 0.1308, "step": 21594 }, { "epoch": 134.13084112149534, "grad_norm": 1.4609375, "learning_rate": 1.2457350692362885e-05, "loss": 0.1333, "step": 21595 }, { "epoch": 134.13707165109034, "grad_norm": 3.265625, "learning_rate": 1.2455219504679683e-05, "loss": 0.2839, "step": 21596 }, { "epoch": 134.14330218068537, "grad_norm": 2.734375, "learning_rate": 1.2453088438833885e-05, "loss": 0.2138, "step": 21597 }, { "epoch": 134.14953271028037, "grad_norm": 2.375, "learning_rate": 1.2450957494846203e-05, "loss": 0.1911, "step": 21598 }, { "epoch": 134.1557632398754, "grad_norm": 2.0625, "learning_rate": 1.2448826672737337e-05, "loss": 0.1209, "step": 21599 }, { "epoch": 134.1619937694704, "grad_norm": 2.453125, "learning_rate": 1.2446695972527977e-05, "loss": 0.192, "step": 21600 }, { "epoch": 134.16822429906543, "grad_norm": 2.609375, "learning_rate": 1.2444565394238806e-05, "loss": 0.2262, "step": 21601 }, { "epoch": 134.17445482866043, "grad_norm": 2.859375, "learning_rate": 1.2442434937890524e-05, "loss": 0.2, "step": 21602 }, { "epoch": 134.18068535825546, "grad_norm": 2.453125, "learning_rate": 1.2440304603503836e-05, "loss": 0.1826, "step": 21603 }, { "epoch": 134.18691588785046, "grad_norm": 1.6796875, "learning_rate": 1.2438174391099408e-05, "loss": 0.1069, "step": 21604 }, { "epoch": 134.1931464174455, "grad_norm": 2.796875, "learning_rate": 1.2436044300697944e-05, "loss": 0.2022, "step": 21605 }, { "epoch": 134.1993769470405, "grad_norm": 2.90625, "learning_rate": 1.2433914332320137e-05, "loss": 0.2666, "step": 21606 }, { "epoch": 134.20560747663552, "grad_norm": 1.4453125, "learning_rate": 1.2431784485986666e-05, "loss": 0.0964, "step": 21607 }, { "epoch": 134.21183800623052, "grad_norm": 2.984375, "learning_rate": 1.2429654761718206e-05, "loss": 0.2172, "step": 21608 }, { "epoch": 134.21806853582555, "grad_norm": 2.90625, "learning_rate": 1.2427525159535455e-05, "loss": 0.2606, "step": 21609 }, { "epoch": 134.22429906542055, "grad_norm": 1.3984375, "learning_rate": 1.2425395679459101e-05, "loss": 0.1314, "step": 21610 }, { "epoch": 134.23052959501558, "grad_norm": 1.921875, "learning_rate": 1.2423266321509807e-05, "loss": 0.153, "step": 21611 }, { "epoch": 134.23676012461058, "grad_norm": 1.5, "learning_rate": 1.2421137085708265e-05, "loss": 0.1303, "step": 21612 }, { "epoch": 134.2429906542056, "grad_norm": 2.59375, "learning_rate": 1.2419007972075165e-05, "loss": 0.2186, "step": 21613 }, { "epoch": 134.2492211838006, "grad_norm": 2.09375, "learning_rate": 1.2416878980631173e-05, "loss": 0.1543, "step": 21614 }, { "epoch": 134.25545171339564, "grad_norm": 0.76171875, "learning_rate": 1.2414750111396959e-05, "loss": 0.0865, "step": 21615 }, { "epoch": 134.26168224299064, "grad_norm": 1.828125, "learning_rate": 1.2412621364393206e-05, "loss": 0.1488, "step": 21616 }, { "epoch": 134.26791277258567, "grad_norm": 3.09375, "learning_rate": 1.2410492739640592e-05, "loss": 0.1973, "step": 21617 }, { "epoch": 134.27414330218068, "grad_norm": 2.453125, "learning_rate": 1.2408364237159797e-05, "loss": 0.1389, "step": 21618 }, { "epoch": 134.2803738317757, "grad_norm": 1.234375, "learning_rate": 1.2406235856971487e-05, "loss": 0.1281, "step": 21619 }, { "epoch": 134.2866043613707, "grad_norm": 2.859375, "learning_rate": 1.2404107599096323e-05, "loss": 0.2403, "step": 21620 }, { "epoch": 134.29283489096574, "grad_norm": 2.296875, "learning_rate": 1.2401979463554982e-05, "loss": 0.2398, "step": 21621 }, { "epoch": 134.29906542056074, "grad_norm": 2.4375, "learning_rate": 1.2399851450368147e-05, "loss": 0.2102, "step": 21622 }, { "epoch": 134.30529595015577, "grad_norm": 1.984375, "learning_rate": 1.2397723559556463e-05, "loss": 0.1489, "step": 21623 }, { "epoch": 134.31152647975077, "grad_norm": 2.34375, "learning_rate": 1.2395595791140604e-05, "loss": 0.1286, "step": 21624 }, { "epoch": 134.3177570093458, "grad_norm": 1.5546875, "learning_rate": 1.2393468145141249e-05, "loss": 0.1247, "step": 21625 }, { "epoch": 134.3239875389408, "grad_norm": 2.640625, "learning_rate": 1.2391340621579054e-05, "loss": 0.1298, "step": 21626 }, { "epoch": 134.33021806853583, "grad_norm": 1.1015625, "learning_rate": 1.2389213220474665e-05, "loss": 0.1033, "step": 21627 }, { "epoch": 134.33644859813083, "grad_norm": 2.8125, "learning_rate": 1.2387085941848764e-05, "loss": 0.2731, "step": 21628 }, { "epoch": 134.34267912772586, "grad_norm": 1.5390625, "learning_rate": 1.2384958785722012e-05, "loss": 0.1293, "step": 21629 }, { "epoch": 134.34890965732086, "grad_norm": 1.9765625, "learning_rate": 1.2382831752115051e-05, "loss": 0.1564, "step": 21630 }, { "epoch": 134.3551401869159, "grad_norm": 1.9609375, "learning_rate": 1.2380704841048552e-05, "loss": 0.1179, "step": 21631 }, { "epoch": 134.3613707165109, "grad_norm": 0.9140625, "learning_rate": 1.237857805254318e-05, "loss": 0.0939, "step": 21632 }, { "epoch": 134.36760124610592, "grad_norm": 1.28125, "learning_rate": 1.2376451386619581e-05, "loss": 0.1029, "step": 21633 }, { "epoch": 134.37383177570092, "grad_norm": 2.859375, "learning_rate": 1.23743248432984e-05, "loss": 0.2455, "step": 21634 }, { "epoch": 134.38006230529595, "grad_norm": 1.7578125, "learning_rate": 1.2372198422600298e-05, "loss": 0.105, "step": 21635 }, { "epoch": 134.38629283489095, "grad_norm": 3.15625, "learning_rate": 1.237007212454594e-05, "loss": 0.2464, "step": 21636 }, { "epoch": 134.39252336448598, "grad_norm": 1.9453125, "learning_rate": 1.2367945949155956e-05, "loss": 0.1591, "step": 21637 }, { "epoch": 134.39875389408098, "grad_norm": 1.0, "learning_rate": 1.2365819896451018e-05, "loss": 0.0955, "step": 21638 }, { "epoch": 134.404984423676, "grad_norm": 1.9765625, "learning_rate": 1.2363693966451751e-05, "loss": 0.1448, "step": 21639 }, { "epoch": 134.41121495327104, "grad_norm": 2.609375, "learning_rate": 1.2361568159178822e-05, "loss": 0.2088, "step": 21640 }, { "epoch": 134.41744548286604, "grad_norm": 1.734375, "learning_rate": 1.2359442474652863e-05, "loss": 0.1348, "step": 21641 }, { "epoch": 134.42367601246107, "grad_norm": 1.125, "learning_rate": 1.2357316912894522e-05, "loss": 0.1224, "step": 21642 }, { "epoch": 134.42990654205607, "grad_norm": 1.828125, "learning_rate": 1.2355191473924448e-05, "loss": 0.267, "step": 21643 }, { "epoch": 134.4361370716511, "grad_norm": 2.6875, "learning_rate": 1.2353066157763304e-05, "loss": 0.3226, "step": 21644 }, { "epoch": 134.4423676012461, "grad_norm": 2.671875, "learning_rate": 1.2350940964431687e-05, "loss": 0.127, "step": 21645 }, { "epoch": 134.44859813084113, "grad_norm": 0.7890625, "learning_rate": 1.2348815893950264e-05, "loss": 0.0946, "step": 21646 }, { "epoch": 134.45482866043614, "grad_norm": 3.046875, "learning_rate": 1.2346690946339665e-05, "loss": 0.1454, "step": 21647 }, { "epoch": 134.46105919003116, "grad_norm": 2.28125, "learning_rate": 1.2344566121620545e-05, "loss": 0.1477, "step": 21648 }, { "epoch": 134.46728971962617, "grad_norm": 1.390625, "learning_rate": 1.234244141981352e-05, "loss": 0.1081, "step": 21649 }, { "epoch": 134.4735202492212, "grad_norm": 2.15625, "learning_rate": 1.2340316840939231e-05, "loss": 0.1051, "step": 21650 }, { "epoch": 134.4797507788162, "grad_norm": 3.484375, "learning_rate": 1.2338192385018326e-05, "loss": 0.3303, "step": 21651 }, { "epoch": 134.48598130841123, "grad_norm": 1.5546875, "learning_rate": 1.2336068052071428e-05, "loss": 0.1337, "step": 21652 }, { "epoch": 134.49221183800623, "grad_norm": 2.4375, "learning_rate": 1.2333943842119162e-05, "loss": 0.2713, "step": 21653 }, { "epoch": 134.49844236760126, "grad_norm": 3.296875, "learning_rate": 1.2331819755182161e-05, "loss": 0.1053, "step": 21654 }, { "epoch": 134.50467289719626, "grad_norm": 2.59375, "learning_rate": 1.232969579128107e-05, "loss": 0.1731, "step": 21655 }, { "epoch": 134.5109034267913, "grad_norm": 1.90625, "learning_rate": 1.2327571950436498e-05, "loss": 0.2139, "step": 21656 }, { "epoch": 134.5171339563863, "grad_norm": 2.15625, "learning_rate": 1.2325448232669087e-05, "loss": 0.1618, "step": 21657 }, { "epoch": 134.52336448598132, "grad_norm": 1.84375, "learning_rate": 1.2323324637999451e-05, "loss": 0.1811, "step": 21658 }, { "epoch": 134.52959501557632, "grad_norm": 1.015625, "learning_rate": 1.2321201166448226e-05, "loss": 0.1018, "step": 21659 }, { "epoch": 134.53582554517135, "grad_norm": 1.078125, "learning_rate": 1.231907781803602e-05, "loss": 0.099, "step": 21660 }, { "epoch": 134.54205607476635, "grad_norm": 1.265625, "learning_rate": 1.2316954592783467e-05, "loss": 0.123, "step": 21661 }, { "epoch": 134.54828660436138, "grad_norm": 1.9609375, "learning_rate": 1.2314831490711195e-05, "loss": 0.1515, "step": 21662 }, { "epoch": 134.55451713395638, "grad_norm": 1.4765625, "learning_rate": 1.2312708511839808e-05, "loss": 0.1034, "step": 21663 }, { "epoch": 134.5607476635514, "grad_norm": 1.2734375, "learning_rate": 1.2310585656189939e-05, "loss": 0.1228, "step": 21664 }, { "epoch": 134.5669781931464, "grad_norm": 2.359375, "learning_rate": 1.2308462923782188e-05, "loss": 0.1843, "step": 21665 }, { "epoch": 134.57320872274144, "grad_norm": 2.484375, "learning_rate": 1.2306340314637194e-05, "loss": 0.265, "step": 21666 }, { "epoch": 134.57943925233644, "grad_norm": 2.640625, "learning_rate": 1.230421782877555e-05, "loss": 0.1809, "step": 21667 }, { "epoch": 134.58566978193147, "grad_norm": 1.3359375, "learning_rate": 1.2302095466217881e-05, "loss": 0.1286, "step": 21668 }, { "epoch": 134.59190031152647, "grad_norm": 2.234375, "learning_rate": 1.2299973226984806e-05, "loss": 0.134, "step": 21669 }, { "epoch": 134.5981308411215, "grad_norm": 2.078125, "learning_rate": 1.2297851111096933e-05, "loss": 0.1657, "step": 21670 }, { "epoch": 134.6043613707165, "grad_norm": 2.609375, "learning_rate": 1.2295729118574859e-05, "loss": 0.2545, "step": 21671 }, { "epoch": 134.61059190031153, "grad_norm": 1.8359375, "learning_rate": 1.2293607249439202e-05, "loss": 0.128, "step": 21672 }, { "epoch": 134.61682242990653, "grad_norm": 1.140625, "learning_rate": 1.2291485503710581e-05, "loss": 0.1215, "step": 21673 }, { "epoch": 134.62305295950156, "grad_norm": 2.265625, "learning_rate": 1.2289363881409585e-05, "loss": 0.1488, "step": 21674 }, { "epoch": 134.62928348909657, "grad_norm": 1.34375, "learning_rate": 1.2287242382556826e-05, "loss": 0.1083, "step": 21675 }, { "epoch": 134.6355140186916, "grad_norm": 1.671875, "learning_rate": 1.228512100717292e-05, "loss": 0.0997, "step": 21676 }, { "epoch": 134.6417445482866, "grad_norm": 1.2734375, "learning_rate": 1.2282999755278462e-05, "loss": 0.1112, "step": 21677 }, { "epoch": 134.64797507788163, "grad_norm": 3.328125, "learning_rate": 1.228087862689404e-05, "loss": 0.2787, "step": 21678 }, { "epoch": 134.65420560747663, "grad_norm": 3.828125, "learning_rate": 1.227875762204027e-05, "loss": 0.3216, "step": 21679 }, { "epoch": 134.66043613707166, "grad_norm": 1.03125, "learning_rate": 1.2276636740737748e-05, "loss": 0.1103, "step": 21680 }, { "epoch": 134.66666666666666, "grad_norm": 1.6484375, "learning_rate": 1.2274515983007084e-05, "loss": 0.1308, "step": 21681 }, { "epoch": 134.6728971962617, "grad_norm": 2.25, "learning_rate": 1.2272395348868853e-05, "loss": 0.0862, "step": 21682 }, { "epoch": 134.6791277258567, "grad_norm": 2.140625, "learning_rate": 1.2270274838343674e-05, "loss": 0.1958, "step": 21683 }, { "epoch": 134.68535825545172, "grad_norm": 3.109375, "learning_rate": 1.2268154451452121e-05, "loss": 0.2123, "step": 21684 }, { "epoch": 134.69158878504672, "grad_norm": 1.59375, "learning_rate": 1.2266034188214804e-05, "loss": 0.1312, "step": 21685 }, { "epoch": 134.69781931464175, "grad_norm": 2.4375, "learning_rate": 1.22639140486523e-05, "loss": 0.2057, "step": 21686 }, { "epoch": 134.70404984423675, "grad_norm": 0.76953125, "learning_rate": 1.226179403278521e-05, "loss": 0.0978, "step": 21687 }, { "epoch": 134.71028037383178, "grad_norm": 2.5625, "learning_rate": 1.2259674140634132e-05, "loss": 0.103, "step": 21688 }, { "epoch": 134.71651090342678, "grad_norm": 2.75, "learning_rate": 1.2257554372219644e-05, "loss": 0.1617, "step": 21689 }, { "epoch": 134.7227414330218, "grad_norm": 2.421875, "learning_rate": 1.2255434727562326e-05, "loss": 0.2103, "step": 21690 }, { "epoch": 134.7289719626168, "grad_norm": 2.4375, "learning_rate": 1.2253315206682774e-05, "loss": 0.2286, "step": 21691 }, { "epoch": 134.73520249221184, "grad_norm": 1.8046875, "learning_rate": 1.2251195809601582e-05, "loss": 0.2454, "step": 21692 }, { "epoch": 134.74143302180684, "grad_norm": 1.1640625, "learning_rate": 1.2249076536339316e-05, "loss": 0.1075, "step": 21693 }, { "epoch": 134.74766355140187, "grad_norm": 2.796875, "learning_rate": 1.2246957386916567e-05, "loss": 0.3507, "step": 21694 }, { "epoch": 134.75389408099687, "grad_norm": 2.875, "learning_rate": 1.2244838361353924e-05, "loss": 0.2041, "step": 21695 }, { "epoch": 134.7601246105919, "grad_norm": 1.2734375, "learning_rate": 1.2242719459671964e-05, "loss": 0.1061, "step": 21696 }, { "epoch": 134.7663551401869, "grad_norm": 3.0625, "learning_rate": 1.2240600681891254e-05, "loss": 0.3134, "step": 21697 }, { "epoch": 134.77258566978193, "grad_norm": 3.703125, "learning_rate": 1.2238482028032378e-05, "loss": 0.2017, "step": 21698 }, { "epoch": 134.77881619937693, "grad_norm": 2.46875, "learning_rate": 1.2236363498115926e-05, "loss": 0.1879, "step": 21699 }, { "epoch": 134.78504672897196, "grad_norm": 0.953125, "learning_rate": 1.2234245092162455e-05, "loss": 0.0907, "step": 21700 }, { "epoch": 134.79127725856696, "grad_norm": 1.796875, "learning_rate": 1.2232126810192546e-05, "loss": 0.1235, "step": 21701 }, { "epoch": 134.797507788162, "grad_norm": 1.0703125, "learning_rate": 1.2230008652226782e-05, "loss": 0.092, "step": 21702 }, { "epoch": 134.80373831775702, "grad_norm": 1.25, "learning_rate": 1.2227890618285728e-05, "loss": 0.1086, "step": 21703 }, { "epoch": 134.80996884735202, "grad_norm": 2.65625, "learning_rate": 1.2225772708389943e-05, "loss": 0.2364, "step": 21704 }, { "epoch": 134.81619937694705, "grad_norm": 1.6875, "learning_rate": 1.222365492256001e-05, "loss": 0.1591, "step": 21705 }, { "epoch": 134.82242990654206, "grad_norm": 2.765625, "learning_rate": 1.22215372608165e-05, "loss": 0.2845, "step": 21706 }, { "epoch": 134.82866043613708, "grad_norm": 2.734375, "learning_rate": 1.2219419723179966e-05, "loss": 0.0925, "step": 21707 }, { "epoch": 134.8348909657321, "grad_norm": 1.9140625, "learning_rate": 1.2217302309670995e-05, "loss": 0.1221, "step": 21708 }, { "epoch": 134.84112149532712, "grad_norm": 2.53125, "learning_rate": 1.2215185020310127e-05, "loss": 0.1761, "step": 21709 }, { "epoch": 134.84735202492212, "grad_norm": 2.375, "learning_rate": 1.2213067855117946e-05, "loss": 0.2168, "step": 21710 }, { "epoch": 134.85358255451715, "grad_norm": 3.09375, "learning_rate": 1.2210950814115e-05, "loss": 0.3019, "step": 21711 }, { "epoch": 134.85981308411215, "grad_norm": 1.8046875, "learning_rate": 1.2208833897321853e-05, "loss": 0.1248, "step": 21712 }, { "epoch": 134.86604361370718, "grad_norm": 1.1171875, "learning_rate": 1.220671710475907e-05, "loss": 0.1098, "step": 21713 }, { "epoch": 134.87227414330218, "grad_norm": 1.9296875, "learning_rate": 1.2204600436447224e-05, "loss": 0.1428, "step": 21714 }, { "epoch": 134.8785046728972, "grad_norm": 2.15625, "learning_rate": 1.2202483892406841e-05, "loss": 0.1877, "step": 21715 }, { "epoch": 134.8847352024922, "grad_norm": 2.5625, "learning_rate": 1.2200367472658492e-05, "loss": 0.1165, "step": 21716 }, { "epoch": 134.89096573208724, "grad_norm": 2.3125, "learning_rate": 1.219825117722273e-05, "loss": 0.1885, "step": 21717 }, { "epoch": 134.89719626168224, "grad_norm": 2.390625, "learning_rate": 1.2196135006120124e-05, "loss": 0.1517, "step": 21718 }, { "epoch": 134.90342679127727, "grad_norm": 1.8359375, "learning_rate": 1.2194018959371203e-05, "loss": 0.1694, "step": 21719 }, { "epoch": 134.90965732087227, "grad_norm": 2.765625, "learning_rate": 1.219190303699653e-05, "loss": 0.2593, "step": 21720 }, { "epoch": 134.9158878504673, "grad_norm": 2.296875, "learning_rate": 1.2189787239016664e-05, "loss": 0.1613, "step": 21721 }, { "epoch": 134.9221183800623, "grad_norm": 3.421875, "learning_rate": 1.2187671565452147e-05, "loss": 0.282, "step": 21722 }, { "epoch": 134.92834890965733, "grad_norm": 2.796875, "learning_rate": 1.2185556016323513e-05, "loss": 0.2009, "step": 21723 }, { "epoch": 134.93457943925233, "grad_norm": 3.0625, "learning_rate": 1.2183440591651321e-05, "loss": 0.2236, "step": 21724 }, { "epoch": 134.94080996884736, "grad_norm": 1.953125, "learning_rate": 1.2181325291456127e-05, "loss": 0.2019, "step": 21725 }, { "epoch": 134.94704049844236, "grad_norm": 3.5, "learning_rate": 1.2179210115758454e-05, "loss": 0.2939, "step": 21726 }, { "epoch": 134.9532710280374, "grad_norm": 1.796875, "learning_rate": 1.2177095064578865e-05, "loss": 0.1168, "step": 21727 }, { "epoch": 134.9595015576324, "grad_norm": 2.765625, "learning_rate": 1.2174980137937884e-05, "loss": 0.2269, "step": 21728 }, { "epoch": 134.96573208722742, "grad_norm": 2.078125, "learning_rate": 1.2172865335856066e-05, "loss": 0.2307, "step": 21729 }, { "epoch": 134.97196261682242, "grad_norm": 1.75, "learning_rate": 1.2170750658353936e-05, "loss": 0.1367, "step": 21730 }, { "epoch": 134.97819314641745, "grad_norm": 1.203125, "learning_rate": 1.2168636105452038e-05, "loss": 0.1382, "step": 21731 }, { "epoch": 134.98442367601245, "grad_norm": 2.34375, "learning_rate": 1.2166521677170922e-05, "loss": 0.165, "step": 21732 }, { "epoch": 134.99065420560748, "grad_norm": 2.890625, "learning_rate": 1.2164407373531104e-05, "loss": 0.2362, "step": 21733 }, { "epoch": 134.99688473520249, "grad_norm": 2.46875, "learning_rate": 1.2162293194553137e-05, "loss": 0.1886, "step": 21734 }, { "epoch": 135.0, "grad_norm": 0.90234375, "learning_rate": 1.2160179140257533e-05, "loss": 0.0792, "step": 21735 }, { "epoch": 135.00623052959503, "grad_norm": 2.09375, "learning_rate": 1.2158065210664848e-05, "loss": 0.1601, "step": 21736 }, { "epoch": 135.01246105919003, "grad_norm": 2.34375, "learning_rate": 1.215595140579559e-05, "loss": 0.2374, "step": 21737 }, { "epoch": 135.01869158878506, "grad_norm": 1.125, "learning_rate": 1.21538377256703e-05, "loss": 0.0918, "step": 21738 }, { "epoch": 135.02492211838006, "grad_norm": 2.453125, "learning_rate": 1.2151724170309506e-05, "loss": 0.2299, "step": 21739 }, { "epoch": 135.0311526479751, "grad_norm": 1.59375, "learning_rate": 1.2149610739733751e-05, "loss": 0.1349, "step": 21740 }, { "epoch": 135.0373831775701, "grad_norm": 1.34375, "learning_rate": 1.2147497433963531e-05, "loss": 0.1312, "step": 21741 }, { "epoch": 135.04361370716512, "grad_norm": 2.890625, "learning_rate": 1.2145384253019382e-05, "loss": 0.1779, "step": 21742 }, { "epoch": 135.04984423676012, "grad_norm": 1.3828125, "learning_rate": 1.2143271196921831e-05, "loss": 0.0922, "step": 21743 }, { "epoch": 135.05607476635515, "grad_norm": 2.859375, "learning_rate": 1.2141158265691413e-05, "loss": 0.2679, "step": 21744 }, { "epoch": 135.06230529595015, "grad_norm": 1.125, "learning_rate": 1.2139045459348622e-05, "loss": 0.1115, "step": 21745 }, { "epoch": 135.06853582554518, "grad_norm": 1.8125, "learning_rate": 1.2136932777914007e-05, "loss": 0.1312, "step": 21746 }, { "epoch": 135.07476635514018, "grad_norm": 1.7109375, "learning_rate": 1.2134820221408059e-05, "loss": 0.1332, "step": 21747 }, { "epoch": 135.0809968847352, "grad_norm": 2.0625, "learning_rate": 1.2132707789851319e-05, "loss": 0.2811, "step": 21748 }, { "epoch": 135.08722741433021, "grad_norm": 1.0859375, "learning_rate": 1.2130595483264282e-05, "loss": 0.107, "step": 21749 }, { "epoch": 135.09345794392524, "grad_norm": 1.8203125, "learning_rate": 1.2128483301667477e-05, "loss": 0.135, "step": 21750 }, { "epoch": 135.09968847352025, "grad_norm": 3.28125, "learning_rate": 1.2126371245081421e-05, "loss": 0.1688, "step": 21751 }, { "epoch": 135.10591900311528, "grad_norm": 1.75, "learning_rate": 1.2124259313526614e-05, "loss": 0.1393, "step": 21752 }, { "epoch": 135.11214953271028, "grad_norm": 1.765625, "learning_rate": 1.2122147507023584e-05, "loss": 0.1112, "step": 21753 }, { "epoch": 135.1183800623053, "grad_norm": 2.953125, "learning_rate": 1.212003582559282e-05, "loss": 0.3276, "step": 21754 }, { "epoch": 135.1246105919003, "grad_norm": 1.5859375, "learning_rate": 1.2117924269254857e-05, "loss": 0.1091, "step": 21755 }, { "epoch": 135.13084112149534, "grad_norm": 2.015625, "learning_rate": 1.2115812838030174e-05, "loss": 0.1382, "step": 21756 }, { "epoch": 135.13707165109034, "grad_norm": 2.796875, "learning_rate": 1.2113701531939295e-05, "loss": 0.1826, "step": 21757 }, { "epoch": 135.14330218068537, "grad_norm": 3.203125, "learning_rate": 1.2111590351002732e-05, "loss": 0.2541, "step": 21758 }, { "epoch": 135.14953271028037, "grad_norm": 1.515625, "learning_rate": 1.210947929524098e-05, "loss": 0.1258, "step": 21759 }, { "epoch": 135.1557632398754, "grad_norm": 1.6171875, "learning_rate": 1.2107368364674532e-05, "loss": 0.1151, "step": 21760 }, { "epoch": 135.1619937694704, "grad_norm": 2.234375, "learning_rate": 1.2105257559323898e-05, "loss": 0.1684, "step": 21761 }, { "epoch": 135.16822429906543, "grad_norm": 1.8828125, "learning_rate": 1.210314687920959e-05, "loss": 0.12, "step": 21762 }, { "epoch": 135.17445482866043, "grad_norm": 1.5234375, "learning_rate": 1.2101036324352089e-05, "loss": 0.1094, "step": 21763 }, { "epoch": 135.18068535825546, "grad_norm": 1.6015625, "learning_rate": 1.20989258947719e-05, "loss": 0.1338, "step": 21764 }, { "epoch": 135.18691588785046, "grad_norm": 1.5625, "learning_rate": 1.2096815590489536e-05, "loss": 0.136, "step": 21765 }, { "epoch": 135.1931464174455, "grad_norm": 1.40625, "learning_rate": 1.2094705411525472e-05, "loss": 0.125, "step": 21766 }, { "epoch": 135.1993769470405, "grad_norm": 1.8046875, "learning_rate": 1.2092595357900205e-05, "loss": 0.1139, "step": 21767 }, { "epoch": 135.20560747663552, "grad_norm": 2.484375, "learning_rate": 1.2090485429634226e-05, "loss": 0.0891, "step": 21768 }, { "epoch": 135.21183800623052, "grad_norm": 1.921875, "learning_rate": 1.2088375626748047e-05, "loss": 0.1589, "step": 21769 }, { "epoch": 135.21806853582555, "grad_norm": 1.984375, "learning_rate": 1.2086265949262135e-05, "loss": 0.1509, "step": 21770 }, { "epoch": 135.22429906542055, "grad_norm": 1.6015625, "learning_rate": 1.2084156397196988e-05, "loss": 0.1324, "step": 21771 }, { "epoch": 135.23052959501558, "grad_norm": 1.140625, "learning_rate": 1.2082046970573107e-05, "loss": 0.0984, "step": 21772 }, { "epoch": 135.23676012461058, "grad_norm": 2.234375, "learning_rate": 1.2079937669410967e-05, "loss": 0.1788, "step": 21773 }, { "epoch": 135.2429906542056, "grad_norm": 3.078125, "learning_rate": 1.2077828493731044e-05, "loss": 0.2215, "step": 21774 }, { "epoch": 135.2492211838006, "grad_norm": 1.5390625, "learning_rate": 1.2075719443553836e-05, "loss": 0.1105, "step": 21775 }, { "epoch": 135.25545171339564, "grad_norm": 1.2734375, "learning_rate": 1.2073610518899825e-05, "loss": 0.1447, "step": 21776 }, { "epoch": 135.26168224299064, "grad_norm": 1.6875, "learning_rate": 1.2071501719789499e-05, "loss": 0.0889, "step": 21777 }, { "epoch": 135.26791277258567, "grad_norm": 1.2109375, "learning_rate": 1.2069393046243336e-05, "loss": 0.1224, "step": 21778 }, { "epoch": 135.27414330218068, "grad_norm": 3.0625, "learning_rate": 1.20672844982818e-05, "loss": 0.297, "step": 21779 }, { "epoch": 135.2803738317757, "grad_norm": 1.265625, "learning_rate": 1.2065176075925386e-05, "loss": 0.1199, "step": 21780 }, { "epoch": 135.2866043613707, "grad_norm": 2.734375, "learning_rate": 1.2063067779194573e-05, "loss": 0.1935, "step": 21781 }, { "epoch": 135.29283489096574, "grad_norm": 1.890625, "learning_rate": 1.2060959608109823e-05, "loss": 0.1886, "step": 21782 }, { "epoch": 135.29906542056074, "grad_norm": 1.109375, "learning_rate": 1.2058851562691622e-05, "loss": 0.1102, "step": 21783 }, { "epoch": 135.30529595015577, "grad_norm": 2.09375, "learning_rate": 1.205674364296045e-05, "loss": 0.1102, "step": 21784 }, { "epoch": 135.31152647975077, "grad_norm": 3.03125, "learning_rate": 1.2054635848936771e-05, "loss": 0.3158, "step": 21785 }, { "epoch": 135.3177570093458, "grad_norm": 1.921875, "learning_rate": 1.2052528180641045e-05, "loss": 0.1511, "step": 21786 }, { "epoch": 135.3239875389408, "grad_norm": 2.140625, "learning_rate": 1.2050420638093757e-05, "loss": 0.0833, "step": 21787 }, { "epoch": 135.33021806853583, "grad_norm": 1.6171875, "learning_rate": 1.2048313221315379e-05, "loss": 0.1797, "step": 21788 }, { "epoch": 135.33644859813083, "grad_norm": 1.09375, "learning_rate": 1.2046205930326363e-05, "loss": 0.0982, "step": 21789 }, { "epoch": 135.34267912772586, "grad_norm": 1.515625, "learning_rate": 1.2044098765147186e-05, "loss": 0.1187, "step": 21790 }, { "epoch": 135.34890965732086, "grad_norm": 2.109375, "learning_rate": 1.2041991725798319e-05, "loss": 0.2228, "step": 21791 }, { "epoch": 135.3551401869159, "grad_norm": 2.171875, "learning_rate": 1.203988481230022e-05, "loss": 0.1612, "step": 21792 }, { "epoch": 135.3613707165109, "grad_norm": 1.15625, "learning_rate": 1.2037778024673341e-05, "loss": 0.1251, "step": 21793 }, { "epoch": 135.36760124610592, "grad_norm": 2.3125, "learning_rate": 1.203567136293815e-05, "loss": 0.1216, "step": 21794 }, { "epoch": 135.37383177570092, "grad_norm": 1.546875, "learning_rate": 1.2033564827115124e-05, "loss": 0.1364, "step": 21795 }, { "epoch": 135.38006230529595, "grad_norm": 2.5, "learning_rate": 1.2031458417224698e-05, "loss": 0.2126, "step": 21796 }, { "epoch": 135.38629283489095, "grad_norm": 1.203125, "learning_rate": 1.2029352133287347e-05, "loss": 0.0993, "step": 21797 }, { "epoch": 135.39252336448598, "grad_norm": 2.984375, "learning_rate": 1.2027245975323511e-05, "loss": 0.3081, "step": 21798 }, { "epoch": 135.39875389408098, "grad_norm": 1.703125, "learning_rate": 1.202513994335367e-05, "loss": 0.1587, "step": 21799 }, { "epoch": 135.404984423676, "grad_norm": 2.109375, "learning_rate": 1.2023034037398247e-05, "loss": 0.1301, "step": 21800 }, { "epoch": 135.41121495327104, "grad_norm": 1.875, "learning_rate": 1.2020928257477718e-05, "loss": 0.1493, "step": 21801 }, { "epoch": 135.41744548286604, "grad_norm": 1.6796875, "learning_rate": 1.2018822603612534e-05, "loss": 0.1283, "step": 21802 }, { "epoch": 135.42367601246107, "grad_norm": 2.03125, "learning_rate": 1.2016717075823131e-05, "loss": 0.1256, "step": 21803 }, { "epoch": 135.42990654205607, "grad_norm": 1.109375, "learning_rate": 1.2014611674129977e-05, "loss": 0.1155, "step": 21804 }, { "epoch": 135.4361370716511, "grad_norm": 1.984375, "learning_rate": 1.20125063985535e-05, "loss": 0.1001, "step": 21805 }, { "epoch": 135.4423676012461, "grad_norm": 1.6171875, "learning_rate": 1.2010401249114167e-05, "loss": 0.1473, "step": 21806 }, { "epoch": 135.44859813084113, "grad_norm": 1.2578125, "learning_rate": 1.2008296225832405e-05, "loss": 0.119, "step": 21807 }, { "epoch": 135.45482866043614, "grad_norm": 1.8828125, "learning_rate": 1.2006191328728666e-05, "loss": 0.0985, "step": 21808 }, { "epoch": 135.46105919003116, "grad_norm": 1.9453125, "learning_rate": 1.2004086557823394e-05, "loss": 0.1794, "step": 21809 }, { "epoch": 135.46728971962617, "grad_norm": 2.90625, "learning_rate": 1.2001981913137051e-05, "loss": 0.2348, "step": 21810 }, { "epoch": 135.4735202492212, "grad_norm": 1.140625, "learning_rate": 1.1999877394690037e-05, "loss": 0.1094, "step": 21811 }, { "epoch": 135.4797507788162, "grad_norm": 1.703125, "learning_rate": 1.1997773002502816e-05, "loss": 0.1135, "step": 21812 }, { "epoch": 135.48598130841123, "grad_norm": 1.59375, "learning_rate": 1.199566873659582e-05, "loss": 0.1139, "step": 21813 }, { "epoch": 135.49221183800623, "grad_norm": 1.2109375, "learning_rate": 1.1993564596989499e-05, "loss": 0.0911, "step": 21814 }, { "epoch": 135.49844236760126, "grad_norm": 2.34375, "learning_rate": 1.1991460583704267e-05, "loss": 0.1809, "step": 21815 }, { "epoch": 135.50467289719626, "grad_norm": 1.7734375, "learning_rate": 1.1989356696760583e-05, "loss": 0.176, "step": 21816 }, { "epoch": 135.5109034267913, "grad_norm": 1.2578125, "learning_rate": 1.1987252936178855e-05, "loss": 0.0987, "step": 21817 }, { "epoch": 135.5171339563863, "grad_norm": 2.40625, "learning_rate": 1.198514930197954e-05, "loss": 0.3391, "step": 21818 }, { "epoch": 135.52336448598132, "grad_norm": 0.88671875, "learning_rate": 1.1983045794183045e-05, "loss": 0.101, "step": 21819 }, { "epoch": 135.52959501557632, "grad_norm": 1.2421875, "learning_rate": 1.1980942412809812e-05, "loss": 0.0983, "step": 21820 }, { "epoch": 135.53582554517135, "grad_norm": 2.578125, "learning_rate": 1.1978839157880278e-05, "loss": 0.1876, "step": 21821 }, { "epoch": 135.54205607476635, "grad_norm": 1.265625, "learning_rate": 1.1976736029414854e-05, "loss": 0.1039, "step": 21822 }, { "epoch": 135.54828660436138, "grad_norm": 2.875, "learning_rate": 1.1974633027433978e-05, "loss": 0.2079, "step": 21823 }, { "epoch": 135.55451713395638, "grad_norm": 2.375, "learning_rate": 1.1972530151958064e-05, "loss": 0.1062, "step": 21824 }, { "epoch": 135.5607476635514, "grad_norm": 1.7421875, "learning_rate": 1.197042740300755e-05, "loss": 0.1225, "step": 21825 }, { "epoch": 135.5669781931464, "grad_norm": 1.046875, "learning_rate": 1.1968324780602838e-05, "loss": 0.1163, "step": 21826 }, { "epoch": 135.57320872274144, "grad_norm": 2.96875, "learning_rate": 1.1966222284764363e-05, "loss": 0.3092, "step": 21827 }, { "epoch": 135.57943925233644, "grad_norm": 1.1875, "learning_rate": 1.196411991551255e-05, "loss": 0.1138, "step": 21828 }, { "epoch": 135.58566978193147, "grad_norm": 2.03125, "learning_rate": 1.1962017672867811e-05, "loss": 0.1102, "step": 21829 }, { "epoch": 135.59190031152647, "grad_norm": 2.21875, "learning_rate": 1.1959915556850557e-05, "loss": 0.1411, "step": 21830 }, { "epoch": 135.5981308411215, "grad_norm": 2.5, "learning_rate": 1.1957813567481205e-05, "loss": 0.2059, "step": 21831 }, { "epoch": 135.6043613707165, "grad_norm": 2.328125, "learning_rate": 1.1955711704780184e-05, "loss": 0.2167, "step": 21832 }, { "epoch": 135.61059190031153, "grad_norm": 2.78125, "learning_rate": 1.1953609968767892e-05, "loss": 0.1905, "step": 21833 }, { "epoch": 135.61682242990653, "grad_norm": 2.0, "learning_rate": 1.1951508359464748e-05, "loss": 0.1942, "step": 21834 }, { "epoch": 135.62305295950156, "grad_norm": 2.65625, "learning_rate": 1.194940687689117e-05, "loss": 0.1935, "step": 21835 }, { "epoch": 135.62928348909657, "grad_norm": 1.3671875, "learning_rate": 1.1947305521067561e-05, "loss": 0.1152, "step": 21836 }, { "epoch": 135.6355140186916, "grad_norm": 1.3046875, "learning_rate": 1.1945204292014322e-05, "loss": 0.0901, "step": 21837 }, { "epoch": 135.6417445482866, "grad_norm": 1.328125, "learning_rate": 1.194310318975187e-05, "loss": 0.1119, "step": 21838 }, { "epoch": 135.64797507788163, "grad_norm": 1.96875, "learning_rate": 1.1941002214300606e-05, "loss": 0.1332, "step": 21839 }, { "epoch": 135.65420560747663, "grad_norm": 1.90625, "learning_rate": 1.1938901365680952e-05, "loss": 0.1571, "step": 21840 }, { "epoch": 135.66043613707166, "grad_norm": 1.6640625, "learning_rate": 1.1936800643913287e-05, "loss": 0.1215, "step": 21841 }, { "epoch": 135.66666666666666, "grad_norm": 1.1953125, "learning_rate": 1.1934700049018035e-05, "loss": 0.0922, "step": 21842 }, { "epoch": 135.6728971962617, "grad_norm": 3.234375, "learning_rate": 1.193259958101558e-05, "loss": 0.299, "step": 21843 }, { "epoch": 135.6791277258567, "grad_norm": 1.53125, "learning_rate": 1.1930499239926337e-05, "loss": 0.1538, "step": 21844 }, { "epoch": 135.68535825545172, "grad_norm": 1.2421875, "learning_rate": 1.1928399025770693e-05, "loss": 0.1347, "step": 21845 }, { "epoch": 135.69158878504672, "grad_norm": 1.875, "learning_rate": 1.1926298938569048e-05, "loss": 0.1262, "step": 21846 }, { "epoch": 135.69781931464175, "grad_norm": 2.546875, "learning_rate": 1.1924198978341809e-05, "loss": 0.2179, "step": 21847 }, { "epoch": 135.70404984423675, "grad_norm": 2.578125, "learning_rate": 1.1922099145109364e-05, "loss": 0.1908, "step": 21848 }, { "epoch": 135.71028037383178, "grad_norm": 2.40625, "learning_rate": 1.1919999438892095e-05, "loss": 0.1937, "step": 21849 }, { "epoch": 135.71651090342678, "grad_norm": 1.9609375, "learning_rate": 1.191789985971041e-05, "loss": 0.1274, "step": 21850 }, { "epoch": 135.7227414330218, "grad_norm": 1.4765625, "learning_rate": 1.1915800407584704e-05, "loss": 0.1471, "step": 21851 }, { "epoch": 135.7289719626168, "grad_norm": 2.578125, "learning_rate": 1.1913701082535353e-05, "loss": 0.2784, "step": 21852 }, { "epoch": 135.73520249221184, "grad_norm": 2.578125, "learning_rate": 1.1911601884582749e-05, "loss": 0.2299, "step": 21853 }, { "epoch": 135.74143302180684, "grad_norm": 2.921875, "learning_rate": 1.1909502813747294e-05, "loss": 0.2232, "step": 21854 }, { "epoch": 135.74766355140187, "grad_norm": 2.453125, "learning_rate": 1.1907403870049365e-05, "loss": 0.225, "step": 21855 }, { "epoch": 135.75389408099687, "grad_norm": 2.609375, "learning_rate": 1.1905305053509341e-05, "loss": 0.107, "step": 21856 }, { "epoch": 135.7601246105919, "grad_norm": 1.59375, "learning_rate": 1.1903206364147609e-05, "loss": 0.1403, "step": 21857 }, { "epoch": 135.7663551401869, "grad_norm": 0.9921875, "learning_rate": 1.1901107801984562e-05, "loss": 0.1121, "step": 21858 }, { "epoch": 135.77258566978193, "grad_norm": 2.03125, "learning_rate": 1.1899009367040567e-05, "loss": 0.0985, "step": 21859 }, { "epoch": 135.77881619937693, "grad_norm": 1.796875, "learning_rate": 1.1896911059336013e-05, "loss": 0.1135, "step": 21860 }, { "epoch": 135.78504672897196, "grad_norm": 1.0625, "learning_rate": 1.1894812878891287e-05, "loss": 0.0978, "step": 21861 }, { "epoch": 135.79127725856696, "grad_norm": 1.96875, "learning_rate": 1.1892714825726756e-05, "loss": 0.1317, "step": 21862 }, { "epoch": 135.797507788162, "grad_norm": 1.6015625, "learning_rate": 1.1890616899862792e-05, "loss": 0.1163, "step": 21863 }, { "epoch": 135.80373831775702, "grad_norm": 1.7421875, "learning_rate": 1.1888519101319776e-05, "loss": 0.1229, "step": 21864 }, { "epoch": 135.80996884735202, "grad_norm": 2.09375, "learning_rate": 1.1886421430118092e-05, "loss": 0.1725, "step": 21865 }, { "epoch": 135.81619937694705, "grad_norm": 2.09375, "learning_rate": 1.1884323886278095e-05, "loss": 0.195, "step": 21866 }, { "epoch": 135.82242990654206, "grad_norm": 2.125, "learning_rate": 1.1882226469820176e-05, "loss": 0.1673, "step": 21867 }, { "epoch": 135.82866043613708, "grad_norm": 1.75, "learning_rate": 1.1880129180764687e-05, "loss": 0.133, "step": 21868 }, { "epoch": 135.8348909657321, "grad_norm": 2.5625, "learning_rate": 1.1878032019132016e-05, "loss": 0.2503, "step": 21869 }, { "epoch": 135.84112149532712, "grad_norm": 2.265625, "learning_rate": 1.187593498494251e-05, "loss": 0.181, "step": 21870 }, { "epoch": 135.84735202492212, "grad_norm": 2.375, "learning_rate": 1.1873838078216547e-05, "loss": 0.1783, "step": 21871 }, { "epoch": 135.85358255451715, "grad_norm": 1.65625, "learning_rate": 1.1871741298974496e-05, "loss": 0.1645, "step": 21872 }, { "epoch": 135.85981308411215, "grad_norm": 2.671875, "learning_rate": 1.1869644647236723e-05, "loss": 0.2135, "step": 21873 }, { "epoch": 135.86604361370718, "grad_norm": 1.5859375, "learning_rate": 1.1867548123023586e-05, "loss": 0.0806, "step": 21874 }, { "epoch": 135.87227414330218, "grad_norm": 1.4453125, "learning_rate": 1.186545172635544e-05, "loss": 0.1251, "step": 21875 }, { "epoch": 135.8785046728972, "grad_norm": 2.1875, "learning_rate": 1.186335545725265e-05, "loss": 0.1139, "step": 21876 }, { "epoch": 135.8847352024922, "grad_norm": 2.859375, "learning_rate": 1.1861259315735587e-05, "loss": 0.242, "step": 21877 }, { "epoch": 135.89096573208724, "grad_norm": 3.609375, "learning_rate": 1.185916330182459e-05, "loss": 0.3262, "step": 21878 }, { "epoch": 135.89719626168224, "grad_norm": 3.046875, "learning_rate": 1.1857067415540029e-05, "loss": 0.0956, "step": 21879 }, { "epoch": 135.90342679127727, "grad_norm": 1.6796875, "learning_rate": 1.1854971656902263e-05, "loss": 0.1126, "step": 21880 }, { "epoch": 135.90965732087227, "grad_norm": 1.6484375, "learning_rate": 1.1852876025931642e-05, "loss": 0.1739, "step": 21881 }, { "epoch": 135.9158878504673, "grad_norm": 1.53125, "learning_rate": 1.1850780522648507e-05, "loss": 0.0886, "step": 21882 }, { "epoch": 135.9221183800623, "grad_norm": 1.3046875, "learning_rate": 1.1848685147073222e-05, "loss": 0.1025, "step": 21883 }, { "epoch": 135.92834890965733, "grad_norm": 2.40625, "learning_rate": 1.1846589899226141e-05, "loss": 0.1768, "step": 21884 }, { "epoch": 135.93457943925233, "grad_norm": 1.8203125, "learning_rate": 1.1844494779127602e-05, "loss": 0.1875, "step": 21885 }, { "epoch": 135.94080996884736, "grad_norm": 1.4140625, "learning_rate": 1.1842399786797967e-05, "loss": 0.1119, "step": 21886 }, { "epoch": 135.94704049844236, "grad_norm": 1.734375, "learning_rate": 1.1840304922257567e-05, "loss": 0.106, "step": 21887 }, { "epoch": 135.9532710280374, "grad_norm": 2.0625, "learning_rate": 1.1838210185526765e-05, "loss": 0.1694, "step": 21888 }, { "epoch": 135.9595015576324, "grad_norm": 1.265625, "learning_rate": 1.1836115576625886e-05, "loss": 0.1228, "step": 21889 }, { "epoch": 135.96573208722742, "grad_norm": 1.6484375, "learning_rate": 1.1834021095575287e-05, "loss": 0.1005, "step": 21890 }, { "epoch": 135.97196261682242, "grad_norm": 2.96875, "learning_rate": 1.1831926742395314e-05, "loss": 0.1866, "step": 21891 }, { "epoch": 135.97819314641745, "grad_norm": 2.5, "learning_rate": 1.1829832517106293e-05, "loss": 0.1885, "step": 21892 }, { "epoch": 135.98442367601245, "grad_norm": 1.1640625, "learning_rate": 1.1827738419728579e-05, "loss": 0.0919, "step": 21893 }, { "epoch": 135.99065420560748, "grad_norm": 1.453125, "learning_rate": 1.1825644450282494e-05, "loss": 0.1255, "step": 21894 }, { "epoch": 135.99688473520249, "grad_norm": 1.3984375, "learning_rate": 1.1823550608788391e-05, "loss": 0.1267, "step": 21895 }, { "epoch": 136.0, "grad_norm": 0.91015625, "learning_rate": 1.1821456895266595e-05, "loss": 0.0557, "step": 21896 }, { "epoch": 136.00623052959503, "grad_norm": 1.3046875, "learning_rate": 1.181936330973744e-05, "loss": 0.1033, "step": 21897 }, { "epoch": 136.01246105919003, "grad_norm": 1.296875, "learning_rate": 1.1817269852221271e-05, "loss": 0.129, "step": 21898 }, { "epoch": 136.01869158878506, "grad_norm": 1.59375, "learning_rate": 1.1815176522738414e-05, "loss": 0.1267, "step": 21899 }, { "epoch": 136.02492211838006, "grad_norm": 3.0, "learning_rate": 1.1813083321309188e-05, "loss": 0.2674, "step": 21900 }, { "epoch": 136.0311526479751, "grad_norm": 2.3125, "learning_rate": 1.1810990247953935e-05, "loss": 0.197, "step": 21901 }, { "epoch": 136.0373831775701, "grad_norm": 1.1171875, "learning_rate": 1.1808897302692987e-05, "loss": 0.089, "step": 21902 }, { "epoch": 136.04361370716512, "grad_norm": 2.71875, "learning_rate": 1.1806804485546658e-05, "loss": 0.2593, "step": 21903 }, { "epoch": 136.04984423676012, "grad_norm": 1.21875, "learning_rate": 1.1804711796535281e-05, "loss": 0.1152, "step": 21904 }, { "epoch": 136.05607476635515, "grad_norm": 1.90625, "learning_rate": 1.180261923567919e-05, "loss": 0.1677, "step": 21905 }, { "epoch": 136.06230529595015, "grad_norm": 0.9765625, "learning_rate": 1.1800526802998702e-05, "loss": 0.0949, "step": 21906 }, { "epoch": 136.06853582554518, "grad_norm": 1.3671875, "learning_rate": 1.1798434498514121e-05, "loss": 0.1449, "step": 21907 }, { "epoch": 136.07476635514018, "grad_norm": 1.140625, "learning_rate": 1.1796342322245788e-05, "loss": 0.1181, "step": 21908 }, { "epoch": 136.0809968847352, "grad_norm": 2.5625, "learning_rate": 1.1794250274214016e-05, "loss": 0.3487, "step": 21909 }, { "epoch": 136.08722741433021, "grad_norm": 2.265625, "learning_rate": 1.1792158354439135e-05, "loss": 0.1734, "step": 21910 }, { "epoch": 136.09345794392524, "grad_norm": 1.203125, "learning_rate": 1.179006656294144e-05, "loss": 0.1324, "step": 21911 }, { "epoch": 136.09968847352025, "grad_norm": 1.578125, "learning_rate": 1.1787974899741272e-05, "loss": 0.1643, "step": 21912 }, { "epoch": 136.10591900311528, "grad_norm": 1.5546875, "learning_rate": 1.178588336485892e-05, "loss": 0.1262, "step": 21913 }, { "epoch": 136.11214953271028, "grad_norm": 2.8125, "learning_rate": 1.1783791958314724e-05, "loss": 0.1805, "step": 21914 }, { "epoch": 136.1183800623053, "grad_norm": 1.203125, "learning_rate": 1.178170068012897e-05, "loss": 0.117, "step": 21915 }, { "epoch": 136.1246105919003, "grad_norm": 2.25, "learning_rate": 1.1779609530321983e-05, "loss": 0.1802, "step": 21916 }, { "epoch": 136.13084112149534, "grad_norm": 2.640625, "learning_rate": 1.1777518508914079e-05, "loss": 0.1528, "step": 21917 }, { "epoch": 136.13707165109034, "grad_norm": 3.765625, "learning_rate": 1.1775427615925557e-05, "loss": 0.3438, "step": 21918 }, { "epoch": 136.14330218068537, "grad_norm": 1.28125, "learning_rate": 1.177333685137672e-05, "loss": 0.0779, "step": 21919 }, { "epoch": 136.14953271028037, "grad_norm": 1.8984375, "learning_rate": 1.1771246215287877e-05, "loss": 0.1736, "step": 21920 }, { "epoch": 136.1557632398754, "grad_norm": 2.015625, "learning_rate": 1.1769155707679344e-05, "loss": 0.2394, "step": 21921 }, { "epoch": 136.1619937694704, "grad_norm": 2.875, "learning_rate": 1.1767065328571408e-05, "loss": 0.2611, "step": 21922 }, { "epoch": 136.16822429906543, "grad_norm": 2.484375, "learning_rate": 1.1764975077984377e-05, "loss": 0.1741, "step": 21923 }, { "epoch": 136.17445482866043, "grad_norm": 2.8125, "learning_rate": 1.1762884955938564e-05, "loss": 0.147, "step": 21924 }, { "epoch": 136.18068535825546, "grad_norm": 3.234375, "learning_rate": 1.1760794962454257e-05, "loss": 0.161, "step": 21925 }, { "epoch": 136.18691588785046, "grad_norm": 1.4765625, "learning_rate": 1.1758705097551748e-05, "loss": 0.127, "step": 21926 }, { "epoch": 136.1931464174455, "grad_norm": 1.6796875, "learning_rate": 1.1756615361251342e-05, "loss": 0.138, "step": 21927 }, { "epoch": 136.1993769470405, "grad_norm": 1.640625, "learning_rate": 1.1754525753573345e-05, "loss": 0.106, "step": 21928 }, { "epoch": 136.20560747663552, "grad_norm": 1.5078125, "learning_rate": 1.1752436274538032e-05, "loss": 0.1047, "step": 21929 }, { "epoch": 136.21183800623052, "grad_norm": 1.9453125, "learning_rate": 1.1750346924165703e-05, "loss": 0.1326, "step": 21930 }, { "epoch": 136.21806853582555, "grad_norm": 0.83984375, "learning_rate": 1.1748257702476664e-05, "loss": 0.0895, "step": 21931 }, { "epoch": 136.22429906542055, "grad_norm": 1.8671875, "learning_rate": 1.1746168609491199e-05, "loss": 0.1128, "step": 21932 }, { "epoch": 136.23052959501558, "grad_norm": 1.828125, "learning_rate": 1.1744079645229581e-05, "loss": 0.1416, "step": 21933 }, { "epoch": 136.23676012461058, "grad_norm": 1.6015625, "learning_rate": 1.1741990809712111e-05, "loss": 0.1439, "step": 21934 }, { "epoch": 136.2429906542056, "grad_norm": 2.546875, "learning_rate": 1.1739902102959077e-05, "loss": 0.294, "step": 21935 }, { "epoch": 136.2492211838006, "grad_norm": 1.515625, "learning_rate": 1.1737813524990774e-05, "loss": 0.0935, "step": 21936 }, { "epoch": 136.25545171339564, "grad_norm": 2.90625, "learning_rate": 1.1735725075827476e-05, "loss": 0.1866, "step": 21937 }, { "epoch": 136.26168224299064, "grad_norm": 2.015625, "learning_rate": 1.173363675548946e-05, "loss": 0.131, "step": 21938 }, { "epoch": 136.26791277258567, "grad_norm": 1.8359375, "learning_rate": 1.1731548563997014e-05, "loss": 0.1545, "step": 21939 }, { "epoch": 136.27414330218068, "grad_norm": 1.8515625, "learning_rate": 1.1729460501370432e-05, "loss": 0.1604, "step": 21940 }, { "epoch": 136.2803738317757, "grad_norm": 3.5625, "learning_rate": 1.172737256762997e-05, "loss": 0.2852, "step": 21941 }, { "epoch": 136.2866043613707, "grad_norm": 1.390625, "learning_rate": 1.1725284762795921e-05, "loss": 0.1258, "step": 21942 }, { "epoch": 136.29283489096574, "grad_norm": 1.9296875, "learning_rate": 1.172319708688857e-05, "loss": 0.1254, "step": 21943 }, { "epoch": 136.29906542056074, "grad_norm": 1.8203125, "learning_rate": 1.1721109539928182e-05, "loss": 0.1335, "step": 21944 }, { "epoch": 136.30529595015577, "grad_norm": 1.8125, "learning_rate": 1.1719022121935022e-05, "loss": 0.1496, "step": 21945 }, { "epoch": 136.31152647975077, "grad_norm": 1.453125, "learning_rate": 1.1716934832929377e-05, "loss": 0.1179, "step": 21946 }, { "epoch": 136.3177570093458, "grad_norm": 1.5390625, "learning_rate": 1.1714847672931522e-05, "loss": 0.1674, "step": 21947 }, { "epoch": 136.3239875389408, "grad_norm": 2.296875, "learning_rate": 1.1712760641961715e-05, "loss": 0.1679, "step": 21948 }, { "epoch": 136.33021806853583, "grad_norm": 2.328125, "learning_rate": 1.1710673740040232e-05, "loss": 0.1828, "step": 21949 }, { "epoch": 136.33644859813083, "grad_norm": 2.46875, "learning_rate": 1.170858696718735e-05, "loss": 0.2054, "step": 21950 }, { "epoch": 136.34267912772586, "grad_norm": 2.6875, "learning_rate": 1.1706500323423331e-05, "loss": 0.2551, "step": 21951 }, { "epoch": 136.34890965732086, "grad_norm": 0.859375, "learning_rate": 1.1704413808768427e-05, "loss": 0.1161, "step": 21952 }, { "epoch": 136.3551401869159, "grad_norm": 4.0, "learning_rate": 1.1702327423242914e-05, "loss": 0.3456, "step": 21953 }, { "epoch": 136.3613707165109, "grad_norm": 3.59375, "learning_rate": 1.1700241166867065e-05, "loss": 0.287, "step": 21954 }, { "epoch": 136.36760124610592, "grad_norm": 1.828125, "learning_rate": 1.1698155039661125e-05, "loss": 0.1644, "step": 21955 }, { "epoch": 136.37383177570092, "grad_norm": 2.390625, "learning_rate": 1.169606904164537e-05, "loss": 0.1335, "step": 21956 }, { "epoch": 136.38006230529595, "grad_norm": 1.2578125, "learning_rate": 1.169398317284004e-05, "loss": 0.094, "step": 21957 }, { "epoch": 136.38629283489095, "grad_norm": 2.1875, "learning_rate": 1.1691897433265417e-05, "loss": 0.1575, "step": 21958 }, { "epoch": 136.39252336448598, "grad_norm": 1.2578125, "learning_rate": 1.1689811822941734e-05, "loss": 0.1116, "step": 21959 }, { "epoch": 136.39875389408098, "grad_norm": 1.828125, "learning_rate": 1.1687726341889263e-05, "loss": 0.1203, "step": 21960 }, { "epoch": 136.404984423676, "grad_norm": 1.453125, "learning_rate": 1.1685640990128264e-05, "loss": 0.1579, "step": 21961 }, { "epoch": 136.41121495327104, "grad_norm": 2.296875, "learning_rate": 1.1683555767678968e-05, "loss": 0.1719, "step": 21962 }, { "epoch": 136.41744548286604, "grad_norm": 1.046875, "learning_rate": 1.1681470674561653e-05, "loss": 0.0944, "step": 21963 }, { "epoch": 136.42367601246107, "grad_norm": 2.546875, "learning_rate": 1.167938571079655e-05, "loss": 0.2107, "step": 21964 }, { "epoch": 136.42990654205607, "grad_norm": 2.328125, "learning_rate": 1.167730087640392e-05, "loss": 0.1341, "step": 21965 }, { "epoch": 136.4361370716511, "grad_norm": 2.515625, "learning_rate": 1.1675216171404002e-05, "loss": 0.1721, "step": 21966 }, { "epoch": 136.4423676012461, "grad_norm": 2.984375, "learning_rate": 1.1673131595817046e-05, "loss": 0.2109, "step": 21967 }, { "epoch": 136.44859813084113, "grad_norm": 1.53125, "learning_rate": 1.16710471496633e-05, "loss": 0.1177, "step": 21968 }, { "epoch": 136.45482866043614, "grad_norm": 2.578125, "learning_rate": 1.1668962832963026e-05, "loss": 0.1339, "step": 21969 }, { "epoch": 136.46105919003116, "grad_norm": 1.140625, "learning_rate": 1.1666878645736435e-05, "loss": 0.1005, "step": 21970 }, { "epoch": 136.46728971962617, "grad_norm": 2.78125, "learning_rate": 1.166479458800378e-05, "loss": 0.222, "step": 21971 }, { "epoch": 136.4735202492212, "grad_norm": 1.6875, "learning_rate": 1.1662710659785306e-05, "loss": 0.1057, "step": 21972 }, { "epoch": 136.4797507788162, "grad_norm": 2.75, "learning_rate": 1.1660626861101263e-05, "loss": 0.2216, "step": 21973 }, { "epoch": 136.48598130841123, "grad_norm": 1.4921875, "learning_rate": 1.1658543191971869e-05, "loss": 0.1447, "step": 21974 }, { "epoch": 136.49221183800623, "grad_norm": 2.03125, "learning_rate": 1.1656459652417379e-05, "loss": 0.1335, "step": 21975 }, { "epoch": 136.49844236760126, "grad_norm": 2.28125, "learning_rate": 1.165437624245801e-05, "loss": 0.158, "step": 21976 }, { "epoch": 136.50467289719626, "grad_norm": 3.09375, "learning_rate": 1.1652292962114016e-05, "loss": 0.1069, "step": 21977 }, { "epoch": 136.5109034267913, "grad_norm": 1.0859375, "learning_rate": 1.1650209811405614e-05, "loss": 0.1026, "step": 21978 }, { "epoch": 136.5171339563863, "grad_norm": 2.484375, "learning_rate": 1.1648126790353042e-05, "loss": 0.185, "step": 21979 }, { "epoch": 136.52336448598132, "grad_norm": 2.484375, "learning_rate": 1.1646043898976539e-05, "loss": 0.0951, "step": 21980 }, { "epoch": 136.52959501557632, "grad_norm": 2.328125, "learning_rate": 1.164396113729632e-05, "loss": 0.2175, "step": 21981 }, { "epoch": 136.53582554517135, "grad_norm": 2.890625, "learning_rate": 1.1641878505332629e-05, "loss": 0.2003, "step": 21982 }, { "epoch": 136.54205607476635, "grad_norm": 2.328125, "learning_rate": 1.1639796003105674e-05, "loss": 0.1191, "step": 21983 }, { "epoch": 136.54828660436138, "grad_norm": 1.46875, "learning_rate": 1.16377136306357e-05, "loss": 0.1154, "step": 21984 }, { "epoch": 136.55451713395638, "grad_norm": 2.75, "learning_rate": 1.1635631387942913e-05, "loss": 0.2469, "step": 21985 }, { "epoch": 136.5607476635514, "grad_norm": 1.6875, "learning_rate": 1.1633549275047548e-05, "loss": 0.1213, "step": 21986 }, { "epoch": 136.5669781931464, "grad_norm": 2.09375, "learning_rate": 1.1631467291969831e-05, "loss": 0.1774, "step": 21987 }, { "epoch": 136.57320872274144, "grad_norm": 2.640625, "learning_rate": 1.1629385438729976e-05, "loss": 0.156, "step": 21988 }, { "epoch": 136.57943925233644, "grad_norm": 2.046875, "learning_rate": 1.1627303715348195e-05, "loss": 0.1753, "step": 21989 }, { "epoch": 136.58566978193147, "grad_norm": 2.953125, "learning_rate": 1.1625222121844712e-05, "loss": 0.161, "step": 21990 }, { "epoch": 136.59190031152647, "grad_norm": 2.125, "learning_rate": 1.1623140658239754e-05, "loss": 0.1111, "step": 21991 }, { "epoch": 136.5981308411215, "grad_norm": 2.6875, "learning_rate": 1.162105932455352e-05, "loss": 0.226, "step": 21992 }, { "epoch": 136.6043613707165, "grad_norm": 2.25, "learning_rate": 1.161897812080623e-05, "loss": 0.195, "step": 21993 }, { "epoch": 136.61059190031153, "grad_norm": 2.796875, "learning_rate": 1.1616897047018113e-05, "loss": 0.1167, "step": 21994 }, { "epoch": 136.61682242990653, "grad_norm": 2.203125, "learning_rate": 1.1614816103209363e-05, "loss": 0.0899, "step": 21995 }, { "epoch": 136.62305295950156, "grad_norm": 2.953125, "learning_rate": 1.1612735289400186e-05, "loss": 0.2379, "step": 21996 }, { "epoch": 136.62928348909657, "grad_norm": 1.5859375, "learning_rate": 1.1610654605610801e-05, "loss": 0.128, "step": 21997 }, { "epoch": 136.6355140186916, "grad_norm": 1.75, "learning_rate": 1.1608574051861427e-05, "loss": 0.1498, "step": 21998 }, { "epoch": 136.6417445482866, "grad_norm": 2.65625, "learning_rate": 1.1606493628172243e-05, "loss": 0.1709, "step": 21999 }, { "epoch": 136.64797507788163, "grad_norm": 3.28125, "learning_rate": 1.1604413334563475e-05, "loss": 0.2601, "step": 22000 }, { "epoch": 136.65420560747663, "grad_norm": 1.453125, "learning_rate": 1.160233317105533e-05, "loss": 0.1153, "step": 22001 }, { "epoch": 136.66043613707166, "grad_norm": 2.5, "learning_rate": 1.1600253137668005e-05, "loss": 0.1545, "step": 22002 }, { "epoch": 136.66666666666666, "grad_norm": 1.7421875, "learning_rate": 1.1598173234421689e-05, "loss": 0.1466, "step": 22003 }, { "epoch": 136.6728971962617, "grad_norm": 2.25, "learning_rate": 1.1596093461336593e-05, "loss": 0.1502, "step": 22004 }, { "epoch": 136.6791277258567, "grad_norm": 1.5, "learning_rate": 1.1594013818432917e-05, "loss": 0.1346, "step": 22005 }, { "epoch": 136.68535825545172, "grad_norm": 1.234375, "learning_rate": 1.159193430573087e-05, "loss": 0.1178, "step": 22006 }, { "epoch": 136.69158878504672, "grad_norm": 2.96875, "learning_rate": 1.1589854923250634e-05, "loss": 0.1903, "step": 22007 }, { "epoch": 136.69781931464175, "grad_norm": 1.7421875, "learning_rate": 1.1587775671012402e-05, "loss": 0.1536, "step": 22008 }, { "epoch": 136.70404984423675, "grad_norm": 1.6171875, "learning_rate": 1.158569654903637e-05, "loss": 0.1204, "step": 22009 }, { "epoch": 136.71028037383178, "grad_norm": 2.515625, "learning_rate": 1.1583617557342746e-05, "loss": 0.207, "step": 22010 }, { "epoch": 136.71651090342678, "grad_norm": 2.84375, "learning_rate": 1.1581538695951702e-05, "loss": 0.2141, "step": 22011 }, { "epoch": 136.7227414330218, "grad_norm": 2.515625, "learning_rate": 1.1579459964883436e-05, "loss": 0.161, "step": 22012 }, { "epoch": 136.7289719626168, "grad_norm": 1.0390625, "learning_rate": 1.1577381364158143e-05, "loss": 0.0951, "step": 22013 }, { "epoch": 136.73520249221184, "grad_norm": 2.40625, "learning_rate": 1.1575302893796008e-05, "loss": 0.2057, "step": 22014 }, { "epoch": 136.74143302180684, "grad_norm": 1.9453125, "learning_rate": 1.1573224553817207e-05, "loss": 0.1148, "step": 22015 }, { "epoch": 136.74766355140187, "grad_norm": 1.7734375, "learning_rate": 1.1571146344241931e-05, "loss": 0.1549, "step": 22016 }, { "epoch": 136.75389408099687, "grad_norm": 2.828125, "learning_rate": 1.1569068265090377e-05, "loss": 0.302, "step": 22017 }, { "epoch": 136.7601246105919, "grad_norm": 2.8125, "learning_rate": 1.1566990316382705e-05, "loss": 0.2196, "step": 22018 }, { "epoch": 136.7663551401869, "grad_norm": 0.98046875, "learning_rate": 1.1564912498139114e-05, "loss": 0.1131, "step": 22019 }, { "epoch": 136.77258566978193, "grad_norm": 3.109375, "learning_rate": 1.1562834810379784e-05, "loss": 0.3303, "step": 22020 }, { "epoch": 136.77881619937693, "grad_norm": 2.4375, "learning_rate": 1.1560757253124885e-05, "loss": 0.1942, "step": 22021 }, { "epoch": 136.78504672897196, "grad_norm": 1.8125, "learning_rate": 1.1558679826394592e-05, "loss": 0.0905, "step": 22022 }, { "epoch": 136.79127725856696, "grad_norm": 1.90625, "learning_rate": 1.1556602530209087e-05, "loss": 0.1335, "step": 22023 }, { "epoch": 136.797507788162, "grad_norm": 1.2109375, "learning_rate": 1.1554525364588555e-05, "loss": 0.1059, "step": 22024 }, { "epoch": 136.80373831775702, "grad_norm": 1.234375, "learning_rate": 1.1552448329553153e-05, "loss": 0.1206, "step": 22025 }, { "epoch": 136.80996884735202, "grad_norm": 1.921875, "learning_rate": 1.1550371425123066e-05, "loss": 0.1048, "step": 22026 }, { "epoch": 136.81619937694705, "grad_norm": 1.921875, "learning_rate": 1.1548294651318452e-05, "loss": 0.1339, "step": 22027 }, { "epoch": 136.82242990654206, "grad_norm": 3.359375, "learning_rate": 1.1546218008159501e-05, "loss": 0.2764, "step": 22028 }, { "epoch": 136.82866043613708, "grad_norm": 1.0625, "learning_rate": 1.1544141495666361e-05, "loss": 0.0926, "step": 22029 }, { "epoch": 136.8348909657321, "grad_norm": 2.53125, "learning_rate": 1.1542065113859207e-05, "loss": 0.2416, "step": 22030 }, { "epoch": 136.84112149532712, "grad_norm": 2.5, "learning_rate": 1.1539988862758209e-05, "loss": 0.18, "step": 22031 }, { "epoch": 136.84735202492212, "grad_norm": 2.71875, "learning_rate": 1.1537912742383536e-05, "loss": 0.2375, "step": 22032 }, { "epoch": 136.85358255451715, "grad_norm": 2.953125, "learning_rate": 1.1535836752755343e-05, "loss": 0.2239, "step": 22033 }, { "epoch": 136.85981308411215, "grad_norm": 1.578125, "learning_rate": 1.1533760893893792e-05, "loss": 0.1211, "step": 22034 }, { "epoch": 136.86604361370718, "grad_norm": 2.171875, "learning_rate": 1.1531685165819041e-05, "loss": 0.2026, "step": 22035 }, { "epoch": 136.87227414330218, "grad_norm": 2.53125, "learning_rate": 1.1529609568551272e-05, "loss": 0.114, "step": 22036 }, { "epoch": 136.8785046728972, "grad_norm": 2.375, "learning_rate": 1.1527534102110612e-05, "loss": 0.1747, "step": 22037 }, { "epoch": 136.8847352024922, "grad_norm": 3.09375, "learning_rate": 1.1525458766517238e-05, "loss": 0.1501, "step": 22038 }, { "epoch": 136.89096573208724, "grad_norm": 1.234375, "learning_rate": 1.152338356179131e-05, "loss": 0.0914, "step": 22039 }, { "epoch": 136.89719626168224, "grad_norm": 3.59375, "learning_rate": 1.1521308487952975e-05, "loss": 0.3597, "step": 22040 }, { "epoch": 136.90342679127727, "grad_norm": 1.0703125, "learning_rate": 1.1519233545022376e-05, "loss": 0.0986, "step": 22041 }, { "epoch": 136.90965732087227, "grad_norm": 2.296875, "learning_rate": 1.1517158733019675e-05, "loss": 0.2436, "step": 22042 }, { "epoch": 136.9158878504673, "grad_norm": 0.90234375, "learning_rate": 1.1515084051965036e-05, "loss": 0.1104, "step": 22043 }, { "epoch": 136.9221183800623, "grad_norm": 2.359375, "learning_rate": 1.1513009501878583e-05, "loss": 0.1543, "step": 22044 }, { "epoch": 136.92834890965733, "grad_norm": 2.265625, "learning_rate": 1.1510935082780489e-05, "loss": 0.2063, "step": 22045 }, { "epoch": 136.93457943925233, "grad_norm": 2.609375, "learning_rate": 1.1508860794690881e-05, "loss": 0.2319, "step": 22046 }, { "epoch": 136.94080996884736, "grad_norm": 2.546875, "learning_rate": 1.1506786637629924e-05, "loss": 0.2166, "step": 22047 }, { "epoch": 136.94704049844236, "grad_norm": 2.234375, "learning_rate": 1.150471261161774e-05, "loss": 0.1599, "step": 22048 }, { "epoch": 136.9532710280374, "grad_norm": 1.5234375, "learning_rate": 1.1502638716674488e-05, "loss": 0.0847, "step": 22049 }, { "epoch": 136.9595015576324, "grad_norm": 2.15625, "learning_rate": 1.1500564952820315e-05, "loss": 0.1825, "step": 22050 }, { "epoch": 136.96573208722742, "grad_norm": 3.0625, "learning_rate": 1.1498491320075342e-05, "loss": 0.3641, "step": 22051 }, { "epoch": 136.97196261682242, "grad_norm": 1.21875, "learning_rate": 1.1496417818459731e-05, "loss": 0.1122, "step": 22052 }, { "epoch": 136.97819314641745, "grad_norm": 2.6875, "learning_rate": 1.1494344447993601e-05, "loss": 0.1894, "step": 22053 }, { "epoch": 136.98442367601245, "grad_norm": 2.4375, "learning_rate": 1.1492271208697106e-05, "loss": 0.1424, "step": 22054 }, { "epoch": 136.99065420560748, "grad_norm": 2.171875, "learning_rate": 1.1490198100590365e-05, "loss": 0.172, "step": 22055 }, { "epoch": 136.99688473520249, "grad_norm": 2.203125, "learning_rate": 1.1488125123693518e-05, "loss": 0.1328, "step": 22056 }, { "epoch": 137.0, "grad_norm": 1.328125, "learning_rate": 1.1486052278026712e-05, "loss": 0.0701, "step": 22057 }, { "epoch": 137.00623052959503, "grad_norm": 2.375, "learning_rate": 1.148397956361007e-05, "loss": 0.2281, "step": 22058 }, { "epoch": 137.01246105919003, "grad_norm": 3.09375, "learning_rate": 1.1481906980463709e-05, "loss": 0.2411, "step": 22059 }, { "epoch": 137.01869158878506, "grad_norm": 2.46875, "learning_rate": 1.147983452860777e-05, "loss": 0.2055, "step": 22060 }, { "epoch": 137.02492211838006, "grad_norm": 2.609375, "learning_rate": 1.1477762208062386e-05, "loss": 0.2337, "step": 22061 }, { "epoch": 137.0311526479751, "grad_norm": 3.09375, "learning_rate": 1.1475690018847674e-05, "loss": 0.3198, "step": 22062 }, { "epoch": 137.0373831775701, "grad_norm": 2.421875, "learning_rate": 1.1473617960983762e-05, "loss": 0.1985, "step": 22063 }, { "epoch": 137.04361370716512, "grad_norm": 2.546875, "learning_rate": 1.1471546034490776e-05, "loss": 0.2053, "step": 22064 }, { "epoch": 137.04984423676012, "grad_norm": 2.890625, "learning_rate": 1.1469474239388858e-05, "loss": 0.2505, "step": 22065 }, { "epoch": 137.05607476635515, "grad_norm": 1.4453125, "learning_rate": 1.1467402575698094e-05, "loss": 0.1366, "step": 22066 }, { "epoch": 137.06230529595015, "grad_norm": 3.0625, "learning_rate": 1.1465331043438621e-05, "loss": 0.2285, "step": 22067 }, { "epoch": 137.06853582554518, "grad_norm": 1.984375, "learning_rate": 1.1463259642630555e-05, "loss": 0.1339, "step": 22068 }, { "epoch": 137.07476635514018, "grad_norm": 1.359375, "learning_rate": 1.146118837329403e-05, "loss": 0.0975, "step": 22069 }, { "epoch": 137.0809968847352, "grad_norm": 1.2578125, "learning_rate": 1.145911723544914e-05, "loss": 0.1387, "step": 22070 }, { "epoch": 137.08722741433021, "grad_norm": 1.5625, "learning_rate": 1.145704622911602e-05, "loss": 0.0825, "step": 22071 }, { "epoch": 137.09345794392524, "grad_norm": 1.4765625, "learning_rate": 1.1454975354314762e-05, "loss": 0.1165, "step": 22072 }, { "epoch": 137.09968847352025, "grad_norm": 2.78125, "learning_rate": 1.1452904611065502e-05, "loss": 0.2256, "step": 22073 }, { "epoch": 137.10591900311528, "grad_norm": 1.984375, "learning_rate": 1.1450833999388332e-05, "loss": 0.1344, "step": 22074 }, { "epoch": 137.11214953271028, "grad_norm": 2.484375, "learning_rate": 1.144876351930337e-05, "loss": 0.1845, "step": 22075 }, { "epoch": 137.1183800623053, "grad_norm": 2.265625, "learning_rate": 1.1446693170830736e-05, "loss": 0.1879, "step": 22076 }, { "epoch": 137.1246105919003, "grad_norm": 3.421875, "learning_rate": 1.1444622953990524e-05, "loss": 0.2351, "step": 22077 }, { "epoch": 137.13084112149534, "grad_norm": 1.453125, "learning_rate": 1.1442552868802836e-05, "loss": 0.1297, "step": 22078 }, { "epoch": 137.13707165109034, "grad_norm": 2.5, "learning_rate": 1.1440482915287786e-05, "loss": 0.1432, "step": 22079 }, { "epoch": 137.14330218068537, "grad_norm": 1.921875, "learning_rate": 1.1438413093465484e-05, "loss": 0.1208, "step": 22080 }, { "epoch": 137.14953271028037, "grad_norm": 1.6875, "learning_rate": 1.1436343403356017e-05, "loss": 0.1061, "step": 22081 }, { "epoch": 137.1557632398754, "grad_norm": 1.859375, "learning_rate": 1.1434273844979493e-05, "loss": 0.149, "step": 22082 }, { "epoch": 137.1619937694704, "grad_norm": 2.09375, "learning_rate": 1.1432204418356022e-05, "loss": 0.1148, "step": 22083 }, { "epoch": 137.16822429906543, "grad_norm": 2.9375, "learning_rate": 1.1430135123505694e-05, "loss": 0.2232, "step": 22084 }, { "epoch": 137.17445482866043, "grad_norm": 1.0234375, "learning_rate": 1.1428065960448598e-05, "loss": 0.0904, "step": 22085 }, { "epoch": 137.18068535825546, "grad_norm": 2.828125, "learning_rate": 1.1425996929204838e-05, "loss": 0.1828, "step": 22086 }, { "epoch": 137.18691588785046, "grad_norm": 1.15625, "learning_rate": 1.142392802979452e-05, "loss": 0.1281, "step": 22087 }, { "epoch": 137.1931464174455, "grad_norm": 2.25, "learning_rate": 1.1421859262237714e-05, "loss": 0.1476, "step": 22088 }, { "epoch": 137.1993769470405, "grad_norm": 1.203125, "learning_rate": 1.1419790626554528e-05, "loss": 0.1102, "step": 22089 }, { "epoch": 137.20560747663552, "grad_norm": 2.875, "learning_rate": 1.141772212276506e-05, "loss": 0.2686, "step": 22090 }, { "epoch": 137.21183800623052, "grad_norm": 1.78125, "learning_rate": 1.1415653750889385e-05, "loss": 0.1599, "step": 22091 }, { "epoch": 137.21806853582555, "grad_norm": 2.59375, "learning_rate": 1.1413585510947591e-05, "loss": 0.1905, "step": 22092 }, { "epoch": 137.22429906542055, "grad_norm": 1.7734375, "learning_rate": 1.141151740295977e-05, "loss": 0.1309, "step": 22093 }, { "epoch": 137.23052959501558, "grad_norm": 2.296875, "learning_rate": 1.1409449426946017e-05, "loss": 0.1868, "step": 22094 }, { "epoch": 137.23676012461058, "grad_norm": 1.71875, "learning_rate": 1.1407381582926399e-05, "loss": 0.1197, "step": 22095 }, { "epoch": 137.2429906542056, "grad_norm": 1.71875, "learning_rate": 1.1405313870921017e-05, "loss": 0.1254, "step": 22096 }, { "epoch": 137.2492211838006, "grad_norm": 1.890625, "learning_rate": 1.1403246290949934e-05, "loss": 0.1104, "step": 22097 }, { "epoch": 137.25545171339564, "grad_norm": 2.546875, "learning_rate": 1.1401178843033253e-05, "loss": 0.1686, "step": 22098 }, { "epoch": 137.26168224299064, "grad_norm": 1.9140625, "learning_rate": 1.1399111527191029e-05, "loss": 0.1595, "step": 22099 }, { "epoch": 137.26791277258567, "grad_norm": 2.34375, "learning_rate": 1.1397044343443352e-05, "loss": 0.1141, "step": 22100 }, { "epoch": 137.27414330218068, "grad_norm": 2.15625, "learning_rate": 1.1394977291810305e-05, "loss": 0.2055, "step": 22101 }, { "epoch": 137.2803738317757, "grad_norm": 2.234375, "learning_rate": 1.1392910372311963e-05, "loss": 0.1177, "step": 22102 }, { "epoch": 137.2866043613707, "grad_norm": 1.6171875, "learning_rate": 1.1390843584968398e-05, "loss": 0.1259, "step": 22103 }, { "epoch": 137.29283489096574, "grad_norm": 2.421875, "learning_rate": 1.1388776929799668e-05, "loss": 0.1954, "step": 22104 }, { "epoch": 137.29906542056074, "grad_norm": 1.9921875, "learning_rate": 1.138671040682586e-05, "loss": 0.1504, "step": 22105 }, { "epoch": 137.30529595015577, "grad_norm": 2.28125, "learning_rate": 1.1384644016067048e-05, "loss": 0.1784, "step": 22106 }, { "epoch": 137.31152647975077, "grad_norm": 2.484375, "learning_rate": 1.1382577757543288e-05, "loss": 0.2576, "step": 22107 }, { "epoch": 137.3177570093458, "grad_norm": 2.265625, "learning_rate": 1.1380511631274654e-05, "loss": 0.1946, "step": 22108 }, { "epoch": 137.3239875389408, "grad_norm": 1.2421875, "learning_rate": 1.1378445637281224e-05, "loss": 0.1055, "step": 22109 }, { "epoch": 137.33021806853583, "grad_norm": 2.078125, "learning_rate": 1.1376379775583051e-05, "loss": 0.1628, "step": 22110 }, { "epoch": 137.33644859813083, "grad_norm": 2.703125, "learning_rate": 1.1374314046200193e-05, "loss": 0.2585, "step": 22111 }, { "epoch": 137.34267912772586, "grad_norm": 2.796875, "learning_rate": 1.1372248449152723e-05, "loss": 0.3535, "step": 22112 }, { "epoch": 137.34890965732086, "grad_norm": 2.46875, "learning_rate": 1.1370182984460706e-05, "loss": 0.2346, "step": 22113 }, { "epoch": 137.3551401869159, "grad_norm": 1.9453125, "learning_rate": 1.1368117652144187e-05, "loss": 0.1842, "step": 22114 }, { "epoch": 137.3613707165109, "grad_norm": 2.671875, "learning_rate": 1.1366052452223243e-05, "loss": 0.2302, "step": 22115 }, { "epoch": 137.36760124610592, "grad_norm": 3.453125, "learning_rate": 1.1363987384717917e-05, "loss": 0.2696, "step": 22116 }, { "epoch": 137.37383177570092, "grad_norm": 2.703125, "learning_rate": 1.136192244964828e-05, "loss": 0.314, "step": 22117 }, { "epoch": 137.38006230529595, "grad_norm": 1.890625, "learning_rate": 1.1359857647034367e-05, "loss": 0.1408, "step": 22118 }, { "epoch": 137.38629283489095, "grad_norm": 2.234375, "learning_rate": 1.1357792976896245e-05, "loss": 0.187, "step": 22119 }, { "epoch": 137.39252336448598, "grad_norm": 1.296875, "learning_rate": 1.1355728439253973e-05, "loss": 0.124, "step": 22120 }, { "epoch": 137.39875389408098, "grad_norm": 2.59375, "learning_rate": 1.1353664034127583e-05, "loss": 0.2479, "step": 22121 }, { "epoch": 137.404984423676, "grad_norm": 1.625, "learning_rate": 1.1351599761537149e-05, "loss": 0.164, "step": 22122 }, { "epoch": 137.41121495327104, "grad_norm": 0.90625, "learning_rate": 1.1349535621502694e-05, "loss": 0.0938, "step": 22123 }, { "epoch": 137.41744548286604, "grad_norm": 2.015625, "learning_rate": 1.1347471614044289e-05, "loss": 0.186, "step": 22124 }, { "epoch": 137.42367601246107, "grad_norm": 1.1875, "learning_rate": 1.1345407739181957e-05, "loss": 0.0948, "step": 22125 }, { "epoch": 137.42990654205607, "grad_norm": 1.9453125, "learning_rate": 1.1343343996935757e-05, "loss": 0.1326, "step": 22126 }, { "epoch": 137.4361370716511, "grad_norm": 3.5, "learning_rate": 1.1341280387325728e-05, "loss": 0.2947, "step": 22127 }, { "epoch": 137.4423676012461, "grad_norm": 1.3203125, "learning_rate": 1.1339216910371933e-05, "loss": 0.1088, "step": 22128 }, { "epoch": 137.44859813084113, "grad_norm": 2.078125, "learning_rate": 1.1337153566094377e-05, "loss": 0.1502, "step": 22129 }, { "epoch": 137.45482866043614, "grad_norm": 1.9609375, "learning_rate": 1.1335090354513118e-05, "loss": 0.1302, "step": 22130 }, { "epoch": 137.46105919003116, "grad_norm": 3.328125, "learning_rate": 1.1333027275648189e-05, "loss": 0.2399, "step": 22131 }, { "epoch": 137.46728971962617, "grad_norm": 1.03125, "learning_rate": 1.1330964329519642e-05, "loss": 0.0976, "step": 22132 }, { "epoch": 137.4735202492212, "grad_norm": 1.953125, "learning_rate": 1.1328901516147492e-05, "loss": 0.1406, "step": 22133 }, { "epoch": 137.4797507788162, "grad_norm": 1.3046875, "learning_rate": 1.1326838835551784e-05, "loss": 0.0888, "step": 22134 }, { "epoch": 137.48598130841123, "grad_norm": 1.8125, "learning_rate": 1.1324776287752562e-05, "loss": 0.1612, "step": 22135 }, { "epoch": 137.49221183800623, "grad_norm": 1.9453125, "learning_rate": 1.1322713872769843e-05, "loss": 0.0942, "step": 22136 }, { "epoch": 137.49844236760126, "grad_norm": 0.96875, "learning_rate": 1.1320651590623652e-05, "loss": 0.105, "step": 22137 }, { "epoch": 137.50467289719626, "grad_norm": 2.984375, "learning_rate": 1.1318589441334029e-05, "loss": 0.2156, "step": 22138 }, { "epoch": 137.5109034267913, "grad_norm": 2.453125, "learning_rate": 1.1316527424921008e-05, "loss": 0.2326, "step": 22139 }, { "epoch": 137.5171339563863, "grad_norm": 0.97265625, "learning_rate": 1.1314465541404597e-05, "loss": 0.0999, "step": 22140 }, { "epoch": 137.52336448598132, "grad_norm": 1.0390625, "learning_rate": 1.1312403790804845e-05, "loss": 0.1064, "step": 22141 }, { "epoch": 137.52959501557632, "grad_norm": 1.4375, "learning_rate": 1.1310342173141753e-05, "loss": 0.1184, "step": 22142 }, { "epoch": 137.53582554517135, "grad_norm": 1.8515625, "learning_rate": 1.1308280688435364e-05, "loss": 0.2465, "step": 22143 }, { "epoch": 137.54205607476635, "grad_norm": 1.1484375, "learning_rate": 1.1306219336705679e-05, "loss": 0.0981, "step": 22144 }, { "epoch": 137.54828660436138, "grad_norm": 2.09375, "learning_rate": 1.1304158117972732e-05, "loss": 0.1921, "step": 22145 }, { "epoch": 137.55451713395638, "grad_norm": 1.9921875, "learning_rate": 1.1302097032256544e-05, "loss": 0.1195, "step": 22146 }, { "epoch": 137.5607476635514, "grad_norm": 2.171875, "learning_rate": 1.1300036079577131e-05, "loss": 0.1641, "step": 22147 }, { "epoch": 137.5669781931464, "grad_norm": 1.4296875, "learning_rate": 1.1297975259954497e-05, "loss": 0.1085, "step": 22148 }, { "epoch": 137.57320872274144, "grad_norm": 1.3828125, "learning_rate": 1.1295914573408667e-05, "loss": 0.1001, "step": 22149 }, { "epoch": 137.57943925233644, "grad_norm": 1.5078125, "learning_rate": 1.129385401995966e-05, "loss": 0.1241, "step": 22150 }, { "epoch": 137.58566978193147, "grad_norm": 2.421875, "learning_rate": 1.1291793599627476e-05, "loss": 0.1542, "step": 22151 }, { "epoch": 137.59190031152647, "grad_norm": 1.71875, "learning_rate": 1.1289733312432133e-05, "loss": 0.1045, "step": 22152 }, { "epoch": 137.5981308411215, "grad_norm": 1.921875, "learning_rate": 1.1287673158393646e-05, "loss": 0.1522, "step": 22153 }, { "epoch": 137.6043613707165, "grad_norm": 1.46875, "learning_rate": 1.1285613137532023e-05, "loss": 0.1288, "step": 22154 }, { "epoch": 137.61059190031153, "grad_norm": 1.203125, "learning_rate": 1.1283553249867255e-05, "loss": 0.1322, "step": 22155 }, { "epoch": 137.61682242990653, "grad_norm": 2.03125, "learning_rate": 1.128149349541936e-05, "loss": 0.1593, "step": 22156 }, { "epoch": 137.62305295950156, "grad_norm": 1.2421875, "learning_rate": 1.1279433874208354e-05, "loss": 0.13, "step": 22157 }, { "epoch": 137.62928348909657, "grad_norm": 2.609375, "learning_rate": 1.1277374386254216e-05, "loss": 0.1922, "step": 22158 }, { "epoch": 137.6355140186916, "grad_norm": 1.171875, "learning_rate": 1.1275315031576963e-05, "loss": 0.0976, "step": 22159 }, { "epoch": 137.6417445482866, "grad_norm": 2.84375, "learning_rate": 1.1273255810196604e-05, "loss": 0.2551, "step": 22160 }, { "epoch": 137.64797507788163, "grad_norm": 2.640625, "learning_rate": 1.1271196722133127e-05, "loss": 0.2222, "step": 22161 }, { "epoch": 137.65420560747663, "grad_norm": 1.7265625, "learning_rate": 1.126913776740652e-05, "loss": 0.2378, "step": 22162 }, { "epoch": 137.66043613707166, "grad_norm": 2.859375, "learning_rate": 1.1267078946036797e-05, "loss": 0.2373, "step": 22163 }, { "epoch": 137.66666666666666, "grad_norm": 2.1875, "learning_rate": 1.1265020258043945e-05, "loss": 0.1263, "step": 22164 }, { "epoch": 137.6728971962617, "grad_norm": 1.3828125, "learning_rate": 1.1262961703447971e-05, "loss": 0.1393, "step": 22165 }, { "epoch": 137.6791277258567, "grad_norm": 1.765625, "learning_rate": 1.1260903282268862e-05, "loss": 0.1722, "step": 22166 }, { "epoch": 137.68535825545172, "grad_norm": 3.890625, "learning_rate": 1.1258844994526596e-05, "loss": 0.1791, "step": 22167 }, { "epoch": 137.69158878504672, "grad_norm": 2.21875, "learning_rate": 1.1256786840241173e-05, "loss": 0.2862, "step": 22168 }, { "epoch": 137.69781931464175, "grad_norm": 1.2734375, "learning_rate": 1.1254728819432595e-05, "loss": 0.12, "step": 22169 }, { "epoch": 137.70404984423675, "grad_norm": 1.96875, "learning_rate": 1.1252670932120827e-05, "loss": 0.1054, "step": 22170 }, { "epoch": 137.71028037383178, "grad_norm": 2.75, "learning_rate": 1.1250613178325866e-05, "loss": 0.2131, "step": 22171 }, { "epoch": 137.71651090342678, "grad_norm": 2.859375, "learning_rate": 1.1248555558067708e-05, "loss": 0.1809, "step": 22172 }, { "epoch": 137.7227414330218, "grad_norm": 1.8984375, "learning_rate": 1.124649807136633e-05, "loss": 0.1474, "step": 22173 }, { "epoch": 137.7289719626168, "grad_norm": 2.1875, "learning_rate": 1.12444407182417e-05, "loss": 0.1658, "step": 22174 }, { "epoch": 137.73520249221184, "grad_norm": 1.8671875, "learning_rate": 1.1242383498713814e-05, "loss": 0.0867, "step": 22175 }, { "epoch": 137.74143302180684, "grad_norm": 1.5859375, "learning_rate": 1.1240326412802656e-05, "loss": 0.1021, "step": 22176 }, { "epoch": 137.74766355140187, "grad_norm": 1.3984375, "learning_rate": 1.1238269460528189e-05, "loss": 0.1153, "step": 22177 }, { "epoch": 137.75389408099687, "grad_norm": 2.265625, "learning_rate": 1.1236212641910401e-05, "loss": 0.2059, "step": 22178 }, { "epoch": 137.7601246105919, "grad_norm": 1.3515625, "learning_rate": 1.1234155956969277e-05, "loss": 0.1349, "step": 22179 }, { "epoch": 137.7663551401869, "grad_norm": 1.3046875, "learning_rate": 1.1232099405724783e-05, "loss": 0.1464, "step": 22180 }, { "epoch": 137.77258566978193, "grad_norm": 2.21875, "learning_rate": 1.1230042988196882e-05, "loss": 0.1655, "step": 22181 }, { "epoch": 137.77881619937693, "grad_norm": 1.3046875, "learning_rate": 1.1227986704405555e-05, "loss": 0.1356, "step": 22182 }, { "epoch": 137.78504672897196, "grad_norm": 1.0625, "learning_rate": 1.1225930554370784e-05, "loss": 0.1102, "step": 22183 }, { "epoch": 137.79127725856696, "grad_norm": 2.171875, "learning_rate": 1.122387453811252e-05, "loss": 0.125, "step": 22184 }, { "epoch": 137.797507788162, "grad_norm": 1.40625, "learning_rate": 1.1221818655650738e-05, "loss": 0.109, "step": 22185 }, { "epoch": 137.80373831775702, "grad_norm": 2.25, "learning_rate": 1.1219762907005418e-05, "loss": 0.1133, "step": 22186 }, { "epoch": 137.80996884735202, "grad_norm": 2.390625, "learning_rate": 1.1217707292196516e-05, "loss": 0.1751, "step": 22187 }, { "epoch": 137.81619937694705, "grad_norm": 1.4375, "learning_rate": 1.1215651811243985e-05, "loss": 0.1185, "step": 22188 }, { "epoch": 137.82242990654206, "grad_norm": 2.078125, "learning_rate": 1.12135964641678e-05, "loss": 0.2052, "step": 22189 }, { "epoch": 137.82866043613708, "grad_norm": 2.0625, "learning_rate": 1.121154125098793e-05, "loss": 0.1929, "step": 22190 }, { "epoch": 137.8348909657321, "grad_norm": 2.90625, "learning_rate": 1.120948617172432e-05, "loss": 0.2529, "step": 22191 }, { "epoch": 137.84112149532712, "grad_norm": 2.328125, "learning_rate": 1.1207431226396947e-05, "loss": 0.1633, "step": 22192 }, { "epoch": 137.84735202492212, "grad_norm": 1.328125, "learning_rate": 1.1205376415025748e-05, "loss": 0.1257, "step": 22193 }, { "epoch": 137.85358255451715, "grad_norm": 2.53125, "learning_rate": 1.1203321737630701e-05, "loss": 0.2111, "step": 22194 }, { "epoch": 137.85981308411215, "grad_norm": 1.0625, "learning_rate": 1.120126719423174e-05, "loss": 0.0957, "step": 22195 }, { "epoch": 137.86604361370718, "grad_norm": 4.125, "learning_rate": 1.1199212784848834e-05, "loss": 0.4315, "step": 22196 }, { "epoch": 137.87227414330218, "grad_norm": 1.6171875, "learning_rate": 1.1197158509501929e-05, "loss": 0.1538, "step": 22197 }, { "epoch": 137.8785046728972, "grad_norm": 2.21875, "learning_rate": 1.1195104368210999e-05, "loss": 0.1134, "step": 22198 }, { "epoch": 137.8847352024922, "grad_norm": 1.859375, "learning_rate": 1.1193050360995958e-05, "loss": 0.1561, "step": 22199 }, { "epoch": 137.89096573208724, "grad_norm": 2.3125, "learning_rate": 1.119099648787677e-05, "loss": 0.2552, "step": 22200 }, { "epoch": 137.89719626168224, "grad_norm": 1.140625, "learning_rate": 1.1188942748873384e-05, "loss": 0.1158, "step": 22201 }, { "epoch": 137.90342679127727, "grad_norm": 1.6171875, "learning_rate": 1.1186889144005758e-05, "loss": 0.1501, "step": 22202 }, { "epoch": 137.90965732087227, "grad_norm": 0.83984375, "learning_rate": 1.1184835673293814e-05, "loss": 0.0909, "step": 22203 }, { "epoch": 137.9158878504673, "grad_norm": 2.65625, "learning_rate": 1.1182782336757511e-05, "loss": 0.2079, "step": 22204 }, { "epoch": 137.9221183800623, "grad_norm": 1.109375, "learning_rate": 1.1180729134416796e-05, "loss": 0.1112, "step": 22205 }, { "epoch": 137.92834890965733, "grad_norm": 2.8125, "learning_rate": 1.1178676066291598e-05, "loss": 0.2343, "step": 22206 }, { "epoch": 137.93457943925233, "grad_norm": 1.421875, "learning_rate": 1.1176623132401856e-05, "loss": 0.1417, "step": 22207 }, { "epoch": 137.94080996884736, "grad_norm": 2.421875, "learning_rate": 1.1174570332767511e-05, "loss": 0.1967, "step": 22208 }, { "epoch": 137.94704049844236, "grad_norm": 2.53125, "learning_rate": 1.1172517667408516e-05, "loss": 0.1572, "step": 22209 }, { "epoch": 137.9532710280374, "grad_norm": 1.1796875, "learning_rate": 1.117046513634478e-05, "loss": 0.1131, "step": 22210 }, { "epoch": 137.9595015576324, "grad_norm": 2.5625, "learning_rate": 1.1168412739596263e-05, "loss": 0.2159, "step": 22211 }, { "epoch": 137.96573208722742, "grad_norm": 2.5, "learning_rate": 1.1166360477182874e-05, "loss": 0.1624, "step": 22212 }, { "epoch": 137.97196261682242, "grad_norm": 1.734375, "learning_rate": 1.116430834912457e-05, "loss": 0.1347, "step": 22213 }, { "epoch": 137.97819314641745, "grad_norm": 1.8828125, "learning_rate": 1.1162256355441258e-05, "loss": 0.1905, "step": 22214 }, { "epoch": 137.98442367601245, "grad_norm": 2.328125, "learning_rate": 1.1160204496152879e-05, "loss": 0.2275, "step": 22215 }, { "epoch": 137.99065420560748, "grad_norm": 1.75, "learning_rate": 1.1158152771279368e-05, "loss": 0.1446, "step": 22216 }, { "epoch": 137.99688473520249, "grad_norm": 2.90625, "learning_rate": 1.1156101180840644e-05, "loss": 0.1977, "step": 22217 }, { "epoch": 138.0, "grad_norm": 1.203125, "learning_rate": 1.1154049724856628e-05, "loss": 0.0614, "step": 22218 }, { "epoch": 138.00623052959503, "grad_norm": 2.734375, "learning_rate": 1.1151998403347244e-05, "loss": 0.158, "step": 22219 }, { "epoch": 138.01246105919003, "grad_norm": 1.796875, "learning_rate": 1.1149947216332432e-05, "loss": 0.1569, "step": 22220 }, { "epoch": 138.01869158878506, "grad_norm": 1.6484375, "learning_rate": 1.1147896163832089e-05, "loss": 0.1277, "step": 22221 }, { "epoch": 138.02492211838006, "grad_norm": 2.21875, "learning_rate": 1.1145845245866149e-05, "loss": 0.1999, "step": 22222 }, { "epoch": 138.0311526479751, "grad_norm": 1.546875, "learning_rate": 1.114379446245453e-05, "loss": 0.1461, "step": 22223 }, { "epoch": 138.0373831775701, "grad_norm": 1.5703125, "learning_rate": 1.1141743813617167e-05, "loss": 0.1022, "step": 22224 }, { "epoch": 138.04361370716512, "grad_norm": 2.984375, "learning_rate": 1.1139693299373941e-05, "loss": 0.2562, "step": 22225 }, { "epoch": 138.04984423676012, "grad_norm": 1.9765625, "learning_rate": 1.1137642919744781e-05, "loss": 0.1654, "step": 22226 }, { "epoch": 138.05607476635515, "grad_norm": 1.328125, "learning_rate": 1.1135592674749607e-05, "loss": 0.1443, "step": 22227 }, { "epoch": 138.06230529595015, "grad_norm": 3.8125, "learning_rate": 1.1133542564408337e-05, "loss": 0.289, "step": 22228 }, { "epoch": 138.06853582554518, "grad_norm": 1.46875, "learning_rate": 1.1131492588740863e-05, "loss": 0.134, "step": 22229 }, { "epoch": 138.07476635514018, "grad_norm": 1.1484375, "learning_rate": 1.1129442747767115e-05, "loss": 0.1204, "step": 22230 }, { "epoch": 138.0809968847352, "grad_norm": 1.578125, "learning_rate": 1.1127393041506983e-05, "loss": 0.1268, "step": 22231 }, { "epoch": 138.08722741433021, "grad_norm": 2.28125, "learning_rate": 1.112534346998039e-05, "loss": 0.1568, "step": 22232 }, { "epoch": 138.09345794392524, "grad_norm": 1.4296875, "learning_rate": 1.112329403320723e-05, "loss": 0.0964, "step": 22233 }, { "epoch": 138.09968847352025, "grad_norm": 2.8125, "learning_rate": 1.1121244731207409e-05, "loss": 0.2074, "step": 22234 }, { "epoch": 138.10591900311528, "grad_norm": 1.6875, "learning_rate": 1.1119195564000842e-05, "loss": 0.1512, "step": 22235 }, { "epoch": 138.11214953271028, "grad_norm": 2.5625, "learning_rate": 1.1117146531607425e-05, "loss": 0.3439, "step": 22236 }, { "epoch": 138.1183800623053, "grad_norm": 2.28125, "learning_rate": 1.1115097634047044e-05, "loss": 0.1469, "step": 22237 }, { "epoch": 138.1246105919003, "grad_norm": 2.28125, "learning_rate": 1.1113048871339613e-05, "loss": 0.2098, "step": 22238 }, { "epoch": 138.13084112149534, "grad_norm": 2.640625, "learning_rate": 1.1111000243505034e-05, "loss": 0.1714, "step": 22239 }, { "epoch": 138.13707165109034, "grad_norm": 2.0, "learning_rate": 1.110895175056319e-05, "loss": 0.15, "step": 22240 }, { "epoch": 138.14330218068537, "grad_norm": 0.9140625, "learning_rate": 1.1106903392533982e-05, "loss": 0.0986, "step": 22241 }, { "epoch": 138.14953271028037, "grad_norm": 2.28125, "learning_rate": 1.1104855169437314e-05, "loss": 0.1829, "step": 22242 }, { "epoch": 138.1557632398754, "grad_norm": 2.359375, "learning_rate": 1.1102807081293073e-05, "loss": 0.1682, "step": 22243 }, { "epoch": 138.1619937694704, "grad_norm": 2.9375, "learning_rate": 1.1100759128121135e-05, "loss": 0.299, "step": 22244 }, { "epoch": 138.16822429906543, "grad_norm": 2.15625, "learning_rate": 1.1098711309941406e-05, "loss": 0.1425, "step": 22245 }, { "epoch": 138.17445482866043, "grad_norm": 2.03125, "learning_rate": 1.1096663626773778e-05, "loss": 0.2167, "step": 22246 }, { "epoch": 138.18068535825546, "grad_norm": 1.171875, "learning_rate": 1.1094616078638123e-05, "loss": 0.125, "step": 22247 }, { "epoch": 138.18691588785046, "grad_norm": 2.265625, "learning_rate": 1.109256866555434e-05, "loss": 0.189, "step": 22248 }, { "epoch": 138.1931464174455, "grad_norm": 1.859375, "learning_rate": 1.1090521387542316e-05, "loss": 0.1344, "step": 22249 }, { "epoch": 138.1993769470405, "grad_norm": 1.3359375, "learning_rate": 1.108847424462193e-05, "loss": 0.1234, "step": 22250 }, { "epoch": 138.20560747663552, "grad_norm": 1.53125, "learning_rate": 1.1086427236813051e-05, "loss": 0.1288, "step": 22251 }, { "epoch": 138.21183800623052, "grad_norm": 2.375, "learning_rate": 1.1084380364135575e-05, "loss": 0.1213, "step": 22252 }, { "epoch": 138.21806853582555, "grad_norm": 1.2109375, "learning_rate": 1.1082333626609386e-05, "loss": 0.1145, "step": 22253 }, { "epoch": 138.22429906542055, "grad_norm": 1.421875, "learning_rate": 1.1080287024254349e-05, "loss": 0.1382, "step": 22254 }, { "epoch": 138.23052959501558, "grad_norm": 2.96875, "learning_rate": 1.1078240557090343e-05, "loss": 0.2479, "step": 22255 }, { "epoch": 138.23676012461058, "grad_norm": 2.3125, "learning_rate": 1.107619422513726e-05, "loss": 0.2268, "step": 22256 }, { "epoch": 138.2429906542056, "grad_norm": 2.453125, "learning_rate": 1.107414802841496e-05, "loss": 0.2314, "step": 22257 }, { "epoch": 138.2492211838006, "grad_norm": 1.8203125, "learning_rate": 1.1072101966943307e-05, "loss": 0.1347, "step": 22258 }, { "epoch": 138.25545171339564, "grad_norm": 2.046875, "learning_rate": 1.1070056040742189e-05, "loss": 0.201, "step": 22259 }, { "epoch": 138.26168224299064, "grad_norm": 2.734375, "learning_rate": 1.1068010249831467e-05, "loss": 0.2086, "step": 22260 }, { "epoch": 138.26791277258567, "grad_norm": 2.21875, "learning_rate": 1.1065964594231023e-05, "loss": 0.1351, "step": 22261 }, { "epoch": 138.27414330218068, "grad_norm": 1.0625, "learning_rate": 1.106391907396072e-05, "loss": 0.0987, "step": 22262 }, { "epoch": 138.2803738317757, "grad_norm": 2.3125, "learning_rate": 1.1061873689040409e-05, "loss": 0.1419, "step": 22263 }, { "epoch": 138.2866043613707, "grad_norm": 2.296875, "learning_rate": 1.1059828439489969e-05, "loss": 0.2118, "step": 22264 }, { "epoch": 138.29283489096574, "grad_norm": 3.484375, "learning_rate": 1.1057783325329268e-05, "loss": 0.3948, "step": 22265 }, { "epoch": 138.29906542056074, "grad_norm": 1.234375, "learning_rate": 1.1055738346578154e-05, "loss": 0.0945, "step": 22266 }, { "epoch": 138.30529595015577, "grad_norm": 3.15625, "learning_rate": 1.1053693503256499e-05, "loss": 0.1183, "step": 22267 }, { "epoch": 138.31152647975077, "grad_norm": 1.59375, "learning_rate": 1.1051648795384167e-05, "loss": 0.1495, "step": 22268 }, { "epoch": 138.3177570093458, "grad_norm": 1.8203125, "learning_rate": 1.1049604222981011e-05, "loss": 0.1882, "step": 22269 }, { "epoch": 138.3239875389408, "grad_norm": 2.671875, "learning_rate": 1.104755978606688e-05, "loss": 0.3363, "step": 22270 }, { "epoch": 138.33021806853583, "grad_norm": 1.828125, "learning_rate": 1.1045515484661635e-05, "loss": 0.1976, "step": 22271 }, { "epoch": 138.33644859813083, "grad_norm": 2.21875, "learning_rate": 1.1043471318785142e-05, "loss": 0.2109, "step": 22272 }, { "epoch": 138.34267912772586, "grad_norm": 2.46875, "learning_rate": 1.1041427288457237e-05, "loss": 0.2008, "step": 22273 }, { "epoch": 138.34890965732086, "grad_norm": 3.234375, "learning_rate": 1.103938339369778e-05, "loss": 0.2272, "step": 22274 }, { "epoch": 138.3551401869159, "grad_norm": 2.703125, "learning_rate": 1.1037339634526633e-05, "loss": 0.2106, "step": 22275 }, { "epoch": 138.3613707165109, "grad_norm": 1.4453125, "learning_rate": 1.1035296010963634e-05, "loss": 0.125, "step": 22276 }, { "epoch": 138.36760124610592, "grad_norm": 2.75, "learning_rate": 1.1033252523028624e-05, "loss": 0.2047, "step": 22277 }, { "epoch": 138.37383177570092, "grad_norm": 2.109375, "learning_rate": 1.1031209170741452e-05, "loss": 0.246, "step": 22278 }, { "epoch": 138.38006230529595, "grad_norm": 1.9765625, "learning_rate": 1.1029165954121983e-05, "loss": 0.1796, "step": 22279 }, { "epoch": 138.38629283489095, "grad_norm": 3.484375, "learning_rate": 1.1027122873190035e-05, "loss": 0.309, "step": 22280 }, { "epoch": 138.39252336448598, "grad_norm": 1.6171875, "learning_rate": 1.1025079927965473e-05, "loss": 0.1031, "step": 22281 }, { "epoch": 138.39875389408098, "grad_norm": 2.203125, "learning_rate": 1.102303711846812e-05, "loss": 0.1564, "step": 22282 }, { "epoch": 138.404984423676, "grad_norm": 2.40625, "learning_rate": 1.1020994444717834e-05, "loss": 0.1779, "step": 22283 }, { "epoch": 138.41121495327104, "grad_norm": 1.28125, "learning_rate": 1.1018951906734431e-05, "loss": 0.1144, "step": 22284 }, { "epoch": 138.41744548286604, "grad_norm": 1.859375, "learning_rate": 1.1016909504537768e-05, "loss": 0.1374, "step": 22285 }, { "epoch": 138.42367601246107, "grad_norm": 1.7578125, "learning_rate": 1.101486723814768e-05, "loss": 0.1301, "step": 22286 }, { "epoch": 138.42990654205607, "grad_norm": 2.4375, "learning_rate": 1.1012825107583998e-05, "loss": 0.2036, "step": 22287 }, { "epoch": 138.4361370716511, "grad_norm": 3.0, "learning_rate": 1.1010783112866545e-05, "loss": 0.2641, "step": 22288 }, { "epoch": 138.4423676012461, "grad_norm": 1.5625, "learning_rate": 1.1008741254015161e-05, "loss": 0.0843, "step": 22289 }, { "epoch": 138.44859813084113, "grad_norm": 2.234375, "learning_rate": 1.1006699531049691e-05, "loss": 0.1753, "step": 22290 }, { "epoch": 138.45482866043614, "grad_norm": 2.53125, "learning_rate": 1.1004657943989943e-05, "loss": 0.1415, "step": 22291 }, { "epoch": 138.46105919003116, "grad_norm": 2.3125, "learning_rate": 1.1002616492855755e-05, "loss": 0.2034, "step": 22292 }, { "epoch": 138.46728971962617, "grad_norm": 1.390625, "learning_rate": 1.1000575177666952e-05, "loss": 0.1122, "step": 22293 }, { "epoch": 138.4735202492212, "grad_norm": 1.7421875, "learning_rate": 1.099853399844338e-05, "loss": 0.1085, "step": 22294 }, { "epoch": 138.4797507788162, "grad_norm": 2.078125, "learning_rate": 1.0996492955204827e-05, "loss": 0.1822, "step": 22295 }, { "epoch": 138.48598130841123, "grad_norm": 2.03125, "learning_rate": 1.0994452047971131e-05, "loss": 0.1247, "step": 22296 }, { "epoch": 138.49221183800623, "grad_norm": 2.765625, "learning_rate": 1.0992411276762122e-05, "loss": 0.2561, "step": 22297 }, { "epoch": 138.49844236760126, "grad_norm": 3.09375, "learning_rate": 1.0990370641597619e-05, "loss": 0.1995, "step": 22298 }, { "epoch": 138.50467289719626, "grad_norm": 2.40625, "learning_rate": 1.0988330142497427e-05, "loss": 0.1853, "step": 22299 }, { "epoch": 138.5109034267913, "grad_norm": 2.390625, "learning_rate": 1.0986289779481387e-05, "loss": 0.2433, "step": 22300 }, { "epoch": 138.5171339563863, "grad_norm": 1.0078125, "learning_rate": 1.098424955256929e-05, "loss": 0.0924, "step": 22301 }, { "epoch": 138.52336448598132, "grad_norm": 2.234375, "learning_rate": 1.0982209461780973e-05, "loss": 0.215, "step": 22302 }, { "epoch": 138.52959501557632, "grad_norm": 2.4375, "learning_rate": 1.0980169507136229e-05, "loss": 0.1837, "step": 22303 }, { "epoch": 138.53582554517135, "grad_norm": 1.9140625, "learning_rate": 1.0978129688654884e-05, "loss": 0.1483, "step": 22304 }, { "epoch": 138.54205607476635, "grad_norm": 1.1484375, "learning_rate": 1.0976090006356754e-05, "loss": 0.125, "step": 22305 }, { "epoch": 138.54828660436138, "grad_norm": 1.3359375, "learning_rate": 1.0974050460261643e-05, "loss": 0.1174, "step": 22306 }, { "epoch": 138.55451713395638, "grad_norm": 2.40625, "learning_rate": 1.0972011050389344e-05, "loss": 0.1838, "step": 22307 }, { "epoch": 138.5607476635514, "grad_norm": 1.7265625, "learning_rate": 1.0969971776759683e-05, "loss": 0.1636, "step": 22308 }, { "epoch": 138.5669781931464, "grad_norm": 3.8125, "learning_rate": 1.0967932639392468e-05, "loss": 0.2574, "step": 22309 }, { "epoch": 138.57320872274144, "grad_norm": 2.359375, "learning_rate": 1.0965893638307483e-05, "loss": 0.1614, "step": 22310 }, { "epoch": 138.57943925233644, "grad_norm": 1.953125, "learning_rate": 1.096385477352455e-05, "loss": 0.139, "step": 22311 }, { "epoch": 138.58566978193147, "grad_norm": 2.140625, "learning_rate": 1.0961816045063472e-05, "loss": 0.1112, "step": 22312 }, { "epoch": 138.59190031152647, "grad_norm": 1.734375, "learning_rate": 1.0959777452944042e-05, "loss": 0.1496, "step": 22313 }, { "epoch": 138.5981308411215, "grad_norm": 1.2890625, "learning_rate": 1.0957738997186052e-05, "loss": 0.1274, "step": 22314 }, { "epoch": 138.6043613707165, "grad_norm": 1.375, "learning_rate": 1.0955700677809308e-05, "loss": 0.1011, "step": 22315 }, { "epoch": 138.61059190031153, "grad_norm": 1.5703125, "learning_rate": 1.0953662494833616e-05, "loss": 0.1603, "step": 22316 }, { "epoch": 138.61682242990653, "grad_norm": 3.234375, "learning_rate": 1.095162444827875e-05, "loss": 0.2694, "step": 22317 }, { "epoch": 138.62305295950156, "grad_norm": 2.328125, "learning_rate": 1.0949586538164519e-05, "loss": 0.2647, "step": 22318 }, { "epoch": 138.62928348909657, "grad_norm": 1.375, "learning_rate": 1.0947548764510718e-05, "loss": 0.1281, "step": 22319 }, { "epoch": 138.6355140186916, "grad_norm": 1.875, "learning_rate": 1.0945511127337135e-05, "loss": 0.1509, "step": 22320 }, { "epoch": 138.6417445482866, "grad_norm": 3.890625, "learning_rate": 1.0943473626663547e-05, "loss": 0.2475, "step": 22321 }, { "epoch": 138.64797507788163, "grad_norm": 2.78125, "learning_rate": 1.0941436262509752e-05, "loss": 0.255, "step": 22322 }, { "epoch": 138.65420560747663, "grad_norm": 1.8515625, "learning_rate": 1.093939903489554e-05, "loss": 0.1395, "step": 22323 }, { "epoch": 138.66043613707166, "grad_norm": 1.578125, "learning_rate": 1.0937361943840705e-05, "loss": 0.1321, "step": 22324 }, { "epoch": 138.66666666666666, "grad_norm": 1.3828125, "learning_rate": 1.0935324989365012e-05, "loss": 0.0871, "step": 22325 }, { "epoch": 138.6728971962617, "grad_norm": 2.4375, "learning_rate": 1.0933288171488262e-05, "loss": 0.2373, "step": 22326 }, { "epoch": 138.6791277258567, "grad_norm": 2.3125, "learning_rate": 1.0931251490230221e-05, "loss": 0.1602, "step": 22327 }, { "epoch": 138.68535825545172, "grad_norm": 2.15625, "learning_rate": 1.0929214945610686e-05, "loss": 0.2165, "step": 22328 }, { "epoch": 138.69158878504672, "grad_norm": 0.82421875, "learning_rate": 1.092717853764942e-05, "loss": 0.0828, "step": 22329 }, { "epoch": 138.69781931464175, "grad_norm": 2.609375, "learning_rate": 1.0925142266366208e-05, "loss": 0.0783, "step": 22330 }, { "epoch": 138.70404984423675, "grad_norm": 2.359375, "learning_rate": 1.0923106131780839e-05, "loss": 0.1735, "step": 22331 }, { "epoch": 138.71028037383178, "grad_norm": 2.078125, "learning_rate": 1.0921070133913076e-05, "loss": 0.2199, "step": 22332 }, { "epoch": 138.71651090342678, "grad_norm": 2.125, "learning_rate": 1.0919034272782686e-05, "loss": 0.1792, "step": 22333 }, { "epoch": 138.7227414330218, "grad_norm": 2.078125, "learning_rate": 1.0916998548409449e-05, "loss": 0.1229, "step": 22334 }, { "epoch": 138.7289719626168, "grad_norm": 3.265625, "learning_rate": 1.0914962960813146e-05, "loss": 0.3558, "step": 22335 }, { "epoch": 138.73520249221184, "grad_norm": 2.484375, "learning_rate": 1.0912927510013529e-05, "loss": 0.2109, "step": 22336 }, { "epoch": 138.74143302180684, "grad_norm": 1.7421875, "learning_rate": 1.0910892196030376e-05, "loss": 0.1431, "step": 22337 }, { "epoch": 138.74766355140187, "grad_norm": 1.8984375, "learning_rate": 1.0908857018883465e-05, "loss": 0.1435, "step": 22338 }, { "epoch": 138.75389408099687, "grad_norm": 1.2890625, "learning_rate": 1.0906821978592549e-05, "loss": 0.11, "step": 22339 }, { "epoch": 138.7601246105919, "grad_norm": 1.8046875, "learning_rate": 1.0904787075177388e-05, "loss": 0.151, "step": 22340 }, { "epoch": 138.7663551401869, "grad_norm": 2.625, "learning_rate": 1.0902752308657751e-05, "loss": 0.2661, "step": 22341 }, { "epoch": 138.77258566978193, "grad_norm": 2.109375, "learning_rate": 1.0900717679053412e-05, "loss": 0.1859, "step": 22342 }, { "epoch": 138.77881619937693, "grad_norm": 2.375, "learning_rate": 1.0898683186384113e-05, "loss": 0.1412, "step": 22343 }, { "epoch": 138.78504672897196, "grad_norm": 2.203125, "learning_rate": 1.089664883066962e-05, "loss": 0.2133, "step": 22344 }, { "epoch": 138.79127725856696, "grad_norm": 1.984375, "learning_rate": 1.0894614611929702e-05, "loss": 0.1353, "step": 22345 }, { "epoch": 138.797507788162, "grad_norm": 2.5625, "learning_rate": 1.0892580530184107e-05, "loss": 0.2117, "step": 22346 }, { "epoch": 138.80373831775702, "grad_norm": 1.9296875, "learning_rate": 1.0890546585452579e-05, "loss": 0.2169, "step": 22347 }, { "epoch": 138.80996884735202, "grad_norm": 2.859375, "learning_rate": 1.0888512777754887e-05, "loss": 0.1541, "step": 22348 }, { "epoch": 138.81619937694705, "grad_norm": 1.4296875, "learning_rate": 1.0886479107110786e-05, "loss": 0.117, "step": 22349 }, { "epoch": 138.82242990654206, "grad_norm": 2.328125, "learning_rate": 1.0884445573540012e-05, "loss": 0.2284, "step": 22350 }, { "epoch": 138.82866043613708, "grad_norm": 1.2890625, "learning_rate": 1.0882412177062335e-05, "loss": 0.1361, "step": 22351 }, { "epoch": 138.8348909657321, "grad_norm": 2.5625, "learning_rate": 1.088037891769748e-05, "loss": 0.211, "step": 22352 }, { "epoch": 138.84112149532712, "grad_norm": 2.546875, "learning_rate": 1.087834579546522e-05, "loss": 0.2909, "step": 22353 }, { "epoch": 138.84735202492212, "grad_norm": 1.2890625, "learning_rate": 1.0876312810385281e-05, "loss": 0.1242, "step": 22354 }, { "epoch": 138.85358255451715, "grad_norm": 2.484375, "learning_rate": 1.087427996247741e-05, "loss": 0.3972, "step": 22355 }, { "epoch": 138.85981308411215, "grad_norm": 2.78125, "learning_rate": 1.0872247251761361e-05, "loss": 0.3383, "step": 22356 }, { "epoch": 138.86604361370718, "grad_norm": 1.546875, "learning_rate": 1.0870214678256885e-05, "loss": 0.1182, "step": 22357 }, { "epoch": 138.87227414330218, "grad_norm": 2.0, "learning_rate": 1.0868182241983694e-05, "loss": 0.131, "step": 22358 }, { "epoch": 138.8785046728972, "grad_norm": 1.8046875, "learning_rate": 1.0866149942961538e-05, "loss": 0.1188, "step": 22359 }, { "epoch": 138.8847352024922, "grad_norm": 3.046875, "learning_rate": 1.0864117781210162e-05, "loss": 0.1353, "step": 22360 }, { "epoch": 138.89096573208724, "grad_norm": 1.921875, "learning_rate": 1.0862085756749308e-05, "loss": 0.1308, "step": 22361 }, { "epoch": 138.89719626168224, "grad_norm": 1.09375, "learning_rate": 1.0860053869598694e-05, "loss": 0.1041, "step": 22362 }, { "epoch": 138.90342679127727, "grad_norm": 2.09375, "learning_rate": 1.085802211977806e-05, "loss": 0.1334, "step": 22363 }, { "epoch": 138.90965732087227, "grad_norm": 2.265625, "learning_rate": 1.0855990507307156e-05, "loss": 0.1692, "step": 22364 }, { "epoch": 138.9158878504673, "grad_norm": 2.890625, "learning_rate": 1.08539590322057e-05, "loss": 0.2473, "step": 22365 }, { "epoch": 138.9221183800623, "grad_norm": 1.078125, "learning_rate": 1.0851927694493409e-05, "loss": 0.1297, "step": 22366 }, { "epoch": 138.92834890965733, "grad_norm": 2.28125, "learning_rate": 1.0849896494190026e-05, "loss": 0.1609, "step": 22367 }, { "epoch": 138.93457943925233, "grad_norm": 3.421875, "learning_rate": 1.0847865431315285e-05, "loss": 0.1507, "step": 22368 }, { "epoch": 138.94080996884736, "grad_norm": 3.34375, "learning_rate": 1.0845834505888896e-05, "loss": 0.1565, "step": 22369 }, { "epoch": 138.94704049844236, "grad_norm": 1.4609375, "learning_rate": 1.08438037179306e-05, "loss": 0.1062, "step": 22370 }, { "epoch": 138.9532710280374, "grad_norm": 1.703125, "learning_rate": 1.0841773067460104e-05, "loss": 0.1461, "step": 22371 }, { "epoch": 138.9595015576324, "grad_norm": 2.34375, "learning_rate": 1.0839742554497145e-05, "loss": 0.2705, "step": 22372 }, { "epoch": 138.96573208722742, "grad_norm": 1.5078125, "learning_rate": 1.083771217906143e-05, "loss": 0.1215, "step": 22373 }, { "epoch": 138.97196261682242, "grad_norm": 1.9375, "learning_rate": 1.0835681941172685e-05, "loss": 0.2334, "step": 22374 }, { "epoch": 138.97819314641745, "grad_norm": 2.3125, "learning_rate": 1.0833651840850637e-05, "loss": 0.158, "step": 22375 }, { "epoch": 138.98442367601245, "grad_norm": 3.0, "learning_rate": 1.0831621878114995e-05, "loss": 0.1006, "step": 22376 }, { "epoch": 138.99065420560748, "grad_norm": 3.03125, "learning_rate": 1.0829592052985464e-05, "loss": 0.1817, "step": 22377 }, { "epoch": 138.99688473520249, "grad_norm": 1.5078125, "learning_rate": 1.0827562365481764e-05, "loss": 0.112, "step": 22378 }, { "epoch": 139.0, "grad_norm": 2.015625, "learning_rate": 1.0825532815623623e-05, "loss": 0.1122, "step": 22379 }, { "epoch": 139.00623052959503, "grad_norm": 1.4765625, "learning_rate": 1.0823503403430734e-05, "loss": 0.1354, "step": 22380 }, { "epoch": 139.01246105919003, "grad_norm": 2.4375, "learning_rate": 1.0821474128922813e-05, "loss": 0.2565, "step": 22381 }, { "epoch": 139.01869158878506, "grad_norm": 1.8984375, "learning_rate": 1.0819444992119576e-05, "loss": 0.1208, "step": 22382 }, { "epoch": 139.02492211838006, "grad_norm": 3.15625, "learning_rate": 1.0817415993040724e-05, "loss": 0.2379, "step": 22383 }, { "epoch": 139.0311526479751, "grad_norm": 1.2734375, "learning_rate": 1.0815387131705957e-05, "loss": 0.1285, "step": 22384 }, { "epoch": 139.0373831775701, "grad_norm": 4.34375, "learning_rate": 1.0813358408134983e-05, "loss": 0.417, "step": 22385 }, { "epoch": 139.04361370716512, "grad_norm": 2.0, "learning_rate": 1.0811329822347519e-05, "loss": 0.1033, "step": 22386 }, { "epoch": 139.04984423676012, "grad_norm": 2.75, "learning_rate": 1.0809301374363246e-05, "loss": 0.1008, "step": 22387 }, { "epoch": 139.05607476635515, "grad_norm": 1.9765625, "learning_rate": 1.0807273064201875e-05, "loss": 0.2668, "step": 22388 }, { "epoch": 139.06230529595015, "grad_norm": 1.5, "learning_rate": 1.0805244891883113e-05, "loss": 0.1205, "step": 22389 }, { "epoch": 139.06853582554518, "grad_norm": 1.6328125, "learning_rate": 1.0803216857426652e-05, "loss": 0.1361, "step": 22390 }, { "epoch": 139.07476635514018, "grad_norm": 1.8671875, "learning_rate": 1.0801188960852176e-05, "loss": 0.1571, "step": 22391 }, { "epoch": 139.0809968847352, "grad_norm": 3.0, "learning_rate": 1.0799161202179392e-05, "loss": 0.27, "step": 22392 }, { "epoch": 139.08722741433021, "grad_norm": 1.28125, "learning_rate": 1.0797133581427993e-05, "loss": 0.1037, "step": 22393 }, { "epoch": 139.09345794392524, "grad_norm": 2.8125, "learning_rate": 1.079510609861768e-05, "loss": 0.26, "step": 22394 }, { "epoch": 139.09968847352025, "grad_norm": 1.5234375, "learning_rate": 1.0793078753768126e-05, "loss": 0.1232, "step": 22395 }, { "epoch": 139.10591900311528, "grad_norm": 1.921875, "learning_rate": 1.0791051546899042e-05, "loss": 0.1636, "step": 22396 }, { "epoch": 139.11214953271028, "grad_norm": 1.03125, "learning_rate": 1.0789024478030096e-05, "loss": 0.0897, "step": 22397 }, { "epoch": 139.1183800623053, "grad_norm": 2.265625, "learning_rate": 1.0786997547180996e-05, "loss": 0.1995, "step": 22398 }, { "epoch": 139.1246105919003, "grad_norm": 2.296875, "learning_rate": 1.0784970754371404e-05, "loss": 0.1357, "step": 22399 }, { "epoch": 139.13084112149534, "grad_norm": 3.640625, "learning_rate": 1.078294409962102e-05, "loss": 0.3813, "step": 22400 }, { "epoch": 139.13707165109034, "grad_norm": 2.203125, "learning_rate": 1.0780917582949535e-05, "loss": 0.1486, "step": 22401 }, { "epoch": 139.14330218068537, "grad_norm": 1.15625, "learning_rate": 1.077889120437662e-05, "loss": 0.0893, "step": 22402 }, { "epoch": 139.14953271028037, "grad_norm": 1.1640625, "learning_rate": 1.0776864963921946e-05, "loss": 0.1011, "step": 22403 }, { "epoch": 139.1557632398754, "grad_norm": 3.59375, "learning_rate": 1.0774838861605202e-05, "loss": 0.2137, "step": 22404 }, { "epoch": 139.1619937694704, "grad_norm": 1.171875, "learning_rate": 1.0772812897446077e-05, "loss": 0.0999, "step": 22405 }, { "epoch": 139.16822429906543, "grad_norm": 2.46875, "learning_rate": 1.077078707146423e-05, "loss": 0.2093, "step": 22406 }, { "epoch": 139.17445482866043, "grad_norm": 2.59375, "learning_rate": 1.0768761383679341e-05, "loss": 0.2007, "step": 22407 }, { "epoch": 139.18068535825546, "grad_norm": 3.71875, "learning_rate": 1.0766735834111094e-05, "loss": 0.3425, "step": 22408 }, { "epoch": 139.18691588785046, "grad_norm": 0.8984375, "learning_rate": 1.0764710422779156e-05, "loss": 0.1011, "step": 22409 }, { "epoch": 139.1931464174455, "grad_norm": 3.109375, "learning_rate": 1.0762685149703185e-05, "loss": 0.1963, "step": 22410 }, { "epoch": 139.1993769470405, "grad_norm": 2.265625, "learning_rate": 1.0760660014902864e-05, "loss": 0.1569, "step": 22411 }, { "epoch": 139.20560747663552, "grad_norm": 1.9140625, "learning_rate": 1.075863501839787e-05, "loss": 0.1178, "step": 22412 }, { "epoch": 139.21183800623052, "grad_norm": 2.421875, "learning_rate": 1.0756610160207848e-05, "loss": 0.1587, "step": 22413 }, { "epoch": 139.21806853582555, "grad_norm": 2.234375, "learning_rate": 1.0754585440352474e-05, "loss": 0.2182, "step": 22414 }, { "epoch": 139.22429906542055, "grad_norm": 4.21875, "learning_rate": 1.0752560858851423e-05, "loss": 0.198, "step": 22415 }, { "epoch": 139.23052959501558, "grad_norm": 1.5390625, "learning_rate": 1.075053641572435e-05, "loss": 0.0871, "step": 22416 }, { "epoch": 139.23676012461058, "grad_norm": 1.375, "learning_rate": 1.0748512110990908e-05, "loss": 0.1126, "step": 22417 }, { "epoch": 139.2429906542056, "grad_norm": 3.1875, "learning_rate": 1.0746487944670763e-05, "loss": 0.2411, "step": 22418 }, { "epoch": 139.2492211838006, "grad_norm": 3.3125, "learning_rate": 1.0744463916783575e-05, "loss": 0.3934, "step": 22419 }, { "epoch": 139.25545171339564, "grad_norm": 1.8125, "learning_rate": 1.0742440027349015e-05, "loss": 0.1294, "step": 22420 }, { "epoch": 139.26168224299064, "grad_norm": 2.28125, "learning_rate": 1.0740416276386727e-05, "loss": 0.1739, "step": 22421 }, { "epoch": 139.26791277258567, "grad_norm": 0.85546875, "learning_rate": 1.0738392663916357e-05, "loss": 0.0854, "step": 22422 }, { "epoch": 139.27414330218068, "grad_norm": 2.4375, "learning_rate": 1.073636918995757e-05, "loss": 0.2683, "step": 22423 }, { "epoch": 139.2803738317757, "grad_norm": 1.53125, "learning_rate": 1.0734345854530024e-05, "loss": 0.0985, "step": 22424 }, { "epoch": 139.2866043613707, "grad_norm": 2.796875, "learning_rate": 1.0732322657653354e-05, "loss": 0.1181, "step": 22425 }, { "epoch": 139.29283489096574, "grad_norm": 1.3203125, "learning_rate": 1.0730299599347219e-05, "loss": 0.0875, "step": 22426 }, { "epoch": 139.29906542056074, "grad_norm": 1.265625, "learning_rate": 1.0728276679631274e-05, "loss": 0.0995, "step": 22427 }, { "epoch": 139.30529595015577, "grad_norm": 2.28125, "learning_rate": 1.072625389852516e-05, "loss": 0.112, "step": 22428 }, { "epoch": 139.31152647975077, "grad_norm": 2.140625, "learning_rate": 1.0724231256048512e-05, "loss": 0.1531, "step": 22429 }, { "epoch": 139.3177570093458, "grad_norm": 1.328125, "learning_rate": 1.0722208752220986e-05, "loss": 0.1078, "step": 22430 }, { "epoch": 139.3239875389408, "grad_norm": 2.484375, "learning_rate": 1.0720186387062228e-05, "loss": 0.187, "step": 22431 }, { "epoch": 139.33021806853583, "grad_norm": 1.8671875, "learning_rate": 1.0718164160591868e-05, "loss": 0.1129, "step": 22432 }, { "epoch": 139.33644859813083, "grad_norm": 0.75390625, "learning_rate": 1.0716142072829552e-05, "loss": 0.0848, "step": 22433 }, { "epoch": 139.34267912772586, "grad_norm": 2.46875, "learning_rate": 1.0714120123794927e-05, "loss": 0.1731, "step": 22434 }, { "epoch": 139.34890965732086, "grad_norm": 2.609375, "learning_rate": 1.0712098313507624e-05, "loss": 0.2266, "step": 22435 }, { "epoch": 139.3551401869159, "grad_norm": 2.734375, "learning_rate": 1.0710076641987269e-05, "loss": 0.2572, "step": 22436 }, { "epoch": 139.3613707165109, "grad_norm": 2.265625, "learning_rate": 1.0708055109253506e-05, "loss": 0.1716, "step": 22437 }, { "epoch": 139.36760124610592, "grad_norm": 1.4296875, "learning_rate": 1.0706033715325977e-05, "loss": 0.1191, "step": 22438 }, { "epoch": 139.37383177570092, "grad_norm": 2.328125, "learning_rate": 1.07040124602243e-05, "loss": 0.1988, "step": 22439 }, { "epoch": 139.38006230529595, "grad_norm": 2.0, "learning_rate": 1.0701991343968119e-05, "loss": 0.1133, "step": 22440 }, { "epoch": 139.38629283489095, "grad_norm": 1.6015625, "learning_rate": 1.0699970366577047e-05, "loss": 0.1357, "step": 22441 }, { "epoch": 139.39252336448598, "grad_norm": 2.140625, "learning_rate": 1.069794952807073e-05, "loss": 0.1652, "step": 22442 }, { "epoch": 139.39875389408098, "grad_norm": 3.109375, "learning_rate": 1.0695928828468777e-05, "loss": 0.2806, "step": 22443 }, { "epoch": 139.404984423676, "grad_norm": 2.90625, "learning_rate": 1.0693908267790823e-05, "loss": 0.2832, "step": 22444 }, { "epoch": 139.41121495327104, "grad_norm": 3.265625, "learning_rate": 1.06918878460565e-05, "loss": 0.249, "step": 22445 }, { "epoch": 139.41744548286604, "grad_norm": 2.265625, "learning_rate": 1.0689867563285413e-05, "loss": 0.2037, "step": 22446 }, { "epoch": 139.42367601246107, "grad_norm": 3.21875, "learning_rate": 1.0687847419497204e-05, "loss": 0.2657, "step": 22447 }, { "epoch": 139.42990654205607, "grad_norm": 1.9921875, "learning_rate": 1.0685827414711472e-05, "loss": 0.2056, "step": 22448 }, { "epoch": 139.4361370716511, "grad_norm": 2.171875, "learning_rate": 1.0683807548947852e-05, "loss": 0.1883, "step": 22449 }, { "epoch": 139.4423676012461, "grad_norm": 1.515625, "learning_rate": 1.0681787822225947e-05, "loss": 0.1397, "step": 22450 }, { "epoch": 139.44859813084113, "grad_norm": 1.90625, "learning_rate": 1.0679768234565381e-05, "loss": 0.2243, "step": 22451 }, { "epoch": 139.45482866043614, "grad_norm": 2.390625, "learning_rate": 1.0677748785985769e-05, "loss": 0.2065, "step": 22452 }, { "epoch": 139.46105919003116, "grad_norm": 1.3046875, "learning_rate": 1.0675729476506744e-05, "loss": 0.101, "step": 22453 }, { "epoch": 139.46728971962617, "grad_norm": 4.53125, "learning_rate": 1.0673710306147877e-05, "loss": 0.2511, "step": 22454 }, { "epoch": 139.4735202492212, "grad_norm": 1.765625, "learning_rate": 1.0671691274928799e-05, "loss": 0.1642, "step": 22455 }, { "epoch": 139.4797507788162, "grad_norm": 2.34375, "learning_rate": 1.066967238286912e-05, "loss": 0.1764, "step": 22456 }, { "epoch": 139.48598130841123, "grad_norm": 1.8984375, "learning_rate": 1.0667653629988458e-05, "loss": 0.1758, "step": 22457 }, { "epoch": 139.49221183800623, "grad_norm": 3.390625, "learning_rate": 1.06656350163064e-05, "loss": 0.2747, "step": 22458 }, { "epoch": 139.49844236760126, "grad_norm": 1.8359375, "learning_rate": 1.0663616541842566e-05, "loss": 0.1418, "step": 22459 }, { "epoch": 139.50467289719626, "grad_norm": 1.8125, "learning_rate": 1.0661598206616546e-05, "loss": 0.0974, "step": 22460 }, { "epoch": 139.5109034267913, "grad_norm": 3.125, "learning_rate": 1.065958001064796e-05, "loss": 0.3595, "step": 22461 }, { "epoch": 139.5171339563863, "grad_norm": 2.6875, "learning_rate": 1.065756195395639e-05, "loss": 0.2246, "step": 22462 }, { "epoch": 139.52336448598132, "grad_norm": 2.359375, "learning_rate": 1.0655544036561446e-05, "loss": 0.1883, "step": 22463 }, { "epoch": 139.52959501557632, "grad_norm": 1.9609375, "learning_rate": 1.0653526258482736e-05, "loss": 0.1614, "step": 22464 }, { "epoch": 139.53582554517135, "grad_norm": 1.15625, "learning_rate": 1.0651508619739836e-05, "loss": 0.1034, "step": 22465 }, { "epoch": 139.54205607476635, "grad_norm": 3.09375, "learning_rate": 1.0649491120352363e-05, "loss": 0.2074, "step": 22466 }, { "epoch": 139.54828660436138, "grad_norm": 0.91796875, "learning_rate": 1.0647473760339891e-05, "loss": 0.0898, "step": 22467 }, { "epoch": 139.55451713395638, "grad_norm": 1.375, "learning_rate": 1.064545653972203e-05, "loss": 0.131, "step": 22468 }, { "epoch": 139.5607476635514, "grad_norm": 2.5, "learning_rate": 1.0643439458518356e-05, "loss": 0.2037, "step": 22469 }, { "epoch": 139.5669781931464, "grad_norm": 1.4140625, "learning_rate": 1.064142251674847e-05, "loss": 0.1345, "step": 22470 }, { "epoch": 139.57320872274144, "grad_norm": 1.6015625, "learning_rate": 1.0639405714431966e-05, "loss": 0.1178, "step": 22471 }, { "epoch": 139.57943925233644, "grad_norm": 1.9453125, "learning_rate": 1.0637389051588426e-05, "loss": 0.14, "step": 22472 }, { "epoch": 139.58566978193147, "grad_norm": 3.03125, "learning_rate": 1.0635372528237422e-05, "loss": 0.2019, "step": 22473 }, { "epoch": 139.59190031152647, "grad_norm": 1.3046875, "learning_rate": 1.0633356144398553e-05, "loss": 0.1337, "step": 22474 }, { "epoch": 139.5981308411215, "grad_norm": 2.265625, "learning_rate": 1.0631339900091414e-05, "loss": 0.2351, "step": 22475 }, { "epoch": 139.6043613707165, "grad_norm": 3.234375, "learning_rate": 1.0629323795335561e-05, "loss": 0.2395, "step": 22476 }, { "epoch": 139.61059190031153, "grad_norm": 2.828125, "learning_rate": 1.062730783015059e-05, "loss": 0.169, "step": 22477 }, { "epoch": 139.61682242990653, "grad_norm": 2.390625, "learning_rate": 1.0625292004556086e-05, "loss": 0.1953, "step": 22478 }, { "epoch": 139.62305295950156, "grad_norm": 2.5625, "learning_rate": 1.062327631857162e-05, "loss": 0.3823, "step": 22479 }, { "epoch": 139.62928348909657, "grad_norm": 1.6953125, "learning_rate": 1.062126077221676e-05, "loss": 0.1347, "step": 22480 }, { "epoch": 139.6355140186916, "grad_norm": 2.453125, "learning_rate": 1.0619245365511088e-05, "loss": 0.21, "step": 22481 }, { "epoch": 139.6417445482866, "grad_norm": 2.53125, "learning_rate": 1.0617230098474188e-05, "loss": 0.2841, "step": 22482 }, { "epoch": 139.64797507788163, "grad_norm": 2.703125, "learning_rate": 1.0615214971125616e-05, "loss": 0.2199, "step": 22483 }, { "epoch": 139.65420560747663, "grad_norm": 2.828125, "learning_rate": 1.0613199983484953e-05, "loss": 0.1987, "step": 22484 }, { "epoch": 139.66043613707166, "grad_norm": 1.9765625, "learning_rate": 1.0611185135571777e-05, "loss": 0.1515, "step": 22485 }, { "epoch": 139.66666666666666, "grad_norm": 1.234375, "learning_rate": 1.0609170427405645e-05, "loss": 0.1476, "step": 22486 }, { "epoch": 139.6728971962617, "grad_norm": 1.71875, "learning_rate": 1.0607155859006118e-05, "loss": 0.1054, "step": 22487 }, { "epoch": 139.6791277258567, "grad_norm": 2.34375, "learning_rate": 1.0605141430392773e-05, "loss": 0.2229, "step": 22488 }, { "epoch": 139.68535825545172, "grad_norm": 1.8046875, "learning_rate": 1.060312714158517e-05, "loss": 0.1495, "step": 22489 }, { "epoch": 139.69158878504672, "grad_norm": 1.25, "learning_rate": 1.0601112992602883e-05, "loss": 0.1132, "step": 22490 }, { "epoch": 139.69781931464175, "grad_norm": 1.5625, "learning_rate": 1.0599098983465463e-05, "loss": 0.1541, "step": 22491 }, { "epoch": 139.70404984423675, "grad_norm": 1.5, "learning_rate": 1.0597085114192467e-05, "loss": 0.1022, "step": 22492 }, { "epoch": 139.71028037383178, "grad_norm": 1.65625, "learning_rate": 1.0595071384803456e-05, "loss": 0.1293, "step": 22493 }, { "epoch": 139.71651090342678, "grad_norm": 3.671875, "learning_rate": 1.0593057795318004e-05, "loss": 0.2934, "step": 22494 }, { "epoch": 139.7227414330218, "grad_norm": 1.5234375, "learning_rate": 1.0591044345755643e-05, "loss": 0.102, "step": 22495 }, { "epoch": 139.7289719626168, "grad_norm": 2.34375, "learning_rate": 1.0589031036135941e-05, "loss": 0.1218, "step": 22496 }, { "epoch": 139.73520249221184, "grad_norm": 1.8046875, "learning_rate": 1.0587017866478457e-05, "loss": 0.1008, "step": 22497 }, { "epoch": 139.74143302180684, "grad_norm": 1.703125, "learning_rate": 1.058500483680274e-05, "loss": 0.1525, "step": 22498 }, { "epoch": 139.74766355140187, "grad_norm": 2.46875, "learning_rate": 1.0582991947128324e-05, "loss": 0.1411, "step": 22499 }, { "epoch": 139.75389408099687, "grad_norm": 2.828125, "learning_rate": 1.0580979197474774e-05, "loss": 0.1998, "step": 22500 }, { "epoch": 139.7601246105919, "grad_norm": 2.546875, "learning_rate": 1.0578966587861647e-05, "loss": 0.2899, "step": 22501 }, { "epoch": 139.7663551401869, "grad_norm": 1.8671875, "learning_rate": 1.057695411830847e-05, "loss": 0.1658, "step": 22502 }, { "epoch": 139.77258566978193, "grad_norm": 2.9375, "learning_rate": 1.0574941788834797e-05, "loss": 0.26, "step": 22503 }, { "epoch": 139.77881619937693, "grad_norm": 2.140625, "learning_rate": 1.057292959946018e-05, "loss": 0.1067, "step": 22504 }, { "epoch": 139.78504672897196, "grad_norm": 2.9375, "learning_rate": 1.0570917550204156e-05, "loss": 0.2237, "step": 22505 }, { "epoch": 139.79127725856696, "grad_norm": 1.578125, "learning_rate": 1.0568905641086258e-05, "loss": 0.1334, "step": 22506 }, { "epoch": 139.797507788162, "grad_norm": 1.28125, "learning_rate": 1.0566893872126032e-05, "loss": 0.1164, "step": 22507 }, { "epoch": 139.80373831775702, "grad_norm": 2.53125, "learning_rate": 1.0564882243343025e-05, "loss": 0.2126, "step": 22508 }, { "epoch": 139.80996884735202, "grad_norm": 1.7734375, "learning_rate": 1.056287075475676e-05, "loss": 0.1432, "step": 22509 }, { "epoch": 139.81619937694705, "grad_norm": 2.078125, "learning_rate": 1.056085940638679e-05, "loss": 0.1861, "step": 22510 }, { "epoch": 139.82242990654206, "grad_norm": 2.03125, "learning_rate": 1.0558848198252628e-05, "loss": 0.137, "step": 22511 }, { "epoch": 139.82866043613708, "grad_norm": 3.140625, "learning_rate": 1.055683713037383e-05, "loss": 0.1756, "step": 22512 }, { "epoch": 139.8348909657321, "grad_norm": 2.015625, "learning_rate": 1.0554826202769908e-05, "loss": 0.2221, "step": 22513 }, { "epoch": 139.84112149532712, "grad_norm": 1.3515625, "learning_rate": 1.0552815415460402e-05, "loss": 0.1024, "step": 22514 }, { "epoch": 139.84735202492212, "grad_norm": 3.484375, "learning_rate": 1.0550804768464836e-05, "loss": 0.2521, "step": 22515 }, { "epoch": 139.85358255451715, "grad_norm": 2.359375, "learning_rate": 1.0548794261802758e-05, "loss": 0.1926, "step": 22516 }, { "epoch": 139.85981308411215, "grad_norm": 1.1328125, "learning_rate": 1.0546783895493675e-05, "loss": 0.0961, "step": 22517 }, { "epoch": 139.86604361370718, "grad_norm": 2.234375, "learning_rate": 1.0544773669557106e-05, "loss": 0.1543, "step": 22518 }, { "epoch": 139.87227414330218, "grad_norm": 1.4296875, "learning_rate": 1.0542763584012589e-05, "loss": 0.1085, "step": 22519 }, { "epoch": 139.8785046728972, "grad_norm": 3.046875, "learning_rate": 1.0540753638879648e-05, "loss": 0.2345, "step": 22520 }, { "epoch": 139.8847352024922, "grad_norm": 2.859375, "learning_rate": 1.053874383417779e-05, "loss": 0.1681, "step": 22521 }, { "epoch": 139.89096573208724, "grad_norm": 2.171875, "learning_rate": 1.053673416992654e-05, "loss": 0.1389, "step": 22522 }, { "epoch": 139.89719626168224, "grad_norm": 2.203125, "learning_rate": 1.053472464614543e-05, "loss": 0.2547, "step": 22523 }, { "epoch": 139.90342679127727, "grad_norm": 1.828125, "learning_rate": 1.0532715262853967e-05, "loss": 0.168, "step": 22524 }, { "epoch": 139.90965732087227, "grad_norm": 1.53125, "learning_rate": 1.0530706020071654e-05, "loss": 0.1251, "step": 22525 }, { "epoch": 139.9158878504673, "grad_norm": 3.25, "learning_rate": 1.0528696917818017e-05, "loss": 0.2782, "step": 22526 }, { "epoch": 139.9221183800623, "grad_norm": 2.328125, "learning_rate": 1.052668795611258e-05, "loss": 0.2276, "step": 22527 }, { "epoch": 139.92834890965733, "grad_norm": 1.8359375, "learning_rate": 1.052467913497483e-05, "loss": 0.102, "step": 22528 }, { "epoch": 139.93457943925233, "grad_norm": 2.5625, "learning_rate": 1.0522670454424302e-05, "loss": 0.2941, "step": 22529 }, { "epoch": 139.94080996884736, "grad_norm": 2.578125, "learning_rate": 1.052066191448048e-05, "loss": 0.296, "step": 22530 }, { "epoch": 139.94704049844236, "grad_norm": 1.15625, "learning_rate": 1.0518653515162897e-05, "loss": 0.0906, "step": 22531 }, { "epoch": 139.9532710280374, "grad_norm": 3.640625, "learning_rate": 1.0516645256491034e-05, "loss": 0.2599, "step": 22532 }, { "epoch": 139.9595015576324, "grad_norm": 1.21875, "learning_rate": 1.0514637138484407e-05, "loss": 0.1084, "step": 22533 }, { "epoch": 139.96573208722742, "grad_norm": 1.9609375, "learning_rate": 1.0512629161162529e-05, "loss": 0.1113, "step": 22534 }, { "epoch": 139.97196261682242, "grad_norm": 2.9375, "learning_rate": 1.0510621324544883e-05, "loss": 0.2113, "step": 22535 }, { "epoch": 139.97819314641745, "grad_norm": 1.453125, "learning_rate": 1.050861362865099e-05, "loss": 0.1066, "step": 22536 }, { "epoch": 139.98442367601245, "grad_norm": 2.859375, "learning_rate": 1.050660607350033e-05, "loss": 0.2415, "step": 22537 }, { "epoch": 139.99065420560748, "grad_norm": 2.0625, "learning_rate": 1.0504598659112417e-05, "loss": 0.1804, "step": 22538 }, { "epoch": 139.99688473520249, "grad_norm": 2.28125, "learning_rate": 1.050259138550673e-05, "loss": 0.21, "step": 22539 }, { "epoch": 140.0, "grad_norm": 1.2109375, "learning_rate": 1.0500584252702771e-05, "loss": 0.0476, "step": 22540 }, { "epoch": 140.00623052959503, "grad_norm": 2.671875, "learning_rate": 1.049857726072005e-05, "loss": 0.2179, "step": 22541 }, { "epoch": 140.01246105919003, "grad_norm": 3.640625, "learning_rate": 1.0496570409578042e-05, "loss": 0.1717, "step": 22542 }, { "epoch": 140.01869158878506, "grad_norm": 2.0625, "learning_rate": 1.0494563699296234e-05, "loss": 0.1971, "step": 22543 }, { "epoch": 140.02492211838006, "grad_norm": 1.7109375, "learning_rate": 1.0492557129894121e-05, "loss": 0.1376, "step": 22544 }, { "epoch": 140.0311526479751, "grad_norm": 1.34375, "learning_rate": 1.0490550701391203e-05, "loss": 0.1362, "step": 22545 }, { "epoch": 140.0373831775701, "grad_norm": 1.7578125, "learning_rate": 1.0488544413806947e-05, "loss": 0.1333, "step": 22546 }, { "epoch": 140.04361370716512, "grad_norm": 1.859375, "learning_rate": 1.048653826716085e-05, "loss": 0.1082, "step": 22547 }, { "epoch": 140.04984423676012, "grad_norm": 1.578125, "learning_rate": 1.0484532261472405e-05, "loss": 0.1324, "step": 22548 }, { "epoch": 140.05607476635515, "grad_norm": 1.0234375, "learning_rate": 1.0482526396761082e-05, "loss": 0.0982, "step": 22549 }, { "epoch": 140.06230529595015, "grad_norm": 2.84375, "learning_rate": 1.0480520673046356e-05, "loss": 0.1749, "step": 22550 }, { "epoch": 140.06853582554518, "grad_norm": 1.5546875, "learning_rate": 1.0478515090347715e-05, "loss": 0.1697, "step": 22551 }, { "epoch": 140.07476635514018, "grad_norm": 2.15625, "learning_rate": 1.047650964868464e-05, "loss": 0.1259, "step": 22552 }, { "epoch": 140.0809968847352, "grad_norm": 2.421875, "learning_rate": 1.0474504348076613e-05, "loss": 0.1163, "step": 22553 }, { "epoch": 140.08722741433021, "grad_norm": 3.578125, "learning_rate": 1.0472499188543097e-05, "loss": 0.4136, "step": 22554 }, { "epoch": 140.09345794392524, "grad_norm": 1.328125, "learning_rate": 1.0470494170103582e-05, "loss": 0.1206, "step": 22555 }, { "epoch": 140.09968847352025, "grad_norm": 1.5234375, "learning_rate": 1.0468489292777519e-05, "loss": 0.1457, "step": 22556 }, { "epoch": 140.10591900311528, "grad_norm": 2.453125, "learning_rate": 1.0466484556584404e-05, "loss": 0.1619, "step": 22557 }, { "epoch": 140.11214953271028, "grad_norm": 2.890625, "learning_rate": 1.0464479961543688e-05, "loss": 0.3142, "step": 22558 }, { "epoch": 140.1183800623053, "grad_norm": 3.09375, "learning_rate": 1.0462475507674848e-05, "loss": 0.3012, "step": 22559 }, { "epoch": 140.1246105919003, "grad_norm": 2.25, "learning_rate": 1.0460471194997362e-05, "loss": 0.2334, "step": 22560 }, { "epoch": 140.13084112149534, "grad_norm": 2.609375, "learning_rate": 1.0458467023530685e-05, "loss": 0.1538, "step": 22561 }, { "epoch": 140.13707165109034, "grad_norm": 1.484375, "learning_rate": 1.0456462993294272e-05, "loss": 0.1279, "step": 22562 }, { "epoch": 140.14330218068537, "grad_norm": 1.8515625, "learning_rate": 1.0454459104307602e-05, "loss": 0.1211, "step": 22563 }, { "epoch": 140.14953271028037, "grad_norm": 2.5, "learning_rate": 1.0452455356590138e-05, "loss": 0.2191, "step": 22564 }, { "epoch": 140.1557632398754, "grad_norm": 2.171875, "learning_rate": 1.0450451750161328e-05, "loss": 0.1341, "step": 22565 }, { "epoch": 140.1619937694704, "grad_norm": 1.9921875, "learning_rate": 1.0448448285040638e-05, "loss": 0.174, "step": 22566 }, { "epoch": 140.16822429906543, "grad_norm": 2.046875, "learning_rate": 1.0446444961247536e-05, "loss": 0.1709, "step": 22567 }, { "epoch": 140.17445482866043, "grad_norm": 2.484375, "learning_rate": 1.0444441778801472e-05, "loss": 0.2622, "step": 22568 }, { "epoch": 140.18068535825546, "grad_norm": 2.15625, "learning_rate": 1.0442438737721888e-05, "loss": 0.1575, "step": 22569 }, { "epoch": 140.18691588785046, "grad_norm": 1.96875, "learning_rate": 1.044043583802825e-05, "loss": 0.1365, "step": 22570 }, { "epoch": 140.1931464174455, "grad_norm": 1.71875, "learning_rate": 1.0438433079740017e-05, "loss": 0.1371, "step": 22571 }, { "epoch": 140.1993769470405, "grad_norm": 2.140625, "learning_rate": 1.0436430462876625e-05, "loss": 0.1819, "step": 22572 }, { "epoch": 140.20560747663552, "grad_norm": 1.34375, "learning_rate": 1.0434427987457532e-05, "loss": 0.103, "step": 22573 }, { "epoch": 140.21183800623052, "grad_norm": 2.359375, "learning_rate": 1.0432425653502196e-05, "loss": 0.1155, "step": 22574 }, { "epoch": 140.21806853582555, "grad_norm": 1.734375, "learning_rate": 1.0430423461030054e-05, "loss": 0.1305, "step": 22575 }, { "epoch": 140.22429906542055, "grad_norm": 2.09375, "learning_rate": 1.0428421410060541e-05, "loss": 0.1314, "step": 22576 }, { "epoch": 140.23052959501558, "grad_norm": 2.6875, "learning_rate": 1.042641950061311e-05, "loss": 0.2742, "step": 22577 }, { "epoch": 140.23676012461058, "grad_norm": 1.828125, "learning_rate": 1.0424417732707217e-05, "loss": 0.1641, "step": 22578 }, { "epoch": 140.2429906542056, "grad_norm": 1.796875, "learning_rate": 1.0422416106362284e-05, "loss": 0.255, "step": 22579 }, { "epoch": 140.2492211838006, "grad_norm": 1.3671875, "learning_rate": 1.0420414621597769e-05, "loss": 0.1234, "step": 22580 }, { "epoch": 140.25545171339564, "grad_norm": 2.34375, "learning_rate": 1.0418413278433091e-05, "loss": 0.1971, "step": 22581 }, { "epoch": 140.26168224299064, "grad_norm": 1.890625, "learning_rate": 1.041641207688771e-05, "loss": 0.1094, "step": 22582 }, { "epoch": 140.26791277258567, "grad_norm": 1.7890625, "learning_rate": 1.041441101698104e-05, "loss": 0.1438, "step": 22583 }, { "epoch": 140.27414330218068, "grad_norm": 1.5390625, "learning_rate": 1.0412410098732525e-05, "loss": 0.1063, "step": 22584 }, { "epoch": 140.2803738317757, "grad_norm": 1.6484375, "learning_rate": 1.04104093221616e-05, "loss": 0.1478, "step": 22585 }, { "epoch": 140.2866043613707, "grad_norm": 2.46875, "learning_rate": 1.0408408687287704e-05, "loss": 0.2222, "step": 22586 }, { "epoch": 140.29283489096574, "grad_norm": 1.4140625, "learning_rate": 1.0406408194130259e-05, "loss": 0.1002, "step": 22587 }, { "epoch": 140.29906542056074, "grad_norm": 1.890625, "learning_rate": 1.0404407842708689e-05, "loss": 0.1232, "step": 22588 }, { "epoch": 140.30529595015577, "grad_norm": 1.875, "learning_rate": 1.0402407633042424e-05, "loss": 0.1462, "step": 22589 }, { "epoch": 140.31152647975077, "grad_norm": 1.234375, "learning_rate": 1.0400407565150905e-05, "loss": 0.1199, "step": 22590 }, { "epoch": 140.3177570093458, "grad_norm": 1.046875, "learning_rate": 1.0398407639053537e-05, "loss": 0.09, "step": 22591 }, { "epoch": 140.3239875389408, "grad_norm": 1.515625, "learning_rate": 1.0396407854769751e-05, "loss": 0.1385, "step": 22592 }, { "epoch": 140.33021806853583, "grad_norm": 2.21875, "learning_rate": 1.0394408212318981e-05, "loss": 0.1841, "step": 22593 }, { "epoch": 140.33644859813083, "grad_norm": 2.21875, "learning_rate": 1.039240871172064e-05, "loss": 0.1542, "step": 22594 }, { "epoch": 140.34267912772586, "grad_norm": 1.7421875, "learning_rate": 1.0390409352994134e-05, "loss": 0.1373, "step": 22595 }, { "epoch": 140.34890965732086, "grad_norm": 2.96875, "learning_rate": 1.0388410136158894e-05, "loss": 0.2511, "step": 22596 }, { "epoch": 140.3551401869159, "grad_norm": 1.7734375, "learning_rate": 1.0386411061234347e-05, "loss": 0.1583, "step": 22597 }, { "epoch": 140.3613707165109, "grad_norm": 2.953125, "learning_rate": 1.0384412128239885e-05, "loss": 0.1957, "step": 22598 }, { "epoch": 140.36760124610592, "grad_norm": 3.796875, "learning_rate": 1.0382413337194944e-05, "loss": 0.1655, "step": 22599 }, { "epoch": 140.37383177570092, "grad_norm": 1.59375, "learning_rate": 1.0380414688118917e-05, "loss": 0.1677, "step": 22600 }, { "epoch": 140.38006230529595, "grad_norm": 2.15625, "learning_rate": 1.0378416181031233e-05, "loss": 0.1536, "step": 22601 }, { "epoch": 140.38629283489095, "grad_norm": 2.0625, "learning_rate": 1.0376417815951286e-05, "loss": 0.0844, "step": 22602 }, { "epoch": 140.39252336448598, "grad_norm": 1.40625, "learning_rate": 1.0374419592898493e-05, "loss": 0.1039, "step": 22603 }, { "epoch": 140.39875389408098, "grad_norm": 2.890625, "learning_rate": 1.037242151189227e-05, "loss": 0.2625, "step": 22604 }, { "epoch": 140.404984423676, "grad_norm": 1.59375, "learning_rate": 1.0370423572952001e-05, "loss": 0.159, "step": 22605 }, { "epoch": 140.41121495327104, "grad_norm": 1.078125, "learning_rate": 1.0368425776097115e-05, "loss": 0.1109, "step": 22606 }, { "epoch": 140.41744548286604, "grad_norm": 2.65625, "learning_rate": 1.0366428121346992e-05, "loss": 0.2242, "step": 22607 }, { "epoch": 140.42367601246107, "grad_norm": 1.828125, "learning_rate": 1.0364430608721055e-05, "loss": 0.1491, "step": 22608 }, { "epoch": 140.42990654205607, "grad_norm": 1.5, "learning_rate": 1.036243323823868e-05, "loss": 0.1273, "step": 22609 }, { "epoch": 140.4361370716511, "grad_norm": 1.6640625, "learning_rate": 1.0360436009919283e-05, "loss": 0.1225, "step": 22610 }, { "epoch": 140.4423676012461, "grad_norm": 2.875, "learning_rate": 1.035843892378226e-05, "loss": 0.2325, "step": 22611 }, { "epoch": 140.44859813084113, "grad_norm": 2.265625, "learning_rate": 1.035644197984702e-05, "loss": 0.1547, "step": 22612 }, { "epoch": 140.45482866043614, "grad_norm": 2.90625, "learning_rate": 1.0354445178132925e-05, "loss": 0.1515, "step": 22613 }, { "epoch": 140.46105919003116, "grad_norm": 2.28125, "learning_rate": 1.0352448518659386e-05, "loss": 0.0991, "step": 22614 }, { "epoch": 140.46728971962617, "grad_norm": 1.2265625, "learning_rate": 1.0350452001445796e-05, "loss": 0.0975, "step": 22615 }, { "epoch": 140.4735202492212, "grad_norm": 2.515625, "learning_rate": 1.0348455626511553e-05, "loss": 0.236, "step": 22616 }, { "epoch": 140.4797507788162, "grad_norm": 2.109375, "learning_rate": 1.0346459393876026e-05, "loss": 0.1591, "step": 22617 }, { "epoch": 140.48598130841123, "grad_norm": 1.6171875, "learning_rate": 1.0344463303558627e-05, "loss": 0.1207, "step": 22618 }, { "epoch": 140.49221183800623, "grad_norm": 2.34375, "learning_rate": 1.034246735557872e-05, "loss": 0.1553, "step": 22619 }, { "epoch": 140.49844236760126, "grad_norm": 1.4765625, "learning_rate": 1.0340471549955707e-05, "loss": 0.1288, "step": 22620 }, { "epoch": 140.50467289719626, "grad_norm": 2.0625, "learning_rate": 1.0338475886708959e-05, "loss": 0.1623, "step": 22621 }, { "epoch": 140.5109034267913, "grad_norm": 1.9921875, "learning_rate": 1.0336480365857862e-05, "loss": 0.1011, "step": 22622 }, { "epoch": 140.5171339563863, "grad_norm": 3.140625, "learning_rate": 1.0334484987421808e-05, "loss": 0.2083, "step": 22623 }, { "epoch": 140.52336448598132, "grad_norm": 2.6875, "learning_rate": 1.0332489751420161e-05, "loss": 0.237, "step": 22624 }, { "epoch": 140.52959501557632, "grad_norm": 2.390625, "learning_rate": 1.0330494657872312e-05, "loss": 0.1735, "step": 22625 }, { "epoch": 140.53582554517135, "grad_norm": 2.15625, "learning_rate": 1.0328499706797624e-05, "loss": 0.1878, "step": 22626 }, { "epoch": 140.54205607476635, "grad_norm": 2.765625, "learning_rate": 1.0326504898215487e-05, "loss": 0.174, "step": 22627 }, { "epoch": 140.54828660436138, "grad_norm": 2.265625, "learning_rate": 1.0324510232145258e-05, "loss": 0.2009, "step": 22628 }, { "epoch": 140.55451713395638, "grad_norm": 1.296875, "learning_rate": 1.032251570860632e-05, "loss": 0.1343, "step": 22629 }, { "epoch": 140.5607476635514, "grad_norm": 1.78125, "learning_rate": 1.0320521327618053e-05, "loss": 0.1251, "step": 22630 }, { "epoch": 140.5669781931464, "grad_norm": 1.9453125, "learning_rate": 1.0318527089199815e-05, "loss": 0.1659, "step": 22631 }, { "epoch": 140.57320872274144, "grad_norm": 2.046875, "learning_rate": 1.031653299337097e-05, "loss": 0.1998, "step": 22632 }, { "epoch": 140.57943925233644, "grad_norm": 1.15625, "learning_rate": 1.0314539040150892e-05, "loss": 0.1432, "step": 22633 }, { "epoch": 140.58566978193147, "grad_norm": 1.421875, "learning_rate": 1.0312545229558954e-05, "loss": 0.096, "step": 22634 }, { "epoch": 140.59190031152647, "grad_norm": 3.8125, "learning_rate": 1.0310551561614505e-05, "loss": 0.2582, "step": 22635 }, { "epoch": 140.5981308411215, "grad_norm": 1.90625, "learning_rate": 1.0308558036336912e-05, "loss": 0.1622, "step": 22636 }, { "epoch": 140.6043613707165, "grad_norm": 3.171875, "learning_rate": 1.0306564653745553e-05, "loss": 0.2444, "step": 22637 }, { "epoch": 140.61059190031153, "grad_norm": 1.5, "learning_rate": 1.0304571413859776e-05, "loss": 0.129, "step": 22638 }, { "epoch": 140.61682242990653, "grad_norm": 1.7265625, "learning_rate": 1.0302578316698927e-05, "loss": 0.1191, "step": 22639 }, { "epoch": 140.62305295950156, "grad_norm": 1.28125, "learning_rate": 1.0300585362282378e-05, "loss": 0.1062, "step": 22640 }, { "epoch": 140.62928348909657, "grad_norm": 0.98828125, "learning_rate": 1.0298592550629488e-05, "loss": 0.1006, "step": 22641 }, { "epoch": 140.6355140186916, "grad_norm": 2.015625, "learning_rate": 1.02965998817596e-05, "loss": 0.1366, "step": 22642 }, { "epoch": 140.6417445482866, "grad_norm": 1.5234375, "learning_rate": 1.0294607355692071e-05, "loss": 0.1259, "step": 22643 }, { "epoch": 140.64797507788163, "grad_norm": 1.453125, "learning_rate": 1.0292614972446265e-05, "loss": 0.1008, "step": 22644 }, { "epoch": 140.65420560747663, "grad_norm": 2.4375, "learning_rate": 1.029062273204152e-05, "loss": 0.2054, "step": 22645 }, { "epoch": 140.66043613707166, "grad_norm": 2.8125, "learning_rate": 1.0288630634497182e-05, "loss": 0.2278, "step": 22646 }, { "epoch": 140.66666666666666, "grad_norm": 2.296875, "learning_rate": 1.0286638679832602e-05, "loss": 0.1676, "step": 22647 }, { "epoch": 140.6728971962617, "grad_norm": 2.09375, "learning_rate": 1.028464686806713e-05, "loss": 0.1749, "step": 22648 }, { "epoch": 140.6791277258567, "grad_norm": 2.515625, "learning_rate": 1.0282655199220116e-05, "loss": 0.3363, "step": 22649 }, { "epoch": 140.68535825545172, "grad_norm": 1.21875, "learning_rate": 1.0280663673310898e-05, "loss": 0.1255, "step": 22650 }, { "epoch": 140.69158878504672, "grad_norm": 3.5625, "learning_rate": 1.0278672290358807e-05, "loss": 0.1902, "step": 22651 }, { "epoch": 140.69781931464175, "grad_norm": 2.75, "learning_rate": 1.0276681050383194e-05, "loss": 0.1564, "step": 22652 }, { "epoch": 140.70404984423675, "grad_norm": 2.5625, "learning_rate": 1.0274689953403407e-05, "loss": 0.172, "step": 22653 }, { "epoch": 140.71028037383178, "grad_norm": 2.5, "learning_rate": 1.0272698999438766e-05, "loss": 0.2122, "step": 22654 }, { "epoch": 140.71651090342678, "grad_norm": 2.171875, "learning_rate": 1.0270708188508612e-05, "loss": 0.1492, "step": 22655 }, { "epoch": 140.7227414330218, "grad_norm": 1.71875, "learning_rate": 1.0268717520632298e-05, "loss": 0.1549, "step": 22656 }, { "epoch": 140.7289719626168, "grad_norm": 1.8046875, "learning_rate": 1.0266726995829144e-05, "loss": 0.1414, "step": 22657 }, { "epoch": 140.73520249221184, "grad_norm": 3.0, "learning_rate": 1.0264736614118472e-05, "loss": 0.3226, "step": 22658 }, { "epoch": 140.74143302180684, "grad_norm": 2.265625, "learning_rate": 1.0262746375519624e-05, "loss": 0.1532, "step": 22659 }, { "epoch": 140.74766355140187, "grad_norm": 1.4609375, "learning_rate": 1.026075628005194e-05, "loss": 0.118, "step": 22660 }, { "epoch": 140.75389408099687, "grad_norm": 1.3359375, "learning_rate": 1.0258766327734726e-05, "loss": 0.1091, "step": 22661 }, { "epoch": 140.7601246105919, "grad_norm": 1.8515625, "learning_rate": 1.0256776518587319e-05, "loss": 0.1634, "step": 22662 }, { "epoch": 140.7663551401869, "grad_norm": 3.046875, "learning_rate": 1.0254786852629056e-05, "loss": 0.3119, "step": 22663 }, { "epoch": 140.77258566978193, "grad_norm": 1.6015625, "learning_rate": 1.0252797329879251e-05, "loss": 0.1281, "step": 22664 }, { "epoch": 140.77881619937693, "grad_norm": 1.0390625, "learning_rate": 1.0250807950357219e-05, "loss": 0.0989, "step": 22665 }, { "epoch": 140.78504672897196, "grad_norm": 2.3125, "learning_rate": 1.0248818714082287e-05, "loss": 0.1468, "step": 22666 }, { "epoch": 140.79127725856696, "grad_norm": 2.828125, "learning_rate": 1.0246829621073784e-05, "loss": 0.2506, "step": 22667 }, { "epoch": 140.797507788162, "grad_norm": 2.046875, "learning_rate": 1.0244840671351011e-05, "loss": 0.1599, "step": 22668 }, { "epoch": 140.80373831775702, "grad_norm": 2.03125, "learning_rate": 1.0242851864933305e-05, "loss": 0.14, "step": 22669 }, { "epoch": 140.80996884735202, "grad_norm": 2.3125, "learning_rate": 1.0240863201839965e-05, "loss": 0.1493, "step": 22670 }, { "epoch": 140.81619937694705, "grad_norm": 1.859375, "learning_rate": 1.0238874682090318e-05, "loss": 0.1493, "step": 22671 }, { "epoch": 140.82242990654206, "grad_norm": 2.15625, "learning_rate": 1.0236886305703664e-05, "loss": 0.1336, "step": 22672 }, { "epoch": 140.82866043613708, "grad_norm": 1.84375, "learning_rate": 1.023489807269932e-05, "loss": 0.1037, "step": 22673 }, { "epoch": 140.8348909657321, "grad_norm": 1.84375, "learning_rate": 1.0232909983096606e-05, "loss": 0.1226, "step": 22674 }, { "epoch": 140.84112149532712, "grad_norm": 2.78125, "learning_rate": 1.0230922036914815e-05, "loss": 0.189, "step": 22675 }, { "epoch": 140.84735202492212, "grad_norm": 2.109375, "learning_rate": 1.0228934234173267e-05, "loss": 0.1331, "step": 22676 }, { "epoch": 140.85358255451715, "grad_norm": 1.3515625, "learning_rate": 1.0226946574891257e-05, "loss": 0.1359, "step": 22677 }, { "epoch": 140.85981308411215, "grad_norm": 1.2734375, "learning_rate": 1.0224959059088101e-05, "loss": 0.1483, "step": 22678 }, { "epoch": 140.86604361370718, "grad_norm": 1.2578125, "learning_rate": 1.022297168678309e-05, "loss": 0.0862, "step": 22679 }, { "epoch": 140.87227414330218, "grad_norm": 1.1171875, "learning_rate": 1.0220984457995531e-05, "loss": 0.1264, "step": 22680 }, { "epoch": 140.8785046728972, "grad_norm": 2.640625, "learning_rate": 1.0218997372744724e-05, "loss": 0.2394, "step": 22681 }, { "epoch": 140.8847352024922, "grad_norm": 1.3984375, "learning_rate": 1.0217010431049986e-05, "loss": 0.1335, "step": 22682 }, { "epoch": 140.89096573208724, "grad_norm": 1.390625, "learning_rate": 1.0215023632930585e-05, "loss": 0.1167, "step": 22683 }, { "epoch": 140.89719626168224, "grad_norm": 2.203125, "learning_rate": 1.0213036978405823e-05, "loss": 0.1411, "step": 22684 }, { "epoch": 140.90342679127727, "grad_norm": 2.484375, "learning_rate": 1.0211050467495006e-05, "loss": 0.275, "step": 22685 }, { "epoch": 140.90965732087227, "grad_norm": 1.8984375, "learning_rate": 1.0209064100217433e-05, "loss": 0.1418, "step": 22686 }, { "epoch": 140.9158878504673, "grad_norm": 2.234375, "learning_rate": 1.0207077876592374e-05, "loss": 0.2864, "step": 22687 }, { "epoch": 140.9221183800623, "grad_norm": 1.734375, "learning_rate": 1.0205091796639144e-05, "loss": 0.1185, "step": 22688 }, { "epoch": 140.92834890965733, "grad_norm": 1.9140625, "learning_rate": 1.0203105860377007e-05, "loss": 0.2415, "step": 22689 }, { "epoch": 140.93457943925233, "grad_norm": 0.8671875, "learning_rate": 1.0201120067825277e-05, "loss": 0.0967, "step": 22690 }, { "epoch": 140.94080996884736, "grad_norm": 1.828125, "learning_rate": 1.0199134419003214e-05, "loss": 0.1087, "step": 22691 }, { "epoch": 140.94704049844236, "grad_norm": 3.09375, "learning_rate": 1.019714891393012e-05, "loss": 0.2813, "step": 22692 }, { "epoch": 140.9532710280374, "grad_norm": 1.9765625, "learning_rate": 1.0195163552625282e-05, "loss": 0.1537, "step": 22693 }, { "epoch": 140.9595015576324, "grad_norm": 2.203125, "learning_rate": 1.0193178335107964e-05, "loss": 0.1384, "step": 22694 }, { "epoch": 140.96573208722742, "grad_norm": 1.5625, "learning_rate": 1.0191193261397468e-05, "loss": 0.1141, "step": 22695 }, { "epoch": 140.97196261682242, "grad_norm": 0.8203125, "learning_rate": 1.0189208331513057e-05, "loss": 0.0915, "step": 22696 }, { "epoch": 140.97819314641745, "grad_norm": 1.71875, "learning_rate": 1.018722354547402e-05, "loss": 0.1307, "step": 22697 }, { "epoch": 140.98442367601245, "grad_norm": 1.6796875, "learning_rate": 1.0185238903299624e-05, "loss": 0.108, "step": 22698 }, { "epoch": 140.99065420560748, "grad_norm": 2.546875, "learning_rate": 1.0183254405009146e-05, "loss": 0.2482, "step": 22699 }, { "epoch": 140.99688473520249, "grad_norm": 1.703125, "learning_rate": 1.0181270050621874e-05, "loss": 0.1519, "step": 22700 }, { "epoch": 141.0, "grad_norm": 1.890625, "learning_rate": 1.0179285840157069e-05, "loss": 0.0545, "step": 22701 }, { "epoch": 141.00623052959503, "grad_norm": 2.59375, "learning_rate": 1.0177301773633993e-05, "loss": 0.2126, "step": 22702 }, { "epoch": 141.01246105919003, "grad_norm": 2.28125, "learning_rate": 1.0175317851071924e-05, "loss": 0.1345, "step": 22703 }, { "epoch": 141.01869158878506, "grad_norm": 0.87890625, "learning_rate": 1.0173334072490143e-05, "loss": 0.0919, "step": 22704 }, { "epoch": 141.02492211838006, "grad_norm": 1.4921875, "learning_rate": 1.0171350437907894e-05, "loss": 0.1476, "step": 22705 }, { "epoch": 141.0311526479751, "grad_norm": 2.484375, "learning_rate": 1.0169366947344459e-05, "loss": 0.1814, "step": 22706 }, { "epoch": 141.0373831775701, "grad_norm": 1.28125, "learning_rate": 1.01673836008191e-05, "loss": 0.1233, "step": 22707 }, { "epoch": 141.04361370716512, "grad_norm": 2.390625, "learning_rate": 1.0165400398351077e-05, "loss": 0.1993, "step": 22708 }, { "epoch": 141.04984423676012, "grad_norm": 1.890625, "learning_rate": 1.0163417339959645e-05, "loss": 0.0881, "step": 22709 }, { "epoch": 141.05607476635515, "grad_norm": 1.671875, "learning_rate": 1.0161434425664065e-05, "loss": 0.1563, "step": 22710 }, { "epoch": 141.06230529595015, "grad_norm": 2.625, "learning_rate": 1.0159451655483605e-05, "loss": 0.2151, "step": 22711 }, { "epoch": 141.06853582554518, "grad_norm": 2.375, "learning_rate": 1.0157469029437521e-05, "loss": 0.1757, "step": 22712 }, { "epoch": 141.07476635514018, "grad_norm": 1.046875, "learning_rate": 1.015548654754506e-05, "loss": 0.0959, "step": 22713 }, { "epoch": 141.0809968847352, "grad_norm": 2.625, "learning_rate": 1.0153504209825487e-05, "loss": 0.1828, "step": 22714 }, { "epoch": 141.08722741433021, "grad_norm": 3.09375, "learning_rate": 1.0151522016298043e-05, "loss": 0.2599, "step": 22715 }, { "epoch": 141.09345794392524, "grad_norm": 1.6484375, "learning_rate": 1.0149539966981991e-05, "loss": 0.2174, "step": 22716 }, { "epoch": 141.09968847352025, "grad_norm": 3.078125, "learning_rate": 1.014755806189657e-05, "loss": 0.2196, "step": 22717 }, { "epoch": 141.10591900311528, "grad_norm": 2.328125, "learning_rate": 1.014557630106103e-05, "loss": 0.2091, "step": 22718 }, { "epoch": 141.11214953271028, "grad_norm": 2.53125, "learning_rate": 1.0143594684494634e-05, "loss": 0.1858, "step": 22719 }, { "epoch": 141.1183800623053, "grad_norm": 1.703125, "learning_rate": 1.0141613212216616e-05, "loss": 0.1563, "step": 22720 }, { "epoch": 141.1246105919003, "grad_norm": 1.546875, "learning_rate": 1.0139631884246212e-05, "loss": 0.1373, "step": 22721 }, { "epoch": 141.13084112149534, "grad_norm": 2.3125, "learning_rate": 1.0137650700602671e-05, "loss": 0.1781, "step": 22722 }, { "epoch": 141.13707165109034, "grad_norm": 1.2578125, "learning_rate": 1.013566966130525e-05, "loss": 0.1128, "step": 22723 }, { "epoch": 141.14330218068537, "grad_norm": 3.21875, "learning_rate": 1.0133688766373164e-05, "loss": 0.291, "step": 22724 }, { "epoch": 141.14953271028037, "grad_norm": 3.515625, "learning_rate": 1.0131708015825667e-05, "loss": 0.3007, "step": 22725 }, { "epoch": 141.1557632398754, "grad_norm": 1.625, "learning_rate": 1.0129727409682003e-05, "loss": 0.0823, "step": 22726 }, { "epoch": 141.1619937694704, "grad_norm": 2.53125, "learning_rate": 1.0127746947961397e-05, "loss": 0.1936, "step": 22727 }, { "epoch": 141.16822429906543, "grad_norm": 2.21875, "learning_rate": 1.0125766630683076e-05, "loss": 0.135, "step": 22728 }, { "epoch": 141.17445482866043, "grad_norm": 1.09375, "learning_rate": 1.0123786457866285e-05, "loss": 0.0875, "step": 22729 }, { "epoch": 141.18068535825546, "grad_norm": 1.40625, "learning_rate": 1.0121806429530262e-05, "loss": 0.131, "step": 22730 }, { "epoch": 141.18691588785046, "grad_norm": 1.578125, "learning_rate": 1.0119826545694219e-05, "loss": 0.1388, "step": 22731 }, { "epoch": 141.1931464174455, "grad_norm": 2.390625, "learning_rate": 1.0117846806377396e-05, "loss": 0.1837, "step": 22732 }, { "epoch": 141.1993769470405, "grad_norm": 2.859375, "learning_rate": 1.0115867211599029e-05, "loss": 0.183, "step": 22733 }, { "epoch": 141.20560747663552, "grad_norm": 1.25, "learning_rate": 1.0113887761378336e-05, "loss": 0.1083, "step": 22734 }, { "epoch": 141.21183800623052, "grad_norm": 1.5625, "learning_rate": 1.0111908455734528e-05, "loss": 0.0888, "step": 22735 }, { "epoch": 141.21806853582555, "grad_norm": 1.8046875, "learning_rate": 1.0109929294686846e-05, "loss": 0.1579, "step": 22736 }, { "epoch": 141.22429906542055, "grad_norm": 2.625, "learning_rate": 1.0107950278254511e-05, "loss": 0.3032, "step": 22737 }, { "epoch": 141.23052959501558, "grad_norm": 1.25, "learning_rate": 1.0105971406456736e-05, "loss": 0.1066, "step": 22738 }, { "epoch": 141.23676012461058, "grad_norm": 1.8515625, "learning_rate": 1.010399267931275e-05, "loss": 0.1635, "step": 22739 }, { "epoch": 141.2429906542056, "grad_norm": 1.3046875, "learning_rate": 1.0102014096841755e-05, "loss": 0.1008, "step": 22740 }, { "epoch": 141.2492211838006, "grad_norm": 3.890625, "learning_rate": 1.0100035659062988e-05, "loss": 0.292, "step": 22741 }, { "epoch": 141.25545171339564, "grad_norm": 2.390625, "learning_rate": 1.0098057365995642e-05, "loss": 0.1657, "step": 22742 }, { "epoch": 141.26168224299064, "grad_norm": 2.28125, "learning_rate": 1.0096079217658943e-05, "loss": 0.2104, "step": 22743 }, { "epoch": 141.26791277258567, "grad_norm": 2.25, "learning_rate": 1.0094101214072102e-05, "loss": 0.139, "step": 22744 }, { "epoch": 141.27414330218068, "grad_norm": 1.6484375, "learning_rate": 1.0092123355254338e-05, "loss": 0.1844, "step": 22745 }, { "epoch": 141.2803738317757, "grad_norm": 3.40625, "learning_rate": 1.0090145641224853e-05, "loss": 0.1516, "step": 22746 }, { "epoch": 141.2866043613707, "grad_norm": 3.109375, "learning_rate": 1.0088168072002844e-05, "loss": 0.2871, "step": 22747 }, { "epoch": 141.29283489096574, "grad_norm": 2.40625, "learning_rate": 1.008619064760753e-05, "loss": 0.2576, "step": 22748 }, { "epoch": 141.29906542056074, "grad_norm": 2.25, "learning_rate": 1.0084213368058118e-05, "loss": 0.1529, "step": 22749 }, { "epoch": 141.30529595015577, "grad_norm": 2.34375, "learning_rate": 1.0082236233373801e-05, "loss": 0.1727, "step": 22750 }, { "epoch": 141.31152647975077, "grad_norm": 3.421875, "learning_rate": 1.0080259243573789e-05, "loss": 0.3732, "step": 22751 }, { "epoch": 141.3177570093458, "grad_norm": 1.7265625, "learning_rate": 1.0078282398677289e-05, "loss": 0.1521, "step": 22752 }, { "epoch": 141.3239875389408, "grad_norm": 1.3203125, "learning_rate": 1.0076305698703493e-05, "loss": 0.087, "step": 22753 }, { "epoch": 141.33021806853583, "grad_norm": 3.015625, "learning_rate": 1.0074329143671591e-05, "loss": 0.2495, "step": 22754 }, { "epoch": 141.33644859813083, "grad_norm": 2.078125, "learning_rate": 1.0072352733600788e-05, "loss": 0.1734, "step": 22755 }, { "epoch": 141.34267912772586, "grad_norm": 2.21875, "learning_rate": 1.0070376468510286e-05, "loss": 0.1428, "step": 22756 }, { "epoch": 141.34890965732086, "grad_norm": 2.5625, "learning_rate": 1.0068400348419266e-05, "loss": 0.1938, "step": 22757 }, { "epoch": 141.3551401869159, "grad_norm": 2.0, "learning_rate": 1.0066424373346935e-05, "loss": 0.1629, "step": 22758 }, { "epoch": 141.3613707165109, "grad_norm": 1.1953125, "learning_rate": 1.0064448543312466e-05, "loss": 0.1154, "step": 22759 }, { "epoch": 141.36760124610592, "grad_norm": 1.2734375, "learning_rate": 1.0062472858335068e-05, "loss": 0.1216, "step": 22760 }, { "epoch": 141.37383177570092, "grad_norm": 1.8515625, "learning_rate": 1.006049731843391e-05, "loss": 0.1653, "step": 22761 }, { "epoch": 141.38006230529595, "grad_norm": 2.265625, "learning_rate": 1.005852192362819e-05, "loss": 0.2055, "step": 22762 }, { "epoch": 141.38629283489095, "grad_norm": 1.609375, "learning_rate": 1.0056546673937098e-05, "loss": 0.1148, "step": 22763 }, { "epoch": 141.39252336448598, "grad_norm": 2.90625, "learning_rate": 1.0054571569379804e-05, "loss": 0.2876, "step": 22764 }, { "epoch": 141.39875389408098, "grad_norm": 2.28125, "learning_rate": 1.0052596609975507e-05, "loss": 0.2037, "step": 22765 }, { "epoch": 141.404984423676, "grad_norm": 2.078125, "learning_rate": 1.005062179574337e-05, "loss": 0.1197, "step": 22766 }, { "epoch": 141.41121495327104, "grad_norm": 2.046875, "learning_rate": 1.0048647126702593e-05, "loss": 0.1336, "step": 22767 }, { "epoch": 141.41744548286604, "grad_norm": 1.7734375, "learning_rate": 1.0046672602872334e-05, "loss": 0.2117, "step": 22768 }, { "epoch": 141.42367601246107, "grad_norm": 1.171875, "learning_rate": 1.0044698224271782e-05, "loss": 0.0914, "step": 22769 }, { "epoch": 141.42990654205607, "grad_norm": 1.7265625, "learning_rate": 1.0042723990920117e-05, "loss": 0.1411, "step": 22770 }, { "epoch": 141.4361370716511, "grad_norm": 1.1640625, "learning_rate": 1.0040749902836507e-05, "loss": 0.0914, "step": 22771 }, { "epoch": 141.4423676012461, "grad_norm": 1.53125, "learning_rate": 1.0038775960040115e-05, "loss": 0.16, "step": 22772 }, { "epoch": 141.44859813084113, "grad_norm": 1.8828125, "learning_rate": 1.0036802162550121e-05, "loss": 0.1387, "step": 22773 }, { "epoch": 141.45482866043614, "grad_norm": 2.96875, "learning_rate": 1.0034828510385707e-05, "loss": 0.3232, "step": 22774 }, { "epoch": 141.46105919003116, "grad_norm": 2.875, "learning_rate": 1.003285500356602e-05, "loss": 0.1914, "step": 22775 }, { "epoch": 141.46728971962617, "grad_norm": 1.1328125, "learning_rate": 1.0030881642110235e-05, "loss": 0.0984, "step": 22776 }, { "epoch": 141.4735202492212, "grad_norm": 1.9609375, "learning_rate": 1.0028908426037523e-05, "loss": 0.1694, "step": 22777 }, { "epoch": 141.4797507788162, "grad_norm": 1.4765625, "learning_rate": 1.0026935355367058e-05, "loss": 0.1087, "step": 22778 }, { "epoch": 141.48598130841123, "grad_norm": 1.4140625, "learning_rate": 1.0024962430117976e-05, "loss": 0.1109, "step": 22779 }, { "epoch": 141.49221183800623, "grad_norm": 0.9375, "learning_rate": 1.0022989650309452e-05, "loss": 0.1187, "step": 22780 }, { "epoch": 141.49844236760126, "grad_norm": 1.203125, "learning_rate": 1.0021017015960646e-05, "loss": 0.1035, "step": 22781 }, { "epoch": 141.50467289719626, "grad_norm": 2.984375, "learning_rate": 1.0019044527090726e-05, "loss": 0.1377, "step": 22782 }, { "epoch": 141.5109034267913, "grad_norm": 1.09375, "learning_rate": 1.0017072183718832e-05, "loss": 0.1051, "step": 22783 }, { "epoch": 141.5171339563863, "grad_norm": 1.5078125, "learning_rate": 1.0015099985864135e-05, "loss": 0.1487, "step": 22784 }, { "epoch": 141.52336448598132, "grad_norm": 2.1875, "learning_rate": 1.0013127933545773e-05, "loss": 0.221, "step": 22785 }, { "epoch": 141.52959501557632, "grad_norm": 1.578125, "learning_rate": 1.0011156026782919e-05, "loss": 0.1479, "step": 22786 }, { "epoch": 141.53582554517135, "grad_norm": 1.921875, "learning_rate": 1.0009184265594706e-05, "loss": 0.127, "step": 22787 }, { "epoch": 141.54205607476635, "grad_norm": 2.234375, "learning_rate": 1.0007212650000291e-05, "loss": 0.2287, "step": 22788 }, { "epoch": 141.54828660436138, "grad_norm": 2.0625, "learning_rate": 1.0005241180018832e-05, "loss": 0.1464, "step": 22789 }, { "epoch": 141.55451713395638, "grad_norm": 2.109375, "learning_rate": 1.0003269855669468e-05, "loss": 0.1277, "step": 22790 }, { "epoch": 141.5607476635514, "grad_norm": 1.984375, "learning_rate": 1.0001298676971335e-05, "loss": 0.1791, "step": 22791 }, { "epoch": 141.5669781931464, "grad_norm": 2.34375, "learning_rate": 9.99932764394359e-06, "loss": 0.1599, "step": 22792 }, { "epoch": 141.57320872274144, "grad_norm": 2.1875, "learning_rate": 9.99735675660538e-06, "loss": 0.211, "step": 22793 }, { "epoch": 141.57943925233644, "grad_norm": 1.8125, "learning_rate": 9.995386014975833e-06, "loss": 0.2077, "step": 22794 }, { "epoch": 141.58566978193147, "grad_norm": 3.25, "learning_rate": 9.993415419074095e-06, "loss": 0.2627, "step": 22795 }, { "epoch": 141.59190031152647, "grad_norm": 2.046875, "learning_rate": 9.991444968919317e-06, "loss": 0.1629, "step": 22796 }, { "epoch": 141.5981308411215, "grad_norm": 2.3125, "learning_rate": 9.989474664530624e-06, "loss": 0.2438, "step": 22797 }, { "epoch": 141.6043613707165, "grad_norm": 1.21875, "learning_rate": 9.987504505927145e-06, "loss": 0.1064, "step": 22798 }, { "epoch": 141.61059190031153, "grad_norm": 1.4609375, "learning_rate": 9.985534493128024e-06, "loss": 0.0956, "step": 22799 }, { "epoch": 141.61682242990653, "grad_norm": 3.421875, "learning_rate": 9.983564626152403e-06, "loss": 0.2128, "step": 22800 }, { "epoch": 141.62305295950156, "grad_norm": 2.359375, "learning_rate": 9.981594905019393e-06, "loss": 0.2274, "step": 22801 }, { "epoch": 141.62928348909657, "grad_norm": 1.625, "learning_rate": 9.979625329748136e-06, "loss": 0.0823, "step": 22802 }, { "epoch": 141.6355140186916, "grad_norm": 2.3125, "learning_rate": 9.977655900357769e-06, "loss": 0.1526, "step": 22803 }, { "epoch": 141.6417445482866, "grad_norm": 2.265625, "learning_rate": 9.97568661686741e-06, "loss": 0.2173, "step": 22804 }, { "epoch": 141.64797507788163, "grad_norm": 3.40625, "learning_rate": 9.973717479296179e-06, "loss": 0.2088, "step": 22805 }, { "epoch": 141.65420560747663, "grad_norm": 2.578125, "learning_rate": 9.971748487663206e-06, "loss": 0.1847, "step": 22806 }, { "epoch": 141.66043613707166, "grad_norm": 2.71875, "learning_rate": 9.969779641987618e-06, "loss": 0.2515, "step": 22807 }, { "epoch": 141.66666666666666, "grad_norm": 3.546875, "learning_rate": 9.967810942288542e-06, "loss": 0.2517, "step": 22808 }, { "epoch": 141.6728971962617, "grad_norm": 3.03125, "learning_rate": 9.96584238858509e-06, "loss": 0.1128, "step": 22809 }, { "epoch": 141.6791277258567, "grad_norm": 2.59375, "learning_rate": 9.963873980896373e-06, "loss": 0.2061, "step": 22810 }, { "epoch": 141.68535825545172, "grad_norm": 3.1875, "learning_rate": 9.96190571924152e-06, "loss": 0.1204, "step": 22811 }, { "epoch": 141.69158878504672, "grad_norm": 2.09375, "learning_rate": 9.959937603639652e-06, "loss": 0.2067, "step": 22812 }, { "epoch": 141.69781931464175, "grad_norm": 2.5, "learning_rate": 9.957969634109868e-06, "loss": 0.1573, "step": 22813 }, { "epoch": 141.70404984423675, "grad_norm": 1.171875, "learning_rate": 9.95600181067129e-06, "loss": 0.0965, "step": 22814 }, { "epoch": 141.71028037383178, "grad_norm": 2.546875, "learning_rate": 9.954034133343038e-06, "loss": 0.1088, "step": 22815 }, { "epoch": 141.71651090342678, "grad_norm": 1.9375, "learning_rate": 9.952066602144214e-06, "loss": 0.1441, "step": 22816 }, { "epoch": 141.7227414330218, "grad_norm": 1.7890625, "learning_rate": 9.95009921709392e-06, "loss": 0.1414, "step": 22817 }, { "epoch": 141.7289719626168, "grad_norm": 1.828125, "learning_rate": 9.948131978211266e-06, "loss": 0.1269, "step": 22818 }, { "epoch": 141.73520249221184, "grad_norm": 3.03125, "learning_rate": 9.946164885515377e-06, "loss": 0.2103, "step": 22819 }, { "epoch": 141.74143302180684, "grad_norm": 3.5625, "learning_rate": 9.944197939025332e-06, "loss": 0.1849, "step": 22820 }, { "epoch": 141.74766355140187, "grad_norm": 3.109375, "learning_rate": 9.942231138760249e-06, "loss": 0.2518, "step": 22821 }, { "epoch": 141.75389408099687, "grad_norm": 1.921875, "learning_rate": 9.940264484739234e-06, "loss": 0.1305, "step": 22822 }, { "epoch": 141.7601246105919, "grad_norm": 2.140625, "learning_rate": 9.93829797698138e-06, "loss": 0.0929, "step": 22823 }, { "epoch": 141.7663551401869, "grad_norm": 1.7734375, "learning_rate": 9.936331615505781e-06, "loss": 0.1487, "step": 22824 }, { "epoch": 141.77258566978193, "grad_norm": 2.640625, "learning_rate": 9.934365400331539e-06, "loss": 0.2236, "step": 22825 }, { "epoch": 141.77881619937693, "grad_norm": 2.015625, "learning_rate": 9.932399331477759e-06, "loss": 0.1631, "step": 22826 }, { "epoch": 141.78504672897196, "grad_norm": 2.09375, "learning_rate": 9.930433408963522e-06, "loss": 0.1849, "step": 22827 }, { "epoch": 141.79127725856696, "grad_norm": 2.5625, "learning_rate": 9.928467632807937e-06, "loss": 0.1759, "step": 22828 }, { "epoch": 141.797507788162, "grad_norm": 2.40625, "learning_rate": 9.926502003030075e-06, "loss": 0.2021, "step": 22829 }, { "epoch": 141.80373831775702, "grad_norm": 2.015625, "learning_rate": 9.92453651964905e-06, "loss": 0.0751, "step": 22830 }, { "epoch": 141.80996884735202, "grad_norm": 0.84375, "learning_rate": 9.922571182683932e-06, "loss": 0.0934, "step": 22831 }, { "epoch": 141.81619937694705, "grad_norm": 0.85546875, "learning_rate": 9.920605992153814e-06, "loss": 0.0935, "step": 22832 }, { "epoch": 141.82242990654206, "grad_norm": 1.4453125, "learning_rate": 9.918640948077796e-06, "loss": 0.1548, "step": 22833 }, { "epoch": 141.82866043613708, "grad_norm": 1.5859375, "learning_rate": 9.916676050474941e-06, "loss": 0.1148, "step": 22834 }, { "epoch": 141.8348909657321, "grad_norm": 1.125, "learning_rate": 9.914711299364354e-06, "loss": 0.1112, "step": 22835 }, { "epoch": 141.84112149532712, "grad_norm": 1.265625, "learning_rate": 9.912746694765095e-06, "loss": 0.1179, "step": 22836 }, { "epoch": 141.84735202492212, "grad_norm": 2.09375, "learning_rate": 9.910782236696267e-06, "loss": 0.1519, "step": 22837 }, { "epoch": 141.85358255451715, "grad_norm": 1.734375, "learning_rate": 9.90881792517693e-06, "loss": 0.1068, "step": 22838 }, { "epoch": 141.85981308411215, "grad_norm": 2.03125, "learning_rate": 9.90685376022617e-06, "loss": 0.1739, "step": 22839 }, { "epoch": 141.86604361370718, "grad_norm": 1.140625, "learning_rate": 9.90488974186306e-06, "loss": 0.102, "step": 22840 }, { "epoch": 141.87227414330218, "grad_norm": 2.203125, "learning_rate": 9.902925870106699e-06, "loss": 0.2582, "step": 22841 }, { "epoch": 141.8785046728972, "grad_norm": 2.75, "learning_rate": 9.900962144976123e-06, "loss": 0.2501, "step": 22842 }, { "epoch": 141.8847352024922, "grad_norm": 2.046875, "learning_rate": 9.898998566490417e-06, "loss": 0.1311, "step": 22843 }, { "epoch": 141.89096573208724, "grad_norm": 2.171875, "learning_rate": 9.897035134668658e-06, "loss": 0.1919, "step": 22844 }, { "epoch": 141.89719626168224, "grad_norm": 1.7578125, "learning_rate": 9.895071849529921e-06, "loss": 0.1826, "step": 22845 }, { "epoch": 141.90342679127727, "grad_norm": 1.546875, "learning_rate": 9.893108711093257e-06, "loss": 0.0832, "step": 22846 }, { "epoch": 141.90965732087227, "grad_norm": 1.828125, "learning_rate": 9.891145719377742e-06, "loss": 0.158, "step": 22847 }, { "epoch": 141.9158878504673, "grad_norm": 2.921875, "learning_rate": 9.889182874402448e-06, "loss": 0.2458, "step": 22848 }, { "epoch": 141.9221183800623, "grad_norm": 2.421875, "learning_rate": 9.887220176186431e-06, "loss": 0.1949, "step": 22849 }, { "epoch": 141.92834890965733, "grad_norm": 1.6875, "learning_rate": 9.885257624748743e-06, "loss": 0.1363, "step": 22850 }, { "epoch": 141.93457943925233, "grad_norm": 1.1328125, "learning_rate": 9.883295220108454e-06, "loss": 0.0877, "step": 22851 }, { "epoch": 141.94080996884736, "grad_norm": 1.5625, "learning_rate": 9.881332962284633e-06, "loss": 0.126, "step": 22852 }, { "epoch": 141.94704049844236, "grad_norm": 1.84375, "learning_rate": 9.879370851296322e-06, "loss": 0.1526, "step": 22853 }, { "epoch": 141.9532710280374, "grad_norm": 1.8828125, "learning_rate": 9.877408887162587e-06, "loss": 0.0998, "step": 22854 }, { "epoch": 141.9595015576324, "grad_norm": 2.765625, "learning_rate": 9.875447069902474e-06, "loss": 0.1978, "step": 22855 }, { "epoch": 141.96573208722742, "grad_norm": 2.625, "learning_rate": 9.873485399535053e-06, "loss": 0.2125, "step": 22856 }, { "epoch": 141.97196261682242, "grad_norm": 2.109375, "learning_rate": 9.871523876079355e-06, "loss": 0.1086, "step": 22857 }, { "epoch": 141.97819314641745, "grad_norm": 3.375, "learning_rate": 9.86956249955444e-06, "loss": 0.2039, "step": 22858 }, { "epoch": 141.98442367601245, "grad_norm": 1.5625, "learning_rate": 9.867601269979371e-06, "loss": 0.1095, "step": 22859 }, { "epoch": 141.99065420560748, "grad_norm": 2.703125, "learning_rate": 9.865640187373182e-06, "loss": 0.2465, "step": 22860 }, { "epoch": 141.99688473520249, "grad_norm": 1.8515625, "learning_rate": 9.86367925175491e-06, "loss": 0.128, "step": 22861 }, { "epoch": 142.0, "grad_norm": 0.6484375, "learning_rate": 9.861718463143612e-06, "loss": 0.0397, "step": 22862 }, { "epoch": 142.00623052959503, "grad_norm": 1.265625, "learning_rate": 9.859757821558337e-06, "loss": 0.1218, "step": 22863 }, { "epoch": 142.01246105919003, "grad_norm": 2.015625, "learning_rate": 9.857797327018114e-06, "loss": 0.152, "step": 22864 }, { "epoch": 142.01869158878506, "grad_norm": 0.703125, "learning_rate": 9.85583697954199e-06, "loss": 0.0765, "step": 22865 }, { "epoch": 142.02492211838006, "grad_norm": 1.890625, "learning_rate": 9.853876779149016e-06, "loss": 0.1358, "step": 22866 }, { "epoch": 142.0311526479751, "grad_norm": 1.75, "learning_rate": 9.851916725858215e-06, "loss": 0.1474, "step": 22867 }, { "epoch": 142.0373831775701, "grad_norm": 1.34375, "learning_rate": 9.849956819688618e-06, "loss": 0.1154, "step": 22868 }, { "epoch": 142.04361370716512, "grad_norm": 1.984375, "learning_rate": 9.84799706065927e-06, "loss": 0.1098, "step": 22869 }, { "epoch": 142.04984423676012, "grad_norm": 2.984375, "learning_rate": 9.84603744878921e-06, "loss": 0.2288, "step": 22870 }, { "epoch": 142.05607476635515, "grad_norm": 2.421875, "learning_rate": 9.844077984097457e-06, "loss": 0.1343, "step": 22871 }, { "epoch": 142.06230529595015, "grad_norm": 2.046875, "learning_rate": 9.842118666603048e-06, "loss": 0.167, "step": 22872 }, { "epoch": 142.06853582554518, "grad_norm": 2.46875, "learning_rate": 9.84015949632502e-06, "loss": 0.1717, "step": 22873 }, { "epoch": 142.07476635514018, "grad_norm": 1.484375, "learning_rate": 9.838200473282397e-06, "loss": 0.1361, "step": 22874 }, { "epoch": 142.0809968847352, "grad_norm": 1.3359375, "learning_rate": 9.83624159749419e-06, "loss": 0.124, "step": 22875 }, { "epoch": 142.08722741433021, "grad_norm": 2.890625, "learning_rate": 9.834282868979436e-06, "loss": 0.2274, "step": 22876 }, { "epoch": 142.09345794392524, "grad_norm": 2.578125, "learning_rate": 9.832324287757158e-06, "loss": 0.2161, "step": 22877 }, { "epoch": 142.09968847352025, "grad_norm": 2.234375, "learning_rate": 9.830365853846388e-06, "loss": 0.1402, "step": 22878 }, { "epoch": 142.10591900311528, "grad_norm": 2.734375, "learning_rate": 9.828407567266137e-06, "loss": 0.2995, "step": 22879 }, { "epoch": 142.11214953271028, "grad_norm": 2.578125, "learning_rate": 9.826449428035413e-06, "loss": 0.2486, "step": 22880 }, { "epoch": 142.1183800623053, "grad_norm": 2.65625, "learning_rate": 9.824491436173249e-06, "loss": 0.1909, "step": 22881 }, { "epoch": 142.1246105919003, "grad_norm": 2.09375, "learning_rate": 9.822533591698665e-06, "loss": 0.1482, "step": 22882 }, { "epoch": 142.13084112149534, "grad_norm": 0.9296875, "learning_rate": 9.820575894630663e-06, "loss": 0.0854, "step": 22883 }, { "epoch": 142.13707165109034, "grad_norm": 1.6484375, "learning_rate": 9.81861834498826e-06, "loss": 0.1299, "step": 22884 }, { "epoch": 142.14330218068537, "grad_norm": 2.703125, "learning_rate": 9.816660942790475e-06, "loss": 0.1783, "step": 22885 }, { "epoch": 142.14953271028037, "grad_norm": 2.8125, "learning_rate": 9.814703688056321e-06, "loss": 0.229, "step": 22886 }, { "epoch": 142.1557632398754, "grad_norm": 1.890625, "learning_rate": 9.812746580804786e-06, "loss": 0.1707, "step": 22887 }, { "epoch": 142.1619937694704, "grad_norm": 2.625, "learning_rate": 9.810789621054895e-06, "loss": 0.2351, "step": 22888 }, { "epoch": 142.16822429906543, "grad_norm": 1.5390625, "learning_rate": 9.808832808825657e-06, "loss": 0.1197, "step": 22889 }, { "epoch": 142.17445482866043, "grad_norm": 1.609375, "learning_rate": 9.806876144136068e-06, "loss": 0.1112, "step": 22890 }, { "epoch": 142.18068535825546, "grad_norm": 1.5703125, "learning_rate": 9.80491962700513e-06, "loss": 0.1404, "step": 22891 }, { "epoch": 142.18691588785046, "grad_norm": 1.4609375, "learning_rate": 9.80296325745186e-06, "loss": 0.1246, "step": 22892 }, { "epoch": 142.1931464174455, "grad_norm": 2.59375, "learning_rate": 9.80100703549525e-06, "loss": 0.1894, "step": 22893 }, { "epoch": 142.1993769470405, "grad_norm": 2.90625, "learning_rate": 9.799050961154288e-06, "loss": 0.2233, "step": 22894 }, { "epoch": 142.20560747663552, "grad_norm": 1.9765625, "learning_rate": 9.797095034447984e-06, "loss": 0.1462, "step": 22895 }, { "epoch": 142.21183800623052, "grad_norm": 1.796875, "learning_rate": 9.795139255395338e-06, "loss": 0.1644, "step": 22896 }, { "epoch": 142.21806853582555, "grad_norm": 2.125, "learning_rate": 9.793183624015332e-06, "loss": 0.1903, "step": 22897 }, { "epoch": 142.22429906542055, "grad_norm": 2.28125, "learning_rate": 9.791228140326975e-06, "loss": 0.1487, "step": 22898 }, { "epoch": 142.23052959501558, "grad_norm": 1.1328125, "learning_rate": 9.789272804349239e-06, "loss": 0.0901, "step": 22899 }, { "epoch": 142.23676012461058, "grad_norm": 1.2421875, "learning_rate": 9.787317616101139e-06, "loss": 0.1316, "step": 22900 }, { "epoch": 142.2429906542056, "grad_norm": 1.8359375, "learning_rate": 9.78536257560164e-06, "loss": 0.1863, "step": 22901 }, { "epoch": 142.2492211838006, "grad_norm": 2.09375, "learning_rate": 9.783407682869742e-06, "loss": 0.1754, "step": 22902 }, { "epoch": 142.25545171339564, "grad_norm": 3.015625, "learning_rate": 9.781452937924429e-06, "loss": 0.2191, "step": 22903 }, { "epoch": 142.26168224299064, "grad_norm": 1.484375, "learning_rate": 9.779498340784695e-06, "loss": 0.1387, "step": 22904 }, { "epoch": 142.26791277258567, "grad_norm": 2.078125, "learning_rate": 9.777543891469517e-06, "loss": 0.1783, "step": 22905 }, { "epoch": 142.27414330218068, "grad_norm": 2.421875, "learning_rate": 9.775589589997869e-06, "loss": 0.2257, "step": 22906 }, { "epoch": 142.2803738317757, "grad_norm": 1.6796875, "learning_rate": 9.773635436388737e-06, "loss": 0.1483, "step": 22907 }, { "epoch": 142.2866043613707, "grad_norm": 3.28125, "learning_rate": 9.771681430661108e-06, "loss": 0.3652, "step": 22908 }, { "epoch": 142.29283489096574, "grad_norm": 2.40625, "learning_rate": 9.769727572833945e-06, "loss": 0.1391, "step": 22909 }, { "epoch": 142.29906542056074, "grad_norm": 1.796875, "learning_rate": 9.767773862926232e-06, "loss": 0.1222, "step": 22910 }, { "epoch": 142.30529595015577, "grad_norm": 1.90625, "learning_rate": 9.765820300956966e-06, "loss": 0.2348, "step": 22911 }, { "epoch": 142.31152647975077, "grad_norm": 1.21875, "learning_rate": 9.763866886945075e-06, "loss": 0.1247, "step": 22912 }, { "epoch": 142.3177570093458, "grad_norm": 2.5, "learning_rate": 9.761913620909558e-06, "loss": 0.1732, "step": 22913 }, { "epoch": 142.3239875389408, "grad_norm": 2.25, "learning_rate": 9.75996050286938e-06, "loss": 0.1718, "step": 22914 }, { "epoch": 142.33021806853583, "grad_norm": 2.875, "learning_rate": 9.758007532843522e-06, "loss": 0.1947, "step": 22915 }, { "epoch": 142.33644859813083, "grad_norm": 1.5546875, "learning_rate": 9.756054710850934e-06, "loss": 0.1332, "step": 22916 }, { "epoch": 142.34267912772586, "grad_norm": 1.4921875, "learning_rate": 9.754102036910586e-06, "loss": 0.1354, "step": 22917 }, { "epoch": 142.34890965732086, "grad_norm": 2.390625, "learning_rate": 9.75214951104146e-06, "loss": 0.2267, "step": 22918 }, { "epoch": 142.3551401869159, "grad_norm": 2.40625, "learning_rate": 9.750197133262503e-06, "loss": 0.1698, "step": 22919 }, { "epoch": 142.3613707165109, "grad_norm": 1.2421875, "learning_rate": 9.748244903592674e-06, "loss": 0.1005, "step": 22920 }, { "epoch": 142.36760124610592, "grad_norm": 1.3828125, "learning_rate": 9.746292822050937e-06, "loss": 0.1149, "step": 22921 }, { "epoch": 142.37383177570092, "grad_norm": 2.453125, "learning_rate": 9.744340888656262e-06, "loss": 0.1999, "step": 22922 }, { "epoch": 142.38006230529595, "grad_norm": 1.6328125, "learning_rate": 9.742389103427588e-06, "loss": 0.1466, "step": 22923 }, { "epoch": 142.38629283489095, "grad_norm": 1.0390625, "learning_rate": 9.740437466383892e-06, "loss": 0.1163, "step": 22924 }, { "epoch": 142.39252336448598, "grad_norm": 2.5625, "learning_rate": 9.738485977544106e-06, "loss": 0.2354, "step": 22925 }, { "epoch": 142.39875389408098, "grad_norm": 1.953125, "learning_rate": 9.736534636927206e-06, "loss": 0.1373, "step": 22926 }, { "epoch": 142.404984423676, "grad_norm": 1.921875, "learning_rate": 9.734583444552124e-06, "loss": 0.1156, "step": 22927 }, { "epoch": 142.41121495327104, "grad_norm": 1.1953125, "learning_rate": 9.732632400437819e-06, "loss": 0.1071, "step": 22928 }, { "epoch": 142.41744548286604, "grad_norm": 3.5625, "learning_rate": 9.730681504603247e-06, "loss": 0.2761, "step": 22929 }, { "epoch": 142.42367601246107, "grad_norm": 1.2890625, "learning_rate": 9.72873075706735e-06, "loss": 0.1373, "step": 22930 }, { "epoch": 142.42990654205607, "grad_norm": 2.765625, "learning_rate": 9.726780157849064e-06, "loss": 0.212, "step": 22931 }, { "epoch": 142.4361370716511, "grad_norm": 1.734375, "learning_rate": 9.724829706967342e-06, "loss": 0.1514, "step": 22932 }, { "epoch": 142.4423676012461, "grad_norm": 1.3828125, "learning_rate": 9.722879404441135e-06, "loss": 0.1288, "step": 22933 }, { "epoch": 142.44859813084113, "grad_norm": 0.70703125, "learning_rate": 9.72092925028937e-06, "loss": 0.0869, "step": 22934 }, { "epoch": 142.45482866043614, "grad_norm": 2.328125, "learning_rate": 9.718979244530992e-06, "loss": 0.1701, "step": 22935 }, { "epoch": 142.46105919003116, "grad_norm": 1.46875, "learning_rate": 9.717029387184945e-06, "loss": 0.1135, "step": 22936 }, { "epoch": 142.46728971962617, "grad_norm": 2.140625, "learning_rate": 9.715079678270181e-06, "loss": 0.0875, "step": 22937 }, { "epoch": 142.4735202492212, "grad_norm": 2.296875, "learning_rate": 9.713130117805599e-06, "loss": 0.197, "step": 22938 }, { "epoch": 142.4797507788162, "grad_norm": 2.53125, "learning_rate": 9.711180705810156e-06, "loss": 0.218, "step": 22939 }, { "epoch": 142.48598130841123, "grad_norm": 2.609375, "learning_rate": 9.709231442302777e-06, "loss": 0.1673, "step": 22940 }, { "epoch": 142.49221183800623, "grad_norm": 1.8125, "learning_rate": 9.707282327302411e-06, "loss": 0.1138, "step": 22941 }, { "epoch": 142.49844236760126, "grad_norm": 3.078125, "learning_rate": 9.705333360827967e-06, "loss": 0.1868, "step": 22942 }, { "epoch": 142.50467289719626, "grad_norm": 2.390625, "learning_rate": 9.703384542898392e-06, "loss": 0.183, "step": 22943 }, { "epoch": 142.5109034267913, "grad_norm": 2.546875, "learning_rate": 9.701435873532594e-06, "loss": 0.2384, "step": 22944 }, { "epoch": 142.5171339563863, "grad_norm": 0.95703125, "learning_rate": 9.69948735274952e-06, "loss": 0.0994, "step": 22945 }, { "epoch": 142.52336448598132, "grad_norm": 1.6171875, "learning_rate": 9.697538980568071e-06, "loss": 0.1124, "step": 22946 }, { "epoch": 142.52959501557632, "grad_norm": 1.8359375, "learning_rate": 9.695590757007186e-06, "loss": 0.1588, "step": 22947 }, { "epoch": 142.53582554517135, "grad_norm": 3.25, "learning_rate": 9.693642682085791e-06, "loss": 0.3216, "step": 22948 }, { "epoch": 142.54205607476635, "grad_norm": 1.828125, "learning_rate": 9.691694755822794e-06, "loss": 0.1218, "step": 22949 }, { "epoch": 142.54828660436138, "grad_norm": 3.15625, "learning_rate": 9.689746978237113e-06, "loss": 0.1749, "step": 22950 }, { "epoch": 142.55451713395638, "grad_norm": 2.0625, "learning_rate": 9.687799349347668e-06, "loss": 0.1537, "step": 22951 }, { "epoch": 142.5607476635514, "grad_norm": 2.234375, "learning_rate": 9.685851869173388e-06, "loss": 0.1412, "step": 22952 }, { "epoch": 142.5669781931464, "grad_norm": 1.421875, "learning_rate": 9.683904537733162e-06, "loss": 0.1237, "step": 22953 }, { "epoch": 142.57320872274144, "grad_norm": 1.578125, "learning_rate": 9.68195735504592e-06, "loss": 0.1651, "step": 22954 }, { "epoch": 142.57943925233644, "grad_norm": 1.859375, "learning_rate": 9.680010321130578e-06, "loss": 0.1395, "step": 22955 }, { "epoch": 142.58566978193147, "grad_norm": 3.1875, "learning_rate": 9.67806343600604e-06, "loss": 0.1619, "step": 22956 }, { "epoch": 142.59190031152647, "grad_norm": 2.25, "learning_rate": 9.676116699691204e-06, "loss": 0.1082, "step": 22957 }, { "epoch": 142.5981308411215, "grad_norm": 1.3203125, "learning_rate": 9.674170112204983e-06, "loss": 0.1266, "step": 22958 }, { "epoch": 142.6043613707165, "grad_norm": 2.46875, "learning_rate": 9.672223673566297e-06, "loss": 0.2254, "step": 22959 }, { "epoch": 142.61059190031153, "grad_norm": 2.453125, "learning_rate": 9.670277383794027e-06, "loss": 0.1876, "step": 22960 }, { "epoch": 142.61682242990653, "grad_norm": 2.03125, "learning_rate": 9.668331242907089e-06, "loss": 0.2022, "step": 22961 }, { "epoch": 142.62305295950156, "grad_norm": 3.125, "learning_rate": 9.666385250924392e-06, "loss": 0.3126, "step": 22962 }, { "epoch": 142.62928348909657, "grad_norm": 2.09375, "learning_rate": 9.664439407864826e-06, "loss": 0.1654, "step": 22963 }, { "epoch": 142.6355140186916, "grad_norm": 2.03125, "learning_rate": 9.66249371374728e-06, "loss": 0.1874, "step": 22964 }, { "epoch": 142.6417445482866, "grad_norm": 3.53125, "learning_rate": 9.660548168590663e-06, "loss": 0.2259, "step": 22965 }, { "epoch": 142.64797507788163, "grad_norm": 1.7578125, "learning_rate": 9.658602772413878e-06, "loss": 0.2224, "step": 22966 }, { "epoch": 142.65420560747663, "grad_norm": 1.609375, "learning_rate": 9.656657525235799e-06, "loss": 0.142, "step": 22967 }, { "epoch": 142.66043613707166, "grad_norm": 1.71875, "learning_rate": 9.654712427075339e-06, "loss": 0.1522, "step": 22968 }, { "epoch": 142.66666666666666, "grad_norm": 2.53125, "learning_rate": 9.65276747795137e-06, "loss": 0.2126, "step": 22969 }, { "epoch": 142.6728971962617, "grad_norm": 2.03125, "learning_rate": 9.650822677882798e-06, "loss": 0.1612, "step": 22970 }, { "epoch": 142.6791277258567, "grad_norm": 1.7265625, "learning_rate": 9.648878026888497e-06, "loss": 0.1152, "step": 22971 }, { "epoch": 142.68535825545172, "grad_norm": 2.265625, "learning_rate": 9.646933524987361e-06, "loss": 0.1655, "step": 22972 }, { "epoch": 142.69158878504672, "grad_norm": 2.6875, "learning_rate": 9.644989172198276e-06, "loss": 0.2268, "step": 22973 }, { "epoch": 142.69781931464175, "grad_norm": 2.296875, "learning_rate": 9.643044968540135e-06, "loss": 0.2208, "step": 22974 }, { "epoch": 142.70404984423675, "grad_norm": 1.8046875, "learning_rate": 9.641100914031812e-06, "loss": 0.1389, "step": 22975 }, { "epoch": 142.71028037383178, "grad_norm": 1.703125, "learning_rate": 9.639157008692176e-06, "loss": 0.2202, "step": 22976 }, { "epoch": 142.71651090342678, "grad_norm": 0.953125, "learning_rate": 9.637213252540119e-06, "loss": 0.0947, "step": 22977 }, { "epoch": 142.7227414330218, "grad_norm": 2.640625, "learning_rate": 9.635269645594527e-06, "loss": 0.2333, "step": 22978 }, { "epoch": 142.7289719626168, "grad_norm": 1.6171875, "learning_rate": 9.633326187874258e-06, "loss": 0.1138, "step": 22979 }, { "epoch": 142.73520249221184, "grad_norm": 2.15625, "learning_rate": 9.631382879398196e-06, "loss": 0.1557, "step": 22980 }, { "epoch": 142.74143302180684, "grad_norm": 1.7734375, "learning_rate": 9.629439720185226e-06, "loss": 0.109, "step": 22981 }, { "epoch": 142.74766355140187, "grad_norm": 2.375, "learning_rate": 9.627496710254207e-06, "loss": 0.2821, "step": 22982 }, { "epoch": 142.75389408099687, "grad_norm": 2.828125, "learning_rate": 9.625553849624009e-06, "loss": 0.2294, "step": 22983 }, { "epoch": 142.7601246105919, "grad_norm": 2.546875, "learning_rate": 9.623611138313501e-06, "loss": 0.1788, "step": 22984 }, { "epoch": 142.7663551401869, "grad_norm": 3.109375, "learning_rate": 9.621668576341567e-06, "loss": 0.2747, "step": 22985 }, { "epoch": 142.77258566978193, "grad_norm": 1.5625, "learning_rate": 9.61972616372705e-06, "loss": 0.1023, "step": 22986 }, { "epoch": 142.77881619937693, "grad_norm": 3.5625, "learning_rate": 9.61778390048883e-06, "loss": 0.2977, "step": 22987 }, { "epoch": 142.78504672897196, "grad_norm": 2.78125, "learning_rate": 9.615841786645774e-06, "loss": 0.1855, "step": 22988 }, { "epoch": 142.79127725856696, "grad_norm": 2.53125, "learning_rate": 9.61389982221674e-06, "loss": 0.2295, "step": 22989 }, { "epoch": 142.797507788162, "grad_norm": 2.296875, "learning_rate": 9.611958007220575e-06, "loss": 0.3598, "step": 22990 }, { "epoch": 142.80373831775702, "grad_norm": 2.46875, "learning_rate": 9.610016341676154e-06, "loss": 0.192, "step": 22991 }, { "epoch": 142.80996884735202, "grad_norm": 2.828125, "learning_rate": 9.608074825602337e-06, "loss": 0.1454, "step": 22992 }, { "epoch": 142.81619937694705, "grad_norm": 1.09375, "learning_rate": 9.606133459017964e-06, "loss": 0.1077, "step": 22993 }, { "epoch": 142.82242990654206, "grad_norm": 1.3203125, "learning_rate": 9.604192241941909e-06, "loss": 0.0945, "step": 22994 }, { "epoch": 142.82866043613708, "grad_norm": 2.515625, "learning_rate": 9.602251174393009e-06, "loss": 0.2221, "step": 22995 }, { "epoch": 142.8348909657321, "grad_norm": 2.46875, "learning_rate": 9.600310256390133e-06, "loss": 0.302, "step": 22996 }, { "epoch": 142.84112149532712, "grad_norm": 2.34375, "learning_rate": 9.598369487952111e-06, "loss": 0.1823, "step": 22997 }, { "epoch": 142.84735202492212, "grad_norm": 2.671875, "learning_rate": 9.596428869097807e-06, "loss": 0.2046, "step": 22998 }, { "epoch": 142.85358255451715, "grad_norm": 2.171875, "learning_rate": 9.59448839984606e-06, "loss": 0.1998, "step": 22999 }, { "epoch": 142.85981308411215, "grad_norm": 1.4765625, "learning_rate": 9.592548080215744e-06, "loss": 0.1506, "step": 23000 }, { "epoch": 142.86604361370718, "grad_norm": 1.203125, "learning_rate": 9.59060791022566e-06, "loss": 0.0946, "step": 23001 }, { "epoch": 142.87227414330218, "grad_norm": 2.71875, "learning_rate": 9.588667889894673e-06, "loss": 0.1944, "step": 23002 }, { "epoch": 142.8785046728972, "grad_norm": 1.1328125, "learning_rate": 9.586728019241623e-06, "loss": 0.1048, "step": 23003 }, { "epoch": 142.8847352024922, "grad_norm": 3.703125, "learning_rate": 9.58478829828536e-06, "loss": 0.2474, "step": 23004 }, { "epoch": 142.89096573208724, "grad_norm": 1.9296875, "learning_rate": 9.582848727044708e-06, "loss": 0.167, "step": 23005 }, { "epoch": 142.89719626168224, "grad_norm": 2.265625, "learning_rate": 9.58090930553851e-06, "loss": 0.1975, "step": 23006 }, { "epoch": 142.90342679127727, "grad_norm": 2.484375, "learning_rate": 9.578970033785619e-06, "loss": 0.2511, "step": 23007 }, { "epoch": 142.90965732087227, "grad_norm": 2.984375, "learning_rate": 9.57703091180484e-06, "loss": 0.2946, "step": 23008 }, { "epoch": 142.9158878504673, "grad_norm": 1.1796875, "learning_rate": 9.575091939615017e-06, "loss": 0.0915, "step": 23009 }, { "epoch": 142.9221183800623, "grad_norm": 0.74609375, "learning_rate": 9.573153117234984e-06, "loss": 0.0851, "step": 23010 }, { "epoch": 142.92834890965733, "grad_norm": 1.75, "learning_rate": 9.571214444683582e-06, "loss": 0.1171, "step": 23011 }, { "epoch": 142.93457943925233, "grad_norm": 2.515625, "learning_rate": 9.56927592197962e-06, "loss": 0.2659, "step": 23012 }, { "epoch": 142.94080996884736, "grad_norm": 2.53125, "learning_rate": 9.567337549141945e-06, "loss": 0.1913, "step": 23013 }, { "epoch": 142.94704049844236, "grad_norm": 2.265625, "learning_rate": 9.565399326189361e-06, "loss": 0.1898, "step": 23014 }, { "epoch": 142.9532710280374, "grad_norm": 1.171875, "learning_rate": 9.563461253140715e-06, "loss": 0.0842, "step": 23015 }, { "epoch": 142.9595015576324, "grad_norm": 2.703125, "learning_rate": 9.56152333001481e-06, "loss": 0.1983, "step": 23016 }, { "epoch": 142.96573208722742, "grad_norm": 3.09375, "learning_rate": 9.559585556830477e-06, "loss": 0.2762, "step": 23017 }, { "epoch": 142.97196261682242, "grad_norm": 1.4765625, "learning_rate": 9.557647933606542e-06, "loss": 0.1173, "step": 23018 }, { "epoch": 142.97819314641745, "grad_norm": 2.734375, "learning_rate": 9.555710460361819e-06, "loss": 0.1512, "step": 23019 }, { "epoch": 142.98442367601245, "grad_norm": 1.9453125, "learning_rate": 9.553773137115116e-06, "loss": 0.1456, "step": 23020 }, { "epoch": 142.99065420560748, "grad_norm": 2.3125, "learning_rate": 9.551835963885253e-06, "loss": 0.1134, "step": 23021 }, { "epoch": 142.99688473520249, "grad_norm": 2.46875, "learning_rate": 9.549898940691057e-06, "loss": 0.2466, "step": 23022 }, { "epoch": 143.0, "grad_norm": 0.6875, "learning_rate": 9.547962067551322e-06, "loss": 0.0534, "step": 23023 }, { "epoch": 143.00623052959503, "grad_norm": 3.59375, "learning_rate": 9.546025344484869e-06, "loss": 0.1218, "step": 23024 }, { "epoch": 143.01246105919003, "grad_norm": 2.671875, "learning_rate": 9.544088771510515e-06, "loss": 0.1023, "step": 23025 }, { "epoch": 143.01869158878506, "grad_norm": 1.5234375, "learning_rate": 9.54215234864706e-06, "loss": 0.1503, "step": 23026 }, { "epoch": 143.02492211838006, "grad_norm": 1.0546875, "learning_rate": 9.540216075913303e-06, "loss": 0.1055, "step": 23027 }, { "epoch": 143.0311526479751, "grad_norm": 2.09375, "learning_rate": 9.538279953328058e-06, "loss": 0.1674, "step": 23028 }, { "epoch": 143.0373831775701, "grad_norm": 3.0, "learning_rate": 9.536343980910136e-06, "loss": 0.2522, "step": 23029 }, { "epoch": 143.04361370716512, "grad_norm": 1.5390625, "learning_rate": 9.534408158678326e-06, "loss": 0.1266, "step": 23030 }, { "epoch": 143.04984423676012, "grad_norm": 1.5625, "learning_rate": 9.532472486651434e-06, "loss": 0.1303, "step": 23031 }, { "epoch": 143.05607476635515, "grad_norm": 1.2734375, "learning_rate": 9.530536964848272e-06, "loss": 0.1154, "step": 23032 }, { "epoch": 143.06230529595015, "grad_norm": 2.828125, "learning_rate": 9.528601593287626e-06, "loss": 0.1908, "step": 23033 }, { "epoch": 143.06853582554518, "grad_norm": 0.89453125, "learning_rate": 9.526666371988283e-06, "loss": 0.1115, "step": 23034 }, { "epoch": 143.07476635514018, "grad_norm": 1.4296875, "learning_rate": 9.52473130096905e-06, "loss": 0.147, "step": 23035 }, { "epoch": 143.0809968847352, "grad_norm": 2.0, "learning_rate": 9.522796380248722e-06, "loss": 0.1185, "step": 23036 }, { "epoch": 143.08722741433021, "grad_norm": 2.234375, "learning_rate": 9.5208616098461e-06, "loss": 0.2022, "step": 23037 }, { "epoch": 143.09345794392524, "grad_norm": 1.2578125, "learning_rate": 9.518926989779961e-06, "loss": 0.1202, "step": 23038 }, { "epoch": 143.09968847352025, "grad_norm": 0.90625, "learning_rate": 9.516992520069087e-06, "loss": 0.0917, "step": 23039 }, { "epoch": 143.10591900311528, "grad_norm": 1.2890625, "learning_rate": 9.51505820073228e-06, "loss": 0.1169, "step": 23040 }, { "epoch": 143.11214953271028, "grad_norm": 2.265625, "learning_rate": 9.513124031788331e-06, "loss": 0.0979, "step": 23041 }, { "epoch": 143.1183800623053, "grad_norm": 1.21875, "learning_rate": 9.51119001325601e-06, "loss": 0.1066, "step": 23042 }, { "epoch": 143.1246105919003, "grad_norm": 2.5, "learning_rate": 9.509256145154105e-06, "loss": 0.2258, "step": 23043 }, { "epoch": 143.13084112149534, "grad_norm": 2.234375, "learning_rate": 9.507322427501411e-06, "loss": 0.1891, "step": 23044 }, { "epoch": 143.13707165109034, "grad_norm": 2.421875, "learning_rate": 9.5053888603167e-06, "loss": 0.2329, "step": 23045 }, { "epoch": 143.14330218068537, "grad_norm": 2.5, "learning_rate": 9.503455443618739e-06, "loss": 0.1685, "step": 23046 }, { "epoch": 143.14953271028037, "grad_norm": 1.5234375, "learning_rate": 9.501522177426317e-06, "loss": 0.1115, "step": 23047 }, { "epoch": 143.1557632398754, "grad_norm": 1.6796875, "learning_rate": 9.499589061758219e-06, "loss": 0.1241, "step": 23048 }, { "epoch": 143.1619937694704, "grad_norm": 2.46875, "learning_rate": 9.4976560966332e-06, "loss": 0.2172, "step": 23049 }, { "epoch": 143.16822429906543, "grad_norm": 1.0078125, "learning_rate": 9.495723282070043e-06, "loss": 0.0909, "step": 23050 }, { "epoch": 143.17445482866043, "grad_norm": 0.80859375, "learning_rate": 9.493790618087534e-06, "loss": 0.0988, "step": 23051 }, { "epoch": 143.18068535825546, "grad_norm": 1.2734375, "learning_rate": 9.491858104704426e-06, "loss": 0.1029, "step": 23052 }, { "epoch": 143.18691588785046, "grad_norm": 1.53125, "learning_rate": 9.489925741939485e-06, "loss": 0.1341, "step": 23053 }, { "epoch": 143.1931464174455, "grad_norm": 1.9921875, "learning_rate": 9.487993529811484e-06, "loss": 0.2045, "step": 23054 }, { "epoch": 143.1993769470405, "grad_norm": 2.5, "learning_rate": 9.486061468339203e-06, "loss": 0.1968, "step": 23055 }, { "epoch": 143.20560747663552, "grad_norm": 1.59375, "learning_rate": 9.484129557541383e-06, "loss": 0.0968, "step": 23056 }, { "epoch": 143.21183800623052, "grad_norm": 1.34375, "learning_rate": 9.482197797436799e-06, "loss": 0.1007, "step": 23057 }, { "epoch": 143.21806853582555, "grad_norm": 2.125, "learning_rate": 9.480266188044218e-06, "loss": 0.1903, "step": 23058 }, { "epoch": 143.22429906542055, "grad_norm": 1.3984375, "learning_rate": 9.478334729382395e-06, "loss": 0.1328, "step": 23059 }, { "epoch": 143.23052959501558, "grad_norm": 2.140625, "learning_rate": 9.476403421470082e-06, "loss": 0.1725, "step": 23060 }, { "epoch": 143.23676012461058, "grad_norm": 2.78125, "learning_rate": 9.474472264326039e-06, "loss": 0.1903, "step": 23061 }, { "epoch": 143.2429906542056, "grad_norm": 1.65625, "learning_rate": 9.472541257969036e-06, "loss": 0.1411, "step": 23062 }, { "epoch": 143.2492211838006, "grad_norm": 2.15625, "learning_rate": 9.470610402417807e-06, "loss": 0.1654, "step": 23063 }, { "epoch": 143.25545171339564, "grad_norm": 3.25, "learning_rate": 9.468679697691124e-06, "loss": 0.3978, "step": 23064 }, { "epoch": 143.26168224299064, "grad_norm": 2.703125, "learning_rate": 9.466749143807715e-06, "loss": 0.2062, "step": 23065 }, { "epoch": 143.26791277258567, "grad_norm": 1.3203125, "learning_rate": 9.464818740786358e-06, "loss": 0.1098, "step": 23066 }, { "epoch": 143.27414330218068, "grad_norm": 1.7890625, "learning_rate": 9.462888488645775e-06, "loss": 0.1054, "step": 23067 }, { "epoch": 143.2803738317757, "grad_norm": 1.703125, "learning_rate": 9.460958387404725e-06, "loss": 0.1159, "step": 23068 }, { "epoch": 143.2866043613707, "grad_norm": 1.328125, "learning_rate": 9.459028437081952e-06, "loss": 0.1095, "step": 23069 }, { "epoch": 143.29283489096574, "grad_norm": 2.1875, "learning_rate": 9.457098637696221e-06, "loss": 0.1571, "step": 23070 }, { "epoch": 143.29906542056074, "grad_norm": 2.59375, "learning_rate": 9.455168989266237e-06, "loss": 0.2749, "step": 23071 }, { "epoch": 143.30529595015577, "grad_norm": 1.9921875, "learning_rate": 9.45323949181076e-06, "loss": 0.1717, "step": 23072 }, { "epoch": 143.31152647975077, "grad_norm": 1.6328125, "learning_rate": 9.451310145348524e-06, "loss": 0.1483, "step": 23073 }, { "epoch": 143.3177570093458, "grad_norm": 2.890625, "learning_rate": 9.449380949898285e-06, "loss": 0.1735, "step": 23074 }, { "epoch": 143.3239875389408, "grad_norm": 2.453125, "learning_rate": 9.447451905478758e-06, "loss": 0.1991, "step": 23075 }, { "epoch": 143.33021806853583, "grad_norm": 1.15625, "learning_rate": 9.445523012108688e-06, "loss": 0.1017, "step": 23076 }, { "epoch": 143.33644859813083, "grad_norm": 1.046875, "learning_rate": 9.443594269806813e-06, "loss": 0.117, "step": 23077 }, { "epoch": 143.34267912772586, "grad_norm": 1.7734375, "learning_rate": 9.441665678591863e-06, "loss": 0.1028, "step": 23078 }, { "epoch": 143.34890965732086, "grad_norm": 1.8125, "learning_rate": 9.439737238482555e-06, "loss": 0.1337, "step": 23079 }, { "epoch": 143.3551401869159, "grad_norm": 1.3515625, "learning_rate": 9.437808949497632e-06, "loss": 0.0943, "step": 23080 }, { "epoch": 143.3613707165109, "grad_norm": 2.609375, "learning_rate": 9.435880811655826e-06, "loss": 0.2227, "step": 23081 }, { "epoch": 143.36760124610592, "grad_norm": 1.7265625, "learning_rate": 9.433952824975847e-06, "loss": 0.1082, "step": 23082 }, { "epoch": 143.37383177570092, "grad_norm": 1.3359375, "learning_rate": 9.43202498947644e-06, "loss": 0.1128, "step": 23083 }, { "epoch": 143.38006230529595, "grad_norm": 1.515625, "learning_rate": 9.430097305176311e-06, "loss": 0.1236, "step": 23084 }, { "epoch": 143.38629283489095, "grad_norm": 2.75, "learning_rate": 9.428169772094197e-06, "loss": 0.3335, "step": 23085 }, { "epoch": 143.39252336448598, "grad_norm": 1.6328125, "learning_rate": 9.426242390248802e-06, "loss": 0.1288, "step": 23086 }, { "epoch": 143.39875389408098, "grad_norm": 2.546875, "learning_rate": 9.424315159658853e-06, "loss": 0.2446, "step": 23087 }, { "epoch": 143.404984423676, "grad_norm": 2.265625, "learning_rate": 9.422388080343078e-06, "loss": 0.2381, "step": 23088 }, { "epoch": 143.41121495327104, "grad_norm": 2.328125, "learning_rate": 9.420461152320184e-06, "loss": 0.1794, "step": 23089 }, { "epoch": 143.41744548286604, "grad_norm": 1.0546875, "learning_rate": 9.418534375608879e-06, "loss": 0.1197, "step": 23090 }, { "epoch": 143.42367601246107, "grad_norm": 1.796875, "learning_rate": 9.416607750227879e-06, "loss": 0.1598, "step": 23091 }, { "epoch": 143.42990654205607, "grad_norm": 1.0859375, "learning_rate": 9.41468127619591e-06, "loss": 0.1185, "step": 23092 }, { "epoch": 143.4361370716511, "grad_norm": 1.5234375, "learning_rate": 9.412754953531663e-06, "loss": 0.1086, "step": 23093 }, { "epoch": 143.4423676012461, "grad_norm": 1.4609375, "learning_rate": 9.410828782253858e-06, "loss": 0.1158, "step": 23094 }, { "epoch": 143.44859813084113, "grad_norm": 2.71875, "learning_rate": 9.408902762381197e-06, "loss": 0.1872, "step": 23095 }, { "epoch": 143.45482866043614, "grad_norm": 1.65625, "learning_rate": 9.406976893932407e-06, "loss": 0.121, "step": 23096 }, { "epoch": 143.46105919003116, "grad_norm": 1.7421875, "learning_rate": 9.405051176926158e-06, "loss": 0.1283, "step": 23097 }, { "epoch": 143.46728971962617, "grad_norm": 1.4921875, "learning_rate": 9.40312561138117e-06, "loss": 0.1262, "step": 23098 }, { "epoch": 143.4735202492212, "grad_norm": 2.078125, "learning_rate": 9.401200197316143e-06, "loss": 0.1784, "step": 23099 }, { "epoch": 143.4797507788162, "grad_norm": 1.5859375, "learning_rate": 9.399274934749786e-06, "loss": 0.1507, "step": 23100 }, { "epoch": 143.48598130841123, "grad_norm": 1.8515625, "learning_rate": 9.397349823700779e-06, "loss": 0.1504, "step": 23101 }, { "epoch": 143.49221183800623, "grad_norm": 2.421875, "learning_rate": 9.39542486418784e-06, "loss": 0.1655, "step": 23102 }, { "epoch": 143.49844236760126, "grad_norm": 2.3125, "learning_rate": 9.393500056229654e-06, "loss": 0.1706, "step": 23103 }, { "epoch": 143.50467289719626, "grad_norm": 1.2265625, "learning_rate": 9.391575399844906e-06, "loss": 0.1094, "step": 23104 }, { "epoch": 143.5109034267913, "grad_norm": 1.2890625, "learning_rate": 9.3896508950523e-06, "loss": 0.0958, "step": 23105 }, { "epoch": 143.5171339563863, "grad_norm": 2.296875, "learning_rate": 9.387726541870523e-06, "loss": 0.261, "step": 23106 }, { "epoch": 143.52336448598132, "grad_norm": 1.4765625, "learning_rate": 9.385802340318275e-06, "loss": 0.2145, "step": 23107 }, { "epoch": 143.52959501557632, "grad_norm": 2.734375, "learning_rate": 9.383878290414228e-06, "loss": 0.1232, "step": 23108 }, { "epoch": 143.53582554517135, "grad_norm": 0.765625, "learning_rate": 9.381954392177083e-06, "loss": 0.0933, "step": 23109 }, { "epoch": 143.54205607476635, "grad_norm": 2.53125, "learning_rate": 9.380030645625514e-06, "loss": 0.1096, "step": 23110 }, { "epoch": 143.54828660436138, "grad_norm": 1.3984375, "learning_rate": 9.378107050778217e-06, "loss": 0.1349, "step": 23111 }, { "epoch": 143.55451713395638, "grad_norm": 1.6875, "learning_rate": 9.376183607653857e-06, "loss": 0.1986, "step": 23112 }, { "epoch": 143.5607476635514, "grad_norm": 2.5, "learning_rate": 9.374260316271127e-06, "loss": 0.2303, "step": 23113 }, { "epoch": 143.5669781931464, "grad_norm": 2.3125, "learning_rate": 9.372337176648712e-06, "loss": 0.1601, "step": 23114 }, { "epoch": 143.57320872274144, "grad_norm": 1.484375, "learning_rate": 9.370414188805281e-06, "loss": 0.1263, "step": 23115 }, { "epoch": 143.57943925233644, "grad_norm": 1.234375, "learning_rate": 9.368491352759506e-06, "loss": 0.1087, "step": 23116 }, { "epoch": 143.58566978193147, "grad_norm": 3.0, "learning_rate": 9.366568668530064e-06, "loss": 0.3772, "step": 23117 }, { "epoch": 143.59190031152647, "grad_norm": 1.578125, "learning_rate": 9.364646136135643e-06, "loss": 0.0916, "step": 23118 }, { "epoch": 143.5981308411215, "grad_norm": 1.6328125, "learning_rate": 9.362723755594896e-06, "loss": 0.1344, "step": 23119 }, { "epoch": 143.6043613707165, "grad_norm": 1.890625, "learning_rate": 9.360801526926502e-06, "loss": 0.1837, "step": 23120 }, { "epoch": 143.61059190031153, "grad_norm": 1.1953125, "learning_rate": 9.358879450149136e-06, "loss": 0.1074, "step": 23121 }, { "epoch": 143.61682242990653, "grad_norm": 2.375, "learning_rate": 9.356957525281463e-06, "loss": 0.2285, "step": 23122 }, { "epoch": 143.62305295950156, "grad_norm": 2.609375, "learning_rate": 9.355035752342135e-06, "loss": 0.2084, "step": 23123 }, { "epoch": 143.62928348909657, "grad_norm": 1.5546875, "learning_rate": 9.353114131349825e-06, "loss": 0.1401, "step": 23124 }, { "epoch": 143.6355140186916, "grad_norm": 2.25, "learning_rate": 9.351192662323208e-06, "loss": 0.1298, "step": 23125 }, { "epoch": 143.6417445482866, "grad_norm": 2.3125, "learning_rate": 9.349271345280927e-06, "loss": 0.2555, "step": 23126 }, { "epoch": 143.64797507788163, "grad_norm": 1.6328125, "learning_rate": 9.347350180241655e-06, "loss": 0.13, "step": 23127 }, { "epoch": 143.65420560747663, "grad_norm": 1.84375, "learning_rate": 9.345429167224051e-06, "loss": 0.1308, "step": 23128 }, { "epoch": 143.66043613707166, "grad_norm": 2.03125, "learning_rate": 9.343508306246771e-06, "loss": 0.175, "step": 23129 }, { "epoch": 143.66666666666666, "grad_norm": 1.6875, "learning_rate": 9.341587597328458e-06, "loss": 0.168, "step": 23130 }, { "epoch": 143.6728971962617, "grad_norm": 2.203125, "learning_rate": 9.339667040487776e-06, "loss": 0.1737, "step": 23131 }, { "epoch": 143.6791277258567, "grad_norm": 3.359375, "learning_rate": 9.33774663574338e-06, "loss": 0.1015, "step": 23132 }, { "epoch": 143.68535825545172, "grad_norm": 0.9453125, "learning_rate": 9.335826383113928e-06, "loss": 0.0884, "step": 23133 }, { "epoch": 143.69158878504672, "grad_norm": 1.8125, "learning_rate": 9.33390628261806e-06, "loss": 0.1496, "step": 23134 }, { "epoch": 143.69781931464175, "grad_norm": 1.3515625, "learning_rate": 9.331986334274417e-06, "loss": 0.119, "step": 23135 }, { "epoch": 143.70404984423675, "grad_norm": 3.390625, "learning_rate": 9.330066538101657e-06, "loss": 0.3193, "step": 23136 }, { "epoch": 143.71028037383178, "grad_norm": 1.9375, "learning_rate": 9.328146894118429e-06, "loss": 0.149, "step": 23137 }, { "epoch": 143.71651090342678, "grad_norm": 1.6796875, "learning_rate": 9.326227402343363e-06, "loss": 0.1544, "step": 23138 }, { "epoch": 143.7227414330218, "grad_norm": 1.5859375, "learning_rate": 9.32430806279511e-06, "loss": 0.1275, "step": 23139 }, { "epoch": 143.7289719626168, "grad_norm": 1.625, "learning_rate": 9.322388875492319e-06, "loss": 0.1421, "step": 23140 }, { "epoch": 143.73520249221184, "grad_norm": 1.765625, "learning_rate": 9.320469840453622e-06, "loss": 0.1025, "step": 23141 }, { "epoch": 143.74143302180684, "grad_norm": 2.6875, "learning_rate": 9.318550957697647e-06, "loss": 0.1973, "step": 23142 }, { "epoch": 143.74766355140187, "grad_norm": 1.421875, "learning_rate": 9.316632227243038e-06, "loss": 0.0976, "step": 23143 }, { "epoch": 143.75389408099687, "grad_norm": 2.8125, "learning_rate": 9.314713649108445e-06, "loss": 0.1606, "step": 23144 }, { "epoch": 143.7601246105919, "grad_norm": 1.71875, "learning_rate": 9.312795223312476e-06, "loss": 0.1272, "step": 23145 }, { "epoch": 143.7663551401869, "grad_norm": 3.234375, "learning_rate": 9.310876949873776e-06, "loss": 0.2026, "step": 23146 }, { "epoch": 143.77258566978193, "grad_norm": 2.984375, "learning_rate": 9.308958828810985e-06, "loss": 0.1876, "step": 23147 }, { "epoch": 143.77881619937693, "grad_norm": 1.6953125, "learning_rate": 9.30704086014272e-06, "loss": 0.1094, "step": 23148 }, { "epoch": 143.78504672897196, "grad_norm": 1.5625, "learning_rate": 9.305123043887606e-06, "loss": 0.0929, "step": 23149 }, { "epoch": 143.79127725856696, "grad_norm": 2.546875, "learning_rate": 9.303205380064273e-06, "loss": 0.2301, "step": 23150 }, { "epoch": 143.797507788162, "grad_norm": 1.9375, "learning_rate": 9.301287868691356e-06, "loss": 0.1714, "step": 23151 }, { "epoch": 143.80373831775702, "grad_norm": 1.5390625, "learning_rate": 9.299370509787461e-06, "loss": 0.1339, "step": 23152 }, { "epoch": 143.80996884735202, "grad_norm": 1.4921875, "learning_rate": 9.297453303371225e-06, "loss": 0.1151, "step": 23153 }, { "epoch": 143.81619937694705, "grad_norm": 1.5859375, "learning_rate": 9.295536249461253e-06, "loss": 0.1357, "step": 23154 }, { "epoch": 143.82242990654206, "grad_norm": 2.6875, "learning_rate": 9.293619348076183e-06, "loss": 0.2682, "step": 23155 }, { "epoch": 143.82866043613708, "grad_norm": 2.515625, "learning_rate": 9.29170259923461e-06, "loss": 0.2582, "step": 23156 }, { "epoch": 143.8348909657321, "grad_norm": 3.421875, "learning_rate": 9.289786002955165e-06, "loss": 0.2191, "step": 23157 }, { "epoch": 143.84112149532712, "grad_norm": 1.75, "learning_rate": 9.287869559256464e-06, "loss": 0.0961, "step": 23158 }, { "epoch": 143.84735202492212, "grad_norm": 1.515625, "learning_rate": 9.285953268157116e-06, "loss": 0.1358, "step": 23159 }, { "epoch": 143.85358255451715, "grad_norm": 3.03125, "learning_rate": 9.28403712967572e-06, "loss": 0.26, "step": 23160 }, { "epoch": 143.85981308411215, "grad_norm": 2.53125, "learning_rate": 9.282121143830902e-06, "loss": 0.2336, "step": 23161 }, { "epoch": 143.86604361370718, "grad_norm": 1.9609375, "learning_rate": 9.280205310641271e-06, "loss": 0.1633, "step": 23162 }, { "epoch": 143.87227414330218, "grad_norm": 1.9296875, "learning_rate": 9.27828963012542e-06, "loss": 0.1336, "step": 23163 }, { "epoch": 143.8785046728972, "grad_norm": 1.3828125, "learning_rate": 9.276374102301963e-06, "loss": 0.114, "step": 23164 }, { "epoch": 143.8847352024922, "grad_norm": 2.390625, "learning_rate": 9.274458727189502e-06, "loss": 0.143, "step": 23165 }, { "epoch": 143.89096573208724, "grad_norm": 1.7109375, "learning_rate": 9.27254350480666e-06, "loss": 0.1448, "step": 23166 }, { "epoch": 143.89719626168224, "grad_norm": 2.234375, "learning_rate": 9.270628435172005e-06, "loss": 0.1759, "step": 23167 }, { "epoch": 143.90342679127727, "grad_norm": 2.359375, "learning_rate": 9.268713518304148e-06, "loss": 0.1814, "step": 23168 }, { "epoch": 143.90965732087227, "grad_norm": 1.1953125, "learning_rate": 9.266798754221689e-06, "loss": 0.1232, "step": 23169 }, { "epoch": 143.9158878504673, "grad_norm": 2.40625, "learning_rate": 9.264884142943237e-06, "loss": 0.3231, "step": 23170 }, { "epoch": 143.9221183800623, "grad_norm": 1.5625, "learning_rate": 9.262969684487363e-06, "loss": 0.1289, "step": 23171 }, { "epoch": 143.92834890965733, "grad_norm": 2.109375, "learning_rate": 9.261055378872684e-06, "loss": 0.1488, "step": 23172 }, { "epoch": 143.93457943925233, "grad_norm": 2.3125, "learning_rate": 9.259141226117774e-06, "loss": 0.2434, "step": 23173 }, { "epoch": 143.94080996884736, "grad_norm": 2.546875, "learning_rate": 9.257227226241239e-06, "loss": 0.2395, "step": 23174 }, { "epoch": 143.94704049844236, "grad_norm": 1.8984375, "learning_rate": 9.255313379261648e-06, "loss": 0.1916, "step": 23175 }, { "epoch": 143.9532710280374, "grad_norm": 1.75, "learning_rate": 9.253399685197605e-06, "loss": 0.1425, "step": 23176 }, { "epoch": 143.9595015576324, "grad_norm": 1.0078125, "learning_rate": 9.251486144067698e-06, "loss": 0.0942, "step": 23177 }, { "epoch": 143.96573208722742, "grad_norm": 2.171875, "learning_rate": 9.249572755890498e-06, "loss": 0.1956, "step": 23178 }, { "epoch": 143.97196261682242, "grad_norm": 2.140625, "learning_rate": 9.247659520684604e-06, "loss": 0.1395, "step": 23179 }, { "epoch": 143.97819314641745, "grad_norm": 1.90625, "learning_rate": 9.245746438468583e-06, "loss": 0.1097, "step": 23180 }, { "epoch": 143.98442367601245, "grad_norm": 2.1875, "learning_rate": 9.243833509261029e-06, "loss": 0.1776, "step": 23181 }, { "epoch": 143.99065420560748, "grad_norm": 1.59375, "learning_rate": 9.241920733080504e-06, "loss": 0.1554, "step": 23182 }, { "epoch": 143.99688473520249, "grad_norm": 2.1875, "learning_rate": 9.240008109945596e-06, "loss": 0.1565, "step": 23183 }, { "epoch": 144.0, "grad_norm": 1.125, "learning_rate": 9.238095639874889e-06, "loss": 0.0635, "step": 23184 }, { "epoch": 144.00623052959503, "grad_norm": 2.578125, "learning_rate": 9.236183322886945e-06, "loss": 0.2171, "step": 23185 }, { "epoch": 144.01246105919003, "grad_norm": 2.78125, "learning_rate": 9.234271159000332e-06, "loss": 0.166, "step": 23186 }, { "epoch": 144.01869158878506, "grad_norm": 1.84375, "learning_rate": 9.23235914823363e-06, "loss": 0.1442, "step": 23187 }, { "epoch": 144.02492211838006, "grad_norm": 1.234375, "learning_rate": 9.230447290605418e-06, "loss": 0.1137, "step": 23188 }, { "epoch": 144.0311526479751, "grad_norm": 3.46875, "learning_rate": 9.228535586134244e-06, "loss": 0.1909, "step": 23189 }, { "epoch": 144.0373831775701, "grad_norm": 1.2578125, "learning_rate": 9.226624034838685e-06, "loss": 0.1132, "step": 23190 }, { "epoch": 144.04361370716512, "grad_norm": 2.078125, "learning_rate": 9.224712636737316e-06, "loss": 0.2387, "step": 23191 }, { "epoch": 144.04984423676012, "grad_norm": 2.96875, "learning_rate": 9.222801391848687e-06, "loss": 0.1831, "step": 23192 }, { "epoch": 144.05607476635515, "grad_norm": 2.546875, "learning_rate": 9.22089030019136e-06, "loss": 0.1938, "step": 23193 }, { "epoch": 144.06230529595015, "grad_norm": 1.515625, "learning_rate": 9.2189793617839e-06, "loss": 0.1304, "step": 23194 }, { "epoch": 144.06853582554518, "grad_norm": 1.953125, "learning_rate": 9.217068576644866e-06, "loss": 0.1492, "step": 23195 }, { "epoch": 144.07476635514018, "grad_norm": 2.5625, "learning_rate": 9.215157944792826e-06, "loss": 0.2542, "step": 23196 }, { "epoch": 144.0809968847352, "grad_norm": 2.40625, "learning_rate": 9.21324746624632e-06, "loss": 0.1529, "step": 23197 }, { "epoch": 144.08722741433021, "grad_norm": 2.03125, "learning_rate": 9.211337141023919e-06, "loss": 0.1792, "step": 23198 }, { "epoch": 144.09345794392524, "grad_norm": 2.171875, "learning_rate": 9.209426969144166e-06, "loss": 0.2349, "step": 23199 }, { "epoch": 144.09968847352025, "grad_norm": 1.9375, "learning_rate": 9.207516950625609e-06, "loss": 0.1626, "step": 23200 }, { "epoch": 144.10591900311528, "grad_norm": 2.046875, "learning_rate": 9.2056070854868e-06, "loss": 0.1067, "step": 23201 }, { "epoch": 144.11214953271028, "grad_norm": 3.0, "learning_rate": 9.203697373746295e-06, "loss": 0.303, "step": 23202 }, { "epoch": 144.1183800623053, "grad_norm": 2.71875, "learning_rate": 9.201787815422647e-06, "loss": 0.1898, "step": 23203 }, { "epoch": 144.1246105919003, "grad_norm": 1.171875, "learning_rate": 9.199878410534394e-06, "loss": 0.0885, "step": 23204 }, { "epoch": 144.13084112149534, "grad_norm": 2.015625, "learning_rate": 9.19796915910007e-06, "loss": 0.1173, "step": 23205 }, { "epoch": 144.13707165109034, "grad_norm": 1.9140625, "learning_rate": 9.196060061138229e-06, "loss": 0.162, "step": 23206 }, { "epoch": 144.14330218068537, "grad_norm": 1.09375, "learning_rate": 9.194151116667421e-06, "loss": 0.115, "step": 23207 }, { "epoch": 144.14953271028037, "grad_norm": 2.5, "learning_rate": 9.192242325706169e-06, "loss": 0.1605, "step": 23208 }, { "epoch": 144.1557632398754, "grad_norm": 3.09375, "learning_rate": 9.190333688273017e-06, "loss": 0.2404, "step": 23209 }, { "epoch": 144.1619937694704, "grad_norm": 1.890625, "learning_rate": 9.188425204386516e-06, "loss": 0.1432, "step": 23210 }, { "epoch": 144.16822429906543, "grad_norm": 1.7421875, "learning_rate": 9.18651687406519e-06, "loss": 0.1312, "step": 23211 }, { "epoch": 144.17445482866043, "grad_norm": 1.109375, "learning_rate": 9.184608697327562e-06, "loss": 0.0816, "step": 23212 }, { "epoch": 144.18068535825546, "grad_norm": 2.046875, "learning_rate": 9.182700674192177e-06, "loss": 0.0976, "step": 23213 }, { "epoch": 144.18691588785046, "grad_norm": 1.6640625, "learning_rate": 9.180792804677573e-06, "loss": 0.1183, "step": 23214 }, { "epoch": 144.1931464174455, "grad_norm": 2.421875, "learning_rate": 9.178885088802263e-06, "loss": 0.1868, "step": 23215 }, { "epoch": 144.1993769470405, "grad_norm": 2.0625, "learning_rate": 9.176977526584785e-06, "loss": 0.1419, "step": 23216 }, { "epoch": 144.20560747663552, "grad_norm": 1.703125, "learning_rate": 9.175070118043671e-06, "loss": 0.1724, "step": 23217 }, { "epoch": 144.21183800623052, "grad_norm": 1.5625, "learning_rate": 9.173162863197443e-06, "loss": 0.1237, "step": 23218 }, { "epoch": 144.21806853582555, "grad_norm": 1.1953125, "learning_rate": 9.171255762064609e-06, "loss": 0.1066, "step": 23219 }, { "epoch": 144.22429906542055, "grad_norm": 2.203125, "learning_rate": 9.169348814663704e-06, "loss": 0.1815, "step": 23220 }, { "epoch": 144.23052959501558, "grad_norm": 2.921875, "learning_rate": 9.167442021013256e-06, "loss": 0.1637, "step": 23221 }, { "epoch": 144.23676012461058, "grad_norm": 1.6015625, "learning_rate": 9.165535381131768e-06, "loss": 0.138, "step": 23222 }, { "epoch": 144.2429906542056, "grad_norm": 2.453125, "learning_rate": 9.163628895037774e-06, "loss": 0.1999, "step": 23223 }, { "epoch": 144.2492211838006, "grad_norm": 1.6953125, "learning_rate": 9.161722562749772e-06, "loss": 0.1593, "step": 23224 }, { "epoch": 144.25545171339564, "grad_norm": 2.234375, "learning_rate": 9.159816384286296e-06, "loss": 0.1726, "step": 23225 }, { "epoch": 144.26168224299064, "grad_norm": 1.3046875, "learning_rate": 9.157910359665842e-06, "loss": 0.1012, "step": 23226 }, { "epoch": 144.26791277258567, "grad_norm": 3.296875, "learning_rate": 9.156004488906927e-06, "loss": 0.2013, "step": 23227 }, { "epoch": 144.27414330218068, "grad_norm": 2.140625, "learning_rate": 9.154098772028067e-06, "loss": 0.1387, "step": 23228 }, { "epoch": 144.2803738317757, "grad_norm": 1.6796875, "learning_rate": 9.15219320904778e-06, "loss": 0.1278, "step": 23229 }, { "epoch": 144.2866043613707, "grad_norm": 2.15625, "learning_rate": 9.150287799984547e-06, "loss": 0.1378, "step": 23230 }, { "epoch": 144.29283489096574, "grad_norm": 1.515625, "learning_rate": 9.148382544856884e-06, "loss": 0.1198, "step": 23231 }, { "epoch": 144.29906542056074, "grad_norm": 2.453125, "learning_rate": 9.146477443683302e-06, "loss": 0.1875, "step": 23232 }, { "epoch": 144.30529595015577, "grad_norm": 1.2265625, "learning_rate": 9.144572496482306e-06, "loss": 0.112, "step": 23233 }, { "epoch": 144.31152647975077, "grad_norm": 2.703125, "learning_rate": 9.142667703272384e-06, "loss": 0.184, "step": 23234 }, { "epoch": 144.3177570093458, "grad_norm": 2.015625, "learning_rate": 9.140763064072045e-06, "loss": 0.1244, "step": 23235 }, { "epoch": 144.3239875389408, "grad_norm": 2.546875, "learning_rate": 9.138858578899789e-06, "loss": 0.2275, "step": 23236 }, { "epoch": 144.33021806853583, "grad_norm": 2.03125, "learning_rate": 9.136954247774113e-06, "loss": 0.0936, "step": 23237 }, { "epoch": 144.33644859813083, "grad_norm": 3.109375, "learning_rate": 9.1350500707135e-06, "loss": 0.2848, "step": 23238 }, { "epoch": 144.34267912772586, "grad_norm": 2.5, "learning_rate": 9.133146047736449e-06, "loss": 0.207, "step": 23239 }, { "epoch": 144.34890965732086, "grad_norm": 1.9921875, "learning_rate": 9.131242178861466e-06, "loss": 0.1857, "step": 23240 }, { "epoch": 144.3551401869159, "grad_norm": 1.0546875, "learning_rate": 9.12933846410702e-06, "loss": 0.1123, "step": 23241 }, { "epoch": 144.3613707165109, "grad_norm": 2.375, "learning_rate": 9.127434903491625e-06, "loss": 0.2347, "step": 23242 }, { "epoch": 144.36760124610592, "grad_norm": 2.109375, "learning_rate": 9.125531497033744e-06, "loss": 0.1018, "step": 23243 }, { "epoch": 144.37383177570092, "grad_norm": 2.390625, "learning_rate": 9.123628244751881e-06, "loss": 0.2509, "step": 23244 }, { "epoch": 144.38006230529595, "grad_norm": 1.6953125, "learning_rate": 9.121725146664506e-06, "loss": 0.09, "step": 23245 }, { "epoch": 144.38629283489095, "grad_norm": 2.59375, "learning_rate": 9.119822202790113e-06, "loss": 0.2471, "step": 23246 }, { "epoch": 144.39252336448598, "grad_norm": 1.0703125, "learning_rate": 9.117919413147186e-06, "loss": 0.1063, "step": 23247 }, { "epoch": 144.39875389408098, "grad_norm": 2.328125, "learning_rate": 9.116016777754195e-06, "loss": 0.2038, "step": 23248 }, { "epoch": 144.404984423676, "grad_norm": 2.234375, "learning_rate": 9.11411429662963e-06, "loss": 0.1613, "step": 23249 }, { "epoch": 144.41121495327104, "grad_norm": 1.7265625, "learning_rate": 9.112211969791956e-06, "loss": 0.1395, "step": 23250 }, { "epoch": 144.41744548286604, "grad_norm": 1.5234375, "learning_rate": 9.110309797259664e-06, "loss": 0.1359, "step": 23251 }, { "epoch": 144.42367601246107, "grad_norm": 1.9453125, "learning_rate": 9.10840777905121e-06, "loss": 0.1781, "step": 23252 }, { "epoch": 144.42990654205607, "grad_norm": 1.625, "learning_rate": 9.106505915185076e-06, "loss": 0.1342, "step": 23253 }, { "epoch": 144.4361370716511, "grad_norm": 1.765625, "learning_rate": 9.104604205679742e-06, "loss": 0.1233, "step": 23254 }, { "epoch": 144.4423676012461, "grad_norm": 2.109375, "learning_rate": 9.102702650553671e-06, "loss": 0.1442, "step": 23255 }, { "epoch": 144.44859813084113, "grad_norm": 1.9921875, "learning_rate": 9.10080124982532e-06, "loss": 0.192, "step": 23256 }, { "epoch": 144.45482866043614, "grad_norm": 1.65625, "learning_rate": 9.098900003513166e-06, "loss": 0.1245, "step": 23257 }, { "epoch": 144.46105919003116, "grad_norm": 2.21875, "learning_rate": 9.096998911635682e-06, "loss": 0.1436, "step": 23258 }, { "epoch": 144.46728971962617, "grad_norm": 2.40625, "learning_rate": 9.095097974211314e-06, "loss": 0.201, "step": 23259 }, { "epoch": 144.4735202492212, "grad_norm": 2.03125, "learning_rate": 9.093197191258535e-06, "loss": 0.1048, "step": 23260 }, { "epoch": 144.4797507788162, "grad_norm": 1.65625, "learning_rate": 9.091296562795815e-06, "loss": 0.1507, "step": 23261 }, { "epoch": 144.48598130841123, "grad_norm": 2.78125, "learning_rate": 9.089396088841598e-06, "loss": 0.2385, "step": 23262 }, { "epoch": 144.49221183800623, "grad_norm": 1.59375, "learning_rate": 9.087495769414341e-06, "loss": 0.1158, "step": 23263 }, { "epoch": 144.49844236760126, "grad_norm": 2.796875, "learning_rate": 9.085595604532506e-06, "loss": 0.2186, "step": 23264 }, { "epoch": 144.50467289719626, "grad_norm": 2.734375, "learning_rate": 9.083695594214548e-06, "loss": 0.1115, "step": 23265 }, { "epoch": 144.5109034267913, "grad_norm": 1.5546875, "learning_rate": 9.081795738478929e-06, "loss": 0.1184, "step": 23266 }, { "epoch": 144.5171339563863, "grad_norm": 2.796875, "learning_rate": 9.079896037344082e-06, "loss": 0.3204, "step": 23267 }, { "epoch": 144.52336448598132, "grad_norm": 2.296875, "learning_rate": 9.077996490828475e-06, "loss": 0.1834, "step": 23268 }, { "epoch": 144.52959501557632, "grad_norm": 1.3515625, "learning_rate": 9.076097098950543e-06, "loss": 0.1129, "step": 23269 }, { "epoch": 144.53582554517135, "grad_norm": 1.3515625, "learning_rate": 9.074197861728746e-06, "loss": 0.1196, "step": 23270 }, { "epoch": 144.54205607476635, "grad_norm": 1.34375, "learning_rate": 9.072298779181513e-06, "loss": 0.1085, "step": 23271 }, { "epoch": 144.54828660436138, "grad_norm": 1.90625, "learning_rate": 9.070399851327299e-06, "loss": 0.2047, "step": 23272 }, { "epoch": 144.55451713395638, "grad_norm": 2.21875, "learning_rate": 9.068501078184555e-06, "loss": 0.1279, "step": 23273 }, { "epoch": 144.5607476635514, "grad_norm": 3.078125, "learning_rate": 9.06660245977171e-06, "loss": 0.2285, "step": 23274 }, { "epoch": 144.5669781931464, "grad_norm": 1.4453125, "learning_rate": 9.064703996107205e-06, "loss": 0.1335, "step": 23275 }, { "epoch": 144.57320872274144, "grad_norm": 2.046875, "learning_rate": 9.062805687209474e-06, "loss": 0.175, "step": 23276 }, { "epoch": 144.57943925233644, "grad_norm": 2.59375, "learning_rate": 9.060907533096971e-06, "loss": 0.2203, "step": 23277 }, { "epoch": 144.58566978193147, "grad_norm": 1.703125, "learning_rate": 9.059009533788112e-06, "loss": 0.1158, "step": 23278 }, { "epoch": 144.59190031152647, "grad_norm": 2.203125, "learning_rate": 9.05711168930134e-06, "loss": 0.2386, "step": 23279 }, { "epoch": 144.5981308411215, "grad_norm": 1.78125, "learning_rate": 9.055213999655093e-06, "loss": 0.1635, "step": 23280 }, { "epoch": 144.6043613707165, "grad_norm": 2.796875, "learning_rate": 9.053316464867794e-06, "loss": 0.1951, "step": 23281 }, { "epoch": 144.61059190031153, "grad_norm": 2.125, "learning_rate": 9.051419084957865e-06, "loss": 0.1304, "step": 23282 }, { "epoch": 144.61682242990653, "grad_norm": 2.28125, "learning_rate": 9.049521859943743e-06, "loss": 0.191, "step": 23283 }, { "epoch": 144.62305295950156, "grad_norm": 3.015625, "learning_rate": 9.047624789843861e-06, "loss": 0.2233, "step": 23284 }, { "epoch": 144.62928348909657, "grad_norm": 1.1015625, "learning_rate": 9.045727874676629e-06, "loss": 0.1083, "step": 23285 }, { "epoch": 144.6355140186916, "grad_norm": 1.5078125, "learning_rate": 9.043831114460476e-06, "loss": 0.1184, "step": 23286 }, { "epoch": 144.6417445482866, "grad_norm": 1.4609375, "learning_rate": 9.041934509213835e-06, "loss": 0.1054, "step": 23287 }, { "epoch": 144.64797507788163, "grad_norm": 2.5625, "learning_rate": 9.040038058955111e-06, "loss": 0.2652, "step": 23288 }, { "epoch": 144.65420560747663, "grad_norm": 1.6953125, "learning_rate": 9.038141763702723e-06, "loss": 0.1038, "step": 23289 }, { "epoch": 144.66043613707166, "grad_norm": 1.34375, "learning_rate": 9.036245623475092e-06, "loss": 0.1199, "step": 23290 }, { "epoch": 144.66666666666666, "grad_norm": 0.9765625, "learning_rate": 9.034349638290642e-06, "loss": 0.1062, "step": 23291 }, { "epoch": 144.6728971962617, "grad_norm": 1.890625, "learning_rate": 9.032453808167774e-06, "loss": 0.1524, "step": 23292 }, { "epoch": 144.6791277258567, "grad_norm": 2.5, "learning_rate": 9.030558133124914e-06, "loss": 0.1759, "step": 23293 }, { "epoch": 144.68535825545172, "grad_norm": 2.28125, "learning_rate": 9.028662613180457e-06, "loss": 0.1408, "step": 23294 }, { "epoch": 144.69158878504672, "grad_norm": 1.9375, "learning_rate": 9.02676724835283e-06, "loss": 0.16, "step": 23295 }, { "epoch": 144.69781931464175, "grad_norm": 1.9921875, "learning_rate": 9.024872038660422e-06, "loss": 0.1487, "step": 23296 }, { "epoch": 144.70404984423675, "grad_norm": 2.1875, "learning_rate": 9.022976984121653e-06, "loss": 0.2178, "step": 23297 }, { "epoch": 144.71028037383178, "grad_norm": 1.65625, "learning_rate": 9.021082084754923e-06, "loss": 0.0969, "step": 23298 }, { "epoch": 144.71651090342678, "grad_norm": 2.8125, "learning_rate": 9.019187340578656e-06, "loss": 0.2831, "step": 23299 }, { "epoch": 144.7227414330218, "grad_norm": 2.25, "learning_rate": 9.01729275161122e-06, "loss": 0.1576, "step": 23300 }, { "epoch": 144.7289719626168, "grad_norm": 1.375, "learning_rate": 9.01539831787103e-06, "loss": 0.1349, "step": 23301 }, { "epoch": 144.73520249221184, "grad_norm": 2.453125, "learning_rate": 9.013504039376485e-06, "loss": 0.2002, "step": 23302 }, { "epoch": 144.74143302180684, "grad_norm": 0.84375, "learning_rate": 9.011609916145997e-06, "loss": 0.1092, "step": 23303 }, { "epoch": 144.74766355140187, "grad_norm": 2.09375, "learning_rate": 9.009715948197938e-06, "loss": 0.1487, "step": 23304 }, { "epoch": 144.75389408099687, "grad_norm": 2.171875, "learning_rate": 9.007822135550717e-06, "loss": 0.2398, "step": 23305 }, { "epoch": 144.7601246105919, "grad_norm": 1.9453125, "learning_rate": 9.00592847822273e-06, "loss": 0.2555, "step": 23306 }, { "epoch": 144.7663551401869, "grad_norm": 1.6171875, "learning_rate": 9.004034976232365e-06, "loss": 0.1127, "step": 23307 }, { "epoch": 144.77258566978193, "grad_norm": 1.96875, "learning_rate": 9.002141629598e-06, "loss": 0.1517, "step": 23308 }, { "epoch": 144.77881619937693, "grad_norm": 3.125, "learning_rate": 9.000248438338033e-06, "loss": 0.2574, "step": 23309 }, { "epoch": 144.78504672897196, "grad_norm": 2.765625, "learning_rate": 8.99835540247086e-06, "loss": 0.1825, "step": 23310 }, { "epoch": 144.79127725856696, "grad_norm": 2.0625, "learning_rate": 8.99646252201485e-06, "loss": 0.2371, "step": 23311 }, { "epoch": 144.797507788162, "grad_norm": 2.25, "learning_rate": 8.994569796988403e-06, "loss": 0.1897, "step": 23312 }, { "epoch": 144.80373831775702, "grad_norm": 0.8671875, "learning_rate": 8.992677227409888e-06, "loss": 0.0956, "step": 23313 }, { "epoch": 144.80996884735202, "grad_norm": 2.921875, "learning_rate": 8.990784813297695e-06, "loss": 0.286, "step": 23314 }, { "epoch": 144.81619937694705, "grad_norm": 1.953125, "learning_rate": 8.988892554670194e-06, "loss": 0.1482, "step": 23315 }, { "epoch": 144.82242990654206, "grad_norm": 3.171875, "learning_rate": 8.987000451545772e-06, "loss": 0.1, "step": 23316 }, { "epoch": 144.82866043613708, "grad_norm": 1.6484375, "learning_rate": 8.985108503942808e-06, "loss": 0.1223, "step": 23317 }, { "epoch": 144.8348909657321, "grad_norm": 2.0, "learning_rate": 8.983216711879663e-06, "loss": 0.1599, "step": 23318 }, { "epoch": 144.84112149532712, "grad_norm": 1.34375, "learning_rate": 8.981325075374728e-06, "loss": 0.1293, "step": 23319 }, { "epoch": 144.84735202492212, "grad_norm": 1.296875, "learning_rate": 8.979433594446356e-06, "loss": 0.1193, "step": 23320 }, { "epoch": 144.85358255451715, "grad_norm": 1.421875, "learning_rate": 8.977542269112938e-06, "loss": 0.1289, "step": 23321 }, { "epoch": 144.85981308411215, "grad_norm": 3.015625, "learning_rate": 8.975651099392823e-06, "loss": 0.192, "step": 23322 }, { "epoch": 144.86604361370718, "grad_norm": 3.46875, "learning_rate": 8.973760085304386e-06, "loss": 0.3725, "step": 23323 }, { "epoch": 144.87227414330218, "grad_norm": 2.359375, "learning_rate": 8.971869226865998e-06, "loss": 0.2198, "step": 23324 }, { "epoch": 144.8785046728972, "grad_norm": 2.515625, "learning_rate": 8.969978524096034e-06, "loss": 0.0995, "step": 23325 }, { "epoch": 144.8847352024922, "grad_norm": 1.9765625, "learning_rate": 8.968087977012829e-06, "loss": 0.1525, "step": 23326 }, { "epoch": 144.89096573208724, "grad_norm": 2.09375, "learning_rate": 8.966197585634755e-06, "loss": 0.1573, "step": 23327 }, { "epoch": 144.89719626168224, "grad_norm": 1.4609375, "learning_rate": 8.964307349980177e-06, "loss": 0.1245, "step": 23328 }, { "epoch": 144.90342679127727, "grad_norm": 2.25, "learning_rate": 8.96241727006746e-06, "loss": 0.1511, "step": 23329 }, { "epoch": 144.90965732087227, "grad_norm": 1.03125, "learning_rate": 8.960527345914943e-06, "loss": 0.0972, "step": 23330 }, { "epoch": 144.9158878504673, "grad_norm": 2.453125, "learning_rate": 8.958637577541e-06, "loss": 0.1273, "step": 23331 }, { "epoch": 144.9221183800623, "grad_norm": 1.1015625, "learning_rate": 8.956747964963966e-06, "loss": 0.0983, "step": 23332 }, { "epoch": 144.92834890965733, "grad_norm": 1.4140625, "learning_rate": 8.954858508202212e-06, "loss": 0.1167, "step": 23333 }, { "epoch": 144.93457943925233, "grad_norm": 2.90625, "learning_rate": 8.95296920727407e-06, "loss": 0.3585, "step": 23334 }, { "epoch": 144.94080996884736, "grad_norm": 2.640625, "learning_rate": 8.9510800621979e-06, "loss": 0.284, "step": 23335 }, { "epoch": 144.94704049844236, "grad_norm": 1.4296875, "learning_rate": 8.949191072992057e-06, "loss": 0.1482, "step": 23336 }, { "epoch": 144.9532710280374, "grad_norm": 2.265625, "learning_rate": 8.947302239674871e-06, "loss": 0.2092, "step": 23337 }, { "epoch": 144.9595015576324, "grad_norm": 2.296875, "learning_rate": 8.945413562264702e-06, "loss": 0.1868, "step": 23338 }, { "epoch": 144.96573208722742, "grad_norm": 2.0625, "learning_rate": 8.943525040779876e-06, "loss": 0.1766, "step": 23339 }, { "epoch": 144.97196261682242, "grad_norm": 2.375, "learning_rate": 8.941636675238754e-06, "loss": 0.1601, "step": 23340 }, { "epoch": 144.97819314641745, "grad_norm": 0.83984375, "learning_rate": 8.93974846565966e-06, "loss": 0.1073, "step": 23341 }, { "epoch": 144.98442367601245, "grad_norm": 2.609375, "learning_rate": 8.937860412060936e-06, "loss": 0.2189, "step": 23342 }, { "epoch": 144.99065420560748, "grad_norm": 0.828125, "learning_rate": 8.935972514460933e-06, "loss": 0.0835, "step": 23343 }, { "epoch": 144.99688473520249, "grad_norm": 2.296875, "learning_rate": 8.934084772877977e-06, "loss": 0.0985, "step": 23344 }, { "epoch": 145.0, "grad_norm": 0.7421875, "learning_rate": 8.93219718733039e-06, "loss": 0.0553, "step": 23345 }, { "epoch": 145.00623052959503, "grad_norm": 1.1484375, "learning_rate": 8.930309757836517e-06, "loss": 0.1266, "step": 23346 }, { "epoch": 145.01246105919003, "grad_norm": 2.71875, "learning_rate": 8.928422484414698e-06, "loss": 0.2218, "step": 23347 }, { "epoch": 145.01869158878506, "grad_norm": 1.953125, "learning_rate": 8.926535367083241e-06, "loss": 0.1296, "step": 23348 }, { "epoch": 145.02492211838006, "grad_norm": 1.5, "learning_rate": 8.924648405860486e-06, "loss": 0.1179, "step": 23349 }, { "epoch": 145.0311526479751, "grad_norm": 1.984375, "learning_rate": 8.92276160076477e-06, "loss": 0.1119, "step": 23350 }, { "epoch": 145.0373831775701, "grad_norm": 1.4375, "learning_rate": 8.920874951814406e-06, "loss": 0.1372, "step": 23351 }, { "epoch": 145.04361370716512, "grad_norm": 1.9296875, "learning_rate": 8.918988459027708e-06, "loss": 0.1122, "step": 23352 }, { "epoch": 145.04984423676012, "grad_norm": 2.0625, "learning_rate": 8.917102122423012e-06, "loss": 0.1061, "step": 23353 }, { "epoch": 145.05607476635515, "grad_norm": 2.46875, "learning_rate": 8.915215942018643e-06, "loss": 0.1042, "step": 23354 }, { "epoch": 145.06230529595015, "grad_norm": 1.46875, "learning_rate": 8.913329917832902e-06, "loss": 0.13, "step": 23355 }, { "epoch": 145.06853582554518, "grad_norm": 2.4375, "learning_rate": 8.911444049884116e-06, "loss": 0.2044, "step": 23356 }, { "epoch": 145.07476635514018, "grad_norm": 2.15625, "learning_rate": 8.90955833819061e-06, "loss": 0.2326, "step": 23357 }, { "epoch": 145.0809968847352, "grad_norm": 1.8671875, "learning_rate": 8.907672782770693e-06, "loss": 0.1351, "step": 23358 }, { "epoch": 145.08722741433021, "grad_norm": 3.375, "learning_rate": 8.905787383642663e-06, "loss": 0.2781, "step": 23359 }, { "epoch": 145.09345794392524, "grad_norm": 2.578125, "learning_rate": 8.903902140824843e-06, "loss": 0.2074, "step": 23360 }, { "epoch": 145.09968847352025, "grad_norm": 3.1875, "learning_rate": 8.902017054335546e-06, "loss": 0.267, "step": 23361 }, { "epoch": 145.10591900311528, "grad_norm": 2.234375, "learning_rate": 8.900132124193084e-06, "loss": 0.1395, "step": 23362 }, { "epoch": 145.11214953271028, "grad_norm": 1.125, "learning_rate": 8.898247350415756e-06, "loss": 0.1093, "step": 23363 }, { "epoch": 145.1183800623053, "grad_norm": 1.890625, "learning_rate": 8.89636273302186e-06, "loss": 0.1217, "step": 23364 }, { "epoch": 145.1246105919003, "grad_norm": 2.671875, "learning_rate": 8.89447827202971e-06, "loss": 0.1889, "step": 23365 }, { "epoch": 145.13084112149534, "grad_norm": 1.28125, "learning_rate": 8.892593967457613e-06, "loss": 0.1127, "step": 23366 }, { "epoch": 145.13707165109034, "grad_norm": 2.28125, "learning_rate": 8.890709819323856e-06, "loss": 0.0973, "step": 23367 }, { "epoch": 145.14330218068537, "grad_norm": 1.984375, "learning_rate": 8.888825827646747e-06, "loss": 0.1533, "step": 23368 }, { "epoch": 145.14953271028037, "grad_norm": 1.4375, "learning_rate": 8.88694199244459e-06, "loss": 0.1296, "step": 23369 }, { "epoch": 145.1557632398754, "grad_norm": 1.859375, "learning_rate": 8.88505831373567e-06, "loss": 0.1615, "step": 23370 }, { "epoch": 145.1619937694704, "grad_norm": 1.8515625, "learning_rate": 8.88317479153828e-06, "loss": 0.1306, "step": 23371 }, { "epoch": 145.16822429906543, "grad_norm": 1.4375, "learning_rate": 8.881291425870717e-06, "loss": 0.1148, "step": 23372 }, { "epoch": 145.17445482866043, "grad_norm": 1.9453125, "learning_rate": 8.879408216751281e-06, "loss": 0.1743, "step": 23373 }, { "epoch": 145.18068535825546, "grad_norm": 2.703125, "learning_rate": 8.877525164198248e-06, "loss": 0.2452, "step": 23374 }, { "epoch": 145.18691588785046, "grad_norm": 1.953125, "learning_rate": 8.875642268229912e-06, "loss": 0.1655, "step": 23375 }, { "epoch": 145.1931464174455, "grad_norm": 1.046875, "learning_rate": 8.87375952886457e-06, "loss": 0.1094, "step": 23376 }, { "epoch": 145.1993769470405, "grad_norm": 3.359375, "learning_rate": 8.8718769461205e-06, "loss": 0.3394, "step": 23377 }, { "epoch": 145.20560747663552, "grad_norm": 1.484375, "learning_rate": 8.869994520015976e-06, "loss": 0.1258, "step": 23378 }, { "epoch": 145.21183800623052, "grad_norm": 1.1015625, "learning_rate": 8.868112250569288e-06, "loss": 0.1222, "step": 23379 }, { "epoch": 145.21806853582555, "grad_norm": 2.25, "learning_rate": 8.86623013779873e-06, "loss": 0.1375, "step": 23380 }, { "epoch": 145.22429906542055, "grad_norm": 1.640625, "learning_rate": 8.864348181722559e-06, "loss": 0.1061, "step": 23381 }, { "epoch": 145.23052959501558, "grad_norm": 2.203125, "learning_rate": 8.862466382359076e-06, "loss": 0.1877, "step": 23382 }, { "epoch": 145.23676012461058, "grad_norm": 2.125, "learning_rate": 8.860584739726532e-06, "loss": 0.2065, "step": 23383 }, { "epoch": 145.2429906542056, "grad_norm": 1.4765625, "learning_rate": 8.858703253843228e-06, "loss": 0.1221, "step": 23384 }, { "epoch": 145.2492211838006, "grad_norm": 2.03125, "learning_rate": 8.856821924727412e-06, "loss": 0.1885, "step": 23385 }, { "epoch": 145.25545171339564, "grad_norm": 2.6875, "learning_rate": 8.854940752397373e-06, "loss": 0.1822, "step": 23386 }, { "epoch": 145.26168224299064, "grad_norm": 1.296875, "learning_rate": 8.853059736871382e-06, "loss": 0.0955, "step": 23387 }, { "epoch": 145.26791277258567, "grad_norm": 1.9921875, "learning_rate": 8.851178878167695e-06, "loss": 0.1742, "step": 23388 }, { "epoch": 145.27414330218068, "grad_norm": 1.203125, "learning_rate": 8.849298176304597e-06, "loss": 0.1154, "step": 23389 }, { "epoch": 145.2803738317757, "grad_norm": 1.1875, "learning_rate": 8.847417631300334e-06, "loss": 0.1087, "step": 23390 }, { "epoch": 145.2866043613707, "grad_norm": 1.28125, "learning_rate": 8.845537243173186e-06, "loss": 0.1343, "step": 23391 }, { "epoch": 145.29283489096574, "grad_norm": 2.078125, "learning_rate": 8.843657011941403e-06, "loss": 0.1941, "step": 23392 }, { "epoch": 145.29906542056074, "grad_norm": 1.609375, "learning_rate": 8.841776937623253e-06, "loss": 0.1293, "step": 23393 }, { "epoch": 145.30529595015577, "grad_norm": 3.734375, "learning_rate": 8.839897020236994e-06, "loss": 0.2907, "step": 23394 }, { "epoch": 145.31152647975077, "grad_norm": 2.140625, "learning_rate": 8.838017259800904e-06, "loss": 0.1598, "step": 23395 }, { "epoch": 145.3177570093458, "grad_norm": 1.2890625, "learning_rate": 8.836137656333201e-06, "loss": 0.1083, "step": 23396 }, { "epoch": 145.3239875389408, "grad_norm": 3.015625, "learning_rate": 8.834258209852165e-06, "loss": 0.196, "step": 23397 }, { "epoch": 145.33021806853583, "grad_norm": 1.5234375, "learning_rate": 8.832378920376041e-06, "loss": 0.1263, "step": 23398 }, { "epoch": 145.33644859813083, "grad_norm": 1.421875, "learning_rate": 8.830499787923094e-06, "loss": 0.135, "step": 23399 }, { "epoch": 145.34267912772586, "grad_norm": 2.578125, "learning_rate": 8.828620812511557e-06, "loss": 0.1941, "step": 23400 }, { "epoch": 145.34890965732086, "grad_norm": 2.3125, "learning_rate": 8.826741994159696e-06, "loss": 0.242, "step": 23401 }, { "epoch": 145.3551401869159, "grad_norm": 1.5, "learning_rate": 8.824863332885742e-06, "loss": 0.1448, "step": 23402 }, { "epoch": 145.3613707165109, "grad_norm": 1.796875, "learning_rate": 8.822984828707957e-06, "loss": 0.1449, "step": 23403 }, { "epoch": 145.36760124610592, "grad_norm": 1.6796875, "learning_rate": 8.821106481644568e-06, "loss": 0.127, "step": 23404 }, { "epoch": 145.37383177570092, "grad_norm": 1.125, "learning_rate": 8.819228291713827e-06, "loss": 0.0977, "step": 23405 }, { "epoch": 145.38006230529595, "grad_norm": 1.71875, "learning_rate": 8.817350258933985e-06, "loss": 0.1406, "step": 23406 }, { "epoch": 145.38629283489095, "grad_norm": 3.21875, "learning_rate": 8.815472383323262e-06, "loss": 0.3559, "step": 23407 }, { "epoch": 145.39252336448598, "grad_norm": 1.9921875, "learning_rate": 8.813594664899913e-06, "loss": 0.1533, "step": 23408 }, { "epoch": 145.39875389408098, "grad_norm": 1.140625, "learning_rate": 8.811717103682163e-06, "loss": 0.1006, "step": 23409 }, { "epoch": 145.404984423676, "grad_norm": 2.34375, "learning_rate": 8.809839699688261e-06, "loss": 0.2112, "step": 23410 }, { "epoch": 145.41121495327104, "grad_norm": 1.09375, "learning_rate": 8.807962452936422e-06, "loss": 0.0995, "step": 23411 }, { "epoch": 145.41744548286604, "grad_norm": 1.84375, "learning_rate": 8.806085363444888e-06, "loss": 0.2404, "step": 23412 }, { "epoch": 145.42367601246107, "grad_norm": 1.484375, "learning_rate": 8.8042084312319e-06, "loss": 0.1367, "step": 23413 }, { "epoch": 145.42990654205607, "grad_norm": 1.1640625, "learning_rate": 8.802331656315677e-06, "loss": 0.1146, "step": 23414 }, { "epoch": 145.4361370716511, "grad_norm": 2.09375, "learning_rate": 8.800455038714439e-06, "loss": 0.1964, "step": 23415 }, { "epoch": 145.4423676012461, "grad_norm": 2.390625, "learning_rate": 8.798578578446417e-06, "loss": 0.1804, "step": 23416 }, { "epoch": 145.44859813084113, "grad_norm": 1.8515625, "learning_rate": 8.79670227552985e-06, "loss": 0.1003, "step": 23417 }, { "epoch": 145.45482866043614, "grad_norm": 1.1875, "learning_rate": 8.794826129982939e-06, "loss": 0.1043, "step": 23418 }, { "epoch": 145.46105919003116, "grad_norm": 1.8671875, "learning_rate": 8.792950141823916e-06, "loss": 0.1485, "step": 23419 }, { "epoch": 145.46728971962617, "grad_norm": 2.1875, "learning_rate": 8.791074311071012e-06, "loss": 0.1803, "step": 23420 }, { "epoch": 145.4735202492212, "grad_norm": 2.8125, "learning_rate": 8.789198637742433e-06, "loss": 0.171, "step": 23421 }, { "epoch": 145.4797507788162, "grad_norm": 2.25, "learning_rate": 8.787323121856388e-06, "loss": 0.1145, "step": 23422 }, { "epoch": 145.48598130841123, "grad_norm": 1.765625, "learning_rate": 8.785447763431101e-06, "loss": 0.119, "step": 23423 }, { "epoch": 145.49221183800623, "grad_norm": 1.6171875, "learning_rate": 8.78357256248479e-06, "loss": 0.1326, "step": 23424 }, { "epoch": 145.49844236760126, "grad_norm": 2.828125, "learning_rate": 8.78169751903567e-06, "loss": 0.2672, "step": 23425 }, { "epoch": 145.50467289719626, "grad_norm": 1.4296875, "learning_rate": 8.779822633101936e-06, "loss": 0.118, "step": 23426 }, { "epoch": 145.5109034267913, "grad_norm": 2.25, "learning_rate": 8.777947904701816e-06, "loss": 0.1974, "step": 23427 }, { "epoch": 145.5171339563863, "grad_norm": 2.203125, "learning_rate": 8.776073333853501e-06, "loss": 0.1967, "step": 23428 }, { "epoch": 145.52336448598132, "grad_norm": 2.53125, "learning_rate": 8.774198920575213e-06, "loss": 0.2083, "step": 23429 }, { "epoch": 145.52959501557632, "grad_norm": 2.484375, "learning_rate": 8.772324664885142e-06, "loss": 0.1046, "step": 23430 }, { "epoch": 145.53582554517135, "grad_norm": 2.5625, "learning_rate": 8.770450566801497e-06, "loss": 0.1237, "step": 23431 }, { "epoch": 145.54205607476635, "grad_norm": 3.359375, "learning_rate": 8.768576626342488e-06, "loss": 0.1365, "step": 23432 }, { "epoch": 145.54828660436138, "grad_norm": 2.234375, "learning_rate": 8.766702843526306e-06, "loss": 0.145, "step": 23433 }, { "epoch": 145.55451713395638, "grad_norm": 2.09375, "learning_rate": 8.764829218371142e-06, "loss": 0.1057, "step": 23434 }, { "epoch": 145.5607476635514, "grad_norm": 2.578125, "learning_rate": 8.762955750895205e-06, "loss": 0.2557, "step": 23435 }, { "epoch": 145.5669781931464, "grad_norm": 1.4375, "learning_rate": 8.761082441116694e-06, "loss": 0.1114, "step": 23436 }, { "epoch": 145.57320872274144, "grad_norm": 3.140625, "learning_rate": 8.759209289053788e-06, "loss": 0.2799, "step": 23437 }, { "epoch": 145.57943925233644, "grad_norm": 1.7734375, "learning_rate": 8.757336294724689e-06, "loss": 0.1701, "step": 23438 }, { "epoch": 145.58566978193147, "grad_norm": 1.34375, "learning_rate": 8.755463458147595e-06, "loss": 0.1295, "step": 23439 }, { "epoch": 145.59190031152647, "grad_norm": 1.796875, "learning_rate": 8.753590779340686e-06, "loss": 0.1629, "step": 23440 }, { "epoch": 145.5981308411215, "grad_norm": 1.7421875, "learning_rate": 8.751718258322144e-06, "loss": 0.1191, "step": 23441 }, { "epoch": 145.6043613707165, "grad_norm": 0.94140625, "learning_rate": 8.749845895110162e-06, "loss": 0.1146, "step": 23442 }, { "epoch": 145.61059190031153, "grad_norm": 1.4296875, "learning_rate": 8.747973689722933e-06, "loss": 0.1034, "step": 23443 }, { "epoch": 145.61682242990653, "grad_norm": 3.0625, "learning_rate": 8.746101642178622e-06, "loss": 0.3033, "step": 23444 }, { "epoch": 145.62305295950156, "grad_norm": 2.078125, "learning_rate": 8.744229752495425e-06, "loss": 0.1552, "step": 23445 }, { "epoch": 145.62928348909657, "grad_norm": 1.3203125, "learning_rate": 8.742358020691524e-06, "loss": 0.1098, "step": 23446 }, { "epoch": 145.6355140186916, "grad_norm": 2.34375, "learning_rate": 8.74048644678509e-06, "loss": 0.1649, "step": 23447 }, { "epoch": 145.6417445482866, "grad_norm": 1.34375, "learning_rate": 8.738615030794297e-06, "loss": 0.1014, "step": 23448 }, { "epoch": 145.64797507788163, "grad_norm": 1.546875, "learning_rate": 8.736743772737324e-06, "loss": 0.1089, "step": 23449 }, { "epoch": 145.65420560747663, "grad_norm": 2.34375, "learning_rate": 8.734872672632355e-06, "loss": 0.2376, "step": 23450 }, { "epoch": 145.66043613707166, "grad_norm": 3.71875, "learning_rate": 8.733001730497545e-06, "loss": 0.3319, "step": 23451 }, { "epoch": 145.66666666666666, "grad_norm": 1.203125, "learning_rate": 8.731130946351083e-06, "loss": 0.1213, "step": 23452 }, { "epoch": 145.6728971962617, "grad_norm": 1.5546875, "learning_rate": 8.72926032021112e-06, "loss": 0.1273, "step": 23453 }, { "epoch": 145.6791277258567, "grad_norm": 1.734375, "learning_rate": 8.727389852095841e-06, "loss": 0.1232, "step": 23454 }, { "epoch": 145.68535825545172, "grad_norm": 2.234375, "learning_rate": 8.725519542023396e-06, "loss": 0.2663, "step": 23455 }, { "epoch": 145.69158878504672, "grad_norm": 2.921875, "learning_rate": 8.723649390011961e-06, "loss": 0.2217, "step": 23456 }, { "epoch": 145.69781931464175, "grad_norm": 1.890625, "learning_rate": 8.721779396079697e-06, "loss": 0.1416, "step": 23457 }, { "epoch": 145.70404984423675, "grad_norm": 1.34375, "learning_rate": 8.719909560244771e-06, "loss": 0.1283, "step": 23458 }, { "epoch": 145.71028037383178, "grad_norm": 1.40625, "learning_rate": 8.718039882525339e-06, "loss": 0.1305, "step": 23459 }, { "epoch": 145.71651090342678, "grad_norm": 2.328125, "learning_rate": 8.71617036293955e-06, "loss": 0.1923, "step": 23460 }, { "epoch": 145.7227414330218, "grad_norm": 0.80078125, "learning_rate": 8.714301001505567e-06, "loss": 0.0887, "step": 23461 }, { "epoch": 145.7289719626168, "grad_norm": 2.59375, "learning_rate": 8.71243179824156e-06, "loss": 0.2864, "step": 23462 }, { "epoch": 145.73520249221184, "grad_norm": 2.78125, "learning_rate": 8.710562753165663e-06, "loss": 0.1644, "step": 23463 }, { "epoch": 145.74143302180684, "grad_norm": 0.984375, "learning_rate": 8.708693866296033e-06, "loss": 0.0984, "step": 23464 }, { "epoch": 145.74766355140187, "grad_norm": 1.171875, "learning_rate": 8.706825137650837e-06, "loss": 0.0994, "step": 23465 }, { "epoch": 145.75389408099687, "grad_norm": 2.9375, "learning_rate": 8.70495656724821e-06, "loss": 0.2602, "step": 23466 }, { "epoch": 145.7601246105919, "grad_norm": 3.875, "learning_rate": 8.703088155106295e-06, "loss": 0.1307, "step": 23467 }, { "epoch": 145.7663551401869, "grad_norm": 1.21875, "learning_rate": 8.701219901243247e-06, "loss": 0.1021, "step": 23468 }, { "epoch": 145.77258566978193, "grad_norm": 2.140625, "learning_rate": 8.699351805677213e-06, "loss": 0.0925, "step": 23469 }, { "epoch": 145.77881619937693, "grad_norm": 1.3203125, "learning_rate": 8.697483868426329e-06, "loss": 0.1245, "step": 23470 }, { "epoch": 145.78504672897196, "grad_norm": 2.15625, "learning_rate": 8.695616089508745e-06, "loss": 0.2441, "step": 23471 }, { "epoch": 145.79127725856696, "grad_norm": 2.421875, "learning_rate": 8.693748468942592e-06, "loss": 0.2451, "step": 23472 }, { "epoch": 145.797507788162, "grad_norm": 2.09375, "learning_rate": 8.69188100674602e-06, "loss": 0.1802, "step": 23473 }, { "epoch": 145.80373831775702, "grad_norm": 2.5625, "learning_rate": 8.69001370293715e-06, "loss": 0.2462, "step": 23474 }, { "epoch": 145.80996884735202, "grad_norm": 3.375, "learning_rate": 8.68814655753413e-06, "loss": 0.3614, "step": 23475 }, { "epoch": 145.81619937694705, "grad_norm": 2.796875, "learning_rate": 8.686279570555097e-06, "loss": 0.2871, "step": 23476 }, { "epoch": 145.82242990654206, "grad_norm": 1.8828125, "learning_rate": 8.684412742018172e-06, "loss": 0.2089, "step": 23477 }, { "epoch": 145.82866043613708, "grad_norm": 1.9921875, "learning_rate": 8.682546071941498e-06, "loss": 0.2313, "step": 23478 }, { "epoch": 145.8348909657321, "grad_norm": 1.359375, "learning_rate": 8.680679560343192e-06, "loss": 0.1173, "step": 23479 }, { "epoch": 145.84112149532712, "grad_norm": 1.9375, "learning_rate": 8.678813207241393e-06, "loss": 0.1406, "step": 23480 }, { "epoch": 145.84735202492212, "grad_norm": 3.1875, "learning_rate": 8.676947012654218e-06, "loss": 0.2215, "step": 23481 }, { "epoch": 145.85358255451715, "grad_norm": 2.625, "learning_rate": 8.675080976599794e-06, "loss": 0.1905, "step": 23482 }, { "epoch": 145.85981308411215, "grad_norm": 1.9140625, "learning_rate": 8.673215099096258e-06, "loss": 0.1258, "step": 23483 }, { "epoch": 145.86604361370718, "grad_norm": 1.984375, "learning_rate": 8.67134938016172e-06, "loss": 0.14, "step": 23484 }, { "epoch": 145.87227414330218, "grad_norm": 2.453125, "learning_rate": 8.669483819814291e-06, "loss": 0.2483, "step": 23485 }, { "epoch": 145.8785046728972, "grad_norm": 3.0, "learning_rate": 8.667618418072099e-06, "loss": 0.1643, "step": 23486 }, { "epoch": 145.8847352024922, "grad_norm": 2.671875, "learning_rate": 8.665753174953274e-06, "loss": 0.2651, "step": 23487 }, { "epoch": 145.89096573208724, "grad_norm": 2.265625, "learning_rate": 8.663888090475905e-06, "loss": 0.195, "step": 23488 }, { "epoch": 145.89719626168224, "grad_norm": 1.4375, "learning_rate": 8.662023164658125e-06, "loss": 0.083, "step": 23489 }, { "epoch": 145.90342679127727, "grad_norm": 1.359375, "learning_rate": 8.660158397518048e-06, "loss": 0.1496, "step": 23490 }, { "epoch": 145.90965732087227, "grad_norm": 1.8203125, "learning_rate": 8.658293789073782e-06, "loss": 0.0969, "step": 23491 }, { "epoch": 145.9158878504673, "grad_norm": 0.86328125, "learning_rate": 8.656429339343422e-06, "loss": 0.0902, "step": 23492 }, { "epoch": 145.9221183800623, "grad_norm": 1.1328125, "learning_rate": 8.654565048345088e-06, "loss": 0.096, "step": 23493 }, { "epoch": 145.92834890965733, "grad_norm": 1.921875, "learning_rate": 8.652700916096884e-06, "loss": 0.1246, "step": 23494 }, { "epoch": 145.93457943925233, "grad_norm": 2.5625, "learning_rate": 8.650836942616928e-06, "loss": 0.1858, "step": 23495 }, { "epoch": 145.94080996884736, "grad_norm": 1.3515625, "learning_rate": 8.6489731279233e-06, "loss": 0.1008, "step": 23496 }, { "epoch": 145.94704049844236, "grad_norm": 2.296875, "learning_rate": 8.647109472034126e-06, "loss": 0.2115, "step": 23497 }, { "epoch": 145.9532710280374, "grad_norm": 1.59375, "learning_rate": 8.645245974967484e-06, "loss": 0.125, "step": 23498 }, { "epoch": 145.9595015576324, "grad_norm": 1.7578125, "learning_rate": 8.643382636741487e-06, "loss": 0.1116, "step": 23499 }, { "epoch": 145.96573208722742, "grad_norm": 1.046875, "learning_rate": 8.641519457374222e-06, "loss": 0.0836, "step": 23500 }, { "epoch": 145.97196261682242, "grad_norm": 1.015625, "learning_rate": 8.639656436883791e-06, "loss": 0.0821, "step": 23501 }, { "epoch": 145.97819314641745, "grad_norm": 3.234375, "learning_rate": 8.637793575288294e-06, "loss": 0.3164, "step": 23502 }, { "epoch": 145.98442367601245, "grad_norm": 1.3046875, "learning_rate": 8.635930872605819e-06, "loss": 0.1078, "step": 23503 }, { "epoch": 145.99065420560748, "grad_norm": 2.71875, "learning_rate": 8.634068328854441e-06, "loss": 0.2904, "step": 23504 }, { "epoch": 145.99688473520249, "grad_norm": 1.640625, "learning_rate": 8.63220594405227e-06, "loss": 0.1504, "step": 23505 }, { "epoch": 146.0, "grad_norm": 2.515625, "learning_rate": 8.630343718217388e-06, "loss": 0.1555, "step": 23506 }, { "epoch": 146.00623052959503, "grad_norm": 1.4375, "learning_rate": 8.628481651367876e-06, "loss": 0.1143, "step": 23507 }, { "epoch": 146.01246105919003, "grad_norm": 1.3515625, "learning_rate": 8.62661974352182e-06, "loss": 0.1123, "step": 23508 }, { "epoch": 146.01869158878506, "grad_norm": 1.7734375, "learning_rate": 8.624757994697316e-06, "loss": 0.1696, "step": 23509 }, { "epoch": 146.02492211838006, "grad_norm": 2.4375, "learning_rate": 8.622896404912436e-06, "loss": 0.1575, "step": 23510 }, { "epoch": 146.0311526479751, "grad_norm": 1.3046875, "learning_rate": 8.62103497418525e-06, "loss": 0.1153, "step": 23511 }, { "epoch": 146.0373831775701, "grad_norm": 0.9765625, "learning_rate": 8.619173702533848e-06, "loss": 0.1073, "step": 23512 }, { "epoch": 146.04361370716512, "grad_norm": 2.625, "learning_rate": 8.617312589976313e-06, "loss": 0.1546, "step": 23513 }, { "epoch": 146.04984423676012, "grad_norm": 2.21875, "learning_rate": 8.615451636530705e-06, "loss": 0.1977, "step": 23514 }, { "epoch": 146.05607476635515, "grad_norm": 2.375, "learning_rate": 8.613590842215104e-06, "loss": 0.1735, "step": 23515 }, { "epoch": 146.06230529595015, "grad_norm": 1.625, "learning_rate": 8.611730207047597e-06, "loss": 0.0966, "step": 23516 }, { "epoch": 146.06853582554518, "grad_norm": 3.828125, "learning_rate": 8.609869731046238e-06, "loss": 0.1427, "step": 23517 }, { "epoch": 146.07476635514018, "grad_norm": 1.6328125, "learning_rate": 8.608009414229095e-06, "loss": 0.148, "step": 23518 }, { "epoch": 146.0809968847352, "grad_norm": 1.9375, "learning_rate": 8.60614925661424e-06, "loss": 0.1453, "step": 23519 }, { "epoch": 146.08722741433021, "grad_norm": 2.0625, "learning_rate": 8.604289258219739e-06, "loss": 0.1318, "step": 23520 }, { "epoch": 146.09345794392524, "grad_norm": 2.40625, "learning_rate": 8.602429419063667e-06, "loss": 0.1523, "step": 23521 }, { "epoch": 146.09968847352025, "grad_norm": 1.59375, "learning_rate": 8.60056973916408e-06, "loss": 0.1046, "step": 23522 }, { "epoch": 146.10591900311528, "grad_norm": 1.9765625, "learning_rate": 8.59871021853903e-06, "loss": 0.1324, "step": 23523 }, { "epoch": 146.11214953271028, "grad_norm": 2.234375, "learning_rate": 8.596850857206584e-06, "loss": 0.2309, "step": 23524 }, { "epoch": 146.1183800623053, "grad_norm": 2.28125, "learning_rate": 8.594991655184808e-06, "loss": 0.1647, "step": 23525 }, { "epoch": 146.1246105919003, "grad_norm": 1.1484375, "learning_rate": 8.593132612491747e-06, "loss": 0.1153, "step": 23526 }, { "epoch": 146.13084112149534, "grad_norm": 2.640625, "learning_rate": 8.591273729145461e-06, "loss": 0.1485, "step": 23527 }, { "epoch": 146.13707165109034, "grad_norm": 1.671875, "learning_rate": 8.589415005164011e-06, "loss": 0.1579, "step": 23528 }, { "epoch": 146.14330218068537, "grad_norm": 1.7265625, "learning_rate": 8.587556440565445e-06, "loss": 0.1303, "step": 23529 }, { "epoch": 146.14953271028037, "grad_norm": 1.5703125, "learning_rate": 8.5856980353678e-06, "loss": 0.135, "step": 23530 }, { "epoch": 146.1557632398754, "grad_norm": 2.1875, "learning_rate": 8.58383978958914e-06, "loss": 0.2236, "step": 23531 }, { "epoch": 146.1619937694704, "grad_norm": 2.9375, "learning_rate": 8.581981703247519e-06, "loss": 0.1604, "step": 23532 }, { "epoch": 146.16822429906543, "grad_norm": 2.3125, "learning_rate": 8.580123776360962e-06, "loss": 0.2466, "step": 23533 }, { "epoch": 146.17445482866043, "grad_norm": 2.21875, "learning_rate": 8.578266008947525e-06, "loss": 0.1891, "step": 23534 }, { "epoch": 146.18068535825546, "grad_norm": 1.5390625, "learning_rate": 8.576408401025262e-06, "loss": 0.1301, "step": 23535 }, { "epoch": 146.18691588785046, "grad_norm": 2.21875, "learning_rate": 8.574550952612202e-06, "loss": 0.1723, "step": 23536 }, { "epoch": 146.1931464174455, "grad_norm": 2.1875, "learning_rate": 8.572693663726377e-06, "loss": 0.0987, "step": 23537 }, { "epoch": 146.1993769470405, "grad_norm": 1.8515625, "learning_rate": 8.570836534385837e-06, "loss": 0.1106, "step": 23538 }, { "epoch": 146.20560747663552, "grad_norm": 2.328125, "learning_rate": 8.568979564608625e-06, "loss": 0.2454, "step": 23539 }, { "epoch": 146.21183800623052, "grad_norm": 1.34375, "learning_rate": 8.567122754412759e-06, "loss": 0.13, "step": 23540 }, { "epoch": 146.21806853582555, "grad_norm": 2.328125, "learning_rate": 8.565266103816291e-06, "loss": 0.1805, "step": 23541 }, { "epoch": 146.22429906542055, "grad_norm": 2.90625, "learning_rate": 8.563409612837236e-06, "loss": 0.2696, "step": 23542 }, { "epoch": 146.23052959501558, "grad_norm": 2.078125, "learning_rate": 8.561553281493639e-06, "loss": 0.1625, "step": 23543 }, { "epoch": 146.23676012461058, "grad_norm": 2.75, "learning_rate": 8.559697109803514e-06, "loss": 0.2281, "step": 23544 }, { "epoch": 146.2429906542056, "grad_norm": 1.3828125, "learning_rate": 8.557841097784899e-06, "loss": 0.1214, "step": 23545 }, { "epoch": 146.2492211838006, "grad_norm": 1.9296875, "learning_rate": 8.555985245455824e-06, "loss": 0.1568, "step": 23546 }, { "epoch": 146.25545171339564, "grad_norm": 1.328125, "learning_rate": 8.5541295528343e-06, "loss": 0.1055, "step": 23547 }, { "epoch": 146.26168224299064, "grad_norm": 2.484375, "learning_rate": 8.552274019938366e-06, "loss": 0.1716, "step": 23548 }, { "epoch": 146.26791277258567, "grad_norm": 2.125, "learning_rate": 8.550418646786027e-06, "loss": 0.1463, "step": 23549 }, { "epoch": 146.27414330218068, "grad_norm": 1.5078125, "learning_rate": 8.548563433395316e-06, "loss": 0.1229, "step": 23550 }, { "epoch": 146.2803738317757, "grad_norm": 1.140625, "learning_rate": 8.546708379784238e-06, "loss": 0.1167, "step": 23551 }, { "epoch": 146.2866043613707, "grad_norm": 0.8984375, "learning_rate": 8.544853485970818e-06, "loss": 0.084, "step": 23552 }, { "epoch": 146.29283489096574, "grad_norm": 2.390625, "learning_rate": 8.542998751973073e-06, "loss": 0.2009, "step": 23553 }, { "epoch": 146.29906542056074, "grad_norm": 2.78125, "learning_rate": 8.541144177809026e-06, "loss": 0.2789, "step": 23554 }, { "epoch": 146.30529595015577, "grad_norm": 1.390625, "learning_rate": 8.53928976349666e-06, "loss": 0.0973, "step": 23555 }, { "epoch": 146.31152647975077, "grad_norm": 3.109375, "learning_rate": 8.537435509054006e-06, "loss": 0.1993, "step": 23556 }, { "epoch": 146.3177570093458, "grad_norm": 2.484375, "learning_rate": 8.535581414499066e-06, "loss": 0.2202, "step": 23557 }, { "epoch": 146.3239875389408, "grad_norm": 2.515625, "learning_rate": 8.53372747984986e-06, "loss": 0.1015, "step": 23558 }, { "epoch": 146.33021806853583, "grad_norm": 2.0, "learning_rate": 8.531873705124378e-06, "loss": 0.1676, "step": 23559 }, { "epoch": 146.33644859813083, "grad_norm": 2.5625, "learning_rate": 8.530020090340635e-06, "loss": 0.2709, "step": 23560 }, { "epoch": 146.34267912772586, "grad_norm": 1.2421875, "learning_rate": 8.52816663551662e-06, "loss": 0.1151, "step": 23561 }, { "epoch": 146.34890965732086, "grad_norm": 2.546875, "learning_rate": 8.526313340670355e-06, "loss": 0.1078, "step": 23562 }, { "epoch": 146.3551401869159, "grad_norm": 1.15625, "learning_rate": 8.524460205819821e-06, "loss": 0.0974, "step": 23563 }, { "epoch": 146.3613707165109, "grad_norm": 1.65625, "learning_rate": 8.522607230983021e-06, "loss": 0.1181, "step": 23564 }, { "epoch": 146.36760124610592, "grad_norm": 1.3984375, "learning_rate": 8.520754416177959e-06, "loss": 0.1207, "step": 23565 }, { "epoch": 146.37383177570092, "grad_norm": 2.15625, "learning_rate": 8.518901761422618e-06, "loss": 0.1658, "step": 23566 }, { "epoch": 146.38006230529595, "grad_norm": 1.890625, "learning_rate": 8.517049266735007e-06, "loss": 0.1862, "step": 23567 }, { "epoch": 146.38629283489095, "grad_norm": 2.109375, "learning_rate": 8.515196932133099e-06, "loss": 0.1382, "step": 23568 }, { "epoch": 146.39252336448598, "grad_norm": 2.84375, "learning_rate": 8.513344757634904e-06, "loss": 0.2101, "step": 23569 }, { "epoch": 146.39875389408098, "grad_norm": 2.671875, "learning_rate": 8.51149274325839e-06, "loss": 0.2034, "step": 23570 }, { "epoch": 146.404984423676, "grad_norm": 3.8125, "learning_rate": 8.509640889021553e-06, "loss": 0.2393, "step": 23571 }, { "epoch": 146.41121495327104, "grad_norm": 0.921875, "learning_rate": 8.50778919494239e-06, "loss": 0.0933, "step": 23572 }, { "epoch": 146.41744548286604, "grad_norm": 2.359375, "learning_rate": 8.505937661038877e-06, "loss": 0.1248, "step": 23573 }, { "epoch": 146.42367601246107, "grad_norm": 2.03125, "learning_rate": 8.504086287328982e-06, "loss": 0.1897, "step": 23574 }, { "epoch": 146.42990654205607, "grad_norm": 1.46875, "learning_rate": 8.5022350738307e-06, "loss": 0.1036, "step": 23575 }, { "epoch": 146.4361370716511, "grad_norm": 2.765625, "learning_rate": 8.500384020562018e-06, "loss": 0.2812, "step": 23576 }, { "epoch": 146.4423676012461, "grad_norm": 1.375, "learning_rate": 8.498533127540897e-06, "loss": 0.1225, "step": 23577 }, { "epoch": 146.44859813084113, "grad_norm": 3.234375, "learning_rate": 8.496682394785318e-06, "loss": 0.2562, "step": 23578 }, { "epoch": 146.45482866043614, "grad_norm": 1.6484375, "learning_rate": 8.49483182231327e-06, "loss": 0.1367, "step": 23579 }, { "epoch": 146.46105919003116, "grad_norm": 1.2734375, "learning_rate": 8.492981410142714e-06, "loss": 0.1318, "step": 23580 }, { "epoch": 146.46728971962617, "grad_norm": 1.0234375, "learning_rate": 8.491131158291612e-06, "loss": 0.1068, "step": 23581 }, { "epoch": 146.4735202492212, "grad_norm": 3.28125, "learning_rate": 8.489281066777946e-06, "loss": 0.3918, "step": 23582 }, { "epoch": 146.4797507788162, "grad_norm": 1.75, "learning_rate": 8.487431135619691e-06, "loss": 0.0962, "step": 23583 }, { "epoch": 146.48598130841123, "grad_norm": 0.98828125, "learning_rate": 8.485581364834797e-06, "loss": 0.089, "step": 23584 }, { "epoch": 146.49221183800623, "grad_norm": 1.75, "learning_rate": 8.483731754441238e-06, "loss": 0.122, "step": 23585 }, { "epoch": 146.49844236760126, "grad_norm": 1.828125, "learning_rate": 8.481882304456987e-06, "loss": 0.168, "step": 23586 }, { "epoch": 146.50467289719626, "grad_norm": 2.484375, "learning_rate": 8.480033014899993e-06, "loss": 0.2314, "step": 23587 }, { "epoch": 146.5109034267913, "grad_norm": 0.88671875, "learning_rate": 8.478183885788216e-06, "loss": 0.092, "step": 23588 }, { "epoch": 146.5171339563863, "grad_norm": 3.109375, "learning_rate": 8.47633491713962e-06, "loss": 0.3034, "step": 23589 }, { "epoch": 146.52336448598132, "grad_norm": 1.9140625, "learning_rate": 8.47448610897216e-06, "loss": 0.1359, "step": 23590 }, { "epoch": 146.52959501557632, "grad_norm": 1.0703125, "learning_rate": 8.472637461303803e-06, "loss": 0.1055, "step": 23591 }, { "epoch": 146.53582554517135, "grad_norm": 2.09375, "learning_rate": 8.470788974152497e-06, "loss": 0.2668, "step": 23592 }, { "epoch": 146.54205607476635, "grad_norm": 1.5234375, "learning_rate": 8.468940647536184e-06, "loss": 0.1347, "step": 23593 }, { "epoch": 146.54828660436138, "grad_norm": 1.0078125, "learning_rate": 8.467092481472821e-06, "loss": 0.1005, "step": 23594 }, { "epoch": 146.55451713395638, "grad_norm": 1.5703125, "learning_rate": 8.465244475980372e-06, "loss": 0.125, "step": 23595 }, { "epoch": 146.5607476635514, "grad_norm": 1.609375, "learning_rate": 8.463396631076764e-06, "loss": 0.1133, "step": 23596 }, { "epoch": 146.5669781931464, "grad_norm": 1.6015625, "learning_rate": 8.461548946779954e-06, "loss": 0.1473, "step": 23597 }, { "epoch": 146.57320872274144, "grad_norm": 1.5234375, "learning_rate": 8.459701423107893e-06, "loss": 0.1285, "step": 23598 }, { "epoch": 146.57943925233644, "grad_norm": 1.234375, "learning_rate": 8.45785406007852e-06, "loss": 0.1236, "step": 23599 }, { "epoch": 146.58566978193147, "grad_norm": 1.1875, "learning_rate": 8.456006857709767e-06, "loss": 0.1077, "step": 23600 }, { "epoch": 146.59190031152647, "grad_norm": 1.59375, "learning_rate": 8.454159816019584e-06, "loss": 0.1411, "step": 23601 }, { "epoch": 146.5981308411215, "grad_norm": 1.421875, "learning_rate": 8.452312935025913e-06, "loss": 0.1137, "step": 23602 }, { "epoch": 146.6043613707165, "grad_norm": 1.5234375, "learning_rate": 8.45046621474668e-06, "loss": 0.1178, "step": 23603 }, { "epoch": 146.61059190031153, "grad_norm": 2.046875, "learning_rate": 8.448619655199827e-06, "loss": 0.1526, "step": 23604 }, { "epoch": 146.61682242990653, "grad_norm": 2.390625, "learning_rate": 8.446773256403298e-06, "loss": 0.1897, "step": 23605 }, { "epoch": 146.62305295950156, "grad_norm": 2.390625, "learning_rate": 8.444927018375014e-06, "loss": 0.1801, "step": 23606 }, { "epoch": 146.62928348909657, "grad_norm": 1.6875, "learning_rate": 8.443080941132903e-06, "loss": 0.152, "step": 23607 }, { "epoch": 146.6355140186916, "grad_norm": 1.65625, "learning_rate": 8.441235024694895e-06, "loss": 0.1206, "step": 23608 }, { "epoch": 146.6417445482866, "grad_norm": 1.0859375, "learning_rate": 8.439389269078934e-06, "loss": 0.1059, "step": 23609 }, { "epoch": 146.64797507788163, "grad_norm": 2.296875, "learning_rate": 8.437543674302929e-06, "loss": 0.1396, "step": 23610 }, { "epoch": 146.65420560747663, "grad_norm": 2.390625, "learning_rate": 8.435698240384812e-06, "loss": 0.1562, "step": 23611 }, { "epoch": 146.66043613707166, "grad_norm": 2.375, "learning_rate": 8.433852967342501e-06, "loss": 0.1804, "step": 23612 }, { "epoch": 146.66666666666666, "grad_norm": 2.46875, "learning_rate": 8.43200785519393e-06, "loss": 0.187, "step": 23613 }, { "epoch": 146.6728971962617, "grad_norm": 1.7109375, "learning_rate": 8.430162903957e-06, "loss": 0.1674, "step": 23614 }, { "epoch": 146.6791277258567, "grad_norm": 2.65625, "learning_rate": 8.42831811364964e-06, "loss": 0.1902, "step": 23615 }, { "epoch": 146.68535825545172, "grad_norm": 1.4140625, "learning_rate": 8.42647348428977e-06, "loss": 0.0979, "step": 23616 }, { "epoch": 146.69158878504672, "grad_norm": 1.40625, "learning_rate": 8.42462901589531e-06, "loss": 0.1208, "step": 23617 }, { "epoch": 146.69781931464175, "grad_norm": 2.078125, "learning_rate": 8.422784708484166e-06, "loss": 0.1767, "step": 23618 }, { "epoch": 146.70404984423675, "grad_norm": 2.4375, "learning_rate": 8.420940562074242e-06, "loss": 0.2163, "step": 23619 }, { "epoch": 146.71028037383178, "grad_norm": 1.8671875, "learning_rate": 8.419096576683457e-06, "loss": 0.1697, "step": 23620 }, { "epoch": 146.71651090342678, "grad_norm": 1.6796875, "learning_rate": 8.41725275232973e-06, "loss": 0.1511, "step": 23621 }, { "epoch": 146.7227414330218, "grad_norm": 2.25, "learning_rate": 8.415409089030949e-06, "loss": 0.1748, "step": 23622 }, { "epoch": 146.7289719626168, "grad_norm": 1.9140625, "learning_rate": 8.41356558680503e-06, "loss": 0.1611, "step": 23623 }, { "epoch": 146.73520249221184, "grad_norm": 1.3671875, "learning_rate": 8.41172224566989e-06, "loss": 0.1172, "step": 23624 }, { "epoch": 146.74143302180684, "grad_norm": 1.8515625, "learning_rate": 8.409879065643413e-06, "loss": 0.1651, "step": 23625 }, { "epoch": 146.74766355140187, "grad_norm": 2.453125, "learning_rate": 8.408036046743503e-06, "loss": 0.1343, "step": 23626 }, { "epoch": 146.75389408099687, "grad_norm": 1.4765625, "learning_rate": 8.406193188988062e-06, "loss": 0.1349, "step": 23627 }, { "epoch": 146.7601246105919, "grad_norm": 2.0625, "learning_rate": 8.404350492395e-06, "loss": 0.1369, "step": 23628 }, { "epoch": 146.7663551401869, "grad_norm": 1.140625, "learning_rate": 8.402507956982195e-06, "loss": 0.1024, "step": 23629 }, { "epoch": 146.77258566978193, "grad_norm": 1.546875, "learning_rate": 8.400665582767556e-06, "loss": 0.1094, "step": 23630 }, { "epoch": 146.77881619937693, "grad_norm": 2.5, "learning_rate": 8.398823369768965e-06, "loss": 0.1884, "step": 23631 }, { "epoch": 146.78504672897196, "grad_norm": 1.8046875, "learning_rate": 8.396981318004326e-06, "loss": 0.1136, "step": 23632 }, { "epoch": 146.79127725856696, "grad_norm": 1.7890625, "learning_rate": 8.395139427491517e-06, "loss": 0.1277, "step": 23633 }, { "epoch": 146.797507788162, "grad_norm": 1.25, "learning_rate": 8.393297698248434e-06, "loss": 0.0871, "step": 23634 }, { "epoch": 146.80373831775702, "grad_norm": 3.21875, "learning_rate": 8.391456130292968e-06, "loss": 0.1712, "step": 23635 }, { "epoch": 146.80996884735202, "grad_norm": 0.99609375, "learning_rate": 8.389614723642993e-06, "loss": 0.1095, "step": 23636 }, { "epoch": 146.81619937694705, "grad_norm": 3.421875, "learning_rate": 8.387773478316413e-06, "loss": 0.2388, "step": 23637 }, { "epoch": 146.82242990654206, "grad_norm": 1.1328125, "learning_rate": 8.385932394331084e-06, "loss": 0.0898, "step": 23638 }, { "epoch": 146.82866043613708, "grad_norm": 2.65625, "learning_rate": 8.38409147170491e-06, "loss": 0.1616, "step": 23639 }, { "epoch": 146.8348909657321, "grad_norm": 2.90625, "learning_rate": 8.382250710455755e-06, "loss": 0.243, "step": 23640 }, { "epoch": 146.84112149532712, "grad_norm": 1.78125, "learning_rate": 8.380410110601503e-06, "loss": 0.1229, "step": 23641 }, { "epoch": 146.84735202492212, "grad_norm": 2.125, "learning_rate": 8.378569672160036e-06, "loss": 0.2189, "step": 23642 }, { "epoch": 146.85358255451715, "grad_norm": 3.046875, "learning_rate": 8.376729395149225e-06, "loss": 0.171, "step": 23643 }, { "epoch": 146.85981308411215, "grad_norm": 1.4375, "learning_rate": 8.374889279586931e-06, "loss": 0.1382, "step": 23644 }, { "epoch": 146.86604361370718, "grad_norm": 1.21875, "learning_rate": 8.373049325491036e-06, "loss": 0.1127, "step": 23645 }, { "epoch": 146.87227414330218, "grad_norm": 1.0, "learning_rate": 8.371209532879418e-06, "loss": 0.0773, "step": 23646 }, { "epoch": 146.8785046728972, "grad_norm": 2.1875, "learning_rate": 8.369369901769927e-06, "loss": 0.1426, "step": 23647 }, { "epoch": 146.8847352024922, "grad_norm": 2.21875, "learning_rate": 8.367530432180443e-06, "loss": 0.2306, "step": 23648 }, { "epoch": 146.89096573208724, "grad_norm": 2.78125, "learning_rate": 8.365691124128827e-06, "loss": 0.2329, "step": 23649 }, { "epoch": 146.89719626168224, "grad_norm": 3.703125, "learning_rate": 8.363851977632962e-06, "loss": 0.2452, "step": 23650 }, { "epoch": 146.90342679127727, "grad_norm": 1.5234375, "learning_rate": 8.362012992710674e-06, "loss": 0.1492, "step": 23651 }, { "epoch": 146.90965732087227, "grad_norm": 1.4609375, "learning_rate": 8.360174169379842e-06, "loss": 0.1178, "step": 23652 }, { "epoch": 146.9158878504673, "grad_norm": 2.109375, "learning_rate": 8.358335507658324e-06, "loss": 0.1852, "step": 23653 }, { "epoch": 146.9221183800623, "grad_norm": 2.859375, "learning_rate": 8.356497007563988e-06, "loss": 0.1683, "step": 23654 }, { "epoch": 146.92834890965733, "grad_norm": 3.1875, "learning_rate": 8.354658669114668e-06, "loss": 0.2822, "step": 23655 }, { "epoch": 146.93457943925233, "grad_norm": 2.625, "learning_rate": 8.352820492328241e-06, "loss": 0.2219, "step": 23656 }, { "epoch": 146.94080996884736, "grad_norm": 2.171875, "learning_rate": 8.350982477222544e-06, "loss": 0.1622, "step": 23657 }, { "epoch": 146.94704049844236, "grad_norm": 1.9921875, "learning_rate": 8.349144623815435e-06, "loss": 0.1707, "step": 23658 }, { "epoch": 146.9532710280374, "grad_norm": 3.390625, "learning_rate": 8.34730693212476e-06, "loss": 0.2657, "step": 23659 }, { "epoch": 146.9595015576324, "grad_norm": 2.421875, "learning_rate": 8.345469402168365e-06, "loss": 0.1572, "step": 23660 }, { "epoch": 146.96573208722742, "grad_norm": 1.9453125, "learning_rate": 8.34363203396411e-06, "loss": 0.1447, "step": 23661 }, { "epoch": 146.97196261682242, "grad_norm": 2.734375, "learning_rate": 8.341794827529831e-06, "loss": 0.2902, "step": 23662 }, { "epoch": 146.97819314641745, "grad_norm": 1.4453125, "learning_rate": 8.33995778288336e-06, "loss": 0.1259, "step": 23663 }, { "epoch": 146.98442367601245, "grad_norm": 1.921875, "learning_rate": 8.338120900042551e-06, "loss": 0.1781, "step": 23664 }, { "epoch": 146.99065420560748, "grad_norm": 2.125, "learning_rate": 8.33628417902525e-06, "loss": 0.1545, "step": 23665 }, { "epoch": 146.99688473520249, "grad_norm": 1.3046875, "learning_rate": 8.334447619849281e-06, "loss": 0.1043, "step": 23666 }, { "epoch": 147.0, "grad_norm": 0.7578125, "learning_rate": 8.332611222532485e-06, "loss": 0.0609, "step": 23667 }, { "epoch": 147.00623052959503, "grad_norm": 1.4921875, "learning_rate": 8.330774987092712e-06, "loss": 0.1412, "step": 23668 }, { "epoch": 147.01246105919003, "grad_norm": 1.1796875, "learning_rate": 8.328938913547781e-06, "loss": 0.1237, "step": 23669 }, { "epoch": 147.01869158878506, "grad_norm": 1.28125, "learning_rate": 8.327103001915523e-06, "loss": 0.1092, "step": 23670 }, { "epoch": 147.02492211838006, "grad_norm": 2.4375, "learning_rate": 8.325267252213773e-06, "loss": 0.192, "step": 23671 }, { "epoch": 147.0311526479751, "grad_norm": 2.84375, "learning_rate": 8.323431664460368e-06, "loss": 0.1675, "step": 23672 }, { "epoch": 147.0373831775701, "grad_norm": 0.99609375, "learning_rate": 8.321596238673121e-06, "loss": 0.1091, "step": 23673 }, { "epoch": 147.04361370716512, "grad_norm": 1.25, "learning_rate": 8.319760974869863e-06, "loss": 0.1008, "step": 23674 }, { "epoch": 147.04984423676012, "grad_norm": 0.75, "learning_rate": 8.317925873068433e-06, "loss": 0.0956, "step": 23675 }, { "epoch": 147.05607476635515, "grad_norm": 1.6953125, "learning_rate": 8.316090933286638e-06, "loss": 0.1459, "step": 23676 }, { "epoch": 147.06230529595015, "grad_norm": 1.25, "learning_rate": 8.314256155542297e-06, "loss": 0.1207, "step": 23677 }, { "epoch": 147.06853582554518, "grad_norm": 1.8125, "learning_rate": 8.312421539853233e-06, "loss": 0.1339, "step": 23678 }, { "epoch": 147.07476635514018, "grad_norm": 1.1796875, "learning_rate": 8.310587086237276e-06, "loss": 0.1092, "step": 23679 }, { "epoch": 147.0809968847352, "grad_norm": 2.421875, "learning_rate": 8.308752794712226e-06, "loss": 0.2915, "step": 23680 }, { "epoch": 147.08722741433021, "grad_norm": 0.9609375, "learning_rate": 8.306918665295912e-06, "loss": 0.0985, "step": 23681 }, { "epoch": 147.09345794392524, "grad_norm": 1.1796875, "learning_rate": 8.305084698006136e-06, "loss": 0.1007, "step": 23682 }, { "epoch": 147.09968847352025, "grad_norm": 1.1171875, "learning_rate": 8.30325089286072e-06, "loss": 0.1225, "step": 23683 }, { "epoch": 147.10591900311528, "grad_norm": 1.2734375, "learning_rate": 8.30141724987746e-06, "loss": 0.1208, "step": 23684 }, { "epoch": 147.11214953271028, "grad_norm": 2.90625, "learning_rate": 8.299583769074176e-06, "loss": 0.2296, "step": 23685 }, { "epoch": 147.1183800623053, "grad_norm": 1.734375, "learning_rate": 8.297750450468673e-06, "loss": 0.1356, "step": 23686 }, { "epoch": 147.1246105919003, "grad_norm": 3.015625, "learning_rate": 8.295917294078761e-06, "loss": 0.2547, "step": 23687 }, { "epoch": 147.13084112149534, "grad_norm": 2.03125, "learning_rate": 8.294084299922243e-06, "loss": 0.1781, "step": 23688 }, { "epoch": 147.13707165109034, "grad_norm": 2.875, "learning_rate": 8.292251468016906e-06, "loss": 0.1932, "step": 23689 }, { "epoch": 147.14330218068537, "grad_norm": 2.046875, "learning_rate": 8.290418798380568e-06, "loss": 0.1259, "step": 23690 }, { "epoch": 147.14953271028037, "grad_norm": 2.6875, "learning_rate": 8.288586291031026e-06, "loss": 0.2192, "step": 23691 }, { "epoch": 147.1557632398754, "grad_norm": 2.03125, "learning_rate": 8.286753945986067e-06, "loss": 0.1676, "step": 23692 }, { "epoch": 147.1619937694704, "grad_norm": 2.109375, "learning_rate": 8.284921763263498e-06, "loss": 0.1617, "step": 23693 }, { "epoch": 147.16822429906543, "grad_norm": 1.7265625, "learning_rate": 8.283089742881114e-06, "loss": 0.1242, "step": 23694 }, { "epoch": 147.17445482866043, "grad_norm": 1.4765625, "learning_rate": 8.281257884856708e-06, "loss": 0.1109, "step": 23695 }, { "epoch": 147.18068535825546, "grad_norm": 1.6875, "learning_rate": 8.279426189208057e-06, "loss": 0.1147, "step": 23696 }, { "epoch": 147.18691588785046, "grad_norm": 2.265625, "learning_rate": 8.27759465595296e-06, "loss": 0.1722, "step": 23697 }, { "epoch": 147.1931464174455, "grad_norm": 2.453125, "learning_rate": 8.275763285109217e-06, "loss": 0.2607, "step": 23698 }, { "epoch": 147.1993769470405, "grad_norm": 2.140625, "learning_rate": 8.273932076694593e-06, "loss": 0.2157, "step": 23699 }, { "epoch": 147.20560747663552, "grad_norm": 2.34375, "learning_rate": 8.272101030726895e-06, "loss": 0.2163, "step": 23700 }, { "epoch": 147.21183800623052, "grad_norm": 2.40625, "learning_rate": 8.270270147223888e-06, "loss": 0.1638, "step": 23701 }, { "epoch": 147.21806853582555, "grad_norm": 2.109375, "learning_rate": 8.268439426203368e-06, "loss": 0.1327, "step": 23702 }, { "epoch": 147.22429906542055, "grad_norm": 2.5, "learning_rate": 8.2666088676831e-06, "loss": 0.2768, "step": 23703 }, { "epoch": 147.23052959501558, "grad_norm": 2.15625, "learning_rate": 8.264778471680873e-06, "loss": 0.251, "step": 23704 }, { "epoch": 147.23676012461058, "grad_norm": 1.6640625, "learning_rate": 8.262948238214472e-06, "loss": 0.0938, "step": 23705 }, { "epoch": 147.2429906542056, "grad_norm": 1.625, "learning_rate": 8.261118167301654e-06, "loss": 0.1412, "step": 23706 }, { "epoch": 147.2492211838006, "grad_norm": 2.296875, "learning_rate": 8.25928825896021e-06, "loss": 0.2365, "step": 23707 }, { "epoch": 147.25545171339564, "grad_norm": 2.171875, "learning_rate": 8.257458513207896e-06, "loss": 0.1472, "step": 23708 }, { "epoch": 147.26168224299064, "grad_norm": 2.140625, "learning_rate": 8.2556289300625e-06, "loss": 0.1428, "step": 23709 }, { "epoch": 147.26791277258567, "grad_norm": 1.8125, "learning_rate": 8.253799509541776e-06, "loss": 0.1189, "step": 23710 }, { "epoch": 147.27414330218068, "grad_norm": 3.0625, "learning_rate": 8.251970251663498e-06, "loss": 0.188, "step": 23711 }, { "epoch": 147.2803738317757, "grad_norm": 2.90625, "learning_rate": 8.250141156445435e-06, "loss": 0.2059, "step": 23712 }, { "epoch": 147.2866043613707, "grad_norm": 1.578125, "learning_rate": 8.248312223905363e-06, "loss": 0.1443, "step": 23713 }, { "epoch": 147.29283489096574, "grad_norm": 1.7578125, "learning_rate": 8.246483454061015e-06, "loss": 0.1435, "step": 23714 }, { "epoch": 147.29906542056074, "grad_norm": 2.3125, "learning_rate": 8.244654846930172e-06, "loss": 0.2194, "step": 23715 }, { "epoch": 147.30529595015577, "grad_norm": 2.5, "learning_rate": 8.242826402530588e-06, "loss": 0.2565, "step": 23716 }, { "epoch": 147.31152647975077, "grad_norm": 2.078125, "learning_rate": 8.24099812088003e-06, "loss": 0.2023, "step": 23717 }, { "epoch": 147.3177570093458, "grad_norm": 2.640625, "learning_rate": 8.239170001996243e-06, "loss": 0.2699, "step": 23718 }, { "epoch": 147.3239875389408, "grad_norm": 2.59375, "learning_rate": 8.237342045896987e-06, "loss": 0.2263, "step": 23719 }, { "epoch": 147.33021806853583, "grad_norm": 1.7734375, "learning_rate": 8.235514252600024e-06, "loss": 0.0993, "step": 23720 }, { "epoch": 147.33644859813083, "grad_norm": 1.640625, "learning_rate": 8.233686622123096e-06, "loss": 0.146, "step": 23721 }, { "epoch": 147.34267912772586, "grad_norm": 2.28125, "learning_rate": 8.231859154483949e-06, "loss": 0.1636, "step": 23722 }, { "epoch": 147.34890965732086, "grad_norm": 0.71875, "learning_rate": 8.230031849700334e-06, "loss": 0.0974, "step": 23723 }, { "epoch": 147.3551401869159, "grad_norm": 2.0, "learning_rate": 8.228204707790014e-06, "loss": 0.12, "step": 23724 }, { "epoch": 147.3613707165109, "grad_norm": 3.3125, "learning_rate": 8.226377728770712e-06, "loss": 0.2467, "step": 23725 }, { "epoch": 147.36760124610592, "grad_norm": 2.890625, "learning_rate": 8.224550912660192e-06, "loss": 0.233, "step": 23726 }, { "epoch": 147.37383177570092, "grad_norm": 1.984375, "learning_rate": 8.22272425947618e-06, "loss": 0.1722, "step": 23727 }, { "epoch": 147.38006230529595, "grad_norm": 1.8515625, "learning_rate": 8.22089776923643e-06, "loss": 0.1461, "step": 23728 }, { "epoch": 147.38629283489095, "grad_norm": 1.6171875, "learning_rate": 8.219071441958668e-06, "loss": 0.1709, "step": 23729 }, { "epoch": 147.39252336448598, "grad_norm": 1.6484375, "learning_rate": 8.217245277660637e-06, "loss": 0.0988, "step": 23730 }, { "epoch": 147.39875389408098, "grad_norm": 1.6953125, "learning_rate": 8.215419276360084e-06, "loss": 0.1073, "step": 23731 }, { "epoch": 147.404984423676, "grad_norm": 2.640625, "learning_rate": 8.213593438074732e-06, "loss": 0.1282, "step": 23732 }, { "epoch": 147.41121495327104, "grad_norm": 2.578125, "learning_rate": 8.211767762822309e-06, "loss": 0.2193, "step": 23733 }, { "epoch": 147.41744548286604, "grad_norm": 2.75, "learning_rate": 8.209942250620556e-06, "loss": 0.2311, "step": 23734 }, { "epoch": 147.42367601246107, "grad_norm": 2.015625, "learning_rate": 8.208116901487205e-06, "loss": 0.1324, "step": 23735 }, { "epoch": 147.42990654205607, "grad_norm": 1.4765625, "learning_rate": 8.206291715439974e-06, "loss": 0.1233, "step": 23736 }, { "epoch": 147.4361370716511, "grad_norm": 1.140625, "learning_rate": 8.204466692496593e-06, "loss": 0.1115, "step": 23737 }, { "epoch": 147.4423676012461, "grad_norm": 0.86328125, "learning_rate": 8.202641832674796e-06, "loss": 0.0943, "step": 23738 }, { "epoch": 147.44859813084113, "grad_norm": 2.84375, "learning_rate": 8.200817135992301e-06, "loss": 0.2115, "step": 23739 }, { "epoch": 147.45482866043614, "grad_norm": 1.359375, "learning_rate": 8.198992602466817e-06, "loss": 0.1086, "step": 23740 }, { "epoch": 147.46105919003116, "grad_norm": 0.98046875, "learning_rate": 8.197168232116078e-06, "loss": 0.0985, "step": 23741 }, { "epoch": 147.46728971962617, "grad_norm": 2.3125, "learning_rate": 8.19534402495781e-06, "loss": 0.0957, "step": 23742 }, { "epoch": 147.4735202492212, "grad_norm": 2.78125, "learning_rate": 8.19351998100971e-06, "loss": 0.2045, "step": 23743 }, { "epoch": 147.4797507788162, "grad_norm": 1.9765625, "learning_rate": 8.191696100289503e-06, "loss": 0.1276, "step": 23744 }, { "epoch": 147.48598130841123, "grad_norm": 1.640625, "learning_rate": 8.189872382814912e-06, "loss": 0.1158, "step": 23745 }, { "epoch": 147.49221183800623, "grad_norm": 2.703125, "learning_rate": 8.18804882860364e-06, "loss": 0.111, "step": 23746 }, { "epoch": 147.49844236760126, "grad_norm": 1.7578125, "learning_rate": 8.186225437673389e-06, "loss": 0.1574, "step": 23747 }, { "epoch": 147.50467289719626, "grad_norm": 3.515625, "learning_rate": 8.18440221004188e-06, "loss": 0.2977, "step": 23748 }, { "epoch": 147.5109034267913, "grad_norm": 1.3125, "learning_rate": 8.182579145726817e-06, "loss": 0.1277, "step": 23749 }, { "epoch": 147.5171339563863, "grad_norm": 3.390625, "learning_rate": 8.180756244745916e-06, "loss": 0.2628, "step": 23750 }, { "epoch": 147.52336448598132, "grad_norm": 1.6796875, "learning_rate": 8.178933507116873e-06, "loss": 0.1188, "step": 23751 }, { "epoch": 147.52959501557632, "grad_norm": 1.171875, "learning_rate": 8.177110932857379e-06, "loss": 0.1138, "step": 23752 }, { "epoch": 147.53582554517135, "grad_norm": 3.015625, "learning_rate": 8.17528852198515e-06, "loss": 0.1857, "step": 23753 }, { "epoch": 147.54205607476635, "grad_norm": 2.046875, "learning_rate": 8.173466274517889e-06, "loss": 0.1954, "step": 23754 }, { "epoch": 147.54828660436138, "grad_norm": 1.6015625, "learning_rate": 8.17164419047328e-06, "loss": 0.1375, "step": 23755 }, { "epoch": 147.55451713395638, "grad_norm": 1.15625, "learning_rate": 8.169822269869023e-06, "loss": 0.1134, "step": 23756 }, { "epoch": 147.5607476635514, "grad_norm": 1.765625, "learning_rate": 8.168000512722826e-06, "loss": 0.1285, "step": 23757 }, { "epoch": 147.5669781931464, "grad_norm": 1.9453125, "learning_rate": 8.166178919052376e-06, "loss": 0.2439, "step": 23758 }, { "epoch": 147.57320872274144, "grad_norm": 2.5, "learning_rate": 8.164357488875348e-06, "loss": 0.1215, "step": 23759 }, { "epoch": 147.57943925233644, "grad_norm": 2.78125, "learning_rate": 8.16253622220945e-06, "loss": 0.2246, "step": 23760 }, { "epoch": 147.58566978193147, "grad_norm": 1.2890625, "learning_rate": 8.16071511907237e-06, "loss": 0.1232, "step": 23761 }, { "epoch": 147.59190031152647, "grad_norm": 2.296875, "learning_rate": 8.158894179481785e-06, "loss": 0.1943, "step": 23762 }, { "epoch": 147.5981308411215, "grad_norm": 2.578125, "learning_rate": 8.157073403455387e-06, "loss": 0.2201, "step": 23763 }, { "epoch": 147.6043613707165, "grad_norm": 1.6875, "learning_rate": 8.155252791010867e-06, "loss": 0.1707, "step": 23764 }, { "epoch": 147.61059190031153, "grad_norm": 2.671875, "learning_rate": 8.153432342165898e-06, "loss": 0.118, "step": 23765 }, { "epoch": 147.61682242990653, "grad_norm": 2.09375, "learning_rate": 8.151612056938154e-06, "loss": 0.2178, "step": 23766 }, { "epoch": 147.62305295950156, "grad_norm": 1.484375, "learning_rate": 8.149791935345322e-06, "loss": 0.1405, "step": 23767 }, { "epoch": 147.62928348909657, "grad_norm": 1.4453125, "learning_rate": 8.147971977405087e-06, "loss": 0.0812, "step": 23768 }, { "epoch": 147.6355140186916, "grad_norm": 1.5546875, "learning_rate": 8.146152183135108e-06, "loss": 0.2337, "step": 23769 }, { "epoch": 147.6417445482866, "grad_norm": 1.7421875, "learning_rate": 8.144332552553066e-06, "loss": 0.1274, "step": 23770 }, { "epoch": 147.64797507788163, "grad_norm": 2.015625, "learning_rate": 8.142513085676648e-06, "loss": 0.186, "step": 23771 }, { "epoch": 147.65420560747663, "grad_norm": 1.515625, "learning_rate": 8.14069378252351e-06, "loss": 0.092, "step": 23772 }, { "epoch": 147.66043613707166, "grad_norm": 2.578125, "learning_rate": 8.138874643111317e-06, "loss": 0.1102, "step": 23773 }, { "epoch": 147.66666666666666, "grad_norm": 1.2890625, "learning_rate": 8.137055667457743e-06, "loss": 0.1076, "step": 23774 }, { "epoch": 147.6728971962617, "grad_norm": 1.65625, "learning_rate": 8.135236855580463e-06, "loss": 0.1589, "step": 23775 }, { "epoch": 147.6791277258567, "grad_norm": 1.34375, "learning_rate": 8.133418207497129e-06, "loss": 0.128, "step": 23776 }, { "epoch": 147.68535825545172, "grad_norm": 2.640625, "learning_rate": 8.131599723225413e-06, "loss": 0.1909, "step": 23777 }, { "epoch": 147.69158878504672, "grad_norm": 2.296875, "learning_rate": 8.129781402782965e-06, "loss": 0.1848, "step": 23778 }, { "epoch": 147.69781931464175, "grad_norm": 2.59375, "learning_rate": 8.127963246187464e-06, "loss": 0.2039, "step": 23779 }, { "epoch": 147.70404984423675, "grad_norm": 2.859375, "learning_rate": 8.126145253456544e-06, "loss": 0.276, "step": 23780 }, { "epoch": 147.71028037383178, "grad_norm": 2.265625, "learning_rate": 8.124327424607877e-06, "loss": 0.1443, "step": 23781 }, { "epoch": 147.71651090342678, "grad_norm": 1.6328125, "learning_rate": 8.122509759659114e-06, "loss": 0.1259, "step": 23782 }, { "epoch": 147.7227414330218, "grad_norm": 3.109375, "learning_rate": 8.120692258627927e-06, "loss": 0.3119, "step": 23783 }, { "epoch": 147.7289719626168, "grad_norm": 1.875, "learning_rate": 8.118874921531936e-06, "loss": 0.18, "step": 23784 }, { "epoch": 147.73520249221184, "grad_norm": 2.859375, "learning_rate": 8.117057748388804e-06, "loss": 0.2572, "step": 23785 }, { "epoch": 147.74143302180684, "grad_norm": 1.6484375, "learning_rate": 8.115240739216181e-06, "loss": 0.1726, "step": 23786 }, { "epoch": 147.74766355140187, "grad_norm": 1.5625, "learning_rate": 8.113423894031726e-06, "loss": 0.1168, "step": 23787 }, { "epoch": 147.75389408099687, "grad_norm": 1.9140625, "learning_rate": 8.111607212853065e-06, "loss": 0.1165, "step": 23788 }, { "epoch": 147.7601246105919, "grad_norm": 2.265625, "learning_rate": 8.10979069569785e-06, "loss": 0.1868, "step": 23789 }, { "epoch": 147.7663551401869, "grad_norm": 2.375, "learning_rate": 8.107974342583732e-06, "loss": 0.2975, "step": 23790 }, { "epoch": 147.77258566978193, "grad_norm": 2.015625, "learning_rate": 8.106158153528346e-06, "loss": 0.2552, "step": 23791 }, { "epoch": 147.77881619937693, "grad_norm": 1.3515625, "learning_rate": 8.10434212854932e-06, "loss": 0.1106, "step": 23792 }, { "epoch": 147.78504672897196, "grad_norm": 1.421875, "learning_rate": 8.102526267664299e-06, "loss": 0.1495, "step": 23793 }, { "epoch": 147.79127725856696, "grad_norm": 1.78125, "learning_rate": 8.100710570890929e-06, "loss": 0.1392, "step": 23794 }, { "epoch": 147.797507788162, "grad_norm": 2.109375, "learning_rate": 8.098895038246828e-06, "loss": 0.1365, "step": 23795 }, { "epoch": 147.80373831775702, "grad_norm": 2.125, "learning_rate": 8.097079669749644e-06, "loss": 0.1544, "step": 23796 }, { "epoch": 147.80996884735202, "grad_norm": 1.5390625, "learning_rate": 8.095264465416996e-06, "loss": 0.1291, "step": 23797 }, { "epoch": 147.81619937694705, "grad_norm": 2.4375, "learning_rate": 8.093449425266525e-06, "loss": 0.2432, "step": 23798 }, { "epoch": 147.82242990654206, "grad_norm": 1.3984375, "learning_rate": 8.091634549315844e-06, "loss": 0.1214, "step": 23799 }, { "epoch": 147.82866043613708, "grad_norm": 2.15625, "learning_rate": 8.089819837582587e-06, "loss": 0.1204, "step": 23800 }, { "epoch": 147.8348909657321, "grad_norm": 2.765625, "learning_rate": 8.088005290084391e-06, "loss": 0.3581, "step": 23801 }, { "epoch": 147.84112149532712, "grad_norm": 2.03125, "learning_rate": 8.08619090683887e-06, "loss": 0.1171, "step": 23802 }, { "epoch": 147.84735202492212, "grad_norm": 2.265625, "learning_rate": 8.084376687863631e-06, "loss": 0.1334, "step": 23803 }, { "epoch": 147.85358255451715, "grad_norm": 1.734375, "learning_rate": 8.08256263317631e-06, "loss": 0.151, "step": 23804 }, { "epoch": 147.85981308411215, "grad_norm": 2.734375, "learning_rate": 8.08074874279453e-06, "loss": 0.1944, "step": 23805 }, { "epoch": 147.86604361370718, "grad_norm": 2.53125, "learning_rate": 8.078935016735891e-06, "loss": 0.2362, "step": 23806 }, { "epoch": 147.87227414330218, "grad_norm": 1.953125, "learning_rate": 8.077121455018019e-06, "loss": 0.1559, "step": 23807 }, { "epoch": 147.8785046728972, "grad_norm": 2.65625, "learning_rate": 8.075308057658526e-06, "loss": 0.2399, "step": 23808 }, { "epoch": 147.8847352024922, "grad_norm": 2.609375, "learning_rate": 8.07349482467504e-06, "loss": 0.1952, "step": 23809 }, { "epoch": 147.89096573208724, "grad_norm": 2.15625, "learning_rate": 8.07168175608514e-06, "loss": 0.1686, "step": 23810 }, { "epoch": 147.89719626168224, "grad_norm": 1.9140625, "learning_rate": 8.06986885190645e-06, "loss": 0.1954, "step": 23811 }, { "epoch": 147.90342679127727, "grad_norm": 2.234375, "learning_rate": 8.068056112156577e-06, "loss": 0.1834, "step": 23812 }, { "epoch": 147.90965732087227, "grad_norm": 3.25, "learning_rate": 8.066243536853138e-06, "loss": 0.2451, "step": 23813 }, { "epoch": 147.9158878504673, "grad_norm": 1.5546875, "learning_rate": 8.064431126013718e-06, "loss": 0.1162, "step": 23814 }, { "epoch": 147.9221183800623, "grad_norm": 1.1953125, "learning_rate": 8.062618879655937e-06, "loss": 0.1118, "step": 23815 }, { "epoch": 147.92834890965733, "grad_norm": 1.3515625, "learning_rate": 8.060806797797377e-06, "loss": 0.1085, "step": 23816 }, { "epoch": 147.93457943925233, "grad_norm": 2.359375, "learning_rate": 8.058994880455658e-06, "loss": 0.2016, "step": 23817 }, { "epoch": 147.94080996884736, "grad_norm": 1.5078125, "learning_rate": 8.057183127648358e-06, "loss": 0.0808, "step": 23818 }, { "epoch": 147.94704049844236, "grad_norm": 1.0859375, "learning_rate": 8.055371539393081e-06, "loss": 0.1071, "step": 23819 }, { "epoch": 147.9532710280374, "grad_norm": 3.203125, "learning_rate": 8.053560115707434e-06, "loss": 0.1925, "step": 23820 }, { "epoch": 147.9595015576324, "grad_norm": 1.75, "learning_rate": 8.051748856608996e-06, "loss": 0.1182, "step": 23821 }, { "epoch": 147.96573208722742, "grad_norm": 3.078125, "learning_rate": 8.049937762115353e-06, "loss": 0.3459, "step": 23822 }, { "epoch": 147.97196261682242, "grad_norm": 1.65625, "learning_rate": 8.048126832244104e-06, "loss": 0.1254, "step": 23823 }, { "epoch": 147.97819314641745, "grad_norm": 1.671875, "learning_rate": 8.046316067012844e-06, "loss": 0.1165, "step": 23824 }, { "epoch": 147.98442367601245, "grad_norm": 1.4921875, "learning_rate": 8.044505466439142e-06, "loss": 0.1081, "step": 23825 }, { "epoch": 147.99065420560748, "grad_norm": 2.265625, "learning_rate": 8.042695030540593e-06, "loss": 0.1532, "step": 23826 }, { "epoch": 147.99688473520249, "grad_norm": 2.0625, "learning_rate": 8.040884759334785e-06, "loss": 0.1644, "step": 23827 }, { "epoch": 148.0, "grad_norm": 2.109375, "learning_rate": 8.039074652839298e-06, "loss": 0.0717, "step": 23828 }, { "epoch": 148.00623052959503, "grad_norm": 1.359375, "learning_rate": 8.0372647110717e-06, "loss": 0.1345, "step": 23829 }, { "epoch": 148.01246105919003, "grad_norm": 2.328125, "learning_rate": 8.035454934049577e-06, "loss": 0.1368, "step": 23830 }, { "epoch": 148.01869158878506, "grad_norm": 1.4609375, "learning_rate": 8.033645321790514e-06, "loss": 0.1099, "step": 23831 }, { "epoch": 148.02492211838006, "grad_norm": 1.3671875, "learning_rate": 8.031835874312071e-06, "loss": 0.106, "step": 23832 }, { "epoch": 148.0311526479751, "grad_norm": 3.578125, "learning_rate": 8.030026591631831e-06, "loss": 0.3508, "step": 23833 }, { "epoch": 148.0373831775701, "grad_norm": 2.25, "learning_rate": 8.028217473767371e-06, "loss": 0.1212, "step": 23834 }, { "epoch": 148.04361370716512, "grad_norm": 2.1875, "learning_rate": 8.02640852073626e-06, "loss": 0.1166, "step": 23835 }, { "epoch": 148.04984423676012, "grad_norm": 1.0546875, "learning_rate": 8.024599732556051e-06, "loss": 0.1081, "step": 23836 }, { "epoch": 148.05607476635515, "grad_norm": 2.015625, "learning_rate": 8.022791109244324e-06, "loss": 0.0994, "step": 23837 }, { "epoch": 148.06230529595015, "grad_norm": 2.265625, "learning_rate": 8.02098265081865e-06, "loss": 0.1776, "step": 23838 }, { "epoch": 148.06853582554518, "grad_norm": 2.78125, "learning_rate": 8.01917435729658e-06, "loss": 0.216, "step": 23839 }, { "epoch": 148.07476635514018, "grad_norm": 1.15625, "learning_rate": 8.01736622869568e-06, "loss": 0.1187, "step": 23840 }, { "epoch": 148.0809968847352, "grad_norm": 1.53125, "learning_rate": 8.015558265033526e-06, "loss": 0.127, "step": 23841 }, { "epoch": 148.08722741433021, "grad_norm": 1.2578125, "learning_rate": 8.013750466327663e-06, "loss": 0.091, "step": 23842 }, { "epoch": 148.09345794392524, "grad_norm": 2.6875, "learning_rate": 8.011942832595642e-06, "loss": 0.2355, "step": 23843 }, { "epoch": 148.09968847352025, "grad_norm": 1.1171875, "learning_rate": 8.010135363855031e-06, "loss": 0.104, "step": 23844 }, { "epoch": 148.10591900311528, "grad_norm": 2.5, "learning_rate": 8.00832806012338e-06, "loss": 0.1955, "step": 23845 }, { "epoch": 148.11214953271028, "grad_norm": 2.890625, "learning_rate": 8.00652092141825e-06, "loss": 0.2995, "step": 23846 }, { "epoch": 148.1183800623053, "grad_norm": 1.7734375, "learning_rate": 8.00471394775719e-06, "loss": 0.1506, "step": 23847 }, { "epoch": 148.1246105919003, "grad_norm": 0.75390625, "learning_rate": 8.002907139157733e-06, "loss": 0.0976, "step": 23848 }, { "epoch": 148.13084112149534, "grad_norm": 2.0625, "learning_rate": 8.001100495637443e-06, "loss": 0.1707, "step": 23849 }, { "epoch": 148.13707165109034, "grad_norm": 2.453125, "learning_rate": 7.999294017213868e-06, "loss": 0.2799, "step": 23850 }, { "epoch": 148.14330218068537, "grad_norm": 1.8046875, "learning_rate": 7.99748770390454e-06, "loss": 0.2031, "step": 23851 }, { "epoch": 148.14953271028037, "grad_norm": 2.28125, "learning_rate": 7.99568155572701e-06, "loss": 0.1226, "step": 23852 }, { "epoch": 148.1557632398754, "grad_norm": 3.0625, "learning_rate": 7.99387557269883e-06, "loss": 0.263, "step": 23853 }, { "epoch": 148.1619937694704, "grad_norm": 2.109375, "learning_rate": 7.99206975483753e-06, "loss": 0.2307, "step": 23854 }, { "epoch": 148.16822429906543, "grad_norm": 1.453125, "learning_rate": 7.990264102160639e-06, "loss": 0.1415, "step": 23855 }, { "epoch": 148.17445482866043, "grad_norm": 1.8203125, "learning_rate": 7.988458614685704e-06, "loss": 0.1397, "step": 23856 }, { "epoch": 148.18068535825546, "grad_norm": 2.296875, "learning_rate": 7.986653292430265e-06, "loss": 0.1346, "step": 23857 }, { "epoch": 148.18691588785046, "grad_norm": 1.5234375, "learning_rate": 7.984848135411845e-06, "loss": 0.144, "step": 23858 }, { "epoch": 148.1931464174455, "grad_norm": 1.6015625, "learning_rate": 7.983043143647981e-06, "loss": 0.1292, "step": 23859 }, { "epoch": 148.1993769470405, "grad_norm": 2.09375, "learning_rate": 7.981238317156211e-06, "loss": 0.268, "step": 23860 }, { "epoch": 148.20560747663552, "grad_norm": 3.015625, "learning_rate": 7.979433655954055e-06, "loss": 0.2675, "step": 23861 }, { "epoch": 148.21183800623052, "grad_norm": 1.921875, "learning_rate": 7.977629160059036e-06, "loss": 0.1608, "step": 23862 }, { "epoch": 148.21806853582555, "grad_norm": 2.484375, "learning_rate": 7.975824829488684e-06, "loss": 0.1007, "step": 23863 }, { "epoch": 148.22429906542055, "grad_norm": 3.203125, "learning_rate": 7.974020664260534e-06, "loss": 0.2493, "step": 23864 }, { "epoch": 148.23052959501558, "grad_norm": 1.9296875, "learning_rate": 7.972216664392088e-06, "loss": 0.111, "step": 23865 }, { "epoch": 148.23676012461058, "grad_norm": 1.2734375, "learning_rate": 7.970412829900886e-06, "loss": 0.1119, "step": 23866 }, { "epoch": 148.2429906542056, "grad_norm": 0.97265625, "learning_rate": 7.968609160804433e-06, "loss": 0.0846, "step": 23867 }, { "epoch": 148.2492211838006, "grad_norm": 1.453125, "learning_rate": 7.966805657120261e-06, "loss": 0.1459, "step": 23868 }, { "epoch": 148.25545171339564, "grad_norm": 2.328125, "learning_rate": 7.965002318865866e-06, "loss": 0.2284, "step": 23869 }, { "epoch": 148.26168224299064, "grad_norm": 2.046875, "learning_rate": 7.963199146058776e-06, "loss": 0.2037, "step": 23870 }, { "epoch": 148.26791277258567, "grad_norm": 0.8828125, "learning_rate": 7.961396138716506e-06, "loss": 0.1068, "step": 23871 }, { "epoch": 148.27414330218068, "grad_norm": 1.5625, "learning_rate": 7.959593296856566e-06, "loss": 0.1502, "step": 23872 }, { "epoch": 148.2803738317757, "grad_norm": 2.65625, "learning_rate": 7.957790620496453e-06, "loss": 0.3844, "step": 23873 }, { "epoch": 148.2866043613707, "grad_norm": 1.1953125, "learning_rate": 7.955988109653684e-06, "loss": 0.1068, "step": 23874 }, { "epoch": 148.29283489096574, "grad_norm": 1.515625, "learning_rate": 7.954185764345775e-06, "loss": 0.1072, "step": 23875 }, { "epoch": 148.29906542056074, "grad_norm": 0.92578125, "learning_rate": 7.952383584590211e-06, "loss": 0.0899, "step": 23876 }, { "epoch": 148.30529595015577, "grad_norm": 3.59375, "learning_rate": 7.950581570404509e-06, "loss": 0.2117, "step": 23877 }, { "epoch": 148.31152647975077, "grad_norm": 1.6484375, "learning_rate": 7.948779721806163e-06, "loss": 0.1223, "step": 23878 }, { "epoch": 148.3177570093458, "grad_norm": 2.328125, "learning_rate": 7.946978038812695e-06, "loss": 0.1479, "step": 23879 }, { "epoch": 148.3239875389408, "grad_norm": 1.671875, "learning_rate": 7.945176521441569e-06, "loss": 0.1215, "step": 23880 }, { "epoch": 148.33021806853583, "grad_norm": 1.6484375, "learning_rate": 7.943375169710296e-06, "loss": 0.1279, "step": 23881 }, { "epoch": 148.33644859813083, "grad_norm": 2.375, "learning_rate": 7.941573983636375e-06, "loss": 0.1741, "step": 23882 }, { "epoch": 148.34267912772586, "grad_norm": 3.109375, "learning_rate": 7.939772963237304e-06, "loss": 0.3052, "step": 23883 }, { "epoch": 148.34890965732086, "grad_norm": 2.703125, "learning_rate": 7.937972108530561e-06, "loss": 0.1552, "step": 23884 }, { "epoch": 148.3551401869159, "grad_norm": 2.0625, "learning_rate": 7.936171419533653e-06, "loss": 0.2272, "step": 23885 }, { "epoch": 148.3613707165109, "grad_norm": 2.5, "learning_rate": 7.934370896264048e-06, "loss": 0.2255, "step": 23886 }, { "epoch": 148.36760124610592, "grad_norm": 2.671875, "learning_rate": 7.932570538739254e-06, "loss": 0.2872, "step": 23887 }, { "epoch": 148.37383177570092, "grad_norm": 1.53125, "learning_rate": 7.930770346976737e-06, "loss": 0.1159, "step": 23888 }, { "epoch": 148.38006230529595, "grad_norm": 2.8125, "learning_rate": 7.928970320993992e-06, "loss": 0.243, "step": 23889 }, { "epoch": 148.38629283489095, "grad_norm": 2.1875, "learning_rate": 7.927170460808509e-06, "loss": 0.1937, "step": 23890 }, { "epoch": 148.39252336448598, "grad_norm": 1.1171875, "learning_rate": 7.925370766437756e-06, "loss": 0.0986, "step": 23891 }, { "epoch": 148.39875389408098, "grad_norm": 2.046875, "learning_rate": 7.92357123789921e-06, "loss": 0.1461, "step": 23892 }, { "epoch": 148.404984423676, "grad_norm": 2.21875, "learning_rate": 7.921771875210354e-06, "loss": 0.0951, "step": 23893 }, { "epoch": 148.41121495327104, "grad_norm": 0.90625, "learning_rate": 7.919972678388668e-06, "loss": 0.0799, "step": 23894 }, { "epoch": 148.41744548286604, "grad_norm": 1.3984375, "learning_rate": 7.918173647451615e-06, "loss": 0.1457, "step": 23895 }, { "epoch": 148.42367601246107, "grad_norm": 2.515625, "learning_rate": 7.916374782416678e-06, "loss": 0.242, "step": 23896 }, { "epoch": 148.42990654205607, "grad_norm": 2.0, "learning_rate": 7.914576083301328e-06, "loss": 0.1396, "step": 23897 }, { "epoch": 148.4361370716511, "grad_norm": 4.03125, "learning_rate": 7.912777550123032e-06, "loss": 0.1648, "step": 23898 }, { "epoch": 148.4423676012461, "grad_norm": 0.96875, "learning_rate": 7.910979182899247e-06, "loss": 0.0954, "step": 23899 }, { "epoch": 148.44859813084113, "grad_norm": 1.7734375, "learning_rate": 7.909180981647449e-06, "loss": 0.1456, "step": 23900 }, { "epoch": 148.45482866043614, "grad_norm": 2.125, "learning_rate": 7.907382946385108e-06, "loss": 0.1569, "step": 23901 }, { "epoch": 148.46105919003116, "grad_norm": 0.9140625, "learning_rate": 7.905585077129673e-06, "loss": 0.1026, "step": 23902 }, { "epoch": 148.46728971962617, "grad_norm": 2.359375, "learning_rate": 7.903787373898613e-06, "loss": 0.1862, "step": 23903 }, { "epoch": 148.4735202492212, "grad_norm": 2.328125, "learning_rate": 7.901989836709398e-06, "loss": 0.1079, "step": 23904 }, { "epoch": 148.4797507788162, "grad_norm": 2.328125, "learning_rate": 7.900192465579473e-06, "loss": 0.1416, "step": 23905 }, { "epoch": 148.48598130841123, "grad_norm": 1.421875, "learning_rate": 7.89839526052629e-06, "loss": 0.1019, "step": 23906 }, { "epoch": 148.49221183800623, "grad_norm": 2.140625, "learning_rate": 7.896598221567311e-06, "loss": 0.1508, "step": 23907 }, { "epoch": 148.49844236760126, "grad_norm": 0.96484375, "learning_rate": 7.894801348719989e-06, "loss": 0.1065, "step": 23908 }, { "epoch": 148.50467289719626, "grad_norm": 1.2421875, "learning_rate": 7.89300464200178e-06, "loss": 0.1283, "step": 23909 }, { "epoch": 148.5109034267913, "grad_norm": 2.25, "learning_rate": 7.891208101430125e-06, "loss": 0.2239, "step": 23910 }, { "epoch": 148.5171339563863, "grad_norm": 3.15625, "learning_rate": 7.889411727022486e-06, "loss": 0.2314, "step": 23911 }, { "epoch": 148.52336448598132, "grad_norm": 2.53125, "learning_rate": 7.887615518796291e-06, "loss": 0.2467, "step": 23912 }, { "epoch": 148.52959501557632, "grad_norm": 2.375, "learning_rate": 7.885819476769002e-06, "loss": 0.1236, "step": 23913 }, { "epoch": 148.53582554517135, "grad_norm": 2.25, "learning_rate": 7.884023600958049e-06, "loss": 0.1899, "step": 23914 }, { "epoch": 148.54205607476635, "grad_norm": 2.59375, "learning_rate": 7.882227891380878e-06, "loss": 0.1547, "step": 23915 }, { "epoch": 148.54828660436138, "grad_norm": 1.28125, "learning_rate": 7.880432348054941e-06, "loss": 0.139, "step": 23916 }, { "epoch": 148.55451713395638, "grad_norm": 1.53125, "learning_rate": 7.878636970997669e-06, "loss": 0.1121, "step": 23917 }, { "epoch": 148.5607476635514, "grad_norm": 2.140625, "learning_rate": 7.876841760226486e-06, "loss": 0.1405, "step": 23918 }, { "epoch": 148.5669781931464, "grad_norm": 1.7421875, "learning_rate": 7.875046715758838e-06, "loss": 0.1376, "step": 23919 }, { "epoch": 148.57320872274144, "grad_norm": 2.359375, "learning_rate": 7.873251837612173e-06, "loss": 0.1126, "step": 23920 }, { "epoch": 148.57943925233644, "grad_norm": 1.5390625, "learning_rate": 7.871457125803896e-06, "loss": 0.1094, "step": 23921 }, { "epoch": 148.58566978193147, "grad_norm": 2.171875, "learning_rate": 7.869662580351452e-06, "loss": 0.2235, "step": 23922 }, { "epoch": 148.59190031152647, "grad_norm": 1.4140625, "learning_rate": 7.86786820127228e-06, "loss": 0.1205, "step": 23923 }, { "epoch": 148.5981308411215, "grad_norm": 0.89453125, "learning_rate": 7.866073988583795e-06, "loss": 0.0935, "step": 23924 }, { "epoch": 148.6043613707165, "grad_norm": 1.59375, "learning_rate": 7.864279942303418e-06, "loss": 0.1285, "step": 23925 }, { "epoch": 148.61059190031153, "grad_norm": 1.3828125, "learning_rate": 7.862486062448576e-06, "loss": 0.1054, "step": 23926 }, { "epoch": 148.61682242990653, "grad_norm": 1.6796875, "learning_rate": 7.860692349036708e-06, "loss": 0.1294, "step": 23927 }, { "epoch": 148.62305295950156, "grad_norm": 3.09375, "learning_rate": 7.85889880208521e-06, "loss": 0.2295, "step": 23928 }, { "epoch": 148.62928348909657, "grad_norm": 1.765625, "learning_rate": 7.857105421611515e-06, "loss": 0.1573, "step": 23929 }, { "epoch": 148.6355140186916, "grad_norm": 1.78125, "learning_rate": 7.855312207633048e-06, "loss": 0.1497, "step": 23930 }, { "epoch": 148.6417445482866, "grad_norm": 2.578125, "learning_rate": 7.853519160167214e-06, "loss": 0.2034, "step": 23931 }, { "epoch": 148.64797507788163, "grad_norm": 2.671875, "learning_rate": 7.851726279231422e-06, "loss": 0.1972, "step": 23932 }, { "epoch": 148.65420560747663, "grad_norm": 1.890625, "learning_rate": 7.849933564843092e-06, "loss": 0.134, "step": 23933 }, { "epoch": 148.66043613707166, "grad_norm": 1.7890625, "learning_rate": 7.848141017019644e-06, "loss": 0.1804, "step": 23934 }, { "epoch": 148.66666666666666, "grad_norm": 2.3125, "learning_rate": 7.846348635778469e-06, "loss": 0.2323, "step": 23935 }, { "epoch": 148.6728971962617, "grad_norm": 1.734375, "learning_rate": 7.844556421136998e-06, "loss": 0.1521, "step": 23936 }, { "epoch": 148.6791277258567, "grad_norm": 1.1796875, "learning_rate": 7.842764373112613e-06, "loss": 0.1029, "step": 23937 }, { "epoch": 148.68535825545172, "grad_norm": 1.640625, "learning_rate": 7.840972491722736e-06, "loss": 0.1506, "step": 23938 }, { "epoch": 148.69158878504672, "grad_norm": 2.5, "learning_rate": 7.839180776984759e-06, "loss": 0.236, "step": 23939 }, { "epoch": 148.69781931464175, "grad_norm": 3.015625, "learning_rate": 7.837389228916089e-06, "loss": 0.2753, "step": 23940 }, { "epoch": 148.70404984423675, "grad_norm": 2.65625, "learning_rate": 7.835597847534123e-06, "loss": 0.2876, "step": 23941 }, { "epoch": 148.71028037383178, "grad_norm": 2.625, "learning_rate": 7.83380663285628e-06, "loss": 0.2196, "step": 23942 }, { "epoch": 148.71651090342678, "grad_norm": 1.2421875, "learning_rate": 7.832015584899921e-06, "loss": 0.115, "step": 23943 }, { "epoch": 148.7227414330218, "grad_norm": 1.7890625, "learning_rate": 7.830224703682459e-06, "loss": 0.1032, "step": 23944 }, { "epoch": 148.7289719626168, "grad_norm": 1.265625, "learning_rate": 7.82843398922129e-06, "loss": 0.1192, "step": 23945 }, { "epoch": 148.73520249221184, "grad_norm": 2.375, "learning_rate": 7.826643441533806e-06, "loss": 0.2039, "step": 23946 }, { "epoch": 148.74143302180684, "grad_norm": 2.4375, "learning_rate": 7.824853060637388e-06, "loss": 0.1523, "step": 23947 }, { "epoch": 148.74766355140187, "grad_norm": 1.4609375, "learning_rate": 7.823062846549431e-06, "loss": 0.1506, "step": 23948 }, { "epoch": 148.75389408099687, "grad_norm": 1.6171875, "learning_rate": 7.821272799287327e-06, "loss": 0.1208, "step": 23949 }, { "epoch": 148.7601246105919, "grad_norm": 2.21875, "learning_rate": 7.819482918868457e-06, "loss": 0.189, "step": 23950 }, { "epoch": 148.7663551401869, "grad_norm": 1.625, "learning_rate": 7.817693205310194e-06, "loss": 0.1615, "step": 23951 }, { "epoch": 148.77258566978193, "grad_norm": 2.34375, "learning_rate": 7.81590365862993e-06, "loss": 0.2025, "step": 23952 }, { "epoch": 148.77881619937693, "grad_norm": 2.15625, "learning_rate": 7.814114278845052e-06, "loss": 0.1691, "step": 23953 }, { "epoch": 148.78504672897196, "grad_norm": 1.0078125, "learning_rate": 7.812325065972924e-06, "loss": 0.1191, "step": 23954 }, { "epoch": 148.79127725856696, "grad_norm": 2.546875, "learning_rate": 7.81053602003094e-06, "loss": 0.2407, "step": 23955 }, { "epoch": 148.797507788162, "grad_norm": 1.8828125, "learning_rate": 7.808747141036456e-06, "loss": 0.1772, "step": 23956 }, { "epoch": 148.80373831775702, "grad_norm": 2.3125, "learning_rate": 7.806958429006867e-06, "loss": 0.133, "step": 23957 }, { "epoch": 148.80996884735202, "grad_norm": 1.390625, "learning_rate": 7.805169883959526e-06, "loss": 0.1167, "step": 23958 }, { "epoch": 148.81619937694705, "grad_norm": 2.46875, "learning_rate": 7.80338150591181e-06, "loss": 0.0929, "step": 23959 }, { "epoch": 148.82242990654206, "grad_norm": 3.0625, "learning_rate": 7.801593294881101e-06, "loss": 0.2122, "step": 23960 }, { "epoch": 148.82866043613708, "grad_norm": 2.828125, "learning_rate": 7.799805250884753e-06, "loss": 0.1622, "step": 23961 }, { "epoch": 148.8348909657321, "grad_norm": 2.03125, "learning_rate": 7.79801737394013e-06, "loss": 0.189, "step": 23962 }, { "epoch": 148.84112149532712, "grad_norm": 3.390625, "learning_rate": 7.7962296640646e-06, "loss": 0.4097, "step": 23963 }, { "epoch": 148.84735202492212, "grad_norm": 2.234375, "learning_rate": 7.794442121275532e-06, "loss": 0.1799, "step": 23964 }, { "epoch": 148.85358255451715, "grad_norm": 1.9140625, "learning_rate": 7.792654745590276e-06, "loss": 0.1315, "step": 23965 }, { "epoch": 148.85981308411215, "grad_norm": 2.546875, "learning_rate": 7.790867537026195e-06, "loss": 0.1576, "step": 23966 }, { "epoch": 148.86604361370718, "grad_norm": 1.21875, "learning_rate": 7.789080495600656e-06, "loss": 0.121, "step": 23967 }, { "epoch": 148.87227414330218, "grad_norm": 1.5234375, "learning_rate": 7.78729362133101e-06, "loss": 0.1096, "step": 23968 }, { "epoch": 148.8785046728972, "grad_norm": 3.0625, "learning_rate": 7.7855069142346e-06, "loss": 0.2123, "step": 23969 }, { "epoch": 148.8847352024922, "grad_norm": 1.9140625, "learning_rate": 7.783720374328787e-06, "loss": 0.1079, "step": 23970 }, { "epoch": 148.89096573208724, "grad_norm": 1.734375, "learning_rate": 7.781934001630934e-06, "loss": 0.1419, "step": 23971 }, { "epoch": 148.89719626168224, "grad_norm": 1.921875, "learning_rate": 7.780147796158368e-06, "loss": 0.1274, "step": 23972 }, { "epoch": 148.90342679127727, "grad_norm": 1.78125, "learning_rate": 7.77836175792845e-06, "loss": 0.1258, "step": 23973 }, { "epoch": 148.90965732087227, "grad_norm": 1.390625, "learning_rate": 7.776575886958537e-06, "loss": 0.115, "step": 23974 }, { "epoch": 148.9158878504673, "grad_norm": 2.5625, "learning_rate": 7.774790183265956e-06, "loss": 0.2041, "step": 23975 }, { "epoch": 148.9221183800623, "grad_norm": 1.625, "learning_rate": 7.773004646868051e-06, "loss": 0.0878, "step": 23976 }, { "epoch": 148.92834890965733, "grad_norm": 1.5859375, "learning_rate": 7.771219277782168e-06, "loss": 0.1276, "step": 23977 }, { "epoch": 148.93457943925233, "grad_norm": 1.21875, "learning_rate": 7.76943407602565e-06, "loss": 0.0988, "step": 23978 }, { "epoch": 148.94080996884736, "grad_norm": 2.078125, "learning_rate": 7.767649041615837e-06, "loss": 0.1273, "step": 23979 }, { "epoch": 148.94704049844236, "grad_norm": 1.5, "learning_rate": 7.765864174570053e-06, "loss": 0.1334, "step": 23980 }, { "epoch": 148.9532710280374, "grad_norm": 1.7890625, "learning_rate": 7.764079474905652e-06, "loss": 0.1816, "step": 23981 }, { "epoch": 148.9595015576324, "grad_norm": 2.84375, "learning_rate": 7.762294942639947e-06, "loss": 0.1392, "step": 23982 }, { "epoch": 148.96573208722742, "grad_norm": 1.0703125, "learning_rate": 7.760510577790286e-06, "loss": 0.1179, "step": 23983 }, { "epoch": 148.97196261682242, "grad_norm": 2.625, "learning_rate": 7.758726380373987e-06, "loss": 0.2236, "step": 23984 }, { "epoch": 148.97819314641745, "grad_norm": 2.3125, "learning_rate": 7.756942350408382e-06, "loss": 0.1693, "step": 23985 }, { "epoch": 148.98442367601245, "grad_norm": 1.9921875, "learning_rate": 7.755158487910807e-06, "loss": 0.1519, "step": 23986 }, { "epoch": 148.99065420560748, "grad_norm": 1.4609375, "learning_rate": 7.753374792898582e-06, "loss": 0.2096, "step": 23987 }, { "epoch": 148.99688473520249, "grad_norm": 1.25, "learning_rate": 7.751591265389021e-06, "loss": 0.1044, "step": 23988 }, { "epoch": 149.0, "grad_norm": 1.0078125, "learning_rate": 7.74980790539945e-06, "loss": 0.0715, "step": 23989 }, { "epoch": 149.00623052959503, "grad_norm": 2.421875, "learning_rate": 7.748024712947205e-06, "loss": 0.1829, "step": 23990 }, { "epoch": 149.01246105919003, "grad_norm": 1.7265625, "learning_rate": 7.746241688049582e-06, "loss": 0.1433, "step": 23991 }, { "epoch": 149.01869158878506, "grad_norm": 1.6015625, "learning_rate": 7.744458830723911e-06, "loss": 0.0969, "step": 23992 }, { "epoch": 149.02492211838006, "grad_norm": 2.65625, "learning_rate": 7.74267614098751e-06, "loss": 0.2289, "step": 23993 }, { "epoch": 149.0311526479751, "grad_norm": 1.578125, "learning_rate": 7.740893618857689e-06, "loss": 0.138, "step": 23994 }, { "epoch": 149.0373831775701, "grad_norm": 2.171875, "learning_rate": 7.739111264351753e-06, "loss": 0.3478, "step": 23995 }, { "epoch": 149.04361370716512, "grad_norm": 1.625, "learning_rate": 7.737329077487018e-06, "loss": 0.1243, "step": 23996 }, { "epoch": 149.04984423676012, "grad_norm": 2.65625, "learning_rate": 7.735547058280801e-06, "loss": 0.2282, "step": 23997 }, { "epoch": 149.05607476635515, "grad_norm": 2.84375, "learning_rate": 7.733765206750393e-06, "loss": 0.1745, "step": 23998 }, { "epoch": 149.06230529595015, "grad_norm": 1.6328125, "learning_rate": 7.731983522913112e-06, "loss": 0.2137, "step": 23999 }, { "epoch": 149.06853582554518, "grad_norm": 2.53125, "learning_rate": 7.730202006786263e-06, "loss": 0.2254, "step": 24000 }, { "epoch": 149.07476635514018, "grad_norm": 1.5546875, "learning_rate": 7.728420658387148e-06, "loss": 0.1237, "step": 24001 }, { "epoch": 149.0809968847352, "grad_norm": 1.6328125, "learning_rate": 7.726639477733052e-06, "loss": 0.1742, "step": 24002 }, { "epoch": 149.08722741433021, "grad_norm": 1.9765625, "learning_rate": 7.72485846484129e-06, "loss": 0.1274, "step": 24003 }, { "epoch": 149.09345794392524, "grad_norm": 1.421875, "learning_rate": 7.723077619729155e-06, "loss": 0.135, "step": 24004 }, { "epoch": 149.09968847352025, "grad_norm": 2.1875, "learning_rate": 7.72129694241395e-06, "loss": 0.2121, "step": 24005 }, { "epoch": 149.10591900311528, "grad_norm": 1.84375, "learning_rate": 7.719516432912962e-06, "loss": 0.1198, "step": 24006 }, { "epoch": 149.11214953271028, "grad_norm": 1.9296875, "learning_rate": 7.71773609124348e-06, "loss": 0.1224, "step": 24007 }, { "epoch": 149.1183800623053, "grad_norm": 2.234375, "learning_rate": 7.715955917422799e-06, "loss": 0.0908, "step": 24008 }, { "epoch": 149.1246105919003, "grad_norm": 1.3671875, "learning_rate": 7.714175911468215e-06, "loss": 0.1287, "step": 24009 }, { "epoch": 149.13084112149534, "grad_norm": 2.109375, "learning_rate": 7.712396073397004e-06, "loss": 0.1854, "step": 24010 }, { "epoch": 149.13707165109034, "grad_norm": 1.1328125, "learning_rate": 7.710616403226459e-06, "loss": 0.1136, "step": 24011 }, { "epoch": 149.14330218068537, "grad_norm": 2.46875, "learning_rate": 7.70883690097387e-06, "loss": 0.1646, "step": 24012 }, { "epoch": 149.14953271028037, "grad_norm": 1.2734375, "learning_rate": 7.707057566656514e-06, "loss": 0.1484, "step": 24013 }, { "epoch": 149.1557632398754, "grad_norm": 1.734375, "learning_rate": 7.705278400291662e-06, "loss": 0.1549, "step": 24014 }, { "epoch": 149.1619937694704, "grad_norm": 1.7265625, "learning_rate": 7.703499401896605e-06, "loss": 0.1646, "step": 24015 }, { "epoch": 149.16822429906543, "grad_norm": 2.859375, "learning_rate": 7.701720571488628e-06, "loss": 0.2465, "step": 24016 }, { "epoch": 149.17445482866043, "grad_norm": 0.9921875, "learning_rate": 7.69994190908499e-06, "loss": 0.0934, "step": 24017 }, { "epoch": 149.18068535825546, "grad_norm": 1.7734375, "learning_rate": 7.698163414702975e-06, "loss": 0.1757, "step": 24018 }, { "epoch": 149.18691588785046, "grad_norm": 1.4765625, "learning_rate": 7.696385088359862e-06, "loss": 0.114, "step": 24019 }, { "epoch": 149.1931464174455, "grad_norm": 2.0, "learning_rate": 7.694606930072917e-06, "loss": 0.1911, "step": 24020 }, { "epoch": 149.1993769470405, "grad_norm": 2.390625, "learning_rate": 7.6928289398594e-06, "loss": 0.2639, "step": 24021 }, { "epoch": 149.20560747663552, "grad_norm": 2.5, "learning_rate": 7.69105111773659e-06, "loss": 0.11, "step": 24022 }, { "epoch": 149.21183800623052, "grad_norm": 1.1796875, "learning_rate": 7.689273463721755e-06, "loss": 0.1099, "step": 24023 }, { "epoch": 149.21806853582555, "grad_norm": 1.21875, "learning_rate": 7.687495977832151e-06, "loss": 0.1107, "step": 24024 }, { "epoch": 149.22429906542055, "grad_norm": 1.6875, "learning_rate": 7.685718660085057e-06, "loss": 0.1199, "step": 24025 }, { "epoch": 149.23052959501558, "grad_norm": 1.2421875, "learning_rate": 7.683941510497714e-06, "loss": 0.1054, "step": 24026 }, { "epoch": 149.23676012461058, "grad_norm": 1.6875, "learning_rate": 7.6821645290874e-06, "loss": 0.1335, "step": 24027 }, { "epoch": 149.2429906542056, "grad_norm": 1.0390625, "learning_rate": 7.680387715871359e-06, "loss": 0.1103, "step": 24028 }, { "epoch": 149.2492211838006, "grad_norm": 2.703125, "learning_rate": 7.678611070866854e-06, "loss": 0.1759, "step": 24029 }, { "epoch": 149.25545171339564, "grad_norm": 2.34375, "learning_rate": 7.67683459409115e-06, "loss": 0.168, "step": 24030 }, { "epoch": 149.26168224299064, "grad_norm": 1.96875, "learning_rate": 7.67505828556149e-06, "loss": 0.1811, "step": 24031 }, { "epoch": 149.26791277258567, "grad_norm": 1.84375, "learning_rate": 7.673282145295118e-06, "loss": 0.1799, "step": 24032 }, { "epoch": 149.27414330218068, "grad_norm": 2.1875, "learning_rate": 7.671506173309295e-06, "loss": 0.1879, "step": 24033 }, { "epoch": 149.2803738317757, "grad_norm": 2.46875, "learning_rate": 7.669730369621277e-06, "loss": 0.2095, "step": 24034 }, { "epoch": 149.2866043613707, "grad_norm": 2.40625, "learning_rate": 7.667954734248292e-06, "loss": 0.213, "step": 24035 }, { "epoch": 149.29283489096574, "grad_norm": 1.8046875, "learning_rate": 7.666179267207596e-06, "loss": 0.1488, "step": 24036 }, { "epoch": 149.29906542056074, "grad_norm": 2.234375, "learning_rate": 7.66440396851643e-06, "loss": 0.1651, "step": 24037 }, { "epoch": 149.30529595015577, "grad_norm": 2.390625, "learning_rate": 7.66262883819206e-06, "loss": 0.2047, "step": 24038 }, { "epoch": 149.31152647975077, "grad_norm": 2.03125, "learning_rate": 7.660853876251684e-06, "loss": 0.1076, "step": 24039 }, { "epoch": 149.3177570093458, "grad_norm": 2.46875, "learning_rate": 7.65907908271256e-06, "loss": 0.1376, "step": 24040 }, { "epoch": 149.3239875389408, "grad_norm": 1.7890625, "learning_rate": 7.657304457591926e-06, "loss": 0.1645, "step": 24041 }, { "epoch": 149.33021806853583, "grad_norm": 1.515625, "learning_rate": 7.65553000090703e-06, "loss": 0.1463, "step": 24042 }, { "epoch": 149.33644859813083, "grad_norm": 1.625, "learning_rate": 7.653755712675082e-06, "loss": 0.1268, "step": 24043 }, { "epoch": 149.34267912772586, "grad_norm": 1.15625, "learning_rate": 7.651981592913334e-06, "loss": 0.1155, "step": 24044 }, { "epoch": 149.34890965732086, "grad_norm": 1.578125, "learning_rate": 7.650207641639003e-06, "loss": 0.147, "step": 24045 }, { "epoch": 149.3551401869159, "grad_norm": 1.34375, "learning_rate": 7.648433858869331e-06, "loss": 0.1331, "step": 24046 }, { "epoch": 149.3613707165109, "grad_norm": 1.5078125, "learning_rate": 7.64666024462153e-06, "loss": 0.1213, "step": 24047 }, { "epoch": 149.36760124610592, "grad_norm": 2.53125, "learning_rate": 7.644886798912831e-06, "loss": 0.1916, "step": 24048 }, { "epoch": 149.37383177570092, "grad_norm": 1.6328125, "learning_rate": 7.643113521760473e-06, "loss": 0.152, "step": 24049 }, { "epoch": 149.38006230529595, "grad_norm": 1.625, "learning_rate": 7.641340413181655e-06, "loss": 0.0926, "step": 24050 }, { "epoch": 149.38629283489095, "grad_norm": 2.09375, "learning_rate": 7.639567473193618e-06, "loss": 0.3228, "step": 24051 }, { "epoch": 149.39252336448598, "grad_norm": 1.4921875, "learning_rate": 7.637794701813562e-06, "loss": 0.0942, "step": 24052 }, { "epoch": 149.39875389408098, "grad_norm": 1.7265625, "learning_rate": 7.636022099058724e-06, "loss": 0.1466, "step": 24053 }, { "epoch": 149.404984423676, "grad_norm": 1.9609375, "learning_rate": 7.634249664946303e-06, "loss": 0.1771, "step": 24054 }, { "epoch": 149.41121495327104, "grad_norm": 1.359375, "learning_rate": 7.632477399493517e-06, "loss": 0.1375, "step": 24055 }, { "epoch": 149.41744548286604, "grad_norm": 1.671875, "learning_rate": 7.630705302717595e-06, "loss": 0.1607, "step": 24056 }, { "epoch": 149.42367601246107, "grad_norm": 1.2109375, "learning_rate": 7.628933374635733e-06, "loss": 0.1407, "step": 24057 }, { "epoch": 149.42990654205607, "grad_norm": 3.09375, "learning_rate": 7.6271616152651355e-06, "loss": 0.2194, "step": 24058 }, { "epoch": 149.4361370716511, "grad_norm": 1.4609375, "learning_rate": 7.625390024623014e-06, "loss": 0.1084, "step": 24059 }, { "epoch": 149.4423676012461, "grad_norm": 1.28125, "learning_rate": 7.62361860272659e-06, "loss": 0.1101, "step": 24060 }, { "epoch": 149.44859813084113, "grad_norm": 1.9140625, "learning_rate": 7.621847349593045e-06, "loss": 0.1221, "step": 24061 }, { "epoch": 149.45482866043614, "grad_norm": 1.0, "learning_rate": 7.6200762652395925e-06, "loss": 0.0804, "step": 24062 }, { "epoch": 149.46105919003116, "grad_norm": 1.390625, "learning_rate": 7.6183053496834415e-06, "loss": 0.0851, "step": 24063 }, { "epoch": 149.46728971962617, "grad_norm": 2.703125, "learning_rate": 7.616534602941783e-06, "loss": 0.1928, "step": 24064 }, { "epoch": 149.4735202492212, "grad_norm": 2.625, "learning_rate": 7.614764025031809e-06, "loss": 0.1951, "step": 24065 }, { "epoch": 149.4797507788162, "grad_norm": 2.296875, "learning_rate": 7.61299361597072e-06, "loss": 0.1889, "step": 24066 }, { "epoch": 149.48598130841123, "grad_norm": 2.5, "learning_rate": 7.611223375775722e-06, "loss": 0.1972, "step": 24067 }, { "epoch": 149.49221183800623, "grad_norm": 3.25, "learning_rate": 7.6094533044639925e-06, "loss": 0.2502, "step": 24068 }, { "epoch": 149.49844236760126, "grad_norm": 1.6015625, "learning_rate": 7.607683402052726e-06, "loss": 0.1355, "step": 24069 }, { "epoch": 149.50467289719626, "grad_norm": 1.125, "learning_rate": 7.6059136685591226e-06, "loss": 0.1126, "step": 24070 }, { "epoch": 149.5109034267913, "grad_norm": 1.296875, "learning_rate": 7.604144104000363e-06, "loss": 0.1118, "step": 24071 }, { "epoch": 149.5171339563863, "grad_norm": 2.109375, "learning_rate": 7.6023747083936285e-06, "loss": 0.1415, "step": 24072 }, { "epoch": 149.52336448598132, "grad_norm": 1.1796875, "learning_rate": 7.600605481756104e-06, "loss": 0.0986, "step": 24073 }, { "epoch": 149.52959501557632, "grad_norm": 1.9296875, "learning_rate": 7.5988364241049795e-06, "loss": 0.1417, "step": 24074 }, { "epoch": 149.53582554517135, "grad_norm": 1.640625, "learning_rate": 7.597067535457442e-06, "loss": 0.1191, "step": 24075 }, { "epoch": 149.54205607476635, "grad_norm": 1.8671875, "learning_rate": 7.595298815830662e-06, "loss": 0.1366, "step": 24076 }, { "epoch": 149.54828660436138, "grad_norm": 2.765625, "learning_rate": 7.593530265241814e-06, "loss": 0.1581, "step": 24077 }, { "epoch": 149.55451713395638, "grad_norm": 1.375, "learning_rate": 7.591761883708076e-06, "loss": 0.1412, "step": 24078 }, { "epoch": 149.5607476635514, "grad_norm": 1.640625, "learning_rate": 7.589993671246637e-06, "loss": 0.1276, "step": 24079 }, { "epoch": 149.5669781931464, "grad_norm": 1.5, "learning_rate": 7.588225627874651e-06, "loss": 0.14, "step": 24080 }, { "epoch": 149.57320872274144, "grad_norm": 2.609375, "learning_rate": 7.586457753609297e-06, "loss": 0.3132, "step": 24081 }, { "epoch": 149.57943925233644, "grad_norm": 2.53125, "learning_rate": 7.584690048467754e-06, "loss": 0.212, "step": 24082 }, { "epoch": 149.58566978193147, "grad_norm": 2.5625, "learning_rate": 7.582922512467183e-06, "loss": 0.1756, "step": 24083 }, { "epoch": 149.59190031152647, "grad_norm": 3.0, "learning_rate": 7.581155145624743e-06, "loss": 0.1503, "step": 24084 }, { "epoch": 149.5981308411215, "grad_norm": 2.5, "learning_rate": 7.579387947957606e-06, "loss": 0.1622, "step": 24085 }, { "epoch": 149.6043613707165, "grad_norm": 1.4375, "learning_rate": 7.5776209194829446e-06, "loss": 0.1079, "step": 24086 }, { "epoch": 149.61059190031153, "grad_norm": 2.265625, "learning_rate": 7.575854060217902e-06, "loss": 0.1728, "step": 24087 }, { "epoch": 149.61682242990653, "grad_norm": 1.9140625, "learning_rate": 7.574087370179647e-06, "loss": 0.1547, "step": 24088 }, { "epoch": 149.62305295950156, "grad_norm": 2.09375, "learning_rate": 7.57232084938535e-06, "loss": 0.1808, "step": 24089 }, { "epoch": 149.62928348909657, "grad_norm": 2.578125, "learning_rate": 7.570554497852153e-06, "loss": 0.1679, "step": 24090 }, { "epoch": 149.6355140186916, "grad_norm": 1.6640625, "learning_rate": 7.5687883155972085e-06, "loss": 0.1653, "step": 24091 }, { "epoch": 149.6417445482866, "grad_norm": 2.96875, "learning_rate": 7.567022302637677e-06, "loss": 0.2653, "step": 24092 }, { "epoch": 149.64797507788163, "grad_norm": 2.328125, "learning_rate": 7.565256458990716e-06, "loss": 0.1756, "step": 24093 }, { "epoch": 149.65420560747663, "grad_norm": 2.234375, "learning_rate": 7.563490784673463e-06, "loss": 0.1815, "step": 24094 }, { "epoch": 149.66043613707166, "grad_norm": 1.8046875, "learning_rate": 7.56172527970308e-06, "loss": 0.1534, "step": 24095 }, { "epoch": 149.66666666666666, "grad_norm": 1.6796875, "learning_rate": 7.559959944096701e-06, "loss": 0.1231, "step": 24096 }, { "epoch": 149.6728971962617, "grad_norm": 1.4375, "learning_rate": 7.558194777871483e-06, "loss": 0.1182, "step": 24097 }, { "epoch": 149.6791277258567, "grad_norm": 1.984375, "learning_rate": 7.556429781044555e-06, "loss": 0.1481, "step": 24098 }, { "epoch": 149.68535825545172, "grad_norm": 2.34375, "learning_rate": 7.55466495363307e-06, "loss": 0.2137, "step": 24099 }, { "epoch": 149.69158878504672, "grad_norm": 1.109375, "learning_rate": 7.552900295654167e-06, "loss": 0.1013, "step": 24100 }, { "epoch": 149.69781931464175, "grad_norm": 2.671875, "learning_rate": 7.55113580712499e-06, "loss": 0.187, "step": 24101 }, { "epoch": 149.70404984423675, "grad_norm": 2.78125, "learning_rate": 7.54937148806267e-06, "loss": 0.2447, "step": 24102 }, { "epoch": 149.71028037383178, "grad_norm": 1.5546875, "learning_rate": 7.547607338484333e-06, "loss": 0.1071, "step": 24103 }, { "epoch": 149.71651090342678, "grad_norm": 3.015625, "learning_rate": 7.545843358407126e-06, "loss": 0.251, "step": 24104 }, { "epoch": 149.7227414330218, "grad_norm": 1.7890625, "learning_rate": 7.5440795478481815e-06, "loss": 0.1257, "step": 24105 }, { "epoch": 149.7289719626168, "grad_norm": 1.9921875, "learning_rate": 7.542315906824621e-06, "loss": 0.2492, "step": 24106 }, { "epoch": 149.73520249221184, "grad_norm": 0.8828125, "learning_rate": 7.540552435353576e-06, "loss": 0.0962, "step": 24107 }, { "epoch": 149.74143302180684, "grad_norm": 2.28125, "learning_rate": 7.538789133452184e-06, "loss": 0.2512, "step": 24108 }, { "epoch": 149.74766355140187, "grad_norm": 1.8828125, "learning_rate": 7.537026001137562e-06, "loss": 0.1413, "step": 24109 }, { "epoch": 149.75389408099687, "grad_norm": 2.515625, "learning_rate": 7.535263038426829e-06, "loss": 0.2275, "step": 24110 }, { "epoch": 149.7601246105919, "grad_norm": 1.2421875, "learning_rate": 7.5335002453371105e-06, "loss": 0.1286, "step": 24111 }, { "epoch": 149.7663551401869, "grad_norm": 1.546875, "learning_rate": 7.531737621885538e-06, "loss": 0.1098, "step": 24112 }, { "epoch": 149.77258566978193, "grad_norm": 2.359375, "learning_rate": 7.5299751680892125e-06, "loss": 0.1756, "step": 24113 }, { "epoch": 149.77881619937693, "grad_norm": 1.8828125, "learning_rate": 7.528212883965269e-06, "loss": 0.2451, "step": 24114 }, { "epoch": 149.78504672897196, "grad_norm": 2.578125, "learning_rate": 7.5264507695308045e-06, "loss": 0.2668, "step": 24115 }, { "epoch": 149.79127725856696, "grad_norm": 2.421875, "learning_rate": 7.524688824802953e-06, "loss": 0.1247, "step": 24116 }, { "epoch": 149.797507788162, "grad_norm": 1.0546875, "learning_rate": 7.522927049798809e-06, "loss": 0.0985, "step": 24117 }, { "epoch": 149.80373831775702, "grad_norm": 1.7890625, "learning_rate": 7.521165444535491e-06, "loss": 0.0983, "step": 24118 }, { "epoch": 149.80996884735202, "grad_norm": 2.75, "learning_rate": 7.5194040090301146e-06, "loss": 0.305, "step": 24119 }, { "epoch": 149.81619937694705, "grad_norm": 1.7421875, "learning_rate": 7.517642743299775e-06, "loss": 0.1343, "step": 24120 }, { "epoch": 149.82242990654206, "grad_norm": 2.515625, "learning_rate": 7.5158816473615885e-06, "loss": 0.1387, "step": 24121 }, { "epoch": 149.82866043613708, "grad_norm": 2.0625, "learning_rate": 7.514120721232651e-06, "loss": 0.1212, "step": 24122 }, { "epoch": 149.8348909657321, "grad_norm": 2.5625, "learning_rate": 7.512359964930074e-06, "loss": 0.1946, "step": 24123 }, { "epoch": 149.84112149532712, "grad_norm": 1.1875, "learning_rate": 7.510599378470945e-06, "loss": 0.1126, "step": 24124 }, { "epoch": 149.84735202492212, "grad_norm": 2.578125, "learning_rate": 7.50883896187237e-06, "loss": 0.2303, "step": 24125 }, { "epoch": 149.85358255451715, "grad_norm": 2.375, "learning_rate": 7.507078715151455e-06, "loss": 0.2227, "step": 24126 }, { "epoch": 149.85981308411215, "grad_norm": 2.71875, "learning_rate": 7.50531863832529e-06, "loss": 0.2035, "step": 24127 }, { "epoch": 149.86604361370718, "grad_norm": 1.9921875, "learning_rate": 7.503558731410959e-06, "loss": 0.1543, "step": 24128 }, { "epoch": 149.87227414330218, "grad_norm": 2.65625, "learning_rate": 7.501798994425563e-06, "loss": 0.2008, "step": 24129 }, { "epoch": 149.8785046728972, "grad_norm": 1.9609375, "learning_rate": 7.5000394273862015e-06, "loss": 0.1584, "step": 24130 }, { "epoch": 149.8847352024922, "grad_norm": 3.015625, "learning_rate": 7.498280030309945e-06, "loss": 0.215, "step": 24131 }, { "epoch": 149.89096573208724, "grad_norm": 1.171875, "learning_rate": 7.496520803213894e-06, "loss": 0.1147, "step": 24132 }, { "epoch": 149.89719626168224, "grad_norm": 2.296875, "learning_rate": 7.4947617461151395e-06, "loss": 0.165, "step": 24133 }, { "epoch": 149.90342679127727, "grad_norm": 1.96875, "learning_rate": 7.493002859030757e-06, "loss": 0.141, "step": 24134 }, { "epoch": 149.90965732087227, "grad_norm": 1.7265625, "learning_rate": 7.491244141977824e-06, "loss": 0.1539, "step": 24135 }, { "epoch": 149.9158878504673, "grad_norm": 1.8125, "learning_rate": 7.489485594973425e-06, "loss": 0.1615, "step": 24136 }, { "epoch": 149.9221183800623, "grad_norm": 0.796875, "learning_rate": 7.487727218034646e-06, "loss": 0.0934, "step": 24137 }, { "epoch": 149.92834890965733, "grad_norm": 1.5078125, "learning_rate": 7.485969011178565e-06, "loss": 0.1295, "step": 24138 }, { "epoch": 149.93457943925233, "grad_norm": 1.9375, "learning_rate": 7.484210974422248e-06, "loss": 0.1218, "step": 24139 }, { "epoch": 149.94080996884736, "grad_norm": 2.171875, "learning_rate": 7.482453107782783e-06, "loss": 0.1494, "step": 24140 }, { "epoch": 149.94704049844236, "grad_norm": 2.171875, "learning_rate": 7.480695411277228e-06, "loss": 0.1674, "step": 24141 }, { "epoch": 149.9532710280374, "grad_norm": 1.421875, "learning_rate": 7.478937884922668e-06, "loss": 0.1581, "step": 24142 }, { "epoch": 149.9595015576324, "grad_norm": 1.3828125, "learning_rate": 7.477180528736158e-06, "loss": 0.1141, "step": 24143 }, { "epoch": 149.96573208722742, "grad_norm": 1.90625, "learning_rate": 7.475423342734775e-06, "loss": 0.1575, "step": 24144 }, { "epoch": 149.97196261682242, "grad_norm": 1.6171875, "learning_rate": 7.473666326935589e-06, "loss": 0.115, "step": 24145 }, { "epoch": 149.97819314641745, "grad_norm": 1.8984375, "learning_rate": 7.47190948135566e-06, "loss": 0.1514, "step": 24146 }, { "epoch": 149.98442367601245, "grad_norm": 2.921875, "learning_rate": 7.470152806012043e-06, "loss": 0.1777, "step": 24147 }, { "epoch": 149.99065420560748, "grad_norm": 2.546875, "learning_rate": 7.468396300921804e-06, "loss": 0.2199, "step": 24148 }, { "epoch": 149.99688473520249, "grad_norm": 2.6875, "learning_rate": 7.4666399661020126e-06, "loss": 0.195, "step": 24149 }, { "epoch": 150.0, "grad_norm": 1.140625, "learning_rate": 7.464883801569714e-06, "loss": 0.0641, "step": 24150 }, { "epoch": 150.00623052959503, "grad_norm": 2.375, "learning_rate": 7.463127807341966e-06, "loss": 0.2351, "step": 24151 }, { "epoch": 150.01246105919003, "grad_norm": 1.78125, "learning_rate": 7.461371983435836e-06, "loss": 0.1577, "step": 24152 }, { "epoch": 150.01869158878506, "grad_norm": 2.25, "learning_rate": 7.4596163298683654e-06, "loss": 0.1718, "step": 24153 }, { "epoch": 150.02492211838006, "grad_norm": 1.6953125, "learning_rate": 7.4578608466566e-06, "loss": 0.1094, "step": 24154 }, { "epoch": 150.0311526479751, "grad_norm": 1.6171875, "learning_rate": 7.456105533817598e-06, "loss": 0.1572, "step": 24155 }, { "epoch": 150.0373831775701, "grad_norm": 2.296875, "learning_rate": 7.454350391368411e-06, "loss": 0.2494, "step": 24156 }, { "epoch": 150.04361370716512, "grad_norm": 2.96875, "learning_rate": 7.452595419326075e-06, "loss": 0.2158, "step": 24157 }, { "epoch": 150.04984423676012, "grad_norm": 2.21875, "learning_rate": 7.450840617707641e-06, "loss": 0.1902, "step": 24158 }, { "epoch": 150.05607476635515, "grad_norm": 2.140625, "learning_rate": 7.449085986530155e-06, "loss": 0.216, "step": 24159 }, { "epoch": 150.06230529595015, "grad_norm": 1.2421875, "learning_rate": 7.447331525810658e-06, "loss": 0.114, "step": 24160 }, { "epoch": 150.06853582554518, "grad_norm": 2.09375, "learning_rate": 7.4455772355661755e-06, "loss": 0.1762, "step": 24161 }, { "epoch": 150.07476635514018, "grad_norm": 2.296875, "learning_rate": 7.443823115813758e-06, "loss": 0.1279, "step": 24162 }, { "epoch": 150.0809968847352, "grad_norm": 3.1875, "learning_rate": 7.442069166570445e-06, "loss": 0.2138, "step": 24163 }, { "epoch": 150.08722741433021, "grad_norm": 3.390625, "learning_rate": 7.440315387853261e-06, "loss": 0.2181, "step": 24164 }, { "epoch": 150.09345794392524, "grad_norm": 2.359375, "learning_rate": 7.4385617796792516e-06, "loss": 0.1945, "step": 24165 }, { "epoch": 150.09968847352025, "grad_norm": 2.109375, "learning_rate": 7.4368083420654296e-06, "loss": 0.1124, "step": 24166 }, { "epoch": 150.10591900311528, "grad_norm": 3.640625, "learning_rate": 7.4350550750288465e-06, "loss": 0.2327, "step": 24167 }, { "epoch": 150.11214953271028, "grad_norm": 2.171875, "learning_rate": 7.43330197858651e-06, "loss": 0.1403, "step": 24168 }, { "epoch": 150.1183800623053, "grad_norm": 1.3125, "learning_rate": 7.431549052755457e-06, "loss": 0.1061, "step": 24169 }, { "epoch": 150.1246105919003, "grad_norm": 2.125, "learning_rate": 7.429796297552713e-06, "loss": 0.201, "step": 24170 }, { "epoch": 150.13084112149534, "grad_norm": 2.390625, "learning_rate": 7.428043712995305e-06, "loss": 0.2411, "step": 24171 }, { "epoch": 150.13707165109034, "grad_norm": 2.40625, "learning_rate": 7.426291299100252e-06, "loss": 0.2034, "step": 24172 }, { "epoch": 150.14330218068537, "grad_norm": 1.6015625, "learning_rate": 7.4245390558845605e-06, "loss": 0.1072, "step": 24173 }, { "epoch": 150.14953271028037, "grad_norm": 1.6875, "learning_rate": 7.4227869833652595e-06, "loss": 0.1546, "step": 24174 }, { "epoch": 150.1557632398754, "grad_norm": 3.453125, "learning_rate": 7.421035081559374e-06, "loss": 0.2641, "step": 24175 }, { "epoch": 150.1619937694704, "grad_norm": 2.1875, "learning_rate": 7.419283350483899e-06, "loss": 0.1258, "step": 24176 }, { "epoch": 150.16822429906543, "grad_norm": 2.625, "learning_rate": 7.4175317901558626e-06, "loss": 0.2291, "step": 24177 }, { "epoch": 150.17445482866043, "grad_norm": 1.015625, "learning_rate": 7.415780400592276e-06, "loss": 0.1114, "step": 24178 }, { "epoch": 150.18068535825546, "grad_norm": 1.828125, "learning_rate": 7.414029181810148e-06, "loss": 0.1452, "step": 24179 }, { "epoch": 150.18691588785046, "grad_norm": 2.421875, "learning_rate": 7.412278133826475e-06, "loss": 0.0984, "step": 24180 }, { "epoch": 150.1931464174455, "grad_norm": 2.890625, "learning_rate": 7.410527256658273e-06, "loss": 0.1574, "step": 24181 }, { "epoch": 150.1993769470405, "grad_norm": 1.328125, "learning_rate": 7.408776550322552e-06, "loss": 0.134, "step": 24182 }, { "epoch": 150.20560747663552, "grad_norm": 1.78125, "learning_rate": 7.4070260148363045e-06, "loss": 0.0962, "step": 24183 }, { "epoch": 150.21183800623052, "grad_norm": 1.3828125, "learning_rate": 7.405275650216542e-06, "loss": 0.1236, "step": 24184 }, { "epoch": 150.21806853582555, "grad_norm": 2.4375, "learning_rate": 7.403525456480254e-06, "loss": 0.1531, "step": 24185 }, { "epoch": 150.22429906542055, "grad_norm": 2.5, "learning_rate": 7.401775433644453e-06, "loss": 0.2078, "step": 24186 }, { "epoch": 150.23052959501558, "grad_norm": 2.671875, "learning_rate": 7.400025581726116e-06, "loss": 0.2102, "step": 24187 }, { "epoch": 150.23676012461058, "grad_norm": 3.328125, "learning_rate": 7.398275900742252e-06, "loss": 0.3137, "step": 24188 }, { "epoch": 150.2429906542056, "grad_norm": 1.25, "learning_rate": 7.396526390709857e-06, "loss": 0.1084, "step": 24189 }, { "epoch": 150.2492211838006, "grad_norm": 2.9375, "learning_rate": 7.394777051645907e-06, "loss": 0.2313, "step": 24190 }, { "epoch": 150.25545171339564, "grad_norm": 2.171875, "learning_rate": 7.39302788356741e-06, "loss": 0.123, "step": 24191 }, { "epoch": 150.26168224299064, "grad_norm": 1.5703125, "learning_rate": 7.39127888649134e-06, "loss": 0.1599, "step": 24192 }, { "epoch": 150.26791277258567, "grad_norm": 1.7421875, "learning_rate": 7.389530060434696e-06, "loss": 0.2222, "step": 24193 }, { "epoch": 150.27414330218068, "grad_norm": 3.375, "learning_rate": 7.387781405414448e-06, "loss": 0.3025, "step": 24194 }, { "epoch": 150.2803738317757, "grad_norm": 2.921875, "learning_rate": 7.386032921447589e-06, "loss": 0.3966, "step": 24195 }, { "epoch": 150.2866043613707, "grad_norm": 2.375, "learning_rate": 7.384284608551098e-06, "loss": 0.169, "step": 24196 }, { "epoch": 150.29283489096574, "grad_norm": 1.6640625, "learning_rate": 7.382536466741974e-06, "loss": 0.1422, "step": 24197 }, { "epoch": 150.29906542056074, "grad_norm": 2.796875, "learning_rate": 7.380788496037161e-06, "loss": 0.3335, "step": 24198 }, { "epoch": 150.30529595015577, "grad_norm": 3.265625, "learning_rate": 7.379040696453652e-06, "loss": 0.202, "step": 24199 }, { "epoch": 150.31152647975077, "grad_norm": 1.828125, "learning_rate": 7.37729306800842e-06, "loss": 0.1376, "step": 24200 }, { "epoch": 150.3177570093458, "grad_norm": 0.890625, "learning_rate": 7.375545610718448e-06, "loss": 0.0801, "step": 24201 }, { "epoch": 150.3239875389408, "grad_norm": 3.265625, "learning_rate": 7.3737983246006945e-06, "loss": 0.2948, "step": 24202 }, { "epoch": 150.33021806853583, "grad_norm": 1.7890625, "learning_rate": 7.372051209672143e-06, "loss": 0.1279, "step": 24203 }, { "epoch": 150.33644859813083, "grad_norm": 2.015625, "learning_rate": 7.370304265949746e-06, "loss": 0.1866, "step": 24204 }, { "epoch": 150.34267912772586, "grad_norm": 1.8203125, "learning_rate": 7.368557493450484e-06, "loss": 0.1184, "step": 24205 }, { "epoch": 150.34890965732086, "grad_norm": 0.9140625, "learning_rate": 7.366810892191311e-06, "loss": 0.0955, "step": 24206 }, { "epoch": 150.3551401869159, "grad_norm": 1.1640625, "learning_rate": 7.365064462189192e-06, "loss": 0.1262, "step": 24207 }, { "epoch": 150.3613707165109, "grad_norm": 2.609375, "learning_rate": 7.363318203461103e-06, "loss": 0.2145, "step": 24208 }, { "epoch": 150.36760124610592, "grad_norm": 2.625, "learning_rate": 7.361572116023985e-06, "loss": 0.1408, "step": 24209 }, { "epoch": 150.37383177570092, "grad_norm": 2.359375, "learning_rate": 7.359826199894812e-06, "loss": 0.1145, "step": 24210 }, { "epoch": 150.38006230529595, "grad_norm": 2.09375, "learning_rate": 7.358080455090524e-06, "loss": 0.1678, "step": 24211 }, { "epoch": 150.38629283489095, "grad_norm": 1.8125, "learning_rate": 7.3563348816280936e-06, "loss": 0.1697, "step": 24212 }, { "epoch": 150.39252336448598, "grad_norm": 2.15625, "learning_rate": 7.354589479524457e-06, "loss": 0.1287, "step": 24213 }, { "epoch": 150.39875389408098, "grad_norm": 1.703125, "learning_rate": 7.352844248796576e-06, "loss": 0.1286, "step": 24214 }, { "epoch": 150.404984423676, "grad_norm": 2.453125, "learning_rate": 7.351099189461408e-06, "loss": 0.1964, "step": 24215 }, { "epoch": 150.41121495327104, "grad_norm": 1.90625, "learning_rate": 7.34935430153589e-06, "loss": 0.1522, "step": 24216 }, { "epoch": 150.41744548286604, "grad_norm": 1.0078125, "learning_rate": 7.3476095850369655e-06, "loss": 0.1054, "step": 24217 }, { "epoch": 150.42367601246107, "grad_norm": 2.765625, "learning_rate": 7.3458650399815835e-06, "loss": 0.2517, "step": 24218 }, { "epoch": 150.42990654205607, "grad_norm": 1.6484375, "learning_rate": 7.3441206663867e-06, "loss": 0.1539, "step": 24219 }, { "epoch": 150.4361370716511, "grad_norm": 1.6875, "learning_rate": 7.342376464269235e-06, "loss": 0.1427, "step": 24220 }, { "epoch": 150.4423676012461, "grad_norm": 1.0625, "learning_rate": 7.340632433646141e-06, "loss": 0.1031, "step": 24221 }, { "epoch": 150.44859813084113, "grad_norm": 1.6953125, "learning_rate": 7.338888574534364e-06, "loss": 0.1314, "step": 24222 }, { "epoch": 150.45482866043614, "grad_norm": 2.1875, "learning_rate": 7.337144886950831e-06, "loss": 0.1805, "step": 24223 }, { "epoch": 150.46105919003116, "grad_norm": 0.86328125, "learning_rate": 7.335401370912468e-06, "loss": 0.071, "step": 24224 }, { "epoch": 150.46728971962617, "grad_norm": 2.34375, "learning_rate": 7.333658026436218e-06, "loss": 0.1974, "step": 24225 }, { "epoch": 150.4735202492212, "grad_norm": 0.88671875, "learning_rate": 7.331914853539021e-06, "loss": 0.0932, "step": 24226 }, { "epoch": 150.4797507788162, "grad_norm": 2.53125, "learning_rate": 7.330171852237791e-06, "loss": 0.2194, "step": 24227 }, { "epoch": 150.48598130841123, "grad_norm": 2.296875, "learning_rate": 7.328429022549466e-06, "loss": 0.2124, "step": 24228 }, { "epoch": 150.49221183800623, "grad_norm": 1.5234375, "learning_rate": 7.326686364490976e-06, "loss": 0.1223, "step": 24229 }, { "epoch": 150.49844236760126, "grad_norm": 2.15625, "learning_rate": 7.324943878079243e-06, "loss": 0.1713, "step": 24230 }, { "epoch": 150.50467289719626, "grad_norm": 1.6953125, "learning_rate": 7.3232015633311775e-06, "loss": 0.1327, "step": 24231 }, { "epoch": 150.5109034267913, "grad_norm": 2.515625, "learning_rate": 7.321459420263715e-06, "loss": 0.1966, "step": 24232 }, { "epoch": 150.5171339563863, "grad_norm": 2.078125, "learning_rate": 7.319717448893771e-06, "loss": 0.1218, "step": 24233 }, { "epoch": 150.52336448598132, "grad_norm": 1.984375, "learning_rate": 7.317975649238273e-06, "loss": 0.1518, "step": 24234 }, { "epoch": 150.52959501557632, "grad_norm": 2.875, "learning_rate": 7.316234021314131e-06, "loss": 0.2401, "step": 24235 }, { "epoch": 150.53582554517135, "grad_norm": 1.2890625, "learning_rate": 7.314492565138251e-06, "loss": 0.0982, "step": 24236 }, { "epoch": 150.54205607476635, "grad_norm": 1.53125, "learning_rate": 7.312751280727553e-06, "loss": 0.1298, "step": 24237 }, { "epoch": 150.54828660436138, "grad_norm": 1.5078125, "learning_rate": 7.3110101680989584e-06, "loss": 0.0934, "step": 24238 }, { "epoch": 150.55451713395638, "grad_norm": 1.7734375, "learning_rate": 7.309269227269363e-06, "loss": 0.117, "step": 24239 }, { "epoch": 150.5607476635514, "grad_norm": 1.4140625, "learning_rate": 7.30752845825568e-06, "loss": 0.1308, "step": 24240 }, { "epoch": 150.5669781931464, "grad_norm": 2.6875, "learning_rate": 7.3057878610748255e-06, "loss": 0.1801, "step": 24241 }, { "epoch": 150.57320872274144, "grad_norm": 2.125, "learning_rate": 7.304047435743697e-06, "loss": 0.1556, "step": 24242 }, { "epoch": 150.57943925233644, "grad_norm": 1.0859375, "learning_rate": 7.302307182279186e-06, "loss": 0.0876, "step": 24243 }, { "epoch": 150.58566978193147, "grad_norm": 3.03125, "learning_rate": 7.300567100698208e-06, "loss": 0.2286, "step": 24244 }, { "epoch": 150.59190031152647, "grad_norm": 1.75, "learning_rate": 7.298827191017668e-06, "loss": 0.1035, "step": 24245 }, { "epoch": 150.5981308411215, "grad_norm": 1.6953125, "learning_rate": 7.297087453254448e-06, "loss": 0.1181, "step": 24246 }, { "epoch": 150.6043613707165, "grad_norm": 2.8125, "learning_rate": 7.295347887425455e-06, "loss": 0.3185, "step": 24247 }, { "epoch": 150.61059190031153, "grad_norm": 1.4453125, "learning_rate": 7.293608493547591e-06, "loss": 0.1033, "step": 24248 }, { "epoch": 150.61682242990653, "grad_norm": 1.484375, "learning_rate": 7.291869271637738e-06, "loss": 0.1273, "step": 24249 }, { "epoch": 150.62305295950156, "grad_norm": 2.53125, "learning_rate": 7.290130221712782e-06, "loss": 0.2392, "step": 24250 }, { "epoch": 150.62928348909657, "grad_norm": 1.734375, "learning_rate": 7.2883913437896244e-06, "loss": 0.1134, "step": 24251 }, { "epoch": 150.6355140186916, "grad_norm": 1.078125, "learning_rate": 7.286652637885158e-06, "loss": 0.0992, "step": 24252 }, { "epoch": 150.6417445482866, "grad_norm": 0.78125, "learning_rate": 7.284914104016252e-06, "loss": 0.0926, "step": 24253 }, { "epoch": 150.64797507788163, "grad_norm": 1.90625, "learning_rate": 7.28317574219981e-06, "loss": 0.1463, "step": 24254 }, { "epoch": 150.65420560747663, "grad_norm": 1.1875, "learning_rate": 7.2814375524526994e-06, "loss": 0.11, "step": 24255 }, { "epoch": 150.66043613707166, "grad_norm": 2.78125, "learning_rate": 7.2796995347918176e-06, "loss": 0.2199, "step": 24256 }, { "epoch": 150.66666666666666, "grad_norm": 1.7578125, "learning_rate": 7.277961689234028e-06, "loss": 0.1205, "step": 24257 }, { "epoch": 150.6728971962617, "grad_norm": 3.53125, "learning_rate": 7.2762240157962155e-06, "loss": 0.3527, "step": 24258 }, { "epoch": 150.6791277258567, "grad_norm": 2.484375, "learning_rate": 7.274486514495268e-06, "loss": 0.1823, "step": 24259 }, { "epoch": 150.68535825545172, "grad_norm": 2.640625, "learning_rate": 7.272749185348043e-06, "loss": 0.2079, "step": 24260 }, { "epoch": 150.69158878504672, "grad_norm": 1.5703125, "learning_rate": 7.271012028371427e-06, "loss": 0.1324, "step": 24261 }, { "epoch": 150.69781931464175, "grad_norm": 1.875, "learning_rate": 7.2692750435822795e-06, "loss": 0.1618, "step": 24262 }, { "epoch": 150.70404984423675, "grad_norm": 2.984375, "learning_rate": 7.267538230997487e-06, "loss": 0.2973, "step": 24263 }, { "epoch": 150.71028037383178, "grad_norm": 1.296875, "learning_rate": 7.2658015906338974e-06, "loss": 0.1225, "step": 24264 }, { "epoch": 150.71651090342678, "grad_norm": 1.6640625, "learning_rate": 7.26406512250839e-06, "loss": 0.1262, "step": 24265 }, { "epoch": 150.7227414330218, "grad_norm": 1.78125, "learning_rate": 7.262328826637826e-06, "loss": 0.2331, "step": 24266 }, { "epoch": 150.7289719626168, "grad_norm": 1.2265625, "learning_rate": 7.2605927030390845e-06, "loss": 0.103, "step": 24267 }, { "epoch": 150.73520249221184, "grad_norm": 2.5625, "learning_rate": 7.2588567517289985e-06, "loss": 0.1461, "step": 24268 }, { "epoch": 150.74143302180684, "grad_norm": 2.234375, "learning_rate": 7.257120972724443e-06, "loss": 0.1857, "step": 24269 }, { "epoch": 150.74766355140187, "grad_norm": 1.578125, "learning_rate": 7.255385366042275e-06, "loss": 0.143, "step": 24270 }, { "epoch": 150.75389408099687, "grad_norm": 2.59375, "learning_rate": 7.253649931699358e-06, "loss": 0.2309, "step": 24271 }, { "epoch": 150.7601246105919, "grad_norm": 3.375, "learning_rate": 7.2519146697125315e-06, "loss": 0.1194, "step": 24272 }, { "epoch": 150.7663551401869, "grad_norm": 1.375, "learning_rate": 7.250179580098668e-06, "loss": 0.1237, "step": 24273 }, { "epoch": 150.77258566978193, "grad_norm": 1.953125, "learning_rate": 7.248444662874598e-06, "loss": 0.1642, "step": 24274 }, { "epoch": 150.77881619937693, "grad_norm": 1.4609375, "learning_rate": 7.246709918057193e-06, "loss": 0.1284, "step": 24275 }, { "epoch": 150.78504672897196, "grad_norm": 1.203125, "learning_rate": 7.244975345663279e-06, "loss": 0.0943, "step": 24276 }, { "epoch": 150.79127725856696, "grad_norm": 2.421875, "learning_rate": 7.243240945709717e-06, "loss": 0.1423, "step": 24277 }, { "epoch": 150.797507788162, "grad_norm": 1.375, "learning_rate": 7.241506718213353e-06, "loss": 0.1434, "step": 24278 }, { "epoch": 150.80373831775702, "grad_norm": 1.40625, "learning_rate": 7.239772663191021e-06, "loss": 0.118, "step": 24279 }, { "epoch": 150.80996884735202, "grad_norm": 1.4375, "learning_rate": 7.238038780659573e-06, "loss": 0.1259, "step": 24280 }, { "epoch": 150.81619937694705, "grad_norm": 1.59375, "learning_rate": 7.236305070635835e-06, "loss": 0.1375, "step": 24281 }, { "epoch": 150.82242990654206, "grad_norm": 3.28125, "learning_rate": 7.234571533136661e-06, "loss": 0.1944, "step": 24282 }, { "epoch": 150.82866043613708, "grad_norm": 2.46875, "learning_rate": 7.232838168178871e-06, "loss": 0.2416, "step": 24283 }, { "epoch": 150.8348909657321, "grad_norm": 2.890625, "learning_rate": 7.23110497577931e-06, "loss": 0.1411, "step": 24284 }, { "epoch": 150.84112149532712, "grad_norm": 2.0625, "learning_rate": 7.229371955954817e-06, "loss": 0.1061, "step": 24285 }, { "epoch": 150.84735202492212, "grad_norm": 1.6484375, "learning_rate": 7.227639108722214e-06, "loss": 0.1385, "step": 24286 }, { "epoch": 150.85358255451715, "grad_norm": 1.796875, "learning_rate": 7.225906434098326e-06, "loss": 0.1145, "step": 24287 }, { "epoch": 150.85981308411215, "grad_norm": 1.265625, "learning_rate": 7.224173932099986e-06, "loss": 0.1215, "step": 24288 }, { "epoch": 150.86604361370718, "grad_norm": 2.96875, "learning_rate": 7.222441602744034e-06, "loss": 0.3565, "step": 24289 }, { "epoch": 150.87227414330218, "grad_norm": 1.7578125, "learning_rate": 7.220709446047272e-06, "loss": 0.1324, "step": 24290 }, { "epoch": 150.8785046728972, "grad_norm": 1.9375, "learning_rate": 7.2189774620265354e-06, "loss": 0.1693, "step": 24291 }, { "epoch": 150.8847352024922, "grad_norm": 2.4375, "learning_rate": 7.217245650698651e-06, "loss": 0.1983, "step": 24292 }, { "epoch": 150.89096573208724, "grad_norm": 1.2265625, "learning_rate": 7.215514012080429e-06, "loss": 0.0837, "step": 24293 }, { "epoch": 150.89719626168224, "grad_norm": 1.640625, "learning_rate": 7.213782546188685e-06, "loss": 0.1324, "step": 24294 }, { "epoch": 150.90342679127727, "grad_norm": 1.7734375, "learning_rate": 7.212051253040242e-06, "loss": 0.1381, "step": 24295 }, { "epoch": 150.90965732087227, "grad_norm": 2.109375, "learning_rate": 7.21032013265191e-06, "loss": 0.19, "step": 24296 }, { "epoch": 150.9158878504673, "grad_norm": 1.7578125, "learning_rate": 7.208589185040513e-06, "loss": 0.1192, "step": 24297 }, { "epoch": 150.9221183800623, "grad_norm": 2.15625, "learning_rate": 7.206858410222847e-06, "loss": 0.22, "step": 24298 }, { "epoch": 150.92834890965733, "grad_norm": 1.1796875, "learning_rate": 7.205127808215737e-06, "loss": 0.0896, "step": 24299 }, { "epoch": 150.93457943925233, "grad_norm": 1.7578125, "learning_rate": 7.2033973790359755e-06, "loss": 0.1402, "step": 24300 }, { "epoch": 150.94080996884736, "grad_norm": 1.59375, "learning_rate": 7.201667122700381e-06, "loss": 0.1316, "step": 24301 }, { "epoch": 150.94704049844236, "grad_norm": 2.015625, "learning_rate": 7.199937039225748e-06, "loss": 0.2197, "step": 24302 }, { "epoch": 150.9532710280374, "grad_norm": 2.84375, "learning_rate": 7.198207128628887e-06, "loss": 0.177, "step": 24303 }, { "epoch": 150.9595015576324, "grad_norm": 1.6484375, "learning_rate": 7.196477390926601e-06, "loss": 0.1381, "step": 24304 }, { "epoch": 150.96573208722742, "grad_norm": 2.265625, "learning_rate": 7.194747826135687e-06, "loss": 0.1545, "step": 24305 }, { "epoch": 150.97196261682242, "grad_norm": 1.765625, "learning_rate": 7.193018434272933e-06, "loss": 0.1313, "step": 24306 }, { "epoch": 150.97819314641745, "grad_norm": 2.6875, "learning_rate": 7.191289215355143e-06, "loss": 0.2279, "step": 24307 }, { "epoch": 150.98442367601245, "grad_norm": 1.6796875, "learning_rate": 7.189560169399121e-06, "loss": 0.1467, "step": 24308 }, { "epoch": 150.99065420560748, "grad_norm": 2.078125, "learning_rate": 7.18783129642164e-06, "loss": 0.1672, "step": 24309 }, { "epoch": 150.99688473520249, "grad_norm": 3.234375, "learning_rate": 7.186102596439506e-06, "loss": 0.3131, "step": 24310 }, { "epoch": 151.0, "grad_norm": 0.54296875, "learning_rate": 7.1843740694695096e-06, "loss": 0.0529, "step": 24311 }, { "epoch": 151.00623052959503, "grad_norm": 3.0, "learning_rate": 7.1826457155284355e-06, "loss": 0.1871, "step": 24312 }, { "epoch": 151.01246105919003, "grad_norm": 1.359375, "learning_rate": 7.1809175346330585e-06, "loss": 0.1149, "step": 24313 }, { "epoch": 151.01869158878506, "grad_norm": 3.125, "learning_rate": 7.179189526800173e-06, "loss": 0.2862, "step": 24314 }, { "epoch": 151.02492211838006, "grad_norm": 2.546875, "learning_rate": 7.177461692046569e-06, "loss": 0.1491, "step": 24315 }, { "epoch": 151.0311526479751, "grad_norm": 1.5625, "learning_rate": 7.1757340303890115e-06, "loss": 0.1189, "step": 24316 }, { "epoch": 151.0373831775701, "grad_norm": 1.890625, "learning_rate": 7.174006541844289e-06, "loss": 0.1238, "step": 24317 }, { "epoch": 151.04361370716512, "grad_norm": 2.6875, "learning_rate": 7.1722792264291846e-06, "loss": 0.2601, "step": 24318 }, { "epoch": 151.04984423676012, "grad_norm": 2.875, "learning_rate": 7.17055208416047e-06, "loss": 0.1814, "step": 24319 }, { "epoch": 151.05607476635515, "grad_norm": 1.3203125, "learning_rate": 7.1688251150549085e-06, "loss": 0.1202, "step": 24320 }, { "epoch": 151.06230529595015, "grad_norm": 1.7109375, "learning_rate": 7.167098319129281e-06, "loss": 0.1266, "step": 24321 }, { "epoch": 151.06853582554518, "grad_norm": 2.09375, "learning_rate": 7.1653716964003695e-06, "loss": 0.2629, "step": 24322 }, { "epoch": 151.07476635514018, "grad_norm": 1.578125, "learning_rate": 7.163645246884923e-06, "loss": 0.1305, "step": 24323 }, { "epoch": 151.0809968847352, "grad_norm": 1.2421875, "learning_rate": 7.161918970599729e-06, "loss": 0.1038, "step": 24324 }, { "epoch": 151.08722741433021, "grad_norm": 2.359375, "learning_rate": 7.1601928675615355e-06, "loss": 0.1678, "step": 24325 }, { "epoch": 151.09345794392524, "grad_norm": 1.5859375, "learning_rate": 7.15846693778712e-06, "loss": 0.1124, "step": 24326 }, { "epoch": 151.09968847352025, "grad_norm": 2.390625, "learning_rate": 7.156741181293236e-06, "loss": 0.1739, "step": 24327 }, { "epoch": 151.10591900311528, "grad_norm": 1.9765625, "learning_rate": 7.155015598096648e-06, "loss": 0.1653, "step": 24328 }, { "epoch": 151.11214953271028, "grad_norm": 2.28125, "learning_rate": 7.153290188214115e-06, "loss": 0.1636, "step": 24329 }, { "epoch": 151.1183800623053, "grad_norm": 2.65625, "learning_rate": 7.151564951662401e-06, "loss": 0.23, "step": 24330 }, { "epoch": 151.1246105919003, "grad_norm": 1.390625, "learning_rate": 7.149839888458257e-06, "loss": 0.1236, "step": 24331 }, { "epoch": 151.13084112149534, "grad_norm": 2.375, "learning_rate": 7.1481149986184295e-06, "loss": 0.2237, "step": 24332 }, { "epoch": 151.13707165109034, "grad_norm": 1.0390625, "learning_rate": 7.146390282159679e-06, "loss": 0.1115, "step": 24333 }, { "epoch": 151.14330218068537, "grad_norm": 1.7109375, "learning_rate": 7.1446657390987605e-06, "loss": 0.1159, "step": 24334 }, { "epoch": 151.14953271028037, "grad_norm": 1.2890625, "learning_rate": 7.142941369452411e-06, "loss": 0.0943, "step": 24335 }, { "epoch": 151.1557632398754, "grad_norm": 2.546875, "learning_rate": 7.1412171732373825e-06, "loss": 0.1787, "step": 24336 }, { "epoch": 151.1619937694704, "grad_norm": 1.96875, "learning_rate": 7.13949315047043e-06, "loss": 0.149, "step": 24337 }, { "epoch": 151.16822429906543, "grad_norm": 1.578125, "learning_rate": 7.137769301168293e-06, "loss": 0.1366, "step": 24338 }, { "epoch": 151.17445482866043, "grad_norm": 3.15625, "learning_rate": 7.136045625347703e-06, "loss": 0.1411, "step": 24339 }, { "epoch": 151.18068535825546, "grad_norm": 2.296875, "learning_rate": 7.134322123025408e-06, "loss": 0.2671, "step": 24340 }, { "epoch": 151.18691588785046, "grad_norm": 0.84765625, "learning_rate": 7.1325987942181565e-06, "loss": 0.0998, "step": 24341 }, { "epoch": 151.1931464174455, "grad_norm": 1.9765625, "learning_rate": 7.130875638942669e-06, "loss": 0.205, "step": 24342 }, { "epoch": 151.1993769470405, "grad_norm": 1.1953125, "learning_rate": 7.129152657215696e-06, "loss": 0.1098, "step": 24343 }, { "epoch": 151.20560747663552, "grad_norm": 2.671875, "learning_rate": 7.127429849053957e-06, "loss": 0.1293, "step": 24344 }, { "epoch": 151.21183800623052, "grad_norm": 1.421875, "learning_rate": 7.125707214474201e-06, "loss": 0.1315, "step": 24345 }, { "epoch": 151.21806853582555, "grad_norm": 1.3046875, "learning_rate": 7.12398475349314e-06, "loss": 0.1305, "step": 24346 }, { "epoch": 151.22429906542055, "grad_norm": 2.859375, "learning_rate": 7.122262466127514e-06, "loss": 0.3246, "step": 24347 }, { "epoch": 151.23052959501558, "grad_norm": 1.84375, "learning_rate": 7.120540352394056e-06, "loss": 0.1299, "step": 24348 }, { "epoch": 151.23676012461058, "grad_norm": 1.65625, "learning_rate": 7.118818412309478e-06, "loss": 0.1365, "step": 24349 }, { "epoch": 151.2429906542056, "grad_norm": 1.7109375, "learning_rate": 7.117096645890514e-06, "loss": 0.1727, "step": 24350 }, { "epoch": 151.2492211838006, "grad_norm": 1.875, "learning_rate": 7.115375053153878e-06, "loss": 0.1148, "step": 24351 }, { "epoch": 151.25545171339564, "grad_norm": 1.7578125, "learning_rate": 7.113653634116299e-06, "loss": 0.1732, "step": 24352 }, { "epoch": 151.26168224299064, "grad_norm": 2.40625, "learning_rate": 7.1119323887944845e-06, "loss": 0.2128, "step": 24353 }, { "epoch": 151.26791277258567, "grad_norm": 2.1875, "learning_rate": 7.110211317205159e-06, "loss": 0.1743, "step": 24354 }, { "epoch": 151.27414330218068, "grad_norm": 1.765625, "learning_rate": 7.1084904193650465e-06, "loss": 0.0953, "step": 24355 }, { "epoch": 151.2803738317757, "grad_norm": 2.46875, "learning_rate": 7.106769695290849e-06, "loss": 0.1502, "step": 24356 }, { "epoch": 151.2866043613707, "grad_norm": 1.15625, "learning_rate": 7.105049144999276e-06, "loss": 0.1137, "step": 24357 }, { "epoch": 151.29283489096574, "grad_norm": 3.515625, "learning_rate": 7.103328768507039e-06, "loss": 0.3426, "step": 24358 }, { "epoch": 151.29906542056074, "grad_norm": 1.34375, "learning_rate": 7.101608565830861e-06, "loss": 0.1317, "step": 24359 }, { "epoch": 151.30529595015577, "grad_norm": 1.78125, "learning_rate": 7.0998885369874315e-06, "loss": 0.098, "step": 24360 }, { "epoch": 151.31152647975077, "grad_norm": 2.140625, "learning_rate": 7.0981686819934615e-06, "loss": 0.2299, "step": 24361 }, { "epoch": 151.3177570093458, "grad_norm": 3.53125, "learning_rate": 7.096449000865663e-06, "loss": 0.1798, "step": 24362 }, { "epoch": 151.3239875389408, "grad_norm": 2.390625, "learning_rate": 7.0947294936207305e-06, "loss": 0.2266, "step": 24363 }, { "epoch": 151.33021806853583, "grad_norm": 2.296875, "learning_rate": 7.093010160275357e-06, "loss": 0.2131, "step": 24364 }, { "epoch": 151.33644859813083, "grad_norm": 2.390625, "learning_rate": 7.091291000846248e-06, "loss": 0.2294, "step": 24365 }, { "epoch": 151.34267912772586, "grad_norm": 2.390625, "learning_rate": 7.089572015350101e-06, "loss": 0.1927, "step": 24366 }, { "epoch": 151.34890965732086, "grad_norm": 2.90625, "learning_rate": 7.087853203803618e-06, "loss": 0.2766, "step": 24367 }, { "epoch": 151.3551401869159, "grad_norm": 1.3984375, "learning_rate": 7.086134566223479e-06, "loss": 0.1228, "step": 24368 }, { "epoch": 151.3613707165109, "grad_norm": 1.515625, "learning_rate": 7.08441610262639e-06, "loss": 0.098, "step": 24369 }, { "epoch": 151.36760124610592, "grad_norm": 1.0078125, "learning_rate": 7.082697813029024e-06, "loss": 0.1119, "step": 24370 }, { "epoch": 151.37383177570092, "grad_norm": 2.65625, "learning_rate": 7.080979697448084e-06, "loss": 0.1716, "step": 24371 }, { "epoch": 151.38006230529595, "grad_norm": 2.390625, "learning_rate": 7.079261755900246e-06, "loss": 0.1299, "step": 24372 }, { "epoch": 151.38629283489095, "grad_norm": 1.8125, "learning_rate": 7.077543988402202e-06, "loss": 0.0858, "step": 24373 }, { "epoch": 151.39252336448598, "grad_norm": 1.640625, "learning_rate": 7.075826394970639e-06, "loss": 0.1538, "step": 24374 }, { "epoch": 151.39875389408098, "grad_norm": 2.140625, "learning_rate": 7.074108975622232e-06, "loss": 0.1405, "step": 24375 }, { "epoch": 151.404984423676, "grad_norm": 1.7265625, "learning_rate": 7.072391730373656e-06, "loss": 0.1399, "step": 24376 }, { "epoch": 151.41121495327104, "grad_norm": 1.4765625, "learning_rate": 7.070674659241594e-06, "loss": 0.2122, "step": 24377 }, { "epoch": 151.41744548286604, "grad_norm": 3.359375, "learning_rate": 7.068957762242731e-06, "loss": 0.2512, "step": 24378 }, { "epoch": 151.42367601246107, "grad_norm": 2.453125, "learning_rate": 7.067241039393729e-06, "loss": 0.3167, "step": 24379 }, { "epoch": 151.42990654205607, "grad_norm": 2.71875, "learning_rate": 7.065524490711265e-06, "loss": 0.3546, "step": 24380 }, { "epoch": 151.4361370716511, "grad_norm": 1.515625, "learning_rate": 7.06380811621202e-06, "loss": 0.1508, "step": 24381 }, { "epoch": 151.4423676012461, "grad_norm": 2.5, "learning_rate": 7.062091915912655e-06, "loss": 0.1689, "step": 24382 }, { "epoch": 151.44859813084113, "grad_norm": 2.546875, "learning_rate": 7.060375889829834e-06, "loss": 0.1894, "step": 24383 }, { "epoch": 151.45482866043614, "grad_norm": 1.53125, "learning_rate": 7.058660037980228e-06, "loss": 0.1482, "step": 24384 }, { "epoch": 151.46105919003116, "grad_norm": 2.234375, "learning_rate": 7.056944360380508e-06, "loss": 0.1115, "step": 24385 }, { "epoch": 151.46728971962617, "grad_norm": 2.546875, "learning_rate": 7.055228857047324e-06, "loss": 0.1672, "step": 24386 }, { "epoch": 151.4735202492212, "grad_norm": 1.6171875, "learning_rate": 7.053513527997344e-06, "loss": 0.1337, "step": 24387 }, { "epoch": 151.4797507788162, "grad_norm": 2.75, "learning_rate": 7.051798373247237e-06, "loss": 0.2153, "step": 24388 }, { "epoch": 151.48598130841123, "grad_norm": 3.359375, "learning_rate": 7.05008339281365e-06, "loss": 0.1038, "step": 24389 }, { "epoch": 151.49221183800623, "grad_norm": 2.3125, "learning_rate": 7.048368586713236e-06, "loss": 0.1387, "step": 24390 }, { "epoch": 151.49844236760126, "grad_norm": 1.4140625, "learning_rate": 7.046653954962654e-06, "loss": 0.1578, "step": 24391 }, { "epoch": 151.50467289719626, "grad_norm": 2.25, "learning_rate": 7.044939497578556e-06, "loss": 0.1822, "step": 24392 }, { "epoch": 151.5109034267913, "grad_norm": 1.265625, "learning_rate": 7.0432252145776036e-06, "loss": 0.1376, "step": 24393 }, { "epoch": 151.5171339563863, "grad_norm": 1.28125, "learning_rate": 7.0415111059764395e-06, "loss": 0.1048, "step": 24394 }, { "epoch": 151.52336448598132, "grad_norm": 3.28125, "learning_rate": 7.039797171791698e-06, "loss": 0.2545, "step": 24395 }, { "epoch": 151.52959501557632, "grad_norm": 1.7109375, "learning_rate": 7.038083412040039e-06, "loss": 0.1029, "step": 24396 }, { "epoch": 151.53582554517135, "grad_norm": 2.734375, "learning_rate": 7.036369826738109e-06, "loss": 0.2032, "step": 24397 }, { "epoch": 151.54205607476635, "grad_norm": 1.5234375, "learning_rate": 7.034656415902544e-06, "loss": 0.1287, "step": 24398 }, { "epoch": 151.54828660436138, "grad_norm": 3.109375, "learning_rate": 7.032943179549984e-06, "loss": 0.2041, "step": 24399 }, { "epoch": 151.55451713395638, "grad_norm": 3.046875, "learning_rate": 7.031230117697077e-06, "loss": 0.2388, "step": 24400 }, { "epoch": 151.5607476635514, "grad_norm": 1.0859375, "learning_rate": 7.029517230360458e-06, "loss": 0.1208, "step": 24401 }, { "epoch": 151.5669781931464, "grad_norm": 1.1640625, "learning_rate": 7.027804517556749e-06, "loss": 0.1353, "step": 24402 }, { "epoch": 151.57320872274144, "grad_norm": 1.796875, "learning_rate": 7.0260919793025985e-06, "loss": 0.1385, "step": 24403 }, { "epoch": 151.57943925233644, "grad_norm": 1.84375, "learning_rate": 7.0243796156146404e-06, "loss": 0.1204, "step": 24404 }, { "epoch": 151.58566978193147, "grad_norm": 1.5859375, "learning_rate": 7.022667426509497e-06, "loss": 0.1487, "step": 24405 }, { "epoch": 151.59190031152647, "grad_norm": 2.34375, "learning_rate": 7.020955412003796e-06, "loss": 0.1478, "step": 24406 }, { "epoch": 151.5981308411215, "grad_norm": 2.046875, "learning_rate": 7.0192435721141815e-06, "loss": 0.144, "step": 24407 }, { "epoch": 151.6043613707165, "grad_norm": 2.390625, "learning_rate": 7.0175319068572664e-06, "loss": 0.2078, "step": 24408 }, { "epoch": 151.61059190031153, "grad_norm": 2.4375, "learning_rate": 7.015820416249669e-06, "loss": 0.2044, "step": 24409 }, { "epoch": 151.61682242990653, "grad_norm": 2.34375, "learning_rate": 7.014109100308019e-06, "loss": 0.1778, "step": 24410 }, { "epoch": 151.62305295950156, "grad_norm": 1.9609375, "learning_rate": 7.012397959048944e-06, "loss": 0.1679, "step": 24411 }, { "epoch": 151.62928348909657, "grad_norm": 1.3203125, "learning_rate": 7.010686992489046e-06, "loss": 0.0974, "step": 24412 }, { "epoch": 151.6355140186916, "grad_norm": 2.109375, "learning_rate": 7.0089762006449624e-06, "loss": 0.137, "step": 24413 }, { "epoch": 151.6417445482866, "grad_norm": 2.046875, "learning_rate": 7.00726558353329e-06, "loss": 0.132, "step": 24414 }, { "epoch": 151.64797507788163, "grad_norm": 2.65625, "learning_rate": 7.005555141170655e-06, "loss": 0.1948, "step": 24415 }, { "epoch": 151.65420560747663, "grad_norm": 3.203125, "learning_rate": 7.003844873573662e-06, "loss": 0.1797, "step": 24416 }, { "epoch": 151.66043613707166, "grad_norm": 2.5625, "learning_rate": 7.0021347807589235e-06, "loss": 0.2047, "step": 24417 }, { "epoch": 151.66666666666666, "grad_norm": 2.1875, "learning_rate": 7.000424862743058e-06, "loss": 0.132, "step": 24418 }, { "epoch": 151.6728971962617, "grad_norm": 3.203125, "learning_rate": 6.998715119542657e-06, "loss": 0.2784, "step": 24419 }, { "epoch": 151.6791277258567, "grad_norm": 3.328125, "learning_rate": 6.99700555117434e-06, "loss": 0.3013, "step": 24420 }, { "epoch": 151.68535825545172, "grad_norm": 1.828125, "learning_rate": 6.9952961576546964e-06, "loss": 0.1196, "step": 24421 }, { "epoch": 151.69158878504672, "grad_norm": 2.828125, "learning_rate": 6.9935869390003435e-06, "loss": 0.2748, "step": 24422 }, { "epoch": 151.69781931464175, "grad_norm": 1.0546875, "learning_rate": 6.991877895227864e-06, "loss": 0.1013, "step": 24423 }, { "epoch": 151.70404984423675, "grad_norm": 0.97265625, "learning_rate": 6.990169026353868e-06, "loss": 0.0911, "step": 24424 }, { "epoch": 151.71028037383178, "grad_norm": 2.03125, "learning_rate": 6.98846033239495e-06, "loss": 0.1352, "step": 24425 }, { "epoch": 151.71651090342678, "grad_norm": 1.296875, "learning_rate": 6.986751813367723e-06, "loss": 0.1036, "step": 24426 }, { "epoch": 151.7227414330218, "grad_norm": 2.265625, "learning_rate": 6.985043469288746e-06, "loss": 0.1877, "step": 24427 }, { "epoch": 151.7289719626168, "grad_norm": 1.9609375, "learning_rate": 6.983335300174626e-06, "loss": 0.1467, "step": 24428 }, { "epoch": 151.73520249221184, "grad_norm": 1.453125, "learning_rate": 6.981627306041958e-06, "loss": 0.127, "step": 24429 }, { "epoch": 151.74143302180684, "grad_norm": 1.4609375, "learning_rate": 6.979919486907335e-06, "loss": 0.2164, "step": 24430 }, { "epoch": 151.74766355140187, "grad_norm": 2.0625, "learning_rate": 6.978211842787327e-06, "loss": 0.1031, "step": 24431 }, { "epoch": 151.75389408099687, "grad_norm": 3.046875, "learning_rate": 6.976504373698528e-06, "loss": 0.2734, "step": 24432 }, { "epoch": 151.7601246105919, "grad_norm": 1.8984375, "learning_rate": 6.97479707965753e-06, "loss": 0.1835, "step": 24433 }, { "epoch": 151.7663551401869, "grad_norm": 1.2109375, "learning_rate": 6.973089960680906e-06, "loss": 0.0796, "step": 24434 }, { "epoch": 151.77258566978193, "grad_norm": 2.421875, "learning_rate": 6.971383016785227e-06, "loss": 0.2018, "step": 24435 }, { "epoch": 151.77881619937693, "grad_norm": 1.9375, "learning_rate": 6.969676247987081e-06, "loss": 0.1888, "step": 24436 }, { "epoch": 151.78504672897196, "grad_norm": 3.03125, "learning_rate": 6.96796965430305e-06, "loss": 0.2926, "step": 24437 }, { "epoch": 151.79127725856696, "grad_norm": 2.3125, "learning_rate": 6.9662632357496956e-06, "loss": 0.2195, "step": 24438 }, { "epoch": 151.797507788162, "grad_norm": 1.515625, "learning_rate": 6.964556992343604e-06, "loss": 0.147, "step": 24439 }, { "epoch": 151.80373831775702, "grad_norm": 3.078125, "learning_rate": 6.962850924101336e-06, "loss": 0.1274, "step": 24440 }, { "epoch": 151.80996884735202, "grad_norm": 2.140625, "learning_rate": 6.961145031039471e-06, "loss": 0.1606, "step": 24441 }, { "epoch": 151.81619937694705, "grad_norm": 1.9921875, "learning_rate": 6.959439313174562e-06, "loss": 0.1453, "step": 24442 }, { "epoch": 151.82242990654206, "grad_norm": 3.046875, "learning_rate": 6.957733770523187e-06, "loss": 0.2804, "step": 24443 }, { "epoch": 151.82866043613708, "grad_norm": 2.953125, "learning_rate": 6.956028403101913e-06, "loss": 0.2753, "step": 24444 }, { "epoch": 151.8348909657321, "grad_norm": 1.796875, "learning_rate": 6.954323210927299e-06, "loss": 0.2382, "step": 24445 }, { "epoch": 151.84112149532712, "grad_norm": 0.90625, "learning_rate": 6.952618194015898e-06, "loss": 0.0994, "step": 24446 }, { "epoch": 151.84735202492212, "grad_norm": 1.921875, "learning_rate": 6.950913352384278e-06, "loss": 0.1758, "step": 24447 }, { "epoch": 151.85358255451715, "grad_norm": 1.375, "learning_rate": 6.949208686049002e-06, "loss": 0.1259, "step": 24448 }, { "epoch": 151.85981308411215, "grad_norm": 1.375, "learning_rate": 6.9475041950266104e-06, "loss": 0.0933, "step": 24449 }, { "epoch": 151.86604361370718, "grad_norm": 1.2421875, "learning_rate": 6.94579987933367e-06, "loss": 0.1053, "step": 24450 }, { "epoch": 151.87227414330218, "grad_norm": 2.25, "learning_rate": 6.944095738986736e-06, "loss": 0.2359, "step": 24451 }, { "epoch": 151.8785046728972, "grad_norm": 2.515625, "learning_rate": 6.942391774002352e-06, "loss": 0.1336, "step": 24452 }, { "epoch": 151.8847352024922, "grad_norm": 2.59375, "learning_rate": 6.940687984397065e-06, "loss": 0.1557, "step": 24453 }, { "epoch": 151.89096573208724, "grad_norm": 1.5703125, "learning_rate": 6.938984370187424e-06, "loss": 0.1228, "step": 24454 }, { "epoch": 151.89719626168224, "grad_norm": 1.8359375, "learning_rate": 6.937280931389983e-06, "loss": 0.1399, "step": 24455 }, { "epoch": 151.90342679127727, "grad_norm": 3.078125, "learning_rate": 6.935577668021276e-06, "loss": 0.3262, "step": 24456 }, { "epoch": 151.90965732087227, "grad_norm": 1.328125, "learning_rate": 6.933874580097846e-06, "loss": 0.1086, "step": 24457 }, { "epoch": 151.9158878504673, "grad_norm": 1.515625, "learning_rate": 6.932171667636247e-06, "loss": 0.1487, "step": 24458 }, { "epoch": 151.9221183800623, "grad_norm": 1.203125, "learning_rate": 6.930468930653009e-06, "loss": 0.1458, "step": 24459 }, { "epoch": 151.92834890965733, "grad_norm": 2.171875, "learning_rate": 6.928766369164661e-06, "loss": 0.1647, "step": 24460 }, { "epoch": 151.93457943925233, "grad_norm": 0.80859375, "learning_rate": 6.927063983187745e-06, "loss": 0.0985, "step": 24461 }, { "epoch": 151.94080996884736, "grad_norm": 1.765625, "learning_rate": 6.925361772738798e-06, "loss": 0.1514, "step": 24462 }, { "epoch": 151.94704049844236, "grad_norm": 1.46875, "learning_rate": 6.9236597378343565e-06, "loss": 0.0888, "step": 24463 }, { "epoch": 151.9532710280374, "grad_norm": 2.046875, "learning_rate": 6.921957878490948e-06, "loss": 0.1532, "step": 24464 }, { "epoch": 151.9595015576324, "grad_norm": 1.859375, "learning_rate": 6.920256194725086e-06, "loss": 0.1324, "step": 24465 }, { "epoch": 151.96573208722742, "grad_norm": 1.7421875, "learning_rate": 6.9185546865533136e-06, "loss": 0.1575, "step": 24466 }, { "epoch": 151.97196261682242, "grad_norm": 2.15625, "learning_rate": 6.916853353992159e-06, "loss": 0.1519, "step": 24467 }, { "epoch": 151.97819314641745, "grad_norm": 1.03125, "learning_rate": 6.915152197058133e-06, "loss": 0.0966, "step": 24468 }, { "epoch": 151.98442367601245, "grad_norm": 2.09375, "learning_rate": 6.9134512157677605e-06, "loss": 0.1931, "step": 24469 }, { "epoch": 151.99065420560748, "grad_norm": 2.421875, "learning_rate": 6.911750410137577e-06, "loss": 0.2143, "step": 24470 }, { "epoch": 151.99688473520249, "grad_norm": 1.3203125, "learning_rate": 6.910049780184086e-06, "loss": 0.1196, "step": 24471 }, { "epoch": 152.0, "grad_norm": 2.328125, "learning_rate": 6.908349325923802e-06, "loss": 0.1668, "step": 24472 }, { "epoch": 152.00623052959503, "grad_norm": 2.53125, "learning_rate": 6.906649047373246e-06, "loss": 0.2419, "step": 24473 }, { "epoch": 152.01246105919003, "grad_norm": 2.203125, "learning_rate": 6.9049489445489394e-06, "loss": 0.1046, "step": 24474 }, { "epoch": 152.01869158878506, "grad_norm": 1.9765625, "learning_rate": 6.9032490174673755e-06, "loss": 0.1292, "step": 24475 }, { "epoch": 152.02492211838006, "grad_norm": 1.328125, "learning_rate": 6.901549266145074e-06, "loss": 0.1274, "step": 24476 }, { "epoch": 152.0311526479751, "grad_norm": 1.875, "learning_rate": 6.899849690598556e-06, "loss": 0.1496, "step": 24477 }, { "epoch": 152.0373831775701, "grad_norm": 2.0625, "learning_rate": 6.898150290844313e-06, "loss": 0.1597, "step": 24478 }, { "epoch": 152.04361370716512, "grad_norm": 2.171875, "learning_rate": 6.896451066898846e-06, "loss": 0.0997, "step": 24479 }, { "epoch": 152.04984423676012, "grad_norm": 1.5, "learning_rate": 6.894752018778664e-06, "loss": 0.1537, "step": 24480 }, { "epoch": 152.05607476635515, "grad_norm": 2.96875, "learning_rate": 6.893053146500278e-06, "loss": 0.3372, "step": 24481 }, { "epoch": 152.06230529595015, "grad_norm": 1.0078125, "learning_rate": 6.89135445008017e-06, "loss": 0.075, "step": 24482 }, { "epoch": 152.06853582554518, "grad_norm": 2.359375, "learning_rate": 6.889655929534858e-06, "loss": 0.2023, "step": 24483 }, { "epoch": 152.07476635514018, "grad_norm": 1.453125, "learning_rate": 6.887957584880819e-06, "loss": 0.1478, "step": 24484 }, { "epoch": 152.0809968847352, "grad_norm": 2.265625, "learning_rate": 6.886259416134564e-06, "loss": 0.2521, "step": 24485 }, { "epoch": 152.08722741433021, "grad_norm": 1.8203125, "learning_rate": 6.884561423312569e-06, "loss": 0.1319, "step": 24486 }, { "epoch": 152.09345794392524, "grad_norm": 1.7890625, "learning_rate": 6.882863606431336e-06, "loss": 0.151, "step": 24487 }, { "epoch": 152.09968847352025, "grad_norm": 1.015625, "learning_rate": 6.881165965507355e-06, "loss": 0.0842, "step": 24488 }, { "epoch": 152.10591900311528, "grad_norm": 3.75, "learning_rate": 6.879468500557118e-06, "loss": 0.4202, "step": 24489 }, { "epoch": 152.11214953271028, "grad_norm": 3.453125, "learning_rate": 6.877771211597106e-06, "loss": 0.3568, "step": 24490 }, { "epoch": 152.1183800623053, "grad_norm": 1.375, "learning_rate": 6.8760740986437924e-06, "loss": 0.075, "step": 24491 }, { "epoch": 152.1246105919003, "grad_norm": 1.875, "learning_rate": 6.874377161713671e-06, "loss": 0.1544, "step": 24492 }, { "epoch": 152.13084112149534, "grad_norm": 2.390625, "learning_rate": 6.872680400823231e-06, "loss": 0.1845, "step": 24493 }, { "epoch": 152.13707165109034, "grad_norm": 1.7578125, "learning_rate": 6.870983815988935e-06, "loss": 0.1251, "step": 24494 }, { "epoch": 152.14330218068537, "grad_norm": 3.6875, "learning_rate": 6.869287407227268e-06, "loss": 0.3666, "step": 24495 }, { "epoch": 152.14953271028037, "grad_norm": 2.25, "learning_rate": 6.867591174554713e-06, "loss": 0.2122, "step": 24496 }, { "epoch": 152.1557632398754, "grad_norm": 2.171875, "learning_rate": 6.865895117987736e-06, "loss": 0.1712, "step": 24497 }, { "epoch": 152.1619937694704, "grad_norm": 1.5625, "learning_rate": 6.864199237542804e-06, "loss": 0.1193, "step": 24498 }, { "epoch": 152.16822429906543, "grad_norm": 1.6171875, "learning_rate": 6.862503533236395e-06, "loss": 0.1256, "step": 24499 }, { "epoch": 152.17445482866043, "grad_norm": 2.625, "learning_rate": 6.860808005084984e-06, "loss": 0.2179, "step": 24500 }, { "epoch": 152.18068535825546, "grad_norm": 2.046875, "learning_rate": 6.859112653105024e-06, "loss": 0.2032, "step": 24501 }, { "epoch": 152.18691588785046, "grad_norm": 1.6640625, "learning_rate": 6.857417477312988e-06, "loss": 0.1654, "step": 24502 }, { "epoch": 152.1931464174455, "grad_norm": 2.5625, "learning_rate": 6.855722477725349e-06, "loss": 0.228, "step": 24503 }, { "epoch": 152.1993769470405, "grad_norm": 1.46875, "learning_rate": 6.854027654358558e-06, "loss": 0.1227, "step": 24504 }, { "epoch": 152.20560747663552, "grad_norm": 1.7890625, "learning_rate": 6.852333007229073e-06, "loss": 0.1825, "step": 24505 }, { "epoch": 152.21183800623052, "grad_norm": 1.40625, "learning_rate": 6.850638536353354e-06, "loss": 0.1111, "step": 24506 }, { "epoch": 152.21806853582555, "grad_norm": 1.3984375, "learning_rate": 6.848944241747873e-06, "loss": 0.1017, "step": 24507 }, { "epoch": 152.22429906542055, "grad_norm": 1.2421875, "learning_rate": 6.847250123429064e-06, "loss": 0.134, "step": 24508 }, { "epoch": 152.23052959501558, "grad_norm": 0.90625, "learning_rate": 6.845556181413398e-06, "loss": 0.0895, "step": 24509 }, { "epoch": 152.23676012461058, "grad_norm": 3.296875, "learning_rate": 6.843862415717311e-06, "loss": 0.318, "step": 24510 }, { "epoch": 152.2429906542056, "grad_norm": 1.1328125, "learning_rate": 6.842168826357273e-06, "loss": 0.1158, "step": 24511 }, { "epoch": 152.2492211838006, "grad_norm": 2.765625, "learning_rate": 6.840475413349709e-06, "loss": 0.1974, "step": 24512 }, { "epoch": 152.25545171339564, "grad_norm": 3.484375, "learning_rate": 6.838782176711081e-06, "loss": 0.4118, "step": 24513 }, { "epoch": 152.26168224299064, "grad_norm": 2.015625, "learning_rate": 6.8370891164578375e-06, "loss": 0.1372, "step": 24514 }, { "epoch": 152.26791277258567, "grad_norm": 1.28125, "learning_rate": 6.835396232606414e-06, "loss": 0.1171, "step": 24515 }, { "epoch": 152.27414330218068, "grad_norm": 1.71875, "learning_rate": 6.8337035251732465e-06, "loss": 0.085, "step": 24516 }, { "epoch": 152.2803738317757, "grad_norm": 2.28125, "learning_rate": 6.832010994174781e-06, "loss": 0.1907, "step": 24517 }, { "epoch": 152.2866043613707, "grad_norm": 2.53125, "learning_rate": 6.830318639627464e-06, "loss": 0.1221, "step": 24518 }, { "epoch": 152.29283489096574, "grad_norm": 1.6875, "learning_rate": 6.828626461547719e-06, "loss": 0.1651, "step": 24519 }, { "epoch": 152.29906542056074, "grad_norm": 1.875, "learning_rate": 6.826934459951983e-06, "loss": 0.1258, "step": 24520 }, { "epoch": 152.30529595015577, "grad_norm": 1.8984375, "learning_rate": 6.825242634856693e-06, "loss": 0.1647, "step": 24521 }, { "epoch": 152.31152647975077, "grad_norm": 3.40625, "learning_rate": 6.823550986278299e-06, "loss": 0.2949, "step": 24522 }, { "epoch": 152.3177570093458, "grad_norm": 3.578125, "learning_rate": 6.821859514233192e-06, "loss": 0.3006, "step": 24523 }, { "epoch": 152.3239875389408, "grad_norm": 2.390625, "learning_rate": 6.820168218737818e-06, "loss": 0.176, "step": 24524 }, { "epoch": 152.33021806853583, "grad_norm": 1.328125, "learning_rate": 6.8184770998086065e-06, "loss": 0.117, "step": 24525 }, { "epoch": 152.33644859813083, "grad_norm": 2.5, "learning_rate": 6.816786157461988e-06, "loss": 0.1303, "step": 24526 }, { "epoch": 152.34267912772586, "grad_norm": 0.86328125, "learning_rate": 6.815095391714368e-06, "loss": 0.0752, "step": 24527 }, { "epoch": 152.34890965732086, "grad_norm": 2.265625, "learning_rate": 6.813404802582185e-06, "loss": 0.1752, "step": 24528 }, { "epoch": 152.3551401869159, "grad_norm": 2.15625, "learning_rate": 6.8117143900818414e-06, "loss": 0.2915, "step": 24529 }, { "epoch": 152.3613707165109, "grad_norm": 3.09375, "learning_rate": 6.810024154229774e-06, "loss": 0.1887, "step": 24530 }, { "epoch": 152.36760124610592, "grad_norm": 2.390625, "learning_rate": 6.8083340950423764e-06, "loss": 0.1958, "step": 24531 }, { "epoch": 152.37383177570092, "grad_norm": 1.96875, "learning_rate": 6.806644212536079e-06, "loss": 0.163, "step": 24532 }, { "epoch": 152.38006230529595, "grad_norm": 1.9453125, "learning_rate": 6.804954506727293e-06, "loss": 0.1676, "step": 24533 }, { "epoch": 152.38629283489095, "grad_norm": 2.8125, "learning_rate": 6.803264977632429e-06, "loss": 0.2459, "step": 24534 }, { "epoch": 152.39252336448598, "grad_norm": 2.359375, "learning_rate": 6.801575625267884e-06, "loss": 0.2226, "step": 24535 }, { "epoch": 152.39875389408098, "grad_norm": 2.203125, "learning_rate": 6.799886449650075e-06, "loss": 0.215, "step": 24536 }, { "epoch": 152.404984423676, "grad_norm": 1.6953125, "learning_rate": 6.7981974507954125e-06, "loss": 0.1382, "step": 24537 }, { "epoch": 152.41121495327104, "grad_norm": 1.4453125, "learning_rate": 6.7965086287202885e-06, "loss": 0.1536, "step": 24538 }, { "epoch": 152.41744548286604, "grad_norm": 1.5, "learning_rate": 6.794819983441112e-06, "loss": 0.1171, "step": 24539 }, { "epoch": 152.42367601246107, "grad_norm": 2.203125, "learning_rate": 6.793131514974291e-06, "loss": 0.1547, "step": 24540 }, { "epoch": 152.42990654205607, "grad_norm": 0.984375, "learning_rate": 6.791443223336217e-06, "loss": 0.0768, "step": 24541 }, { "epoch": 152.4361370716511, "grad_norm": 2.0625, "learning_rate": 6.789755108543275e-06, "loss": 0.1771, "step": 24542 }, { "epoch": 152.4423676012461, "grad_norm": 2.5, "learning_rate": 6.788067170611875e-06, "loss": 0.1768, "step": 24543 }, { "epoch": 152.44859813084113, "grad_norm": 1.9765625, "learning_rate": 6.786379409558413e-06, "loss": 0.1434, "step": 24544 }, { "epoch": 152.45482866043614, "grad_norm": 2.0625, "learning_rate": 6.784691825399267e-06, "loss": 0.1485, "step": 24545 }, { "epoch": 152.46105919003116, "grad_norm": 1.0703125, "learning_rate": 6.783004418150837e-06, "loss": 0.0948, "step": 24546 }, { "epoch": 152.46728971962617, "grad_norm": 2.359375, "learning_rate": 6.781317187829517e-06, "loss": 0.1771, "step": 24547 }, { "epoch": 152.4735202492212, "grad_norm": 1.125, "learning_rate": 6.779630134451684e-06, "loss": 0.1032, "step": 24548 }, { "epoch": 152.4797507788162, "grad_norm": 1.640625, "learning_rate": 6.7779432580337195e-06, "loss": 0.1813, "step": 24549 }, { "epoch": 152.48598130841123, "grad_norm": 2.515625, "learning_rate": 6.776256558592012e-06, "loss": 0.2073, "step": 24550 }, { "epoch": 152.49221183800623, "grad_norm": 2.796875, "learning_rate": 6.774570036142952e-06, "loss": 0.2193, "step": 24551 }, { "epoch": 152.49844236760126, "grad_norm": 1.40625, "learning_rate": 6.7728836907029026e-06, "loss": 0.1403, "step": 24552 }, { "epoch": 152.50467289719626, "grad_norm": 3.203125, "learning_rate": 6.771197522288259e-06, "loss": 0.2094, "step": 24553 }, { "epoch": 152.5109034267913, "grad_norm": 2.140625, "learning_rate": 6.769511530915382e-06, "loss": 0.1902, "step": 24554 }, { "epoch": 152.5171339563863, "grad_norm": 1.3984375, "learning_rate": 6.767825716600659e-06, "loss": 0.1405, "step": 24555 }, { "epoch": 152.52336448598132, "grad_norm": 3.0, "learning_rate": 6.76614007936045e-06, "loss": 0.1974, "step": 24556 }, { "epoch": 152.52959501557632, "grad_norm": 2.671875, "learning_rate": 6.764454619211133e-06, "loss": 0.2603, "step": 24557 }, { "epoch": 152.53582554517135, "grad_norm": 1.328125, "learning_rate": 6.762769336169078e-06, "loss": 0.1189, "step": 24558 }, { "epoch": 152.54205607476635, "grad_norm": 2.109375, "learning_rate": 6.761084230250661e-06, "loss": 0.1648, "step": 24559 }, { "epoch": 152.54828660436138, "grad_norm": 1.8515625, "learning_rate": 6.759399301472239e-06, "loss": 0.1397, "step": 24560 }, { "epoch": 152.55451713395638, "grad_norm": 2.234375, "learning_rate": 6.757714549850172e-06, "loss": 0.1478, "step": 24561 }, { "epoch": 152.5607476635514, "grad_norm": 2.125, "learning_rate": 6.756029975400824e-06, "loss": 0.2547, "step": 24562 }, { "epoch": 152.5669781931464, "grad_norm": 1.796875, "learning_rate": 6.754345578140569e-06, "loss": 0.0958, "step": 24563 }, { "epoch": 152.57320872274144, "grad_norm": 1.40625, "learning_rate": 6.7526613580857515e-06, "loss": 0.1075, "step": 24564 }, { "epoch": 152.57943925233644, "grad_norm": 2.5625, "learning_rate": 6.750977315252732e-06, "loss": 0.2281, "step": 24565 }, { "epoch": 152.58566978193147, "grad_norm": 1.59375, "learning_rate": 6.749293449657879e-06, "loss": 0.1218, "step": 24566 }, { "epoch": 152.59190031152647, "grad_norm": 1.5625, "learning_rate": 6.747609761317533e-06, "loss": 0.1044, "step": 24567 }, { "epoch": 152.5981308411215, "grad_norm": 2.046875, "learning_rate": 6.7459262502480445e-06, "loss": 0.1481, "step": 24568 }, { "epoch": 152.6043613707165, "grad_norm": 2.28125, "learning_rate": 6.744242916465768e-06, "loss": 0.2159, "step": 24569 }, { "epoch": 152.61059190031153, "grad_norm": 2.53125, "learning_rate": 6.74255975998706e-06, "loss": 0.2361, "step": 24570 }, { "epoch": 152.61682242990653, "grad_norm": 1.046875, "learning_rate": 6.7408767808282536e-06, "loss": 0.1218, "step": 24571 }, { "epoch": 152.62305295950156, "grad_norm": 1.765625, "learning_rate": 6.739193979005703e-06, "loss": 0.1291, "step": 24572 }, { "epoch": 152.62928348909657, "grad_norm": 2.28125, "learning_rate": 6.737511354535758e-06, "loss": 0.1171, "step": 24573 }, { "epoch": 152.6355140186916, "grad_norm": 2.890625, "learning_rate": 6.735828907434755e-06, "loss": 0.248, "step": 24574 }, { "epoch": 152.6417445482866, "grad_norm": 1.578125, "learning_rate": 6.734146637719024e-06, "loss": 0.1263, "step": 24575 }, { "epoch": 152.64797507788163, "grad_norm": 1.6328125, "learning_rate": 6.732464545404912e-06, "loss": 0.1489, "step": 24576 }, { "epoch": 152.65420560747663, "grad_norm": 1.53125, "learning_rate": 6.730782630508764e-06, "loss": 0.1341, "step": 24577 }, { "epoch": 152.66043613707166, "grad_norm": 2.765625, "learning_rate": 6.729100893046897e-06, "loss": 0.2594, "step": 24578 }, { "epoch": 152.66666666666666, "grad_norm": 1.03125, "learning_rate": 6.727419333035664e-06, "loss": 0.1055, "step": 24579 }, { "epoch": 152.6728971962617, "grad_norm": 3.34375, "learning_rate": 6.725737950491379e-06, "loss": 0.3743, "step": 24580 }, { "epoch": 152.6791277258567, "grad_norm": 2.484375, "learning_rate": 6.7240567454303874e-06, "loss": 0.2019, "step": 24581 }, { "epoch": 152.68535825545172, "grad_norm": 1.796875, "learning_rate": 6.722375717869003e-06, "loss": 0.1524, "step": 24582 }, { "epoch": 152.69158878504672, "grad_norm": 3.25, "learning_rate": 6.72069486782356e-06, "loss": 0.2233, "step": 24583 }, { "epoch": 152.69781931464175, "grad_norm": 2.03125, "learning_rate": 6.71901419531038e-06, "loss": 0.1163, "step": 24584 }, { "epoch": 152.70404984423675, "grad_norm": 2.4375, "learning_rate": 6.717333700345807e-06, "loss": 0.1949, "step": 24585 }, { "epoch": 152.71028037383178, "grad_norm": 2.375, "learning_rate": 6.715653382946127e-06, "loss": 0.125, "step": 24586 }, { "epoch": 152.71651090342678, "grad_norm": 2.21875, "learning_rate": 6.713973243127675e-06, "loss": 0.1691, "step": 24587 }, { "epoch": 152.7227414330218, "grad_norm": 1.28125, "learning_rate": 6.71229328090677e-06, "loss": 0.1179, "step": 24588 }, { "epoch": 152.7289719626168, "grad_norm": 0.99609375, "learning_rate": 6.710613496299739e-06, "loss": 0.1098, "step": 24589 }, { "epoch": 152.73520249221184, "grad_norm": 3.28125, "learning_rate": 6.708933889322874e-06, "loss": 0.2004, "step": 24590 }, { "epoch": 152.74143302180684, "grad_norm": 2.578125, "learning_rate": 6.707254459992502e-06, "loss": 0.2585, "step": 24591 }, { "epoch": 152.74766355140187, "grad_norm": 1.6953125, "learning_rate": 6.705575208324938e-06, "loss": 0.1084, "step": 24592 }, { "epoch": 152.75389408099687, "grad_norm": 1.8984375, "learning_rate": 6.703896134336485e-06, "loss": 0.1145, "step": 24593 }, { "epoch": 152.7601246105919, "grad_norm": 1.4453125, "learning_rate": 6.702217238043443e-06, "loss": 0.0881, "step": 24594 }, { "epoch": 152.7663551401869, "grad_norm": 2.859375, "learning_rate": 6.700538519462124e-06, "loss": 0.3089, "step": 24595 }, { "epoch": 152.77258566978193, "grad_norm": 3.59375, "learning_rate": 6.69885997860884e-06, "loss": 0.2011, "step": 24596 }, { "epoch": 152.77881619937693, "grad_norm": 1.84375, "learning_rate": 6.6971816154998815e-06, "loss": 0.1495, "step": 24597 }, { "epoch": 152.78504672897196, "grad_norm": 3.234375, "learning_rate": 6.695503430151559e-06, "loss": 0.2445, "step": 24598 }, { "epoch": 152.79127725856696, "grad_norm": 1.9296875, "learning_rate": 6.693825422580158e-06, "loss": 0.1499, "step": 24599 }, { "epoch": 152.797507788162, "grad_norm": 1.265625, "learning_rate": 6.6921475928019925e-06, "loss": 0.1305, "step": 24600 }, { "epoch": 152.80373831775702, "grad_norm": 2.609375, "learning_rate": 6.690469940833341e-06, "loss": 0.13, "step": 24601 }, { "epoch": 152.80996884735202, "grad_norm": 1.78125, "learning_rate": 6.688792466690505e-06, "loss": 0.1113, "step": 24602 }, { "epoch": 152.81619937694705, "grad_norm": 1.5078125, "learning_rate": 6.687115170389788e-06, "loss": 0.1246, "step": 24603 }, { "epoch": 152.82242990654206, "grad_norm": 2.671875, "learning_rate": 6.685438051947465e-06, "loss": 0.2122, "step": 24604 }, { "epoch": 152.82866043613708, "grad_norm": 1.9140625, "learning_rate": 6.683761111379822e-06, "loss": 0.1807, "step": 24605 }, { "epoch": 152.8348909657321, "grad_norm": 2.15625, "learning_rate": 6.682084348703152e-06, "loss": 0.1625, "step": 24606 }, { "epoch": 152.84112149532712, "grad_norm": 2.9375, "learning_rate": 6.6804077639337496e-06, "loss": 0.2541, "step": 24607 }, { "epoch": 152.84735202492212, "grad_norm": 2.578125, "learning_rate": 6.678731357087881e-06, "loss": 0.1777, "step": 24608 }, { "epoch": 152.85358255451715, "grad_norm": 2.328125, "learning_rate": 6.677055128181836e-06, "loss": 0.2943, "step": 24609 }, { "epoch": 152.85981308411215, "grad_norm": 0.94140625, "learning_rate": 6.675379077231902e-06, "loss": 0.0968, "step": 24610 }, { "epoch": 152.86604361370718, "grad_norm": 1.21875, "learning_rate": 6.673703204254347e-06, "loss": 0.0822, "step": 24611 }, { "epoch": 152.87227414330218, "grad_norm": 2.21875, "learning_rate": 6.672027509265447e-06, "loss": 0.1596, "step": 24612 }, { "epoch": 152.8785046728972, "grad_norm": 2.15625, "learning_rate": 6.670351992281476e-06, "loss": 0.1385, "step": 24613 }, { "epoch": 152.8847352024922, "grad_norm": 1.5390625, "learning_rate": 6.6686766533187174e-06, "loss": 0.135, "step": 24614 }, { "epoch": 152.89096573208724, "grad_norm": 2.5625, "learning_rate": 6.66700149239343e-06, "loss": 0.2026, "step": 24615 }, { "epoch": 152.89719626168224, "grad_norm": 3.375, "learning_rate": 6.665326509521888e-06, "loss": 0.3774, "step": 24616 }, { "epoch": 152.90342679127727, "grad_norm": 1.3515625, "learning_rate": 6.663651704720366e-06, "loss": 0.1101, "step": 24617 }, { "epoch": 152.90965732087227, "grad_norm": 1.5546875, "learning_rate": 6.661977078005127e-06, "loss": 0.11, "step": 24618 }, { "epoch": 152.9158878504673, "grad_norm": 2.015625, "learning_rate": 6.660302629392423e-06, "loss": 0.2304, "step": 24619 }, { "epoch": 152.9221183800623, "grad_norm": 2.5, "learning_rate": 6.658628358898527e-06, "loss": 0.1488, "step": 24620 }, { "epoch": 152.92834890965733, "grad_norm": 1.296875, "learning_rate": 6.656954266539697e-06, "loss": 0.0938, "step": 24621 }, { "epoch": 152.93457943925233, "grad_norm": 1.9765625, "learning_rate": 6.655280352332205e-06, "loss": 0.1589, "step": 24622 }, { "epoch": 152.94080996884736, "grad_norm": 2.375, "learning_rate": 6.653606616292294e-06, "loss": 0.185, "step": 24623 }, { "epoch": 152.94704049844236, "grad_norm": 2.703125, "learning_rate": 6.651933058436216e-06, "loss": 0.2314, "step": 24624 }, { "epoch": 152.9532710280374, "grad_norm": 1.078125, "learning_rate": 6.650259678780233e-06, "loss": 0.111, "step": 24625 }, { "epoch": 152.9595015576324, "grad_norm": 0.98046875, "learning_rate": 6.648586477340601e-06, "loss": 0.0999, "step": 24626 }, { "epoch": 152.96573208722742, "grad_norm": 1.1484375, "learning_rate": 6.646913454133561e-06, "loss": 0.1072, "step": 24627 }, { "epoch": 152.97196261682242, "grad_norm": 2.203125, "learning_rate": 6.645240609175363e-06, "loss": 0.2596, "step": 24628 }, { "epoch": 152.97819314641745, "grad_norm": 2.09375, "learning_rate": 6.643567942482268e-06, "loss": 0.1844, "step": 24629 }, { "epoch": 152.98442367601245, "grad_norm": 1.59375, "learning_rate": 6.641895454070507e-06, "loss": 0.1028, "step": 24630 }, { "epoch": 152.99065420560748, "grad_norm": 2.34375, "learning_rate": 6.640223143956323e-06, "loss": 0.1909, "step": 24631 }, { "epoch": 152.99688473520249, "grad_norm": 2.1875, "learning_rate": 6.638551012155961e-06, "loss": 0.1553, "step": 24632 }, { "epoch": 153.0, "grad_norm": 0.8671875, "learning_rate": 6.636879058685669e-06, "loss": 0.0723, "step": 24633 }, { "epoch": 153.00623052959503, "grad_norm": 1.6328125, "learning_rate": 6.63520728356167e-06, "loss": 0.136, "step": 24634 }, { "epoch": 153.01246105919003, "grad_norm": 1.71875, "learning_rate": 6.6335356868002074e-06, "loss": 0.1429, "step": 24635 }, { "epoch": 153.01869158878506, "grad_norm": 3.578125, "learning_rate": 6.631864268417528e-06, "loss": 0.389, "step": 24636 }, { "epoch": 153.02492211838006, "grad_norm": 2.328125, "learning_rate": 6.630193028429854e-06, "loss": 0.1825, "step": 24637 }, { "epoch": 153.0311526479751, "grad_norm": 2.234375, "learning_rate": 6.62852196685341e-06, "loss": 0.1761, "step": 24638 }, { "epoch": 153.0373831775701, "grad_norm": 2.28125, "learning_rate": 6.626851083704433e-06, "loss": 0.1846, "step": 24639 }, { "epoch": 153.04361370716512, "grad_norm": 1.75, "learning_rate": 6.625180378999157e-06, "loss": 0.1573, "step": 24640 }, { "epoch": 153.04984423676012, "grad_norm": 2.015625, "learning_rate": 6.623509852753798e-06, "loss": 0.1497, "step": 24641 }, { "epoch": 153.05607476635515, "grad_norm": 1.734375, "learning_rate": 6.621839504984581e-06, "loss": 0.1297, "step": 24642 }, { "epoch": 153.06230529595015, "grad_norm": 2.328125, "learning_rate": 6.6201693357077435e-06, "loss": 0.1604, "step": 24643 }, { "epoch": 153.06853582554518, "grad_norm": 1.53125, "learning_rate": 6.618499344939491e-06, "loss": 0.1232, "step": 24644 }, { "epoch": 153.07476635514018, "grad_norm": 1.65625, "learning_rate": 6.6168295326960424e-06, "loss": 0.1491, "step": 24645 }, { "epoch": 153.0809968847352, "grad_norm": 2.28125, "learning_rate": 6.615159898993619e-06, "loss": 0.2722, "step": 24646 }, { "epoch": 153.08722741433021, "grad_norm": 1.4140625, "learning_rate": 6.613490443848447e-06, "loss": 0.0799, "step": 24647 }, { "epoch": 153.09345794392524, "grad_norm": 2.4375, "learning_rate": 6.611821167276724e-06, "loss": 0.1595, "step": 24648 }, { "epoch": 153.09968847352025, "grad_norm": 1.4375, "learning_rate": 6.6101520692946755e-06, "loss": 0.13, "step": 24649 }, { "epoch": 153.10591900311528, "grad_norm": 1.3828125, "learning_rate": 6.608483149918499e-06, "loss": 0.1153, "step": 24650 }, { "epoch": 153.11214953271028, "grad_norm": 1.4453125, "learning_rate": 6.606814409164419e-06, "loss": 0.1339, "step": 24651 }, { "epoch": 153.1183800623053, "grad_norm": 2.0625, "learning_rate": 6.605145847048624e-06, "loss": 0.1868, "step": 24652 }, { "epoch": 153.1246105919003, "grad_norm": 1.6640625, "learning_rate": 6.603477463587332e-06, "loss": 0.1357, "step": 24653 }, { "epoch": 153.13084112149534, "grad_norm": 2.8125, "learning_rate": 6.601809258796746e-06, "loss": 0.2456, "step": 24654 }, { "epoch": 153.13707165109034, "grad_norm": 1.7734375, "learning_rate": 6.600141232693077e-06, "loss": 0.1514, "step": 24655 }, { "epoch": 153.14330218068537, "grad_norm": 1.1640625, "learning_rate": 6.598473385292503e-06, "loss": 0.1224, "step": 24656 }, { "epoch": 153.14953271028037, "grad_norm": 1.453125, "learning_rate": 6.596805716611232e-06, "loss": 0.1446, "step": 24657 }, { "epoch": 153.1557632398754, "grad_norm": 2.578125, "learning_rate": 6.595138226665462e-06, "loss": 0.1363, "step": 24658 }, { "epoch": 153.1619937694704, "grad_norm": 2.703125, "learning_rate": 6.593470915471398e-06, "loss": 0.1909, "step": 24659 }, { "epoch": 153.16822429906543, "grad_norm": 2.40625, "learning_rate": 6.591803783045214e-06, "loss": 0.2259, "step": 24660 }, { "epoch": 153.17445482866043, "grad_norm": 2.109375, "learning_rate": 6.590136829403113e-06, "loss": 0.0997, "step": 24661 }, { "epoch": 153.18068535825546, "grad_norm": 3.125, "learning_rate": 6.588470054561291e-06, "loss": 0.2069, "step": 24662 }, { "epoch": 153.18691588785046, "grad_norm": 1.4375, "learning_rate": 6.586803458535926e-06, "loss": 0.0983, "step": 24663 }, { "epoch": 153.1931464174455, "grad_norm": 2.703125, "learning_rate": 6.585137041343201e-06, "loss": 0.1492, "step": 24664 }, { "epoch": 153.1993769470405, "grad_norm": 2.5, "learning_rate": 6.583470802999306e-06, "loss": 0.1497, "step": 24665 }, { "epoch": 153.20560747663552, "grad_norm": 0.7890625, "learning_rate": 6.581804743520431e-06, "loss": 0.0945, "step": 24666 }, { "epoch": 153.21183800623052, "grad_norm": 2.40625, "learning_rate": 6.580138862922741e-06, "loss": 0.1807, "step": 24667 }, { "epoch": 153.21806853582555, "grad_norm": 2.484375, "learning_rate": 6.578473161222434e-06, "loss": 0.2006, "step": 24668 }, { "epoch": 153.22429906542055, "grad_norm": 1.859375, "learning_rate": 6.57680763843567e-06, "loss": 0.19, "step": 24669 }, { "epoch": 153.23052959501558, "grad_norm": 2.609375, "learning_rate": 6.57514229457864e-06, "loss": 0.19, "step": 24670 }, { "epoch": 153.23676012461058, "grad_norm": 1.9375, "learning_rate": 6.5734771296675045e-06, "loss": 0.1419, "step": 24671 }, { "epoch": 153.2429906542056, "grad_norm": 1.375, "learning_rate": 6.571812143718442e-06, "loss": 0.1257, "step": 24672 }, { "epoch": 153.2492211838006, "grad_norm": 1.703125, "learning_rate": 6.570147336747632e-06, "loss": 0.1291, "step": 24673 }, { "epoch": 153.25545171339564, "grad_norm": 0.8359375, "learning_rate": 6.568482708771234e-06, "loss": 0.0869, "step": 24674 }, { "epoch": 153.26168224299064, "grad_norm": 1.828125, "learning_rate": 6.56681825980541e-06, "loss": 0.1781, "step": 24675 }, { "epoch": 153.26791277258567, "grad_norm": 1.8359375, "learning_rate": 6.565153989866332e-06, "loss": 0.1934, "step": 24676 }, { "epoch": 153.27414330218068, "grad_norm": 1.25, "learning_rate": 6.563489898970168e-06, "loss": 0.0959, "step": 24677 }, { "epoch": 153.2803738317757, "grad_norm": 2.671875, "learning_rate": 6.561825987133071e-06, "loss": 0.1921, "step": 24678 }, { "epoch": 153.2866043613707, "grad_norm": 1.2578125, "learning_rate": 6.5601622543712045e-06, "loss": 0.1095, "step": 24679 }, { "epoch": 153.29283489096574, "grad_norm": 1.9453125, "learning_rate": 6.55849870070073e-06, "loss": 0.2003, "step": 24680 }, { "epoch": 153.29906542056074, "grad_norm": 2.03125, "learning_rate": 6.556835326137814e-06, "loss": 0.1257, "step": 24681 }, { "epoch": 153.30529595015577, "grad_norm": 1.796875, "learning_rate": 6.555172130698589e-06, "loss": 0.1346, "step": 24682 }, { "epoch": 153.31152647975077, "grad_norm": 1.5859375, "learning_rate": 6.5535091143992175e-06, "loss": 0.124, "step": 24683 }, { "epoch": 153.3177570093458, "grad_norm": 2.1875, "learning_rate": 6.551846277255852e-06, "loss": 0.1946, "step": 24684 }, { "epoch": 153.3239875389408, "grad_norm": 3.0, "learning_rate": 6.55018361928465e-06, "loss": 0.3147, "step": 24685 }, { "epoch": 153.33021806853583, "grad_norm": 2.234375, "learning_rate": 6.5485211405017485e-06, "loss": 0.1494, "step": 24686 }, { "epoch": 153.33644859813083, "grad_norm": 1.9140625, "learning_rate": 6.546858840923304e-06, "loss": 0.1332, "step": 24687 }, { "epoch": 153.34267912772586, "grad_norm": 2.8125, "learning_rate": 6.5451967205654466e-06, "loss": 0.1316, "step": 24688 }, { "epoch": 153.34890965732086, "grad_norm": 1.59375, "learning_rate": 6.543534779444335e-06, "loss": 0.0975, "step": 24689 }, { "epoch": 153.3551401869159, "grad_norm": 2.859375, "learning_rate": 6.5418730175760974e-06, "loss": 0.2048, "step": 24690 }, { "epoch": 153.3613707165109, "grad_norm": 2.03125, "learning_rate": 6.540211434976878e-06, "loss": 0.1429, "step": 24691 }, { "epoch": 153.36760124610592, "grad_norm": 2.25, "learning_rate": 6.538550031662821e-06, "loss": 0.2474, "step": 24692 }, { "epoch": 153.37383177570092, "grad_norm": 1.8125, "learning_rate": 6.53688880765006e-06, "loss": 0.139, "step": 24693 }, { "epoch": 153.38006230529595, "grad_norm": 1.34375, "learning_rate": 6.535227762954716e-06, "loss": 0.1098, "step": 24694 }, { "epoch": 153.38629283489095, "grad_norm": 2.21875, "learning_rate": 6.53356689759293e-06, "loss": 0.1166, "step": 24695 }, { "epoch": 153.39252336448598, "grad_norm": 1.5, "learning_rate": 6.531906211580843e-06, "loss": 0.1272, "step": 24696 }, { "epoch": 153.39875389408098, "grad_norm": 2.234375, "learning_rate": 6.530245704934565e-06, "loss": 0.245, "step": 24697 }, { "epoch": 153.404984423676, "grad_norm": 3.4375, "learning_rate": 6.5285853776702346e-06, "loss": 0.3219, "step": 24698 }, { "epoch": 153.41121495327104, "grad_norm": 1.3515625, "learning_rate": 6.526925229803982e-06, "loss": 0.1098, "step": 24699 }, { "epoch": 153.41744548286604, "grad_norm": 1.8828125, "learning_rate": 6.525265261351924e-06, "loss": 0.1457, "step": 24700 }, { "epoch": 153.42367601246107, "grad_norm": 2.640625, "learning_rate": 6.523605472330177e-06, "loss": 0.1787, "step": 24701 }, { "epoch": 153.42990654205607, "grad_norm": 1.296875, "learning_rate": 6.521945862754866e-06, "loss": 0.1039, "step": 24702 }, { "epoch": 153.4361370716511, "grad_norm": 0.80859375, "learning_rate": 6.520286432642117e-06, "loss": 0.1061, "step": 24703 }, { "epoch": 153.4423676012461, "grad_norm": 3.03125, "learning_rate": 6.518627182008033e-06, "loss": 0.2441, "step": 24704 }, { "epoch": 153.44859813084113, "grad_norm": 1.625, "learning_rate": 6.5169681108687375e-06, "loss": 0.1427, "step": 24705 }, { "epoch": 153.45482866043614, "grad_norm": 2.40625, "learning_rate": 6.515309219240348e-06, "loss": 0.1563, "step": 24706 }, { "epoch": 153.46105919003116, "grad_norm": 2.203125, "learning_rate": 6.513650507138971e-06, "loss": 0.1486, "step": 24707 }, { "epoch": 153.46728971962617, "grad_norm": 1.859375, "learning_rate": 6.511991974580708e-06, "loss": 0.1471, "step": 24708 }, { "epoch": 153.4735202492212, "grad_norm": 1.6015625, "learning_rate": 6.510333621581677e-06, "loss": 0.1645, "step": 24709 }, { "epoch": 153.4797507788162, "grad_norm": 1.296875, "learning_rate": 6.5086754481579885e-06, "loss": 0.1141, "step": 24710 }, { "epoch": 153.48598130841123, "grad_norm": 2.625, "learning_rate": 6.507017454325731e-06, "loss": 0.2344, "step": 24711 }, { "epoch": 153.49221183800623, "grad_norm": 1.65625, "learning_rate": 6.505359640101016e-06, "loss": 0.0998, "step": 24712 }, { "epoch": 153.49844236760126, "grad_norm": 2.5625, "learning_rate": 6.503702005499957e-06, "loss": 0.2211, "step": 24713 }, { "epoch": 153.50467289719626, "grad_norm": 1.7421875, "learning_rate": 6.502044550538638e-06, "loss": 0.1295, "step": 24714 }, { "epoch": 153.5109034267913, "grad_norm": 2.34375, "learning_rate": 6.500387275233153e-06, "loss": 0.1706, "step": 24715 }, { "epoch": 153.5171339563863, "grad_norm": 1.46875, "learning_rate": 6.498730179599605e-06, "loss": 0.116, "step": 24716 }, { "epoch": 153.52336448598132, "grad_norm": 1.3515625, "learning_rate": 6.497073263654088e-06, "loss": 0.1338, "step": 24717 }, { "epoch": 153.52959501557632, "grad_norm": 1.3125, "learning_rate": 6.495416527412701e-06, "loss": 0.0972, "step": 24718 }, { "epoch": 153.53582554517135, "grad_norm": 2.171875, "learning_rate": 6.493759970891527e-06, "loss": 0.2027, "step": 24719 }, { "epoch": 153.54205607476635, "grad_norm": 1.625, "learning_rate": 6.492103594106649e-06, "loss": 0.148, "step": 24720 }, { "epoch": 153.54828660436138, "grad_norm": 2.390625, "learning_rate": 6.4904473970741616e-06, "loss": 0.183, "step": 24721 }, { "epoch": 153.55451713395638, "grad_norm": 1.7578125, "learning_rate": 6.4887913798101555e-06, "loss": 0.14, "step": 24722 }, { "epoch": 153.5607476635514, "grad_norm": 2.40625, "learning_rate": 6.4871355423307025e-06, "loss": 0.1581, "step": 24723 }, { "epoch": 153.5669781931464, "grad_norm": 1.3046875, "learning_rate": 6.485479884651885e-06, "loss": 0.1293, "step": 24724 }, { "epoch": 153.57320872274144, "grad_norm": 1.59375, "learning_rate": 6.483824406789798e-06, "loss": 0.1167, "step": 24725 }, { "epoch": 153.57943925233644, "grad_norm": 2.5625, "learning_rate": 6.482169108760511e-06, "loss": 0.1222, "step": 24726 }, { "epoch": 153.58566978193147, "grad_norm": 1.171875, "learning_rate": 6.480513990580089e-06, "loss": 0.1053, "step": 24727 }, { "epoch": 153.59190031152647, "grad_norm": 1.2265625, "learning_rate": 6.478859052264622e-06, "loss": 0.1006, "step": 24728 }, { "epoch": 153.5981308411215, "grad_norm": 2.40625, "learning_rate": 6.47720429383018e-06, "loss": 0.2005, "step": 24729 }, { "epoch": 153.6043613707165, "grad_norm": 1.484375, "learning_rate": 6.475549715292828e-06, "loss": 0.1351, "step": 24730 }, { "epoch": 153.61059190031153, "grad_norm": 2.125, "learning_rate": 6.473895316668638e-06, "loss": 0.205, "step": 24731 }, { "epoch": 153.61682242990653, "grad_norm": 2.28125, "learning_rate": 6.4722410979736914e-06, "loss": 0.1964, "step": 24732 }, { "epoch": 153.62305295950156, "grad_norm": 1.1015625, "learning_rate": 6.4705870592240395e-06, "loss": 0.0928, "step": 24733 }, { "epoch": 153.62928348909657, "grad_norm": 2.59375, "learning_rate": 6.468933200435742e-06, "loss": 0.2059, "step": 24734 }, { "epoch": 153.6355140186916, "grad_norm": 2.53125, "learning_rate": 6.467279521624872e-06, "loss": 0.1996, "step": 24735 }, { "epoch": 153.6417445482866, "grad_norm": 2.0, "learning_rate": 6.4656260228074944e-06, "loss": 0.1584, "step": 24736 }, { "epoch": 153.64797507788163, "grad_norm": 2.46875, "learning_rate": 6.463972703999655e-06, "loss": 0.1672, "step": 24737 }, { "epoch": 153.65420560747663, "grad_norm": 1.890625, "learning_rate": 6.462319565217423e-06, "loss": 0.1124, "step": 24738 }, { "epoch": 153.66043613707166, "grad_norm": 1.796875, "learning_rate": 6.460666606476845e-06, "loss": 0.129, "step": 24739 }, { "epoch": 153.66666666666666, "grad_norm": 3.09375, "learning_rate": 6.459013827793983e-06, "loss": 0.1697, "step": 24740 }, { "epoch": 153.6728971962617, "grad_norm": 1.5703125, "learning_rate": 6.45736122918488e-06, "loss": 0.136, "step": 24741 }, { "epoch": 153.6791277258567, "grad_norm": 2.59375, "learning_rate": 6.4557088106655895e-06, "loss": 0.2249, "step": 24742 }, { "epoch": 153.68535825545172, "grad_norm": 1.859375, "learning_rate": 6.45405657225217e-06, "loss": 0.1698, "step": 24743 }, { "epoch": 153.69158878504672, "grad_norm": 2.421875, "learning_rate": 6.452404513960661e-06, "loss": 0.1806, "step": 24744 }, { "epoch": 153.69781931464175, "grad_norm": 2.625, "learning_rate": 6.4507526358071e-06, "loss": 0.1998, "step": 24745 }, { "epoch": 153.70404984423675, "grad_norm": 2.109375, "learning_rate": 6.449100937807537e-06, "loss": 0.1537, "step": 24746 }, { "epoch": 153.71028037383178, "grad_norm": 2.953125, "learning_rate": 6.447449419978022e-06, "loss": 0.2099, "step": 24747 }, { "epoch": 153.71651090342678, "grad_norm": 1.7734375, "learning_rate": 6.445798082334578e-06, "loss": 0.145, "step": 24748 }, { "epoch": 153.7227414330218, "grad_norm": 1.6796875, "learning_rate": 6.4441469248932515e-06, "loss": 0.1646, "step": 24749 }, { "epoch": 153.7289719626168, "grad_norm": 1.828125, "learning_rate": 6.44249594767008e-06, "loss": 0.1262, "step": 24750 }, { "epoch": 153.73520249221184, "grad_norm": 1.6796875, "learning_rate": 6.440845150681113e-06, "loss": 0.1335, "step": 24751 }, { "epoch": 153.74143302180684, "grad_norm": 1.6875, "learning_rate": 6.439194533942353e-06, "loss": 0.1356, "step": 24752 }, { "epoch": 153.74766355140187, "grad_norm": 1.09375, "learning_rate": 6.437544097469844e-06, "loss": 0.1175, "step": 24753 }, { "epoch": 153.75389408099687, "grad_norm": 1.6328125, "learning_rate": 6.435893841279616e-06, "loss": 0.1442, "step": 24754 }, { "epoch": 153.7601246105919, "grad_norm": 2.984375, "learning_rate": 6.434243765387709e-06, "loss": 0.2091, "step": 24755 }, { "epoch": 153.7663551401869, "grad_norm": 0.97265625, "learning_rate": 6.43259386981013e-06, "loss": 0.0855, "step": 24756 }, { "epoch": 153.77258566978193, "grad_norm": 0.98046875, "learning_rate": 6.430944154562918e-06, "loss": 0.0869, "step": 24757 }, { "epoch": 153.77881619937693, "grad_norm": 2.765625, "learning_rate": 6.4292946196620796e-06, "loss": 0.2263, "step": 24758 }, { "epoch": 153.78504672897196, "grad_norm": 0.87109375, "learning_rate": 6.427645265123653e-06, "loss": 0.1016, "step": 24759 }, { "epoch": 153.79127725856696, "grad_norm": 1.5390625, "learning_rate": 6.4259960909636406e-06, "loss": 0.1061, "step": 24760 }, { "epoch": 153.797507788162, "grad_norm": 1.125, "learning_rate": 6.424347097198063e-06, "loss": 0.101, "step": 24761 }, { "epoch": 153.80373831775702, "grad_norm": 2.359375, "learning_rate": 6.422698283842951e-06, "loss": 0.1852, "step": 24762 }, { "epoch": 153.80996884735202, "grad_norm": 2.671875, "learning_rate": 6.421049650914299e-06, "loss": 0.1841, "step": 24763 }, { "epoch": 153.81619937694705, "grad_norm": 2.984375, "learning_rate": 6.419401198428135e-06, "loss": 0.3073, "step": 24764 }, { "epoch": 153.82242990654206, "grad_norm": 2.09375, "learning_rate": 6.417752926400453e-06, "loss": 0.2043, "step": 24765 }, { "epoch": 153.82866043613708, "grad_norm": 2.171875, "learning_rate": 6.416104834847275e-06, "loss": 0.1879, "step": 24766 }, { "epoch": 153.8348909657321, "grad_norm": 1.7421875, "learning_rate": 6.414456923784593e-06, "loss": 0.1413, "step": 24767 }, { "epoch": 153.84112149532712, "grad_norm": 2.46875, "learning_rate": 6.412809193228422e-06, "loss": 0.1415, "step": 24768 }, { "epoch": 153.84735202492212, "grad_norm": 1.65625, "learning_rate": 6.411161643194771e-06, "loss": 0.0981, "step": 24769 }, { "epoch": 153.85358255451715, "grad_norm": 3.15625, "learning_rate": 6.409514273699635e-06, "loss": 0.3119, "step": 24770 }, { "epoch": 153.85981308411215, "grad_norm": 2.296875, "learning_rate": 6.4078670847590064e-06, "loss": 0.2014, "step": 24771 }, { "epoch": 153.86604361370718, "grad_norm": 0.91015625, "learning_rate": 6.406220076388886e-06, "loss": 0.0959, "step": 24772 }, { "epoch": 153.87227414330218, "grad_norm": 2.28125, "learning_rate": 6.404573248605284e-06, "loss": 0.2208, "step": 24773 }, { "epoch": 153.8785046728972, "grad_norm": 1.2421875, "learning_rate": 6.4029266014241725e-06, "loss": 0.1026, "step": 24774 }, { "epoch": 153.8847352024922, "grad_norm": 1.3515625, "learning_rate": 6.401280134861556e-06, "loss": 0.1238, "step": 24775 }, { "epoch": 153.89096573208724, "grad_norm": 1.9296875, "learning_rate": 6.399633848933434e-06, "loss": 0.1595, "step": 24776 }, { "epoch": 153.89719626168224, "grad_norm": 1.40625, "learning_rate": 6.397987743655784e-06, "loss": 0.1265, "step": 24777 }, { "epoch": 153.90342679127727, "grad_norm": 1.890625, "learning_rate": 6.396341819044591e-06, "loss": 0.1661, "step": 24778 }, { "epoch": 153.90965732087227, "grad_norm": 2.140625, "learning_rate": 6.394696075115841e-06, "loss": 0.1139, "step": 24779 }, { "epoch": 153.9158878504673, "grad_norm": 1.140625, "learning_rate": 6.393050511885526e-06, "loss": 0.1019, "step": 24780 }, { "epoch": 153.9221183800623, "grad_norm": 2.5625, "learning_rate": 6.391405129369629e-06, "loss": 0.181, "step": 24781 }, { "epoch": 153.92834890965733, "grad_norm": 1.9453125, "learning_rate": 6.389759927584118e-06, "loss": 0.1235, "step": 24782 }, { "epoch": 153.93457943925233, "grad_norm": 1.84375, "learning_rate": 6.388114906544987e-06, "loss": 0.1346, "step": 24783 }, { "epoch": 153.94080996884736, "grad_norm": 2.3125, "learning_rate": 6.386470066268196e-06, "loss": 0.2227, "step": 24784 }, { "epoch": 153.94704049844236, "grad_norm": 1.7265625, "learning_rate": 6.384825406769737e-06, "loss": 0.1389, "step": 24785 }, { "epoch": 153.9532710280374, "grad_norm": 1.0, "learning_rate": 6.383180928065566e-06, "loss": 0.1023, "step": 24786 }, { "epoch": 153.9595015576324, "grad_norm": 2.1875, "learning_rate": 6.381536630171664e-06, "loss": 0.124, "step": 24787 }, { "epoch": 153.96573208722742, "grad_norm": 2.78125, "learning_rate": 6.379892513104008e-06, "loss": 0.1497, "step": 24788 }, { "epoch": 153.97196261682242, "grad_norm": 1.8515625, "learning_rate": 6.378248576878556e-06, "loss": 0.1698, "step": 24789 }, { "epoch": 153.97819314641745, "grad_norm": 1.7734375, "learning_rate": 6.3766048215112704e-06, "loss": 0.1594, "step": 24790 }, { "epoch": 153.98442367601245, "grad_norm": 1.890625, "learning_rate": 6.374961247018121e-06, "loss": 0.0997, "step": 24791 }, { "epoch": 153.99065420560748, "grad_norm": 2.9375, "learning_rate": 6.373317853415078e-06, "loss": 0.2206, "step": 24792 }, { "epoch": 153.99688473520249, "grad_norm": 1.8671875, "learning_rate": 6.37167464071809e-06, "loss": 0.1174, "step": 24793 }, { "epoch": 154.0, "grad_norm": 1.7578125, "learning_rate": 6.37003160894312e-06, "loss": 0.1182, "step": 24794 }, { "epoch": 154.00623052959503, "grad_norm": 2.265625, "learning_rate": 6.368388758106133e-06, "loss": 0.18, "step": 24795 }, { "epoch": 154.01246105919003, "grad_norm": 2.484375, "learning_rate": 6.366746088223083e-06, "loss": 0.1552, "step": 24796 }, { "epoch": 154.01869158878506, "grad_norm": 1.6875, "learning_rate": 6.365103599309908e-06, "loss": 0.1388, "step": 24797 }, { "epoch": 154.02492211838006, "grad_norm": 2.203125, "learning_rate": 6.363461291382575e-06, "loss": 0.1691, "step": 24798 }, { "epoch": 154.0311526479751, "grad_norm": 1.796875, "learning_rate": 6.361819164457036e-06, "loss": 0.186, "step": 24799 }, { "epoch": 154.0373831775701, "grad_norm": 1.046875, "learning_rate": 6.3601772185492296e-06, "loss": 0.1233, "step": 24800 }, { "epoch": 154.04361370716512, "grad_norm": 2.546875, "learning_rate": 6.358535453675107e-06, "loss": 0.1435, "step": 24801 }, { "epoch": 154.04984423676012, "grad_norm": 2.296875, "learning_rate": 6.3568938698506216e-06, "loss": 0.1525, "step": 24802 }, { "epoch": 154.05607476635515, "grad_norm": 2.21875, "learning_rate": 6.35525246709171e-06, "loss": 0.1145, "step": 24803 }, { "epoch": 154.06230529595015, "grad_norm": 0.66015625, "learning_rate": 6.353611245414306e-06, "loss": 0.0754, "step": 24804 }, { "epoch": 154.06853582554518, "grad_norm": 1.7265625, "learning_rate": 6.351970204834359e-06, "loss": 0.1753, "step": 24805 }, { "epoch": 154.07476635514018, "grad_norm": 0.984375, "learning_rate": 6.350329345367811e-06, "loss": 0.0933, "step": 24806 }, { "epoch": 154.0809968847352, "grad_norm": 2.03125, "learning_rate": 6.348688667030583e-06, "loss": 0.1649, "step": 24807 }, { "epoch": 154.08722741433021, "grad_norm": 1.1953125, "learning_rate": 6.347048169838629e-06, "loss": 0.0921, "step": 24808 }, { "epoch": 154.09345794392524, "grad_norm": 1.0, "learning_rate": 6.3454078538078635e-06, "loss": 0.1059, "step": 24809 }, { "epoch": 154.09968847352025, "grad_norm": 1.9921875, "learning_rate": 6.343767718954233e-06, "loss": 0.1502, "step": 24810 }, { "epoch": 154.10591900311528, "grad_norm": 2.375, "learning_rate": 6.342127765293654e-06, "loss": 0.19, "step": 24811 }, { "epoch": 154.11214953271028, "grad_norm": 1.3359375, "learning_rate": 6.340487992842059e-06, "loss": 0.1167, "step": 24812 }, { "epoch": 154.1183800623053, "grad_norm": 2.09375, "learning_rate": 6.338848401615372e-06, "loss": 0.2008, "step": 24813 }, { "epoch": 154.1246105919003, "grad_norm": 2.421875, "learning_rate": 6.337208991629537e-06, "loss": 0.1833, "step": 24814 }, { "epoch": 154.13084112149534, "grad_norm": 1.3984375, "learning_rate": 6.335569762900445e-06, "loss": 0.1344, "step": 24815 }, { "epoch": 154.13707165109034, "grad_norm": 2.03125, "learning_rate": 6.333930715444028e-06, "loss": 0.1362, "step": 24816 }, { "epoch": 154.14330218068537, "grad_norm": 1.5234375, "learning_rate": 6.332291849276209e-06, "loss": 0.1061, "step": 24817 }, { "epoch": 154.14953271028037, "grad_norm": 1.7890625, "learning_rate": 6.330653164412908e-06, "loss": 0.1664, "step": 24818 }, { "epoch": 154.1557632398754, "grad_norm": 2.25, "learning_rate": 6.329014660870031e-06, "loss": 0.2061, "step": 24819 }, { "epoch": 154.1619937694704, "grad_norm": 2.609375, "learning_rate": 6.327376338663493e-06, "loss": 0.2049, "step": 24820 }, { "epoch": 154.16822429906543, "grad_norm": 1.4140625, "learning_rate": 6.325738197809217e-06, "loss": 0.0935, "step": 24821 }, { "epoch": 154.17445482866043, "grad_norm": 1.0390625, "learning_rate": 6.324100238323105e-06, "loss": 0.0966, "step": 24822 }, { "epoch": 154.18068535825546, "grad_norm": 1.8671875, "learning_rate": 6.322462460221057e-06, "loss": 0.1392, "step": 24823 }, { "epoch": 154.18691588785046, "grad_norm": 2.03125, "learning_rate": 6.3208248635189896e-06, "loss": 0.2356, "step": 24824 }, { "epoch": 154.1931464174455, "grad_norm": 0.9609375, "learning_rate": 6.319187448232811e-06, "loss": 0.1118, "step": 24825 }, { "epoch": 154.1993769470405, "grad_norm": 2.296875, "learning_rate": 6.317550214378409e-06, "loss": 0.2367, "step": 24826 }, { "epoch": 154.20560747663552, "grad_norm": 2.015625, "learning_rate": 6.315913161971701e-06, "loss": 0.2523, "step": 24827 }, { "epoch": 154.21183800623052, "grad_norm": 2.5, "learning_rate": 6.314276291028573e-06, "loss": 0.1432, "step": 24828 }, { "epoch": 154.21806853582555, "grad_norm": 2.078125, "learning_rate": 6.312639601564937e-06, "loss": 0.1872, "step": 24829 }, { "epoch": 154.22429906542055, "grad_norm": 2.53125, "learning_rate": 6.311003093596673e-06, "loss": 0.1955, "step": 24830 }, { "epoch": 154.23052959501558, "grad_norm": 2.453125, "learning_rate": 6.309366767139682e-06, "loss": 0.2471, "step": 24831 }, { "epoch": 154.23676012461058, "grad_norm": 1.953125, "learning_rate": 6.307730622209864e-06, "loss": 0.1563, "step": 24832 }, { "epoch": 154.2429906542056, "grad_norm": 2.4375, "learning_rate": 6.306094658823097e-06, "loss": 0.1295, "step": 24833 }, { "epoch": 154.2492211838006, "grad_norm": 1.9296875, "learning_rate": 6.304458876995284e-06, "loss": 0.1538, "step": 24834 }, { "epoch": 154.25545171339564, "grad_norm": 2.703125, "learning_rate": 6.3028232767422924e-06, "loss": 0.2156, "step": 24835 }, { "epoch": 154.26168224299064, "grad_norm": 2.8125, "learning_rate": 6.301187858080029e-06, "loss": 0.2638, "step": 24836 }, { "epoch": 154.26791277258567, "grad_norm": 2.75, "learning_rate": 6.29955262102436e-06, "loss": 0.2062, "step": 24837 }, { "epoch": 154.27414330218068, "grad_norm": 1.8984375, "learning_rate": 6.297917565591172e-06, "loss": 0.1638, "step": 24838 }, { "epoch": 154.2803738317757, "grad_norm": 1.0, "learning_rate": 6.296282691796354e-06, "loss": 0.1038, "step": 24839 }, { "epoch": 154.2866043613707, "grad_norm": 2.453125, "learning_rate": 6.294647999655778e-06, "loss": 0.2273, "step": 24840 }, { "epoch": 154.29283489096574, "grad_norm": 1.9765625, "learning_rate": 6.2930134891853146e-06, "loss": 0.1649, "step": 24841 }, { "epoch": 154.29906542056074, "grad_norm": 1.4375, "learning_rate": 6.2913791604008425e-06, "loss": 0.1174, "step": 24842 }, { "epoch": 154.30529595015577, "grad_norm": 1.3125, "learning_rate": 6.28974501331824e-06, "loss": 0.1262, "step": 24843 }, { "epoch": 154.31152647975077, "grad_norm": 1.390625, "learning_rate": 6.288111047953371e-06, "loss": 0.1219, "step": 24844 }, { "epoch": 154.3177570093458, "grad_norm": 1.203125, "learning_rate": 6.286477264322105e-06, "loss": 0.1236, "step": 24845 }, { "epoch": 154.3239875389408, "grad_norm": 1.6484375, "learning_rate": 6.2848436624403205e-06, "loss": 0.1384, "step": 24846 }, { "epoch": 154.33021806853583, "grad_norm": 1.6015625, "learning_rate": 6.283210242323878e-06, "loss": 0.1229, "step": 24847 }, { "epoch": 154.33644859813083, "grad_norm": 2.75, "learning_rate": 6.281577003988629e-06, "loss": 0.175, "step": 24848 }, { "epoch": 154.34267912772586, "grad_norm": 2.0625, "learning_rate": 6.279943947450445e-06, "loss": 0.13, "step": 24849 }, { "epoch": 154.34890965732086, "grad_norm": 2.3125, "learning_rate": 6.278311072725191e-06, "loss": 0.2121, "step": 24850 }, { "epoch": 154.3551401869159, "grad_norm": 2.28125, "learning_rate": 6.276678379828727e-06, "loss": 0.2809, "step": 24851 }, { "epoch": 154.3613707165109, "grad_norm": 2.40625, "learning_rate": 6.2750458687768995e-06, "loss": 0.2235, "step": 24852 }, { "epoch": 154.36760124610592, "grad_norm": 3.0, "learning_rate": 6.273413539585573e-06, "loss": 0.2307, "step": 24853 }, { "epoch": 154.37383177570092, "grad_norm": 1.0390625, "learning_rate": 6.271781392270596e-06, "loss": 0.1147, "step": 24854 }, { "epoch": 154.38006230529595, "grad_norm": 2.4375, "learning_rate": 6.270149426847824e-06, "loss": 0.1325, "step": 24855 }, { "epoch": 154.38629283489095, "grad_norm": 2.046875, "learning_rate": 6.268517643333102e-06, "loss": 0.1074, "step": 24856 }, { "epoch": 154.39252336448598, "grad_norm": 1.671875, "learning_rate": 6.2668860417422784e-06, "loss": 0.1104, "step": 24857 }, { "epoch": 154.39875389408098, "grad_norm": 2.65625, "learning_rate": 6.26525462209121e-06, "loss": 0.1579, "step": 24858 }, { "epoch": 154.404984423676, "grad_norm": 1.7578125, "learning_rate": 6.2636233843957335e-06, "loss": 0.1227, "step": 24859 }, { "epoch": 154.41121495327104, "grad_norm": 1.875, "learning_rate": 6.261992328671684e-06, "loss": 0.2275, "step": 24860 }, { "epoch": 154.41744548286604, "grad_norm": 1.90625, "learning_rate": 6.260361454934913e-06, "loss": 0.133, "step": 24861 }, { "epoch": 154.42367601246107, "grad_norm": 2.9375, "learning_rate": 6.2587307632012624e-06, "loss": 0.2519, "step": 24862 }, { "epoch": 154.42990654205607, "grad_norm": 1.875, "learning_rate": 6.257100253486558e-06, "loss": 0.1569, "step": 24863 }, { "epoch": 154.4361370716511, "grad_norm": 2.671875, "learning_rate": 6.255469925806643e-06, "loss": 0.2296, "step": 24864 }, { "epoch": 154.4423676012461, "grad_norm": 3.15625, "learning_rate": 6.2538397801773594e-06, "loss": 0.1772, "step": 24865 }, { "epoch": 154.44859813084113, "grad_norm": 2.296875, "learning_rate": 6.252209816614529e-06, "loss": 0.1539, "step": 24866 }, { "epoch": 154.45482866043614, "grad_norm": 2.828125, "learning_rate": 6.250580035133979e-06, "loss": 0.2308, "step": 24867 }, { "epoch": 154.46105919003116, "grad_norm": 3.09375, "learning_rate": 6.248950435751544e-06, "loss": 0.2528, "step": 24868 }, { "epoch": 154.46728971962617, "grad_norm": 2.296875, "learning_rate": 6.247321018483057e-06, "loss": 0.1829, "step": 24869 }, { "epoch": 154.4735202492212, "grad_norm": 2.296875, "learning_rate": 6.245691783344329e-06, "loss": 0.2397, "step": 24870 }, { "epoch": 154.4797507788162, "grad_norm": 2.28125, "learning_rate": 6.244062730351191e-06, "loss": 0.1216, "step": 24871 }, { "epoch": 154.48598130841123, "grad_norm": 2.203125, "learning_rate": 6.2424338595194724e-06, "loss": 0.1235, "step": 24872 }, { "epoch": 154.49221183800623, "grad_norm": 3.5, "learning_rate": 6.240805170864989e-06, "loss": 0.306, "step": 24873 }, { "epoch": 154.49844236760126, "grad_norm": 2.171875, "learning_rate": 6.2391766644035455e-06, "loss": 0.1262, "step": 24874 }, { "epoch": 154.50467289719626, "grad_norm": 1.890625, "learning_rate": 6.237548340150967e-06, "loss": 0.2009, "step": 24875 }, { "epoch": 154.5109034267913, "grad_norm": 1.5859375, "learning_rate": 6.235920198123071e-06, "loss": 0.1403, "step": 24876 }, { "epoch": 154.5171339563863, "grad_norm": 2.5, "learning_rate": 6.234292238335676e-06, "loss": 0.2214, "step": 24877 }, { "epoch": 154.52336448598132, "grad_norm": 2.5625, "learning_rate": 6.232664460804588e-06, "loss": 0.2428, "step": 24878 }, { "epoch": 154.52959501557632, "grad_norm": 1.90625, "learning_rate": 6.231036865545606e-06, "loss": 0.1405, "step": 24879 }, { "epoch": 154.53582554517135, "grad_norm": 1.8671875, "learning_rate": 6.229409452574544e-06, "loss": 0.1284, "step": 24880 }, { "epoch": 154.54205607476635, "grad_norm": 2.578125, "learning_rate": 6.227782221907219e-06, "loss": 0.2494, "step": 24881 }, { "epoch": 154.54828660436138, "grad_norm": 2.921875, "learning_rate": 6.226155173559417e-06, "loss": 0.3112, "step": 24882 }, { "epoch": 154.55451713395638, "grad_norm": 3.234375, "learning_rate": 6.2245283075469475e-06, "loss": 0.2834, "step": 24883 }, { "epoch": 154.5607476635514, "grad_norm": 1.2578125, "learning_rate": 6.2229016238856244e-06, "loss": 0.1234, "step": 24884 }, { "epoch": 154.5669781931464, "grad_norm": 1.390625, "learning_rate": 6.22127512259123e-06, "loss": 0.1359, "step": 24885 }, { "epoch": 154.57320872274144, "grad_norm": 1.6640625, "learning_rate": 6.219648803679559e-06, "loss": 0.139, "step": 24886 }, { "epoch": 154.57943925233644, "grad_norm": 1.34375, "learning_rate": 6.218022667166412e-06, "loss": 0.1004, "step": 24887 }, { "epoch": 154.58566978193147, "grad_norm": 2.25, "learning_rate": 6.216396713067593e-06, "loss": 0.1838, "step": 24888 }, { "epoch": 154.59190031152647, "grad_norm": 3.828125, "learning_rate": 6.2147709413988735e-06, "loss": 0.2614, "step": 24889 }, { "epoch": 154.5981308411215, "grad_norm": 2.671875, "learning_rate": 6.213145352176056e-06, "loss": 0.1761, "step": 24890 }, { "epoch": 154.6043613707165, "grad_norm": 1.15625, "learning_rate": 6.211519945414934e-06, "loss": 0.0948, "step": 24891 }, { "epoch": 154.61059190031153, "grad_norm": 2.25, "learning_rate": 6.2098947211312844e-06, "loss": 0.2602, "step": 24892 }, { "epoch": 154.61682242990653, "grad_norm": 1.515625, "learning_rate": 6.208269679340886e-06, "loss": 0.0961, "step": 24893 }, { "epoch": 154.62305295950156, "grad_norm": 1.71875, "learning_rate": 6.206644820059529e-06, "loss": 0.1493, "step": 24894 }, { "epoch": 154.62928348909657, "grad_norm": 2.390625, "learning_rate": 6.205020143303003e-06, "loss": 0.0863, "step": 24895 }, { "epoch": 154.6355140186916, "grad_norm": 2.75, "learning_rate": 6.203395649087071e-06, "loss": 0.3011, "step": 24896 }, { "epoch": 154.6417445482866, "grad_norm": 2.203125, "learning_rate": 6.201771337427523e-06, "loss": 0.1817, "step": 24897 }, { "epoch": 154.64797507788163, "grad_norm": 2.234375, "learning_rate": 6.200147208340124e-06, "loss": 0.1801, "step": 24898 }, { "epoch": 154.65420560747663, "grad_norm": 3.125, "learning_rate": 6.198523261840661e-06, "loss": 0.3028, "step": 24899 }, { "epoch": 154.66043613707166, "grad_norm": 1.3828125, "learning_rate": 6.196899497944888e-06, "loss": 0.1099, "step": 24900 }, { "epoch": 154.66666666666666, "grad_norm": 2.6875, "learning_rate": 6.195275916668589e-06, "loss": 0.1358, "step": 24901 }, { "epoch": 154.6728971962617, "grad_norm": 1.140625, "learning_rate": 6.193652518027535e-06, "loss": 0.106, "step": 24902 }, { "epoch": 154.6791277258567, "grad_norm": 1.8984375, "learning_rate": 6.19202930203748e-06, "loss": 0.1871, "step": 24903 }, { "epoch": 154.68535825545172, "grad_norm": 2.640625, "learning_rate": 6.190406268714205e-06, "loss": 0.1966, "step": 24904 }, { "epoch": 154.69158878504672, "grad_norm": 2.015625, "learning_rate": 6.188783418073454e-06, "loss": 0.2148, "step": 24905 }, { "epoch": 154.69781931464175, "grad_norm": 1.1875, "learning_rate": 6.187160750131007e-06, "loss": 0.1146, "step": 24906 }, { "epoch": 154.70404984423675, "grad_norm": 1.9296875, "learning_rate": 6.18553826490261e-06, "loss": 0.2016, "step": 24907 }, { "epoch": 154.71028037383178, "grad_norm": 2.9375, "learning_rate": 6.183915962404024e-06, "loss": 0.2455, "step": 24908 }, { "epoch": 154.71651090342678, "grad_norm": 2.34375, "learning_rate": 6.1822938426510104e-06, "loss": 0.148, "step": 24909 }, { "epoch": 154.7227414330218, "grad_norm": 2.203125, "learning_rate": 6.180671905659333e-06, "loss": 0.1165, "step": 24910 }, { "epoch": 154.7289719626168, "grad_norm": 1.125, "learning_rate": 6.179050151444715e-06, "loss": 0.1023, "step": 24911 }, { "epoch": 154.73520249221184, "grad_norm": 3.40625, "learning_rate": 6.177428580022929e-06, "loss": 0.445, "step": 24912 }, { "epoch": 154.74143302180684, "grad_norm": 2.140625, "learning_rate": 6.175807191409716e-06, "loss": 0.2552, "step": 24913 }, { "epoch": 154.74766355140187, "grad_norm": 1.6796875, "learning_rate": 6.174185985620834e-06, "loss": 0.1449, "step": 24914 }, { "epoch": 154.75389408099687, "grad_norm": 2.234375, "learning_rate": 6.1725649626720145e-06, "loss": 0.2063, "step": 24915 }, { "epoch": 154.7601246105919, "grad_norm": 1.015625, "learning_rate": 6.170944122579014e-06, "loss": 0.098, "step": 24916 }, { "epoch": 154.7663551401869, "grad_norm": 1.6015625, "learning_rate": 6.1693234653575625e-06, "loss": 0.1332, "step": 24917 }, { "epoch": 154.77258566978193, "grad_norm": 2.40625, "learning_rate": 6.167702991023411e-06, "loss": 0.2427, "step": 24918 }, { "epoch": 154.77881619937693, "grad_norm": 2.171875, "learning_rate": 6.166082699592288e-06, "loss": 0.1466, "step": 24919 }, { "epoch": 154.78504672897196, "grad_norm": 1.9140625, "learning_rate": 6.164462591079931e-06, "loss": 0.1339, "step": 24920 }, { "epoch": 154.79127725856696, "grad_norm": 1.0859375, "learning_rate": 6.162842665502089e-06, "loss": 0.1022, "step": 24921 }, { "epoch": 154.797507788162, "grad_norm": 2.71875, "learning_rate": 6.161222922874476e-06, "loss": 0.2583, "step": 24922 }, { "epoch": 154.80373831775702, "grad_norm": 3.09375, "learning_rate": 6.159603363212837e-06, "loss": 0.3634, "step": 24923 }, { "epoch": 154.80996884735202, "grad_norm": 2.0, "learning_rate": 6.157983986532892e-06, "loss": 0.1378, "step": 24924 }, { "epoch": 154.81619937694705, "grad_norm": 2.171875, "learning_rate": 6.156364792850381e-06, "loss": 0.1405, "step": 24925 }, { "epoch": 154.82242990654206, "grad_norm": 1.71875, "learning_rate": 6.154745782181009e-06, "loss": 0.1229, "step": 24926 }, { "epoch": 154.82866043613708, "grad_norm": 2.578125, "learning_rate": 6.153126954540519e-06, "loss": 0.0945, "step": 24927 }, { "epoch": 154.8348909657321, "grad_norm": 1.828125, "learning_rate": 6.151508309944631e-06, "loss": 0.1357, "step": 24928 }, { "epoch": 154.84112149532712, "grad_norm": 2.75, "learning_rate": 6.149889848409066e-06, "loss": 0.1987, "step": 24929 }, { "epoch": 154.84735202492212, "grad_norm": 1.5703125, "learning_rate": 6.148271569949526e-06, "loss": 0.1077, "step": 24930 }, { "epoch": 154.85358255451715, "grad_norm": 1.671875, "learning_rate": 6.146653474581745e-06, "loss": 0.1514, "step": 24931 }, { "epoch": 154.85981308411215, "grad_norm": 1.2109375, "learning_rate": 6.145035562321441e-06, "loss": 0.1079, "step": 24932 }, { "epoch": 154.86604361370718, "grad_norm": 1.4765625, "learning_rate": 6.143417833184312e-06, "loss": 0.1495, "step": 24933 }, { "epoch": 154.87227414330218, "grad_norm": 1.40625, "learning_rate": 6.1418002871860805e-06, "loss": 0.1307, "step": 24934 }, { "epoch": 154.8785046728972, "grad_norm": 3.6875, "learning_rate": 6.140182924342461e-06, "loss": 0.3083, "step": 24935 }, { "epoch": 154.8847352024922, "grad_norm": 2.078125, "learning_rate": 6.138565744669153e-06, "loss": 0.1159, "step": 24936 }, { "epoch": 154.89096573208724, "grad_norm": 3.609375, "learning_rate": 6.136948748181859e-06, "loss": 0.3158, "step": 24937 }, { "epoch": 154.89719626168224, "grad_norm": 2.65625, "learning_rate": 6.135331934896288e-06, "loss": 0.1979, "step": 24938 }, { "epoch": 154.90342679127727, "grad_norm": 1.8671875, "learning_rate": 6.1337153048281524e-06, "loss": 0.0972, "step": 24939 }, { "epoch": 154.90965732087227, "grad_norm": 2.5, "learning_rate": 6.132098857993137e-06, "loss": 0.2082, "step": 24940 }, { "epoch": 154.9158878504673, "grad_norm": 2.03125, "learning_rate": 6.130482594406953e-06, "loss": 0.1281, "step": 24941 }, { "epoch": 154.9221183800623, "grad_norm": 1.0625, "learning_rate": 6.128866514085296e-06, "loss": 0.0886, "step": 24942 }, { "epoch": 154.92834890965733, "grad_norm": 2.921875, "learning_rate": 6.1272506170438616e-06, "loss": 0.2143, "step": 24943 }, { "epoch": 154.93457943925233, "grad_norm": 2.09375, "learning_rate": 6.1256349032983355e-06, "loss": 0.1644, "step": 24944 }, { "epoch": 154.94080996884736, "grad_norm": 2.578125, "learning_rate": 6.124019372864417e-06, "loss": 0.1719, "step": 24945 }, { "epoch": 154.94704049844236, "grad_norm": 1.7734375, "learning_rate": 6.122404025757794e-06, "loss": 0.1657, "step": 24946 }, { "epoch": 154.9532710280374, "grad_norm": 1.3984375, "learning_rate": 6.120788861994164e-06, "loss": 0.108, "step": 24947 }, { "epoch": 154.9595015576324, "grad_norm": 1.8984375, "learning_rate": 6.119173881589205e-06, "loss": 0.1415, "step": 24948 }, { "epoch": 154.96573208722742, "grad_norm": 3.296875, "learning_rate": 6.1175590845585995e-06, "loss": 0.2792, "step": 24949 }, { "epoch": 154.97196261682242, "grad_norm": 2.1875, "learning_rate": 6.115944470918031e-06, "loss": 0.2164, "step": 24950 }, { "epoch": 154.97819314641745, "grad_norm": 1.0234375, "learning_rate": 6.1143300406831945e-06, "loss": 0.0959, "step": 24951 }, { "epoch": 154.98442367601245, "grad_norm": 2.578125, "learning_rate": 6.112715793869753e-06, "loss": 0.2097, "step": 24952 }, { "epoch": 154.99065420560748, "grad_norm": 2.375, "learning_rate": 6.111101730493391e-06, "loss": 0.1876, "step": 24953 }, { "epoch": 154.99688473520249, "grad_norm": 1.6796875, "learning_rate": 6.109487850569789e-06, "loss": 0.118, "step": 24954 }, { "epoch": 155.0, "grad_norm": 1.7421875, "learning_rate": 6.107874154114621e-06, "loss": 0.0878, "step": 24955 }, { "epoch": 155.00623052959503, "grad_norm": 1.265625, "learning_rate": 6.106260641143546e-06, "loss": 0.1293, "step": 24956 }, { "epoch": 155.01246105919003, "grad_norm": 2.25, "learning_rate": 6.104647311672246e-06, "loss": 0.1183, "step": 24957 }, { "epoch": 155.01869158878506, "grad_norm": 1.765625, "learning_rate": 6.1030341657163925e-06, "loss": 0.1009, "step": 24958 }, { "epoch": 155.02492211838006, "grad_norm": 1.8671875, "learning_rate": 6.101421203291643e-06, "loss": 0.1729, "step": 24959 }, { "epoch": 155.0311526479751, "grad_norm": 1.046875, "learning_rate": 6.099808424413669e-06, "loss": 0.096, "step": 24960 }, { "epoch": 155.0373831775701, "grad_norm": 2.15625, "learning_rate": 6.0981958290981435e-06, "loss": 0.202, "step": 24961 }, { "epoch": 155.04361370716512, "grad_norm": 1.1328125, "learning_rate": 6.096583417360715e-06, "loss": 0.1151, "step": 24962 }, { "epoch": 155.04984423676012, "grad_norm": 2.1875, "learning_rate": 6.094971189217042e-06, "loss": 0.1843, "step": 24963 }, { "epoch": 155.05607476635515, "grad_norm": 2.015625, "learning_rate": 6.0933591446827865e-06, "loss": 0.1737, "step": 24964 }, { "epoch": 155.06230529595015, "grad_norm": 1.3671875, "learning_rate": 6.091747283773613e-06, "loss": 0.1135, "step": 24965 }, { "epoch": 155.06853582554518, "grad_norm": 1.453125, "learning_rate": 6.090135606505162e-06, "loss": 0.1053, "step": 24966 }, { "epoch": 155.07476635514018, "grad_norm": 2.9375, "learning_rate": 6.088524112893104e-06, "loss": 0.1015, "step": 24967 }, { "epoch": 155.0809968847352, "grad_norm": 1.4765625, "learning_rate": 6.086912802953071e-06, "loss": 0.1126, "step": 24968 }, { "epoch": 155.08722741433021, "grad_norm": 3.296875, "learning_rate": 6.085301676700731e-06, "loss": 0.2427, "step": 24969 }, { "epoch": 155.09345794392524, "grad_norm": 1.078125, "learning_rate": 6.083690734151712e-06, "loss": 0.0988, "step": 24970 }, { "epoch": 155.09968847352025, "grad_norm": 2.53125, "learning_rate": 6.08207997532167e-06, "loss": 0.1237, "step": 24971 }, { "epoch": 155.10591900311528, "grad_norm": 1.34375, "learning_rate": 6.080469400226249e-06, "loss": 0.1122, "step": 24972 }, { "epoch": 155.11214953271028, "grad_norm": 1.4765625, "learning_rate": 6.0788590088811e-06, "loss": 0.1338, "step": 24973 }, { "epoch": 155.1183800623053, "grad_norm": 1.8203125, "learning_rate": 6.0772488013018515e-06, "loss": 0.1605, "step": 24974 }, { "epoch": 155.1246105919003, "grad_norm": 2.71875, "learning_rate": 6.075638777504142e-06, "loss": 0.2598, "step": 24975 }, { "epoch": 155.13084112149534, "grad_norm": 1.6796875, "learning_rate": 6.074028937503609e-06, "loss": 0.1531, "step": 24976 }, { "epoch": 155.13707165109034, "grad_norm": 1.6640625, "learning_rate": 6.0724192813159e-06, "loss": 0.1279, "step": 24977 }, { "epoch": 155.14330218068537, "grad_norm": 2.0625, "learning_rate": 6.070809808956629e-06, "loss": 0.1028, "step": 24978 }, { "epoch": 155.14953271028037, "grad_norm": 2.15625, "learning_rate": 6.069200520441437e-06, "loss": 0.2469, "step": 24979 }, { "epoch": 155.1557632398754, "grad_norm": 1.8984375, "learning_rate": 6.067591415785961e-06, "loss": 0.1703, "step": 24980 }, { "epoch": 155.1619937694704, "grad_norm": 1.328125, "learning_rate": 6.065982495005823e-06, "loss": 0.1245, "step": 24981 }, { "epoch": 155.16822429906543, "grad_norm": 1.359375, "learning_rate": 6.064373758116637e-06, "loss": 0.1272, "step": 24982 }, { "epoch": 155.17445482866043, "grad_norm": 0.984375, "learning_rate": 6.062765205134041e-06, "loss": 0.1178, "step": 24983 }, { "epoch": 155.18068535825546, "grad_norm": 1.796875, "learning_rate": 6.061156836073664e-06, "loss": 0.1623, "step": 24984 }, { "epoch": 155.18691588785046, "grad_norm": 2.375, "learning_rate": 6.059548650951108e-06, "loss": 0.2459, "step": 24985 }, { "epoch": 155.1931464174455, "grad_norm": 2.515625, "learning_rate": 6.057940649782009e-06, "loss": 0.2161, "step": 24986 }, { "epoch": 155.1993769470405, "grad_norm": 2.1875, "learning_rate": 6.0563328325819705e-06, "loss": 0.2108, "step": 24987 }, { "epoch": 155.20560747663552, "grad_norm": 1.2578125, "learning_rate": 6.054725199366621e-06, "loss": 0.1205, "step": 24988 }, { "epoch": 155.21183800623052, "grad_norm": 3.203125, "learning_rate": 6.053117750151557e-06, "loss": 0.2283, "step": 24989 }, { "epoch": 155.21806853582555, "grad_norm": 0.94140625, "learning_rate": 6.0515104849524055e-06, "loss": 0.1135, "step": 24990 }, { "epoch": 155.22429906542055, "grad_norm": 1.546875, "learning_rate": 6.049903403784776e-06, "loss": 0.119, "step": 24991 }, { "epoch": 155.23052959501558, "grad_norm": 1.5546875, "learning_rate": 6.0482965066642695e-06, "loss": 0.1113, "step": 24992 }, { "epoch": 155.23676012461058, "grad_norm": 1.9140625, "learning_rate": 6.0466897936065e-06, "loss": 0.134, "step": 24993 }, { "epoch": 155.2429906542056, "grad_norm": 1.2421875, "learning_rate": 6.04508326462706e-06, "loss": 0.1121, "step": 24994 }, { "epoch": 155.2492211838006, "grad_norm": 3.0625, "learning_rate": 6.043476919741569e-06, "loss": 0.2262, "step": 24995 }, { "epoch": 155.25545171339564, "grad_norm": 2.15625, "learning_rate": 6.041870758965612e-06, "loss": 0.224, "step": 24996 }, { "epoch": 155.26168224299064, "grad_norm": 2.828125, "learning_rate": 6.040264782314795e-06, "loss": 0.3081, "step": 24997 }, { "epoch": 155.26791277258567, "grad_norm": 2.078125, "learning_rate": 6.038658989804724e-06, "loss": 0.2013, "step": 24998 }, { "epoch": 155.27414330218068, "grad_norm": 2.453125, "learning_rate": 6.037053381450989e-06, "loss": 0.1785, "step": 24999 }, { "epoch": 155.2803738317757, "grad_norm": 2.75, "learning_rate": 6.035447957269172e-06, "loss": 0.3209, "step": 25000 }, { "epoch": 155.2866043613707, "grad_norm": 2.453125, "learning_rate": 6.033842717274876e-06, "loss": 0.1686, "step": 25001 }, { "epoch": 155.29283489096574, "grad_norm": 1.109375, "learning_rate": 6.0322376614837004e-06, "loss": 0.1131, "step": 25002 }, { "epoch": 155.29906542056074, "grad_norm": 2.03125, "learning_rate": 6.030632789911214e-06, "loss": 0.1645, "step": 25003 }, { "epoch": 155.30529595015577, "grad_norm": 2.15625, "learning_rate": 6.029028102573017e-06, "loss": 0.2046, "step": 25004 }, { "epoch": 155.31152647975077, "grad_norm": 2.078125, "learning_rate": 6.027423599484697e-06, "loss": 0.171, "step": 25005 }, { "epoch": 155.3177570093458, "grad_norm": 1.3515625, "learning_rate": 6.025819280661832e-06, "loss": 0.1001, "step": 25006 }, { "epoch": 155.3239875389408, "grad_norm": 2.15625, "learning_rate": 6.024215146119994e-06, "loss": 0.1106, "step": 25007 }, { "epoch": 155.33021806853583, "grad_norm": 1.734375, "learning_rate": 6.022611195874772e-06, "loss": 0.1245, "step": 25008 }, { "epoch": 155.33644859813083, "grad_norm": 1.546875, "learning_rate": 6.021007429941744e-06, "loss": 0.122, "step": 25009 }, { "epoch": 155.34267912772586, "grad_norm": 1.7421875, "learning_rate": 6.019403848336494e-06, "loss": 0.1226, "step": 25010 }, { "epoch": 155.34890965732086, "grad_norm": 2.40625, "learning_rate": 6.017800451074579e-06, "loss": 0.1717, "step": 25011 }, { "epoch": 155.3551401869159, "grad_norm": 1.40625, "learning_rate": 6.016197238171589e-06, "loss": 0.1373, "step": 25012 }, { "epoch": 155.3613707165109, "grad_norm": 2.484375, "learning_rate": 6.014594209643079e-06, "loss": 0.1221, "step": 25013 }, { "epoch": 155.36760124610592, "grad_norm": 1.28125, "learning_rate": 6.012991365504633e-06, "loss": 0.0989, "step": 25014 }, { "epoch": 155.37383177570092, "grad_norm": 1.984375, "learning_rate": 6.0113887057718e-06, "loss": 0.2101, "step": 25015 }, { "epoch": 155.38006230529595, "grad_norm": 2.421875, "learning_rate": 6.00978623046016e-06, "loss": 0.2029, "step": 25016 }, { "epoch": 155.38629283489095, "grad_norm": 2.734375, "learning_rate": 6.008183939585277e-06, "loss": 0.1386, "step": 25017 }, { "epoch": 155.39252336448598, "grad_norm": 2.6875, "learning_rate": 6.006581833162708e-06, "loss": 0.1653, "step": 25018 }, { "epoch": 155.39875389408098, "grad_norm": 2.4375, "learning_rate": 6.004979911208006e-06, "loss": 0.3128, "step": 25019 }, { "epoch": 155.404984423676, "grad_norm": 2.40625, "learning_rate": 6.003378173736734e-06, "loss": 0.169, "step": 25020 }, { "epoch": 155.41121495327104, "grad_norm": 1.9375, "learning_rate": 6.001776620764463e-06, "loss": 0.1218, "step": 25021 }, { "epoch": 155.41744548286604, "grad_norm": 2.234375, "learning_rate": 6.000175252306723e-06, "loss": 0.2207, "step": 25022 }, { "epoch": 155.42367601246107, "grad_norm": 1.7890625, "learning_rate": 5.998574068379081e-06, "loss": 0.1221, "step": 25023 }, { "epoch": 155.42990654205607, "grad_norm": 1.640625, "learning_rate": 5.996973068997091e-06, "loss": 0.1297, "step": 25024 }, { "epoch": 155.4361370716511, "grad_norm": 3.21875, "learning_rate": 5.995372254176301e-06, "loss": 0.1462, "step": 25025 }, { "epoch": 155.4423676012461, "grad_norm": 3.109375, "learning_rate": 5.9937716239322455e-06, "loss": 0.2916, "step": 25026 }, { "epoch": 155.44859813084113, "grad_norm": 2.1875, "learning_rate": 5.992171178280478e-06, "loss": 0.2035, "step": 25027 }, { "epoch": 155.45482866043614, "grad_norm": 1.2734375, "learning_rate": 5.990570917236552e-06, "loss": 0.1205, "step": 25028 }, { "epoch": 155.46105919003116, "grad_norm": 1.2890625, "learning_rate": 5.988970840815994e-06, "loss": 0.128, "step": 25029 }, { "epoch": 155.46728971962617, "grad_norm": 1.421875, "learning_rate": 5.9873709490343514e-06, "loss": 0.14, "step": 25030 }, { "epoch": 155.4735202492212, "grad_norm": 1.6171875, "learning_rate": 5.98577124190717e-06, "loss": 0.1374, "step": 25031 }, { "epoch": 155.4797507788162, "grad_norm": 2.203125, "learning_rate": 5.984171719449979e-06, "loss": 0.1075, "step": 25032 }, { "epoch": 155.48598130841123, "grad_norm": 1.7265625, "learning_rate": 5.9825723816783054e-06, "loss": 0.19, "step": 25033 }, { "epoch": 155.49221183800623, "grad_norm": 2.609375, "learning_rate": 5.980973228607692e-06, "loss": 0.2344, "step": 25034 }, { "epoch": 155.49844236760126, "grad_norm": 2.140625, "learning_rate": 5.9793742602536755e-06, "loss": 0.1825, "step": 25035 }, { "epoch": 155.50467289719626, "grad_norm": 1.3359375, "learning_rate": 5.97777547663177e-06, "loss": 0.1555, "step": 25036 }, { "epoch": 155.5109034267913, "grad_norm": 2.75, "learning_rate": 5.97617687775752e-06, "loss": 0.2037, "step": 25037 }, { "epoch": 155.5171339563863, "grad_norm": 1.34375, "learning_rate": 5.9745784636464325e-06, "loss": 0.1211, "step": 25038 }, { "epoch": 155.52336448598132, "grad_norm": 1.578125, "learning_rate": 5.972980234314051e-06, "loss": 0.1641, "step": 25039 }, { "epoch": 155.52959501557632, "grad_norm": 1.578125, "learning_rate": 5.9713821897758845e-06, "loss": 0.1349, "step": 25040 }, { "epoch": 155.53582554517135, "grad_norm": 0.875, "learning_rate": 5.969784330047456e-06, "loss": 0.0894, "step": 25041 }, { "epoch": 155.54205607476635, "grad_norm": 1.75, "learning_rate": 5.968186655144286e-06, "loss": 0.1226, "step": 25042 }, { "epoch": 155.54828660436138, "grad_norm": 2.28125, "learning_rate": 5.966589165081898e-06, "loss": 0.2189, "step": 25043 }, { "epoch": 155.55451713395638, "grad_norm": 1.796875, "learning_rate": 5.9649918598758014e-06, "loss": 0.2028, "step": 25044 }, { "epoch": 155.5607476635514, "grad_norm": 1.9765625, "learning_rate": 5.963394739541503e-06, "loss": 0.1303, "step": 25045 }, { "epoch": 155.5669781931464, "grad_norm": 2.40625, "learning_rate": 5.961797804094518e-06, "loss": 0.1496, "step": 25046 }, { "epoch": 155.57320872274144, "grad_norm": 2.03125, "learning_rate": 5.960201053550365e-06, "loss": 0.1754, "step": 25047 }, { "epoch": 155.57943925233644, "grad_norm": 0.6484375, "learning_rate": 5.9586044879245406e-06, "loss": 0.0833, "step": 25048 }, { "epoch": 155.58566978193147, "grad_norm": 2.828125, "learning_rate": 5.957008107232556e-06, "loss": 0.2975, "step": 25049 }, { "epoch": 155.59190031152647, "grad_norm": 3.03125, "learning_rate": 5.9554119114899225e-06, "loss": 0.2149, "step": 25050 }, { "epoch": 155.5981308411215, "grad_norm": 2.453125, "learning_rate": 5.953815900712131e-06, "loss": 0.1346, "step": 25051 }, { "epoch": 155.6043613707165, "grad_norm": 1.8203125, "learning_rate": 5.952220074914686e-06, "loss": 0.1775, "step": 25052 }, { "epoch": 155.61059190031153, "grad_norm": 3.59375, "learning_rate": 5.950624434113084e-06, "loss": 0.2541, "step": 25053 }, { "epoch": 155.61682242990653, "grad_norm": 1.7109375, "learning_rate": 5.949028978322832e-06, "loss": 0.1323, "step": 25054 }, { "epoch": 155.62305295950156, "grad_norm": 1.953125, "learning_rate": 5.947433707559411e-06, "loss": 0.175, "step": 25055 }, { "epoch": 155.62928348909657, "grad_norm": 2.359375, "learning_rate": 5.9458386218383315e-06, "loss": 0.1747, "step": 25056 }, { "epoch": 155.6355140186916, "grad_norm": 2.421875, "learning_rate": 5.944243721175066e-06, "loss": 0.1579, "step": 25057 }, { "epoch": 155.6417445482866, "grad_norm": 1.3984375, "learning_rate": 5.942649005585124e-06, "loss": 0.133, "step": 25058 }, { "epoch": 155.64797507788163, "grad_norm": 2.21875, "learning_rate": 5.941054475083979e-06, "loss": 0.1641, "step": 25059 }, { "epoch": 155.65420560747663, "grad_norm": 1.1640625, "learning_rate": 5.939460129687122e-06, "loss": 0.1166, "step": 25060 }, { "epoch": 155.66043613707166, "grad_norm": 1.328125, "learning_rate": 5.937865969410042e-06, "loss": 0.0769, "step": 25061 }, { "epoch": 155.66666666666666, "grad_norm": 2.28125, "learning_rate": 5.936271994268214e-06, "loss": 0.1657, "step": 25062 }, { "epoch": 155.6728971962617, "grad_norm": 1.3359375, "learning_rate": 5.934678204277131e-06, "loss": 0.1024, "step": 25063 }, { "epoch": 155.6791277258567, "grad_norm": 1.4375, "learning_rate": 5.933084599452254e-06, "loss": 0.1223, "step": 25064 }, { "epoch": 155.68535825545172, "grad_norm": 2.734375, "learning_rate": 5.931491179809082e-06, "loss": 0.2725, "step": 25065 }, { "epoch": 155.69158878504672, "grad_norm": 1.1953125, "learning_rate": 5.929897945363069e-06, "loss": 0.1147, "step": 25066 }, { "epoch": 155.69781931464175, "grad_norm": 2.828125, "learning_rate": 5.9283048961297e-06, "loss": 0.2029, "step": 25067 }, { "epoch": 155.70404984423675, "grad_norm": 1.390625, "learning_rate": 5.926712032124446e-06, "loss": 0.1213, "step": 25068 }, { "epoch": 155.71028037383178, "grad_norm": 1.546875, "learning_rate": 5.925119353362793e-06, "loss": 0.1353, "step": 25069 }, { "epoch": 155.71651090342678, "grad_norm": 1.5390625, "learning_rate": 5.923526859860179e-06, "loss": 0.1475, "step": 25070 }, { "epoch": 155.7227414330218, "grad_norm": 1.5390625, "learning_rate": 5.921934551632085e-06, "loss": 0.1256, "step": 25071 }, { "epoch": 155.7289719626168, "grad_norm": 2.828125, "learning_rate": 5.920342428693976e-06, "loss": 0.229, "step": 25072 }, { "epoch": 155.73520249221184, "grad_norm": 2.515625, "learning_rate": 5.918750491061323e-06, "loss": 0.238, "step": 25073 }, { "epoch": 155.74143302180684, "grad_norm": 2.3125, "learning_rate": 5.917158738749573e-06, "loss": 0.1582, "step": 25074 }, { "epoch": 155.74766355140187, "grad_norm": 1.9140625, "learning_rate": 5.915567171774198e-06, "loss": 0.1409, "step": 25075 }, { "epoch": 155.75389408099687, "grad_norm": 1.0703125, "learning_rate": 5.9139757901506415e-06, "loss": 0.1283, "step": 25076 }, { "epoch": 155.7601246105919, "grad_norm": 2.546875, "learning_rate": 5.912384593894377e-06, "loss": 0.286, "step": 25077 }, { "epoch": 155.7663551401869, "grad_norm": 2.09375, "learning_rate": 5.910793583020841e-06, "loss": 0.1646, "step": 25078 }, { "epoch": 155.77258566978193, "grad_norm": 1.0625, "learning_rate": 5.9092027575454924e-06, "loss": 0.1016, "step": 25079 }, { "epoch": 155.77881619937693, "grad_norm": 1.390625, "learning_rate": 5.90761211748379e-06, "loss": 0.1278, "step": 25080 }, { "epoch": 155.78504672897196, "grad_norm": 3.109375, "learning_rate": 5.906021662851169e-06, "loss": 0.2063, "step": 25081 }, { "epoch": 155.79127725856696, "grad_norm": 2.171875, "learning_rate": 5.9044313936630875e-06, "loss": 0.1825, "step": 25082 }, { "epoch": 155.797507788162, "grad_norm": 1.0234375, "learning_rate": 5.902841309934981e-06, "loss": 0.0859, "step": 25083 }, { "epoch": 155.80373831775702, "grad_norm": 1.5234375, "learning_rate": 5.901251411682304e-06, "loss": 0.0925, "step": 25084 }, { "epoch": 155.80996884735202, "grad_norm": 2.484375, "learning_rate": 5.899661698920484e-06, "loss": 0.1427, "step": 25085 }, { "epoch": 155.81619937694705, "grad_norm": 1.375, "learning_rate": 5.898072171664964e-06, "loss": 0.1012, "step": 25086 }, { "epoch": 155.82242990654206, "grad_norm": 2.1875, "learning_rate": 5.896482829931196e-06, "loss": 0.1463, "step": 25087 }, { "epoch": 155.82866043613708, "grad_norm": 2.90625, "learning_rate": 5.894893673734606e-06, "loss": 0.1908, "step": 25088 }, { "epoch": 155.8348909657321, "grad_norm": 1.2890625, "learning_rate": 5.893304703090619e-06, "loss": 0.1111, "step": 25089 }, { "epoch": 155.84112149532712, "grad_norm": 2.375, "learning_rate": 5.891715918014676e-06, "loss": 0.205, "step": 25090 }, { "epoch": 155.84735202492212, "grad_norm": 2.03125, "learning_rate": 5.890127318522215e-06, "loss": 0.1246, "step": 25091 }, { "epoch": 155.85358255451715, "grad_norm": 2.625, "learning_rate": 5.888538904628652e-06, "loss": 0.274, "step": 25092 }, { "epoch": 155.85981308411215, "grad_norm": 1.671875, "learning_rate": 5.886950676349418e-06, "loss": 0.1349, "step": 25093 }, { "epoch": 155.86604361370718, "grad_norm": 1.7890625, "learning_rate": 5.885362633699945e-06, "loss": 0.0937, "step": 25094 }, { "epoch": 155.87227414330218, "grad_norm": 2.765625, "learning_rate": 5.883774776695652e-06, "loss": 0.2636, "step": 25095 }, { "epoch": 155.8785046728972, "grad_norm": 1.9765625, "learning_rate": 5.882187105351955e-06, "loss": 0.1334, "step": 25096 }, { "epoch": 155.8847352024922, "grad_norm": 1.53125, "learning_rate": 5.880599619684277e-06, "loss": 0.1273, "step": 25097 }, { "epoch": 155.89096573208724, "grad_norm": 2.6875, "learning_rate": 5.879012319708044e-06, "loss": 0.3524, "step": 25098 }, { "epoch": 155.89719626168224, "grad_norm": 1.40625, "learning_rate": 5.877425205438658e-06, "loss": 0.123, "step": 25099 }, { "epoch": 155.90342679127727, "grad_norm": 2.890625, "learning_rate": 5.875838276891543e-06, "loss": 0.2505, "step": 25100 }, { "epoch": 155.90965732087227, "grad_norm": 1.2890625, "learning_rate": 5.874251534082114e-06, "loss": 0.1353, "step": 25101 }, { "epoch": 155.9158878504673, "grad_norm": 2.0625, "learning_rate": 5.8726649770257795e-06, "loss": 0.1453, "step": 25102 }, { "epoch": 155.9221183800623, "grad_norm": 1.53125, "learning_rate": 5.871078605737937e-06, "loss": 0.107, "step": 25103 }, { "epoch": 155.92834890965733, "grad_norm": 2.203125, "learning_rate": 5.869492420234004e-06, "loss": 0.1807, "step": 25104 }, { "epoch": 155.93457943925233, "grad_norm": 2.109375, "learning_rate": 5.867906420529387e-06, "loss": 0.1239, "step": 25105 }, { "epoch": 155.94080996884736, "grad_norm": 1.53125, "learning_rate": 5.866320606639492e-06, "loss": 0.1282, "step": 25106 }, { "epoch": 155.94704049844236, "grad_norm": 2.28125, "learning_rate": 5.864734978579714e-06, "loss": 0.1868, "step": 25107 }, { "epoch": 155.9532710280374, "grad_norm": 1.6875, "learning_rate": 5.863149536365451e-06, "loss": 0.1195, "step": 25108 }, { "epoch": 155.9595015576324, "grad_norm": 1.6796875, "learning_rate": 5.861564280012105e-06, "loss": 0.1078, "step": 25109 }, { "epoch": 155.96573208722742, "grad_norm": 1.4140625, "learning_rate": 5.859979209535077e-06, "loss": 0.162, "step": 25110 }, { "epoch": 155.97196261682242, "grad_norm": 1.1796875, "learning_rate": 5.858394324949753e-06, "loss": 0.123, "step": 25111 }, { "epoch": 155.97819314641745, "grad_norm": 1.828125, "learning_rate": 5.856809626271529e-06, "loss": 0.1094, "step": 25112 }, { "epoch": 155.98442367601245, "grad_norm": 2.203125, "learning_rate": 5.8552251135158e-06, "loss": 0.2109, "step": 25113 }, { "epoch": 155.99065420560748, "grad_norm": 1.6171875, "learning_rate": 5.853640786697956e-06, "loss": 0.1402, "step": 25114 }, { "epoch": 155.99688473520249, "grad_norm": 3.65625, "learning_rate": 5.852056645833371e-06, "loss": 0.1407, "step": 25115 }, { "epoch": 156.0, "grad_norm": 0.83984375, "learning_rate": 5.850472690937439e-06, "loss": 0.0423, "step": 25116 }, { "epoch": 156.00623052959503, "grad_norm": 2.015625, "learning_rate": 5.848888922025553e-06, "loss": 0.2342, "step": 25117 }, { "epoch": 156.01246105919003, "grad_norm": 1.3984375, "learning_rate": 5.847305339113076e-06, "loss": 0.1223, "step": 25118 }, { "epoch": 156.01869158878506, "grad_norm": 1.9453125, "learning_rate": 5.845721942215401e-06, "loss": 0.1124, "step": 25119 }, { "epoch": 156.02492211838006, "grad_norm": 1.3515625, "learning_rate": 5.844138731347909e-06, "loss": 0.0882, "step": 25120 }, { "epoch": 156.0311526479751, "grad_norm": 3.0625, "learning_rate": 5.842555706525973e-06, "loss": 0.3235, "step": 25121 }, { "epoch": 156.0373831775701, "grad_norm": 0.953125, "learning_rate": 5.840972867764954e-06, "loss": 0.0905, "step": 25122 }, { "epoch": 156.04361370716512, "grad_norm": 1.484375, "learning_rate": 5.8393902150802396e-06, "loss": 0.1242, "step": 25123 }, { "epoch": 156.04984423676012, "grad_norm": 1.8515625, "learning_rate": 5.8378077484872066e-06, "loss": 0.1554, "step": 25124 }, { "epoch": 156.05607476635515, "grad_norm": 2.046875, "learning_rate": 5.836225468001205e-06, "loss": 0.1892, "step": 25125 }, { "epoch": 156.06230529595015, "grad_norm": 1.4609375, "learning_rate": 5.834643373637619e-06, "loss": 0.1017, "step": 25126 }, { "epoch": 156.06853582554518, "grad_norm": 1.3359375, "learning_rate": 5.8330614654118016e-06, "loss": 0.1168, "step": 25127 }, { "epoch": 156.07476635514018, "grad_norm": 3.0, "learning_rate": 5.83147974333913e-06, "loss": 0.2485, "step": 25128 }, { "epoch": 156.0809968847352, "grad_norm": 2.078125, "learning_rate": 5.82989820743495e-06, "loss": 0.1568, "step": 25129 }, { "epoch": 156.08722741433021, "grad_norm": 1.421875, "learning_rate": 5.828316857714633e-06, "loss": 0.1339, "step": 25130 }, { "epoch": 156.09345794392524, "grad_norm": 1.2265625, "learning_rate": 5.826735694193542e-06, "loss": 0.1232, "step": 25131 }, { "epoch": 156.09968847352025, "grad_norm": 1.78125, "learning_rate": 5.825154716887018e-06, "loss": 0.1385, "step": 25132 }, { "epoch": 156.10591900311528, "grad_norm": 0.72265625, "learning_rate": 5.8235739258104324e-06, "loss": 0.0953, "step": 25133 }, { "epoch": 156.11214953271028, "grad_norm": 1.8515625, "learning_rate": 5.821993320979122e-06, "loss": 0.2271, "step": 25134 }, { "epoch": 156.1183800623053, "grad_norm": 2.40625, "learning_rate": 5.8204129024084535e-06, "loss": 0.1942, "step": 25135 }, { "epoch": 156.1246105919003, "grad_norm": 2.03125, "learning_rate": 5.8188326701137635e-06, "loss": 0.1857, "step": 25136 }, { "epoch": 156.13084112149534, "grad_norm": 1.0234375, "learning_rate": 5.817252624110403e-06, "loss": 0.1264, "step": 25137 }, { "epoch": 156.13707165109034, "grad_norm": 1.1015625, "learning_rate": 5.815672764413721e-06, "loss": 0.1155, "step": 25138 }, { "epoch": 156.14330218068537, "grad_norm": 2.203125, "learning_rate": 5.8140930910390715e-06, "loss": 0.1606, "step": 25139 }, { "epoch": 156.14953271028037, "grad_norm": 2.109375, "learning_rate": 5.812513604001773e-06, "loss": 0.1716, "step": 25140 }, { "epoch": 156.1557632398754, "grad_norm": 2.125, "learning_rate": 5.810934303317178e-06, "loss": 0.2044, "step": 25141 }, { "epoch": 156.1619937694704, "grad_norm": 2.796875, "learning_rate": 5.809355189000623e-06, "loss": 0.1654, "step": 25142 }, { "epoch": 156.16822429906543, "grad_norm": 2.015625, "learning_rate": 5.8077762610674576e-06, "loss": 0.1581, "step": 25143 }, { "epoch": 156.17445482866043, "grad_norm": 2.3125, "learning_rate": 5.806197519532993e-06, "loss": 0.1639, "step": 25144 }, { "epoch": 156.18068535825546, "grad_norm": 1.484375, "learning_rate": 5.804618964412586e-06, "loss": 0.1213, "step": 25145 }, { "epoch": 156.18691588785046, "grad_norm": 1.5390625, "learning_rate": 5.8030405957215486e-06, "loss": 0.1124, "step": 25146 }, { "epoch": 156.1931464174455, "grad_norm": 2.359375, "learning_rate": 5.8014624134752255e-06, "loss": 0.0959, "step": 25147 }, { "epoch": 156.1993769470405, "grad_norm": 2.671875, "learning_rate": 5.799884417688931e-06, "loss": 0.2147, "step": 25148 }, { "epoch": 156.20560747663552, "grad_norm": 2.28125, "learning_rate": 5.7983066083779955e-06, "loss": 0.1412, "step": 25149 }, { "epoch": 156.21183800623052, "grad_norm": 0.79296875, "learning_rate": 5.79672898555775e-06, "loss": 0.0902, "step": 25150 }, { "epoch": 156.21806853582555, "grad_norm": 2.765625, "learning_rate": 5.795151549243508e-06, "loss": 0.2216, "step": 25151 }, { "epoch": 156.22429906542055, "grad_norm": 0.9296875, "learning_rate": 5.7935742994505994e-06, "loss": 0.0964, "step": 25152 }, { "epoch": 156.23052959501558, "grad_norm": 2.375, "learning_rate": 5.791997236194327e-06, "loss": 0.1882, "step": 25153 }, { "epoch": 156.23676012461058, "grad_norm": 1.109375, "learning_rate": 5.790420359490026e-06, "loss": 0.1088, "step": 25154 }, { "epoch": 156.2429906542056, "grad_norm": 1.0546875, "learning_rate": 5.7888436693529945e-06, "loss": 0.1003, "step": 25155 }, { "epoch": 156.2492211838006, "grad_norm": 2.734375, "learning_rate": 5.7872671657985546e-06, "loss": 0.2115, "step": 25156 }, { "epoch": 156.25545171339564, "grad_norm": 2.734375, "learning_rate": 5.785690848842024e-06, "loss": 0.2421, "step": 25157 }, { "epoch": 156.26168224299064, "grad_norm": 2.71875, "learning_rate": 5.784114718498701e-06, "loss": 0.1764, "step": 25158 }, { "epoch": 156.26791277258567, "grad_norm": 2.0, "learning_rate": 5.782538774783894e-06, "loss": 0.1396, "step": 25159 }, { "epoch": 156.27414330218068, "grad_norm": 1.65625, "learning_rate": 5.780963017712909e-06, "loss": 0.1285, "step": 25160 }, { "epoch": 156.2803738317757, "grad_norm": 2.75, "learning_rate": 5.779387447301063e-06, "loss": 0.153, "step": 25161 }, { "epoch": 156.2866043613707, "grad_norm": 2.078125, "learning_rate": 5.77781206356364e-06, "loss": 0.1856, "step": 25162 }, { "epoch": 156.29283489096574, "grad_norm": 1.078125, "learning_rate": 5.776236866515947e-06, "loss": 0.1017, "step": 25163 }, { "epoch": 156.29906542056074, "grad_norm": 1.5703125, "learning_rate": 5.774661856173286e-06, "loss": 0.1399, "step": 25164 }, { "epoch": 156.30529595015577, "grad_norm": 2.546875, "learning_rate": 5.773087032550966e-06, "loss": 0.1737, "step": 25165 }, { "epoch": 156.31152647975077, "grad_norm": 1.6171875, "learning_rate": 5.771512395664255e-06, "loss": 0.1152, "step": 25166 }, { "epoch": 156.3177570093458, "grad_norm": 2.21875, "learning_rate": 5.769937945528458e-06, "loss": 0.192, "step": 25167 }, { "epoch": 156.3239875389408, "grad_norm": 2.75, "learning_rate": 5.768363682158867e-06, "loss": 0.2035, "step": 25168 }, { "epoch": 156.33021806853583, "grad_norm": 2.015625, "learning_rate": 5.76678960557078e-06, "loss": 0.1166, "step": 25169 }, { "epoch": 156.33644859813083, "grad_norm": 1.40625, "learning_rate": 5.765215715779471e-06, "loss": 0.0987, "step": 25170 }, { "epoch": 156.34267912772586, "grad_norm": 2.203125, "learning_rate": 5.763642012800238e-06, "loss": 0.2487, "step": 25171 }, { "epoch": 156.34890965732086, "grad_norm": 2.6875, "learning_rate": 5.762068496648354e-06, "loss": 0.1629, "step": 25172 }, { "epoch": 156.3551401869159, "grad_norm": 2.0625, "learning_rate": 5.7604951673391114e-06, "loss": 0.1644, "step": 25173 }, { "epoch": 156.3613707165109, "grad_norm": 2.640625, "learning_rate": 5.758922024887778e-06, "loss": 0.1955, "step": 25174 }, { "epoch": 156.36760124610592, "grad_norm": 2.0, "learning_rate": 5.75734906930964e-06, "loss": 0.1696, "step": 25175 }, { "epoch": 156.37383177570092, "grad_norm": 1.953125, "learning_rate": 5.755776300619983e-06, "loss": 0.2719, "step": 25176 }, { "epoch": 156.38006230529595, "grad_norm": 1.734375, "learning_rate": 5.7542037188340725e-06, "loss": 0.1668, "step": 25177 }, { "epoch": 156.38629283489095, "grad_norm": 1.484375, "learning_rate": 5.752631323967175e-06, "loss": 0.129, "step": 25178 }, { "epoch": 156.39252336448598, "grad_norm": 3.234375, "learning_rate": 5.751059116034571e-06, "loss": 0.2289, "step": 25179 }, { "epoch": 156.39875389408098, "grad_norm": 1.7421875, "learning_rate": 5.7494870950515335e-06, "loss": 0.1454, "step": 25180 }, { "epoch": 156.404984423676, "grad_norm": 1.6953125, "learning_rate": 5.747915261033321e-06, "loss": 0.1617, "step": 25181 }, { "epoch": 156.41121495327104, "grad_norm": 1.84375, "learning_rate": 5.746343613995203e-06, "loss": 0.1381, "step": 25182 }, { "epoch": 156.41744548286604, "grad_norm": 1.140625, "learning_rate": 5.744772153952452e-06, "loss": 0.1039, "step": 25183 }, { "epoch": 156.42367601246107, "grad_norm": 1.9296875, "learning_rate": 5.743200880920322e-06, "loss": 0.1296, "step": 25184 }, { "epoch": 156.42990654205607, "grad_norm": 1.84375, "learning_rate": 5.741629794914069e-06, "loss": 0.1453, "step": 25185 }, { "epoch": 156.4361370716511, "grad_norm": 1.671875, "learning_rate": 5.740058895948955e-06, "loss": 0.1321, "step": 25186 }, { "epoch": 156.4423676012461, "grad_norm": 2.21875, "learning_rate": 5.738488184040247e-06, "loss": 0.2046, "step": 25187 }, { "epoch": 156.44859813084113, "grad_norm": 1.6328125, "learning_rate": 5.736917659203184e-06, "loss": 0.1384, "step": 25188 }, { "epoch": 156.45482866043614, "grad_norm": 1.3203125, "learning_rate": 5.735347321453027e-06, "loss": 0.119, "step": 25189 }, { "epoch": 156.46105919003116, "grad_norm": 2.359375, "learning_rate": 5.733777170805038e-06, "loss": 0.1523, "step": 25190 }, { "epoch": 156.46728971962617, "grad_norm": 1.59375, "learning_rate": 5.7322072072744534e-06, "loss": 0.0854, "step": 25191 }, { "epoch": 156.4735202492212, "grad_norm": 1.65625, "learning_rate": 5.730637430876518e-06, "loss": 0.136, "step": 25192 }, { "epoch": 156.4797507788162, "grad_norm": 2.1875, "learning_rate": 5.729067841626484e-06, "loss": 0.1178, "step": 25193 }, { "epoch": 156.48598130841123, "grad_norm": 2.578125, "learning_rate": 5.727498439539603e-06, "loss": 0.164, "step": 25194 }, { "epoch": 156.49221183800623, "grad_norm": 1.1328125, "learning_rate": 5.7259292246310995e-06, "loss": 0.1189, "step": 25195 }, { "epoch": 156.49844236760126, "grad_norm": 2.109375, "learning_rate": 5.7243601969162345e-06, "loss": 0.1905, "step": 25196 }, { "epoch": 156.50467289719626, "grad_norm": 1.40625, "learning_rate": 5.722791356410226e-06, "loss": 0.1461, "step": 25197 }, { "epoch": 156.5109034267913, "grad_norm": 1.7421875, "learning_rate": 5.721222703128332e-06, "loss": 0.1317, "step": 25198 }, { "epoch": 156.5171339563863, "grad_norm": 1.8046875, "learning_rate": 5.719654237085767e-06, "loss": 0.1441, "step": 25199 }, { "epoch": 156.52336448598132, "grad_norm": 1.2890625, "learning_rate": 5.718085958297775e-06, "loss": 0.1194, "step": 25200 }, { "epoch": 156.52959501557632, "grad_norm": 1.8359375, "learning_rate": 5.716517866779588e-06, "loss": 0.1623, "step": 25201 }, { "epoch": 156.53582554517135, "grad_norm": 1.4453125, "learning_rate": 5.7149499625464395e-06, "loss": 0.0898, "step": 25202 }, { "epoch": 156.54205607476635, "grad_norm": 0.88671875, "learning_rate": 5.713382245613552e-06, "loss": 0.0936, "step": 25203 }, { "epoch": 156.54828660436138, "grad_norm": 2.34375, "learning_rate": 5.711814715996147e-06, "loss": 0.2113, "step": 25204 }, { "epoch": 156.55451713395638, "grad_norm": 2.203125, "learning_rate": 5.710247373709448e-06, "loss": 0.1892, "step": 25205 }, { "epoch": 156.5607476635514, "grad_norm": 1.703125, "learning_rate": 5.708680218768694e-06, "loss": 0.1139, "step": 25206 }, { "epoch": 156.5669781931464, "grad_norm": 2.140625, "learning_rate": 5.707113251189086e-06, "loss": 0.1773, "step": 25207 }, { "epoch": 156.57320872274144, "grad_norm": 3.125, "learning_rate": 5.705546470985851e-06, "loss": 0.2449, "step": 25208 }, { "epoch": 156.57943925233644, "grad_norm": 2.40625, "learning_rate": 5.703979878174212e-06, "loss": 0.1234, "step": 25209 }, { "epoch": 156.58566978193147, "grad_norm": 1.234375, "learning_rate": 5.702413472769377e-06, "loss": 0.1255, "step": 25210 }, { "epoch": 156.59190031152647, "grad_norm": 2.125, "learning_rate": 5.700847254786551e-06, "loss": 0.1588, "step": 25211 }, { "epoch": 156.5981308411215, "grad_norm": 2.1875, "learning_rate": 5.6992812242409575e-06, "loss": 0.2902, "step": 25212 }, { "epoch": 156.6043613707165, "grad_norm": 1.953125, "learning_rate": 5.697715381147808e-06, "loss": 0.1142, "step": 25213 }, { "epoch": 156.61059190031153, "grad_norm": 2.625, "learning_rate": 5.696149725522301e-06, "loss": 0.2958, "step": 25214 }, { "epoch": 156.61682242990653, "grad_norm": 3.390625, "learning_rate": 5.694584257379651e-06, "loss": 0.2477, "step": 25215 }, { "epoch": 156.62305295950156, "grad_norm": 1.6875, "learning_rate": 5.693018976735051e-06, "loss": 0.1378, "step": 25216 }, { "epoch": 156.62928348909657, "grad_norm": 1.8203125, "learning_rate": 5.691453883603717e-06, "loss": 0.1976, "step": 25217 }, { "epoch": 156.6355140186916, "grad_norm": 3.140625, "learning_rate": 5.689888978000838e-06, "loss": 0.2473, "step": 25218 }, { "epoch": 156.6417445482866, "grad_norm": 2.109375, "learning_rate": 5.688324259941613e-06, "loss": 0.156, "step": 25219 }, { "epoch": 156.64797507788163, "grad_norm": 1.546875, "learning_rate": 5.6867597294412545e-06, "loss": 0.1136, "step": 25220 }, { "epoch": 156.65420560747663, "grad_norm": 2.21875, "learning_rate": 5.685195386514935e-06, "loss": 0.2128, "step": 25221 }, { "epoch": 156.66043613707166, "grad_norm": 3.21875, "learning_rate": 5.683631231177869e-06, "loss": 0.1995, "step": 25222 }, { "epoch": 156.66666666666666, "grad_norm": 1.796875, "learning_rate": 5.682067263445229e-06, "loss": 0.1627, "step": 25223 }, { "epoch": 156.6728971962617, "grad_norm": 1.21875, "learning_rate": 5.6805034833322225e-06, "loss": 0.1086, "step": 25224 }, { "epoch": 156.6791277258567, "grad_norm": 1.78125, "learning_rate": 5.678939890854018e-06, "loss": 0.1509, "step": 25225 }, { "epoch": 156.68535825545172, "grad_norm": 2.84375, "learning_rate": 5.677376486025815e-06, "loss": 0.3109, "step": 25226 }, { "epoch": 156.69158878504672, "grad_norm": 2.53125, "learning_rate": 5.675813268862798e-06, "loss": 0.1315, "step": 25227 }, { "epoch": 156.69781931464175, "grad_norm": 1.0625, "learning_rate": 5.674250239380146e-06, "loss": 0.0936, "step": 25228 }, { "epoch": 156.70404984423675, "grad_norm": 2.234375, "learning_rate": 5.672687397593032e-06, "loss": 0.2331, "step": 25229 }, { "epoch": 156.71028037383178, "grad_norm": 1.2265625, "learning_rate": 5.671124743516643e-06, "loss": 0.094, "step": 25230 }, { "epoch": 156.71651090342678, "grad_norm": 2.046875, "learning_rate": 5.66956227716616e-06, "loss": 0.1586, "step": 25231 }, { "epoch": 156.7227414330218, "grad_norm": 1.671875, "learning_rate": 5.6679999985567446e-06, "loss": 0.1259, "step": 25232 }, { "epoch": 156.7289719626168, "grad_norm": 2.890625, "learning_rate": 5.666437907703579e-06, "loss": 0.2241, "step": 25233 }, { "epoch": 156.73520249221184, "grad_norm": 2.390625, "learning_rate": 5.664876004621833e-06, "loss": 0.185, "step": 25234 }, { "epoch": 156.74143302180684, "grad_norm": 1.546875, "learning_rate": 5.6633142893266906e-06, "loss": 0.13, "step": 25235 }, { "epoch": 156.74766355140187, "grad_norm": 3.21875, "learning_rate": 5.661752761833289e-06, "loss": 0.2512, "step": 25236 }, { "epoch": 156.75389408099687, "grad_norm": 2.53125, "learning_rate": 5.66019142215681e-06, "loss": 0.2016, "step": 25237 }, { "epoch": 156.7601246105919, "grad_norm": 1.5390625, "learning_rate": 5.658630270312421e-06, "loss": 0.1153, "step": 25238 }, { "epoch": 156.7663551401869, "grad_norm": 1.5390625, "learning_rate": 5.657069306315285e-06, "loss": 0.0959, "step": 25239 }, { "epoch": 156.77258566978193, "grad_norm": 1.0859375, "learning_rate": 5.655508530180553e-06, "loss": 0.0984, "step": 25240 }, { "epoch": 156.77881619937693, "grad_norm": 1.890625, "learning_rate": 5.653947941923396e-06, "loss": 0.1586, "step": 25241 }, { "epoch": 156.78504672897196, "grad_norm": 3.0, "learning_rate": 5.652387541558957e-06, "loss": 0.3089, "step": 25242 }, { "epoch": 156.79127725856696, "grad_norm": 1.5546875, "learning_rate": 5.650827329102404e-06, "loss": 0.1403, "step": 25243 }, { "epoch": 156.797507788162, "grad_norm": 1.078125, "learning_rate": 5.649267304568879e-06, "loss": 0.1005, "step": 25244 }, { "epoch": 156.80373831775702, "grad_norm": 2.6875, "learning_rate": 5.647707467973534e-06, "loss": 0.1944, "step": 25245 }, { "epoch": 156.80996884735202, "grad_norm": 1.0, "learning_rate": 5.646147819331532e-06, "loss": 0.1051, "step": 25246 }, { "epoch": 156.81619937694705, "grad_norm": 2.765625, "learning_rate": 5.644588358658012e-06, "loss": 0.1806, "step": 25247 }, { "epoch": 156.82242990654206, "grad_norm": 3.09375, "learning_rate": 5.64302908596811e-06, "loss": 0.2493, "step": 25248 }, { "epoch": 156.82866043613708, "grad_norm": 1.515625, "learning_rate": 5.64147000127698e-06, "loss": 0.1087, "step": 25249 }, { "epoch": 156.8348909657321, "grad_norm": 2.8125, "learning_rate": 5.639911104599771e-06, "loss": 0.1805, "step": 25250 }, { "epoch": 156.84112149532712, "grad_norm": 0.91796875, "learning_rate": 5.638352395951607e-06, "loss": 0.0933, "step": 25251 }, { "epoch": 156.84735202492212, "grad_norm": 1.9296875, "learning_rate": 5.636793875347635e-06, "loss": 0.0977, "step": 25252 }, { "epoch": 156.85358255451715, "grad_norm": 1.5546875, "learning_rate": 5.635235542802999e-06, "loss": 0.1376, "step": 25253 }, { "epoch": 156.85981308411215, "grad_norm": 1.8125, "learning_rate": 5.633677398332826e-06, "loss": 0.1427, "step": 25254 }, { "epoch": 156.86604361370718, "grad_norm": 3.984375, "learning_rate": 5.632119441952241e-06, "loss": 0.2551, "step": 25255 }, { "epoch": 156.87227414330218, "grad_norm": 2.15625, "learning_rate": 5.630561673676385e-06, "loss": 0.1364, "step": 25256 }, { "epoch": 156.8785046728972, "grad_norm": 2.171875, "learning_rate": 5.629004093520393e-06, "loss": 0.1075, "step": 25257 }, { "epoch": 156.8847352024922, "grad_norm": 1.3984375, "learning_rate": 5.627446701499378e-06, "loss": 0.0867, "step": 25258 }, { "epoch": 156.89096573208724, "grad_norm": 1.3359375, "learning_rate": 5.625889497628472e-06, "loss": 0.1134, "step": 25259 }, { "epoch": 156.89719626168224, "grad_norm": 3.59375, "learning_rate": 5.624332481922809e-06, "loss": 0.3539, "step": 25260 }, { "epoch": 156.90342679127727, "grad_norm": 1.2734375, "learning_rate": 5.622775654397502e-06, "loss": 0.1142, "step": 25261 }, { "epoch": 156.90965732087227, "grad_norm": 1.125, "learning_rate": 5.621219015067663e-06, "loss": 0.1146, "step": 25262 }, { "epoch": 156.9158878504673, "grad_norm": 1.96875, "learning_rate": 5.619662563948419e-06, "loss": 0.1201, "step": 25263 }, { "epoch": 156.9221183800623, "grad_norm": 2.453125, "learning_rate": 5.618106301054887e-06, "loss": 0.1339, "step": 25264 }, { "epoch": 156.92834890965733, "grad_norm": 1.9140625, "learning_rate": 5.616550226402187e-06, "loss": 0.1353, "step": 25265 }, { "epoch": 156.93457943925233, "grad_norm": 1.5, "learning_rate": 5.614994340005428e-06, "loss": 0.1453, "step": 25266 }, { "epoch": 156.94080996884736, "grad_norm": 2.171875, "learning_rate": 5.613438641879709e-06, "loss": 0.1231, "step": 25267 }, { "epoch": 156.94704049844236, "grad_norm": 3.125, "learning_rate": 5.611883132040158e-06, "loss": 0.2348, "step": 25268 }, { "epoch": 156.9532710280374, "grad_norm": 1.8203125, "learning_rate": 5.610327810501864e-06, "loss": 0.1511, "step": 25269 }, { "epoch": 156.9595015576324, "grad_norm": 3.578125, "learning_rate": 5.608772677279947e-06, "loss": 0.2351, "step": 25270 }, { "epoch": 156.96573208722742, "grad_norm": 1.4765625, "learning_rate": 5.607217732389503e-06, "loss": 0.1207, "step": 25271 }, { "epoch": 156.97196261682242, "grad_norm": 3.0, "learning_rate": 5.605662975845643e-06, "loss": 0.2161, "step": 25272 }, { "epoch": 156.97819314641745, "grad_norm": 2.90625, "learning_rate": 5.604108407663464e-06, "loss": 0.3096, "step": 25273 }, { "epoch": 156.98442367601245, "grad_norm": 1.203125, "learning_rate": 5.602554027858051e-06, "loss": 0.1352, "step": 25274 }, { "epoch": 156.99065420560748, "grad_norm": 2.1875, "learning_rate": 5.600999836444512e-06, "loss": 0.096, "step": 25275 }, { "epoch": 156.99688473520249, "grad_norm": 2.875, "learning_rate": 5.599445833437947e-06, "loss": 0.2067, "step": 25276 }, { "epoch": 157.0, "grad_norm": 2.640625, "learning_rate": 5.597892018853437e-06, "loss": 0.1862, "step": 25277 }, { "epoch": 157.00623052959503, "grad_norm": 1.6328125, "learning_rate": 5.596338392706077e-06, "loss": 0.1282, "step": 25278 }, { "epoch": 157.01246105919003, "grad_norm": 2.625, "learning_rate": 5.594784955010965e-06, "loss": 0.2284, "step": 25279 }, { "epoch": 157.01869158878506, "grad_norm": 2.34375, "learning_rate": 5.593231705783178e-06, "loss": 0.1594, "step": 25280 }, { "epoch": 157.02492211838006, "grad_norm": 2.4375, "learning_rate": 5.5916786450377985e-06, "loss": 0.2431, "step": 25281 }, { "epoch": 157.0311526479751, "grad_norm": 2.0625, "learning_rate": 5.590125772789917e-06, "loss": 0.1147, "step": 25282 }, { "epoch": 157.0373831775701, "grad_norm": 1.28125, "learning_rate": 5.588573089054619e-06, "loss": 0.1335, "step": 25283 }, { "epoch": 157.04361370716512, "grad_norm": 1.5234375, "learning_rate": 5.5870205938469725e-06, "loss": 0.1214, "step": 25284 }, { "epoch": 157.04984423676012, "grad_norm": 0.953125, "learning_rate": 5.585468287182071e-06, "loss": 0.0755, "step": 25285 }, { "epoch": 157.05607476635515, "grad_norm": 1.5, "learning_rate": 5.5839161690749765e-06, "loss": 0.1109, "step": 25286 }, { "epoch": 157.06230529595015, "grad_norm": 2.40625, "learning_rate": 5.582364239540774e-06, "loss": 0.1355, "step": 25287 }, { "epoch": 157.06853582554518, "grad_norm": 1.6328125, "learning_rate": 5.580812498594526e-06, "loss": 0.146, "step": 25288 }, { "epoch": 157.07476635514018, "grad_norm": 2.453125, "learning_rate": 5.5792609462513076e-06, "loss": 0.1747, "step": 25289 }, { "epoch": 157.0809968847352, "grad_norm": 1.765625, "learning_rate": 5.577709582526194e-06, "loss": 0.0992, "step": 25290 }, { "epoch": 157.08722741433021, "grad_norm": 2.265625, "learning_rate": 5.5761584074342455e-06, "loss": 0.2079, "step": 25291 }, { "epoch": 157.09345794392524, "grad_norm": 1.734375, "learning_rate": 5.574607420990532e-06, "loss": 0.1154, "step": 25292 }, { "epoch": 157.09968847352025, "grad_norm": 3.28125, "learning_rate": 5.57305662321011e-06, "loss": 0.234, "step": 25293 }, { "epoch": 157.10591900311528, "grad_norm": 1.9296875, "learning_rate": 5.571506014108049e-06, "loss": 0.1579, "step": 25294 }, { "epoch": 157.11214953271028, "grad_norm": 1.640625, "learning_rate": 5.569955593699402e-06, "loss": 0.1365, "step": 25295 }, { "epoch": 157.1183800623053, "grad_norm": 1.203125, "learning_rate": 5.56840536199923e-06, "loss": 0.0919, "step": 25296 }, { "epoch": 157.1246105919003, "grad_norm": 3.390625, "learning_rate": 5.566855319022587e-06, "loss": 0.2052, "step": 25297 }, { "epoch": 157.13084112149534, "grad_norm": 0.828125, "learning_rate": 5.565305464784545e-06, "loss": 0.0771, "step": 25298 }, { "epoch": 157.13707165109034, "grad_norm": 1.2890625, "learning_rate": 5.563755799300127e-06, "loss": 0.1224, "step": 25299 }, { "epoch": 157.14330218068537, "grad_norm": 1.3828125, "learning_rate": 5.562206322584398e-06, "loss": 0.1187, "step": 25300 }, { "epoch": 157.14953271028037, "grad_norm": 3.34375, "learning_rate": 5.560657034652406e-06, "loss": 0.3421, "step": 25301 }, { "epoch": 157.1557632398754, "grad_norm": 1.953125, "learning_rate": 5.559107935519206e-06, "loss": 0.1518, "step": 25302 }, { "epoch": 157.1619937694704, "grad_norm": 1.796875, "learning_rate": 5.557559025199829e-06, "loss": 0.14, "step": 25303 }, { "epoch": 157.16822429906543, "grad_norm": 2.46875, "learning_rate": 5.556010303709328e-06, "loss": 0.1016, "step": 25304 }, { "epoch": 157.17445482866043, "grad_norm": 3.0625, "learning_rate": 5.554461771062747e-06, "loss": 0.2822, "step": 25305 }, { "epoch": 157.18068535825546, "grad_norm": 1.7890625, "learning_rate": 5.552913427275119e-06, "loss": 0.1553, "step": 25306 }, { "epoch": 157.18691588785046, "grad_norm": 3.046875, "learning_rate": 5.551365272361475e-06, "loss": 0.2201, "step": 25307 }, { "epoch": 157.1931464174455, "grad_norm": 2.609375, "learning_rate": 5.549817306336863e-06, "loss": 0.2185, "step": 25308 }, { "epoch": 157.1993769470405, "grad_norm": 1.109375, "learning_rate": 5.5482695292163184e-06, "loss": 0.1149, "step": 25309 }, { "epoch": 157.20560747663552, "grad_norm": 1.6171875, "learning_rate": 5.54672194101486e-06, "loss": 0.152, "step": 25310 }, { "epoch": 157.21183800623052, "grad_norm": 1.28125, "learning_rate": 5.545174541747536e-06, "loss": 0.1138, "step": 25311 }, { "epoch": 157.21806853582555, "grad_norm": 1.5546875, "learning_rate": 5.5436273314293575e-06, "loss": 0.1397, "step": 25312 }, { "epoch": 157.22429906542055, "grad_norm": 3.0625, "learning_rate": 5.542080310075368e-06, "loss": 0.2157, "step": 25313 }, { "epoch": 157.23052959501558, "grad_norm": 2.5, "learning_rate": 5.540533477700577e-06, "loss": 0.2122, "step": 25314 }, { "epoch": 157.23676012461058, "grad_norm": 1.5625, "learning_rate": 5.538986834320012e-06, "loss": 0.1259, "step": 25315 }, { "epoch": 157.2429906542056, "grad_norm": 2.109375, "learning_rate": 5.537440379948708e-06, "loss": 0.1168, "step": 25316 }, { "epoch": 157.2492211838006, "grad_norm": 1.6796875, "learning_rate": 5.53589411460167e-06, "loss": 0.1383, "step": 25317 }, { "epoch": 157.25545171339564, "grad_norm": 1.15625, "learning_rate": 5.5343480382939125e-06, "loss": 0.111, "step": 25318 }, { "epoch": 157.26168224299064, "grad_norm": 1.640625, "learning_rate": 5.5328021510404616e-06, "loss": 0.1377, "step": 25319 }, { "epoch": 157.26791277258567, "grad_norm": 1.96875, "learning_rate": 5.5312564528563316e-06, "loss": 0.1415, "step": 25320 }, { "epoch": 157.27414330218068, "grad_norm": 1.4140625, "learning_rate": 5.529710943756525e-06, "loss": 0.0959, "step": 25321 }, { "epoch": 157.2803738317757, "grad_norm": 2.3125, "learning_rate": 5.528165623756057e-06, "loss": 0.1629, "step": 25322 }, { "epoch": 157.2866043613707, "grad_norm": 2.90625, "learning_rate": 5.526620492869944e-06, "loss": 0.2175, "step": 25323 }, { "epoch": 157.29283489096574, "grad_norm": 1.6484375, "learning_rate": 5.5250755511131845e-06, "loss": 0.1466, "step": 25324 }, { "epoch": 157.29906542056074, "grad_norm": 2.015625, "learning_rate": 5.5235307985007784e-06, "loss": 0.25, "step": 25325 }, { "epoch": 157.30529595015577, "grad_norm": 1.9140625, "learning_rate": 5.521986235047733e-06, "loss": 0.185, "step": 25326 }, { "epoch": 157.31152647975077, "grad_norm": 2.46875, "learning_rate": 5.520441860769057e-06, "loss": 0.0881, "step": 25327 }, { "epoch": 157.3177570093458, "grad_norm": 1.4609375, "learning_rate": 5.5188976756797405e-06, "loss": 0.1082, "step": 25328 }, { "epoch": 157.3239875389408, "grad_norm": 1.578125, "learning_rate": 5.517353679794781e-06, "loss": 0.129, "step": 25329 }, { "epoch": 157.33021806853583, "grad_norm": 1.890625, "learning_rate": 5.5158098731291856e-06, "loss": 0.1806, "step": 25330 }, { "epoch": 157.33644859813083, "grad_norm": 2.421875, "learning_rate": 5.514266255697939e-06, "loss": 0.1826, "step": 25331 }, { "epoch": 157.34267912772586, "grad_norm": 2.625, "learning_rate": 5.512722827516026e-06, "loss": 0.1991, "step": 25332 }, { "epoch": 157.34890965732086, "grad_norm": 2.265625, "learning_rate": 5.511179588598445e-06, "loss": 0.171, "step": 25333 }, { "epoch": 157.3551401869159, "grad_norm": 2.390625, "learning_rate": 5.509636538960184e-06, "loss": 0.1369, "step": 25334 }, { "epoch": 157.3613707165109, "grad_norm": 1.7734375, "learning_rate": 5.5080936786162325e-06, "loss": 0.2443, "step": 25335 }, { "epoch": 157.36760124610592, "grad_norm": 1.8828125, "learning_rate": 5.506551007581576e-06, "loss": 0.1444, "step": 25336 }, { "epoch": 157.37383177570092, "grad_norm": 1.9296875, "learning_rate": 5.505008525871183e-06, "loss": 0.1431, "step": 25337 }, { "epoch": 157.38006230529595, "grad_norm": 2.875, "learning_rate": 5.503466233500046e-06, "loss": 0.2507, "step": 25338 }, { "epoch": 157.38629283489095, "grad_norm": 1.2578125, "learning_rate": 5.501924130483149e-06, "loss": 0.0989, "step": 25339 }, { "epoch": 157.39252336448598, "grad_norm": 1.625, "learning_rate": 5.500382216835454e-06, "loss": 0.1339, "step": 25340 }, { "epoch": 157.39875389408098, "grad_norm": 2.1875, "learning_rate": 5.498840492571947e-06, "loss": 0.2111, "step": 25341 }, { "epoch": 157.404984423676, "grad_norm": 1.625, "learning_rate": 5.497298957707606e-06, "loss": 0.1531, "step": 25342 }, { "epoch": 157.41121495327104, "grad_norm": 1.640625, "learning_rate": 5.495757612257394e-06, "loss": 0.1558, "step": 25343 }, { "epoch": 157.41744548286604, "grad_norm": 2.1875, "learning_rate": 5.494216456236276e-06, "loss": 0.167, "step": 25344 }, { "epoch": 157.42367601246107, "grad_norm": 2.46875, "learning_rate": 5.49267548965923e-06, "loss": 0.2336, "step": 25345 }, { "epoch": 157.42990654205607, "grad_norm": 1.15625, "learning_rate": 5.491134712541224e-06, "loss": 0.1122, "step": 25346 }, { "epoch": 157.4361370716511, "grad_norm": 1.78125, "learning_rate": 5.48959412489721e-06, "loss": 0.197, "step": 25347 }, { "epoch": 157.4423676012461, "grad_norm": 2.078125, "learning_rate": 5.488053726742157e-06, "loss": 0.1538, "step": 25348 }, { "epoch": 157.44859813084113, "grad_norm": 1.890625, "learning_rate": 5.486513518091036e-06, "loss": 0.124, "step": 25349 }, { "epoch": 157.45482866043614, "grad_norm": 1.3359375, "learning_rate": 5.484973498958795e-06, "loss": 0.1054, "step": 25350 }, { "epoch": 157.46105919003116, "grad_norm": 1.140625, "learning_rate": 5.483433669360385e-06, "loss": 0.1357, "step": 25351 }, { "epoch": 157.46728971962617, "grad_norm": 1.2421875, "learning_rate": 5.481894029310769e-06, "loss": 0.1413, "step": 25352 }, { "epoch": 157.4735202492212, "grad_norm": 1.6953125, "learning_rate": 5.480354578824906e-06, "loss": 0.1486, "step": 25353 }, { "epoch": 157.4797507788162, "grad_norm": 2.171875, "learning_rate": 5.478815317917738e-06, "loss": 0.1522, "step": 25354 }, { "epoch": 157.48598130841123, "grad_norm": 2.546875, "learning_rate": 5.477276246604221e-06, "loss": 0.1327, "step": 25355 }, { "epoch": 157.49221183800623, "grad_norm": 3.390625, "learning_rate": 5.475737364899294e-06, "loss": 0.3615, "step": 25356 }, { "epoch": 157.49844236760126, "grad_norm": 1.3359375, "learning_rate": 5.474198672817915e-06, "loss": 0.1229, "step": 25357 }, { "epoch": 157.50467289719626, "grad_norm": 2.671875, "learning_rate": 5.472660170375013e-06, "loss": 0.1759, "step": 25358 }, { "epoch": 157.5109034267913, "grad_norm": 1.3984375, "learning_rate": 5.471121857585543e-06, "loss": 0.1079, "step": 25359 }, { "epoch": 157.5171339563863, "grad_norm": 1.296875, "learning_rate": 5.469583734464437e-06, "loss": 0.1038, "step": 25360 }, { "epoch": 157.52336448598132, "grad_norm": 2.3125, "learning_rate": 5.468045801026647e-06, "loss": 0.2848, "step": 25361 }, { "epoch": 157.52959501557632, "grad_norm": 2.671875, "learning_rate": 5.466508057287101e-06, "loss": 0.1387, "step": 25362 }, { "epoch": 157.53582554517135, "grad_norm": 2.453125, "learning_rate": 5.464970503260725e-06, "loss": 0.243, "step": 25363 }, { "epoch": 157.54205607476635, "grad_norm": 2.15625, "learning_rate": 5.463433138962468e-06, "loss": 0.1646, "step": 25364 }, { "epoch": 157.54828660436138, "grad_norm": 1.3203125, "learning_rate": 5.461895964407246e-06, "loss": 0.1324, "step": 25365 }, { "epoch": 157.55451713395638, "grad_norm": 1.9453125, "learning_rate": 5.460358979609997e-06, "loss": 0.1916, "step": 25366 }, { "epoch": 157.5607476635514, "grad_norm": 1.0390625, "learning_rate": 5.458822184585649e-06, "loss": 0.1057, "step": 25367 }, { "epoch": 157.5669781931464, "grad_norm": 1.7109375, "learning_rate": 5.4572855793491365e-06, "loss": 0.1496, "step": 25368 }, { "epoch": 157.57320872274144, "grad_norm": 2.40625, "learning_rate": 5.45574916391536e-06, "loss": 0.1989, "step": 25369 }, { "epoch": 157.57943925233644, "grad_norm": 2.703125, "learning_rate": 5.454212938299255e-06, "loss": 0.2332, "step": 25370 }, { "epoch": 157.58566978193147, "grad_norm": 2.1875, "learning_rate": 5.452676902515744e-06, "loss": 0.1666, "step": 25371 }, { "epoch": 157.59190031152647, "grad_norm": 2.265625, "learning_rate": 5.451141056579745e-06, "loss": 0.1964, "step": 25372 }, { "epoch": 157.5981308411215, "grad_norm": 1.0, "learning_rate": 5.449605400506166e-06, "loss": 0.0924, "step": 25373 }, { "epoch": 157.6043613707165, "grad_norm": 2.109375, "learning_rate": 5.448069934309929e-06, "loss": 0.1432, "step": 25374 }, { "epoch": 157.61059190031153, "grad_norm": 2.671875, "learning_rate": 5.44653465800595e-06, "loss": 0.1469, "step": 25375 }, { "epoch": 157.61682242990653, "grad_norm": 1.3984375, "learning_rate": 5.444999571609136e-06, "loss": 0.1058, "step": 25376 }, { "epoch": 157.62305295950156, "grad_norm": 1.3671875, "learning_rate": 5.443464675134388e-06, "loss": 0.1212, "step": 25377 }, { "epoch": 157.62928348909657, "grad_norm": 2.875, "learning_rate": 5.441929968596621e-06, "loss": 0.1923, "step": 25378 }, { "epoch": 157.6355140186916, "grad_norm": 1.1171875, "learning_rate": 5.440395452010747e-06, "loss": 0.0982, "step": 25379 }, { "epoch": 157.6417445482866, "grad_norm": 1.4375, "learning_rate": 5.438861125391654e-06, "loss": 0.1031, "step": 25380 }, { "epoch": 157.64797507788163, "grad_norm": 2.625, "learning_rate": 5.43732698875426e-06, "loss": 0.1309, "step": 25381 }, { "epoch": 157.65420560747663, "grad_norm": 1.359375, "learning_rate": 5.435793042113449e-06, "loss": 0.1078, "step": 25382 }, { "epoch": 157.66043613707166, "grad_norm": 2.6875, "learning_rate": 5.434259285484134e-06, "loss": 0.1373, "step": 25383 }, { "epoch": 157.66666666666666, "grad_norm": 1.8984375, "learning_rate": 5.4327257188811955e-06, "loss": 0.1913, "step": 25384 }, { "epoch": 157.6728971962617, "grad_norm": 2.9375, "learning_rate": 5.431192342319538e-06, "loss": 0.2333, "step": 25385 }, { "epoch": 157.6791277258567, "grad_norm": 1.3046875, "learning_rate": 5.42965915581406e-06, "loss": 0.0963, "step": 25386 }, { "epoch": 157.68535825545172, "grad_norm": 3.828125, "learning_rate": 5.428126159379643e-06, "loss": 0.3807, "step": 25387 }, { "epoch": 157.69158878504672, "grad_norm": 1.2109375, "learning_rate": 5.426593353031171e-06, "loss": 0.1398, "step": 25388 }, { "epoch": 157.69781931464175, "grad_norm": 2.0625, "learning_rate": 5.425060736783536e-06, "loss": 0.109, "step": 25389 }, { "epoch": 157.70404984423675, "grad_norm": 2.40625, "learning_rate": 5.423528310651632e-06, "loss": 0.1522, "step": 25390 }, { "epoch": 157.71028037383178, "grad_norm": 1.28125, "learning_rate": 5.421996074650326e-06, "loss": 0.087, "step": 25391 }, { "epoch": 157.71651090342678, "grad_norm": 3.890625, "learning_rate": 5.420464028794509e-06, "loss": 0.1919, "step": 25392 }, { "epoch": 157.7227414330218, "grad_norm": 2.46875, "learning_rate": 5.418932173099059e-06, "loss": 0.1794, "step": 25393 }, { "epoch": 157.7289719626168, "grad_norm": 2.625, "learning_rate": 5.417400507578865e-06, "loss": 0.3203, "step": 25394 }, { "epoch": 157.73520249221184, "grad_norm": 1.8046875, "learning_rate": 5.4158690322487825e-06, "loss": 0.1712, "step": 25395 }, { "epoch": 157.74143302180684, "grad_norm": 1.3515625, "learning_rate": 5.41433774712369e-06, "loss": 0.119, "step": 25396 }, { "epoch": 157.74766355140187, "grad_norm": 2.21875, "learning_rate": 5.412806652218469e-06, "loss": 0.1975, "step": 25397 }, { "epoch": 157.75389408099687, "grad_norm": 2.875, "learning_rate": 5.411275747547989e-06, "loss": 0.2526, "step": 25398 }, { "epoch": 157.7601246105919, "grad_norm": 2.109375, "learning_rate": 5.4097450331271086e-06, "loss": 0.1295, "step": 25399 }, { "epoch": 157.7663551401869, "grad_norm": 2.5, "learning_rate": 5.408214508970707e-06, "loss": 0.1893, "step": 25400 }, { "epoch": 157.77258566978193, "grad_norm": 2.859375, "learning_rate": 5.406684175093638e-06, "loss": 0.2337, "step": 25401 }, { "epoch": 157.77881619937693, "grad_norm": 1.8984375, "learning_rate": 5.405154031510773e-06, "loss": 0.1069, "step": 25402 }, { "epoch": 157.78504672897196, "grad_norm": 3.3125, "learning_rate": 5.403624078236963e-06, "loss": 0.1192, "step": 25403 }, { "epoch": 157.79127725856696, "grad_norm": 1.28125, "learning_rate": 5.402094315287073e-06, "loss": 0.0993, "step": 25404 }, { "epoch": 157.797507788162, "grad_norm": 0.96875, "learning_rate": 5.400564742675968e-06, "loss": 0.1016, "step": 25405 }, { "epoch": 157.80373831775702, "grad_norm": 2.5, "learning_rate": 5.399035360418494e-06, "loss": 0.2075, "step": 25406 }, { "epoch": 157.80996884735202, "grad_norm": 1.96875, "learning_rate": 5.397506168529504e-06, "loss": 0.1467, "step": 25407 }, { "epoch": 157.81619937694705, "grad_norm": 1.3515625, "learning_rate": 5.3959771670238494e-06, "loss": 0.1171, "step": 25408 }, { "epoch": 157.82242990654206, "grad_norm": 2.9375, "learning_rate": 5.39444835591639e-06, "loss": 0.2788, "step": 25409 }, { "epoch": 157.82866043613708, "grad_norm": 1.6953125, "learning_rate": 5.392919735221963e-06, "loss": 0.1017, "step": 25410 }, { "epoch": 157.8348909657321, "grad_norm": 2.578125, "learning_rate": 5.391391304955415e-06, "loss": 0.2663, "step": 25411 }, { "epoch": 157.84112149532712, "grad_norm": 2.4375, "learning_rate": 5.3898630651316035e-06, "loss": 0.2413, "step": 25412 }, { "epoch": 157.84735202492212, "grad_norm": 2.03125, "learning_rate": 5.388335015765361e-06, "loss": 0.1048, "step": 25413 }, { "epoch": 157.85358255451715, "grad_norm": 1.1171875, "learning_rate": 5.386807156871523e-06, "loss": 0.1121, "step": 25414 }, { "epoch": 157.85981308411215, "grad_norm": 1.6328125, "learning_rate": 5.3852794884649364e-06, "loss": 0.1141, "step": 25415 }, { "epoch": 157.86604361370718, "grad_norm": 1.359375, "learning_rate": 5.383752010560441e-06, "loss": 0.1238, "step": 25416 }, { "epoch": 157.87227414330218, "grad_norm": 2.34375, "learning_rate": 5.38222472317286e-06, "loss": 0.2999, "step": 25417 }, { "epoch": 157.8785046728972, "grad_norm": 2.96875, "learning_rate": 5.380697626317036e-06, "loss": 0.3517, "step": 25418 }, { "epoch": 157.8847352024922, "grad_norm": 1.21875, "learning_rate": 5.3791707200078035e-06, "loss": 0.1164, "step": 25419 }, { "epoch": 157.89096573208724, "grad_norm": 2.390625, "learning_rate": 5.377644004259988e-06, "loss": 0.209, "step": 25420 }, { "epoch": 157.89719626168224, "grad_norm": 1.4921875, "learning_rate": 5.376117479088408e-06, "loss": 0.1129, "step": 25421 }, { "epoch": 157.90342679127727, "grad_norm": 2.703125, "learning_rate": 5.374591144507898e-06, "loss": 0.1836, "step": 25422 }, { "epoch": 157.90965732087227, "grad_norm": 1.9296875, "learning_rate": 5.37306500053329e-06, "loss": 0.1376, "step": 25423 }, { "epoch": 157.9158878504673, "grad_norm": 2.390625, "learning_rate": 5.3715390471793914e-06, "loss": 0.1441, "step": 25424 }, { "epoch": 157.9221183800623, "grad_norm": 1.1171875, "learning_rate": 5.370013284461029e-06, "loss": 0.1039, "step": 25425 }, { "epoch": 157.92834890965733, "grad_norm": 1.4453125, "learning_rate": 5.368487712393027e-06, "loss": 0.1141, "step": 25426 }, { "epoch": 157.93457943925233, "grad_norm": 1.453125, "learning_rate": 5.366962330990199e-06, "loss": 0.1213, "step": 25427 }, { "epoch": 157.94080996884736, "grad_norm": 2.453125, "learning_rate": 5.365437140267349e-06, "loss": 0.222, "step": 25428 }, { "epoch": 157.94704049844236, "grad_norm": 1.5859375, "learning_rate": 5.363912140239297e-06, "loss": 0.1471, "step": 25429 }, { "epoch": 157.9532710280374, "grad_norm": 2.4375, "learning_rate": 5.362387330920857e-06, "loss": 0.0853, "step": 25430 }, { "epoch": 157.9595015576324, "grad_norm": 1.5078125, "learning_rate": 5.360862712326845e-06, "loss": 0.1173, "step": 25431 }, { "epoch": 157.96573208722742, "grad_norm": 2.90625, "learning_rate": 5.359338284472057e-06, "loss": 0.1578, "step": 25432 }, { "epoch": 157.97196261682242, "grad_norm": 2.734375, "learning_rate": 5.357814047371298e-06, "loss": 0.1531, "step": 25433 }, { "epoch": 157.97819314641745, "grad_norm": 2.46875, "learning_rate": 5.356290001039374e-06, "loss": 0.1672, "step": 25434 }, { "epoch": 157.98442367601245, "grad_norm": 1.7734375, "learning_rate": 5.354766145491094e-06, "loss": 0.1209, "step": 25435 }, { "epoch": 157.99065420560748, "grad_norm": 1.3515625, "learning_rate": 5.353242480741247e-06, "loss": 0.1336, "step": 25436 }, { "epoch": 157.99688473520249, "grad_norm": 1.3984375, "learning_rate": 5.351719006804634e-06, "loss": 0.1328, "step": 25437 }, { "epoch": 158.0, "grad_norm": 1.09375, "learning_rate": 5.350195723696064e-06, "loss": 0.0478, "step": 25438 }, { "epoch": 158.00623052959503, "grad_norm": 2.171875, "learning_rate": 5.348672631430318e-06, "loss": 0.1666, "step": 25439 }, { "epoch": 158.01246105919003, "grad_norm": 3.140625, "learning_rate": 5.347149730022185e-06, "loss": 0.1734, "step": 25440 }, { "epoch": 158.01869158878506, "grad_norm": 2.1875, "learning_rate": 5.345627019486465e-06, "loss": 0.0982, "step": 25441 }, { "epoch": 158.02492211838006, "grad_norm": 1.8984375, "learning_rate": 5.344104499837948e-06, "loss": 0.1654, "step": 25442 }, { "epoch": 158.0311526479751, "grad_norm": 1.453125, "learning_rate": 5.342582171091412e-06, "loss": 0.1116, "step": 25443 }, { "epoch": 158.0373831775701, "grad_norm": 2.171875, "learning_rate": 5.3410600332616465e-06, "loss": 0.2845, "step": 25444 }, { "epoch": 158.04361370716512, "grad_norm": 1.75, "learning_rate": 5.339538086363444e-06, "loss": 0.1484, "step": 25445 }, { "epoch": 158.04984423676012, "grad_norm": 3.109375, "learning_rate": 5.338016330411577e-06, "loss": 0.2649, "step": 25446 }, { "epoch": 158.05607476635515, "grad_norm": 2.859375, "learning_rate": 5.336494765420819e-06, "loss": 0.2942, "step": 25447 }, { "epoch": 158.06230529595015, "grad_norm": 2.765625, "learning_rate": 5.3349733914059536e-06, "loss": 0.23, "step": 25448 }, { "epoch": 158.06853582554518, "grad_norm": 1.96875, "learning_rate": 5.333452208381765e-06, "loss": 0.1729, "step": 25449 }, { "epoch": 158.07476635514018, "grad_norm": 1.59375, "learning_rate": 5.331931216363012e-06, "loss": 0.1296, "step": 25450 }, { "epoch": 158.0809968847352, "grad_norm": 1.9453125, "learning_rate": 5.330410415364484e-06, "loss": 0.186, "step": 25451 }, { "epoch": 158.08722741433021, "grad_norm": 3.0625, "learning_rate": 5.328889805400933e-06, "loss": 0.3048, "step": 25452 }, { "epoch": 158.09345794392524, "grad_norm": 1.421875, "learning_rate": 5.327369386487144e-06, "loss": 0.1029, "step": 25453 }, { "epoch": 158.09968847352025, "grad_norm": 2.046875, "learning_rate": 5.325849158637869e-06, "loss": 0.2203, "step": 25454 }, { "epoch": 158.10591900311528, "grad_norm": 1.890625, "learning_rate": 5.32432912186788e-06, "loss": 0.1341, "step": 25455 }, { "epoch": 158.11214953271028, "grad_norm": 2.59375, "learning_rate": 5.32280927619194e-06, "loss": 0.2332, "step": 25456 }, { "epoch": 158.1183800623053, "grad_norm": 2.9375, "learning_rate": 5.321289621624823e-06, "loss": 0.2911, "step": 25457 }, { "epoch": 158.1246105919003, "grad_norm": 2.171875, "learning_rate": 5.319770158181259e-06, "loss": 0.1967, "step": 25458 }, { "epoch": 158.13084112149534, "grad_norm": 2.546875, "learning_rate": 5.318250885876022e-06, "loss": 0.1319, "step": 25459 }, { "epoch": 158.13707165109034, "grad_norm": 2.71875, "learning_rate": 5.316731804723876e-06, "loss": 0.2966, "step": 25460 }, { "epoch": 158.14330218068537, "grad_norm": 1.75, "learning_rate": 5.315212914739556e-06, "loss": 0.1799, "step": 25461 }, { "epoch": 158.14953271028037, "grad_norm": 1.2578125, "learning_rate": 5.313694215937826e-06, "loss": 0.0995, "step": 25462 }, { "epoch": 158.1557632398754, "grad_norm": 3.015625, "learning_rate": 5.31217570833343e-06, "loss": 0.0849, "step": 25463 }, { "epoch": 158.1619937694704, "grad_norm": 2.71875, "learning_rate": 5.310657391941135e-06, "loss": 0.2082, "step": 25464 }, { "epoch": 158.16822429906543, "grad_norm": 2.78125, "learning_rate": 5.3091392667756565e-06, "loss": 0.2216, "step": 25465 }, { "epoch": 158.17445482866043, "grad_norm": 2.265625, "learning_rate": 5.307621332851753e-06, "loss": 0.1892, "step": 25466 }, { "epoch": 158.18068535825546, "grad_norm": 2.765625, "learning_rate": 5.306103590184169e-06, "loss": 0.2044, "step": 25467 }, { "epoch": 158.18691588785046, "grad_norm": 3.390625, "learning_rate": 5.304586038787649e-06, "loss": 0.2448, "step": 25468 }, { "epoch": 158.1931464174455, "grad_norm": 1.3359375, "learning_rate": 5.30306867867692e-06, "loss": 0.1098, "step": 25469 }, { "epoch": 158.1993769470405, "grad_norm": 1.484375, "learning_rate": 5.301551509866731e-06, "loss": 0.2095, "step": 25470 }, { "epoch": 158.20560747663552, "grad_norm": 1.7421875, "learning_rate": 5.3000345323718055e-06, "loss": 0.1344, "step": 25471 }, { "epoch": 158.21183800623052, "grad_norm": 2.484375, "learning_rate": 5.298517746206891e-06, "loss": 0.1249, "step": 25472 }, { "epoch": 158.21806853582555, "grad_norm": 2.390625, "learning_rate": 5.297001151386702e-06, "loss": 0.1354, "step": 25473 }, { "epoch": 158.22429906542055, "grad_norm": 1.7421875, "learning_rate": 5.2954847479259775e-06, "loss": 0.0958, "step": 25474 }, { "epoch": 158.23052959501558, "grad_norm": 3.125, "learning_rate": 5.29396853583945e-06, "loss": 0.3802, "step": 25475 }, { "epoch": 158.23676012461058, "grad_norm": 0.98046875, "learning_rate": 5.292452515141838e-06, "loss": 0.1096, "step": 25476 }, { "epoch": 158.2429906542056, "grad_norm": 0.91015625, "learning_rate": 5.290936685847861e-06, "loss": 0.1019, "step": 25477 }, { "epoch": 158.2492211838006, "grad_norm": 2.359375, "learning_rate": 5.2894210479722485e-06, "loss": 0.2226, "step": 25478 }, { "epoch": 158.25545171339564, "grad_norm": 2.453125, "learning_rate": 5.287905601529724e-06, "loss": 0.1711, "step": 25479 }, { "epoch": 158.26168224299064, "grad_norm": 2.609375, "learning_rate": 5.286390346534992e-06, "loss": 0.1607, "step": 25480 }, { "epoch": 158.26791277258567, "grad_norm": 1.15625, "learning_rate": 5.28487528300278e-06, "loss": 0.1224, "step": 25481 }, { "epoch": 158.27414330218068, "grad_norm": 1.734375, "learning_rate": 5.283360410947807e-06, "loss": 0.1248, "step": 25482 }, { "epoch": 158.2803738317757, "grad_norm": 1.65625, "learning_rate": 5.28184573038478e-06, "loss": 0.1381, "step": 25483 }, { "epoch": 158.2866043613707, "grad_norm": 2.1875, "learning_rate": 5.280331241328398e-06, "loss": 0.1707, "step": 25484 }, { "epoch": 158.29283489096574, "grad_norm": 1.7890625, "learning_rate": 5.278816943793383e-06, "loss": 0.1151, "step": 25485 }, { "epoch": 158.29906542056074, "grad_norm": 1.0859375, "learning_rate": 5.277302837794449e-06, "loss": 0.1128, "step": 25486 }, { "epoch": 158.30529595015577, "grad_norm": 1.5, "learning_rate": 5.275788923346284e-06, "loss": 0.1447, "step": 25487 }, { "epoch": 158.31152647975077, "grad_norm": 2.40625, "learning_rate": 5.274275200463599e-06, "loss": 0.2024, "step": 25488 }, { "epoch": 158.3177570093458, "grad_norm": 1.0703125, "learning_rate": 5.272761669161103e-06, "loss": 0.095, "step": 25489 }, { "epoch": 158.3239875389408, "grad_norm": 0.90234375, "learning_rate": 5.27124832945349e-06, "loss": 0.0992, "step": 25490 }, { "epoch": 158.33021806853583, "grad_norm": 2.59375, "learning_rate": 5.269735181355448e-06, "loss": 0.2018, "step": 25491 }, { "epoch": 158.33644859813083, "grad_norm": 2.546875, "learning_rate": 5.268222224881686e-06, "loss": 0.2177, "step": 25492 }, { "epoch": 158.34267912772586, "grad_norm": 1.3671875, "learning_rate": 5.26670946004689e-06, "loss": 0.1176, "step": 25493 }, { "epoch": 158.34890965732086, "grad_norm": 1.1953125, "learning_rate": 5.265196886865764e-06, "loss": 0.1132, "step": 25494 }, { "epoch": 158.3551401869159, "grad_norm": 2.140625, "learning_rate": 5.263684505352987e-06, "loss": 0.1199, "step": 25495 }, { "epoch": 158.3613707165109, "grad_norm": 2.96875, "learning_rate": 5.262172315523256e-06, "loss": 0.1813, "step": 25496 }, { "epoch": 158.36760124610592, "grad_norm": 1.34375, "learning_rate": 5.260660317391245e-06, "loss": 0.098, "step": 25497 }, { "epoch": 158.37383177570092, "grad_norm": 1.7109375, "learning_rate": 5.259148510971654e-06, "loss": 0.1786, "step": 25498 }, { "epoch": 158.38006230529595, "grad_norm": 1.9765625, "learning_rate": 5.257636896279153e-06, "loss": 0.1101, "step": 25499 }, { "epoch": 158.38629283489095, "grad_norm": 2.390625, "learning_rate": 5.25612547332843e-06, "loss": 0.1421, "step": 25500 }, { "epoch": 158.39252336448598, "grad_norm": 2.265625, "learning_rate": 5.254614242134168e-06, "loss": 0.1841, "step": 25501 }, { "epoch": 158.39875389408098, "grad_norm": 1.140625, "learning_rate": 5.25310320271104e-06, "loss": 0.1123, "step": 25502 }, { "epoch": 158.404984423676, "grad_norm": 2.671875, "learning_rate": 5.251592355073715e-06, "loss": 0.2871, "step": 25503 }, { "epoch": 158.41121495327104, "grad_norm": 1.2265625, "learning_rate": 5.25008169923687e-06, "loss": 0.128, "step": 25504 }, { "epoch": 158.41744548286604, "grad_norm": 3.40625, "learning_rate": 5.248571235215189e-06, "loss": 0.3066, "step": 25505 }, { "epoch": 158.42367601246107, "grad_norm": 1.3515625, "learning_rate": 5.247060963023323e-06, "loss": 0.1243, "step": 25506 }, { "epoch": 158.42990654205607, "grad_norm": 2.546875, "learning_rate": 5.245550882675954e-06, "loss": 0.2251, "step": 25507 }, { "epoch": 158.4361370716511, "grad_norm": 1.7890625, "learning_rate": 5.244040994187746e-06, "loss": 0.0979, "step": 25508 }, { "epoch": 158.4423676012461, "grad_norm": 2.390625, "learning_rate": 5.242531297573364e-06, "loss": 0.1486, "step": 25509 }, { "epoch": 158.44859813084113, "grad_norm": 1.3671875, "learning_rate": 5.241021792847459e-06, "loss": 0.1588, "step": 25510 }, { "epoch": 158.45482866043614, "grad_norm": 2.390625, "learning_rate": 5.239512480024699e-06, "loss": 0.1083, "step": 25511 }, { "epoch": 158.46105919003116, "grad_norm": 1.75, "learning_rate": 5.238003359119753e-06, "loss": 0.1295, "step": 25512 }, { "epoch": 158.46728971962617, "grad_norm": 0.91015625, "learning_rate": 5.236494430147259e-06, "loss": 0.1099, "step": 25513 }, { "epoch": 158.4735202492212, "grad_norm": 4.4375, "learning_rate": 5.234985693121886e-06, "loss": 0.5122, "step": 25514 }, { "epoch": 158.4797507788162, "grad_norm": 0.94921875, "learning_rate": 5.233477148058286e-06, "loss": 0.1142, "step": 25515 }, { "epoch": 158.48598130841123, "grad_norm": 2.515625, "learning_rate": 5.231968794971109e-06, "loss": 0.1949, "step": 25516 }, { "epoch": 158.49221183800623, "grad_norm": 1.53125, "learning_rate": 5.230460633874995e-06, "loss": 0.0861, "step": 25517 }, { "epoch": 158.49844236760126, "grad_norm": 2.171875, "learning_rate": 5.2289526647846e-06, "loss": 0.1743, "step": 25518 }, { "epoch": 158.50467289719626, "grad_norm": 2.4375, "learning_rate": 5.227444887714575e-06, "loss": 0.2219, "step": 25519 }, { "epoch": 158.5109034267913, "grad_norm": 1.171875, "learning_rate": 5.225937302679551e-06, "loss": 0.1121, "step": 25520 }, { "epoch": 158.5171339563863, "grad_norm": 2.15625, "learning_rate": 5.224429909694184e-06, "loss": 0.1632, "step": 25521 }, { "epoch": 158.52336448598132, "grad_norm": 2.109375, "learning_rate": 5.222922708773101e-06, "loss": 0.1523, "step": 25522 }, { "epoch": 158.52959501557632, "grad_norm": 2.484375, "learning_rate": 5.221415699930951e-06, "loss": 0.178, "step": 25523 }, { "epoch": 158.53582554517135, "grad_norm": 1.4765625, "learning_rate": 5.219908883182359e-06, "loss": 0.1249, "step": 25524 }, { "epoch": 158.54205607476635, "grad_norm": 2.8125, "learning_rate": 5.218402258541966e-06, "loss": 0.2003, "step": 25525 }, { "epoch": 158.54828660436138, "grad_norm": 1.2421875, "learning_rate": 5.216895826024404e-06, "loss": 0.1238, "step": 25526 }, { "epoch": 158.55451713395638, "grad_norm": 1.390625, "learning_rate": 5.2153895856443175e-06, "loss": 0.121, "step": 25527 }, { "epoch": 158.5607476635514, "grad_norm": 2.828125, "learning_rate": 5.213883537416312e-06, "loss": 0.284, "step": 25528 }, { "epoch": 158.5669781931464, "grad_norm": 1.8125, "learning_rate": 5.212377681355021e-06, "loss": 0.1216, "step": 25529 }, { "epoch": 158.57320872274144, "grad_norm": 2.265625, "learning_rate": 5.2108720174750745e-06, "loss": 0.1939, "step": 25530 }, { "epoch": 158.57943925233644, "grad_norm": 1.859375, "learning_rate": 5.2093665457911e-06, "loss": 0.1284, "step": 25531 }, { "epoch": 158.58566978193147, "grad_norm": 1.59375, "learning_rate": 5.20786126631771e-06, "loss": 0.1519, "step": 25532 }, { "epoch": 158.59190031152647, "grad_norm": 2.234375, "learning_rate": 5.206356179069527e-06, "loss": 0.1579, "step": 25533 }, { "epoch": 158.5981308411215, "grad_norm": 1.5, "learning_rate": 5.204851284061174e-06, "loss": 0.1186, "step": 25534 }, { "epoch": 158.6043613707165, "grad_norm": 2.328125, "learning_rate": 5.203346581307264e-06, "loss": 0.2078, "step": 25535 }, { "epoch": 158.61059190031153, "grad_norm": 3.015625, "learning_rate": 5.201842070822402e-06, "loss": 0.2673, "step": 25536 }, { "epoch": 158.61682242990653, "grad_norm": 2.703125, "learning_rate": 5.2003377526212095e-06, "loss": 0.229, "step": 25537 }, { "epoch": 158.62305295950156, "grad_norm": 1.7734375, "learning_rate": 5.1988336267183e-06, "loss": 0.1549, "step": 25538 }, { "epoch": 158.62928348909657, "grad_norm": 2.28125, "learning_rate": 5.1973296931282664e-06, "loss": 0.1883, "step": 25539 }, { "epoch": 158.6355140186916, "grad_norm": 1.8671875, "learning_rate": 5.195825951865737e-06, "loss": 0.1095, "step": 25540 }, { "epoch": 158.6417445482866, "grad_norm": 2.046875, "learning_rate": 5.194322402945296e-06, "loss": 0.1609, "step": 25541 }, { "epoch": 158.64797507788163, "grad_norm": 1.8515625, "learning_rate": 5.1928190463815616e-06, "loss": 0.0948, "step": 25542 }, { "epoch": 158.65420560747663, "grad_norm": 2.84375, "learning_rate": 5.191315882189121e-06, "loss": 0.2305, "step": 25543 }, { "epoch": 158.66043613707166, "grad_norm": 1.4921875, "learning_rate": 5.18981291038258e-06, "loss": 0.1183, "step": 25544 }, { "epoch": 158.66666666666666, "grad_norm": 1.6171875, "learning_rate": 5.188310130976543e-06, "loss": 0.1361, "step": 25545 }, { "epoch": 158.6728971962617, "grad_norm": 1.9765625, "learning_rate": 5.186807543985597e-06, "loss": 0.1069, "step": 25546 }, { "epoch": 158.6791277258567, "grad_norm": 1.9609375, "learning_rate": 5.185305149424332e-06, "loss": 0.1116, "step": 25547 }, { "epoch": 158.68535825545172, "grad_norm": 1.671875, "learning_rate": 5.183802947307345e-06, "loss": 0.1687, "step": 25548 }, { "epoch": 158.69158878504672, "grad_norm": 1.734375, "learning_rate": 5.182300937649232e-06, "loss": 0.1475, "step": 25549 }, { "epoch": 158.69781931464175, "grad_norm": 2.90625, "learning_rate": 5.1807991204645635e-06, "loss": 0.3516, "step": 25550 }, { "epoch": 158.70404984423675, "grad_norm": 0.89453125, "learning_rate": 5.1792974957679375e-06, "loss": 0.0792, "step": 25551 }, { "epoch": 158.71028037383178, "grad_norm": 1.40625, "learning_rate": 5.177796063573936e-06, "loss": 0.1086, "step": 25552 }, { "epoch": 158.71651090342678, "grad_norm": 1.3515625, "learning_rate": 5.176294823897157e-06, "loss": 0.0935, "step": 25553 }, { "epoch": 158.7227414330218, "grad_norm": 2.359375, "learning_rate": 5.174793776752149e-06, "loss": 0.2722, "step": 25554 }, { "epoch": 158.7289719626168, "grad_norm": 1.984375, "learning_rate": 5.173292922153508e-06, "loss": 0.1664, "step": 25555 }, { "epoch": 158.73520249221184, "grad_norm": 1.7890625, "learning_rate": 5.171792260115818e-06, "loss": 0.2491, "step": 25556 }, { "epoch": 158.74143302180684, "grad_norm": 3.609375, "learning_rate": 5.170291790653636e-06, "loss": 0.1663, "step": 25557 }, { "epoch": 158.74766355140187, "grad_norm": 2.953125, "learning_rate": 5.168791513781546e-06, "loss": 0.2092, "step": 25558 }, { "epoch": 158.75389408099687, "grad_norm": 1.765625, "learning_rate": 5.167291429514121e-06, "loss": 0.1333, "step": 25559 }, { "epoch": 158.7601246105919, "grad_norm": 2.09375, "learning_rate": 5.1657915378659276e-06, "loss": 0.177, "step": 25560 }, { "epoch": 158.7663551401869, "grad_norm": 1.84375, "learning_rate": 5.164291838851526e-06, "loss": 0.1273, "step": 25561 }, { "epoch": 158.77258566978193, "grad_norm": 1.3984375, "learning_rate": 5.162792332485486e-06, "loss": 0.1315, "step": 25562 }, { "epoch": 158.77881619937693, "grad_norm": 1.90625, "learning_rate": 5.16129301878237e-06, "loss": 0.1049, "step": 25563 }, { "epoch": 158.78504672897196, "grad_norm": 0.984375, "learning_rate": 5.1597938977567514e-06, "loss": 0.1099, "step": 25564 }, { "epoch": 158.79127725856696, "grad_norm": 2.125, "learning_rate": 5.158294969423174e-06, "loss": 0.21, "step": 25565 }, { "epoch": 158.797507788162, "grad_norm": 2.75, "learning_rate": 5.156796233796207e-06, "loss": 0.2167, "step": 25566 }, { "epoch": 158.80373831775702, "grad_norm": 2.609375, "learning_rate": 5.155297690890395e-06, "loss": 0.1852, "step": 25567 }, { "epoch": 158.80996884735202, "grad_norm": 0.93359375, "learning_rate": 5.153799340720309e-06, "loss": 0.1097, "step": 25568 }, { "epoch": 158.81619937694705, "grad_norm": 1.734375, "learning_rate": 5.152301183300481e-06, "loss": 0.1551, "step": 25569 }, { "epoch": 158.82242990654206, "grad_norm": 1.921875, "learning_rate": 5.15080321864547e-06, "loss": 0.1542, "step": 25570 }, { "epoch": 158.82866043613708, "grad_norm": 1.8984375, "learning_rate": 5.1493054467698365e-06, "loss": 0.1898, "step": 25571 }, { "epoch": 158.8348909657321, "grad_norm": 1.1015625, "learning_rate": 5.147807867688115e-06, "loss": 0.0875, "step": 25572 }, { "epoch": 158.84112149532712, "grad_norm": 1.765625, "learning_rate": 5.146310481414842e-06, "loss": 0.168, "step": 25573 }, { "epoch": 158.84735202492212, "grad_norm": 2.515625, "learning_rate": 5.144813287964576e-06, "loss": 0.215, "step": 25574 }, { "epoch": 158.85358255451715, "grad_norm": 2.328125, "learning_rate": 5.143316287351854e-06, "loss": 0.1684, "step": 25575 }, { "epoch": 158.85981308411215, "grad_norm": 2.140625, "learning_rate": 5.141819479591209e-06, "loss": 0.0995, "step": 25576 }, { "epoch": 158.86604361370718, "grad_norm": 2.28125, "learning_rate": 5.140322864697183e-06, "loss": 0.1909, "step": 25577 }, { "epoch": 158.87227414330218, "grad_norm": 2.21875, "learning_rate": 5.138826442684316e-06, "loss": 0.2043, "step": 25578 }, { "epoch": 158.8785046728972, "grad_norm": 0.96484375, "learning_rate": 5.13733021356714e-06, "loss": 0.0957, "step": 25579 }, { "epoch": 158.8847352024922, "grad_norm": 2.53125, "learning_rate": 5.135834177360174e-06, "loss": 0.2067, "step": 25580 }, { "epoch": 158.89096573208724, "grad_norm": 2.0, "learning_rate": 5.13433833407796e-06, "loss": 0.1821, "step": 25581 }, { "epoch": 158.89719626168224, "grad_norm": 2.75, "learning_rate": 5.132842683735026e-06, "loss": 0.1472, "step": 25582 }, { "epoch": 158.90342679127727, "grad_norm": 2.203125, "learning_rate": 5.131347226345892e-06, "loss": 0.1534, "step": 25583 }, { "epoch": 158.90965732087227, "grad_norm": 3.109375, "learning_rate": 5.129851961925083e-06, "loss": 0.2198, "step": 25584 }, { "epoch": 158.9158878504673, "grad_norm": 1.703125, "learning_rate": 5.128356890487132e-06, "loss": 0.1593, "step": 25585 }, { "epoch": 158.9221183800623, "grad_norm": 2.21875, "learning_rate": 5.126862012046552e-06, "loss": 0.2221, "step": 25586 }, { "epoch": 158.92834890965733, "grad_norm": 2.109375, "learning_rate": 5.1253673266178525e-06, "loss": 0.1297, "step": 25587 }, { "epoch": 158.93457943925233, "grad_norm": 2.453125, "learning_rate": 5.123872834215557e-06, "loss": 0.1201, "step": 25588 }, { "epoch": 158.94080996884736, "grad_norm": 1.96875, "learning_rate": 5.122378534854183e-06, "loss": 0.1444, "step": 25589 }, { "epoch": 158.94704049844236, "grad_norm": 2.375, "learning_rate": 5.120884428548251e-06, "loss": 0.2404, "step": 25590 }, { "epoch": 158.9532710280374, "grad_norm": 1.7734375, "learning_rate": 5.119390515312259e-06, "loss": 0.1409, "step": 25591 }, { "epoch": 158.9595015576324, "grad_norm": 1.234375, "learning_rate": 5.117896795160718e-06, "loss": 0.1219, "step": 25592 }, { "epoch": 158.96573208722742, "grad_norm": 1.1015625, "learning_rate": 5.116403268108133e-06, "loss": 0.0947, "step": 25593 }, { "epoch": 158.97196261682242, "grad_norm": 0.78515625, "learning_rate": 5.1149099341690235e-06, "loss": 0.0974, "step": 25594 }, { "epoch": 158.97819314641745, "grad_norm": 2.46875, "learning_rate": 5.113416793357878e-06, "loss": 0.355, "step": 25595 }, { "epoch": 158.98442367601245, "grad_norm": 1.8984375, "learning_rate": 5.111923845689201e-06, "loss": 0.1901, "step": 25596 }, { "epoch": 158.99065420560748, "grad_norm": 2.140625, "learning_rate": 5.110431091177506e-06, "loss": 0.229, "step": 25597 }, { "epoch": 158.99688473520249, "grad_norm": 0.99609375, "learning_rate": 5.108938529837276e-06, "loss": 0.1093, "step": 25598 }, { "epoch": 159.0, "grad_norm": 1.2421875, "learning_rate": 5.107446161683008e-06, "loss": 0.0739, "step": 25599 }, { "epoch": 159.00623052959503, "grad_norm": 0.8359375, "learning_rate": 5.105953986729195e-06, "loss": 0.0777, "step": 25600 }, { "epoch": 159.01246105919003, "grad_norm": 1.0078125, "learning_rate": 5.104462004990343e-06, "loss": 0.0863, "step": 25601 }, { "epoch": 159.01869158878506, "grad_norm": 2.171875, "learning_rate": 5.102970216480926e-06, "loss": 0.1611, "step": 25602 }, { "epoch": 159.02492211838006, "grad_norm": 1.796875, "learning_rate": 5.101478621215442e-06, "loss": 0.147, "step": 25603 }, { "epoch": 159.0311526479751, "grad_norm": 2.90625, "learning_rate": 5.099987219208377e-06, "loss": 0.1775, "step": 25604 }, { "epoch": 159.0373831775701, "grad_norm": 2.703125, "learning_rate": 5.098496010474218e-06, "loss": 0.3311, "step": 25605 }, { "epoch": 159.04361370716512, "grad_norm": 2.296875, "learning_rate": 5.0970049950274355e-06, "loss": 0.2499, "step": 25606 }, { "epoch": 159.04984423676012, "grad_norm": 2.578125, "learning_rate": 5.095514172882521e-06, "loss": 0.1999, "step": 25607 }, { "epoch": 159.05607476635515, "grad_norm": 1.6953125, "learning_rate": 5.094023544053955e-06, "loss": 0.1334, "step": 25608 }, { "epoch": 159.06230529595015, "grad_norm": 1.71875, "learning_rate": 5.092533108556208e-06, "loss": 0.1426, "step": 25609 }, { "epoch": 159.06853582554518, "grad_norm": 1.8828125, "learning_rate": 5.091042866403764e-06, "loss": 0.1846, "step": 25610 }, { "epoch": 159.07476635514018, "grad_norm": 1.6640625, "learning_rate": 5.089552817611085e-06, "loss": 0.1298, "step": 25611 }, { "epoch": 159.0809968847352, "grad_norm": 1.078125, "learning_rate": 5.088062962192655e-06, "loss": 0.0939, "step": 25612 }, { "epoch": 159.08722741433021, "grad_norm": 2.328125, "learning_rate": 5.086573300162931e-06, "loss": 0.1562, "step": 25613 }, { "epoch": 159.09345794392524, "grad_norm": 1.671875, "learning_rate": 5.085083831536389e-06, "loss": 0.0938, "step": 25614 }, { "epoch": 159.09968847352025, "grad_norm": 2.328125, "learning_rate": 5.0835945563275e-06, "loss": 0.1683, "step": 25615 }, { "epoch": 159.10591900311528, "grad_norm": 1.4609375, "learning_rate": 5.0821054745507215e-06, "loss": 0.1296, "step": 25616 }, { "epoch": 159.11214953271028, "grad_norm": 1.3046875, "learning_rate": 5.0806165862205106e-06, "loss": 0.1301, "step": 25617 }, { "epoch": 159.1183800623053, "grad_norm": 1.796875, "learning_rate": 5.079127891351332e-06, "loss": 0.1668, "step": 25618 }, { "epoch": 159.1246105919003, "grad_norm": 1.0859375, "learning_rate": 5.077639389957653e-06, "loss": 0.1061, "step": 25619 }, { "epoch": 159.13084112149534, "grad_norm": 1.8984375, "learning_rate": 5.0761510820539145e-06, "loss": 0.1798, "step": 25620 }, { "epoch": 159.13707165109034, "grad_norm": 2.203125, "learning_rate": 5.074662967654581e-06, "loss": 0.1915, "step": 25621 }, { "epoch": 159.14330218068537, "grad_norm": 3.40625, "learning_rate": 5.073175046774104e-06, "loss": 0.2515, "step": 25622 }, { "epoch": 159.14953271028037, "grad_norm": 2.515625, "learning_rate": 5.071687319426946e-06, "loss": 0.1727, "step": 25623 }, { "epoch": 159.1557632398754, "grad_norm": 1.5234375, "learning_rate": 5.07019978562753e-06, "loss": 0.1415, "step": 25624 }, { "epoch": 159.1619937694704, "grad_norm": 2.84375, "learning_rate": 5.068712445390319e-06, "loss": 0.2417, "step": 25625 }, { "epoch": 159.16822429906543, "grad_norm": 3.296875, "learning_rate": 5.067225298729758e-06, "loss": 0.1544, "step": 25626 }, { "epoch": 159.17445482866043, "grad_norm": 1.9375, "learning_rate": 5.065738345660295e-06, "loss": 0.2846, "step": 25627 }, { "epoch": 159.18068535825546, "grad_norm": 2.390625, "learning_rate": 5.064251586196359e-06, "loss": 0.1615, "step": 25628 }, { "epoch": 159.18691588785046, "grad_norm": 2.265625, "learning_rate": 5.062765020352406e-06, "loss": 0.2414, "step": 25629 }, { "epoch": 159.1931464174455, "grad_norm": 2.203125, "learning_rate": 5.061278648142853e-06, "loss": 0.2809, "step": 25630 }, { "epoch": 159.1993769470405, "grad_norm": 1.328125, "learning_rate": 5.059792469582159e-06, "loss": 0.1149, "step": 25631 }, { "epoch": 159.20560747663552, "grad_norm": 1.3203125, "learning_rate": 5.058306484684735e-06, "loss": 0.125, "step": 25632 }, { "epoch": 159.21183800623052, "grad_norm": 1.8046875, "learning_rate": 5.056820693465028e-06, "loss": 0.1728, "step": 25633 }, { "epoch": 159.21806853582555, "grad_norm": 2.671875, "learning_rate": 5.055335095937472e-06, "loss": 0.2691, "step": 25634 }, { "epoch": 159.22429906542055, "grad_norm": 2.15625, "learning_rate": 5.05384969211648e-06, "loss": 0.1678, "step": 25635 }, { "epoch": 159.23052959501558, "grad_norm": 2.28125, "learning_rate": 5.052364482016494e-06, "loss": 0.1589, "step": 25636 }, { "epoch": 159.23676012461058, "grad_norm": 2.265625, "learning_rate": 5.050879465651925e-06, "loss": 0.1673, "step": 25637 }, { "epoch": 159.2429906542056, "grad_norm": 2.28125, "learning_rate": 5.0493946430372115e-06, "loss": 0.2251, "step": 25638 }, { "epoch": 159.2492211838006, "grad_norm": 0.921875, "learning_rate": 5.047910014186754e-06, "loss": 0.0935, "step": 25639 }, { "epoch": 159.25545171339564, "grad_norm": 1.609375, "learning_rate": 5.046425579114986e-06, "loss": 0.1529, "step": 25640 }, { "epoch": 159.26168224299064, "grad_norm": 1.0859375, "learning_rate": 5.044941337836328e-06, "loss": 0.0947, "step": 25641 }, { "epoch": 159.26791277258567, "grad_norm": 2.421875, "learning_rate": 5.043457290365186e-06, "loss": 0.1716, "step": 25642 }, { "epoch": 159.27414330218068, "grad_norm": 1.890625, "learning_rate": 5.0419734367159715e-06, "loss": 0.1516, "step": 25643 }, { "epoch": 159.2803738317757, "grad_norm": 2.34375, "learning_rate": 5.040489776903101e-06, "loss": 0.1742, "step": 25644 }, { "epoch": 159.2866043613707, "grad_norm": 1.765625, "learning_rate": 5.0390063109409915e-06, "loss": 0.1322, "step": 25645 }, { "epoch": 159.29283489096574, "grad_norm": 2.28125, "learning_rate": 5.037523038844033e-06, "loss": 0.2001, "step": 25646 }, { "epoch": 159.29906542056074, "grad_norm": 1.8203125, "learning_rate": 5.0360399606266424e-06, "loss": 0.171, "step": 25647 }, { "epoch": 159.30529595015577, "grad_norm": 1.125, "learning_rate": 5.034557076303228e-06, "loss": 0.0969, "step": 25648 }, { "epoch": 159.31152647975077, "grad_norm": 3.078125, "learning_rate": 5.033074385888189e-06, "loss": 0.2961, "step": 25649 }, { "epoch": 159.3177570093458, "grad_norm": 3.09375, "learning_rate": 5.031591889395915e-06, "loss": 0.2211, "step": 25650 }, { "epoch": 159.3239875389408, "grad_norm": 2.203125, "learning_rate": 5.030109586840812e-06, "loss": 0.1248, "step": 25651 }, { "epoch": 159.33021806853583, "grad_norm": 2.09375, "learning_rate": 5.028627478237283e-06, "loss": 0.1236, "step": 25652 }, { "epoch": 159.33644859813083, "grad_norm": 0.9609375, "learning_rate": 5.0271455635997086e-06, "loss": 0.0972, "step": 25653 }, { "epoch": 159.34267912772586, "grad_norm": 1.8125, "learning_rate": 5.0256638429424876e-06, "loss": 0.242, "step": 25654 }, { "epoch": 159.34890965732086, "grad_norm": 2.328125, "learning_rate": 5.024182316280024e-06, "loss": 0.1712, "step": 25655 }, { "epoch": 159.3551401869159, "grad_norm": 1.640625, "learning_rate": 5.022700983626691e-06, "loss": 0.1403, "step": 25656 }, { "epoch": 159.3613707165109, "grad_norm": 2.21875, "learning_rate": 5.021219844996875e-06, "loss": 0.1327, "step": 25657 }, { "epoch": 159.36760124610592, "grad_norm": 3.296875, "learning_rate": 5.0197389004049635e-06, "loss": 0.455, "step": 25658 }, { "epoch": 159.37383177570092, "grad_norm": 1.265625, "learning_rate": 5.01825814986534e-06, "loss": 0.1319, "step": 25659 }, { "epoch": 159.38006230529595, "grad_norm": 2.359375, "learning_rate": 5.016777593392399e-06, "loss": 0.2482, "step": 25660 }, { "epoch": 159.38629283489095, "grad_norm": 1.34375, "learning_rate": 5.015297231000504e-06, "loss": 0.1034, "step": 25661 }, { "epoch": 159.39252336448598, "grad_norm": 1.375, "learning_rate": 5.013817062704035e-06, "loss": 0.1059, "step": 25662 }, { "epoch": 159.39875389408098, "grad_norm": 2.8125, "learning_rate": 5.012337088517366e-06, "loss": 0.2234, "step": 25663 }, { "epoch": 159.404984423676, "grad_norm": 2.4375, "learning_rate": 5.010857308454883e-06, "loss": 0.3516, "step": 25664 }, { "epoch": 159.41121495327104, "grad_norm": 2.15625, "learning_rate": 5.009377722530945e-06, "loss": 0.1528, "step": 25665 }, { "epoch": 159.41744548286604, "grad_norm": 2.625, "learning_rate": 5.0078983307599245e-06, "loss": 0.1573, "step": 25666 }, { "epoch": 159.42367601246107, "grad_norm": 0.953125, "learning_rate": 5.006419133156198e-06, "loss": 0.113, "step": 25667 }, { "epoch": 159.42990654205607, "grad_norm": 1.703125, "learning_rate": 5.0049401297341255e-06, "loss": 0.1041, "step": 25668 }, { "epoch": 159.4361370716511, "grad_norm": 2.390625, "learning_rate": 5.0034613205080655e-06, "loss": 0.1954, "step": 25669 }, { "epoch": 159.4423676012461, "grad_norm": 1.25, "learning_rate": 5.001982705492386e-06, "loss": 0.1028, "step": 25670 }, { "epoch": 159.44859813084113, "grad_norm": 1.7734375, "learning_rate": 5.000504284701455e-06, "loss": 0.1421, "step": 25671 }, { "epoch": 159.45482866043614, "grad_norm": 2.203125, "learning_rate": 4.999026058149617e-06, "loss": 0.1307, "step": 25672 }, { "epoch": 159.46105919003116, "grad_norm": 2.640625, "learning_rate": 4.997548025851237e-06, "loss": 0.1871, "step": 25673 }, { "epoch": 159.46728971962617, "grad_norm": 1.890625, "learning_rate": 4.996070187820673e-06, "loss": 0.208, "step": 25674 }, { "epoch": 159.4735202492212, "grad_norm": 1.84375, "learning_rate": 4.994592544072277e-06, "loss": 0.1331, "step": 25675 }, { "epoch": 159.4797507788162, "grad_norm": 1.7734375, "learning_rate": 4.99311509462039e-06, "loss": 0.1111, "step": 25676 }, { "epoch": 159.48598130841123, "grad_norm": 2.984375, "learning_rate": 4.9916378394793685e-06, "loss": 0.2625, "step": 25677 }, { "epoch": 159.49221183800623, "grad_norm": 0.984375, "learning_rate": 4.990160778663564e-06, "loss": 0.0949, "step": 25678 }, { "epoch": 159.49844236760126, "grad_norm": 2.5625, "learning_rate": 4.988683912187314e-06, "loss": 0.262, "step": 25679 }, { "epoch": 159.50467289719626, "grad_norm": 2.5625, "learning_rate": 4.987207240064973e-06, "loss": 0.2667, "step": 25680 }, { "epoch": 159.5109034267913, "grad_norm": 2.53125, "learning_rate": 4.985730762310867e-06, "loss": 0.2944, "step": 25681 }, { "epoch": 159.5171339563863, "grad_norm": 2.453125, "learning_rate": 4.984254478939351e-06, "loss": 0.1903, "step": 25682 }, { "epoch": 159.52336448598132, "grad_norm": 3.4375, "learning_rate": 4.982778389964754e-06, "loss": 0.3447, "step": 25683 }, { "epoch": 159.52959501557632, "grad_norm": 1.7265625, "learning_rate": 4.98130249540141e-06, "loss": 0.1399, "step": 25684 }, { "epoch": 159.53582554517135, "grad_norm": 1.2265625, "learning_rate": 4.979826795263662e-06, "loss": 0.1241, "step": 25685 }, { "epoch": 159.54205607476635, "grad_norm": 2.8125, "learning_rate": 4.978351289565847e-06, "loss": 0.2148, "step": 25686 }, { "epoch": 159.54828660436138, "grad_norm": 1.328125, "learning_rate": 4.976875978322276e-06, "loss": 0.134, "step": 25687 }, { "epoch": 159.55451713395638, "grad_norm": 2.703125, "learning_rate": 4.975400861547289e-06, "loss": 0.1628, "step": 25688 }, { "epoch": 159.5607476635514, "grad_norm": 2.15625, "learning_rate": 4.973925939255209e-06, "loss": 0.1229, "step": 25689 }, { "epoch": 159.5669781931464, "grad_norm": 2.328125, "learning_rate": 4.972451211460371e-06, "loss": 0.1419, "step": 25690 }, { "epoch": 159.57320872274144, "grad_norm": 2.90625, "learning_rate": 4.970976678177083e-06, "loss": 0.2713, "step": 25691 }, { "epoch": 159.57943925233644, "grad_norm": 2.375, "learning_rate": 4.969502339419674e-06, "loss": 0.1955, "step": 25692 }, { "epoch": 159.58566978193147, "grad_norm": 2.546875, "learning_rate": 4.96802819520247e-06, "loss": 0.243, "step": 25693 }, { "epoch": 159.59190031152647, "grad_norm": 1.5546875, "learning_rate": 4.9665542455397785e-06, "loss": 0.1495, "step": 25694 }, { "epoch": 159.5981308411215, "grad_norm": 0.98046875, "learning_rate": 4.9650804904459115e-06, "loss": 0.0877, "step": 25695 }, { "epoch": 159.6043613707165, "grad_norm": 1.8984375, "learning_rate": 4.963606929935188e-06, "loss": 0.1434, "step": 25696 }, { "epoch": 159.61059190031153, "grad_norm": 2.484375, "learning_rate": 4.962133564021923e-06, "loss": 0.1828, "step": 25697 }, { "epoch": 159.61682242990653, "grad_norm": 1.4140625, "learning_rate": 4.960660392720418e-06, "loss": 0.1287, "step": 25698 }, { "epoch": 159.62305295950156, "grad_norm": 3.21875, "learning_rate": 4.959187416044991e-06, "loss": 0.2631, "step": 25699 }, { "epoch": 159.62928348909657, "grad_norm": 2.046875, "learning_rate": 4.957714634009936e-06, "loss": 0.2559, "step": 25700 }, { "epoch": 159.6355140186916, "grad_norm": 1.1796875, "learning_rate": 4.9562420466295694e-06, "loss": 0.1214, "step": 25701 }, { "epoch": 159.6417445482866, "grad_norm": 1.84375, "learning_rate": 4.9547696539181815e-06, "loss": 0.1569, "step": 25702 }, { "epoch": 159.64797507788163, "grad_norm": 3.5625, "learning_rate": 4.953297455890074e-06, "loss": 0.2608, "step": 25703 }, { "epoch": 159.65420560747663, "grad_norm": 2.15625, "learning_rate": 4.951825452559558e-06, "loss": 0.1153, "step": 25704 }, { "epoch": 159.66043613707166, "grad_norm": 2.265625, "learning_rate": 4.9503536439409136e-06, "loss": 0.186, "step": 25705 }, { "epoch": 159.66666666666666, "grad_norm": 1.6171875, "learning_rate": 4.948882030048449e-06, "loss": 0.1202, "step": 25706 }, { "epoch": 159.6728971962617, "grad_norm": 1.421875, "learning_rate": 4.947410610896444e-06, "loss": 0.1366, "step": 25707 }, { "epoch": 159.6791277258567, "grad_norm": 1.8828125, "learning_rate": 4.945939386499201e-06, "loss": 0.1112, "step": 25708 }, { "epoch": 159.68535825545172, "grad_norm": 2.453125, "learning_rate": 4.944468356870999e-06, "loss": 0.2011, "step": 25709 }, { "epoch": 159.69158878504672, "grad_norm": 1.6796875, "learning_rate": 4.942997522026127e-06, "loss": 0.1518, "step": 25710 }, { "epoch": 159.69781931464175, "grad_norm": 1.703125, "learning_rate": 4.941526881978881e-06, "loss": 0.1298, "step": 25711 }, { "epoch": 159.70404984423675, "grad_norm": 2.234375, "learning_rate": 4.940056436743534e-06, "loss": 0.2594, "step": 25712 }, { "epoch": 159.71028037383178, "grad_norm": 3.640625, "learning_rate": 4.938586186334362e-06, "loss": 0.286, "step": 25713 }, { "epoch": 159.71651090342678, "grad_norm": 2.046875, "learning_rate": 4.937116130765654e-06, "loss": 0.1254, "step": 25714 }, { "epoch": 159.7227414330218, "grad_norm": 1.6796875, "learning_rate": 4.935646270051689e-06, "loss": 0.145, "step": 25715 }, { "epoch": 159.7289719626168, "grad_norm": 1.5859375, "learning_rate": 4.934176604206731e-06, "loss": 0.1527, "step": 25716 }, { "epoch": 159.73520249221184, "grad_norm": 1.6484375, "learning_rate": 4.9327071332450625e-06, "loss": 0.1615, "step": 25717 }, { "epoch": 159.74143302180684, "grad_norm": 2.140625, "learning_rate": 4.931237857180959e-06, "loss": 0.1619, "step": 25718 }, { "epoch": 159.74766355140187, "grad_norm": 2.28125, "learning_rate": 4.929768776028687e-06, "loss": 0.1424, "step": 25719 }, { "epoch": 159.75389408099687, "grad_norm": 3.546875, "learning_rate": 4.928299889802504e-06, "loss": 0.2712, "step": 25720 }, { "epoch": 159.7601246105919, "grad_norm": 2.328125, "learning_rate": 4.926831198516685e-06, "loss": 0.1027, "step": 25721 }, { "epoch": 159.7663551401869, "grad_norm": 1.125, "learning_rate": 4.925362702185493e-06, "loss": 0.1112, "step": 25722 }, { "epoch": 159.77258566978193, "grad_norm": 1.5546875, "learning_rate": 4.9238944008232e-06, "loss": 0.1211, "step": 25723 }, { "epoch": 159.77881619937693, "grad_norm": 1.859375, "learning_rate": 4.922426294444049e-06, "loss": 0.1712, "step": 25724 }, { "epoch": 159.78504672897196, "grad_norm": 1.5703125, "learning_rate": 4.920958383062316e-06, "loss": 0.1393, "step": 25725 }, { "epoch": 159.79127725856696, "grad_norm": 3.078125, "learning_rate": 4.919490666692242e-06, "loss": 0.2413, "step": 25726 }, { "epoch": 159.797507788162, "grad_norm": 1.4453125, "learning_rate": 4.918023145348094e-06, "loss": 0.122, "step": 25727 }, { "epoch": 159.80373831775702, "grad_norm": 1.8203125, "learning_rate": 4.916555819044114e-06, "loss": 0.1207, "step": 25728 }, { "epoch": 159.80996884735202, "grad_norm": 2.734375, "learning_rate": 4.915088687794558e-06, "loss": 0.2849, "step": 25729 }, { "epoch": 159.81619937694705, "grad_norm": 2.171875, "learning_rate": 4.913621751613681e-06, "loss": 0.2038, "step": 25730 }, { "epoch": 159.82242990654206, "grad_norm": 2.5, "learning_rate": 4.912155010515726e-06, "loss": 0.1828, "step": 25731 }, { "epoch": 159.82866043613708, "grad_norm": 1.7421875, "learning_rate": 4.910688464514932e-06, "loss": 0.1326, "step": 25732 }, { "epoch": 159.8348909657321, "grad_norm": 2.390625, "learning_rate": 4.9092221136255444e-06, "loss": 0.142, "step": 25733 }, { "epoch": 159.84112149532712, "grad_norm": 1.0625, "learning_rate": 4.907755957861818e-06, "loss": 0.0967, "step": 25734 }, { "epoch": 159.84735202492212, "grad_norm": 3.25, "learning_rate": 4.906289997237973e-06, "loss": 0.2275, "step": 25735 }, { "epoch": 159.85358255451715, "grad_norm": 1.3671875, "learning_rate": 4.904824231768254e-06, "loss": 0.1117, "step": 25736 }, { "epoch": 159.85981308411215, "grad_norm": 2.390625, "learning_rate": 4.9033586614669085e-06, "loss": 0.2734, "step": 25737 }, { "epoch": 159.86604361370718, "grad_norm": 3.25, "learning_rate": 4.901893286348161e-06, "loss": 0.2971, "step": 25738 }, { "epoch": 159.87227414330218, "grad_norm": 2.40625, "learning_rate": 4.900428106426236e-06, "loss": 0.2127, "step": 25739 }, { "epoch": 159.8785046728972, "grad_norm": 1.3828125, "learning_rate": 4.898963121715372e-06, "loss": 0.1344, "step": 25740 }, { "epoch": 159.8847352024922, "grad_norm": 2.015625, "learning_rate": 4.897498332229802e-06, "loss": 0.1569, "step": 25741 }, { "epoch": 159.89096573208724, "grad_norm": 2.21875, "learning_rate": 4.896033737983741e-06, "loss": 0.1591, "step": 25742 }, { "epoch": 159.89719626168224, "grad_norm": 3.421875, "learning_rate": 4.894569338991417e-06, "loss": 0.2936, "step": 25743 }, { "epoch": 159.90342679127727, "grad_norm": 2.78125, "learning_rate": 4.893105135267062e-06, "loss": 0.1542, "step": 25744 }, { "epoch": 159.90965732087227, "grad_norm": 0.93359375, "learning_rate": 4.891641126824889e-06, "loss": 0.0958, "step": 25745 }, { "epoch": 159.9158878504673, "grad_norm": 2.921875, "learning_rate": 4.890177313679112e-06, "loss": 0.3258, "step": 25746 }, { "epoch": 159.9221183800623, "grad_norm": 2.046875, "learning_rate": 4.888713695843952e-06, "loss": 0.1721, "step": 25747 }, { "epoch": 159.92834890965733, "grad_norm": 1.0546875, "learning_rate": 4.8872502733336295e-06, "loss": 0.0917, "step": 25748 }, { "epoch": 159.93457943925233, "grad_norm": 2.625, "learning_rate": 4.885787046162349e-06, "loss": 0.1786, "step": 25749 }, { "epoch": 159.94080996884736, "grad_norm": 2.0, "learning_rate": 4.88432401434433e-06, "loss": 0.2188, "step": 25750 }, { "epoch": 159.94704049844236, "grad_norm": 1.8203125, "learning_rate": 4.882861177893772e-06, "loss": 0.2438, "step": 25751 }, { "epoch": 159.9532710280374, "grad_norm": 2.671875, "learning_rate": 4.881398536824893e-06, "loss": 0.1967, "step": 25752 }, { "epoch": 159.9595015576324, "grad_norm": 1.5859375, "learning_rate": 4.879936091151888e-06, "loss": 0.1375, "step": 25753 }, { "epoch": 159.96573208722742, "grad_norm": 1.8359375, "learning_rate": 4.878473840888964e-06, "loss": 0.1401, "step": 25754 }, { "epoch": 159.97196261682242, "grad_norm": 1.5546875, "learning_rate": 4.877011786050326e-06, "loss": 0.1481, "step": 25755 }, { "epoch": 159.97819314641745, "grad_norm": 1.1953125, "learning_rate": 4.875549926650178e-06, "loss": 0.1238, "step": 25756 }, { "epoch": 159.98442367601245, "grad_norm": 2.25, "learning_rate": 4.874088262702711e-06, "loss": 0.1224, "step": 25757 }, { "epoch": 159.99065420560748, "grad_norm": 2.34375, "learning_rate": 4.872626794222115e-06, "loss": 0.2266, "step": 25758 }, { "epoch": 159.99688473520249, "grad_norm": 2.640625, "learning_rate": 4.8711655212225924e-06, "loss": 0.1969, "step": 25759 }, { "epoch": 160.0, "grad_norm": 1.03125, "learning_rate": 4.86970444371834e-06, "loss": 0.0527, "step": 25760 }, { "epoch": 160.00623052959503, "grad_norm": 1.984375, "learning_rate": 4.868243561723535e-06, "loss": 0.1876, "step": 25761 }, { "epoch": 160.01246105919003, "grad_norm": 0.90234375, "learning_rate": 4.866782875252374e-06, "loss": 0.1036, "step": 25762 }, { "epoch": 160.01869158878506, "grad_norm": 1.6015625, "learning_rate": 4.865322384319046e-06, "loss": 0.1268, "step": 25763 }, { "epoch": 160.02492211838006, "grad_norm": 2.90625, "learning_rate": 4.863862088937735e-06, "loss": 0.3002, "step": 25764 }, { "epoch": 160.0311526479751, "grad_norm": 1.3984375, "learning_rate": 4.862401989122611e-06, "loss": 0.0981, "step": 25765 }, { "epoch": 160.0373831775701, "grad_norm": 1.15625, "learning_rate": 4.860942084887868e-06, "loss": 0.1038, "step": 25766 }, { "epoch": 160.04361370716512, "grad_norm": 1.0546875, "learning_rate": 4.8594823762476845e-06, "loss": 0.0954, "step": 25767 }, { "epoch": 160.04984423676012, "grad_norm": 2.203125, "learning_rate": 4.85802286321623e-06, "loss": 0.151, "step": 25768 }, { "epoch": 160.05607476635515, "grad_norm": 2.078125, "learning_rate": 4.856563545807688e-06, "loss": 0.1034, "step": 25769 }, { "epoch": 160.06230529595015, "grad_norm": 1.4921875, "learning_rate": 4.855104424036222e-06, "loss": 0.1075, "step": 25770 }, { "epoch": 160.06853582554518, "grad_norm": 3.078125, "learning_rate": 4.853645497916018e-06, "loss": 0.2323, "step": 25771 }, { "epoch": 160.07476635514018, "grad_norm": 1.7578125, "learning_rate": 4.852186767461225e-06, "loss": 0.1463, "step": 25772 }, { "epoch": 160.0809968847352, "grad_norm": 2.015625, "learning_rate": 4.850728232686025e-06, "loss": 0.1614, "step": 25773 }, { "epoch": 160.08722741433021, "grad_norm": 1.5546875, "learning_rate": 4.849269893604585e-06, "loss": 0.1338, "step": 25774 }, { "epoch": 160.09345794392524, "grad_norm": 2.375, "learning_rate": 4.847811750231057e-06, "loss": 0.1661, "step": 25775 }, { "epoch": 160.09968847352025, "grad_norm": 2.015625, "learning_rate": 4.846353802579618e-06, "loss": 0.1383, "step": 25776 }, { "epoch": 160.10591900311528, "grad_norm": 1.2265625, "learning_rate": 4.84489605066441e-06, "loss": 0.1118, "step": 25777 }, { "epoch": 160.11214953271028, "grad_norm": 2.390625, "learning_rate": 4.843438494499608e-06, "loss": 0.2491, "step": 25778 }, { "epoch": 160.1183800623053, "grad_norm": 1.265625, "learning_rate": 4.841981134099355e-06, "loss": 0.1042, "step": 25779 }, { "epoch": 160.1246105919003, "grad_norm": 2.6875, "learning_rate": 4.840523969477812e-06, "loss": 0.1778, "step": 25780 }, { "epoch": 160.13084112149534, "grad_norm": 1.453125, "learning_rate": 4.839067000649128e-06, "loss": 0.1087, "step": 25781 }, { "epoch": 160.13707165109034, "grad_norm": 1.4453125, "learning_rate": 4.837610227627468e-06, "loss": 0.1439, "step": 25782 }, { "epoch": 160.14330218068537, "grad_norm": 1.453125, "learning_rate": 4.836153650426955e-06, "loss": 0.1623, "step": 25783 }, { "epoch": 160.14953271028037, "grad_norm": 1.96875, "learning_rate": 4.83469726906175e-06, "loss": 0.1295, "step": 25784 }, { "epoch": 160.1557632398754, "grad_norm": 1.203125, "learning_rate": 4.833241083545995e-06, "loss": 0.1075, "step": 25785 }, { "epoch": 160.1619937694704, "grad_norm": 2.140625, "learning_rate": 4.83178509389384e-06, "loss": 0.1382, "step": 25786 }, { "epoch": 160.16822429906543, "grad_norm": 2.375, "learning_rate": 4.830329300119413e-06, "loss": 0.1603, "step": 25787 }, { "epoch": 160.17445482866043, "grad_norm": 2.265625, "learning_rate": 4.828873702236869e-06, "loss": 0.1718, "step": 25788 }, { "epoch": 160.18068535825546, "grad_norm": 3.3125, "learning_rate": 4.827418300260325e-06, "loss": 0.138, "step": 25789 }, { "epoch": 160.18691588785046, "grad_norm": 2.46875, "learning_rate": 4.825963094203936e-06, "loss": 0.2419, "step": 25790 }, { "epoch": 160.1931464174455, "grad_norm": 2.484375, "learning_rate": 4.824508084081819e-06, "loss": 0.2838, "step": 25791 }, { "epoch": 160.1993769470405, "grad_norm": 2.21875, "learning_rate": 4.823053269908112e-06, "loss": 0.1489, "step": 25792 }, { "epoch": 160.20560747663552, "grad_norm": 1.9609375, "learning_rate": 4.821598651696954e-06, "loss": 0.1552, "step": 25793 }, { "epoch": 160.21183800623052, "grad_norm": 1.5625, "learning_rate": 4.820144229462453e-06, "loss": 0.1357, "step": 25794 }, { "epoch": 160.21806853582555, "grad_norm": 1.2578125, "learning_rate": 4.8186900032187534e-06, "loss": 0.1297, "step": 25795 }, { "epoch": 160.22429906542055, "grad_norm": 2.703125, "learning_rate": 4.817235972979966e-06, "loss": 0.204, "step": 25796 }, { "epoch": 160.23052959501558, "grad_norm": 2.046875, "learning_rate": 4.815782138760222e-06, "loss": 0.2723, "step": 25797 }, { "epoch": 160.23676012461058, "grad_norm": 2.46875, "learning_rate": 4.814328500573631e-06, "loss": 0.1021, "step": 25798 }, { "epoch": 160.2429906542056, "grad_norm": 2.1875, "learning_rate": 4.812875058434315e-06, "loss": 0.1615, "step": 25799 }, { "epoch": 160.2492211838006, "grad_norm": 1.9921875, "learning_rate": 4.811421812356401e-06, "loss": 0.1704, "step": 25800 }, { "epoch": 160.25545171339564, "grad_norm": 1.7734375, "learning_rate": 4.809968762353992e-06, "loss": 0.1194, "step": 25801 }, { "epoch": 160.26168224299064, "grad_norm": 0.890625, "learning_rate": 4.808515908441194e-06, "loss": 0.0968, "step": 25802 }, { "epoch": 160.26791277258567, "grad_norm": 1.6875, "learning_rate": 4.807063250632129e-06, "loss": 0.1356, "step": 25803 }, { "epoch": 160.27414330218068, "grad_norm": 1.4609375, "learning_rate": 4.805610788940906e-06, "loss": 0.1503, "step": 25804 }, { "epoch": 160.2803738317757, "grad_norm": 1.734375, "learning_rate": 4.804158523381624e-06, "loss": 0.1628, "step": 25805 }, { "epoch": 160.2866043613707, "grad_norm": 2.140625, "learning_rate": 4.802706453968389e-06, "loss": 0.1773, "step": 25806 }, { "epoch": 160.29283489096574, "grad_norm": 2.34375, "learning_rate": 4.801254580715314e-06, "loss": 0.1734, "step": 25807 }, { "epoch": 160.29906542056074, "grad_norm": 2.1875, "learning_rate": 4.799802903636494e-06, "loss": 0.1799, "step": 25808 }, { "epoch": 160.30529595015577, "grad_norm": 2.078125, "learning_rate": 4.7983514227460165e-06, "loss": 0.2317, "step": 25809 }, { "epoch": 160.31152647975077, "grad_norm": 1.6875, "learning_rate": 4.796900138057989e-06, "loss": 0.1407, "step": 25810 }, { "epoch": 160.3177570093458, "grad_norm": 1.7421875, "learning_rate": 4.795449049586515e-06, "loss": 0.1788, "step": 25811 }, { "epoch": 160.3239875389408, "grad_norm": 3.671875, "learning_rate": 4.793998157345669e-06, "loss": 0.3203, "step": 25812 }, { "epoch": 160.33021806853583, "grad_norm": 2.125, "learning_rate": 4.792547461349553e-06, "loss": 0.1571, "step": 25813 }, { "epoch": 160.33644859813083, "grad_norm": 1.8671875, "learning_rate": 4.79109696161226e-06, "loss": 0.1522, "step": 25814 }, { "epoch": 160.34267912772586, "grad_norm": 2.5, "learning_rate": 4.789646658147875e-06, "loss": 0.1916, "step": 25815 }, { "epoch": 160.34890965732086, "grad_norm": 2.828125, "learning_rate": 4.788196550970475e-06, "loss": 0.2417, "step": 25816 }, { "epoch": 160.3551401869159, "grad_norm": 2.703125, "learning_rate": 4.78674664009415e-06, "loss": 0.1357, "step": 25817 }, { "epoch": 160.3613707165109, "grad_norm": 2.171875, "learning_rate": 4.785296925532984e-06, "loss": 0.1716, "step": 25818 }, { "epoch": 160.36760124610592, "grad_norm": 1.71875, "learning_rate": 4.783847407301059e-06, "loss": 0.1038, "step": 25819 }, { "epoch": 160.37383177570092, "grad_norm": 2.609375, "learning_rate": 4.7823980854124516e-06, "loss": 0.2815, "step": 25820 }, { "epoch": 160.38006230529595, "grad_norm": 2.671875, "learning_rate": 4.7809489598812265e-06, "loss": 0.1427, "step": 25821 }, { "epoch": 160.38629283489095, "grad_norm": 1.6328125, "learning_rate": 4.779500030721468e-06, "loss": 0.1297, "step": 25822 }, { "epoch": 160.39252336448598, "grad_norm": 1.9375, "learning_rate": 4.778051297947253e-06, "loss": 0.1379, "step": 25823 }, { "epoch": 160.39875389408098, "grad_norm": 2.359375, "learning_rate": 4.77660276157264e-06, "loss": 0.1785, "step": 25824 }, { "epoch": 160.404984423676, "grad_norm": 1.609375, "learning_rate": 4.775154421611705e-06, "loss": 0.1625, "step": 25825 }, { "epoch": 160.41121495327104, "grad_norm": 3.078125, "learning_rate": 4.773706278078519e-06, "loss": 0.3438, "step": 25826 }, { "epoch": 160.41744548286604, "grad_norm": 2.34375, "learning_rate": 4.772258330987139e-06, "loss": 0.1173, "step": 25827 }, { "epoch": 160.42367601246107, "grad_norm": 2.125, "learning_rate": 4.770810580351625e-06, "loss": 0.149, "step": 25828 }, { "epoch": 160.42990654205607, "grad_norm": 1.765625, "learning_rate": 4.769363026186041e-06, "loss": 0.1421, "step": 25829 }, { "epoch": 160.4361370716511, "grad_norm": 2.140625, "learning_rate": 4.767915668504455e-06, "loss": 0.1542, "step": 25830 }, { "epoch": 160.4423676012461, "grad_norm": 1.609375, "learning_rate": 4.766468507320909e-06, "loss": 0.129, "step": 25831 }, { "epoch": 160.44859813084113, "grad_norm": 1.6953125, "learning_rate": 4.765021542649464e-06, "loss": 0.0818, "step": 25832 }, { "epoch": 160.45482866043614, "grad_norm": 2.4375, "learning_rate": 4.763574774504184e-06, "loss": 0.2177, "step": 25833 }, { "epoch": 160.46105919003116, "grad_norm": 2.71875, "learning_rate": 4.76212820289911e-06, "loss": 0.2172, "step": 25834 }, { "epoch": 160.46728971962617, "grad_norm": 2.0, "learning_rate": 4.760681827848284e-06, "loss": 0.1722, "step": 25835 }, { "epoch": 160.4735202492212, "grad_norm": 1.6796875, "learning_rate": 4.759235649365765e-06, "loss": 0.1293, "step": 25836 }, { "epoch": 160.4797507788162, "grad_norm": 1.3515625, "learning_rate": 4.757789667465601e-06, "loss": 0.1314, "step": 25837 }, { "epoch": 160.48598130841123, "grad_norm": 2.703125, "learning_rate": 4.756343882161823e-06, "loss": 0.1606, "step": 25838 }, { "epoch": 160.49221183800623, "grad_norm": 1.6328125, "learning_rate": 4.754898293468485e-06, "loss": 0.1983, "step": 25839 }, { "epoch": 160.49844236760126, "grad_norm": 2.734375, "learning_rate": 4.753452901399616e-06, "loss": 0.1677, "step": 25840 }, { "epoch": 160.50467289719626, "grad_norm": 1.875, "learning_rate": 4.752007705969269e-06, "loss": 0.1744, "step": 25841 }, { "epoch": 160.5109034267913, "grad_norm": 2.578125, "learning_rate": 4.750562707191461e-06, "loss": 0.2407, "step": 25842 }, { "epoch": 160.5171339563863, "grad_norm": 2.09375, "learning_rate": 4.749117905080236e-06, "loss": 0.1741, "step": 25843 }, { "epoch": 160.52336448598132, "grad_norm": 1.8671875, "learning_rate": 4.747673299649633e-06, "loss": 0.1667, "step": 25844 }, { "epoch": 160.52959501557632, "grad_norm": 1.71875, "learning_rate": 4.746228890913671e-06, "loss": 0.18, "step": 25845 }, { "epoch": 160.53582554517135, "grad_norm": 1.421875, "learning_rate": 4.744784678886386e-06, "loss": 0.1129, "step": 25846 }, { "epoch": 160.54205607476635, "grad_norm": 1.53125, "learning_rate": 4.743340663581794e-06, "loss": 0.1168, "step": 25847 }, { "epoch": 160.54828660436138, "grad_norm": 1.1796875, "learning_rate": 4.741896845013935e-06, "loss": 0.1238, "step": 25848 }, { "epoch": 160.55451713395638, "grad_norm": 2.109375, "learning_rate": 4.740453223196817e-06, "loss": 0.1263, "step": 25849 }, { "epoch": 160.5607476635514, "grad_norm": 1.3046875, "learning_rate": 4.739009798144467e-06, "loss": 0.1325, "step": 25850 }, { "epoch": 160.5669781931464, "grad_norm": 2.109375, "learning_rate": 4.737566569870905e-06, "loss": 0.2253, "step": 25851 }, { "epoch": 160.57320872274144, "grad_norm": 3.5625, "learning_rate": 4.7361235383901585e-06, "loss": 0.2637, "step": 25852 }, { "epoch": 160.57943925233644, "grad_norm": 3.03125, "learning_rate": 4.734680703716215e-06, "loss": 0.1636, "step": 25853 }, { "epoch": 160.58566978193147, "grad_norm": 2.34375, "learning_rate": 4.733238065863107e-06, "loss": 0.2039, "step": 25854 }, { "epoch": 160.59190031152647, "grad_norm": 2.53125, "learning_rate": 4.731795624844843e-06, "loss": 0.2305, "step": 25855 }, { "epoch": 160.5981308411215, "grad_norm": 1.796875, "learning_rate": 4.730353380675437e-06, "loss": 0.1346, "step": 25856 }, { "epoch": 160.6043613707165, "grad_norm": 1.4140625, "learning_rate": 4.728911333368882e-06, "loss": 0.1234, "step": 25857 }, { "epoch": 160.61059190031153, "grad_norm": 2.1875, "learning_rate": 4.7274694829392e-06, "loss": 0.195, "step": 25858 }, { "epoch": 160.61682242990653, "grad_norm": 2.203125, "learning_rate": 4.726027829400382e-06, "loss": 0.1551, "step": 25859 }, { "epoch": 160.62305295950156, "grad_norm": 0.8125, "learning_rate": 4.72458637276644e-06, "loss": 0.0894, "step": 25860 }, { "epoch": 160.62928348909657, "grad_norm": 2.484375, "learning_rate": 4.723145113051361e-06, "loss": 0.2107, "step": 25861 }, { "epoch": 160.6355140186916, "grad_norm": 1.421875, "learning_rate": 4.721704050269149e-06, "loss": 0.1202, "step": 25862 }, { "epoch": 160.6417445482866, "grad_norm": 2.40625, "learning_rate": 4.720263184433809e-06, "loss": 0.2343, "step": 25863 }, { "epoch": 160.64797507788163, "grad_norm": 2.6875, "learning_rate": 4.7188225155593205e-06, "loss": 0.3115, "step": 25864 }, { "epoch": 160.65420560747663, "grad_norm": 2.15625, "learning_rate": 4.717382043659688e-06, "loss": 0.1765, "step": 25865 }, { "epoch": 160.66043613707166, "grad_norm": 1.9921875, "learning_rate": 4.715941768748888e-06, "loss": 0.186, "step": 25866 }, { "epoch": 160.66666666666666, "grad_norm": 1.984375, "learning_rate": 4.714501690840922e-06, "loss": 0.1626, "step": 25867 }, { "epoch": 160.6728971962617, "grad_norm": 2.390625, "learning_rate": 4.7130618099497675e-06, "loss": 0.1438, "step": 25868 }, { "epoch": 160.6791277258567, "grad_norm": 1.59375, "learning_rate": 4.71162212608941e-06, "loss": 0.119, "step": 25869 }, { "epoch": 160.68535825545172, "grad_norm": 1.0546875, "learning_rate": 4.710182639273841e-06, "loss": 0.1218, "step": 25870 }, { "epoch": 160.69158878504672, "grad_norm": 2.1875, "learning_rate": 4.708743349517034e-06, "loss": 0.1528, "step": 25871 }, { "epoch": 160.69781931464175, "grad_norm": 1.4453125, "learning_rate": 4.707304256832959e-06, "loss": 0.1375, "step": 25872 }, { "epoch": 160.70404984423675, "grad_norm": 1.6640625, "learning_rate": 4.705865361235603e-06, "loss": 0.1321, "step": 25873 }, { "epoch": 160.71028037383178, "grad_norm": 1.7265625, "learning_rate": 4.704426662738948e-06, "loss": 0.0888, "step": 25874 }, { "epoch": 160.71651090342678, "grad_norm": 1.21875, "learning_rate": 4.702988161356947e-06, "loss": 0.1164, "step": 25875 }, { "epoch": 160.7227414330218, "grad_norm": 2.34375, "learning_rate": 4.701549857103588e-06, "loss": 0.2031, "step": 25876 }, { "epoch": 160.7289719626168, "grad_norm": 1.5546875, "learning_rate": 4.700111749992836e-06, "loss": 0.1152, "step": 25877 }, { "epoch": 160.73520249221184, "grad_norm": 1.9453125, "learning_rate": 4.698673840038659e-06, "loss": 0.1458, "step": 25878 }, { "epoch": 160.74143302180684, "grad_norm": 1.8828125, "learning_rate": 4.697236127255011e-06, "loss": 0.1337, "step": 25879 }, { "epoch": 160.74766355140187, "grad_norm": 1.515625, "learning_rate": 4.695798611655869e-06, "loss": 0.0985, "step": 25880 }, { "epoch": 160.75389408099687, "grad_norm": 0.98828125, "learning_rate": 4.694361293255184e-06, "loss": 0.1225, "step": 25881 }, { "epoch": 160.7601246105919, "grad_norm": 2.28125, "learning_rate": 4.692924172066931e-06, "loss": 0.2155, "step": 25882 }, { "epoch": 160.7663551401869, "grad_norm": 1.1015625, "learning_rate": 4.69148724810505e-06, "loss": 0.1278, "step": 25883 }, { "epoch": 160.77258566978193, "grad_norm": 2.828125, "learning_rate": 4.6900505213835126e-06, "loss": 0.1812, "step": 25884 }, { "epoch": 160.77881619937693, "grad_norm": 2.1875, "learning_rate": 4.6886139919162575e-06, "loss": 0.1295, "step": 25885 }, { "epoch": 160.78504672897196, "grad_norm": 1.5, "learning_rate": 4.687177659717248e-06, "loss": 0.1274, "step": 25886 }, { "epoch": 160.79127725856696, "grad_norm": 2.515625, "learning_rate": 4.685741524800424e-06, "loss": 0.1851, "step": 25887 }, { "epoch": 160.797507788162, "grad_norm": 1.890625, "learning_rate": 4.684305587179741e-06, "loss": 0.1548, "step": 25888 }, { "epoch": 160.80373831775702, "grad_norm": 2.390625, "learning_rate": 4.682869846869148e-06, "loss": 0.1997, "step": 25889 }, { "epoch": 160.80996884735202, "grad_norm": 2.34375, "learning_rate": 4.681434303882584e-06, "loss": 0.2343, "step": 25890 }, { "epoch": 160.81619937694705, "grad_norm": 1.7890625, "learning_rate": 4.6799989582339834e-06, "loss": 0.1315, "step": 25891 }, { "epoch": 160.82242990654206, "grad_norm": 1.53125, "learning_rate": 4.678563809937298e-06, "loss": 0.1063, "step": 25892 }, { "epoch": 160.82866043613708, "grad_norm": 1.3828125, "learning_rate": 4.677128859006469e-06, "loss": 0.098, "step": 25893 }, { "epoch": 160.8348909657321, "grad_norm": 1.9765625, "learning_rate": 4.675694105455419e-06, "loss": 0.1637, "step": 25894 }, { "epoch": 160.84112149532712, "grad_norm": 1.875, "learning_rate": 4.67425954929809e-06, "loss": 0.1823, "step": 25895 }, { "epoch": 160.84735202492212, "grad_norm": 2.453125, "learning_rate": 4.672825190548425e-06, "loss": 0.2072, "step": 25896 }, { "epoch": 160.85358255451715, "grad_norm": 1.53125, "learning_rate": 4.671391029220343e-06, "loss": 0.1479, "step": 25897 }, { "epoch": 160.85981308411215, "grad_norm": 2.59375, "learning_rate": 4.669957065327768e-06, "loss": 0.11, "step": 25898 }, { "epoch": 160.86604361370718, "grad_norm": 1.375, "learning_rate": 4.668523298884636e-06, "loss": 0.1154, "step": 25899 }, { "epoch": 160.87227414330218, "grad_norm": 2.421875, "learning_rate": 4.667089729904875e-06, "loss": 0.2264, "step": 25900 }, { "epoch": 160.8785046728972, "grad_norm": 1.609375, "learning_rate": 4.665656358402395e-06, "loss": 0.0971, "step": 25901 }, { "epoch": 160.8847352024922, "grad_norm": 2.8125, "learning_rate": 4.664223184391128e-06, "loss": 0.2513, "step": 25902 }, { "epoch": 160.89096573208724, "grad_norm": 2.84375, "learning_rate": 4.6627902078849955e-06, "loss": 0.2715, "step": 25903 }, { "epoch": 160.89719626168224, "grad_norm": 2.734375, "learning_rate": 4.661357428897911e-06, "loss": 0.1985, "step": 25904 }, { "epoch": 160.90342679127727, "grad_norm": 1.34375, "learning_rate": 4.659924847443781e-06, "loss": 0.129, "step": 25905 }, { "epoch": 160.90965732087227, "grad_norm": 2.109375, "learning_rate": 4.658492463536527e-06, "loss": 0.1906, "step": 25906 }, { "epoch": 160.9158878504673, "grad_norm": 1.8203125, "learning_rate": 4.657060277190067e-06, "loss": 0.1403, "step": 25907 }, { "epoch": 160.9221183800623, "grad_norm": 1.78125, "learning_rate": 4.655628288418298e-06, "loss": 0.1401, "step": 25908 }, { "epoch": 160.92834890965733, "grad_norm": 3.0625, "learning_rate": 4.654196497235139e-06, "loss": 0.32, "step": 25909 }, { "epoch": 160.93457943925233, "grad_norm": 1.6171875, "learning_rate": 4.652764903654485e-06, "loss": 0.1219, "step": 25910 }, { "epoch": 160.94080996884736, "grad_norm": 1.7734375, "learning_rate": 4.6513335076902534e-06, "loss": 0.1279, "step": 25911 }, { "epoch": 160.94704049844236, "grad_norm": 0.77734375, "learning_rate": 4.64990230935633e-06, "loss": 0.0897, "step": 25912 }, { "epoch": 160.9532710280374, "grad_norm": 2.34375, "learning_rate": 4.648471308666624e-06, "loss": 0.183, "step": 25913 }, { "epoch": 160.9595015576324, "grad_norm": 1.359375, "learning_rate": 4.647040505635031e-06, "loss": 0.1446, "step": 25914 }, { "epoch": 160.96573208722742, "grad_norm": 2.078125, "learning_rate": 4.645609900275458e-06, "loss": 0.1408, "step": 25915 }, { "epoch": 160.97196261682242, "grad_norm": 2.078125, "learning_rate": 4.644179492601792e-06, "loss": 0.1975, "step": 25916 }, { "epoch": 160.97819314641745, "grad_norm": 3.078125, "learning_rate": 4.642749282627912e-06, "loss": 0.2723, "step": 25917 }, { "epoch": 160.98442367601245, "grad_norm": 2.171875, "learning_rate": 4.641319270367725e-06, "loss": 0.2787, "step": 25918 }, { "epoch": 160.99065420560748, "grad_norm": 2.015625, "learning_rate": 4.63988945583512e-06, "loss": 0.1232, "step": 25919 }, { "epoch": 160.99688473520249, "grad_norm": 2.59375, "learning_rate": 4.638459839043971e-06, "loss": 0.248, "step": 25920 }, { "epoch": 161.0, "grad_norm": 2.265625, "learning_rate": 4.63703042000817e-06, "loss": 0.1457, "step": 25921 }, { "epoch": 161.00623052959503, "grad_norm": 2.453125, "learning_rate": 4.6356011987416075e-06, "loss": 0.2052, "step": 25922 }, { "epoch": 161.01246105919003, "grad_norm": 2.125, "learning_rate": 4.634172175258159e-06, "loss": 0.149, "step": 25923 }, { "epoch": 161.01869158878506, "grad_norm": 3.0625, "learning_rate": 4.6327433495716915e-06, "loss": 0.2087, "step": 25924 }, { "epoch": 161.02492211838006, "grad_norm": 1.6875, "learning_rate": 4.631314721696092e-06, "loss": 0.1442, "step": 25925 }, { "epoch": 161.0311526479751, "grad_norm": 1.265625, "learning_rate": 4.629886291645244e-06, "loss": 0.1157, "step": 25926 }, { "epoch": 161.0373831775701, "grad_norm": 2.09375, "learning_rate": 4.628458059433005e-06, "loss": 0.1623, "step": 25927 }, { "epoch": 161.04361370716512, "grad_norm": 1.265625, "learning_rate": 4.627030025073259e-06, "loss": 0.1337, "step": 25928 }, { "epoch": 161.04984423676012, "grad_norm": 2.625, "learning_rate": 4.625602188579867e-06, "loss": 0.2565, "step": 25929 }, { "epoch": 161.05607476635515, "grad_norm": 1.3359375, "learning_rate": 4.6241745499667005e-06, "loss": 0.1229, "step": 25930 }, { "epoch": 161.06230529595015, "grad_norm": 0.921875, "learning_rate": 4.622747109247622e-06, "loss": 0.0886, "step": 25931 }, { "epoch": 161.06853582554518, "grad_norm": 2.03125, "learning_rate": 4.621319866436494e-06, "loss": 0.2039, "step": 25932 }, { "epoch": 161.07476635514018, "grad_norm": 2.203125, "learning_rate": 4.619892821547189e-06, "loss": 0.1355, "step": 25933 }, { "epoch": 161.0809968847352, "grad_norm": 2.15625, "learning_rate": 4.618465974593553e-06, "loss": 0.1745, "step": 25934 }, { "epoch": 161.08722741433021, "grad_norm": 2.6875, "learning_rate": 4.617039325589453e-06, "loss": 0.1988, "step": 25935 }, { "epoch": 161.09345794392524, "grad_norm": 1.0859375, "learning_rate": 4.615612874548739e-06, "loss": 0.104, "step": 25936 }, { "epoch": 161.09968847352025, "grad_norm": 3.921875, "learning_rate": 4.61418662148527e-06, "loss": 0.2132, "step": 25937 }, { "epoch": 161.10591900311528, "grad_norm": 1.515625, "learning_rate": 4.612760566412891e-06, "loss": 0.1449, "step": 25938 }, { "epoch": 161.11214953271028, "grad_norm": 2.921875, "learning_rate": 4.611334709345458e-06, "loss": 0.1937, "step": 25939 }, { "epoch": 161.1183800623053, "grad_norm": 2.71875, "learning_rate": 4.609909050296823e-06, "loss": 0.2256, "step": 25940 }, { "epoch": 161.1246105919003, "grad_norm": 2.21875, "learning_rate": 4.608483589280827e-06, "loss": 0.1058, "step": 25941 }, { "epoch": 161.13084112149534, "grad_norm": 1.6328125, "learning_rate": 4.6070583263113085e-06, "loss": 0.0927, "step": 25942 }, { "epoch": 161.13707165109034, "grad_norm": 1.859375, "learning_rate": 4.605633261402114e-06, "loss": 0.1642, "step": 25943 }, { "epoch": 161.14330218068537, "grad_norm": 1.8203125, "learning_rate": 4.6042083945670975e-06, "loss": 0.1363, "step": 25944 }, { "epoch": 161.14953271028037, "grad_norm": 1.4140625, "learning_rate": 4.602783725820076e-06, "loss": 0.1164, "step": 25945 }, { "epoch": 161.1557632398754, "grad_norm": 1.6484375, "learning_rate": 4.601359255174898e-06, "loss": 0.1207, "step": 25946 }, { "epoch": 161.1619937694704, "grad_norm": 1.78125, "learning_rate": 4.599934982645398e-06, "loss": 0.1199, "step": 25947 }, { "epoch": 161.16822429906543, "grad_norm": 1.6328125, "learning_rate": 4.598510908245413e-06, "loss": 0.1589, "step": 25948 }, { "epoch": 161.17445482866043, "grad_norm": 1.5390625, "learning_rate": 4.597087031988759e-06, "loss": 0.1385, "step": 25949 }, { "epoch": 161.18068535825546, "grad_norm": 2.171875, "learning_rate": 4.5956633538892764e-06, "loss": 0.1627, "step": 25950 }, { "epoch": 161.18691588785046, "grad_norm": 1.9765625, "learning_rate": 4.5942398739607904e-06, "loss": 0.1958, "step": 25951 }, { "epoch": 161.1931464174455, "grad_norm": 0.94140625, "learning_rate": 4.592816592217128e-06, "loss": 0.0979, "step": 25952 }, { "epoch": 161.1993769470405, "grad_norm": 1.734375, "learning_rate": 4.591393508672107e-06, "loss": 0.1401, "step": 25953 }, { "epoch": 161.20560747663552, "grad_norm": 1.28125, "learning_rate": 4.589970623339557e-06, "loss": 0.11, "step": 25954 }, { "epoch": 161.21183800623052, "grad_norm": 1.8828125, "learning_rate": 4.588547936233287e-06, "loss": 0.1066, "step": 25955 }, { "epoch": 161.21806853582555, "grad_norm": 1.5859375, "learning_rate": 4.587125447367127e-06, "loss": 0.1383, "step": 25956 }, { "epoch": 161.22429906542055, "grad_norm": 2.171875, "learning_rate": 4.585703156754875e-06, "loss": 0.1818, "step": 25957 }, { "epoch": 161.23052959501558, "grad_norm": 1.0390625, "learning_rate": 4.584281064410359e-06, "loss": 0.1024, "step": 25958 }, { "epoch": 161.23676012461058, "grad_norm": 1.90625, "learning_rate": 4.582859170347389e-06, "loss": 0.1635, "step": 25959 }, { "epoch": 161.2429906542056, "grad_norm": 2.5625, "learning_rate": 4.581437474579775e-06, "loss": 0.1739, "step": 25960 }, { "epoch": 161.2492211838006, "grad_norm": 2.0625, "learning_rate": 4.580015977121313e-06, "loss": 0.1903, "step": 25961 }, { "epoch": 161.25545171339564, "grad_norm": 2.359375, "learning_rate": 4.57859467798582e-06, "loss": 0.2087, "step": 25962 }, { "epoch": 161.26168224299064, "grad_norm": 2.03125, "learning_rate": 4.577173577187102e-06, "loss": 0.1414, "step": 25963 }, { "epoch": 161.26791277258567, "grad_norm": 1.71875, "learning_rate": 4.575752674738951e-06, "loss": 0.1316, "step": 25964 }, { "epoch": 161.27414330218068, "grad_norm": 1.390625, "learning_rate": 4.574331970655172e-06, "loss": 0.0823, "step": 25965 }, { "epoch": 161.2803738317757, "grad_norm": 1.9921875, "learning_rate": 4.572911464949572e-06, "loss": 0.1574, "step": 25966 }, { "epoch": 161.2866043613707, "grad_norm": 1.65625, "learning_rate": 4.571491157635937e-06, "loss": 0.135, "step": 25967 }, { "epoch": 161.29283489096574, "grad_norm": 3.0625, "learning_rate": 4.570071048728058e-06, "loss": 0.2081, "step": 25968 }, { "epoch": 161.29906542056074, "grad_norm": 2.34375, "learning_rate": 4.568651138239735e-06, "loss": 0.1525, "step": 25969 }, { "epoch": 161.30529595015577, "grad_norm": 2.359375, "learning_rate": 4.5672314261847594e-06, "loss": 0.2322, "step": 25970 }, { "epoch": 161.31152647975077, "grad_norm": 1.265625, "learning_rate": 4.565811912576912e-06, "loss": 0.1062, "step": 25971 }, { "epoch": 161.3177570093458, "grad_norm": 1.953125, "learning_rate": 4.564392597429984e-06, "loss": 0.164, "step": 25972 }, { "epoch": 161.3239875389408, "grad_norm": 2.0625, "learning_rate": 4.562973480757765e-06, "loss": 0.2138, "step": 25973 }, { "epoch": 161.33021806853583, "grad_norm": 1.5390625, "learning_rate": 4.561554562574036e-06, "loss": 0.1027, "step": 25974 }, { "epoch": 161.33644859813083, "grad_norm": 2.90625, "learning_rate": 4.560135842892565e-06, "loss": 0.2598, "step": 25975 }, { "epoch": 161.34267912772586, "grad_norm": 0.94140625, "learning_rate": 4.558717321727144e-06, "loss": 0.0915, "step": 25976 }, { "epoch": 161.34890965732086, "grad_norm": 1.9375, "learning_rate": 4.557298999091541e-06, "loss": 0.1505, "step": 25977 }, { "epoch": 161.3551401869159, "grad_norm": 2.609375, "learning_rate": 4.555880874999546e-06, "loss": 0.149, "step": 25978 }, { "epoch": 161.3613707165109, "grad_norm": 1.34375, "learning_rate": 4.554462949464924e-06, "loss": 0.1474, "step": 25979 }, { "epoch": 161.36760124610592, "grad_norm": 2.53125, "learning_rate": 4.553045222501437e-06, "loss": 0.1685, "step": 25980 }, { "epoch": 161.37383177570092, "grad_norm": 1.875, "learning_rate": 4.551627694122862e-06, "loss": 0.1186, "step": 25981 }, { "epoch": 161.38006230529595, "grad_norm": 2.734375, "learning_rate": 4.550210364342975e-06, "loss": 0.1868, "step": 25982 }, { "epoch": 161.38629283489095, "grad_norm": 2.4375, "learning_rate": 4.548793233175525e-06, "loss": 0.1868, "step": 25983 }, { "epoch": 161.39252336448598, "grad_norm": 2.328125, "learning_rate": 4.547376300634282e-06, "loss": 0.2503, "step": 25984 }, { "epoch": 161.39875389408098, "grad_norm": 2.375, "learning_rate": 4.545959566733016e-06, "loss": 0.1973, "step": 25985 }, { "epoch": 161.404984423676, "grad_norm": 1.2890625, "learning_rate": 4.544543031485479e-06, "loss": 0.1022, "step": 25986 }, { "epoch": 161.41121495327104, "grad_norm": 1.125, "learning_rate": 4.5431266949054226e-06, "loss": 0.1113, "step": 25987 }, { "epoch": 161.41744548286604, "grad_norm": 2.53125, "learning_rate": 4.5417105570066095e-06, "loss": 0.2116, "step": 25988 }, { "epoch": 161.42367601246107, "grad_norm": 2.296875, "learning_rate": 4.540294617802798e-06, "loss": 0.1376, "step": 25989 }, { "epoch": 161.42990654205607, "grad_norm": 0.828125, "learning_rate": 4.5388788773077315e-06, "loss": 0.0994, "step": 25990 }, { "epoch": 161.4361370716511, "grad_norm": 3.015625, "learning_rate": 4.537463335535161e-06, "loss": 0.1846, "step": 25991 }, { "epoch": 161.4423676012461, "grad_norm": 2.28125, "learning_rate": 4.536047992498843e-06, "loss": 0.1528, "step": 25992 }, { "epoch": 161.44859813084113, "grad_norm": 1.828125, "learning_rate": 4.534632848212519e-06, "loss": 0.136, "step": 25993 }, { "epoch": 161.45482866043614, "grad_norm": 2.59375, "learning_rate": 4.533217902689926e-06, "loss": 0.1449, "step": 25994 }, { "epoch": 161.46105919003116, "grad_norm": 1.953125, "learning_rate": 4.531803155944811e-06, "loss": 0.1442, "step": 25995 }, { "epoch": 161.46728971962617, "grad_norm": 1.34375, "learning_rate": 4.530388607990921e-06, "loss": 0.1377, "step": 25996 }, { "epoch": 161.4735202492212, "grad_norm": 1.4609375, "learning_rate": 4.5289742588419835e-06, "loss": 0.1189, "step": 25997 }, { "epoch": 161.4797507788162, "grad_norm": 2.21875, "learning_rate": 4.5275601085117445e-06, "loss": 0.1534, "step": 25998 }, { "epoch": 161.48598130841123, "grad_norm": 2.203125, "learning_rate": 4.5261461570139305e-06, "loss": 0.2396, "step": 25999 }, { "epoch": 161.49221183800623, "grad_norm": 2.375, "learning_rate": 4.5247324043622805e-06, "loss": 0.2268, "step": 26000 }, { "epoch": 161.49844236760126, "grad_norm": 1.6953125, "learning_rate": 4.523318850570518e-06, "loss": 0.1404, "step": 26001 }, { "epoch": 161.50467289719626, "grad_norm": 1.59375, "learning_rate": 4.521905495652376e-06, "loss": 0.1161, "step": 26002 }, { "epoch": 161.5109034267913, "grad_norm": 1.4609375, "learning_rate": 4.520492339621588e-06, "loss": 0.1122, "step": 26003 }, { "epoch": 161.5171339563863, "grad_norm": 1.640625, "learning_rate": 4.519079382491867e-06, "loss": 0.1276, "step": 26004 }, { "epoch": 161.52336448598132, "grad_norm": 2.140625, "learning_rate": 4.517666624276945e-06, "loss": 0.1305, "step": 26005 }, { "epoch": 161.52959501557632, "grad_norm": 3.03125, "learning_rate": 4.516254064990533e-06, "loss": 0.342, "step": 26006 }, { "epoch": 161.53582554517135, "grad_norm": 1.59375, "learning_rate": 4.514841704646364e-06, "loss": 0.1597, "step": 26007 }, { "epoch": 161.54205607476635, "grad_norm": 1.1953125, "learning_rate": 4.513429543258141e-06, "loss": 0.1171, "step": 26008 }, { "epoch": 161.54828660436138, "grad_norm": 1.453125, "learning_rate": 4.512017580839583e-06, "loss": 0.1007, "step": 26009 }, { "epoch": 161.55451713395638, "grad_norm": 2.171875, "learning_rate": 4.5106058174044064e-06, "loss": 0.1815, "step": 26010 }, { "epoch": 161.5607476635514, "grad_norm": 2.234375, "learning_rate": 4.5091942529663366e-06, "loss": 0.1814, "step": 26011 }, { "epoch": 161.5669781931464, "grad_norm": 2.875, "learning_rate": 4.507782887539056e-06, "loss": 0.1878, "step": 26012 }, { "epoch": 161.57320872274144, "grad_norm": 1.171875, "learning_rate": 4.5063717211362825e-06, "loss": 0.1033, "step": 26013 }, { "epoch": 161.57943925233644, "grad_norm": 3.140625, "learning_rate": 4.504960753771723e-06, "loss": 0.2888, "step": 26014 }, { "epoch": 161.58566978193147, "grad_norm": 2.125, "learning_rate": 4.503549985459088e-06, "loss": 0.1972, "step": 26015 }, { "epoch": 161.59190031152647, "grad_norm": 2.265625, "learning_rate": 4.502139416212067e-06, "loss": 0.277, "step": 26016 }, { "epoch": 161.5981308411215, "grad_norm": 2.265625, "learning_rate": 4.5007290460443715e-06, "loss": 0.1719, "step": 26017 }, { "epoch": 161.6043613707165, "grad_norm": 3.234375, "learning_rate": 4.499318874969685e-06, "loss": 0.314, "step": 26018 }, { "epoch": 161.61059190031153, "grad_norm": 1.6484375, "learning_rate": 4.4979089030017216e-06, "loss": 0.1244, "step": 26019 }, { "epoch": 161.61682242990653, "grad_norm": 2.21875, "learning_rate": 4.496499130154158e-06, "loss": 0.1693, "step": 26020 }, { "epoch": 161.62305295950156, "grad_norm": 0.9140625, "learning_rate": 4.49508955644069e-06, "loss": 0.0941, "step": 26021 }, { "epoch": 161.62928348909657, "grad_norm": 1.3046875, "learning_rate": 4.4936801818750205e-06, "loss": 0.0817, "step": 26022 }, { "epoch": 161.6355140186916, "grad_norm": 3.078125, "learning_rate": 4.492271006470822e-06, "loss": 0.3213, "step": 26023 }, { "epoch": 161.6417445482866, "grad_norm": 2.703125, "learning_rate": 4.490862030241793e-06, "loss": 0.1767, "step": 26024 }, { "epoch": 161.64797507788163, "grad_norm": 1.7890625, "learning_rate": 4.489453253201606e-06, "loss": 0.1933, "step": 26025 }, { "epoch": 161.65420560747663, "grad_norm": 1.875, "learning_rate": 4.4880446753639536e-06, "loss": 0.142, "step": 26026 }, { "epoch": 161.66043613707166, "grad_norm": 1.5390625, "learning_rate": 4.486636296742506e-06, "loss": 0.1101, "step": 26027 }, { "epoch": 161.66666666666666, "grad_norm": 0.8671875, "learning_rate": 4.485228117350948e-06, "loss": 0.1042, "step": 26028 }, { "epoch": 161.6728971962617, "grad_norm": 1.828125, "learning_rate": 4.48382013720296e-06, "loss": 0.1384, "step": 26029 }, { "epoch": 161.6791277258567, "grad_norm": 2.0, "learning_rate": 4.482412356312213e-06, "loss": 0.1657, "step": 26030 }, { "epoch": 161.68535825545172, "grad_norm": 1.8359375, "learning_rate": 4.481004774692371e-06, "loss": 0.2378, "step": 26031 }, { "epoch": 161.69158878504672, "grad_norm": 1.265625, "learning_rate": 4.479597392357113e-06, "loss": 0.1573, "step": 26032 }, { "epoch": 161.69781931464175, "grad_norm": 1.4921875, "learning_rate": 4.478190209320116e-06, "loss": 0.1034, "step": 26033 }, { "epoch": 161.70404984423675, "grad_norm": 1.8359375, "learning_rate": 4.476783225595027e-06, "loss": 0.1466, "step": 26034 }, { "epoch": 161.71028037383178, "grad_norm": 2.359375, "learning_rate": 4.475376441195525e-06, "loss": 0.2663, "step": 26035 }, { "epoch": 161.71651090342678, "grad_norm": 2.765625, "learning_rate": 4.4739698561352735e-06, "loss": 0.2056, "step": 26036 }, { "epoch": 161.7227414330218, "grad_norm": 1.6953125, "learning_rate": 4.472563470427932e-06, "loss": 0.1388, "step": 26037 }, { "epoch": 161.7289719626168, "grad_norm": 2.078125, "learning_rate": 4.471157284087152e-06, "loss": 0.1655, "step": 26038 }, { "epoch": 161.73520249221184, "grad_norm": 2.234375, "learning_rate": 4.469751297126596e-06, "loss": 0.1886, "step": 26039 }, { "epoch": 161.74143302180684, "grad_norm": 1.890625, "learning_rate": 4.468345509559923e-06, "loss": 0.1457, "step": 26040 }, { "epoch": 161.74766355140187, "grad_norm": 1.3671875, "learning_rate": 4.4669399214007785e-06, "loss": 0.1156, "step": 26041 }, { "epoch": 161.75389408099687, "grad_norm": 2.171875, "learning_rate": 4.465534532662815e-06, "loss": 0.1605, "step": 26042 }, { "epoch": 161.7601246105919, "grad_norm": 2.078125, "learning_rate": 4.464129343359694e-06, "loss": 0.1314, "step": 26043 }, { "epoch": 161.7663551401869, "grad_norm": 3.25, "learning_rate": 4.462724353505055e-06, "loss": 0.2692, "step": 26044 }, { "epoch": 161.77258566978193, "grad_norm": 2.140625, "learning_rate": 4.461319563112532e-06, "loss": 0.1437, "step": 26045 }, { "epoch": 161.77881619937693, "grad_norm": 2.9375, "learning_rate": 4.459914972195781e-06, "loss": 0.1991, "step": 26046 }, { "epoch": 161.78504672897196, "grad_norm": 1.6328125, "learning_rate": 4.458510580768443e-06, "loss": 0.1123, "step": 26047 }, { "epoch": 161.79127725856696, "grad_norm": 1.6640625, "learning_rate": 4.457106388844162e-06, "loss": 0.1517, "step": 26048 }, { "epoch": 161.797507788162, "grad_norm": 2.359375, "learning_rate": 4.45570239643657e-06, "loss": 0.092, "step": 26049 }, { "epoch": 161.80373831775702, "grad_norm": 2.09375, "learning_rate": 4.454298603559295e-06, "loss": 0.2285, "step": 26050 }, { "epoch": 161.80996884735202, "grad_norm": 2.546875, "learning_rate": 4.4528950102259796e-06, "loss": 0.2262, "step": 26051 }, { "epoch": 161.81619937694705, "grad_norm": 1.59375, "learning_rate": 4.4514916164502655e-06, "loss": 0.1197, "step": 26052 }, { "epoch": 161.82242990654206, "grad_norm": 1.7265625, "learning_rate": 4.45008842224576e-06, "loss": 0.1179, "step": 26053 }, { "epoch": 161.82866043613708, "grad_norm": 1.7421875, "learning_rate": 4.448685427626109e-06, "loss": 0.1109, "step": 26054 }, { "epoch": 161.8348909657321, "grad_norm": 2.71875, "learning_rate": 4.447282632604935e-06, "loss": 0.3068, "step": 26055 }, { "epoch": 161.84112149532712, "grad_norm": 1.2734375, "learning_rate": 4.445880037195862e-06, "loss": 0.121, "step": 26056 }, { "epoch": 161.84735202492212, "grad_norm": 1.9453125, "learning_rate": 4.444477641412506e-06, "loss": 0.1543, "step": 26057 }, { "epoch": 161.85358255451715, "grad_norm": 1.3515625, "learning_rate": 4.4430754452684904e-06, "loss": 0.1124, "step": 26058 }, { "epoch": 161.85981308411215, "grad_norm": 1.765625, "learning_rate": 4.4416734487774445e-06, "loss": 0.1567, "step": 26059 }, { "epoch": 161.86604361370718, "grad_norm": 2.484375, "learning_rate": 4.440271651952965e-06, "loss": 0.1902, "step": 26060 }, { "epoch": 161.87227414330218, "grad_norm": 3.0, "learning_rate": 4.43887005480868e-06, "loss": 0.266, "step": 26061 }, { "epoch": 161.8785046728972, "grad_norm": 2.140625, "learning_rate": 4.437468657358204e-06, "loss": 0.1417, "step": 26062 }, { "epoch": 161.8847352024922, "grad_norm": 1.1484375, "learning_rate": 4.436067459615145e-06, "loss": 0.1286, "step": 26063 }, { "epoch": 161.89096573208724, "grad_norm": 1.078125, "learning_rate": 4.434666461593101e-06, "loss": 0.1123, "step": 26064 }, { "epoch": 161.89719626168224, "grad_norm": 1.3828125, "learning_rate": 4.433265663305688e-06, "loss": 0.1247, "step": 26065 }, { "epoch": 161.90342679127727, "grad_norm": 2.015625, "learning_rate": 4.431865064766516e-06, "loss": 0.1562, "step": 26066 }, { "epoch": 161.90965732087227, "grad_norm": 2.421875, "learning_rate": 4.4304646659891755e-06, "loss": 0.2294, "step": 26067 }, { "epoch": 161.9158878504673, "grad_norm": 1.3359375, "learning_rate": 4.429064466987281e-06, "loss": 0.1191, "step": 26068 }, { "epoch": 161.9221183800623, "grad_norm": 1.3828125, "learning_rate": 4.427664467774418e-06, "loss": 0.1157, "step": 26069 }, { "epoch": 161.92834890965733, "grad_norm": 2.859375, "learning_rate": 4.426264668364197e-06, "loss": 0.1545, "step": 26070 }, { "epoch": 161.93457943925233, "grad_norm": 1.890625, "learning_rate": 4.424865068770198e-06, "loss": 0.1596, "step": 26071 }, { "epoch": 161.94080996884736, "grad_norm": 2.65625, "learning_rate": 4.423465669006024e-06, "loss": 0.1988, "step": 26072 }, { "epoch": 161.94704049844236, "grad_norm": 1.640625, "learning_rate": 4.422066469085262e-06, "loss": 0.1206, "step": 26073 }, { "epoch": 161.9532710280374, "grad_norm": 1.953125, "learning_rate": 4.420667469021514e-06, "loss": 0.2079, "step": 26074 }, { "epoch": 161.9595015576324, "grad_norm": 1.1171875, "learning_rate": 4.4192686688283555e-06, "loss": 0.108, "step": 26075 }, { "epoch": 161.96573208722742, "grad_norm": 2.1875, "learning_rate": 4.417870068519367e-06, "loss": 0.2983, "step": 26076 }, { "epoch": 161.97196261682242, "grad_norm": 1.15625, "learning_rate": 4.416471668108138e-06, "loss": 0.1111, "step": 26077 }, { "epoch": 161.97819314641745, "grad_norm": 1.2109375, "learning_rate": 4.415073467608258e-06, "loss": 0.1419, "step": 26078 }, { "epoch": 161.98442367601245, "grad_norm": 1.9609375, "learning_rate": 4.413675467033293e-06, "loss": 0.188, "step": 26079 }, { "epoch": 161.99065420560748, "grad_norm": 2.671875, "learning_rate": 4.4122776663968275e-06, "loss": 0.1916, "step": 26080 }, { "epoch": 161.99688473520249, "grad_norm": 3.09375, "learning_rate": 4.4108800657124415e-06, "loss": 0.2592, "step": 26081 }, { "epoch": 162.0, "grad_norm": 1.375, "learning_rate": 4.409482664993706e-06, "loss": 0.0768, "step": 26082 }, { "epoch": 162.00623052959503, "grad_norm": 0.75, "learning_rate": 4.408085464254183e-06, "loss": 0.0899, "step": 26083 }, { "epoch": 162.01246105919003, "grad_norm": 3.34375, "learning_rate": 4.40668846350745e-06, "loss": 0.2785, "step": 26084 }, { "epoch": 162.01869158878506, "grad_norm": 3.046875, "learning_rate": 4.405291662767083e-06, "loss": 0.3378, "step": 26085 }, { "epoch": 162.02492211838006, "grad_norm": 2.28125, "learning_rate": 4.403895062046631e-06, "loss": 0.2067, "step": 26086 }, { "epoch": 162.0311526479751, "grad_norm": 1.5546875, "learning_rate": 4.402498661359672e-06, "loss": 0.1374, "step": 26087 }, { "epoch": 162.0373831775701, "grad_norm": 2.5625, "learning_rate": 4.401102460719766e-06, "loss": 0.2016, "step": 26088 }, { "epoch": 162.04361370716512, "grad_norm": 1.9765625, "learning_rate": 4.399706460140471e-06, "loss": 0.1249, "step": 26089 }, { "epoch": 162.04984423676012, "grad_norm": 1.90625, "learning_rate": 4.398310659635338e-06, "loss": 0.175, "step": 26090 }, { "epoch": 162.05607476635515, "grad_norm": 1.9140625, "learning_rate": 4.396915059217932e-06, "loss": 0.1651, "step": 26091 }, { "epoch": 162.06230529595015, "grad_norm": 1.5234375, "learning_rate": 4.395519658901812e-06, "loss": 0.11, "step": 26092 }, { "epoch": 162.06853582554518, "grad_norm": 2.15625, "learning_rate": 4.394124458700516e-06, "loss": 0.1885, "step": 26093 }, { "epoch": 162.07476635514018, "grad_norm": 2.03125, "learning_rate": 4.392729458627606e-06, "loss": 0.1494, "step": 26094 }, { "epoch": 162.0809968847352, "grad_norm": 2.953125, "learning_rate": 4.391334658696624e-06, "loss": 0.1884, "step": 26095 }, { "epoch": 162.08722741433021, "grad_norm": 2.65625, "learning_rate": 4.389940058921127e-06, "loss": 0.1611, "step": 26096 }, { "epoch": 162.09345794392524, "grad_norm": 2.078125, "learning_rate": 4.3885456593146415e-06, "loss": 0.2637, "step": 26097 }, { "epoch": 162.09968847352025, "grad_norm": 1.9296875, "learning_rate": 4.387151459890723e-06, "loss": 0.1428, "step": 26098 }, { "epoch": 162.10591900311528, "grad_norm": 1.875, "learning_rate": 4.3857574606629175e-06, "loss": 0.1666, "step": 26099 }, { "epoch": 162.11214953271028, "grad_norm": 1.8828125, "learning_rate": 4.384363661644756e-06, "loss": 0.1386, "step": 26100 }, { "epoch": 162.1183800623053, "grad_norm": 1.4296875, "learning_rate": 4.382970062849767e-06, "loss": 0.1113, "step": 26101 }, { "epoch": 162.1246105919003, "grad_norm": 2.59375, "learning_rate": 4.381576664291495e-06, "loss": 0.1964, "step": 26102 }, { "epoch": 162.13084112149534, "grad_norm": 1.8203125, "learning_rate": 4.380183465983479e-06, "loss": 0.1903, "step": 26103 }, { "epoch": 162.13707165109034, "grad_norm": 1.71875, "learning_rate": 4.3787904679392365e-06, "loss": 0.1228, "step": 26104 }, { "epoch": 162.14330218068537, "grad_norm": 1.1640625, "learning_rate": 4.377397670172301e-06, "loss": 0.1114, "step": 26105 }, { "epoch": 162.14953271028037, "grad_norm": 1.921875, "learning_rate": 4.376005072696204e-06, "loss": 0.1623, "step": 26106 }, { "epoch": 162.1557632398754, "grad_norm": 1.3203125, "learning_rate": 4.374612675524481e-06, "loss": 0.1244, "step": 26107 }, { "epoch": 162.1619937694704, "grad_norm": 2.703125, "learning_rate": 4.373220478670628e-06, "loss": 0.1811, "step": 26108 }, { "epoch": 162.16822429906543, "grad_norm": 2.671875, "learning_rate": 4.3718284821481825e-06, "loss": 0.2472, "step": 26109 }, { "epoch": 162.17445482866043, "grad_norm": 1.921875, "learning_rate": 4.370436685970664e-06, "loss": 0.1844, "step": 26110 }, { "epoch": 162.18068535825546, "grad_norm": 1.3203125, "learning_rate": 4.369045090151591e-06, "loss": 0.0972, "step": 26111 }, { "epoch": 162.18691588785046, "grad_norm": 2.15625, "learning_rate": 4.3676536947044736e-06, "loss": 0.1704, "step": 26112 }, { "epoch": 162.1931464174455, "grad_norm": 2.671875, "learning_rate": 4.3662624996428316e-06, "loss": 0.2042, "step": 26113 }, { "epoch": 162.1993769470405, "grad_norm": 2.203125, "learning_rate": 4.364871504980167e-06, "loss": 0.2586, "step": 26114 }, { "epoch": 162.20560747663552, "grad_norm": 2.5, "learning_rate": 4.3634807107300005e-06, "loss": 0.1354, "step": 26115 }, { "epoch": 162.21183800623052, "grad_norm": 1.8671875, "learning_rate": 4.362090116905828e-06, "loss": 0.1599, "step": 26116 }, { "epoch": 162.21806853582555, "grad_norm": 1.546875, "learning_rate": 4.360699723521164e-06, "loss": 0.1498, "step": 26117 }, { "epoch": 162.22429906542055, "grad_norm": 1.7265625, "learning_rate": 4.359309530589514e-06, "loss": 0.1621, "step": 26118 }, { "epoch": 162.23052959501558, "grad_norm": 1.234375, "learning_rate": 4.357919538124377e-06, "loss": 0.1056, "step": 26119 }, { "epoch": 162.23676012461058, "grad_norm": 2.3125, "learning_rate": 4.356529746139246e-06, "loss": 0.1634, "step": 26120 }, { "epoch": 162.2429906542056, "grad_norm": 2.03125, "learning_rate": 4.355140154647624e-06, "loss": 0.2189, "step": 26121 }, { "epoch": 162.2492211838006, "grad_norm": 2.09375, "learning_rate": 4.353750763663014e-06, "loss": 0.1574, "step": 26122 }, { "epoch": 162.25545171339564, "grad_norm": 2.953125, "learning_rate": 4.352361573198899e-06, "loss": 0.2882, "step": 26123 }, { "epoch": 162.26168224299064, "grad_norm": 2.359375, "learning_rate": 4.3509725832687726e-06, "loss": 0.1608, "step": 26124 }, { "epoch": 162.26791277258567, "grad_norm": 1.3828125, "learning_rate": 4.349583793886136e-06, "loss": 0.1066, "step": 26125 }, { "epoch": 162.27414330218068, "grad_norm": 1.65625, "learning_rate": 4.348195205064471e-06, "loss": 0.1095, "step": 26126 }, { "epoch": 162.2803738317757, "grad_norm": 3.15625, "learning_rate": 4.346806816817254e-06, "loss": 0.2036, "step": 26127 }, { "epoch": 162.2866043613707, "grad_norm": 2.75, "learning_rate": 4.34541862915798e-06, "loss": 0.2611, "step": 26128 }, { "epoch": 162.29283489096574, "grad_norm": 3.265625, "learning_rate": 4.344030642100133e-06, "loss": 0.3055, "step": 26129 }, { "epoch": 162.29906542056074, "grad_norm": 2.5, "learning_rate": 4.342642855657184e-06, "loss": 0.1394, "step": 26130 }, { "epoch": 162.30529595015577, "grad_norm": 2.71875, "learning_rate": 4.3412552698426145e-06, "loss": 0.1962, "step": 26131 }, { "epoch": 162.31152647975077, "grad_norm": 1.921875, "learning_rate": 4.339867884669913e-06, "loss": 0.1325, "step": 26132 }, { "epoch": 162.3177570093458, "grad_norm": 1.875, "learning_rate": 4.338480700152542e-06, "loss": 0.1519, "step": 26133 }, { "epoch": 162.3239875389408, "grad_norm": 1.2421875, "learning_rate": 4.337093716303972e-06, "loss": 0.1062, "step": 26134 }, { "epoch": 162.33021806853583, "grad_norm": 1.9140625, "learning_rate": 4.335706933137679e-06, "loss": 0.1576, "step": 26135 }, { "epoch": 162.33644859813083, "grad_norm": 2.296875, "learning_rate": 4.334320350667137e-06, "loss": 0.2221, "step": 26136 }, { "epoch": 162.34267912772586, "grad_norm": 2.984375, "learning_rate": 4.3329339689058e-06, "loss": 0.2475, "step": 26137 }, { "epoch": 162.34890965732086, "grad_norm": 2.28125, "learning_rate": 4.331547787867144e-06, "loss": 0.1916, "step": 26138 }, { "epoch": 162.3551401869159, "grad_norm": 1.890625, "learning_rate": 4.3301618075646225e-06, "loss": 0.1145, "step": 26139 }, { "epoch": 162.3613707165109, "grad_norm": 1.09375, "learning_rate": 4.3287760280117075e-06, "loss": 0.1229, "step": 26140 }, { "epoch": 162.36760124610592, "grad_norm": 1.875, "learning_rate": 4.327390449221846e-06, "loss": 0.1315, "step": 26141 }, { "epoch": 162.37383177570092, "grad_norm": 1.046875, "learning_rate": 4.3260050712085e-06, "loss": 0.1024, "step": 26142 }, { "epoch": 162.38006230529595, "grad_norm": 1.4296875, "learning_rate": 4.3246198939851265e-06, "loss": 0.1221, "step": 26143 }, { "epoch": 162.38629283489095, "grad_norm": 1.8203125, "learning_rate": 4.3232349175651835e-06, "loss": 0.2222, "step": 26144 }, { "epoch": 162.39252336448598, "grad_norm": 2.625, "learning_rate": 4.321850141962116e-06, "loss": 0.1992, "step": 26145 }, { "epoch": 162.39875389408098, "grad_norm": 2.4375, "learning_rate": 4.320465567189369e-06, "loss": 0.2134, "step": 26146 }, { "epoch": 162.404984423676, "grad_norm": 1.53125, "learning_rate": 4.319081193260391e-06, "loss": 0.1475, "step": 26147 }, { "epoch": 162.41121495327104, "grad_norm": 3.109375, "learning_rate": 4.317697020188638e-06, "loss": 0.2535, "step": 26148 }, { "epoch": 162.41744548286604, "grad_norm": 1.6015625, "learning_rate": 4.316313047987539e-06, "loss": 0.1256, "step": 26149 }, { "epoch": 162.42367601246107, "grad_norm": 1.5859375, "learning_rate": 4.314929276670543e-06, "loss": 0.1151, "step": 26150 }, { "epoch": 162.42990654205607, "grad_norm": 1.8203125, "learning_rate": 4.313545706251096e-06, "loss": 0.1608, "step": 26151 }, { "epoch": 162.4361370716511, "grad_norm": 1.3515625, "learning_rate": 4.312162336742626e-06, "loss": 0.111, "step": 26152 }, { "epoch": 162.4423676012461, "grad_norm": 2.28125, "learning_rate": 4.3107791681585655e-06, "loss": 0.1762, "step": 26153 }, { "epoch": 162.44859813084113, "grad_norm": 1.625, "learning_rate": 4.309396200512355e-06, "loss": 0.1249, "step": 26154 }, { "epoch": 162.45482866043614, "grad_norm": 0.9609375, "learning_rate": 4.308013433817429e-06, "loss": 0.0936, "step": 26155 }, { "epoch": 162.46105919003116, "grad_norm": 3.015625, "learning_rate": 4.306630868087208e-06, "loss": 0.1311, "step": 26156 }, { "epoch": 162.46728971962617, "grad_norm": 1.8125, "learning_rate": 4.305248503335121e-06, "loss": 0.1092, "step": 26157 }, { "epoch": 162.4735202492212, "grad_norm": 1.40625, "learning_rate": 4.303866339574608e-06, "loss": 0.1283, "step": 26158 }, { "epoch": 162.4797507788162, "grad_norm": 2.6875, "learning_rate": 4.30248437681908e-06, "loss": 0.2, "step": 26159 }, { "epoch": 162.48598130841123, "grad_norm": 0.9140625, "learning_rate": 4.301102615081956e-06, "loss": 0.1092, "step": 26160 }, { "epoch": 162.49221183800623, "grad_norm": 1.2109375, "learning_rate": 4.2997210543766595e-06, "loss": 0.1202, "step": 26161 }, { "epoch": 162.49844236760126, "grad_norm": 2.703125, "learning_rate": 4.298339694716616e-06, "loss": 0.1492, "step": 26162 }, { "epoch": 162.50467289719626, "grad_norm": 2.03125, "learning_rate": 4.296958536115231e-06, "loss": 0.143, "step": 26163 }, { "epoch": 162.5109034267913, "grad_norm": 1.1328125, "learning_rate": 4.295577578585927e-06, "loss": 0.1124, "step": 26164 }, { "epoch": 162.5171339563863, "grad_norm": 1.5078125, "learning_rate": 4.294196822142108e-06, "loss": 0.1111, "step": 26165 }, { "epoch": 162.52336448598132, "grad_norm": 1.1875, "learning_rate": 4.292816266797195e-06, "loss": 0.132, "step": 26166 }, { "epoch": 162.52959501557632, "grad_norm": 3.59375, "learning_rate": 4.291435912564584e-06, "loss": 0.3124, "step": 26167 }, { "epoch": 162.53582554517135, "grad_norm": 1.9375, "learning_rate": 4.290055759457684e-06, "loss": 0.1574, "step": 26168 }, { "epoch": 162.54205607476635, "grad_norm": 1.4921875, "learning_rate": 4.288675807489906e-06, "loss": 0.0948, "step": 26169 }, { "epoch": 162.54828660436138, "grad_norm": 2.5, "learning_rate": 4.287296056674658e-06, "loss": 0.1569, "step": 26170 }, { "epoch": 162.55451713395638, "grad_norm": 2.703125, "learning_rate": 4.28591650702532e-06, "loss": 0.2514, "step": 26171 }, { "epoch": 162.5607476635514, "grad_norm": 1.625, "learning_rate": 4.284537158555302e-06, "loss": 0.1413, "step": 26172 }, { "epoch": 162.5669781931464, "grad_norm": 1.4921875, "learning_rate": 4.283158011277999e-06, "loss": 0.1277, "step": 26173 }, { "epoch": 162.57320872274144, "grad_norm": 2.90625, "learning_rate": 4.2817790652068116e-06, "loss": 0.21, "step": 26174 }, { "epoch": 162.57943925233644, "grad_norm": 1.9609375, "learning_rate": 4.280400320355124e-06, "loss": 0.1344, "step": 26175 }, { "epoch": 162.58566978193147, "grad_norm": 1.9296875, "learning_rate": 4.27902177673633e-06, "loss": 0.1879, "step": 26176 }, { "epoch": 162.59190031152647, "grad_norm": 1.28125, "learning_rate": 4.277643434363824e-06, "loss": 0.1002, "step": 26177 }, { "epoch": 162.5981308411215, "grad_norm": 2.1875, "learning_rate": 4.276265293250986e-06, "loss": 0.2405, "step": 26178 }, { "epoch": 162.6043613707165, "grad_norm": 2.8125, "learning_rate": 4.274887353411197e-06, "loss": 0.1665, "step": 26179 }, { "epoch": 162.61059190031153, "grad_norm": 2.515625, "learning_rate": 4.2735096148578455e-06, "loss": 0.2056, "step": 26180 }, { "epoch": 162.61682242990653, "grad_norm": 2.84375, "learning_rate": 4.2721320776043186e-06, "loss": 0.1813, "step": 26181 }, { "epoch": 162.62305295950156, "grad_norm": 1.484375, "learning_rate": 4.27075474166398e-06, "loss": 0.1523, "step": 26182 }, { "epoch": 162.62928348909657, "grad_norm": 1.6953125, "learning_rate": 4.2693776070502205e-06, "loss": 0.1389, "step": 26183 }, { "epoch": 162.6355140186916, "grad_norm": 1.8828125, "learning_rate": 4.268000673776406e-06, "loss": 0.1504, "step": 26184 }, { "epoch": 162.6417445482866, "grad_norm": 2.421875, "learning_rate": 4.266623941855919e-06, "loss": 0.1519, "step": 26185 }, { "epoch": 162.64797507788163, "grad_norm": 2.40625, "learning_rate": 4.2652474113021176e-06, "loss": 0.1732, "step": 26186 }, { "epoch": 162.65420560747663, "grad_norm": 1.8046875, "learning_rate": 4.263871082128379e-06, "loss": 0.1697, "step": 26187 }, { "epoch": 162.66043613707166, "grad_norm": 0.99609375, "learning_rate": 4.262494954348076e-06, "loss": 0.0837, "step": 26188 }, { "epoch": 162.66666666666666, "grad_norm": 1.8046875, "learning_rate": 4.261119027974567e-06, "loss": 0.1324, "step": 26189 }, { "epoch": 162.6728971962617, "grad_norm": 3.453125, "learning_rate": 4.259743303021207e-06, "loss": 0.3655, "step": 26190 }, { "epoch": 162.6791277258567, "grad_norm": 2.15625, "learning_rate": 4.258367779501371e-06, "loss": 0.1742, "step": 26191 }, { "epoch": 162.68535825545172, "grad_norm": 0.8046875, "learning_rate": 4.256992457428416e-06, "loss": 0.0915, "step": 26192 }, { "epoch": 162.69158878504672, "grad_norm": 1.8515625, "learning_rate": 4.25561733681569e-06, "loss": 0.184, "step": 26193 }, { "epoch": 162.69781931464175, "grad_norm": 1.71875, "learning_rate": 4.254242417676557e-06, "loss": 0.2575, "step": 26194 }, { "epoch": 162.70404984423675, "grad_norm": 2.234375, "learning_rate": 4.252867700024374e-06, "loss": 0.231, "step": 26195 }, { "epoch": 162.71028037383178, "grad_norm": 1.9375, "learning_rate": 4.251493183872487e-06, "loss": 0.1681, "step": 26196 }, { "epoch": 162.71651090342678, "grad_norm": 1.3203125, "learning_rate": 4.25011886923424e-06, "loss": 0.1261, "step": 26197 }, { "epoch": 162.7227414330218, "grad_norm": 2.125, "learning_rate": 4.248744756122986e-06, "loss": 0.1887, "step": 26198 }, { "epoch": 162.7289719626168, "grad_norm": 2.96875, "learning_rate": 4.247370844552076e-06, "loss": 0.2146, "step": 26199 }, { "epoch": 162.73520249221184, "grad_norm": 1.0078125, "learning_rate": 4.245997134534846e-06, "loss": 0.1033, "step": 26200 }, { "epoch": 162.74143302180684, "grad_norm": 1.65625, "learning_rate": 4.244623626084637e-06, "loss": 0.102, "step": 26201 }, { "epoch": 162.74766355140187, "grad_norm": 2.859375, "learning_rate": 4.2432503192148e-06, "loss": 0.1108, "step": 26202 }, { "epoch": 162.75389408099687, "grad_norm": 0.734375, "learning_rate": 4.241877213938666e-06, "loss": 0.0716, "step": 26203 }, { "epoch": 162.7601246105919, "grad_norm": 1.578125, "learning_rate": 4.240504310269561e-06, "loss": 0.111, "step": 26204 }, { "epoch": 162.7663551401869, "grad_norm": 1.359375, "learning_rate": 4.239131608220831e-06, "loss": 0.1149, "step": 26205 }, { "epoch": 162.77258566978193, "grad_norm": 1.4921875, "learning_rate": 4.237759107805802e-06, "loss": 0.1356, "step": 26206 }, { "epoch": 162.77881619937693, "grad_norm": 1.390625, "learning_rate": 4.236386809037815e-06, "loss": 0.1212, "step": 26207 }, { "epoch": 162.78504672897196, "grad_norm": 2.75, "learning_rate": 4.235014711930188e-06, "loss": 0.1855, "step": 26208 }, { "epoch": 162.79127725856696, "grad_norm": 2.0625, "learning_rate": 4.233642816496244e-06, "loss": 0.1894, "step": 26209 }, { "epoch": 162.797507788162, "grad_norm": 1.09375, "learning_rate": 4.232271122749313e-06, "loss": 0.1053, "step": 26210 }, { "epoch": 162.80373831775702, "grad_norm": 1.171875, "learning_rate": 4.230899630702722e-06, "loss": 0.1013, "step": 26211 }, { "epoch": 162.80996884735202, "grad_norm": 1.2421875, "learning_rate": 4.229528340369779e-06, "loss": 0.1323, "step": 26212 }, { "epoch": 162.81619937694705, "grad_norm": 1.8046875, "learning_rate": 4.228157251763809e-06, "loss": 0.1377, "step": 26213 }, { "epoch": 162.82242990654206, "grad_norm": 1.515625, "learning_rate": 4.226786364898133e-06, "loss": 0.1206, "step": 26214 }, { "epoch": 162.82866043613708, "grad_norm": 1.453125, "learning_rate": 4.22541567978606e-06, "loss": 0.1498, "step": 26215 }, { "epoch": 162.8348909657321, "grad_norm": 1.515625, "learning_rate": 4.224045196440898e-06, "loss": 0.1374, "step": 26216 }, { "epoch": 162.84112149532712, "grad_norm": 1.8359375, "learning_rate": 4.222674914875962e-06, "loss": 0.1812, "step": 26217 }, { "epoch": 162.84735202492212, "grad_norm": 1.15625, "learning_rate": 4.221304835104567e-06, "loss": 0.1272, "step": 26218 }, { "epoch": 162.85358255451715, "grad_norm": 0.8203125, "learning_rate": 4.219934957140007e-06, "loss": 0.1133, "step": 26219 }, { "epoch": 162.85981308411215, "grad_norm": 1.6015625, "learning_rate": 4.21856528099559e-06, "loss": 0.1006, "step": 26220 }, { "epoch": 162.86604361370718, "grad_norm": 1.46875, "learning_rate": 4.217195806684629e-06, "loss": 0.1262, "step": 26221 }, { "epoch": 162.87227414330218, "grad_norm": 1.8359375, "learning_rate": 4.215826534220416e-06, "loss": 0.1133, "step": 26222 }, { "epoch": 162.8785046728972, "grad_norm": 1.84375, "learning_rate": 4.214457463616245e-06, "loss": 0.1632, "step": 26223 }, { "epoch": 162.8847352024922, "grad_norm": 2.03125, "learning_rate": 4.21308859488542e-06, "loss": 0.1735, "step": 26224 }, { "epoch": 162.89096573208724, "grad_norm": 2.28125, "learning_rate": 4.211719928041235e-06, "loss": 0.2154, "step": 26225 }, { "epoch": 162.89719626168224, "grad_norm": 2.875, "learning_rate": 4.210351463096979e-06, "loss": 0.1784, "step": 26226 }, { "epoch": 162.90342679127727, "grad_norm": 1.875, "learning_rate": 4.2089832000659465e-06, "loss": 0.1605, "step": 26227 }, { "epoch": 162.90965732087227, "grad_norm": 1.515625, "learning_rate": 4.20761513896143e-06, "loss": 0.1362, "step": 26228 }, { "epoch": 162.9158878504673, "grad_norm": 1.1640625, "learning_rate": 4.2062472797967104e-06, "loss": 0.1022, "step": 26229 }, { "epoch": 162.9221183800623, "grad_norm": 2.5, "learning_rate": 4.204879622585067e-06, "loss": 0.1678, "step": 26230 }, { "epoch": 162.92834890965733, "grad_norm": 1.765625, "learning_rate": 4.20351216733979e-06, "loss": 0.1314, "step": 26231 }, { "epoch": 162.93457943925233, "grad_norm": 2.40625, "learning_rate": 4.202144914074169e-06, "loss": 0.1604, "step": 26232 }, { "epoch": 162.94080996884736, "grad_norm": 1.9140625, "learning_rate": 4.200777862801467e-06, "loss": 0.1315, "step": 26233 }, { "epoch": 162.94704049844236, "grad_norm": 1.7734375, "learning_rate": 4.199411013534971e-06, "loss": 0.1624, "step": 26234 }, { "epoch": 162.9532710280374, "grad_norm": 2.0625, "learning_rate": 4.1980443662879485e-06, "loss": 0.1718, "step": 26235 }, { "epoch": 162.9595015576324, "grad_norm": 2.234375, "learning_rate": 4.196677921073686e-06, "loss": 0.1703, "step": 26236 }, { "epoch": 162.96573208722742, "grad_norm": 2.53125, "learning_rate": 4.195311677905436e-06, "loss": 0.2086, "step": 26237 }, { "epoch": 162.97196261682242, "grad_norm": 1.59375, "learning_rate": 4.193945636796482e-06, "loss": 0.1154, "step": 26238 }, { "epoch": 162.97819314641745, "grad_norm": 1.5546875, "learning_rate": 4.192579797760085e-06, "loss": 0.108, "step": 26239 }, { "epoch": 162.98442367601245, "grad_norm": 2.34375, "learning_rate": 4.191214160809523e-06, "loss": 0.1971, "step": 26240 }, { "epoch": 162.99065420560748, "grad_norm": 2.609375, "learning_rate": 4.189848725958037e-06, "loss": 0.2036, "step": 26241 }, { "epoch": 162.99688473520249, "grad_norm": 1.4609375, "learning_rate": 4.1884834932189e-06, "loss": 0.1082, "step": 26242 }, { "epoch": 163.0, "grad_norm": 1.4140625, "learning_rate": 4.187118462605375e-06, "loss": 0.1467, "step": 26243 }, { "epoch": 163.00623052959503, "grad_norm": 3.28125, "learning_rate": 4.185753634130718e-06, "loss": 0.3301, "step": 26244 }, { "epoch": 163.01246105919003, "grad_norm": 2.703125, "learning_rate": 4.1843890078081784e-06, "loss": 0.1736, "step": 26245 }, { "epoch": 163.01869158878506, "grad_norm": 1.671875, "learning_rate": 4.183024583651013e-06, "loss": 0.1249, "step": 26246 }, { "epoch": 163.02492211838006, "grad_norm": 1.0859375, "learning_rate": 4.18166036167248e-06, "loss": 0.1035, "step": 26247 }, { "epoch": 163.0311526479751, "grad_norm": 1.3984375, "learning_rate": 4.180296341885823e-06, "loss": 0.0916, "step": 26248 }, { "epoch": 163.0373831775701, "grad_norm": 1.8828125, "learning_rate": 4.178932524304283e-06, "loss": 0.119, "step": 26249 }, { "epoch": 163.04361370716512, "grad_norm": 2.703125, "learning_rate": 4.177568908941113e-06, "loss": 0.2061, "step": 26250 }, { "epoch": 163.04984423676012, "grad_norm": 3.28125, "learning_rate": 4.176205495809563e-06, "loss": 0.241, "step": 26251 }, { "epoch": 163.05607476635515, "grad_norm": 2.375, "learning_rate": 4.174842284922861e-06, "loss": 0.2359, "step": 26252 }, { "epoch": 163.06230529595015, "grad_norm": 2.203125, "learning_rate": 4.173479276294259e-06, "loss": 0.1616, "step": 26253 }, { "epoch": 163.06853582554518, "grad_norm": 0.8046875, "learning_rate": 4.172116469936985e-06, "loss": 0.0907, "step": 26254 }, { "epoch": 163.07476635514018, "grad_norm": 1.828125, "learning_rate": 4.170753865864285e-06, "loss": 0.1478, "step": 26255 }, { "epoch": 163.0809968847352, "grad_norm": 1.4765625, "learning_rate": 4.169391464089379e-06, "loss": 0.117, "step": 26256 }, { "epoch": 163.08722741433021, "grad_norm": 1.671875, "learning_rate": 4.1680292646255084e-06, "loss": 0.1189, "step": 26257 }, { "epoch": 163.09345794392524, "grad_norm": 1.4375, "learning_rate": 4.1666672674859075e-06, "loss": 0.1461, "step": 26258 }, { "epoch": 163.09968847352025, "grad_norm": 1.1796875, "learning_rate": 4.1653054726838005e-06, "loss": 0.1239, "step": 26259 }, { "epoch": 163.10591900311528, "grad_norm": 2.734375, "learning_rate": 4.163943880232405e-06, "loss": 0.2569, "step": 26260 }, { "epoch": 163.11214953271028, "grad_norm": 1.578125, "learning_rate": 4.162582490144948e-06, "loss": 0.1646, "step": 26261 }, { "epoch": 163.1183800623053, "grad_norm": 1.71875, "learning_rate": 4.161221302434665e-06, "loss": 0.106, "step": 26262 }, { "epoch": 163.1246105919003, "grad_norm": 2.96875, "learning_rate": 4.159860317114758e-06, "loss": 0.1874, "step": 26263 }, { "epoch": 163.13084112149534, "grad_norm": 0.98046875, "learning_rate": 4.158499534198457e-06, "loss": 0.105, "step": 26264 }, { "epoch": 163.13707165109034, "grad_norm": 3.015625, "learning_rate": 4.157138953698972e-06, "loss": 0.253, "step": 26265 }, { "epoch": 163.14330218068537, "grad_norm": 2.203125, "learning_rate": 4.155778575629532e-06, "loss": 0.1955, "step": 26266 }, { "epoch": 163.14953271028037, "grad_norm": 0.79296875, "learning_rate": 4.154418400003324e-06, "loss": 0.1061, "step": 26267 }, { "epoch": 163.1557632398754, "grad_norm": 3.328125, "learning_rate": 4.153058426833573e-06, "loss": 0.305, "step": 26268 }, { "epoch": 163.1619937694704, "grad_norm": 1.9453125, "learning_rate": 4.1516986561334845e-06, "loss": 0.1597, "step": 26269 }, { "epoch": 163.16822429906543, "grad_norm": 1.7734375, "learning_rate": 4.150339087916274e-06, "loss": 0.1268, "step": 26270 }, { "epoch": 163.17445482866043, "grad_norm": 2.890625, "learning_rate": 4.148979722195131e-06, "loss": 0.2559, "step": 26271 }, { "epoch": 163.18068535825546, "grad_norm": 2.609375, "learning_rate": 4.14762055898327e-06, "loss": 0.1964, "step": 26272 }, { "epoch": 163.18691588785046, "grad_norm": 1.1015625, "learning_rate": 4.1462615982938826e-06, "loss": 0.1071, "step": 26273 }, { "epoch": 163.1931464174455, "grad_norm": 2.3125, "learning_rate": 4.144902840140177e-06, "loss": 0.2198, "step": 26274 }, { "epoch": 163.1993769470405, "grad_norm": 2.671875, "learning_rate": 4.143544284535336e-06, "loss": 0.1232, "step": 26275 }, { "epoch": 163.20560747663552, "grad_norm": 1.6171875, "learning_rate": 4.142185931492562e-06, "loss": 0.0969, "step": 26276 }, { "epoch": 163.21183800623052, "grad_norm": 1.21875, "learning_rate": 4.140827781025055e-06, "loss": 0.1078, "step": 26277 }, { "epoch": 163.21806853582555, "grad_norm": 3.09375, "learning_rate": 4.139469833145998e-06, "loss": 0.2818, "step": 26278 }, { "epoch": 163.22429906542055, "grad_norm": 1.9453125, "learning_rate": 4.138112087868576e-06, "loss": 0.1341, "step": 26279 }, { "epoch": 163.23052959501558, "grad_norm": 1.9921875, "learning_rate": 4.136754545205976e-06, "loss": 0.1858, "step": 26280 }, { "epoch": 163.23676012461058, "grad_norm": 2.734375, "learning_rate": 4.135397205171393e-06, "loss": 0.228, "step": 26281 }, { "epoch": 163.2429906542056, "grad_norm": 2.328125, "learning_rate": 4.1340400677779975e-06, "loss": 0.1616, "step": 26282 }, { "epoch": 163.2492211838006, "grad_norm": 0.84765625, "learning_rate": 4.132683133038978e-06, "loss": 0.0929, "step": 26283 }, { "epoch": 163.25545171339564, "grad_norm": 2.40625, "learning_rate": 4.131326400967514e-06, "loss": 0.2101, "step": 26284 }, { "epoch": 163.26168224299064, "grad_norm": 1.8828125, "learning_rate": 4.12996987157678e-06, "loss": 0.1579, "step": 26285 }, { "epoch": 163.26791277258567, "grad_norm": 1.0546875, "learning_rate": 4.1286135448799475e-06, "loss": 0.1094, "step": 26286 }, { "epoch": 163.27414330218068, "grad_norm": 3.125, "learning_rate": 4.1272574208901905e-06, "loss": 0.1828, "step": 26287 }, { "epoch": 163.2803738317757, "grad_norm": 2.265625, "learning_rate": 4.125901499620688e-06, "loss": 0.182, "step": 26288 }, { "epoch": 163.2866043613707, "grad_norm": 2.234375, "learning_rate": 4.124545781084596e-06, "loss": 0.263, "step": 26289 }, { "epoch": 163.29283489096574, "grad_norm": 2.84375, "learning_rate": 4.123190265295088e-06, "loss": 0.2368, "step": 26290 }, { "epoch": 163.29906542056074, "grad_norm": 2.109375, "learning_rate": 4.121834952265335e-06, "loss": 0.1318, "step": 26291 }, { "epoch": 163.30529595015577, "grad_norm": 1.6796875, "learning_rate": 4.120479842008496e-06, "loss": 0.1577, "step": 26292 }, { "epoch": 163.31152647975077, "grad_norm": 1.5625, "learning_rate": 4.119124934537725e-06, "loss": 0.1476, "step": 26293 }, { "epoch": 163.3177570093458, "grad_norm": 1.2734375, "learning_rate": 4.117770229866186e-06, "loss": 0.1016, "step": 26294 }, { "epoch": 163.3239875389408, "grad_norm": 2.4375, "learning_rate": 4.116415728007042e-06, "loss": 0.1429, "step": 26295 }, { "epoch": 163.33021806853583, "grad_norm": 2.421875, "learning_rate": 4.115061428973438e-06, "loss": 0.1837, "step": 26296 }, { "epoch": 163.33644859813083, "grad_norm": 1.453125, "learning_rate": 4.11370733277853e-06, "loss": 0.146, "step": 26297 }, { "epoch": 163.34267912772586, "grad_norm": 1.1640625, "learning_rate": 4.112353439435479e-06, "loss": 0.0882, "step": 26298 }, { "epoch": 163.34890965732086, "grad_norm": 1.9140625, "learning_rate": 4.110999748957428e-06, "loss": 0.1543, "step": 26299 }, { "epoch": 163.3551401869159, "grad_norm": 2.109375, "learning_rate": 4.1096462613575135e-06, "loss": 0.1181, "step": 26300 }, { "epoch": 163.3613707165109, "grad_norm": 2.5625, "learning_rate": 4.108292976648895e-06, "loss": 0.1892, "step": 26301 }, { "epoch": 163.36760124610592, "grad_norm": 0.74609375, "learning_rate": 4.106939894844708e-06, "loss": 0.0758, "step": 26302 }, { "epoch": 163.37383177570092, "grad_norm": 1.390625, "learning_rate": 4.105587015958104e-06, "loss": 0.1015, "step": 26303 }, { "epoch": 163.38006230529595, "grad_norm": 2.6875, "learning_rate": 4.104234340002217e-06, "loss": 0.2324, "step": 26304 }, { "epoch": 163.38629283489095, "grad_norm": 1.859375, "learning_rate": 4.102881866990174e-06, "loss": 0.1335, "step": 26305 }, { "epoch": 163.39252336448598, "grad_norm": 1.46875, "learning_rate": 4.101529596935122e-06, "loss": 0.1186, "step": 26306 }, { "epoch": 163.39875389408098, "grad_norm": 1.84375, "learning_rate": 4.100177529850199e-06, "loss": 0.1647, "step": 26307 }, { "epoch": 163.404984423676, "grad_norm": 1.4140625, "learning_rate": 4.098825665748521e-06, "loss": 0.1313, "step": 26308 }, { "epoch": 163.41121495327104, "grad_norm": 1.7265625, "learning_rate": 4.097474004643229e-06, "loss": 0.1558, "step": 26309 }, { "epoch": 163.41744548286604, "grad_norm": 2.921875, "learning_rate": 4.096122546547451e-06, "loss": 0.3141, "step": 26310 }, { "epoch": 163.42367601246107, "grad_norm": 1.671875, "learning_rate": 4.094771291474312e-06, "loss": 0.1147, "step": 26311 }, { "epoch": 163.42990654205607, "grad_norm": 0.734375, "learning_rate": 4.093420239436924e-06, "loss": 0.0917, "step": 26312 }, { "epoch": 163.4361370716511, "grad_norm": 2.765625, "learning_rate": 4.0920693904484196e-06, "loss": 0.1813, "step": 26313 }, { "epoch": 163.4423676012461, "grad_norm": 1.921875, "learning_rate": 4.090718744521924e-06, "loss": 0.1539, "step": 26314 }, { "epoch": 163.44859813084113, "grad_norm": 1.796875, "learning_rate": 4.089368301670538e-06, "loss": 0.1566, "step": 26315 }, { "epoch": 163.45482866043614, "grad_norm": 1.4296875, "learning_rate": 4.088018061907389e-06, "loss": 0.1142, "step": 26316 }, { "epoch": 163.46105919003116, "grad_norm": 3.890625, "learning_rate": 4.086668025245596e-06, "loss": 0.2838, "step": 26317 }, { "epoch": 163.46728971962617, "grad_norm": 1.6328125, "learning_rate": 4.085318191698262e-06, "loss": 0.1416, "step": 26318 }, { "epoch": 163.4735202492212, "grad_norm": 2.71875, "learning_rate": 4.083968561278495e-06, "loss": 0.2481, "step": 26319 }, { "epoch": 163.4797507788162, "grad_norm": 2.03125, "learning_rate": 4.082619133999402e-06, "loss": 0.1465, "step": 26320 }, { "epoch": 163.48598130841123, "grad_norm": 2.1875, "learning_rate": 4.081269909874103e-06, "loss": 0.2248, "step": 26321 }, { "epoch": 163.49221183800623, "grad_norm": 1.2578125, "learning_rate": 4.079920888915683e-06, "loss": 0.1386, "step": 26322 }, { "epoch": 163.49844236760126, "grad_norm": 2.53125, "learning_rate": 4.078572071137263e-06, "loss": 0.3082, "step": 26323 }, { "epoch": 163.50467289719626, "grad_norm": 2.515625, "learning_rate": 4.077223456551926e-06, "loss": 0.2766, "step": 26324 }, { "epoch": 163.5109034267913, "grad_norm": 1.1875, "learning_rate": 4.0758750451727806e-06, "loss": 0.1144, "step": 26325 }, { "epoch": 163.5171339563863, "grad_norm": 2.34375, "learning_rate": 4.074526837012915e-06, "loss": 0.2009, "step": 26326 }, { "epoch": 163.52336448598132, "grad_norm": 1.046875, "learning_rate": 4.073178832085428e-06, "loss": 0.1177, "step": 26327 }, { "epoch": 163.52959501557632, "grad_norm": 3.15625, "learning_rate": 4.07183103040342e-06, "loss": 0.14, "step": 26328 }, { "epoch": 163.53582554517135, "grad_norm": 0.8671875, "learning_rate": 4.070483431979969e-06, "loss": 0.0923, "step": 26329 }, { "epoch": 163.54205607476635, "grad_norm": 2.5, "learning_rate": 4.0691360368281615e-06, "loss": 0.2245, "step": 26330 }, { "epoch": 163.54828660436138, "grad_norm": 2.28125, "learning_rate": 4.067788844961088e-06, "loss": 0.2293, "step": 26331 }, { "epoch": 163.55451713395638, "grad_norm": 1.9140625, "learning_rate": 4.066441856391839e-06, "loss": 0.19, "step": 26332 }, { "epoch": 163.5607476635514, "grad_norm": 1.2578125, "learning_rate": 4.065095071133487e-06, "loss": 0.1002, "step": 26333 }, { "epoch": 163.5669781931464, "grad_norm": 3.15625, "learning_rate": 4.063748489199118e-06, "loss": 0.281, "step": 26334 }, { "epoch": 163.57320872274144, "grad_norm": 1.5234375, "learning_rate": 4.062402110601804e-06, "loss": 0.1345, "step": 26335 }, { "epoch": 163.57943925233644, "grad_norm": 1.3671875, "learning_rate": 4.061055935354643e-06, "loss": 0.1203, "step": 26336 }, { "epoch": 163.58566978193147, "grad_norm": 1.6796875, "learning_rate": 4.059709963470676e-06, "loss": 0.1061, "step": 26337 }, { "epoch": 163.59190031152647, "grad_norm": 2.375, "learning_rate": 4.058364194962996e-06, "loss": 0.1976, "step": 26338 }, { "epoch": 163.5981308411215, "grad_norm": 1.734375, "learning_rate": 4.057018629844666e-06, "loss": 0.1501, "step": 26339 }, { "epoch": 163.6043613707165, "grad_norm": 2.46875, "learning_rate": 4.055673268128765e-06, "loss": 0.1501, "step": 26340 }, { "epoch": 163.61059190031153, "grad_norm": 1.875, "learning_rate": 4.054328109828345e-06, "loss": 0.1806, "step": 26341 }, { "epoch": 163.61682242990653, "grad_norm": 1.1484375, "learning_rate": 4.052983154956483e-06, "loss": 0.1094, "step": 26342 }, { "epoch": 163.62305295950156, "grad_norm": 1.2265625, "learning_rate": 4.051638403526231e-06, "loss": 0.0953, "step": 26343 }, { "epoch": 163.62928348909657, "grad_norm": 1.6171875, "learning_rate": 4.050293855550661e-06, "loss": 0.1691, "step": 26344 }, { "epoch": 163.6355140186916, "grad_norm": 1.890625, "learning_rate": 4.048949511042816e-06, "loss": 0.1153, "step": 26345 }, { "epoch": 163.6417445482866, "grad_norm": 2.578125, "learning_rate": 4.047605370015764e-06, "loss": 0.2439, "step": 26346 }, { "epoch": 163.64797507788163, "grad_norm": 2.78125, "learning_rate": 4.046261432482562e-06, "loss": 0.1532, "step": 26347 }, { "epoch": 163.65420560747663, "grad_norm": 2.65625, "learning_rate": 4.044917698456249e-06, "loss": 0.2685, "step": 26348 }, { "epoch": 163.66043613707166, "grad_norm": 1.6953125, "learning_rate": 4.043574167949893e-06, "loss": 0.1039, "step": 26349 }, { "epoch": 163.66666666666666, "grad_norm": 2.28125, "learning_rate": 4.04223084097653e-06, "loss": 0.2497, "step": 26350 }, { "epoch": 163.6728971962617, "grad_norm": 0.83203125, "learning_rate": 4.0408877175492136e-06, "loss": 0.0811, "step": 26351 }, { "epoch": 163.6791277258567, "grad_norm": 2.5625, "learning_rate": 4.039544797680978e-06, "loss": 0.1813, "step": 26352 }, { "epoch": 163.68535825545172, "grad_norm": 1.765625, "learning_rate": 4.038202081384876e-06, "loss": 0.1411, "step": 26353 }, { "epoch": 163.69158878504672, "grad_norm": 1.1484375, "learning_rate": 4.03685956867395e-06, "loss": 0.1288, "step": 26354 }, { "epoch": 163.69781931464175, "grad_norm": 2.359375, "learning_rate": 4.035517259561234e-06, "loss": 0.185, "step": 26355 }, { "epoch": 163.70404984423675, "grad_norm": 1.2109375, "learning_rate": 4.0341751540597604e-06, "loss": 0.1017, "step": 26356 }, { "epoch": 163.71028037383178, "grad_norm": 2.953125, "learning_rate": 4.032833252182569e-06, "loss": 0.3355, "step": 26357 }, { "epoch": 163.71651090342678, "grad_norm": 1.015625, "learning_rate": 4.031491553942696e-06, "loss": 0.1055, "step": 26358 }, { "epoch": 163.7227414330218, "grad_norm": 1.2421875, "learning_rate": 4.030150059353163e-06, "loss": 0.1175, "step": 26359 }, { "epoch": 163.7289719626168, "grad_norm": 2.34375, "learning_rate": 4.028808768427006e-06, "loss": 0.2032, "step": 26360 }, { "epoch": 163.73520249221184, "grad_norm": 2.859375, "learning_rate": 4.027467681177255e-06, "loss": 0.1696, "step": 26361 }, { "epoch": 163.74143302180684, "grad_norm": 1.7109375, "learning_rate": 4.026126797616933e-06, "loss": 0.1308, "step": 26362 }, { "epoch": 163.74766355140187, "grad_norm": 1.53125, "learning_rate": 4.024786117759052e-06, "loss": 0.1461, "step": 26363 }, { "epoch": 163.75389408099687, "grad_norm": 2.359375, "learning_rate": 4.023445641616641e-06, "loss": 0.1672, "step": 26364 }, { "epoch": 163.7601246105919, "grad_norm": 1.4140625, "learning_rate": 4.022105369202719e-06, "loss": 0.108, "step": 26365 }, { "epoch": 163.7663551401869, "grad_norm": 2.15625, "learning_rate": 4.0207653005303074e-06, "loss": 0.1807, "step": 26366 }, { "epoch": 163.77258566978193, "grad_norm": 2.171875, "learning_rate": 4.019425435612414e-06, "loss": 0.1748, "step": 26367 }, { "epoch": 163.77881619937693, "grad_norm": 2.3125, "learning_rate": 4.018085774462061e-06, "loss": 0.2273, "step": 26368 }, { "epoch": 163.78504672897196, "grad_norm": 1.6015625, "learning_rate": 4.016746317092243e-06, "loss": 0.1368, "step": 26369 }, { "epoch": 163.79127725856696, "grad_norm": 2.5, "learning_rate": 4.015407063515988e-06, "loss": 0.21, "step": 26370 }, { "epoch": 163.797507788162, "grad_norm": 2.03125, "learning_rate": 4.014068013746289e-06, "loss": 0.1619, "step": 26371 }, { "epoch": 163.80373831775702, "grad_norm": 1.6640625, "learning_rate": 4.012729167796156e-06, "loss": 0.1506, "step": 26372 }, { "epoch": 163.80996884735202, "grad_norm": 1.8125, "learning_rate": 4.011390525678596e-06, "loss": 0.182, "step": 26373 }, { "epoch": 163.81619937694705, "grad_norm": 1.0, "learning_rate": 4.01005208740661e-06, "loss": 0.0999, "step": 26374 }, { "epoch": 163.82242990654206, "grad_norm": 1.140625, "learning_rate": 4.008713852993187e-06, "loss": 0.1025, "step": 26375 }, { "epoch": 163.82866043613708, "grad_norm": 3.25, "learning_rate": 4.007375822451331e-06, "loss": 0.2158, "step": 26376 }, { "epoch": 163.8348909657321, "grad_norm": 1.203125, "learning_rate": 4.006037995794043e-06, "loss": 0.127, "step": 26377 }, { "epoch": 163.84112149532712, "grad_norm": 2.203125, "learning_rate": 4.004700373034306e-06, "loss": 0.1691, "step": 26378 }, { "epoch": 163.84735202492212, "grad_norm": 1.25, "learning_rate": 4.003362954185114e-06, "loss": 0.1295, "step": 26379 }, { "epoch": 163.85358255451715, "grad_norm": 2.21875, "learning_rate": 4.002025739259466e-06, "loss": 0.1874, "step": 26380 }, { "epoch": 163.85981308411215, "grad_norm": 0.91796875, "learning_rate": 4.0006887282703385e-06, "loss": 0.1003, "step": 26381 }, { "epoch": 163.86604361370718, "grad_norm": 3.0625, "learning_rate": 3.9993519212307154e-06, "loss": 0.2364, "step": 26382 }, { "epoch": 163.87227414330218, "grad_norm": 1.59375, "learning_rate": 3.998015318153586e-06, "loss": 0.0908, "step": 26383 }, { "epoch": 163.8785046728972, "grad_norm": 1.40625, "learning_rate": 3.996678919051933e-06, "loss": 0.1274, "step": 26384 }, { "epoch": 163.8847352024922, "grad_norm": 1.4453125, "learning_rate": 3.99534272393873e-06, "loss": 0.1086, "step": 26385 }, { "epoch": 163.89096573208724, "grad_norm": 2.28125, "learning_rate": 3.9940067328269545e-06, "loss": 0.1429, "step": 26386 }, { "epoch": 163.89719626168224, "grad_norm": 2.703125, "learning_rate": 3.992670945729593e-06, "loss": 0.1791, "step": 26387 }, { "epoch": 163.90342679127727, "grad_norm": 1.15625, "learning_rate": 3.9913353626596095e-06, "loss": 0.1095, "step": 26388 }, { "epoch": 163.90965732087227, "grad_norm": 1.7109375, "learning_rate": 3.989999983629969e-06, "loss": 0.1079, "step": 26389 }, { "epoch": 163.9158878504673, "grad_norm": 1.984375, "learning_rate": 3.98866480865365e-06, "loss": 0.1391, "step": 26390 }, { "epoch": 163.9221183800623, "grad_norm": 1.359375, "learning_rate": 3.987329837743625e-06, "loss": 0.1268, "step": 26391 }, { "epoch": 163.92834890965733, "grad_norm": 1.3828125, "learning_rate": 3.9859950709128454e-06, "loss": 0.1202, "step": 26392 }, { "epoch": 163.93457943925233, "grad_norm": 2.65625, "learning_rate": 3.98466050817429e-06, "loss": 0.1288, "step": 26393 }, { "epoch": 163.94080996884736, "grad_norm": 1.375, "learning_rate": 3.9833261495409054e-06, "loss": 0.113, "step": 26394 }, { "epoch": 163.94704049844236, "grad_norm": 1.9765625, "learning_rate": 3.9819919950256634e-06, "loss": 0.1662, "step": 26395 }, { "epoch": 163.9532710280374, "grad_norm": 2.34375, "learning_rate": 3.980658044641514e-06, "loss": 0.1204, "step": 26396 }, { "epoch": 163.9595015576324, "grad_norm": 2.078125, "learning_rate": 3.979324298401413e-06, "loss": 0.1394, "step": 26397 }, { "epoch": 163.96573208722742, "grad_norm": 1.5703125, "learning_rate": 3.977990756318315e-06, "loss": 0.119, "step": 26398 }, { "epoch": 163.97196261682242, "grad_norm": 1.8671875, "learning_rate": 3.976657418405188e-06, "loss": 0.2333, "step": 26399 }, { "epoch": 163.97819314641745, "grad_norm": 1.25, "learning_rate": 3.9753242846749526e-06, "loss": 0.106, "step": 26400 }, { "epoch": 163.98442367601245, "grad_norm": 1.265625, "learning_rate": 3.9739913551405715e-06, "loss": 0.1013, "step": 26401 }, { "epoch": 163.99065420560748, "grad_norm": 1.4609375, "learning_rate": 3.972658629814988e-06, "loss": 0.1321, "step": 26402 }, { "epoch": 163.99688473520249, "grad_norm": 1.4921875, "learning_rate": 3.971326108711154e-06, "loss": 0.1503, "step": 26403 }, { "epoch": 164.0, "grad_norm": 2.421875, "learning_rate": 3.969993791842e-06, "loss": 0.2041, "step": 26404 }, { "epoch": 164.00623052959503, "grad_norm": 2.546875, "learning_rate": 3.968661679220468e-06, "loss": 0.2264, "step": 26405 }, { "epoch": 164.01246105919003, "grad_norm": 1.8984375, "learning_rate": 3.967329770859504e-06, "loss": 0.1205, "step": 26406 }, { "epoch": 164.01869158878506, "grad_norm": 1.8359375, "learning_rate": 3.965998066772039e-06, "loss": 0.1426, "step": 26407 }, { "epoch": 164.02492211838006, "grad_norm": 2.296875, "learning_rate": 3.964666566970998e-06, "loss": 0.2248, "step": 26408 }, { "epoch": 164.0311526479751, "grad_norm": 2.765625, "learning_rate": 3.963335271469321e-06, "loss": 0.2349, "step": 26409 }, { "epoch": 164.0373831775701, "grad_norm": 2.359375, "learning_rate": 3.962004180279943e-06, "loss": 0.1361, "step": 26410 }, { "epoch": 164.04361370716512, "grad_norm": 1.5, "learning_rate": 3.9606732934157784e-06, "loss": 0.1152, "step": 26411 }, { "epoch": 164.04984423676012, "grad_norm": 2.84375, "learning_rate": 3.959342610889769e-06, "loss": 0.2215, "step": 26412 }, { "epoch": 164.05607476635515, "grad_norm": 1.1171875, "learning_rate": 3.9580121327148226e-06, "loss": 0.1004, "step": 26413 }, { "epoch": 164.06230529595015, "grad_norm": 2.125, "learning_rate": 3.956681858903874e-06, "loss": 0.188, "step": 26414 }, { "epoch": 164.06853582554518, "grad_norm": 3.234375, "learning_rate": 3.955351789469833e-06, "loss": 0.249, "step": 26415 }, { "epoch": 164.07476635514018, "grad_norm": 2.421875, "learning_rate": 3.9540219244256214e-06, "loss": 0.2856, "step": 26416 }, { "epoch": 164.0809968847352, "grad_norm": 1.9453125, "learning_rate": 3.952692263784161e-06, "loss": 0.1579, "step": 26417 }, { "epoch": 164.08722741433021, "grad_norm": 1.5625, "learning_rate": 3.951362807558357e-06, "loss": 0.1275, "step": 26418 }, { "epoch": 164.09345794392524, "grad_norm": 1.890625, "learning_rate": 3.950033555761129e-06, "loss": 0.1555, "step": 26419 }, { "epoch": 164.09968847352025, "grad_norm": 1.953125, "learning_rate": 3.948704508405376e-06, "loss": 0.1652, "step": 26420 }, { "epoch": 164.10591900311528, "grad_norm": 1.6015625, "learning_rate": 3.947375665504022e-06, "loss": 0.11, "step": 26421 }, { "epoch": 164.11214953271028, "grad_norm": 2.546875, "learning_rate": 3.946047027069954e-06, "loss": 0.2151, "step": 26422 }, { "epoch": 164.1183800623053, "grad_norm": 2.734375, "learning_rate": 3.94471859311609e-06, "loss": 0.1507, "step": 26423 }, { "epoch": 164.1246105919003, "grad_norm": 1.5078125, "learning_rate": 3.94339036365533e-06, "loss": 0.1354, "step": 26424 }, { "epoch": 164.13084112149534, "grad_norm": 1.0859375, "learning_rate": 3.942062338700575e-06, "loss": 0.1138, "step": 26425 }, { "epoch": 164.13707165109034, "grad_norm": 1.0703125, "learning_rate": 3.940734518264713e-06, "loss": 0.1161, "step": 26426 }, { "epoch": 164.14330218068537, "grad_norm": 1.9765625, "learning_rate": 3.939406902360645e-06, "loss": 0.1426, "step": 26427 }, { "epoch": 164.14953271028037, "grad_norm": 1.7890625, "learning_rate": 3.9380794910012755e-06, "loss": 0.1481, "step": 26428 }, { "epoch": 164.1557632398754, "grad_norm": 3.21875, "learning_rate": 3.9367522841994825e-06, "loss": 0.2263, "step": 26429 }, { "epoch": 164.1619937694704, "grad_norm": 3.09375, "learning_rate": 3.9354252819681605e-06, "loss": 0.3148, "step": 26430 }, { "epoch": 164.16822429906543, "grad_norm": 2.59375, "learning_rate": 3.934098484320206e-06, "loss": 0.2292, "step": 26431 }, { "epoch": 164.17445482866043, "grad_norm": 1.7578125, "learning_rate": 3.932771891268497e-06, "loss": 0.1468, "step": 26432 }, { "epoch": 164.18068535825546, "grad_norm": 2.09375, "learning_rate": 3.931445502825912e-06, "loss": 0.147, "step": 26433 }, { "epoch": 164.18691588785046, "grad_norm": 2.15625, "learning_rate": 3.930119319005341e-06, "loss": 0.1548, "step": 26434 }, { "epoch": 164.1931464174455, "grad_norm": 1.328125, "learning_rate": 3.928793339819661e-06, "loss": 0.1148, "step": 26435 }, { "epoch": 164.1993769470405, "grad_norm": 1.953125, "learning_rate": 3.927467565281761e-06, "loss": 0.1918, "step": 26436 }, { "epoch": 164.20560747663552, "grad_norm": 1.6953125, "learning_rate": 3.9261419954045e-06, "loss": 0.1301, "step": 26437 }, { "epoch": 164.21183800623052, "grad_norm": 1.9140625, "learning_rate": 3.92481663020077e-06, "loss": 0.2471, "step": 26438 }, { "epoch": 164.21806853582555, "grad_norm": 1.15625, "learning_rate": 3.923491469683424e-06, "loss": 0.1074, "step": 26439 }, { "epoch": 164.22429906542055, "grad_norm": 1.7578125, "learning_rate": 3.9221665138653495e-06, "loss": 0.1469, "step": 26440 }, { "epoch": 164.23052959501558, "grad_norm": 1.7421875, "learning_rate": 3.9208417627594e-06, "loss": 0.1666, "step": 26441 }, { "epoch": 164.23676012461058, "grad_norm": 2.34375, "learning_rate": 3.919517216378452e-06, "loss": 0.0946, "step": 26442 }, { "epoch": 164.2429906542056, "grad_norm": 1.5078125, "learning_rate": 3.918192874735371e-06, "loss": 0.1264, "step": 26443 }, { "epoch": 164.2492211838006, "grad_norm": 1.96875, "learning_rate": 3.916868737843013e-06, "loss": 0.1726, "step": 26444 }, { "epoch": 164.25545171339564, "grad_norm": 1.5078125, "learning_rate": 3.915544805714236e-06, "loss": 0.1189, "step": 26445 }, { "epoch": 164.26168224299064, "grad_norm": 2.21875, "learning_rate": 3.914221078361904e-06, "loss": 0.0852, "step": 26446 }, { "epoch": 164.26791277258567, "grad_norm": 1.90625, "learning_rate": 3.912897555798875e-06, "loss": 0.2, "step": 26447 }, { "epoch": 164.27414330218068, "grad_norm": 1.3046875, "learning_rate": 3.911574238037996e-06, "loss": 0.0927, "step": 26448 }, { "epoch": 164.2803738317757, "grad_norm": 3.234375, "learning_rate": 3.9102511250921234e-06, "loss": 0.1488, "step": 26449 }, { "epoch": 164.2866043613707, "grad_norm": 1.96875, "learning_rate": 3.9089282169741114e-06, "loss": 0.1436, "step": 26450 }, { "epoch": 164.29283489096574, "grad_norm": 1.4453125, "learning_rate": 3.907605513696808e-06, "loss": 0.1395, "step": 26451 }, { "epoch": 164.29906542056074, "grad_norm": 1.921875, "learning_rate": 3.906283015273049e-06, "loss": 0.1665, "step": 26452 }, { "epoch": 164.30529595015577, "grad_norm": 3.125, "learning_rate": 3.9049607217156836e-06, "loss": 0.2681, "step": 26453 }, { "epoch": 164.31152647975077, "grad_norm": 1.3359375, "learning_rate": 3.903638633037565e-06, "loss": 0.1141, "step": 26454 }, { "epoch": 164.3177570093458, "grad_norm": 1.6328125, "learning_rate": 3.902316749251519e-06, "loss": 0.1489, "step": 26455 }, { "epoch": 164.3239875389408, "grad_norm": 1.0703125, "learning_rate": 3.900995070370389e-06, "loss": 0.0799, "step": 26456 }, { "epoch": 164.33021806853583, "grad_norm": 2.40625, "learning_rate": 3.899673596407019e-06, "loss": 0.1597, "step": 26457 }, { "epoch": 164.33644859813083, "grad_norm": 1.1953125, "learning_rate": 3.8983523273742354e-06, "loss": 0.118, "step": 26458 }, { "epoch": 164.34267912772586, "grad_norm": 1.3359375, "learning_rate": 3.897031263284867e-06, "loss": 0.1097, "step": 26459 }, { "epoch": 164.34890965732086, "grad_norm": 3.171875, "learning_rate": 3.895710404151751e-06, "loss": 0.2104, "step": 26460 }, { "epoch": 164.3551401869159, "grad_norm": 1.1640625, "learning_rate": 3.8943897499877144e-06, "loss": 0.112, "step": 26461 }, { "epoch": 164.3613707165109, "grad_norm": 1.90625, "learning_rate": 3.893069300805588e-06, "loss": 0.1903, "step": 26462 }, { "epoch": 164.36760124610592, "grad_norm": 2.546875, "learning_rate": 3.8917490566181905e-06, "loss": 0.1806, "step": 26463 }, { "epoch": 164.37383177570092, "grad_norm": 1.8984375, "learning_rate": 3.890429017438341e-06, "loss": 0.1842, "step": 26464 }, { "epoch": 164.38006230529595, "grad_norm": 2.171875, "learning_rate": 3.889109183278866e-06, "loss": 0.1499, "step": 26465 }, { "epoch": 164.38629283489095, "grad_norm": 3.484375, "learning_rate": 3.887789554152587e-06, "loss": 0.3935, "step": 26466 }, { "epoch": 164.39252336448598, "grad_norm": 2.375, "learning_rate": 3.8864701300723085e-06, "loss": 0.1405, "step": 26467 }, { "epoch": 164.39875389408098, "grad_norm": 1.125, "learning_rate": 3.885150911050856e-06, "loss": 0.1135, "step": 26468 }, { "epoch": 164.404984423676, "grad_norm": 2.53125, "learning_rate": 3.8838318971010425e-06, "loss": 0.1942, "step": 26469 }, { "epoch": 164.41121495327104, "grad_norm": 1.53125, "learning_rate": 3.882513088235676e-06, "loss": 0.1629, "step": 26470 }, { "epoch": 164.41744548286604, "grad_norm": 2.375, "learning_rate": 3.881194484467557e-06, "loss": 0.1714, "step": 26471 }, { "epoch": 164.42367601246107, "grad_norm": 1.40625, "learning_rate": 3.879876085809497e-06, "loss": 0.127, "step": 26472 }, { "epoch": 164.42990654205607, "grad_norm": 1.5390625, "learning_rate": 3.878557892274312e-06, "loss": 0.1278, "step": 26473 }, { "epoch": 164.4361370716511, "grad_norm": 1.8828125, "learning_rate": 3.8772399038747855e-06, "loss": 0.2552, "step": 26474 }, { "epoch": 164.4423676012461, "grad_norm": 2.65625, "learning_rate": 3.875922120623729e-06, "loss": 0.4019, "step": 26475 }, { "epoch": 164.44859813084113, "grad_norm": 2.296875, "learning_rate": 3.874604542533944e-06, "loss": 0.2287, "step": 26476 }, { "epoch": 164.45482866043614, "grad_norm": 2.1875, "learning_rate": 3.873287169618226e-06, "loss": 0.1645, "step": 26477 }, { "epoch": 164.46105919003116, "grad_norm": 1.078125, "learning_rate": 3.871970001889358e-06, "loss": 0.1171, "step": 26478 }, { "epoch": 164.46728971962617, "grad_norm": 2.421875, "learning_rate": 3.870653039360139e-06, "loss": 0.2316, "step": 26479 }, { "epoch": 164.4735202492212, "grad_norm": 2.953125, "learning_rate": 3.86933628204337e-06, "loss": 0.3026, "step": 26480 }, { "epoch": 164.4797507788162, "grad_norm": 0.8203125, "learning_rate": 3.8680197299518244e-06, "loss": 0.0964, "step": 26481 }, { "epoch": 164.48598130841123, "grad_norm": 2.5, "learning_rate": 3.8667033830983e-06, "loss": 0.1815, "step": 26482 }, { "epoch": 164.49221183800623, "grad_norm": 1.2734375, "learning_rate": 3.865387241495572e-06, "loss": 0.1064, "step": 26483 }, { "epoch": 164.49844236760126, "grad_norm": 2.15625, "learning_rate": 3.864071305156433e-06, "loss": 0.1613, "step": 26484 }, { "epoch": 164.50467289719626, "grad_norm": 1.8359375, "learning_rate": 3.862755574093654e-06, "loss": 0.1949, "step": 26485 }, { "epoch": 164.5109034267913, "grad_norm": 1.7578125, "learning_rate": 3.861440048320017e-06, "loss": 0.1368, "step": 26486 }, { "epoch": 164.5171339563863, "grad_norm": 2.09375, "learning_rate": 3.860124727848305e-06, "loss": 0.2257, "step": 26487 }, { "epoch": 164.52336448598132, "grad_norm": 1.125, "learning_rate": 3.858809612691283e-06, "loss": 0.1054, "step": 26488 }, { "epoch": 164.52959501557632, "grad_norm": 2.34375, "learning_rate": 3.857494702861733e-06, "loss": 0.1449, "step": 26489 }, { "epoch": 164.53582554517135, "grad_norm": 2.359375, "learning_rate": 3.856179998372414e-06, "loss": 0.2095, "step": 26490 }, { "epoch": 164.54205607476635, "grad_norm": 1.828125, "learning_rate": 3.85486549923611e-06, "loss": 0.2135, "step": 26491 }, { "epoch": 164.54828660436138, "grad_norm": 1.109375, "learning_rate": 3.853551205465572e-06, "loss": 0.1102, "step": 26492 }, { "epoch": 164.55451713395638, "grad_norm": 2.296875, "learning_rate": 3.852237117073571e-06, "loss": 0.1939, "step": 26493 }, { "epoch": 164.5607476635514, "grad_norm": 1.8046875, "learning_rate": 3.85092323407287e-06, "loss": 0.163, "step": 26494 }, { "epoch": 164.5669781931464, "grad_norm": 2.625, "learning_rate": 3.849609556476244e-06, "loss": 0.1233, "step": 26495 }, { "epoch": 164.57320872274144, "grad_norm": 2.609375, "learning_rate": 3.848296084296424e-06, "loss": 0.2016, "step": 26496 }, { "epoch": 164.57943925233644, "grad_norm": 2.828125, "learning_rate": 3.8469828175461825e-06, "loss": 0.2361, "step": 26497 }, { "epoch": 164.58566978193147, "grad_norm": 2.09375, "learning_rate": 3.845669756238271e-06, "loss": 0.2038, "step": 26498 }, { "epoch": 164.59190031152647, "grad_norm": 1.7890625, "learning_rate": 3.844356900385452e-06, "loss": 0.1504, "step": 26499 }, { "epoch": 164.5981308411215, "grad_norm": 1.4296875, "learning_rate": 3.843044250000458e-06, "loss": 0.136, "step": 26500 }, { "epoch": 164.6043613707165, "grad_norm": 2.5, "learning_rate": 3.841731805096055e-06, "loss": 0.1426, "step": 26501 }, { "epoch": 164.61059190031153, "grad_norm": 2.078125, "learning_rate": 3.840419565684975e-06, "loss": 0.1735, "step": 26502 }, { "epoch": 164.61682242990653, "grad_norm": 0.89453125, "learning_rate": 3.839107531779978e-06, "loss": 0.1023, "step": 26503 }, { "epoch": 164.62305295950156, "grad_norm": 3.3125, "learning_rate": 3.8377957033937925e-06, "loss": 0.2427, "step": 26504 }, { "epoch": 164.62928348909657, "grad_norm": 1.9140625, "learning_rate": 3.836484080539163e-06, "loss": 0.179, "step": 26505 }, { "epoch": 164.6355140186916, "grad_norm": 2.609375, "learning_rate": 3.835172663228839e-06, "loss": 0.1951, "step": 26506 }, { "epoch": 164.6417445482866, "grad_norm": 1.890625, "learning_rate": 3.833861451475543e-06, "loss": 0.1375, "step": 26507 }, { "epoch": 164.64797507788163, "grad_norm": 2.453125, "learning_rate": 3.832550445292019e-06, "loss": 0.1644, "step": 26508 }, { "epoch": 164.65420560747663, "grad_norm": 2.625, "learning_rate": 3.831239644690993e-06, "loss": 0.2358, "step": 26509 }, { "epoch": 164.66043613707166, "grad_norm": 2.734375, "learning_rate": 3.8299290496852016e-06, "loss": 0.2499, "step": 26510 }, { "epoch": 164.66666666666666, "grad_norm": 2.5, "learning_rate": 3.82861866028737e-06, "loss": 0.1362, "step": 26511 }, { "epoch": 164.6728971962617, "grad_norm": 1.3828125, "learning_rate": 3.827308476510222e-06, "loss": 0.1211, "step": 26512 }, { "epoch": 164.6791277258567, "grad_norm": 2.515625, "learning_rate": 3.825998498366496e-06, "loss": 0.1892, "step": 26513 }, { "epoch": 164.68535825545172, "grad_norm": 1.859375, "learning_rate": 3.824688725868902e-06, "loss": 0.1474, "step": 26514 }, { "epoch": 164.69158878504672, "grad_norm": 2.296875, "learning_rate": 3.82337915903016e-06, "loss": 0.1465, "step": 26515 }, { "epoch": 164.69781931464175, "grad_norm": 2.34375, "learning_rate": 3.82206979786299e-06, "loss": 0.1728, "step": 26516 }, { "epoch": 164.70404984423675, "grad_norm": 3.765625, "learning_rate": 3.82076064238012e-06, "loss": 0.3539, "step": 26517 }, { "epoch": 164.71028037383178, "grad_norm": 2.484375, "learning_rate": 3.8194516925942505e-06, "loss": 0.2655, "step": 26518 }, { "epoch": 164.71651090342678, "grad_norm": 1.796875, "learning_rate": 3.8181429485181e-06, "loss": 0.1524, "step": 26519 }, { "epoch": 164.7227414330218, "grad_norm": 1.2890625, "learning_rate": 3.8168344101643825e-06, "loss": 0.1318, "step": 26520 }, { "epoch": 164.7289719626168, "grad_norm": 1.859375, "learning_rate": 3.815526077545806e-06, "loss": 0.1478, "step": 26521 }, { "epoch": 164.73520249221184, "grad_norm": 1.2734375, "learning_rate": 3.8142179506750695e-06, "loss": 0.1327, "step": 26522 }, { "epoch": 164.74143302180684, "grad_norm": 2.296875, "learning_rate": 3.812910029564881e-06, "loss": 0.2593, "step": 26523 }, { "epoch": 164.74766355140187, "grad_norm": 1.515625, "learning_rate": 3.811602314227955e-06, "loss": 0.133, "step": 26524 }, { "epoch": 164.75389408099687, "grad_norm": 3.046875, "learning_rate": 3.810294804676973e-06, "loss": 0.3045, "step": 26525 }, { "epoch": 164.7601246105919, "grad_norm": 1.5546875, "learning_rate": 3.808987500924646e-06, "loss": 0.1057, "step": 26526 }, { "epoch": 164.7663551401869, "grad_norm": 1.59375, "learning_rate": 3.8076804029836736e-06, "loss": 0.1143, "step": 26527 }, { "epoch": 164.77258566978193, "grad_norm": 1.9765625, "learning_rate": 3.8063735108667455e-06, "loss": 0.1816, "step": 26528 }, { "epoch": 164.77881619937693, "grad_norm": 1.5703125, "learning_rate": 3.805066824586548e-06, "loss": 0.0979, "step": 26529 }, { "epoch": 164.78504672897196, "grad_norm": 2.359375, "learning_rate": 3.8037603441557796e-06, "loss": 0.2395, "step": 26530 }, { "epoch": 164.79127725856696, "grad_norm": 2.890625, "learning_rate": 3.8024540695871274e-06, "loss": 0.2561, "step": 26531 }, { "epoch": 164.797507788162, "grad_norm": 2.3125, "learning_rate": 3.8011480008932843e-06, "loss": 0.1672, "step": 26532 }, { "epoch": 164.80373831775702, "grad_norm": 2.96875, "learning_rate": 3.7998421380869287e-06, "loss": 0.2858, "step": 26533 }, { "epoch": 164.80996884735202, "grad_norm": 2.515625, "learning_rate": 3.7985364811807367e-06, "loss": 0.1038, "step": 26534 }, { "epoch": 164.81619937694705, "grad_norm": 2.109375, "learning_rate": 3.797231030187398e-06, "loss": 0.1413, "step": 26535 }, { "epoch": 164.82242990654206, "grad_norm": 1.625, "learning_rate": 3.7959257851195974e-06, "loss": 0.1329, "step": 26536 }, { "epoch": 164.82866043613708, "grad_norm": 2.265625, "learning_rate": 3.7946207459899963e-06, "loss": 0.176, "step": 26537 }, { "epoch": 164.8348909657321, "grad_norm": 3.125, "learning_rate": 3.7933159128112796e-06, "loss": 0.295, "step": 26538 }, { "epoch": 164.84112149532712, "grad_norm": 0.97265625, "learning_rate": 3.79201128559612e-06, "loss": 0.1256, "step": 26539 }, { "epoch": 164.84735202492212, "grad_norm": 1.1953125, "learning_rate": 3.7907068643571907e-06, "loss": 0.1202, "step": 26540 }, { "epoch": 164.85358255451715, "grad_norm": 2.4375, "learning_rate": 3.789402649107146e-06, "loss": 0.1196, "step": 26541 }, { "epoch": 164.85981308411215, "grad_norm": 1.8125, "learning_rate": 3.7880986398586644e-06, "loss": 0.1656, "step": 26542 }, { "epoch": 164.86604361370718, "grad_norm": 1.796875, "learning_rate": 3.786794836624416e-06, "loss": 0.1558, "step": 26543 }, { "epoch": 164.87227414330218, "grad_norm": 2.328125, "learning_rate": 3.7854912394170495e-06, "loss": 0.1595, "step": 26544 }, { "epoch": 164.8785046728972, "grad_norm": 2.625, "learning_rate": 3.784187848249232e-06, "loss": 0.2645, "step": 26545 }, { "epoch": 164.8847352024922, "grad_norm": 2.90625, "learning_rate": 3.782884663133632e-06, "loss": 0.2077, "step": 26546 }, { "epoch": 164.89096573208724, "grad_norm": 2.40625, "learning_rate": 3.781581684082894e-06, "loss": 0.1891, "step": 26547 }, { "epoch": 164.89719626168224, "grad_norm": 1.265625, "learning_rate": 3.780278911109672e-06, "loss": 0.1309, "step": 26548 }, { "epoch": 164.90342679127727, "grad_norm": 2.34375, "learning_rate": 3.778976344226623e-06, "loss": 0.2416, "step": 26549 }, { "epoch": 164.90965732087227, "grad_norm": 2.03125, "learning_rate": 3.777673983446403e-06, "loss": 0.2355, "step": 26550 }, { "epoch": 164.9158878504673, "grad_norm": 2.84375, "learning_rate": 3.7763718287816495e-06, "loss": 0.2063, "step": 26551 }, { "epoch": 164.9221183800623, "grad_norm": 2.09375, "learning_rate": 3.7750698802450217e-06, "loss": 0.1792, "step": 26552 }, { "epoch": 164.92834890965733, "grad_norm": 1.40625, "learning_rate": 3.7737681378491515e-06, "loss": 0.1015, "step": 26553 }, { "epoch": 164.93457943925233, "grad_norm": 2.0625, "learning_rate": 3.7724666016066947e-06, "loss": 0.2541, "step": 26554 }, { "epoch": 164.94080996884736, "grad_norm": 2.359375, "learning_rate": 3.7711652715302777e-06, "loss": 0.1994, "step": 26555 }, { "epoch": 164.94704049844236, "grad_norm": 2.125, "learning_rate": 3.7698641476325487e-06, "loss": 0.2065, "step": 26556 }, { "epoch": 164.9532710280374, "grad_norm": 1.421875, "learning_rate": 3.7685632299261447e-06, "loss": 0.1007, "step": 26557 }, { "epoch": 164.9595015576324, "grad_norm": 1.5546875, "learning_rate": 3.7672625184237034e-06, "loss": 0.1212, "step": 26558 }, { "epoch": 164.96573208722742, "grad_norm": 2.953125, "learning_rate": 3.7659620131378527e-06, "loss": 0.1252, "step": 26559 }, { "epoch": 164.97196261682242, "grad_norm": 1.9140625, "learning_rate": 3.7646617140812164e-06, "loss": 0.1018, "step": 26560 }, { "epoch": 164.97819314641745, "grad_norm": 1.5390625, "learning_rate": 3.7633616212664315e-06, "loss": 0.1152, "step": 26561 }, { "epoch": 164.98442367601245, "grad_norm": 1.328125, "learning_rate": 3.7620617347061293e-06, "loss": 0.0867, "step": 26562 }, { "epoch": 164.99065420560748, "grad_norm": 1.6953125, "learning_rate": 3.7607620544129225e-06, "loss": 0.1603, "step": 26563 }, { "epoch": 164.99688473520249, "grad_norm": 2.921875, "learning_rate": 3.759462580399442e-06, "loss": 0.2159, "step": 26564 }, { "epoch": 165.0, "grad_norm": 1.1328125, "learning_rate": 3.758163312678312e-06, "loss": 0.0484, "step": 26565 }, { "epoch": 165.00623052959503, "grad_norm": 2.265625, "learning_rate": 3.756864251262143e-06, "loss": 0.2678, "step": 26566 }, { "epoch": 165.01246105919003, "grad_norm": 2.1875, "learning_rate": 3.755565396163552e-06, "loss": 0.1764, "step": 26567 }, { "epoch": 165.01869158878506, "grad_norm": 1.703125, "learning_rate": 3.7542667473951578e-06, "loss": 0.1794, "step": 26568 }, { "epoch": 165.02492211838006, "grad_norm": 1.578125, "learning_rate": 3.752968304969576e-06, "loss": 0.1203, "step": 26569 }, { "epoch": 165.0311526479751, "grad_norm": 2.59375, "learning_rate": 3.7516700688994074e-06, "loss": 0.34, "step": 26570 }, { "epoch": 165.0373831775701, "grad_norm": 2.90625, "learning_rate": 3.7503720391972723e-06, "loss": 0.2785, "step": 26571 }, { "epoch": 165.04361370716512, "grad_norm": 1.2890625, "learning_rate": 3.749074215875767e-06, "loss": 0.0989, "step": 26572 }, { "epoch": 165.04984423676012, "grad_norm": 2.578125, "learning_rate": 3.7477765989475054e-06, "loss": 0.2075, "step": 26573 }, { "epoch": 165.05607476635515, "grad_norm": 2.9375, "learning_rate": 3.7464791884250783e-06, "loss": 0.1265, "step": 26574 }, { "epoch": 165.06230529595015, "grad_norm": 1.7734375, "learning_rate": 3.745181984321097e-06, "loss": 0.1336, "step": 26575 }, { "epoch": 165.06853582554518, "grad_norm": 2.25, "learning_rate": 3.743884986648161e-06, "loss": 0.1721, "step": 26576 }, { "epoch": 165.07476635514018, "grad_norm": 1.3203125, "learning_rate": 3.7425881954188587e-06, "loss": 0.1198, "step": 26577 }, { "epoch": 165.0809968847352, "grad_norm": 1.1640625, "learning_rate": 3.741291610645792e-06, "loss": 0.1121, "step": 26578 }, { "epoch": 165.08722741433021, "grad_norm": 3.515625, "learning_rate": 3.7399952323415477e-06, "loss": 0.3313, "step": 26579 }, { "epoch": 165.09345794392524, "grad_norm": 1.1015625, "learning_rate": 3.7386990605187245e-06, "loss": 0.0943, "step": 26580 }, { "epoch": 165.09968847352025, "grad_norm": 2.359375, "learning_rate": 3.737403095189901e-06, "loss": 0.1679, "step": 26581 }, { "epoch": 165.10591900311528, "grad_norm": 2.15625, "learning_rate": 3.736107336367667e-06, "loss": 0.1035, "step": 26582 }, { "epoch": 165.11214953271028, "grad_norm": 1.2421875, "learning_rate": 3.734811784064618e-06, "loss": 0.1116, "step": 26583 }, { "epoch": 165.1183800623053, "grad_norm": 2.171875, "learning_rate": 3.7335164382933246e-06, "loss": 0.145, "step": 26584 }, { "epoch": 165.1246105919003, "grad_norm": 2.375, "learning_rate": 3.7322212990663658e-06, "loss": 0.1799, "step": 26585 }, { "epoch": 165.13084112149534, "grad_norm": 2.34375, "learning_rate": 3.7309263663963257e-06, "loss": 0.2101, "step": 26586 }, { "epoch": 165.13707165109034, "grad_norm": 1.953125, "learning_rate": 3.729631640295786e-06, "loss": 0.164, "step": 26587 }, { "epoch": 165.14330218068537, "grad_norm": 2.015625, "learning_rate": 3.728337120777309e-06, "loss": 0.1477, "step": 26588 }, { "epoch": 165.14953271028037, "grad_norm": 1.8828125, "learning_rate": 3.727042807853476e-06, "loss": 0.1324, "step": 26589 }, { "epoch": 165.1557632398754, "grad_norm": 2.140625, "learning_rate": 3.72574870153686e-06, "loss": 0.1956, "step": 26590 }, { "epoch": 165.1619937694704, "grad_norm": 1.953125, "learning_rate": 3.7244548018400245e-06, "loss": 0.1264, "step": 26591 }, { "epoch": 165.16822429906543, "grad_norm": 2.125, "learning_rate": 3.723161108775533e-06, "loss": 0.2534, "step": 26592 }, { "epoch": 165.17445482866043, "grad_norm": 2.140625, "learning_rate": 3.721867622355951e-06, "loss": 0.2322, "step": 26593 }, { "epoch": 165.18068535825546, "grad_norm": 1.578125, "learning_rate": 3.7205743425938466e-06, "loss": 0.1291, "step": 26594 }, { "epoch": 165.18691588785046, "grad_norm": 1.6953125, "learning_rate": 3.7192812695017843e-06, "loss": 0.1397, "step": 26595 }, { "epoch": 165.1931464174455, "grad_norm": 2.828125, "learning_rate": 3.7179884030923094e-06, "loss": 0.2236, "step": 26596 }, { "epoch": 165.1993769470405, "grad_norm": 2.125, "learning_rate": 3.7166957433779904e-06, "loss": 0.1735, "step": 26597 }, { "epoch": 165.20560747663552, "grad_norm": 3.546875, "learning_rate": 3.715403290371372e-06, "loss": 0.3005, "step": 26598 }, { "epoch": 165.21183800623052, "grad_norm": 2.46875, "learning_rate": 3.7141110440850167e-06, "loss": 0.1799, "step": 26599 }, { "epoch": 165.21806853582555, "grad_norm": 1.4375, "learning_rate": 3.7128190045314646e-06, "loss": 0.1168, "step": 26600 }, { "epoch": 165.22429906542055, "grad_norm": 2.34375, "learning_rate": 3.711527171723267e-06, "loss": 0.2687, "step": 26601 }, { "epoch": 165.23052959501558, "grad_norm": 1.140625, "learning_rate": 3.7102355456729827e-06, "loss": 0.1078, "step": 26602 }, { "epoch": 165.23676012461058, "grad_norm": 1.734375, "learning_rate": 3.7089441263931435e-06, "loss": 0.2089, "step": 26603 }, { "epoch": 165.2429906542056, "grad_norm": 1.9296875, "learning_rate": 3.7076529138962924e-06, "loss": 0.2031, "step": 26604 }, { "epoch": 165.2492211838006, "grad_norm": 3.453125, "learning_rate": 3.7063619081949693e-06, "loss": 0.1927, "step": 26605 }, { "epoch": 165.25545171339564, "grad_norm": 1.6171875, "learning_rate": 3.705071109301722e-06, "loss": 0.1876, "step": 26606 }, { "epoch": 165.26168224299064, "grad_norm": 1.796875, "learning_rate": 3.7037805172290775e-06, "loss": 0.1474, "step": 26607 }, { "epoch": 165.26791277258567, "grad_norm": 3.03125, "learning_rate": 3.702490131989572e-06, "loss": 0.183, "step": 26608 }, { "epoch": 165.27414330218068, "grad_norm": 0.91015625, "learning_rate": 3.7011999535957436e-06, "loss": 0.1037, "step": 26609 }, { "epoch": 165.2803738317757, "grad_norm": 1.9140625, "learning_rate": 3.6999099820601206e-06, "loss": 0.1067, "step": 26610 }, { "epoch": 165.2866043613707, "grad_norm": 1.9375, "learning_rate": 3.698620217395224e-06, "loss": 0.1699, "step": 26611 }, { "epoch": 165.29283489096574, "grad_norm": 1.15625, "learning_rate": 3.697330659613588e-06, "loss": 0.1148, "step": 26612 }, { "epoch": 165.29906542056074, "grad_norm": 2.265625, "learning_rate": 3.696041308727738e-06, "loss": 0.2278, "step": 26613 }, { "epoch": 165.30529595015577, "grad_norm": 1.9765625, "learning_rate": 3.694752164750187e-06, "loss": 0.2387, "step": 26614 }, { "epoch": 165.31152647975077, "grad_norm": 2.421875, "learning_rate": 3.6934632276934637e-06, "loss": 0.2558, "step": 26615 }, { "epoch": 165.3177570093458, "grad_norm": 1.2578125, "learning_rate": 3.692174497570089e-06, "loss": 0.0879, "step": 26616 }, { "epoch": 165.3239875389408, "grad_norm": 2.28125, "learning_rate": 3.690885974392577e-06, "loss": 0.1553, "step": 26617 }, { "epoch": 165.33021806853583, "grad_norm": 2.0, "learning_rate": 3.6895976581734325e-06, "loss": 0.1736, "step": 26618 }, { "epoch": 165.33644859813083, "grad_norm": 3.890625, "learning_rate": 3.6883095489251758e-06, "loss": 0.3205, "step": 26619 }, { "epoch": 165.34267912772586, "grad_norm": 2.9375, "learning_rate": 3.6870216466603215e-06, "loss": 0.2475, "step": 26620 }, { "epoch": 165.34890965732086, "grad_norm": 3.28125, "learning_rate": 3.685733951391371e-06, "loss": 0.3142, "step": 26621 }, { "epoch": 165.3551401869159, "grad_norm": 2.921875, "learning_rate": 3.684446463130836e-06, "loss": 0.3118, "step": 26622 }, { "epoch": 165.3613707165109, "grad_norm": 1.5078125, "learning_rate": 3.683159181891213e-06, "loss": 0.1169, "step": 26623 }, { "epoch": 165.36760124610592, "grad_norm": 3.09375, "learning_rate": 3.6818721076850137e-06, "loss": 0.3925, "step": 26624 }, { "epoch": 165.37383177570092, "grad_norm": 1.3046875, "learning_rate": 3.6805852405247282e-06, "loss": 0.1203, "step": 26625 }, { "epoch": 165.38006230529595, "grad_norm": 1.4921875, "learning_rate": 3.67929858042286e-06, "loss": 0.1014, "step": 26626 }, { "epoch": 165.38629283489095, "grad_norm": 2.265625, "learning_rate": 3.6780121273919084e-06, "loss": 0.2015, "step": 26627 }, { "epoch": 165.39252336448598, "grad_norm": 2.640625, "learning_rate": 3.6767258814443705e-06, "loss": 0.2119, "step": 26628 }, { "epoch": 165.39875389408098, "grad_norm": 2.46875, "learning_rate": 3.6754398425927315e-06, "loss": 0.2279, "step": 26629 }, { "epoch": 165.404984423676, "grad_norm": 1.9453125, "learning_rate": 3.674154010849479e-06, "loss": 0.1606, "step": 26630 }, { "epoch": 165.41121495327104, "grad_norm": 1.484375, "learning_rate": 3.6728683862271047e-06, "loss": 0.0919, "step": 26631 }, { "epoch": 165.41744548286604, "grad_norm": 2.4375, "learning_rate": 3.671582968738102e-06, "loss": 0.1424, "step": 26632 }, { "epoch": 165.42367601246107, "grad_norm": 1.9765625, "learning_rate": 3.6702977583949443e-06, "loss": 0.2548, "step": 26633 }, { "epoch": 165.42990654205607, "grad_norm": 1.84375, "learning_rate": 3.669012755210119e-06, "loss": 0.1703, "step": 26634 }, { "epoch": 165.4361370716511, "grad_norm": 2.65625, "learning_rate": 3.66772795919611e-06, "loss": 0.1898, "step": 26635 }, { "epoch": 165.4423676012461, "grad_norm": 2.1875, "learning_rate": 3.666443370365391e-06, "loss": 0.1189, "step": 26636 }, { "epoch": 165.44859813084113, "grad_norm": 2.25, "learning_rate": 3.665158988730433e-06, "loss": 0.2115, "step": 26637 }, { "epoch": 165.45482866043614, "grad_norm": 1.8046875, "learning_rate": 3.6638748143037165e-06, "loss": 0.1401, "step": 26638 }, { "epoch": 165.46105919003116, "grad_norm": 2.515625, "learning_rate": 3.6625908470977186e-06, "loss": 0.227, "step": 26639 }, { "epoch": 165.46728971962617, "grad_norm": 0.9609375, "learning_rate": 3.6613070871248963e-06, "loss": 0.0889, "step": 26640 }, { "epoch": 165.4735202492212, "grad_norm": 1.6015625, "learning_rate": 3.660023534397733e-06, "loss": 0.1243, "step": 26641 }, { "epoch": 165.4797507788162, "grad_norm": 2.046875, "learning_rate": 3.6587401889286775e-06, "loss": 0.1544, "step": 26642 }, { "epoch": 165.48598130841123, "grad_norm": 1.75, "learning_rate": 3.657457050730212e-06, "loss": 0.1474, "step": 26643 }, { "epoch": 165.49221183800623, "grad_norm": 2.1875, "learning_rate": 3.6561741198147844e-06, "loss": 0.2052, "step": 26644 }, { "epoch": 165.49844236760126, "grad_norm": 1.8203125, "learning_rate": 3.6548913961948596e-06, "loss": 0.1423, "step": 26645 }, { "epoch": 165.50467289719626, "grad_norm": 2.28125, "learning_rate": 3.653608879882903e-06, "loss": 0.1988, "step": 26646 }, { "epoch": 165.5109034267913, "grad_norm": 1.921875, "learning_rate": 3.6523265708913567e-06, "loss": 0.1862, "step": 26647 }, { "epoch": 165.5171339563863, "grad_norm": 2.296875, "learning_rate": 3.6510444692326894e-06, "loss": 0.1914, "step": 26648 }, { "epoch": 165.52336448598132, "grad_norm": 1.734375, "learning_rate": 3.649762574919338e-06, "loss": 0.1348, "step": 26649 }, { "epoch": 165.52959501557632, "grad_norm": 2.0625, "learning_rate": 3.6484808879637678e-06, "loss": 0.1489, "step": 26650 }, { "epoch": 165.53582554517135, "grad_norm": 2.21875, "learning_rate": 3.6471994083784128e-06, "loss": 0.2171, "step": 26651 }, { "epoch": 165.54205607476635, "grad_norm": 0.99609375, "learning_rate": 3.6459181361757244e-06, "loss": 0.1003, "step": 26652 }, { "epoch": 165.54828660436138, "grad_norm": 1.7421875, "learning_rate": 3.644637071368151e-06, "loss": 0.143, "step": 26653 }, { "epoch": 165.55451713395638, "grad_norm": 2.765625, "learning_rate": 3.6433562139681383e-06, "loss": 0.196, "step": 26654 }, { "epoch": 165.5607476635514, "grad_norm": 2.203125, "learning_rate": 3.642075563988112e-06, "loss": 0.1674, "step": 26655 }, { "epoch": 165.5669781931464, "grad_norm": 2.1875, "learning_rate": 3.640795121440513e-06, "loss": 0.2937, "step": 26656 }, { "epoch": 165.57320872274144, "grad_norm": 2.9375, "learning_rate": 3.6395148863377858e-06, "loss": 0.1154, "step": 26657 }, { "epoch": 165.57943925233644, "grad_norm": 1.625, "learning_rate": 3.6382348586923632e-06, "loss": 0.2354, "step": 26658 }, { "epoch": 165.58566978193147, "grad_norm": 1.6953125, "learning_rate": 3.636955038516668e-06, "loss": 0.12, "step": 26659 }, { "epoch": 165.59190031152647, "grad_norm": 2.796875, "learning_rate": 3.63567542582314e-06, "loss": 0.2399, "step": 26660 }, { "epoch": 165.5981308411215, "grad_norm": 3.109375, "learning_rate": 3.6343960206242004e-06, "loss": 0.2034, "step": 26661 }, { "epoch": 165.6043613707165, "grad_norm": 1.4921875, "learning_rate": 3.633116822932281e-06, "loss": 0.1199, "step": 26662 }, { "epoch": 165.61059190031153, "grad_norm": 1.6875, "learning_rate": 3.631837832759799e-06, "loss": 0.1384, "step": 26663 }, { "epoch": 165.61682242990653, "grad_norm": 2.171875, "learning_rate": 3.6305590501191753e-06, "loss": 0.1854, "step": 26664 }, { "epoch": 165.62305295950156, "grad_norm": 2.25, "learning_rate": 3.629280475022842e-06, "loss": 0.1698, "step": 26665 }, { "epoch": 165.62928348909657, "grad_norm": 2.203125, "learning_rate": 3.6280021074832026e-06, "loss": 0.2227, "step": 26666 }, { "epoch": 165.6355140186916, "grad_norm": 1.359375, "learning_rate": 3.6267239475126863e-06, "loss": 0.1203, "step": 26667 }, { "epoch": 165.6417445482866, "grad_norm": 2.1875, "learning_rate": 3.6254459951236915e-06, "loss": 0.2438, "step": 26668 }, { "epoch": 165.64797507788163, "grad_norm": 2.09375, "learning_rate": 3.624168250328644e-06, "loss": 0.1801, "step": 26669 }, { "epoch": 165.65420560747663, "grad_norm": 3.125, "learning_rate": 3.6228907131399427e-06, "loss": 0.3367, "step": 26670 }, { "epoch": 165.66043613707166, "grad_norm": 2.875, "learning_rate": 3.6216133835699995e-06, "loss": 0.2195, "step": 26671 }, { "epoch": 165.66666666666666, "grad_norm": 2.78125, "learning_rate": 3.6203362616312268e-06, "loss": 0.2148, "step": 26672 }, { "epoch": 165.6728971962617, "grad_norm": 2.125, "learning_rate": 3.6190593473360207e-06, "loss": 0.1376, "step": 26673 }, { "epoch": 165.6791277258567, "grad_norm": 2.765625, "learning_rate": 3.617782640696779e-06, "loss": 0.2624, "step": 26674 }, { "epoch": 165.68535825545172, "grad_norm": 1.9765625, "learning_rate": 3.616506141725906e-06, "loss": 0.1388, "step": 26675 }, { "epoch": 165.69158878504672, "grad_norm": 1.59375, "learning_rate": 3.6152298504358084e-06, "loss": 0.1135, "step": 26676 }, { "epoch": 165.69781931464175, "grad_norm": 3.21875, "learning_rate": 3.6139537668388653e-06, "loss": 0.2364, "step": 26677 }, { "epoch": 165.70404984423675, "grad_norm": 2.5625, "learning_rate": 3.6126778909474773e-06, "loss": 0.1754, "step": 26678 }, { "epoch": 165.71028037383178, "grad_norm": 1.5078125, "learning_rate": 3.6114022227740433e-06, "loss": 0.1048, "step": 26679 }, { "epoch": 165.71651090342678, "grad_norm": 2.171875, "learning_rate": 3.6101267623309454e-06, "loss": 0.1723, "step": 26680 }, { "epoch": 165.7227414330218, "grad_norm": 1.8671875, "learning_rate": 3.6088515096305674e-06, "loss": 0.1968, "step": 26681 }, { "epoch": 165.7289719626168, "grad_norm": 2.015625, "learning_rate": 3.6075764646852998e-06, "loss": 0.1764, "step": 26682 }, { "epoch": 165.73520249221184, "grad_norm": 1.84375, "learning_rate": 3.606301627507533e-06, "loss": 0.1431, "step": 26683 }, { "epoch": 165.74143302180684, "grad_norm": 1.71875, "learning_rate": 3.605026998109634e-06, "loss": 0.1239, "step": 26684 }, { "epoch": 165.74766355140187, "grad_norm": 1.4375, "learning_rate": 3.603752576503988e-06, "loss": 0.1289, "step": 26685 }, { "epoch": 165.75389408099687, "grad_norm": 2.375, "learning_rate": 3.602478362702982e-06, "loss": 0.2204, "step": 26686 }, { "epoch": 165.7601246105919, "grad_norm": 3.078125, "learning_rate": 3.6012043567189845e-06, "loss": 0.1833, "step": 26687 }, { "epoch": 165.7663551401869, "grad_norm": 1.4140625, "learning_rate": 3.59993055856436e-06, "loss": 0.1466, "step": 26688 }, { "epoch": 165.77258566978193, "grad_norm": 2.390625, "learning_rate": 3.5986569682514876e-06, "loss": 0.1534, "step": 26689 }, { "epoch": 165.77881619937693, "grad_norm": 2.203125, "learning_rate": 3.597383585792738e-06, "loss": 0.2033, "step": 26690 }, { "epoch": 165.78504672897196, "grad_norm": 1.5390625, "learning_rate": 3.5961104112004816e-06, "loss": 0.1096, "step": 26691 }, { "epoch": 165.79127725856696, "grad_norm": 1.6875, "learning_rate": 3.5948374444870814e-06, "loss": 0.1155, "step": 26692 }, { "epoch": 165.797507788162, "grad_norm": 1.9375, "learning_rate": 3.5935646856648907e-06, "loss": 0.1281, "step": 26693 }, { "epoch": 165.80373831775702, "grad_norm": 1.5859375, "learning_rate": 3.592292134746281e-06, "loss": 0.1129, "step": 26694 }, { "epoch": 165.80996884735202, "grad_norm": 1.671875, "learning_rate": 3.5910197917436135e-06, "loss": 0.0924, "step": 26695 }, { "epoch": 165.81619937694705, "grad_norm": 0.9453125, "learning_rate": 3.5897476566692346e-06, "loss": 0.1001, "step": 26696 }, { "epoch": 165.82242990654206, "grad_norm": 2.625, "learning_rate": 3.5884757295355095e-06, "loss": 0.2064, "step": 26697 }, { "epoch": 165.82866043613708, "grad_norm": 2.140625, "learning_rate": 3.5872040103547917e-06, "loss": 0.2473, "step": 26698 }, { "epoch": 165.8348909657321, "grad_norm": 1.3359375, "learning_rate": 3.5859324991394298e-06, "loss": 0.1165, "step": 26699 }, { "epoch": 165.84112149532712, "grad_norm": 1.015625, "learning_rate": 3.5846611959017645e-06, "loss": 0.107, "step": 26700 }, { "epoch": 165.84735202492212, "grad_norm": 1.2265625, "learning_rate": 3.583390100654152e-06, "loss": 0.1115, "step": 26701 }, { "epoch": 165.85358255451715, "grad_norm": 1.5078125, "learning_rate": 3.582119213408941e-06, "loss": 0.1064, "step": 26702 }, { "epoch": 165.85981308411215, "grad_norm": 1.296875, "learning_rate": 3.5808485341784658e-06, "loss": 0.1194, "step": 26703 }, { "epoch": 165.86604361370718, "grad_norm": 1.7734375, "learning_rate": 3.57957806297507e-06, "loss": 0.1678, "step": 26704 }, { "epoch": 165.87227414330218, "grad_norm": 1.375, "learning_rate": 3.578307799811098e-06, "loss": 0.1286, "step": 26705 }, { "epoch": 165.8785046728972, "grad_norm": 2.640625, "learning_rate": 3.5770377446988855e-06, "loss": 0.1762, "step": 26706 }, { "epoch": 165.8847352024922, "grad_norm": 2.0, "learning_rate": 3.5757678976507585e-06, "loss": 0.184, "step": 26707 }, { "epoch": 165.89096573208724, "grad_norm": 1.078125, "learning_rate": 3.574498258679054e-06, "loss": 0.1118, "step": 26708 }, { "epoch": 165.89719626168224, "grad_norm": 1.3515625, "learning_rate": 3.5732288277961127e-06, "loss": 0.1209, "step": 26709 }, { "epoch": 165.90342679127727, "grad_norm": 2.171875, "learning_rate": 3.5719596050142523e-06, "loss": 0.1047, "step": 26710 }, { "epoch": 165.90965732087227, "grad_norm": 1.671875, "learning_rate": 3.570690590345807e-06, "loss": 0.1113, "step": 26711 }, { "epoch": 165.9158878504673, "grad_norm": 2.671875, "learning_rate": 3.5694217838030923e-06, "loss": 0.2404, "step": 26712 }, { "epoch": 165.9221183800623, "grad_norm": 2.28125, "learning_rate": 3.568153185398443e-06, "loss": 0.1146, "step": 26713 }, { "epoch": 165.92834890965733, "grad_norm": 2.390625, "learning_rate": 3.566884795144171e-06, "loss": 0.1806, "step": 26714 }, { "epoch": 165.93457943925233, "grad_norm": 1.3203125, "learning_rate": 3.565616613052594e-06, "loss": 0.1396, "step": 26715 }, { "epoch": 165.94080996884736, "grad_norm": 1.890625, "learning_rate": 3.564348639136042e-06, "loss": 0.1285, "step": 26716 }, { "epoch": 165.94704049844236, "grad_norm": 2.4375, "learning_rate": 3.5630808734068128e-06, "loss": 0.226, "step": 26717 }, { "epoch": 165.9532710280374, "grad_norm": 2.046875, "learning_rate": 3.5618133158772352e-06, "loss": 0.179, "step": 26718 }, { "epoch": 165.9595015576324, "grad_norm": 1.796875, "learning_rate": 3.560545966559606e-06, "loss": 0.1557, "step": 26719 }, { "epoch": 165.96573208722742, "grad_norm": 2.90625, "learning_rate": 3.5592788254662447e-06, "loss": 0.3143, "step": 26720 }, { "epoch": 165.97196261682242, "grad_norm": 2.171875, "learning_rate": 3.5580118926094504e-06, "loss": 0.1766, "step": 26721 }, { "epoch": 165.97819314641745, "grad_norm": 1.015625, "learning_rate": 3.556745168001527e-06, "loss": 0.1038, "step": 26722 }, { "epoch": 165.98442367601245, "grad_norm": 1.90625, "learning_rate": 3.555478651654784e-06, "loss": 0.1424, "step": 26723 }, { "epoch": 165.99065420560748, "grad_norm": 2.328125, "learning_rate": 3.554212343581528e-06, "loss": 0.206, "step": 26724 }, { "epoch": 165.99688473520249, "grad_norm": 1.671875, "learning_rate": 3.5529462437940392e-06, "loss": 0.1407, "step": 26725 }, { "epoch": 166.0, "grad_norm": 1.578125, "learning_rate": 3.5516803523046207e-06, "loss": 0.0753, "step": 26726 }, { "epoch": 166.00623052959503, "grad_norm": 1.4609375, "learning_rate": 3.5504146691255736e-06, "loss": 0.1205, "step": 26727 }, { "epoch": 166.01246105919003, "grad_norm": 1.921875, "learning_rate": 3.5491491942691913e-06, "loss": 0.2166, "step": 26728 }, { "epoch": 166.01869158878506, "grad_norm": 2.484375, "learning_rate": 3.547883927747753e-06, "loss": 0.1554, "step": 26729 }, { "epoch": 166.02492211838006, "grad_norm": 2.1875, "learning_rate": 3.546618869573562e-06, "loss": 0.209, "step": 26730 }, { "epoch": 166.0311526479751, "grad_norm": 2.4375, "learning_rate": 3.5453540197588896e-06, "loss": 0.1683, "step": 26731 }, { "epoch": 166.0373831775701, "grad_norm": 3.703125, "learning_rate": 3.5440893783160347e-06, "loss": 0.1865, "step": 26732 }, { "epoch": 166.04361370716512, "grad_norm": 1.28125, "learning_rate": 3.542824945257264e-06, "loss": 0.1144, "step": 26733 }, { "epoch": 166.04984423676012, "grad_norm": 1.9609375, "learning_rate": 3.541560720594869e-06, "loss": 0.164, "step": 26734 }, { "epoch": 166.05607476635515, "grad_norm": 3.609375, "learning_rate": 3.5402967043411338e-06, "loss": 0.3734, "step": 26735 }, { "epoch": 166.06230529595015, "grad_norm": 1.1015625, "learning_rate": 3.539032896508318e-06, "loss": 0.0984, "step": 26736 }, { "epoch": 166.06853582554518, "grad_norm": 3.328125, "learning_rate": 3.5377692971087117e-06, "loss": 0.3358, "step": 26737 }, { "epoch": 166.07476635514018, "grad_norm": 1.5390625, "learning_rate": 3.536505906154572e-06, "loss": 0.2285, "step": 26738 }, { "epoch": 166.0809968847352, "grad_norm": 2.375, "learning_rate": 3.535242723658186e-06, "loss": 0.2838, "step": 26739 }, { "epoch": 166.08722741433021, "grad_norm": 1.359375, "learning_rate": 3.533979749631805e-06, "loss": 0.1131, "step": 26740 }, { "epoch": 166.09345794392524, "grad_norm": 2.265625, "learning_rate": 3.5327169840877055e-06, "loss": 0.1655, "step": 26741 }, { "epoch": 166.09968847352025, "grad_norm": 2.140625, "learning_rate": 3.5314544270381554e-06, "loss": 0.1764, "step": 26742 }, { "epoch": 166.10591900311528, "grad_norm": 1.625, "learning_rate": 3.5301920784954117e-06, "loss": 0.1643, "step": 26743 }, { "epoch": 166.11214953271028, "grad_norm": 2.625, "learning_rate": 3.5289299384717285e-06, "loss": 0.2274, "step": 26744 }, { "epoch": 166.1183800623053, "grad_norm": 1.5703125, "learning_rate": 3.5276680069793703e-06, "loss": 0.1367, "step": 26745 }, { "epoch": 166.1246105919003, "grad_norm": 1.34375, "learning_rate": 3.526406284030595e-06, "loss": 0.0883, "step": 26746 }, { "epoch": 166.13084112149534, "grad_norm": 1.640625, "learning_rate": 3.525144769637653e-06, "loss": 0.1171, "step": 26747 }, { "epoch": 166.13707165109034, "grad_norm": 2.25, "learning_rate": 3.5238834638127964e-06, "loss": 0.1514, "step": 26748 }, { "epoch": 166.14330218068537, "grad_norm": 2.96875, "learning_rate": 3.522622366568276e-06, "loss": 0.2942, "step": 26749 }, { "epoch": 166.14953271028037, "grad_norm": 1.4453125, "learning_rate": 3.5213614779163518e-06, "loss": 0.1208, "step": 26750 }, { "epoch": 166.1557632398754, "grad_norm": 2.4375, "learning_rate": 3.520100797869247e-06, "loss": 0.1186, "step": 26751 }, { "epoch": 166.1619937694704, "grad_norm": 2.375, "learning_rate": 3.5188403264392154e-06, "loss": 0.1729, "step": 26752 }, { "epoch": 166.16822429906543, "grad_norm": 2.671875, "learning_rate": 3.517580063638504e-06, "loss": 0.3333, "step": 26753 }, { "epoch": 166.17445482866043, "grad_norm": 1.9609375, "learning_rate": 3.5163200094793514e-06, "loss": 0.1018, "step": 26754 }, { "epoch": 166.18068535825546, "grad_norm": 1.078125, "learning_rate": 3.5150601639739905e-06, "loss": 0.0819, "step": 26755 }, { "epoch": 166.18691588785046, "grad_norm": 1.7109375, "learning_rate": 3.5138005271346643e-06, "loss": 0.1302, "step": 26756 }, { "epoch": 166.1931464174455, "grad_norm": 2.609375, "learning_rate": 3.512541098973596e-06, "loss": 0.1685, "step": 26757 }, { "epoch": 166.1993769470405, "grad_norm": 1.7109375, "learning_rate": 3.511281879503031e-06, "loss": 0.1223, "step": 26758 }, { "epoch": 166.20560747663552, "grad_norm": 2.875, "learning_rate": 3.5100228687351854e-06, "loss": 0.2376, "step": 26759 }, { "epoch": 166.21183800623052, "grad_norm": 2.15625, "learning_rate": 3.5087640666822936e-06, "loss": 0.1885, "step": 26760 }, { "epoch": 166.21806853582555, "grad_norm": 2.671875, "learning_rate": 3.507505473356587e-06, "loss": 0.195, "step": 26761 }, { "epoch": 166.22429906542055, "grad_norm": 2.40625, "learning_rate": 3.5062470887702843e-06, "loss": 0.1123, "step": 26762 }, { "epoch": 166.23052959501558, "grad_norm": 1.796875, "learning_rate": 3.5049889129356e-06, "loss": 0.1209, "step": 26763 }, { "epoch": 166.23676012461058, "grad_norm": 2.15625, "learning_rate": 3.503730945864761e-06, "loss": 0.1729, "step": 26764 }, { "epoch": 166.2429906542056, "grad_norm": 2.109375, "learning_rate": 3.5024731875699877e-06, "loss": 0.0981, "step": 26765 }, { "epoch": 166.2492211838006, "grad_norm": 1.125, "learning_rate": 3.5012156380634875e-06, "loss": 0.109, "step": 26766 }, { "epoch": 166.25545171339564, "grad_norm": 1.84375, "learning_rate": 3.4999582973574777e-06, "loss": 0.1541, "step": 26767 }, { "epoch": 166.26168224299064, "grad_norm": 2.609375, "learning_rate": 3.498701165464177e-06, "loss": 0.2542, "step": 26768 }, { "epoch": 166.26791277258567, "grad_norm": 1.109375, "learning_rate": 3.4974442423957892e-06, "loss": 0.1201, "step": 26769 }, { "epoch": 166.27414330218068, "grad_norm": 2.453125, "learning_rate": 3.4961875281645156e-06, "loss": 0.192, "step": 26770 }, { "epoch": 166.2803738317757, "grad_norm": 2.078125, "learning_rate": 3.494931022782569e-06, "loss": 0.1882, "step": 26771 }, { "epoch": 166.2866043613707, "grad_norm": 1.65625, "learning_rate": 3.4936747262621526e-06, "loss": 0.1328, "step": 26772 }, { "epoch": 166.29283489096574, "grad_norm": 1.6796875, "learning_rate": 3.4924186386154633e-06, "loss": 0.1268, "step": 26773 }, { "epoch": 166.29906542056074, "grad_norm": 2.078125, "learning_rate": 3.491162759854702e-06, "loss": 0.1355, "step": 26774 }, { "epoch": 166.30529595015577, "grad_norm": 2.03125, "learning_rate": 3.4899070899920755e-06, "loss": 0.1554, "step": 26775 }, { "epoch": 166.31152647975077, "grad_norm": 2.3125, "learning_rate": 3.4886516290397685e-06, "loss": 0.1723, "step": 26776 }, { "epoch": 166.3177570093458, "grad_norm": 1.3828125, "learning_rate": 3.4873963770099714e-06, "loss": 0.1379, "step": 26777 }, { "epoch": 166.3239875389408, "grad_norm": 2.25, "learning_rate": 3.486141333914883e-06, "loss": 0.1613, "step": 26778 }, { "epoch": 166.33021806853583, "grad_norm": 2.0, "learning_rate": 3.484886499766696e-06, "loss": 0.1491, "step": 26779 }, { "epoch": 166.33644859813083, "grad_norm": 2.078125, "learning_rate": 3.4836318745775835e-06, "loss": 0.141, "step": 26780 }, { "epoch": 166.34267912772586, "grad_norm": 1.1640625, "learning_rate": 3.482377458359748e-06, "loss": 0.1125, "step": 26781 }, { "epoch": 166.34890965732086, "grad_norm": 1.4609375, "learning_rate": 3.481123251125357e-06, "loss": 0.1337, "step": 26782 }, { "epoch": 166.3551401869159, "grad_norm": 1.734375, "learning_rate": 3.4798692528866057e-06, "loss": 0.1473, "step": 26783 }, { "epoch": 166.3613707165109, "grad_norm": 1.6484375, "learning_rate": 3.478615463655657e-06, "loss": 0.083, "step": 26784 }, { "epoch": 166.36760124610592, "grad_norm": 3.515625, "learning_rate": 3.4773618834446993e-06, "loss": 0.3055, "step": 26785 }, { "epoch": 166.37383177570092, "grad_norm": 1.3984375, "learning_rate": 3.476108512265905e-06, "loss": 0.1036, "step": 26786 }, { "epoch": 166.38006230529595, "grad_norm": 1.9921875, "learning_rate": 3.4748553501314536e-06, "loss": 0.1608, "step": 26787 }, { "epoch": 166.38629283489095, "grad_norm": 1.2109375, "learning_rate": 3.4736023970535102e-06, "loss": 0.1075, "step": 26788 }, { "epoch": 166.39252336448598, "grad_norm": 2.03125, "learning_rate": 3.4723496530442352e-06, "loss": 0.1677, "step": 26789 }, { "epoch": 166.39875389408098, "grad_norm": 1.1875, "learning_rate": 3.471097118115807e-06, "loss": 0.1082, "step": 26790 }, { "epoch": 166.404984423676, "grad_norm": 3.03125, "learning_rate": 3.4698447922803883e-06, "loss": 0.2368, "step": 26791 }, { "epoch": 166.41121495327104, "grad_norm": 1.9375, "learning_rate": 3.468592675550139e-06, "loss": 0.1225, "step": 26792 }, { "epoch": 166.41744548286604, "grad_norm": 1.6484375, "learning_rate": 3.4673407679372216e-06, "loss": 0.1353, "step": 26793 }, { "epoch": 166.42367601246107, "grad_norm": 1.9140625, "learning_rate": 3.4660890694537985e-06, "loss": 0.1746, "step": 26794 }, { "epoch": 166.42990654205607, "grad_norm": 1.4375, "learning_rate": 3.4648375801120237e-06, "loss": 0.1114, "step": 26795 }, { "epoch": 166.4361370716511, "grad_norm": 2.953125, "learning_rate": 3.4635862999240457e-06, "loss": 0.2246, "step": 26796 }, { "epoch": 166.4423676012461, "grad_norm": 1.7734375, "learning_rate": 3.462335228902022e-06, "loss": 0.1213, "step": 26797 }, { "epoch": 166.44859813084113, "grad_norm": 1.1875, "learning_rate": 3.4610843670581094e-06, "loss": 0.0901, "step": 26798 }, { "epoch": 166.45482866043614, "grad_norm": 1.5234375, "learning_rate": 3.459833714404448e-06, "loss": 0.1323, "step": 26799 }, { "epoch": 166.46105919003116, "grad_norm": 1.953125, "learning_rate": 3.458583270953189e-06, "loss": 0.1446, "step": 26800 }, { "epoch": 166.46728971962617, "grad_norm": 2.203125, "learning_rate": 3.457333036716473e-06, "loss": 0.1778, "step": 26801 }, { "epoch": 166.4735202492212, "grad_norm": 3.171875, "learning_rate": 3.4560830117064485e-06, "loss": 0.3215, "step": 26802 }, { "epoch": 166.4797507788162, "grad_norm": 0.89453125, "learning_rate": 3.4548331959352474e-06, "loss": 0.0941, "step": 26803 }, { "epoch": 166.48598130841123, "grad_norm": 1.9375, "learning_rate": 3.453583589415013e-06, "loss": 0.2038, "step": 26804 }, { "epoch": 166.49221183800623, "grad_norm": 2.8125, "learning_rate": 3.4523341921578885e-06, "loss": 0.2239, "step": 26805 }, { "epoch": 166.49844236760126, "grad_norm": 1.828125, "learning_rate": 3.451085004175994e-06, "loss": 0.1475, "step": 26806 }, { "epoch": 166.50467289719626, "grad_norm": 1.84375, "learning_rate": 3.449836025481476e-06, "loss": 0.083, "step": 26807 }, { "epoch": 166.5109034267913, "grad_norm": 1.75, "learning_rate": 3.4485872560864522e-06, "loss": 0.1366, "step": 26808 }, { "epoch": 166.5171339563863, "grad_norm": 2.984375, "learning_rate": 3.4473386960030635e-06, "loss": 0.2831, "step": 26809 }, { "epoch": 166.52336448598132, "grad_norm": 1.3046875, "learning_rate": 3.4460903452434245e-06, "loss": 0.1184, "step": 26810 }, { "epoch": 166.52959501557632, "grad_norm": 2.0625, "learning_rate": 3.444842203819662e-06, "loss": 0.1778, "step": 26811 }, { "epoch": 166.53582554517135, "grad_norm": 1.7421875, "learning_rate": 3.4435942717439105e-06, "loss": 0.129, "step": 26812 }, { "epoch": 166.54205607476635, "grad_norm": 0.89453125, "learning_rate": 3.4423465490282776e-06, "loss": 0.1035, "step": 26813 }, { "epoch": 166.54828660436138, "grad_norm": 1.984375, "learning_rate": 3.4410990356848778e-06, "loss": 0.0894, "step": 26814 }, { "epoch": 166.55451713395638, "grad_norm": 1.5390625, "learning_rate": 3.439851731725835e-06, "loss": 0.1449, "step": 26815 }, { "epoch": 166.5607476635514, "grad_norm": 2.0, "learning_rate": 3.43860463716327e-06, "loss": 0.1623, "step": 26816 }, { "epoch": 166.5669781931464, "grad_norm": 3.09375, "learning_rate": 3.437357752009279e-06, "loss": 0.2016, "step": 26817 }, { "epoch": 166.57320872274144, "grad_norm": 1.375, "learning_rate": 3.4361110762759773e-06, "loss": 0.1246, "step": 26818 }, { "epoch": 166.57943925233644, "grad_norm": 1.5, "learning_rate": 3.43486460997548e-06, "loss": 0.0911, "step": 26819 }, { "epoch": 166.58566978193147, "grad_norm": 1.7265625, "learning_rate": 3.4336183531198967e-06, "loss": 0.0973, "step": 26820 }, { "epoch": 166.59190031152647, "grad_norm": 1.3828125, "learning_rate": 3.4323723057213127e-06, "loss": 0.1199, "step": 26821 }, { "epoch": 166.5981308411215, "grad_norm": 1.5625, "learning_rate": 3.43112646779184e-06, "loss": 0.0913, "step": 26822 }, { "epoch": 166.6043613707165, "grad_norm": 2.21875, "learning_rate": 3.4298808393435773e-06, "loss": 0.2564, "step": 26823 }, { "epoch": 166.61059190031153, "grad_norm": 3.46875, "learning_rate": 3.4286354203886317e-06, "loss": 0.2732, "step": 26824 }, { "epoch": 166.61682242990653, "grad_norm": 2.421875, "learning_rate": 3.427390210939083e-06, "loss": 0.1958, "step": 26825 }, { "epoch": 166.62305295950156, "grad_norm": 2.125, "learning_rate": 3.4261452110070396e-06, "loss": 0.213, "step": 26826 }, { "epoch": 166.62928348909657, "grad_norm": 1.1953125, "learning_rate": 3.424900420604582e-06, "loss": 0.1186, "step": 26827 }, { "epoch": 166.6355140186916, "grad_norm": 1.7578125, "learning_rate": 3.4236558397438084e-06, "loss": 0.1471, "step": 26828 }, { "epoch": 166.6417445482866, "grad_norm": 1.9921875, "learning_rate": 3.422411468436798e-06, "loss": 0.1331, "step": 26829 }, { "epoch": 166.64797507788163, "grad_norm": 1.9609375, "learning_rate": 3.4211673066956382e-06, "loss": 0.173, "step": 26830 }, { "epoch": 166.65420560747663, "grad_norm": 2.203125, "learning_rate": 3.4199233545324228e-06, "loss": 0.1507, "step": 26831 }, { "epoch": 166.66043613707166, "grad_norm": 1.5625, "learning_rate": 3.4186796119592244e-06, "loss": 0.1194, "step": 26832 }, { "epoch": 166.66666666666666, "grad_norm": 1.046875, "learning_rate": 3.4174360789881206e-06, "loss": 0.1099, "step": 26833 }, { "epoch": 166.6728971962617, "grad_norm": 2.28125, "learning_rate": 3.4161927556311896e-06, "loss": 0.1707, "step": 26834 }, { "epoch": 166.6791277258567, "grad_norm": 2.515625, "learning_rate": 3.414949641900514e-06, "loss": 0.2763, "step": 26835 }, { "epoch": 166.68535825545172, "grad_norm": 1.1484375, "learning_rate": 3.413706737808156e-06, "loss": 0.1282, "step": 26836 }, { "epoch": 166.69158878504672, "grad_norm": 3.34375, "learning_rate": 3.4124640433661953e-06, "loss": 0.2708, "step": 26837 }, { "epoch": 166.69781931464175, "grad_norm": 1.65625, "learning_rate": 3.411221558586705e-06, "loss": 0.1306, "step": 26838 }, { "epoch": 166.70404984423675, "grad_norm": 2.453125, "learning_rate": 3.4099792834817456e-06, "loss": 0.1828, "step": 26839 }, { "epoch": 166.71028037383178, "grad_norm": 1.7734375, "learning_rate": 3.4087372180633765e-06, "loss": 0.1578, "step": 26840 }, { "epoch": 166.71651090342678, "grad_norm": 1.7265625, "learning_rate": 3.4074953623436683e-06, "loss": 0.1293, "step": 26841 }, { "epoch": 166.7227414330218, "grad_norm": 1.5390625, "learning_rate": 3.406253716334687e-06, "loss": 0.1333, "step": 26842 }, { "epoch": 166.7289719626168, "grad_norm": 1.171875, "learning_rate": 3.405012280048481e-06, "loss": 0.1189, "step": 26843 }, { "epoch": 166.73520249221184, "grad_norm": 0.92578125, "learning_rate": 3.4037710534971125e-06, "loss": 0.0929, "step": 26844 }, { "epoch": 166.74143302180684, "grad_norm": 2.984375, "learning_rate": 3.4025300366926414e-06, "loss": 0.1929, "step": 26845 }, { "epoch": 166.74766355140187, "grad_norm": 1.2734375, "learning_rate": 3.4012892296471172e-06, "loss": 0.1067, "step": 26846 }, { "epoch": 166.75389408099687, "grad_norm": 1.828125, "learning_rate": 3.400048632372582e-06, "loss": 0.1129, "step": 26847 }, { "epoch": 166.7601246105919, "grad_norm": 2.21875, "learning_rate": 3.3988082448810936e-06, "loss": 0.2821, "step": 26848 }, { "epoch": 166.7663551401869, "grad_norm": 1.515625, "learning_rate": 3.397568067184698e-06, "loss": 0.1503, "step": 26849 }, { "epoch": 166.77258566978193, "grad_norm": 1.8046875, "learning_rate": 3.3963280992954425e-06, "loss": 0.1321, "step": 26850 }, { "epoch": 166.77881619937693, "grad_norm": 1.5703125, "learning_rate": 3.395088341225369e-06, "loss": 0.1317, "step": 26851 }, { "epoch": 166.78504672897196, "grad_norm": 1.484375, "learning_rate": 3.393848792986512e-06, "loss": 0.1542, "step": 26852 }, { "epoch": 166.79127725856696, "grad_norm": 2.15625, "learning_rate": 3.3926094545909115e-06, "loss": 0.1814, "step": 26853 }, { "epoch": 166.797507788162, "grad_norm": 1.328125, "learning_rate": 3.391370326050616e-06, "loss": 0.1021, "step": 26854 }, { "epoch": 166.80373831775702, "grad_norm": 2.859375, "learning_rate": 3.390131407377642e-06, "loss": 0.2192, "step": 26855 }, { "epoch": 166.80996884735202, "grad_norm": 1.390625, "learning_rate": 3.388892698584034e-06, "loss": 0.1431, "step": 26856 }, { "epoch": 166.81619937694705, "grad_norm": 3.171875, "learning_rate": 3.3876541996818253e-06, "loss": 0.2206, "step": 26857 }, { "epoch": 166.82242990654206, "grad_norm": 1.640625, "learning_rate": 3.3864159106830383e-06, "loss": 0.1421, "step": 26858 }, { "epoch": 166.82866043613708, "grad_norm": 1.7109375, "learning_rate": 3.385177831599698e-06, "loss": 0.1345, "step": 26859 }, { "epoch": 166.8348909657321, "grad_norm": 1.046875, "learning_rate": 3.3839399624438273e-06, "loss": 0.1038, "step": 26860 }, { "epoch": 166.84112149532712, "grad_norm": 2.21875, "learning_rate": 3.382702303227461e-06, "loss": 0.2369, "step": 26861 }, { "epoch": 166.84735202492212, "grad_norm": 1.3046875, "learning_rate": 3.3814648539626043e-06, "loss": 0.1044, "step": 26862 }, { "epoch": 166.85358255451715, "grad_norm": 1.203125, "learning_rate": 3.3802276146612827e-06, "loss": 0.1191, "step": 26863 }, { "epoch": 166.85981308411215, "grad_norm": 1.3046875, "learning_rate": 3.3789905853355176e-06, "loss": 0.1114, "step": 26864 }, { "epoch": 166.86604361370718, "grad_norm": 1.109375, "learning_rate": 3.3777537659973186e-06, "loss": 0.0933, "step": 26865 }, { "epoch": 166.87227414330218, "grad_norm": 1.8515625, "learning_rate": 3.3765171566586903e-06, "loss": 0.1079, "step": 26866 }, { "epoch": 166.8785046728972, "grad_norm": 2.5, "learning_rate": 3.3752807573316503e-06, "loss": 0.224, "step": 26867 }, { "epoch": 166.8847352024922, "grad_norm": 2.09375, "learning_rate": 3.374044568028212e-06, "loss": 0.1691, "step": 26868 }, { "epoch": 166.89096573208724, "grad_norm": 3.015625, "learning_rate": 3.3728085887603704e-06, "loss": 0.2093, "step": 26869 }, { "epoch": 166.89719626168224, "grad_norm": 1.734375, "learning_rate": 3.371572819540142e-06, "loss": 0.1229, "step": 26870 }, { "epoch": 166.90342679127727, "grad_norm": 1.6015625, "learning_rate": 3.3703372603795138e-06, "loss": 0.1368, "step": 26871 }, { "epoch": 166.90965732087227, "grad_norm": 1.859375, "learning_rate": 3.369101911290501e-06, "loss": 0.156, "step": 26872 }, { "epoch": 166.9158878504673, "grad_norm": 1.4375, "learning_rate": 3.367866772285086e-06, "loss": 0.1082, "step": 26873 }, { "epoch": 166.9221183800623, "grad_norm": 2.40625, "learning_rate": 3.366631843375276e-06, "loss": 0.1744, "step": 26874 }, { "epoch": 166.92834890965733, "grad_norm": 1.9296875, "learning_rate": 3.3653971245730666e-06, "loss": 0.1458, "step": 26875 }, { "epoch": 166.93457943925233, "grad_norm": 1.46875, "learning_rate": 3.36416261589044e-06, "loss": 0.1495, "step": 26876 }, { "epoch": 166.94080996884736, "grad_norm": 1.328125, "learning_rate": 3.362928317339395e-06, "loss": 0.1013, "step": 26877 }, { "epoch": 166.94704049844236, "grad_norm": 2.5, "learning_rate": 3.361694228931911e-06, "loss": 0.1716, "step": 26878 }, { "epoch": 166.9532710280374, "grad_norm": 2.0625, "learning_rate": 3.360460350679984e-06, "loss": 0.1562, "step": 26879 }, { "epoch": 166.9595015576324, "grad_norm": 1.34375, "learning_rate": 3.359226682595587e-06, "loss": 0.1422, "step": 26880 }, { "epoch": 166.96573208722742, "grad_norm": 2.109375, "learning_rate": 3.357993224690703e-06, "loss": 0.2058, "step": 26881 }, { "epoch": 166.97196261682242, "grad_norm": 1.0625, "learning_rate": 3.3567599769773166e-06, "loss": 0.0909, "step": 26882 }, { "epoch": 166.97819314641745, "grad_norm": 2.71875, "learning_rate": 3.3555269394674155e-06, "loss": 0.1895, "step": 26883 }, { "epoch": 166.98442367601245, "grad_norm": 2.171875, "learning_rate": 3.3542941121729536e-06, "loss": 0.1452, "step": 26884 }, { "epoch": 166.99065420560748, "grad_norm": 2.703125, "learning_rate": 3.3530614951059108e-06, "loss": 0.2272, "step": 26885 }, { "epoch": 166.99688473520249, "grad_norm": 2.25, "learning_rate": 3.351829088278266e-06, "loss": 0.1846, "step": 26886 }, { "epoch": 167.0, "grad_norm": 0.70703125, "learning_rate": 3.3505968917019874e-06, "loss": 0.0515, "step": 26887 }, { "epoch": 167.00623052959503, "grad_norm": 1.0625, "learning_rate": 3.3493649053890326e-06, "loss": 0.0972, "step": 26888 }, { "epoch": 167.01246105919003, "grad_norm": 2.75, "learning_rate": 3.348133129351377e-06, "loss": 0.2396, "step": 26889 }, { "epoch": 167.01869158878506, "grad_norm": 1.4296875, "learning_rate": 3.3469015636009846e-06, "loss": 0.1393, "step": 26890 }, { "epoch": 167.02492211838006, "grad_norm": 1.6484375, "learning_rate": 3.345670208149812e-06, "loss": 0.1619, "step": 26891 }, { "epoch": 167.0311526479751, "grad_norm": 2.671875, "learning_rate": 3.3444390630098133e-06, "loss": 0.23, "step": 26892 }, { "epoch": 167.0373831775701, "grad_norm": 2.890625, "learning_rate": 3.343208128192951e-06, "loss": 0.257, "step": 26893 }, { "epoch": 167.04361370716512, "grad_norm": 2.5, "learning_rate": 3.3419774037111886e-06, "loss": 0.191, "step": 26894 }, { "epoch": 167.04984423676012, "grad_norm": 2.078125, "learning_rate": 3.3407468895764625e-06, "loss": 0.2261, "step": 26895 }, { "epoch": 167.05607476635515, "grad_norm": 2.140625, "learning_rate": 3.3395165858007395e-06, "loss": 0.1436, "step": 26896 }, { "epoch": 167.06230529595015, "grad_norm": 1.4453125, "learning_rate": 3.338286492395953e-06, "loss": 0.1094, "step": 26897 }, { "epoch": 167.06853582554518, "grad_norm": 2.28125, "learning_rate": 3.3370566093740674e-06, "loss": 0.175, "step": 26898 }, { "epoch": 167.07476635514018, "grad_norm": 2.09375, "learning_rate": 3.335826936747011e-06, "loss": 0.1807, "step": 26899 }, { "epoch": 167.0809968847352, "grad_norm": 2.828125, "learning_rate": 3.334597474526735e-06, "loss": 0.2322, "step": 26900 }, { "epoch": 167.08722741433021, "grad_norm": 1.5390625, "learning_rate": 3.3333682227251833e-06, "loss": 0.122, "step": 26901 }, { "epoch": 167.09345794392524, "grad_norm": 2.0, "learning_rate": 3.3321391813542906e-06, "loss": 0.1739, "step": 26902 }, { "epoch": 167.09968847352025, "grad_norm": 2.015625, "learning_rate": 3.330910350425989e-06, "loss": 0.1357, "step": 26903 }, { "epoch": 167.10591900311528, "grad_norm": 2.625, "learning_rate": 3.3296817299522164e-06, "loss": 0.2002, "step": 26904 }, { "epoch": 167.11214953271028, "grad_norm": 2.28125, "learning_rate": 3.328453319944913e-06, "loss": 0.1885, "step": 26905 }, { "epoch": 167.1183800623053, "grad_norm": 1.6015625, "learning_rate": 3.327225120415997e-06, "loss": 0.1152, "step": 26906 }, { "epoch": 167.1246105919003, "grad_norm": 1.9375, "learning_rate": 3.3259971313774034e-06, "loss": 0.1538, "step": 26907 }, { "epoch": 167.13084112149534, "grad_norm": 1.8046875, "learning_rate": 3.3247693528410645e-06, "loss": 0.1086, "step": 26908 }, { "epoch": 167.13707165109034, "grad_norm": 2.625, "learning_rate": 3.3235417848188983e-06, "loss": 0.2354, "step": 26909 }, { "epoch": 167.14330218068537, "grad_norm": 1.3671875, "learning_rate": 3.32231442732282e-06, "loss": 0.1083, "step": 26910 }, { "epoch": 167.14953271028037, "grad_norm": 2.65625, "learning_rate": 3.321087280364757e-06, "loss": 0.2198, "step": 26911 }, { "epoch": 167.1557632398754, "grad_norm": 1.515625, "learning_rate": 3.319860343956635e-06, "loss": 0.1327, "step": 26912 }, { "epoch": 167.1619937694704, "grad_norm": 3.25, "learning_rate": 3.3186336181103587e-06, "loss": 0.2057, "step": 26913 }, { "epoch": 167.16822429906543, "grad_norm": 2.15625, "learning_rate": 3.317407102837844e-06, "loss": 0.1523, "step": 26914 }, { "epoch": 167.17445482866043, "grad_norm": 2.328125, "learning_rate": 3.316180798151014e-06, "loss": 0.1884, "step": 26915 }, { "epoch": 167.18068535825546, "grad_norm": 1.6796875, "learning_rate": 3.314954704061768e-06, "loss": 0.1394, "step": 26916 }, { "epoch": 167.18691588785046, "grad_norm": 2.015625, "learning_rate": 3.313728820582013e-06, "loss": 0.1793, "step": 26917 }, { "epoch": 167.1931464174455, "grad_norm": 2.28125, "learning_rate": 3.3125031477236585e-06, "loss": 0.2046, "step": 26918 }, { "epoch": 167.1993769470405, "grad_norm": 2.171875, "learning_rate": 3.3112776854986066e-06, "loss": 0.1568, "step": 26919 }, { "epoch": 167.20560747663552, "grad_norm": 1.59375, "learning_rate": 3.31005243391877e-06, "loss": 0.1273, "step": 26920 }, { "epoch": 167.21183800623052, "grad_norm": 1.6875, "learning_rate": 3.308827392996039e-06, "loss": 0.1317, "step": 26921 }, { "epoch": 167.21806853582555, "grad_norm": 1.9609375, "learning_rate": 3.307602562742304e-06, "loss": 0.1724, "step": 26922 }, { "epoch": 167.22429906542055, "grad_norm": 2.796875, "learning_rate": 3.3063779431694726e-06, "loss": 0.3186, "step": 26923 }, { "epoch": 167.23052959501558, "grad_norm": 1.5078125, "learning_rate": 3.3051535342894374e-06, "loss": 0.1367, "step": 26924 }, { "epoch": 167.23676012461058, "grad_norm": 1.9140625, "learning_rate": 3.303929336114087e-06, "loss": 0.1557, "step": 26925 }, { "epoch": 167.2429906542056, "grad_norm": 2.96875, "learning_rate": 3.302705348655308e-06, "loss": 0.2056, "step": 26926 }, { "epoch": 167.2492211838006, "grad_norm": 3.328125, "learning_rate": 3.301481571924997e-06, "loss": 0.3319, "step": 26927 }, { "epoch": 167.25545171339564, "grad_norm": 1.8984375, "learning_rate": 3.3002580059350363e-06, "loss": 0.1205, "step": 26928 }, { "epoch": 167.26168224299064, "grad_norm": 1.4609375, "learning_rate": 3.2990346506973023e-06, "loss": 0.1031, "step": 26929 }, { "epoch": 167.26791277258567, "grad_norm": 2.140625, "learning_rate": 3.2978115062236803e-06, "loss": 0.1946, "step": 26930 }, { "epoch": 167.27414330218068, "grad_norm": 2.828125, "learning_rate": 3.2965885725260574e-06, "loss": 0.2024, "step": 26931 }, { "epoch": 167.2803738317757, "grad_norm": 1.8671875, "learning_rate": 3.2953658496163e-06, "loss": 0.1518, "step": 26932 }, { "epoch": 167.2866043613707, "grad_norm": 1.5390625, "learning_rate": 3.2941433375062863e-06, "loss": 0.1335, "step": 26933 }, { "epoch": 167.29283489096574, "grad_norm": 3.421875, "learning_rate": 3.2929210362078995e-06, "loss": 0.3055, "step": 26934 }, { "epoch": 167.29906542056074, "grad_norm": 1.7109375, "learning_rate": 3.2916989457329987e-06, "loss": 0.1689, "step": 26935 }, { "epoch": 167.30529595015577, "grad_norm": 2.0625, "learning_rate": 3.290477066093456e-06, "loss": 0.1289, "step": 26936 }, { "epoch": 167.31152647975077, "grad_norm": 1.4375, "learning_rate": 3.2892553973011358e-06, "loss": 0.1243, "step": 26937 }, { "epoch": 167.3177570093458, "grad_norm": 2.984375, "learning_rate": 3.2880339393679156e-06, "loss": 0.182, "step": 26938 }, { "epoch": 167.3239875389408, "grad_norm": 2.75, "learning_rate": 3.2868126923056437e-06, "loss": 0.179, "step": 26939 }, { "epoch": 167.33021806853583, "grad_norm": 3.078125, "learning_rate": 3.285591656126191e-06, "loss": 0.3344, "step": 26940 }, { "epoch": 167.33644859813083, "grad_norm": 2.3125, "learning_rate": 3.28437083084141e-06, "loss": 0.2163, "step": 26941 }, { "epoch": 167.34267912772586, "grad_norm": 1.0390625, "learning_rate": 3.2831502164631627e-06, "loss": 0.1057, "step": 26942 }, { "epoch": 167.34890965732086, "grad_norm": 1.96875, "learning_rate": 3.2819298130032983e-06, "loss": 0.1516, "step": 26943 }, { "epoch": 167.3551401869159, "grad_norm": 1.859375, "learning_rate": 3.280709620473674e-06, "loss": 0.1645, "step": 26944 }, { "epoch": 167.3613707165109, "grad_norm": 2.65625, "learning_rate": 3.279489638886138e-06, "loss": 0.2859, "step": 26945 }, { "epoch": 167.36760124610592, "grad_norm": 2.625, "learning_rate": 3.2782698682525454e-06, "loss": 0.2022, "step": 26946 }, { "epoch": 167.37383177570092, "grad_norm": 1.7421875, "learning_rate": 3.277050308584739e-06, "loss": 0.1663, "step": 26947 }, { "epoch": 167.38006230529595, "grad_norm": 2.71875, "learning_rate": 3.2758309598945547e-06, "loss": 0.2626, "step": 26948 }, { "epoch": 167.38629283489095, "grad_norm": 1.4921875, "learning_rate": 3.274611822193846e-06, "loss": 0.1263, "step": 26949 }, { "epoch": 167.39252336448598, "grad_norm": 1.4765625, "learning_rate": 3.273392895494454e-06, "loss": 0.1238, "step": 26950 }, { "epoch": 167.39875389408098, "grad_norm": 0.8828125, "learning_rate": 3.2721741798082073e-06, "loss": 0.1041, "step": 26951 }, { "epoch": 167.404984423676, "grad_norm": 1.78125, "learning_rate": 3.270955675146947e-06, "loss": 0.1559, "step": 26952 }, { "epoch": 167.41121495327104, "grad_norm": 2.75, "learning_rate": 3.269737381522517e-06, "loss": 0.2202, "step": 26953 }, { "epoch": 167.41744548286604, "grad_norm": 1.734375, "learning_rate": 3.26851929894674e-06, "loss": 0.1627, "step": 26954 }, { "epoch": 167.42367601246107, "grad_norm": 2.6875, "learning_rate": 3.2673014274314432e-06, "loss": 0.1313, "step": 26955 }, { "epoch": 167.42990654205607, "grad_norm": 2.078125, "learning_rate": 3.2660837669884592e-06, "loss": 0.1253, "step": 26956 }, { "epoch": 167.4361370716511, "grad_norm": 1.8359375, "learning_rate": 3.264866317629617e-06, "loss": 0.1511, "step": 26957 }, { "epoch": 167.4423676012461, "grad_norm": 1.1953125, "learning_rate": 3.2636490793667346e-06, "loss": 0.1177, "step": 26958 }, { "epoch": 167.44859813084113, "grad_norm": 1.78125, "learning_rate": 3.2624320522116364e-06, "loss": 0.2502, "step": 26959 }, { "epoch": 167.45482866043614, "grad_norm": 2.125, "learning_rate": 3.2612152361761515e-06, "loss": 0.1803, "step": 26960 }, { "epoch": 167.46105919003116, "grad_norm": 2.359375, "learning_rate": 3.2599986312720877e-06, "loss": 0.2114, "step": 26961 }, { "epoch": 167.46728971962617, "grad_norm": 3.203125, "learning_rate": 3.258782237511257e-06, "loss": 0.2333, "step": 26962 }, { "epoch": 167.4735202492212, "grad_norm": 1.609375, "learning_rate": 3.257566054905478e-06, "loss": 0.1501, "step": 26963 }, { "epoch": 167.4797507788162, "grad_norm": 1.5390625, "learning_rate": 3.256350083466572e-06, "loss": 0.1571, "step": 26964 }, { "epoch": 167.48598130841123, "grad_norm": 1.921875, "learning_rate": 3.2551343232063326e-06, "loss": 0.121, "step": 26965 }, { "epoch": 167.49221183800623, "grad_norm": 2.359375, "learning_rate": 3.253918774136583e-06, "loss": 0.1888, "step": 26966 }, { "epoch": 167.49844236760126, "grad_norm": 2.015625, "learning_rate": 3.2527034362691138e-06, "loss": 0.1596, "step": 26967 }, { "epoch": 167.50467289719626, "grad_norm": 2.65625, "learning_rate": 3.2514883096157438e-06, "loss": 0.1161, "step": 26968 }, { "epoch": 167.5109034267913, "grad_norm": 2.71875, "learning_rate": 3.2502733941882606e-06, "loss": 0.2386, "step": 26969 }, { "epoch": 167.5171339563863, "grad_norm": 2.0625, "learning_rate": 3.249058689998469e-06, "loss": 0.1793, "step": 26970 }, { "epoch": 167.52336448598132, "grad_norm": 1.1015625, "learning_rate": 3.2478441970581723e-06, "loss": 0.1224, "step": 26971 }, { "epoch": 167.52959501557632, "grad_norm": 1.6953125, "learning_rate": 3.246629915379162e-06, "loss": 0.1319, "step": 26972 }, { "epoch": 167.53582554517135, "grad_norm": 1.6875, "learning_rate": 3.2454158449732237e-06, "loss": 0.1627, "step": 26973 }, { "epoch": 167.54205607476635, "grad_norm": 3.0, "learning_rate": 3.2442019858521576e-06, "loss": 0.2339, "step": 26974 }, { "epoch": 167.54828660436138, "grad_norm": 1.8359375, "learning_rate": 3.2429883380277526e-06, "loss": 0.1528, "step": 26975 }, { "epoch": 167.55451713395638, "grad_norm": 3.1875, "learning_rate": 3.2417749015117933e-06, "loss": 0.309, "step": 26976 }, { "epoch": 167.5607476635514, "grad_norm": 3.140625, "learning_rate": 3.2405616763160617e-06, "loss": 0.1616, "step": 26977 }, { "epoch": 167.5669781931464, "grad_norm": 2.71875, "learning_rate": 3.2393486624523463e-06, "loss": 0.284, "step": 26978 }, { "epoch": 167.57320872274144, "grad_norm": 1.2890625, "learning_rate": 3.238135859932437e-06, "loss": 0.1374, "step": 26979 }, { "epoch": 167.57943925233644, "grad_norm": 2.203125, "learning_rate": 3.236923268768091e-06, "loss": 0.1245, "step": 26980 }, { "epoch": 167.58566978193147, "grad_norm": 3.328125, "learning_rate": 3.235710888971097e-06, "loss": 0.2186, "step": 26981 }, { "epoch": 167.59190031152647, "grad_norm": 1.7578125, "learning_rate": 3.234498720553231e-06, "loss": 0.1308, "step": 26982 }, { "epoch": 167.5981308411215, "grad_norm": 1.25, "learning_rate": 3.2332867635262666e-06, "loss": 0.1061, "step": 26983 }, { "epoch": 167.6043613707165, "grad_norm": 2.078125, "learning_rate": 3.23207501790197e-06, "loss": 0.2267, "step": 26984 }, { "epoch": 167.61059190031153, "grad_norm": 2.96875, "learning_rate": 3.230863483692115e-06, "loss": 0.2881, "step": 26985 }, { "epoch": 167.61682242990653, "grad_norm": 1.1171875, "learning_rate": 3.2296521609084613e-06, "loss": 0.0808, "step": 26986 }, { "epoch": 167.62305295950156, "grad_norm": 3.34375, "learning_rate": 3.228441049562783e-06, "loss": 0.2293, "step": 26987 }, { "epoch": 167.62928348909657, "grad_norm": 1.859375, "learning_rate": 3.2272301496668324e-06, "loss": 0.1554, "step": 26988 }, { "epoch": 167.6355140186916, "grad_norm": 1.0703125, "learning_rate": 3.226019461232374e-06, "loss": 0.103, "step": 26989 }, { "epoch": 167.6417445482866, "grad_norm": 2.59375, "learning_rate": 3.2248089842711717e-06, "loss": 0.269, "step": 26990 }, { "epoch": 167.64797507788163, "grad_norm": 2.171875, "learning_rate": 3.223598718794979e-06, "loss": 0.2025, "step": 26991 }, { "epoch": 167.65420560747663, "grad_norm": 2.203125, "learning_rate": 3.2223886648155426e-06, "loss": 0.1402, "step": 26992 }, { "epoch": 167.66043613707166, "grad_norm": 2.21875, "learning_rate": 3.22117882234462e-06, "loss": 0.161, "step": 26993 }, { "epoch": 167.66666666666666, "grad_norm": 2.09375, "learning_rate": 3.219969191393968e-06, "loss": 0.1482, "step": 26994 }, { "epoch": 167.6728971962617, "grad_norm": 1.078125, "learning_rate": 3.2187597719753244e-06, "loss": 0.101, "step": 26995 }, { "epoch": 167.6791277258567, "grad_norm": 2.234375, "learning_rate": 3.2175505641004384e-06, "loss": 0.2188, "step": 26996 }, { "epoch": 167.68535825545172, "grad_norm": 2.015625, "learning_rate": 3.2163415677810615e-06, "loss": 0.1654, "step": 26997 }, { "epoch": 167.69158878504672, "grad_norm": 1.3515625, "learning_rate": 3.2151327830289316e-06, "loss": 0.1209, "step": 26998 }, { "epoch": 167.69781931464175, "grad_norm": 1.5859375, "learning_rate": 3.213924209855779e-06, "loss": 0.1663, "step": 26999 }, { "epoch": 167.70404984423675, "grad_norm": 1.4140625, "learning_rate": 3.212715848273351e-06, "loss": 0.1289, "step": 27000 }, { "epoch": 167.71028037383178, "grad_norm": 1.875, "learning_rate": 3.2115076982933866e-06, "loss": 0.24, "step": 27001 }, { "epoch": 167.71651090342678, "grad_norm": 1.6953125, "learning_rate": 3.2102997599276096e-06, "loss": 0.1402, "step": 27002 }, { "epoch": 167.7227414330218, "grad_norm": 1.3125, "learning_rate": 3.209092033187755e-06, "loss": 0.0995, "step": 27003 }, { "epoch": 167.7289719626168, "grad_norm": 1.8046875, "learning_rate": 3.2078845180855632e-06, "loss": 0.1106, "step": 27004 }, { "epoch": 167.73520249221184, "grad_norm": 2.671875, "learning_rate": 3.20667721463275e-06, "loss": 0.2357, "step": 27005 }, { "epoch": 167.74143302180684, "grad_norm": 1.53125, "learning_rate": 3.205470122841042e-06, "loss": 0.1335, "step": 27006 }, { "epoch": 167.74766355140187, "grad_norm": 2.0625, "learning_rate": 3.2042632427221636e-06, "loss": 0.2149, "step": 27007 }, { "epoch": 167.75389408099687, "grad_norm": 1.609375, "learning_rate": 3.203056574287841e-06, "loss": 0.1382, "step": 27008 }, { "epoch": 167.7601246105919, "grad_norm": 0.7734375, "learning_rate": 3.201850117549787e-06, "loss": 0.0871, "step": 27009 }, { "epoch": 167.7663551401869, "grad_norm": 2.5, "learning_rate": 3.2006438725197206e-06, "loss": 0.15, "step": 27010 }, { "epoch": 167.77258566978193, "grad_norm": 1.796875, "learning_rate": 3.1994378392093653e-06, "loss": 0.1376, "step": 27011 }, { "epoch": 167.77881619937693, "grad_norm": 1.6796875, "learning_rate": 3.198232017630426e-06, "loss": 0.1589, "step": 27012 }, { "epoch": 167.78504672897196, "grad_norm": 2.078125, "learning_rate": 3.1970264077946095e-06, "loss": 0.147, "step": 27013 }, { "epoch": 167.79127725856696, "grad_norm": 2.265625, "learning_rate": 3.1958210097136314e-06, "loss": 0.1027, "step": 27014 }, { "epoch": 167.797507788162, "grad_norm": 2.203125, "learning_rate": 3.1946158233991996e-06, "loss": 0.2504, "step": 27015 }, { "epoch": 167.80373831775702, "grad_norm": 2.40625, "learning_rate": 3.193410848863024e-06, "loss": 0.1822, "step": 27016 }, { "epoch": 167.80996884735202, "grad_norm": 0.9765625, "learning_rate": 3.1922060861167973e-06, "loss": 0.1179, "step": 27017 }, { "epoch": 167.81619937694705, "grad_norm": 2.46875, "learning_rate": 3.1910015351722223e-06, "loss": 0.3484, "step": 27018 }, { "epoch": 167.82242990654206, "grad_norm": 2.453125, "learning_rate": 3.189797196041e-06, "loss": 0.2011, "step": 27019 }, { "epoch": 167.82866043613708, "grad_norm": 0.8046875, "learning_rate": 3.1885930687348296e-06, "loss": 0.0898, "step": 27020 }, { "epoch": 167.8348909657321, "grad_norm": 1.765625, "learning_rate": 3.1873891532654022e-06, "loss": 0.1289, "step": 27021 }, { "epoch": 167.84112149532712, "grad_norm": 2.40625, "learning_rate": 3.1861854496444076e-06, "loss": 0.2116, "step": 27022 }, { "epoch": 167.84735202492212, "grad_norm": 2.046875, "learning_rate": 3.184981957883548e-06, "loss": 0.1373, "step": 27023 }, { "epoch": 167.85358255451715, "grad_norm": 0.8125, "learning_rate": 3.1837786779945035e-06, "loss": 0.0823, "step": 27024 }, { "epoch": 167.85981308411215, "grad_norm": 2.765625, "learning_rate": 3.182575609988955e-06, "loss": 0.1442, "step": 27025 }, { "epoch": 167.86604361370718, "grad_norm": 1.5703125, "learning_rate": 3.181372753878595e-06, "loss": 0.091, "step": 27026 }, { "epoch": 167.87227414330218, "grad_norm": 1.625, "learning_rate": 3.1801701096751098e-06, "loss": 0.1282, "step": 27027 }, { "epoch": 167.8785046728972, "grad_norm": 2.375, "learning_rate": 3.1789676773901683e-06, "loss": 0.1441, "step": 27028 }, { "epoch": 167.8847352024922, "grad_norm": 1.921875, "learning_rate": 3.1777654570354558e-06, "loss": 0.1436, "step": 27029 }, { "epoch": 167.89096573208724, "grad_norm": 2.125, "learning_rate": 3.176563448622652e-06, "loss": 0.1009, "step": 27030 }, { "epoch": 167.89719626168224, "grad_norm": 1.7109375, "learning_rate": 3.1753616521634248e-06, "loss": 0.1306, "step": 27031 }, { "epoch": 167.90342679127727, "grad_norm": 1.40625, "learning_rate": 3.1741600676694433e-06, "loss": 0.1337, "step": 27032 }, { "epoch": 167.90965732087227, "grad_norm": 1.7578125, "learning_rate": 3.172958695152384e-06, "loss": 0.1272, "step": 27033 }, { "epoch": 167.9158878504673, "grad_norm": 2.015625, "learning_rate": 3.1717575346239174e-06, "loss": 0.153, "step": 27034 }, { "epoch": 167.9221183800623, "grad_norm": 1.703125, "learning_rate": 3.170556586095699e-06, "loss": 0.1122, "step": 27035 }, { "epoch": 167.92834890965733, "grad_norm": 2.078125, "learning_rate": 3.169355849579403e-06, "loss": 0.2391, "step": 27036 }, { "epoch": 167.93457943925233, "grad_norm": 2.015625, "learning_rate": 3.168155325086683e-06, "loss": 0.1307, "step": 27037 }, { "epoch": 167.94080996884736, "grad_norm": 2.203125, "learning_rate": 3.166955012629208e-06, "loss": 0.1984, "step": 27038 }, { "epoch": 167.94704049844236, "grad_norm": 2.96875, "learning_rate": 3.165754912218624e-06, "loss": 0.2865, "step": 27039 }, { "epoch": 167.9532710280374, "grad_norm": 0.7890625, "learning_rate": 3.1645550238665945e-06, "loss": 0.1051, "step": 27040 }, { "epoch": 167.9595015576324, "grad_norm": 1.71875, "learning_rate": 3.1633553475847704e-06, "loss": 0.1521, "step": 27041 }, { "epoch": 167.96573208722742, "grad_norm": 0.96484375, "learning_rate": 3.1621558833848153e-06, "loss": 0.1054, "step": 27042 }, { "epoch": 167.97196261682242, "grad_norm": 1.796875, "learning_rate": 3.1609566312783556e-06, "loss": 0.1699, "step": 27043 }, { "epoch": 167.97819314641745, "grad_norm": 2.953125, "learning_rate": 3.1597575912770516e-06, "loss": 0.198, "step": 27044 }, { "epoch": 167.98442367601245, "grad_norm": 2.125, "learning_rate": 3.1585587633925466e-06, "loss": 0.2337, "step": 27045 }, { "epoch": 167.99065420560748, "grad_norm": 2.046875, "learning_rate": 3.1573601476364927e-06, "loss": 0.167, "step": 27046 }, { "epoch": 167.99688473520249, "grad_norm": 3.203125, "learning_rate": 3.1561617440205165e-06, "loss": 0.4145, "step": 27047 }, { "epoch": 168.0, "grad_norm": 0.6796875, "learning_rate": 3.154963552556264e-06, "loss": 0.0397, "step": 27048 }, { "epoch": 168.00623052959503, "grad_norm": 2.03125, "learning_rate": 3.1537655732553768e-06, "loss": 0.1316, "step": 27049 }, { "epoch": 168.01246105919003, "grad_norm": 1.5859375, "learning_rate": 3.152567806129486e-06, "loss": 0.2154, "step": 27050 }, { "epoch": 168.01869158878506, "grad_norm": 2.265625, "learning_rate": 3.1513702511902192e-06, "loss": 0.2151, "step": 27051 }, { "epoch": 168.02492211838006, "grad_norm": 2.78125, "learning_rate": 3.1501729084492086e-06, "loss": 0.2713, "step": 27052 }, { "epoch": 168.0311526479751, "grad_norm": 1.546875, "learning_rate": 3.148975777918095e-06, "loss": 0.1214, "step": 27053 }, { "epoch": 168.0373831775701, "grad_norm": 2.140625, "learning_rate": 3.1477788596084913e-06, "loss": 0.1798, "step": 27054 }, { "epoch": 168.04361370716512, "grad_norm": 2.578125, "learning_rate": 3.1465821535320324e-06, "loss": 0.1988, "step": 27055 }, { "epoch": 168.04984423676012, "grad_norm": 1.4296875, "learning_rate": 3.1453856597003312e-06, "loss": 0.1462, "step": 27056 }, { "epoch": 168.05607476635515, "grad_norm": 3.578125, "learning_rate": 3.1441893781250175e-06, "loss": 0.1893, "step": 27057 }, { "epoch": 168.06230529595015, "grad_norm": 2.03125, "learning_rate": 3.142993308817699e-06, "loss": 0.1369, "step": 27058 }, { "epoch": 168.06853582554518, "grad_norm": 3.125, "learning_rate": 3.1417974517900017e-06, "loss": 0.3179, "step": 27059 }, { "epoch": 168.07476635514018, "grad_norm": 1.125, "learning_rate": 3.140601807053539e-06, "loss": 0.1081, "step": 27060 }, { "epoch": 168.0809968847352, "grad_norm": 1.7421875, "learning_rate": 3.1394063746199244e-06, "loss": 0.1226, "step": 27061 }, { "epoch": 168.08722741433021, "grad_norm": 2.5, "learning_rate": 3.138211154500756e-06, "loss": 0.28, "step": 27062 }, { "epoch": 168.09345794392524, "grad_norm": 1.8671875, "learning_rate": 3.1370161467076525e-06, "loss": 0.1716, "step": 27063 }, { "epoch": 168.09968847352025, "grad_norm": 1.75, "learning_rate": 3.1358213512522246e-06, "loss": 0.1077, "step": 27064 }, { "epoch": 168.10591900311528, "grad_norm": 2.6875, "learning_rate": 3.134626768146065e-06, "loss": 0.2843, "step": 27065 }, { "epoch": 168.11214953271028, "grad_norm": 2.15625, "learning_rate": 3.1334323974007817e-06, "loss": 0.1681, "step": 27066 }, { "epoch": 168.1183800623053, "grad_norm": 1.9375, "learning_rate": 3.1322382390279764e-06, "loss": 0.1081, "step": 27067 }, { "epoch": 168.1246105919003, "grad_norm": 1.859375, "learning_rate": 3.131044293039248e-06, "loss": 0.1849, "step": 27068 }, { "epoch": 168.13084112149534, "grad_norm": 1.109375, "learning_rate": 3.129850559446182e-06, "loss": 0.1187, "step": 27069 }, { "epoch": 168.13707165109034, "grad_norm": 1.78125, "learning_rate": 3.128657038260377e-06, "loss": 0.1636, "step": 27070 }, { "epoch": 168.14330218068537, "grad_norm": 1.7421875, "learning_rate": 3.127463729493435e-06, "loss": 0.1812, "step": 27071 }, { "epoch": 168.14953271028037, "grad_norm": 2.09375, "learning_rate": 3.126270633156933e-06, "loss": 0.146, "step": 27072 }, { "epoch": 168.1557632398754, "grad_norm": 1.6484375, "learning_rate": 3.1250777492624616e-06, "loss": 0.1214, "step": 27073 }, { "epoch": 168.1619937694704, "grad_norm": 2.46875, "learning_rate": 3.1238850778216116e-06, "loss": 0.1402, "step": 27074 }, { "epoch": 168.16822429906543, "grad_norm": 2.6875, "learning_rate": 3.1226926188459656e-06, "loss": 0.2836, "step": 27075 }, { "epoch": 168.17445482866043, "grad_norm": 1.6484375, "learning_rate": 3.1215003723470944e-06, "loss": 0.2323, "step": 27076 }, { "epoch": 168.18068535825546, "grad_norm": 2.453125, "learning_rate": 3.120308338336586e-06, "loss": 0.0919, "step": 27077 }, { "epoch": 168.18691588785046, "grad_norm": 1.8671875, "learning_rate": 3.119116516826015e-06, "loss": 0.263, "step": 27078 }, { "epoch": 168.1931464174455, "grad_norm": 1.2578125, "learning_rate": 3.1179249078269664e-06, "loss": 0.1154, "step": 27079 }, { "epoch": 168.1993769470405, "grad_norm": 1.21875, "learning_rate": 3.1167335113509997e-06, "loss": 0.1418, "step": 27080 }, { "epoch": 168.20560747663552, "grad_norm": 1.34375, "learning_rate": 3.115542327409696e-06, "loss": 0.1357, "step": 27081 }, { "epoch": 168.21183800623052, "grad_norm": 1.8359375, "learning_rate": 3.1143513560146138e-06, "loss": 0.1203, "step": 27082 }, { "epoch": 168.21806853582555, "grad_norm": 1.65625, "learning_rate": 3.113160597177331e-06, "loss": 0.1463, "step": 27083 }, { "epoch": 168.22429906542055, "grad_norm": 1.8203125, "learning_rate": 3.1119700509094047e-06, "loss": 0.1687, "step": 27084 }, { "epoch": 168.23052959501558, "grad_norm": 3.046875, "learning_rate": 3.1107797172223978e-06, "loss": 0.305, "step": 27085 }, { "epoch": 168.23676012461058, "grad_norm": 2.0, "learning_rate": 3.1095895961278793e-06, "loss": 0.1717, "step": 27086 }, { "epoch": 168.2429906542056, "grad_norm": 2.671875, "learning_rate": 3.1083996876374063e-06, "loss": 0.1878, "step": 27087 }, { "epoch": 168.2492211838006, "grad_norm": 2.453125, "learning_rate": 3.107209991762522e-06, "loss": 0.1191, "step": 27088 }, { "epoch": 168.25545171339564, "grad_norm": 1.25, "learning_rate": 3.1060205085147926e-06, "loss": 0.1127, "step": 27089 }, { "epoch": 168.26168224299064, "grad_norm": 1.0, "learning_rate": 3.1048312379057726e-06, "loss": 0.1053, "step": 27090 }, { "epoch": 168.26791277258567, "grad_norm": 2.421875, "learning_rate": 3.1036421799470057e-06, "loss": 0.2622, "step": 27091 }, { "epoch": 168.27414330218068, "grad_norm": 2.328125, "learning_rate": 3.1024533346500405e-06, "loss": 0.2276, "step": 27092 }, { "epoch": 168.2803738317757, "grad_norm": 1.453125, "learning_rate": 3.1012647020264295e-06, "loss": 0.0822, "step": 27093 }, { "epoch": 168.2866043613707, "grad_norm": 0.921875, "learning_rate": 3.100076282087716e-06, "loss": 0.0994, "step": 27094 }, { "epoch": 168.29283489096574, "grad_norm": 2.515625, "learning_rate": 3.0988880748454353e-06, "loss": 0.3704, "step": 27095 }, { "epoch": 168.29906542056074, "grad_norm": 2.78125, "learning_rate": 3.0977000803111278e-06, "loss": 0.293, "step": 27096 }, { "epoch": 168.30529595015577, "grad_norm": 1.640625, "learning_rate": 3.096512298496343e-06, "loss": 0.1131, "step": 27097 }, { "epoch": 168.31152647975077, "grad_norm": 2.875, "learning_rate": 3.095324729412602e-06, "loss": 0.2287, "step": 27098 }, { "epoch": 168.3177570093458, "grad_norm": 1.8203125, "learning_rate": 3.0941373730714486e-06, "loss": 0.1187, "step": 27099 }, { "epoch": 168.3239875389408, "grad_norm": 2.21875, "learning_rate": 3.092950229484415e-06, "loss": 0.1972, "step": 27100 }, { "epoch": 168.33021806853583, "grad_norm": 2.203125, "learning_rate": 3.0917632986630305e-06, "loss": 0.1784, "step": 27101 }, { "epoch": 168.33644859813083, "grad_norm": 2.484375, "learning_rate": 3.090576580618812e-06, "loss": 0.2354, "step": 27102 }, { "epoch": 168.34267912772586, "grad_norm": 2.109375, "learning_rate": 3.089390075363294e-06, "loss": 0.1568, "step": 27103 }, { "epoch": 168.34890965732086, "grad_norm": 3.0625, "learning_rate": 3.0882037829080056e-06, "loss": 0.1888, "step": 27104 }, { "epoch": 168.3551401869159, "grad_norm": 2.140625, "learning_rate": 3.0870177032644553e-06, "loss": 0.1844, "step": 27105 }, { "epoch": 168.3613707165109, "grad_norm": 1.125, "learning_rate": 3.0858318364441778e-06, "loss": 0.1178, "step": 27106 }, { "epoch": 168.36760124610592, "grad_norm": 0.98046875, "learning_rate": 3.0846461824586752e-06, "loss": 0.091, "step": 27107 }, { "epoch": 168.37383177570092, "grad_norm": 2.75, "learning_rate": 3.0834607413194746e-06, "loss": 0.197, "step": 27108 }, { "epoch": 168.38006230529595, "grad_norm": 1.3125, "learning_rate": 3.0822755130380772e-06, "loss": 0.1146, "step": 27109 }, { "epoch": 168.38629283489095, "grad_norm": 1.7734375, "learning_rate": 3.0810904976260045e-06, "loss": 0.1079, "step": 27110 }, { "epoch": 168.39252336448598, "grad_norm": 2.59375, "learning_rate": 3.0799056950947592e-06, "loss": 0.1773, "step": 27111 }, { "epoch": 168.39875389408098, "grad_norm": 1.1015625, "learning_rate": 3.0787211054558647e-06, "loss": 0.1321, "step": 27112 }, { "epoch": 168.404984423676, "grad_norm": 1.203125, "learning_rate": 3.0775367287208005e-06, "loss": 0.1054, "step": 27113 }, { "epoch": 168.41121495327104, "grad_norm": 2.34375, "learning_rate": 3.076352564901083e-06, "loss": 0.2623, "step": 27114 }, { "epoch": 168.41744548286604, "grad_norm": 1.421875, "learning_rate": 3.0751686140082116e-06, "loss": 0.1086, "step": 27115 }, { "epoch": 168.42367601246107, "grad_norm": 1.90625, "learning_rate": 3.07398487605369e-06, "loss": 0.1662, "step": 27116 }, { "epoch": 168.42990654205607, "grad_norm": 1.6484375, "learning_rate": 3.0728013510490067e-06, "loss": 0.1075, "step": 27117 }, { "epoch": 168.4361370716511, "grad_norm": 2.28125, "learning_rate": 3.071618039005658e-06, "loss": 0.1309, "step": 27118 }, { "epoch": 168.4423676012461, "grad_norm": 1.3125, "learning_rate": 3.0704349399351435e-06, "loss": 0.1094, "step": 27119 }, { "epoch": 168.44859813084113, "grad_norm": 2.375, "learning_rate": 3.0692520538489507e-06, "loss": 0.1593, "step": 27120 }, { "epoch": 168.45482866043614, "grad_norm": 4.25, "learning_rate": 3.0680693807585593e-06, "loss": 0.1587, "step": 27121 }, { "epoch": 168.46105919003116, "grad_norm": 0.72265625, "learning_rate": 3.0668869206754603e-06, "loss": 0.0951, "step": 27122 }, { "epoch": 168.46728971962617, "grad_norm": 1.359375, "learning_rate": 3.065704673611147e-06, "loss": 0.1069, "step": 27123 }, { "epoch": 168.4735202492212, "grad_norm": 2.25, "learning_rate": 3.0645226395770887e-06, "loss": 0.2288, "step": 27124 }, { "epoch": 168.4797507788162, "grad_norm": 1.8359375, "learning_rate": 3.0633408185847784e-06, "loss": 0.1782, "step": 27125 }, { "epoch": 168.48598130841123, "grad_norm": 2.09375, "learning_rate": 3.0621592106456794e-06, "loss": 0.1847, "step": 27126 }, { "epoch": 168.49221183800623, "grad_norm": 2.890625, "learning_rate": 3.060977815771282e-06, "loss": 0.1194, "step": 27127 }, { "epoch": 168.49844236760126, "grad_norm": 2.21875, "learning_rate": 3.059796633973047e-06, "loss": 0.1627, "step": 27128 }, { "epoch": 168.50467289719626, "grad_norm": 0.96484375, "learning_rate": 3.0586156652624542e-06, "loss": 0.1041, "step": 27129 }, { "epoch": 168.5109034267913, "grad_norm": 1.390625, "learning_rate": 3.0574349096509775e-06, "loss": 0.118, "step": 27130 }, { "epoch": 168.5171339563863, "grad_norm": 1.3046875, "learning_rate": 3.056254367150077e-06, "loss": 0.122, "step": 27131 }, { "epoch": 168.52336448598132, "grad_norm": 1.515625, "learning_rate": 3.0550740377712188e-06, "loss": 0.1084, "step": 27132 }, { "epoch": 168.52959501557632, "grad_norm": 1.28125, "learning_rate": 3.053893921525866e-06, "loss": 0.144, "step": 27133 }, { "epoch": 168.53582554517135, "grad_norm": 2.328125, "learning_rate": 3.052714018425487e-06, "loss": 0.195, "step": 27134 }, { "epoch": 168.54205607476635, "grad_norm": 2.15625, "learning_rate": 3.051534328481534e-06, "loss": 0.1124, "step": 27135 }, { "epoch": 168.54828660436138, "grad_norm": 3.5625, "learning_rate": 3.0503548517054644e-06, "loss": 0.4115, "step": 27136 }, { "epoch": 168.55451713395638, "grad_norm": 1.4921875, "learning_rate": 3.049175588108738e-06, "loss": 0.1329, "step": 27137 }, { "epoch": 168.5607476635514, "grad_norm": 2.53125, "learning_rate": 3.0479965377028158e-06, "loss": 0.1765, "step": 27138 }, { "epoch": 168.5669781931464, "grad_norm": 2.0625, "learning_rate": 3.0468177004991273e-06, "loss": 0.0878, "step": 27139 }, { "epoch": 168.57320872274144, "grad_norm": 1.453125, "learning_rate": 3.0456390765091354e-06, "loss": 0.155, "step": 27140 }, { "epoch": 168.57943925233644, "grad_norm": 1.609375, "learning_rate": 3.044460665744284e-06, "loss": 0.0994, "step": 27141 }, { "epoch": 168.58566978193147, "grad_norm": 1.546875, "learning_rate": 3.043282468216027e-06, "loss": 0.1465, "step": 27142 }, { "epoch": 168.59190031152647, "grad_norm": 1.7890625, "learning_rate": 3.0421044839357926e-06, "loss": 0.1203, "step": 27143 }, { "epoch": 168.5981308411215, "grad_norm": 1.9296875, "learning_rate": 3.0409267129150348e-06, "loss": 0.1618, "step": 27144 }, { "epoch": 168.6043613707165, "grad_norm": 3.328125, "learning_rate": 3.0397491551651807e-06, "loss": 0.3774, "step": 27145 }, { "epoch": 168.61059190031153, "grad_norm": 2.546875, "learning_rate": 3.0385718106976795e-06, "loss": 0.1962, "step": 27146 }, { "epoch": 168.61682242990653, "grad_norm": 2.203125, "learning_rate": 3.03739467952395e-06, "loss": 0.2123, "step": 27147 }, { "epoch": 168.62305295950156, "grad_norm": 0.83984375, "learning_rate": 3.036217761655438e-06, "loss": 0.0917, "step": 27148 }, { "epoch": 168.62928348909657, "grad_norm": 1.875, "learning_rate": 3.0350410571035713e-06, "loss": 0.1446, "step": 27149 }, { "epoch": 168.6355140186916, "grad_norm": 2.03125, "learning_rate": 3.0338645658797737e-06, "loss": 0.1698, "step": 27150 }, { "epoch": 168.6417445482866, "grad_norm": 1.1015625, "learning_rate": 3.0326882879954778e-06, "loss": 0.1144, "step": 27151 }, { "epoch": 168.64797507788163, "grad_norm": 1.359375, "learning_rate": 3.031512223462099e-06, "loss": 0.1247, "step": 27152 }, { "epoch": 168.65420560747663, "grad_norm": 1.9609375, "learning_rate": 3.0303363722910704e-06, "loss": 0.1362, "step": 27153 }, { "epoch": 168.66043613707166, "grad_norm": 2.515625, "learning_rate": 3.0291607344938022e-06, "loss": 0.2589, "step": 27154 }, { "epoch": 168.66666666666666, "grad_norm": 2.765625, "learning_rate": 3.0279853100817183e-06, "loss": 0.2413, "step": 27155 }, { "epoch": 168.6728971962617, "grad_norm": 2.171875, "learning_rate": 3.0268100990662353e-06, "loss": 0.2417, "step": 27156 }, { "epoch": 168.6791277258567, "grad_norm": 1.1953125, "learning_rate": 3.025635101458768e-06, "loss": 0.106, "step": 27157 }, { "epoch": 168.68535825545172, "grad_norm": 1.5078125, "learning_rate": 3.0244603172707163e-06, "loss": 0.1437, "step": 27158 }, { "epoch": 168.69158878504672, "grad_norm": 1.4140625, "learning_rate": 3.0232857465135015e-06, "loss": 0.1292, "step": 27159 }, { "epoch": 168.69781931464175, "grad_norm": 3.53125, "learning_rate": 3.0221113891985343e-06, "loss": 0.3463, "step": 27160 }, { "epoch": 168.70404984423675, "grad_norm": 2.296875, "learning_rate": 3.0209372453372077e-06, "loss": 0.177, "step": 27161 }, { "epoch": 168.71028037383178, "grad_norm": 1.5625, "learning_rate": 3.019763314940932e-06, "loss": 0.1366, "step": 27162 }, { "epoch": 168.71651090342678, "grad_norm": 1.4296875, "learning_rate": 3.018589598021115e-06, "loss": 0.1192, "step": 27163 }, { "epoch": 168.7227414330218, "grad_norm": 2.53125, "learning_rate": 3.017416094589151e-06, "loss": 0.2759, "step": 27164 }, { "epoch": 168.7289719626168, "grad_norm": 1.9921875, "learning_rate": 3.016242804656427e-06, "loss": 0.1721, "step": 27165 }, { "epoch": 168.73520249221184, "grad_norm": 2.4375, "learning_rate": 3.0150697282343516e-06, "loss": 0.2325, "step": 27166 }, { "epoch": 168.74143302180684, "grad_norm": 1.4296875, "learning_rate": 3.0138968653343176e-06, "loss": 0.1364, "step": 27167 }, { "epoch": 168.74766355140187, "grad_norm": 2.078125, "learning_rate": 3.012724215967705e-06, "loss": 0.1699, "step": 27168 }, { "epoch": 168.75389408099687, "grad_norm": 2.40625, "learning_rate": 3.0115517801459104e-06, "loss": 0.1772, "step": 27169 }, { "epoch": 168.7601246105919, "grad_norm": 1.828125, "learning_rate": 3.010379557880327e-06, "loss": 0.1223, "step": 27170 }, { "epoch": 168.7663551401869, "grad_norm": 2.078125, "learning_rate": 3.009207549182333e-06, "loss": 0.1651, "step": 27171 }, { "epoch": 168.77258566978193, "grad_norm": 2.515625, "learning_rate": 3.008035754063304e-06, "loss": 0.1846, "step": 27172 }, { "epoch": 168.77881619937693, "grad_norm": 1.3046875, "learning_rate": 3.006864172534632e-06, "loss": 0.124, "step": 27173 }, { "epoch": 168.78504672897196, "grad_norm": 2.0625, "learning_rate": 3.0056928046076877e-06, "loss": 0.159, "step": 27174 }, { "epoch": 168.79127725856696, "grad_norm": 2.765625, "learning_rate": 3.0045216502938567e-06, "loss": 0.2679, "step": 27175 }, { "epoch": 168.797507788162, "grad_norm": 1.796875, "learning_rate": 3.0033507096045106e-06, "loss": 0.1662, "step": 27176 }, { "epoch": 168.80373831775702, "grad_norm": 1.6171875, "learning_rate": 3.002179982551012e-06, "loss": 0.1376, "step": 27177 }, { "epoch": 168.80996884735202, "grad_norm": 2.75, "learning_rate": 3.001009469144739e-06, "loss": 0.192, "step": 27178 }, { "epoch": 168.81619937694705, "grad_norm": 1.4609375, "learning_rate": 2.999839169397067e-06, "loss": 0.1233, "step": 27179 }, { "epoch": 168.82242990654206, "grad_norm": 2.09375, "learning_rate": 2.998669083319347e-06, "loss": 0.1523, "step": 27180 }, { "epoch": 168.82866043613708, "grad_norm": 1.1796875, "learning_rate": 2.997499210922952e-06, "loss": 0.0835, "step": 27181 }, { "epoch": 168.8348909657321, "grad_norm": 1.1875, "learning_rate": 2.9963295522192486e-06, "loss": 0.1285, "step": 27182 }, { "epoch": 168.84112149532712, "grad_norm": 1.2890625, "learning_rate": 2.9951601072195886e-06, "loss": 0.1093, "step": 27183 }, { "epoch": 168.84735202492212, "grad_norm": 2.46875, "learning_rate": 2.993990875935329e-06, "loss": 0.27, "step": 27184 }, { "epoch": 168.85358255451715, "grad_norm": 3.15625, "learning_rate": 2.992821858377828e-06, "loss": 0.2269, "step": 27185 }, { "epoch": 168.85981308411215, "grad_norm": 1.6640625, "learning_rate": 2.9916530545584465e-06, "loss": 0.1393, "step": 27186 }, { "epoch": 168.86604361370718, "grad_norm": 2.265625, "learning_rate": 2.9904844644885245e-06, "loss": 0.19, "step": 27187 }, { "epoch": 168.87227414330218, "grad_norm": 1.296875, "learning_rate": 2.989316088179417e-06, "loss": 0.1021, "step": 27188 }, { "epoch": 168.8785046728972, "grad_norm": 2.21875, "learning_rate": 2.988147925642476e-06, "loss": 0.2281, "step": 27189 }, { "epoch": 168.8847352024922, "grad_norm": 1.8671875, "learning_rate": 2.9869799768890425e-06, "loss": 0.1765, "step": 27190 }, { "epoch": 168.89096573208724, "grad_norm": 2.78125, "learning_rate": 2.9858122419304552e-06, "loss": 0.3179, "step": 27191 }, { "epoch": 168.89719626168224, "grad_norm": 1.7265625, "learning_rate": 2.9846447207780597e-06, "loss": 0.1329, "step": 27192 }, { "epoch": 168.90342679127727, "grad_norm": 2.375, "learning_rate": 2.9834774134432032e-06, "loss": 0.1887, "step": 27193 }, { "epoch": 168.90965732087227, "grad_norm": 1.71875, "learning_rate": 2.9823103199372065e-06, "loss": 0.1313, "step": 27194 }, { "epoch": 168.9158878504673, "grad_norm": 2.140625, "learning_rate": 2.9811434402714196e-06, "loss": 0.1887, "step": 27195 }, { "epoch": 168.9221183800623, "grad_norm": 1.65625, "learning_rate": 2.9799767744571632e-06, "loss": 0.1302, "step": 27196 }, { "epoch": 168.92834890965733, "grad_norm": 1.265625, "learning_rate": 2.9788103225057788e-06, "loss": 0.1436, "step": 27197 }, { "epoch": 168.93457943925233, "grad_norm": 1.3125, "learning_rate": 2.977644084428588e-06, "loss": 0.1469, "step": 27198 }, { "epoch": 168.94080996884736, "grad_norm": 2.125, "learning_rate": 2.9764780602369175e-06, "loss": 0.1916, "step": 27199 }, { "epoch": 168.94704049844236, "grad_norm": 1.140625, "learning_rate": 2.9753122499421e-06, "loss": 0.0834, "step": 27200 }, { "epoch": 168.9532710280374, "grad_norm": 3.125, "learning_rate": 2.974146653555454e-06, "loss": 0.1487, "step": 27201 }, { "epoch": 168.9595015576324, "grad_norm": 2.734375, "learning_rate": 2.9729812710882906e-06, "loss": 0.2405, "step": 27202 }, { "epoch": 168.96573208722742, "grad_norm": 2.703125, "learning_rate": 2.9718161025519388e-06, "loss": 0.2196, "step": 27203 }, { "epoch": 168.97196261682242, "grad_norm": 1.671875, "learning_rate": 2.9706511479577174e-06, "loss": 0.245, "step": 27204 }, { "epoch": 168.97819314641745, "grad_norm": 1.3984375, "learning_rate": 2.9694864073169293e-06, "loss": 0.1239, "step": 27205 }, { "epoch": 168.98442367601245, "grad_norm": 1.5390625, "learning_rate": 2.9683218806408924e-06, "loss": 0.1325, "step": 27206 }, { "epoch": 168.99065420560748, "grad_norm": 1.9921875, "learning_rate": 2.9671575679409203e-06, "loss": 0.158, "step": 27207 }, { "epoch": 168.99688473520249, "grad_norm": 1.546875, "learning_rate": 2.9659934692283258e-06, "loss": 0.1353, "step": 27208 }, { "epoch": 169.0, "grad_norm": 1.1484375, "learning_rate": 2.9648295845143973e-06, "loss": 0.0489, "step": 27209 }, { "epoch": 169.00623052959503, "grad_norm": 1.6953125, "learning_rate": 2.9636659138104513e-06, "loss": 0.1307, "step": 27210 }, { "epoch": 169.01246105919003, "grad_norm": 1.1015625, "learning_rate": 2.9625024571277833e-06, "loss": 0.1078, "step": 27211 }, { "epoch": 169.01869158878506, "grad_norm": 1.78125, "learning_rate": 2.9613392144777042e-06, "loss": 0.1073, "step": 27212 }, { "epoch": 169.02492211838006, "grad_norm": 1.4609375, "learning_rate": 2.9601761858714965e-06, "loss": 0.1316, "step": 27213 }, { "epoch": 169.0311526479751, "grad_norm": 1.828125, "learning_rate": 2.959013371320471e-06, "loss": 0.1346, "step": 27214 }, { "epoch": 169.0373831775701, "grad_norm": 1.9921875, "learning_rate": 2.957850770835907e-06, "loss": 0.0951, "step": 27215 }, { "epoch": 169.04361370716512, "grad_norm": 2.0, "learning_rate": 2.95668838442911e-06, "loss": 0.1928, "step": 27216 }, { "epoch": 169.04984423676012, "grad_norm": 1.875, "learning_rate": 2.9555262121113536e-06, "loss": 0.1709, "step": 27217 }, { "epoch": 169.05607476635515, "grad_norm": 2.5625, "learning_rate": 2.9543642538939347e-06, "loss": 0.1079, "step": 27218 }, { "epoch": 169.06230529595015, "grad_norm": 2.203125, "learning_rate": 2.9532025097881418e-06, "loss": 0.1727, "step": 27219 }, { "epoch": 169.06853582554518, "grad_norm": 1.203125, "learning_rate": 2.9520409798052483e-06, "loss": 0.1, "step": 27220 }, { "epoch": 169.07476635514018, "grad_norm": 2.765625, "learning_rate": 2.9508796639565433e-06, "loss": 0.2294, "step": 27221 }, { "epoch": 169.0809968847352, "grad_norm": 2.765625, "learning_rate": 2.9497185622533006e-06, "loss": 0.2231, "step": 27222 }, { "epoch": 169.08722741433021, "grad_norm": 2.296875, "learning_rate": 2.9485576747068007e-06, "loss": 0.1551, "step": 27223 }, { "epoch": 169.09345794392524, "grad_norm": 1.875, "learning_rate": 2.947397001328314e-06, "loss": 0.155, "step": 27224 }, { "epoch": 169.09968847352025, "grad_norm": 2.09375, "learning_rate": 2.9462365421291162e-06, "loss": 0.1742, "step": 27225 }, { "epoch": 169.10591900311528, "grad_norm": 1.6640625, "learning_rate": 2.9450762971204806e-06, "loss": 0.2477, "step": 27226 }, { "epoch": 169.11214953271028, "grad_norm": 1.8515625, "learning_rate": 2.943916266313676e-06, "loss": 0.1781, "step": 27227 }, { "epoch": 169.1183800623053, "grad_norm": 2.359375, "learning_rate": 2.9427564497199576e-06, "loss": 0.1919, "step": 27228 }, { "epoch": 169.1246105919003, "grad_norm": 1.3046875, "learning_rate": 2.941596847350597e-06, "loss": 0.1083, "step": 27229 }, { "epoch": 169.13084112149534, "grad_norm": 2.609375, "learning_rate": 2.940437459216863e-06, "loss": 0.2619, "step": 27230 }, { "epoch": 169.13707165109034, "grad_norm": 2.25, "learning_rate": 2.9392782853300077e-06, "loss": 0.0962, "step": 27231 }, { "epoch": 169.14330218068537, "grad_norm": 1.828125, "learning_rate": 2.9381193257012886e-06, "loss": 0.1552, "step": 27232 }, { "epoch": 169.14953271028037, "grad_norm": 2.421875, "learning_rate": 2.9369605803419715e-06, "loss": 0.2278, "step": 27233 }, { "epoch": 169.1557632398754, "grad_norm": 2.046875, "learning_rate": 2.935802049263303e-06, "loss": 0.1903, "step": 27234 }, { "epoch": 169.1619937694704, "grad_norm": 2.0, "learning_rate": 2.9346437324765302e-06, "loss": 0.1446, "step": 27235 }, { "epoch": 169.16822429906543, "grad_norm": 1.703125, "learning_rate": 2.9334856299929075e-06, "loss": 0.1351, "step": 27236 }, { "epoch": 169.17445482866043, "grad_norm": 2.234375, "learning_rate": 2.9323277418236845e-06, "loss": 0.2314, "step": 27237 }, { "epoch": 169.18068535825546, "grad_norm": 2.078125, "learning_rate": 2.9311700679801102e-06, "loss": 0.1523, "step": 27238 }, { "epoch": 169.18691588785046, "grad_norm": 2.71875, "learning_rate": 2.9300126084734176e-06, "loss": 0.1957, "step": 27239 }, { "epoch": 169.1931464174455, "grad_norm": 3.09375, "learning_rate": 2.928855363314861e-06, "loss": 0.2543, "step": 27240 }, { "epoch": 169.1993769470405, "grad_norm": 2.25, "learning_rate": 2.927698332515666e-06, "loss": 0.1882, "step": 27241 }, { "epoch": 169.20560747663552, "grad_norm": 2.765625, "learning_rate": 2.9265415160870836e-06, "loss": 0.1675, "step": 27242 }, { "epoch": 169.21183800623052, "grad_norm": 1.1875, "learning_rate": 2.9253849140403383e-06, "loss": 0.1067, "step": 27243 }, { "epoch": 169.21806853582555, "grad_norm": 1.6640625, "learning_rate": 2.9242285263866657e-06, "loss": 0.1611, "step": 27244 }, { "epoch": 169.22429906542055, "grad_norm": 1.5703125, "learning_rate": 2.9230723531373045e-06, "loss": 0.1373, "step": 27245 }, { "epoch": 169.23052959501558, "grad_norm": 2.53125, "learning_rate": 2.9219163943034782e-06, "loss": 0.2452, "step": 27246 }, { "epoch": 169.23676012461058, "grad_norm": 3.125, "learning_rate": 2.9207606498964064e-06, "loss": 0.2599, "step": 27247 }, { "epoch": 169.2429906542056, "grad_norm": 2.8125, "learning_rate": 2.919605119927321e-06, "loss": 0.2499, "step": 27248 }, { "epoch": 169.2492211838006, "grad_norm": 1.8828125, "learning_rate": 2.9184498044074496e-06, "loss": 0.1673, "step": 27249 }, { "epoch": 169.25545171339564, "grad_norm": 1.28125, "learning_rate": 2.917294703348003e-06, "loss": 0.1056, "step": 27250 }, { "epoch": 169.26168224299064, "grad_norm": 1.859375, "learning_rate": 2.916139816760205e-06, "loss": 0.131, "step": 27251 }, { "epoch": 169.26791277258567, "grad_norm": 1.375, "learning_rate": 2.9149851446552773e-06, "loss": 0.1209, "step": 27252 }, { "epoch": 169.27414330218068, "grad_norm": 2.3125, "learning_rate": 2.9138306870444277e-06, "loss": 0.1994, "step": 27253 }, { "epoch": 169.2803738317757, "grad_norm": 1.6171875, "learning_rate": 2.912676443938864e-06, "loss": 0.1359, "step": 27254 }, { "epoch": 169.2866043613707, "grad_norm": 2.6875, "learning_rate": 2.911522415349802e-06, "loss": 0.2059, "step": 27255 }, { "epoch": 169.29283489096574, "grad_norm": 2.0, "learning_rate": 2.9103686012884546e-06, "loss": 0.1491, "step": 27256 }, { "epoch": 169.29906542056074, "grad_norm": 1.0078125, "learning_rate": 2.909215001766019e-06, "loss": 0.088, "step": 27257 }, { "epoch": 169.30529595015577, "grad_norm": 2.15625, "learning_rate": 2.9080616167937e-06, "loss": 0.1505, "step": 27258 }, { "epoch": 169.31152647975077, "grad_norm": 1.6171875, "learning_rate": 2.906908446382711e-06, "loss": 0.1186, "step": 27259 }, { "epoch": 169.3177570093458, "grad_norm": 2.09375, "learning_rate": 2.905755490544243e-06, "loss": 0.1666, "step": 27260 }, { "epoch": 169.3239875389408, "grad_norm": 2.46875, "learning_rate": 2.9046027492894894e-06, "loss": 0.1814, "step": 27261 }, { "epoch": 169.33021806853583, "grad_norm": 2.09375, "learning_rate": 2.90345022262965e-06, "loss": 0.14, "step": 27262 }, { "epoch": 169.33644859813083, "grad_norm": 1.2734375, "learning_rate": 2.902297910575924e-06, "loss": 0.1177, "step": 27263 }, { "epoch": 169.34267912772586, "grad_norm": 1.40625, "learning_rate": 2.901145813139491e-06, "loss": 0.1063, "step": 27264 }, { "epoch": 169.34890965732086, "grad_norm": 1.5546875, "learning_rate": 2.8999939303315545e-06, "loss": 0.1468, "step": 27265 }, { "epoch": 169.3551401869159, "grad_norm": 1.96875, "learning_rate": 2.89884226216329e-06, "loss": 0.1297, "step": 27266 }, { "epoch": 169.3613707165109, "grad_norm": 3.53125, "learning_rate": 2.897690808645892e-06, "loss": 0.2356, "step": 27267 }, { "epoch": 169.36760124610592, "grad_norm": 1.7890625, "learning_rate": 2.896539569790535e-06, "loss": 0.1133, "step": 27268 }, { "epoch": 169.37383177570092, "grad_norm": 2.125, "learning_rate": 2.895388545608402e-06, "loss": 0.1292, "step": 27269 }, { "epoch": 169.38006230529595, "grad_norm": 1.9453125, "learning_rate": 2.894237736110675e-06, "loss": 0.2251, "step": 27270 }, { "epoch": 169.38629283489095, "grad_norm": 2.515625, "learning_rate": 2.89308714130854e-06, "loss": 0.2151, "step": 27271 }, { "epoch": 169.39252336448598, "grad_norm": 2.375, "learning_rate": 2.8919367612131517e-06, "loss": 0.255, "step": 27272 }, { "epoch": 169.39875389408098, "grad_norm": 1.3125, "learning_rate": 2.890786595835693e-06, "loss": 0.1325, "step": 27273 }, { "epoch": 169.404984423676, "grad_norm": 2.171875, "learning_rate": 2.8896366451873324e-06, "loss": 0.1901, "step": 27274 }, { "epoch": 169.41121495327104, "grad_norm": 1.3359375, "learning_rate": 2.8884869092792475e-06, "loss": 0.1188, "step": 27275 }, { "epoch": 169.41744548286604, "grad_norm": 2.515625, "learning_rate": 2.8873373881225936e-06, "loss": 0.2017, "step": 27276 }, { "epoch": 169.42367601246107, "grad_norm": 1.6875, "learning_rate": 2.8861880817285384e-06, "loss": 0.1307, "step": 27277 }, { "epoch": 169.42990654205607, "grad_norm": 1.2265625, "learning_rate": 2.8850389901082492e-06, "loss": 0.1007, "step": 27278 }, { "epoch": 169.4361370716511, "grad_norm": 1.125, "learning_rate": 2.8838901132728827e-06, "loss": 0.1049, "step": 27279 }, { "epoch": 169.4423676012461, "grad_norm": 2.125, "learning_rate": 2.8827414512335923e-06, "loss": 0.1831, "step": 27280 }, { "epoch": 169.44859813084113, "grad_norm": 2.703125, "learning_rate": 2.881593004001537e-06, "loss": 0.2917, "step": 27281 }, { "epoch": 169.45482866043614, "grad_norm": 2.515625, "learning_rate": 2.880444771587876e-06, "loss": 0.2036, "step": 27282 }, { "epoch": 169.46105919003116, "grad_norm": 1.3984375, "learning_rate": 2.8792967540037545e-06, "loss": 0.1382, "step": 27283 }, { "epoch": 169.46728971962617, "grad_norm": 1.3203125, "learning_rate": 2.8781489512603286e-06, "loss": 0.1139, "step": 27284 }, { "epoch": 169.4735202492212, "grad_norm": 2.671875, "learning_rate": 2.8770013633687364e-06, "loss": 0.1829, "step": 27285 }, { "epoch": 169.4797507788162, "grad_norm": 1.671875, "learning_rate": 2.875853990340133e-06, "loss": 0.1356, "step": 27286 }, { "epoch": 169.48598130841123, "grad_norm": 2.28125, "learning_rate": 2.8747068321856556e-06, "loss": 0.1887, "step": 27287 }, { "epoch": 169.49221183800623, "grad_norm": 1.578125, "learning_rate": 2.8735598889164465e-06, "loss": 0.1192, "step": 27288 }, { "epoch": 169.49844236760126, "grad_norm": 1.796875, "learning_rate": 2.872413160543655e-06, "loss": 0.125, "step": 27289 }, { "epoch": 169.50467289719626, "grad_norm": 1.546875, "learning_rate": 2.8712666470784024e-06, "loss": 0.1709, "step": 27290 }, { "epoch": 169.5109034267913, "grad_norm": 2.0625, "learning_rate": 2.870120348531835e-06, "loss": 0.2216, "step": 27291 }, { "epoch": 169.5171339563863, "grad_norm": 1.671875, "learning_rate": 2.868974264915081e-06, "loss": 0.1237, "step": 27292 }, { "epoch": 169.52336448598132, "grad_norm": 2.34375, "learning_rate": 2.867828396239275e-06, "loss": 0.1583, "step": 27293 }, { "epoch": 169.52959501557632, "grad_norm": 1.59375, "learning_rate": 2.866682742515542e-06, "loss": 0.1158, "step": 27294 }, { "epoch": 169.53582554517135, "grad_norm": 2.40625, "learning_rate": 2.865537303755006e-06, "loss": 0.1374, "step": 27295 }, { "epoch": 169.54205607476635, "grad_norm": 0.96875, "learning_rate": 2.8643920799688056e-06, "loss": 0.0957, "step": 27296 }, { "epoch": 169.54828660436138, "grad_norm": 1.6875, "learning_rate": 2.8632470711680515e-06, "loss": 0.0862, "step": 27297 }, { "epoch": 169.55451713395638, "grad_norm": 1.453125, "learning_rate": 2.8621022773638623e-06, "loss": 0.1288, "step": 27298 }, { "epoch": 169.5607476635514, "grad_norm": 2.796875, "learning_rate": 2.8609576985673624e-06, "loss": 0.3149, "step": 27299 }, { "epoch": 169.5669781931464, "grad_norm": 1.7734375, "learning_rate": 2.859813334789671e-06, "loss": 0.2408, "step": 27300 }, { "epoch": 169.57320872274144, "grad_norm": 2.625, "learning_rate": 2.858669186041893e-06, "loss": 0.1859, "step": 27301 }, { "epoch": 169.57943925233644, "grad_norm": 2.390625, "learning_rate": 2.857525252335144e-06, "loss": 0.2195, "step": 27302 }, { "epoch": 169.58566978193147, "grad_norm": 1.28125, "learning_rate": 2.8563815336805437e-06, "loss": 0.1235, "step": 27303 }, { "epoch": 169.59190031152647, "grad_norm": 1.84375, "learning_rate": 2.855238030089191e-06, "loss": 0.1064, "step": 27304 }, { "epoch": 169.5981308411215, "grad_norm": 1.8515625, "learning_rate": 2.8540947415721913e-06, "loss": 0.0932, "step": 27305 }, { "epoch": 169.6043613707165, "grad_norm": 2.71875, "learning_rate": 2.8529516681406465e-06, "loss": 0.1956, "step": 27306 }, { "epoch": 169.61059190031153, "grad_norm": 2.921875, "learning_rate": 2.8518088098056643e-06, "loss": 0.2446, "step": 27307 }, { "epoch": 169.61682242990653, "grad_norm": 1.203125, "learning_rate": 2.8506661665783473e-06, "loss": 0.1309, "step": 27308 }, { "epoch": 169.62305295950156, "grad_norm": 1.96875, "learning_rate": 2.849523738469781e-06, "loss": 0.1501, "step": 27309 }, { "epoch": 169.62928348909657, "grad_norm": 3.078125, "learning_rate": 2.8483815254910757e-06, "loss": 0.2141, "step": 27310 }, { "epoch": 169.6355140186916, "grad_norm": 1.2421875, "learning_rate": 2.847239527653309e-06, "loss": 0.1224, "step": 27311 }, { "epoch": 169.6417445482866, "grad_norm": 1.453125, "learning_rate": 2.8460977449675878e-06, "loss": 0.142, "step": 27312 }, { "epoch": 169.64797507788163, "grad_norm": 1.3125, "learning_rate": 2.844956177444988e-06, "loss": 0.1488, "step": 27313 }, { "epoch": 169.65420560747663, "grad_norm": 3.234375, "learning_rate": 2.8438148250966025e-06, "loss": 0.396, "step": 27314 }, { "epoch": 169.66043613707166, "grad_norm": 1.0625, "learning_rate": 2.84267368793352e-06, "loss": 0.1212, "step": 27315 }, { "epoch": 169.66666666666666, "grad_norm": 2.890625, "learning_rate": 2.84153276596682e-06, "loss": 0.1609, "step": 27316 }, { "epoch": 169.6728971962617, "grad_norm": 1.4375, "learning_rate": 2.8403920592075807e-06, "loss": 0.1134, "step": 27317 }, { "epoch": 169.6791277258567, "grad_norm": 1.5625, "learning_rate": 2.839251567666881e-06, "loss": 0.1133, "step": 27318 }, { "epoch": 169.68535825545172, "grad_norm": 2.015625, "learning_rate": 2.838111291355805e-06, "loss": 0.1625, "step": 27319 }, { "epoch": 169.69158878504672, "grad_norm": 1.9453125, "learning_rate": 2.836971230285418e-06, "loss": 0.1639, "step": 27320 }, { "epoch": 169.69781931464175, "grad_norm": 2.09375, "learning_rate": 2.8358313844667973e-06, "loss": 0.159, "step": 27321 }, { "epoch": 169.70404984423675, "grad_norm": 2.859375, "learning_rate": 2.834691753911017e-06, "loss": 0.224, "step": 27322 }, { "epoch": 169.71028037383178, "grad_norm": 1.4765625, "learning_rate": 2.8335523386291417e-06, "loss": 0.0926, "step": 27323 }, { "epoch": 169.71651090342678, "grad_norm": 1.671875, "learning_rate": 2.832413138632231e-06, "loss": 0.1304, "step": 27324 }, { "epoch": 169.7227414330218, "grad_norm": 3.625, "learning_rate": 2.831274153931357e-06, "loss": 0.3339, "step": 27325 }, { "epoch": 169.7289719626168, "grad_norm": 2.296875, "learning_rate": 2.830135384537583e-06, "loss": 0.182, "step": 27326 }, { "epoch": 169.73520249221184, "grad_norm": 2.15625, "learning_rate": 2.828996830461961e-06, "loss": 0.1875, "step": 27327 }, { "epoch": 169.74143302180684, "grad_norm": 2.296875, "learning_rate": 2.8278584917155543e-06, "loss": 0.2954, "step": 27328 }, { "epoch": 169.74766355140187, "grad_norm": 1.7578125, "learning_rate": 2.8267203683094213e-06, "loss": 0.1481, "step": 27329 }, { "epoch": 169.75389408099687, "grad_norm": 1.6953125, "learning_rate": 2.825582460254614e-06, "loss": 0.116, "step": 27330 }, { "epoch": 169.7601246105919, "grad_norm": 2.65625, "learning_rate": 2.8244447675621786e-06, "loss": 0.1996, "step": 27331 }, { "epoch": 169.7663551401869, "grad_norm": 2.078125, "learning_rate": 2.8233072902431655e-06, "loss": 0.2365, "step": 27332 }, { "epoch": 169.77258566978193, "grad_norm": 1.90625, "learning_rate": 2.8221700283086266e-06, "loss": 0.1089, "step": 27333 }, { "epoch": 169.77881619937693, "grad_norm": 3.3125, "learning_rate": 2.8210329817696115e-06, "loss": 0.3436, "step": 27334 }, { "epoch": 169.78504672897196, "grad_norm": 1.21875, "learning_rate": 2.8198961506371578e-06, "loss": 0.1252, "step": 27335 }, { "epoch": 169.79127725856696, "grad_norm": 1.046875, "learning_rate": 2.8187595349222994e-06, "loss": 0.0888, "step": 27336 }, { "epoch": 169.797507788162, "grad_norm": 2.078125, "learning_rate": 2.8176231346360826e-06, "loss": 0.1912, "step": 27337 }, { "epoch": 169.80373831775702, "grad_norm": 1.0703125, "learning_rate": 2.8164869497895512e-06, "loss": 0.1076, "step": 27338 }, { "epoch": 169.80996884735202, "grad_norm": 2.890625, "learning_rate": 2.8153509803937274e-06, "loss": 0.2033, "step": 27339 }, { "epoch": 169.81619937694705, "grad_norm": 2.921875, "learning_rate": 2.814215226459646e-06, "loss": 0.2968, "step": 27340 }, { "epoch": 169.82242990654206, "grad_norm": 2.609375, "learning_rate": 2.813079687998349e-06, "loss": 0.2411, "step": 27341 }, { "epoch": 169.82866043613708, "grad_norm": 2.4375, "learning_rate": 2.8119443650208578e-06, "loss": 0.2079, "step": 27342 }, { "epoch": 169.8348909657321, "grad_norm": 2.09375, "learning_rate": 2.810809257538194e-06, "loss": 0.1831, "step": 27343 }, { "epoch": 169.84112149532712, "grad_norm": 1.984375, "learning_rate": 2.809674365561382e-06, "loss": 0.1191, "step": 27344 }, { "epoch": 169.84735202492212, "grad_norm": 1.9375, "learning_rate": 2.808539689101458e-06, "loss": 0.1567, "step": 27345 }, { "epoch": 169.85358255451715, "grad_norm": 1.515625, "learning_rate": 2.8074052281694243e-06, "loss": 0.0947, "step": 27346 }, { "epoch": 169.85981308411215, "grad_norm": 2.734375, "learning_rate": 2.80627098277631e-06, "loss": 0.1978, "step": 27347 }, { "epoch": 169.86604361370718, "grad_norm": 1.90625, "learning_rate": 2.8051369529331294e-06, "loss": 0.2178, "step": 27348 }, { "epoch": 169.87227414330218, "grad_norm": 2.03125, "learning_rate": 2.8040031386508986e-06, "loss": 0.1658, "step": 27349 }, { "epoch": 169.8785046728972, "grad_norm": 2.6875, "learning_rate": 2.8028695399406195e-06, "loss": 0.1352, "step": 27350 }, { "epoch": 169.8847352024922, "grad_norm": 2.5625, "learning_rate": 2.801736156813312e-06, "loss": 0.1204, "step": 27351 }, { "epoch": 169.89096573208724, "grad_norm": 3.59375, "learning_rate": 2.800602989279982e-06, "loss": 0.2513, "step": 27352 }, { "epoch": 169.89719626168224, "grad_norm": 1.4296875, "learning_rate": 2.7994700373516285e-06, "loss": 0.1235, "step": 27353 }, { "epoch": 169.90342679127727, "grad_norm": 2.015625, "learning_rate": 2.798337301039264e-06, "loss": 0.1939, "step": 27354 }, { "epoch": 169.90965732087227, "grad_norm": 0.89453125, "learning_rate": 2.797204780353882e-06, "loss": 0.0703, "step": 27355 }, { "epoch": 169.9158878504673, "grad_norm": 2.578125, "learning_rate": 2.796072475306491e-06, "loss": 0.21, "step": 27356 }, { "epoch": 169.9221183800623, "grad_norm": 2.140625, "learning_rate": 2.7949403859080765e-06, "loss": 0.251, "step": 27357 }, { "epoch": 169.92834890965733, "grad_norm": 2.828125, "learning_rate": 2.7938085121696406e-06, "loss": 0.3027, "step": 27358 }, { "epoch": 169.93457943925233, "grad_norm": 2.921875, "learning_rate": 2.79267685410218e-06, "loss": 0.2353, "step": 27359 }, { "epoch": 169.94080996884736, "grad_norm": 1.2734375, "learning_rate": 2.791545411716676e-06, "loss": 0.1251, "step": 27360 }, { "epoch": 169.94704049844236, "grad_norm": 1.1015625, "learning_rate": 2.790414185024126e-06, "loss": 0.1107, "step": 27361 }, { "epoch": 169.9532710280374, "grad_norm": 1.5546875, "learning_rate": 2.7892831740355117e-06, "loss": 0.1105, "step": 27362 }, { "epoch": 169.9595015576324, "grad_norm": 1.28125, "learning_rate": 2.7881523787618215e-06, "loss": 0.1271, "step": 27363 }, { "epoch": 169.96573208722742, "grad_norm": 1.40625, "learning_rate": 2.787021799214029e-06, "loss": 0.1035, "step": 27364 }, { "epoch": 169.97196261682242, "grad_norm": 1.34375, "learning_rate": 2.7858914354031236e-06, "loss": 0.1064, "step": 27365 }, { "epoch": 169.97819314641745, "grad_norm": 3.328125, "learning_rate": 2.784761287340079e-06, "loss": 0.2245, "step": 27366 }, { "epoch": 169.98442367601245, "grad_norm": 2.265625, "learning_rate": 2.7836313550358844e-06, "loss": 0.2135, "step": 27367 }, { "epoch": 169.99065420560748, "grad_norm": 0.74609375, "learning_rate": 2.7825016385014946e-06, "loss": 0.0822, "step": 27368 }, { "epoch": 169.99688473520249, "grad_norm": 0.85546875, "learning_rate": 2.7813721377478867e-06, "loss": 0.0964, "step": 27369 }, { "epoch": 170.0, "grad_norm": 1.609375, "learning_rate": 2.7802428527860386e-06, "loss": 0.0703, "step": 27370 }, { "epoch": 170.00623052959503, "grad_norm": 1.5390625, "learning_rate": 2.779113783626916e-06, "loss": 0.1459, "step": 27371 }, { "epoch": 170.01246105919003, "grad_norm": 1.734375, "learning_rate": 2.777984930281477e-06, "loss": 0.1161, "step": 27372 }, { "epoch": 170.01869158878506, "grad_norm": 2.859375, "learning_rate": 2.776856292760696e-06, "loss": 0.2581, "step": 27373 }, { "epoch": 170.02492211838006, "grad_norm": 2.484375, "learning_rate": 2.775727871075523e-06, "loss": 0.2487, "step": 27374 }, { "epoch": 170.0311526479751, "grad_norm": 1.4140625, "learning_rate": 2.7745996652369295e-06, "loss": 0.0993, "step": 27375 }, { "epoch": 170.0373831775701, "grad_norm": 1.640625, "learning_rate": 2.773471675255859e-06, "loss": 0.1329, "step": 27376 }, { "epoch": 170.04361370716512, "grad_norm": 1.9296875, "learning_rate": 2.772343901143279e-06, "loss": 0.1489, "step": 27377 }, { "epoch": 170.04984423676012, "grad_norm": 1.625, "learning_rate": 2.7712163429101407e-06, "loss": 0.1487, "step": 27378 }, { "epoch": 170.05607476635515, "grad_norm": 1.4296875, "learning_rate": 2.7700890005673887e-06, "loss": 0.1434, "step": 27379 }, { "epoch": 170.06230529595015, "grad_norm": 1.3203125, "learning_rate": 2.7689618741259805e-06, "loss": 0.1051, "step": 27380 }, { "epoch": 170.06853582554518, "grad_norm": 2.953125, "learning_rate": 2.767834963596852e-06, "loss": 0.1385, "step": 27381 }, { "epoch": 170.07476635514018, "grad_norm": 2.09375, "learning_rate": 2.7667082689909613e-06, "loss": 0.1259, "step": 27382 }, { "epoch": 170.0809968847352, "grad_norm": 2.203125, "learning_rate": 2.765581790319241e-06, "loss": 0.1614, "step": 27383 }, { "epoch": 170.08722741433021, "grad_norm": 1.921875, "learning_rate": 2.7644555275926326e-06, "loss": 0.1668, "step": 27384 }, { "epoch": 170.09345794392524, "grad_norm": 1.7578125, "learning_rate": 2.763329480822083e-06, "loss": 0.1358, "step": 27385 }, { "epoch": 170.09968847352025, "grad_norm": 3.84375, "learning_rate": 2.7622036500185215e-06, "loss": 0.4353, "step": 27386 }, { "epoch": 170.10591900311528, "grad_norm": 1.5859375, "learning_rate": 2.7610780351928768e-06, "loss": 0.1252, "step": 27387 }, { "epoch": 170.11214953271028, "grad_norm": 1.3125, "learning_rate": 2.7599526363560892e-06, "loss": 0.1147, "step": 27388 }, { "epoch": 170.1183800623053, "grad_norm": 2.703125, "learning_rate": 2.7588274535190945e-06, "loss": 0.2955, "step": 27389 }, { "epoch": 170.1246105919003, "grad_norm": 2.765625, "learning_rate": 2.7577024866928062e-06, "loss": 0.128, "step": 27390 }, { "epoch": 170.13084112149534, "grad_norm": 1.7578125, "learning_rate": 2.756577735888158e-06, "loss": 0.113, "step": 27391 }, { "epoch": 170.13707165109034, "grad_norm": 2.84375, "learning_rate": 2.7554532011160762e-06, "loss": 0.2346, "step": 27392 }, { "epoch": 170.14330218068537, "grad_norm": 1.6796875, "learning_rate": 2.754328882387483e-06, "loss": 0.1346, "step": 27393 }, { "epoch": 170.14953271028037, "grad_norm": 1.921875, "learning_rate": 2.7532047797132867e-06, "loss": 0.1521, "step": 27394 }, { "epoch": 170.1557632398754, "grad_norm": 2.328125, "learning_rate": 2.752080893104414e-06, "loss": 0.2634, "step": 27395 }, { "epoch": 170.1619937694704, "grad_norm": 1.4296875, "learning_rate": 2.7509572225717846e-06, "loss": 0.101, "step": 27396 }, { "epoch": 170.16822429906543, "grad_norm": 2.171875, "learning_rate": 2.7498337681263002e-06, "loss": 0.1673, "step": 27397 }, { "epoch": 170.17445482866043, "grad_norm": 2.453125, "learning_rate": 2.7487105297788777e-06, "loss": 0.1867, "step": 27398 }, { "epoch": 170.18068535825546, "grad_norm": 2.390625, "learning_rate": 2.7475875075404306e-06, "loss": 0.21, "step": 27399 }, { "epoch": 170.18691588785046, "grad_norm": 1.75, "learning_rate": 2.7464647014218635e-06, "loss": 0.1667, "step": 27400 }, { "epoch": 170.1931464174455, "grad_norm": 1.0703125, "learning_rate": 2.7453421114340737e-06, "loss": 0.1063, "step": 27401 }, { "epoch": 170.1993769470405, "grad_norm": 1.953125, "learning_rate": 2.744219737587972e-06, "loss": 0.1969, "step": 27402 }, { "epoch": 170.20560747663552, "grad_norm": 2.328125, "learning_rate": 2.7430975798944548e-06, "loss": 0.1469, "step": 27403 }, { "epoch": 170.21183800623052, "grad_norm": 2.703125, "learning_rate": 2.74197563836443e-06, "loss": 0.1957, "step": 27404 }, { "epoch": 170.21806853582555, "grad_norm": 2.234375, "learning_rate": 2.740853913008784e-06, "loss": 0.1153, "step": 27405 }, { "epoch": 170.22429906542055, "grad_norm": 2.671875, "learning_rate": 2.7397324038384103e-06, "loss": 0.2329, "step": 27406 }, { "epoch": 170.23052959501558, "grad_norm": 1.5390625, "learning_rate": 2.738611110864206e-06, "loss": 0.1469, "step": 27407 }, { "epoch": 170.23676012461058, "grad_norm": 2.484375, "learning_rate": 2.7374900340970675e-06, "loss": 0.1757, "step": 27408 }, { "epoch": 170.2429906542056, "grad_norm": 3.859375, "learning_rate": 2.7363691735478673e-06, "loss": 0.1435, "step": 27409 }, { "epoch": 170.2492211838006, "grad_norm": 1.2265625, "learning_rate": 2.735248529227502e-06, "loss": 0.0952, "step": 27410 }, { "epoch": 170.25545171339564, "grad_norm": 1.8671875, "learning_rate": 2.73412810114686e-06, "loss": 0.1324, "step": 27411 }, { "epoch": 170.26168224299064, "grad_norm": 1.4296875, "learning_rate": 2.7330078893168132e-06, "loss": 0.1204, "step": 27412 }, { "epoch": 170.26791277258567, "grad_norm": 2.421875, "learning_rate": 2.731887893748242e-06, "loss": 0.1979, "step": 27413 }, { "epoch": 170.27414330218068, "grad_norm": 2.78125, "learning_rate": 2.730768114452029e-06, "loss": 0.2239, "step": 27414 }, { "epoch": 170.2803738317757, "grad_norm": 1.046875, "learning_rate": 2.7296485514390486e-06, "loss": 0.1146, "step": 27415 }, { "epoch": 170.2866043613707, "grad_norm": 2.078125, "learning_rate": 2.7285292047201706e-06, "loss": 0.1295, "step": 27416 }, { "epoch": 170.29283489096574, "grad_norm": 0.69140625, "learning_rate": 2.727410074306266e-06, "loss": 0.0802, "step": 27417 }, { "epoch": 170.29906542056074, "grad_norm": 2.25, "learning_rate": 2.7262911602082165e-06, "loss": 0.1452, "step": 27418 }, { "epoch": 170.30529595015577, "grad_norm": 1.390625, "learning_rate": 2.725172462436876e-06, "loss": 0.0979, "step": 27419 }, { "epoch": 170.31152647975077, "grad_norm": 1.125, "learning_rate": 2.724053981003111e-06, "loss": 0.1061, "step": 27420 }, { "epoch": 170.3177570093458, "grad_norm": 2.03125, "learning_rate": 2.722935715917785e-06, "loss": 0.1869, "step": 27421 }, { "epoch": 170.3239875389408, "grad_norm": 2.84375, "learning_rate": 2.7218176671917646e-06, "loss": 0.179, "step": 27422 }, { "epoch": 170.33021806853583, "grad_norm": 2.375, "learning_rate": 2.7206998348359018e-06, "loss": 0.1181, "step": 27423 }, { "epoch": 170.33644859813083, "grad_norm": 2.03125, "learning_rate": 2.719582218861058e-06, "loss": 0.192, "step": 27424 }, { "epoch": 170.34267912772586, "grad_norm": 1.2265625, "learning_rate": 2.7184648192780825e-06, "loss": 0.1235, "step": 27425 }, { "epoch": 170.34890965732086, "grad_norm": 1.4375, "learning_rate": 2.7173476360978334e-06, "loss": 0.1152, "step": 27426 }, { "epoch": 170.3551401869159, "grad_norm": 3.109375, "learning_rate": 2.7162306693311544e-06, "loss": 0.3132, "step": 27427 }, { "epoch": 170.3613707165109, "grad_norm": 2.34375, "learning_rate": 2.7151139189888986e-06, "loss": 0.2366, "step": 27428 }, { "epoch": 170.36760124610592, "grad_norm": 1.96875, "learning_rate": 2.7139973850819122e-06, "loss": 0.1917, "step": 27429 }, { "epoch": 170.37383177570092, "grad_norm": 2.6875, "learning_rate": 2.71288106762104e-06, "loss": 0.1414, "step": 27430 }, { "epoch": 170.38006230529595, "grad_norm": 2.375, "learning_rate": 2.7117649666171226e-06, "loss": 0.1923, "step": 27431 }, { "epoch": 170.38629283489095, "grad_norm": 1.5625, "learning_rate": 2.710649082080996e-06, "loss": 0.1109, "step": 27432 }, { "epoch": 170.39252336448598, "grad_norm": 2.5625, "learning_rate": 2.709533414023499e-06, "loss": 0.3148, "step": 27433 }, { "epoch": 170.39875389408098, "grad_norm": 2.296875, "learning_rate": 2.708417962455476e-06, "loss": 0.1525, "step": 27434 }, { "epoch": 170.404984423676, "grad_norm": 2.15625, "learning_rate": 2.7073027273877477e-06, "loss": 0.1413, "step": 27435 }, { "epoch": 170.41121495327104, "grad_norm": 0.8828125, "learning_rate": 2.706187708831151e-06, "loss": 0.0984, "step": 27436 }, { "epoch": 170.41744548286604, "grad_norm": 0.8125, "learning_rate": 2.705072906796527e-06, "loss": 0.0906, "step": 27437 }, { "epoch": 170.42367601246107, "grad_norm": 2.640625, "learning_rate": 2.703958321294681e-06, "loss": 0.2363, "step": 27438 }, { "epoch": 170.42990654205607, "grad_norm": 1.875, "learning_rate": 2.702843952336448e-06, "loss": 0.1151, "step": 27439 }, { "epoch": 170.4361370716511, "grad_norm": 3.640625, "learning_rate": 2.7017297999326537e-06, "loss": 0.3706, "step": 27440 }, { "epoch": 170.4423676012461, "grad_norm": 1.7421875, "learning_rate": 2.7006158640941193e-06, "loss": 0.0952, "step": 27441 }, { "epoch": 170.44859813084113, "grad_norm": 2.703125, "learning_rate": 2.6995021448316588e-06, "loss": 0.2254, "step": 27442 }, { "epoch": 170.45482866043614, "grad_norm": 2.203125, "learning_rate": 2.6983886421560937e-06, "loss": 0.2098, "step": 27443 }, { "epoch": 170.46105919003116, "grad_norm": 1.828125, "learning_rate": 2.6972753560782328e-06, "loss": 0.2252, "step": 27444 }, { "epoch": 170.46728971962617, "grad_norm": 3.1875, "learning_rate": 2.696162286608897e-06, "loss": 0.1459, "step": 27445 }, { "epoch": 170.4735202492212, "grad_norm": 2.28125, "learning_rate": 2.6950494337588866e-06, "loss": 0.2214, "step": 27446 }, { "epoch": 170.4797507788162, "grad_norm": 1.125, "learning_rate": 2.693936797539012e-06, "loss": 0.0977, "step": 27447 }, { "epoch": 170.48598130841123, "grad_norm": 2.21875, "learning_rate": 2.6928243779600903e-06, "loss": 0.1602, "step": 27448 }, { "epoch": 170.49221183800623, "grad_norm": 1.7578125, "learning_rate": 2.691712175032912e-06, "loss": 0.1756, "step": 27449 }, { "epoch": 170.49844236760126, "grad_norm": 2.8125, "learning_rate": 2.6906001887682885e-06, "loss": 0.161, "step": 27450 }, { "epoch": 170.50467289719626, "grad_norm": 2.625, "learning_rate": 2.6894884191770107e-06, "loss": 0.2135, "step": 27451 }, { "epoch": 170.5109034267913, "grad_norm": 3.140625, "learning_rate": 2.6883768662698873e-06, "loss": 0.1836, "step": 27452 }, { "epoch": 170.5171339563863, "grad_norm": 1.8125, "learning_rate": 2.6872655300577005e-06, "loss": 0.1691, "step": 27453 }, { "epoch": 170.52336448598132, "grad_norm": 2.171875, "learning_rate": 2.686154410551253e-06, "loss": 0.242, "step": 27454 }, { "epoch": 170.52959501557632, "grad_norm": 2.21875, "learning_rate": 2.6850435077613366e-06, "loss": 0.1764, "step": 27455 }, { "epoch": 170.53582554517135, "grad_norm": 1.4296875, "learning_rate": 2.6839328216987427e-06, "loss": 0.1251, "step": 27456 }, { "epoch": 170.54205607476635, "grad_norm": 1.7890625, "learning_rate": 2.682822352374245e-06, "loss": 0.1277, "step": 27457 }, { "epoch": 170.54828660436138, "grad_norm": 1.96875, "learning_rate": 2.681712099798639e-06, "loss": 0.1441, "step": 27458 }, { "epoch": 170.55451713395638, "grad_norm": 1.6484375, "learning_rate": 2.680602063982712e-06, "loss": 0.1574, "step": 27459 }, { "epoch": 170.5607476635514, "grad_norm": 2.421875, "learning_rate": 2.6794922449372345e-06, "loss": 0.1696, "step": 27460 }, { "epoch": 170.5669781931464, "grad_norm": 1.0703125, "learning_rate": 2.6783826426729914e-06, "loss": 0.1136, "step": 27461 }, { "epoch": 170.57320872274144, "grad_norm": 2.53125, "learning_rate": 2.6772732572007604e-06, "loss": 0.1576, "step": 27462 }, { "epoch": 170.57943925233644, "grad_norm": 1.3359375, "learning_rate": 2.6761640885313133e-06, "loss": 0.1115, "step": 27463 }, { "epoch": 170.58566978193147, "grad_norm": 2.765625, "learning_rate": 2.6750551366754193e-06, "loss": 0.2331, "step": 27464 }, { "epoch": 170.59190031152647, "grad_norm": 2.0, "learning_rate": 2.673946401643854e-06, "loss": 0.177, "step": 27465 }, { "epoch": 170.5981308411215, "grad_norm": 3.0625, "learning_rate": 2.672837883447382e-06, "loss": 0.2443, "step": 27466 }, { "epoch": 170.6043613707165, "grad_norm": 2.6875, "learning_rate": 2.6717295820967796e-06, "loss": 0.2882, "step": 27467 }, { "epoch": 170.61059190031153, "grad_norm": 3.171875, "learning_rate": 2.6706214976027934e-06, "loss": 0.1362, "step": 27468 }, { "epoch": 170.61682242990653, "grad_norm": 1.9296875, "learning_rate": 2.669513629976203e-06, "loss": 0.2081, "step": 27469 }, { "epoch": 170.62305295950156, "grad_norm": 2.71875, "learning_rate": 2.668405979227753e-06, "loss": 0.235, "step": 27470 }, { "epoch": 170.62928348909657, "grad_norm": 2.015625, "learning_rate": 2.6672985453682158e-06, "loss": 0.1023, "step": 27471 }, { "epoch": 170.6355140186916, "grad_norm": 1.4140625, "learning_rate": 2.6661913284083324e-06, "loss": 0.136, "step": 27472 }, { "epoch": 170.6417445482866, "grad_norm": 1.6875, "learning_rate": 2.6650843283588633e-06, "loss": 0.1703, "step": 27473 }, { "epoch": 170.64797507788163, "grad_norm": 2.3125, "learning_rate": 2.6639775452305645e-06, "loss": 0.2176, "step": 27474 }, { "epoch": 170.65420560747663, "grad_norm": 1.6015625, "learning_rate": 2.6628709790341823e-06, "loss": 0.1081, "step": 27475 }, { "epoch": 170.66043613707166, "grad_norm": 0.7578125, "learning_rate": 2.6617646297804553e-06, "loss": 0.0759, "step": 27476 }, { "epoch": 170.66666666666666, "grad_norm": 1.28125, "learning_rate": 2.6606584974801367e-06, "loss": 0.122, "step": 27477 }, { "epoch": 170.6728971962617, "grad_norm": 1.8984375, "learning_rate": 2.6595525821439732e-06, "loss": 0.1705, "step": 27478 }, { "epoch": 170.6791277258567, "grad_norm": 2.578125, "learning_rate": 2.6584468837826946e-06, "loss": 0.2415, "step": 27479 }, { "epoch": 170.68535825545172, "grad_norm": 3.109375, "learning_rate": 2.6573414024070485e-06, "loss": 0.2942, "step": 27480 }, { "epoch": 170.69158878504672, "grad_norm": 1.921875, "learning_rate": 2.6562361380277705e-06, "loss": 0.1675, "step": 27481 }, { "epoch": 170.69781931464175, "grad_norm": 2.0625, "learning_rate": 2.6551310906555964e-06, "loss": 0.1036, "step": 27482 }, { "epoch": 170.70404984423675, "grad_norm": 1.890625, "learning_rate": 2.654026260301251e-06, "loss": 0.1283, "step": 27483 }, { "epoch": 170.71028037383178, "grad_norm": 1.6015625, "learning_rate": 2.6529216469754677e-06, "loss": 0.1231, "step": 27484 }, { "epoch": 170.71651090342678, "grad_norm": 1.859375, "learning_rate": 2.6518172506889815e-06, "loss": 0.1593, "step": 27485 }, { "epoch": 170.7227414330218, "grad_norm": 1.9296875, "learning_rate": 2.6507130714525095e-06, "loss": 0.1571, "step": 27486 }, { "epoch": 170.7289719626168, "grad_norm": 3.125, "learning_rate": 2.6496091092767794e-06, "loss": 0.3339, "step": 27487 }, { "epoch": 170.73520249221184, "grad_norm": 1.2421875, "learning_rate": 2.6485053641725183e-06, "loss": 0.1018, "step": 27488 }, { "epoch": 170.74143302180684, "grad_norm": 2.421875, "learning_rate": 2.64740183615044e-06, "loss": 0.1915, "step": 27489 }, { "epoch": 170.74766355140187, "grad_norm": 1.5703125, "learning_rate": 2.646298525221258e-06, "loss": 0.1359, "step": 27490 }, { "epoch": 170.75389408099687, "grad_norm": 1.859375, "learning_rate": 2.6451954313956946e-06, "loss": 0.11, "step": 27491 }, { "epoch": 170.7601246105919, "grad_norm": 1.4765625, "learning_rate": 2.644092554684469e-06, "loss": 0.1284, "step": 27492 }, { "epoch": 170.7663551401869, "grad_norm": 1.703125, "learning_rate": 2.6429898950982772e-06, "loss": 0.1409, "step": 27493 }, { "epoch": 170.77258566978193, "grad_norm": 3.4375, "learning_rate": 2.64188745264784e-06, "loss": 0.1648, "step": 27494 }, { "epoch": 170.77881619937693, "grad_norm": 2.9375, "learning_rate": 2.6407852273438586e-06, "loss": 0.2947, "step": 27495 }, { "epoch": 170.78504672897196, "grad_norm": 2.09375, "learning_rate": 2.639683219197048e-06, "loss": 0.1522, "step": 27496 }, { "epoch": 170.79127725856696, "grad_norm": 1.4140625, "learning_rate": 2.6385814282180953e-06, "loss": 0.1175, "step": 27497 }, { "epoch": 170.797507788162, "grad_norm": 2.015625, "learning_rate": 2.6374798544177098e-06, "loss": 0.1905, "step": 27498 }, { "epoch": 170.80373831775702, "grad_norm": 2.484375, "learning_rate": 2.636378497806591e-06, "loss": 0.2187, "step": 27499 }, { "epoch": 170.80996884735202, "grad_norm": 2.0, "learning_rate": 2.6352773583954384e-06, "loss": 0.2278, "step": 27500 }, { "epoch": 170.81619937694705, "grad_norm": 2.140625, "learning_rate": 2.634176436194946e-06, "loss": 0.2041, "step": 27501 }, { "epoch": 170.82242990654206, "grad_norm": 2.03125, "learning_rate": 2.633075731215795e-06, "loss": 0.1482, "step": 27502 }, { "epoch": 170.82866043613708, "grad_norm": 2.3125, "learning_rate": 2.6319752434686845e-06, "loss": 0.2076, "step": 27503 }, { "epoch": 170.8348909657321, "grad_norm": 2.5625, "learning_rate": 2.6308749729643063e-06, "loss": 0.1378, "step": 27504 }, { "epoch": 170.84112149532712, "grad_norm": 2.96875, "learning_rate": 2.6297749197133377e-06, "loss": 0.247, "step": 27505 }, { "epoch": 170.84735202492212, "grad_norm": 1.4375, "learning_rate": 2.6286750837264644e-06, "loss": 0.1151, "step": 27506 }, { "epoch": 170.85358255451715, "grad_norm": 1.1953125, "learning_rate": 2.6275754650143787e-06, "loss": 0.111, "step": 27507 }, { "epoch": 170.85981308411215, "grad_norm": 2.75, "learning_rate": 2.6264760635877518e-06, "loss": 0.1699, "step": 27508 }, { "epoch": 170.86604361370718, "grad_norm": 1.7578125, "learning_rate": 2.625376879457253e-06, "loss": 0.1225, "step": 27509 }, { "epoch": 170.87227414330218, "grad_norm": 1.5703125, "learning_rate": 2.6242779126335717e-06, "loss": 0.1222, "step": 27510 }, { "epoch": 170.8785046728972, "grad_norm": 2.953125, "learning_rate": 2.623179163127379e-06, "loss": 0.2034, "step": 27511 }, { "epoch": 170.8847352024922, "grad_norm": 1.3046875, "learning_rate": 2.622080630949336e-06, "loss": 0.127, "step": 27512 }, { "epoch": 170.89096573208724, "grad_norm": 1.2734375, "learning_rate": 2.6209823161101268e-06, "loss": 0.0921, "step": 27513 }, { "epoch": 170.89719626168224, "grad_norm": 1.015625, "learning_rate": 2.6198842186204053e-06, "loss": 0.0866, "step": 27514 }, { "epoch": 170.90342679127727, "grad_norm": 2.453125, "learning_rate": 2.618786338490847e-06, "loss": 0.1448, "step": 27515 }, { "epoch": 170.90965732087227, "grad_norm": 1.7890625, "learning_rate": 2.617688675732105e-06, "loss": 0.1209, "step": 27516 }, { "epoch": 170.9158878504673, "grad_norm": 1.796875, "learning_rate": 2.616591230354845e-06, "loss": 0.1427, "step": 27517 }, { "epoch": 170.9221183800623, "grad_norm": 1.6640625, "learning_rate": 2.6154940023697284e-06, "loss": 0.1071, "step": 27518 }, { "epoch": 170.92834890965733, "grad_norm": 1.859375, "learning_rate": 2.6143969917874074e-06, "loss": 0.1453, "step": 27519 }, { "epoch": 170.93457943925233, "grad_norm": 2.515625, "learning_rate": 2.61330019861854e-06, "loss": 0.1536, "step": 27520 }, { "epoch": 170.94080996884736, "grad_norm": 2.515625, "learning_rate": 2.612203622873771e-06, "loss": 0.2019, "step": 27521 }, { "epoch": 170.94704049844236, "grad_norm": 1.890625, "learning_rate": 2.6111072645637614e-06, "loss": 0.1477, "step": 27522 }, { "epoch": 170.9532710280374, "grad_norm": 1.9140625, "learning_rate": 2.6100111236991463e-06, "loss": 0.1563, "step": 27523 }, { "epoch": 170.9595015576324, "grad_norm": 2.9375, "learning_rate": 2.608915200290582e-06, "loss": 0.2135, "step": 27524 }, { "epoch": 170.96573208722742, "grad_norm": 2.765625, "learning_rate": 2.6078194943487066e-06, "loss": 0.2478, "step": 27525 }, { "epoch": 170.97196261682242, "grad_norm": 2.734375, "learning_rate": 2.606724005884176e-06, "loss": 0.1901, "step": 27526 }, { "epoch": 170.97819314641745, "grad_norm": 1.9765625, "learning_rate": 2.605628734907606e-06, "loss": 0.1506, "step": 27527 }, { "epoch": 170.98442367601245, "grad_norm": 1.8125, "learning_rate": 2.6045336814296493e-06, "loss": 0.1442, "step": 27528 }, { "epoch": 170.99065420560748, "grad_norm": 3.328125, "learning_rate": 2.6034388454609372e-06, "loss": 0.319, "step": 27529 }, { "epoch": 170.99688473520249, "grad_norm": 2.140625, "learning_rate": 2.602344227012107e-06, "loss": 0.1799, "step": 27530 }, { "epoch": 171.0, "grad_norm": 1.8359375, "learning_rate": 2.601249826093785e-06, "loss": 0.0996, "step": 27531 }, { "epoch": 171.00623052959503, "grad_norm": 0.8203125, "learning_rate": 2.6001556427166063e-06, "loss": 0.0836, "step": 27532 }, { "epoch": 171.01246105919003, "grad_norm": 0.8828125, "learning_rate": 2.599061676891193e-06, "loss": 0.0874, "step": 27533 }, { "epoch": 171.01869158878506, "grad_norm": 1.3125, "learning_rate": 2.5979679286281644e-06, "loss": 0.1067, "step": 27534 }, { "epoch": 171.02492211838006, "grad_norm": 2.765625, "learning_rate": 2.596874397938151e-06, "loss": 0.1564, "step": 27535 }, { "epoch": 171.0311526479751, "grad_norm": 2.25, "learning_rate": 2.595781084831772e-06, "loss": 0.2199, "step": 27536 }, { "epoch": 171.0373831775701, "grad_norm": 1.7109375, "learning_rate": 2.5946879893196498e-06, "loss": 0.1596, "step": 27537 }, { "epoch": 171.04361370716512, "grad_norm": 1.78125, "learning_rate": 2.5935951114123917e-06, "loss": 0.1405, "step": 27538 }, { "epoch": 171.04984423676012, "grad_norm": 1.9453125, "learning_rate": 2.5925024511206207e-06, "loss": 0.117, "step": 27539 }, { "epoch": 171.05607476635515, "grad_norm": 0.91015625, "learning_rate": 2.591410008454942e-06, "loss": 0.0817, "step": 27540 }, { "epoch": 171.06230529595015, "grad_norm": 1.6171875, "learning_rate": 2.590317783425972e-06, "loss": 0.1464, "step": 27541 }, { "epoch": 171.06853582554518, "grad_norm": 1.5390625, "learning_rate": 2.58922577604431e-06, "loss": 0.151, "step": 27542 }, { "epoch": 171.07476635514018, "grad_norm": 2.109375, "learning_rate": 2.5881339863205707e-06, "loss": 0.1784, "step": 27543 }, { "epoch": 171.0809968847352, "grad_norm": 2.1875, "learning_rate": 2.587042414265356e-06, "loss": 0.207, "step": 27544 }, { "epoch": 171.08722741433021, "grad_norm": 4.0, "learning_rate": 2.585951059889266e-06, "loss": 0.4375, "step": 27545 }, { "epoch": 171.09345794392524, "grad_norm": 1.2578125, "learning_rate": 2.5848599232028953e-06, "loss": 0.1321, "step": 27546 }, { "epoch": 171.09968847352025, "grad_norm": 1.15625, "learning_rate": 2.583769004216846e-06, "loss": 0.1188, "step": 27547 }, { "epoch": 171.10591900311528, "grad_norm": 2.0625, "learning_rate": 2.5826783029417157e-06, "loss": 0.156, "step": 27548 }, { "epoch": 171.11214953271028, "grad_norm": 1.5, "learning_rate": 2.5815878193880927e-06, "loss": 0.1533, "step": 27549 }, { "epoch": 171.1183800623053, "grad_norm": 2.296875, "learning_rate": 2.580497553566569e-06, "loss": 0.1754, "step": 27550 }, { "epoch": 171.1246105919003, "grad_norm": 2.453125, "learning_rate": 2.579407505487741e-06, "loss": 0.1304, "step": 27551 }, { "epoch": 171.13084112149534, "grad_norm": 1.2421875, "learning_rate": 2.5783176751621867e-06, "loss": 0.1174, "step": 27552 }, { "epoch": 171.13707165109034, "grad_norm": 1.25, "learning_rate": 2.5772280626004896e-06, "loss": 0.1091, "step": 27553 }, { "epoch": 171.14330218068537, "grad_norm": 1.1796875, "learning_rate": 2.5761386678132355e-06, "loss": 0.0887, "step": 27554 }, { "epoch": 171.14953271028037, "grad_norm": 1.9765625, "learning_rate": 2.57504949081101e-06, "loss": 0.1086, "step": 27555 }, { "epoch": 171.1557632398754, "grad_norm": 2.34375, "learning_rate": 2.573960531604383e-06, "loss": 0.1684, "step": 27556 }, { "epoch": 171.1619937694704, "grad_norm": 2.90625, "learning_rate": 2.5728717902039316e-06, "loss": 0.2616, "step": 27557 }, { "epoch": 171.16822429906543, "grad_norm": 2.015625, "learning_rate": 2.5717832666202396e-06, "loss": 0.1336, "step": 27558 }, { "epoch": 171.17445482866043, "grad_norm": 1.1953125, "learning_rate": 2.5706949608638704e-06, "loss": 0.1086, "step": 27559 }, { "epoch": 171.18068535825546, "grad_norm": 2.875, "learning_rate": 2.5696068729453903e-06, "loss": 0.2478, "step": 27560 }, { "epoch": 171.18691588785046, "grad_norm": 2.890625, "learning_rate": 2.568519002875372e-06, "loss": 0.29, "step": 27561 }, { "epoch": 171.1931464174455, "grad_norm": 3.125, "learning_rate": 2.5674313506643815e-06, "loss": 0.2536, "step": 27562 }, { "epoch": 171.1993769470405, "grad_norm": 2.25, "learning_rate": 2.5663439163229884e-06, "loss": 0.1802, "step": 27563 }, { "epoch": 171.20560747663552, "grad_norm": 1.3671875, "learning_rate": 2.5652566998617455e-06, "loss": 0.1082, "step": 27564 }, { "epoch": 171.21183800623052, "grad_norm": 2.953125, "learning_rate": 2.564169701291211e-06, "loss": 0.2189, "step": 27565 }, { "epoch": 171.21806853582555, "grad_norm": 2.1875, "learning_rate": 2.5630829206219424e-06, "loss": 0.2448, "step": 27566 }, { "epoch": 171.22429906542055, "grad_norm": 2.203125, "learning_rate": 2.5619963578645047e-06, "loss": 0.2479, "step": 27567 }, { "epoch": 171.23052959501558, "grad_norm": 1.1796875, "learning_rate": 2.5609100130294383e-06, "loss": 0.0876, "step": 27568 }, { "epoch": 171.23676012461058, "grad_norm": 2.171875, "learning_rate": 2.5598238861272994e-06, "loss": 0.137, "step": 27569 }, { "epoch": 171.2429906542056, "grad_norm": 1.796875, "learning_rate": 2.55873797716864e-06, "loss": 0.1274, "step": 27570 }, { "epoch": 171.2492211838006, "grad_norm": 2.28125, "learning_rate": 2.5576522861640055e-06, "loss": 0.125, "step": 27571 }, { "epoch": 171.25545171339564, "grad_norm": 2.921875, "learning_rate": 2.556566813123931e-06, "loss": 0.2155, "step": 27572 }, { "epoch": 171.26168224299064, "grad_norm": 2.140625, "learning_rate": 2.5554815580589668e-06, "loss": 0.1555, "step": 27573 }, { "epoch": 171.26791277258567, "grad_norm": 1.28125, "learning_rate": 2.55439652097966e-06, "loss": 0.1214, "step": 27574 }, { "epoch": 171.27414330218068, "grad_norm": 2.359375, "learning_rate": 2.5533117018965325e-06, "loss": 0.1638, "step": 27575 }, { "epoch": 171.2803738317757, "grad_norm": 1.125, "learning_rate": 2.5522271008201286e-06, "loss": 0.1328, "step": 27576 }, { "epoch": 171.2866043613707, "grad_norm": 2.53125, "learning_rate": 2.5511427177609904e-06, "loss": 0.1762, "step": 27577 }, { "epoch": 171.29283489096574, "grad_norm": 3.234375, "learning_rate": 2.5500585527296396e-06, "loss": 0.3586, "step": 27578 }, { "epoch": 171.29906542056074, "grad_norm": 2.0, "learning_rate": 2.5489746057366037e-06, "loss": 0.1816, "step": 27579 }, { "epoch": 171.30529595015577, "grad_norm": 2.765625, "learning_rate": 2.547890876792414e-06, "loss": 0.1575, "step": 27580 }, { "epoch": 171.31152647975077, "grad_norm": 2.484375, "learning_rate": 2.5468073659076e-06, "loss": 0.2057, "step": 27581 }, { "epoch": 171.3177570093458, "grad_norm": 1.7265625, "learning_rate": 2.5457240730926756e-06, "loss": 0.1453, "step": 27582 }, { "epoch": 171.3239875389408, "grad_norm": 2.28125, "learning_rate": 2.5446409983581747e-06, "loss": 0.2838, "step": 27583 }, { "epoch": 171.33021806853583, "grad_norm": 0.703125, "learning_rate": 2.5435581417146025e-06, "loss": 0.082, "step": 27584 }, { "epoch": 171.33644859813083, "grad_norm": 2.453125, "learning_rate": 2.5424755031724893e-06, "loss": 0.2262, "step": 27585 }, { "epoch": 171.34267912772586, "grad_norm": 0.953125, "learning_rate": 2.5413930827423384e-06, "loss": 0.1086, "step": 27586 }, { "epoch": 171.34890965732086, "grad_norm": 1.0390625, "learning_rate": 2.5403108804346653e-06, "loss": 0.1111, "step": 27587 }, { "epoch": 171.3551401869159, "grad_norm": 1.03125, "learning_rate": 2.5392288962599874e-06, "loss": 0.0911, "step": 27588 }, { "epoch": 171.3613707165109, "grad_norm": 1.7890625, "learning_rate": 2.5381471302288047e-06, "loss": 0.1547, "step": 27589 }, { "epoch": 171.36760124610592, "grad_norm": 2.46875, "learning_rate": 2.5370655823516306e-06, "loss": 0.1658, "step": 27590 }, { "epoch": 171.37383177570092, "grad_norm": 1.875, "learning_rate": 2.5359842526389623e-06, "loss": 0.2153, "step": 27591 }, { "epoch": 171.38006230529595, "grad_norm": 2.90625, "learning_rate": 2.5349031411013106e-06, "loss": 0.1075, "step": 27592 }, { "epoch": 171.38629283489095, "grad_norm": 2.328125, "learning_rate": 2.533822247749165e-06, "loss": 0.1209, "step": 27593 }, { "epoch": 171.39252336448598, "grad_norm": 2.75, "learning_rate": 2.532741572593028e-06, "loss": 0.2525, "step": 27594 }, { "epoch": 171.39875389408098, "grad_norm": 2.515625, "learning_rate": 2.531661115643397e-06, "loss": 0.1489, "step": 27595 }, { "epoch": 171.404984423676, "grad_norm": 2.59375, "learning_rate": 2.5305808769107735e-06, "loss": 0.2051, "step": 27596 }, { "epoch": 171.41121495327104, "grad_norm": 1.2109375, "learning_rate": 2.529500856405631e-06, "loss": 0.0968, "step": 27597 }, { "epoch": 171.41744548286604, "grad_norm": 1.7421875, "learning_rate": 2.5284210541384663e-06, "loss": 0.1404, "step": 27598 }, { "epoch": 171.42367601246107, "grad_norm": 3.03125, "learning_rate": 2.5273414701197707e-06, "loss": 0.2099, "step": 27599 }, { "epoch": 171.42990654205607, "grad_norm": 1.046875, "learning_rate": 2.5262621043600305e-06, "loss": 0.116, "step": 27600 }, { "epoch": 171.4361370716511, "grad_norm": 2.640625, "learning_rate": 2.5251829568697207e-06, "loss": 0.2211, "step": 27601 }, { "epoch": 171.4423676012461, "grad_norm": 2.25, "learning_rate": 2.5241040276593296e-06, "loss": 0.2033, "step": 27602 }, { "epoch": 171.44859813084113, "grad_norm": 2.171875, "learning_rate": 2.5230253167393303e-06, "loss": 0.1877, "step": 27603 }, { "epoch": 171.45482866043614, "grad_norm": 1.3671875, "learning_rate": 2.5219468241202054e-06, "loss": 0.1352, "step": 27604 }, { "epoch": 171.46105919003116, "grad_norm": 2.84375, "learning_rate": 2.520868549812422e-06, "loss": 0.279, "step": 27605 }, { "epoch": 171.46728971962617, "grad_norm": 1.8671875, "learning_rate": 2.519790493826457e-06, "loss": 0.112, "step": 27606 }, { "epoch": 171.4735202492212, "grad_norm": 1.8203125, "learning_rate": 2.5187126561727865e-06, "loss": 0.1637, "step": 27607 }, { "epoch": 171.4797507788162, "grad_norm": 1.796875, "learning_rate": 2.5176350368618652e-06, "loss": 0.1512, "step": 27608 }, { "epoch": 171.48598130841123, "grad_norm": 2.390625, "learning_rate": 2.5165576359041737e-06, "loss": 0.1772, "step": 27609 }, { "epoch": 171.49221183800623, "grad_norm": 2.6875, "learning_rate": 2.5154804533101623e-06, "loss": 0.2081, "step": 27610 }, { "epoch": 171.49844236760126, "grad_norm": 1.609375, "learning_rate": 2.514403489090306e-06, "loss": 0.1567, "step": 27611 }, { "epoch": 171.50467289719626, "grad_norm": 2.203125, "learning_rate": 2.5133267432550512e-06, "loss": 0.1222, "step": 27612 }, { "epoch": 171.5109034267913, "grad_norm": 1.1328125, "learning_rate": 2.5122502158148626e-06, "loss": 0.122, "step": 27613 }, { "epoch": 171.5171339563863, "grad_norm": 1.703125, "learning_rate": 2.511173906780201e-06, "loss": 0.1286, "step": 27614 }, { "epoch": 171.52336448598132, "grad_norm": 1.8203125, "learning_rate": 2.5100978161615135e-06, "loss": 0.1714, "step": 27615 }, { "epoch": 171.52959501557632, "grad_norm": 1.484375, "learning_rate": 2.5090219439692447e-06, "loss": 0.1357, "step": 27616 }, { "epoch": 171.53582554517135, "grad_norm": 3.140625, "learning_rate": 2.5079462902138527e-06, "loss": 0.2903, "step": 27617 }, { "epoch": 171.54205607476635, "grad_norm": 2.078125, "learning_rate": 2.506870854905788e-06, "loss": 0.2025, "step": 27618 }, { "epoch": 171.54828660436138, "grad_norm": 1.921875, "learning_rate": 2.505795638055483e-06, "loss": 0.1428, "step": 27619 }, { "epoch": 171.55451713395638, "grad_norm": 0.92578125, "learning_rate": 2.5047206396733886e-06, "loss": 0.1077, "step": 27620 }, { "epoch": 171.5607476635514, "grad_norm": 1.4140625, "learning_rate": 2.503645859769943e-06, "loss": 0.1254, "step": 27621 }, { "epoch": 171.5669781931464, "grad_norm": 1.359375, "learning_rate": 2.5025712983555965e-06, "loss": 0.1171, "step": 27622 }, { "epoch": 171.57320872274144, "grad_norm": 1.7890625, "learning_rate": 2.501496955440763e-06, "loss": 0.1363, "step": 27623 }, { "epoch": 171.57943925233644, "grad_norm": 2.609375, "learning_rate": 2.5004228310358898e-06, "loss": 0.2007, "step": 27624 }, { "epoch": 171.58566978193147, "grad_norm": 2.109375, "learning_rate": 2.4993489251514095e-06, "loss": 0.1916, "step": 27625 }, { "epoch": 171.59190031152647, "grad_norm": 3.453125, "learning_rate": 2.498275237797751e-06, "loss": 0.3079, "step": 27626 }, { "epoch": 171.5981308411215, "grad_norm": 1.8359375, "learning_rate": 2.497201768985341e-06, "loss": 0.1311, "step": 27627 }, { "epoch": 171.6043613707165, "grad_norm": 2.171875, "learning_rate": 2.496128518724608e-06, "loss": 0.1466, "step": 27628 }, { "epoch": 171.61059190031153, "grad_norm": 1.984375, "learning_rate": 2.4950554870259713e-06, "loss": 0.1304, "step": 27629 }, { "epoch": 171.61682242990653, "grad_norm": 1.3359375, "learning_rate": 2.493982673899853e-06, "loss": 0.1235, "step": 27630 }, { "epoch": 171.62305295950156, "grad_norm": 1.9921875, "learning_rate": 2.492910079356672e-06, "loss": 0.1496, "step": 27631 }, { "epoch": 171.62928348909657, "grad_norm": 2.8125, "learning_rate": 2.491837703406849e-06, "loss": 0.2169, "step": 27632 }, { "epoch": 171.6355140186916, "grad_norm": 2.90625, "learning_rate": 2.4907655460607997e-06, "loss": 0.3167, "step": 27633 }, { "epoch": 171.6417445482866, "grad_norm": 2.484375, "learning_rate": 2.4896936073289383e-06, "loss": 0.2224, "step": 27634 }, { "epoch": 171.64797507788163, "grad_norm": 1.734375, "learning_rate": 2.4886218872216648e-06, "loss": 0.101, "step": 27635 }, { "epoch": 171.65420560747663, "grad_norm": 1.4140625, "learning_rate": 2.487550385749396e-06, "loss": 0.1227, "step": 27636 }, { "epoch": 171.66043613707166, "grad_norm": 2.125, "learning_rate": 2.4864791029225456e-06, "loss": 0.2164, "step": 27637 }, { "epoch": 171.66666666666666, "grad_norm": 2.71875, "learning_rate": 2.485408038751502e-06, "loss": 0.1889, "step": 27638 }, { "epoch": 171.6728971962617, "grad_norm": 1.671875, "learning_rate": 2.4843371932466774e-06, "loss": 0.1719, "step": 27639 }, { "epoch": 171.6791277258567, "grad_norm": 1.8984375, "learning_rate": 2.4832665664184767e-06, "loss": 0.1751, "step": 27640 }, { "epoch": 171.68535825545172, "grad_norm": 1.828125, "learning_rate": 2.482196158277289e-06, "loss": 0.1386, "step": 27641 }, { "epoch": 171.69158878504672, "grad_norm": 1.546875, "learning_rate": 2.4811259688335114e-06, "loss": 0.144, "step": 27642 }, { "epoch": 171.69781931464175, "grad_norm": 2.234375, "learning_rate": 2.4800559980975384e-06, "loss": 0.1264, "step": 27643 }, { "epoch": 171.70404984423675, "grad_norm": 2.5, "learning_rate": 2.4789862460797704e-06, "loss": 0.1245, "step": 27644 }, { "epoch": 171.71028037383178, "grad_norm": 1.0, "learning_rate": 2.4779167127905817e-06, "loss": 0.0938, "step": 27645 }, { "epoch": 171.71651090342678, "grad_norm": 1.890625, "learning_rate": 2.47684739824037e-06, "loss": 0.15, "step": 27646 }, { "epoch": 171.7227414330218, "grad_norm": 2.0, "learning_rate": 2.475778302439524e-06, "loss": 0.1639, "step": 27647 }, { "epoch": 171.7289719626168, "grad_norm": 2.796875, "learning_rate": 2.4747094253984193e-06, "loss": 0.2847, "step": 27648 }, { "epoch": 171.73520249221184, "grad_norm": 2.140625, "learning_rate": 2.4736407671274358e-06, "loss": 0.12, "step": 27649 }, { "epoch": 171.74143302180684, "grad_norm": 2.5625, "learning_rate": 2.4725723276369573e-06, "loss": 0.2142, "step": 27650 }, { "epoch": 171.74766355140187, "grad_norm": 2.078125, "learning_rate": 2.471504106937364e-06, "loss": 0.1982, "step": 27651 }, { "epoch": 171.75389408099687, "grad_norm": 3.234375, "learning_rate": 2.47043610503902e-06, "loss": 0.342, "step": 27652 }, { "epoch": 171.7601246105919, "grad_norm": 1.09375, "learning_rate": 2.4693683219523088e-06, "loss": 0.1044, "step": 27653 }, { "epoch": 171.7663551401869, "grad_norm": 1.9375, "learning_rate": 2.4683007576875915e-06, "loss": 0.1368, "step": 27654 }, { "epoch": 171.77258566978193, "grad_norm": 2.234375, "learning_rate": 2.467233412255246e-06, "loss": 0.2201, "step": 27655 }, { "epoch": 171.77881619937693, "grad_norm": 2.84375, "learning_rate": 2.4661662856656304e-06, "loss": 0.1281, "step": 27656 }, { "epoch": 171.78504672897196, "grad_norm": 1.46875, "learning_rate": 2.4650993779291114e-06, "loss": 0.0885, "step": 27657 }, { "epoch": 171.79127725856696, "grad_norm": 2.1875, "learning_rate": 2.464032689056053e-06, "loss": 0.198, "step": 27658 }, { "epoch": 171.797507788162, "grad_norm": 1.3671875, "learning_rate": 2.462966219056817e-06, "loss": 0.1315, "step": 27659 }, { "epoch": 171.80373831775702, "grad_norm": 2.171875, "learning_rate": 2.461899967941758e-06, "loss": 0.204, "step": 27660 }, { "epoch": 171.80996884735202, "grad_norm": 1.9375, "learning_rate": 2.4608339357212297e-06, "loss": 0.1813, "step": 27661 }, { "epoch": 171.81619937694705, "grad_norm": 2.015625, "learning_rate": 2.459768122405587e-06, "loss": 0.1267, "step": 27662 }, { "epoch": 171.82242990654206, "grad_norm": 2.453125, "learning_rate": 2.4587025280051857e-06, "loss": 0.162, "step": 27663 }, { "epoch": 171.82866043613708, "grad_norm": 1.1640625, "learning_rate": 2.4576371525303647e-06, "loss": 0.1177, "step": 27664 }, { "epoch": 171.8348909657321, "grad_norm": 2.265625, "learning_rate": 2.45657199599148e-06, "loss": 0.1704, "step": 27665 }, { "epoch": 171.84112149532712, "grad_norm": 1.6015625, "learning_rate": 2.455507058398879e-06, "loss": 0.1579, "step": 27666 }, { "epoch": 171.84735202492212, "grad_norm": 2.640625, "learning_rate": 2.4544423397628994e-06, "loss": 0.1961, "step": 27667 }, { "epoch": 171.85358255451715, "grad_norm": 1.0859375, "learning_rate": 2.453377840093879e-06, "loss": 0.1097, "step": 27668 }, { "epoch": 171.85981308411215, "grad_norm": 3.15625, "learning_rate": 2.4523135594021585e-06, "loss": 0.0964, "step": 27669 }, { "epoch": 171.86604361370718, "grad_norm": 1.453125, "learning_rate": 2.451249497698083e-06, "loss": 0.1632, "step": 27670 }, { "epoch": 171.87227414330218, "grad_norm": 3.875, "learning_rate": 2.450185654991974e-06, "loss": 0.2788, "step": 27671 }, { "epoch": 171.8785046728972, "grad_norm": 2.671875, "learning_rate": 2.449122031294168e-06, "loss": 0.2202, "step": 27672 }, { "epoch": 171.8847352024922, "grad_norm": 2.1875, "learning_rate": 2.4480586266150018e-06, "loss": 0.1402, "step": 27673 }, { "epoch": 171.89096573208724, "grad_norm": 2.796875, "learning_rate": 2.4469954409647998e-06, "loss": 0.2534, "step": 27674 }, { "epoch": 171.89719626168224, "grad_norm": 2.671875, "learning_rate": 2.4459324743538794e-06, "loss": 0.2934, "step": 27675 }, { "epoch": 171.90342679127727, "grad_norm": 1.6953125, "learning_rate": 2.444869726792573e-06, "loss": 0.137, "step": 27676 }, { "epoch": 171.90965732087227, "grad_norm": 3.21875, "learning_rate": 2.4438071982912064e-06, "loss": 0.1389, "step": 27677 }, { "epoch": 171.9158878504673, "grad_norm": 1.859375, "learning_rate": 2.442744888860085e-06, "loss": 0.1377, "step": 27678 }, { "epoch": 171.9221183800623, "grad_norm": 2.390625, "learning_rate": 2.4416827985095423e-06, "loss": 0.2356, "step": 27679 }, { "epoch": 171.92834890965733, "grad_norm": 1.2734375, "learning_rate": 2.4406209272498787e-06, "loss": 0.1233, "step": 27680 }, { "epoch": 171.93457943925233, "grad_norm": 1.2578125, "learning_rate": 2.4395592750914187e-06, "loss": 0.0953, "step": 27681 }, { "epoch": 171.94080996884736, "grad_norm": 2.421875, "learning_rate": 2.4384978420444655e-06, "loss": 0.1946, "step": 27682 }, { "epoch": 171.94704049844236, "grad_norm": 1.1328125, "learning_rate": 2.43743662811933e-06, "loss": 0.0903, "step": 27683 }, { "epoch": 171.9532710280374, "grad_norm": 1.21875, "learning_rate": 2.4363756333263267e-06, "loss": 0.1139, "step": 27684 }, { "epoch": 171.9595015576324, "grad_norm": 2.0625, "learning_rate": 2.435314857675752e-06, "loss": 0.1997, "step": 27685 }, { "epoch": 171.96573208722742, "grad_norm": 2.859375, "learning_rate": 2.4342543011779073e-06, "loss": 0.3598, "step": 27686 }, { "epoch": 171.97196261682242, "grad_norm": 3.71875, "learning_rate": 2.4331939638430918e-06, "loss": 0.3286, "step": 27687 }, { "epoch": 171.97819314641745, "grad_norm": 1.0859375, "learning_rate": 2.4321338456816166e-06, "loss": 0.105, "step": 27688 }, { "epoch": 171.98442367601245, "grad_norm": 1.671875, "learning_rate": 2.4310739467037603e-06, "loss": 0.1208, "step": 27689 }, { "epoch": 171.99065420560748, "grad_norm": 2.046875, "learning_rate": 2.4300142669198277e-06, "loss": 0.1654, "step": 27690 }, { "epoch": 171.99688473520249, "grad_norm": 1.2109375, "learning_rate": 2.4289548063401084e-06, "loss": 0.1163, "step": 27691 }, { "epoch": 172.0, "grad_norm": 1.734375, "learning_rate": 2.4278955649749e-06, "loss": 0.1777, "step": 27692 }, { "epoch": 172.00623052959503, "grad_norm": 1.0390625, "learning_rate": 2.4268365428344736e-06, "loss": 0.1011, "step": 27693 }, { "epoch": 172.01246105919003, "grad_norm": 2.6875, "learning_rate": 2.4257777399291223e-06, "loss": 0.2486, "step": 27694 }, { "epoch": 172.01869158878506, "grad_norm": 1.0390625, "learning_rate": 2.4247191562691285e-06, "loss": 0.1174, "step": 27695 }, { "epoch": 172.02492211838006, "grad_norm": 2.921875, "learning_rate": 2.423660791864782e-06, "loss": 0.261, "step": 27696 }, { "epoch": 172.0311526479751, "grad_norm": 1.7109375, "learning_rate": 2.422602646726349e-06, "loss": 0.1637, "step": 27697 }, { "epoch": 172.0373831775701, "grad_norm": 2.359375, "learning_rate": 2.4215447208641164e-06, "loss": 0.0995, "step": 27698 }, { "epoch": 172.04361370716512, "grad_norm": 2.25, "learning_rate": 2.42048701428835e-06, "loss": 0.1601, "step": 27699 }, { "epoch": 172.04984423676012, "grad_norm": 2.53125, "learning_rate": 2.419429527009334e-06, "loss": 0.2438, "step": 27700 }, { "epoch": 172.05607476635515, "grad_norm": 2.5625, "learning_rate": 2.418372259037327e-06, "loss": 0.2652, "step": 27701 }, { "epoch": 172.06230529595015, "grad_norm": 1.9609375, "learning_rate": 2.4173152103826006e-06, "loss": 0.1663, "step": 27702 }, { "epoch": 172.06853582554518, "grad_norm": 1.6015625, "learning_rate": 2.416258381055431e-06, "loss": 0.0961, "step": 27703 }, { "epoch": 172.07476635514018, "grad_norm": 2.3125, "learning_rate": 2.4152017710660723e-06, "loss": 0.1948, "step": 27704 }, { "epoch": 172.0809968847352, "grad_norm": 2.71875, "learning_rate": 2.4141453804247842e-06, "loss": 0.2215, "step": 27705 }, { "epoch": 172.08722741433021, "grad_norm": 1.125, "learning_rate": 2.4130892091418305e-06, "loss": 0.1173, "step": 27706 }, { "epoch": 172.09345794392524, "grad_norm": 2.015625, "learning_rate": 2.412033257227475e-06, "loss": 0.1976, "step": 27707 }, { "epoch": 172.09968847352025, "grad_norm": 1.8046875, "learning_rate": 2.4109775246919625e-06, "loss": 0.1818, "step": 27708 }, { "epoch": 172.10591900311528, "grad_norm": 1.5625, "learning_rate": 2.4099220115455514e-06, "loss": 0.1378, "step": 27709 }, { "epoch": 172.11214953271028, "grad_norm": 1.1953125, "learning_rate": 2.4088667177985e-06, "loss": 0.1435, "step": 27710 }, { "epoch": 172.1183800623053, "grad_norm": 1.6640625, "learning_rate": 2.40781164346105e-06, "loss": 0.105, "step": 27711 }, { "epoch": 172.1246105919003, "grad_norm": 0.9609375, "learning_rate": 2.406756788543443e-06, "loss": 0.0981, "step": 27712 }, { "epoch": 172.13084112149534, "grad_norm": 1.5859375, "learning_rate": 2.40570215305593e-06, "loss": 0.1354, "step": 27713 }, { "epoch": 172.13707165109034, "grad_norm": 1.71875, "learning_rate": 2.404647737008761e-06, "loss": 0.1271, "step": 27714 }, { "epoch": 172.14330218068537, "grad_norm": 2.625, "learning_rate": 2.4035935404121663e-06, "loss": 0.1887, "step": 27715 }, { "epoch": 172.14953271028037, "grad_norm": 1.1796875, "learning_rate": 2.4025395632763846e-06, "loss": 0.1314, "step": 27716 }, { "epoch": 172.1557632398754, "grad_norm": 2.078125, "learning_rate": 2.401485805611664e-06, "loss": 0.1718, "step": 27717 }, { "epoch": 172.1619937694704, "grad_norm": 2.4375, "learning_rate": 2.400432267428229e-06, "loss": 0.2267, "step": 27718 }, { "epoch": 172.16822429906543, "grad_norm": 1.9375, "learning_rate": 2.3993789487363085e-06, "loss": 0.1942, "step": 27719 }, { "epoch": 172.17445482866043, "grad_norm": 1.234375, "learning_rate": 2.3983258495461377e-06, "loss": 0.1209, "step": 27720 }, { "epoch": 172.18068535825546, "grad_norm": 3.21875, "learning_rate": 2.397272969867942e-06, "loss": 0.2954, "step": 27721 }, { "epoch": 172.18691588785046, "grad_norm": 2.3125, "learning_rate": 2.3962203097119578e-06, "loss": 0.2096, "step": 27722 }, { "epoch": 172.1931464174455, "grad_norm": 2.828125, "learning_rate": 2.3951678690883995e-06, "loss": 0.2322, "step": 27723 }, { "epoch": 172.1993769470405, "grad_norm": 3.59375, "learning_rate": 2.394115648007483e-06, "loss": 0.2699, "step": 27724 }, { "epoch": 172.20560747663552, "grad_norm": 2.4375, "learning_rate": 2.39306364647944e-06, "loss": 0.1351, "step": 27725 }, { "epoch": 172.21183800623052, "grad_norm": 1.8203125, "learning_rate": 2.3920118645144764e-06, "loss": 0.1538, "step": 27726 }, { "epoch": 172.21806853582555, "grad_norm": 2.65625, "learning_rate": 2.3909603021228133e-06, "loss": 0.3259, "step": 27727 }, { "epoch": 172.22429906542055, "grad_norm": 1.640625, "learning_rate": 2.3899089593146628e-06, "loss": 0.1372, "step": 27728 }, { "epoch": 172.23052959501558, "grad_norm": 1.9296875, "learning_rate": 2.3888578361002417e-06, "loss": 0.1592, "step": 27729 }, { "epoch": 172.23676012461058, "grad_norm": 2.328125, "learning_rate": 2.3878069324897527e-06, "loss": 0.2198, "step": 27730 }, { "epoch": 172.2429906542056, "grad_norm": 1.34375, "learning_rate": 2.386756248493399e-06, "loss": 0.0937, "step": 27731 }, { "epoch": 172.2492211838006, "grad_norm": 1.6328125, "learning_rate": 2.3857057841213892e-06, "loss": 0.0956, "step": 27732 }, { "epoch": 172.25545171339564, "grad_norm": 2.40625, "learning_rate": 2.3846555393839287e-06, "loss": 0.1721, "step": 27733 }, { "epoch": 172.26168224299064, "grad_norm": 2.0625, "learning_rate": 2.3836055142912093e-06, "loss": 0.1401, "step": 27734 }, { "epoch": 172.26791277258567, "grad_norm": 2.234375, "learning_rate": 2.3825557088534345e-06, "loss": 0.1342, "step": 27735 }, { "epoch": 172.27414330218068, "grad_norm": 2.984375, "learning_rate": 2.381506123080807e-06, "loss": 0.1716, "step": 27736 }, { "epoch": 172.2803738317757, "grad_norm": 0.78125, "learning_rate": 2.3804567569835096e-06, "loss": 0.0873, "step": 27737 }, { "epoch": 172.2866043613707, "grad_norm": 1.21875, "learning_rate": 2.379407610571735e-06, "loss": 0.119, "step": 27738 }, { "epoch": 172.29283489096574, "grad_norm": 1.65625, "learning_rate": 2.378358683855675e-06, "loss": 0.1399, "step": 27739 }, { "epoch": 172.29906542056074, "grad_norm": 2.484375, "learning_rate": 2.377309976845524e-06, "loss": 0.2327, "step": 27740 }, { "epoch": 172.30529595015577, "grad_norm": 1.5, "learning_rate": 2.376261489551454e-06, "loss": 0.1015, "step": 27741 }, { "epoch": 172.31152647975077, "grad_norm": 2.140625, "learning_rate": 2.375213221983655e-06, "loss": 0.0947, "step": 27742 }, { "epoch": 172.3177570093458, "grad_norm": 1.6796875, "learning_rate": 2.3741651741523124e-06, "loss": 0.1296, "step": 27743 }, { "epoch": 172.3239875389408, "grad_norm": 2.109375, "learning_rate": 2.3731173460675994e-06, "loss": 0.153, "step": 27744 }, { "epoch": 172.33021806853583, "grad_norm": 3.515625, "learning_rate": 2.3720697377396905e-06, "loss": 0.2911, "step": 27745 }, { "epoch": 172.33644859813083, "grad_norm": 2.875, "learning_rate": 2.371022349178764e-06, "loss": 0.2353, "step": 27746 }, { "epoch": 172.34267912772586, "grad_norm": 2.625, "learning_rate": 2.3699751803949955e-06, "loss": 0.1775, "step": 27747 }, { "epoch": 172.34890965732086, "grad_norm": 1.21875, "learning_rate": 2.3689282313985454e-06, "loss": 0.1201, "step": 27748 }, { "epoch": 172.3551401869159, "grad_norm": 2.984375, "learning_rate": 2.3678815021995955e-06, "loss": 0.301, "step": 27749 }, { "epoch": 172.3613707165109, "grad_norm": 2.875, "learning_rate": 2.3668349928083005e-06, "loss": 0.1596, "step": 27750 }, { "epoch": 172.36760124610592, "grad_norm": 2.9375, "learning_rate": 2.3657887032348306e-06, "loss": 0.2223, "step": 27751 }, { "epoch": 172.37383177570092, "grad_norm": 1.7265625, "learning_rate": 2.364742633489342e-06, "loss": 0.1153, "step": 27752 }, { "epoch": 172.38006230529595, "grad_norm": 2.59375, "learning_rate": 2.363696783581998e-06, "loss": 0.2362, "step": 27753 }, { "epoch": 172.38629283489095, "grad_norm": 2.25, "learning_rate": 2.3626511535229543e-06, "loss": 0.2324, "step": 27754 }, { "epoch": 172.39252336448598, "grad_norm": 1.96875, "learning_rate": 2.3616057433223788e-06, "loss": 0.1781, "step": 27755 }, { "epoch": 172.39875389408098, "grad_norm": 0.90625, "learning_rate": 2.3605605529904013e-06, "loss": 0.0827, "step": 27756 }, { "epoch": 172.404984423676, "grad_norm": 1.1953125, "learning_rate": 2.359515582537189e-06, "loss": 0.1184, "step": 27757 }, { "epoch": 172.41121495327104, "grad_norm": 1.859375, "learning_rate": 2.358470831972884e-06, "loss": 0.1788, "step": 27758 }, { "epoch": 172.41744548286604, "grad_norm": 2.421875, "learning_rate": 2.357426301307644e-06, "loss": 0.2066, "step": 27759 }, { "epoch": 172.42367601246107, "grad_norm": 1.7421875, "learning_rate": 2.356381990551598e-06, "loss": 0.1714, "step": 27760 }, { "epoch": 172.42990654205607, "grad_norm": 1.8671875, "learning_rate": 2.355337899714899e-06, "loss": 0.181, "step": 27761 }, { "epoch": 172.4361370716511, "grad_norm": 2.0, "learning_rate": 2.3542940288076883e-06, "loss": 0.1659, "step": 27762 }, { "epoch": 172.4423676012461, "grad_norm": 1.8359375, "learning_rate": 2.353250377840102e-06, "loss": 0.1478, "step": 27763 }, { "epoch": 172.44859813084113, "grad_norm": 1.8984375, "learning_rate": 2.352206946822269e-06, "loss": 0.1405, "step": 27764 }, { "epoch": 172.45482866043614, "grad_norm": 2.1875, "learning_rate": 2.351163735764328e-06, "loss": 0.1248, "step": 27765 }, { "epoch": 172.46105919003116, "grad_norm": 2.140625, "learning_rate": 2.3501207446764206e-06, "loss": 0.2336, "step": 27766 }, { "epoch": 172.46728971962617, "grad_norm": 1.8046875, "learning_rate": 2.349077973568661e-06, "loss": 0.1369, "step": 27767 }, { "epoch": 172.4735202492212, "grad_norm": 2.6875, "learning_rate": 2.3480354224511892e-06, "loss": 0.1179, "step": 27768 }, { "epoch": 172.4797507788162, "grad_norm": 1.2109375, "learning_rate": 2.346993091334121e-06, "loss": 0.0992, "step": 27769 }, { "epoch": 172.48598130841123, "grad_norm": 1.578125, "learning_rate": 2.3459509802275877e-06, "loss": 0.1319, "step": 27770 }, { "epoch": 172.49221183800623, "grad_norm": 1.6484375, "learning_rate": 2.344909089141703e-06, "loss": 0.2111, "step": 27771 }, { "epoch": 172.49844236760126, "grad_norm": 2.78125, "learning_rate": 2.34386741808659e-06, "loss": 0.224, "step": 27772 }, { "epoch": 172.50467289719626, "grad_norm": 2.96875, "learning_rate": 2.3428259670723706e-06, "loss": 0.1808, "step": 27773 }, { "epoch": 172.5109034267913, "grad_norm": 2.046875, "learning_rate": 2.341784736109154e-06, "loss": 0.1325, "step": 27774 }, { "epoch": 172.5171339563863, "grad_norm": 2.65625, "learning_rate": 2.340743725207048e-06, "loss": 0.1999, "step": 27775 }, { "epoch": 172.52336448598132, "grad_norm": 3.0625, "learning_rate": 2.33970293437617e-06, "loss": 0.1408, "step": 27776 }, { "epoch": 172.52959501557632, "grad_norm": 2.375, "learning_rate": 2.3386623636266287e-06, "loss": 0.1343, "step": 27777 }, { "epoch": 172.53582554517135, "grad_norm": 2.265625, "learning_rate": 2.3376220129685268e-06, "loss": 0.2426, "step": 27778 }, { "epoch": 172.54205607476635, "grad_norm": 1.609375, "learning_rate": 2.336581882411967e-06, "loss": 0.1453, "step": 27779 }, { "epoch": 172.54828660436138, "grad_norm": 2.0, "learning_rate": 2.335541971967062e-06, "loss": 0.1327, "step": 27780 }, { "epoch": 172.55451713395638, "grad_norm": 1.3828125, "learning_rate": 2.334502281643902e-06, "loss": 0.1313, "step": 27781 }, { "epoch": 172.5607476635514, "grad_norm": 1.75, "learning_rate": 2.333462811452583e-06, "loss": 0.1424, "step": 27782 }, { "epoch": 172.5669781931464, "grad_norm": 1.8046875, "learning_rate": 2.3324235614032048e-06, "loss": 0.173, "step": 27783 }, { "epoch": 172.57320872274144, "grad_norm": 1.6796875, "learning_rate": 2.3313845315058618e-06, "loss": 0.1403, "step": 27784 }, { "epoch": 172.57943925233644, "grad_norm": 1.4609375, "learning_rate": 2.330345721770641e-06, "loss": 0.1163, "step": 27785 }, { "epoch": 172.58566978193147, "grad_norm": 1.84375, "learning_rate": 2.3293071322076365e-06, "loss": 0.2057, "step": 27786 }, { "epoch": 172.59190031152647, "grad_norm": 2.375, "learning_rate": 2.328268762826935e-06, "loss": 0.2747, "step": 27787 }, { "epoch": 172.5981308411215, "grad_norm": 1.390625, "learning_rate": 2.3272306136386174e-06, "loss": 0.1054, "step": 27788 }, { "epoch": 172.6043613707165, "grad_norm": 1.8828125, "learning_rate": 2.326192684652767e-06, "loss": 0.1415, "step": 27789 }, { "epoch": 172.61059190031153, "grad_norm": 3.140625, "learning_rate": 2.325154975879462e-06, "loss": 0.2825, "step": 27790 }, { "epoch": 172.61682242990653, "grad_norm": 2.109375, "learning_rate": 2.324117487328789e-06, "loss": 0.1614, "step": 27791 }, { "epoch": 172.62305295950156, "grad_norm": 1.25, "learning_rate": 2.32308021901082e-06, "loss": 0.1081, "step": 27792 }, { "epoch": 172.62928348909657, "grad_norm": 1.6328125, "learning_rate": 2.322043170935631e-06, "loss": 0.2234, "step": 27793 }, { "epoch": 172.6355140186916, "grad_norm": 1.6796875, "learning_rate": 2.3210063431132882e-06, "loss": 0.132, "step": 27794 }, { "epoch": 172.6417445482866, "grad_norm": 2.546875, "learning_rate": 2.3199697355538645e-06, "loss": 0.136, "step": 27795 }, { "epoch": 172.64797507788163, "grad_norm": 2.1875, "learning_rate": 2.3189333482674326e-06, "loss": 0.2234, "step": 27796 }, { "epoch": 172.65420560747663, "grad_norm": 1.484375, "learning_rate": 2.3178971812640504e-06, "loss": 0.1462, "step": 27797 }, { "epoch": 172.66043613707166, "grad_norm": 0.98046875, "learning_rate": 2.3168612345537825e-06, "loss": 0.0955, "step": 27798 }, { "epoch": 172.66666666666666, "grad_norm": 2.53125, "learning_rate": 2.315825508146699e-06, "loss": 0.1721, "step": 27799 }, { "epoch": 172.6728971962617, "grad_norm": 2.1875, "learning_rate": 2.3147900020528525e-06, "loss": 0.1485, "step": 27800 }, { "epoch": 172.6791277258567, "grad_norm": 1.6015625, "learning_rate": 2.313754716282296e-06, "loss": 0.0915, "step": 27801 }, { "epoch": 172.68535825545172, "grad_norm": 1.1640625, "learning_rate": 2.3127196508450884e-06, "loss": 0.1182, "step": 27802 }, { "epoch": 172.69158878504672, "grad_norm": 2.53125, "learning_rate": 2.311684805751288e-06, "loss": 0.1676, "step": 27803 }, { "epoch": 172.69781931464175, "grad_norm": 3.765625, "learning_rate": 2.3106501810109316e-06, "loss": 0.3563, "step": 27804 }, { "epoch": 172.70404984423675, "grad_norm": 2.71875, "learning_rate": 2.3096157766340803e-06, "loss": 0.2172, "step": 27805 }, { "epoch": 172.71028037383178, "grad_norm": 2.703125, "learning_rate": 2.308581592630779e-06, "loss": 0.1807, "step": 27806 }, { "epoch": 172.71651090342678, "grad_norm": 0.89453125, "learning_rate": 2.3075476290110697e-06, "loss": 0.1013, "step": 27807 }, { "epoch": 172.7227414330218, "grad_norm": 2.90625, "learning_rate": 2.3065138857849888e-06, "loss": 0.1722, "step": 27808 }, { "epoch": 172.7289719626168, "grad_norm": 1.1015625, "learning_rate": 2.305480362962584e-06, "loss": 0.1086, "step": 27809 }, { "epoch": 172.73520249221184, "grad_norm": 1.234375, "learning_rate": 2.304447060553891e-06, "loss": 0.1205, "step": 27810 }, { "epoch": 172.74143302180684, "grad_norm": 2.0, "learning_rate": 2.303413978568944e-06, "loss": 0.1605, "step": 27811 }, { "epoch": 172.74766355140187, "grad_norm": 2.859375, "learning_rate": 2.302381117017774e-06, "loss": 0.2558, "step": 27812 }, { "epoch": 172.75389408099687, "grad_norm": 2.234375, "learning_rate": 2.3013484759104224e-06, "loss": 0.1962, "step": 27813 }, { "epoch": 172.7601246105919, "grad_norm": 1.3671875, "learning_rate": 2.3003160552569124e-06, "loss": 0.109, "step": 27814 }, { "epoch": 172.7663551401869, "grad_norm": 1.484375, "learning_rate": 2.2992838550672635e-06, "loss": 0.0992, "step": 27815 }, { "epoch": 172.77258566978193, "grad_norm": 1.1484375, "learning_rate": 2.298251875351509e-06, "loss": 0.0926, "step": 27816 }, { "epoch": 172.77881619937693, "grad_norm": 2.703125, "learning_rate": 2.297220116119669e-06, "loss": 0.2249, "step": 27817 }, { "epoch": 172.78504672897196, "grad_norm": 1.9296875, "learning_rate": 2.2961885773817723e-06, "loss": 0.1219, "step": 27818 }, { "epoch": 172.79127725856696, "grad_norm": 3.0625, "learning_rate": 2.2951572591478286e-06, "loss": 0.2516, "step": 27819 }, { "epoch": 172.797507788162, "grad_norm": 2.1875, "learning_rate": 2.294126161427854e-06, "loss": 0.0924, "step": 27820 }, { "epoch": 172.80373831775702, "grad_norm": 2.015625, "learning_rate": 2.2930952842318663e-06, "loss": 0.1483, "step": 27821 }, { "epoch": 172.80996884735202, "grad_norm": 1.1796875, "learning_rate": 2.2920646275698733e-06, "loss": 0.1037, "step": 27822 }, { "epoch": 172.81619937694705, "grad_norm": 1.5078125, "learning_rate": 2.291034191451888e-06, "loss": 0.1365, "step": 27823 }, { "epoch": 172.82242990654206, "grad_norm": 1.4609375, "learning_rate": 2.2900039758879195e-06, "loss": 0.1275, "step": 27824 }, { "epoch": 172.82866043613708, "grad_norm": 1.21875, "learning_rate": 2.288973980887979e-06, "loss": 0.1363, "step": 27825 }, { "epoch": 172.8348909657321, "grad_norm": 2.390625, "learning_rate": 2.2879442064620533e-06, "loss": 0.2046, "step": 27826 }, { "epoch": 172.84112149532712, "grad_norm": 2.109375, "learning_rate": 2.2869146526201563e-06, "loss": 0.1162, "step": 27827 }, { "epoch": 172.84735202492212, "grad_norm": 1.09375, "learning_rate": 2.285885319372283e-06, "loss": 0.1193, "step": 27828 }, { "epoch": 172.85358255451715, "grad_norm": 1.7734375, "learning_rate": 2.284856206728436e-06, "loss": 0.2387, "step": 27829 }, { "epoch": 172.85981308411215, "grad_norm": 1.7890625, "learning_rate": 2.283827314698603e-06, "loss": 0.1434, "step": 27830 }, { "epoch": 172.86604361370718, "grad_norm": 2.421875, "learning_rate": 2.282798643292777e-06, "loss": 0.1645, "step": 27831 }, { "epoch": 172.87227414330218, "grad_norm": 1.3671875, "learning_rate": 2.28177019252096e-06, "loss": 0.1003, "step": 27832 }, { "epoch": 172.8785046728972, "grad_norm": 1.03125, "learning_rate": 2.280741962393129e-06, "loss": 0.1124, "step": 27833 }, { "epoch": 172.8847352024922, "grad_norm": 2.03125, "learning_rate": 2.279713952919271e-06, "loss": 0.2298, "step": 27834 }, { "epoch": 172.89096573208724, "grad_norm": 1.90625, "learning_rate": 2.278686164109373e-06, "loss": 0.1116, "step": 27835 }, { "epoch": 172.89719626168224, "grad_norm": 2.015625, "learning_rate": 2.2776585959734196e-06, "loss": 0.167, "step": 27836 }, { "epoch": 172.90342679127727, "grad_norm": 1.1640625, "learning_rate": 2.2766312485213854e-06, "loss": 0.1102, "step": 27837 }, { "epoch": 172.90965732087227, "grad_norm": 1.7265625, "learning_rate": 2.275604121763253e-06, "loss": 0.1035, "step": 27838 }, { "epoch": 172.9158878504673, "grad_norm": 1.96875, "learning_rate": 2.2745772157089922e-06, "loss": 0.1262, "step": 27839 }, { "epoch": 172.9221183800623, "grad_norm": 2.515625, "learning_rate": 2.2735505303685873e-06, "loss": 0.1766, "step": 27840 }, { "epoch": 172.92834890965733, "grad_norm": 2.125, "learning_rate": 2.2725240657519935e-06, "loss": 0.2059, "step": 27841 }, { "epoch": 172.93457943925233, "grad_norm": 2.140625, "learning_rate": 2.2714978218691924e-06, "loss": 0.1968, "step": 27842 }, { "epoch": 172.94080996884736, "grad_norm": 1.734375, "learning_rate": 2.2704717987301505e-06, "loss": 0.132, "step": 27843 }, { "epoch": 172.94704049844236, "grad_norm": 1.265625, "learning_rate": 2.2694459963448324e-06, "loss": 0.1384, "step": 27844 }, { "epoch": 172.9532710280374, "grad_norm": 1.3046875, "learning_rate": 2.2684204147231937e-06, "loss": 0.1079, "step": 27845 }, { "epoch": 172.9595015576324, "grad_norm": 1.953125, "learning_rate": 2.2673950538751987e-06, "loss": 0.1421, "step": 27846 }, { "epoch": 172.96573208722742, "grad_norm": 1.9921875, "learning_rate": 2.2663699138108117e-06, "loss": 0.1822, "step": 27847 }, { "epoch": 172.97196261682242, "grad_norm": 2.828125, "learning_rate": 2.265344994539978e-06, "loss": 0.2497, "step": 27848 }, { "epoch": 172.97819314641745, "grad_norm": 1.71875, "learning_rate": 2.264320296072661e-06, "loss": 0.1121, "step": 27849 }, { "epoch": 172.98442367601245, "grad_norm": 2.046875, "learning_rate": 2.263295818418809e-06, "loss": 0.1484, "step": 27850 }, { "epoch": 172.99065420560748, "grad_norm": 1.2890625, "learning_rate": 2.2622715615883834e-06, "loss": 0.1333, "step": 27851 }, { "epoch": 172.99688473520249, "grad_norm": 2.109375, "learning_rate": 2.261247525591312e-06, "loss": 0.14, "step": 27852 }, { "epoch": 173.0, "grad_norm": 1.140625, "learning_rate": 2.260223710437548e-06, "loss": 0.0537, "step": 27853 }, { "epoch": 173.00623052959503, "grad_norm": 2.671875, "learning_rate": 2.2592001161370392e-06, "loss": 0.2596, "step": 27854 }, { "epoch": 173.01246105919003, "grad_norm": 1.53125, "learning_rate": 2.2581767426997275e-06, "loss": 0.1527, "step": 27855 }, { "epoch": 173.01869158878506, "grad_norm": 2.71875, "learning_rate": 2.2571535901355466e-06, "loss": 0.2587, "step": 27856 }, { "epoch": 173.02492211838006, "grad_norm": 1.1484375, "learning_rate": 2.256130658454439e-06, "loss": 0.0968, "step": 27857 }, { "epoch": 173.0311526479751, "grad_norm": 1.875, "learning_rate": 2.255107947666335e-06, "loss": 0.183, "step": 27858 }, { "epoch": 173.0373831775701, "grad_norm": 1.8046875, "learning_rate": 2.2540854577811717e-06, "loss": 0.1594, "step": 27859 }, { "epoch": 173.04361370716512, "grad_norm": 0.7890625, "learning_rate": 2.2530631888088736e-06, "loss": 0.0988, "step": 27860 }, { "epoch": 173.04984423676012, "grad_norm": 1.0390625, "learning_rate": 2.25204114075937e-06, "loss": 0.0978, "step": 27861 }, { "epoch": 173.05607476635515, "grad_norm": 3.15625, "learning_rate": 2.2510193136425967e-06, "loss": 0.3938, "step": 27862 }, { "epoch": 173.06230529595015, "grad_norm": 2.984375, "learning_rate": 2.2499977074684735e-06, "loss": 0.2304, "step": 27863 }, { "epoch": 173.06853582554518, "grad_norm": 2.015625, "learning_rate": 2.2489763222469125e-06, "loss": 0.1045, "step": 27864 }, { "epoch": 173.07476635514018, "grad_norm": 3.1875, "learning_rate": 2.247955157987841e-06, "loss": 0.2936, "step": 27865 }, { "epoch": 173.0809968847352, "grad_norm": 1.515625, "learning_rate": 2.246934214701185e-06, "loss": 0.1495, "step": 27866 }, { "epoch": 173.08722741433021, "grad_norm": 1.15625, "learning_rate": 2.245913492396845e-06, "loss": 0.0989, "step": 27867 }, { "epoch": 173.09345794392524, "grad_norm": 2.578125, "learning_rate": 2.244892991084743e-06, "loss": 0.1629, "step": 27868 }, { "epoch": 173.09968847352025, "grad_norm": 1.6640625, "learning_rate": 2.2438727107747913e-06, "loss": 0.1556, "step": 27869 }, { "epoch": 173.10591900311528, "grad_norm": 1.5234375, "learning_rate": 2.2428526514768982e-06, "loss": 0.1664, "step": 27870 }, { "epoch": 173.11214953271028, "grad_norm": 1.1640625, "learning_rate": 2.2418328132009643e-06, "loss": 0.1063, "step": 27871 }, { "epoch": 173.1183800623053, "grad_norm": 1.9921875, "learning_rate": 2.2408131959569006e-06, "loss": 0.2102, "step": 27872 }, { "epoch": 173.1246105919003, "grad_norm": 2.359375, "learning_rate": 2.239793799754611e-06, "loss": 0.2465, "step": 27873 }, { "epoch": 173.13084112149534, "grad_norm": 1.3828125, "learning_rate": 2.23877462460399e-06, "loss": 0.1107, "step": 27874 }, { "epoch": 173.13707165109034, "grad_norm": 1.390625, "learning_rate": 2.2377556705149405e-06, "loss": 0.1117, "step": 27875 }, { "epoch": 173.14330218068537, "grad_norm": 2.078125, "learning_rate": 2.2367369374973606e-06, "loss": 0.2144, "step": 27876 }, { "epoch": 173.14953271028037, "grad_norm": 2.671875, "learning_rate": 2.235718425561145e-06, "loss": 0.2312, "step": 27877 }, { "epoch": 173.1557632398754, "grad_norm": 2.640625, "learning_rate": 2.234700134716175e-06, "loss": 0.167, "step": 27878 }, { "epoch": 173.1619937694704, "grad_norm": 2.359375, "learning_rate": 2.2336820649723507e-06, "loss": 0.1394, "step": 27879 }, { "epoch": 173.16822429906543, "grad_norm": 2.0625, "learning_rate": 2.232664216339558e-06, "loss": 0.1429, "step": 27880 }, { "epoch": 173.17445482866043, "grad_norm": 2.5, "learning_rate": 2.2316465888276795e-06, "loss": 0.193, "step": 27881 }, { "epoch": 173.18068535825546, "grad_norm": 1.3515625, "learning_rate": 2.2306291824466003e-06, "loss": 0.1367, "step": 27882 }, { "epoch": 173.18691588785046, "grad_norm": 1.125, "learning_rate": 2.2296119972062072e-06, "loss": 0.1014, "step": 27883 }, { "epoch": 173.1931464174455, "grad_norm": 2.0, "learning_rate": 2.2285950331163737e-06, "loss": 0.1697, "step": 27884 }, { "epoch": 173.1993769470405, "grad_norm": 2.71875, "learning_rate": 2.2275782901869714e-06, "loss": 0.3133, "step": 27885 }, { "epoch": 173.20560747663552, "grad_norm": 2.203125, "learning_rate": 2.226561768427882e-06, "loss": 0.1824, "step": 27886 }, { "epoch": 173.21183800623052, "grad_norm": 1.4765625, "learning_rate": 2.2255454678489773e-06, "loss": 0.2247, "step": 27887 }, { "epoch": 173.21806853582555, "grad_norm": 2.234375, "learning_rate": 2.224529388460131e-06, "loss": 0.1724, "step": 27888 }, { "epoch": 173.22429906542055, "grad_norm": 1.9453125, "learning_rate": 2.2235135302712095e-06, "loss": 0.1345, "step": 27889 }, { "epoch": 173.23052959501558, "grad_norm": 1.1015625, "learning_rate": 2.222497893292072e-06, "loss": 0.1025, "step": 27890 }, { "epoch": 173.23676012461058, "grad_norm": 1.0859375, "learning_rate": 2.221482477532588e-06, "loss": 0.0905, "step": 27891 }, { "epoch": 173.2429906542056, "grad_norm": 1.34375, "learning_rate": 2.220467283002625e-06, "loss": 0.1082, "step": 27892 }, { "epoch": 173.2492211838006, "grad_norm": 2.0625, "learning_rate": 2.2194523097120305e-06, "loss": 0.1504, "step": 27893 }, { "epoch": 173.25545171339564, "grad_norm": 2.203125, "learning_rate": 2.2184375576706718e-06, "loss": 0.195, "step": 27894 }, { "epoch": 173.26168224299064, "grad_norm": 1.578125, "learning_rate": 2.2174230268884046e-06, "loss": 0.1548, "step": 27895 }, { "epoch": 173.26791277258567, "grad_norm": 1.6328125, "learning_rate": 2.216408717375079e-06, "loss": 0.1237, "step": 27896 }, { "epoch": 173.27414330218068, "grad_norm": 1.8046875, "learning_rate": 2.215394629140541e-06, "loss": 0.1095, "step": 27897 }, { "epoch": 173.2803738317757, "grad_norm": 3.0, "learning_rate": 2.2143807621946454e-06, "loss": 0.27, "step": 27898 }, { "epoch": 173.2866043613707, "grad_norm": 1.78125, "learning_rate": 2.2133671165472434e-06, "loss": 0.153, "step": 27899 }, { "epoch": 173.29283489096574, "grad_norm": 1.2890625, "learning_rate": 2.212353692208172e-06, "loss": 0.108, "step": 27900 }, { "epoch": 173.29906542056074, "grad_norm": 2.140625, "learning_rate": 2.2113404891872745e-06, "loss": 0.1655, "step": 27901 }, { "epoch": 173.30529595015577, "grad_norm": 2.453125, "learning_rate": 2.2103275074943972e-06, "loss": 0.2076, "step": 27902 }, { "epoch": 173.31152647975077, "grad_norm": 2.15625, "learning_rate": 2.2093147471393767e-06, "loss": 0.1474, "step": 27903 }, { "epoch": 173.3177570093458, "grad_norm": 2.671875, "learning_rate": 2.208302208132043e-06, "loss": 0.2341, "step": 27904 }, { "epoch": 173.3239875389408, "grad_norm": 2.65625, "learning_rate": 2.2072898904822336e-06, "loss": 0.2077, "step": 27905 }, { "epoch": 173.33021806853583, "grad_norm": 2.25, "learning_rate": 2.2062777941997847e-06, "loss": 0.1069, "step": 27906 }, { "epoch": 173.33644859813083, "grad_norm": 2.5, "learning_rate": 2.205265919294519e-06, "loss": 0.2301, "step": 27907 }, { "epoch": 173.34267912772586, "grad_norm": 2.75, "learning_rate": 2.2042542657762704e-06, "loss": 0.1184, "step": 27908 }, { "epoch": 173.34890965732086, "grad_norm": 1.4296875, "learning_rate": 2.2032428336548587e-06, "loss": 0.1321, "step": 27909 }, { "epoch": 173.3551401869159, "grad_norm": 0.80859375, "learning_rate": 2.2022316229401126e-06, "loss": 0.091, "step": 27910 }, { "epoch": 173.3613707165109, "grad_norm": 2.671875, "learning_rate": 2.2012206336418455e-06, "loss": 0.3321, "step": 27911 }, { "epoch": 173.36760124610592, "grad_norm": 1.9140625, "learning_rate": 2.200209865769881e-06, "loss": 0.1428, "step": 27912 }, { "epoch": 173.37383177570092, "grad_norm": 2.328125, "learning_rate": 2.1991993193340416e-06, "loss": 0.1787, "step": 27913 }, { "epoch": 173.38006230529595, "grad_norm": 1.3828125, "learning_rate": 2.198188994344133e-06, "loss": 0.1198, "step": 27914 }, { "epoch": 173.38629283489095, "grad_norm": 2.59375, "learning_rate": 2.19717889080997e-06, "loss": 0.2147, "step": 27915 }, { "epoch": 173.39252336448598, "grad_norm": 1.9375, "learning_rate": 2.196169008741361e-06, "loss": 0.1851, "step": 27916 }, { "epoch": 173.39875389408098, "grad_norm": 1.390625, "learning_rate": 2.1951593481481237e-06, "loss": 0.1123, "step": 27917 }, { "epoch": 173.404984423676, "grad_norm": 1.921875, "learning_rate": 2.1941499090400524e-06, "loss": 0.1908, "step": 27918 }, { "epoch": 173.41121495327104, "grad_norm": 3.765625, "learning_rate": 2.1931406914269564e-06, "loss": 0.2232, "step": 27919 }, { "epoch": 173.41744548286604, "grad_norm": 1.1953125, "learning_rate": 2.192131695318636e-06, "loss": 0.1152, "step": 27920 }, { "epoch": 173.42367601246107, "grad_norm": 2.78125, "learning_rate": 2.1911229207249e-06, "loss": 0.1593, "step": 27921 }, { "epoch": 173.42990654205607, "grad_norm": 2.5, "learning_rate": 2.1901143676555295e-06, "loss": 0.1781, "step": 27922 }, { "epoch": 173.4361370716511, "grad_norm": 2.71875, "learning_rate": 2.189106036120328e-06, "loss": 0.327, "step": 27923 }, { "epoch": 173.4423676012461, "grad_norm": 2.734375, "learning_rate": 2.18809792612909e-06, "loss": 0.2508, "step": 27924 }, { "epoch": 173.44859813084113, "grad_norm": 2.0625, "learning_rate": 2.1870900376916083e-06, "loss": 0.2088, "step": 27925 }, { "epoch": 173.45482866043614, "grad_norm": 2.96875, "learning_rate": 2.1860823708176633e-06, "loss": 0.2685, "step": 27926 }, { "epoch": 173.46105919003116, "grad_norm": 1.265625, "learning_rate": 2.18507492551705e-06, "loss": 0.1151, "step": 27927 }, { "epoch": 173.46728971962617, "grad_norm": 2.46875, "learning_rate": 2.184067701799547e-06, "loss": 0.1759, "step": 27928 }, { "epoch": 173.4735202492212, "grad_norm": 2.125, "learning_rate": 2.1830606996749435e-06, "loss": 0.2081, "step": 27929 }, { "epoch": 173.4797507788162, "grad_norm": 2.421875, "learning_rate": 2.1820539191530098e-06, "loss": 0.2352, "step": 27930 }, { "epoch": 173.48598130841123, "grad_norm": 1.625, "learning_rate": 2.1810473602435324e-06, "loss": 0.1295, "step": 27931 }, { "epoch": 173.49221183800623, "grad_norm": 1.796875, "learning_rate": 2.180041022956286e-06, "loss": 0.1047, "step": 27932 }, { "epoch": 173.49844236760126, "grad_norm": 1.75, "learning_rate": 2.1790349073010447e-06, "loss": 0.1281, "step": 27933 }, { "epoch": 173.50467289719626, "grad_norm": 3.125, "learning_rate": 2.1780290132875718e-06, "loss": 0.2237, "step": 27934 }, { "epoch": 173.5109034267913, "grad_norm": 1.2109375, "learning_rate": 2.177023340925646e-06, "loss": 0.1246, "step": 27935 }, { "epoch": 173.5171339563863, "grad_norm": 1.9296875, "learning_rate": 2.176017890225035e-06, "loss": 0.2155, "step": 27936 }, { "epoch": 173.52336448598132, "grad_norm": 2.4375, "learning_rate": 2.175012661195494e-06, "loss": 0.2339, "step": 27937 }, { "epoch": 173.52959501557632, "grad_norm": 1.46875, "learning_rate": 2.174007653846796e-06, "loss": 0.1626, "step": 27938 }, { "epoch": 173.53582554517135, "grad_norm": 1.609375, "learning_rate": 2.1730028681887003e-06, "loss": 0.1558, "step": 27939 }, { "epoch": 173.54205607476635, "grad_norm": 1.015625, "learning_rate": 2.1719983042309655e-06, "loss": 0.113, "step": 27940 }, { "epoch": 173.54828660436138, "grad_norm": 1.1484375, "learning_rate": 2.1709939619833418e-06, "loss": 0.1183, "step": 27941 }, { "epoch": 173.55451713395638, "grad_norm": 1.4296875, "learning_rate": 2.1699898414555884e-06, "loss": 0.1475, "step": 27942 }, { "epoch": 173.5607476635514, "grad_norm": 1.0625, "learning_rate": 2.168985942657459e-06, "loss": 0.1139, "step": 27943 }, { "epoch": 173.5669781931464, "grad_norm": 1.125, "learning_rate": 2.1679822655987004e-06, "loss": 0.1183, "step": 27944 }, { "epoch": 173.57320872274144, "grad_norm": 3.015625, "learning_rate": 2.1669788102890608e-06, "loss": 0.1377, "step": 27945 }, { "epoch": 173.57943925233644, "grad_norm": 2.109375, "learning_rate": 2.165975576738294e-06, "loss": 0.1317, "step": 27946 }, { "epoch": 173.58566978193147, "grad_norm": 2.65625, "learning_rate": 2.164972564956133e-06, "loss": 0.2344, "step": 27947 }, { "epoch": 173.59190031152647, "grad_norm": 1.5546875, "learning_rate": 2.1639697749523206e-06, "loss": 0.1062, "step": 27948 }, { "epoch": 173.5981308411215, "grad_norm": 1.9609375, "learning_rate": 2.162967206736599e-06, "loss": 0.1638, "step": 27949 }, { "epoch": 173.6043613707165, "grad_norm": 1.5703125, "learning_rate": 2.161964860318705e-06, "loss": 0.1436, "step": 27950 }, { "epoch": 173.61059190031153, "grad_norm": 1.3046875, "learning_rate": 2.1609627357083772e-06, "loss": 0.1021, "step": 27951 }, { "epoch": 173.61682242990653, "grad_norm": 2.21875, "learning_rate": 2.1599608329153392e-06, "loss": 0.1687, "step": 27952 }, { "epoch": 173.62305295950156, "grad_norm": 1.2109375, "learning_rate": 2.158959151949333e-06, "loss": 0.1325, "step": 27953 }, { "epoch": 173.62928348909657, "grad_norm": 1.703125, "learning_rate": 2.1579576928200784e-06, "loss": 0.1103, "step": 27954 }, { "epoch": 173.6355140186916, "grad_norm": 1.65625, "learning_rate": 2.156956455537307e-06, "loss": 0.1593, "step": 27955 }, { "epoch": 173.6417445482866, "grad_norm": 2.578125, "learning_rate": 2.155955440110738e-06, "loss": 0.3182, "step": 27956 }, { "epoch": 173.64797507788163, "grad_norm": 2.0, "learning_rate": 2.1549546465500947e-06, "loss": 0.1779, "step": 27957 }, { "epoch": 173.65420560747663, "grad_norm": 2.078125, "learning_rate": 2.153954074865103e-06, "loss": 0.1734, "step": 27958 }, { "epoch": 173.66043613707166, "grad_norm": 2.453125, "learning_rate": 2.152953725065476e-06, "loss": 0.1179, "step": 27959 }, { "epoch": 173.66666666666666, "grad_norm": 2.03125, "learning_rate": 2.1519535971609274e-06, "loss": 0.1713, "step": 27960 }, { "epoch": 173.6728971962617, "grad_norm": 1.421875, "learning_rate": 2.15095369116117e-06, "loss": 0.1403, "step": 27961 }, { "epoch": 173.6791277258567, "grad_norm": 2.9375, "learning_rate": 2.1499540070759216e-06, "loss": 0.1782, "step": 27962 }, { "epoch": 173.68535825545172, "grad_norm": 1.546875, "learning_rate": 2.1489545449148857e-06, "loss": 0.1393, "step": 27963 }, { "epoch": 173.69158878504672, "grad_norm": 1.28125, "learning_rate": 2.147955304687771e-06, "loss": 0.1129, "step": 27964 }, { "epoch": 173.69781931464175, "grad_norm": 0.75390625, "learning_rate": 2.1469562864042837e-06, "loss": 0.0913, "step": 27965 }, { "epoch": 173.70404984423675, "grad_norm": 1.8828125, "learning_rate": 2.14595749007413e-06, "loss": 0.1471, "step": 27966 }, { "epoch": 173.71028037383178, "grad_norm": 3.5, "learning_rate": 2.144958915706996e-06, "loss": 0.287, "step": 27967 }, { "epoch": 173.71651090342678, "grad_norm": 2.21875, "learning_rate": 2.143960563312594e-06, "loss": 0.1785, "step": 27968 }, { "epoch": 173.7227414330218, "grad_norm": 2.78125, "learning_rate": 2.1429624329006164e-06, "loss": 0.3174, "step": 27969 }, { "epoch": 173.7289719626168, "grad_norm": 1.234375, "learning_rate": 2.141964524480755e-06, "loss": 0.1255, "step": 27970 }, { "epoch": 173.73520249221184, "grad_norm": 2.65625, "learning_rate": 2.1409668380627026e-06, "loss": 0.1969, "step": 27971 }, { "epoch": 173.74143302180684, "grad_norm": 2.125, "learning_rate": 2.1399693736561532e-06, "loss": 0.1978, "step": 27972 }, { "epoch": 173.74766355140187, "grad_norm": 0.9765625, "learning_rate": 2.1389721312707918e-06, "loss": 0.0886, "step": 27973 }, { "epoch": 173.75389408099687, "grad_norm": 1.6796875, "learning_rate": 2.137975110916299e-06, "loss": 0.1501, "step": 27974 }, { "epoch": 173.7601246105919, "grad_norm": 1.578125, "learning_rate": 2.1369783126023613e-06, "loss": 0.1291, "step": 27975 }, { "epoch": 173.7663551401869, "grad_norm": 1.96875, "learning_rate": 2.135981736338666e-06, "loss": 0.143, "step": 27976 }, { "epoch": 173.77258566978193, "grad_norm": 1.5546875, "learning_rate": 2.13498538213488e-06, "loss": 0.1814, "step": 27977 }, { "epoch": 173.77881619937693, "grad_norm": 2.59375, "learning_rate": 2.1339892500006924e-06, "loss": 0.24, "step": 27978 }, { "epoch": 173.78504672897196, "grad_norm": 2.671875, "learning_rate": 2.1329933399457684e-06, "loss": 0.3539, "step": 27979 }, { "epoch": 173.79127725856696, "grad_norm": 1.6875, "learning_rate": 2.1319976519797864e-06, "loss": 0.123, "step": 27980 }, { "epoch": 173.797507788162, "grad_norm": 2.5, "learning_rate": 2.1310021861124103e-06, "loss": 0.2254, "step": 27981 }, { "epoch": 173.80373831775702, "grad_norm": 1.6171875, "learning_rate": 2.130006942353313e-06, "loss": 0.1462, "step": 27982 }, { "epoch": 173.80996884735202, "grad_norm": 1.7265625, "learning_rate": 2.1290119207121594e-06, "loss": 0.1268, "step": 27983 }, { "epoch": 173.81619937694705, "grad_norm": 2.484375, "learning_rate": 2.128017121198625e-06, "loss": 0.2736, "step": 27984 }, { "epoch": 173.82242990654206, "grad_norm": 1.671875, "learning_rate": 2.127022543822349e-06, "loss": 0.1164, "step": 27985 }, { "epoch": 173.82866043613708, "grad_norm": 2.5625, "learning_rate": 2.1260281885930046e-06, "loss": 0.2007, "step": 27986 }, { "epoch": 173.8348909657321, "grad_norm": 2.65625, "learning_rate": 2.1250340555202446e-06, "loss": 0.1668, "step": 27987 }, { "epoch": 173.84112149532712, "grad_norm": 2.53125, "learning_rate": 2.1240401446137315e-06, "loss": 0.2625, "step": 27988 }, { "epoch": 173.84735202492212, "grad_norm": 1.0859375, "learning_rate": 2.1230464558831097e-06, "loss": 0.09, "step": 27989 }, { "epoch": 173.85358255451715, "grad_norm": 2.0, "learning_rate": 2.122052989338033e-06, "loss": 0.1334, "step": 27990 }, { "epoch": 173.85981308411215, "grad_norm": 1.8671875, "learning_rate": 2.1210597449881546e-06, "loss": 0.1125, "step": 27991 }, { "epoch": 173.86604361370718, "grad_norm": 2.0, "learning_rate": 2.120066722843117e-06, "loss": 0.1373, "step": 27992 }, { "epoch": 173.87227414330218, "grad_norm": 2.71875, "learning_rate": 2.119073922912562e-06, "loss": 0.227, "step": 27993 }, { "epoch": 173.8785046728972, "grad_norm": 1.796875, "learning_rate": 2.1180813452061353e-06, "loss": 0.1726, "step": 27994 }, { "epoch": 173.8847352024922, "grad_norm": 1.7890625, "learning_rate": 2.117088989733479e-06, "loss": 0.1313, "step": 27995 }, { "epoch": 173.89096573208724, "grad_norm": 1.8203125, "learning_rate": 2.1160968565042264e-06, "loss": 0.1271, "step": 27996 }, { "epoch": 173.89719626168224, "grad_norm": 2.21875, "learning_rate": 2.1151049455280182e-06, "loss": 0.2123, "step": 27997 }, { "epoch": 173.90342679127727, "grad_norm": 2.6875, "learning_rate": 2.1141132568144815e-06, "loss": 0.2052, "step": 27998 }, { "epoch": 173.90965732087227, "grad_norm": 1.3515625, "learning_rate": 2.113121790373257e-06, "loss": 0.1407, "step": 27999 }, { "epoch": 173.9158878504673, "grad_norm": 1.46875, "learning_rate": 2.112130546213964e-06, "loss": 0.1364, "step": 28000 }, { "epoch": 173.9221183800623, "grad_norm": 2.59375, "learning_rate": 2.111139524346234e-06, "loss": 0.1992, "step": 28001 }, { "epoch": 173.92834890965733, "grad_norm": 2.09375, "learning_rate": 2.110148724779695e-06, "loss": 0.1815, "step": 28002 }, { "epoch": 173.93457943925233, "grad_norm": 3.078125, "learning_rate": 2.1091581475239652e-06, "loss": 0.1951, "step": 28003 }, { "epoch": 173.94080996884736, "grad_norm": 2.265625, "learning_rate": 2.1081677925886723e-06, "loss": 0.2422, "step": 28004 }, { "epoch": 173.94704049844236, "grad_norm": 1.390625, "learning_rate": 2.1071776599834225e-06, "loss": 0.1099, "step": 28005 }, { "epoch": 173.9532710280374, "grad_norm": 0.96484375, "learning_rate": 2.1061877497178445e-06, "loss": 0.0975, "step": 28006 }, { "epoch": 173.9595015576324, "grad_norm": 1.5390625, "learning_rate": 2.105198061801544e-06, "loss": 0.1193, "step": 28007 }, { "epoch": 173.96573208722742, "grad_norm": 2.359375, "learning_rate": 2.1042085962441365e-06, "loss": 0.1953, "step": 28008 }, { "epoch": 173.97196261682242, "grad_norm": 1.3046875, "learning_rate": 2.1032193530552357e-06, "loss": 0.1097, "step": 28009 }, { "epoch": 173.97819314641745, "grad_norm": 2.03125, "learning_rate": 2.102230332244448e-06, "loss": 0.1518, "step": 28010 }, { "epoch": 173.98442367601245, "grad_norm": 1.3359375, "learning_rate": 2.101241533821369e-06, "loss": 0.1033, "step": 28011 }, { "epoch": 173.99065420560748, "grad_norm": 1.359375, "learning_rate": 2.100252957795612e-06, "loss": 0.1075, "step": 28012 }, { "epoch": 173.99688473520249, "grad_norm": 2.21875, "learning_rate": 2.0992646041767786e-06, "loss": 0.136, "step": 28013 }, { "epoch": 174.0, "grad_norm": 1.984375, "learning_rate": 2.0982764729744586e-06, "loss": 0.0936, "step": 28014 }, { "epoch": 174.00623052959503, "grad_norm": 2.453125, "learning_rate": 2.09728856419826e-06, "loss": 0.1125, "step": 28015 }, { "epoch": 174.01246105919003, "grad_norm": 1.5390625, "learning_rate": 2.0963008778577736e-06, "loss": 0.171, "step": 28016 }, { "epoch": 174.01869158878506, "grad_norm": 2.203125, "learning_rate": 2.0953134139625936e-06, "loss": 0.1504, "step": 28017 }, { "epoch": 174.02492211838006, "grad_norm": 1.921875, "learning_rate": 2.094326172522304e-06, "loss": 0.1406, "step": 28018 }, { "epoch": 174.0311526479751, "grad_norm": 1.984375, "learning_rate": 2.0933391535464976e-06, "loss": 0.174, "step": 28019 }, { "epoch": 174.0373831775701, "grad_norm": 2.09375, "learning_rate": 2.092352357044758e-06, "loss": 0.1652, "step": 28020 }, { "epoch": 174.04361370716512, "grad_norm": 2.578125, "learning_rate": 2.0913657830266774e-06, "loss": 0.1787, "step": 28021 }, { "epoch": 174.04984423676012, "grad_norm": 1.375, "learning_rate": 2.0903794315018267e-06, "loss": 0.1025, "step": 28022 }, { "epoch": 174.05607476635515, "grad_norm": 1.8203125, "learning_rate": 2.0893933024797975e-06, "loss": 0.1517, "step": 28023 }, { "epoch": 174.06230529595015, "grad_norm": 2.171875, "learning_rate": 2.0884073959701518e-06, "loss": 0.185, "step": 28024 }, { "epoch": 174.06853582554518, "grad_norm": 2.9375, "learning_rate": 2.0874217119824792e-06, "loss": 0.2945, "step": 28025 }, { "epoch": 174.07476635514018, "grad_norm": 2.609375, "learning_rate": 2.086436250526344e-06, "loss": 0.1455, "step": 28026 }, { "epoch": 174.0809968847352, "grad_norm": 1.65625, "learning_rate": 2.08545101161132e-06, "loss": 0.1524, "step": 28027 }, { "epoch": 174.08722741433021, "grad_norm": 3.0625, "learning_rate": 2.0844659952469814e-06, "loss": 0.3235, "step": 28028 }, { "epoch": 174.09345794392524, "grad_norm": 1.3671875, "learning_rate": 2.0834812014428885e-06, "loss": 0.1205, "step": 28029 }, { "epoch": 174.09968847352025, "grad_norm": 2.65625, "learning_rate": 2.0824966302086024e-06, "loss": 0.1913, "step": 28030 }, { "epoch": 174.10591900311528, "grad_norm": 1.765625, "learning_rate": 2.081512281553691e-06, "loss": 0.1078, "step": 28031 }, { "epoch": 174.11214953271028, "grad_norm": 1.703125, "learning_rate": 2.080528155487715e-06, "loss": 0.1301, "step": 28032 }, { "epoch": 174.1183800623053, "grad_norm": 1.484375, "learning_rate": 2.0795442520202294e-06, "loss": 0.1341, "step": 28033 }, { "epoch": 174.1246105919003, "grad_norm": 1.765625, "learning_rate": 2.0785605711607924e-06, "loss": 0.1551, "step": 28034 }, { "epoch": 174.13084112149534, "grad_norm": 2.09375, "learning_rate": 2.0775771129189602e-06, "loss": 0.1882, "step": 28035 }, { "epoch": 174.13707165109034, "grad_norm": 2.296875, "learning_rate": 2.0765938773042808e-06, "loss": 0.123, "step": 28036 }, { "epoch": 174.14330218068537, "grad_norm": 2.21875, "learning_rate": 2.075610864326302e-06, "loss": 0.1906, "step": 28037 }, { "epoch": 174.14953271028037, "grad_norm": 3.28125, "learning_rate": 2.0746280739945697e-06, "loss": 0.2125, "step": 28038 }, { "epoch": 174.1557632398754, "grad_norm": 1.5859375, "learning_rate": 2.073645506318639e-06, "loss": 0.1462, "step": 28039 }, { "epoch": 174.1619937694704, "grad_norm": 1.3671875, "learning_rate": 2.0726631613080415e-06, "loss": 0.1057, "step": 28040 }, { "epoch": 174.16822429906543, "grad_norm": 1.7890625, "learning_rate": 2.07168103897232e-06, "loss": 0.1461, "step": 28041 }, { "epoch": 174.17445482866043, "grad_norm": 2.234375, "learning_rate": 2.0706991393210224e-06, "loss": 0.2233, "step": 28042 }, { "epoch": 174.18068535825546, "grad_norm": 2.421875, "learning_rate": 2.0697174623636794e-06, "loss": 0.2165, "step": 28043 }, { "epoch": 174.18691588785046, "grad_norm": 1.6953125, "learning_rate": 2.0687360081098174e-06, "loss": 0.0934, "step": 28044 }, { "epoch": 174.1931464174455, "grad_norm": 1.890625, "learning_rate": 2.0677547765689755e-06, "loss": 0.1602, "step": 28045 }, { "epoch": 174.1993769470405, "grad_norm": 1.109375, "learning_rate": 2.066773767750682e-06, "loss": 0.105, "step": 28046 }, { "epoch": 174.20560747663552, "grad_norm": 1.5078125, "learning_rate": 2.0657929816644716e-06, "loss": 0.1344, "step": 28047 }, { "epoch": 174.21183800623052, "grad_norm": 2.796875, "learning_rate": 2.0648124183198643e-06, "loss": 0.2239, "step": 28048 }, { "epoch": 174.21806853582555, "grad_norm": 2.71875, "learning_rate": 2.0638320777263797e-06, "loss": 0.2573, "step": 28049 }, { "epoch": 174.22429906542055, "grad_norm": 1.3125, "learning_rate": 2.0628519598935415e-06, "loss": 0.1057, "step": 28050 }, { "epoch": 174.23052959501558, "grad_norm": 2.171875, "learning_rate": 2.0618720648308775e-06, "loss": 0.1487, "step": 28051 }, { "epoch": 174.23676012461058, "grad_norm": 1.375, "learning_rate": 2.060892392547889e-06, "loss": 0.1033, "step": 28052 }, { "epoch": 174.2429906542056, "grad_norm": 1.640625, "learning_rate": 2.0599129430541013e-06, "loss": 0.1379, "step": 28053 }, { "epoch": 174.2492211838006, "grad_norm": 3.0, "learning_rate": 2.0589337163590295e-06, "loss": 0.2495, "step": 28054 }, { "epoch": 174.25545171339564, "grad_norm": 1.8359375, "learning_rate": 2.0579547124721765e-06, "loss": 0.2231, "step": 28055 }, { "epoch": 174.26168224299064, "grad_norm": 1.25, "learning_rate": 2.0569759314030517e-06, "loss": 0.1153, "step": 28056 }, { "epoch": 174.26791277258567, "grad_norm": 1.3203125, "learning_rate": 2.0559973731611616e-06, "loss": 0.1085, "step": 28057 }, { "epoch": 174.27414330218068, "grad_norm": 2.40625, "learning_rate": 2.0550190377560147e-06, "loss": 0.2171, "step": 28058 }, { "epoch": 174.2803738317757, "grad_norm": 1.5390625, "learning_rate": 2.0540409251971065e-06, "loss": 0.1093, "step": 28059 }, { "epoch": 174.2866043613707, "grad_norm": 2.078125, "learning_rate": 2.0530630354939406e-06, "loss": 0.1306, "step": 28060 }, { "epoch": 174.29283489096574, "grad_norm": 2.125, "learning_rate": 2.0520853686560178e-06, "loss": 0.1885, "step": 28061 }, { "epoch": 174.29906542056074, "grad_norm": 1.2109375, "learning_rate": 2.0511079246928273e-06, "loss": 0.1238, "step": 28062 }, { "epoch": 174.30529595015577, "grad_norm": 2.078125, "learning_rate": 2.050130703613859e-06, "loss": 0.148, "step": 28063 }, { "epoch": 174.31152647975077, "grad_norm": 2.28125, "learning_rate": 2.049153705428611e-06, "loss": 0.1892, "step": 28064 }, { "epoch": 174.3177570093458, "grad_norm": 2.09375, "learning_rate": 2.0481769301465753e-06, "loss": 0.167, "step": 28065 }, { "epoch": 174.3239875389408, "grad_norm": 0.859375, "learning_rate": 2.047200377777225e-06, "loss": 0.0884, "step": 28066 }, { "epoch": 174.33021806853583, "grad_norm": 1.9296875, "learning_rate": 2.0462240483300616e-06, "loss": 0.161, "step": 28067 }, { "epoch": 174.33644859813083, "grad_norm": 1.78125, "learning_rate": 2.0452479418145514e-06, "loss": 0.1082, "step": 28068 }, { "epoch": 174.34267912772586, "grad_norm": 2.28125, "learning_rate": 2.0442720582401873e-06, "loss": 0.115, "step": 28069 }, { "epoch": 174.34890965732086, "grad_norm": 1.609375, "learning_rate": 2.043296397616437e-06, "loss": 0.1185, "step": 28070 }, { "epoch": 174.3551401869159, "grad_norm": 1.171875, "learning_rate": 2.0423209599527814e-06, "loss": 0.1019, "step": 28071 }, { "epoch": 174.3613707165109, "grad_norm": 2.421875, "learning_rate": 2.0413457452586966e-06, "loss": 0.1372, "step": 28072 }, { "epoch": 174.36760124610592, "grad_norm": 2.453125, "learning_rate": 2.0403707535436495e-06, "loss": 0.1736, "step": 28073 }, { "epoch": 174.37383177570092, "grad_norm": 2.03125, "learning_rate": 2.0393959848171134e-06, "loss": 0.2202, "step": 28074 }, { "epoch": 174.38006230529595, "grad_norm": 1.0, "learning_rate": 2.0384214390885478e-06, "loss": 0.0942, "step": 28075 }, { "epoch": 174.38629283489095, "grad_norm": 3.03125, "learning_rate": 2.03744711636743e-06, "loss": 0.3253, "step": 28076 }, { "epoch": 174.39252336448598, "grad_norm": 2.046875, "learning_rate": 2.0364730166632122e-06, "loss": 0.1804, "step": 28077 }, { "epoch": 174.39875389408098, "grad_norm": 2.3125, "learning_rate": 2.0354991399853557e-06, "loss": 0.2731, "step": 28078 }, { "epoch": 174.404984423676, "grad_norm": 1.40625, "learning_rate": 2.034525486343325e-06, "loss": 0.1073, "step": 28079 }, { "epoch": 174.41121495327104, "grad_norm": 2.21875, "learning_rate": 2.0335520557465792e-06, "loss": 0.1679, "step": 28080 }, { "epoch": 174.41744548286604, "grad_norm": 2.34375, "learning_rate": 2.032578848204561e-06, "loss": 0.1847, "step": 28081 }, { "epoch": 174.42367601246107, "grad_norm": 2.03125, "learning_rate": 2.0316058637267267e-06, "loss": 0.1162, "step": 28082 }, { "epoch": 174.42990654205607, "grad_norm": 0.82421875, "learning_rate": 2.0306331023225265e-06, "loss": 0.085, "step": 28083 }, { "epoch": 174.4361370716511, "grad_norm": 1.640625, "learning_rate": 2.029660564001415e-06, "loss": 0.0941, "step": 28084 }, { "epoch": 174.4423676012461, "grad_norm": 1.6875, "learning_rate": 2.028688248772828e-06, "loss": 0.1599, "step": 28085 }, { "epoch": 174.44859813084113, "grad_norm": 1.53125, "learning_rate": 2.0277161566462174e-06, "loss": 0.1322, "step": 28086 }, { "epoch": 174.45482866043614, "grad_norm": 0.890625, "learning_rate": 2.026744287631013e-06, "loss": 0.0924, "step": 28087 }, { "epoch": 174.46105919003116, "grad_norm": 2.703125, "learning_rate": 2.0257726417366645e-06, "loss": 0.1488, "step": 28088 }, { "epoch": 174.46728971962617, "grad_norm": 1.015625, "learning_rate": 2.0248012189726022e-06, "loss": 0.106, "step": 28089 }, { "epoch": 174.4735202492212, "grad_norm": 1.765625, "learning_rate": 2.0238300193482607e-06, "loss": 0.1317, "step": 28090 }, { "epoch": 174.4797507788162, "grad_norm": 1.3515625, "learning_rate": 2.022859042873082e-06, "loss": 0.1196, "step": 28091 }, { "epoch": 174.48598130841123, "grad_norm": 1.1171875, "learning_rate": 2.021888289556484e-06, "loss": 0.0931, "step": 28092 }, { "epoch": 174.49221183800623, "grad_norm": 2.3125, "learning_rate": 2.0209177594079033e-06, "loss": 0.2488, "step": 28093 }, { "epoch": 174.49844236760126, "grad_norm": 1.9140625, "learning_rate": 2.01994745243676e-06, "loss": 0.1554, "step": 28094 }, { "epoch": 174.50467289719626, "grad_norm": 1.953125, "learning_rate": 2.018977368652483e-06, "loss": 0.2333, "step": 28095 }, { "epoch": 174.5109034267913, "grad_norm": 1.953125, "learning_rate": 2.01800750806449e-06, "loss": 0.1337, "step": 28096 }, { "epoch": 174.5171339563863, "grad_norm": 2.46875, "learning_rate": 2.0170378706822006e-06, "loss": 0.2205, "step": 28097 }, { "epoch": 174.52336448598132, "grad_norm": 1.421875, "learning_rate": 2.0160684565150384e-06, "loss": 0.135, "step": 28098 }, { "epoch": 174.52959501557632, "grad_norm": 3.21875, "learning_rate": 2.0150992655724123e-06, "loss": 0.2431, "step": 28099 }, { "epoch": 174.53582554517135, "grad_norm": 2.15625, "learning_rate": 2.014130297863734e-06, "loss": 0.1403, "step": 28100 }, { "epoch": 174.54205607476635, "grad_norm": 3.0, "learning_rate": 2.0131615533984153e-06, "loss": 0.2401, "step": 28101 }, { "epoch": 174.54828660436138, "grad_norm": 1.9609375, "learning_rate": 2.0121930321858718e-06, "loss": 0.1493, "step": 28102 }, { "epoch": 174.55451713395638, "grad_norm": 1.3046875, "learning_rate": 2.0112247342355006e-06, "loss": 0.1287, "step": 28103 }, { "epoch": 174.5607476635514, "grad_norm": 1.6875, "learning_rate": 2.010256659556711e-06, "loss": 0.1364, "step": 28104 }, { "epoch": 174.5669781931464, "grad_norm": 1.1640625, "learning_rate": 2.0092888081589072e-06, "loss": 0.1108, "step": 28105 }, { "epoch": 174.57320872274144, "grad_norm": 2.4375, "learning_rate": 2.0083211800514865e-06, "loss": 0.2665, "step": 28106 }, { "epoch": 174.57943925233644, "grad_norm": 0.5859375, "learning_rate": 2.007353775243842e-06, "loss": 0.0746, "step": 28107 }, { "epoch": 174.58566978193147, "grad_norm": 1.9765625, "learning_rate": 2.0063865937453766e-06, "loss": 0.1448, "step": 28108 }, { "epoch": 174.59190031152647, "grad_norm": 1.578125, "learning_rate": 2.005419635565481e-06, "loss": 0.1318, "step": 28109 }, { "epoch": 174.5981308411215, "grad_norm": 2.375, "learning_rate": 2.004452900713544e-06, "loss": 0.2132, "step": 28110 }, { "epoch": 174.6043613707165, "grad_norm": 0.9609375, "learning_rate": 2.003486389198958e-06, "loss": 0.0853, "step": 28111 }, { "epoch": 174.61059190031153, "grad_norm": 2.28125, "learning_rate": 2.0025201010311135e-06, "loss": 0.2281, "step": 28112 }, { "epoch": 174.61682242990653, "grad_norm": 3.109375, "learning_rate": 2.0015540362193885e-06, "loss": 0.2733, "step": 28113 }, { "epoch": 174.62305295950156, "grad_norm": 2.640625, "learning_rate": 2.0005881947731646e-06, "loss": 0.1924, "step": 28114 }, { "epoch": 174.62928348909657, "grad_norm": 1.765625, "learning_rate": 1.999622576701826e-06, "loss": 0.1968, "step": 28115 }, { "epoch": 174.6355140186916, "grad_norm": 1.1328125, "learning_rate": 1.9986571820147514e-06, "loss": 0.1206, "step": 28116 }, { "epoch": 174.6417445482866, "grad_norm": 1.9140625, "learning_rate": 1.9976920107213193e-06, "loss": 0.1718, "step": 28117 }, { "epoch": 174.64797507788163, "grad_norm": 2.390625, "learning_rate": 1.9967270628309025e-06, "loss": 0.138, "step": 28118 }, { "epoch": 174.65420560747663, "grad_norm": 2.0, "learning_rate": 1.9957623383528633e-06, "loss": 0.1516, "step": 28119 }, { "epoch": 174.66043613707166, "grad_norm": 1.6015625, "learning_rate": 1.99479783729658e-06, "loss": 0.1363, "step": 28120 }, { "epoch": 174.66666666666666, "grad_norm": 2.28125, "learning_rate": 1.9938335596714262e-06, "loss": 0.204, "step": 28121 }, { "epoch": 174.6728971962617, "grad_norm": 2.65625, "learning_rate": 1.9928695054867524e-06, "loss": 0.3558, "step": 28122 }, { "epoch": 174.6791277258567, "grad_norm": 2.53125, "learning_rate": 1.991905674751929e-06, "loss": 0.2192, "step": 28123 }, { "epoch": 174.68535825545172, "grad_norm": 1.96875, "learning_rate": 1.99094206747632e-06, "loss": 0.193, "step": 28124 }, { "epoch": 174.69158878504672, "grad_norm": 1.46875, "learning_rate": 1.989978683669283e-06, "loss": 0.1255, "step": 28125 }, { "epoch": 174.69781931464175, "grad_norm": 1.9765625, "learning_rate": 1.989015523340168e-06, "loss": 0.1518, "step": 28126 }, { "epoch": 174.70404984423675, "grad_norm": 2.203125, "learning_rate": 1.9880525864983347e-06, "loss": 0.234, "step": 28127 }, { "epoch": 174.71028037383178, "grad_norm": 2.703125, "learning_rate": 1.9870898731531364e-06, "loss": 0.2327, "step": 28128 }, { "epoch": 174.71651090342678, "grad_norm": 1.5625, "learning_rate": 1.986127383313921e-06, "loss": 0.1265, "step": 28129 }, { "epoch": 174.7227414330218, "grad_norm": 1.9375, "learning_rate": 1.9851651169900343e-06, "loss": 0.1442, "step": 28130 }, { "epoch": 174.7289719626168, "grad_norm": 2.53125, "learning_rate": 1.9842030741908297e-06, "loss": 0.2067, "step": 28131 }, { "epoch": 174.73520249221184, "grad_norm": 1.109375, "learning_rate": 1.9832412549256442e-06, "loss": 0.1129, "step": 28132 }, { "epoch": 174.74143302180684, "grad_norm": 1.9609375, "learning_rate": 1.9822796592038174e-06, "loss": 0.1811, "step": 28133 }, { "epoch": 174.74766355140187, "grad_norm": 1.6796875, "learning_rate": 1.981318287034692e-06, "loss": 0.1261, "step": 28134 }, { "epoch": 174.75389408099687, "grad_norm": 1.1015625, "learning_rate": 1.9803571384276075e-06, "loss": 0.1125, "step": 28135 }, { "epoch": 174.7601246105919, "grad_norm": 1.7109375, "learning_rate": 1.979396213391893e-06, "loss": 0.1393, "step": 28136 }, { "epoch": 174.7663551401869, "grad_norm": 2.375, "learning_rate": 1.978435511936888e-06, "loss": 0.1819, "step": 28137 }, { "epoch": 174.77258566978193, "grad_norm": 1.765625, "learning_rate": 1.9774750340719156e-06, "loss": 0.1656, "step": 28138 }, { "epoch": 174.77881619937693, "grad_norm": 1.40625, "learning_rate": 1.9765147798063127e-06, "loss": 0.1306, "step": 28139 }, { "epoch": 174.78504672897196, "grad_norm": 1.3203125, "learning_rate": 1.975554749149394e-06, "loss": 0.1103, "step": 28140 }, { "epoch": 174.79127725856696, "grad_norm": 1.9609375, "learning_rate": 1.9745949421104914e-06, "loss": 0.1357, "step": 28141 }, { "epoch": 174.797507788162, "grad_norm": 1.3984375, "learning_rate": 1.9736353586989248e-06, "loss": 0.1183, "step": 28142 }, { "epoch": 174.80373831775702, "grad_norm": 1.578125, "learning_rate": 1.97267599892402e-06, "loss": 0.1319, "step": 28143 }, { "epoch": 174.80996884735202, "grad_norm": 2.546875, "learning_rate": 1.9717168627950896e-06, "loss": 0.2132, "step": 28144 }, { "epoch": 174.81619937694705, "grad_norm": 3.265625, "learning_rate": 1.9707579503214418e-06, "loss": 0.3799, "step": 28145 }, { "epoch": 174.82242990654206, "grad_norm": 1.2109375, "learning_rate": 1.9697992615123977e-06, "loss": 0.1094, "step": 28146 }, { "epoch": 174.82866043613708, "grad_norm": 1.453125, "learning_rate": 1.968840796377269e-06, "loss": 0.1279, "step": 28147 }, { "epoch": 174.8348909657321, "grad_norm": 2.25, "learning_rate": 1.96788255492536e-06, "loss": 0.2082, "step": 28148 }, { "epoch": 174.84112149532712, "grad_norm": 0.890625, "learning_rate": 1.9669245371659788e-06, "loss": 0.094, "step": 28149 }, { "epoch": 174.84735202492212, "grad_norm": 2.53125, "learning_rate": 1.9659667431084355e-06, "loss": 0.1761, "step": 28150 }, { "epoch": 174.85358255451715, "grad_norm": 1.59375, "learning_rate": 1.9650091727620277e-06, "loss": 0.1275, "step": 28151 }, { "epoch": 174.85981308411215, "grad_norm": 1.390625, "learning_rate": 1.964051826136051e-06, "loss": 0.1296, "step": 28152 }, { "epoch": 174.86604361370718, "grad_norm": 1.1796875, "learning_rate": 1.9630947032398067e-06, "loss": 0.1289, "step": 28153 }, { "epoch": 174.87227414330218, "grad_norm": 1.6015625, "learning_rate": 1.9621378040825977e-06, "loss": 0.1093, "step": 28154 }, { "epoch": 174.8785046728972, "grad_norm": 3.640625, "learning_rate": 1.961181128673706e-06, "loss": 0.3163, "step": 28155 }, { "epoch": 174.8847352024922, "grad_norm": 1.5703125, "learning_rate": 1.960224677022435e-06, "loss": 0.1226, "step": 28156 }, { "epoch": 174.89096573208724, "grad_norm": 3.21875, "learning_rate": 1.9592684491380603e-06, "loss": 0.3069, "step": 28157 }, { "epoch": 174.89719626168224, "grad_norm": 1.921875, "learning_rate": 1.9583124450298834e-06, "loss": 0.091, "step": 28158 }, { "epoch": 174.90342679127727, "grad_norm": 1.375, "learning_rate": 1.9573566647071747e-06, "loss": 0.12, "step": 28159 }, { "epoch": 174.90965732087227, "grad_norm": 2.359375, "learning_rate": 1.956401108179226e-06, "loss": 0.1909, "step": 28160 }, { "epoch": 174.9158878504673, "grad_norm": 1.8515625, "learning_rate": 1.9554457754553197e-06, "loss": 0.1593, "step": 28161 }, { "epoch": 174.9221183800623, "grad_norm": 1.2265625, "learning_rate": 1.9544906665447255e-06, "loss": 0.1126, "step": 28162 }, { "epoch": 174.92834890965733, "grad_norm": 1.8359375, "learning_rate": 1.953535781456731e-06, "loss": 0.0933, "step": 28163 }, { "epoch": 174.93457943925233, "grad_norm": 3.0, "learning_rate": 1.952581120200597e-06, "loss": 0.104, "step": 28164 }, { "epoch": 174.94080996884736, "grad_norm": 2.625, "learning_rate": 1.9516266827856096e-06, "loss": 0.2482, "step": 28165 }, { "epoch": 174.94704049844236, "grad_norm": 3.296875, "learning_rate": 1.9506724692210237e-06, "loss": 0.3104, "step": 28166 }, { "epoch": 174.9532710280374, "grad_norm": 2.03125, "learning_rate": 1.949718479516116e-06, "loss": 0.1219, "step": 28167 }, { "epoch": 174.9595015576324, "grad_norm": 2.25, "learning_rate": 1.948764713680154e-06, "loss": 0.1574, "step": 28168 }, { "epoch": 174.96573208722742, "grad_norm": 3.265625, "learning_rate": 1.947811171722397e-06, "loss": 0.1684, "step": 28169 }, { "epoch": 174.97196261682242, "grad_norm": 2.265625, "learning_rate": 1.946857853652101e-06, "loss": 0.1213, "step": 28170 }, { "epoch": 174.97819314641745, "grad_norm": 1.4921875, "learning_rate": 1.945904759478531e-06, "loss": 0.0969, "step": 28171 }, { "epoch": 174.98442367601245, "grad_norm": 1.21875, "learning_rate": 1.944951889210947e-06, "loss": 0.1291, "step": 28172 }, { "epoch": 174.99065420560748, "grad_norm": 0.8671875, "learning_rate": 1.9439992428585936e-06, "loss": 0.1029, "step": 28173 }, { "epoch": 174.99688473520249, "grad_norm": 2.109375, "learning_rate": 1.943046820430727e-06, "loss": 0.1383, "step": 28174 }, { "epoch": 175.0, "grad_norm": 1.8515625, "learning_rate": 1.942094621936605e-06, "loss": 0.1279, "step": 28175 }, { "epoch": 175.00623052959503, "grad_norm": 1.375, "learning_rate": 1.9411426473854688e-06, "loss": 0.1068, "step": 28176 }, { "epoch": 175.01246105919003, "grad_norm": 1.734375, "learning_rate": 1.9401908967865617e-06, "loss": 0.1374, "step": 28177 }, { "epoch": 175.01869158878506, "grad_norm": 2.40625, "learning_rate": 1.9392393701491264e-06, "loss": 0.1585, "step": 28178 }, { "epoch": 175.02492211838006, "grad_norm": 1.15625, "learning_rate": 1.9382880674824104e-06, "loss": 0.1324, "step": 28179 }, { "epoch": 175.0311526479751, "grad_norm": 2.578125, "learning_rate": 1.937336988795657e-06, "loss": 0.2279, "step": 28180 }, { "epoch": 175.0373831775701, "grad_norm": 1.609375, "learning_rate": 1.9363861340980895e-06, "loss": 0.1451, "step": 28181 }, { "epoch": 175.04361370716512, "grad_norm": 1.6484375, "learning_rate": 1.9354355033989553e-06, "loss": 0.1768, "step": 28182 }, { "epoch": 175.04984423676012, "grad_norm": 1.7265625, "learning_rate": 1.934485096707478e-06, "loss": 0.0995, "step": 28183 }, { "epoch": 175.05607476635515, "grad_norm": 1.7890625, "learning_rate": 1.9335349140328975e-06, "loss": 0.1678, "step": 28184 }, { "epoch": 175.06230529595015, "grad_norm": 1.5078125, "learning_rate": 1.932584955384431e-06, "loss": 0.0934, "step": 28185 }, { "epoch": 175.06853582554518, "grad_norm": 2.203125, "learning_rate": 1.9316352207713105e-06, "loss": 0.2056, "step": 28186 }, { "epoch": 175.07476635514018, "grad_norm": 1.0078125, "learning_rate": 1.930685710202765e-06, "loss": 0.0882, "step": 28187 }, { "epoch": 175.0809968847352, "grad_norm": 2.5, "learning_rate": 1.929736423688011e-06, "loss": 0.2067, "step": 28188 }, { "epoch": 175.08722741433021, "grad_norm": 1.8359375, "learning_rate": 1.9287873612362615e-06, "loss": 0.1355, "step": 28189 }, { "epoch": 175.09345794392524, "grad_norm": 1.8828125, "learning_rate": 1.927838522856745e-06, "loss": 0.1404, "step": 28190 }, { "epoch": 175.09968847352025, "grad_norm": 1.59375, "learning_rate": 1.9268899085586766e-06, "loss": 0.1478, "step": 28191 }, { "epoch": 175.10591900311528, "grad_norm": 2.578125, "learning_rate": 1.9259415183512596e-06, "loss": 0.1872, "step": 28192 }, { "epoch": 175.11214953271028, "grad_norm": 2.890625, "learning_rate": 1.924993352243712e-06, "loss": 0.2663, "step": 28193 }, { "epoch": 175.1183800623053, "grad_norm": 1.171875, "learning_rate": 1.9240454102452456e-06, "loss": 0.0907, "step": 28194 }, { "epoch": 175.1246105919003, "grad_norm": 2.4375, "learning_rate": 1.923097692365064e-06, "loss": 0.1278, "step": 28195 }, { "epoch": 175.13084112149534, "grad_norm": 1.9296875, "learning_rate": 1.922150198612366e-06, "loss": 0.1669, "step": 28196 }, { "epoch": 175.13707165109034, "grad_norm": 1.703125, "learning_rate": 1.921202928996357e-06, "loss": 0.1199, "step": 28197 }, { "epoch": 175.14330218068537, "grad_norm": 1.8359375, "learning_rate": 1.9202558835262423e-06, "loss": 0.1538, "step": 28198 }, { "epoch": 175.14953271028037, "grad_norm": 1.2109375, "learning_rate": 1.9193090622112136e-06, "loss": 0.1139, "step": 28199 }, { "epoch": 175.1557632398754, "grad_norm": 2.640625, "learning_rate": 1.918362465060469e-06, "loss": 0.1334, "step": 28200 }, { "epoch": 175.1619937694704, "grad_norm": 2.78125, "learning_rate": 1.9174160920832045e-06, "loss": 0.2719, "step": 28201 }, { "epoch": 175.16822429906543, "grad_norm": 2.921875, "learning_rate": 1.916469943288612e-06, "loss": 0.3065, "step": 28202 }, { "epoch": 175.17445482866043, "grad_norm": 2.0625, "learning_rate": 1.9155240186858716e-06, "loss": 0.0983, "step": 28203 }, { "epoch": 175.18068535825546, "grad_norm": 2.03125, "learning_rate": 1.914578318284177e-06, "loss": 0.139, "step": 28204 }, { "epoch": 175.18691588785046, "grad_norm": 1.8203125, "learning_rate": 1.9136328420927163e-06, "loss": 0.149, "step": 28205 }, { "epoch": 175.1931464174455, "grad_norm": 2.25, "learning_rate": 1.9126875901206624e-06, "loss": 0.1649, "step": 28206 }, { "epoch": 175.1993769470405, "grad_norm": 3.03125, "learning_rate": 1.911742562377208e-06, "loss": 0.1866, "step": 28207 }, { "epoch": 175.20560747663552, "grad_norm": 2.328125, "learning_rate": 1.9107977588715216e-06, "loss": 0.2545, "step": 28208 }, { "epoch": 175.21183800623052, "grad_norm": 1.53125, "learning_rate": 1.909853179612786e-06, "loss": 0.1556, "step": 28209 }, { "epoch": 175.21806853582555, "grad_norm": 1.1796875, "learning_rate": 1.9089088246101673e-06, "loss": 0.1053, "step": 28210 }, { "epoch": 175.22429906542055, "grad_norm": 1.875, "learning_rate": 1.9079646938728407e-06, "loss": 0.1781, "step": 28211 }, { "epoch": 175.23052959501558, "grad_norm": 1.2734375, "learning_rate": 1.9070207874099799e-06, "loss": 0.1402, "step": 28212 }, { "epoch": 175.23676012461058, "grad_norm": 2.0, "learning_rate": 1.9060771052307524e-06, "loss": 0.2312, "step": 28213 }, { "epoch": 175.2429906542056, "grad_norm": 1.265625, "learning_rate": 1.9051336473443203e-06, "loss": 0.1119, "step": 28214 }, { "epoch": 175.2492211838006, "grad_norm": 1.5625, "learning_rate": 1.9041904137598432e-06, "loss": 0.1258, "step": 28215 }, { "epoch": 175.25545171339564, "grad_norm": 1.5234375, "learning_rate": 1.9032474044864828e-06, "loss": 0.1092, "step": 28216 }, { "epoch": 175.26168224299064, "grad_norm": 2.703125, "learning_rate": 1.902304619533407e-06, "loss": 0.1992, "step": 28217 }, { "epoch": 175.26791277258567, "grad_norm": 2.828125, "learning_rate": 1.9013620589097614e-06, "loss": 0.3206, "step": 28218 }, { "epoch": 175.27414330218068, "grad_norm": 3.4375, "learning_rate": 1.900419722624705e-06, "loss": 0.2787, "step": 28219 }, { "epoch": 175.2803738317757, "grad_norm": 2.765625, "learning_rate": 1.8994776106873946e-06, "loss": 0.2403, "step": 28220 }, { "epoch": 175.2866043613707, "grad_norm": 1.3515625, "learning_rate": 1.8985357231069756e-06, "loss": 0.1314, "step": 28221 }, { "epoch": 175.29283489096574, "grad_norm": 1.5078125, "learning_rate": 1.8975940598925906e-06, "loss": 0.1236, "step": 28222 }, { "epoch": 175.29906542056074, "grad_norm": 2.4375, "learning_rate": 1.8966526210533908e-06, "loss": 0.2169, "step": 28223 }, { "epoch": 175.30529595015577, "grad_norm": 1.8046875, "learning_rate": 1.8957114065985215e-06, "loss": 0.1603, "step": 28224 }, { "epoch": 175.31152647975077, "grad_norm": 1.96875, "learning_rate": 1.8947704165371172e-06, "loss": 0.1019, "step": 28225 }, { "epoch": 175.3177570093458, "grad_norm": 2.234375, "learning_rate": 1.893829650878326e-06, "loss": 0.1641, "step": 28226 }, { "epoch": 175.3239875389408, "grad_norm": 1.4375, "learning_rate": 1.8928891096312768e-06, "loss": 0.1318, "step": 28227 }, { "epoch": 175.33021806853583, "grad_norm": 3.828125, "learning_rate": 1.8919487928051122e-06, "loss": 0.3482, "step": 28228 }, { "epoch": 175.33644859813083, "grad_norm": 3.03125, "learning_rate": 1.8910087004089555e-06, "loss": 0.3048, "step": 28229 }, { "epoch": 175.34267912772586, "grad_norm": 2.40625, "learning_rate": 1.8900688324519411e-06, "loss": 0.2068, "step": 28230 }, { "epoch": 175.34890965732086, "grad_norm": 1.09375, "learning_rate": 1.8891291889432007e-06, "loss": 0.0932, "step": 28231 }, { "epoch": 175.3551401869159, "grad_norm": 1.6171875, "learning_rate": 1.8881897698918544e-06, "loss": 0.1239, "step": 28232 }, { "epoch": 175.3613707165109, "grad_norm": 1.625, "learning_rate": 1.887250575307034e-06, "loss": 0.1051, "step": 28233 }, { "epoch": 175.36760124610592, "grad_norm": 1.3203125, "learning_rate": 1.886311605197852e-06, "loss": 0.1242, "step": 28234 }, { "epoch": 175.37383177570092, "grad_norm": 2.21875, "learning_rate": 1.8853728595734365e-06, "loss": 0.1798, "step": 28235 }, { "epoch": 175.38006230529595, "grad_norm": 1.125, "learning_rate": 1.8844343384428948e-06, "loss": 0.0966, "step": 28236 }, { "epoch": 175.38629283489095, "grad_norm": 1.140625, "learning_rate": 1.883496041815347e-06, "loss": 0.1244, "step": 28237 }, { "epoch": 175.39252336448598, "grad_norm": 2.03125, "learning_rate": 1.8825579696999079e-06, "loss": 0.1652, "step": 28238 }, { "epoch": 175.39875389408098, "grad_norm": 2.53125, "learning_rate": 1.8816201221056956e-06, "loss": 0.2188, "step": 28239 }, { "epoch": 175.404984423676, "grad_norm": 2.15625, "learning_rate": 1.8806824990417998e-06, "loss": 0.1469, "step": 28240 }, { "epoch": 175.41121495327104, "grad_norm": 1.453125, "learning_rate": 1.8797451005173384e-06, "loss": 0.1378, "step": 28241 }, { "epoch": 175.41744548286604, "grad_norm": 2.375, "learning_rate": 1.878807926541412e-06, "loss": 0.119, "step": 28242 }, { "epoch": 175.42367601246107, "grad_norm": 1.234375, "learning_rate": 1.8778709771231335e-06, "loss": 0.1132, "step": 28243 }, { "epoch": 175.42990654205607, "grad_norm": 1.7578125, "learning_rate": 1.8769342522715866e-06, "loss": 0.1397, "step": 28244 }, { "epoch": 175.4361370716511, "grad_norm": 2.21875, "learning_rate": 1.875997751995881e-06, "loss": 0.1798, "step": 28245 }, { "epoch": 175.4423676012461, "grad_norm": 2.90625, "learning_rate": 1.8750614763051039e-06, "loss": 0.2215, "step": 28246 }, { "epoch": 175.44859813084113, "grad_norm": 1.359375, "learning_rate": 1.8741254252083561e-06, "loss": 0.1125, "step": 28247 }, { "epoch": 175.45482866043614, "grad_norm": 1.296875, "learning_rate": 1.8731895987147197e-06, "loss": 0.1077, "step": 28248 }, { "epoch": 175.46105919003116, "grad_norm": 2.046875, "learning_rate": 1.8722539968332924e-06, "loss": 0.1733, "step": 28249 }, { "epoch": 175.46728971962617, "grad_norm": 1.3125, "learning_rate": 1.8713186195731591e-06, "loss": 0.1153, "step": 28250 }, { "epoch": 175.4735202492212, "grad_norm": 2.015625, "learning_rate": 1.8703834669433985e-06, "loss": 0.1772, "step": 28251 }, { "epoch": 175.4797507788162, "grad_norm": 2.25, "learning_rate": 1.8694485389531003e-06, "loss": 0.1029, "step": 28252 }, { "epoch": 175.48598130841123, "grad_norm": 2.0, "learning_rate": 1.8685138356113408e-06, "loss": 0.1631, "step": 28253 }, { "epoch": 175.49221183800623, "grad_norm": 1.0234375, "learning_rate": 1.8675793569272016e-06, "loss": 0.11, "step": 28254 }, { "epoch": 175.49844236760126, "grad_norm": 1.6953125, "learning_rate": 1.8666451029097504e-06, "loss": 0.1328, "step": 28255 }, { "epoch": 175.50467289719626, "grad_norm": 2.328125, "learning_rate": 1.865711073568066e-06, "loss": 0.169, "step": 28256 }, { "epoch": 175.5109034267913, "grad_norm": 1.40625, "learning_rate": 1.8647772689112247e-06, "loss": 0.1403, "step": 28257 }, { "epoch": 175.5171339563863, "grad_norm": 1.8203125, "learning_rate": 1.8638436889482912e-06, "loss": 0.147, "step": 28258 }, { "epoch": 175.52336448598132, "grad_norm": 1.515625, "learning_rate": 1.8629103336883308e-06, "loss": 0.1614, "step": 28259 }, { "epoch": 175.52959501557632, "grad_norm": 3.21875, "learning_rate": 1.8619772031404081e-06, "loss": 0.2483, "step": 28260 }, { "epoch": 175.53582554517135, "grad_norm": 2.40625, "learning_rate": 1.861044297313591e-06, "loss": 0.2488, "step": 28261 }, { "epoch": 175.54205607476635, "grad_norm": 1.953125, "learning_rate": 1.8601116162169336e-06, "loss": 0.1294, "step": 28262 }, { "epoch": 175.54828660436138, "grad_norm": 3.828125, "learning_rate": 1.8591791598595004e-06, "loss": 0.1956, "step": 28263 }, { "epoch": 175.55451713395638, "grad_norm": 2.34375, "learning_rate": 1.8582469282503457e-06, "loss": 0.1031, "step": 28264 }, { "epoch": 175.5607476635514, "grad_norm": 1.796875, "learning_rate": 1.8573149213985258e-06, "loss": 0.1477, "step": 28265 }, { "epoch": 175.5669781931464, "grad_norm": 2.265625, "learning_rate": 1.8563831393130838e-06, "loss": 0.2169, "step": 28266 }, { "epoch": 175.57320872274144, "grad_norm": 2.578125, "learning_rate": 1.8554515820030732e-06, "loss": 0.1878, "step": 28267 }, { "epoch": 175.57943925233644, "grad_norm": 1.09375, "learning_rate": 1.8545202494775509e-06, "loss": 0.1222, "step": 28268 }, { "epoch": 175.58566978193147, "grad_norm": 1.203125, "learning_rate": 1.8535891417455487e-06, "loss": 0.0907, "step": 28269 }, { "epoch": 175.59190031152647, "grad_norm": 1.9921875, "learning_rate": 1.8526582588161146e-06, "loss": 0.1083, "step": 28270 }, { "epoch": 175.5981308411215, "grad_norm": 1.34375, "learning_rate": 1.851727600698297e-06, "loss": 0.131, "step": 28271 }, { "epoch": 175.6043613707165, "grad_norm": 1.0078125, "learning_rate": 1.8507971674011276e-06, "loss": 0.0981, "step": 28272 }, { "epoch": 175.61059190031153, "grad_norm": 0.97265625, "learning_rate": 1.849866958933638e-06, "loss": 0.0873, "step": 28273 }, { "epoch": 175.61682242990653, "grad_norm": 2.609375, "learning_rate": 1.8489369753048685e-06, "loss": 0.3155, "step": 28274 }, { "epoch": 175.62305295950156, "grad_norm": 2.421875, "learning_rate": 1.8480072165238533e-06, "loss": 0.1525, "step": 28275 }, { "epoch": 175.62928348909657, "grad_norm": 1.3515625, "learning_rate": 1.8470776825996212e-06, "loss": 0.0952, "step": 28276 }, { "epoch": 175.6355140186916, "grad_norm": 1.9375, "learning_rate": 1.8461483735412011e-06, "loss": 0.1612, "step": 28277 }, { "epoch": 175.6417445482866, "grad_norm": 1.8203125, "learning_rate": 1.8452192893576109e-06, "loss": 0.1308, "step": 28278 }, { "epoch": 175.64797507788163, "grad_norm": 1.1953125, "learning_rate": 1.8442904300578795e-06, "loss": 0.1119, "step": 28279 }, { "epoch": 175.65420560747663, "grad_norm": 2.09375, "learning_rate": 1.843361795651033e-06, "loss": 0.158, "step": 28280 }, { "epoch": 175.66043613707166, "grad_norm": 1.8515625, "learning_rate": 1.8424333861460807e-06, "loss": 0.1258, "step": 28281 }, { "epoch": 175.66666666666666, "grad_norm": 3.125, "learning_rate": 1.8415052015520433e-06, "loss": 0.3037, "step": 28282 }, { "epoch": 175.6728971962617, "grad_norm": 1.359375, "learning_rate": 1.8405772418779415e-06, "loss": 0.0899, "step": 28283 }, { "epoch": 175.6791277258567, "grad_norm": 3.015625, "learning_rate": 1.8396495071327845e-06, "loss": 0.2321, "step": 28284 }, { "epoch": 175.68535825545172, "grad_norm": 1.7734375, "learning_rate": 1.8387219973255765e-06, "loss": 0.1357, "step": 28285 }, { "epoch": 175.69158878504672, "grad_norm": 1.8828125, "learning_rate": 1.8377947124653295e-06, "loss": 0.1187, "step": 28286 }, { "epoch": 175.69781931464175, "grad_norm": 3.0, "learning_rate": 1.8368676525610557e-06, "loss": 0.3532, "step": 28287 }, { "epoch": 175.70404984423675, "grad_norm": 2.28125, "learning_rate": 1.835940817621748e-06, "loss": 0.1979, "step": 28288 }, { "epoch": 175.71028037383178, "grad_norm": 1.7734375, "learning_rate": 1.8350142076564158e-06, "loss": 0.1703, "step": 28289 }, { "epoch": 175.71651090342678, "grad_norm": 1.71875, "learning_rate": 1.8340878226740576e-06, "loss": 0.1304, "step": 28290 }, { "epoch": 175.7227414330218, "grad_norm": 1.234375, "learning_rate": 1.8331616626836718e-06, "loss": 0.1346, "step": 28291 }, { "epoch": 175.7289719626168, "grad_norm": 2.546875, "learning_rate": 1.8322357276942453e-06, "loss": 0.2564, "step": 28292 }, { "epoch": 175.73520249221184, "grad_norm": 1.3046875, "learning_rate": 1.831310017714774e-06, "loss": 0.0853, "step": 28293 }, { "epoch": 175.74143302180684, "grad_norm": 0.71484375, "learning_rate": 1.830384532754259e-06, "loss": 0.0694, "step": 28294 }, { "epoch": 175.74766355140187, "grad_norm": 2.125, "learning_rate": 1.8294592728216765e-06, "loss": 0.1551, "step": 28295 }, { "epoch": 175.75389408099687, "grad_norm": 1.7578125, "learning_rate": 1.828534237926019e-06, "loss": 0.1231, "step": 28296 }, { "epoch": 175.7601246105919, "grad_norm": 1.234375, "learning_rate": 1.8276094280762658e-06, "loss": 0.1273, "step": 28297 }, { "epoch": 175.7663551401869, "grad_norm": 0.984375, "learning_rate": 1.8266848432814066e-06, "loss": 0.1203, "step": 28298 }, { "epoch": 175.77258566978193, "grad_norm": 1.1328125, "learning_rate": 1.8257604835504122e-06, "loss": 0.0878, "step": 28299 }, { "epoch": 175.77881619937693, "grad_norm": 1.5625, "learning_rate": 1.8248363488922643e-06, "loss": 0.1489, "step": 28300 }, { "epoch": 175.78504672897196, "grad_norm": 0.86328125, "learning_rate": 1.8239124393159445e-06, "loss": 0.1084, "step": 28301 }, { "epoch": 175.79127725856696, "grad_norm": 0.9296875, "learning_rate": 1.8229887548304125e-06, "loss": 0.1222, "step": 28302 }, { "epoch": 175.797507788162, "grad_norm": 2.125, "learning_rate": 1.8220652954446526e-06, "loss": 0.1238, "step": 28303 }, { "epoch": 175.80373831775702, "grad_norm": 1.2421875, "learning_rate": 1.8211420611676245e-06, "loss": 0.1017, "step": 28304 }, { "epoch": 175.80996884735202, "grad_norm": 2.109375, "learning_rate": 1.8202190520082984e-06, "loss": 0.1528, "step": 28305 }, { "epoch": 175.81619937694705, "grad_norm": 2.984375, "learning_rate": 1.8192962679756371e-06, "loss": 0.2115, "step": 28306 }, { "epoch": 175.82242990654206, "grad_norm": 3.171875, "learning_rate": 1.8183737090786051e-06, "loss": 0.2642, "step": 28307 }, { "epoch": 175.82866043613708, "grad_norm": 2.5, "learning_rate": 1.8174513753261596e-06, "loss": 0.203, "step": 28308 }, { "epoch": 175.8348909657321, "grad_norm": 1.25, "learning_rate": 1.816529266727271e-06, "loss": 0.0997, "step": 28309 }, { "epoch": 175.84112149532712, "grad_norm": 2.3125, "learning_rate": 1.8156073832908737e-06, "loss": 0.1833, "step": 28310 }, { "epoch": 175.84735202492212, "grad_norm": 1.609375, "learning_rate": 1.8146857250259331e-06, "loss": 0.0927, "step": 28311 }, { "epoch": 175.85358255451715, "grad_norm": 2.15625, "learning_rate": 1.8137642919413971e-06, "loss": 0.1898, "step": 28312 }, { "epoch": 175.85981308411215, "grad_norm": 1.21875, "learning_rate": 1.8128430840462258e-06, "loss": 0.1174, "step": 28313 }, { "epoch": 175.86604361370718, "grad_norm": 2.53125, "learning_rate": 1.8119221013493505e-06, "loss": 0.233, "step": 28314 }, { "epoch": 175.87227414330218, "grad_norm": 2.046875, "learning_rate": 1.8110013438597279e-06, "loss": 0.0984, "step": 28315 }, { "epoch": 175.8785046728972, "grad_norm": 2.703125, "learning_rate": 1.81008081158629e-06, "loss": 0.2172, "step": 28316 }, { "epoch": 175.8847352024922, "grad_norm": 1.6796875, "learning_rate": 1.8091605045379877e-06, "loss": 0.1713, "step": 28317 }, { "epoch": 175.89096573208724, "grad_norm": 2.265625, "learning_rate": 1.8082404227237505e-06, "loss": 0.1274, "step": 28318 }, { "epoch": 175.89719626168224, "grad_norm": 1.59375, "learning_rate": 1.8073205661525177e-06, "loss": 0.1385, "step": 28319 }, { "epoch": 175.90342679127727, "grad_norm": 2.65625, "learning_rate": 1.80640093483323e-06, "loss": 0.1726, "step": 28320 }, { "epoch": 175.90965732087227, "grad_norm": 3.5, "learning_rate": 1.805481528774805e-06, "loss": 0.2732, "step": 28321 }, { "epoch": 175.9158878504673, "grad_norm": 1.59375, "learning_rate": 1.8045623479861883e-06, "loss": 0.1498, "step": 28322 }, { "epoch": 175.9221183800623, "grad_norm": 2.125, "learning_rate": 1.8036433924762924e-06, "loss": 0.1371, "step": 28323 }, { "epoch": 175.92834890965733, "grad_norm": 1.75, "learning_rate": 1.8027246622540517e-06, "loss": 0.126, "step": 28324 }, { "epoch": 175.93457943925233, "grad_norm": 2.484375, "learning_rate": 1.801806157328384e-06, "loss": 0.2197, "step": 28325 }, { "epoch": 175.94080996884736, "grad_norm": 1.8828125, "learning_rate": 1.80088787770821e-06, "loss": 0.1222, "step": 28326 }, { "epoch": 175.94704049844236, "grad_norm": 1.984375, "learning_rate": 1.7999698234024559e-06, "loss": 0.1469, "step": 28327 }, { "epoch": 175.9532710280374, "grad_norm": 1.953125, "learning_rate": 1.7990519944200313e-06, "loss": 0.1484, "step": 28328 }, { "epoch": 175.9595015576324, "grad_norm": 1.703125, "learning_rate": 1.7981343907698484e-06, "loss": 0.0934, "step": 28329 }, { "epoch": 175.96573208722742, "grad_norm": 2.671875, "learning_rate": 1.7972170124608224e-06, "loss": 0.1691, "step": 28330 }, { "epoch": 175.97196261682242, "grad_norm": 2.4375, "learning_rate": 1.7962998595018682e-06, "loss": 0.1869, "step": 28331 }, { "epoch": 175.97819314641745, "grad_norm": 2.46875, "learning_rate": 1.7953829319018817e-06, "loss": 0.2182, "step": 28332 }, { "epoch": 175.98442367601245, "grad_norm": 1.5859375, "learning_rate": 1.7944662296697751e-06, "loss": 0.1156, "step": 28333 }, { "epoch": 175.99065420560748, "grad_norm": 2.15625, "learning_rate": 1.7935497528144524e-06, "loss": 0.1652, "step": 28334 }, { "epoch": 175.99688473520249, "grad_norm": 2.09375, "learning_rate": 1.7926335013448204e-06, "loss": 0.169, "step": 28335 }, { "epoch": 176.0, "grad_norm": 1.71875, "learning_rate": 1.7917174752697636e-06, "loss": 0.1823, "step": 28336 }, { "epoch": 176.00623052959503, "grad_norm": 2.03125, "learning_rate": 1.790801674598186e-06, "loss": 0.1403, "step": 28337 }, { "epoch": 176.01246105919003, "grad_norm": 1.0546875, "learning_rate": 1.7898860993389803e-06, "loss": 0.0993, "step": 28338 }, { "epoch": 176.01869158878506, "grad_norm": 2.359375, "learning_rate": 1.7889707495010454e-06, "loss": 0.2246, "step": 28339 }, { "epoch": 176.02492211838006, "grad_norm": 2.5625, "learning_rate": 1.788055625093263e-06, "loss": 0.2526, "step": 28340 }, { "epoch": 176.0311526479751, "grad_norm": 1.1328125, "learning_rate": 1.7871407261245282e-06, "loss": 0.0914, "step": 28341 }, { "epoch": 176.0373831775701, "grad_norm": 1.734375, "learning_rate": 1.7862260526037179e-06, "loss": 0.1394, "step": 28342 }, { "epoch": 176.04361370716512, "grad_norm": 1.328125, "learning_rate": 1.7853116045397246e-06, "loss": 0.0967, "step": 28343 }, { "epoch": 176.04984423676012, "grad_norm": 1.75, "learning_rate": 1.7843973819414218e-06, "loss": 0.1422, "step": 28344 }, { "epoch": 176.05607476635515, "grad_norm": 1.5078125, "learning_rate": 1.783483384817694e-06, "loss": 0.1311, "step": 28345 }, { "epoch": 176.06230529595015, "grad_norm": 2.453125, "learning_rate": 1.7825696131774177e-06, "loss": 0.2283, "step": 28346 }, { "epoch": 176.06853582554518, "grad_norm": 2.546875, "learning_rate": 1.7816560670294691e-06, "loss": 0.3034, "step": 28347 }, { "epoch": 176.07476635514018, "grad_norm": 2.796875, "learning_rate": 1.780742746382713e-06, "loss": 0.2335, "step": 28348 }, { "epoch": 176.0809968847352, "grad_norm": 2.078125, "learning_rate": 1.7798296512460232e-06, "loss": 0.1542, "step": 28349 }, { "epoch": 176.08722741433021, "grad_norm": 2.203125, "learning_rate": 1.778916781628276e-06, "loss": 0.1766, "step": 28350 }, { "epoch": 176.09345794392524, "grad_norm": 3.015625, "learning_rate": 1.778004137538325e-06, "loss": 0.2339, "step": 28351 }, { "epoch": 176.09968847352025, "grad_norm": 1.4296875, "learning_rate": 1.7770917189850412e-06, "loss": 0.0951, "step": 28352 }, { "epoch": 176.10591900311528, "grad_norm": 1.984375, "learning_rate": 1.7761795259772867e-06, "loss": 0.153, "step": 28353 }, { "epoch": 176.11214953271028, "grad_norm": 1.21875, "learning_rate": 1.7752675585239215e-06, "loss": 0.1225, "step": 28354 }, { "epoch": 176.1183800623053, "grad_norm": 1.078125, "learning_rate": 1.7743558166337936e-06, "loss": 0.0968, "step": 28355 }, { "epoch": 176.1246105919003, "grad_norm": 1.640625, "learning_rate": 1.7734443003157658e-06, "loss": 0.2261, "step": 28356 }, { "epoch": 176.13084112149534, "grad_norm": 1.7265625, "learning_rate": 1.7725330095786946e-06, "loss": 0.2098, "step": 28357 }, { "epoch": 176.13707165109034, "grad_norm": 3.46875, "learning_rate": 1.7716219444314202e-06, "loss": 0.3335, "step": 28358 }, { "epoch": 176.14330218068537, "grad_norm": 2.125, "learning_rate": 1.7707111048827996e-06, "loss": 0.2095, "step": 28359 }, { "epoch": 176.14953271028037, "grad_norm": 1.1640625, "learning_rate": 1.7698004909416781e-06, "loss": 0.0866, "step": 28360 }, { "epoch": 176.1557632398754, "grad_norm": 2.421875, "learning_rate": 1.7688901026168992e-06, "loss": 0.2347, "step": 28361 }, { "epoch": 176.1619937694704, "grad_norm": 1.6796875, "learning_rate": 1.767979939917297e-06, "loss": 0.12, "step": 28362 }, { "epoch": 176.16822429906543, "grad_norm": 3.140625, "learning_rate": 1.76707000285172e-06, "loss": 0.3162, "step": 28363 }, { "epoch": 176.17445482866043, "grad_norm": 2.0625, "learning_rate": 1.7661602914290059e-06, "loss": 0.221, "step": 28364 }, { "epoch": 176.18068535825546, "grad_norm": 2.09375, "learning_rate": 1.7652508056579837e-06, "loss": 0.1603, "step": 28365 }, { "epoch": 176.18691588785046, "grad_norm": 2.171875, "learning_rate": 1.764341545547496e-06, "loss": 0.1788, "step": 28366 }, { "epoch": 176.1931464174455, "grad_norm": 1.4140625, "learning_rate": 1.7634325111063609e-06, "loss": 0.1465, "step": 28367 }, { "epoch": 176.1993769470405, "grad_norm": 1.7265625, "learning_rate": 1.7625237023434215e-06, "loss": 0.139, "step": 28368 }, { "epoch": 176.20560747663552, "grad_norm": 1.03125, "learning_rate": 1.7616151192674901e-06, "loss": 0.1123, "step": 28369 }, { "epoch": 176.21183800623052, "grad_norm": 1.453125, "learning_rate": 1.7607067618873985e-06, "loss": 0.1118, "step": 28370 }, { "epoch": 176.21806853582555, "grad_norm": 1.3515625, "learning_rate": 1.7597986302119702e-06, "loss": 0.1286, "step": 28371 }, { "epoch": 176.22429906542055, "grad_norm": 2.921875, "learning_rate": 1.7588907242500257e-06, "loss": 0.2923, "step": 28372 }, { "epoch": 176.23052959501558, "grad_norm": 2.015625, "learning_rate": 1.7579830440103807e-06, "loss": 0.1811, "step": 28373 }, { "epoch": 176.23676012461058, "grad_norm": 2.140625, "learning_rate": 1.7570755895018443e-06, "loss": 0.1867, "step": 28374 }, { "epoch": 176.2429906542056, "grad_norm": 2.28125, "learning_rate": 1.7561683607332375e-06, "loss": 0.1837, "step": 28375 }, { "epoch": 176.2492211838006, "grad_norm": 1.1484375, "learning_rate": 1.7552613577133752e-06, "loss": 0.1104, "step": 28376 }, { "epoch": 176.25545171339564, "grad_norm": 2.78125, "learning_rate": 1.7543545804510563e-06, "loss": 0.2417, "step": 28377 }, { "epoch": 176.26168224299064, "grad_norm": 0.8671875, "learning_rate": 1.7534480289550904e-06, "loss": 0.0984, "step": 28378 }, { "epoch": 176.26791277258567, "grad_norm": 1.421875, "learning_rate": 1.7525417032342895e-06, "loss": 0.0947, "step": 28379 }, { "epoch": 176.27414330218068, "grad_norm": 1.5546875, "learning_rate": 1.7516356032974523e-06, "loss": 0.1544, "step": 28380 }, { "epoch": 176.2803738317757, "grad_norm": 1.0859375, "learning_rate": 1.750729729153372e-06, "loss": 0.1203, "step": 28381 }, { "epoch": 176.2866043613707, "grad_norm": 1.7265625, "learning_rate": 1.7498240808108524e-06, "loss": 0.1463, "step": 28382 }, { "epoch": 176.29283489096574, "grad_norm": 1.2109375, "learning_rate": 1.748918658278692e-06, "loss": 0.1217, "step": 28383 }, { "epoch": 176.29906542056074, "grad_norm": 2.28125, "learning_rate": 1.7480134615656785e-06, "loss": 0.1204, "step": 28384 }, { "epoch": 176.30529595015577, "grad_norm": 1.359375, "learning_rate": 1.74710849068061e-06, "loss": 0.1082, "step": 28385 }, { "epoch": 176.31152647975077, "grad_norm": 3.078125, "learning_rate": 1.746203745632266e-06, "loss": 0.2974, "step": 28386 }, { "epoch": 176.3177570093458, "grad_norm": 2.109375, "learning_rate": 1.7452992264294476e-06, "loss": 0.2279, "step": 28387 }, { "epoch": 176.3239875389408, "grad_norm": 1.3515625, "learning_rate": 1.7443949330809257e-06, "loss": 0.1288, "step": 28388 }, { "epoch": 176.33021806853583, "grad_norm": 1.8359375, "learning_rate": 1.7434908655954874e-06, "loss": 0.1238, "step": 28389 }, { "epoch": 176.33644859813083, "grad_norm": 2.484375, "learning_rate": 1.7425870239819203e-06, "loss": 0.3209, "step": 28390 }, { "epoch": 176.34267912772586, "grad_norm": 2.0, "learning_rate": 1.7416834082489925e-06, "loss": 0.1416, "step": 28391 }, { "epoch": 176.34890965732086, "grad_norm": 1.421875, "learning_rate": 1.7407800184054885e-06, "loss": 0.1259, "step": 28392 }, { "epoch": 176.3551401869159, "grad_norm": 2.265625, "learning_rate": 1.7398768544601762e-06, "loss": 0.1611, "step": 28393 }, { "epoch": 176.3613707165109, "grad_norm": 1.3515625, "learning_rate": 1.7389739164218294e-06, "loss": 0.161, "step": 28394 }, { "epoch": 176.36760124610592, "grad_norm": 1.140625, "learning_rate": 1.7380712042992187e-06, "loss": 0.1134, "step": 28395 }, { "epoch": 176.37383177570092, "grad_norm": 1.5234375, "learning_rate": 1.7371687181011064e-06, "loss": 0.1461, "step": 28396 }, { "epoch": 176.38006230529595, "grad_norm": 1.96875, "learning_rate": 1.7362664578362693e-06, "loss": 0.1374, "step": 28397 }, { "epoch": 176.38629283489095, "grad_norm": 1.4375, "learning_rate": 1.735364423513458e-06, "loss": 0.0948, "step": 28398 }, { "epoch": 176.39252336448598, "grad_norm": 2.0, "learning_rate": 1.7344626151414383e-06, "loss": 0.1671, "step": 28399 }, { "epoch": 176.39875389408098, "grad_norm": 0.9921875, "learning_rate": 1.733561032728967e-06, "loss": 0.0859, "step": 28400 }, { "epoch": 176.404984423676, "grad_norm": 1.28125, "learning_rate": 1.7326596762848064e-06, "loss": 0.1222, "step": 28401 }, { "epoch": 176.41121495327104, "grad_norm": 1.21875, "learning_rate": 1.7317585458176993e-06, "loss": 0.1105, "step": 28402 }, { "epoch": 176.41744548286604, "grad_norm": 1.5078125, "learning_rate": 1.730857641336406e-06, "loss": 0.1197, "step": 28403 }, { "epoch": 176.42367601246107, "grad_norm": 1.875, "learning_rate": 1.729956962849677e-06, "loss": 0.1952, "step": 28404 }, { "epoch": 176.42990654205607, "grad_norm": 2.421875, "learning_rate": 1.7290565103662614e-06, "loss": 0.2032, "step": 28405 }, { "epoch": 176.4361370716511, "grad_norm": 1.8359375, "learning_rate": 1.7281562838948966e-06, "loss": 0.1305, "step": 28406 }, { "epoch": 176.4423676012461, "grad_norm": 2.484375, "learning_rate": 1.7272562834443284e-06, "loss": 0.2916, "step": 28407 }, { "epoch": 176.44859813084113, "grad_norm": 1.5546875, "learning_rate": 1.7263565090232998e-06, "loss": 0.1324, "step": 28408 }, { "epoch": 176.45482866043614, "grad_norm": 2.625, "learning_rate": 1.7254569606405506e-06, "loss": 0.2043, "step": 28409 }, { "epoch": 176.46105919003116, "grad_norm": 2.328125, "learning_rate": 1.7245576383048135e-06, "loss": 0.1211, "step": 28410 }, { "epoch": 176.46728971962617, "grad_norm": 2.109375, "learning_rate": 1.7236585420248308e-06, "loss": 0.1782, "step": 28411 }, { "epoch": 176.4735202492212, "grad_norm": 3.125, "learning_rate": 1.7227596718093236e-06, "loss": 0.1485, "step": 28412 }, { "epoch": 176.4797507788162, "grad_norm": 1.1953125, "learning_rate": 1.7218610276670323e-06, "loss": 0.1184, "step": 28413 }, { "epoch": 176.48598130841123, "grad_norm": 2.359375, "learning_rate": 1.7209626096066743e-06, "loss": 0.1839, "step": 28414 }, { "epoch": 176.49221183800623, "grad_norm": 2.09375, "learning_rate": 1.7200644176369824e-06, "loss": 0.206, "step": 28415 }, { "epoch": 176.49844236760126, "grad_norm": 1.8515625, "learning_rate": 1.7191664517666823e-06, "loss": 0.2423, "step": 28416 }, { "epoch": 176.50467289719626, "grad_norm": 2.1875, "learning_rate": 1.7182687120044894e-06, "loss": 0.2203, "step": 28417 }, { "epoch": 176.5109034267913, "grad_norm": 0.953125, "learning_rate": 1.7173711983591246e-06, "loss": 0.1043, "step": 28418 }, { "epoch": 176.5171339563863, "grad_norm": 1.734375, "learning_rate": 1.7164739108393002e-06, "loss": 0.1461, "step": 28419 }, { "epoch": 176.52336448598132, "grad_norm": 1.7734375, "learning_rate": 1.7155768494537428e-06, "loss": 0.1026, "step": 28420 }, { "epoch": 176.52959501557632, "grad_norm": 1.53125, "learning_rate": 1.7146800142111535e-06, "loss": 0.1147, "step": 28421 }, { "epoch": 176.53582554517135, "grad_norm": 0.84765625, "learning_rate": 1.713783405120245e-06, "loss": 0.0955, "step": 28422 }, { "epoch": 176.54205607476635, "grad_norm": 3.125, "learning_rate": 1.7128870221897325e-06, "loss": 0.3676, "step": 28423 }, { "epoch": 176.54828660436138, "grad_norm": 1.3515625, "learning_rate": 1.7119908654283146e-06, "loss": 0.1019, "step": 28424 }, { "epoch": 176.55451713395638, "grad_norm": 1.3984375, "learning_rate": 1.7110949348446924e-06, "loss": 0.1115, "step": 28425 }, { "epoch": 176.5607476635514, "grad_norm": 2.359375, "learning_rate": 1.7101992304475734e-06, "loss": 0.2168, "step": 28426 }, { "epoch": 176.5669781931464, "grad_norm": 1.25, "learning_rate": 1.7093037522456583e-06, "loss": 0.1248, "step": 28427 }, { "epoch": 176.57320872274144, "grad_norm": 1.7890625, "learning_rate": 1.7084085002476353e-06, "loss": 0.1045, "step": 28428 }, { "epoch": 176.57943925233644, "grad_norm": 1.7109375, "learning_rate": 1.707513474462205e-06, "loss": 0.1554, "step": 28429 }, { "epoch": 176.58566978193147, "grad_norm": 1.84375, "learning_rate": 1.7066186748980668e-06, "loss": 0.1559, "step": 28430 }, { "epoch": 176.59190031152647, "grad_norm": 1.1875, "learning_rate": 1.705724101563902e-06, "loss": 0.099, "step": 28431 }, { "epoch": 176.5981308411215, "grad_norm": 1.7265625, "learning_rate": 1.7048297544683983e-06, "loss": 0.1213, "step": 28432 }, { "epoch": 176.6043613707165, "grad_norm": 2.703125, "learning_rate": 1.7039356336202434e-06, "loss": 0.2717, "step": 28433 }, { "epoch": 176.61059190031153, "grad_norm": 1.1796875, "learning_rate": 1.7030417390281216e-06, "loss": 0.1023, "step": 28434 }, { "epoch": 176.61682242990653, "grad_norm": 1.875, "learning_rate": 1.702148070700721e-06, "loss": 0.1866, "step": 28435 }, { "epoch": 176.62305295950156, "grad_norm": 2.90625, "learning_rate": 1.7012546286467174e-06, "loss": 0.256, "step": 28436 }, { "epoch": 176.62928348909657, "grad_norm": 3.359375, "learning_rate": 1.700361412874779e-06, "loss": 0.3539, "step": 28437 }, { "epoch": 176.6355140186916, "grad_norm": 2.0625, "learning_rate": 1.6994684233935882e-06, "loss": 0.1499, "step": 28438 }, { "epoch": 176.6417445482866, "grad_norm": 1.2109375, "learning_rate": 1.6985756602118236e-06, "loss": 0.1086, "step": 28439 }, { "epoch": 176.64797507788163, "grad_norm": 1.4140625, "learning_rate": 1.6976831233381452e-06, "loss": 0.1144, "step": 28440 }, { "epoch": 176.65420560747663, "grad_norm": 1.7890625, "learning_rate": 1.6967908127812265e-06, "loss": 0.111, "step": 28441 }, { "epoch": 176.66043613707166, "grad_norm": 2.859375, "learning_rate": 1.6958987285497358e-06, "loss": 0.1124, "step": 28442 }, { "epoch": 176.66666666666666, "grad_norm": 1.7265625, "learning_rate": 1.6950068706523382e-06, "loss": 0.1437, "step": 28443 }, { "epoch": 176.6728971962617, "grad_norm": 2.09375, "learning_rate": 1.6941152390976878e-06, "loss": 0.1611, "step": 28444 }, { "epoch": 176.6791277258567, "grad_norm": 2.390625, "learning_rate": 1.6932238338944472e-06, "loss": 0.2707, "step": 28445 }, { "epoch": 176.68535825545172, "grad_norm": 2.09375, "learning_rate": 1.6923326550512792e-06, "loss": 0.17, "step": 28446 }, { "epoch": 176.69158878504672, "grad_norm": 1.8125, "learning_rate": 1.6914417025768348e-06, "loss": 0.1265, "step": 28447 }, { "epoch": 176.69781931464175, "grad_norm": 2.015625, "learning_rate": 1.6905509764797656e-06, "loss": 0.1479, "step": 28448 }, { "epoch": 176.70404984423675, "grad_norm": 1.609375, "learning_rate": 1.6896604767687284e-06, "loss": 0.1009, "step": 28449 }, { "epoch": 176.71028037383178, "grad_norm": 1.90625, "learning_rate": 1.6887702034523694e-06, "loss": 0.1808, "step": 28450 }, { "epoch": 176.71651090342678, "grad_norm": 1.1875, "learning_rate": 1.6878801565393287e-06, "loss": 0.1101, "step": 28451 }, { "epoch": 176.7227414330218, "grad_norm": 1.96875, "learning_rate": 1.6869903360382577e-06, "loss": 0.1595, "step": 28452 }, { "epoch": 176.7289719626168, "grad_norm": 1.2421875, "learning_rate": 1.6861007419577995e-06, "loss": 0.1327, "step": 28453 }, { "epoch": 176.73520249221184, "grad_norm": 2.875, "learning_rate": 1.685211374306589e-06, "loss": 0.2395, "step": 28454 }, { "epoch": 176.74143302180684, "grad_norm": 2.703125, "learning_rate": 1.6843222330932661e-06, "loss": 0.2041, "step": 28455 }, { "epoch": 176.74766355140187, "grad_norm": 2.0625, "learning_rate": 1.6834333183264662e-06, "loss": 0.1608, "step": 28456 }, { "epoch": 176.75389408099687, "grad_norm": 2.203125, "learning_rate": 1.6825446300148261e-06, "loss": 0.1069, "step": 28457 }, { "epoch": 176.7601246105919, "grad_norm": 2.203125, "learning_rate": 1.68165616816697e-06, "loss": 0.1885, "step": 28458 }, { "epoch": 176.7663551401869, "grad_norm": 2.515625, "learning_rate": 1.6807679327915298e-06, "loss": 0.2629, "step": 28459 }, { "epoch": 176.77258566978193, "grad_norm": 2.96875, "learning_rate": 1.6798799238971346e-06, "loss": 0.3628, "step": 28460 }, { "epoch": 176.77881619937693, "grad_norm": 3.109375, "learning_rate": 1.6789921414924054e-06, "loss": 0.2081, "step": 28461 }, { "epoch": 176.78504672897196, "grad_norm": 1.859375, "learning_rate": 1.6781045855859711e-06, "loss": 0.1153, "step": 28462 }, { "epoch": 176.79127725856696, "grad_norm": 1.25, "learning_rate": 1.6772172561864418e-06, "loss": 0.0959, "step": 28463 }, { "epoch": 176.797507788162, "grad_norm": 1.328125, "learning_rate": 1.6763301533024439e-06, "loss": 0.1031, "step": 28464 }, { "epoch": 176.80373831775702, "grad_norm": 1.1484375, "learning_rate": 1.675443276942587e-06, "loss": 0.1065, "step": 28465 }, { "epoch": 176.80996884735202, "grad_norm": 1.6953125, "learning_rate": 1.6745566271154895e-06, "loss": 0.1364, "step": 28466 }, { "epoch": 176.81619937694705, "grad_norm": 1.296875, "learning_rate": 1.6736702038297581e-06, "loss": 0.1288, "step": 28467 }, { "epoch": 176.82242990654206, "grad_norm": 1.0859375, "learning_rate": 1.672784007094011e-06, "loss": 0.1096, "step": 28468 }, { "epoch": 176.82866043613708, "grad_norm": 1.96875, "learning_rate": 1.671898036916844e-06, "loss": 0.2311, "step": 28469 }, { "epoch": 176.8348909657321, "grad_norm": 1.7734375, "learning_rate": 1.6710122933068646e-06, "loss": 0.1213, "step": 28470 }, { "epoch": 176.84112149532712, "grad_norm": 2.5625, "learning_rate": 1.670126776272679e-06, "loss": 0.1741, "step": 28471 }, { "epoch": 176.84735202492212, "grad_norm": 3.140625, "learning_rate": 1.6692414858228867e-06, "loss": 0.2353, "step": 28472 }, { "epoch": 176.85358255451715, "grad_norm": 2.171875, "learning_rate": 1.668356421966083e-06, "loss": 0.149, "step": 28473 }, { "epoch": 176.85981308411215, "grad_norm": 1.84375, "learning_rate": 1.6674715847108641e-06, "loss": 0.1482, "step": 28474 }, { "epoch": 176.86604361370718, "grad_norm": 2.671875, "learning_rate": 1.6665869740658312e-06, "loss": 0.1186, "step": 28475 }, { "epoch": 176.87227414330218, "grad_norm": 1.75, "learning_rate": 1.6657025900395694e-06, "loss": 0.1578, "step": 28476 }, { "epoch": 176.8785046728972, "grad_norm": 1.7734375, "learning_rate": 1.6648184326406634e-06, "loss": 0.1426, "step": 28477 }, { "epoch": 176.8847352024922, "grad_norm": 1.9609375, "learning_rate": 1.6639345018777063e-06, "loss": 0.1286, "step": 28478 }, { "epoch": 176.89096573208724, "grad_norm": 2.3125, "learning_rate": 1.6630507977592858e-06, "loss": 0.2003, "step": 28479 }, { "epoch": 176.89719626168224, "grad_norm": 1.8046875, "learning_rate": 1.662167320293978e-06, "loss": 0.1175, "step": 28480 }, { "epoch": 176.90342679127727, "grad_norm": 2.25, "learning_rate": 1.6612840694903681e-06, "loss": 0.265, "step": 28481 }, { "epoch": 176.90965732087227, "grad_norm": 1.8515625, "learning_rate": 1.6604010453570296e-06, "loss": 0.1551, "step": 28482 }, { "epoch": 176.9158878504673, "grad_norm": 1.4765625, "learning_rate": 1.6595182479025472e-06, "loss": 0.1083, "step": 28483 }, { "epoch": 176.9221183800623, "grad_norm": 1.3984375, "learning_rate": 1.6586356771354838e-06, "loss": 0.1316, "step": 28484 }, { "epoch": 176.92834890965733, "grad_norm": 1.9921875, "learning_rate": 1.6577533330644156e-06, "loss": 0.1393, "step": 28485 }, { "epoch": 176.93457943925233, "grad_norm": 1.7421875, "learning_rate": 1.6568712156979167e-06, "loss": 0.1107, "step": 28486 }, { "epoch": 176.94080996884736, "grad_norm": 2.5625, "learning_rate": 1.6559893250445518e-06, "loss": 0.1941, "step": 28487 }, { "epoch": 176.94704049844236, "grad_norm": 2.296875, "learning_rate": 1.6551076611128813e-06, "loss": 0.1974, "step": 28488 }, { "epoch": 176.9532710280374, "grad_norm": 2.734375, "learning_rate": 1.654226223911473e-06, "loss": 0.2623, "step": 28489 }, { "epoch": 176.9595015576324, "grad_norm": 1.0625, "learning_rate": 1.6533450134488898e-06, "loss": 0.0952, "step": 28490 }, { "epoch": 176.96573208722742, "grad_norm": 2.296875, "learning_rate": 1.6524640297336802e-06, "loss": 0.2656, "step": 28491 }, { "epoch": 176.97196261682242, "grad_norm": 1.5078125, "learning_rate": 1.6515832727744097e-06, "loss": 0.1047, "step": 28492 }, { "epoch": 176.97819314641745, "grad_norm": 3.171875, "learning_rate": 1.6507027425796324e-06, "loss": 0.2936, "step": 28493 }, { "epoch": 176.98442367601245, "grad_norm": 1.203125, "learning_rate": 1.6498224391579e-06, "loss": 0.1215, "step": 28494 }, { "epoch": 176.99065420560748, "grad_norm": 1.6328125, "learning_rate": 1.6489423625177524e-06, "loss": 0.1204, "step": 28495 }, { "epoch": 176.99688473520249, "grad_norm": 2.21875, "learning_rate": 1.6480625126677446e-06, "loss": 0.189, "step": 28496 }, { "epoch": 177.0, "grad_norm": 0.9453125, "learning_rate": 1.6471828896164275e-06, "loss": 0.0733, "step": 28497 }, { "epoch": 177.00623052959503, "grad_norm": 2.21875, "learning_rate": 1.6463034933723337e-06, "loss": 0.1358, "step": 28498 }, { "epoch": 177.01246105919003, "grad_norm": 1.9375, "learning_rate": 1.6454243239440087e-06, "loss": 0.1706, "step": 28499 }, { "epoch": 177.01869158878506, "grad_norm": 1.390625, "learning_rate": 1.644545381339993e-06, "loss": 0.1397, "step": 28500 }, { "epoch": 177.02492211838006, "grad_norm": 1.609375, "learning_rate": 1.6436666655688243e-06, "loss": 0.1331, "step": 28501 }, { "epoch": 177.0311526479751, "grad_norm": 1.8125, "learning_rate": 1.642788176639029e-06, "loss": 0.1835, "step": 28502 }, { "epoch": 177.0373831775701, "grad_norm": 1.1953125, "learning_rate": 1.6419099145591422e-06, "loss": 0.0962, "step": 28503 }, { "epoch": 177.04361370716512, "grad_norm": 1.1875, "learning_rate": 1.6410318793376982e-06, "loss": 0.1254, "step": 28504 }, { "epoch": 177.04984423676012, "grad_norm": 1.8515625, "learning_rate": 1.6401540709832242e-06, "loss": 0.1826, "step": 28505 }, { "epoch": 177.05607476635515, "grad_norm": 1.2421875, "learning_rate": 1.6392764895042434e-06, "loss": 0.1037, "step": 28506 }, { "epoch": 177.06230529595015, "grad_norm": 1.8046875, "learning_rate": 1.6383991349092743e-06, "loss": 0.212, "step": 28507 }, { "epoch": 177.06853582554518, "grad_norm": 2.203125, "learning_rate": 1.6375220072068432e-06, "loss": 0.2097, "step": 28508 }, { "epoch": 177.07476635514018, "grad_norm": 2.03125, "learning_rate": 1.6366451064054739e-06, "loss": 0.1462, "step": 28509 }, { "epoch": 177.0809968847352, "grad_norm": 1.9296875, "learning_rate": 1.6357684325136735e-06, "loss": 0.1804, "step": 28510 }, { "epoch": 177.08722741433021, "grad_norm": 2.484375, "learning_rate": 1.6348919855399603e-06, "loss": 0.1977, "step": 28511 }, { "epoch": 177.09345794392524, "grad_norm": 1.7421875, "learning_rate": 1.6340157654928496e-06, "loss": 0.0968, "step": 28512 }, { "epoch": 177.09968847352025, "grad_norm": 2.59375, "learning_rate": 1.6331397723808484e-06, "loss": 0.1911, "step": 28513 }, { "epoch": 177.10591900311528, "grad_norm": 3.21875, "learning_rate": 1.6322640062124612e-06, "loss": 0.2827, "step": 28514 }, { "epoch": 177.11214953271028, "grad_norm": 3.0, "learning_rate": 1.6313884669961976e-06, "loss": 0.2344, "step": 28515 }, { "epoch": 177.1183800623053, "grad_norm": 2.28125, "learning_rate": 1.6305131547405621e-06, "loss": 0.2812, "step": 28516 }, { "epoch": 177.1246105919003, "grad_norm": 1.2734375, "learning_rate": 1.6296380694540537e-06, "loss": 0.1086, "step": 28517 }, { "epoch": 177.13084112149534, "grad_norm": 3.171875, "learning_rate": 1.6287632111451707e-06, "loss": 0.2634, "step": 28518 }, { "epoch": 177.13707165109034, "grad_norm": 1.359375, "learning_rate": 1.6278885798224147e-06, "loss": 0.1323, "step": 28519 }, { "epoch": 177.14330218068537, "grad_norm": 1.671875, "learning_rate": 1.6270141754942765e-06, "loss": 0.1412, "step": 28520 }, { "epoch": 177.14953271028037, "grad_norm": 1.765625, "learning_rate": 1.626139998169246e-06, "loss": 0.2091, "step": 28521 }, { "epoch": 177.1557632398754, "grad_norm": 2.234375, "learning_rate": 1.6252660478558168e-06, "loss": 0.1882, "step": 28522 }, { "epoch": 177.1619937694704, "grad_norm": 1.53125, "learning_rate": 1.6243923245624792e-06, "loss": 0.1242, "step": 28523 }, { "epoch": 177.16822429906543, "grad_norm": 1.359375, "learning_rate": 1.6235188282977127e-06, "loss": 0.1108, "step": 28524 }, { "epoch": 177.17445482866043, "grad_norm": 1.9140625, "learning_rate": 1.6226455590700074e-06, "loss": 0.1393, "step": 28525 }, { "epoch": 177.18068535825546, "grad_norm": 1.6171875, "learning_rate": 1.6217725168878372e-06, "loss": 0.1093, "step": 28526 }, { "epoch": 177.18691588785046, "grad_norm": 0.828125, "learning_rate": 1.62089970175969e-06, "loss": 0.101, "step": 28527 }, { "epoch": 177.1931464174455, "grad_norm": 1.1953125, "learning_rate": 1.6200271136940337e-06, "loss": 0.0938, "step": 28528 }, { "epoch": 177.1993769470405, "grad_norm": 2.9375, "learning_rate": 1.6191547526993477e-06, "loss": 0.3106, "step": 28529 }, { "epoch": 177.20560747663552, "grad_norm": 2.140625, "learning_rate": 1.618282618784106e-06, "loss": 0.1851, "step": 28530 }, { "epoch": 177.21183800623052, "grad_norm": 1.6328125, "learning_rate": 1.6174107119567794e-06, "loss": 0.1813, "step": 28531 }, { "epoch": 177.21806853582555, "grad_norm": 2.453125, "learning_rate": 1.6165390322258334e-06, "loss": 0.1413, "step": 28532 }, { "epoch": 177.22429906542055, "grad_norm": 2.0, "learning_rate": 1.6156675795997333e-06, "loss": 0.1967, "step": 28533 }, { "epoch": 177.23052959501558, "grad_norm": 2.109375, "learning_rate": 1.614796354086942e-06, "loss": 0.2609, "step": 28534 }, { "epoch": 177.23676012461058, "grad_norm": 2.203125, "learning_rate": 1.6139253556959304e-06, "loss": 0.1341, "step": 28535 }, { "epoch": 177.2429906542056, "grad_norm": 3.234375, "learning_rate": 1.6130545844351419e-06, "loss": 0.1351, "step": 28536 }, { "epoch": 177.2492211838006, "grad_norm": 1.453125, "learning_rate": 1.6121840403130444e-06, "loss": 0.1312, "step": 28537 }, { "epoch": 177.25545171339564, "grad_norm": 2.265625, "learning_rate": 1.6113137233380954e-06, "loss": 0.1728, "step": 28538 }, { "epoch": 177.26168224299064, "grad_norm": 1.8046875, "learning_rate": 1.6104436335187434e-06, "loss": 0.141, "step": 28539 }, { "epoch": 177.26791277258567, "grad_norm": 1.0546875, "learning_rate": 1.6095737708634317e-06, "loss": 0.0923, "step": 28540 }, { "epoch": 177.27414330218068, "grad_norm": 3.046875, "learning_rate": 1.6087041353806177e-06, "loss": 0.3527, "step": 28541 }, { "epoch": 177.2803738317757, "grad_norm": 3.125, "learning_rate": 1.607834727078747e-06, "loss": 0.3354, "step": 28542 }, { "epoch": 177.2866043613707, "grad_norm": 1.90625, "learning_rate": 1.6069655459662576e-06, "loss": 0.153, "step": 28543 }, { "epoch": 177.29283489096574, "grad_norm": 2.03125, "learning_rate": 1.606096592051598e-06, "loss": 0.1792, "step": 28544 }, { "epoch": 177.29906542056074, "grad_norm": 1.71875, "learning_rate": 1.6052278653432062e-06, "loss": 0.1434, "step": 28545 }, { "epoch": 177.30529595015577, "grad_norm": 1.828125, "learning_rate": 1.604359365849517e-06, "loss": 0.163, "step": 28546 }, { "epoch": 177.31152647975077, "grad_norm": 1.5390625, "learning_rate": 1.6034910935789627e-06, "loss": 0.1365, "step": 28547 }, { "epoch": 177.3177570093458, "grad_norm": 1.796875, "learning_rate": 1.6026230485399806e-06, "loss": 0.1102, "step": 28548 }, { "epoch": 177.3239875389408, "grad_norm": 2.3125, "learning_rate": 1.6017552307410033e-06, "loss": 0.1605, "step": 28549 }, { "epoch": 177.33021806853583, "grad_norm": 2.359375, "learning_rate": 1.6008876401904544e-06, "loss": 0.2703, "step": 28550 }, { "epoch": 177.33644859813083, "grad_norm": 1.71875, "learning_rate": 1.6000202768967659e-06, "loss": 0.1478, "step": 28551 }, { "epoch": 177.34267912772586, "grad_norm": 2.5625, "learning_rate": 1.5991531408683535e-06, "loss": 0.1798, "step": 28552 }, { "epoch": 177.34890965732086, "grad_norm": 1.734375, "learning_rate": 1.5982862321136467e-06, "loss": 0.1289, "step": 28553 }, { "epoch": 177.3551401869159, "grad_norm": 1.328125, "learning_rate": 1.5974195506410578e-06, "loss": 0.111, "step": 28554 }, { "epoch": 177.3613707165109, "grad_norm": 1.8203125, "learning_rate": 1.596553096459011e-06, "loss": 0.1407, "step": 28555 }, { "epoch": 177.36760124610592, "grad_norm": 1.703125, "learning_rate": 1.5956868695759186e-06, "loss": 0.132, "step": 28556 }, { "epoch": 177.37383177570092, "grad_norm": 1.9140625, "learning_rate": 1.5948208700001966e-06, "loss": 0.1879, "step": 28557 }, { "epoch": 177.38006230529595, "grad_norm": 0.8828125, "learning_rate": 1.593955097740249e-06, "loss": 0.1211, "step": 28558 }, { "epoch": 177.38629283489095, "grad_norm": 0.87109375, "learning_rate": 1.593089552804486e-06, "loss": 0.0891, "step": 28559 }, { "epoch": 177.39252336448598, "grad_norm": 2.328125, "learning_rate": 1.5922242352013227e-06, "loss": 0.1543, "step": 28560 }, { "epoch": 177.39875389408098, "grad_norm": 1.6640625, "learning_rate": 1.591359144939153e-06, "loss": 0.1347, "step": 28561 }, { "epoch": 177.404984423676, "grad_norm": 1.0078125, "learning_rate": 1.5904942820263807e-06, "loss": 0.1007, "step": 28562 }, { "epoch": 177.41121495327104, "grad_norm": 2.34375, "learning_rate": 1.5896296464714077e-06, "loss": 0.1678, "step": 28563 }, { "epoch": 177.41744548286604, "grad_norm": 1.5078125, "learning_rate": 1.5887652382826384e-06, "loss": 0.1383, "step": 28564 }, { "epoch": 177.42367601246107, "grad_norm": 1.2734375, "learning_rate": 1.587901057468455e-06, "loss": 0.1204, "step": 28565 }, { "epoch": 177.42990654205607, "grad_norm": 2.859375, "learning_rate": 1.587037104037256e-06, "loss": 0.1717, "step": 28566 }, { "epoch": 177.4361370716511, "grad_norm": 1.9140625, "learning_rate": 1.5861733779974326e-06, "loss": 0.1735, "step": 28567 }, { "epoch": 177.4423676012461, "grad_norm": 1.5390625, "learning_rate": 1.5853098793573772e-06, "loss": 0.1019, "step": 28568 }, { "epoch": 177.44859813084113, "grad_norm": 2.65625, "learning_rate": 1.58444660812547e-06, "loss": 0.2613, "step": 28569 }, { "epoch": 177.45482866043614, "grad_norm": 3.09375, "learning_rate": 1.583583564310101e-06, "loss": 0.242, "step": 28570 }, { "epoch": 177.46105919003116, "grad_norm": 1.6953125, "learning_rate": 1.5827207479196443e-06, "loss": 0.1585, "step": 28571 }, { "epoch": 177.46728971962617, "grad_norm": 1.9453125, "learning_rate": 1.5818581589624904e-06, "loss": 0.2301, "step": 28572 }, { "epoch": 177.4735202492212, "grad_norm": 1.5703125, "learning_rate": 1.5809957974470073e-06, "loss": 0.1013, "step": 28573 }, { "epoch": 177.4797507788162, "grad_norm": 2.578125, "learning_rate": 1.5801336633815722e-06, "loss": 0.2387, "step": 28574 }, { "epoch": 177.48598130841123, "grad_norm": 2.859375, "learning_rate": 1.5792717567745668e-06, "loss": 0.156, "step": 28575 }, { "epoch": 177.49221183800623, "grad_norm": 2.359375, "learning_rate": 1.578410077634354e-06, "loss": 0.1778, "step": 28576 }, { "epoch": 177.49844236760126, "grad_norm": 2.015625, "learning_rate": 1.5775486259693024e-06, "loss": 0.1335, "step": 28577 }, { "epoch": 177.50467289719626, "grad_norm": 2.703125, "learning_rate": 1.5766874017877798e-06, "loss": 0.2054, "step": 28578 }, { "epoch": 177.5109034267913, "grad_norm": 1.453125, "learning_rate": 1.5758264050981548e-06, "loss": 0.1337, "step": 28579 }, { "epoch": 177.5171339563863, "grad_norm": 2.90625, "learning_rate": 1.5749656359087816e-06, "loss": 0.2471, "step": 28580 }, { "epoch": 177.52336448598132, "grad_norm": 2.375, "learning_rate": 1.574105094228026e-06, "loss": 0.1634, "step": 28581 }, { "epoch": 177.52959501557632, "grad_norm": 2.71875, "learning_rate": 1.573244780064248e-06, "loss": 0.2189, "step": 28582 }, { "epoch": 177.53582554517135, "grad_norm": 1.8203125, "learning_rate": 1.5723846934258018e-06, "loss": 0.1507, "step": 28583 }, { "epoch": 177.54205607476635, "grad_norm": 2.578125, "learning_rate": 1.571524834321031e-06, "loss": 0.2686, "step": 28584 }, { "epoch": 177.54828660436138, "grad_norm": 1.8046875, "learning_rate": 1.5706652027582953e-06, "loss": 0.1322, "step": 28585 }, { "epoch": 177.55451713395638, "grad_norm": 1.09375, "learning_rate": 1.5698057987459492e-06, "loss": 0.1142, "step": 28586 }, { "epoch": 177.5607476635514, "grad_norm": 1.0234375, "learning_rate": 1.5689466222923249e-06, "loss": 0.116, "step": 28587 }, { "epoch": 177.5669781931464, "grad_norm": 2.1875, "learning_rate": 1.568087673405777e-06, "loss": 0.1446, "step": 28588 }, { "epoch": 177.57320872274144, "grad_norm": 1.7421875, "learning_rate": 1.5672289520946487e-06, "loss": 0.1466, "step": 28589 }, { "epoch": 177.57943925233644, "grad_norm": 1.9765625, "learning_rate": 1.5663704583672779e-06, "loss": 0.2722, "step": 28590 }, { "epoch": 177.58566978193147, "grad_norm": 1.4296875, "learning_rate": 1.5655121922319965e-06, "loss": 0.1139, "step": 28591 }, { "epoch": 177.59190031152647, "grad_norm": 3.421875, "learning_rate": 1.5646541536971454e-06, "loss": 0.3305, "step": 28592 }, { "epoch": 177.5981308411215, "grad_norm": 2.6875, "learning_rate": 1.5637963427710622e-06, "loss": 0.094, "step": 28593 }, { "epoch": 177.6043613707165, "grad_norm": 1.7578125, "learning_rate": 1.5629387594620681e-06, "loss": 0.1448, "step": 28594 }, { "epoch": 177.61059190031153, "grad_norm": 2.546875, "learning_rate": 1.5620814037785036e-06, "loss": 0.3323, "step": 28595 }, { "epoch": 177.61682242990653, "grad_norm": 1.4921875, "learning_rate": 1.561224275728687e-06, "loss": 0.1484, "step": 28596 }, { "epoch": 177.62305295950156, "grad_norm": 1.2265625, "learning_rate": 1.560367375320948e-06, "loss": 0.1081, "step": 28597 }, { "epoch": 177.62928348909657, "grad_norm": 2.5625, "learning_rate": 1.5595107025636019e-06, "loss": 0.1767, "step": 28598 }, { "epoch": 177.6355140186916, "grad_norm": 2.1875, "learning_rate": 1.5586542574649726e-06, "loss": 0.155, "step": 28599 }, { "epoch": 177.6417445482866, "grad_norm": 1.2421875, "learning_rate": 1.5577980400333814e-06, "loss": 0.0912, "step": 28600 }, { "epoch": 177.64797507788163, "grad_norm": 2.3125, "learning_rate": 1.5569420502771436e-06, "loss": 0.1543, "step": 28601 }, { "epoch": 177.65420560747663, "grad_norm": 1.875, "learning_rate": 1.5560862882045723e-06, "loss": 0.1271, "step": 28602 }, { "epoch": 177.66043613707166, "grad_norm": 1.3359375, "learning_rate": 1.5552307538239719e-06, "loss": 0.1045, "step": 28603 }, { "epoch": 177.66666666666666, "grad_norm": 2.921875, "learning_rate": 1.5543754471436577e-06, "loss": 0.2292, "step": 28604 }, { "epoch": 177.6728971962617, "grad_norm": 1.7578125, "learning_rate": 1.55352036817194e-06, "loss": 0.1892, "step": 28605 }, { "epoch": 177.6791277258567, "grad_norm": 1.7890625, "learning_rate": 1.552665516917115e-06, "loss": 0.1618, "step": 28606 }, { "epoch": 177.68535825545172, "grad_norm": 1.1953125, "learning_rate": 1.5518108933874898e-06, "loss": 0.1127, "step": 28607 }, { "epoch": 177.69158878504672, "grad_norm": 1.453125, "learning_rate": 1.5509564975913687e-06, "loss": 0.1013, "step": 28608 }, { "epoch": 177.69781931464175, "grad_norm": 2.625, "learning_rate": 1.5501023295370454e-06, "loss": 0.2169, "step": 28609 }, { "epoch": 177.70404984423675, "grad_norm": 3.34375, "learning_rate": 1.5492483892328102e-06, "loss": 0.3208, "step": 28610 }, { "epoch": 177.71028037383178, "grad_norm": 2.765625, "learning_rate": 1.548394676686965e-06, "loss": 0.2096, "step": 28611 }, { "epoch": 177.71651090342678, "grad_norm": 1.515625, "learning_rate": 1.5475411919078004e-06, "loss": 0.1019, "step": 28612 }, { "epoch": 177.7227414330218, "grad_norm": 1.1640625, "learning_rate": 1.5466879349036012e-06, "loss": 0.1259, "step": 28613 }, { "epoch": 177.7289719626168, "grad_norm": 2.203125, "learning_rate": 1.5458349056826554e-06, "loss": 0.1615, "step": 28614 }, { "epoch": 177.73520249221184, "grad_norm": 1.84375, "learning_rate": 1.5449821042532564e-06, "loss": 0.1294, "step": 28615 }, { "epoch": 177.74143302180684, "grad_norm": 1.6484375, "learning_rate": 1.5441295306236781e-06, "loss": 0.1549, "step": 28616 }, { "epoch": 177.74766355140187, "grad_norm": 1.9453125, "learning_rate": 1.543277184802197e-06, "loss": 0.1357, "step": 28617 }, { "epoch": 177.75389408099687, "grad_norm": 0.8359375, "learning_rate": 1.5424250667970986e-06, "loss": 0.0994, "step": 28618 }, { "epoch": 177.7601246105919, "grad_norm": 1.4453125, "learning_rate": 1.5415731766166591e-06, "loss": 0.1172, "step": 28619 }, { "epoch": 177.7663551401869, "grad_norm": 1.5546875, "learning_rate": 1.5407215142691473e-06, "loss": 0.0808, "step": 28620 }, { "epoch": 177.77258566978193, "grad_norm": 1.375, "learning_rate": 1.5398700797628424e-06, "loss": 0.0943, "step": 28621 }, { "epoch": 177.77881619937693, "grad_norm": 1.7421875, "learning_rate": 1.5390188731060047e-06, "loss": 0.1452, "step": 28622 }, { "epoch": 177.78504672897196, "grad_norm": 2.484375, "learning_rate": 1.5381678943069077e-06, "loss": 0.1229, "step": 28623 }, { "epoch": 177.79127725856696, "grad_norm": 2.84375, "learning_rate": 1.537317143373812e-06, "loss": 0.1995, "step": 28624 }, { "epoch": 177.797507788162, "grad_norm": 1.078125, "learning_rate": 1.536466620314983e-06, "loss": 0.1093, "step": 28625 }, { "epoch": 177.80373831775702, "grad_norm": 2.28125, "learning_rate": 1.535616325138678e-06, "loss": 0.1743, "step": 28626 }, { "epoch": 177.80996884735202, "grad_norm": 1.2421875, "learning_rate": 1.534766257853168e-06, "loss": 0.1265, "step": 28627 }, { "epoch": 177.81619937694705, "grad_norm": 1.515625, "learning_rate": 1.5339164184666881e-06, "loss": 0.1399, "step": 28628 }, { "epoch": 177.82242990654206, "grad_norm": 1.8984375, "learning_rate": 1.5330668069875043e-06, "loss": 0.1094, "step": 28629 }, { "epoch": 177.82866043613708, "grad_norm": 1.7734375, "learning_rate": 1.532217423423868e-06, "loss": 0.1391, "step": 28630 }, { "epoch": 177.8348909657321, "grad_norm": 1.2265625, "learning_rate": 1.531368267784028e-06, "loss": 0.1045, "step": 28631 }, { "epoch": 177.84112149532712, "grad_norm": 1.828125, "learning_rate": 1.530519340076228e-06, "loss": 0.1566, "step": 28632 }, { "epoch": 177.84735202492212, "grad_norm": 2.390625, "learning_rate": 1.5296706403087168e-06, "loss": 0.2498, "step": 28633 }, { "epoch": 177.85358255451715, "grad_norm": 1.171875, "learning_rate": 1.528822168489738e-06, "loss": 0.1405, "step": 28634 }, { "epoch": 177.85981308411215, "grad_norm": 2.359375, "learning_rate": 1.527973924627532e-06, "loss": 0.2228, "step": 28635 }, { "epoch": 177.86604361370718, "grad_norm": 1.5, "learning_rate": 1.5271259087303314e-06, "loss": 0.1236, "step": 28636 }, { "epoch": 177.87227414330218, "grad_norm": 0.77734375, "learning_rate": 1.5262781208063736e-06, "loss": 0.0922, "step": 28637 }, { "epoch": 177.8785046728972, "grad_norm": 2.140625, "learning_rate": 1.5254305608638996e-06, "loss": 0.1939, "step": 28638 }, { "epoch": 177.8847352024922, "grad_norm": 1.59375, "learning_rate": 1.524583228911136e-06, "loss": 0.1143, "step": 28639 }, { "epoch": 177.89096573208724, "grad_norm": 1.859375, "learning_rate": 1.5237361249563126e-06, "loss": 0.2, "step": 28640 }, { "epoch": 177.89719626168224, "grad_norm": 2.046875, "learning_rate": 1.5228892490076558e-06, "loss": 0.2675, "step": 28641 }, { "epoch": 177.90342679127727, "grad_norm": 1.9140625, "learning_rate": 1.5220426010733923e-06, "loss": 0.2008, "step": 28642 }, { "epoch": 177.90965732087227, "grad_norm": 2.421875, "learning_rate": 1.5211961811617436e-06, "loss": 0.2234, "step": 28643 }, { "epoch": 177.9158878504673, "grad_norm": 2.71875, "learning_rate": 1.520349989280928e-06, "loss": 0.1672, "step": 28644 }, { "epoch": 177.9221183800623, "grad_norm": 1.1015625, "learning_rate": 1.5195040254391723e-06, "loss": 0.1026, "step": 28645 }, { "epoch": 177.92834890965733, "grad_norm": 2.078125, "learning_rate": 1.5186582896446867e-06, "loss": 0.1072, "step": 28646 }, { "epoch": 177.93457943925233, "grad_norm": 1.8125, "learning_rate": 1.5178127819056809e-06, "loss": 0.1772, "step": 28647 }, { "epoch": 177.94080996884736, "grad_norm": 1.0859375, "learning_rate": 1.5169675022303736e-06, "loss": 0.1247, "step": 28648 }, { "epoch": 177.94704049844236, "grad_norm": 0.90234375, "learning_rate": 1.516122450626975e-06, "loss": 0.0933, "step": 28649 }, { "epoch": 177.9532710280374, "grad_norm": 0.79296875, "learning_rate": 1.515277627103684e-06, "loss": 0.1019, "step": 28650 }, { "epoch": 177.9595015576324, "grad_norm": 1.7890625, "learning_rate": 1.5144330316687134e-06, "loss": 0.1018, "step": 28651 }, { "epoch": 177.96573208722742, "grad_norm": 1.9765625, "learning_rate": 1.5135886643302676e-06, "loss": 0.1089, "step": 28652 }, { "epoch": 177.97196261682242, "grad_norm": 1.0078125, "learning_rate": 1.5127445250965406e-06, "loss": 0.104, "step": 28653 }, { "epoch": 177.97819314641745, "grad_norm": 1.390625, "learning_rate": 1.5119006139757335e-06, "loss": 0.0989, "step": 28654 }, { "epoch": 177.98442367601245, "grad_norm": 2.28125, "learning_rate": 1.5110569309760404e-06, "loss": 0.1501, "step": 28655 }, { "epoch": 177.99065420560748, "grad_norm": 1.984375, "learning_rate": 1.5102134761056624e-06, "loss": 0.1547, "step": 28656 }, { "epoch": 177.99688473520249, "grad_norm": 1.765625, "learning_rate": 1.5093702493727852e-06, "loss": 0.1773, "step": 28657 }, { "epoch": 178.0, "grad_norm": 1.8125, "learning_rate": 1.5085272507855964e-06, "loss": 0.0891, "step": 28658 }, { "epoch": 178.00623052959503, "grad_norm": 1.5, "learning_rate": 1.5076844803522922e-06, "loss": 0.143, "step": 28659 }, { "epoch": 178.01246105919003, "grad_norm": 2.828125, "learning_rate": 1.5068419380810522e-06, "loss": 0.2786, "step": 28660 }, { "epoch": 178.01869158878506, "grad_norm": 1.6015625, "learning_rate": 1.5059996239800562e-06, "loss": 0.1141, "step": 28661 }, { "epoch": 178.02492211838006, "grad_norm": 1.5625, "learning_rate": 1.5051575380574861e-06, "loss": 0.1366, "step": 28662 }, { "epoch": 178.0311526479751, "grad_norm": 3.875, "learning_rate": 1.5043156803215247e-06, "loss": 0.152, "step": 28663 }, { "epoch": 178.0373831775701, "grad_norm": 1.140625, "learning_rate": 1.5034740507803486e-06, "loss": 0.0975, "step": 28664 }, { "epoch": 178.04361370716512, "grad_norm": 2.5625, "learning_rate": 1.5026326494421262e-06, "loss": 0.1224, "step": 28665 }, { "epoch": 178.04984423676012, "grad_norm": 1.09375, "learning_rate": 1.501791476315037e-06, "loss": 0.1098, "step": 28666 }, { "epoch": 178.05607476635515, "grad_norm": 1.515625, "learning_rate": 1.5009505314072415e-06, "loss": 0.1318, "step": 28667 }, { "epoch": 178.06230529595015, "grad_norm": 1.515625, "learning_rate": 1.5001098147269132e-06, "loss": 0.0939, "step": 28668 }, { "epoch": 178.06853582554518, "grad_norm": 1.4609375, "learning_rate": 1.4992693262822156e-06, "loss": 0.1076, "step": 28669 }, { "epoch": 178.07476635514018, "grad_norm": 2.171875, "learning_rate": 1.498429066081311e-06, "loss": 0.175, "step": 28670 }, { "epoch": 178.0809968847352, "grad_norm": 2.421875, "learning_rate": 1.4975890341323655e-06, "loss": 0.2095, "step": 28671 }, { "epoch": 178.08722741433021, "grad_norm": 1.28125, "learning_rate": 1.4967492304435338e-06, "loss": 0.1158, "step": 28672 }, { "epoch": 178.09345794392524, "grad_norm": 2.15625, "learning_rate": 1.4959096550229645e-06, "loss": 0.143, "step": 28673 }, { "epoch": 178.09968847352025, "grad_norm": 2.765625, "learning_rate": 1.4950703078788236e-06, "loss": 0.1383, "step": 28674 }, { "epoch": 178.10591900311528, "grad_norm": 2.125, "learning_rate": 1.49423118901926e-06, "loss": 0.1575, "step": 28675 }, { "epoch": 178.11214953271028, "grad_norm": 2.90625, "learning_rate": 1.4933922984524174e-06, "loss": 0.1636, "step": 28676 }, { "epoch": 178.1183800623053, "grad_norm": 2.328125, "learning_rate": 1.4925536361864473e-06, "loss": 0.1922, "step": 28677 }, { "epoch": 178.1246105919003, "grad_norm": 1.265625, "learning_rate": 1.4917152022295016e-06, "loss": 0.1457, "step": 28678 }, { "epoch": 178.13084112149534, "grad_norm": 0.953125, "learning_rate": 1.4908769965897157e-06, "loss": 0.0946, "step": 28679 }, { "epoch": 178.13707165109034, "grad_norm": 2.65625, "learning_rate": 1.4900390192752272e-06, "loss": 0.2991, "step": 28680 }, { "epoch": 178.14330218068537, "grad_norm": 1.328125, "learning_rate": 1.4892012702941799e-06, "loss": 0.1115, "step": 28681 }, { "epoch": 178.14953271028037, "grad_norm": 1.578125, "learning_rate": 1.4883637496547142e-06, "loss": 0.1212, "step": 28682 }, { "epoch": 178.1557632398754, "grad_norm": 1.03125, "learning_rate": 1.4875264573649544e-06, "loss": 0.0886, "step": 28683 }, { "epoch": 178.1619937694704, "grad_norm": 2.796875, "learning_rate": 1.4866893934330384e-06, "loss": 0.3238, "step": 28684 }, { "epoch": 178.16822429906543, "grad_norm": 0.8515625, "learning_rate": 1.4858525578671013e-06, "loss": 0.0876, "step": 28685 }, { "epoch": 178.17445482866043, "grad_norm": 3.453125, "learning_rate": 1.4850159506752615e-06, "loss": 0.2971, "step": 28686 }, { "epoch": 178.18068535825546, "grad_norm": 1.453125, "learning_rate": 1.4841795718656431e-06, "loss": 0.1217, "step": 28687 }, { "epoch": 178.18691588785046, "grad_norm": 2.875, "learning_rate": 1.483343421446376e-06, "loss": 0.1042, "step": 28688 }, { "epoch": 178.1931464174455, "grad_norm": 1.75, "learning_rate": 1.482507499425581e-06, "loss": 0.1405, "step": 28689 }, { "epoch": 178.1993769470405, "grad_norm": 2.09375, "learning_rate": 1.4816718058113689e-06, "loss": 0.1647, "step": 28690 }, { "epoch": 178.20560747663552, "grad_norm": 1.5703125, "learning_rate": 1.4808363406118686e-06, "loss": 0.1394, "step": 28691 }, { "epoch": 178.21183800623052, "grad_norm": 2.171875, "learning_rate": 1.4800011038351797e-06, "loss": 0.1982, "step": 28692 }, { "epoch": 178.21806853582555, "grad_norm": 1.546875, "learning_rate": 1.4791660954894287e-06, "loss": 0.1263, "step": 28693 }, { "epoch": 178.22429906542055, "grad_norm": 1.484375, "learning_rate": 1.4783313155827122e-06, "loss": 0.1468, "step": 28694 }, { "epoch": 178.23052959501558, "grad_norm": 2.3125, "learning_rate": 1.4774967641231458e-06, "loss": 0.1541, "step": 28695 }, { "epoch": 178.23676012461058, "grad_norm": 0.75, "learning_rate": 1.4766624411188313e-06, "loss": 0.0954, "step": 28696 }, { "epoch": 178.2429906542056, "grad_norm": 1.1640625, "learning_rate": 1.4758283465778817e-06, "loss": 0.0853, "step": 28697 }, { "epoch": 178.2492211838006, "grad_norm": 1.5703125, "learning_rate": 1.4749944805083821e-06, "loss": 0.1326, "step": 28698 }, { "epoch": 178.25545171339564, "grad_norm": 2.9375, "learning_rate": 1.4741608429184372e-06, "loss": 0.1749, "step": 28699 }, { "epoch": 178.26168224299064, "grad_norm": 1.9453125, "learning_rate": 1.4733274338161463e-06, "loss": 0.1353, "step": 28700 }, { "epoch": 178.26791277258567, "grad_norm": 1.6640625, "learning_rate": 1.4724942532096054e-06, "loss": 0.185, "step": 28701 }, { "epoch": 178.27414330218068, "grad_norm": 1.4765625, "learning_rate": 1.4716613011068997e-06, "loss": 0.0847, "step": 28702 }, { "epoch": 178.2803738317757, "grad_norm": 1.5390625, "learning_rate": 1.470828577516123e-06, "loss": 0.1133, "step": 28703 }, { "epoch": 178.2866043613707, "grad_norm": 2.203125, "learning_rate": 1.4699960824453658e-06, "loss": 0.132, "step": 28704 }, { "epoch": 178.29283489096574, "grad_norm": 1.171875, "learning_rate": 1.4691638159027105e-06, "loss": 0.1289, "step": 28705 }, { "epoch": 178.29906542056074, "grad_norm": 1.5703125, "learning_rate": 1.4683317778962368e-06, "loss": 0.1088, "step": 28706 }, { "epoch": 178.30529595015577, "grad_norm": 1.3671875, "learning_rate": 1.4674999684340273e-06, "loss": 0.1036, "step": 28707 }, { "epoch": 178.31152647975077, "grad_norm": 1.9296875, "learning_rate": 1.4666683875241671e-06, "loss": 0.2009, "step": 28708 }, { "epoch": 178.3177570093458, "grad_norm": 2.28125, "learning_rate": 1.4658370351747247e-06, "loss": 0.2117, "step": 28709 }, { "epoch": 178.3239875389408, "grad_norm": 1.5625, "learning_rate": 1.4650059113937798e-06, "loss": 0.146, "step": 28710 }, { "epoch": 178.33021806853583, "grad_norm": 1.75, "learning_rate": 1.464175016189398e-06, "loss": 0.2215, "step": 28711 }, { "epoch": 178.33644859813083, "grad_norm": 2.015625, "learning_rate": 1.4633443495696592e-06, "loss": 0.1669, "step": 28712 }, { "epoch": 178.34267912772586, "grad_norm": 2.125, "learning_rate": 1.462513911542618e-06, "loss": 0.2035, "step": 28713 }, { "epoch": 178.34890965732086, "grad_norm": 1.2578125, "learning_rate": 1.4616837021163481e-06, "loss": 0.1485, "step": 28714 }, { "epoch": 178.3551401869159, "grad_norm": 1.671875, "learning_rate": 1.460853721298916e-06, "loss": 0.1028, "step": 28715 }, { "epoch": 178.3613707165109, "grad_norm": 3.53125, "learning_rate": 1.4600239690983787e-06, "loss": 0.2839, "step": 28716 }, { "epoch": 178.36760124610592, "grad_norm": 3.4375, "learning_rate": 1.459194445522788e-06, "loss": 0.3039, "step": 28717 }, { "epoch": 178.37383177570092, "grad_norm": 4.5625, "learning_rate": 1.458365150580207e-06, "loss": 0.5326, "step": 28718 }, { "epoch": 178.38006230529595, "grad_norm": 1.4140625, "learning_rate": 1.4575360842786934e-06, "loss": 0.1217, "step": 28719 }, { "epoch": 178.38629283489095, "grad_norm": 2.734375, "learning_rate": 1.4567072466262932e-06, "loss": 0.213, "step": 28720 }, { "epoch": 178.39252336448598, "grad_norm": 1.296875, "learning_rate": 1.4558786376310584e-06, "loss": 0.0918, "step": 28721 }, { "epoch": 178.39875389408098, "grad_norm": 2.046875, "learning_rate": 1.4550502573010354e-06, "loss": 0.2218, "step": 28722 }, { "epoch": 178.404984423676, "grad_norm": 3.046875, "learning_rate": 1.4542221056442762e-06, "loss": 0.2739, "step": 28723 }, { "epoch": 178.41121495327104, "grad_norm": 1.015625, "learning_rate": 1.4533941826688157e-06, "loss": 0.1007, "step": 28724 }, { "epoch": 178.41744548286604, "grad_norm": 3.40625, "learning_rate": 1.452566488382695e-06, "loss": 0.2297, "step": 28725 }, { "epoch": 178.42367601246107, "grad_norm": 2.578125, "learning_rate": 1.4517390227939548e-06, "loss": 0.19, "step": 28726 }, { "epoch": 178.42990654205607, "grad_norm": 2.46875, "learning_rate": 1.4509117859106386e-06, "loss": 0.2107, "step": 28727 }, { "epoch": 178.4361370716511, "grad_norm": 2.09375, "learning_rate": 1.4500847777407678e-06, "loss": 0.1748, "step": 28728 }, { "epoch": 178.4423676012461, "grad_norm": 1.75, "learning_rate": 1.4492579982923887e-06, "loss": 0.1449, "step": 28729 }, { "epoch": 178.44859813084113, "grad_norm": 1.40625, "learning_rate": 1.4484314475735171e-06, "loss": 0.1555, "step": 28730 }, { "epoch": 178.45482866043614, "grad_norm": 1.09375, "learning_rate": 1.4476051255921912e-06, "loss": 0.0972, "step": 28731 }, { "epoch": 178.46105919003116, "grad_norm": 2.875, "learning_rate": 1.4467790323564296e-06, "loss": 0.3051, "step": 28732 }, { "epoch": 178.46728971962617, "grad_norm": 2.453125, "learning_rate": 1.4459531678742588e-06, "loss": 0.125, "step": 28733 }, { "epoch": 178.4735202492212, "grad_norm": 2.5625, "learning_rate": 1.4451275321537006e-06, "loss": 0.2748, "step": 28734 }, { "epoch": 178.4797507788162, "grad_norm": 3.109375, "learning_rate": 1.4443021252027704e-06, "loss": 0.2445, "step": 28735 }, { "epoch": 178.48598130841123, "grad_norm": 2.296875, "learning_rate": 1.44347694702949e-06, "loss": 0.196, "step": 28736 }, { "epoch": 178.49221183800623, "grad_norm": 1.90625, "learning_rate": 1.4426519976418668e-06, "loss": 0.1342, "step": 28737 }, { "epoch": 178.49844236760126, "grad_norm": 1.96875, "learning_rate": 1.441827277047919e-06, "loss": 0.22, "step": 28738 }, { "epoch": 178.50467289719626, "grad_norm": 1.5390625, "learning_rate": 1.4410027852556518e-06, "loss": 0.1102, "step": 28739 }, { "epoch": 178.5109034267913, "grad_norm": 1.78125, "learning_rate": 1.4401785222730724e-06, "loss": 0.1686, "step": 28740 }, { "epoch": 178.5171339563863, "grad_norm": 2.03125, "learning_rate": 1.4393544881081939e-06, "loss": 0.1656, "step": 28741 }, { "epoch": 178.52336448598132, "grad_norm": 1.921875, "learning_rate": 1.4385306827690153e-06, "loss": 0.127, "step": 28742 }, { "epoch": 178.52959501557632, "grad_norm": 2.15625, "learning_rate": 1.4377071062635305e-06, "loss": 0.1267, "step": 28743 }, { "epoch": 178.53582554517135, "grad_norm": 3.328125, "learning_rate": 1.4368837585997469e-06, "loss": 0.2809, "step": 28744 }, { "epoch": 178.54205607476635, "grad_norm": 3.046875, "learning_rate": 1.4360606397856607e-06, "loss": 0.3021, "step": 28745 }, { "epoch": 178.54828660436138, "grad_norm": 1.890625, "learning_rate": 1.4352377498292602e-06, "loss": 0.1626, "step": 28746 }, { "epoch": 178.55451713395638, "grad_norm": 2.65625, "learning_rate": 1.4344150887385417e-06, "loss": 0.1885, "step": 28747 }, { "epoch": 178.5607476635514, "grad_norm": 1.6640625, "learning_rate": 1.4335926565214985e-06, "loss": 0.1159, "step": 28748 }, { "epoch": 178.5669781931464, "grad_norm": 2.28125, "learning_rate": 1.4327704531861136e-06, "loss": 0.1737, "step": 28749 }, { "epoch": 178.57320872274144, "grad_norm": 2.40625, "learning_rate": 1.4319484787403692e-06, "loss": 0.1031, "step": 28750 }, { "epoch": 178.57943925233644, "grad_norm": 2.640625, "learning_rate": 1.4311267331922534e-06, "loss": 0.2052, "step": 28751 }, { "epoch": 178.58566978193147, "grad_norm": 1.875, "learning_rate": 1.430305216549749e-06, "loss": 0.1628, "step": 28752 }, { "epoch": 178.59190031152647, "grad_norm": 1.1640625, "learning_rate": 1.4294839288208268e-06, "loss": 0.0998, "step": 28753 }, { "epoch": 178.5981308411215, "grad_norm": 2.015625, "learning_rate": 1.4286628700134697e-06, "loss": 0.1382, "step": 28754 }, { "epoch": 178.6043613707165, "grad_norm": 2.46875, "learning_rate": 1.4278420401356546e-06, "loss": 0.2119, "step": 28755 }, { "epoch": 178.61059190031153, "grad_norm": 0.83984375, "learning_rate": 1.4270214391953502e-06, "loss": 0.0896, "step": 28756 }, { "epoch": 178.61682242990653, "grad_norm": 1.7109375, "learning_rate": 1.4262010672005193e-06, "loss": 0.1685, "step": 28757 }, { "epoch": 178.62305295950156, "grad_norm": 1.5546875, "learning_rate": 1.4253809241591393e-06, "loss": 0.1342, "step": 28758 }, { "epoch": 178.62928348909657, "grad_norm": 2.796875, "learning_rate": 1.4245610100791702e-06, "loss": 0.2185, "step": 28759 }, { "epoch": 178.6355140186916, "grad_norm": 2.4375, "learning_rate": 1.4237413249685805e-06, "loss": 0.1118, "step": 28760 }, { "epoch": 178.6417445482866, "grad_norm": 1.3125, "learning_rate": 1.4229218688353306e-06, "loss": 0.1152, "step": 28761 }, { "epoch": 178.64797507788163, "grad_norm": 2.4375, "learning_rate": 1.42210264168737e-06, "loss": 0.2127, "step": 28762 }, { "epoch": 178.65420560747663, "grad_norm": 1.7890625, "learning_rate": 1.4212836435326614e-06, "loss": 0.145, "step": 28763 }, { "epoch": 178.66043613707166, "grad_norm": 1.7578125, "learning_rate": 1.4204648743791655e-06, "loss": 0.0983, "step": 28764 }, { "epoch": 178.66666666666666, "grad_norm": 2.578125, "learning_rate": 1.4196463342348226e-06, "loss": 0.3102, "step": 28765 }, { "epoch": 178.6728971962617, "grad_norm": 3.484375, "learning_rate": 1.4188280231075907e-06, "loss": 0.2271, "step": 28766 }, { "epoch": 178.6791277258567, "grad_norm": 3.671875, "learning_rate": 1.418009941005416e-06, "loss": 0.308, "step": 28767 }, { "epoch": 178.68535825545172, "grad_norm": 2.8125, "learning_rate": 1.4171920879362422e-06, "loss": 0.162, "step": 28768 }, { "epoch": 178.69158878504672, "grad_norm": 1.890625, "learning_rate": 1.4163744639080101e-06, "loss": 0.2099, "step": 28769 }, { "epoch": 178.69781931464175, "grad_norm": 2.921875, "learning_rate": 1.4155570689286634e-06, "loss": 0.2545, "step": 28770 }, { "epoch": 178.70404984423675, "grad_norm": 1.2734375, "learning_rate": 1.4147399030061459e-06, "loss": 0.1162, "step": 28771 }, { "epoch": 178.71028037383178, "grad_norm": 2.4375, "learning_rate": 1.4139229661483843e-06, "loss": 0.166, "step": 28772 }, { "epoch": 178.71651090342678, "grad_norm": 2.90625, "learning_rate": 1.413106258363317e-06, "loss": 0.2395, "step": 28773 }, { "epoch": 178.7227414330218, "grad_norm": 1.96875, "learning_rate": 1.4122897796588818e-06, "loss": 0.1752, "step": 28774 }, { "epoch": 178.7289719626168, "grad_norm": 1.578125, "learning_rate": 1.4114735300430033e-06, "loss": 0.1252, "step": 28775 }, { "epoch": 178.73520249221184, "grad_norm": 0.8984375, "learning_rate": 1.4106575095236023e-06, "loss": 0.1023, "step": 28776 }, { "epoch": 178.74143302180684, "grad_norm": 1.2109375, "learning_rate": 1.4098417181086148e-06, "loss": 0.1, "step": 28777 }, { "epoch": 178.74766355140187, "grad_norm": 2.546875, "learning_rate": 1.409026155805962e-06, "loss": 0.216, "step": 28778 }, { "epoch": 178.75389408099687, "grad_norm": 1.7421875, "learning_rate": 1.4082108226235596e-06, "loss": 0.1474, "step": 28779 }, { "epoch": 178.7601246105919, "grad_norm": 2.0, "learning_rate": 1.4073957185693321e-06, "loss": 0.1802, "step": 28780 }, { "epoch": 178.7663551401869, "grad_norm": 1.109375, "learning_rate": 1.4065808436511924e-06, "loss": 0.1221, "step": 28781 }, { "epoch": 178.77258566978193, "grad_norm": 3.234375, "learning_rate": 1.4057661978770592e-06, "loss": 0.1811, "step": 28782 }, { "epoch": 178.77881619937693, "grad_norm": 1.75, "learning_rate": 1.4049517812548347e-06, "loss": 0.1426, "step": 28783 }, { "epoch": 178.78504672897196, "grad_norm": 2.15625, "learning_rate": 1.4041375937924374e-06, "loss": 0.185, "step": 28784 }, { "epoch": 178.79127725856696, "grad_norm": 2.765625, "learning_rate": 1.4033236354977746e-06, "loss": 0.3274, "step": 28785 }, { "epoch": 178.797507788162, "grad_norm": 1.390625, "learning_rate": 1.402509906378749e-06, "loss": 0.1276, "step": 28786 }, { "epoch": 178.80373831775702, "grad_norm": 1.515625, "learning_rate": 1.401696406443262e-06, "loss": 0.0941, "step": 28787 }, { "epoch": 178.80996884735202, "grad_norm": 1.9375, "learning_rate": 1.4008831356992186e-06, "loss": 0.1593, "step": 28788 }, { "epoch": 178.81619937694705, "grad_norm": 1.171875, "learning_rate": 1.4000700941545152e-06, "loss": 0.109, "step": 28789 }, { "epoch": 178.82242990654206, "grad_norm": 3.046875, "learning_rate": 1.3992572818170484e-06, "loss": 0.2029, "step": 28790 }, { "epoch": 178.82866043613708, "grad_norm": 2.765625, "learning_rate": 1.3984446986947093e-06, "loss": 0.1206, "step": 28791 }, { "epoch": 178.8348909657321, "grad_norm": 1.625, "learning_rate": 1.3976323447953965e-06, "loss": 0.1155, "step": 28792 }, { "epoch": 178.84112149532712, "grad_norm": 2.046875, "learning_rate": 1.3968202201270014e-06, "loss": 0.2308, "step": 28793 }, { "epoch": 178.84735202492212, "grad_norm": 1.546875, "learning_rate": 1.396008324697401e-06, "loss": 0.1245, "step": 28794 }, { "epoch": 178.85358255451715, "grad_norm": 1.8828125, "learning_rate": 1.3951966585144833e-06, "loss": 0.255, "step": 28795 }, { "epoch": 178.85981308411215, "grad_norm": 1.75, "learning_rate": 1.3943852215861364e-06, "loss": 0.1459, "step": 28796 }, { "epoch": 178.86604361370718, "grad_norm": 1.4609375, "learning_rate": 1.393574013920243e-06, "loss": 0.1263, "step": 28797 }, { "epoch": 178.87227414330218, "grad_norm": 1.2265625, "learning_rate": 1.3927630355246718e-06, "loss": 0.1205, "step": 28798 }, { "epoch": 178.8785046728972, "grad_norm": 2.71875, "learning_rate": 1.391952286407311e-06, "loss": 0.2602, "step": 28799 }, { "epoch": 178.8847352024922, "grad_norm": 2.078125, "learning_rate": 1.3911417665760235e-06, "loss": 0.1551, "step": 28800 }, { "epoch": 178.89096573208724, "grad_norm": 1.2734375, "learning_rate": 1.3903314760386894e-06, "loss": 0.1259, "step": 28801 }, { "epoch": 178.89719626168224, "grad_norm": 1.5390625, "learning_rate": 1.3895214148031744e-06, "loss": 0.1218, "step": 28802 }, { "epoch": 178.90342679127727, "grad_norm": 1.7109375, "learning_rate": 1.3887115828773445e-06, "loss": 0.1386, "step": 28803 }, { "epoch": 178.90965732087227, "grad_norm": 2.375, "learning_rate": 1.3879019802690713e-06, "loss": 0.143, "step": 28804 }, { "epoch": 178.9158878504673, "grad_norm": 2.34375, "learning_rate": 1.3870926069862122e-06, "loss": 0.2306, "step": 28805 }, { "epoch": 178.9221183800623, "grad_norm": 2.84375, "learning_rate": 1.3862834630366333e-06, "loss": 0.2159, "step": 28806 }, { "epoch": 178.92834890965733, "grad_norm": 1.4765625, "learning_rate": 1.385474548428184e-06, "loss": 0.1306, "step": 28807 }, { "epoch": 178.93457943925233, "grad_norm": 2.59375, "learning_rate": 1.3846658631687326e-06, "loss": 0.2788, "step": 28808 }, { "epoch": 178.94080996884736, "grad_norm": 1.890625, "learning_rate": 1.3838574072661231e-06, "loss": 0.1387, "step": 28809 }, { "epoch": 178.94704049844236, "grad_norm": 2.0, "learning_rate": 1.3830491807282076e-06, "loss": 0.1838, "step": 28810 }, { "epoch": 178.9532710280374, "grad_norm": 2.140625, "learning_rate": 1.3822411835628463e-06, "loss": 0.2029, "step": 28811 }, { "epoch": 178.9595015576324, "grad_norm": 1.9375, "learning_rate": 1.3814334157778801e-06, "loss": 0.258, "step": 28812 }, { "epoch": 178.96573208722742, "grad_norm": 1.078125, "learning_rate": 1.3806258773811476e-06, "loss": 0.1127, "step": 28813 }, { "epoch": 178.97196261682242, "grad_norm": 2.75, "learning_rate": 1.3798185683805005e-06, "loss": 0.1441, "step": 28814 }, { "epoch": 178.97819314641745, "grad_norm": 0.875, "learning_rate": 1.3790114887837797e-06, "loss": 0.0907, "step": 28815 }, { "epoch": 178.98442367601245, "grad_norm": 1.375, "learning_rate": 1.3782046385988179e-06, "loss": 0.1091, "step": 28816 }, { "epoch": 178.99065420560748, "grad_norm": 1.328125, "learning_rate": 1.3773980178334534e-06, "loss": 0.1182, "step": 28817 }, { "epoch": 178.99688473520249, "grad_norm": 2.1875, "learning_rate": 1.376591626495527e-06, "loss": 0.1538, "step": 28818 }, { "epoch": 179.0, "grad_norm": 2.171875, "learning_rate": 1.3757854645928632e-06, "loss": 0.1733, "step": 28819 }, { "epoch": 179.00623052959503, "grad_norm": 1.2109375, "learning_rate": 1.3749795321332887e-06, "loss": 0.1069, "step": 28820 }, { "epoch": 179.01246105919003, "grad_norm": 1.3203125, "learning_rate": 1.3741738291246391e-06, "loss": 0.1175, "step": 28821 }, { "epoch": 179.01869158878506, "grad_norm": 1.8828125, "learning_rate": 1.3733683555747334e-06, "loss": 0.1257, "step": 28822 }, { "epoch": 179.02492211838006, "grad_norm": 1.109375, "learning_rate": 1.3725631114914012e-06, "loss": 0.1072, "step": 28823 }, { "epoch": 179.0311526479751, "grad_norm": 2.984375, "learning_rate": 1.3717580968824556e-06, "loss": 0.247, "step": 28824 }, { "epoch": 179.0373831775701, "grad_norm": 2.4375, "learning_rate": 1.3709533117557211e-06, "loss": 0.1102, "step": 28825 }, { "epoch": 179.04361370716512, "grad_norm": 2.296875, "learning_rate": 1.370148756119008e-06, "loss": 0.1753, "step": 28826 }, { "epoch": 179.04984423676012, "grad_norm": 1.9765625, "learning_rate": 1.3693444299801377e-06, "loss": 0.2077, "step": 28827 }, { "epoch": 179.05607476635515, "grad_norm": 1.96875, "learning_rate": 1.3685403333469155e-06, "loss": 0.1512, "step": 28828 }, { "epoch": 179.06230529595015, "grad_norm": 2.65625, "learning_rate": 1.367736466227154e-06, "loss": 0.1882, "step": 28829 }, { "epoch": 179.06853582554518, "grad_norm": 1.6328125, "learning_rate": 1.3669328286286614e-06, "loss": 0.15, "step": 28830 }, { "epoch": 179.07476635514018, "grad_norm": 2.796875, "learning_rate": 1.3661294205592424e-06, "loss": 0.154, "step": 28831 }, { "epoch": 179.0809968847352, "grad_norm": 2.921875, "learning_rate": 1.3653262420266932e-06, "loss": 0.3535, "step": 28832 }, { "epoch": 179.08722741433021, "grad_norm": 1.265625, "learning_rate": 1.3645232930388218e-06, "loss": 0.0844, "step": 28833 }, { "epoch": 179.09345794392524, "grad_norm": 2.484375, "learning_rate": 1.3637205736034302e-06, "loss": 0.1115, "step": 28834 }, { "epoch": 179.09968847352025, "grad_norm": 1.0625, "learning_rate": 1.362918083728304e-06, "loss": 0.0982, "step": 28835 }, { "epoch": 179.10591900311528, "grad_norm": 1.8671875, "learning_rate": 1.3621158234212421e-06, "loss": 0.2113, "step": 28836 }, { "epoch": 179.11214953271028, "grad_norm": 2.5625, "learning_rate": 1.3613137926900388e-06, "loss": 0.2465, "step": 28837 }, { "epoch": 179.1183800623053, "grad_norm": 1.453125, "learning_rate": 1.360511991542482e-06, "loss": 0.1292, "step": 28838 }, { "epoch": 179.1246105919003, "grad_norm": 0.99609375, "learning_rate": 1.3597104199863543e-06, "loss": 0.1146, "step": 28839 }, { "epoch": 179.13084112149534, "grad_norm": 1.234375, "learning_rate": 1.3589090780294466e-06, "loss": 0.1091, "step": 28840 }, { "epoch": 179.13707165109034, "grad_norm": 2.328125, "learning_rate": 1.358107965679542e-06, "loss": 0.2067, "step": 28841 }, { "epoch": 179.14330218068537, "grad_norm": 2.125, "learning_rate": 1.3573070829444146e-06, "loss": 0.244, "step": 28842 }, { "epoch": 179.14953271028037, "grad_norm": 2.5, "learning_rate": 1.356506429831847e-06, "loss": 0.2507, "step": 28843 }, { "epoch": 179.1557632398754, "grad_norm": 2.28125, "learning_rate": 1.3557060063496192e-06, "loss": 0.1561, "step": 28844 }, { "epoch": 179.1619937694704, "grad_norm": 1.984375, "learning_rate": 1.3549058125055025e-06, "loss": 0.1322, "step": 28845 }, { "epoch": 179.16822429906543, "grad_norm": 2.15625, "learning_rate": 1.3541058483072605e-06, "loss": 0.1975, "step": 28846 }, { "epoch": 179.17445482866043, "grad_norm": 2.21875, "learning_rate": 1.353306113762673e-06, "loss": 0.1824, "step": 28847 }, { "epoch": 179.18068535825546, "grad_norm": 2.90625, "learning_rate": 1.352506608879503e-06, "loss": 0.2618, "step": 28848 }, { "epoch": 179.18691588785046, "grad_norm": 1.4140625, "learning_rate": 1.351707333665514e-06, "loss": 0.1255, "step": 28849 }, { "epoch": 179.1931464174455, "grad_norm": 1.59375, "learning_rate": 1.3509082881284746e-06, "loss": 0.1195, "step": 28850 }, { "epoch": 179.1993769470405, "grad_norm": 3.484375, "learning_rate": 1.350109472276137e-06, "loss": 0.2188, "step": 28851 }, { "epoch": 179.20560747663552, "grad_norm": 1.328125, "learning_rate": 1.3493108861162674e-06, "loss": 0.1308, "step": 28852 }, { "epoch": 179.21183800623052, "grad_norm": 2.484375, "learning_rate": 1.3485125296566148e-06, "loss": 0.2656, "step": 28853 }, { "epoch": 179.21806853582555, "grad_norm": 1.5546875, "learning_rate": 1.3477144029049371e-06, "loss": 0.0905, "step": 28854 }, { "epoch": 179.22429906542055, "grad_norm": 3.109375, "learning_rate": 1.3469165058689836e-06, "loss": 0.2648, "step": 28855 }, { "epoch": 179.23052959501558, "grad_norm": 1.0234375, "learning_rate": 1.3461188385565149e-06, "loss": 0.0947, "step": 28856 }, { "epoch": 179.23676012461058, "grad_norm": 2.453125, "learning_rate": 1.345321400975258e-06, "loss": 0.1335, "step": 28857 }, { "epoch": 179.2429906542056, "grad_norm": 1.703125, "learning_rate": 1.3445241931329706e-06, "loss": 0.1586, "step": 28858 }, { "epoch": 179.2492211838006, "grad_norm": 1.96875, "learning_rate": 1.3437272150373909e-06, "loss": 0.259, "step": 28859 }, { "epoch": 179.25545171339564, "grad_norm": 1.65625, "learning_rate": 1.3429304666962656e-06, "loss": 0.1516, "step": 28860 }, { "epoch": 179.26168224299064, "grad_norm": 2.21875, "learning_rate": 1.3421339481173246e-06, "loss": 0.1609, "step": 28861 }, { "epoch": 179.26791277258567, "grad_norm": 1.109375, "learning_rate": 1.3413376593083089e-06, "loss": 0.1166, "step": 28862 }, { "epoch": 179.27414330218068, "grad_norm": 1.3828125, "learning_rate": 1.3405416002769538e-06, "loss": 0.1158, "step": 28863 }, { "epoch": 179.2803738317757, "grad_norm": 1.015625, "learning_rate": 1.3397457710309897e-06, "loss": 0.1011, "step": 28864 }, { "epoch": 179.2866043613707, "grad_norm": 1.3359375, "learning_rate": 1.3389501715781377e-06, "loss": 0.0949, "step": 28865 }, { "epoch": 179.29283489096574, "grad_norm": 2.21875, "learning_rate": 1.3381548019261335e-06, "loss": 0.112, "step": 28866 }, { "epoch": 179.29906542056074, "grad_norm": 1.7890625, "learning_rate": 1.3373596620827016e-06, "loss": 0.1006, "step": 28867 }, { "epoch": 179.30529595015577, "grad_norm": 0.91015625, "learning_rate": 1.3365647520555608e-06, "loss": 0.0878, "step": 28868 }, { "epoch": 179.31152647975077, "grad_norm": 2.5625, "learning_rate": 1.335770071852438e-06, "loss": 0.2105, "step": 28869 }, { "epoch": 179.3177570093458, "grad_norm": 0.88671875, "learning_rate": 1.3349756214810411e-06, "loss": 0.077, "step": 28870 }, { "epoch": 179.3239875389408, "grad_norm": 1.203125, "learning_rate": 1.3341814009490972e-06, "loss": 0.1135, "step": 28871 }, { "epoch": 179.33021806853583, "grad_norm": 1.515625, "learning_rate": 1.3333874102643085e-06, "loss": 0.1093, "step": 28872 }, { "epoch": 179.33644859813083, "grad_norm": 3.234375, "learning_rate": 1.332593649434391e-06, "loss": 0.2776, "step": 28873 }, { "epoch": 179.34267912772586, "grad_norm": 1.21875, "learning_rate": 1.3318001184670609e-06, "loss": 0.0977, "step": 28874 }, { "epoch": 179.34890965732086, "grad_norm": 1.3984375, "learning_rate": 1.3310068173700146e-06, "loss": 0.1055, "step": 28875 }, { "epoch": 179.3551401869159, "grad_norm": 1.9296875, "learning_rate": 1.3302137461509657e-06, "loss": 0.1646, "step": 28876 }, { "epoch": 179.3613707165109, "grad_norm": 2.578125, "learning_rate": 1.3294209048176104e-06, "loss": 0.1893, "step": 28877 }, { "epoch": 179.36760124610592, "grad_norm": 1.2109375, "learning_rate": 1.3286282933776512e-06, "loss": 0.1029, "step": 28878 }, { "epoch": 179.37383177570092, "grad_norm": 1.4609375, "learning_rate": 1.3278359118387845e-06, "loss": 0.1076, "step": 28879 }, { "epoch": 179.38006230529595, "grad_norm": 3.0, "learning_rate": 1.327043760208707e-06, "loss": 0.3241, "step": 28880 }, { "epoch": 179.38629283489095, "grad_norm": 2.4375, "learning_rate": 1.3262518384951156e-06, "loss": 0.1873, "step": 28881 }, { "epoch": 179.39252336448598, "grad_norm": 1.0625, "learning_rate": 1.3254601467056982e-06, "loss": 0.0941, "step": 28882 }, { "epoch": 179.39875389408098, "grad_norm": 0.984375, "learning_rate": 1.3246686848481403e-06, "loss": 0.1163, "step": 28883 }, { "epoch": 179.404984423676, "grad_norm": 2.890625, "learning_rate": 1.3238774529301363e-06, "loss": 0.2479, "step": 28884 }, { "epoch": 179.41121495327104, "grad_norm": 1.8203125, "learning_rate": 1.3230864509593682e-06, "loss": 0.1824, "step": 28885 }, { "epoch": 179.41744548286604, "grad_norm": 2.3125, "learning_rate": 1.3222956789435136e-06, "loss": 0.1671, "step": 28886 }, { "epoch": 179.42367601246107, "grad_norm": 2.890625, "learning_rate": 1.3215051368902553e-06, "loss": 0.3031, "step": 28887 }, { "epoch": 179.42990654205607, "grad_norm": 2.515625, "learning_rate": 1.3207148248072786e-06, "loss": 0.163, "step": 28888 }, { "epoch": 179.4361370716511, "grad_norm": 1.640625, "learning_rate": 1.3199247427022526e-06, "loss": 0.1466, "step": 28889 }, { "epoch": 179.4423676012461, "grad_norm": 2.171875, "learning_rate": 1.319134890582846e-06, "loss": 0.1856, "step": 28890 }, { "epoch": 179.44859813084113, "grad_norm": 2.34375, "learning_rate": 1.318345268456736e-06, "loss": 0.2161, "step": 28891 }, { "epoch": 179.45482866043614, "grad_norm": 0.95703125, "learning_rate": 1.3175558763315888e-06, "loss": 0.0962, "step": 28892 }, { "epoch": 179.46105919003116, "grad_norm": 2.109375, "learning_rate": 1.3167667142150786e-06, "loss": 0.1699, "step": 28893 }, { "epoch": 179.46728971962617, "grad_norm": 2.328125, "learning_rate": 1.3159777821148605e-06, "loss": 0.1662, "step": 28894 }, { "epoch": 179.4735202492212, "grad_norm": 2.578125, "learning_rate": 1.3151890800386034e-06, "loss": 0.1943, "step": 28895 }, { "epoch": 179.4797507788162, "grad_norm": 2.0625, "learning_rate": 1.3144006079939625e-06, "loss": 0.1778, "step": 28896 }, { "epoch": 179.48598130841123, "grad_norm": 1.9765625, "learning_rate": 1.3136123659885979e-06, "loss": 0.1572, "step": 28897 }, { "epoch": 179.49221183800623, "grad_norm": 1.8984375, "learning_rate": 1.312824354030165e-06, "loss": 0.1552, "step": 28898 }, { "epoch": 179.49844236760126, "grad_norm": 2.234375, "learning_rate": 1.3120365721263157e-06, "loss": 0.1288, "step": 28899 }, { "epoch": 179.50467289719626, "grad_norm": 2.4375, "learning_rate": 1.311249020284705e-06, "loss": 0.1754, "step": 28900 }, { "epoch": 179.5109034267913, "grad_norm": 2.203125, "learning_rate": 1.3104616985129796e-06, "loss": 0.3719, "step": 28901 }, { "epoch": 179.5171339563863, "grad_norm": 2.578125, "learning_rate": 1.309674606818781e-06, "loss": 0.178, "step": 28902 }, { "epoch": 179.52336448598132, "grad_norm": 1.6640625, "learning_rate": 1.3088877452097582e-06, "loss": 0.1317, "step": 28903 }, { "epoch": 179.52959501557632, "grad_norm": 3.40625, "learning_rate": 1.308101113693558e-06, "loss": 0.2553, "step": 28904 }, { "epoch": 179.53582554517135, "grad_norm": 2.1875, "learning_rate": 1.3073147122778128e-06, "loss": 0.1446, "step": 28905 }, { "epoch": 179.54205607476635, "grad_norm": 2.09375, "learning_rate": 1.3065285409701616e-06, "loss": 0.1826, "step": 28906 }, { "epoch": 179.54828660436138, "grad_norm": 1.6328125, "learning_rate": 1.3057425997782453e-06, "loss": 0.1369, "step": 28907 }, { "epoch": 179.55451713395638, "grad_norm": 2.453125, "learning_rate": 1.3049568887096936e-06, "loss": 0.1143, "step": 28908 }, { "epoch": 179.5607476635514, "grad_norm": 2.09375, "learning_rate": 1.304171407772134e-06, "loss": 0.1098, "step": 28909 }, { "epoch": 179.5669781931464, "grad_norm": 3.015625, "learning_rate": 1.3033861569731992e-06, "loss": 0.257, "step": 28910 }, { "epoch": 179.57320872274144, "grad_norm": 2.125, "learning_rate": 1.3026011363205165e-06, "loss": 0.184, "step": 28911 }, { "epoch": 179.57943925233644, "grad_norm": 1.4296875, "learning_rate": 1.3018163458217076e-06, "loss": 0.1153, "step": 28912 }, { "epoch": 179.58566978193147, "grad_norm": 1.578125, "learning_rate": 1.301031785484394e-06, "loss": 0.1468, "step": 28913 }, { "epoch": 179.59190031152647, "grad_norm": 1.734375, "learning_rate": 1.3002474553162003e-06, "loss": 0.123, "step": 28914 }, { "epoch": 179.5981308411215, "grad_norm": 1.6953125, "learning_rate": 1.29946335532474e-06, "loss": 0.1472, "step": 28915 }, { "epoch": 179.6043613707165, "grad_norm": 2.25, "learning_rate": 1.298679485517626e-06, "loss": 0.2058, "step": 28916 }, { "epoch": 179.61059190031153, "grad_norm": 1.25, "learning_rate": 1.2978958459024748e-06, "loss": 0.102, "step": 28917 }, { "epoch": 179.61682242990653, "grad_norm": 1.453125, "learning_rate": 1.2971124364868998e-06, "loss": 0.137, "step": 28918 }, { "epoch": 179.62305295950156, "grad_norm": 1.25, "learning_rate": 1.2963292572785058e-06, "loss": 0.1278, "step": 28919 }, { "epoch": 179.62928348909657, "grad_norm": 2.34375, "learning_rate": 1.2955463082849033e-06, "loss": 0.2074, "step": 28920 }, { "epoch": 179.6355140186916, "grad_norm": 1.9609375, "learning_rate": 1.2947635895136895e-06, "loss": 0.1291, "step": 28921 }, { "epoch": 179.6417445482866, "grad_norm": 1.4453125, "learning_rate": 1.2939811009724718e-06, "loss": 0.1412, "step": 28922 }, { "epoch": 179.64797507788163, "grad_norm": 2.15625, "learning_rate": 1.2931988426688496e-06, "loss": 0.2072, "step": 28923 }, { "epoch": 179.65420560747663, "grad_norm": 0.90625, "learning_rate": 1.2924168146104144e-06, "loss": 0.1124, "step": 28924 }, { "epoch": 179.66043613707166, "grad_norm": 2.234375, "learning_rate": 1.2916350168047681e-06, "loss": 0.1231, "step": 28925 }, { "epoch": 179.66666666666666, "grad_norm": 1.6015625, "learning_rate": 1.2908534492595076e-06, "loss": 0.1248, "step": 28926 }, { "epoch": 179.6728971962617, "grad_norm": 2.203125, "learning_rate": 1.290072111982213e-06, "loss": 0.1134, "step": 28927 }, { "epoch": 179.6791277258567, "grad_norm": 2.296875, "learning_rate": 1.289291004980478e-06, "loss": 0.2181, "step": 28928 }, { "epoch": 179.68535825545172, "grad_norm": 1.8046875, "learning_rate": 1.2885101282618856e-06, "loss": 0.1317, "step": 28929 }, { "epoch": 179.69158878504672, "grad_norm": 1.4765625, "learning_rate": 1.2877294818340268e-06, "loss": 0.1346, "step": 28930 }, { "epoch": 179.69781931464175, "grad_norm": 1.53125, "learning_rate": 1.2869490657044764e-06, "loss": 0.1183, "step": 28931 }, { "epoch": 179.70404984423675, "grad_norm": 1.3359375, "learning_rate": 1.286168879880817e-06, "loss": 0.1438, "step": 28932 }, { "epoch": 179.71028037383178, "grad_norm": 2.078125, "learning_rate": 1.2853889243706314e-06, "loss": 0.2145, "step": 28933 }, { "epoch": 179.71651090342678, "grad_norm": 1.53125, "learning_rate": 1.2846091991814857e-06, "loss": 0.1328, "step": 28934 }, { "epoch": 179.7227414330218, "grad_norm": 1.71875, "learning_rate": 1.2838297043209519e-06, "loss": 0.1169, "step": 28935 }, { "epoch": 179.7289719626168, "grad_norm": 2.953125, "learning_rate": 1.283050439796607e-06, "loss": 0.2831, "step": 28936 }, { "epoch": 179.73520249221184, "grad_norm": 1.171875, "learning_rate": 1.28227140561602e-06, "loss": 0.0996, "step": 28937 }, { "epoch": 179.74143302180684, "grad_norm": 1.6015625, "learning_rate": 1.2814926017867518e-06, "loss": 0.1185, "step": 28938 }, { "epoch": 179.74766355140187, "grad_norm": 1.4375, "learning_rate": 1.280714028316371e-06, "loss": 0.1382, "step": 28939 }, { "epoch": 179.75389408099687, "grad_norm": 1.0078125, "learning_rate": 1.2799356852124329e-06, "loss": 0.0948, "step": 28940 }, { "epoch": 179.7601246105919, "grad_norm": 1.7109375, "learning_rate": 1.2791575724825062e-06, "loss": 0.1229, "step": 28941 }, { "epoch": 179.7663551401869, "grad_norm": 1.140625, "learning_rate": 1.2783796901341378e-06, "loss": 0.1003, "step": 28942 }, { "epoch": 179.77258566978193, "grad_norm": 1.8203125, "learning_rate": 1.2776020381748882e-06, "loss": 0.1447, "step": 28943 }, { "epoch": 179.77881619937693, "grad_norm": 2.15625, "learning_rate": 1.2768246166123126e-06, "loss": 0.1419, "step": 28944 }, { "epoch": 179.78504672897196, "grad_norm": 1.3203125, "learning_rate": 1.2760474254539552e-06, "loss": 0.1069, "step": 28945 }, { "epoch": 179.79127725856696, "grad_norm": 2.078125, "learning_rate": 1.2752704647073732e-06, "loss": 0.1247, "step": 28946 }, { "epoch": 179.797507788162, "grad_norm": 1.2734375, "learning_rate": 1.2744937343801e-06, "loss": 0.1213, "step": 28947 }, { "epoch": 179.80373831775702, "grad_norm": 2.421875, "learning_rate": 1.2737172344796905e-06, "loss": 0.1861, "step": 28948 }, { "epoch": 179.80996884735202, "grad_norm": 2.203125, "learning_rate": 1.2729409650136804e-06, "loss": 0.1551, "step": 28949 }, { "epoch": 179.81619937694705, "grad_norm": 1.390625, "learning_rate": 1.2721649259896106e-06, "loss": 0.0993, "step": 28950 }, { "epoch": 179.82242990654206, "grad_norm": 1.1328125, "learning_rate": 1.2713891174150172e-06, "loss": 0.094, "step": 28951 }, { "epoch": 179.82866043613708, "grad_norm": 2.0, "learning_rate": 1.2706135392974412e-06, "loss": 0.1553, "step": 28952 }, { "epoch": 179.8348909657321, "grad_norm": 1.625, "learning_rate": 1.269838191644404e-06, "loss": 0.0962, "step": 28953 }, { "epoch": 179.84112149532712, "grad_norm": 2.03125, "learning_rate": 1.2690630744634447e-06, "loss": 0.1402, "step": 28954 }, { "epoch": 179.84735202492212, "grad_norm": 2.703125, "learning_rate": 1.2682881877620845e-06, "loss": 0.2265, "step": 28955 }, { "epoch": 179.85358255451715, "grad_norm": 0.90234375, "learning_rate": 1.2675135315478592e-06, "loss": 0.1011, "step": 28956 }, { "epoch": 179.85981308411215, "grad_norm": 1.3984375, "learning_rate": 1.2667391058282823e-06, "loss": 0.1183, "step": 28957 }, { "epoch": 179.86604361370718, "grad_norm": 1.9140625, "learning_rate": 1.265964910610884e-06, "loss": 0.1595, "step": 28958 }, { "epoch": 179.87227414330218, "grad_norm": 2.53125, "learning_rate": 1.2651909459031746e-06, "loss": 0.2406, "step": 28959 }, { "epoch": 179.8785046728972, "grad_norm": 2.53125, "learning_rate": 1.2644172117126761e-06, "loss": 0.2262, "step": 28960 }, { "epoch": 179.8847352024922, "grad_norm": 2.421875, "learning_rate": 1.2636437080469021e-06, "loss": 0.2339, "step": 28961 }, { "epoch": 179.89096573208724, "grad_norm": 3.171875, "learning_rate": 1.262870434913363e-06, "loss": 0.1555, "step": 28962 }, { "epoch": 179.89719626168224, "grad_norm": 0.94140625, "learning_rate": 1.2620973923195777e-06, "loss": 0.1021, "step": 28963 }, { "epoch": 179.90342679127727, "grad_norm": 2.375, "learning_rate": 1.2613245802730406e-06, "loss": 0.1746, "step": 28964 }, { "epoch": 179.90965732087227, "grad_norm": 1.8984375, "learning_rate": 1.2605519987812704e-06, "loss": 0.1088, "step": 28965 }, { "epoch": 179.9158878504673, "grad_norm": 1.3984375, "learning_rate": 1.259779647851761e-06, "loss": 0.1237, "step": 28966 }, { "epoch": 179.9221183800623, "grad_norm": 2.34375, "learning_rate": 1.2590075274920205e-06, "loss": 0.1178, "step": 28967 }, { "epoch": 179.92834890965733, "grad_norm": 2.234375, "learning_rate": 1.25823563770954e-06, "loss": 0.2024, "step": 28968 }, { "epoch": 179.93457943925233, "grad_norm": 2.328125, "learning_rate": 1.257463978511822e-06, "loss": 0.1902, "step": 28969 }, { "epoch": 179.94080996884736, "grad_norm": 2.046875, "learning_rate": 1.256692549906363e-06, "loss": 0.1669, "step": 28970 }, { "epoch": 179.94704049844236, "grad_norm": 2.71875, "learning_rate": 1.2559213519006518e-06, "loss": 0.2278, "step": 28971 }, { "epoch": 179.9532710280374, "grad_norm": 1.84375, "learning_rate": 1.2551503845021767e-06, "loss": 0.1668, "step": 28972 }, { "epoch": 179.9595015576324, "grad_norm": 0.8203125, "learning_rate": 1.254379647718426e-06, "loss": 0.0945, "step": 28973 }, { "epoch": 179.96573208722742, "grad_norm": 2.265625, "learning_rate": 1.2536091415568913e-06, "loss": 0.1478, "step": 28974 }, { "epoch": 179.97196261682242, "grad_norm": 2.125, "learning_rate": 1.2528388660250468e-06, "loss": 0.2221, "step": 28975 }, { "epoch": 179.97819314641745, "grad_norm": 2.359375, "learning_rate": 1.2520688211303782e-06, "loss": 0.1858, "step": 28976 }, { "epoch": 179.98442367601245, "grad_norm": 1.265625, "learning_rate": 1.2512990068803688e-06, "loss": 0.1151, "step": 28977 }, { "epoch": 179.99065420560748, "grad_norm": 1.0390625, "learning_rate": 1.25052942328249e-06, "loss": 0.108, "step": 28978 }, { "epoch": 179.99688473520249, "grad_norm": 2.578125, "learning_rate": 1.2497600703442135e-06, "loss": 0.2051, "step": 28979 }, { "epoch": 180.0, "grad_norm": 1.3671875, "learning_rate": 1.2489909480730145e-06, "loss": 0.0947, "step": 28980 }, { "epoch": 180.00623052959503, "grad_norm": 1.2578125, "learning_rate": 1.248222056476367e-06, "loss": 0.1031, "step": 28981 }, { "epoch": 180.01246105919003, "grad_norm": 2.015625, "learning_rate": 1.2474533955617319e-06, "loss": 0.1693, "step": 28982 }, { "epoch": 180.01869158878506, "grad_norm": 1.140625, "learning_rate": 1.2466849653365781e-06, "loss": 0.1136, "step": 28983 }, { "epoch": 180.02492211838006, "grad_norm": 2.078125, "learning_rate": 1.2459167658083692e-06, "loss": 0.1675, "step": 28984 }, { "epoch": 180.0311526479751, "grad_norm": 2.40625, "learning_rate": 1.245148796984566e-06, "loss": 0.2123, "step": 28985 }, { "epoch": 180.0373831775701, "grad_norm": 2.25, "learning_rate": 1.2443810588726235e-06, "loss": 0.2555, "step": 28986 }, { "epoch": 180.04361370716512, "grad_norm": 2.890625, "learning_rate": 1.2436135514799996e-06, "loss": 0.2956, "step": 28987 }, { "epoch": 180.04984423676012, "grad_norm": 2.296875, "learning_rate": 1.2428462748141523e-06, "loss": 0.1682, "step": 28988 }, { "epoch": 180.05607476635515, "grad_norm": 3.671875, "learning_rate": 1.2420792288825312e-06, "loss": 0.3729, "step": 28989 }, { "epoch": 180.06230529595015, "grad_norm": 2.359375, "learning_rate": 1.2413124136925885e-06, "loss": 0.185, "step": 28990 }, { "epoch": 180.06853582554518, "grad_norm": 2.28125, "learning_rate": 1.240545829251763e-06, "loss": 0.1205, "step": 28991 }, { "epoch": 180.07476635514018, "grad_norm": 3.171875, "learning_rate": 1.239779475567507e-06, "loss": 0.2635, "step": 28992 }, { "epoch": 180.0809968847352, "grad_norm": 0.7265625, "learning_rate": 1.239013352647264e-06, "loss": 0.0944, "step": 28993 }, { "epoch": 180.08722741433021, "grad_norm": 2.25, "learning_rate": 1.2382474604984734e-06, "loss": 0.1639, "step": 28994 }, { "epoch": 180.09345794392524, "grad_norm": 2.5, "learning_rate": 1.2374817991285702e-06, "loss": 0.2598, "step": 28995 }, { "epoch": 180.09968847352025, "grad_norm": 1.25, "learning_rate": 1.236716368544999e-06, "loss": 0.1175, "step": 28996 }, { "epoch": 180.10591900311528, "grad_norm": 1.4921875, "learning_rate": 1.2359511687551866e-06, "loss": 0.1473, "step": 28997 }, { "epoch": 180.11214953271028, "grad_norm": 1.296875, "learning_rate": 1.2351861997665636e-06, "loss": 0.1376, "step": 28998 }, { "epoch": 180.1183800623053, "grad_norm": 0.9921875, "learning_rate": 1.2344214615865657e-06, "loss": 0.0898, "step": 28999 }, { "epoch": 180.1246105919003, "grad_norm": 2.203125, "learning_rate": 1.2336569542226173e-06, "loss": 0.2048, "step": 29000 }, { "epoch": 180.13084112149534, "grad_norm": 1.671875, "learning_rate": 1.2328926776821403e-06, "loss": 0.1454, "step": 29001 }, { "epoch": 180.13707165109034, "grad_norm": 2.59375, "learning_rate": 1.2321286319725622e-06, "loss": 0.198, "step": 29002 }, { "epoch": 180.14330218068537, "grad_norm": 2.046875, "learning_rate": 1.2313648171013049e-06, "loss": 0.1943, "step": 29003 }, { "epoch": 180.14953271028037, "grad_norm": 2.515625, "learning_rate": 1.2306012330757843e-06, "loss": 0.2024, "step": 29004 }, { "epoch": 180.1557632398754, "grad_norm": 1.9140625, "learning_rate": 1.2298378799034116e-06, "loss": 0.135, "step": 29005 }, { "epoch": 180.1619937694704, "grad_norm": 2.171875, "learning_rate": 1.2290747575916055e-06, "loss": 0.2144, "step": 29006 }, { "epoch": 180.16822429906543, "grad_norm": 2.234375, "learning_rate": 1.22831186614778e-06, "loss": 0.1658, "step": 29007 }, { "epoch": 180.17445482866043, "grad_norm": 1.8203125, "learning_rate": 1.2275492055793368e-06, "loss": 0.1341, "step": 29008 }, { "epoch": 180.18068535825546, "grad_norm": 2.359375, "learning_rate": 1.2267867758936929e-06, "loss": 0.1251, "step": 29009 }, { "epoch": 180.18691588785046, "grad_norm": 2.078125, "learning_rate": 1.226024577098245e-06, "loss": 0.1986, "step": 29010 }, { "epoch": 180.1931464174455, "grad_norm": 1.6640625, "learning_rate": 1.2252626092004005e-06, "loss": 0.1319, "step": 29011 }, { "epoch": 180.1993769470405, "grad_norm": 1.828125, "learning_rate": 1.224500872207554e-06, "loss": 0.1677, "step": 29012 }, { "epoch": 180.20560747663552, "grad_norm": 1.453125, "learning_rate": 1.2237393661271107e-06, "loss": 0.1336, "step": 29013 }, { "epoch": 180.21183800623052, "grad_norm": 1.6015625, "learning_rate": 1.2229780909664618e-06, "loss": 0.2439, "step": 29014 }, { "epoch": 180.21806853582555, "grad_norm": 2.171875, "learning_rate": 1.2222170467330068e-06, "loss": 0.1621, "step": 29015 }, { "epoch": 180.22429906542055, "grad_norm": 1.578125, "learning_rate": 1.2214562334341317e-06, "loss": 0.0984, "step": 29016 }, { "epoch": 180.23052959501558, "grad_norm": 2.8125, "learning_rate": 1.2206956510772221e-06, "loss": 0.1939, "step": 29017 }, { "epoch": 180.23676012461058, "grad_norm": 3.390625, "learning_rate": 1.2199352996696723e-06, "loss": 0.3652, "step": 29018 }, { "epoch": 180.2429906542056, "grad_norm": 2.78125, "learning_rate": 1.2191751792188678e-06, "loss": 0.2927, "step": 29019 }, { "epoch": 180.2492211838006, "grad_norm": 1.8515625, "learning_rate": 1.2184152897321833e-06, "loss": 0.1307, "step": 29020 }, { "epoch": 180.25545171339564, "grad_norm": 1.140625, "learning_rate": 1.2176556312170046e-06, "loss": 0.1284, "step": 29021 }, { "epoch": 180.26168224299064, "grad_norm": 0.88671875, "learning_rate": 1.2168962036807092e-06, "loss": 0.0871, "step": 29022 }, { "epoch": 180.26791277258567, "grad_norm": 1.71875, "learning_rate": 1.2161370071306743e-06, "loss": 0.13, "step": 29023 }, { "epoch": 180.27414330218068, "grad_norm": 1.0546875, "learning_rate": 1.2153780415742666e-06, "loss": 0.1018, "step": 29024 }, { "epoch": 180.2803738317757, "grad_norm": 1.96875, "learning_rate": 1.2146193070188632e-06, "loss": 0.1585, "step": 29025 }, { "epoch": 180.2866043613707, "grad_norm": 2.953125, "learning_rate": 1.2138608034718335e-06, "loss": 0.2106, "step": 29026 }, { "epoch": 180.29283489096574, "grad_norm": 1.6953125, "learning_rate": 1.2131025309405409e-06, "loss": 0.1616, "step": 29027 }, { "epoch": 180.29906542056074, "grad_norm": 2.140625, "learning_rate": 1.2123444894323516e-06, "loss": 0.2466, "step": 29028 }, { "epoch": 180.30529595015577, "grad_norm": 1.953125, "learning_rate": 1.2115866789546266e-06, "loss": 0.163, "step": 29029 }, { "epoch": 180.31152647975077, "grad_norm": 2.703125, "learning_rate": 1.2108290995147291e-06, "loss": 0.2073, "step": 29030 }, { "epoch": 180.3177570093458, "grad_norm": 2.0625, "learning_rate": 1.210071751120015e-06, "loss": 0.2616, "step": 29031 }, { "epoch": 180.3239875389408, "grad_norm": 2.75, "learning_rate": 1.2093146337778361e-06, "loss": 0.178, "step": 29032 }, { "epoch": 180.33021806853583, "grad_norm": 2.921875, "learning_rate": 1.2085577474955533e-06, "loss": 0.1952, "step": 29033 }, { "epoch": 180.33644859813083, "grad_norm": 1.6796875, "learning_rate": 1.2078010922805083e-06, "loss": 0.1215, "step": 29034 }, { "epoch": 180.34267912772586, "grad_norm": 1.3125, "learning_rate": 1.2070446681400616e-06, "loss": 0.1324, "step": 29035 }, { "epoch": 180.34890965732086, "grad_norm": 2.34375, "learning_rate": 1.206288475081549e-06, "loss": 0.1883, "step": 29036 }, { "epoch": 180.3551401869159, "grad_norm": 2.4375, "learning_rate": 1.2055325131123202e-06, "loss": 0.2401, "step": 29037 }, { "epoch": 180.3613707165109, "grad_norm": 2.1875, "learning_rate": 1.2047767822397139e-06, "loss": 0.2368, "step": 29038 }, { "epoch": 180.36760124610592, "grad_norm": 1.375, "learning_rate": 1.204021282471074e-06, "loss": 0.0862, "step": 29039 }, { "epoch": 180.37383177570092, "grad_norm": 1.8671875, "learning_rate": 1.2032660138137363e-06, "loss": 0.1097, "step": 29040 }, { "epoch": 180.38006230529595, "grad_norm": 1.9609375, "learning_rate": 1.2025109762750398e-06, "loss": 0.1736, "step": 29041 }, { "epoch": 180.38629283489095, "grad_norm": 1.1953125, "learning_rate": 1.2017561698623058e-06, "loss": 0.1018, "step": 29042 }, { "epoch": 180.39252336448598, "grad_norm": 1.578125, "learning_rate": 1.2010015945828763e-06, "loss": 0.1197, "step": 29043 }, { "epoch": 180.39875389408098, "grad_norm": 2.21875, "learning_rate": 1.200247250444081e-06, "loss": 0.1428, "step": 29044 }, { "epoch": 180.404984423676, "grad_norm": 2.09375, "learning_rate": 1.1994931374532364e-06, "loss": 0.1047, "step": 29045 }, { "epoch": 180.41121495327104, "grad_norm": 2.484375, "learning_rate": 1.1987392556176702e-06, "loss": 0.3801, "step": 29046 }, { "epoch": 180.41744548286604, "grad_norm": 1.5078125, "learning_rate": 1.1979856049447125e-06, "loss": 0.1159, "step": 29047 }, { "epoch": 180.42367601246107, "grad_norm": 2.46875, "learning_rate": 1.1972321854416768e-06, "loss": 0.2125, "step": 29048 }, { "epoch": 180.42990654205607, "grad_norm": 2.265625, "learning_rate": 1.1964789971158742e-06, "loss": 0.2284, "step": 29049 }, { "epoch": 180.4361370716511, "grad_norm": 1.6171875, "learning_rate": 1.1957260399746261e-06, "loss": 0.1347, "step": 29050 }, { "epoch": 180.4423676012461, "grad_norm": 1.90625, "learning_rate": 1.1949733140252466e-06, "loss": 0.1346, "step": 29051 }, { "epoch": 180.44859813084113, "grad_norm": 1.8515625, "learning_rate": 1.1942208192750493e-06, "loss": 0.1717, "step": 29052 }, { "epoch": 180.45482866043614, "grad_norm": 1.7734375, "learning_rate": 1.1934685557313336e-06, "loss": 0.1435, "step": 29053 }, { "epoch": 180.46105919003116, "grad_norm": 3.671875, "learning_rate": 1.192716523401413e-06, "loss": 0.4155, "step": 29054 }, { "epoch": 180.46728971962617, "grad_norm": 2.359375, "learning_rate": 1.191964722292585e-06, "loss": 0.1622, "step": 29055 }, { "epoch": 180.4735202492212, "grad_norm": 1.09375, "learning_rate": 1.1912131524121573e-06, "loss": 0.1023, "step": 29056 }, { "epoch": 180.4797507788162, "grad_norm": 3.328125, "learning_rate": 1.1904618137674239e-06, "loss": 0.1545, "step": 29057 }, { "epoch": 180.48598130841123, "grad_norm": 1.4453125, "learning_rate": 1.1897107063656848e-06, "loss": 0.1092, "step": 29058 }, { "epoch": 180.49221183800623, "grad_norm": 2.546875, "learning_rate": 1.1889598302142396e-06, "loss": 0.2002, "step": 29059 }, { "epoch": 180.49844236760126, "grad_norm": 1.4921875, "learning_rate": 1.188209185320377e-06, "loss": 0.1388, "step": 29060 }, { "epoch": 180.50467289719626, "grad_norm": 1.5390625, "learning_rate": 1.18745877169138e-06, "loss": 0.1175, "step": 29061 }, { "epoch": 180.5109034267913, "grad_norm": 2.421875, "learning_rate": 1.1867085893345453e-06, "loss": 0.2114, "step": 29062 }, { "epoch": 180.5171339563863, "grad_norm": 0.921875, "learning_rate": 1.185958638257162e-06, "loss": 0.0932, "step": 29063 }, { "epoch": 180.52336448598132, "grad_norm": 2.109375, "learning_rate": 1.1852089184665015e-06, "loss": 0.1661, "step": 29064 }, { "epoch": 180.52959501557632, "grad_norm": 1.640625, "learning_rate": 1.1844594299698558e-06, "loss": 0.138, "step": 29065 }, { "epoch": 180.53582554517135, "grad_norm": 0.9765625, "learning_rate": 1.183710172774502e-06, "loss": 0.1001, "step": 29066 }, { "epoch": 180.54205607476635, "grad_norm": 2.40625, "learning_rate": 1.1829611468877177e-06, "loss": 0.1844, "step": 29067 }, { "epoch": 180.54828660436138, "grad_norm": 1.84375, "learning_rate": 1.1822123523167721e-06, "loss": 0.1183, "step": 29068 }, { "epoch": 180.55451713395638, "grad_norm": 1.53125, "learning_rate": 1.18146378906894e-06, "loss": 0.2129, "step": 29069 }, { "epoch": 180.5607476635514, "grad_norm": 3.109375, "learning_rate": 1.1807154571514961e-06, "loss": 0.2657, "step": 29070 }, { "epoch": 180.5669781931464, "grad_norm": 2.65625, "learning_rate": 1.1799673565717012e-06, "loss": 0.1525, "step": 29071 }, { "epoch": 180.57320872274144, "grad_norm": 0.83203125, "learning_rate": 1.1792194873368272e-06, "loss": 0.0738, "step": 29072 }, { "epoch": 180.57943925233644, "grad_norm": 1.7265625, "learning_rate": 1.1784718494541353e-06, "loss": 0.1762, "step": 29073 }, { "epoch": 180.58566978193147, "grad_norm": 1.65625, "learning_rate": 1.1777244429308887e-06, "loss": 0.0992, "step": 29074 }, { "epoch": 180.59190031152647, "grad_norm": 2.46875, "learning_rate": 1.1769772677743428e-06, "loss": 0.1738, "step": 29075 }, { "epoch": 180.5981308411215, "grad_norm": 1.4609375, "learning_rate": 1.1762303239917532e-06, "loss": 0.138, "step": 29076 }, { "epoch": 180.6043613707165, "grad_norm": 3.171875, "learning_rate": 1.1754836115903807e-06, "loss": 0.2416, "step": 29077 }, { "epoch": 180.61059190031153, "grad_norm": 2.421875, "learning_rate": 1.1747371305774691e-06, "loss": 0.1995, "step": 29078 }, { "epoch": 180.61682242990653, "grad_norm": 1.0234375, "learning_rate": 1.1739908809602796e-06, "loss": 0.1016, "step": 29079 }, { "epoch": 180.62305295950156, "grad_norm": 1.484375, "learning_rate": 1.173244862746048e-06, "loss": 0.1369, "step": 29080 }, { "epoch": 180.62928348909657, "grad_norm": 0.81640625, "learning_rate": 1.1724990759420295e-06, "loss": 0.0886, "step": 29081 }, { "epoch": 180.6355140186916, "grad_norm": 1.0546875, "learning_rate": 1.1717535205554574e-06, "loss": 0.0924, "step": 29082 }, { "epoch": 180.6417445482866, "grad_norm": 2.234375, "learning_rate": 1.1710081965935816e-06, "loss": 0.1313, "step": 29083 }, { "epoch": 180.64797507788163, "grad_norm": 1.09375, "learning_rate": 1.1702631040636347e-06, "loss": 0.0857, "step": 29084 }, { "epoch": 180.65420560747663, "grad_norm": 2.640625, "learning_rate": 1.1695182429728613e-06, "loss": 0.1874, "step": 29085 }, { "epoch": 180.66043613707166, "grad_norm": 2.578125, "learning_rate": 1.1687736133284887e-06, "loss": 0.1979, "step": 29086 }, { "epoch": 180.66666666666666, "grad_norm": 1.2578125, "learning_rate": 1.1680292151377475e-06, "loss": 0.1451, "step": 29087 }, { "epoch": 180.6728971962617, "grad_norm": 2.5625, "learning_rate": 1.1672850484078707e-06, "loss": 0.1457, "step": 29088 }, { "epoch": 180.6791277258567, "grad_norm": 1.984375, "learning_rate": 1.1665411131460913e-06, "loss": 0.1524, "step": 29089 }, { "epoch": 180.68535825545172, "grad_norm": 2.40625, "learning_rate": 1.165797409359623e-06, "loss": 0.0827, "step": 29090 }, { "epoch": 180.69158878504672, "grad_norm": 2.09375, "learning_rate": 1.1650539370556934e-06, "loss": 0.1922, "step": 29091 }, { "epoch": 180.69781931464175, "grad_norm": 1.515625, "learning_rate": 1.16431069624153e-06, "loss": 0.1219, "step": 29092 }, { "epoch": 180.70404984423675, "grad_norm": 2.484375, "learning_rate": 1.1635676869243438e-06, "loss": 0.1111, "step": 29093 }, { "epoch": 180.71028037383178, "grad_norm": 3.0625, "learning_rate": 1.1628249091113513e-06, "loss": 0.2431, "step": 29094 }, { "epoch": 180.71651090342678, "grad_norm": 1.7578125, "learning_rate": 1.1620823628097688e-06, "loss": 0.1283, "step": 29095 }, { "epoch": 180.7227414330218, "grad_norm": 2.109375, "learning_rate": 1.1613400480268099e-06, "loss": 0.1102, "step": 29096 }, { "epoch": 180.7289719626168, "grad_norm": 2.90625, "learning_rate": 1.1605979647696802e-06, "loss": 0.2909, "step": 29097 }, { "epoch": 180.73520249221184, "grad_norm": 2.296875, "learning_rate": 1.1598561130455903e-06, "loss": 0.2766, "step": 29098 }, { "epoch": 180.74143302180684, "grad_norm": 1.9765625, "learning_rate": 1.1591144928617375e-06, "loss": 0.119, "step": 29099 }, { "epoch": 180.74766355140187, "grad_norm": 1.3203125, "learning_rate": 1.1583731042253381e-06, "loss": 0.1222, "step": 29100 }, { "epoch": 180.75389408099687, "grad_norm": 2.46875, "learning_rate": 1.1576319471435781e-06, "loss": 0.1882, "step": 29101 }, { "epoch": 180.7601246105919, "grad_norm": 1.6796875, "learning_rate": 1.1568910216236655e-06, "loss": 0.123, "step": 29102 }, { "epoch": 180.7663551401869, "grad_norm": 1.734375, "learning_rate": 1.1561503276727948e-06, "loss": 0.1194, "step": 29103 }, { "epoch": 180.77258566978193, "grad_norm": 1.6484375, "learning_rate": 1.1554098652981572e-06, "loss": 0.0968, "step": 29104 }, { "epoch": 180.77881619937693, "grad_norm": 2.421875, "learning_rate": 1.1546696345069468e-06, "loss": 0.1977, "step": 29105 }, { "epoch": 180.78504672897196, "grad_norm": 1.2890625, "learning_rate": 1.1539296353063472e-06, "loss": 0.1444, "step": 29106 }, { "epoch": 180.79127725856696, "grad_norm": 2.71875, "learning_rate": 1.1531898677035551e-06, "loss": 0.2857, "step": 29107 }, { "epoch": 180.797507788162, "grad_norm": 2.328125, "learning_rate": 1.1524503317057456e-06, "loss": 0.1906, "step": 29108 }, { "epoch": 180.80373831775702, "grad_norm": 1.984375, "learning_rate": 1.151711027320107e-06, "loss": 0.115, "step": 29109 }, { "epoch": 180.80996884735202, "grad_norm": 1.296875, "learning_rate": 1.1509719545538172e-06, "loss": 0.1222, "step": 29110 }, { "epoch": 180.81619937694705, "grad_norm": 2.203125, "learning_rate": 1.1502331134140593e-06, "loss": 0.1162, "step": 29111 }, { "epoch": 180.82242990654206, "grad_norm": 1.8203125, "learning_rate": 1.1494945039080024e-06, "loss": 0.1569, "step": 29112 }, { "epoch": 180.82866043613708, "grad_norm": 1.125, "learning_rate": 1.1487561260428188e-06, "loss": 0.0991, "step": 29113 }, { "epoch": 180.8348909657321, "grad_norm": 2.71875, "learning_rate": 1.1480179798256858e-06, "loss": 0.1397, "step": 29114 }, { "epoch": 180.84112149532712, "grad_norm": 1.8984375, "learning_rate": 1.1472800652637726e-06, "loss": 0.1361, "step": 29115 }, { "epoch": 180.84735202492212, "grad_norm": 1.7578125, "learning_rate": 1.1465423823642406e-06, "loss": 0.143, "step": 29116 }, { "epoch": 180.85358255451715, "grad_norm": 1.34375, "learning_rate": 1.1458049311342616e-06, "loss": 0.1375, "step": 29117 }, { "epoch": 180.85981308411215, "grad_norm": 2.140625, "learning_rate": 1.1450677115809883e-06, "loss": 0.0977, "step": 29118 }, { "epoch": 180.86604361370718, "grad_norm": 1.890625, "learning_rate": 1.1443307237115896e-06, "loss": 0.1648, "step": 29119 }, { "epoch": 180.87227414330218, "grad_norm": 2.390625, "learning_rate": 1.143593967533216e-06, "loss": 0.1529, "step": 29120 }, { "epoch": 180.8785046728972, "grad_norm": 1.8203125, "learning_rate": 1.1428574430530276e-06, "loss": 0.0987, "step": 29121 }, { "epoch": 180.8847352024922, "grad_norm": 2.484375, "learning_rate": 1.1421211502781803e-06, "loss": 0.1865, "step": 29122 }, { "epoch": 180.89096573208724, "grad_norm": 2.671875, "learning_rate": 1.1413850892158157e-06, "loss": 0.2136, "step": 29123 }, { "epoch": 180.89719626168224, "grad_norm": 2.546875, "learning_rate": 1.1406492598730945e-06, "loss": 0.198, "step": 29124 }, { "epoch": 180.90342679127727, "grad_norm": 2.046875, "learning_rate": 1.1399136622571527e-06, "loss": 0.0924, "step": 29125 }, { "epoch": 180.90965732087227, "grad_norm": 1.125, "learning_rate": 1.1391782963751402e-06, "loss": 0.1002, "step": 29126 }, { "epoch": 180.9158878504673, "grad_norm": 1.4765625, "learning_rate": 1.1384431622341958e-06, "loss": 0.1458, "step": 29127 }, { "epoch": 180.9221183800623, "grad_norm": 1.9453125, "learning_rate": 1.1377082598414634e-06, "loss": 0.1061, "step": 29128 }, { "epoch": 180.92834890965733, "grad_norm": 2.09375, "learning_rate": 1.1369735892040796e-06, "loss": 0.1166, "step": 29129 }, { "epoch": 180.93457943925233, "grad_norm": 2.953125, "learning_rate": 1.1362391503291796e-06, "loss": 0.2834, "step": 29130 }, { "epoch": 180.94080996884736, "grad_norm": 2.828125, "learning_rate": 1.135504943223889e-06, "loss": 0.2259, "step": 29131 }, { "epoch": 180.94704049844236, "grad_norm": 0.796875, "learning_rate": 1.134770967895349e-06, "loss": 0.1015, "step": 29132 }, { "epoch": 180.9532710280374, "grad_norm": 2.65625, "learning_rate": 1.1340372243506841e-06, "loss": 0.2864, "step": 29133 }, { "epoch": 180.9595015576324, "grad_norm": 1.265625, "learning_rate": 1.133303712597017e-06, "loss": 0.0954, "step": 29134 }, { "epoch": 180.96573208722742, "grad_norm": 1.3671875, "learning_rate": 1.1325704326414776e-06, "loss": 0.1218, "step": 29135 }, { "epoch": 180.97196261682242, "grad_norm": 1.984375, "learning_rate": 1.1318373844911855e-06, "loss": 0.1732, "step": 29136 }, { "epoch": 180.97819314641745, "grad_norm": 2.46875, "learning_rate": 1.1311045681532629e-06, "loss": 0.2871, "step": 29137 }, { "epoch": 180.98442367601245, "grad_norm": 1.8359375, "learning_rate": 1.1303719836348204e-06, "loss": 0.1201, "step": 29138 }, { "epoch": 180.99065420560748, "grad_norm": 3.078125, "learning_rate": 1.1296396309429746e-06, "loss": 0.2206, "step": 29139 }, { "epoch": 180.99688473520249, "grad_norm": 0.8984375, "learning_rate": 1.128907510084845e-06, "loss": 0.0865, "step": 29140 }, { "epoch": 181.0, "grad_norm": 0.93359375, "learning_rate": 1.1281756210675342e-06, "loss": 0.0627, "step": 29141 }, { "epoch": 181.00623052959503, "grad_norm": 1.359375, "learning_rate": 1.1274439638981532e-06, "loss": 0.1254, "step": 29142 }, { "epoch": 181.01246105919003, "grad_norm": 2.5, "learning_rate": 1.1267125385838102e-06, "loss": 0.1628, "step": 29143 }, { "epoch": 181.01869158878506, "grad_norm": 1.15625, "learning_rate": 1.1259813451316104e-06, "loss": 0.1081, "step": 29144 }, { "epoch": 181.02492211838006, "grad_norm": 1.921875, "learning_rate": 1.1252503835486456e-06, "loss": 0.1293, "step": 29145 }, { "epoch": 181.0311526479751, "grad_norm": 1.78125, "learning_rate": 1.1245196538420238e-06, "loss": 0.1028, "step": 29146 }, { "epoch": 181.0373831775701, "grad_norm": 2.21875, "learning_rate": 1.1237891560188367e-06, "loss": 0.2039, "step": 29147 }, { "epoch": 181.04361370716512, "grad_norm": 2.5, "learning_rate": 1.1230588900861899e-06, "loss": 0.197, "step": 29148 }, { "epoch": 181.04984423676012, "grad_norm": 1.234375, "learning_rate": 1.1223288560511664e-06, "loss": 0.095, "step": 29149 }, { "epoch": 181.05607476635515, "grad_norm": 3.109375, "learning_rate": 1.1215990539208521e-06, "loss": 0.1599, "step": 29150 }, { "epoch": 181.06230529595015, "grad_norm": 2.40625, "learning_rate": 1.1208694837023441e-06, "loss": 0.3352, "step": 29151 }, { "epoch": 181.06853582554518, "grad_norm": 2.078125, "learning_rate": 1.120140145402726e-06, "loss": 0.1305, "step": 29152 }, { "epoch": 181.07476635514018, "grad_norm": 1.8359375, "learning_rate": 1.119411039029078e-06, "loss": 0.1708, "step": 29153 }, { "epoch": 181.0809968847352, "grad_norm": 1.40625, "learning_rate": 1.118682164588486e-06, "loss": 0.1357, "step": 29154 }, { "epoch": 181.08722741433021, "grad_norm": 2.6875, "learning_rate": 1.1179535220880278e-06, "loss": 0.2775, "step": 29155 }, { "epoch": 181.09345794392524, "grad_norm": 2.0625, "learning_rate": 1.117225111534778e-06, "loss": 0.2286, "step": 29156 }, { "epoch": 181.09968847352025, "grad_norm": 1.9375, "learning_rate": 1.1164969329358093e-06, "loss": 0.2676, "step": 29157 }, { "epoch": 181.10591900311528, "grad_norm": 1.796875, "learning_rate": 1.1157689862981962e-06, "loss": 0.1355, "step": 29158 }, { "epoch": 181.11214953271028, "grad_norm": 1.25, "learning_rate": 1.1150412716290137e-06, "loss": 0.1163, "step": 29159 }, { "epoch": 181.1183800623053, "grad_norm": 1.8359375, "learning_rate": 1.11431378893532e-06, "loss": 0.1839, "step": 29160 }, { "epoch": 181.1246105919003, "grad_norm": 2.40625, "learning_rate": 1.1135865382241873e-06, "loss": 0.1855, "step": 29161 }, { "epoch": 181.13084112149534, "grad_norm": 1.1171875, "learning_rate": 1.112859519502679e-06, "loss": 0.1183, "step": 29162 }, { "epoch": 181.13707165109034, "grad_norm": 2.109375, "learning_rate": 1.1121327327778541e-06, "loss": 0.1147, "step": 29163 }, { "epoch": 181.14330218068537, "grad_norm": 3.25, "learning_rate": 1.1114061780567703e-06, "loss": 0.2803, "step": 29164 }, { "epoch": 181.14953271028037, "grad_norm": 1.3515625, "learning_rate": 1.1106798553464804e-06, "loss": 0.1363, "step": 29165 }, { "epoch": 181.1557632398754, "grad_norm": 1.53125, "learning_rate": 1.1099537646540508e-06, "loss": 0.1173, "step": 29166 }, { "epoch": 181.1619937694704, "grad_norm": 1.5, "learning_rate": 1.1092279059865206e-06, "loss": 0.1202, "step": 29167 }, { "epoch": 181.16822429906543, "grad_norm": 2.515625, "learning_rate": 1.1085022793509454e-06, "loss": 0.2363, "step": 29168 }, { "epoch": 181.17445482866043, "grad_norm": 1.9765625, "learning_rate": 1.1077768847543718e-06, "loss": 0.172, "step": 29169 }, { "epoch": 181.18068535825546, "grad_norm": 2.25, "learning_rate": 1.1070517222038473e-06, "loss": 0.1769, "step": 29170 }, { "epoch": 181.18691588785046, "grad_norm": 2.875, "learning_rate": 1.1063267917064106e-06, "loss": 0.2531, "step": 29171 }, { "epoch": 181.1931464174455, "grad_norm": 1.5546875, "learning_rate": 1.1056020932691036e-06, "loss": 0.1418, "step": 29172 }, { "epoch": 181.1993769470405, "grad_norm": 1.7890625, "learning_rate": 1.1048776268989674e-06, "loss": 0.1179, "step": 29173 }, { "epoch": 181.20560747663552, "grad_norm": 1.1875, "learning_rate": 1.104153392603033e-06, "loss": 0.1149, "step": 29174 }, { "epoch": 181.21183800623052, "grad_norm": 1.703125, "learning_rate": 1.1034293903883413e-06, "loss": 0.138, "step": 29175 }, { "epoch": 181.21806853582555, "grad_norm": 1.609375, "learning_rate": 1.102705620261918e-06, "loss": 0.1781, "step": 29176 }, { "epoch": 181.22429906542055, "grad_norm": 1.125, "learning_rate": 1.1019820822307985e-06, "loss": 0.1159, "step": 29177 }, { "epoch": 181.23052959501558, "grad_norm": 1.8984375, "learning_rate": 1.101258776302e-06, "loss": 0.1604, "step": 29178 }, { "epoch": 181.23676012461058, "grad_norm": 2.796875, "learning_rate": 1.1005357024825553e-06, "loss": 0.2395, "step": 29179 }, { "epoch": 181.2429906542056, "grad_norm": 2.84375, "learning_rate": 1.0998128607794866e-06, "loss": 0.208, "step": 29180 }, { "epoch": 181.2492211838006, "grad_norm": 1.1640625, "learning_rate": 1.0990902511998192e-06, "loss": 0.1081, "step": 29181 }, { "epoch": 181.25545171339564, "grad_norm": 2.375, "learning_rate": 1.0983678737505554e-06, "loss": 0.1399, "step": 29182 }, { "epoch": 181.26168224299064, "grad_norm": 2.03125, "learning_rate": 1.0976457284387232e-06, "loss": 0.2188, "step": 29183 }, { "epoch": 181.26791277258567, "grad_norm": 1.9609375, "learning_rate": 1.0969238152713334e-06, "loss": 0.1482, "step": 29184 }, { "epoch": 181.27414330218068, "grad_norm": 2.03125, "learning_rate": 1.0962021342554002e-06, "loss": 0.1561, "step": 29185 }, { "epoch": 181.2803738317757, "grad_norm": 2.46875, "learning_rate": 1.095480685397926e-06, "loss": 0.2056, "step": 29186 }, { "epoch": 181.2866043613707, "grad_norm": 1.9921875, "learning_rate": 1.0947594687059277e-06, "loss": 0.2555, "step": 29187 }, { "epoch": 181.29283489096574, "grad_norm": 2.78125, "learning_rate": 1.0940384841863992e-06, "loss": 0.3082, "step": 29188 }, { "epoch": 181.29906542056074, "grad_norm": 2.328125, "learning_rate": 1.0933177318463494e-06, "loss": 0.1719, "step": 29189 }, { "epoch": 181.30529595015577, "grad_norm": 1.640625, "learning_rate": 1.0925972116927752e-06, "loss": 0.1077, "step": 29190 }, { "epoch": 181.31152647975077, "grad_norm": 1.84375, "learning_rate": 1.0918769237326764e-06, "loss": 0.1384, "step": 29191 }, { "epoch": 181.3177570093458, "grad_norm": 0.9375, "learning_rate": 1.0911568679730504e-06, "loss": 0.1052, "step": 29192 }, { "epoch": 181.3239875389408, "grad_norm": 1.8828125, "learning_rate": 1.090437044420886e-06, "loss": 0.1181, "step": 29193 }, { "epoch": 181.33021806853583, "grad_norm": 2.0, "learning_rate": 1.0897174530831778e-06, "loss": 0.1748, "step": 29194 }, { "epoch": 181.33644859813083, "grad_norm": 1.9609375, "learning_rate": 1.0889980939669114e-06, "loss": 0.2298, "step": 29195 }, { "epoch": 181.34267912772586, "grad_norm": 2.34375, "learning_rate": 1.0882789670790789e-06, "loss": 0.2421, "step": 29196 }, { "epoch": 181.34890965732086, "grad_norm": 2.265625, "learning_rate": 1.0875600724266577e-06, "loss": 0.2067, "step": 29197 }, { "epoch": 181.3551401869159, "grad_norm": 2.453125, "learning_rate": 1.0868414100166313e-06, "loss": 0.2169, "step": 29198 }, { "epoch": 181.3613707165109, "grad_norm": 2.65625, "learning_rate": 1.0861229798559858e-06, "loss": 0.2074, "step": 29199 }, { "epoch": 181.36760124610592, "grad_norm": 2.609375, "learning_rate": 1.0854047819516933e-06, "loss": 0.1224, "step": 29200 }, { "epoch": 181.37383177570092, "grad_norm": 1.015625, "learning_rate": 1.084686816310726e-06, "loss": 0.1061, "step": 29201 }, { "epoch": 181.38006230529595, "grad_norm": 2.171875, "learning_rate": 1.0839690829400645e-06, "loss": 0.1174, "step": 29202 }, { "epoch": 181.38629283489095, "grad_norm": 2.734375, "learning_rate": 1.0832515818466754e-06, "loss": 0.1625, "step": 29203 }, { "epoch": 181.39252336448598, "grad_norm": 3.125, "learning_rate": 1.082534313037528e-06, "loss": 0.1947, "step": 29204 }, { "epoch": 181.39875389408098, "grad_norm": 1.1171875, "learning_rate": 1.0818172765195867e-06, "loss": 0.1195, "step": 29205 }, { "epoch": 181.404984423676, "grad_norm": 1.3046875, "learning_rate": 1.0811004722998146e-06, "loss": 0.088, "step": 29206 }, { "epoch": 181.41121495327104, "grad_norm": 2.9375, "learning_rate": 1.0803839003851845e-06, "loss": 0.2587, "step": 29207 }, { "epoch": 181.41744548286604, "grad_norm": 1.78125, "learning_rate": 1.0796675607826435e-06, "loss": 0.1855, "step": 29208 }, { "epoch": 181.42367601246107, "grad_norm": 2.921875, "learning_rate": 1.0789514534991497e-06, "loss": 0.2281, "step": 29209 }, { "epoch": 181.42990654205607, "grad_norm": 1.6796875, "learning_rate": 1.0782355785416615e-06, "loss": 0.1294, "step": 29210 }, { "epoch": 181.4361370716511, "grad_norm": 1.9296875, "learning_rate": 1.0775199359171345e-06, "loss": 0.1926, "step": 29211 }, { "epoch": 181.4423676012461, "grad_norm": 1.8984375, "learning_rate": 1.0768045256325132e-06, "loss": 0.1194, "step": 29212 }, { "epoch": 181.44859813084113, "grad_norm": 1.9375, "learning_rate": 1.0760893476947531e-06, "loss": 0.1738, "step": 29213 }, { "epoch": 181.45482866043614, "grad_norm": 2.46875, "learning_rate": 1.0753744021107902e-06, "loss": 0.1952, "step": 29214 }, { "epoch": 181.46105919003116, "grad_norm": 1.7734375, "learning_rate": 1.0746596888875776e-06, "loss": 0.1732, "step": 29215 }, { "epoch": 181.46728971962617, "grad_norm": 0.85546875, "learning_rate": 1.0739452080320484e-06, "loss": 0.101, "step": 29216 }, { "epoch": 181.4735202492212, "grad_norm": 2.171875, "learning_rate": 1.0732309595511469e-06, "loss": 0.2932, "step": 29217 }, { "epoch": 181.4797507788162, "grad_norm": 2.109375, "learning_rate": 1.0725169434518123e-06, "loss": 0.2047, "step": 29218 }, { "epoch": 181.48598130841123, "grad_norm": 2.234375, "learning_rate": 1.071803159740975e-06, "loss": 0.1336, "step": 29219 }, { "epoch": 181.49221183800623, "grad_norm": 1.5, "learning_rate": 1.0710896084255684e-06, "loss": 0.1342, "step": 29220 }, { "epoch": 181.49844236760126, "grad_norm": 1.4296875, "learning_rate": 1.0703762895125202e-06, "loss": 0.1412, "step": 29221 }, { "epoch": 181.50467289719626, "grad_norm": 3.40625, "learning_rate": 1.069663203008764e-06, "loss": 0.362, "step": 29222 }, { "epoch": 181.5109034267913, "grad_norm": 2.53125, "learning_rate": 1.0689503489212215e-06, "loss": 0.1798, "step": 29223 }, { "epoch": 181.5171339563863, "grad_norm": 1.71875, "learning_rate": 1.0682377272568157e-06, "loss": 0.1716, "step": 29224 }, { "epoch": 181.52336448598132, "grad_norm": 1.484375, "learning_rate": 1.0675253380224709e-06, "loss": 0.0968, "step": 29225 }, { "epoch": 181.52959501557632, "grad_norm": 1.0546875, "learning_rate": 1.0668131812251043e-06, "loss": 0.1038, "step": 29226 }, { "epoch": 181.53582554517135, "grad_norm": 1.6171875, "learning_rate": 1.0661012568716295e-06, "loss": 0.0988, "step": 29227 }, { "epoch": 181.54205607476635, "grad_norm": 2.40625, "learning_rate": 1.065389564968966e-06, "loss": 0.1776, "step": 29228 }, { "epoch": 181.54828660436138, "grad_norm": 2.328125, "learning_rate": 1.0646781055240223e-06, "loss": 0.2022, "step": 29229 }, { "epoch": 181.55451713395638, "grad_norm": 2.265625, "learning_rate": 1.0639668785437096e-06, "loss": 0.1286, "step": 29230 }, { "epoch": 181.5607476635514, "grad_norm": 2.046875, "learning_rate": 1.0632558840349332e-06, "loss": 0.1893, "step": 29231 }, { "epoch": 181.5669781931464, "grad_norm": 1.671875, "learning_rate": 1.0625451220046045e-06, "loss": 0.2102, "step": 29232 }, { "epoch": 181.57320872274144, "grad_norm": 2.296875, "learning_rate": 1.0618345924596234e-06, "loss": 0.1858, "step": 29233 }, { "epoch": 181.57943925233644, "grad_norm": 1.6328125, "learning_rate": 1.0611242954068873e-06, "loss": 0.1293, "step": 29234 }, { "epoch": 181.58566978193147, "grad_norm": 2.578125, "learning_rate": 1.0604142308532961e-06, "loss": 0.19, "step": 29235 }, { "epoch": 181.59190031152647, "grad_norm": 2.109375, "learning_rate": 1.0597043988057497e-06, "loss": 0.1825, "step": 29236 }, { "epoch": 181.5981308411215, "grad_norm": 2.828125, "learning_rate": 1.0589947992711374e-06, "loss": 0.2022, "step": 29237 }, { "epoch": 181.6043613707165, "grad_norm": 2.28125, "learning_rate": 1.0582854322563562e-06, "loss": 0.1659, "step": 29238 }, { "epoch": 181.61059190031153, "grad_norm": 3.21875, "learning_rate": 1.0575762977682895e-06, "loss": 0.3052, "step": 29239 }, { "epoch": 181.61682242990653, "grad_norm": 1.296875, "learning_rate": 1.056867395813832e-06, "loss": 0.1137, "step": 29240 }, { "epoch": 181.62305295950156, "grad_norm": 2.40625, "learning_rate": 1.0561587263998613e-06, "loss": 0.1869, "step": 29241 }, { "epoch": 181.62928348909657, "grad_norm": 0.8515625, "learning_rate": 1.0554502895332607e-06, "loss": 0.0953, "step": 29242 }, { "epoch": 181.6355140186916, "grad_norm": 1.078125, "learning_rate": 1.0547420852209167e-06, "loss": 0.1077, "step": 29243 }, { "epoch": 181.6417445482866, "grad_norm": 2.0625, "learning_rate": 1.054034113469704e-06, "loss": 0.2164, "step": 29244 }, { "epoch": 181.64797507788163, "grad_norm": 2.015625, "learning_rate": 1.0533263742865007e-06, "loss": 0.1677, "step": 29245 }, { "epoch": 181.65420560747663, "grad_norm": 2.671875, "learning_rate": 1.0526188676781761e-06, "loss": 0.252, "step": 29246 }, { "epoch": 181.66043613707166, "grad_norm": 1.3828125, "learning_rate": 1.0519115936516028e-06, "loss": 0.1126, "step": 29247 }, { "epoch": 181.66666666666666, "grad_norm": 1.8515625, "learning_rate": 1.0512045522136554e-06, "loss": 0.1002, "step": 29248 }, { "epoch": 181.6728971962617, "grad_norm": 3.21875, "learning_rate": 1.0504977433711926e-06, "loss": 0.136, "step": 29249 }, { "epoch": 181.6791277258567, "grad_norm": 1.3828125, "learning_rate": 1.049791167131084e-06, "loss": 0.1256, "step": 29250 }, { "epoch": 181.68535825545172, "grad_norm": 1.5078125, "learning_rate": 1.0490848235001932e-06, "loss": 0.1121, "step": 29251 }, { "epoch": 181.69158878504672, "grad_norm": 2.4375, "learning_rate": 1.048378712485379e-06, "loss": 0.1266, "step": 29252 }, { "epoch": 181.69781931464175, "grad_norm": 1.96875, "learning_rate": 1.0476728340934939e-06, "loss": 0.1473, "step": 29253 }, { "epoch": 181.70404984423675, "grad_norm": 2.90625, "learning_rate": 1.0469671883313992e-06, "loss": 0.3005, "step": 29254 }, { "epoch": 181.71028037383178, "grad_norm": 2.546875, "learning_rate": 1.0462617752059505e-06, "loss": 0.2207, "step": 29255 }, { "epoch": 181.71651090342678, "grad_norm": 2.109375, "learning_rate": 1.0455565947239926e-06, "loss": 0.126, "step": 29256 }, { "epoch": 181.7227414330218, "grad_norm": 2.390625, "learning_rate": 1.0448516468923808e-06, "loss": 0.1664, "step": 29257 }, { "epoch": 181.7289719626168, "grad_norm": 2.484375, "learning_rate": 1.0441469317179542e-06, "loss": 0.1437, "step": 29258 }, { "epoch": 181.73520249221184, "grad_norm": 0.9140625, "learning_rate": 1.0434424492075656e-06, "loss": 0.0939, "step": 29259 }, { "epoch": 181.74143302180684, "grad_norm": 1.859375, "learning_rate": 1.0427381993680512e-06, "loss": 0.1518, "step": 29260 }, { "epoch": 181.74766355140187, "grad_norm": 2.375, "learning_rate": 1.0420341822062502e-06, "loss": 0.2195, "step": 29261 }, { "epoch": 181.75389408099687, "grad_norm": 2.46875, "learning_rate": 1.041330397729004e-06, "loss": 0.2371, "step": 29262 }, { "epoch": 181.7601246105919, "grad_norm": 1.6484375, "learning_rate": 1.040626845943146e-06, "loss": 0.1141, "step": 29263 }, { "epoch": 181.7663551401869, "grad_norm": 2.15625, "learning_rate": 1.0399235268555103e-06, "loss": 0.1048, "step": 29264 }, { "epoch": 181.77258566978193, "grad_norm": 1.96875, "learning_rate": 1.0392204404729266e-06, "loss": 0.145, "step": 29265 }, { "epoch": 181.77881619937693, "grad_norm": 1.2734375, "learning_rate": 1.0385175868022234e-06, "loss": 0.1028, "step": 29266 }, { "epoch": 181.78504672897196, "grad_norm": 1.328125, "learning_rate": 1.0378149658502256e-06, "loss": 0.0991, "step": 29267 }, { "epoch": 181.79127725856696, "grad_norm": 2.03125, "learning_rate": 1.0371125776237584e-06, "loss": 0.181, "step": 29268 }, { "epoch": 181.797507788162, "grad_norm": 1.96875, "learning_rate": 1.0364104221296495e-06, "loss": 0.215, "step": 29269 }, { "epoch": 181.80373831775702, "grad_norm": 1.0703125, "learning_rate": 1.0357084993747101e-06, "loss": 0.1145, "step": 29270 }, { "epoch": 181.80996884735202, "grad_norm": 1.875, "learning_rate": 1.0350068093657572e-06, "loss": 0.1147, "step": 29271 }, { "epoch": 181.81619937694705, "grad_norm": 2.1875, "learning_rate": 1.0343053521096102e-06, "loss": 0.2408, "step": 29272 }, { "epoch": 181.82242990654206, "grad_norm": 2.4375, "learning_rate": 1.033604127613083e-06, "loss": 0.1521, "step": 29273 }, { "epoch": 181.82866043613708, "grad_norm": 1.578125, "learning_rate": 1.0329031358829788e-06, "loss": 0.128, "step": 29274 }, { "epoch": 181.8348909657321, "grad_norm": 2.25, "learning_rate": 1.0322023769261113e-06, "loss": 0.1964, "step": 29275 }, { "epoch": 181.84112149532712, "grad_norm": 2.125, "learning_rate": 1.0315018507492836e-06, "loss": 0.2324, "step": 29276 }, { "epoch": 181.84735202492212, "grad_norm": 2.46875, "learning_rate": 1.0308015573593094e-06, "loss": 0.2147, "step": 29277 }, { "epoch": 181.85358255451715, "grad_norm": 2.8125, "learning_rate": 1.0301014967629751e-06, "loss": 0.1671, "step": 29278 }, { "epoch": 181.85981308411215, "grad_norm": 1.21875, "learning_rate": 1.0294016689670865e-06, "loss": 0.1166, "step": 29279 }, { "epoch": 181.86604361370718, "grad_norm": 1.671875, "learning_rate": 1.0287020739784408e-06, "loss": 0.1466, "step": 29280 }, { "epoch": 181.87227414330218, "grad_norm": 1.3359375, "learning_rate": 1.0280027118038355e-06, "loss": 0.1208, "step": 29281 }, { "epoch": 181.8785046728972, "grad_norm": 1.5, "learning_rate": 1.0273035824500565e-06, "loss": 0.1265, "step": 29282 }, { "epoch": 181.8847352024922, "grad_norm": 1.6484375, "learning_rate": 1.0266046859238986e-06, "loss": 0.1496, "step": 29283 }, { "epoch": 181.89096573208724, "grad_norm": 2.859375, "learning_rate": 1.0259060222321483e-06, "loss": 0.3409, "step": 29284 }, { "epoch": 181.89719626168224, "grad_norm": 2.359375, "learning_rate": 1.0252075913815912e-06, "loss": 0.2358, "step": 29285 }, { "epoch": 181.90342679127727, "grad_norm": 0.91015625, "learning_rate": 1.0245093933790085e-06, "loss": 0.0971, "step": 29286 }, { "epoch": 181.90965732087227, "grad_norm": 2.265625, "learning_rate": 1.0238114282311835e-06, "loss": 0.122, "step": 29287 }, { "epoch": 181.9158878504673, "grad_norm": 1.7890625, "learning_rate": 1.0231136959448966e-06, "loss": 0.1225, "step": 29288 }, { "epoch": 181.9221183800623, "grad_norm": 2.21875, "learning_rate": 1.0224161965269236e-06, "loss": 0.1073, "step": 29289 }, { "epoch": 181.92834890965733, "grad_norm": 2.703125, "learning_rate": 1.0217189299840336e-06, "loss": 0.1518, "step": 29290 }, { "epoch": 181.93457943925233, "grad_norm": 1.78125, "learning_rate": 1.0210218963230044e-06, "loss": 0.1106, "step": 29291 }, { "epoch": 181.94080996884736, "grad_norm": 3.65625, "learning_rate": 1.020325095550606e-06, "loss": 0.2287, "step": 29292 }, { "epoch": 181.94704049844236, "grad_norm": 2.4375, "learning_rate": 1.0196285276735995e-06, "loss": 0.1095, "step": 29293 }, { "epoch": 181.9532710280374, "grad_norm": 1.5234375, "learning_rate": 1.0189321926987543e-06, "loss": 0.0995, "step": 29294 }, { "epoch": 181.9595015576324, "grad_norm": 0.921875, "learning_rate": 1.0182360906328375e-06, "loss": 0.0907, "step": 29295 }, { "epoch": 181.96573208722742, "grad_norm": 1.5234375, "learning_rate": 1.0175402214826045e-06, "loss": 0.1284, "step": 29296 }, { "epoch": 181.97196261682242, "grad_norm": 3.109375, "learning_rate": 1.0168445852548142e-06, "loss": 0.2238, "step": 29297 }, { "epoch": 181.97819314641745, "grad_norm": 1.4921875, "learning_rate": 1.016149181956222e-06, "loss": 0.1223, "step": 29298 }, { "epoch": 181.98442367601245, "grad_norm": 1.640625, "learning_rate": 1.0154540115935863e-06, "loss": 0.1379, "step": 29299 }, { "epoch": 181.99065420560748, "grad_norm": 3.015625, "learning_rate": 1.0147590741736517e-06, "loss": 0.2854, "step": 29300 }, { "epoch": 181.99688473520249, "grad_norm": 1.5859375, "learning_rate": 1.0140643697031716e-06, "loss": 0.1289, "step": 29301 }, { "epoch": 182.0, "grad_norm": 1.2578125, "learning_rate": 1.0133698981888955e-06, "loss": 0.0767, "step": 29302 }, { "epoch": 182.00623052959503, "grad_norm": 3.15625, "learning_rate": 1.0126756596375686e-06, "loss": 0.3905, "step": 29303 }, { "epoch": 182.01246105919003, "grad_norm": 1.5703125, "learning_rate": 1.011981654055924e-06, "loss": 0.1384, "step": 29304 }, { "epoch": 182.01869158878506, "grad_norm": 2.5, "learning_rate": 1.011287881450712e-06, "loss": 0.3055, "step": 29305 }, { "epoch": 182.02492211838006, "grad_norm": 2.265625, "learning_rate": 1.0105943418286661e-06, "loss": 0.1875, "step": 29306 }, { "epoch": 182.0311526479751, "grad_norm": 2.875, "learning_rate": 1.0099010351965255e-06, "loss": 0.1903, "step": 29307 }, { "epoch": 182.0373831775701, "grad_norm": 1.578125, "learning_rate": 1.0092079615610234e-06, "loss": 0.1422, "step": 29308 }, { "epoch": 182.04361370716512, "grad_norm": 2.9375, "learning_rate": 1.008515120928885e-06, "loss": 0.1417, "step": 29309 }, { "epoch": 182.04984423676012, "grad_norm": 2.625, "learning_rate": 1.0078225133068442e-06, "loss": 0.1927, "step": 29310 }, { "epoch": 182.05607476635515, "grad_norm": 1.9921875, "learning_rate": 1.0071301387016284e-06, "loss": 0.1189, "step": 29311 }, { "epoch": 182.06230529595015, "grad_norm": 2.71875, "learning_rate": 1.0064379971199605e-06, "loss": 0.2083, "step": 29312 }, { "epoch": 182.06853582554518, "grad_norm": 1.453125, "learning_rate": 1.00574608856856e-06, "loss": 0.1415, "step": 29313 }, { "epoch": 182.07476635514018, "grad_norm": 2.6875, "learning_rate": 1.0050544130541545e-06, "loss": 0.2783, "step": 29314 }, { "epoch": 182.0809968847352, "grad_norm": 2.734375, "learning_rate": 1.0043629705834556e-06, "loss": 0.185, "step": 29315 }, { "epoch": 182.08722741433021, "grad_norm": 1.453125, "learning_rate": 1.0036717611631802e-06, "loss": 0.1129, "step": 29316 }, { "epoch": 182.09345794392524, "grad_norm": 1.796875, "learning_rate": 1.0029807848000395e-06, "loss": 0.1141, "step": 29317 }, { "epoch": 182.09968847352025, "grad_norm": 1.125, "learning_rate": 1.0022900415007503e-06, "loss": 0.1045, "step": 29318 }, { "epoch": 182.10591900311528, "grad_norm": 1.953125, "learning_rate": 1.0015995312720127e-06, "loss": 0.122, "step": 29319 }, { "epoch": 182.11214953271028, "grad_norm": 1.859375, "learning_rate": 1.0009092541205384e-06, "loss": 0.1202, "step": 29320 }, { "epoch": 182.1183800623053, "grad_norm": 1.53125, "learning_rate": 1.0002192100530327e-06, "loss": 0.1502, "step": 29321 }, { "epoch": 182.1246105919003, "grad_norm": 2.703125, "learning_rate": 9.99529399076196e-07, "loss": 0.1379, "step": 29322 }, { "epoch": 182.13084112149534, "grad_norm": 2.78125, "learning_rate": 9.98839821196726e-07, "loss": 0.2313, "step": 29323 }, { "epoch": 182.13707165109034, "grad_norm": 1.3984375, "learning_rate": 9.981504764213194e-07, "loss": 0.1029, "step": 29324 }, { "epoch": 182.14330218068537, "grad_norm": 2.671875, "learning_rate": 9.974613647566772e-07, "loss": 0.2702, "step": 29325 }, { "epoch": 182.14953271028037, "grad_norm": 1.0859375, "learning_rate": 9.96772486209485e-07, "loss": 0.0952, "step": 29326 }, { "epoch": 182.1557632398754, "grad_norm": 2.015625, "learning_rate": 9.96083840786438e-07, "loss": 0.1205, "step": 29327 }, { "epoch": 182.1619937694704, "grad_norm": 2.40625, "learning_rate": 9.953954284942253e-07, "loss": 0.1856, "step": 29328 }, { "epoch": 182.16822429906543, "grad_norm": 1.140625, "learning_rate": 9.947072493395299e-07, "loss": 0.1148, "step": 29329 }, { "epoch": 182.17445482866043, "grad_norm": 1.03125, "learning_rate": 9.940193033290329e-07, "loss": 0.0938, "step": 29330 }, { "epoch": 182.18068535825546, "grad_norm": 3.484375, "learning_rate": 9.93331590469418e-07, "loss": 0.3086, "step": 29331 }, { "epoch": 182.18691588785046, "grad_norm": 1.9296875, "learning_rate": 9.926441107673711e-07, "loss": 0.1492, "step": 29332 }, { "epoch": 182.1931464174455, "grad_norm": 2.609375, "learning_rate": 9.919568642295595e-07, "loss": 0.2457, "step": 29333 }, { "epoch": 182.1993769470405, "grad_norm": 2.9375, "learning_rate": 9.91269850862664e-07, "loss": 0.2249, "step": 29334 }, { "epoch": 182.20560747663552, "grad_norm": 2.890625, "learning_rate": 9.905830706733538e-07, "loss": 0.3859, "step": 29335 }, { "epoch": 182.21183800623052, "grad_norm": 2.28125, "learning_rate": 9.898965236683016e-07, "loss": 0.1727, "step": 29336 }, { "epoch": 182.21806853582555, "grad_norm": 2.1875, "learning_rate": 9.892102098541716e-07, "loss": 0.1856, "step": 29337 }, { "epoch": 182.22429906542055, "grad_norm": 2.1875, "learning_rate": 9.885241292376335e-07, "loss": 0.1526, "step": 29338 }, { "epoch": 182.23052959501558, "grad_norm": 1.90625, "learning_rate": 9.878382818253457e-07, "loss": 0.187, "step": 29339 }, { "epoch": 182.23676012461058, "grad_norm": 2.140625, "learning_rate": 9.871526676239806e-07, "loss": 0.1707, "step": 29340 }, { "epoch": 182.2429906542056, "grad_norm": 3.15625, "learning_rate": 9.86467286640183e-07, "loss": 0.3306, "step": 29341 }, { "epoch": 182.2492211838006, "grad_norm": 1.734375, "learning_rate": 9.857821388806144e-07, "loss": 0.1465, "step": 29342 }, { "epoch": 182.25545171339564, "grad_norm": 2.78125, "learning_rate": 9.850972243519301e-07, "loss": 0.1875, "step": 29343 }, { "epoch": 182.26168224299064, "grad_norm": 2.5625, "learning_rate": 9.844125430607836e-07, "loss": 0.2335, "step": 29344 }, { "epoch": 182.26791277258567, "grad_norm": 0.81640625, "learning_rate": 9.837280950138223e-07, "loss": 0.0909, "step": 29345 }, { "epoch": 182.27414330218068, "grad_norm": 2.4375, "learning_rate": 9.830438802176933e-07, "loss": 0.0912, "step": 29346 }, { "epoch": 182.2803738317757, "grad_norm": 3.34375, "learning_rate": 9.823598986790472e-07, "loss": 0.2777, "step": 29347 }, { "epoch": 182.2866043613707, "grad_norm": 2.84375, "learning_rate": 9.81676150404523e-07, "loss": 0.2423, "step": 29348 }, { "epoch": 182.29283489096574, "grad_norm": 1.5234375, "learning_rate": 9.80992635400757e-07, "loss": 0.1221, "step": 29349 }, { "epoch": 182.29906542056074, "grad_norm": 0.8828125, "learning_rate": 9.803093536743912e-07, "loss": 0.0987, "step": 29350 }, { "epoch": 182.30529595015577, "grad_norm": 2.015625, "learning_rate": 9.796263052320675e-07, "loss": 0.2199, "step": 29351 }, { "epoch": 182.31152647975077, "grad_norm": 3.59375, "learning_rate": 9.789434900804085e-07, "loss": 0.1201, "step": 29352 }, { "epoch": 182.3177570093458, "grad_norm": 2.71875, "learning_rate": 9.782609082260586e-07, "loss": 0.2318, "step": 29353 }, { "epoch": 182.3239875389408, "grad_norm": 2.375, "learning_rate": 9.775785596756348e-07, "loss": 0.2016, "step": 29354 }, { "epoch": 182.33021806853583, "grad_norm": 4.1875, "learning_rate": 9.768964444357764e-07, "loss": 0.2743, "step": 29355 }, { "epoch": 182.33644859813083, "grad_norm": 2.4375, "learning_rate": 9.762145625130948e-07, "loss": 0.2529, "step": 29356 }, { "epoch": 182.34267912772586, "grad_norm": 2.625, "learning_rate": 9.755329139142233e-07, "loss": 0.2107, "step": 29357 }, { "epoch": 182.34890965732086, "grad_norm": 2.46875, "learning_rate": 9.74851498645779e-07, "loss": 0.2482, "step": 29358 }, { "epoch": 182.3551401869159, "grad_norm": 1.3046875, "learning_rate": 9.74170316714379e-07, "loss": 0.1155, "step": 29359 }, { "epoch": 182.3613707165109, "grad_norm": 1.5, "learning_rate": 9.7348936812664e-07, "loss": 0.1038, "step": 29360 }, { "epoch": 182.36760124610592, "grad_norm": 1.5078125, "learning_rate": 9.728086528891706e-07, "loss": 0.124, "step": 29361 }, { "epoch": 182.37383177570092, "grad_norm": 1.9140625, "learning_rate": 9.721281710085934e-07, "loss": 0.1222, "step": 29362 }, { "epoch": 182.38006230529595, "grad_norm": 1.8046875, "learning_rate": 9.714479224915058e-07, "loss": 0.0771, "step": 29363 }, { "epoch": 182.38629283489095, "grad_norm": 3.0625, "learning_rate": 9.707679073445193e-07, "loss": 0.3176, "step": 29364 }, { "epoch": 182.39252336448598, "grad_norm": 1.9609375, "learning_rate": 9.700881255742395e-07, "loss": 0.1511, "step": 29365 }, { "epoch": 182.39875389408098, "grad_norm": 2.359375, "learning_rate": 9.694085771872695e-07, "loss": 0.2182, "step": 29366 }, { "epoch": 182.404984423676, "grad_norm": 0.9765625, "learning_rate": 9.687292621902044e-07, "loss": 0.091, "step": 29367 }, { "epoch": 182.41121495327104, "grad_norm": 2.125, "learning_rate": 9.680501805896442e-07, "loss": 0.1717, "step": 29368 }, { "epoch": 182.41744548286604, "grad_norm": 2.171875, "learning_rate": 9.673713323921862e-07, "loss": 0.12, "step": 29369 }, { "epoch": 182.42367601246107, "grad_norm": 1.75, "learning_rate": 9.6669271760442e-07, "loss": 0.1293, "step": 29370 }, { "epoch": 182.42990654205607, "grad_norm": 2.765625, "learning_rate": 9.66014336232937e-07, "loss": 0.2901, "step": 29371 }, { "epoch": 182.4361370716511, "grad_norm": 1.4140625, "learning_rate": 9.653361882843298e-07, "loss": 0.1392, "step": 29372 }, { "epoch": 182.4423676012461, "grad_norm": 2.140625, "learning_rate": 9.646582737651844e-07, "loss": 0.1742, "step": 29373 }, { "epoch": 182.44859813084113, "grad_norm": 1.6953125, "learning_rate": 9.63980592682079e-07, "loss": 0.1335, "step": 29374 }, { "epoch": 182.45482866043614, "grad_norm": 2.875, "learning_rate": 9.63303145041597e-07, "loss": 0.1754, "step": 29375 }, { "epoch": 182.46105919003116, "grad_norm": 2.0625, "learning_rate": 9.626259308503194e-07, "loss": 0.1722, "step": 29376 }, { "epoch": 182.46728971962617, "grad_norm": 1.4296875, "learning_rate": 9.619489501148299e-07, "loss": 0.1039, "step": 29377 }, { "epoch": 182.4735202492212, "grad_norm": 2.5625, "learning_rate": 9.612722028416926e-07, "loss": 0.313, "step": 29378 }, { "epoch": 182.4797507788162, "grad_norm": 1.9765625, "learning_rate": 9.605956890374858e-07, "loss": 0.1588, "step": 29379 }, { "epoch": 182.48598130841123, "grad_norm": 1.203125, "learning_rate": 9.599194087087788e-07, "loss": 0.0945, "step": 29380 }, { "epoch": 182.49221183800623, "grad_norm": 1.515625, "learning_rate": 9.592433618621416e-07, "loss": 0.1359, "step": 29381 }, { "epoch": 182.49844236760126, "grad_norm": 1.5, "learning_rate": 9.585675485041356e-07, "loss": 0.1201, "step": 29382 }, { "epoch": 182.50467289719626, "grad_norm": 1.8046875, "learning_rate": 9.578919686413279e-07, "loss": 0.1614, "step": 29383 }, { "epoch": 182.5109034267913, "grad_norm": 1.953125, "learning_rate": 9.572166222802824e-07, "loss": 0.1269, "step": 29384 }, { "epoch": 182.5171339563863, "grad_norm": 3.0, "learning_rate": 9.565415094275553e-07, "loss": 0.2527, "step": 29385 }, { "epoch": 182.52336448598132, "grad_norm": 1.2890625, "learning_rate": 9.558666300896996e-07, "loss": 0.1016, "step": 29386 }, { "epoch": 182.52959501557632, "grad_norm": 1.6015625, "learning_rate": 9.551919842732737e-07, "loss": 0.1091, "step": 29387 }, { "epoch": 182.53582554517135, "grad_norm": 1.78125, "learning_rate": 9.545175719848338e-07, "loss": 0.1672, "step": 29388 }, { "epoch": 182.54205607476635, "grad_norm": 2.40625, "learning_rate": 9.538433932309216e-07, "loss": 0.2129, "step": 29389 }, { "epoch": 182.54828660436138, "grad_norm": 1.4375, "learning_rate": 9.531694480180875e-07, "loss": 0.111, "step": 29390 }, { "epoch": 182.55451713395638, "grad_norm": 2.65625, "learning_rate": 9.524957363528847e-07, "loss": 0.1702, "step": 29391 }, { "epoch": 182.5607476635514, "grad_norm": 1.0859375, "learning_rate": 9.518222582418496e-07, "loss": 0.0874, "step": 29392 }, { "epoch": 182.5669781931464, "grad_norm": 1.4921875, "learning_rate": 9.511490136915186e-07, "loss": 0.1014, "step": 29393 }, { "epoch": 182.57320872274144, "grad_norm": 3.140625, "learning_rate": 9.504760027084364e-07, "loss": 0.2667, "step": 29394 }, { "epoch": 182.57943925233644, "grad_norm": 1.984375, "learning_rate": 9.498032252991451e-07, "loss": 0.1111, "step": 29395 }, { "epoch": 182.58566978193147, "grad_norm": 2.265625, "learning_rate": 9.491306814701644e-07, "loss": 0.2017, "step": 29396 }, { "epoch": 182.59190031152647, "grad_norm": 1.75, "learning_rate": 9.484583712280365e-07, "loss": 0.1666, "step": 29397 }, { "epoch": 182.5981308411215, "grad_norm": 2.515625, "learning_rate": 9.477862945792921e-07, "loss": 0.1719, "step": 29398 }, { "epoch": 182.6043613707165, "grad_norm": 1.1640625, "learning_rate": 9.471144515304564e-07, "loss": 0.1207, "step": 29399 }, { "epoch": 182.61059190031153, "grad_norm": 3.25, "learning_rate": 9.464428420880494e-07, "loss": 0.216, "step": 29400 }, { "epoch": 182.61682242990653, "grad_norm": 1.8359375, "learning_rate": 9.457714662585965e-07, "loss": 0.1722, "step": 29401 }, { "epoch": 182.62305295950156, "grad_norm": 2.171875, "learning_rate": 9.451003240486201e-07, "loss": 0.2124, "step": 29402 }, { "epoch": 182.62928348909657, "grad_norm": 2.734375, "learning_rate": 9.444294154646399e-07, "loss": 0.207, "step": 29403 }, { "epoch": 182.6355140186916, "grad_norm": 1.78125, "learning_rate": 9.437587405131704e-07, "loss": 0.1632, "step": 29404 }, { "epoch": 182.6417445482866, "grad_norm": 2.90625, "learning_rate": 9.430882992007228e-07, "loss": 0.2103, "step": 29405 }, { "epoch": 182.64797507788163, "grad_norm": 1.8125, "learning_rate": 9.424180915338088e-07, "loss": 0.1535, "step": 29406 }, { "epoch": 182.65420560747663, "grad_norm": 2.015625, "learning_rate": 9.417481175189452e-07, "loss": 0.1575, "step": 29407 }, { "epoch": 182.66043613707166, "grad_norm": 1.2265625, "learning_rate": 9.410783771626269e-07, "loss": 0.1018, "step": 29408 }, { "epoch": 182.66666666666666, "grad_norm": 2.796875, "learning_rate": 9.404088704713654e-07, "loss": 0.1167, "step": 29409 }, { "epoch": 182.6728971962617, "grad_norm": 1.3828125, "learning_rate": 9.397395974516665e-07, "loss": 0.1185, "step": 29410 }, { "epoch": 182.6791277258567, "grad_norm": 1.4453125, "learning_rate": 9.390705581100251e-07, "loss": 0.1165, "step": 29411 }, { "epoch": 182.68535825545172, "grad_norm": 1.203125, "learning_rate": 9.384017524529387e-07, "loss": 0.127, "step": 29412 }, { "epoch": 182.69158878504672, "grad_norm": 2.296875, "learning_rate": 9.377331804869022e-07, "loss": 0.1954, "step": 29413 }, { "epoch": 182.69781931464175, "grad_norm": 1.015625, "learning_rate": 9.370648422184159e-07, "loss": 0.0965, "step": 29414 }, { "epoch": 182.70404984423675, "grad_norm": 2.109375, "learning_rate": 9.363967376539606e-07, "loss": 0.1493, "step": 29415 }, { "epoch": 182.71028037383178, "grad_norm": 1.8984375, "learning_rate": 9.357288668000342e-07, "loss": 0.2102, "step": 29416 }, { "epoch": 182.71651090342678, "grad_norm": 1.1875, "learning_rate": 9.350612296631172e-07, "loss": 0.1333, "step": 29417 }, { "epoch": 182.7227414330218, "grad_norm": 2.25, "learning_rate": 9.343938262496993e-07, "loss": 0.1386, "step": 29418 }, { "epoch": 182.7289719626168, "grad_norm": 1.3671875, "learning_rate": 9.337266565662556e-07, "loss": 0.1214, "step": 29419 }, { "epoch": 182.73520249221184, "grad_norm": 2.109375, "learning_rate": 9.330597206192698e-07, "loss": 0.1413, "step": 29420 }, { "epoch": 182.74143302180684, "grad_norm": 1.484375, "learning_rate": 9.32393018415223e-07, "loss": 0.1288, "step": 29421 }, { "epoch": 182.74766355140187, "grad_norm": 0.76171875, "learning_rate": 9.31726549960582e-07, "loss": 0.0887, "step": 29422 }, { "epoch": 182.75389408099687, "grad_norm": 2.8125, "learning_rate": 9.310603152618252e-07, "loss": 0.1804, "step": 29423 }, { "epoch": 182.7601246105919, "grad_norm": 1.7109375, "learning_rate": 9.303943143254224e-07, "loss": 0.1225, "step": 29424 }, { "epoch": 182.7663551401869, "grad_norm": 1.9453125, "learning_rate": 9.297285471578404e-07, "loss": 0.1722, "step": 29425 }, { "epoch": 182.77258566978193, "grad_norm": 2.484375, "learning_rate": 9.290630137655465e-07, "loss": 0.2262, "step": 29426 }, { "epoch": 182.77881619937693, "grad_norm": 1.7421875, "learning_rate": 9.283977141550021e-07, "loss": 0.1631, "step": 29427 }, { "epoch": 182.78504672897196, "grad_norm": 0.9921875, "learning_rate": 9.277326483326743e-07, "loss": 0.1102, "step": 29428 }, { "epoch": 182.79127725856696, "grad_norm": 2.015625, "learning_rate": 9.270678163050217e-07, "loss": 0.1497, "step": 29429 }, { "epoch": 182.797507788162, "grad_norm": 1.546875, "learning_rate": 9.264032180784921e-07, "loss": 0.1032, "step": 29430 }, { "epoch": 182.80373831775702, "grad_norm": 2.203125, "learning_rate": 9.257388536595496e-07, "loss": 0.1419, "step": 29431 }, { "epoch": 182.80996884735202, "grad_norm": 2.515625, "learning_rate": 9.250747230546447e-07, "loss": 0.2004, "step": 29432 }, { "epoch": 182.81619937694705, "grad_norm": 1.9765625, "learning_rate": 9.244108262702222e-07, "loss": 0.1292, "step": 29433 }, { "epoch": 182.82242990654206, "grad_norm": 3.453125, "learning_rate": 9.237471633127381e-07, "loss": 0.1396, "step": 29434 }, { "epoch": 182.82866043613708, "grad_norm": 3.046875, "learning_rate": 9.230837341886317e-07, "loss": 0.3017, "step": 29435 }, { "epoch": 182.8348909657321, "grad_norm": 1.765625, "learning_rate": 9.224205389043561e-07, "loss": 0.181, "step": 29436 }, { "epoch": 182.84112149532712, "grad_norm": 2.5, "learning_rate": 9.217575774663367e-07, "loss": 0.1141, "step": 29437 }, { "epoch": 182.84735202492212, "grad_norm": 2.21875, "learning_rate": 9.21094849881024e-07, "loss": 0.2249, "step": 29438 }, { "epoch": 182.85358255451715, "grad_norm": 2.03125, "learning_rate": 9.204323561548517e-07, "loss": 0.1599, "step": 29439 }, { "epoch": 182.85981308411215, "grad_norm": 2.515625, "learning_rate": 9.197700962942534e-07, "loss": 0.1394, "step": 29440 }, { "epoch": 182.86604361370718, "grad_norm": 1.0546875, "learning_rate": 9.191080703056604e-07, "loss": 0.1089, "step": 29441 }, { "epoch": 182.87227414330218, "grad_norm": 3.21875, "learning_rate": 9.184462781955061e-07, "loss": 0.1886, "step": 29442 }, { "epoch": 182.8785046728972, "grad_norm": 1.734375, "learning_rate": 9.177847199702133e-07, "loss": 0.1458, "step": 29443 }, { "epoch": 182.8847352024922, "grad_norm": 1.9765625, "learning_rate": 9.171233956362129e-07, "loss": 0.2373, "step": 29444 }, { "epoch": 182.89096573208724, "grad_norm": 2.578125, "learning_rate": 9.164623051999193e-07, "loss": 0.1676, "step": 29445 }, { "epoch": 182.89719626168224, "grad_norm": 1.7421875, "learning_rate": 9.158014486677579e-07, "loss": 0.1508, "step": 29446 }, { "epoch": 182.90342679127727, "grad_norm": 1.9453125, "learning_rate": 9.151408260461513e-07, "loss": 0.1538, "step": 29447 }, { "epoch": 182.90965732087227, "grad_norm": 0.79296875, "learning_rate": 9.144804373415112e-07, "loss": 0.1008, "step": 29448 }, { "epoch": 182.9158878504673, "grad_norm": 0.97265625, "learning_rate": 9.138202825602488e-07, "loss": 0.1186, "step": 29449 }, { "epoch": 182.9221183800623, "grad_norm": 1.7109375, "learning_rate": 9.131603617087786e-07, "loss": 0.1243, "step": 29450 }, { "epoch": 182.92834890965733, "grad_norm": 0.921875, "learning_rate": 9.125006747935122e-07, "loss": 0.112, "step": 29451 }, { "epoch": 182.93457943925233, "grad_norm": 2.5, "learning_rate": 9.118412218208528e-07, "loss": 0.2176, "step": 29452 }, { "epoch": 182.94080996884736, "grad_norm": 1.3515625, "learning_rate": 9.111820027972035e-07, "loss": 0.1343, "step": 29453 }, { "epoch": 182.94704049844236, "grad_norm": 1.15625, "learning_rate": 9.105230177289759e-07, "loss": 0.1305, "step": 29454 }, { "epoch": 182.9532710280374, "grad_norm": 1.703125, "learning_rate": 9.098642666225621e-07, "loss": 0.1232, "step": 29455 }, { "epoch": 182.9595015576324, "grad_norm": 1.90625, "learning_rate": 9.092057494843598e-07, "loss": 0.1035, "step": 29456 }, { "epoch": 182.96573208722742, "grad_norm": 1.1484375, "learning_rate": 9.085474663207666e-07, "loss": 0.1035, "step": 29457 }, { "epoch": 182.97196261682242, "grad_norm": 1.65625, "learning_rate": 9.078894171381803e-07, "loss": 0.0859, "step": 29458 }, { "epoch": 182.97819314641745, "grad_norm": 2.46875, "learning_rate": 9.072316019429844e-07, "loss": 0.1161, "step": 29459 }, { "epoch": 182.98442367601245, "grad_norm": 2.703125, "learning_rate": 9.065740207415712e-07, "loss": 0.2112, "step": 29460 }, { "epoch": 182.99065420560748, "grad_norm": 1.5, "learning_rate": 9.059166735403302e-07, "loss": 0.1377, "step": 29461 }, { "epoch": 182.99688473520249, "grad_norm": 1.328125, "learning_rate": 9.05259560345642e-07, "loss": 0.1382, "step": 29462 }, { "epoch": 183.0, "grad_norm": 0.75, "learning_rate": 9.04602681163888e-07, "loss": 0.0506, "step": 29463 }, { "epoch": 183.00623052959503, "grad_norm": 3.046875, "learning_rate": 9.03946036001449e-07, "loss": 0.2575, "step": 29464 }, { "epoch": 183.01246105919003, "grad_norm": 1.796875, "learning_rate": 9.032896248647033e-07, "loss": 0.1616, "step": 29465 }, { "epoch": 183.01869158878506, "grad_norm": 1.078125, "learning_rate": 9.026334477600262e-07, "loss": 0.0924, "step": 29466 }, { "epoch": 183.02492211838006, "grad_norm": 2.40625, "learning_rate": 9.019775046937879e-07, "loss": 0.2278, "step": 29467 }, { "epoch": 183.0311526479751, "grad_norm": 2.34375, "learning_rate": 9.013217956723636e-07, "loss": 0.1678, "step": 29468 }, { "epoch": 183.0373831775701, "grad_norm": 1.515625, "learning_rate": 9.006663207021205e-07, "loss": 0.118, "step": 29469 }, { "epoch": 183.04361370716512, "grad_norm": 3.171875, "learning_rate": 9.000110797894201e-07, "loss": 0.2444, "step": 29470 }, { "epoch": 183.04984423676012, "grad_norm": 0.82421875, "learning_rate": 8.993560729406297e-07, "loss": 0.1044, "step": 29471 }, { "epoch": 183.05607476635515, "grad_norm": 1.4609375, "learning_rate": 8.987013001621108e-07, "loss": 0.1439, "step": 29472 }, { "epoch": 183.06230529595015, "grad_norm": 3.03125, "learning_rate": 8.980467614602279e-07, "loss": 0.3082, "step": 29473 }, { "epoch": 183.06853582554518, "grad_norm": 1.46875, "learning_rate": 8.973924568413283e-07, "loss": 0.1282, "step": 29474 }, { "epoch": 183.07476635514018, "grad_norm": 1.8125, "learning_rate": 8.967383863117712e-07, "loss": 0.112, "step": 29475 }, { "epoch": 183.0809968847352, "grad_norm": 1.9765625, "learning_rate": 8.960845498779069e-07, "loss": 0.1679, "step": 29476 }, { "epoch": 183.08722741433021, "grad_norm": 1.203125, "learning_rate": 8.954309475460943e-07, "loss": 0.1341, "step": 29477 }, { "epoch": 183.09345794392524, "grad_norm": 2.65625, "learning_rate": 8.947775793226698e-07, "loss": 0.3045, "step": 29478 }, { "epoch": 183.09968847352025, "grad_norm": 2.78125, "learning_rate": 8.941244452139813e-07, "loss": 0.0988, "step": 29479 }, { "epoch": 183.10591900311528, "grad_norm": 2.0625, "learning_rate": 8.93471545226382e-07, "loss": 0.1594, "step": 29480 }, { "epoch": 183.11214953271028, "grad_norm": 1.234375, "learning_rate": 8.928188793662029e-07, "loss": 0.1114, "step": 29481 }, { "epoch": 183.1183800623053, "grad_norm": 2.4375, "learning_rate": 8.921664476397834e-07, "loss": 0.1793, "step": 29482 }, { "epoch": 183.1246105919003, "grad_norm": 1.1640625, "learning_rate": 8.915142500534629e-07, "loss": 0.087, "step": 29483 }, { "epoch": 183.13084112149534, "grad_norm": 1.546875, "learning_rate": 8.90862286613578e-07, "loss": 0.136, "step": 29484 }, { "epoch": 183.13707165109034, "grad_norm": 1.5859375, "learning_rate": 8.902105573264569e-07, "loss": 0.1383, "step": 29485 }, { "epoch": 183.14330218068537, "grad_norm": 2.90625, "learning_rate": 8.895590621984279e-07, "loss": 0.2215, "step": 29486 }, { "epoch": 183.14953271028037, "grad_norm": 1.890625, "learning_rate": 8.889078012358249e-07, "loss": 0.1802, "step": 29487 }, { "epoch": 183.1557632398754, "grad_norm": 1.8515625, "learning_rate": 8.882567744449677e-07, "loss": 0.1189, "step": 29488 }, { "epoch": 183.1619937694704, "grad_norm": 2.703125, "learning_rate": 8.876059818321791e-07, "loss": 0.2141, "step": 29489 }, { "epoch": 183.16822429906543, "grad_norm": 1.890625, "learning_rate": 8.869554234037791e-07, "loss": 0.1149, "step": 29490 }, { "epoch": 183.17445482866043, "grad_norm": 1.578125, "learning_rate": 8.863050991660932e-07, "loss": 0.1277, "step": 29491 }, { "epoch": 183.18068535825546, "grad_norm": 1.796875, "learning_rate": 8.856550091254301e-07, "loss": 0.1714, "step": 29492 }, { "epoch": 183.18691588785046, "grad_norm": 1.953125, "learning_rate": 8.850051532881071e-07, "loss": 0.1493, "step": 29493 }, { "epoch": 183.1931464174455, "grad_norm": 2.5, "learning_rate": 8.84355531660433e-07, "loss": 0.3396, "step": 29494 }, { "epoch": 183.1993769470405, "grad_norm": 2.59375, "learning_rate": 8.837061442487193e-07, "loss": 0.1751, "step": 29495 }, { "epoch": 183.20560747663552, "grad_norm": 2.984375, "learning_rate": 8.830569910592696e-07, "loss": 0.2892, "step": 29496 }, { "epoch": 183.21183800623052, "grad_norm": 1.9765625, "learning_rate": 8.824080720983924e-07, "loss": 0.2387, "step": 29497 }, { "epoch": 183.21806853582555, "grad_norm": 2.515625, "learning_rate": 8.817593873723912e-07, "loss": 0.1921, "step": 29498 }, { "epoch": 183.22429906542055, "grad_norm": 1.5390625, "learning_rate": 8.811109368875664e-07, "loss": 0.1412, "step": 29499 }, { "epoch": 183.23052959501558, "grad_norm": 3.21875, "learning_rate": 8.804627206502075e-07, "loss": 0.36, "step": 29500 }, { "epoch": 183.23676012461058, "grad_norm": 2.640625, "learning_rate": 8.798147386666178e-07, "loss": 0.1625, "step": 29501 }, { "epoch": 183.2429906542056, "grad_norm": 1.515625, "learning_rate": 8.791669909430894e-07, "loss": 0.0919, "step": 29502 }, { "epoch": 183.2492211838006, "grad_norm": 1.1484375, "learning_rate": 8.785194774859146e-07, "loss": 0.0925, "step": 29503 }, { "epoch": 183.25545171339564, "grad_norm": 1.0, "learning_rate": 8.778721983013771e-07, "loss": 0.0856, "step": 29504 }, { "epoch": 183.26168224299064, "grad_norm": 2.625, "learning_rate": 8.772251533957693e-07, "loss": 0.233, "step": 29505 }, { "epoch": 183.26791277258567, "grad_norm": 1.1875, "learning_rate": 8.765783427753721e-07, "loss": 0.0985, "step": 29506 }, { "epoch": 183.27414330218068, "grad_norm": 1.0390625, "learning_rate": 8.759317664464695e-07, "loss": 0.0865, "step": 29507 }, { "epoch": 183.2803738317757, "grad_norm": 1.8359375, "learning_rate": 8.752854244153369e-07, "loss": 0.1438, "step": 29508 }, { "epoch": 183.2866043613707, "grad_norm": 1.7265625, "learning_rate": 8.746393166882555e-07, "loss": 0.2536, "step": 29509 }, { "epoch": 183.29283489096574, "grad_norm": 2.328125, "learning_rate": 8.739934432715035e-07, "loss": 0.1799, "step": 29510 }, { "epoch": 183.29906542056074, "grad_norm": 1.640625, "learning_rate": 8.733478041713455e-07, "loss": 0.1081, "step": 29511 }, { "epoch": 183.30529595015577, "grad_norm": 1.2265625, "learning_rate": 8.727023993940598e-07, "loss": 0.0977, "step": 29512 }, { "epoch": 183.31152647975077, "grad_norm": 2.5, "learning_rate": 8.720572289459078e-07, "loss": 0.2188, "step": 29513 }, { "epoch": 183.3177570093458, "grad_norm": 2.515625, "learning_rate": 8.714122928331625e-07, "loss": 0.1748, "step": 29514 }, { "epoch": 183.3239875389408, "grad_norm": 1.4453125, "learning_rate": 8.7076759106208e-07, "loss": 0.1076, "step": 29515 }, { "epoch": 183.33021806853583, "grad_norm": 1.796875, "learning_rate": 8.701231236389274e-07, "loss": 0.136, "step": 29516 }, { "epoch": 183.33644859813083, "grad_norm": 1.8046875, "learning_rate": 8.694788905699663e-07, "loss": 0.1329, "step": 29517 }, { "epoch": 183.34267912772586, "grad_norm": 2.671875, "learning_rate": 8.688348918614475e-07, "loss": 0.2296, "step": 29518 }, { "epoch": 183.34890965732086, "grad_norm": 2.8125, "learning_rate": 8.68191127519627e-07, "loss": 0.1832, "step": 29519 }, { "epoch": 183.3551401869159, "grad_norm": 2.6875, "learning_rate": 8.67547597550758e-07, "loss": 0.2659, "step": 29520 }, { "epoch": 183.3613707165109, "grad_norm": 2.734375, "learning_rate": 8.669043019610912e-07, "loss": 0.197, "step": 29521 }, { "epoch": 183.36760124610592, "grad_norm": 2.375, "learning_rate": 8.662612407568743e-07, "loss": 0.1882, "step": 29522 }, { "epoch": 183.37383177570092, "grad_norm": 3.28125, "learning_rate": 8.656184139443496e-07, "loss": 0.4208, "step": 29523 }, { "epoch": 183.38006230529595, "grad_norm": 0.88671875, "learning_rate": 8.649758215297676e-07, "loss": 0.0942, "step": 29524 }, { "epoch": 183.38629283489095, "grad_norm": 1.765625, "learning_rate": 8.643334635193651e-07, "loss": 0.1429, "step": 29525 }, { "epoch": 183.39252336448598, "grad_norm": 2.546875, "learning_rate": 8.636913399193786e-07, "loss": 0.2514, "step": 29526 }, { "epoch": 183.39875389408098, "grad_norm": 1.21875, "learning_rate": 8.630494507360448e-07, "loss": 0.1176, "step": 29527 }, { "epoch": 183.404984423676, "grad_norm": 2.53125, "learning_rate": 8.624077959756032e-07, "loss": 0.2757, "step": 29528 }, { "epoch": 183.41121495327104, "grad_norm": 1.96875, "learning_rate": 8.617663756442796e-07, "loss": 0.1471, "step": 29529 }, { "epoch": 183.41744548286604, "grad_norm": 1.609375, "learning_rate": 8.611251897483047e-07, "loss": 0.1474, "step": 29530 }, { "epoch": 183.42367601246107, "grad_norm": 3.109375, "learning_rate": 8.604842382939099e-07, "loss": 0.2707, "step": 29531 }, { "epoch": 183.42990654205607, "grad_norm": 2.703125, "learning_rate": 8.598435212873207e-07, "loss": 0.1159, "step": 29532 }, { "epoch": 183.4361370716511, "grad_norm": 1.6328125, "learning_rate": 8.592030387347516e-07, "loss": 0.1077, "step": 29533 }, { "epoch": 183.4423676012461, "grad_norm": 2.109375, "learning_rate": 8.585627906424282e-07, "loss": 0.1594, "step": 29534 }, { "epoch": 183.44859813084113, "grad_norm": 1.53125, "learning_rate": 8.579227770165677e-07, "loss": 0.1148, "step": 29535 }, { "epoch": 183.45482866043614, "grad_norm": 2.71875, "learning_rate": 8.572829978633928e-07, "loss": 0.1951, "step": 29536 }, { "epoch": 183.46105919003116, "grad_norm": 1.125, "learning_rate": 8.566434531891072e-07, "loss": 0.1152, "step": 29537 }, { "epoch": 183.46728971962617, "grad_norm": 1.6796875, "learning_rate": 8.560041429999277e-07, "loss": 0.1403, "step": 29538 }, { "epoch": 183.4735202492212, "grad_norm": 2.5625, "learning_rate": 8.553650673020608e-07, "loss": 0.1515, "step": 29539 }, { "epoch": 183.4797507788162, "grad_norm": 1.3046875, "learning_rate": 8.547262261017181e-07, "loss": 0.1454, "step": 29540 }, { "epoch": 183.48598130841123, "grad_norm": 1.6875, "learning_rate": 8.540876194050973e-07, "loss": 0.1437, "step": 29541 }, { "epoch": 183.49221183800623, "grad_norm": 2.125, "learning_rate": 8.534492472184047e-07, "loss": 0.1209, "step": 29542 }, { "epoch": 183.49844236760126, "grad_norm": 2.421875, "learning_rate": 8.528111095478436e-07, "loss": 0.3032, "step": 29543 }, { "epoch": 183.50467289719626, "grad_norm": 1.5546875, "learning_rate": 8.521732063996063e-07, "loss": 0.1009, "step": 29544 }, { "epoch": 183.5109034267913, "grad_norm": 2.78125, "learning_rate": 8.515355377798878e-07, "loss": 0.2388, "step": 29545 }, { "epoch": 183.5171339563863, "grad_norm": 2.265625, "learning_rate": 8.508981036948833e-07, "loss": 0.1853, "step": 29546 }, { "epoch": 183.52336448598132, "grad_norm": 2.375, "learning_rate": 8.502609041507848e-07, "loss": 0.1712, "step": 29547 }, { "epoch": 183.52959501557632, "grad_norm": 1.03125, "learning_rate": 8.496239391537792e-07, "loss": 0.1058, "step": 29548 }, { "epoch": 183.53582554517135, "grad_norm": 1.90625, "learning_rate": 8.489872087100531e-07, "loss": 0.1919, "step": 29549 }, { "epoch": 183.54205607476635, "grad_norm": 2.375, "learning_rate": 8.483507128257934e-07, "loss": 0.2198, "step": 29550 }, { "epoch": 183.54828660436138, "grad_norm": 1.4453125, "learning_rate": 8.47714451507181e-07, "loss": 0.1393, "step": 29551 }, { "epoch": 183.55451713395638, "grad_norm": 2.0625, "learning_rate": 8.470784247603891e-07, "loss": 0.201, "step": 29552 }, { "epoch": 183.5607476635514, "grad_norm": 2.09375, "learning_rate": 8.464426325915986e-07, "loss": 0.2011, "step": 29553 }, { "epoch": 183.5669781931464, "grad_norm": 3.0, "learning_rate": 8.458070750069907e-07, "loss": 0.3395, "step": 29554 }, { "epoch": 183.57320872274144, "grad_norm": 2.640625, "learning_rate": 8.451717520127273e-07, "loss": 0.1997, "step": 29555 }, { "epoch": 183.57943925233644, "grad_norm": 2.453125, "learning_rate": 8.445366636149865e-07, "loss": 0.1755, "step": 29556 }, { "epoch": 183.58566978193147, "grad_norm": 1.1796875, "learning_rate": 8.439018098199358e-07, "loss": 0.1151, "step": 29557 }, { "epoch": 183.59190031152647, "grad_norm": 1.8046875, "learning_rate": 8.432671906337425e-07, "loss": 0.1493, "step": 29558 }, { "epoch": 183.5981308411215, "grad_norm": 1.7890625, "learning_rate": 8.426328060625599e-07, "loss": 0.1372, "step": 29559 }, { "epoch": 183.6043613707165, "grad_norm": 2.796875, "learning_rate": 8.419986561125581e-07, "loss": 0.3143, "step": 29560 }, { "epoch": 183.61059190031153, "grad_norm": 1.359375, "learning_rate": 8.413647407898989e-07, "loss": 0.1245, "step": 29561 }, { "epoch": 183.61682242990653, "grad_norm": 2.359375, "learning_rate": 8.407310601007301e-07, "loss": 0.1544, "step": 29562 }, { "epoch": 183.62305295950156, "grad_norm": 2.8125, "learning_rate": 8.400976140512135e-07, "loss": 0.2846, "step": 29563 }, { "epoch": 183.62928348909657, "grad_norm": 2.75, "learning_rate": 8.394644026474968e-07, "loss": 0.3273, "step": 29564 }, { "epoch": 183.6355140186916, "grad_norm": 2.5, "learning_rate": 8.388314258957336e-07, "loss": 0.261, "step": 29565 }, { "epoch": 183.6417445482866, "grad_norm": 1.7421875, "learning_rate": 8.381986838020662e-07, "loss": 0.1488, "step": 29566 }, { "epoch": 183.64797507788163, "grad_norm": 1.3125, "learning_rate": 8.375661763726422e-07, "loss": 0.123, "step": 29567 }, { "epoch": 183.65420560747663, "grad_norm": 1.3125, "learning_rate": 8.36933903613607e-07, "loss": 0.1702, "step": 29568 }, { "epoch": 183.66043613707166, "grad_norm": 1.9296875, "learning_rate": 8.363018655311028e-07, "loss": 0.1643, "step": 29569 }, { "epoch": 183.66666666666666, "grad_norm": 1.5625, "learning_rate": 8.356700621312607e-07, "loss": 0.0896, "step": 29570 }, { "epoch": 183.6728971962617, "grad_norm": 3.25, "learning_rate": 8.350384934202232e-07, "loss": 0.2595, "step": 29571 }, { "epoch": 183.6791277258567, "grad_norm": 2.390625, "learning_rate": 8.344071594041214e-07, "loss": 0.2363, "step": 29572 }, { "epoch": 183.68535825545172, "grad_norm": 1.875, "learning_rate": 8.337760600890893e-07, "loss": 0.1373, "step": 29573 }, { "epoch": 183.69158878504672, "grad_norm": 1.8359375, "learning_rate": 8.331451954812525e-07, "loss": 0.1028, "step": 29574 }, { "epoch": 183.69781931464175, "grad_norm": 2.203125, "learning_rate": 8.325145655867422e-07, "loss": 0.1932, "step": 29575 }, { "epoch": 183.70404984423675, "grad_norm": 2.75, "learning_rate": 8.318841704116842e-07, "loss": 0.2522, "step": 29576 }, { "epoch": 183.71028037383178, "grad_norm": 1.234375, "learning_rate": 8.312540099621957e-07, "loss": 0.1155, "step": 29577 }, { "epoch": 183.71651090342678, "grad_norm": 2.25, "learning_rate": 8.306240842443969e-07, "loss": 0.1371, "step": 29578 }, { "epoch": 183.7227414330218, "grad_norm": 1.921875, "learning_rate": 8.299943932644105e-07, "loss": 0.2128, "step": 29579 }, { "epoch": 183.7289719626168, "grad_norm": 1.4453125, "learning_rate": 8.293649370283513e-07, "loss": 0.1036, "step": 29580 }, { "epoch": 183.73520249221184, "grad_norm": 1.234375, "learning_rate": 8.287357155423309e-07, "loss": 0.111, "step": 29581 }, { "epoch": 183.74143302180684, "grad_norm": 1.8359375, "learning_rate": 8.281067288124611e-07, "loss": 0.0982, "step": 29582 }, { "epoch": 183.74766355140187, "grad_norm": 2.984375, "learning_rate": 8.274779768448482e-07, "loss": 0.2621, "step": 29583 }, { "epoch": 183.75389408099687, "grad_norm": 1.9609375, "learning_rate": 8.268494596456039e-07, "loss": 0.1667, "step": 29584 }, { "epoch": 183.7601246105919, "grad_norm": 1.1484375, "learning_rate": 8.262211772208289e-07, "loss": 0.1273, "step": 29585 }, { "epoch": 183.7663551401869, "grad_norm": 1.25, "learning_rate": 8.25593129576624e-07, "loss": 0.1148, "step": 29586 }, { "epoch": 183.77258566978193, "grad_norm": 2.65625, "learning_rate": 8.249653167190952e-07, "loss": 0.2419, "step": 29587 }, { "epoch": 183.77881619937693, "grad_norm": 3.09375, "learning_rate": 8.243377386543349e-07, "loss": 0.3826, "step": 29588 }, { "epoch": 183.78504672897196, "grad_norm": 2.171875, "learning_rate": 8.237103953884356e-07, "loss": 0.1319, "step": 29589 }, { "epoch": 183.79127725856696, "grad_norm": 2.96875, "learning_rate": 8.230832869274952e-07, "loss": 0.2729, "step": 29590 }, { "epoch": 183.797507788162, "grad_norm": 2.625, "learning_rate": 8.224564132776058e-07, "loss": 0.2289, "step": 29591 }, { "epoch": 183.80373831775702, "grad_norm": 1.59375, "learning_rate": 8.218297744448489e-07, "loss": 0.1267, "step": 29592 }, { "epoch": 183.80996884735202, "grad_norm": 2.125, "learning_rate": 8.21203370435314e-07, "loss": 0.1036, "step": 29593 }, { "epoch": 183.81619937694705, "grad_norm": 1.4140625, "learning_rate": 8.205772012550849e-07, "loss": 0.1077, "step": 29594 }, { "epoch": 183.82242990654206, "grad_norm": 2.46875, "learning_rate": 8.199512669102516e-07, "loss": 0.1796, "step": 29595 }, { "epoch": 183.82866043613708, "grad_norm": 1.109375, "learning_rate": 8.193255674068783e-07, "loss": 0.1071, "step": 29596 }, { "epoch": 183.8348909657321, "grad_norm": 2.3125, "learning_rate": 8.187001027510466e-07, "loss": 0.1781, "step": 29597 }, { "epoch": 183.84112149532712, "grad_norm": 1.1015625, "learning_rate": 8.180748729488346e-07, "loss": 0.1015, "step": 29598 }, { "epoch": 183.84735202492212, "grad_norm": 2.265625, "learning_rate": 8.174498780063156e-07, "loss": 0.169, "step": 29599 }, { "epoch": 183.85358255451715, "grad_norm": 1.5703125, "learning_rate": 8.168251179295566e-07, "loss": 0.1186, "step": 29600 }, { "epoch": 183.85981308411215, "grad_norm": 1.2890625, "learning_rate": 8.16200592724628e-07, "loss": 0.1039, "step": 29601 }, { "epoch": 183.86604361370718, "grad_norm": 2.453125, "learning_rate": 8.155763023975888e-07, "loss": 0.3145, "step": 29602 }, { "epoch": 183.87227414330218, "grad_norm": 2.515625, "learning_rate": 8.149522469545118e-07, "loss": 0.1557, "step": 29603 }, { "epoch": 183.8785046728972, "grad_norm": 2.9375, "learning_rate": 8.143284264014506e-07, "loss": 0.1842, "step": 29604 }, { "epoch": 183.8847352024922, "grad_norm": 2.859375, "learning_rate": 8.137048407444642e-07, "loss": 0.2621, "step": 29605 }, { "epoch": 183.89096573208724, "grad_norm": 1.796875, "learning_rate": 8.130814899896172e-07, "loss": 0.1139, "step": 29606 }, { "epoch": 183.89719626168224, "grad_norm": 1.859375, "learning_rate": 8.12458374142952e-07, "loss": 0.2356, "step": 29607 }, { "epoch": 183.90342679127727, "grad_norm": 2.171875, "learning_rate": 8.118354932105305e-07, "loss": 0.1465, "step": 29608 }, { "epoch": 183.90965732087227, "grad_norm": 2.28125, "learning_rate": 8.11212847198392e-07, "loss": 0.2182, "step": 29609 }, { "epoch": 183.9158878504673, "grad_norm": 1.796875, "learning_rate": 8.10590436112596e-07, "loss": 0.116, "step": 29610 }, { "epoch": 183.9221183800623, "grad_norm": 1.265625, "learning_rate": 8.099682599591762e-07, "loss": 0.0969, "step": 29611 }, { "epoch": 183.92834890965733, "grad_norm": 2.90625, "learning_rate": 8.093463187441807e-07, "loss": 0.2131, "step": 29612 }, { "epoch": 183.93457943925233, "grad_norm": 2.09375, "learning_rate": 8.087246124736519e-07, "loss": 0.1733, "step": 29613 }, { "epoch": 183.94080996884736, "grad_norm": 1.9375, "learning_rate": 8.081031411536266e-07, "loss": 0.1882, "step": 29614 }, { "epoch": 183.94704049844236, "grad_norm": 2.25, "learning_rate": 8.07481904790136e-07, "loss": 0.2526, "step": 29615 }, { "epoch": 183.9532710280374, "grad_norm": 3.046875, "learning_rate": 8.068609033892172e-07, "loss": 0.3135, "step": 29616 }, { "epoch": 183.9595015576324, "grad_norm": 2.765625, "learning_rate": 8.062401369569039e-07, "loss": 0.2228, "step": 29617 }, { "epoch": 183.96573208722742, "grad_norm": 2.3125, "learning_rate": 8.056196054992193e-07, "loss": 0.2134, "step": 29618 }, { "epoch": 183.97196261682242, "grad_norm": 2.46875, "learning_rate": 8.049993090221946e-07, "loss": 0.1784, "step": 29619 }, { "epoch": 183.97819314641745, "grad_norm": 2.25, "learning_rate": 8.043792475318557e-07, "loss": 0.1215, "step": 29620 }, { "epoch": 183.98442367601245, "grad_norm": 2.96875, "learning_rate": 8.037594210342197e-07, "loss": 0.2773, "step": 29621 }, { "epoch": 183.99065420560748, "grad_norm": 1.90625, "learning_rate": 8.031398295353098e-07, "loss": 0.181, "step": 29622 }, { "epoch": 183.99688473520249, "grad_norm": 1.6484375, "learning_rate": 8.025204730411407e-07, "loss": 0.117, "step": 29623 }, { "epoch": 184.0, "grad_norm": 2.453125, "learning_rate": 8.019013515577323e-07, "loss": 0.1269, "step": 29624 }, { "epoch": 184.00623052959503, "grad_norm": 1.96875, "learning_rate": 8.012824650910938e-07, "loss": 0.1728, "step": 29625 }, { "epoch": 184.01246105919003, "grad_norm": 3.0, "learning_rate": 8.006638136472372e-07, "loss": 0.3274, "step": 29626 }, { "epoch": 184.01869158878506, "grad_norm": 2.421875, "learning_rate": 8.000453972321714e-07, "loss": 0.1548, "step": 29627 }, { "epoch": 184.02492211838006, "grad_norm": 1.5234375, "learning_rate": 7.994272158519056e-07, "loss": 0.1381, "step": 29628 }, { "epoch": 184.0311526479751, "grad_norm": 2.8125, "learning_rate": 7.98809269512435e-07, "loss": 0.1596, "step": 29629 }, { "epoch": 184.0373831775701, "grad_norm": 2.59375, "learning_rate": 7.981915582197658e-07, "loss": 0.1315, "step": 29630 }, { "epoch": 184.04361370716512, "grad_norm": 2.796875, "learning_rate": 7.975740819798988e-07, "loss": 0.366, "step": 29631 }, { "epoch": 184.04984423676012, "grad_norm": 2.203125, "learning_rate": 7.969568407988348e-07, "loss": 0.1908, "step": 29632 }, { "epoch": 184.05607476635515, "grad_norm": 2.859375, "learning_rate": 7.963398346825635e-07, "loss": 0.117, "step": 29633 }, { "epoch": 184.06230529595015, "grad_norm": 2.859375, "learning_rate": 7.957230636370744e-07, "loss": 0.2716, "step": 29634 }, { "epoch": 184.06853582554518, "grad_norm": 2.75, "learning_rate": 7.9510652766836e-07, "loss": 0.2228, "step": 29635 }, { "epoch": 184.07476635514018, "grad_norm": 2.765625, "learning_rate": 7.944902267824155e-07, "loss": 0.1992, "step": 29636 }, { "epoch": 184.0809968847352, "grad_norm": 1.6171875, "learning_rate": 7.938741609852141e-07, "loss": 0.1746, "step": 29637 }, { "epoch": 184.08722741433021, "grad_norm": 1.875, "learning_rate": 7.932583302827479e-07, "loss": 0.1041, "step": 29638 }, { "epoch": 184.09345794392524, "grad_norm": 2.765625, "learning_rate": 7.926427346809983e-07, "loss": 0.1635, "step": 29639 }, { "epoch": 184.09968847352025, "grad_norm": 2.046875, "learning_rate": 7.920273741859413e-07, "loss": 0.1538, "step": 29640 }, { "epoch": 184.10591900311528, "grad_norm": 2.859375, "learning_rate": 7.914122488035497e-07, "loss": 0.3537, "step": 29641 }, { "epoch": 184.11214953271028, "grad_norm": 2.109375, "learning_rate": 7.907973585398021e-07, "loss": 0.0859, "step": 29642 }, { "epoch": 184.1183800623053, "grad_norm": 1.453125, "learning_rate": 7.901827034006742e-07, "loss": 0.089, "step": 29643 }, { "epoch": 184.1246105919003, "grad_norm": 1.90625, "learning_rate": 7.895682833921253e-07, "loss": 0.1749, "step": 29644 }, { "epoch": 184.13084112149534, "grad_norm": 2.375, "learning_rate": 7.88954098520131e-07, "loss": 0.2273, "step": 29645 }, { "epoch": 184.13707165109034, "grad_norm": 1.046875, "learning_rate": 7.883401487906561e-07, "loss": 0.1213, "step": 29646 }, { "epoch": 184.14330218068537, "grad_norm": 2.0625, "learning_rate": 7.877264342096624e-07, "loss": 0.131, "step": 29647 }, { "epoch": 184.14953271028037, "grad_norm": 1.8359375, "learning_rate": 7.871129547831063e-07, "loss": 0.1717, "step": 29648 }, { "epoch": 184.1557632398754, "grad_norm": 2.671875, "learning_rate": 7.864997105169497e-07, "loss": 0.2086, "step": 29649 }, { "epoch": 184.1619937694704, "grad_norm": 1.8671875, "learning_rate": 7.858867014171517e-07, "loss": 0.1437, "step": 29650 }, { "epoch": 184.16822429906543, "grad_norm": 1.3984375, "learning_rate": 7.852739274896576e-07, "loss": 0.1139, "step": 29651 }, { "epoch": 184.17445482866043, "grad_norm": 1.9921875, "learning_rate": 7.846613887404264e-07, "loss": 0.1499, "step": 29652 }, { "epoch": 184.18068535825546, "grad_norm": 1.7578125, "learning_rate": 7.840490851754035e-07, "loss": 0.1515, "step": 29653 }, { "epoch": 184.18691588785046, "grad_norm": 2.46875, "learning_rate": 7.834370168005367e-07, "loss": 0.2759, "step": 29654 }, { "epoch": 184.1931464174455, "grad_norm": 2.140625, "learning_rate": 7.828251836217687e-07, "loss": 0.2221, "step": 29655 }, { "epoch": 184.1993769470405, "grad_norm": 1.6171875, "learning_rate": 7.822135856450446e-07, "loss": 0.1295, "step": 29656 }, { "epoch": 184.20560747663552, "grad_norm": 1.984375, "learning_rate": 7.816022228763043e-07, "loss": 0.1354, "step": 29657 }, { "epoch": 184.21183800623052, "grad_norm": 2.46875, "learning_rate": 7.809910953214816e-07, "loss": 0.138, "step": 29658 }, { "epoch": 184.21806853582555, "grad_norm": 0.73828125, "learning_rate": 7.803802029865193e-07, "loss": 0.0934, "step": 29659 }, { "epoch": 184.22429906542055, "grad_norm": 1.515625, "learning_rate": 7.79769545877343e-07, "loss": 0.1394, "step": 29660 }, { "epoch": 184.23052959501558, "grad_norm": 3.453125, "learning_rate": 7.791591239998869e-07, "loss": 0.2602, "step": 29661 }, { "epoch": 184.23676012461058, "grad_norm": 1.421875, "learning_rate": 7.785489373600769e-07, "loss": 0.1326, "step": 29662 }, { "epoch": 184.2429906542056, "grad_norm": 3.484375, "learning_rate": 7.779389859638414e-07, "loss": 0.1138, "step": 29663 }, { "epoch": 184.2492211838006, "grad_norm": 2.71875, "learning_rate": 7.773292698171065e-07, "loss": 0.2343, "step": 29664 }, { "epoch": 184.25545171339564, "grad_norm": 0.8984375, "learning_rate": 7.767197889257948e-07, "loss": 0.1211, "step": 29665 }, { "epoch": 184.26168224299064, "grad_norm": 1.9140625, "learning_rate": 7.761105432958187e-07, "loss": 0.1676, "step": 29666 }, { "epoch": 184.26791277258567, "grad_norm": 1.5859375, "learning_rate": 7.755015329330984e-07, "loss": 0.1348, "step": 29667 }, { "epoch": 184.27414330218068, "grad_norm": 2.046875, "learning_rate": 7.748927578435511e-07, "loss": 0.2487, "step": 29668 }, { "epoch": 184.2803738317757, "grad_norm": 2.125, "learning_rate": 7.742842180330889e-07, "loss": 0.0956, "step": 29669 }, { "epoch": 184.2866043613707, "grad_norm": 2.21875, "learning_rate": 7.736759135076182e-07, "loss": 0.2688, "step": 29670 }, { "epoch": 184.29283489096574, "grad_norm": 1.9296875, "learning_rate": 7.730678442730538e-07, "loss": 0.1443, "step": 29671 }, { "epoch": 184.29906542056074, "grad_norm": 1.734375, "learning_rate": 7.724600103352936e-07, "loss": 0.1621, "step": 29672 }, { "epoch": 184.30529595015577, "grad_norm": 2.0625, "learning_rate": 7.718524117002468e-07, "loss": 0.1662, "step": 29673 }, { "epoch": 184.31152647975077, "grad_norm": 3.0, "learning_rate": 7.712450483738115e-07, "loss": 0.1998, "step": 29674 }, { "epoch": 184.3177570093458, "grad_norm": 1.0390625, "learning_rate": 7.706379203618857e-07, "loss": 0.0926, "step": 29675 }, { "epoch": 184.3239875389408, "grad_norm": 1.984375, "learning_rate": 7.700310276703704e-07, "loss": 0.098, "step": 29676 }, { "epoch": 184.33021806853583, "grad_norm": 2.625, "learning_rate": 7.694243703051579e-07, "loss": 0.2469, "step": 29677 }, { "epoch": 184.33644859813083, "grad_norm": 1.5, "learning_rate": 7.688179482721381e-07, "loss": 0.1136, "step": 29678 }, { "epoch": 184.34267912772586, "grad_norm": 0.76171875, "learning_rate": 7.682117615772005e-07, "loss": 0.0815, "step": 29679 }, { "epoch": 184.34890965732086, "grad_norm": 0.7578125, "learning_rate": 7.676058102262379e-07, "loss": 0.1021, "step": 29680 }, { "epoch": 184.3551401869159, "grad_norm": 2.5625, "learning_rate": 7.670000942251287e-07, "loss": 0.2361, "step": 29681 }, { "epoch": 184.3613707165109, "grad_norm": 2.328125, "learning_rate": 7.663946135797601e-07, "loss": 0.1549, "step": 29682 }, { "epoch": 184.36760124610592, "grad_norm": 1.6328125, "learning_rate": 7.657893682960104e-07, "loss": 0.1296, "step": 29683 }, { "epoch": 184.37383177570092, "grad_norm": 1.28125, "learning_rate": 7.651843583797613e-07, "loss": 0.1188, "step": 29684 }, { "epoch": 184.38006230529595, "grad_norm": 1.6171875, "learning_rate": 7.64579583836883e-07, "loss": 0.1181, "step": 29685 }, { "epoch": 184.38629283489095, "grad_norm": 1.0234375, "learning_rate": 7.639750446732513e-07, "loss": 0.0904, "step": 29686 }, { "epoch": 184.39252336448598, "grad_norm": 1.59375, "learning_rate": 7.633707408947421e-07, "loss": 0.1197, "step": 29687 }, { "epoch": 184.39875389408098, "grad_norm": 1.328125, "learning_rate": 7.627666725072202e-07, "loss": 0.106, "step": 29688 }, { "epoch": 184.404984423676, "grad_norm": 1.25, "learning_rate": 7.621628395165503e-07, "loss": 0.1004, "step": 29689 }, { "epoch": 184.41121495327104, "grad_norm": 1.6328125, "learning_rate": 7.615592419286055e-07, "loss": 0.1539, "step": 29690 }, { "epoch": 184.41744548286604, "grad_norm": 1.9453125, "learning_rate": 7.609558797492394e-07, "loss": 0.143, "step": 29691 }, { "epoch": 184.42367601246107, "grad_norm": 1.0546875, "learning_rate": 7.603527529843141e-07, "loss": 0.0917, "step": 29692 }, { "epoch": 184.42990654205607, "grad_norm": 1.8125, "learning_rate": 7.597498616396886e-07, "loss": 0.1889, "step": 29693 }, { "epoch": 184.4361370716511, "grad_norm": 1.875, "learning_rate": 7.591472057212168e-07, "loss": 0.116, "step": 29694 }, { "epoch": 184.4423676012461, "grad_norm": 2.265625, "learning_rate": 7.585447852347577e-07, "loss": 0.2012, "step": 29695 }, { "epoch": 184.44859813084113, "grad_norm": 2.0625, "learning_rate": 7.579426001861512e-07, "loss": 0.1325, "step": 29696 }, { "epoch": 184.45482866043614, "grad_norm": 2.203125, "learning_rate": 7.573406505812591e-07, "loss": 0.2243, "step": 29697 }, { "epoch": 184.46105919003116, "grad_norm": 1.4609375, "learning_rate": 7.567389364259131e-07, "loss": 0.1252, "step": 29698 }, { "epoch": 184.46728971962617, "grad_norm": 1.5625, "learning_rate": 7.561374577259722e-07, "loss": 0.1105, "step": 29699 }, { "epoch": 184.4735202492212, "grad_norm": 2.28125, "learning_rate": 7.555362144872652e-07, "loss": 0.1982, "step": 29700 }, { "epoch": 184.4797507788162, "grad_norm": 1.9765625, "learning_rate": 7.549352067156374e-07, "loss": 0.1936, "step": 29701 }, { "epoch": 184.48598130841123, "grad_norm": 1.625, "learning_rate": 7.543344344169285e-07, "loss": 0.1012, "step": 29702 }, { "epoch": 184.49221183800623, "grad_norm": 2.546875, "learning_rate": 7.537338975969671e-07, "loss": 0.1887, "step": 29703 }, { "epoch": 184.49844236760126, "grad_norm": 1.1484375, "learning_rate": 7.531335962615904e-07, "loss": 0.1187, "step": 29704 }, { "epoch": 184.50467289719626, "grad_norm": 2.390625, "learning_rate": 7.525335304166242e-07, "loss": 0.1436, "step": 29705 }, { "epoch": 184.5109034267913, "grad_norm": 2.03125, "learning_rate": 7.519337000679027e-07, "loss": 0.2091, "step": 29706 }, { "epoch": 184.5171339563863, "grad_norm": 2.65625, "learning_rate": 7.513341052212463e-07, "loss": 0.1621, "step": 29707 }, { "epoch": 184.52336448598132, "grad_norm": 1.734375, "learning_rate": 7.50734745882481e-07, "loss": 0.1126, "step": 29708 }, { "epoch": 184.52959501557632, "grad_norm": 1.3203125, "learning_rate": 7.501356220574296e-07, "loss": 0.1121, "step": 29709 }, { "epoch": 184.53582554517135, "grad_norm": 1.9296875, "learning_rate": 7.495367337519072e-07, "loss": 0.1671, "step": 29710 }, { "epoch": 184.54205607476635, "grad_norm": 1.6328125, "learning_rate": 7.489380809717284e-07, "loss": 0.1756, "step": 29711 }, { "epoch": 184.54828660436138, "grad_norm": 1.9921875, "learning_rate": 7.483396637227108e-07, "loss": 0.1784, "step": 29712 }, { "epoch": 184.55451713395638, "grad_norm": 1.828125, "learning_rate": 7.477414820106721e-07, "loss": 0.1838, "step": 29713 }, { "epoch": 184.5607476635514, "grad_norm": 2.296875, "learning_rate": 7.471435358414103e-07, "loss": 0.2038, "step": 29714 }, { "epoch": 184.5669781931464, "grad_norm": 1.5078125, "learning_rate": 7.465458252207375e-07, "loss": 0.1318, "step": 29715 }, { "epoch": 184.57320872274144, "grad_norm": 1.7265625, "learning_rate": 7.459483501544656e-07, "loss": 0.1523, "step": 29716 }, { "epoch": 184.57943925233644, "grad_norm": 2.40625, "learning_rate": 7.453511106483902e-07, "loss": 0.1315, "step": 29717 }, { "epoch": 184.58566978193147, "grad_norm": 2.296875, "learning_rate": 7.447541067083092e-07, "loss": 0.2325, "step": 29718 }, { "epoch": 184.59190031152647, "grad_norm": 2.015625, "learning_rate": 7.441573383400263e-07, "loss": 0.1322, "step": 29719 }, { "epoch": 184.5981308411215, "grad_norm": 1.6875, "learning_rate": 7.435608055493398e-07, "loss": 0.1243, "step": 29720 }, { "epoch": 184.6043613707165, "grad_norm": 2.4375, "learning_rate": 7.429645083420367e-07, "loss": 0.1609, "step": 29721 }, { "epoch": 184.61059190031153, "grad_norm": 2.03125, "learning_rate": 7.423684467239122e-07, "loss": 0.161, "step": 29722 }, { "epoch": 184.61682242990653, "grad_norm": 1.6875, "learning_rate": 7.417726207007508e-07, "loss": 0.1449, "step": 29723 }, { "epoch": 184.62305295950156, "grad_norm": 1.5625, "learning_rate": 7.411770302783477e-07, "loss": 0.1139, "step": 29724 }, { "epoch": 184.62928348909657, "grad_norm": 1.5859375, "learning_rate": 7.405816754624789e-07, "loss": 0.1252, "step": 29725 }, { "epoch": 184.6355140186916, "grad_norm": 1.0703125, "learning_rate": 7.399865562589315e-07, "loss": 0.0832, "step": 29726 }, { "epoch": 184.6417445482866, "grad_norm": 2.734375, "learning_rate": 7.393916726734812e-07, "loss": 0.2087, "step": 29727 }, { "epoch": 184.64797507788163, "grad_norm": 1.9375, "learning_rate": 7.387970247119152e-07, "loss": 0.1465, "step": 29728 }, { "epoch": 184.65420560747663, "grad_norm": 1.625, "learning_rate": 7.382026123799984e-07, "loss": 0.1213, "step": 29729 }, { "epoch": 184.66043613707166, "grad_norm": 3.515625, "learning_rate": 7.376084356835067e-07, "loss": 0.2436, "step": 29730 }, { "epoch": 184.66666666666666, "grad_norm": 1.078125, "learning_rate": 7.370144946282132e-07, "loss": 0.1005, "step": 29731 }, { "epoch": 184.6728971962617, "grad_norm": 4.53125, "learning_rate": 7.364207892198855e-07, "loss": 0.2998, "step": 29732 }, { "epoch": 184.6791277258567, "grad_norm": 2.90625, "learning_rate": 7.358273194642884e-07, "loss": 0.2278, "step": 29733 }, { "epoch": 184.68535825545172, "grad_norm": 2.15625, "learning_rate": 7.352340853671841e-07, "loss": 0.1316, "step": 29734 }, { "epoch": 184.69158878504672, "grad_norm": 1.9765625, "learning_rate": 7.346410869343429e-07, "loss": 0.1391, "step": 29735 }, { "epoch": 184.69781931464175, "grad_norm": 1.7734375, "learning_rate": 7.340483241715157e-07, "loss": 0.1317, "step": 29736 }, { "epoch": 184.70404984423675, "grad_norm": 1.671875, "learning_rate": 7.334557970844591e-07, "loss": 0.1525, "step": 29737 }, { "epoch": 184.71028037383178, "grad_norm": 1.53125, "learning_rate": 7.328635056789323e-07, "loss": 0.1388, "step": 29738 }, { "epoch": 184.71651090342678, "grad_norm": 2.375, "learning_rate": 7.322714499606864e-07, "loss": 0.1768, "step": 29739 }, { "epoch": 184.7227414330218, "grad_norm": 1.421875, "learning_rate": 7.316796299354695e-07, "loss": 0.1639, "step": 29740 }, { "epoch": 184.7289719626168, "grad_norm": 1.375, "learning_rate": 7.310880456090352e-07, "loss": 0.1067, "step": 29741 }, { "epoch": 184.73520249221184, "grad_norm": 1.3984375, "learning_rate": 7.304966969871208e-07, "loss": 0.1134, "step": 29742 }, { "epoch": 184.74143302180684, "grad_norm": 3.171875, "learning_rate": 7.29905584075477e-07, "loss": 0.2655, "step": 29743 }, { "epoch": 184.74766355140187, "grad_norm": 3.546875, "learning_rate": 7.293147068798384e-07, "loss": 0.3199, "step": 29744 }, { "epoch": 184.75389408099687, "grad_norm": 1.5234375, "learning_rate": 7.287240654059474e-07, "loss": 0.1063, "step": 29745 }, { "epoch": 184.7601246105919, "grad_norm": 2.453125, "learning_rate": 7.28133659659544e-07, "loss": 0.2301, "step": 29746 }, { "epoch": 184.7663551401869, "grad_norm": 2.234375, "learning_rate": 7.27543489646354e-07, "loss": 0.2047, "step": 29747 }, { "epoch": 184.77258566978193, "grad_norm": 2.953125, "learning_rate": 7.269535553721174e-07, "loss": 0.2649, "step": 29748 }, { "epoch": 184.77881619937693, "grad_norm": 2.40625, "learning_rate": 7.263638568425546e-07, "loss": 0.1914, "step": 29749 }, { "epoch": 184.78504672897196, "grad_norm": 2.0625, "learning_rate": 7.257743940634054e-07, "loss": 0.0801, "step": 29750 }, { "epoch": 184.79127725856696, "grad_norm": 2.0625, "learning_rate": 7.25185167040382e-07, "loss": 0.1804, "step": 29751 }, { "epoch": 184.797507788162, "grad_norm": 2.859375, "learning_rate": 7.24596175779213e-07, "loss": 0.2291, "step": 29752 }, { "epoch": 184.80373831775702, "grad_norm": 1.40625, "learning_rate": 7.240074202856217e-07, "loss": 0.131, "step": 29753 }, { "epoch": 184.80996884735202, "grad_norm": 3.171875, "learning_rate": 7.23418900565323e-07, "loss": 0.1635, "step": 29754 }, { "epoch": 184.81619937694705, "grad_norm": 1.8046875, "learning_rate": 7.22830616624029e-07, "loss": 0.1401, "step": 29755 }, { "epoch": 184.82242990654206, "grad_norm": 1.6953125, "learning_rate": 7.222425684674571e-07, "loss": 0.1281, "step": 29756 }, { "epoch": 184.82866043613708, "grad_norm": 1.1640625, "learning_rate": 7.216547561013226e-07, "loss": 0.1196, "step": 29757 }, { "epoch": 184.8348909657321, "grad_norm": 2.1875, "learning_rate": 7.210671795313262e-07, "loss": 0.1836, "step": 29758 }, { "epoch": 184.84112149532712, "grad_norm": 2.03125, "learning_rate": 7.204798387631773e-07, "loss": 0.1548, "step": 29759 }, { "epoch": 184.84735202492212, "grad_norm": 1.2109375, "learning_rate": 7.19892733802588e-07, "loss": 0.0841, "step": 29760 }, { "epoch": 184.85358255451715, "grad_norm": 2.5625, "learning_rate": 7.193058646552508e-07, "loss": 0.2157, "step": 29761 }, { "epoch": 184.85981308411215, "grad_norm": 1.84375, "learning_rate": 7.18719231326867e-07, "loss": 0.1073, "step": 29762 }, { "epoch": 184.86604361370718, "grad_norm": 1.9296875, "learning_rate": 7.181328338231375e-07, "loss": 0.1704, "step": 29763 }, { "epoch": 184.87227414330218, "grad_norm": 2.390625, "learning_rate": 7.17546672149752e-07, "loss": 0.1218, "step": 29764 }, { "epoch": 184.8785046728972, "grad_norm": 1.4921875, "learning_rate": 7.169607463124117e-07, "loss": 0.12, "step": 29765 }, { "epoch": 184.8847352024922, "grad_norm": 1.6171875, "learning_rate": 7.163750563168009e-07, "loss": 0.135, "step": 29766 }, { "epoch": 184.89096573208724, "grad_norm": 2.703125, "learning_rate": 7.157896021686122e-07, "loss": 0.2122, "step": 29767 }, { "epoch": 184.89719626168224, "grad_norm": 2.390625, "learning_rate": 7.152043838735273e-07, "loss": 0.1467, "step": 29768 }, { "epoch": 184.90342679127727, "grad_norm": 2.625, "learning_rate": 7.146194014372332e-07, "loss": 0.1208, "step": 29769 }, { "epoch": 184.90965732087227, "grad_norm": 2.453125, "learning_rate": 7.140346548654059e-07, "loss": 0.1797, "step": 29770 }, { "epoch": 184.9158878504673, "grad_norm": 2.875, "learning_rate": 7.134501441637325e-07, "loss": 0.2421, "step": 29771 }, { "epoch": 184.9221183800623, "grad_norm": 1.4921875, "learning_rate": 7.128658693378865e-07, "loss": 0.1224, "step": 29772 }, { "epoch": 184.92834890965733, "grad_norm": 2.578125, "learning_rate": 7.122818303935408e-07, "loss": 0.142, "step": 29773 }, { "epoch": 184.93457943925233, "grad_norm": 1.359375, "learning_rate": 7.116980273363688e-07, "loss": 0.143, "step": 29774 }, { "epoch": 184.94080996884736, "grad_norm": 2.125, "learning_rate": 7.111144601720382e-07, "loss": 0.231, "step": 29775 }, { "epoch": 184.94704049844236, "grad_norm": 2.640625, "learning_rate": 7.105311289062249e-07, "loss": 0.153, "step": 29776 }, { "epoch": 184.9532710280374, "grad_norm": 3.0, "learning_rate": 7.099480335445829e-07, "loss": 0.214, "step": 29777 }, { "epoch": 184.9595015576324, "grad_norm": 2.65625, "learning_rate": 7.093651740927825e-07, "loss": 0.1087, "step": 29778 }, { "epoch": 184.96573208722742, "grad_norm": 2.390625, "learning_rate": 7.08782550556486e-07, "loss": 0.1195, "step": 29779 }, { "epoch": 184.97196261682242, "grad_norm": 1.5703125, "learning_rate": 7.082001629413499e-07, "loss": 0.0973, "step": 29780 }, { "epoch": 184.97819314641745, "grad_norm": 3.265625, "learning_rate": 7.076180112530251e-07, "loss": 0.3051, "step": 29781 }, { "epoch": 184.98442367601245, "grad_norm": 2.015625, "learning_rate": 7.070360954971711e-07, "loss": 0.1457, "step": 29782 }, { "epoch": 184.99065420560748, "grad_norm": 1.953125, "learning_rate": 7.064544156794417e-07, "loss": 0.1537, "step": 29783 }, { "epoch": 184.99688473520249, "grad_norm": 1.8828125, "learning_rate": 7.058729718054796e-07, "loss": 0.155, "step": 29784 }, { "epoch": 185.0, "grad_norm": 0.71875, "learning_rate": 7.052917638809359e-07, "loss": 0.0431, "step": 29785 }, { "epoch": 185.00623052959503, "grad_norm": 1.25, "learning_rate": 7.047107919114588e-07, "loss": 0.1313, "step": 29786 }, { "epoch": 185.01246105919003, "grad_norm": 0.7734375, "learning_rate": 7.041300559026854e-07, "loss": 0.0809, "step": 29787 }, { "epoch": 185.01869158878506, "grad_norm": 2.765625, "learning_rate": 7.035495558602557e-07, "loss": 0.2376, "step": 29788 }, { "epoch": 185.02492211838006, "grad_norm": 1.4921875, "learning_rate": 7.029692917898095e-07, "loss": 0.118, "step": 29789 }, { "epoch": 185.0311526479751, "grad_norm": 3.0, "learning_rate": 7.023892636969814e-07, "loss": 0.2338, "step": 29790 }, { "epoch": 185.0373831775701, "grad_norm": 2.15625, "learning_rate": 7.018094715874085e-07, "loss": 0.1823, "step": 29791 }, { "epoch": 185.04361370716512, "grad_norm": 1.6953125, "learning_rate": 7.012299154667195e-07, "loss": 0.1676, "step": 29792 }, { "epoch": 185.04984423676012, "grad_norm": 2.15625, "learning_rate": 7.006505953405407e-07, "loss": 0.1338, "step": 29793 }, { "epoch": 185.05607476635515, "grad_norm": 2.75, "learning_rate": 7.000715112145007e-07, "loss": 0.1867, "step": 29794 }, { "epoch": 185.06230529595015, "grad_norm": 2.1875, "learning_rate": 6.994926630942255e-07, "loss": 0.1748, "step": 29795 }, { "epoch": 185.06853582554518, "grad_norm": 2.765625, "learning_rate": 6.989140509853331e-07, "loss": 0.2267, "step": 29796 }, { "epoch": 185.07476635514018, "grad_norm": 1.7265625, "learning_rate": 6.983356748934466e-07, "loss": 0.1281, "step": 29797 }, { "epoch": 185.0809968847352, "grad_norm": 3.578125, "learning_rate": 6.977575348241811e-07, "loss": 0.1835, "step": 29798 }, { "epoch": 185.08722741433021, "grad_norm": 1.828125, "learning_rate": 6.97179630783154e-07, "loss": 0.1489, "step": 29799 }, { "epoch": 185.09345794392524, "grad_norm": 2.4375, "learning_rate": 6.966019627759751e-07, "loss": 0.2237, "step": 29800 }, { "epoch": 185.09968847352025, "grad_norm": 1.3046875, "learning_rate": 6.960245308082536e-07, "loss": 0.1415, "step": 29801 }, { "epoch": 185.10591900311528, "grad_norm": 1.390625, "learning_rate": 6.954473348856044e-07, "loss": 0.1205, "step": 29802 }, { "epoch": 185.11214953271028, "grad_norm": 2.28125, "learning_rate": 6.948703750136259e-07, "loss": 0.1664, "step": 29803 }, { "epoch": 185.1183800623053, "grad_norm": 2.90625, "learning_rate": 6.942936511979248e-07, "loss": 0.1999, "step": 29804 }, { "epoch": 185.1246105919003, "grad_norm": 1.109375, "learning_rate": 6.937171634441048e-07, "loss": 0.1157, "step": 29805 }, { "epoch": 185.13084112149534, "grad_norm": 1.8203125, "learning_rate": 6.931409117577642e-07, "loss": 0.2467, "step": 29806 }, { "epoch": 185.13707165109034, "grad_norm": 2.34375, "learning_rate": 6.92564896144493e-07, "loss": 0.201, "step": 29807 }, { "epoch": 185.14330218068537, "grad_norm": 2.296875, "learning_rate": 6.919891166098924e-07, "loss": 0.1741, "step": 29808 }, { "epoch": 185.14953271028037, "grad_norm": 3.1875, "learning_rate": 6.91413573159555e-07, "loss": 0.1379, "step": 29809 }, { "epoch": 185.1557632398754, "grad_norm": 2.5625, "learning_rate": 6.908382657990625e-07, "loss": 0.2091, "step": 29810 }, { "epoch": 185.1619937694704, "grad_norm": 1.7578125, "learning_rate": 6.902631945340132e-07, "loss": 0.1317, "step": 29811 }, { "epoch": 185.16822429906543, "grad_norm": 2.25, "learning_rate": 6.89688359369986e-07, "loss": 0.1081, "step": 29812 }, { "epoch": 185.17445482866043, "grad_norm": 1.78125, "learning_rate": 6.891137603125653e-07, "loss": 0.134, "step": 29813 }, { "epoch": 185.18068535825546, "grad_norm": 2.0625, "learning_rate": 6.885393973673298e-07, "loss": 0.2338, "step": 29814 }, { "epoch": 185.18691588785046, "grad_norm": 1.7109375, "learning_rate": 6.879652705398587e-07, "loss": 0.1347, "step": 29815 }, { "epoch": 185.1931464174455, "grad_norm": 3.0625, "learning_rate": 6.873913798357306e-07, "loss": 0.2211, "step": 29816 }, { "epoch": 185.1993769470405, "grad_norm": 1.71875, "learning_rate": 6.868177252605163e-07, "loss": 0.1191, "step": 29817 }, { "epoch": 185.20560747663552, "grad_norm": 2.21875, "learning_rate": 6.862443068197888e-07, "loss": 0.1743, "step": 29818 }, { "epoch": 185.21183800623052, "grad_norm": 2.828125, "learning_rate": 6.856711245191161e-07, "loss": 0.2621, "step": 29819 }, { "epoch": 185.21806853582555, "grad_norm": 2.484375, "learning_rate": 6.850981783640686e-07, "loss": 0.2088, "step": 29820 }, { "epoch": 185.22429906542055, "grad_norm": 1.75, "learning_rate": 6.84525468360206e-07, "loss": 0.1027, "step": 29821 }, { "epoch": 185.23052959501558, "grad_norm": 1.3203125, "learning_rate": 6.839529945130901e-07, "loss": 0.0952, "step": 29822 }, { "epoch": 185.23676012461058, "grad_norm": 1.2578125, "learning_rate": 6.833807568282863e-07, "loss": 0.1251, "step": 29823 }, { "epoch": 185.2429906542056, "grad_norm": 2.296875, "learning_rate": 6.828087553113538e-07, "loss": 0.1961, "step": 29824 }, { "epoch": 185.2492211838006, "grad_norm": 1.8984375, "learning_rate": 6.822369899678382e-07, "loss": 0.1295, "step": 29825 }, { "epoch": 185.25545171339564, "grad_norm": 0.8359375, "learning_rate": 6.816654608032963e-07, "loss": 0.0875, "step": 29826 }, { "epoch": 185.26168224299064, "grad_norm": 1.8515625, "learning_rate": 6.810941678232819e-07, "loss": 0.1551, "step": 29827 }, { "epoch": 185.26791277258567, "grad_norm": 1.6796875, "learning_rate": 6.80523111033346e-07, "loss": 0.114, "step": 29828 }, { "epoch": 185.27414330218068, "grad_norm": 2.890625, "learning_rate": 6.79952290439026e-07, "loss": 0.2318, "step": 29829 }, { "epoch": 185.2803738317757, "grad_norm": 2.4375, "learning_rate": 6.793817060458757e-07, "loss": 0.195, "step": 29830 }, { "epoch": 185.2866043613707, "grad_norm": 1.875, "learning_rate": 6.788113578594268e-07, "loss": 0.13, "step": 29831 }, { "epoch": 185.29283489096574, "grad_norm": 2.125, "learning_rate": 6.782412458852277e-07, "loss": 0.2034, "step": 29832 }, { "epoch": 185.29906542056074, "grad_norm": 2.078125, "learning_rate": 6.776713701288101e-07, "loss": 0.1829, "step": 29833 }, { "epoch": 185.30529595015577, "grad_norm": 1.7578125, "learning_rate": 6.771017305957084e-07, "loss": 0.1359, "step": 29834 }, { "epoch": 185.31152647975077, "grad_norm": 1.3515625, "learning_rate": 6.765323272914598e-07, "loss": 0.1296, "step": 29835 }, { "epoch": 185.3177570093458, "grad_norm": 2.15625, "learning_rate": 6.759631602215877e-07, "loss": 0.1153, "step": 29836 }, { "epoch": 185.3239875389408, "grad_norm": 3.40625, "learning_rate": 6.753942293916266e-07, "loss": 0.2895, "step": 29837 }, { "epoch": 185.33021806853583, "grad_norm": 1.5390625, "learning_rate": 6.74825534807097e-07, "loss": 0.0804, "step": 29838 }, { "epoch": 185.33644859813083, "grad_norm": 2.171875, "learning_rate": 6.742570764735278e-07, "loss": 0.1587, "step": 29839 }, { "epoch": 185.34267912772586, "grad_norm": 2.03125, "learning_rate": 6.736888543964343e-07, "loss": 0.1997, "step": 29840 }, { "epoch": 185.34890965732086, "grad_norm": 3.0, "learning_rate": 6.731208685813367e-07, "loss": 0.303, "step": 29841 }, { "epoch": 185.3551401869159, "grad_norm": 2.71875, "learning_rate": 6.725531190337531e-07, "loss": 0.1198, "step": 29842 }, { "epoch": 185.3613707165109, "grad_norm": 2.625, "learning_rate": 6.719856057591983e-07, "loss": 0.1724, "step": 29843 }, { "epoch": 185.36760124610592, "grad_norm": 0.984375, "learning_rate": 6.714183287631793e-07, "loss": 0.0945, "step": 29844 }, { "epoch": 185.37383177570092, "grad_norm": 1.59375, "learning_rate": 6.70851288051208e-07, "loss": 0.1504, "step": 29845 }, { "epoch": 185.38006230529595, "grad_norm": 1.5, "learning_rate": 6.702844836287969e-07, "loss": 0.156, "step": 29846 }, { "epoch": 185.38629283489095, "grad_norm": 1.484375, "learning_rate": 6.697179155014416e-07, "loss": 0.158, "step": 29847 }, { "epoch": 185.39252336448598, "grad_norm": 1.8515625, "learning_rate": 6.691515836746487e-07, "loss": 0.1492, "step": 29848 }, { "epoch": 185.39875389408098, "grad_norm": 1.3203125, "learning_rate": 6.68585488153925e-07, "loss": 0.1237, "step": 29849 }, { "epoch": 185.404984423676, "grad_norm": 1.953125, "learning_rate": 6.680196289447605e-07, "loss": 0.2129, "step": 29850 }, { "epoch": 185.41121495327104, "grad_norm": 1.953125, "learning_rate": 6.674540060526507e-07, "loss": 0.1627, "step": 29851 }, { "epoch": 185.41744548286604, "grad_norm": 2.234375, "learning_rate": 6.668886194830914e-07, "loss": 0.1947, "step": 29852 }, { "epoch": 185.42367601246107, "grad_norm": 1.390625, "learning_rate": 6.663234692415782e-07, "loss": 0.1332, "step": 29853 }, { "epoch": 185.42990654205607, "grad_norm": 2.484375, "learning_rate": 6.657585553335926e-07, "loss": 0.138, "step": 29854 }, { "epoch": 185.4361370716511, "grad_norm": 2.5625, "learning_rate": 6.651938777646222e-07, "loss": 0.2076, "step": 29855 }, { "epoch": 185.4423676012461, "grad_norm": 1.09375, "learning_rate": 6.646294365401595e-07, "loss": 0.1068, "step": 29856 }, { "epoch": 185.44859813084113, "grad_norm": 1.578125, "learning_rate": 6.640652316656781e-07, "loss": 0.1679, "step": 29857 }, { "epoch": 185.45482866043614, "grad_norm": 1.984375, "learning_rate": 6.63501263146657e-07, "loss": 0.1614, "step": 29858 }, { "epoch": 185.46105919003116, "grad_norm": 1.4453125, "learning_rate": 6.629375309885749e-07, "loss": 0.1231, "step": 29859 }, { "epoch": 185.46728971962617, "grad_norm": 1.8828125, "learning_rate": 6.62374035196911e-07, "loss": 0.1464, "step": 29860 }, { "epoch": 185.4735202492212, "grad_norm": 1.9765625, "learning_rate": 6.618107757771386e-07, "loss": 0.1492, "step": 29861 }, { "epoch": 185.4797507788162, "grad_norm": 1.9765625, "learning_rate": 6.612477527347228e-07, "loss": 0.154, "step": 29862 }, { "epoch": 185.48598130841123, "grad_norm": 1.546875, "learning_rate": 6.606849660751313e-07, "loss": 0.1469, "step": 29863 }, { "epoch": 185.49221183800623, "grad_norm": 2.296875, "learning_rate": 6.601224158038349e-07, "loss": 0.2182, "step": 29864 }, { "epoch": 185.49844236760126, "grad_norm": 1.203125, "learning_rate": 6.595601019262959e-07, "loss": 0.1252, "step": 29865 }, { "epoch": 185.50467289719626, "grad_norm": 1.890625, "learning_rate": 6.589980244479738e-07, "loss": 0.2491, "step": 29866 }, { "epoch": 185.5109034267913, "grad_norm": 3.421875, "learning_rate": 6.58436183374328e-07, "loss": 0.244, "step": 29867 }, { "epoch": 185.5171339563863, "grad_norm": 2.453125, "learning_rate": 6.578745787108154e-07, "loss": 0.2151, "step": 29868 }, { "epoch": 185.52336448598132, "grad_norm": 2.4375, "learning_rate": 6.573132104628954e-07, "loss": 0.1155, "step": 29869 }, { "epoch": 185.52959501557632, "grad_norm": 1.375, "learning_rate": 6.567520786360109e-07, "loss": 0.0853, "step": 29870 }, { "epoch": 185.53582554517135, "grad_norm": 1.5546875, "learning_rate": 6.56191183235616e-07, "loss": 0.1005, "step": 29871 }, { "epoch": 185.54205607476635, "grad_norm": 1.6171875, "learning_rate": 6.556305242671618e-07, "loss": 0.1152, "step": 29872 }, { "epoch": 185.54828660436138, "grad_norm": 1.6875, "learning_rate": 6.550701017360856e-07, "loss": 0.1738, "step": 29873 }, { "epoch": 185.55451713395638, "grad_norm": 2.078125, "learning_rate": 6.545099156478384e-07, "loss": 0.173, "step": 29874 }, { "epoch": 185.5607476635514, "grad_norm": 3.28125, "learning_rate": 6.539499660078579e-07, "loss": 0.3171, "step": 29875 }, { "epoch": 185.5669781931464, "grad_norm": 3.21875, "learning_rate": 6.533902528215813e-07, "loss": 0.2894, "step": 29876 }, { "epoch": 185.57320872274144, "grad_norm": 2.59375, "learning_rate": 6.52830776094443e-07, "loss": 0.2218, "step": 29877 }, { "epoch": 185.57943925233644, "grad_norm": 2.203125, "learning_rate": 6.522715358318803e-07, "loss": 0.1908, "step": 29878 }, { "epoch": 185.58566978193147, "grad_norm": 1.5703125, "learning_rate": 6.51712532039328e-07, "loss": 0.1399, "step": 29879 }, { "epoch": 185.59190031152647, "grad_norm": 1.0625, "learning_rate": 6.511537647222038e-07, "loss": 0.108, "step": 29880 }, { "epoch": 185.5981308411215, "grad_norm": 2.4375, "learning_rate": 6.505952338859478e-07, "loss": 0.0872, "step": 29881 }, { "epoch": 185.6043613707165, "grad_norm": 1.9140625, "learning_rate": 6.500369395359724e-07, "loss": 0.1335, "step": 29882 }, { "epoch": 185.61059190031153, "grad_norm": 1.234375, "learning_rate": 6.494788816777092e-07, "loss": 0.1015, "step": 29883 }, { "epoch": 185.61682242990653, "grad_norm": 1.2578125, "learning_rate": 6.489210603165735e-07, "loss": 0.1181, "step": 29884 }, { "epoch": 185.62305295950156, "grad_norm": 1.6640625, "learning_rate": 6.483634754579832e-07, "loss": 0.1564, "step": 29885 }, { "epoch": 185.62928348909657, "grad_norm": 0.83203125, "learning_rate": 6.47806127107356e-07, "loss": 0.0912, "step": 29886 }, { "epoch": 185.6355140186916, "grad_norm": 2.5625, "learning_rate": 6.472490152701044e-07, "loss": 0.1361, "step": 29887 }, { "epoch": 185.6417445482866, "grad_norm": 1.4921875, "learning_rate": 6.466921399516407e-07, "loss": 0.125, "step": 29888 }, { "epoch": 185.64797507788163, "grad_norm": 1.8203125, "learning_rate": 6.461355011573661e-07, "loss": 0.1382, "step": 29889 }, { "epoch": 185.65420560747663, "grad_norm": 1.6484375, "learning_rate": 6.45579098892693e-07, "loss": 0.1348, "step": 29890 }, { "epoch": 185.66043613707166, "grad_norm": 1.9765625, "learning_rate": 6.450229331630253e-07, "loss": 0.1083, "step": 29891 }, { "epoch": 185.66666666666666, "grad_norm": 2.015625, "learning_rate": 6.444670039737616e-07, "loss": 0.1889, "step": 29892 }, { "epoch": 185.6728971962617, "grad_norm": 1.2578125, "learning_rate": 6.43911311330303e-07, "loss": 0.1081, "step": 29893 }, { "epoch": 185.6791277258567, "grad_norm": 1.71875, "learning_rate": 6.433558552380509e-07, "loss": 0.1516, "step": 29894 }, { "epoch": 185.68535825545172, "grad_norm": 1.734375, "learning_rate": 6.428006357023897e-07, "loss": 0.1435, "step": 29895 }, { "epoch": 185.69158878504672, "grad_norm": 0.8125, "learning_rate": 6.422456527287207e-07, "loss": 0.1005, "step": 29896 }, { "epoch": 185.69781931464175, "grad_norm": 1.1875, "learning_rate": 6.416909063224286e-07, "loss": 0.113, "step": 29897 }, { "epoch": 185.70404984423675, "grad_norm": 1.8125, "learning_rate": 6.411363964889061e-07, "loss": 0.1613, "step": 29898 }, { "epoch": 185.71028037383178, "grad_norm": 2.078125, "learning_rate": 6.405821232335351e-07, "loss": 0.2886, "step": 29899 }, { "epoch": 185.71651090342678, "grad_norm": 1.7890625, "learning_rate": 6.400280865617003e-07, "loss": 0.1895, "step": 29900 }, { "epoch": 185.7227414330218, "grad_norm": 2.921875, "learning_rate": 6.394742864787806e-07, "loss": 0.2224, "step": 29901 }, { "epoch": 185.7289719626168, "grad_norm": 1.3203125, "learning_rate": 6.389207229901606e-07, "loss": 0.1452, "step": 29902 }, { "epoch": 185.73520249221184, "grad_norm": 1.390625, "learning_rate": 6.383673961012054e-07, "loss": 0.1092, "step": 29903 }, { "epoch": 185.74143302180684, "grad_norm": 1.0546875, "learning_rate": 6.378143058172997e-07, "loss": 0.1076, "step": 29904 }, { "epoch": 185.74766355140187, "grad_norm": 2.203125, "learning_rate": 6.372614521438114e-07, "loss": 0.1613, "step": 29905 }, { "epoch": 185.75389408099687, "grad_norm": 1.9921875, "learning_rate": 6.367088350861111e-07, "loss": 0.1358, "step": 29906 }, { "epoch": 185.7601246105919, "grad_norm": 2.109375, "learning_rate": 6.361564546495641e-07, "loss": 0.132, "step": 29907 }, { "epoch": 185.7663551401869, "grad_norm": 2.125, "learning_rate": 6.356043108395326e-07, "loss": 0.1152, "step": 29908 }, { "epoch": 185.77258566978193, "grad_norm": 1.8828125, "learning_rate": 6.350524036613876e-07, "loss": 0.1073, "step": 29909 }, { "epoch": 185.77881619937693, "grad_norm": 2.9375, "learning_rate": 6.3450073312048e-07, "loss": 0.1677, "step": 29910 }, { "epoch": 185.78504672897196, "grad_norm": 1.4140625, "learning_rate": 6.339492992221724e-07, "loss": 0.1246, "step": 29911 }, { "epoch": 185.79127725856696, "grad_norm": 2.578125, "learning_rate": 6.333981019718216e-07, "loss": 0.2065, "step": 29912 }, { "epoch": 185.797507788162, "grad_norm": 1.7265625, "learning_rate": 6.328471413747788e-07, "loss": 0.1632, "step": 29913 }, { "epoch": 185.80373831775702, "grad_norm": 1.75, "learning_rate": 6.322964174363955e-07, "loss": 0.1584, "step": 29914 }, { "epoch": 185.80996884735202, "grad_norm": 2.234375, "learning_rate": 6.317459301620171e-07, "loss": 0.2151, "step": 29915 }, { "epoch": 185.81619937694705, "grad_norm": 1.3984375, "learning_rate": 6.311956795569978e-07, "loss": 0.1059, "step": 29916 }, { "epoch": 185.82242990654206, "grad_norm": 3.34375, "learning_rate": 6.306456656266752e-07, "loss": 0.1977, "step": 29917 }, { "epoch": 185.82866043613708, "grad_norm": 3.9375, "learning_rate": 6.300958883763919e-07, "loss": 0.3651, "step": 29918 }, { "epoch": 185.8348909657321, "grad_norm": 2.015625, "learning_rate": 6.29546347811491e-07, "loss": 0.1254, "step": 29919 }, { "epoch": 185.84112149532712, "grad_norm": 2.0625, "learning_rate": 6.289970439373099e-07, "loss": 0.2261, "step": 29920 }, { "epoch": 185.84735202492212, "grad_norm": 2.734375, "learning_rate": 6.284479767591778e-07, "loss": 0.2428, "step": 29921 }, { "epoch": 185.85358255451715, "grad_norm": 1.8046875, "learning_rate": 6.278991462824291e-07, "loss": 0.1854, "step": 29922 }, { "epoch": 185.85981308411215, "grad_norm": 4.03125, "learning_rate": 6.273505525123958e-07, "loss": 0.3986, "step": 29923 }, { "epoch": 185.86604361370718, "grad_norm": 1.7734375, "learning_rate": 6.268021954544096e-07, "loss": 0.1645, "step": 29924 }, { "epoch": 185.87227414330218, "grad_norm": 2.09375, "learning_rate": 6.262540751137914e-07, "loss": 0.2397, "step": 29925 }, { "epoch": 185.8785046728972, "grad_norm": 1.3359375, "learning_rate": 6.257061914958673e-07, "loss": 0.1114, "step": 29926 }, { "epoch": 185.8847352024922, "grad_norm": 1.9765625, "learning_rate": 6.251585446059527e-07, "loss": 0.168, "step": 29927 }, { "epoch": 185.89096573208724, "grad_norm": 2.640625, "learning_rate": 6.246111344493738e-07, "loss": 0.2229, "step": 29928 }, { "epoch": 185.89719626168224, "grad_norm": 2.296875, "learning_rate": 6.24063961031443e-07, "loss": 0.1133, "step": 29929 }, { "epoch": 185.90342679127727, "grad_norm": 2.5, "learning_rate": 6.235170243574756e-07, "loss": 0.1822, "step": 29930 }, { "epoch": 185.90965732087227, "grad_norm": 1.6171875, "learning_rate": 6.229703244327867e-07, "loss": 0.1593, "step": 29931 }, { "epoch": 185.9158878504673, "grad_norm": 2.59375, "learning_rate": 6.224238612626804e-07, "loss": 0.3609, "step": 29932 }, { "epoch": 185.9221183800623, "grad_norm": 2.25, "learning_rate": 6.218776348524663e-07, "loss": 0.2043, "step": 29933 }, { "epoch": 185.92834890965733, "grad_norm": 1.5390625, "learning_rate": 6.213316452074485e-07, "loss": 0.144, "step": 29934 }, { "epoch": 185.93457943925233, "grad_norm": 2.4375, "learning_rate": 6.20785892332934e-07, "loss": 0.1076, "step": 29935 }, { "epoch": 185.94080996884736, "grad_norm": 2.375, "learning_rate": 6.202403762342157e-07, "loss": 0.2178, "step": 29936 }, { "epoch": 185.94704049844236, "grad_norm": 1.1015625, "learning_rate": 6.196950969166005e-07, "loss": 0.1088, "step": 29937 }, { "epoch": 185.9532710280374, "grad_norm": 2.34375, "learning_rate": 6.191500543853784e-07, "loss": 0.2393, "step": 29938 }, { "epoch": 185.9595015576324, "grad_norm": 3.703125, "learning_rate": 6.186052486458454e-07, "loss": 0.2582, "step": 29939 }, { "epoch": 185.96573208722742, "grad_norm": 2.25, "learning_rate": 6.180606797032917e-07, "loss": 0.1309, "step": 29940 }, { "epoch": 185.97196261682242, "grad_norm": 1.640625, "learning_rate": 6.175163475630047e-07, "loss": 0.2113, "step": 29941 }, { "epoch": 185.97819314641745, "grad_norm": 1.15625, "learning_rate": 6.169722522302774e-07, "loss": 0.0951, "step": 29942 }, { "epoch": 185.98442367601245, "grad_norm": 1.171875, "learning_rate": 6.164283937103831e-07, "loss": 0.124, "step": 29943 }, { "epoch": 185.99065420560748, "grad_norm": 2.171875, "learning_rate": 6.158847720086153e-07, "loss": 0.2382, "step": 29944 }, { "epoch": 185.99688473520249, "grad_norm": 1.390625, "learning_rate": 6.153413871302471e-07, "loss": 0.1388, "step": 29945 }, { "epoch": 186.0, "grad_norm": 1.3671875, "learning_rate": 6.147982390805607e-07, "loss": 0.0507, "step": 29946 }, { "epoch": 186.00623052959503, "grad_norm": 1.484375, "learning_rate": 6.142553278648239e-07, "loss": 0.1189, "step": 29947 }, { "epoch": 186.01246105919003, "grad_norm": 2.234375, "learning_rate": 6.137126534883159e-07, "loss": 0.2382, "step": 29948 }, { "epoch": 186.01869158878506, "grad_norm": 1.03125, "learning_rate": 6.131702159563075e-07, "loss": 0.1131, "step": 29949 }, { "epoch": 186.02492211838006, "grad_norm": 2.171875, "learning_rate": 6.12628015274061e-07, "loss": 0.1272, "step": 29950 }, { "epoch": 186.0311526479751, "grad_norm": 0.8203125, "learning_rate": 6.120860514468502e-07, "loss": 0.0896, "step": 29951 }, { "epoch": 186.0373831775701, "grad_norm": 1.5, "learning_rate": 6.11544324479929e-07, "loss": 0.1081, "step": 29952 }, { "epoch": 186.04361370716512, "grad_norm": 1.0390625, "learning_rate": 6.110028343785712e-07, "loss": 0.0893, "step": 29953 }, { "epoch": 186.04984423676012, "grad_norm": 1.90625, "learning_rate": 6.104615811480252e-07, "loss": 0.1427, "step": 29954 }, { "epoch": 186.05607476635515, "grad_norm": 2.203125, "learning_rate": 6.099205647935508e-07, "loss": 0.2156, "step": 29955 }, { "epoch": 186.06230529595015, "grad_norm": 1.578125, "learning_rate": 6.09379785320402e-07, "loss": 0.16, "step": 29956 }, { "epoch": 186.06853582554518, "grad_norm": 3.0625, "learning_rate": 6.088392427338385e-07, "loss": 0.2604, "step": 29957 }, { "epoch": 186.07476635514018, "grad_norm": 1.5703125, "learning_rate": 6.082989370391007e-07, "loss": 0.1597, "step": 29958 }, { "epoch": 186.0809968847352, "grad_norm": 1.7109375, "learning_rate": 6.077588682414398e-07, "loss": 0.1723, "step": 29959 }, { "epoch": 186.08722741433021, "grad_norm": 2.765625, "learning_rate": 6.072190363460989e-07, "loss": 0.2651, "step": 29960 }, { "epoch": 186.09345794392524, "grad_norm": 2.421875, "learning_rate": 6.066794413583237e-07, "loss": 0.1284, "step": 29961 }, { "epoch": 186.09968847352025, "grad_norm": 2.390625, "learning_rate": 6.061400832833547e-07, "loss": 0.2111, "step": 29962 }, { "epoch": 186.10591900311528, "grad_norm": 2.734375, "learning_rate": 6.056009621264264e-07, "loss": 0.2527, "step": 29963 }, { "epoch": 186.11214953271028, "grad_norm": 2.65625, "learning_rate": 6.050620778927818e-07, "loss": 0.2188, "step": 29964 }, { "epoch": 186.1183800623053, "grad_norm": 1.8125, "learning_rate": 6.045234305876501e-07, "loss": 0.1052, "step": 29965 }, { "epoch": 186.1246105919003, "grad_norm": 2.140625, "learning_rate": 6.039850202162633e-07, "loss": 0.2084, "step": 29966 }, { "epoch": 186.13084112149534, "grad_norm": 1.25, "learning_rate": 6.034468467838478e-07, "loss": 0.0896, "step": 29967 }, { "epoch": 186.13707165109034, "grad_norm": 1.515625, "learning_rate": 6.029089102956381e-07, "loss": 0.1118, "step": 29968 }, { "epoch": 186.14330218068537, "grad_norm": 1.5859375, "learning_rate": 6.023712107568497e-07, "loss": 0.1479, "step": 29969 }, { "epoch": 186.14953271028037, "grad_norm": 2.671875, "learning_rate": 6.018337481727115e-07, "loss": 0.2715, "step": 29970 }, { "epoch": 186.1557632398754, "grad_norm": 2.5, "learning_rate": 6.012965225484418e-07, "loss": 0.1421, "step": 29971 }, { "epoch": 186.1619937694704, "grad_norm": 2.140625, "learning_rate": 6.007595338892557e-07, "loss": 0.2072, "step": 29972 }, { "epoch": 186.16822429906543, "grad_norm": 2.453125, "learning_rate": 6.002227822003687e-07, "loss": 0.2039, "step": 29973 }, { "epoch": 186.17445482866043, "grad_norm": 2.78125, "learning_rate": 5.996862674869985e-07, "loss": 0.2691, "step": 29974 }, { "epoch": 186.18068535825546, "grad_norm": 1.96875, "learning_rate": 5.991499897543524e-07, "loss": 0.1968, "step": 29975 }, { "epoch": 186.18691588785046, "grad_norm": 1.1328125, "learning_rate": 5.98613949007637e-07, "loss": 0.1172, "step": 29976 }, { "epoch": 186.1931464174455, "grad_norm": 1.3984375, "learning_rate": 5.980781452520651e-07, "loss": 0.1306, "step": 29977 }, { "epoch": 186.1993769470405, "grad_norm": 1.8828125, "learning_rate": 5.975425784928296e-07, "loss": 0.1917, "step": 29978 }, { "epoch": 186.20560747663552, "grad_norm": 1.671875, "learning_rate": 5.970072487351458e-07, "loss": 0.1301, "step": 29979 }, { "epoch": 186.21183800623052, "grad_norm": 1.671875, "learning_rate": 5.964721559841985e-07, "loss": 0.1016, "step": 29980 }, { "epoch": 186.21806853582555, "grad_norm": 1.9609375, "learning_rate": 5.959373002451945e-07, "loss": 0.093, "step": 29981 }, { "epoch": 186.22429906542055, "grad_norm": 2.6875, "learning_rate": 5.954026815233272e-07, "loss": 0.2184, "step": 29982 }, { "epoch": 186.23052959501558, "grad_norm": 2.609375, "learning_rate": 5.948682998237892e-07, "loss": 0.2508, "step": 29983 }, { "epoch": 186.23676012461058, "grad_norm": 1.5390625, "learning_rate": 5.943341551517656e-07, "loss": 0.131, "step": 29984 }, { "epoch": 186.2429906542056, "grad_norm": 1.0234375, "learning_rate": 5.938002475124465e-07, "loss": 0.1375, "step": 29985 }, { "epoch": 186.2492211838006, "grad_norm": 2.25, "learning_rate": 5.932665769110224e-07, "loss": 0.1903, "step": 29986 }, { "epoch": 186.25545171339564, "grad_norm": 1.59375, "learning_rate": 5.927331433526667e-07, "loss": 0.1194, "step": 29987 }, { "epoch": 186.26168224299064, "grad_norm": 3.03125, "learning_rate": 5.921999468425671e-07, "loss": 0.1983, "step": 29988 }, { "epoch": 186.26791277258567, "grad_norm": 0.9921875, "learning_rate": 5.916669873859027e-07, "loss": 0.1068, "step": 29989 }, { "epoch": 186.27414330218068, "grad_norm": 2.34375, "learning_rate": 5.911342649878499e-07, "loss": 0.1789, "step": 29990 }, { "epoch": 186.2803738317757, "grad_norm": 1.8515625, "learning_rate": 5.906017796535768e-07, "loss": 0.1666, "step": 29991 }, { "epoch": 186.2866043613707, "grad_norm": 1.6171875, "learning_rate": 5.90069531388257e-07, "loss": 0.1367, "step": 29992 }, { "epoch": 186.29283489096574, "grad_norm": 1.609375, "learning_rate": 5.895375201970616e-07, "loss": 0.1622, "step": 29993 }, { "epoch": 186.29906542056074, "grad_norm": 0.7890625, "learning_rate": 5.890057460851611e-07, "loss": 0.088, "step": 29994 }, { "epoch": 186.30529595015577, "grad_norm": 1.6875, "learning_rate": 5.884742090577128e-07, "loss": 0.1739, "step": 29995 }, { "epoch": 186.31152647975077, "grad_norm": 1.765625, "learning_rate": 5.879429091198846e-07, "loss": 0.1771, "step": 29996 }, { "epoch": 186.3177570093458, "grad_norm": 0.86328125, "learning_rate": 5.874118462768335e-07, "loss": 0.0922, "step": 29997 }, { "epoch": 186.3239875389408, "grad_norm": 1.5546875, "learning_rate": 5.868810205337221e-07, "loss": 0.146, "step": 29998 }, { "epoch": 186.33021806853583, "grad_norm": 2.046875, "learning_rate": 5.863504318956992e-07, "loss": 0.1777, "step": 29999 }, { "epoch": 186.33644859813083, "grad_norm": 2.703125, "learning_rate": 5.858200803679187e-07, "loss": 0.217, "step": 30000 }, { "epoch": 186.34267912772586, "grad_norm": 1.859375, "learning_rate": 5.852899659555377e-07, "loss": 0.1375, "step": 30001 }, { "epoch": 186.34890965732086, "grad_norm": 1.390625, "learning_rate": 5.847600886637023e-07, "loss": 0.1303, "step": 30002 }, { "epoch": 186.3551401869159, "grad_norm": 1.8984375, "learning_rate": 5.842304484975553e-07, "loss": 0.1352, "step": 30003 }, { "epoch": 186.3613707165109, "grad_norm": 2.390625, "learning_rate": 5.8370104546224e-07, "loss": 0.1876, "step": 30004 }, { "epoch": 186.36760124610592, "grad_norm": 1.3125, "learning_rate": 5.831718795629049e-07, "loss": 0.1231, "step": 30005 }, { "epoch": 186.37383177570092, "grad_norm": 2.734375, "learning_rate": 5.826429508046849e-07, "loss": 0.2673, "step": 30006 }, { "epoch": 186.38006230529595, "grad_norm": 1.5546875, "learning_rate": 5.821142591927148e-07, "loss": 0.1022, "step": 30007 }, { "epoch": 186.38629283489095, "grad_norm": 1.0859375, "learning_rate": 5.815858047321349e-07, "loss": 0.0906, "step": 30008 }, { "epoch": 186.39252336448598, "grad_norm": 2.390625, "learning_rate": 5.810575874280744e-07, "loss": 0.2027, "step": 30009 }, { "epoch": 186.39875389408098, "grad_norm": 1.2265625, "learning_rate": 5.805296072856625e-07, "loss": 0.1266, "step": 30010 }, { "epoch": 186.404984423676, "grad_norm": 1.03125, "learning_rate": 5.800018643100286e-07, "loss": 0.1109, "step": 30011 }, { "epoch": 186.41121495327104, "grad_norm": 1.7734375, "learning_rate": 5.79474358506299e-07, "loss": 0.1366, "step": 30012 }, { "epoch": 186.41744548286604, "grad_norm": 1.6640625, "learning_rate": 5.789470898795946e-07, "loss": 0.1635, "step": 30013 }, { "epoch": 186.42367601246107, "grad_norm": 1.390625, "learning_rate": 5.784200584350392e-07, "loss": 0.1126, "step": 30014 }, { "epoch": 186.42990654205607, "grad_norm": 2.25, "learning_rate": 5.778932641777507e-07, "loss": 0.1405, "step": 30015 }, { "epoch": 186.4361370716511, "grad_norm": 1.375, "learning_rate": 5.773667071128447e-07, "loss": 0.1227, "step": 30016 }, { "epoch": 186.4423676012461, "grad_norm": 1.984375, "learning_rate": 5.768403872454309e-07, "loss": 0.1875, "step": 30017 }, { "epoch": 186.44859813084113, "grad_norm": 1.265625, "learning_rate": 5.763143045806274e-07, "loss": 0.1004, "step": 30018 }, { "epoch": 186.45482866043614, "grad_norm": 2.359375, "learning_rate": 5.757884591235413e-07, "loss": 0.1946, "step": 30019 }, { "epoch": 186.46105919003116, "grad_norm": 0.83984375, "learning_rate": 5.752628508792823e-07, "loss": 0.0985, "step": 30020 }, { "epoch": 186.46728971962617, "grad_norm": 1.1484375, "learning_rate": 5.747374798529548e-07, "loss": 0.1079, "step": 30021 }, { "epoch": 186.4735202492212, "grad_norm": 1.625, "learning_rate": 5.742123460496546e-07, "loss": 0.1257, "step": 30022 }, { "epoch": 186.4797507788162, "grad_norm": 2.453125, "learning_rate": 5.736874494744887e-07, "loss": 0.239, "step": 30023 }, { "epoch": 186.48598130841123, "grad_norm": 2.140625, "learning_rate": 5.731627901325531e-07, "loss": 0.1857, "step": 30024 }, { "epoch": 186.49221183800623, "grad_norm": 1.9140625, "learning_rate": 5.726383680289437e-07, "loss": 0.1702, "step": 30025 }, { "epoch": 186.49844236760126, "grad_norm": 2.5, "learning_rate": 5.72114183168751e-07, "loss": 0.1728, "step": 30026 }, { "epoch": 186.50467289719626, "grad_norm": 2.21875, "learning_rate": 5.715902355570707e-07, "loss": 0.1286, "step": 30027 }, { "epoch": 186.5109034267913, "grad_norm": 2.46875, "learning_rate": 5.710665251989905e-07, "loss": 0.1959, "step": 30028 }, { "epoch": 186.5171339563863, "grad_norm": 1.5703125, "learning_rate": 5.705430520995924e-07, "loss": 0.1197, "step": 30029 }, { "epoch": 186.52336448598132, "grad_norm": 1.71875, "learning_rate": 5.700198162639642e-07, "loss": 0.1522, "step": 30030 }, { "epoch": 186.52959501557632, "grad_norm": 1.15625, "learning_rate": 5.694968176971876e-07, "loss": 0.1251, "step": 30031 }, { "epoch": 186.53582554517135, "grad_norm": 1.9921875, "learning_rate": 5.689740564043422e-07, "loss": 0.0971, "step": 30032 }, { "epoch": 186.54205607476635, "grad_norm": 1.5, "learning_rate": 5.684515323905015e-07, "loss": 0.1206, "step": 30033 }, { "epoch": 186.54828660436138, "grad_norm": 2.046875, "learning_rate": 5.679292456607476e-07, "loss": 0.1704, "step": 30034 }, { "epoch": 186.55451713395638, "grad_norm": 2.703125, "learning_rate": 5.674071962201489e-07, "loss": 0.2033, "step": 30035 }, { "epoch": 186.5607476635514, "grad_norm": 1.453125, "learning_rate": 5.668853840737704e-07, "loss": 0.1373, "step": 30036 }, { "epoch": 186.5669781931464, "grad_norm": 1.625, "learning_rate": 5.663638092266859e-07, "loss": 0.1302, "step": 30037 }, { "epoch": 186.57320872274144, "grad_norm": 1.9453125, "learning_rate": 5.658424716839639e-07, "loss": 0.1375, "step": 30038 }, { "epoch": 186.57943925233644, "grad_norm": 1.65625, "learning_rate": 5.653213714506612e-07, "loss": 0.1251, "step": 30039 }, { "epoch": 186.58566978193147, "grad_norm": 2.203125, "learning_rate": 5.648005085318459e-07, "loss": 0.1879, "step": 30040 }, { "epoch": 186.59190031152647, "grad_norm": 1.21875, "learning_rate": 5.642798829325696e-07, "loss": 0.0951, "step": 30041 }, { "epoch": 186.5981308411215, "grad_norm": 3.421875, "learning_rate": 5.637594946578922e-07, "loss": 0.4282, "step": 30042 }, { "epoch": 186.6043613707165, "grad_norm": 1.71875, "learning_rate": 5.632393437128652e-07, "loss": 0.1366, "step": 30043 }, { "epoch": 186.61059190031153, "grad_norm": 0.88671875, "learning_rate": 5.627194301025429e-07, "loss": 0.1001, "step": 30044 }, { "epoch": 186.61682242990653, "grad_norm": 1.515625, "learning_rate": 5.621997538319768e-07, "loss": 0.1105, "step": 30045 }, { "epoch": 186.62305295950156, "grad_norm": 2.25, "learning_rate": 5.6168031490621e-07, "loss": 0.2039, "step": 30046 }, { "epoch": 186.62928348909657, "grad_norm": 2.34375, "learning_rate": 5.611611133302913e-07, "loss": 0.101, "step": 30047 }, { "epoch": 186.6355140186916, "grad_norm": 2.515625, "learning_rate": 5.606421491092583e-07, "loss": 0.1352, "step": 30048 }, { "epoch": 186.6417445482866, "grad_norm": 3.984375, "learning_rate": 5.601234222481571e-07, "loss": 0.2848, "step": 30049 }, { "epoch": 186.64797507788163, "grad_norm": 1.3984375, "learning_rate": 5.596049327520198e-07, "loss": 0.1065, "step": 30050 }, { "epoch": 186.65420560747663, "grad_norm": 1.0234375, "learning_rate": 5.590866806258838e-07, "loss": 0.1002, "step": 30051 }, { "epoch": 186.66043613707166, "grad_norm": 1.6953125, "learning_rate": 5.585686658747841e-07, "loss": 0.1074, "step": 30052 }, { "epoch": 186.66666666666666, "grad_norm": 1.34375, "learning_rate": 5.580508885037583e-07, "loss": 0.1145, "step": 30053 }, { "epoch": 186.6728971962617, "grad_norm": 2.625, "learning_rate": 5.57533348517822e-07, "loss": 0.209, "step": 30054 }, { "epoch": 186.6791277258567, "grad_norm": 2.71875, "learning_rate": 5.57016045922007e-07, "loss": 0.1546, "step": 30055 }, { "epoch": 186.68535825545172, "grad_norm": 2.296875, "learning_rate": 5.564989807213372e-07, "loss": 0.3351, "step": 30056 }, { "epoch": 186.69158878504672, "grad_norm": 2.234375, "learning_rate": 5.559821529208392e-07, "loss": 0.189, "step": 30057 }, { "epoch": 186.69781931464175, "grad_norm": 1.9140625, "learning_rate": 5.554655625255284e-07, "loss": 0.1514, "step": 30058 }, { "epoch": 186.70404984423675, "grad_norm": 1.59375, "learning_rate": 5.549492095404202e-07, "loss": 0.1042, "step": 30059 }, { "epoch": 186.71028037383178, "grad_norm": 2.140625, "learning_rate": 5.544330939705328e-07, "loss": 0.1725, "step": 30060 }, { "epoch": 186.71651090342678, "grad_norm": 1.765625, "learning_rate": 5.539172158208816e-07, "loss": 0.1439, "step": 30061 }, { "epoch": 186.7227414330218, "grad_norm": 1.3515625, "learning_rate": 5.534015750964683e-07, "loss": 0.1276, "step": 30062 }, { "epoch": 186.7289719626168, "grad_norm": 2.046875, "learning_rate": 5.528861718023054e-07, "loss": 0.1398, "step": 30063 }, { "epoch": 186.73520249221184, "grad_norm": 1.3984375, "learning_rate": 5.523710059434e-07, "loss": 0.2157, "step": 30064 }, { "epoch": 186.74143302180684, "grad_norm": 1.484375, "learning_rate": 5.518560775247538e-07, "loss": 0.1325, "step": 30065 }, { "epoch": 186.74766355140187, "grad_norm": 3.484375, "learning_rate": 5.513413865513711e-07, "loss": 0.3687, "step": 30066 }, { "epoch": 186.75389408099687, "grad_norm": 1.3984375, "learning_rate": 5.50826933028245e-07, "loss": 0.0902, "step": 30067 }, { "epoch": 186.7601246105919, "grad_norm": 2.078125, "learning_rate": 5.5031271696038e-07, "loss": 0.1632, "step": 30068 }, { "epoch": 186.7663551401869, "grad_norm": 1.5546875, "learning_rate": 5.497987383527608e-07, "loss": 0.1463, "step": 30069 }, { "epoch": 186.77258566978193, "grad_norm": 2.5625, "learning_rate": 5.492849972103836e-07, "loss": 0.2496, "step": 30070 }, { "epoch": 186.77881619937693, "grad_norm": 2.359375, "learning_rate": 5.487714935382388e-07, "loss": 0.2068, "step": 30071 }, { "epoch": 186.78504672897196, "grad_norm": 2.109375, "learning_rate": 5.482582273413139e-07, "loss": 0.1731, "step": 30072 }, { "epoch": 186.79127725856696, "grad_norm": 2.03125, "learning_rate": 5.477451986245941e-07, "loss": 0.1455, "step": 30073 }, { "epoch": 186.797507788162, "grad_norm": 1.65625, "learning_rate": 5.472324073930557e-07, "loss": 0.1229, "step": 30074 }, { "epoch": 186.80373831775702, "grad_norm": 1.53125, "learning_rate": 5.467198536516893e-07, "loss": 0.1203, "step": 30075 }, { "epoch": 186.80996884735202, "grad_norm": 2.15625, "learning_rate": 5.46207537405466e-07, "loss": 0.1711, "step": 30076 }, { "epoch": 186.81619937694705, "grad_norm": 1.171875, "learning_rate": 5.456954586593621e-07, "loss": 0.0914, "step": 30077 }, { "epoch": 186.82242990654206, "grad_norm": 1.25, "learning_rate": 5.451836174183516e-07, "loss": 0.0954, "step": 30078 }, { "epoch": 186.82866043613708, "grad_norm": 2.390625, "learning_rate": 5.446720136874111e-07, "loss": 0.2336, "step": 30079 }, { "epoch": 186.8348909657321, "grad_norm": 2.3125, "learning_rate": 5.441606474715006e-07, "loss": 0.1847, "step": 30080 }, { "epoch": 186.84112149532712, "grad_norm": 1.4921875, "learning_rate": 5.436495187755908e-07, "loss": 0.1295, "step": 30081 }, { "epoch": 186.84735202492212, "grad_norm": 1.734375, "learning_rate": 5.431386276046473e-07, "loss": 0.1271, "step": 30082 }, { "epoch": 186.85358255451715, "grad_norm": 1.328125, "learning_rate": 5.426279739636302e-07, "loss": 0.1299, "step": 30083 }, { "epoch": 186.85981308411215, "grad_norm": 1.671875, "learning_rate": 5.421175578574966e-07, "loss": 0.0879, "step": 30084 }, { "epoch": 186.86604361370718, "grad_norm": 2.53125, "learning_rate": 5.416073792912091e-07, "loss": 0.2855, "step": 30085 }, { "epoch": 186.87227414330218, "grad_norm": 2.203125, "learning_rate": 5.410974382697192e-07, "loss": 0.1883, "step": 30086 }, { "epoch": 186.8785046728972, "grad_norm": 2.046875, "learning_rate": 5.405877347979788e-07, "loss": 0.1466, "step": 30087 }, { "epoch": 186.8847352024922, "grad_norm": 2.046875, "learning_rate": 5.400782688809391e-07, "loss": 0.1474, "step": 30088 }, { "epoch": 186.89096573208724, "grad_norm": 2.046875, "learning_rate": 5.395690405235493e-07, "loss": 0.1628, "step": 30089 }, { "epoch": 186.89719626168224, "grad_norm": 1.671875, "learning_rate": 5.39060049730758e-07, "loss": 0.0861, "step": 30090 }, { "epoch": 186.90342679127727, "grad_norm": 2.03125, "learning_rate": 5.385512965075029e-07, "loss": 0.1849, "step": 30091 }, { "epoch": 186.90965732087227, "grad_norm": 2.625, "learning_rate": 5.380427808587246e-07, "loss": 0.2007, "step": 30092 }, { "epoch": 186.9158878504673, "grad_norm": 1.8046875, "learning_rate": 5.375345027893664e-07, "loss": 0.1565, "step": 30093 }, { "epoch": 186.9221183800623, "grad_norm": 2.578125, "learning_rate": 5.370264623043658e-07, "loss": 0.2417, "step": 30094 }, { "epoch": 186.92834890965733, "grad_norm": 1.8828125, "learning_rate": 5.365186594086497e-07, "loss": 0.1571, "step": 30095 }, { "epoch": 186.93457943925233, "grad_norm": 2.8125, "learning_rate": 5.360110941071555e-07, "loss": 0.207, "step": 30096 }, { "epoch": 186.94080996884736, "grad_norm": 2.03125, "learning_rate": 5.355037664048157e-07, "loss": 0.1336, "step": 30097 }, { "epoch": 186.94704049844236, "grad_norm": 1.3046875, "learning_rate": 5.349966763065511e-07, "loss": 0.1249, "step": 30098 }, { "epoch": 186.9532710280374, "grad_norm": 1.078125, "learning_rate": 5.344898238172885e-07, "loss": 0.1188, "step": 30099 }, { "epoch": 186.9595015576324, "grad_norm": 1.1796875, "learning_rate": 5.339832089419516e-07, "loss": 0.1091, "step": 30100 }, { "epoch": 186.96573208722742, "grad_norm": 2.1875, "learning_rate": 5.334768316854616e-07, "loss": 0.1318, "step": 30101 }, { "epoch": 186.97196261682242, "grad_norm": 2.390625, "learning_rate": 5.329706920527338e-07, "loss": 0.2263, "step": 30102 }, { "epoch": 186.97819314641745, "grad_norm": 2.03125, "learning_rate": 5.324647900486868e-07, "loss": 0.1402, "step": 30103 }, { "epoch": 186.98442367601245, "grad_norm": 1.375, "learning_rate": 5.319591256782331e-07, "loss": 0.1248, "step": 30104 }, { "epoch": 186.99065420560748, "grad_norm": 1.2109375, "learning_rate": 5.314536989462854e-07, "loss": 0.1168, "step": 30105 }, { "epoch": 186.99688473520249, "grad_norm": 1.4296875, "learning_rate": 5.309485098577454e-07, "loss": 0.1005, "step": 30106 }, { "epoch": 187.0, "grad_norm": 2.09375, "learning_rate": 5.30443558417526e-07, "loss": 0.1374, "step": 30107 }, { "epoch": 187.00623052959503, "grad_norm": 3.203125, "learning_rate": 5.299388446305343e-07, "loss": 0.1407, "step": 30108 }, { "epoch": 187.01246105919003, "grad_norm": 2.171875, "learning_rate": 5.294343685016634e-07, "loss": 0.2031, "step": 30109 }, { "epoch": 187.01869158878506, "grad_norm": 1.734375, "learning_rate": 5.289301300358207e-07, "loss": 0.1033, "step": 30110 }, { "epoch": 187.02492211838006, "grad_norm": 1.5078125, "learning_rate": 5.284261292378967e-07, "loss": 0.1375, "step": 30111 }, { "epoch": 187.0311526479751, "grad_norm": 1.9140625, "learning_rate": 5.27922366112793e-07, "loss": 0.1665, "step": 30112 }, { "epoch": 187.0373831775701, "grad_norm": 1.8671875, "learning_rate": 5.274188406653974e-07, "loss": 0.1939, "step": 30113 }, { "epoch": 187.04361370716512, "grad_norm": 1.875, "learning_rate": 5.269155529006003e-07, "loss": 0.2085, "step": 30114 }, { "epoch": 187.04984423676012, "grad_norm": 2.625, "learning_rate": 5.264125028232925e-07, "loss": 0.2255, "step": 30115 }, { "epoch": 187.05607476635515, "grad_norm": 1.4140625, "learning_rate": 5.259096904383615e-07, "loss": 0.1302, "step": 30116 }, { "epoch": 187.06230529595015, "grad_norm": 2.140625, "learning_rate": 5.254071157506868e-07, "loss": 0.1862, "step": 30117 }, { "epoch": 187.06853582554518, "grad_norm": 1.2734375, "learning_rate": 5.249047787651506e-07, "loss": 0.1187, "step": 30118 }, { "epoch": 187.07476635514018, "grad_norm": 2.328125, "learning_rate": 5.244026794866297e-07, "loss": 0.1715, "step": 30119 }, { "epoch": 187.0809968847352, "grad_norm": 1.53125, "learning_rate": 5.239008179200061e-07, "loss": 0.0892, "step": 30120 }, { "epoch": 187.08722741433021, "grad_norm": 1.1015625, "learning_rate": 5.233991940701455e-07, "loss": 0.0897, "step": 30121 }, { "epoch": 187.09345794392524, "grad_norm": 2.359375, "learning_rate": 5.228978079419272e-07, "loss": 0.1098, "step": 30122 }, { "epoch": 187.09968847352025, "grad_norm": 2.890625, "learning_rate": 5.223966595402196e-07, "loss": 0.1766, "step": 30123 }, { "epoch": 187.10591900311528, "grad_norm": 2.859375, "learning_rate": 5.218957488698883e-07, "loss": 0.2852, "step": 30124 }, { "epoch": 187.11214953271028, "grad_norm": 1.109375, "learning_rate": 5.213950759357961e-07, "loss": 0.1025, "step": 30125 }, { "epoch": 187.1183800623053, "grad_norm": 1.765625, "learning_rate": 5.208946407428112e-07, "loss": 0.1086, "step": 30126 }, { "epoch": 187.1246105919003, "grad_norm": 1.546875, "learning_rate": 5.20394443295788e-07, "loss": 0.1177, "step": 30127 }, { "epoch": 187.13084112149534, "grad_norm": 1.8203125, "learning_rate": 5.198944835995895e-07, "loss": 0.1617, "step": 30128 }, { "epoch": 187.13707165109034, "grad_norm": 2.75, "learning_rate": 5.193947616590672e-07, "loss": 0.302, "step": 30129 }, { "epoch": 187.14330218068537, "grad_norm": 1.46875, "learning_rate": 5.188952774790784e-07, "loss": 0.0932, "step": 30130 }, { "epoch": 187.14953271028037, "grad_norm": 1.078125, "learning_rate": 5.183960310644748e-07, "loss": 0.1175, "step": 30131 }, { "epoch": 187.1557632398754, "grad_norm": 1.625, "learning_rate": 5.178970224200968e-07, "loss": 0.1695, "step": 30132 }, { "epoch": 187.1619937694704, "grad_norm": 1.578125, "learning_rate": 5.173982515507991e-07, "loss": 0.1229, "step": 30133 }, { "epoch": 187.16822429906543, "grad_norm": 1.1328125, "learning_rate": 5.168997184614249e-07, "loss": 0.1074, "step": 30134 }, { "epoch": 187.17445482866043, "grad_norm": 1.5625, "learning_rate": 5.164014231568148e-07, "loss": 0.111, "step": 30135 }, { "epoch": 187.18068535825546, "grad_norm": 1.3359375, "learning_rate": 5.159033656418066e-07, "loss": 0.1196, "step": 30136 }, { "epoch": 187.18691588785046, "grad_norm": 2.703125, "learning_rate": 5.154055459212409e-07, "loss": 0.2194, "step": 30137 }, { "epoch": 187.1931464174455, "grad_norm": 2.078125, "learning_rate": 5.149079639999499e-07, "loss": 0.16, "step": 30138 }, { "epoch": 187.1993769470405, "grad_norm": 1.3828125, "learning_rate": 5.144106198827659e-07, "loss": 0.0872, "step": 30139 }, { "epoch": 187.20560747663552, "grad_norm": 1.9453125, "learning_rate": 5.13913513574521e-07, "loss": 0.1736, "step": 30140 }, { "epoch": 187.21183800623052, "grad_norm": 2.5625, "learning_rate": 5.134166450800449e-07, "loss": 0.1702, "step": 30141 }, { "epoch": 187.21806853582555, "grad_norm": 2.078125, "learning_rate": 5.129200144041613e-07, "loss": 0.2374, "step": 30142 }, { "epoch": 187.22429906542055, "grad_norm": 1.2421875, "learning_rate": 5.124236215516914e-07, "loss": 0.124, "step": 30143 }, { "epoch": 187.23052959501558, "grad_norm": 2.234375, "learning_rate": 5.119274665274564e-07, "loss": 0.2367, "step": 30144 }, { "epoch": 187.23676012461058, "grad_norm": 1.8125, "learning_rate": 5.114315493362832e-07, "loss": 0.1383, "step": 30145 }, { "epoch": 187.2429906542056, "grad_norm": 1.890625, "learning_rate": 5.109358699829759e-07, "loss": 0.1348, "step": 30146 }, { "epoch": 187.2492211838006, "grad_norm": 2.09375, "learning_rate": 5.104404284723558e-07, "loss": 0.1649, "step": 30147 }, { "epoch": 187.25545171339564, "grad_norm": 3.0625, "learning_rate": 5.099452248092329e-07, "loss": 0.3691, "step": 30148 }, { "epoch": 187.26168224299064, "grad_norm": 2.359375, "learning_rate": 5.094502589984229e-07, "loss": 0.1418, "step": 30149 }, { "epoch": 187.26791277258567, "grad_norm": 1.2890625, "learning_rate": 5.089555310447247e-07, "loss": 0.1227, "step": 30150 }, { "epoch": 187.27414330218068, "grad_norm": 2.5625, "learning_rate": 5.084610409529428e-07, "loss": 0.1435, "step": 30151 }, { "epoch": 187.2803738317757, "grad_norm": 1.75, "learning_rate": 5.079667887278844e-07, "loss": 0.1359, "step": 30152 }, { "epoch": 187.2866043613707, "grad_norm": 2.328125, "learning_rate": 5.074727743743512e-07, "loss": 0.1901, "step": 30153 }, { "epoch": 187.29283489096574, "grad_norm": 2.1875, "learning_rate": 5.069789978971368e-07, "loss": 0.1243, "step": 30154 }, { "epoch": 187.29906542056074, "grad_norm": 2.0625, "learning_rate": 5.064854593010371e-07, "loss": 0.1811, "step": 30155 }, { "epoch": 187.30529595015577, "grad_norm": 1.96875, "learning_rate": 5.059921585908484e-07, "loss": 0.1318, "step": 30156 }, { "epoch": 187.31152647975077, "grad_norm": 2.25, "learning_rate": 5.054990957713612e-07, "loss": 0.1746, "step": 30157 }, { "epoch": 187.3177570093458, "grad_norm": 2.578125, "learning_rate": 5.050062708473608e-07, "loss": 0.1833, "step": 30158 }, { "epoch": 187.3239875389408, "grad_norm": 2.578125, "learning_rate": 5.045136838236347e-07, "loss": 0.1464, "step": 30159 }, { "epoch": 187.33021806853583, "grad_norm": 1.453125, "learning_rate": 5.040213347049711e-07, "loss": 0.1366, "step": 30160 }, { "epoch": 187.33644859813083, "grad_norm": 2.84375, "learning_rate": 5.035292234961492e-07, "loss": 0.2737, "step": 30161 }, { "epoch": 187.34267912772586, "grad_norm": 2.40625, "learning_rate": 5.030373502019486e-07, "loss": 0.1809, "step": 30162 }, { "epoch": 187.34890965732086, "grad_norm": 1.34375, "learning_rate": 5.025457148271434e-07, "loss": 0.1099, "step": 30163 }, { "epoch": 187.3551401869159, "grad_norm": 2.859375, "learning_rate": 5.020543173765129e-07, "loss": 0.1273, "step": 30164 }, { "epoch": 187.3613707165109, "grad_norm": 2.265625, "learning_rate": 5.015631578548257e-07, "loss": 0.2213, "step": 30165 }, { "epoch": 187.36760124610592, "grad_norm": 1.2734375, "learning_rate": 5.010722362668558e-07, "loss": 0.1133, "step": 30166 }, { "epoch": 187.37383177570092, "grad_norm": 2.0625, "learning_rate": 5.005815526173713e-07, "loss": 0.1052, "step": 30167 }, { "epoch": 187.38006230529595, "grad_norm": 1.640625, "learning_rate": 5.000911069111353e-07, "loss": 0.1451, "step": 30168 }, { "epoch": 187.38629283489095, "grad_norm": 2.640625, "learning_rate": 4.996008991529105e-07, "loss": 0.189, "step": 30169 }, { "epoch": 187.39252336448598, "grad_norm": 1.875, "learning_rate": 4.9911092934746e-07, "loss": 0.2723, "step": 30170 }, { "epoch": 187.39875389408098, "grad_norm": 1.3828125, "learning_rate": 4.986211974995409e-07, "loss": 0.1042, "step": 30171 }, { "epoch": 187.404984423676, "grad_norm": 2.28125, "learning_rate": 4.981317036139105e-07, "loss": 0.1628, "step": 30172 }, { "epoch": 187.41121495327104, "grad_norm": 1.46875, "learning_rate": 4.976424476953235e-07, "loss": 0.1339, "step": 30173 }, { "epoch": 187.41744548286604, "grad_norm": 1.796875, "learning_rate": 4.971534297485314e-07, "loss": 0.1527, "step": 30174 }, { "epoch": 187.42367601246107, "grad_norm": 2.453125, "learning_rate": 4.966646497782862e-07, "loss": 0.1632, "step": 30175 }, { "epoch": 187.42990654205607, "grad_norm": 1.859375, "learning_rate": 4.961761077893257e-07, "loss": 0.2228, "step": 30176 }, { "epoch": 187.4361370716511, "grad_norm": 2.65625, "learning_rate": 4.956878037864043e-07, "loss": 0.2253, "step": 30177 }, { "epoch": 187.4423676012461, "grad_norm": 1.1953125, "learning_rate": 4.951997377742629e-07, "loss": 0.1495, "step": 30178 }, { "epoch": 187.44859813084113, "grad_norm": 1.65625, "learning_rate": 4.947119097576364e-07, "loss": 0.1375, "step": 30179 }, { "epoch": 187.45482866043614, "grad_norm": 2.328125, "learning_rate": 4.942243197412683e-07, "loss": 0.1885, "step": 30180 }, { "epoch": 187.46105919003116, "grad_norm": 1.609375, "learning_rate": 4.937369677298937e-07, "loss": 0.122, "step": 30181 }, { "epoch": 187.46728971962617, "grad_norm": 2.640625, "learning_rate": 4.932498537282421e-07, "loss": 0.2721, "step": 30182 }, { "epoch": 187.4735202492212, "grad_norm": 2.390625, "learning_rate": 4.927629777410459e-07, "loss": 0.1551, "step": 30183 }, { "epoch": 187.4797507788162, "grad_norm": 2.078125, "learning_rate": 4.922763397730373e-07, "loss": 0.1863, "step": 30184 }, { "epoch": 187.48598130841123, "grad_norm": 2.484375, "learning_rate": 4.917899398289377e-07, "loss": 0.198, "step": 30185 }, { "epoch": 187.49221183800623, "grad_norm": 1.21875, "learning_rate": 4.913037779134738e-07, "loss": 0.1221, "step": 30186 }, { "epoch": 187.49844236760126, "grad_norm": 3.46875, "learning_rate": 4.908178540313695e-07, "loss": 0.2823, "step": 30187 }, { "epoch": 187.50467289719626, "grad_norm": 1.3125, "learning_rate": 4.90332168187338e-07, "loss": 0.1219, "step": 30188 }, { "epoch": 187.5109034267913, "grad_norm": 2.109375, "learning_rate": 4.898467203861001e-07, "loss": 0.1645, "step": 30189 }, { "epoch": 187.5171339563863, "grad_norm": 1.5390625, "learning_rate": 4.893615106323718e-07, "loss": 0.1439, "step": 30190 }, { "epoch": 187.52336448598132, "grad_norm": 1.4609375, "learning_rate": 4.888765389308631e-07, "loss": 0.1195, "step": 30191 }, { "epoch": 187.52959501557632, "grad_norm": 1.84375, "learning_rate": 4.88391805286284e-07, "loss": 0.12, "step": 30192 }, { "epoch": 187.53582554517135, "grad_norm": 2.40625, "learning_rate": 4.879073097033476e-07, "loss": 0.1086, "step": 30193 }, { "epoch": 187.54205607476635, "grad_norm": 1.9765625, "learning_rate": 4.874230521867528e-07, "loss": 0.1578, "step": 30194 }, { "epoch": 187.54828660436138, "grad_norm": 2.25, "learning_rate": 4.869390327412043e-07, "loss": 0.1561, "step": 30195 }, { "epoch": 187.55451713395638, "grad_norm": 1.8671875, "learning_rate": 4.864552513714038e-07, "loss": 0.1526, "step": 30196 }, { "epoch": 187.5607476635514, "grad_norm": 1.8359375, "learning_rate": 4.859717080820531e-07, "loss": 0.1675, "step": 30197 }, { "epoch": 187.5669781931464, "grad_norm": 0.93359375, "learning_rate": 4.854884028778428e-07, "loss": 0.0951, "step": 30198 }, { "epoch": 187.57320872274144, "grad_norm": 1.3203125, "learning_rate": 4.850053357634693e-07, "loss": 0.1039, "step": 30199 }, { "epoch": 187.57943925233644, "grad_norm": 1.5625, "learning_rate": 4.84522506743626e-07, "loss": 0.1226, "step": 30200 }, { "epoch": 187.58566978193147, "grad_norm": 1.6953125, "learning_rate": 4.840399158230008e-07, "loss": 0.1297, "step": 30201 }, { "epoch": 187.59190031152647, "grad_norm": 2.3125, "learning_rate": 4.835575630062789e-07, "loss": 0.2531, "step": 30202 }, { "epoch": 187.5981308411215, "grad_norm": 1.59375, "learning_rate": 4.830754482981481e-07, "loss": 0.1267, "step": 30203 }, { "epoch": 187.6043613707165, "grad_norm": 1.125, "learning_rate": 4.825935717032881e-07, "loss": 0.1187, "step": 30204 }, { "epoch": 187.61059190031153, "grad_norm": 2.4375, "learning_rate": 4.821119332263813e-07, "loss": 0.0929, "step": 30205 }, { "epoch": 187.61682242990653, "grad_norm": 2.3125, "learning_rate": 4.816305328721044e-07, "loss": 0.1404, "step": 30206 }, { "epoch": 187.62305295950156, "grad_norm": 2.375, "learning_rate": 4.811493706451314e-07, "loss": 0.2843, "step": 30207 }, { "epoch": 187.62928348909657, "grad_norm": 2.09375, "learning_rate": 4.806684465501393e-07, "loss": 0.1771, "step": 30208 }, { "epoch": 187.6355140186916, "grad_norm": 2.03125, "learning_rate": 4.801877605917964e-07, "loss": 0.2336, "step": 30209 }, { "epoch": 187.6417445482866, "grad_norm": 1.375, "learning_rate": 4.797073127747686e-07, "loss": 0.1369, "step": 30210 }, { "epoch": 187.64797507788163, "grad_norm": 1.4296875, "learning_rate": 4.792271031037271e-07, "loss": 0.1503, "step": 30211 }, { "epoch": 187.65420560747663, "grad_norm": 1.65625, "learning_rate": 4.787471315833347e-07, "loss": 0.1278, "step": 30212 }, { "epoch": 187.66043613707166, "grad_norm": 2.3125, "learning_rate": 4.78267398218249e-07, "loss": 0.162, "step": 30213 }, { "epoch": 187.66666666666666, "grad_norm": 2.0625, "learning_rate": 4.777879030131327e-07, "loss": 0.1582, "step": 30214 }, { "epoch": 187.6728971962617, "grad_norm": 2.09375, "learning_rate": 4.773086459726434e-07, "loss": 0.1834, "step": 30215 }, { "epoch": 187.6791277258567, "grad_norm": 2.125, "learning_rate": 4.768296271014355e-07, "loss": 0.1494, "step": 30216 }, { "epoch": 187.68535825545172, "grad_norm": 1.7734375, "learning_rate": 4.763508464041583e-07, "loss": 0.1525, "step": 30217 }, { "epoch": 187.69158878504672, "grad_norm": 2.265625, "learning_rate": 4.7587230388546344e-07, "loss": 0.135, "step": 30218 }, { "epoch": 187.69781931464175, "grad_norm": 2.78125, "learning_rate": 4.7539399955000006e-07, "loss": 0.303, "step": 30219 }, { "epoch": 187.70404984423675, "grad_norm": 1.8671875, "learning_rate": 4.749159334024145e-07, "loss": 0.138, "step": 30220 }, { "epoch": 187.71028037383178, "grad_norm": 2.03125, "learning_rate": 4.744381054473446e-07, "loss": 0.1575, "step": 30221 }, { "epoch": 187.71651090342678, "grad_norm": 1.9765625, "learning_rate": 4.739605156894339e-07, "loss": 0.1854, "step": 30222 }, { "epoch": 187.7227414330218, "grad_norm": 2.375, "learning_rate": 4.734831641333259e-07, "loss": 0.1289, "step": 30223 }, { "epoch": 187.7289719626168, "grad_norm": 1.859375, "learning_rate": 4.7300605078364756e-07, "loss": 0.2081, "step": 30224 }, { "epoch": 187.73520249221184, "grad_norm": 2.75, "learning_rate": 4.725291756450395e-07, "loss": 0.245, "step": 30225 }, { "epoch": 187.74143302180684, "grad_norm": 1.90625, "learning_rate": 4.720525387221314e-07, "loss": 0.1736, "step": 30226 }, { "epoch": 187.74766355140187, "grad_norm": 1.09375, "learning_rate": 4.7157614001955007e-07, "loss": 0.1159, "step": 30227 }, { "epoch": 187.75389408099687, "grad_norm": 1.2734375, "learning_rate": 4.710999795419252e-07, "loss": 0.1181, "step": 30228 }, { "epoch": 187.7601246105919, "grad_norm": 2.796875, "learning_rate": 4.706240572938808e-07, "loss": 0.1461, "step": 30229 }, { "epoch": 187.7663551401869, "grad_norm": 2.546875, "learning_rate": 4.7014837328004104e-07, "loss": 0.2237, "step": 30230 }, { "epoch": 187.77258566978193, "grad_norm": 1.7109375, "learning_rate": 4.6967292750502156e-07, "loss": 0.124, "step": 30231 }, { "epoch": 187.77881619937693, "grad_norm": 2.0625, "learning_rate": 4.69197719973441e-07, "loss": 0.154, "step": 30232 }, { "epoch": 187.78504672897196, "grad_norm": 2.109375, "learning_rate": 4.6872275068991776e-07, "loss": 0.1934, "step": 30233 }, { "epoch": 187.79127725856696, "grad_norm": 2.03125, "learning_rate": 4.682480196590622e-07, "loss": 0.1602, "step": 30234 }, { "epoch": 187.797507788162, "grad_norm": 0.96875, "learning_rate": 4.6777352688548437e-07, "loss": 0.104, "step": 30235 }, { "epoch": 187.80373831775702, "grad_norm": 2.078125, "learning_rate": 4.6729927237379454e-07, "loss": 0.1111, "step": 30236 }, { "epoch": 187.80996884735202, "grad_norm": 3.03125, "learning_rate": 4.668252561286002e-07, "loss": 0.2704, "step": 30237 }, { "epoch": 187.81619937694705, "grad_norm": 0.8203125, "learning_rate": 4.663514781545031e-07, "loss": 0.1078, "step": 30238 }, { "epoch": 187.82242990654206, "grad_norm": 2.1875, "learning_rate": 4.658779384561024e-07, "loss": 0.1494, "step": 30239 }, { "epoch": 187.82866043613708, "grad_norm": 1.4921875, "learning_rate": 4.6540463703799997e-07, "loss": 0.1174, "step": 30240 }, { "epoch": 187.8348909657321, "grad_norm": 1.0703125, "learning_rate": 4.6493157390479213e-07, "loss": 0.1188, "step": 30241 }, { "epoch": 187.84112149532712, "grad_norm": 1.6953125, "learning_rate": 4.6445874906107243e-07, "loss": 0.1531, "step": 30242 }, { "epoch": 187.84735202492212, "grad_norm": 1.9375, "learning_rate": 4.6398616251143443e-07, "loss": 0.1665, "step": 30243 }, { "epoch": 187.85358255451715, "grad_norm": 1.2890625, "learning_rate": 4.635138142604689e-07, "loss": 0.1414, "step": 30244 }, { "epoch": 187.85981308411215, "grad_norm": 1.4453125, "learning_rate": 4.6304170431276383e-07, "loss": 0.1208, "step": 30245 }, { "epoch": 187.86604361370718, "grad_norm": 1.1953125, "learning_rate": 4.6256983267289887e-07, "loss": 0.0994, "step": 30246 }, { "epoch": 187.87227414330218, "grad_norm": 2.78125, "learning_rate": 4.620981993454593e-07, "loss": 0.2543, "step": 30247 }, { "epoch": 187.8785046728972, "grad_norm": 1.390625, "learning_rate": 4.6162680433503026e-07, "loss": 0.1185, "step": 30248 }, { "epoch": 187.8847352024922, "grad_norm": 2.90625, "learning_rate": 4.611556476461859e-07, "loss": 0.2321, "step": 30249 }, { "epoch": 187.89096573208724, "grad_norm": 1.84375, "learning_rate": 4.6068472928350317e-07, "loss": 0.0939, "step": 30250 }, { "epoch": 187.89719626168224, "grad_norm": 1.125, "learning_rate": 4.6021404925155607e-07, "loss": 0.1061, "step": 30251 }, { "epoch": 187.90342679127727, "grad_norm": 1.421875, "learning_rate": 4.5974360755491333e-07, "loss": 0.1249, "step": 30252 }, { "epoch": 187.90965732087227, "grad_norm": 2.484375, "learning_rate": 4.592734041981517e-07, "loss": 0.1681, "step": 30253 }, { "epoch": 187.9158878504673, "grad_norm": 2.484375, "learning_rate": 4.588034391858259e-07, "loss": 0.1415, "step": 30254 }, { "epoch": 187.9221183800623, "grad_norm": 1.75, "learning_rate": 4.5833371252251013e-07, "loss": 0.1129, "step": 30255 }, { "epoch": 187.92834890965733, "grad_norm": 2.84375, "learning_rate": 4.578642242127618e-07, "loss": 0.1726, "step": 30256 }, { "epoch": 187.93457943925233, "grad_norm": 1.2421875, "learning_rate": 4.5739497426114384e-07, "loss": 0.1185, "step": 30257 }, { "epoch": 187.94080996884736, "grad_norm": 1.59375, "learning_rate": 4.569259626722111e-07, "loss": 0.0903, "step": 30258 }, { "epoch": 187.94704049844236, "grad_norm": 2.390625, "learning_rate": 4.564571894505182e-07, "loss": 0.176, "step": 30259 }, { "epoch": 187.9532710280374, "grad_norm": 1.859375, "learning_rate": 4.5598865460061977e-07, "loss": 0.2502, "step": 30260 }, { "epoch": 187.9595015576324, "grad_norm": 2.171875, "learning_rate": 4.555203581270651e-07, "loss": 0.2138, "step": 30261 }, { "epoch": 187.96573208722742, "grad_norm": 1.671875, "learning_rate": 4.550523000344031e-07, "loss": 0.1342, "step": 30262 }, { "epoch": 187.97196261682242, "grad_norm": 1.1640625, "learning_rate": 4.5458448032718314e-07, "loss": 0.1071, "step": 30263 }, { "epoch": 187.97819314641745, "grad_norm": 1.7734375, "learning_rate": 4.5411689900994313e-07, "loss": 0.1507, "step": 30264 }, { "epoch": 187.98442367601245, "grad_norm": 1.5, "learning_rate": 4.536495560872267e-07, "loss": 0.1506, "step": 30265 }, { "epoch": 187.99065420560748, "grad_norm": 1.609375, "learning_rate": 4.5318245156356907e-07, "loss": 0.1185, "step": 30266 }, { "epoch": 187.99688473520249, "grad_norm": 2.09375, "learning_rate": 4.527155854435139e-07, "loss": 0.2892, "step": 30267 }, { "epoch": 188.0, "grad_norm": 0.6796875, "learning_rate": 4.522489577315908e-07, "loss": 0.0498, "step": 30268 }, { "epoch": 188.00623052959503, "grad_norm": 2.34375, "learning_rate": 4.517825684323324e-07, "loss": 0.1144, "step": 30269 }, { "epoch": 188.01246105919003, "grad_norm": 1.4609375, "learning_rate": 4.51316417550271e-07, "loss": 0.1234, "step": 30270 }, { "epoch": 188.01869158878506, "grad_norm": 2.15625, "learning_rate": 4.508505050899309e-07, "loss": 0.1909, "step": 30271 }, { "epoch": 188.02492211838006, "grad_norm": 1.484375, "learning_rate": 4.503848310558362e-07, "loss": 0.1276, "step": 30272 }, { "epoch": 188.0311526479751, "grad_norm": 1.5546875, "learning_rate": 4.4991939545251107e-07, "loss": 0.147, "step": 30273 }, { "epoch": 188.0373831775701, "grad_norm": 1.1484375, "learning_rate": 4.4945419828447964e-07, "loss": 0.0877, "step": 30274 }, { "epoch": 188.04361370716512, "grad_norm": 1.21875, "learning_rate": 4.489892395562523e-07, "loss": 0.1243, "step": 30275 }, { "epoch": 188.04984423676012, "grad_norm": 1.625, "learning_rate": 4.4852451927235304e-07, "loss": 0.1315, "step": 30276 }, { "epoch": 188.05607476635515, "grad_norm": 1.2578125, "learning_rate": 4.4806003743728677e-07, "loss": 0.1205, "step": 30277 }, { "epoch": 188.06230529595015, "grad_norm": 2.046875, "learning_rate": 4.4759579405557473e-07, "loss": 0.1907, "step": 30278 }, { "epoch": 188.06853582554518, "grad_norm": 1.828125, "learning_rate": 4.4713178913171615e-07, "loss": 0.1372, "step": 30279 }, { "epoch": 188.07476635514018, "grad_norm": 2.015625, "learning_rate": 4.466680226702213e-07, "loss": 0.1496, "step": 30280 }, { "epoch": 188.0809968847352, "grad_norm": 1.84375, "learning_rate": 4.462044946755978e-07, "loss": 0.1578, "step": 30281 }, { "epoch": 188.08722741433021, "grad_norm": 3.84375, "learning_rate": 4.4574120515234464e-07, "loss": 0.4391, "step": 30282 }, { "epoch": 188.09345794392524, "grad_norm": 0.7421875, "learning_rate": 4.452781541049611e-07, "loss": 0.0954, "step": 30283 }, { "epoch": 188.09968847352025, "grad_norm": 2.984375, "learning_rate": 4.448153415379408e-07, "loss": 0.2197, "step": 30284 }, { "epoch": 188.10591900311528, "grad_norm": 1.5078125, "learning_rate": 4.4435276745578293e-07, "loss": 0.1336, "step": 30285 }, { "epoch": 188.11214953271028, "grad_norm": 1.9375, "learning_rate": 4.4389043186298386e-07, "loss": 0.1749, "step": 30286 }, { "epoch": 188.1183800623053, "grad_norm": 1.25, "learning_rate": 4.4342833476402614e-07, "loss": 0.1121, "step": 30287 }, { "epoch": 188.1246105919003, "grad_norm": 2.453125, "learning_rate": 4.429664761634006e-07, "loss": 0.1513, "step": 30288 }, { "epoch": 188.13084112149534, "grad_norm": 0.828125, "learning_rate": 4.425048560655981e-07, "loss": 0.1017, "step": 30289 }, { "epoch": 188.13707165109034, "grad_norm": 2.0, "learning_rate": 4.4204347447509285e-07, "loss": 0.1586, "step": 30290 }, { "epoch": 188.14330218068537, "grad_norm": 1.484375, "learning_rate": 4.4158233139637293e-07, "loss": 0.1254, "step": 30291 }, { "epoch": 188.14953271028037, "grad_norm": 2.53125, "learning_rate": 4.4112142683391257e-07, "loss": 0.1289, "step": 30292 }, { "epoch": 188.1557632398754, "grad_norm": 2.484375, "learning_rate": 4.406607607921914e-07, "loss": 0.1962, "step": 30293 }, { "epoch": 188.1619937694704, "grad_norm": 2.203125, "learning_rate": 4.40200333275681e-07, "loss": 0.1904, "step": 30294 }, { "epoch": 188.16822429906543, "grad_norm": 1.875, "learning_rate": 4.397401442888555e-07, "loss": 0.2016, "step": 30295 }, { "epoch": 188.17445482866043, "grad_norm": 2.171875, "learning_rate": 4.3928019383618347e-07, "loss": 0.1967, "step": 30296 }, { "epoch": 188.18068535825546, "grad_norm": 1.2734375, "learning_rate": 4.3882048192213365e-07, "loss": 0.1216, "step": 30297 }, { "epoch": 188.18691588785046, "grad_norm": 1.625, "learning_rate": 4.3836100855116356e-07, "loss": 0.1605, "step": 30298 }, { "epoch": 188.1931464174455, "grad_norm": 1.2578125, "learning_rate": 4.3790177372774466e-07, "loss": 0.1256, "step": 30299 }, { "epoch": 188.1993769470405, "grad_norm": 2.296875, "learning_rate": 4.3744277745633444e-07, "loss": 0.1282, "step": 30300 }, { "epoch": 188.20560747663552, "grad_norm": 1.0234375, "learning_rate": 4.369840197413905e-07, "loss": 0.1102, "step": 30301 }, { "epoch": 188.21183800623052, "grad_norm": 1.203125, "learning_rate": 4.3652550058736477e-07, "loss": 0.1184, "step": 30302 }, { "epoch": 188.21806853582555, "grad_norm": 1.71875, "learning_rate": 4.36067219998712e-07, "loss": 0.1223, "step": 30303 }, { "epoch": 188.22429906542055, "grad_norm": 2.3125, "learning_rate": 4.3560917797988985e-07, "loss": 0.2472, "step": 30304 }, { "epoch": 188.23052959501558, "grad_norm": 2.8125, "learning_rate": 4.3515137453533915e-07, "loss": 0.2974, "step": 30305 }, { "epoch": 188.23676012461058, "grad_norm": 1.1484375, "learning_rate": 4.3469380966950637e-07, "loss": 0.0974, "step": 30306 }, { "epoch": 188.2429906542056, "grad_norm": 2.296875, "learning_rate": 4.3423648338684064e-07, "loss": 0.1839, "step": 30307 }, { "epoch": 188.2492211838006, "grad_norm": 2.21875, "learning_rate": 4.3377939569178296e-07, "loss": 0.2326, "step": 30308 }, { "epoch": 188.25545171339564, "grad_norm": 2.78125, "learning_rate": 4.333225465887658e-07, "loss": 0.2874, "step": 30309 }, { "epoch": 188.26168224299064, "grad_norm": 1.7265625, "learning_rate": 4.3286593608223294e-07, "loss": 0.1697, "step": 30310 }, { "epoch": 188.26791277258567, "grad_norm": 2.75, "learning_rate": 4.324095641766168e-07, "loss": 0.2355, "step": 30311 }, { "epoch": 188.27414330218068, "grad_norm": 1.921875, "learning_rate": 4.319534308763501e-07, "loss": 0.1653, "step": 30312 }, { "epoch": 188.2803738317757, "grad_norm": 2.0, "learning_rate": 4.314975361858625e-07, "loss": 0.0981, "step": 30313 }, { "epoch": 188.2866043613707, "grad_norm": 2.546875, "learning_rate": 4.310418801095839e-07, "loss": 0.2889, "step": 30314 }, { "epoch": 188.29283489096574, "grad_norm": 3.03125, "learning_rate": 4.305864626519329e-07, "loss": 0.4037, "step": 30315 }, { "epoch": 188.29906542056074, "grad_norm": 2.3125, "learning_rate": 4.301312838173421e-07, "loss": 0.1339, "step": 30316 }, { "epoch": 188.30529595015577, "grad_norm": 2.421875, "learning_rate": 4.2967634361022746e-07, "loss": 0.141, "step": 30317 }, { "epoch": 188.31152647975077, "grad_norm": 1.7421875, "learning_rate": 4.292216420350048e-07, "loss": 0.1274, "step": 30318 }, { "epoch": 188.3177570093458, "grad_norm": 2.796875, "learning_rate": 4.2876717909609565e-07, "loss": 0.216, "step": 30319 }, { "epoch": 188.3239875389408, "grad_norm": 1.6015625, "learning_rate": 4.2831295479791035e-07, "loss": 0.1107, "step": 30320 }, { "epoch": 188.33021806853583, "grad_norm": 3.21875, "learning_rate": 4.27858969144862e-07, "loss": 0.265, "step": 30321 }, { "epoch": 188.33644859813083, "grad_norm": 2.109375, "learning_rate": 4.274052221413583e-07, "loss": 0.2393, "step": 30322 }, { "epoch": 188.34267912772586, "grad_norm": 1.8515625, "learning_rate": 4.2695171379180943e-07, "loss": 0.1768, "step": 30323 }, { "epoch": 188.34890965732086, "grad_norm": 1.7421875, "learning_rate": 4.264984441006148e-07, "loss": 0.0981, "step": 30324 }, { "epoch": 188.3551401869159, "grad_norm": 3.09375, "learning_rate": 4.2604541307217914e-07, "loss": 0.2564, "step": 30325 }, { "epoch": 188.3613707165109, "grad_norm": 3.546875, "learning_rate": 4.2559262071090455e-07, "loss": 0.2755, "step": 30326 }, { "epoch": 188.36760124610592, "grad_norm": 1.1796875, "learning_rate": 4.2514006702118746e-07, "loss": 0.1196, "step": 30327 }, { "epoch": 188.37383177570092, "grad_norm": 2.015625, "learning_rate": 4.246877520074188e-07, "loss": 0.0875, "step": 30328 }, { "epoch": 188.38006230529595, "grad_norm": 1.8125, "learning_rate": 4.242356756739979e-07, "loss": 0.1526, "step": 30329 }, { "epoch": 188.38629283489095, "grad_norm": 2.265625, "learning_rate": 4.2378383802531284e-07, "loss": 0.1929, "step": 30330 }, { "epoch": 188.39252336448598, "grad_norm": 0.9765625, "learning_rate": 4.23332239065749e-07, "loss": 0.1138, "step": 30331 }, { "epoch": 188.39875389408098, "grad_norm": 1.7890625, "learning_rate": 4.228808787996974e-07, "loss": 0.1624, "step": 30332 }, { "epoch": 188.404984423676, "grad_norm": 1.4921875, "learning_rate": 4.2242975723154056e-07, "loss": 0.1215, "step": 30333 }, { "epoch": 188.41121495327104, "grad_norm": 1.9921875, "learning_rate": 4.2197887436566107e-07, "loss": 0.1274, "step": 30334 }, { "epoch": 188.41744548286604, "grad_norm": 2.421875, "learning_rate": 4.2152823020643053e-07, "loss": 0.2138, "step": 30335 }, { "epoch": 188.42367601246107, "grad_norm": 2.3125, "learning_rate": 4.2107782475823145e-07, "loss": 0.1695, "step": 30336 }, { "epoch": 188.42990654205607, "grad_norm": 2.25, "learning_rate": 4.206276580254409e-07, "loss": 0.1464, "step": 30337 }, { "epoch": 188.4361370716511, "grad_norm": 2.109375, "learning_rate": 4.201777300124249e-07, "loss": 0.1351, "step": 30338 }, { "epoch": 188.4423676012461, "grad_norm": 3.28125, "learning_rate": 4.1972804072355767e-07, "loss": 0.4026, "step": 30339 }, { "epoch": 188.44859813084113, "grad_norm": 1.71875, "learning_rate": 4.192785901632079e-07, "loss": 0.1384, "step": 30340 }, { "epoch": 188.45482866043614, "grad_norm": 1.2421875, "learning_rate": 4.188293783357361e-07, "loss": 0.1081, "step": 30341 }, { "epoch": 188.46105919003116, "grad_norm": 1.84375, "learning_rate": 4.1838040524550536e-07, "loss": 0.1602, "step": 30342 }, { "epoch": 188.46728971962617, "grad_norm": 1.7578125, "learning_rate": 4.1793167089687613e-07, "loss": 0.1763, "step": 30343 }, { "epoch": 188.4735202492212, "grad_norm": 1.0625, "learning_rate": 4.1748317529421153e-07, "loss": 0.0952, "step": 30344 }, { "epoch": 188.4797507788162, "grad_norm": 2.234375, "learning_rate": 4.170349184418637e-07, "loss": 0.256, "step": 30345 }, { "epoch": 188.48598130841123, "grad_norm": 1.2734375, "learning_rate": 4.1658690034419023e-07, "loss": 0.1226, "step": 30346 }, { "epoch": 188.49221183800623, "grad_norm": 1.90625, "learning_rate": 4.161391210055349e-07, "loss": 0.1127, "step": 30347 }, { "epoch": 188.49844236760126, "grad_norm": 1.34375, "learning_rate": 4.156915804302497e-07, "loss": 0.1346, "step": 30348 }, { "epoch": 188.50467289719626, "grad_norm": 1.8125, "learning_rate": 4.15244278622684e-07, "loss": 0.1442, "step": 30349 }, { "epoch": 188.5109034267913, "grad_norm": 2.03125, "learning_rate": 4.1479721558717874e-07, "loss": 0.1828, "step": 30350 }, { "epoch": 188.5171339563863, "grad_norm": 1.9609375, "learning_rate": 4.143503913280777e-07, "loss": 0.1762, "step": 30351 }, { "epoch": 188.52336448598132, "grad_norm": 2.375, "learning_rate": 4.1390380584972466e-07, "loss": 0.1833, "step": 30352 }, { "epoch": 188.52959501557632, "grad_norm": 2.484375, "learning_rate": 4.134574591564494e-07, "loss": 0.2677, "step": 30353 }, { "epoch": 188.53582554517135, "grad_norm": 1.515625, "learning_rate": 4.130113512525874e-07, "loss": 0.1452, "step": 30354 }, { "epoch": 188.54205607476635, "grad_norm": 0.85546875, "learning_rate": 4.1256548214247683e-07, "loss": 0.0869, "step": 30355 }, { "epoch": 188.54828660436138, "grad_norm": 1.125, "learning_rate": 4.1211985183044476e-07, "loss": 0.1057, "step": 30356 }, { "epoch": 188.55451713395638, "grad_norm": 0.9765625, "learning_rate": 4.1167446032082115e-07, "loss": 0.0957, "step": 30357 }, { "epoch": 188.5607476635514, "grad_norm": 1.5859375, "learning_rate": 4.1122930761792746e-07, "loss": 0.102, "step": 30358 }, { "epoch": 188.5669781931464, "grad_norm": 2.203125, "learning_rate": 4.107843937260908e-07, "loss": 0.1689, "step": 30359 }, { "epoch": 188.57320872274144, "grad_norm": 1.2890625, "learning_rate": 4.103397186496355e-07, "loss": 0.1018, "step": 30360 }, { "epoch": 188.57943925233644, "grad_norm": 2.125, "learning_rate": 4.098952823928692e-07, "loss": 0.1861, "step": 30361 }, { "epoch": 188.58566978193147, "grad_norm": 2.421875, "learning_rate": 4.0945108496011906e-07, "loss": 0.1673, "step": 30362 }, { "epoch": 188.59190031152647, "grad_norm": 2.03125, "learning_rate": 4.090071263556955e-07, "loss": 0.1812, "step": 30363 }, { "epoch": 188.5981308411215, "grad_norm": 4.34375, "learning_rate": 4.0856340658391166e-07, "loss": 0.2788, "step": 30364 }, { "epoch": 188.6043613707165, "grad_norm": 2.390625, "learning_rate": 4.081199256490753e-07, "loss": 0.2162, "step": 30365 }, { "epoch": 188.61059190031153, "grad_norm": 2.40625, "learning_rate": 4.076766835554913e-07, "loss": 0.1687, "step": 30366 }, { "epoch": 188.61682242990653, "grad_norm": 1.5703125, "learning_rate": 4.0723368030747e-07, "loss": 0.1314, "step": 30367 }, { "epoch": 188.62305295950156, "grad_norm": 1.6640625, "learning_rate": 4.0679091590930807e-07, "loss": 0.1465, "step": 30368 }, { "epoch": 188.62928348909657, "grad_norm": 1.890625, "learning_rate": 4.0634839036531035e-07, "loss": 0.1103, "step": 30369 }, { "epoch": 188.6355140186916, "grad_norm": 1.015625, "learning_rate": 4.059061036797762e-07, "loss": 0.0986, "step": 30370 }, { "epoch": 188.6417445482866, "grad_norm": 2.078125, "learning_rate": 4.0546405585699667e-07, "loss": 0.165, "step": 30371 }, { "epoch": 188.64797507788163, "grad_norm": 1.3046875, "learning_rate": 4.0502224690126545e-07, "loss": 0.1199, "step": 30372 }, { "epoch": 188.65420560747663, "grad_norm": 1.5234375, "learning_rate": 4.045806768168736e-07, "loss": 0.1405, "step": 30373 }, { "epoch": 188.66043613707166, "grad_norm": 1.1796875, "learning_rate": 4.041393456081122e-07, "loss": 0.1098, "step": 30374 }, { "epoch": 188.66666666666666, "grad_norm": 1.2421875, "learning_rate": 4.0369825327926656e-07, "loss": 0.1173, "step": 30375 }, { "epoch": 188.6728971962617, "grad_norm": 1.390625, "learning_rate": 4.0325739983461953e-07, "loss": 0.125, "step": 30376 }, { "epoch": 188.6791277258567, "grad_norm": 1.7734375, "learning_rate": 4.028167852784509e-07, "loss": 0.246, "step": 30377 }, { "epoch": 188.68535825545172, "grad_norm": 2.765625, "learning_rate": 4.0237640961504897e-07, "loss": 0.2552, "step": 30378 }, { "epoch": 188.69158878504672, "grad_norm": 2.65625, "learning_rate": 4.019362728486797e-07, "loss": 0.1338, "step": 30379 }, { "epoch": 188.69781931464175, "grad_norm": 1.1015625, "learning_rate": 4.014963749836231e-07, "loss": 0.098, "step": 30380 }, { "epoch": 188.70404984423675, "grad_norm": 2.25, "learning_rate": 4.01056716024148e-07, "loss": 0.0979, "step": 30381 }, { "epoch": 188.71028037383178, "grad_norm": 3.546875, "learning_rate": 4.0061729597453134e-07, "loss": 0.3862, "step": 30382 }, { "epoch": 188.71651090342678, "grad_norm": 1.9296875, "learning_rate": 4.0017811483903655e-07, "loss": 0.1483, "step": 30383 }, { "epoch": 188.7227414330218, "grad_norm": 1.5, "learning_rate": 3.997391726219296e-07, "loss": 0.1293, "step": 30384 }, { "epoch": 188.7289719626168, "grad_norm": 1.578125, "learning_rate": 3.9930046932747103e-07, "loss": 0.1667, "step": 30385 }, { "epoch": 188.73520249221184, "grad_norm": 1.453125, "learning_rate": 3.9886200495992675e-07, "loss": 0.0999, "step": 30386 }, { "epoch": 188.74143302180684, "grad_norm": 3.0625, "learning_rate": 3.9842377952354904e-07, "loss": 0.3332, "step": 30387 }, { "epoch": 188.74766355140187, "grad_norm": 1.53125, "learning_rate": 3.9798579302259833e-07, "loss": 0.0976, "step": 30388 }, { "epoch": 188.75389408099687, "grad_norm": 0.91015625, "learning_rate": 3.975480454613295e-07, "loss": 0.0964, "step": 30389 }, { "epoch": 188.7601246105919, "grad_norm": 2.484375, "learning_rate": 3.9711053684398927e-07, "loss": 0.2023, "step": 30390 }, { "epoch": 188.7663551401869, "grad_norm": 2.59375, "learning_rate": 3.966732671748297e-07, "loss": 0.1439, "step": 30391 }, { "epoch": 188.77258566978193, "grad_norm": 2.46875, "learning_rate": 3.9623623645810015e-07, "loss": 0.1665, "step": 30392 }, { "epoch": 188.77881619937693, "grad_norm": 2.84375, "learning_rate": 3.9579944469804174e-07, "loss": 0.0911, "step": 30393 }, { "epoch": 188.78504672897196, "grad_norm": 2.15625, "learning_rate": 3.953628918988955e-07, "loss": 0.1154, "step": 30394 }, { "epoch": 188.79127725856696, "grad_norm": 1.2109375, "learning_rate": 3.949265780649025e-07, "loss": 0.1207, "step": 30395 }, { "epoch": 188.797507788162, "grad_norm": 0.98046875, "learning_rate": 3.9449050320030376e-07, "loss": 0.0868, "step": 30396 }, { "epoch": 188.80373831775702, "grad_norm": 2.71875, "learning_rate": 3.9405466730933204e-07, "loss": 0.2313, "step": 30397 }, { "epoch": 188.80996884735202, "grad_norm": 3.0625, "learning_rate": 3.9361907039621725e-07, "loss": 0.2582, "step": 30398 }, { "epoch": 188.81619937694705, "grad_norm": 2.0625, "learning_rate": 3.9318371246519494e-07, "loss": 0.1864, "step": 30399 }, { "epoch": 188.82242990654206, "grad_norm": 1.8828125, "learning_rate": 3.927485935204922e-07, "loss": 0.1259, "step": 30400 }, { "epoch": 188.82866043613708, "grad_norm": 0.94921875, "learning_rate": 3.9231371356633075e-07, "loss": 0.1019, "step": 30401 }, { "epoch": 188.8348909657321, "grad_norm": 1.875, "learning_rate": 3.9187907260693765e-07, "loss": 0.1529, "step": 30402 }, { "epoch": 188.84112149532712, "grad_norm": 1.5390625, "learning_rate": 3.9144467064653746e-07, "loss": 0.1356, "step": 30403 }, { "epoch": 188.84735202492212, "grad_norm": 1.9140625, "learning_rate": 3.9101050768934336e-07, "loss": 0.1194, "step": 30404 }, { "epoch": 188.85358255451715, "grad_norm": 1.1796875, "learning_rate": 3.9057658373957415e-07, "loss": 0.1161, "step": 30405 }, { "epoch": 188.85981308411215, "grad_norm": 2.828125, "learning_rate": 3.9014289880144606e-07, "loss": 0.1938, "step": 30406 }, { "epoch": 188.86604361370718, "grad_norm": 1.6171875, "learning_rate": 3.897094528791667e-07, "loss": 0.1077, "step": 30407 }, { "epoch": 188.87227414330218, "grad_norm": 1.8515625, "learning_rate": 3.892762459769522e-07, "loss": 0.1305, "step": 30408 }, { "epoch": 188.8785046728972, "grad_norm": 2.140625, "learning_rate": 3.8884327809900477e-07, "loss": 0.1786, "step": 30409 }, { "epoch": 188.8847352024922, "grad_norm": 2.28125, "learning_rate": 3.8841054924953215e-07, "loss": 0.171, "step": 30410 }, { "epoch": 188.89096573208724, "grad_norm": 2.25, "learning_rate": 3.879780594327337e-07, "loss": 0.1683, "step": 30411 }, { "epoch": 188.89719626168224, "grad_norm": 2.5, "learning_rate": 3.875458086528144e-07, "loss": 0.1902, "step": 30412 }, { "epoch": 188.90342679127727, "grad_norm": 1.2890625, "learning_rate": 3.871137969139682e-07, "loss": 0.1126, "step": 30413 }, { "epoch": 188.90965732087227, "grad_norm": 1.546875, "learning_rate": 3.866820242203945e-07, "loss": 0.1571, "step": 30414 }, { "epoch": 188.9158878504673, "grad_norm": 1.8984375, "learning_rate": 3.8625049057628425e-07, "loss": 0.1076, "step": 30415 }, { "epoch": 188.9221183800623, "grad_norm": 0.83203125, "learning_rate": 3.858191959858315e-07, "loss": 0.1068, "step": 30416 }, { "epoch": 188.92834890965733, "grad_norm": 1.53125, "learning_rate": 3.853881404532189e-07, "loss": 0.119, "step": 30417 }, { "epoch": 188.93457943925233, "grad_norm": 2.046875, "learning_rate": 3.849573239826404e-07, "loss": 0.1461, "step": 30418 }, { "epoch": 188.94080996884736, "grad_norm": 2.171875, "learning_rate": 3.845267465782759e-07, "loss": 0.1697, "step": 30419 }, { "epoch": 188.94704049844236, "grad_norm": 1.4375, "learning_rate": 3.840964082443083e-07, "loss": 0.1352, "step": 30420 }, { "epoch": 188.9532710280374, "grad_norm": 1.3671875, "learning_rate": 3.836663089849146e-07, "loss": 0.117, "step": 30421 }, { "epoch": 188.9595015576324, "grad_norm": 1.484375, "learning_rate": 3.8323644880427776e-07, "loss": 0.1166, "step": 30422 }, { "epoch": 188.96573208722742, "grad_norm": 2.0625, "learning_rate": 3.8280682770656937e-07, "loss": 0.1648, "step": 30423 }, { "epoch": 188.97196261682242, "grad_norm": 0.81640625, "learning_rate": 3.8237744569596114e-07, "loss": 0.0948, "step": 30424 }, { "epoch": 188.97819314641745, "grad_norm": 2.71875, "learning_rate": 3.8194830277662183e-07, "loss": 0.2221, "step": 30425 }, { "epoch": 188.98442367601245, "grad_norm": 2.234375, "learning_rate": 3.8151939895272326e-07, "loss": 0.2189, "step": 30426 }, { "epoch": 188.99065420560748, "grad_norm": 3.140625, "learning_rate": 3.810907342284287e-07, "loss": 0.2481, "step": 30427 }, { "epoch": 188.99688473520249, "grad_norm": 1.6875, "learning_rate": 3.806623086079042e-07, "loss": 0.1309, "step": 30428 }, { "epoch": 189.0, "grad_norm": 1.703125, "learning_rate": 3.802341220953076e-07, "loss": 0.09, "step": 30429 }, { "epoch": 189.00623052959503, "grad_norm": 2.46875, "learning_rate": 3.7980617469479953e-07, "loss": 0.1695, "step": 30430 }, { "epoch": 189.01246105919003, "grad_norm": 2.4375, "learning_rate": 3.793784664105349e-07, "loss": 0.2132, "step": 30431 }, { "epoch": 189.01869158878506, "grad_norm": 1.96875, "learning_rate": 3.7895099724666595e-07, "loss": 0.1478, "step": 30432 }, { "epoch": 189.02492211838006, "grad_norm": 2.265625, "learning_rate": 3.7852376720735047e-07, "loss": 0.2647, "step": 30433 }, { "epoch": 189.0311526479751, "grad_norm": 1.859375, "learning_rate": 3.780967762967297e-07, "loss": 0.1329, "step": 30434 }, { "epoch": 189.0373831775701, "grad_norm": 1.546875, "learning_rate": 3.776700245189585e-07, "loss": 0.1404, "step": 30435 }, { "epoch": 189.04361370716512, "grad_norm": 2.75, "learning_rate": 3.772435118781753e-07, "loss": 0.1711, "step": 30436 }, { "epoch": 189.04984423676012, "grad_norm": 1.2265625, "learning_rate": 3.768172383785268e-07, "loss": 0.1115, "step": 30437 }, { "epoch": 189.05607476635515, "grad_norm": 1.8203125, "learning_rate": 3.763912040241513e-07, "loss": 0.1538, "step": 30438 }, { "epoch": 189.06230529595015, "grad_norm": 1.65625, "learning_rate": 3.759654088191844e-07, "loss": 0.1288, "step": 30439 }, { "epoch": 189.06853582554518, "grad_norm": 1.3359375, "learning_rate": 3.755398527677645e-07, "loss": 0.1175, "step": 30440 }, { "epoch": 189.07476635514018, "grad_norm": 1.2578125, "learning_rate": 3.751145358740271e-07, "loss": 0.1073, "step": 30441 }, { "epoch": 189.0809968847352, "grad_norm": 1.3515625, "learning_rate": 3.746894581420968e-07, "loss": 0.0991, "step": 30442 }, { "epoch": 189.08722741433021, "grad_norm": 3.234375, "learning_rate": 3.7426461957610347e-07, "loss": 0.2251, "step": 30443 }, { "epoch": 189.09345794392524, "grad_norm": 2.109375, "learning_rate": 3.738400201801773e-07, "loss": 0.1572, "step": 30444 }, { "epoch": 189.09968847352025, "grad_norm": 3.734375, "learning_rate": 3.7341565995843985e-07, "loss": 0.2428, "step": 30445 }, { "epoch": 189.10591900311528, "grad_norm": 2.859375, "learning_rate": 3.729915389150102e-07, "loss": 0.1768, "step": 30446 }, { "epoch": 189.11214953271028, "grad_norm": 1.2890625, "learning_rate": 3.7256765705400987e-07, "loss": 0.115, "step": 30447 }, { "epoch": 189.1183800623053, "grad_norm": 1.59375, "learning_rate": 3.7214401437955795e-07, "loss": 0.1258, "step": 30448 }, { "epoch": 189.1246105919003, "grad_norm": 1.3984375, "learning_rate": 3.717206108957649e-07, "loss": 0.1318, "step": 30449 }, { "epoch": 189.13084112149534, "grad_norm": 2.03125, "learning_rate": 3.7129744660674425e-07, "loss": 0.162, "step": 30450 }, { "epoch": 189.13707165109034, "grad_norm": 2.359375, "learning_rate": 3.7087452151660373e-07, "loss": 0.1855, "step": 30451 }, { "epoch": 189.14330218068537, "grad_norm": 2.34375, "learning_rate": 3.7045183562945673e-07, "loss": 0.1857, "step": 30452 }, { "epoch": 189.14953271028037, "grad_norm": 2.03125, "learning_rate": 3.7002938894940274e-07, "loss": 0.1058, "step": 30453 }, { "epoch": 189.1557632398754, "grad_norm": 2.578125, "learning_rate": 3.6960718148054686e-07, "loss": 0.2328, "step": 30454 }, { "epoch": 189.1619937694704, "grad_norm": 2.0, "learning_rate": 3.6918521322699126e-07, "loss": 0.1922, "step": 30455 }, { "epoch": 189.16822429906543, "grad_norm": 1.1171875, "learning_rate": 3.687634841928328e-07, "loss": 0.0963, "step": 30456 }, { "epoch": 189.17445482866043, "grad_norm": 1.5546875, "learning_rate": 3.6834199438216253e-07, "loss": 0.1272, "step": 30457 }, { "epoch": 189.18068535825546, "grad_norm": 1.78125, "learning_rate": 3.679207437990828e-07, "loss": 0.147, "step": 30458 }, { "epoch": 189.18691588785046, "grad_norm": 2.328125, "learning_rate": 3.674997324476792e-07, "loss": 0.1276, "step": 30459 }, { "epoch": 189.1931464174455, "grad_norm": 1.3359375, "learning_rate": 3.670789603320429e-07, "loss": 0.1056, "step": 30460 }, { "epoch": 189.1993769470405, "grad_norm": 2.421875, "learning_rate": 3.666584274562596e-07, "loss": 0.1902, "step": 30461 }, { "epoch": 189.20560747663552, "grad_norm": 2.21875, "learning_rate": 3.6623813382441196e-07, "loss": 0.2124, "step": 30462 }, { "epoch": 189.21183800623052, "grad_norm": 2.3125, "learning_rate": 3.6581807944058577e-07, "loss": 0.1899, "step": 30463 }, { "epoch": 189.21806853582555, "grad_norm": 2.203125, "learning_rate": 3.6539826430885826e-07, "loss": 0.1792, "step": 30464 }, { "epoch": 189.22429906542055, "grad_norm": 2.65625, "learning_rate": 3.6497868843330673e-07, "loss": 0.1581, "step": 30465 }, { "epoch": 189.23052959501558, "grad_norm": 1.671875, "learning_rate": 3.6455935181800847e-07, "loss": 0.1096, "step": 30466 }, { "epoch": 189.23676012461058, "grad_norm": 1.0546875, "learning_rate": 3.6414025446703247e-07, "loss": 0.1065, "step": 30467 }, { "epoch": 189.2429906542056, "grad_norm": 1.3828125, "learning_rate": 3.637213963844505e-07, "loss": 0.0923, "step": 30468 }, { "epoch": 189.2492211838006, "grad_norm": 2.109375, "learning_rate": 3.633027775743314e-07, "loss": 0.1978, "step": 30469 }, { "epoch": 189.25545171339564, "grad_norm": 1.6796875, "learning_rate": 3.628843980407415e-07, "loss": 0.242, "step": 30470 }, { "epoch": 189.26168224299064, "grad_norm": 2.625, "learning_rate": 3.6246625778774135e-07, "loss": 0.1791, "step": 30471 }, { "epoch": 189.26791277258567, "grad_norm": 2.765625, "learning_rate": 3.6204835681939443e-07, "loss": 0.2352, "step": 30472 }, { "epoch": 189.27414330218068, "grad_norm": 1.53125, "learning_rate": 3.6163069513976136e-07, "loss": 0.0942, "step": 30473 }, { "epoch": 189.2803738317757, "grad_norm": 2.1875, "learning_rate": 3.612132727528944e-07, "loss": 0.2011, "step": 30474 }, { "epoch": 189.2866043613707, "grad_norm": 1.6171875, "learning_rate": 3.607960896628487e-07, "loss": 0.1075, "step": 30475 }, { "epoch": 189.29283489096574, "grad_norm": 1.875, "learning_rate": 3.603791458736766e-07, "loss": 0.1026, "step": 30476 }, { "epoch": 189.29906542056074, "grad_norm": 2.46875, "learning_rate": 3.599624413894276e-07, "loss": 0.127, "step": 30477 }, { "epoch": 189.30529595015577, "grad_norm": 2.390625, "learning_rate": 3.5954597621414843e-07, "loss": 0.2, "step": 30478 }, { "epoch": 189.31152647975077, "grad_norm": 2.484375, "learning_rate": 3.591297503518859e-07, "loss": 0.1776, "step": 30479 }, { "epoch": 189.3177570093458, "grad_norm": 2.671875, "learning_rate": 3.587137638066812e-07, "loss": 0.2189, "step": 30480 }, { "epoch": 189.3239875389408, "grad_norm": 1.6953125, "learning_rate": 3.5829801658257277e-07, "loss": 0.1321, "step": 30481 }, { "epoch": 189.33021806853583, "grad_norm": 3.109375, "learning_rate": 3.5788250868359907e-07, "loss": 0.4257, "step": 30482 }, { "epoch": 189.33644859813083, "grad_norm": 2.640625, "learning_rate": 3.574672401137985e-07, "loss": 0.1713, "step": 30483 }, { "epoch": 189.34267912772586, "grad_norm": 1.59375, "learning_rate": 3.5705221087719844e-07, "loss": 0.0899, "step": 30484 }, { "epoch": 189.34890965732086, "grad_norm": 3.609375, "learning_rate": 3.5663742097783724e-07, "loss": 0.3385, "step": 30485 }, { "epoch": 189.3551401869159, "grad_norm": 1.421875, "learning_rate": 3.562228704197396e-07, "loss": 0.1388, "step": 30486 }, { "epoch": 189.3613707165109, "grad_norm": 1.7109375, "learning_rate": 3.558085592069299e-07, "loss": 0.1267, "step": 30487 }, { "epoch": 189.36760124610592, "grad_norm": 1.46875, "learning_rate": 3.553944873434328e-07, "loss": 0.12, "step": 30488 }, { "epoch": 189.37383177570092, "grad_norm": 2.640625, "learning_rate": 3.549806548332729e-07, "loss": 0.2023, "step": 30489 }, { "epoch": 189.38006230529595, "grad_norm": 1.4296875, "learning_rate": 3.545670616804636e-07, "loss": 0.1188, "step": 30490 }, { "epoch": 189.38629283489095, "grad_norm": 1.96875, "learning_rate": 3.5415370788902945e-07, "loss": 0.1593, "step": 30491 }, { "epoch": 189.39252336448598, "grad_norm": 2.5625, "learning_rate": 3.5374059346298114e-07, "loss": 0.1783, "step": 30492 }, { "epoch": 189.39875389408098, "grad_norm": 2.828125, "learning_rate": 3.5332771840632936e-07, "loss": 0.2688, "step": 30493 }, { "epoch": 189.404984423676, "grad_norm": 1.78125, "learning_rate": 3.5291508272308483e-07, "loss": 0.1437, "step": 30494 }, { "epoch": 189.41121495327104, "grad_norm": 1.1640625, "learning_rate": 3.525026864172554e-07, "loss": 0.1156, "step": 30495 }, { "epoch": 189.41744548286604, "grad_norm": 2.078125, "learning_rate": 3.520905294928489e-07, "loss": 0.1464, "step": 30496 }, { "epoch": 189.42367601246107, "grad_norm": 2.015625, "learning_rate": 3.516786119538651e-07, "loss": 0.2119, "step": 30497 }, { "epoch": 189.42990654205607, "grad_norm": 1.4609375, "learning_rate": 3.512669338043062e-07, "loss": 0.1379, "step": 30498 }, { "epoch": 189.4361370716511, "grad_norm": 1.7109375, "learning_rate": 3.5085549504817185e-07, "loss": 0.1674, "step": 30499 }, { "epoch": 189.4423676012461, "grad_norm": 3.0625, "learning_rate": 3.5044429568945324e-07, "loss": 0.2586, "step": 30500 }, { "epoch": 189.44859813084113, "grad_norm": 1.1484375, "learning_rate": 3.5003333573214727e-07, "loss": 0.0874, "step": 30501 }, { "epoch": 189.45482866043614, "grad_norm": 1.921875, "learning_rate": 3.496226151802451e-07, "loss": 0.1154, "step": 30502 }, { "epoch": 189.46105919003116, "grad_norm": 1.328125, "learning_rate": 3.492121340377352e-07, "loss": 0.0911, "step": 30503 }, { "epoch": 189.46728971962617, "grad_norm": 1.6875, "learning_rate": 3.488018923086089e-07, "loss": 0.1456, "step": 30504 }, { "epoch": 189.4735202492212, "grad_norm": 2.875, "learning_rate": 3.4839188999684357e-07, "loss": 0.1313, "step": 30505 }, { "epoch": 189.4797507788162, "grad_norm": 1.953125, "learning_rate": 3.479821271064221e-07, "loss": 0.1391, "step": 30506 }, { "epoch": 189.48598130841123, "grad_norm": 1.6328125, "learning_rate": 3.475726036413274e-07, "loss": 0.1286, "step": 30507 }, { "epoch": 189.49221183800623, "grad_norm": 1.5, "learning_rate": 3.4716331960553683e-07, "loss": 0.1336, "step": 30508 }, { "epoch": 189.49844236760126, "grad_norm": 2.90625, "learning_rate": 3.4675427500302227e-07, "loss": 0.2155, "step": 30509 }, { "epoch": 189.50467289719626, "grad_norm": 2.171875, "learning_rate": 3.463454698377555e-07, "loss": 0.1121, "step": 30510 }, { "epoch": 189.5109034267913, "grad_norm": 1.875, "learning_rate": 3.459369041137167e-07, "loss": 0.1573, "step": 30511 }, { "epoch": 189.5171339563863, "grad_norm": 2.09375, "learning_rate": 3.455285778348638e-07, "loss": 0.2195, "step": 30512 }, { "epoch": 189.52336448598132, "grad_norm": 1.6328125, "learning_rate": 3.451204910051631e-07, "loss": 0.147, "step": 30513 }, { "epoch": 189.52959501557632, "grad_norm": 1.6328125, "learning_rate": 3.447126436285836e-07, "loss": 0.137, "step": 30514 }, { "epoch": 189.53582554517135, "grad_norm": 2.59375, "learning_rate": 3.4430503570908324e-07, "loss": 0.235, "step": 30515 }, { "epoch": 189.54205607476635, "grad_norm": 2.4375, "learning_rate": 3.438976672506172e-07, "loss": 0.1974, "step": 30516 }, { "epoch": 189.54828660436138, "grad_norm": 2.390625, "learning_rate": 3.4349053825714905e-07, "loss": 0.1652, "step": 30517 }, { "epoch": 189.55451713395638, "grad_norm": 2.59375, "learning_rate": 3.430836487326311e-07, "loss": 0.1717, "step": 30518 }, { "epoch": 189.5607476635514, "grad_norm": 2.265625, "learning_rate": 3.426769986810102e-07, "loss": 0.1862, "step": 30519 }, { "epoch": 189.5669781931464, "grad_norm": 1.2265625, "learning_rate": 3.4227058810624156e-07, "loss": 0.0984, "step": 30520 }, { "epoch": 189.57320872274144, "grad_norm": 1.875, "learning_rate": 3.4186441701226647e-07, "loss": 0.1103, "step": 30521 }, { "epoch": 189.57943925233644, "grad_norm": 1.6484375, "learning_rate": 3.414584854030373e-07, "loss": 0.143, "step": 30522 }, { "epoch": 189.58566978193147, "grad_norm": 2.015625, "learning_rate": 3.4105279328248974e-07, "loss": 0.2053, "step": 30523 }, { "epoch": 189.59190031152647, "grad_norm": 1.65625, "learning_rate": 3.406473406545707e-07, "loss": 0.1408, "step": 30524 }, { "epoch": 189.5981308411215, "grad_norm": 1.6328125, "learning_rate": 3.4024212752321037e-07, "loss": 0.0984, "step": 30525 }, { "epoch": 189.6043613707165, "grad_norm": 2.921875, "learning_rate": 3.3983715389234725e-07, "loss": 0.1586, "step": 30526 }, { "epoch": 189.61059190031153, "grad_norm": 2.203125, "learning_rate": 3.394324197659171e-07, "loss": 0.147, "step": 30527 }, { "epoch": 189.61682242990653, "grad_norm": 1.1796875, "learning_rate": 3.390279251478473e-07, "loss": 0.1098, "step": 30528 }, { "epoch": 189.62305295950156, "grad_norm": 1.9921875, "learning_rate": 3.386236700420708e-07, "loss": 0.1864, "step": 30529 }, { "epoch": 189.62928348909657, "grad_norm": 1.09375, "learning_rate": 3.382196544525068e-07, "loss": 0.106, "step": 30530 }, { "epoch": 189.6355140186916, "grad_norm": 1.59375, "learning_rate": 3.3781587838308814e-07, "loss": 0.1358, "step": 30531 }, { "epoch": 189.6417445482866, "grad_norm": 2.0, "learning_rate": 3.374123418377284e-07, "loss": 0.1997, "step": 30532 }, { "epoch": 189.64797507788163, "grad_norm": 0.98828125, "learning_rate": 3.3700904482034945e-07, "loss": 0.0981, "step": 30533 }, { "epoch": 189.65420560747663, "grad_norm": 1.9453125, "learning_rate": 3.3660598733487034e-07, "loss": 0.188, "step": 30534 }, { "epoch": 189.66043613707166, "grad_norm": 1.796875, "learning_rate": 3.362031693852019e-07, "loss": 0.1364, "step": 30535 }, { "epoch": 189.66666666666666, "grad_norm": 2.96875, "learning_rate": 3.358005909752604e-07, "loss": 0.2268, "step": 30536 }, { "epoch": 189.6728971962617, "grad_norm": 2.0625, "learning_rate": 3.353982521089538e-07, "loss": 0.0947, "step": 30537 }, { "epoch": 189.6791277258567, "grad_norm": 2.03125, "learning_rate": 3.3499615279019014e-07, "loss": 0.1344, "step": 30538 }, { "epoch": 189.68535825545172, "grad_norm": 2.578125, "learning_rate": 3.3459429302287184e-07, "loss": 0.1572, "step": 30539 }, { "epoch": 189.69158878504672, "grad_norm": 2.6875, "learning_rate": 3.3419267281090404e-07, "loss": 0.2402, "step": 30540 }, { "epoch": 189.69781931464175, "grad_norm": 1.453125, "learning_rate": 3.3379129215819203e-07, "loss": 0.0957, "step": 30541 }, { "epoch": 189.70404984423675, "grad_norm": 1.546875, "learning_rate": 3.3339015106862715e-07, "loss": 0.1092, "step": 30542 }, { "epoch": 189.71028037383178, "grad_norm": 2.078125, "learning_rate": 3.32989249546109e-07, "loss": 0.2068, "step": 30543 }, { "epoch": 189.71651090342678, "grad_norm": 1.8671875, "learning_rate": 3.325885875945262e-07, "loss": 0.168, "step": 30544 }, { "epoch": 189.7227414330218, "grad_norm": 1.953125, "learning_rate": 3.321881652177783e-07, "loss": 0.129, "step": 30545 }, { "epoch": 189.7289719626168, "grad_norm": 1.8984375, "learning_rate": 3.317879824197484e-07, "loss": 0.1273, "step": 30546 }, { "epoch": 189.73520249221184, "grad_norm": 1.7109375, "learning_rate": 3.313880392043223e-07, "loss": 0.1657, "step": 30547 }, { "epoch": 189.74143302180684, "grad_norm": 1.9140625, "learning_rate": 3.309883355753912e-07, "loss": 0.2595, "step": 30548 }, { "epoch": 189.74766355140187, "grad_norm": 1.4375, "learning_rate": 3.305888715368299e-07, "loss": 0.1118, "step": 30549 }, { "epoch": 189.75389408099687, "grad_norm": 1.515625, "learning_rate": 3.301896470925214e-07, "loss": 0.1211, "step": 30550 }, { "epoch": 189.7601246105919, "grad_norm": 1.515625, "learning_rate": 3.297906622463404e-07, "loss": 0.1318, "step": 30551 }, { "epoch": 189.7663551401869, "grad_norm": 0.90625, "learning_rate": 3.29391917002167e-07, "loss": 0.1081, "step": 30552 }, { "epoch": 189.77258566978193, "grad_norm": 2.5625, "learning_rate": 3.289934113638704e-07, "loss": 0.1797, "step": 30553 }, { "epoch": 189.77881619937693, "grad_norm": 2.09375, "learning_rate": 3.285951453353198e-07, "loss": 0.197, "step": 30554 }, { "epoch": 189.78504672897196, "grad_norm": 1.2421875, "learning_rate": 3.2819711892038697e-07, "loss": 0.1235, "step": 30555 }, { "epoch": 189.79127725856696, "grad_norm": 3.03125, "learning_rate": 3.277993321229356e-07, "loss": 0.3013, "step": 30556 }, { "epoch": 189.797507788162, "grad_norm": 2.0625, "learning_rate": 3.274017849468264e-07, "loss": 0.2138, "step": 30557 }, { "epoch": 189.80373831775702, "grad_norm": 2.34375, "learning_rate": 3.27004477395923e-07, "loss": 0.2075, "step": 30558 }, { "epoch": 189.80996884735202, "grad_norm": 0.8359375, "learning_rate": 3.266074094740862e-07, "loss": 0.1004, "step": 30559 }, { "epoch": 189.81619937694705, "grad_norm": 2.71875, "learning_rate": 3.2621058118517125e-07, "loss": 0.2043, "step": 30560 }, { "epoch": 189.82242990654206, "grad_norm": 1.125, "learning_rate": 3.258139925330278e-07, "loss": 0.1239, "step": 30561 }, { "epoch": 189.82866043613708, "grad_norm": 2.640625, "learning_rate": 3.2541764352151394e-07, "loss": 0.145, "step": 30562 }, { "epoch": 189.8348909657321, "grad_norm": 1.21875, "learning_rate": 3.2502153415447656e-07, "loss": 0.1199, "step": 30563 }, { "epoch": 189.84112149532712, "grad_norm": 2.4375, "learning_rate": 3.246256644357598e-07, "loss": 0.1798, "step": 30564 }, { "epoch": 189.84735202492212, "grad_norm": 1.734375, "learning_rate": 3.242300343692134e-07, "loss": 0.1455, "step": 30565 }, { "epoch": 189.85358255451715, "grad_norm": 3.34375, "learning_rate": 3.238346439586787e-07, "loss": 0.2589, "step": 30566 }, { "epoch": 189.85981308411215, "grad_norm": 1.265625, "learning_rate": 3.234394932079915e-07, "loss": 0.091, "step": 30567 }, { "epoch": 189.86604361370718, "grad_norm": 2.140625, "learning_rate": 3.230445821209932e-07, "loss": 0.1815, "step": 30568 }, { "epoch": 189.87227414330218, "grad_norm": 3.0, "learning_rate": 3.226499107015196e-07, "loss": 0.2851, "step": 30569 }, { "epoch": 189.8785046728972, "grad_norm": 1.9609375, "learning_rate": 3.222554789534038e-07, "loss": 0.187, "step": 30570 }, { "epoch": 189.8847352024922, "grad_norm": 1.703125, "learning_rate": 3.2186128688047323e-07, "loss": 0.1288, "step": 30571 }, { "epoch": 189.89096573208724, "grad_norm": 2.296875, "learning_rate": 3.21467334486561e-07, "loss": 0.149, "step": 30572 }, { "epoch": 189.89719626168224, "grad_norm": 1.609375, "learning_rate": 3.2107362177548895e-07, "loss": 0.1318, "step": 30573 }, { "epoch": 189.90342679127727, "grad_norm": 1.9921875, "learning_rate": 3.2068014875108465e-07, "loss": 0.1777, "step": 30574 }, { "epoch": 189.90965732087227, "grad_norm": 2.03125, "learning_rate": 3.2028691541717e-07, "loss": 0.2058, "step": 30575 }, { "epoch": 189.9158878504673, "grad_norm": 1.8359375, "learning_rate": 3.1989392177755875e-07, "loss": 0.1422, "step": 30576 }, { "epoch": 189.9221183800623, "grad_norm": 2.671875, "learning_rate": 3.195011678360726e-07, "loss": 0.3169, "step": 30577 }, { "epoch": 189.92834890965733, "grad_norm": 3.03125, "learning_rate": 3.1910865359652543e-07, "loss": 0.2405, "step": 30578 }, { "epoch": 189.93457943925233, "grad_norm": 1.890625, "learning_rate": 3.187163790627279e-07, "loss": 0.1586, "step": 30579 }, { "epoch": 189.94080996884736, "grad_norm": 2.453125, "learning_rate": 3.183243442384909e-07, "loss": 0.1702, "step": 30580 }, { "epoch": 189.94704049844236, "grad_norm": 1.390625, "learning_rate": 3.179325491276225e-07, "loss": 0.1209, "step": 30581 }, { "epoch": 189.9532710280374, "grad_norm": 1.4609375, "learning_rate": 3.175409937339252e-07, "loss": 0.2209, "step": 30582 }, { "epoch": 189.9595015576324, "grad_norm": 1.8515625, "learning_rate": 3.171496780612043e-07, "loss": 0.1719, "step": 30583 }, { "epoch": 189.96573208722742, "grad_norm": 2.359375, "learning_rate": 3.167586021132596e-07, "loss": 0.2183, "step": 30584 }, { "epoch": 189.97196261682242, "grad_norm": 1.4765625, "learning_rate": 3.163677658938907e-07, "loss": 0.1131, "step": 30585 }, { "epoch": 189.97819314641745, "grad_norm": 2.40625, "learning_rate": 3.1597716940688917e-07, "loss": 0.1021, "step": 30586 }, { "epoch": 189.98442367601245, "grad_norm": 1.4140625, "learning_rate": 3.155868126560546e-07, "loss": 0.1461, "step": 30587 }, { "epoch": 189.99065420560748, "grad_norm": 1.0859375, "learning_rate": 3.151966956451757e-07, "loss": 0.093, "step": 30588 }, { "epoch": 189.99688473520249, "grad_norm": 2.28125, "learning_rate": 3.1480681837803837e-07, "loss": 0.2031, "step": 30589 }, { "epoch": 190.0, "grad_norm": 2.125, "learning_rate": 3.14417180858434e-07, "loss": 0.1291, "step": 30590 }, { "epoch": 190.00623052959503, "grad_norm": 1.3828125, "learning_rate": 3.140277830901428e-07, "loss": 0.1211, "step": 30591 }, { "epoch": 190.01246105919003, "grad_norm": 1.484375, "learning_rate": 3.1363862507694795e-07, "loss": 0.1417, "step": 30592 }, { "epoch": 190.01869158878506, "grad_norm": 1.9296875, "learning_rate": 3.132497068226298e-07, "loss": 0.1466, "step": 30593 }, { "epoch": 190.02492211838006, "grad_norm": 2.71875, "learning_rate": 3.128610283309685e-07, "loss": 0.2508, "step": 30594 }, { "epoch": 190.0311526479751, "grad_norm": 2.65625, "learning_rate": 3.124725896057307e-07, "loss": 0.1977, "step": 30595 }, { "epoch": 190.0373831775701, "grad_norm": 2.03125, "learning_rate": 3.120843906506993e-07, "loss": 0.1332, "step": 30596 }, { "epoch": 190.04361370716512, "grad_norm": 1.0078125, "learning_rate": 3.1169643146963524e-07, "loss": 0.0962, "step": 30597 }, { "epoch": 190.04984423676012, "grad_norm": 2.203125, "learning_rate": 3.113087120663105e-07, "loss": 0.2086, "step": 30598 }, { "epoch": 190.05607476635515, "grad_norm": 1.5078125, "learning_rate": 3.1092123244449155e-07, "loss": 0.1268, "step": 30599 }, { "epoch": 190.06230529595015, "grad_norm": 1.8203125, "learning_rate": 3.10533992607942e-07, "loss": 0.1307, "step": 30600 }, { "epoch": 190.06853582554518, "grad_norm": 1.234375, "learning_rate": 3.101469925604228e-07, "loss": 0.0888, "step": 30601 }, { "epoch": 190.07476635514018, "grad_norm": 1.7109375, "learning_rate": 3.097602323056892e-07, "loss": 0.1002, "step": 30602 }, { "epoch": 190.0809968847352, "grad_norm": 2.484375, "learning_rate": 3.093737118474993e-07, "loss": 0.183, "step": 30603 }, { "epoch": 190.08722741433021, "grad_norm": 1.4140625, "learning_rate": 3.089874311896085e-07, "loss": 0.0963, "step": 30604 }, { "epoch": 190.09345794392524, "grad_norm": 2.34375, "learning_rate": 3.086013903357665e-07, "loss": 0.2009, "step": 30605 }, { "epoch": 190.09968847352025, "grad_norm": 0.9765625, "learning_rate": 3.0821558928972315e-07, "loss": 0.1004, "step": 30606 }, { "epoch": 190.10591900311528, "grad_norm": 1.984375, "learning_rate": 3.078300280552282e-07, "loss": 0.1896, "step": 30607 }, { "epoch": 190.11214953271028, "grad_norm": 2.484375, "learning_rate": 3.074447066360231e-07, "loss": 0.182, "step": 30608 }, { "epoch": 190.1183800623053, "grad_norm": 2.3125, "learning_rate": 3.070596250358493e-07, "loss": 0.2632, "step": 30609 }, { "epoch": 190.1246105919003, "grad_norm": 2.515625, "learning_rate": 3.0667478325844825e-07, "loss": 0.0964, "step": 30610 }, { "epoch": 190.13084112149534, "grad_norm": 2.75, "learning_rate": 3.0629018130756136e-07, "loss": 0.1978, "step": 30611 }, { "epoch": 190.13707165109034, "grad_norm": 3.640625, "learning_rate": 3.059058191869163e-07, "loss": 0.2585, "step": 30612 }, { "epoch": 190.14330218068537, "grad_norm": 2.796875, "learning_rate": 3.0552169690025167e-07, "loss": 0.1896, "step": 30613 }, { "epoch": 190.14953271028037, "grad_norm": 2.875, "learning_rate": 3.0513781445129783e-07, "loss": 0.1878, "step": 30614 }, { "epoch": 190.1557632398754, "grad_norm": 2.4375, "learning_rate": 3.047541718437824e-07, "loss": 0.2225, "step": 30615 }, { "epoch": 190.1619937694704, "grad_norm": 2.421875, "learning_rate": 3.0437076908142734e-07, "loss": 0.1373, "step": 30616 }, { "epoch": 190.16822429906543, "grad_norm": 3.390625, "learning_rate": 3.0398760616796304e-07, "loss": 0.2603, "step": 30617 }, { "epoch": 190.17445482866043, "grad_norm": 1.2109375, "learning_rate": 3.03604683107106e-07, "loss": 0.1355, "step": 30618 }, { "epoch": 190.18068535825546, "grad_norm": 2.5625, "learning_rate": 3.032219999025782e-07, "loss": 0.2035, "step": 30619 }, { "epoch": 190.18691588785046, "grad_norm": 1.6796875, "learning_rate": 3.0283955655809604e-07, "loss": 0.0939, "step": 30620 }, { "epoch": 190.1931464174455, "grad_norm": 2.25, "learning_rate": 3.0245735307737053e-07, "loss": 0.1252, "step": 30621 }, { "epoch": 190.1993769470405, "grad_norm": 2.109375, "learning_rate": 3.0207538946411817e-07, "loss": 0.2187, "step": 30622 }, { "epoch": 190.20560747663552, "grad_norm": 1.1015625, "learning_rate": 3.016936657220443e-07, "loss": 0.121, "step": 30623 }, { "epoch": 190.21183800623052, "grad_norm": 2.3125, "learning_rate": 3.0131218185485976e-07, "loss": 0.2167, "step": 30624 }, { "epoch": 190.21806853582555, "grad_norm": 3.265625, "learning_rate": 3.0093093786627e-07, "loss": 0.3792, "step": 30625 }, { "epoch": 190.22429906542055, "grad_norm": 1.109375, "learning_rate": 3.0054993375997765e-07, "loss": 0.0989, "step": 30626 }, { "epoch": 190.23052959501558, "grad_norm": 2.34375, "learning_rate": 3.001691695396769e-07, "loss": 0.2099, "step": 30627 }, { "epoch": 190.23676012461058, "grad_norm": 1.4453125, "learning_rate": 2.997886452090731e-07, "loss": 0.1274, "step": 30628 }, { "epoch": 190.2429906542056, "grad_norm": 1.9296875, "learning_rate": 2.9940836077185786e-07, "loss": 0.1515, "step": 30629 }, { "epoch": 190.2492211838006, "grad_norm": 2.375, "learning_rate": 2.9902831623172813e-07, "loss": 0.173, "step": 30630 }, { "epoch": 190.25545171339564, "grad_norm": 2.78125, "learning_rate": 2.9864851159237264e-07, "loss": 0.1969, "step": 30631 }, { "epoch": 190.26168224299064, "grad_norm": 1.3203125, "learning_rate": 2.9826894685748006e-07, "loss": 0.125, "step": 30632 }, { "epoch": 190.26791277258567, "grad_norm": 2.421875, "learning_rate": 2.9788962203073923e-07, "loss": 0.2029, "step": 30633 }, { "epoch": 190.27414330218068, "grad_norm": 1.875, "learning_rate": 2.9751053711582764e-07, "loss": 0.1818, "step": 30634 }, { "epoch": 190.2803738317757, "grad_norm": 1.6640625, "learning_rate": 2.971316921164341e-07, "loss": 0.1473, "step": 30635 }, { "epoch": 190.2866043613707, "grad_norm": 2.015625, "learning_rate": 2.9675308703623616e-07, "loss": 0.1662, "step": 30636 }, { "epoch": 190.29283489096574, "grad_norm": 1.4921875, "learning_rate": 2.963747218789115e-07, "loss": 0.1413, "step": 30637 }, { "epoch": 190.29906542056074, "grad_norm": 1.4375, "learning_rate": 2.9599659664812927e-07, "loss": 0.126, "step": 30638 }, { "epoch": 190.30529595015577, "grad_norm": 2.1875, "learning_rate": 2.9561871134757e-07, "loss": 0.183, "step": 30639 }, { "epoch": 190.31152647975077, "grad_norm": 2.796875, "learning_rate": 2.952410659809002e-07, "loss": 0.2044, "step": 30640 }, { "epoch": 190.3177570093458, "grad_norm": 1.515625, "learning_rate": 2.9486366055178624e-07, "loss": 0.1637, "step": 30641 }, { "epoch": 190.3239875389408, "grad_norm": 3.5, "learning_rate": 2.944864950638948e-07, "loss": 0.2361, "step": 30642 }, { "epoch": 190.33021806853583, "grad_norm": 2.25, "learning_rate": 2.9410956952088953e-07, "loss": 0.1662, "step": 30643 }, { "epoch": 190.33644859813083, "grad_norm": 2.25, "learning_rate": 2.937328839264314e-07, "loss": 0.1906, "step": 30644 }, { "epoch": 190.34267912772586, "grad_norm": 1.03125, "learning_rate": 2.9335643828417857e-07, "loss": 0.0991, "step": 30645 }, { "epoch": 190.34890965732086, "grad_norm": 3.4375, "learning_rate": 2.9298023259778653e-07, "loss": 0.1986, "step": 30646 }, { "epoch": 190.3551401869159, "grad_norm": 1.296875, "learning_rate": 2.926042668709078e-07, "loss": 0.142, "step": 30647 }, { "epoch": 190.3613707165109, "grad_norm": 2.15625, "learning_rate": 2.922285411071951e-07, "loss": 0.1568, "step": 30648 }, { "epoch": 190.36760124610592, "grad_norm": 1.4609375, "learning_rate": 2.9185305531029827e-07, "loss": 0.1291, "step": 30649 }, { "epoch": 190.37383177570092, "grad_norm": 0.9921875, "learning_rate": 2.914778094838644e-07, "loss": 0.1084, "step": 30650 }, { "epoch": 190.38006230529595, "grad_norm": 1.640625, "learning_rate": 2.9110280363153777e-07, "loss": 0.1242, "step": 30651 }, { "epoch": 190.38629283489095, "grad_norm": 2.1875, "learning_rate": 2.9072803775695987e-07, "loss": 0.183, "step": 30652 }, { "epoch": 190.39252336448598, "grad_norm": 1.625, "learning_rate": 2.9035351186376956e-07, "loss": 0.1237, "step": 30653 }, { "epoch": 190.39875389408098, "grad_norm": 3.125, "learning_rate": 2.899792259556056e-07, "loss": 0.1959, "step": 30654 }, { "epoch": 190.404984423676, "grad_norm": 2.671875, "learning_rate": 2.896051800361038e-07, "loss": 0.1825, "step": 30655 }, { "epoch": 190.41121495327104, "grad_norm": 2.65625, "learning_rate": 2.89231374108892e-07, "loss": 0.2248, "step": 30656 }, { "epoch": 190.41744548286604, "grad_norm": 1.734375, "learning_rate": 2.888578081776089e-07, "loss": 0.165, "step": 30657 }, { "epoch": 190.42367601246107, "grad_norm": 2.140625, "learning_rate": 2.884844822458765e-07, "loss": 0.1869, "step": 30658 }, { "epoch": 190.42990654205607, "grad_norm": 1.1484375, "learning_rate": 2.881113963173254e-07, "loss": 0.1002, "step": 30659 }, { "epoch": 190.4361370716511, "grad_norm": 2.3125, "learning_rate": 2.8773855039557205e-07, "loss": 0.1865, "step": 30660 }, { "epoch": 190.4423676012461, "grad_norm": 1.03125, "learning_rate": 2.8736594448424415e-07, "loss": 0.0907, "step": 30661 }, { "epoch": 190.44859813084113, "grad_norm": 1.4609375, "learning_rate": 2.8699357858695816e-07, "loss": 0.1687, "step": 30662 }, { "epoch": 190.45482866043614, "grad_norm": 2.5, "learning_rate": 2.8662145270732797e-07, "loss": 0.1911, "step": 30663 }, { "epoch": 190.46105919003116, "grad_norm": 2.109375, "learning_rate": 2.8624956684897564e-07, "loss": 0.1151, "step": 30664 }, { "epoch": 190.46728971962617, "grad_norm": 2.234375, "learning_rate": 2.858779210155038e-07, "loss": 0.2718, "step": 30665 }, { "epoch": 190.4735202492212, "grad_norm": 1.5, "learning_rate": 2.855065152105263e-07, "loss": 0.1397, "step": 30666 }, { "epoch": 190.4797507788162, "grad_norm": 1.0, "learning_rate": 2.851353494376485e-07, "loss": 0.1128, "step": 30667 }, { "epoch": 190.48598130841123, "grad_norm": 2.03125, "learning_rate": 2.8476442370047874e-07, "loss": 0.1147, "step": 30668 }, { "epoch": 190.49221183800623, "grad_norm": 1.1875, "learning_rate": 2.8439373800261683e-07, "loss": 0.1127, "step": 30669 }, { "epoch": 190.49844236760126, "grad_norm": 2.1875, "learning_rate": 2.8402329234766266e-07, "loss": 0.1239, "step": 30670 }, { "epoch": 190.50467289719626, "grad_norm": 2.640625, "learning_rate": 2.83653086739219e-07, "loss": 0.2544, "step": 30671 }, { "epoch": 190.5109034267913, "grad_norm": 2.265625, "learning_rate": 2.8328312118087173e-07, "loss": 0.152, "step": 30672 }, { "epoch": 190.5171339563863, "grad_norm": 2.96875, "learning_rate": 2.829133956762209e-07, "loss": 0.3334, "step": 30673 }, { "epoch": 190.52336448598132, "grad_norm": 2.4375, "learning_rate": 2.8254391022885796e-07, "loss": 0.2203, "step": 30674 }, { "epoch": 190.52959501557632, "grad_norm": 1.390625, "learning_rate": 2.82174664842369e-07, "loss": 0.1179, "step": 30675 }, { "epoch": 190.53582554517135, "grad_norm": 2.78125, "learning_rate": 2.8180565952034e-07, "loss": 0.2588, "step": 30676 }, { "epoch": 190.54205607476635, "grad_norm": 3.234375, "learning_rate": 2.81436894266357e-07, "loss": 0.302, "step": 30677 }, { "epoch": 190.54828660436138, "grad_norm": 1.2890625, "learning_rate": 2.810683690839977e-07, "loss": 0.1205, "step": 30678 }, { "epoch": 190.55451713395638, "grad_norm": 1.4453125, "learning_rate": 2.8070008397684535e-07, "loss": 0.1156, "step": 30679 }, { "epoch": 190.5607476635514, "grad_norm": 1.390625, "learning_rate": 2.8033203894847203e-07, "loss": 0.1445, "step": 30680 }, { "epoch": 190.5669781931464, "grad_norm": 1.3671875, "learning_rate": 2.799642340024583e-07, "loss": 0.1173, "step": 30681 }, { "epoch": 190.57320872274144, "grad_norm": 2.265625, "learning_rate": 2.7959666914237346e-07, "loss": 0.1061, "step": 30682 }, { "epoch": 190.57943925233644, "grad_norm": 2.5, "learning_rate": 2.7922934437178695e-07, "loss": 0.1882, "step": 30683 }, { "epoch": 190.58566978193147, "grad_norm": 2.34375, "learning_rate": 2.7886225969426525e-07, "loss": 0.1901, "step": 30684 }, { "epoch": 190.59190031152647, "grad_norm": 2.546875, "learning_rate": 2.7849541511337785e-07, "loss": 0.0942, "step": 30685 }, { "epoch": 190.5981308411215, "grad_norm": 2.015625, "learning_rate": 2.7812881063268294e-07, "loss": 0.2212, "step": 30686 }, { "epoch": 190.6043613707165, "grad_norm": 1.859375, "learning_rate": 2.7776244625574153e-07, "loss": 0.1589, "step": 30687 }, { "epoch": 190.61059190031153, "grad_norm": 2.609375, "learning_rate": 2.773963219861148e-07, "loss": 0.2326, "step": 30688 }, { "epoch": 190.61682242990653, "grad_norm": 1.609375, "learning_rate": 2.770304378273553e-07, "loss": 0.1317, "step": 30689 }, { "epoch": 190.62305295950156, "grad_norm": 1.296875, "learning_rate": 2.7666479378302136e-07, "loss": 0.0952, "step": 30690 }, { "epoch": 190.62928348909657, "grad_norm": 1.2109375, "learning_rate": 2.7629938985665737e-07, "loss": 0.1257, "step": 30691 }, { "epoch": 190.6355140186916, "grad_norm": 1.765625, "learning_rate": 2.7593422605181887e-07, "loss": 0.1324, "step": 30692 }, { "epoch": 190.6417445482866, "grad_norm": 1.96875, "learning_rate": 2.7556930237204734e-07, "loss": 0.1916, "step": 30693 }, { "epoch": 190.64797507788163, "grad_norm": 1.9609375, "learning_rate": 2.752046188208873e-07, "loss": 0.1969, "step": 30694 }, { "epoch": 190.65420560747663, "grad_norm": 1.5234375, "learning_rate": 2.748401754018859e-07, "loss": 0.1404, "step": 30695 }, { "epoch": 190.66043613707166, "grad_norm": 3.046875, "learning_rate": 2.744759721185791e-07, "loss": 0.2797, "step": 30696 }, { "epoch": 190.66666666666666, "grad_norm": 1.4765625, "learning_rate": 2.741120089745031e-07, "loss": 0.1137, "step": 30697 }, { "epoch": 190.6728971962617, "grad_norm": 1.7890625, "learning_rate": 2.737482859731938e-07, "loss": 0.1225, "step": 30698 }, { "epoch": 190.6791277258567, "grad_norm": 1.40625, "learning_rate": 2.7338480311818183e-07, "loss": 0.1319, "step": 30699 }, { "epoch": 190.68535825545172, "grad_norm": 2.15625, "learning_rate": 2.730215604130032e-07, "loss": 0.2394, "step": 30700 }, { "epoch": 190.69158878504672, "grad_norm": 1.9453125, "learning_rate": 2.726585578611801e-07, "loss": 0.1483, "step": 30701 }, { "epoch": 190.69781931464175, "grad_norm": 1.109375, "learning_rate": 2.722957954662403e-07, "loss": 0.0975, "step": 30702 }, { "epoch": 190.70404984423675, "grad_norm": 1.328125, "learning_rate": 2.719332732317059e-07, "loss": 0.1236, "step": 30703 }, { "epoch": 190.71028037383178, "grad_norm": 2.28125, "learning_rate": 2.7157099116110195e-07, "loss": 0.2329, "step": 30704 }, { "epoch": 190.71651090342678, "grad_norm": 2.25, "learning_rate": 2.712089492579395e-07, "loss": 0.2089, "step": 30705 }, { "epoch": 190.7227414330218, "grad_norm": 1.3203125, "learning_rate": 2.708471475257407e-07, "loss": 0.1039, "step": 30706 }, { "epoch": 190.7289719626168, "grad_norm": 2.875, "learning_rate": 2.7048558596801663e-07, "loss": 0.1743, "step": 30707 }, { "epoch": 190.73520249221184, "grad_norm": 1.8203125, "learning_rate": 2.701242645882812e-07, "loss": 0.1913, "step": 30708 }, { "epoch": 190.74143302180684, "grad_norm": 3.28125, "learning_rate": 2.697631833900427e-07, "loss": 0.2729, "step": 30709 }, { "epoch": 190.74766355140187, "grad_norm": 2.0625, "learning_rate": 2.694023423768066e-07, "loss": 0.1934, "step": 30710 }, { "epoch": 190.75389408099687, "grad_norm": 2.09375, "learning_rate": 2.690417415520785e-07, "loss": 0.1928, "step": 30711 }, { "epoch": 190.7601246105919, "grad_norm": 2.796875, "learning_rate": 2.686813809193611e-07, "loss": 0.254, "step": 30712 }, { "epoch": 190.7663551401869, "grad_norm": 2.53125, "learning_rate": 2.6832126048215167e-07, "loss": 0.2537, "step": 30713 }, { "epoch": 190.77258566978193, "grad_norm": 2.3125, "learning_rate": 2.679613802439529e-07, "loss": 0.1742, "step": 30714 }, { "epoch": 190.77881619937693, "grad_norm": 1.28125, "learning_rate": 2.6760174020825644e-07, "loss": 0.1325, "step": 30715 }, { "epoch": 190.78504672897196, "grad_norm": 1.6640625, "learning_rate": 2.6724234037855403e-07, "loss": 0.1541, "step": 30716 }, { "epoch": 190.79127725856696, "grad_norm": 2.046875, "learning_rate": 2.6688318075834e-07, "loss": 0.2127, "step": 30717 }, { "epoch": 190.797507788162, "grad_norm": 1.4921875, "learning_rate": 2.6652426135110043e-07, "loss": 0.1565, "step": 30718 }, { "epoch": 190.80373831775702, "grad_norm": 1.7578125, "learning_rate": 2.6616558216031874e-07, "loss": 0.1899, "step": 30719 }, { "epoch": 190.80996884735202, "grad_norm": 1.6875, "learning_rate": 2.6580714318948374e-07, "loss": 0.1705, "step": 30720 }, { "epoch": 190.81619937694705, "grad_norm": 2.328125, "learning_rate": 2.6544894444207324e-07, "loss": 0.2118, "step": 30721 }, { "epoch": 190.82242990654206, "grad_norm": 2.734375, "learning_rate": 2.650909859215678e-07, "loss": 0.2925, "step": 30722 }, { "epoch": 190.82866043613708, "grad_norm": 2.6875, "learning_rate": 2.6473326763144237e-07, "loss": 0.1084, "step": 30723 }, { "epoch": 190.8348909657321, "grad_norm": 2.359375, "learning_rate": 2.6437578957517197e-07, "loss": 0.14, "step": 30724 }, { "epoch": 190.84112149532712, "grad_norm": 1.6796875, "learning_rate": 2.640185517562288e-07, "loss": 0.1475, "step": 30725 }, { "epoch": 190.84735202492212, "grad_norm": 2.421875, "learning_rate": 2.6366155417807956e-07, "loss": 0.2187, "step": 30726 }, { "epoch": 190.85358255451715, "grad_norm": 1.796875, "learning_rate": 2.633047968441965e-07, "loss": 0.1098, "step": 30727 }, { "epoch": 190.85981308411215, "grad_norm": 1.359375, "learning_rate": 2.6294827975804347e-07, "loss": 0.0811, "step": 30728 }, { "epoch": 190.86604361370718, "grad_norm": 2.390625, "learning_rate": 2.6259200292307886e-07, "loss": 0.2939, "step": 30729 }, { "epoch": 190.87227414330218, "grad_norm": 2.5, "learning_rate": 2.6223596634276647e-07, "loss": 0.1321, "step": 30730 }, { "epoch": 190.8785046728972, "grad_norm": 3.78125, "learning_rate": 2.61880170020562e-07, "loss": 0.2671, "step": 30731 }, { "epoch": 190.8847352024922, "grad_norm": 2.96875, "learning_rate": 2.615246139599237e-07, "loss": 0.2987, "step": 30732 }, { "epoch": 190.89096573208724, "grad_norm": 2.875, "learning_rate": 2.611692981643044e-07, "loss": 0.1955, "step": 30733 }, { "epoch": 190.89719626168224, "grad_norm": 3.234375, "learning_rate": 2.6081422263715404e-07, "loss": 0.3337, "step": 30734 }, { "epoch": 190.90342679127727, "grad_norm": 2.5625, "learning_rate": 2.6045938738192e-07, "loss": 0.1879, "step": 30735 }, { "epoch": 190.90965732087227, "grad_norm": 2.40625, "learning_rate": 2.6010479240204945e-07, "loss": 0.2815, "step": 30736 }, { "epoch": 190.9158878504673, "grad_norm": 1.1171875, "learning_rate": 2.597504377009868e-07, "loss": 0.1007, "step": 30737 }, { "epoch": 190.9221183800623, "grad_norm": 3.515625, "learning_rate": 2.593963232821739e-07, "loss": 0.3618, "step": 30738 }, { "epoch": 190.92834890965733, "grad_norm": 3.609375, "learning_rate": 2.5904244914904954e-07, "loss": 0.2919, "step": 30739 }, { "epoch": 190.93457943925233, "grad_norm": 1.9453125, "learning_rate": 2.5868881530504985e-07, "loss": 0.2241, "step": 30740 }, { "epoch": 190.94080996884736, "grad_norm": 2.15625, "learning_rate": 2.58335421753611e-07, "loss": 0.1813, "step": 30741 }, { "epoch": 190.94704049844236, "grad_norm": 3.5, "learning_rate": 2.5798226849816364e-07, "loss": 0.225, "step": 30742 }, { "epoch": 190.9532710280374, "grad_norm": 0.953125, "learning_rate": 2.576293555421383e-07, "loss": 0.0974, "step": 30743 }, { "epoch": 190.9595015576324, "grad_norm": 1.8125, "learning_rate": 2.572766828889628e-07, "loss": 0.1332, "step": 30744 }, { "epoch": 190.96573208722742, "grad_norm": 1.5, "learning_rate": 2.5692425054206213e-07, "loss": 0.0986, "step": 30745 }, { "epoch": 190.97196261682242, "grad_norm": 2.640625, "learning_rate": 2.5657205850485866e-07, "loss": 0.2194, "step": 30746 }, { "epoch": 190.97819314641745, "grad_norm": 1.90625, "learning_rate": 2.5622010678077455e-07, "loss": 0.1376, "step": 30747 }, { "epoch": 190.98442367601245, "grad_norm": 2.171875, "learning_rate": 2.558683953732266e-07, "loss": 0.1711, "step": 30748 }, { "epoch": 190.99065420560748, "grad_norm": 1.8125, "learning_rate": 2.5551692428563143e-07, "loss": 0.1402, "step": 30749 }, { "epoch": 190.99688473520249, "grad_norm": 2.515625, "learning_rate": 2.551656935214003e-07, "loss": 0.2109, "step": 30750 }, { "epoch": 191.0, "grad_norm": 2.0, "learning_rate": 2.548147030839498e-07, "loss": 0.1499, "step": 30751 }, { "epoch": 191.00623052959503, "grad_norm": 1.59375, "learning_rate": 2.544639529766829e-07, "loss": 0.1041, "step": 30752 }, { "epoch": 191.01246105919003, "grad_norm": 2.609375, "learning_rate": 2.541134432030079e-07, "loss": 0.2034, "step": 30753 }, { "epoch": 191.01869158878506, "grad_norm": 1.3515625, "learning_rate": 2.537631737663304e-07, "loss": 0.1152, "step": 30754 }, { "epoch": 191.02492211838006, "grad_norm": 1.515625, "learning_rate": 2.5341314467005326e-07, "loss": 0.1411, "step": 30755 }, { "epoch": 191.0311526479751, "grad_norm": 2.546875, "learning_rate": 2.530633559175738e-07, "loss": 0.2747, "step": 30756 }, { "epoch": 191.0373831775701, "grad_norm": 2.953125, "learning_rate": 2.5271380751228646e-07, "loss": 0.263, "step": 30757 }, { "epoch": 191.04361370716512, "grad_norm": 2.28125, "learning_rate": 2.523644994575941e-07, "loss": 0.2042, "step": 30758 }, { "epoch": 191.04984423676012, "grad_norm": 2.796875, "learning_rate": 2.520154317568829e-07, "loss": 0.2149, "step": 30759 }, { "epoch": 191.05607476635515, "grad_norm": 1.890625, "learning_rate": 2.516666044135446e-07, "loss": 0.1673, "step": 30760 }, { "epoch": 191.06230529595015, "grad_norm": 1.734375, "learning_rate": 2.5131801743096815e-07, "loss": 0.151, "step": 30761 }, { "epoch": 191.06853582554518, "grad_norm": 0.9296875, "learning_rate": 2.50969670812537e-07, "loss": 0.1026, "step": 30762 }, { "epoch": 191.07476635514018, "grad_norm": 1.6953125, "learning_rate": 2.506215645616372e-07, "loss": 0.0968, "step": 30763 }, { "epoch": 191.0809968847352, "grad_norm": 1.484375, "learning_rate": 2.5027369868164394e-07, "loss": 0.1537, "step": 30764 }, { "epoch": 191.08722741433021, "grad_norm": 1.546875, "learning_rate": 2.499260731759434e-07, "loss": 0.1331, "step": 30765 }, { "epoch": 191.09345794392524, "grad_norm": 1.375, "learning_rate": 2.495786880479106e-07, "loss": 0.114, "step": 30766 }, { "epoch": 191.09968847352025, "grad_norm": 1.8828125, "learning_rate": 2.492315433009151e-07, "loss": 0.1667, "step": 30767 }, { "epoch": 191.10591900311528, "grad_norm": 2.46875, "learning_rate": 2.4888463893832926e-07, "loss": 0.1325, "step": 30768 }, { "epoch": 191.11214953271028, "grad_norm": 1.5546875, "learning_rate": 2.485379749635225e-07, "loss": 0.1234, "step": 30769 }, { "epoch": 191.1183800623053, "grad_norm": 2.234375, "learning_rate": 2.481915513798644e-07, "loss": 0.2055, "step": 30770 }, { "epoch": 191.1246105919003, "grad_norm": 2.1875, "learning_rate": 2.4784536819071616e-07, "loss": 0.1067, "step": 30771 }, { "epoch": 191.13084112149534, "grad_norm": 2.421875, "learning_rate": 2.474994253994445e-07, "loss": 0.1614, "step": 30772 }, { "epoch": 191.13707165109034, "grad_norm": 2.6875, "learning_rate": 2.471537230094051e-07, "loss": 0.1778, "step": 30773 }, { "epoch": 191.14330218068537, "grad_norm": 2.578125, "learning_rate": 2.468082610239564e-07, "loss": 0.2649, "step": 30774 }, { "epoch": 191.14953271028037, "grad_norm": 1.8125, "learning_rate": 2.46463039446454e-07, "loss": 0.1315, "step": 30775 }, { "epoch": 191.1557632398754, "grad_norm": 2.0625, "learning_rate": 2.4611805828024803e-07, "loss": 0.2331, "step": 30776 }, { "epoch": 191.1619937694704, "grad_norm": 2.390625, "learning_rate": 2.457733175286969e-07, "loss": 0.1586, "step": 30777 }, { "epoch": 191.16822429906543, "grad_norm": 1.5234375, "learning_rate": 2.4542881719513965e-07, "loss": 0.1401, "step": 30778 }, { "epoch": 191.17445482866043, "grad_norm": 1.25, "learning_rate": 2.450845572829291e-07, "loss": 0.0899, "step": 30779 }, { "epoch": 191.18068535825546, "grad_norm": 1.5078125, "learning_rate": 2.447405377954015e-07, "loss": 0.0835, "step": 30780 }, { "epoch": 191.18691588785046, "grad_norm": 2.6875, "learning_rate": 2.4439675873590697e-07, "loss": 0.2569, "step": 30781 }, { "epoch": 191.1931464174455, "grad_norm": 2.546875, "learning_rate": 2.440532201077761e-07, "loss": 0.1649, "step": 30782 }, { "epoch": 191.1993769470405, "grad_norm": 1.40625, "learning_rate": 2.437099219143507e-07, "loss": 0.1202, "step": 30783 }, { "epoch": 191.20560747663552, "grad_norm": 3.0, "learning_rate": 2.433668641589615e-07, "loss": 0.3265, "step": 30784 }, { "epoch": 191.21183800623052, "grad_norm": 1.5703125, "learning_rate": 2.430240468449446e-07, "loss": 0.1324, "step": 30785 }, { "epoch": 191.21806853582555, "grad_norm": 2.1875, "learning_rate": 2.4268146997562524e-07, "loss": 0.2002, "step": 30786 }, { "epoch": 191.22429906542055, "grad_norm": 1.96875, "learning_rate": 2.42339133554334e-07, "loss": 0.169, "step": 30787 }, { "epoch": 191.23052959501558, "grad_norm": 2.015625, "learning_rate": 2.419970375843933e-07, "loss": 0.1377, "step": 30788 }, { "epoch": 191.23676012461058, "grad_norm": 1.828125, "learning_rate": 2.416551820691254e-07, "loss": 0.1679, "step": 30789 }, { "epoch": 191.2429906542056, "grad_norm": 1.265625, "learning_rate": 2.4131356701184994e-07, "loss": 0.1344, "step": 30790 }, { "epoch": 191.2492211838006, "grad_norm": 2.46875, "learning_rate": 2.409721924158892e-07, "loss": 0.1804, "step": 30791 }, { "epoch": 191.25545171339564, "grad_norm": 2.28125, "learning_rate": 2.4063105828455726e-07, "loss": 0.2022, "step": 30792 }, { "epoch": 191.26168224299064, "grad_norm": 1.6171875, "learning_rate": 2.402901646211625e-07, "loss": 0.1674, "step": 30793 }, { "epoch": 191.26791277258567, "grad_norm": 2.359375, "learning_rate": 2.39949511429019e-07, "loss": 0.1607, "step": 30794 }, { "epoch": 191.27414330218068, "grad_norm": 0.99609375, "learning_rate": 2.396090987114352e-07, "loss": 0.0881, "step": 30795 }, { "epoch": 191.2803738317757, "grad_norm": 1.890625, "learning_rate": 2.392689264717196e-07, "loss": 0.148, "step": 30796 }, { "epoch": 191.2866043613707, "grad_norm": 1.4296875, "learning_rate": 2.389289947131723e-07, "loss": 0.12, "step": 30797 }, { "epoch": 191.29283489096574, "grad_norm": 1.8203125, "learning_rate": 2.3858930343909625e-07, "loss": 0.1491, "step": 30798 }, { "epoch": 191.29906542056074, "grad_norm": 1.53125, "learning_rate": 2.3824985265279154e-07, "loss": 0.1267, "step": 30799 }, { "epoch": 191.30529595015577, "grad_norm": 1.671875, "learning_rate": 2.379106423575528e-07, "loss": 0.1459, "step": 30800 }, { "epoch": 191.31152647975077, "grad_norm": 1.7109375, "learning_rate": 2.3757167255667734e-07, "loss": 0.1351, "step": 30801 }, { "epoch": 191.3177570093458, "grad_norm": 0.9921875, "learning_rate": 2.3723294325345425e-07, "loss": 0.0945, "step": 30802 }, { "epoch": 191.3239875389408, "grad_norm": 2.53125, "learning_rate": 2.368944544511753e-07, "loss": 0.1412, "step": 30803 }, { "epoch": 191.33021806853583, "grad_norm": 1.6484375, "learning_rate": 2.3655620615312958e-07, "loss": 0.1568, "step": 30804 }, { "epoch": 191.33644859813083, "grad_norm": 3.890625, "learning_rate": 2.3621819836259496e-07, "loss": 0.3305, "step": 30805 }, { "epoch": 191.34267912772586, "grad_norm": 2.390625, "learning_rate": 2.358804310828633e-07, "loss": 0.2305, "step": 30806 }, { "epoch": 191.34890965732086, "grad_norm": 1.3984375, "learning_rate": 2.355429043172097e-07, "loss": 0.1113, "step": 30807 }, { "epoch": 191.3551401869159, "grad_norm": 1.8203125, "learning_rate": 2.3520561806891496e-07, "loss": 0.1539, "step": 30808 }, { "epoch": 191.3613707165109, "grad_norm": 1.4453125, "learning_rate": 2.3486857234125138e-07, "loss": 0.1289, "step": 30809 }, { "epoch": 191.36760124610592, "grad_norm": 0.859375, "learning_rate": 2.3453176713749693e-07, "loss": 0.0942, "step": 30810 }, { "epoch": 191.37383177570092, "grad_norm": 2.578125, "learning_rate": 2.341952024609212e-07, "loss": 0.146, "step": 30811 }, { "epoch": 191.38006230529595, "grad_norm": 2.625, "learning_rate": 2.3385887831479102e-07, "loss": 0.2963, "step": 30812 }, { "epoch": 191.38629283489095, "grad_norm": 2.65625, "learning_rate": 2.3352279470237048e-07, "loss": 0.2996, "step": 30813 }, { "epoch": 191.39252336448598, "grad_norm": 1.515625, "learning_rate": 2.3318695162693193e-07, "loss": 0.0858, "step": 30814 }, { "epoch": 191.39875389408098, "grad_norm": 1.0234375, "learning_rate": 2.3285134909173112e-07, "loss": 0.1005, "step": 30815 }, { "epoch": 191.404984423676, "grad_norm": 0.73046875, "learning_rate": 2.3251598710002932e-07, "loss": 0.097, "step": 30816 }, { "epoch": 191.41121495327104, "grad_norm": 1.9296875, "learning_rate": 2.3218086565508222e-07, "loss": 0.1658, "step": 30817 }, { "epoch": 191.41744548286604, "grad_norm": 2.109375, "learning_rate": 2.318459847601484e-07, "loss": 0.1841, "step": 30818 }, { "epoch": 191.42367601246107, "grad_norm": 1.1328125, "learning_rate": 2.3151134441847244e-07, "loss": 0.1118, "step": 30819 }, { "epoch": 191.42990654205607, "grad_norm": 2.28125, "learning_rate": 2.3117694463331286e-07, "loss": 0.1185, "step": 30820 }, { "epoch": 191.4361370716511, "grad_norm": 2.375, "learning_rate": 2.3084278540791427e-07, "loss": 0.1802, "step": 30821 }, { "epoch": 191.4423676012461, "grad_norm": 1.8359375, "learning_rate": 2.3050886674551852e-07, "loss": 0.13, "step": 30822 }, { "epoch": 191.44859813084113, "grad_norm": 1.9921875, "learning_rate": 2.3017518864937583e-07, "loss": 0.129, "step": 30823 }, { "epoch": 191.45482866043614, "grad_norm": 1.5390625, "learning_rate": 2.2984175112272245e-07, "loss": 0.1443, "step": 30824 }, { "epoch": 191.46105919003116, "grad_norm": 1.0703125, "learning_rate": 2.2950855416879746e-07, "loss": 0.108, "step": 30825 }, { "epoch": 191.46728971962617, "grad_norm": 2.0625, "learning_rate": 2.291755977908344e-07, "loss": 0.1384, "step": 30826 }, { "epoch": 191.4735202492212, "grad_norm": 2.328125, "learning_rate": 2.288428819920696e-07, "loss": 0.2329, "step": 30827 }, { "epoch": 191.4797507788162, "grad_norm": 1.3046875, "learning_rate": 2.2851040677573653e-07, "loss": 0.1249, "step": 30828 }, { "epoch": 191.48598130841123, "grad_norm": 1.140625, "learning_rate": 2.2817817214506044e-07, "loss": 0.0916, "step": 30829 }, { "epoch": 191.49221183800623, "grad_norm": 2.109375, "learning_rate": 2.2784617810327203e-07, "loss": 0.1922, "step": 30830 }, { "epoch": 191.49844236760126, "grad_norm": 2.734375, "learning_rate": 2.2751442465359096e-07, "loss": 0.2193, "step": 30831 }, { "epoch": 191.50467289719626, "grad_norm": 0.94140625, "learning_rate": 2.2718291179924245e-07, "loss": 0.1022, "step": 30832 }, { "epoch": 191.5109034267913, "grad_norm": 1.75, "learning_rate": 2.2685163954344613e-07, "loss": 0.163, "step": 30833 }, { "epoch": 191.5171339563863, "grad_norm": 1.3046875, "learning_rate": 2.2652060788941887e-07, "loss": 0.1198, "step": 30834 }, { "epoch": 191.52336448598132, "grad_norm": 1.0390625, "learning_rate": 2.2618981684037478e-07, "loss": 0.1026, "step": 30835 }, { "epoch": 191.52959501557632, "grad_norm": 2.046875, "learning_rate": 2.2585926639952791e-07, "loss": 0.1882, "step": 30836 }, { "epoch": 191.53582554517135, "grad_norm": 1.2578125, "learning_rate": 2.2552895657008966e-07, "loss": 0.1207, "step": 30837 }, { "epoch": 191.54205607476635, "grad_norm": 1.515625, "learning_rate": 2.2519888735526295e-07, "loss": 0.1125, "step": 30838 }, { "epoch": 191.54828660436138, "grad_norm": 2.21875, "learning_rate": 2.2486905875825916e-07, "loss": 0.1886, "step": 30839 }, { "epoch": 191.55451713395638, "grad_norm": 2.34375, "learning_rate": 2.245394707822812e-07, "loss": 0.2009, "step": 30840 }, { "epoch": 191.5607476635514, "grad_norm": 1.421875, "learning_rate": 2.2421012343052384e-07, "loss": 0.1307, "step": 30841 }, { "epoch": 191.5669781931464, "grad_norm": 1.6640625, "learning_rate": 2.2388101670619556e-07, "loss": 0.107, "step": 30842 }, { "epoch": 191.57320872274144, "grad_norm": 2.203125, "learning_rate": 2.235521506124827e-07, "loss": 0.215, "step": 30843 }, { "epoch": 191.57943925233644, "grad_norm": 2.25, "learning_rate": 2.2322352515258826e-07, "loss": 0.1601, "step": 30844 }, { "epoch": 191.58566978193147, "grad_norm": 2.53125, "learning_rate": 2.2289514032969584e-07, "loss": 0.2069, "step": 30845 }, { "epoch": 191.59190031152647, "grad_norm": 2.328125, "learning_rate": 2.2256699614700005e-07, "loss": 0.2194, "step": 30846 }, { "epoch": 191.5981308411215, "grad_norm": 3.21875, "learning_rate": 2.2223909260769005e-07, "loss": 0.2803, "step": 30847 }, { "epoch": 191.6043613707165, "grad_norm": 1.7265625, "learning_rate": 2.2191142971494105e-07, "loss": 0.1502, "step": 30848 }, { "epoch": 191.61059190031153, "grad_norm": 3.0625, "learning_rate": 2.2158400747194495e-07, "loss": 0.2293, "step": 30849 }, { "epoch": 191.61682242990653, "grad_norm": 0.95703125, "learning_rate": 2.21256825881877e-07, "loss": 0.0955, "step": 30850 }, { "epoch": 191.62305295950156, "grad_norm": 2.328125, "learning_rate": 2.20929884947918e-07, "loss": 0.192, "step": 30851 }, { "epoch": 191.62928348909657, "grad_norm": 0.94140625, "learning_rate": 2.206031846732376e-07, "loss": 0.1086, "step": 30852 }, { "epoch": 191.6355140186916, "grad_norm": 3.6875, "learning_rate": 2.202767250610138e-07, "loss": 0.1619, "step": 30853 }, { "epoch": 191.6417445482866, "grad_norm": 2.234375, "learning_rate": 2.1995050611441635e-07, "loss": 0.2075, "step": 30854 }, { "epoch": 191.64797507788163, "grad_norm": 1.796875, "learning_rate": 2.196245278366149e-07, "loss": 0.1446, "step": 30855 }, { "epoch": 191.65420560747663, "grad_norm": 0.84765625, "learning_rate": 2.1929879023077083e-07, "loss": 0.1012, "step": 30856 }, { "epoch": 191.66043613707166, "grad_norm": 1.4921875, "learning_rate": 2.18973293300051e-07, "loss": 0.1228, "step": 30857 }, { "epoch": 191.66666666666666, "grad_norm": 1.578125, "learning_rate": 2.1864803704761682e-07, "loss": 0.1227, "step": 30858 }, { "epoch": 191.6728971962617, "grad_norm": 1.9921875, "learning_rate": 2.183230214766241e-07, "loss": 0.1732, "step": 30859 }, { "epoch": 191.6791277258567, "grad_norm": 1.71875, "learning_rate": 2.1799824659023417e-07, "loss": 0.1269, "step": 30860 }, { "epoch": 191.68535825545172, "grad_norm": 1.5390625, "learning_rate": 2.176737123915973e-07, "loss": 0.1532, "step": 30861 }, { "epoch": 191.69158878504672, "grad_norm": 1.78125, "learning_rate": 2.1734941888387206e-07, "loss": 0.1663, "step": 30862 }, { "epoch": 191.69781931464175, "grad_norm": 3.53125, "learning_rate": 2.170253660702004e-07, "loss": 0.1861, "step": 30863 }, { "epoch": 191.70404984423675, "grad_norm": 1.859375, "learning_rate": 2.1670155395372982e-07, "loss": 0.1725, "step": 30864 }, { "epoch": 191.71028037383178, "grad_norm": 2.109375, "learning_rate": 2.1637798253761054e-07, "loss": 0.1923, "step": 30865 }, { "epoch": 191.71651090342678, "grad_norm": 3.171875, "learning_rate": 2.1605465182498176e-07, "loss": 0.348, "step": 30866 }, { "epoch": 191.7227414330218, "grad_norm": 3.296875, "learning_rate": 2.1573156181898534e-07, "loss": 0.3337, "step": 30867 }, { "epoch": 191.7289719626168, "grad_norm": 1.125, "learning_rate": 2.1540871252275775e-07, "loss": 0.1023, "step": 30868 }, { "epoch": 191.73520249221184, "grad_norm": 1.7734375, "learning_rate": 2.1508610393943252e-07, "loss": 0.1554, "step": 30869 }, { "epoch": 191.74143302180684, "grad_norm": 2.703125, "learning_rate": 2.1476373607214884e-07, "loss": 0.2745, "step": 30870 }, { "epoch": 191.74766355140187, "grad_norm": 2.859375, "learning_rate": 2.14441608924032e-07, "loss": 0.1978, "step": 30871 }, { "epoch": 191.75389408099687, "grad_norm": 1.921875, "learning_rate": 2.1411972249821555e-07, "loss": 0.2508, "step": 30872 }, { "epoch": 191.7601246105919, "grad_norm": 2.90625, "learning_rate": 2.1379807679782206e-07, "loss": 0.2123, "step": 30873 }, { "epoch": 191.7663551401869, "grad_norm": 1.09375, "learning_rate": 2.1347667182597399e-07, "loss": 0.0898, "step": 30874 }, { "epoch": 191.77258566978193, "grad_norm": 2.375, "learning_rate": 2.1315550758579662e-07, "loss": 0.1893, "step": 30875 }, { "epoch": 191.77881619937693, "grad_norm": 2.609375, "learning_rate": 2.1283458408040692e-07, "loss": 0.2213, "step": 30876 }, { "epoch": 191.78504672897196, "grad_norm": 2.046875, "learning_rate": 2.125139013129246e-07, "loss": 0.1491, "step": 30877 }, { "epoch": 191.79127725856696, "grad_norm": 1.0703125, "learning_rate": 2.1219345928646106e-07, "loss": 0.1097, "step": 30878 }, { "epoch": 191.797507788162, "grad_norm": 2.53125, "learning_rate": 2.118732580041305e-07, "loss": 0.2441, "step": 30879 }, { "epoch": 191.80373831775702, "grad_norm": 1.234375, "learning_rate": 2.115532974690415e-07, "loss": 0.0975, "step": 30880 }, { "epoch": 191.80996884735202, "grad_norm": 2.046875, "learning_rate": 2.1123357768430275e-07, "loss": 0.205, "step": 30881 }, { "epoch": 191.81619937694705, "grad_norm": 2.109375, "learning_rate": 2.1091409865301726e-07, "loss": 0.1814, "step": 30882 }, { "epoch": 191.82242990654206, "grad_norm": 2.46875, "learning_rate": 2.1059486037828813e-07, "loss": 0.1889, "step": 30883 }, { "epoch": 191.82866043613708, "grad_norm": 2.03125, "learning_rate": 2.1027586286321842e-07, "loss": 0.1081, "step": 30884 }, { "epoch": 191.8348909657321, "grad_norm": 1.265625, "learning_rate": 2.099571061109057e-07, "loss": 0.1268, "step": 30885 }, { "epoch": 191.84112149532712, "grad_norm": 3.5, "learning_rate": 2.0963859012444188e-07, "loss": 0.2967, "step": 30886 }, { "epoch": 191.84735202492212, "grad_norm": 2.28125, "learning_rate": 2.0932031490692728e-07, "loss": 0.1507, "step": 30887 }, { "epoch": 191.85358255451715, "grad_norm": 1.203125, "learning_rate": 2.0900228046144832e-07, "loss": 0.1132, "step": 30888 }, { "epoch": 191.85981308411215, "grad_norm": 2.390625, "learning_rate": 2.086844867910942e-07, "loss": 0.283, "step": 30889 }, { "epoch": 191.86604361370718, "grad_norm": 1.984375, "learning_rate": 2.0836693389895135e-07, "loss": 0.1448, "step": 30890 }, { "epoch": 191.87227414330218, "grad_norm": 3.34375, "learning_rate": 2.0804962178810615e-07, "loss": 0.2818, "step": 30891 }, { "epoch": 191.8785046728972, "grad_norm": 2.203125, "learning_rate": 2.0773255046163953e-07, "loss": 0.273, "step": 30892 }, { "epoch": 191.8847352024922, "grad_norm": 1.2265625, "learning_rate": 2.0741571992262955e-07, "loss": 0.1196, "step": 30893 }, { "epoch": 191.89096573208724, "grad_norm": 1.90625, "learning_rate": 2.070991301741515e-07, "loss": 0.1937, "step": 30894 }, { "epoch": 191.89719626168224, "grad_norm": 3.390625, "learning_rate": 2.0678278121928352e-07, "loss": 0.2992, "step": 30895 }, { "epoch": 191.90342679127727, "grad_norm": 2.4375, "learning_rate": 2.064666730611009e-07, "loss": 0.1484, "step": 30896 }, { "epoch": 191.90965732087227, "grad_norm": 2.109375, "learning_rate": 2.0615080570266509e-07, "loss": 0.2781, "step": 30897 }, { "epoch": 191.9158878504673, "grad_norm": 2.0, "learning_rate": 2.058351791470514e-07, "loss": 0.1683, "step": 30898 }, { "epoch": 191.9221183800623, "grad_norm": 1.6484375, "learning_rate": 2.055197933973241e-07, "loss": 0.1705, "step": 30899 }, { "epoch": 191.92834890965733, "grad_norm": 1.4609375, "learning_rate": 2.0520464845654452e-07, "loss": 0.1118, "step": 30900 }, { "epoch": 191.93457943925233, "grad_norm": 1.0859375, "learning_rate": 2.0488974432777142e-07, "loss": 0.0978, "step": 30901 }, { "epoch": 191.94080996884736, "grad_norm": 1.1328125, "learning_rate": 2.045750810140662e-07, "loss": 0.1004, "step": 30902 }, { "epoch": 191.94704049844236, "grad_norm": 2.359375, "learning_rate": 2.0426065851848476e-07, "loss": 0.1886, "step": 30903 }, { "epoch": 191.9532710280374, "grad_norm": 2.609375, "learning_rate": 2.039464768440802e-07, "loss": 0.2001, "step": 30904 }, { "epoch": 191.9595015576324, "grad_norm": 2.234375, "learning_rate": 2.0363253599390286e-07, "loss": 0.1945, "step": 30905 }, { "epoch": 191.96573208722742, "grad_norm": 0.95703125, "learning_rate": 2.0331883597100586e-07, "loss": 0.1, "step": 30906 }, { "epoch": 191.97196261682242, "grad_norm": 1.796875, "learning_rate": 2.03005376778434e-07, "loss": 0.13, "step": 30907 }, { "epoch": 191.97819314641745, "grad_norm": 1.84375, "learning_rate": 2.026921584192265e-07, "loss": 0.1547, "step": 30908 }, { "epoch": 191.98442367601245, "grad_norm": 1.046875, "learning_rate": 2.0237918089643094e-07, "loss": 0.1047, "step": 30909 }, { "epoch": 191.99065420560748, "grad_norm": 1.4296875, "learning_rate": 2.0206644421308652e-07, "loss": 0.125, "step": 30910 }, { "epoch": 191.99688473520249, "grad_norm": 1.328125, "learning_rate": 2.017539483722297e-07, "loss": 0.1054, "step": 30911 }, { "epoch": 192.0, "grad_norm": 1.953125, "learning_rate": 2.01441693376897e-07, "loss": 0.1157, "step": 30912 }, { "epoch": 192.00623052959503, "grad_norm": 2.5, "learning_rate": 2.011296792301165e-07, "loss": 0.1733, "step": 30913 }, { "epoch": 192.01246105919003, "grad_norm": 2.59375, "learning_rate": 2.0081790593492468e-07, "loss": 0.1179, "step": 30914 }, { "epoch": 192.01869158878506, "grad_norm": 1.5078125, "learning_rate": 2.0050637349434686e-07, "loss": 0.1278, "step": 30915 }, { "epoch": 192.02492211838006, "grad_norm": 3.21875, "learning_rate": 2.0019508191140568e-07, "loss": 0.1412, "step": 30916 }, { "epoch": 192.0311526479751, "grad_norm": 3.390625, "learning_rate": 1.9988403118912925e-07, "loss": 0.2111, "step": 30917 }, { "epoch": 192.0373831775701, "grad_norm": 3.109375, "learning_rate": 1.995732213305346e-07, "loss": 0.3389, "step": 30918 }, { "epoch": 192.04361370716512, "grad_norm": 3.078125, "learning_rate": 1.9926265233864429e-07, "loss": 0.2452, "step": 30919 }, { "epoch": 192.04984423676012, "grad_norm": 1.4375, "learning_rate": 1.989523242164726e-07, "loss": 0.1029, "step": 30920 }, { "epoch": 192.05607476635515, "grad_norm": 2.671875, "learning_rate": 1.9864223696703377e-07, "loss": 0.3546, "step": 30921 }, { "epoch": 192.06230529595015, "grad_norm": 2.140625, "learning_rate": 1.9833239059333654e-07, "loss": 0.3194, "step": 30922 }, { "epoch": 192.06853582554518, "grad_norm": 0.90625, "learning_rate": 1.9802278509839512e-07, "loss": 0.0846, "step": 30923 }, { "epoch": 192.07476635514018, "grad_norm": 3.09375, "learning_rate": 1.977134204852127e-07, "loss": 0.0941, "step": 30924 }, { "epoch": 192.0809968847352, "grad_norm": 2.1875, "learning_rate": 1.9740429675680072e-07, "loss": 0.1984, "step": 30925 }, { "epoch": 192.08722741433021, "grad_norm": 2.703125, "learning_rate": 1.9709541391615128e-07, "loss": 0.2637, "step": 30926 }, { "epoch": 192.09345794392524, "grad_norm": 1.9375, "learning_rate": 1.967867719662675e-07, "loss": 0.1193, "step": 30927 }, { "epoch": 192.09968847352025, "grad_norm": 3.078125, "learning_rate": 1.9647837091015253e-07, "loss": 0.122, "step": 30928 }, { "epoch": 192.10591900311528, "grad_norm": 2.5, "learning_rate": 1.9617021075079567e-07, "loss": 0.1816, "step": 30929 }, { "epoch": 192.11214953271028, "grad_norm": 1.1328125, "learning_rate": 1.9586229149119174e-07, "loss": 0.1166, "step": 30930 }, { "epoch": 192.1183800623053, "grad_norm": 1.796875, "learning_rate": 1.9555461313433276e-07, "loss": 0.2371, "step": 30931 }, { "epoch": 192.1246105919003, "grad_norm": 1.328125, "learning_rate": 1.9524717568320526e-07, "loss": 0.1207, "step": 30932 }, { "epoch": 192.13084112149534, "grad_norm": 1.0703125, "learning_rate": 1.9493997914079853e-07, "loss": 0.0955, "step": 30933 }, { "epoch": 192.13707165109034, "grad_norm": 1.8984375, "learning_rate": 1.9463302351008795e-07, "loss": 0.1198, "step": 30934 }, { "epoch": 192.14330218068537, "grad_norm": 2.28125, "learning_rate": 1.9432630879406278e-07, "loss": 0.2081, "step": 30935 }, { "epoch": 192.14953271028037, "grad_norm": 2.671875, "learning_rate": 1.9401983499569842e-07, "loss": 0.2166, "step": 30936 }, { "epoch": 192.1557632398754, "grad_norm": 1.359375, "learning_rate": 1.937136021179703e-07, "loss": 0.1415, "step": 30937 }, { "epoch": 192.1619937694704, "grad_norm": 1.0234375, "learning_rate": 1.9340761016385934e-07, "loss": 0.1082, "step": 30938 }, { "epoch": 192.16822429906543, "grad_norm": 2.3125, "learning_rate": 1.9310185913632706e-07, "loss": 0.1518, "step": 30939 }, { "epoch": 192.17445482866043, "grad_norm": 1.2734375, "learning_rate": 1.9279634903835164e-07, "loss": 0.1149, "step": 30940 }, { "epoch": 192.18068535825546, "grad_norm": 1.328125, "learning_rate": 1.924910798728946e-07, "loss": 0.123, "step": 30941 }, { "epoch": 192.18691588785046, "grad_norm": 1.625, "learning_rate": 1.9218605164292302e-07, "loss": 0.1362, "step": 30942 }, { "epoch": 192.1931464174455, "grad_norm": 2.34375, "learning_rate": 1.9188126435140395e-07, "loss": 0.2622, "step": 30943 }, { "epoch": 192.1993769470405, "grad_norm": 1.40625, "learning_rate": 1.9157671800129062e-07, "loss": 0.1162, "step": 30944 }, { "epoch": 192.20560747663552, "grad_norm": 1.6015625, "learning_rate": 1.9127241259554175e-07, "loss": 0.1411, "step": 30945 }, { "epoch": 192.21183800623052, "grad_norm": 1.875, "learning_rate": 1.9096834813711328e-07, "loss": 0.1334, "step": 30946 }, { "epoch": 192.21806853582555, "grad_norm": 2.78125, "learning_rate": 1.90664524628964e-07, "loss": 0.183, "step": 30947 }, { "epoch": 192.22429906542055, "grad_norm": 2.46875, "learning_rate": 1.90360942074036e-07, "loss": 0.1, "step": 30948 }, { "epoch": 192.23052959501558, "grad_norm": 2.046875, "learning_rate": 1.9005760047528241e-07, "loss": 0.1701, "step": 30949 }, { "epoch": 192.23676012461058, "grad_norm": 1.921875, "learning_rate": 1.8975449983565097e-07, "loss": 0.1698, "step": 30950 }, { "epoch": 192.2429906542056, "grad_norm": 2.359375, "learning_rate": 1.894516401580837e-07, "loss": 0.1448, "step": 30951 }, { "epoch": 192.2492211838006, "grad_norm": 1.640625, "learning_rate": 1.8914902144551717e-07, "loss": 0.153, "step": 30952 }, { "epoch": 192.25545171339564, "grad_norm": 1.5078125, "learning_rate": 1.8884664370089898e-07, "loss": 0.1169, "step": 30953 }, { "epoch": 192.26168224299064, "grad_norm": 3.0625, "learning_rate": 1.8854450692716018e-07, "loss": 0.3235, "step": 30954 }, { "epoch": 192.26791277258567, "grad_norm": 1.4765625, "learning_rate": 1.8824261112723728e-07, "loss": 0.1071, "step": 30955 }, { "epoch": 192.27414330218068, "grad_norm": 1.3515625, "learning_rate": 1.8794095630406127e-07, "loss": 0.1203, "step": 30956 }, { "epoch": 192.2803738317757, "grad_norm": 3.03125, "learning_rate": 1.8763954246056314e-07, "loss": 0.0914, "step": 30957 }, { "epoch": 192.2866043613707, "grad_norm": 2.359375, "learning_rate": 1.873383695996711e-07, "loss": 0.2059, "step": 30958 }, { "epoch": 192.29283489096574, "grad_norm": 1.1640625, "learning_rate": 1.870374377243078e-07, "loss": 0.0837, "step": 30959 }, { "epoch": 192.29906542056074, "grad_norm": 2.53125, "learning_rate": 1.8673674683739595e-07, "loss": 0.1943, "step": 30960 }, { "epoch": 192.30529595015577, "grad_norm": 1.8046875, "learning_rate": 1.8643629694185538e-07, "loss": 0.1458, "step": 30961 }, { "epoch": 192.31152647975077, "grad_norm": 2.546875, "learning_rate": 1.861360880406088e-07, "loss": 0.1972, "step": 30962 }, { "epoch": 192.3177570093458, "grad_norm": 2.1875, "learning_rate": 1.8583612013656771e-07, "loss": 0.2115, "step": 30963 }, { "epoch": 192.3239875389408, "grad_norm": 2.828125, "learning_rate": 1.855363932326465e-07, "loss": 0.2232, "step": 30964 }, { "epoch": 192.33021806853583, "grad_norm": 1.796875, "learning_rate": 1.852369073317567e-07, "loss": 0.1766, "step": 30965 }, { "epoch": 192.33644859813083, "grad_norm": 2.203125, "learning_rate": 1.849376624368071e-07, "loss": 0.1709, "step": 30966 }, { "epoch": 192.34267912772586, "grad_norm": 2.484375, "learning_rate": 1.8463865855070095e-07, "loss": 0.1406, "step": 30967 }, { "epoch": 192.34890965732086, "grad_norm": 1.125, "learning_rate": 1.8433989567634425e-07, "loss": 0.1185, "step": 30968 }, { "epoch": 192.3551401869159, "grad_norm": 1.484375, "learning_rate": 1.8404137381664022e-07, "loss": 0.1373, "step": 30969 }, { "epoch": 192.3613707165109, "grad_norm": 1.9609375, "learning_rate": 1.8374309297448933e-07, "loss": 0.1003, "step": 30970 }, { "epoch": 192.36760124610592, "grad_norm": 3.453125, "learning_rate": 1.8344505315278372e-07, "loss": 0.1932, "step": 30971 }, { "epoch": 192.37383177570092, "grad_norm": 1.84375, "learning_rate": 1.8314725435442105e-07, "loss": 0.1103, "step": 30972 }, { "epoch": 192.38006230529595, "grad_norm": 2.421875, "learning_rate": 1.8284969658229344e-07, "loss": 0.2281, "step": 30973 }, { "epoch": 192.38629283489095, "grad_norm": 1.1953125, "learning_rate": 1.8255237983928752e-07, "loss": 0.1084, "step": 30974 }, { "epoch": 192.39252336448598, "grad_norm": 2.796875, "learning_rate": 1.822553041282954e-07, "loss": 0.2099, "step": 30975 }, { "epoch": 192.39875389408098, "grad_norm": 1.8828125, "learning_rate": 1.8195846945220363e-07, "loss": 0.1621, "step": 30976 }, { "epoch": 192.404984423676, "grad_norm": 2.171875, "learning_rate": 1.8166187581389048e-07, "loss": 0.2435, "step": 30977 }, { "epoch": 192.41121495327104, "grad_norm": 1.7421875, "learning_rate": 1.81365523216237e-07, "loss": 0.1724, "step": 30978 }, { "epoch": 192.41744548286604, "grad_norm": 1.625, "learning_rate": 1.810694116621242e-07, "loss": 0.134, "step": 30979 }, { "epoch": 192.42367601246107, "grad_norm": 2.0625, "learning_rate": 1.807735411544248e-07, "loss": 0.1522, "step": 30980 }, { "epoch": 192.42990654205607, "grad_norm": 1.078125, "learning_rate": 1.8047791169601426e-07, "loss": 0.113, "step": 30981 }, { "epoch": 192.4361370716511, "grad_norm": 2.484375, "learning_rate": 1.8018252328976527e-07, "loss": 0.2048, "step": 30982 }, { "epoch": 192.4423676012461, "grad_norm": 3.53125, "learning_rate": 1.79887375938545e-07, "loss": 0.3108, "step": 30983 }, { "epoch": 192.44859813084113, "grad_norm": 2.453125, "learning_rate": 1.795924696452206e-07, "loss": 0.2742, "step": 30984 }, { "epoch": 192.45482866043614, "grad_norm": 1.953125, "learning_rate": 1.7929780441265365e-07, "loss": 0.1509, "step": 30985 }, { "epoch": 192.46105919003116, "grad_norm": 2.1875, "learning_rate": 1.7900338024370579e-07, "loss": 0.1957, "step": 30986 }, { "epoch": 192.46728971962617, "grad_norm": 1.4921875, "learning_rate": 1.7870919714124136e-07, "loss": 0.1295, "step": 30987 }, { "epoch": 192.4735202492212, "grad_norm": 0.9375, "learning_rate": 1.784152551081164e-07, "loss": 0.1005, "step": 30988 }, { "epoch": 192.4797507788162, "grad_norm": 0.84375, "learning_rate": 1.7812155414718424e-07, "loss": 0.0924, "step": 30989 }, { "epoch": 192.48598130841123, "grad_norm": 2.640625, "learning_rate": 1.778280942612953e-07, "loss": 0.195, "step": 30990 }, { "epoch": 192.49221183800623, "grad_norm": 2.03125, "learning_rate": 1.7753487545330017e-07, "loss": 0.1772, "step": 30991 }, { "epoch": 192.49844236760126, "grad_norm": 0.90234375, "learning_rate": 1.7724189772605204e-07, "loss": 0.0931, "step": 30992 }, { "epoch": 192.50467289719626, "grad_norm": 2.28125, "learning_rate": 1.7694916108239035e-07, "loss": 0.2065, "step": 30993 }, { "epoch": 192.5109034267913, "grad_norm": 1.5234375, "learning_rate": 1.7665666552516002e-07, "loss": 0.0925, "step": 30994 }, { "epoch": 192.5171339563863, "grad_norm": 1.1484375, "learning_rate": 1.7636441105720324e-07, "loss": 0.1169, "step": 30995 }, { "epoch": 192.52336448598132, "grad_norm": 1.6796875, "learning_rate": 1.7607239768135942e-07, "loss": 0.1762, "step": 30996 }, { "epoch": 192.52959501557632, "grad_norm": 1.703125, "learning_rate": 1.757806254004596e-07, "loss": 0.1446, "step": 30997 }, { "epoch": 192.53582554517135, "grad_norm": 2.34375, "learning_rate": 1.7548909421734038e-07, "loss": 0.1458, "step": 30998 }, { "epoch": 192.54205607476635, "grad_norm": 2.359375, "learning_rate": 1.7519780413483566e-07, "loss": 0.1764, "step": 30999 }, { "epoch": 192.54828660436138, "grad_norm": 2.484375, "learning_rate": 1.7490675515577092e-07, "loss": 0.206, "step": 31000 }, { "epoch": 192.55451713395638, "grad_norm": 3.09375, "learning_rate": 1.746159472829745e-07, "loss": 0.3209, "step": 31001 }, { "epoch": 192.5607476635514, "grad_norm": 1.8203125, "learning_rate": 1.7432538051927182e-07, "loss": 0.1551, "step": 31002 }, { "epoch": 192.5669781931464, "grad_norm": 0.8515625, "learning_rate": 1.7403505486748294e-07, "loss": 0.0803, "step": 31003 }, { "epoch": 192.57320872274144, "grad_norm": 1.2421875, "learning_rate": 1.7374497033042503e-07, "loss": 0.0837, "step": 31004 }, { "epoch": 192.57943925233644, "grad_norm": 2.046875, "learning_rate": 1.734551269109208e-07, "loss": 0.1592, "step": 31005 }, { "epoch": 192.58566978193147, "grad_norm": 1.28125, "learning_rate": 1.731655246117847e-07, "loss": 0.1168, "step": 31006 }, { "epoch": 192.59190031152647, "grad_norm": 1.40625, "learning_rate": 1.7287616343582558e-07, "loss": 0.0866, "step": 31007 }, { "epoch": 192.5981308411215, "grad_norm": 2.625, "learning_rate": 1.7258704338585784e-07, "loss": 0.2202, "step": 31008 }, { "epoch": 192.6043613707165, "grad_norm": 1.359375, "learning_rate": 1.722981644646876e-07, "loss": 0.1251, "step": 31009 }, { "epoch": 192.61059190031153, "grad_norm": 1.296875, "learning_rate": 1.7200952667512093e-07, "loss": 0.1093, "step": 31010 }, { "epoch": 192.61682242990653, "grad_norm": 2.09375, "learning_rate": 1.7172113001995837e-07, "loss": 0.2248, "step": 31011 }, { "epoch": 192.62305295950156, "grad_norm": 2.328125, "learning_rate": 1.7143297450200324e-07, "loss": 0.1869, "step": 31012 }, { "epoch": 192.62928348909657, "grad_norm": 1.5625, "learning_rate": 1.7114506012405607e-07, "loss": 0.1641, "step": 31013 }, { "epoch": 192.6355140186916, "grad_norm": 2.234375, "learning_rate": 1.7085738688891185e-07, "loss": 0.1629, "step": 31014 }, { "epoch": 192.6417445482866, "grad_norm": 1.8515625, "learning_rate": 1.7056995479936277e-07, "loss": 0.1592, "step": 31015 }, { "epoch": 192.64797507788163, "grad_norm": 1.7578125, "learning_rate": 1.7028276385820108e-07, "loss": 0.1353, "step": 31016 }, { "epoch": 192.65420560747663, "grad_norm": 0.77734375, "learning_rate": 1.699958140682162e-07, "loss": 0.0856, "step": 31017 }, { "epoch": 192.66043613707166, "grad_norm": 2.484375, "learning_rate": 1.6970910543219754e-07, "loss": 0.2028, "step": 31018 }, { "epoch": 192.66666666666666, "grad_norm": 1.4375, "learning_rate": 1.6942263795292623e-07, "loss": 0.128, "step": 31019 }, { "epoch": 192.6728971962617, "grad_norm": 0.83984375, "learning_rate": 1.6913641163318338e-07, "loss": 0.097, "step": 31020 }, { "epoch": 192.6791277258567, "grad_norm": 1.9375, "learning_rate": 1.6885042647575566e-07, "loss": 0.1567, "step": 31021 }, { "epoch": 192.68535825545172, "grad_norm": 2.59375, "learning_rate": 1.6856468248341307e-07, "loss": 0.2343, "step": 31022 }, { "epoch": 192.69158878504672, "grad_norm": 2.125, "learning_rate": 1.682791796589367e-07, "loss": 0.1401, "step": 31023 }, { "epoch": 192.69781931464175, "grad_norm": 2.78125, "learning_rate": 1.6799391800509377e-07, "loss": 0.2527, "step": 31024 }, { "epoch": 192.70404984423675, "grad_norm": 4.40625, "learning_rate": 1.677088975246599e-07, "loss": 0.2528, "step": 31025 }, { "epoch": 192.71028037383178, "grad_norm": 2.046875, "learning_rate": 1.674241182203995e-07, "loss": 0.1758, "step": 31026 }, { "epoch": 192.71651090342678, "grad_norm": 1.8125, "learning_rate": 1.6713958009508258e-07, "loss": 0.1122, "step": 31027 }, { "epoch": 192.7227414330218, "grad_norm": 2.71875, "learning_rate": 1.6685528315146803e-07, "loss": 0.238, "step": 31028 }, { "epoch": 192.7289719626168, "grad_norm": 2.765625, "learning_rate": 1.6657122739232035e-07, "loss": 0.2491, "step": 31029 }, { "epoch": 192.73520249221184, "grad_norm": 2.03125, "learning_rate": 1.6628741282039283e-07, "loss": 0.1673, "step": 31030 }, { "epoch": 192.74143302180684, "grad_norm": 2.546875, "learning_rate": 1.6600383943845e-07, "loss": 0.2264, "step": 31031 }, { "epoch": 192.74766355140187, "grad_norm": 0.75, "learning_rate": 1.6572050724924237e-07, "loss": 0.0875, "step": 31032 }, { "epoch": 192.75389408099687, "grad_norm": 1.7578125, "learning_rate": 1.6543741625552057e-07, "loss": 0.1485, "step": 31033 }, { "epoch": 192.7601246105919, "grad_norm": 2.15625, "learning_rate": 1.6515456646003512e-07, "loss": 0.2014, "step": 31034 }, { "epoch": 192.7663551401869, "grad_norm": 2.5, "learning_rate": 1.648719578655311e-07, "loss": 0.205, "step": 31035 }, { "epoch": 192.77258566978193, "grad_norm": 1.8359375, "learning_rate": 1.6458959047475632e-07, "loss": 0.1699, "step": 31036 }, { "epoch": 192.77881619937693, "grad_norm": 2.0625, "learning_rate": 1.6430746429045295e-07, "loss": 0.1738, "step": 31037 }, { "epoch": 192.78504672897196, "grad_norm": 2.796875, "learning_rate": 1.6402557931535778e-07, "loss": 0.2136, "step": 31038 }, { "epoch": 192.79127725856696, "grad_norm": 1.9375, "learning_rate": 1.63743935552213e-07, "loss": 0.1787, "step": 31039 }, { "epoch": 192.797507788162, "grad_norm": 1.8203125, "learning_rate": 1.6346253300375258e-07, "loss": 0.1451, "step": 31040 }, { "epoch": 192.80373831775702, "grad_norm": 1.453125, "learning_rate": 1.6318137167270764e-07, "loss": 0.1107, "step": 31041 }, { "epoch": 192.80996884735202, "grad_norm": 2.171875, "learning_rate": 1.6290045156180932e-07, "loss": 0.1244, "step": 31042 }, { "epoch": 192.81619937694705, "grad_norm": 1.109375, "learning_rate": 1.6261977267378602e-07, "loss": 0.0938, "step": 31043 }, { "epoch": 192.82242990654206, "grad_norm": 2.640625, "learning_rate": 1.623393350113661e-07, "loss": 0.2073, "step": 31044 }, { "epoch": 192.82866043613708, "grad_norm": 1.890625, "learning_rate": 1.620591385772696e-07, "loss": 0.1406, "step": 31045 }, { "epoch": 192.8348909657321, "grad_norm": 1.765625, "learning_rate": 1.6177918337422216e-07, "loss": 0.1165, "step": 31046 }, { "epoch": 192.84112149532712, "grad_norm": 1.8203125, "learning_rate": 1.614994694049382e-07, "loss": 0.1405, "step": 31047 }, { "epoch": 192.84735202492212, "grad_norm": 2.25, "learning_rate": 1.6121999667213782e-07, "loss": 0.1092, "step": 31048 }, { "epoch": 192.85358255451715, "grad_norm": 1.6171875, "learning_rate": 1.6094076517853274e-07, "loss": 0.0981, "step": 31049 }, { "epoch": 192.85981308411215, "grad_norm": 1.515625, "learning_rate": 1.6066177492683741e-07, "loss": 0.1292, "step": 31050 }, { "epoch": 192.86604361370718, "grad_norm": 1.640625, "learning_rate": 1.6038302591975806e-07, "loss": 0.1688, "step": 31051 }, { "epoch": 192.87227414330218, "grad_norm": 0.9765625, "learning_rate": 1.6010451816000637e-07, "loss": 0.107, "step": 31052 }, { "epoch": 192.8785046728972, "grad_norm": 1.2734375, "learning_rate": 1.5982625165028575e-07, "loss": 0.1086, "step": 31053 }, { "epoch": 192.8847352024922, "grad_norm": 1.09375, "learning_rate": 1.5954822639329682e-07, "loss": 0.098, "step": 31054 }, { "epoch": 192.89096573208724, "grad_norm": 2.890625, "learning_rate": 1.5927044239174015e-07, "loss": 0.3286, "step": 31055 }, { "epoch": 192.89719626168224, "grad_norm": 1.53125, "learning_rate": 1.5899289964831365e-07, "loss": 0.1239, "step": 31056 }, { "epoch": 192.90342679127727, "grad_norm": 2.796875, "learning_rate": 1.5871559816571514e-07, "loss": 0.2548, "step": 31057 }, { "epoch": 192.90965732087227, "grad_norm": 2.484375, "learning_rate": 1.5843853794663687e-07, "loss": 0.1741, "step": 31058 }, { "epoch": 192.9158878504673, "grad_norm": 2.640625, "learning_rate": 1.5816171899377118e-07, "loss": 0.2181, "step": 31059 }, { "epoch": 192.9221183800623, "grad_norm": 2.4375, "learning_rate": 1.5788514130980203e-07, "loss": 0.1604, "step": 31060 }, { "epoch": 192.92834890965733, "grad_norm": 1.21875, "learning_rate": 1.576088048974189e-07, "loss": 0.0996, "step": 31061 }, { "epoch": 192.93457943925233, "grad_norm": 1.078125, "learning_rate": 1.5733270975930304e-07, "loss": 0.1072, "step": 31062 }, { "epoch": 192.94080996884736, "grad_norm": 2.640625, "learning_rate": 1.5705685589814112e-07, "loss": 0.2371, "step": 31063 }, { "epoch": 192.94704049844236, "grad_norm": 1.8828125, "learning_rate": 1.5678124331660605e-07, "loss": 0.1543, "step": 31064 }, { "epoch": 192.9532710280374, "grad_norm": 1.3359375, "learning_rate": 1.56505872017379e-07, "loss": 0.1241, "step": 31065 }, { "epoch": 192.9595015576324, "grad_norm": 3.09375, "learning_rate": 1.562307420031328e-07, "loss": 0.2838, "step": 31066 }, { "epoch": 192.96573208722742, "grad_norm": 1.9140625, "learning_rate": 1.559558532765404e-07, "loss": 0.1875, "step": 31067 }, { "epoch": 192.97196261682242, "grad_norm": 2.265625, "learning_rate": 1.556812058402718e-07, "loss": 0.1996, "step": 31068 }, { "epoch": 192.97819314641745, "grad_norm": 1.6484375, "learning_rate": 1.5540679969699156e-07, "loss": 0.2024, "step": 31069 }, { "epoch": 192.98442367601245, "grad_norm": 2.40625, "learning_rate": 1.55132634849367e-07, "loss": 0.2154, "step": 31070 }, { "epoch": 192.99065420560748, "grad_norm": 1.921875, "learning_rate": 1.5485871130005992e-07, "loss": 0.1286, "step": 31071 }, { "epoch": 192.99688473520249, "grad_norm": 1.5546875, "learning_rate": 1.5458502905173476e-07, "loss": 0.1399, "step": 31072 }, { "epoch": 193.0, "grad_norm": 1.9296875, "learning_rate": 1.5431158810704506e-07, "loss": 0.119, "step": 31073 }, { "epoch": 193.00623052959503, "grad_norm": 2.171875, "learning_rate": 1.5403838846864694e-07, "loss": 0.1735, "step": 31074 }, { "epoch": 193.01246105919003, "grad_norm": 2.9375, "learning_rate": 1.5376543013919388e-07, "loss": 0.2006, "step": 31075 }, { "epoch": 193.01869158878506, "grad_norm": 1.21875, "learning_rate": 1.5349271312133927e-07, "loss": 0.1125, "step": 31076 }, { "epoch": 193.02492211838006, "grad_norm": 2.828125, "learning_rate": 1.5322023741772828e-07, "loss": 0.2281, "step": 31077 }, { "epoch": 193.0311526479751, "grad_norm": 2.140625, "learning_rate": 1.5294800303101153e-07, "loss": 0.1505, "step": 31078 }, { "epoch": 193.0373831775701, "grad_norm": 1.171875, "learning_rate": 1.526760099638286e-07, "loss": 0.1224, "step": 31079 }, { "epoch": 193.04361370716512, "grad_norm": 1.640625, "learning_rate": 1.5240425821882453e-07, "loss": 0.1088, "step": 31080 }, { "epoch": 193.04984423676012, "grad_norm": 2.328125, "learning_rate": 1.5213274779863617e-07, "loss": 0.1047, "step": 31081 }, { "epoch": 193.05607476635515, "grad_norm": 2.546875, "learning_rate": 1.5186147870590307e-07, "loss": 0.188, "step": 31082 }, { "epoch": 193.06230529595015, "grad_norm": 1.1640625, "learning_rate": 1.515904509432564e-07, "loss": 0.0915, "step": 31083 }, { "epoch": 193.06853582554518, "grad_norm": 1.2890625, "learning_rate": 1.5131966451333302e-07, "loss": 0.1038, "step": 31084 }, { "epoch": 193.07476635514018, "grad_norm": 1.2265625, "learning_rate": 1.5104911941875578e-07, "loss": 0.1238, "step": 31085 }, { "epoch": 193.0809968847352, "grad_norm": 2.203125, "learning_rate": 1.5077881566215868e-07, "loss": 0.2215, "step": 31086 }, { "epoch": 193.08722741433021, "grad_norm": 2.140625, "learning_rate": 1.5050875324616743e-07, "loss": 0.2031, "step": 31087 }, { "epoch": 193.09345794392524, "grad_norm": 2.640625, "learning_rate": 1.5023893217339936e-07, "loss": 0.1696, "step": 31088 }, { "epoch": 193.09968847352025, "grad_norm": 1.0859375, "learning_rate": 1.4996935244648014e-07, "loss": 0.1016, "step": 31089 }, { "epoch": 193.10591900311528, "grad_norm": 0.86328125, "learning_rate": 1.4970001406802436e-07, "loss": 0.1084, "step": 31090 }, { "epoch": 193.11214953271028, "grad_norm": 1.6953125, "learning_rate": 1.4943091704065216e-07, "loss": 0.1695, "step": 31091 }, { "epoch": 193.1183800623053, "grad_norm": 1.546875, "learning_rate": 1.4916206136697253e-07, "loss": 0.1455, "step": 31092 }, { "epoch": 193.1246105919003, "grad_norm": 2.484375, "learning_rate": 1.488934470495973e-07, "loss": 0.2602, "step": 31093 }, { "epoch": 193.13084112149534, "grad_norm": 2.390625, "learning_rate": 1.4862507409113823e-07, "loss": 0.1765, "step": 31094 }, { "epoch": 193.13707165109034, "grad_norm": 2.21875, "learning_rate": 1.4835694249420162e-07, "loss": 0.2001, "step": 31095 }, { "epoch": 193.14330218068537, "grad_norm": 1.3515625, "learning_rate": 1.4808905226138815e-07, "loss": 0.1372, "step": 31096 }, { "epoch": 193.14953271028037, "grad_norm": 1.453125, "learning_rate": 1.4782140339530127e-07, "loss": 0.1449, "step": 31097 }, { "epoch": 193.1557632398754, "grad_norm": 2.15625, "learning_rate": 1.475539958985417e-07, "loss": 0.1761, "step": 31098 }, { "epoch": 193.1619937694704, "grad_norm": 2.40625, "learning_rate": 1.4728682977370457e-07, "loss": 0.2282, "step": 31099 }, { "epoch": 193.16822429906543, "grad_norm": 1.7890625, "learning_rate": 1.470199050233878e-07, "loss": 0.1773, "step": 31100 }, { "epoch": 193.17445482866043, "grad_norm": 1.78125, "learning_rate": 1.4675322165017825e-07, "loss": 0.1166, "step": 31101 }, { "epoch": 193.18068535825546, "grad_norm": 2.0625, "learning_rate": 1.4648677965667378e-07, "loss": 0.1285, "step": 31102 }, { "epoch": 193.18691588785046, "grad_norm": 2.546875, "learning_rate": 1.462205790454557e-07, "loss": 0.2108, "step": 31103 }, { "epoch": 193.1931464174455, "grad_norm": 2.171875, "learning_rate": 1.4595461981911086e-07, "loss": 0.1869, "step": 31104 }, { "epoch": 193.1993769470405, "grad_norm": 2.59375, "learning_rate": 1.4568890198022323e-07, "loss": 0.1621, "step": 31105 }, { "epoch": 193.20560747663552, "grad_norm": 2.640625, "learning_rate": 1.4542342553137413e-07, "loss": 0.1163, "step": 31106 }, { "epoch": 193.21183800623052, "grad_norm": 1.6875, "learning_rate": 1.4515819047513922e-07, "loss": 0.1176, "step": 31107 }, { "epoch": 193.21806853582555, "grad_norm": 1.09375, "learning_rate": 1.4489319681409707e-07, "loss": 0.0947, "step": 31108 }, { "epoch": 193.22429906542055, "grad_norm": 1.015625, "learning_rate": 1.446284445508206e-07, "loss": 0.0945, "step": 31109 }, { "epoch": 193.23052959501558, "grad_norm": 1.6640625, "learning_rate": 1.443639336878827e-07, "loss": 0.1226, "step": 31110 }, { "epoch": 193.23676012461058, "grad_norm": 1.4921875, "learning_rate": 1.4409966422785083e-07, "loss": 0.113, "step": 31111 }, { "epoch": 193.2429906542056, "grad_norm": 2.953125, "learning_rate": 1.4383563617328954e-07, "loss": 0.2338, "step": 31112 }, { "epoch": 193.2492211838006, "grad_norm": 2.03125, "learning_rate": 1.435718495267663e-07, "loss": 0.1606, "step": 31113 }, { "epoch": 193.25545171339564, "grad_norm": 2.453125, "learning_rate": 1.4330830429084286e-07, "loss": 0.1853, "step": 31114 }, { "epoch": 193.26168224299064, "grad_norm": 2.5625, "learning_rate": 1.4304500046807556e-07, "loss": 0.2085, "step": 31115 }, { "epoch": 193.26791277258567, "grad_norm": 1.78125, "learning_rate": 1.4278193806102624e-07, "loss": 0.1571, "step": 31116 }, { "epoch": 193.27414330218068, "grad_norm": 2.875, "learning_rate": 1.425191170722484e-07, "loss": 0.202, "step": 31117 }, { "epoch": 193.2803738317757, "grad_norm": 2.65625, "learning_rate": 1.4225653750429003e-07, "loss": 0.2312, "step": 31118 }, { "epoch": 193.2866043613707, "grad_norm": 3.265625, "learning_rate": 1.4199419935970737e-07, "loss": 0.2889, "step": 31119 }, { "epoch": 193.29283489096574, "grad_norm": 2.796875, "learning_rate": 1.4173210264104564e-07, "loss": 0.1843, "step": 31120 }, { "epoch": 193.29906542056074, "grad_norm": 2.5625, "learning_rate": 1.414702473508528e-07, "loss": 0.1492, "step": 31121 }, { "epoch": 193.30529595015577, "grad_norm": 3.0625, "learning_rate": 1.4120863349166845e-07, "loss": 0.1111, "step": 31122 }, { "epoch": 193.31152647975077, "grad_norm": 2.578125, "learning_rate": 1.4094726106603505e-07, "loss": 0.1782, "step": 31123 }, { "epoch": 193.3177570093458, "grad_norm": 2.578125, "learning_rate": 1.4068613007648945e-07, "loss": 0.2003, "step": 31124 }, { "epoch": 193.3239875389408, "grad_norm": 1.1796875, "learning_rate": 1.4042524052557128e-07, "loss": 0.1207, "step": 31125 }, { "epoch": 193.33021806853583, "grad_norm": 1.203125, "learning_rate": 1.4016459241581182e-07, "loss": 0.0956, "step": 31126 }, { "epoch": 193.33644859813083, "grad_norm": 1.6875, "learning_rate": 1.3990418574974517e-07, "loss": 0.1122, "step": 31127 }, { "epoch": 193.34267912772586, "grad_norm": 1.0859375, "learning_rate": 1.396440205298971e-07, "loss": 0.1138, "step": 31128 }, { "epoch": 193.34890965732086, "grad_norm": 1.46875, "learning_rate": 1.3938409675879615e-07, "loss": 0.1292, "step": 31129 }, { "epoch": 193.3551401869159, "grad_norm": 1.8125, "learning_rate": 1.3912441443896528e-07, "loss": 0.1811, "step": 31130 }, { "epoch": 193.3613707165109, "grad_norm": 3.171875, "learning_rate": 1.388649735729275e-07, "loss": 0.2496, "step": 31131 }, { "epoch": 193.36760124610592, "grad_norm": 1.890625, "learning_rate": 1.3860577416320574e-07, "loss": 0.1582, "step": 31132 }, { "epoch": 193.37383177570092, "grad_norm": 2.0625, "learning_rate": 1.3834681621231194e-07, "loss": 0.167, "step": 31133 }, { "epoch": 193.38006230529595, "grad_norm": 2.078125, "learning_rate": 1.3808809972276072e-07, "loss": 0.2417, "step": 31134 }, { "epoch": 193.38629283489095, "grad_norm": 2.734375, "learning_rate": 1.3782962469707227e-07, "loss": 0.1857, "step": 31135 }, { "epoch": 193.39252336448598, "grad_norm": 1.1796875, "learning_rate": 1.3757139113775297e-07, "loss": 0.1038, "step": 31136 }, { "epoch": 193.39875389408098, "grad_norm": 1.7109375, "learning_rate": 1.3731339904730634e-07, "loss": 0.1234, "step": 31137 }, { "epoch": 193.404984423676, "grad_norm": 1.7734375, "learning_rate": 1.3705564842824424e-07, "loss": 0.1258, "step": 31138 }, { "epoch": 193.41121495327104, "grad_norm": 1.59375, "learning_rate": 1.367981392830675e-07, "loss": 0.2462, "step": 31139 }, { "epoch": 193.41744548286604, "grad_norm": 2.578125, "learning_rate": 1.3654087161427686e-07, "loss": 0.2167, "step": 31140 }, { "epoch": 193.42367601246107, "grad_norm": 2.703125, "learning_rate": 1.3628384542437033e-07, "loss": 0.3248, "step": 31141 }, { "epoch": 193.42990654205607, "grad_norm": 1.78125, "learning_rate": 1.3602706071584592e-07, "loss": 0.1189, "step": 31142 }, { "epoch": 193.4361370716511, "grad_norm": 2.046875, "learning_rate": 1.3577051749119885e-07, "loss": 0.2389, "step": 31143 }, { "epoch": 193.4423676012461, "grad_norm": 1.9140625, "learning_rate": 1.35514215752916e-07, "loss": 0.1459, "step": 31144 }, { "epoch": 193.44859813084113, "grad_norm": 3.125, "learning_rate": 1.3525815550349262e-07, "loss": 0.2084, "step": 31145 }, { "epoch": 193.45482866043614, "grad_norm": 3.3125, "learning_rate": 1.3500233674541007e-07, "loss": 0.1693, "step": 31146 }, { "epoch": 193.46105919003116, "grad_norm": 2.546875, "learning_rate": 1.3474675948115522e-07, "loss": 0.1898, "step": 31147 }, { "epoch": 193.46728971962617, "grad_norm": 2.6875, "learning_rate": 1.344914237132122e-07, "loss": 0.1657, "step": 31148 }, { "epoch": 193.4735202492212, "grad_norm": 3.484375, "learning_rate": 1.3423632944405685e-07, "loss": 0.339, "step": 31149 }, { "epoch": 193.4797507788162, "grad_norm": 2.171875, "learning_rate": 1.3398147667617046e-07, "loss": 0.1717, "step": 31150 }, { "epoch": 193.48598130841123, "grad_norm": 2.515625, "learning_rate": 1.3372686541202327e-07, "loss": 0.2109, "step": 31151 }, { "epoch": 193.49221183800623, "grad_norm": 2.0, "learning_rate": 1.3347249565409392e-07, "loss": 0.1986, "step": 31152 }, { "epoch": 193.49844236760126, "grad_norm": 1.046875, "learning_rate": 1.3321836740484983e-07, "loss": 0.1104, "step": 31153 }, { "epoch": 193.50467289719626, "grad_norm": 1.546875, "learning_rate": 1.3296448066676125e-07, "loss": 0.1331, "step": 31154 }, { "epoch": 193.5109034267913, "grad_norm": 1.53125, "learning_rate": 1.327108354422929e-07, "loss": 0.1141, "step": 31155 }, { "epoch": 193.5171339563863, "grad_norm": 2.140625, "learning_rate": 1.324574317339039e-07, "loss": 0.1478, "step": 31156 }, { "epoch": 193.52336448598132, "grad_norm": 1.40625, "learning_rate": 1.3220426954406174e-07, "loss": 0.1434, "step": 31157 }, { "epoch": 193.52959501557632, "grad_norm": 2.515625, "learning_rate": 1.3195134887522277e-07, "loss": 0.2365, "step": 31158 }, { "epoch": 193.53582554517135, "grad_norm": 2.828125, "learning_rate": 1.3169866972984336e-07, "loss": 0.2978, "step": 31159 }, { "epoch": 193.54205607476635, "grad_norm": 1.7890625, "learning_rate": 1.314462321103771e-07, "loss": 0.1411, "step": 31160 }, { "epoch": 193.54828660436138, "grad_norm": 1.3359375, "learning_rate": 1.311940360192776e-07, "loss": 0.1095, "step": 31161 }, { "epoch": 193.55451713395638, "grad_norm": 2.25, "learning_rate": 1.309420814589929e-07, "loss": 0.1738, "step": 31162 }, { "epoch": 193.5607476635514, "grad_norm": 2.734375, "learning_rate": 1.3069036843196825e-07, "loss": 0.169, "step": 31163 }, { "epoch": 193.5669781931464, "grad_norm": 1.34375, "learning_rate": 1.3043889694064892e-07, "loss": 0.1367, "step": 31164 }, { "epoch": 193.57320872274144, "grad_norm": 2.34375, "learning_rate": 1.301876669874802e-07, "loss": 0.2406, "step": 31165 }, { "epoch": 193.57943925233644, "grad_norm": 2.265625, "learning_rate": 1.29936678574899e-07, "loss": 0.1431, "step": 31166 }, { "epoch": 193.58566978193147, "grad_norm": 2.453125, "learning_rate": 1.2968593170534504e-07, "loss": 0.2471, "step": 31167 }, { "epoch": 193.59190031152647, "grad_norm": 2.09375, "learning_rate": 1.294354263812525e-07, "loss": 0.1472, "step": 31168 }, { "epoch": 193.5981308411215, "grad_norm": 2.640625, "learning_rate": 1.2918516260505552e-07, "loss": 0.1752, "step": 31169 }, { "epoch": 193.6043613707165, "grad_norm": 1.140625, "learning_rate": 1.2893514037918276e-07, "loss": 0.1053, "step": 31170 }, { "epoch": 193.61059190031153, "grad_norm": 2.421875, "learning_rate": 1.2868535970606554e-07, "loss": 0.169, "step": 31171 }, { "epoch": 193.61682242990653, "grad_norm": 2.03125, "learning_rate": 1.2843582058812699e-07, "loss": 0.2171, "step": 31172 }, { "epoch": 193.62305295950156, "grad_norm": 2.046875, "learning_rate": 1.2818652302779012e-07, "loss": 0.228, "step": 31173 }, { "epoch": 193.62928348909657, "grad_norm": 2.59375, "learning_rate": 1.27937467027478e-07, "loss": 0.1391, "step": 31174 }, { "epoch": 193.6355140186916, "grad_norm": 3.28125, "learning_rate": 1.2768865258960815e-07, "loss": 0.1794, "step": 31175 }, { "epoch": 193.6417445482866, "grad_norm": 1.890625, "learning_rate": 1.274400797166009e-07, "loss": 0.2469, "step": 31176 }, { "epoch": 193.64797507788163, "grad_norm": 1.265625, "learning_rate": 1.2719174841086535e-07, "loss": 0.1348, "step": 31177 }, { "epoch": 193.65420560747663, "grad_norm": 1.609375, "learning_rate": 1.269436586748135e-07, "loss": 0.1647, "step": 31178 }, { "epoch": 193.66043613707166, "grad_norm": 0.99609375, "learning_rate": 1.266958105108601e-07, "loss": 0.0822, "step": 31179 }, { "epoch": 193.66666666666666, "grad_norm": 3.203125, "learning_rate": 1.2644820392140878e-07, "loss": 0.1455, "step": 31180 }, { "epoch": 193.6728971962617, "grad_norm": 1.4453125, "learning_rate": 1.2620083890886315e-07, "loss": 0.1284, "step": 31181 }, { "epoch": 193.6791277258567, "grad_norm": 1.875, "learning_rate": 1.2595371547562686e-07, "loss": 0.2147, "step": 31182 }, { "epoch": 193.68535825545172, "grad_norm": 1.9296875, "learning_rate": 1.2570683362410352e-07, "loss": 0.1569, "step": 31183 }, { "epoch": 193.69158878504672, "grad_norm": 2.203125, "learning_rate": 1.2546019335668568e-07, "loss": 0.1188, "step": 31184 }, { "epoch": 193.69781931464175, "grad_norm": 1.703125, "learning_rate": 1.2521379467576865e-07, "loss": 0.2008, "step": 31185 }, { "epoch": 193.70404984423675, "grad_norm": 3.25, "learning_rate": 1.249676375837505e-07, "loss": 0.1795, "step": 31186 }, { "epoch": 193.71028037383178, "grad_norm": 1.90625, "learning_rate": 1.2472172208301825e-07, "loss": 0.1388, "step": 31187 }, { "epoch": 193.71651090342678, "grad_norm": 1.1171875, "learning_rate": 1.244760481759616e-07, "loss": 0.104, "step": 31188 }, { "epoch": 193.7227414330218, "grad_norm": 1.4140625, "learning_rate": 1.2423061586496477e-07, "loss": 0.1028, "step": 31189 }, { "epoch": 193.7289719626168, "grad_norm": 1.296875, "learning_rate": 1.2398542515241197e-07, "loss": 0.1175, "step": 31190 }, { "epoch": 193.73520249221184, "grad_norm": 1.7265625, "learning_rate": 1.237404760406874e-07, "loss": 0.1395, "step": 31191 }, { "epoch": 193.74143302180684, "grad_norm": 1.78125, "learning_rate": 1.2349576853216972e-07, "loss": 0.12, "step": 31192 }, { "epoch": 193.74766355140187, "grad_norm": 2.375, "learning_rate": 1.23251302629232e-07, "loss": 0.195, "step": 31193 }, { "epoch": 193.75389408099687, "grad_norm": 1.8984375, "learning_rate": 1.230070783342474e-07, "loss": 0.1609, "step": 31194 }, { "epoch": 193.7601246105919, "grad_norm": 2.796875, "learning_rate": 1.2276309564959454e-07, "loss": 0.1189, "step": 31195 }, { "epoch": 193.7663551401869, "grad_norm": 1.859375, "learning_rate": 1.225193545776382e-07, "loss": 0.2693, "step": 31196 }, { "epoch": 193.77258566978193, "grad_norm": 1.515625, "learning_rate": 1.222758551207459e-07, "loss": 0.0825, "step": 31197 }, { "epoch": 193.77881619937693, "grad_norm": 2.3125, "learning_rate": 1.2203259728128525e-07, "loss": 0.1758, "step": 31198 }, { "epoch": 193.78504672897196, "grad_norm": 2.140625, "learning_rate": 1.217895810616182e-07, "loss": 0.2295, "step": 31199 }, { "epoch": 193.79127725856696, "grad_norm": 2.515625, "learning_rate": 1.2154680646410122e-07, "loss": 0.1776, "step": 31200 }, { "epoch": 193.797507788162, "grad_norm": 1.640625, "learning_rate": 1.2130427349109353e-07, "loss": 0.1251, "step": 31201 }, { "epoch": 193.80373831775702, "grad_norm": 2.234375, "learning_rate": 1.2106198214495435e-07, "loss": 0.1792, "step": 31202 }, { "epoch": 193.80996884735202, "grad_norm": 1.5390625, "learning_rate": 1.2081993242803458e-07, "loss": 0.1274, "step": 31203 }, { "epoch": 193.81619937694705, "grad_norm": 2.296875, "learning_rate": 1.2057812434268233e-07, "loss": 0.1542, "step": 31204 }, { "epoch": 193.82242990654206, "grad_norm": 1.328125, "learning_rate": 1.203365578912513e-07, "loss": 0.1135, "step": 31205 }, { "epoch": 193.82866043613708, "grad_norm": 1.6328125, "learning_rate": 1.2009523307608406e-07, "loss": 0.1144, "step": 31206 }, { "epoch": 193.8348909657321, "grad_norm": 2.21875, "learning_rate": 1.1985414989952593e-07, "loss": 0.2078, "step": 31207 }, { "epoch": 193.84112149532712, "grad_norm": 2.265625, "learning_rate": 1.196133083639167e-07, "loss": 0.1377, "step": 31208 }, { "epoch": 193.84735202492212, "grad_norm": 1.453125, "learning_rate": 1.193727084715962e-07, "loss": 0.0971, "step": 31209 }, { "epoch": 193.85358255451715, "grad_norm": 1.1796875, "learning_rate": 1.1913235022490143e-07, "loss": 0.1178, "step": 31210 }, { "epoch": 193.85981308411215, "grad_norm": 2.515625, "learning_rate": 1.1889223362616663e-07, "loss": 0.202, "step": 31211 }, { "epoch": 193.86604361370718, "grad_norm": 1.84375, "learning_rate": 1.1865235867772607e-07, "loss": 0.1457, "step": 31212 }, { "epoch": 193.87227414330218, "grad_norm": 2.640625, "learning_rate": 1.184127253819084e-07, "loss": 0.2639, "step": 31213 }, { "epoch": 193.8785046728972, "grad_norm": 0.87890625, "learning_rate": 1.1817333374103678e-07, "loss": 0.0995, "step": 31214 }, { "epoch": 193.8847352024922, "grad_norm": 1.28125, "learning_rate": 1.1793418375743992e-07, "loss": 0.0967, "step": 31215 }, { "epoch": 193.89096573208724, "grad_norm": 1.390625, "learning_rate": 1.1769527543344095e-07, "loss": 0.1259, "step": 31216 }, { "epoch": 193.89719626168224, "grad_norm": 2.34375, "learning_rate": 1.1745660877136023e-07, "loss": 0.1752, "step": 31217 }, { "epoch": 193.90342679127727, "grad_norm": 1.203125, "learning_rate": 1.1721818377351534e-07, "loss": 0.1485, "step": 31218 }, { "epoch": 193.90965732087227, "grad_norm": 1.421875, "learning_rate": 1.1698000044221834e-07, "loss": 0.1266, "step": 31219 }, { "epoch": 193.9158878504673, "grad_norm": 3.046875, "learning_rate": 1.1674205877978683e-07, "loss": 0.1784, "step": 31220 }, { "epoch": 193.9221183800623, "grad_norm": 2.390625, "learning_rate": 1.1650435878853283e-07, "loss": 0.1957, "step": 31221 }, { "epoch": 193.92834890965733, "grad_norm": 2.515625, "learning_rate": 1.1626690047076005e-07, "loss": 0.2284, "step": 31222 }, { "epoch": 193.93457943925233, "grad_norm": 1.6953125, "learning_rate": 1.1602968382877777e-07, "loss": 0.1207, "step": 31223 }, { "epoch": 193.94080996884736, "grad_norm": 1.859375, "learning_rate": 1.1579270886488968e-07, "loss": 0.235, "step": 31224 }, { "epoch": 193.94704049844236, "grad_norm": 0.87890625, "learning_rate": 1.1555597558139952e-07, "loss": 0.0941, "step": 31225 }, { "epoch": 193.9532710280374, "grad_norm": 2.890625, "learning_rate": 1.1531948398059989e-07, "loss": 0.0995, "step": 31226 }, { "epoch": 193.9595015576324, "grad_norm": 1.2421875, "learning_rate": 1.1508323406479449e-07, "loss": 0.1185, "step": 31227 }, { "epoch": 193.96573208722742, "grad_norm": 1.3046875, "learning_rate": 1.1484722583627316e-07, "loss": 0.104, "step": 31228 }, { "epoch": 193.97196261682242, "grad_norm": 1.8828125, "learning_rate": 1.146114592973313e-07, "loss": 0.118, "step": 31229 }, { "epoch": 193.97819314641745, "grad_norm": 2.0, "learning_rate": 1.1437593445025596e-07, "loss": 0.1548, "step": 31230 }, { "epoch": 193.98442367601245, "grad_norm": 1.9375, "learning_rate": 1.1414065129733697e-07, "loss": 0.1276, "step": 31231 }, { "epoch": 193.99065420560748, "grad_norm": 2.71875, "learning_rate": 1.1390560984086141e-07, "loss": 0.2533, "step": 31232 }, { "epoch": 193.99688473520249, "grad_norm": 1.03125, "learning_rate": 1.136708100831052e-07, "loss": 0.1088, "step": 31233 }, { "epoch": 194.0, "grad_norm": 1.9609375, "learning_rate": 1.1343625202635266e-07, "loss": 0.1058, "step": 31234 }, { "epoch": 194.00623052959503, "grad_norm": 2.28125, "learning_rate": 1.1320193567288529e-07, "loss": 0.1602, "step": 31235 }, { "epoch": 194.01246105919003, "grad_norm": 2.390625, "learning_rate": 1.1296786102497348e-07, "loss": 0.1473, "step": 31236 }, { "epoch": 194.01869158878506, "grad_norm": 2.015625, "learning_rate": 1.127340280848932e-07, "loss": 0.1392, "step": 31237 }, { "epoch": 194.02492211838006, "grad_norm": 2.578125, "learning_rate": 1.1250043685491484e-07, "loss": 0.2461, "step": 31238 }, { "epoch": 194.0311526479751, "grad_norm": 1.5859375, "learning_rate": 1.1226708733730606e-07, "loss": 0.1288, "step": 31239 }, { "epoch": 194.0373831775701, "grad_norm": 3.0, "learning_rate": 1.1203397953433448e-07, "loss": 0.1939, "step": 31240 }, { "epoch": 194.04361370716512, "grad_norm": 1.671875, "learning_rate": 1.1180111344826493e-07, "loss": 0.1427, "step": 31241 }, { "epoch": 194.04984423676012, "grad_norm": 2.953125, "learning_rate": 1.1156848908135676e-07, "loss": 0.2313, "step": 31242 }, { "epoch": 194.05607476635515, "grad_norm": 2.15625, "learning_rate": 1.11336106435872e-07, "loss": 0.1975, "step": 31243 }, { "epoch": 194.06230529595015, "grad_norm": 1.0703125, "learning_rate": 1.1110396551406721e-07, "loss": 0.0991, "step": 31244 }, { "epoch": 194.06853582554518, "grad_norm": 2.75, "learning_rate": 1.1087206631819336e-07, "loss": 0.1784, "step": 31245 }, { "epoch": 194.07476635514018, "grad_norm": 2.09375, "learning_rate": 1.1064040885050697e-07, "loss": 0.1375, "step": 31246 }, { "epoch": 194.0809968847352, "grad_norm": 1.53125, "learning_rate": 1.1040899311325347e-07, "loss": 0.1075, "step": 31247 }, { "epoch": 194.08722741433021, "grad_norm": 1.3046875, "learning_rate": 1.1017781910868386e-07, "loss": 0.1138, "step": 31248 }, { "epoch": 194.09345794392524, "grad_norm": 2.15625, "learning_rate": 1.0994688683904353e-07, "loss": 0.1716, "step": 31249 }, { "epoch": 194.09968847352025, "grad_norm": 1.3984375, "learning_rate": 1.0971619630657792e-07, "loss": 0.1206, "step": 31250 }, { "epoch": 194.10591900311528, "grad_norm": 2.1875, "learning_rate": 1.0948574751351858e-07, "loss": 0.1615, "step": 31251 }, { "epoch": 194.11214953271028, "grad_norm": 1.3203125, "learning_rate": 1.0925554046211096e-07, "loss": 0.1295, "step": 31252 }, { "epoch": 194.1183800623053, "grad_norm": 1.875, "learning_rate": 1.0902557515458933e-07, "loss": 0.1547, "step": 31253 }, { "epoch": 194.1246105919003, "grad_norm": 2.203125, "learning_rate": 1.0879585159318806e-07, "loss": 0.1758, "step": 31254 }, { "epoch": 194.13084112149534, "grad_norm": 2.015625, "learning_rate": 1.085663697801359e-07, "loss": 0.1579, "step": 31255 }, { "epoch": 194.13707165109034, "grad_norm": 2.578125, "learning_rate": 1.083371297176644e-07, "loss": 0.1952, "step": 31256 }, { "epoch": 194.14330218068537, "grad_norm": 2.078125, "learning_rate": 1.0810813140799681e-07, "loss": 0.2054, "step": 31257 }, { "epoch": 194.14953271028037, "grad_norm": 2.265625, "learning_rate": 1.0787937485335908e-07, "loss": 0.1969, "step": 31258 }, { "epoch": 194.1557632398754, "grad_norm": 1.359375, "learning_rate": 1.076508600559717e-07, "loss": 0.1527, "step": 31259 }, { "epoch": 194.1619937694704, "grad_norm": 2.125, "learning_rate": 1.0742258701805785e-07, "loss": 0.1791, "step": 31260 }, { "epoch": 194.16822429906543, "grad_norm": 2.625, "learning_rate": 1.0719455574182968e-07, "loss": 0.2186, "step": 31261 }, { "epoch": 194.17445482866043, "grad_norm": 2.03125, "learning_rate": 1.0696676622950485e-07, "loss": 0.1817, "step": 31262 }, { "epoch": 194.18068535825546, "grad_norm": 1.75, "learning_rate": 1.0673921848329271e-07, "loss": 0.1618, "step": 31263 }, { "epoch": 194.18691588785046, "grad_norm": 1.578125, "learning_rate": 1.0651191250540815e-07, "loss": 0.137, "step": 31264 }, { "epoch": 194.1931464174455, "grad_norm": 1.21875, "learning_rate": 1.0628484829805497e-07, "loss": 0.1091, "step": 31265 }, { "epoch": 194.1993769470405, "grad_norm": 2.5, "learning_rate": 1.0605802586343972e-07, "loss": 0.2179, "step": 31266 }, { "epoch": 194.20560747663552, "grad_norm": 2.21875, "learning_rate": 1.0583144520376343e-07, "loss": 0.1837, "step": 31267 }, { "epoch": 194.21183800623052, "grad_norm": 1.7578125, "learning_rate": 1.0560510632122989e-07, "loss": 0.1302, "step": 31268 }, { "epoch": 194.21806853582555, "grad_norm": 3.34375, "learning_rate": 1.0537900921803734e-07, "loss": 0.2892, "step": 31269 }, { "epoch": 194.22429906542055, "grad_norm": 2.59375, "learning_rate": 1.0515315389637848e-07, "loss": 0.2053, "step": 31270 }, { "epoch": 194.23052959501558, "grad_norm": 1.4453125, "learning_rate": 1.0492754035844598e-07, "loss": 0.1398, "step": 31271 }, { "epoch": 194.23676012461058, "grad_norm": 1.1328125, "learning_rate": 1.0470216860643812e-07, "loss": 0.0813, "step": 31272 }, { "epoch": 194.2429906542056, "grad_norm": 2.46875, "learning_rate": 1.0447703864253644e-07, "loss": 0.237, "step": 31273 }, { "epoch": 194.2492211838006, "grad_norm": 2.265625, "learning_rate": 1.0425215046893089e-07, "loss": 0.1211, "step": 31274 }, { "epoch": 194.25545171339564, "grad_norm": 3.015625, "learning_rate": 1.0402750408780582e-07, "loss": 0.1456, "step": 31275 }, { "epoch": 194.26168224299064, "grad_norm": 1.703125, "learning_rate": 1.0380309950134282e-07, "loss": 0.1426, "step": 31276 }, { "epoch": 194.26791277258567, "grad_norm": 2.359375, "learning_rate": 1.0357893671171792e-07, "loss": 0.2049, "step": 31277 }, { "epoch": 194.27414330218068, "grad_norm": 1.875, "learning_rate": 1.0335501572111273e-07, "loss": 0.2162, "step": 31278 }, { "epoch": 194.2803738317757, "grad_norm": 2.421875, "learning_rate": 1.0313133653170049e-07, "loss": 0.1749, "step": 31279 }, { "epoch": 194.2866043613707, "grad_norm": 1.8203125, "learning_rate": 1.0290789914565446e-07, "loss": 0.1456, "step": 31280 }, { "epoch": 194.29283489096574, "grad_norm": 2.921875, "learning_rate": 1.0268470356514237e-07, "loss": 0.2069, "step": 31281 }, { "epoch": 194.29906542056074, "grad_norm": 1.28125, "learning_rate": 1.024617497923347e-07, "loss": 0.1057, "step": 31282 }, { "epoch": 194.30529595015577, "grad_norm": 3.0, "learning_rate": 1.0223903782939359e-07, "loss": 0.1488, "step": 31283 }, { "epoch": 194.31152647975077, "grad_norm": 1.8359375, "learning_rate": 1.0201656767848677e-07, "loss": 0.1128, "step": 31284 }, { "epoch": 194.3177570093458, "grad_norm": 1.4921875, "learning_rate": 1.0179433934176807e-07, "loss": 0.1154, "step": 31285 }, { "epoch": 194.3239875389408, "grad_norm": 2.703125, "learning_rate": 1.0157235282140243e-07, "loss": 0.203, "step": 31286 }, { "epoch": 194.33021806853583, "grad_norm": 2.546875, "learning_rate": 1.0135060811954367e-07, "loss": 0.2402, "step": 31287 }, { "epoch": 194.33644859813083, "grad_norm": 1.8828125, "learning_rate": 1.0112910523834284e-07, "loss": 0.1187, "step": 31288 }, { "epoch": 194.34267912772586, "grad_norm": 1.4921875, "learning_rate": 1.009078441799538e-07, "loss": 0.0964, "step": 31289 }, { "epoch": 194.34890965732086, "grad_norm": 1.5546875, "learning_rate": 1.0068682494652482e-07, "loss": 0.1425, "step": 31290 }, { "epoch": 194.3551401869159, "grad_norm": 2.3125, "learning_rate": 1.0046604754020416e-07, "loss": 0.2356, "step": 31291 }, { "epoch": 194.3613707165109, "grad_norm": 2.828125, "learning_rate": 1.0024551196313458e-07, "loss": 0.2484, "step": 31292 }, { "epoch": 194.36760124610592, "grad_norm": 0.98828125, "learning_rate": 1.0002521821745602e-07, "loss": 0.0809, "step": 31293 }, { "epoch": 194.37383177570092, "grad_norm": 1.171875, "learning_rate": 9.980516630531123e-08, "loss": 0.1042, "step": 31294 }, { "epoch": 194.38006230529595, "grad_norm": 1.484375, "learning_rate": 9.958535622883735e-08, "loss": 0.1395, "step": 31295 }, { "epoch": 194.38629283489095, "grad_norm": 1.7578125, "learning_rate": 9.936578799016604e-08, "loss": 0.1174, "step": 31296 }, { "epoch": 194.39252336448598, "grad_norm": 1.0, "learning_rate": 9.91464615914317e-08, "loss": 0.1097, "step": 31297 }, { "epoch": 194.39875389408098, "grad_norm": 1.9453125, "learning_rate": 9.892737703476595e-08, "loss": 0.1907, "step": 31298 }, { "epoch": 194.404984423676, "grad_norm": 2.046875, "learning_rate": 9.870853432229488e-08, "loss": 0.1183, "step": 31299 }, { "epoch": 194.41121495327104, "grad_norm": 2.46875, "learning_rate": 9.848993345614177e-08, "loss": 0.1457, "step": 31300 }, { "epoch": 194.41744548286604, "grad_norm": 1.03125, "learning_rate": 9.827157443843549e-08, "loss": 0.0858, "step": 31301 }, { "epoch": 194.42367601246107, "grad_norm": 2.796875, "learning_rate": 9.805345727129101e-08, "loss": 0.2138, "step": 31302 }, { "epoch": 194.42990654205607, "grad_norm": 2.234375, "learning_rate": 9.783558195683162e-08, "loss": 0.1583, "step": 31303 }, { "epoch": 194.4361370716511, "grad_norm": 1.890625, "learning_rate": 9.761794849716676e-08, "loss": 0.1609, "step": 31304 }, { "epoch": 194.4423676012461, "grad_norm": 1.3046875, "learning_rate": 9.740055689441973e-08, "loss": 0.1393, "step": 31305 }, { "epoch": 194.44859813084113, "grad_norm": 2.21875, "learning_rate": 9.718340715069163e-08, "loss": 0.1155, "step": 31306 }, { "epoch": 194.45482866043614, "grad_norm": 2.1875, "learning_rate": 9.69664992681002e-08, "loss": 0.1347, "step": 31307 }, { "epoch": 194.46105919003116, "grad_norm": 2.09375, "learning_rate": 9.674983324874376e-08, "loss": 0.1237, "step": 31308 }, { "epoch": 194.46728971962617, "grad_norm": 1.7734375, "learning_rate": 9.653340909473452e-08, "loss": 0.1462, "step": 31309 }, { "epoch": 194.4735202492212, "grad_norm": 1.46875, "learning_rate": 9.631722680817079e-08, "loss": 0.1333, "step": 31310 }, { "epoch": 194.4797507788162, "grad_norm": 2.28125, "learning_rate": 9.610128639115091e-08, "loss": 0.2605, "step": 31311 }, { "epoch": 194.48598130841123, "grad_norm": 0.9765625, "learning_rate": 9.588558784577317e-08, "loss": 0.1082, "step": 31312 }, { "epoch": 194.49221183800623, "grad_norm": 2.1875, "learning_rate": 9.567013117413315e-08, "loss": 0.2021, "step": 31313 }, { "epoch": 194.49844236760126, "grad_norm": 2.375, "learning_rate": 9.54549163783236e-08, "loss": 0.2352, "step": 31314 }, { "epoch": 194.50467289719626, "grad_norm": 2.078125, "learning_rate": 9.523994346043453e-08, "loss": 0.2183, "step": 31315 }, { "epoch": 194.5109034267913, "grad_norm": 1.5546875, "learning_rate": 9.502521242255314e-08, "loss": 0.1391, "step": 31316 }, { "epoch": 194.5171339563863, "grad_norm": 1.4296875, "learning_rate": 9.481072326676388e-08, "loss": 0.1276, "step": 31317 }, { "epoch": 194.52336448598132, "grad_norm": 1.4765625, "learning_rate": 9.459647599515398e-08, "loss": 0.11, "step": 31318 }, { "epoch": 194.52959501557632, "grad_norm": 1.5, "learning_rate": 9.438247060979955e-08, "loss": 0.1404, "step": 31319 }, { "epoch": 194.53582554517135, "grad_norm": 3.09375, "learning_rate": 9.416870711278225e-08, "loss": 0.2611, "step": 31320 }, { "epoch": 194.54205607476635, "grad_norm": 1.1171875, "learning_rate": 9.395518550617821e-08, "loss": 0.1289, "step": 31321 }, { "epoch": 194.54828660436138, "grad_norm": 2.125, "learning_rate": 9.374190579205799e-08, "loss": 0.1823, "step": 31322 }, { "epoch": 194.55451713395638, "grad_norm": 2.640625, "learning_rate": 9.352886797249494e-08, "loss": 0.1746, "step": 31323 }, { "epoch": 194.5607476635514, "grad_norm": 1.8828125, "learning_rate": 9.331607204955684e-08, "loss": 0.1443, "step": 31324 }, { "epoch": 194.5669781931464, "grad_norm": 2.171875, "learning_rate": 9.310351802531425e-08, "loss": 0.1406, "step": 31325 }, { "epoch": 194.57320872274144, "grad_norm": 2.03125, "learning_rate": 9.289120590182942e-08, "loss": 0.2225, "step": 31326 }, { "epoch": 194.57943925233644, "grad_norm": 2.125, "learning_rate": 9.267913568116183e-08, "loss": 0.1753, "step": 31327 }, { "epoch": 194.58566978193147, "grad_norm": 2.828125, "learning_rate": 9.246730736537646e-08, "loss": 0.2536, "step": 31328 }, { "epoch": 194.59190031152647, "grad_norm": 2.921875, "learning_rate": 9.225572095652446e-08, "loss": 0.2531, "step": 31329 }, { "epoch": 194.5981308411215, "grad_norm": 2.203125, "learning_rate": 9.204437645666531e-08, "loss": 0.1636, "step": 31330 }, { "epoch": 194.6043613707165, "grad_norm": 2.265625, "learning_rate": 9.183327386785013e-08, "loss": 0.1579, "step": 31331 }, { "epoch": 194.61059190031153, "grad_norm": 2.1875, "learning_rate": 9.162241319213005e-08, "loss": 0.1231, "step": 31332 }, { "epoch": 194.61682242990653, "grad_norm": 2.734375, "learning_rate": 9.141179443155068e-08, "loss": 0.1556, "step": 31333 }, { "epoch": 194.62305295950156, "grad_norm": 1.4609375, "learning_rate": 9.120141758816037e-08, "loss": 0.1054, "step": 31334 }, { "epoch": 194.62928348909657, "grad_norm": 2.140625, "learning_rate": 9.099128266399915e-08, "loss": 0.1269, "step": 31335 }, { "epoch": 194.6355140186916, "grad_norm": 0.80078125, "learning_rate": 9.078138966111261e-08, "loss": 0.0981, "step": 31336 }, { "epoch": 194.6417445482866, "grad_norm": 2.984375, "learning_rate": 9.057173858153246e-08, "loss": 0.2552, "step": 31337 }, { "epoch": 194.64797507788163, "grad_norm": 2.328125, "learning_rate": 9.03623294273015e-08, "loss": 0.1708, "step": 31338 }, { "epoch": 194.65420560747663, "grad_norm": 1.671875, "learning_rate": 9.015316220045145e-08, "loss": 0.1143, "step": 31339 }, { "epoch": 194.66043613707166, "grad_norm": 1.9765625, "learning_rate": 8.994423690301401e-08, "loss": 0.0983, "step": 31340 }, { "epoch": 194.66666666666666, "grad_norm": 1.0859375, "learning_rate": 8.973555353701535e-08, "loss": 0.0965, "step": 31341 }, { "epoch": 194.6728971962617, "grad_norm": 1.203125, "learning_rate": 8.952711210448439e-08, "loss": 0.1077, "step": 31342 }, { "epoch": 194.6791277258567, "grad_norm": 1.3671875, "learning_rate": 8.931891260744729e-08, "loss": 0.1038, "step": 31343 }, { "epoch": 194.68535825545172, "grad_norm": 2.109375, "learning_rate": 8.911095504792189e-08, "loss": 0.1044, "step": 31344 }, { "epoch": 194.69158878504672, "grad_norm": 1.1953125, "learning_rate": 8.890323942793432e-08, "loss": 0.0929, "step": 31345 }, { "epoch": 194.69781931464175, "grad_norm": 1.953125, "learning_rate": 8.86957657494969e-08, "loss": 0.125, "step": 31346 }, { "epoch": 194.70404984423675, "grad_norm": 1.859375, "learning_rate": 8.848853401462464e-08, "loss": 0.176, "step": 31347 }, { "epoch": 194.71028037383178, "grad_norm": 1.03125, "learning_rate": 8.828154422532986e-08, "loss": 0.1023, "step": 31348 }, { "epoch": 194.71651090342678, "grad_norm": 1.6328125, "learning_rate": 8.807479638362759e-08, "loss": 0.1284, "step": 31349 }, { "epoch": 194.7227414330218, "grad_norm": 2.9375, "learning_rate": 8.78682904915218e-08, "loss": 0.2122, "step": 31350 }, { "epoch": 194.7289719626168, "grad_norm": 1.5234375, "learning_rate": 8.766202655101641e-08, "loss": 0.1605, "step": 31351 }, { "epoch": 194.73520249221184, "grad_norm": 1.609375, "learning_rate": 8.745600456412095e-08, "loss": 0.1191, "step": 31352 }, { "epoch": 194.74143302180684, "grad_norm": 1.6953125, "learning_rate": 8.725022453283105e-08, "loss": 0.1059, "step": 31353 }, { "epoch": 194.74766355140187, "grad_norm": 1.9765625, "learning_rate": 8.704468645914788e-08, "loss": 0.1238, "step": 31354 }, { "epoch": 194.75389408099687, "grad_norm": 3.234375, "learning_rate": 8.683939034506428e-08, "loss": 0.2946, "step": 31355 }, { "epoch": 194.7601246105919, "grad_norm": 1.5859375, "learning_rate": 8.663433619257865e-08, "loss": 0.1219, "step": 31356 }, { "epoch": 194.7663551401869, "grad_norm": 1.71875, "learning_rate": 8.64295240036811e-08, "loss": 0.1373, "step": 31357 }, { "epoch": 194.77258566978193, "grad_norm": 1.2734375, "learning_rate": 8.622495378035889e-08, "loss": 0.1018, "step": 31358 }, { "epoch": 194.77881619937693, "grad_norm": 2.3125, "learning_rate": 8.602062552459932e-08, "loss": 0.1249, "step": 31359 }, { "epoch": 194.78504672897196, "grad_norm": 1.796875, "learning_rate": 8.581653923838972e-08, "loss": 0.1228, "step": 31360 }, { "epoch": 194.79127725856696, "grad_norm": 1.2265625, "learning_rate": 8.561269492370905e-08, "loss": 0.1225, "step": 31361 }, { "epoch": 194.797507788162, "grad_norm": 1.9921875, "learning_rate": 8.540909258253904e-08, "loss": 0.178, "step": 31362 }, { "epoch": 194.80373831775702, "grad_norm": 2.046875, "learning_rate": 8.520573221685312e-08, "loss": 0.1916, "step": 31363 }, { "epoch": 194.80996884735202, "grad_norm": 1.75, "learning_rate": 8.500261382863306e-08, "loss": 0.1145, "step": 31364 }, { "epoch": 194.81619937694705, "grad_norm": 2.5625, "learning_rate": 8.479973741984671e-08, "loss": 0.2029, "step": 31365 }, { "epoch": 194.82242990654206, "grad_norm": 1.3125, "learning_rate": 8.459710299246748e-08, "loss": 0.1205, "step": 31366 }, { "epoch": 194.82866043613708, "grad_norm": 2.765625, "learning_rate": 8.439471054845771e-08, "loss": 0.2181, "step": 31367 }, { "epoch": 194.8348909657321, "grad_norm": 1.7890625, "learning_rate": 8.419256008979082e-08, "loss": 0.1686, "step": 31368 }, { "epoch": 194.84112149532712, "grad_norm": 1.234375, "learning_rate": 8.399065161842356e-08, "loss": 0.1448, "step": 31369 }, { "epoch": 194.84735202492212, "grad_norm": 2.53125, "learning_rate": 8.378898513632105e-08, "loss": 0.2389, "step": 31370 }, { "epoch": 194.85358255451715, "grad_norm": 2.4375, "learning_rate": 8.358756064544004e-08, "loss": 0.2009, "step": 31371 }, { "epoch": 194.85981308411215, "grad_norm": 1.234375, "learning_rate": 8.338637814774008e-08, "loss": 0.0976, "step": 31372 }, { "epoch": 194.86604361370718, "grad_norm": 1.3671875, "learning_rate": 8.318543764516961e-08, "loss": 0.1151, "step": 31373 }, { "epoch": 194.87227414330218, "grad_norm": 1.640625, "learning_rate": 8.298473913968263e-08, "loss": 0.1628, "step": 31374 }, { "epoch": 194.8785046728972, "grad_norm": 1.859375, "learning_rate": 8.278428263322758e-08, "loss": 0.1835, "step": 31375 }, { "epoch": 194.8847352024922, "grad_norm": 2.6875, "learning_rate": 8.258406812775565e-08, "loss": 0.2184, "step": 31376 }, { "epoch": 194.89096573208724, "grad_norm": 1.7109375, "learning_rate": 8.238409562520699e-08, "loss": 0.1416, "step": 31377 }, { "epoch": 194.89719626168224, "grad_norm": 3.03125, "learning_rate": 8.218436512752448e-08, "loss": 0.2829, "step": 31378 }, { "epoch": 194.90342679127727, "grad_norm": 2.34375, "learning_rate": 8.198487663664822e-08, "loss": 0.2271, "step": 31379 }, { "epoch": 194.90965732087227, "grad_norm": 1.6171875, "learning_rate": 8.178563015451834e-08, "loss": 0.0915, "step": 31380 }, { "epoch": 194.9158878504673, "grad_norm": 1.8515625, "learning_rate": 8.158662568306385e-08, "loss": 0.1459, "step": 31381 }, { "epoch": 194.9221183800623, "grad_norm": 1.765625, "learning_rate": 8.13878632242221e-08, "loss": 0.1269, "step": 31382 }, { "epoch": 194.92834890965733, "grad_norm": 2.078125, "learning_rate": 8.118934277992485e-08, "loss": 0.1065, "step": 31383 }, { "epoch": 194.93457943925233, "grad_norm": 1.9140625, "learning_rate": 8.099106435209558e-08, "loss": 0.1598, "step": 31384 }, { "epoch": 194.94080996884736, "grad_norm": 1.5703125, "learning_rate": 8.07930279426633e-08, "loss": 0.14, "step": 31385 }, { "epoch": 194.94704049844236, "grad_norm": 2.21875, "learning_rate": 8.05952335535487e-08, "loss": 0.2385, "step": 31386 }, { "epoch": 194.9532710280374, "grad_norm": 2.859375, "learning_rate": 8.039768118667801e-08, "loss": 0.3078, "step": 31387 }, { "epoch": 194.9595015576324, "grad_norm": 1.734375, "learning_rate": 8.020037084396359e-08, "loss": 0.1429, "step": 31388 }, { "epoch": 194.96573208722742, "grad_norm": 1.734375, "learning_rate": 8.000330252732335e-08, "loss": 0.1757, "step": 31389 }, { "epoch": 194.97196261682242, "grad_norm": 2.171875, "learning_rate": 7.980647623867521e-08, "loss": 0.1795, "step": 31390 }, { "epoch": 194.97819314641745, "grad_norm": 2.765625, "learning_rate": 7.960989197992597e-08, "loss": 0.1381, "step": 31391 }, { "epoch": 194.98442367601245, "grad_norm": 3.8125, "learning_rate": 7.941354975298798e-08, "loss": 0.2538, "step": 31392 }, { "epoch": 194.99065420560748, "grad_norm": 1.6640625, "learning_rate": 7.921744955976806e-08, "loss": 0.1183, "step": 31393 }, { "epoch": 194.99688473520249, "grad_norm": 1.8515625, "learning_rate": 7.902159140216747e-08, "loss": 0.1526, "step": 31394 }, { "epoch": 195.0, "grad_norm": 1.0, "learning_rate": 7.882597528209302e-08, "loss": 0.0706, "step": 31395 }, { "epoch": 195.00623052959503, "grad_norm": 1.7265625, "learning_rate": 7.863060120144317e-08, "loss": 0.1393, "step": 31396 }, { "epoch": 195.01246105919003, "grad_norm": 1.7734375, "learning_rate": 7.843546916211086e-08, "loss": 0.1578, "step": 31397 }, { "epoch": 195.01869158878506, "grad_norm": 1.2265625, "learning_rate": 7.824057916599736e-08, "loss": 0.1025, "step": 31398 }, { "epoch": 195.02492211838006, "grad_norm": 2.1875, "learning_rate": 7.804593121499281e-08, "loss": 0.1302, "step": 31399 }, { "epoch": 195.0311526479751, "grad_norm": 1.7421875, "learning_rate": 7.785152531098738e-08, "loss": 0.1424, "step": 31400 }, { "epoch": 195.0373831775701, "grad_norm": 1.5703125, "learning_rate": 7.765736145587122e-08, "loss": 0.1375, "step": 31401 }, { "epoch": 195.04361370716512, "grad_norm": 1.8671875, "learning_rate": 7.746343965152892e-08, "loss": 0.142, "step": 31402 }, { "epoch": 195.04984423676012, "grad_norm": 3.046875, "learning_rate": 7.72697598998423e-08, "loss": 0.1917, "step": 31403 }, { "epoch": 195.05607476635515, "grad_norm": 2.859375, "learning_rate": 7.707632220269324e-08, "loss": 0.3162, "step": 31404 }, { "epoch": 195.06230529595015, "grad_norm": 1.6875, "learning_rate": 7.688312656196072e-08, "loss": 0.1422, "step": 31405 }, { "epoch": 195.06853582554518, "grad_norm": 1.4921875, "learning_rate": 7.669017297952385e-08, "loss": 0.1292, "step": 31406 }, { "epoch": 195.07476635514018, "grad_norm": 1.5390625, "learning_rate": 7.649746145725056e-08, "loss": 0.1023, "step": 31407 }, { "epoch": 195.0809968847352, "grad_norm": 1.4296875, "learning_rate": 7.630499199701713e-08, "loss": 0.1358, "step": 31408 }, { "epoch": 195.08722741433021, "grad_norm": 0.9609375, "learning_rate": 7.611276460069428e-08, "loss": 0.1049, "step": 31409 }, { "epoch": 195.09345794392524, "grad_norm": 2.578125, "learning_rate": 7.59207792701444e-08, "loss": 0.1925, "step": 31410 }, { "epoch": 195.09968847352025, "grad_norm": 0.9375, "learning_rate": 7.572903600723546e-08, "loss": 0.113, "step": 31411 }, { "epoch": 195.10591900311528, "grad_norm": 1.0625, "learning_rate": 7.553753481382709e-08, "loss": 0.1223, "step": 31412 }, { "epoch": 195.11214953271028, "grad_norm": 2.265625, "learning_rate": 7.534627569177887e-08, "loss": 0.1929, "step": 31413 }, { "epoch": 195.1183800623053, "grad_norm": 1.8359375, "learning_rate": 7.515525864295325e-08, "loss": 0.1612, "step": 31414 }, { "epoch": 195.1246105919003, "grad_norm": 2.265625, "learning_rate": 7.49644836692015e-08, "loss": 0.2798, "step": 31415 }, { "epoch": 195.13084112149534, "grad_norm": 1.59375, "learning_rate": 7.477395077237493e-08, "loss": 0.12, "step": 31416 }, { "epoch": 195.13707165109034, "grad_norm": 2.15625, "learning_rate": 7.458365995432759e-08, "loss": 0.1354, "step": 31417 }, { "epoch": 195.14330218068537, "grad_norm": 1.4609375, "learning_rate": 7.439361121690525e-08, "loss": 0.148, "step": 31418 }, { "epoch": 195.14953271028037, "grad_norm": 1.6171875, "learning_rate": 7.420380456195641e-08, "loss": 0.1245, "step": 31419 }, { "epoch": 195.1557632398754, "grad_norm": 2.453125, "learning_rate": 7.401423999132407e-08, "loss": 0.229, "step": 31420 }, { "epoch": 195.1619937694704, "grad_norm": 2.046875, "learning_rate": 7.382491750684562e-08, "loss": 0.142, "step": 31421 }, { "epoch": 195.16822429906543, "grad_norm": 1.421875, "learning_rate": 7.363583711036126e-08, "loss": 0.1029, "step": 31422 }, { "epoch": 195.17445482866043, "grad_norm": 1.9140625, "learning_rate": 7.34469988037112e-08, "loss": 0.1693, "step": 31423 }, { "epoch": 195.18068535825546, "grad_norm": 0.9765625, "learning_rate": 7.325840258872452e-08, "loss": 0.0964, "step": 31424 }, { "epoch": 195.18691588785046, "grad_norm": 2.484375, "learning_rate": 7.307004846723586e-08, "loss": 0.2719, "step": 31425 }, { "epoch": 195.1931464174455, "grad_norm": 1.4765625, "learning_rate": 7.288193644107155e-08, "loss": 0.1476, "step": 31426 }, { "epoch": 195.1993769470405, "grad_norm": 2.328125, "learning_rate": 7.269406651206345e-08, "loss": 0.1664, "step": 31427 }, { "epoch": 195.20560747663552, "grad_norm": 1.296875, "learning_rate": 7.250643868203233e-08, "loss": 0.1369, "step": 31428 }, { "epoch": 195.21183800623052, "grad_norm": 1.8046875, "learning_rate": 7.231905295280173e-08, "loss": 0.2171, "step": 31429 }, { "epoch": 195.21806853582555, "grad_norm": 2.765625, "learning_rate": 7.21319093261924e-08, "loss": 0.1814, "step": 31430 }, { "epoch": 195.22429906542055, "grad_norm": 1.5625, "learning_rate": 7.194500780401958e-08, "loss": 0.143, "step": 31431 }, { "epoch": 195.23052959501558, "grad_norm": 4.53125, "learning_rate": 7.175834838810125e-08, "loss": 0.1282, "step": 31432 }, { "epoch": 195.23676012461058, "grad_norm": 2.625, "learning_rate": 7.157193108024707e-08, "loss": 0.1208, "step": 31433 }, { "epoch": 195.2429906542056, "grad_norm": 1.9921875, "learning_rate": 7.138575588226948e-08, "loss": 0.1475, "step": 31434 }, { "epoch": 195.2492211838006, "grad_norm": 3.515625, "learning_rate": 7.119982279597815e-08, "loss": 0.2531, "step": 31435 }, { "epoch": 195.25545171339564, "grad_norm": 1.9453125, "learning_rate": 7.101413182317718e-08, "loss": 0.1021, "step": 31436 }, { "epoch": 195.26168224299064, "grad_norm": 2.03125, "learning_rate": 7.082868296566791e-08, "loss": 0.182, "step": 31437 }, { "epoch": 195.26791277258567, "grad_norm": 1.2890625, "learning_rate": 7.064347622525724e-08, "loss": 0.1354, "step": 31438 }, { "epoch": 195.27414330218068, "grad_norm": 1.40625, "learning_rate": 7.045851160373817e-08, "loss": 0.0964, "step": 31439 }, { "epoch": 195.2803738317757, "grad_norm": 4.28125, "learning_rate": 7.027378910290928e-08, "loss": 0.3964, "step": 31440 }, { "epoch": 195.2866043613707, "grad_norm": 1.21875, "learning_rate": 7.008930872456354e-08, "loss": 0.1172, "step": 31441 }, { "epoch": 195.29283489096574, "grad_norm": 1.328125, "learning_rate": 6.990507047049676e-08, "loss": 0.0962, "step": 31442 }, { "epoch": 195.29906542056074, "grad_norm": 1.640625, "learning_rate": 6.972107434249364e-08, "loss": 0.1409, "step": 31443 }, { "epoch": 195.30529595015577, "grad_norm": 1.4296875, "learning_rate": 6.95373203423444e-08, "loss": 0.1254, "step": 31444 }, { "epoch": 195.31152647975077, "grad_norm": 1.5546875, "learning_rate": 6.935380847182815e-08, "loss": 0.1325, "step": 31445 }, { "epoch": 195.3177570093458, "grad_norm": 1.3671875, "learning_rate": 6.917053873273515e-08, "loss": 0.1234, "step": 31446 }, { "epoch": 195.3239875389408, "grad_norm": 1.8046875, "learning_rate": 6.898751112684176e-08, "loss": 0.1619, "step": 31447 }, { "epoch": 195.33021806853583, "grad_norm": 3.015625, "learning_rate": 6.880472565592432e-08, "loss": 0.2058, "step": 31448 }, { "epoch": 195.33644859813083, "grad_norm": 2.859375, "learning_rate": 6.862218232175644e-08, "loss": 0.2589, "step": 31449 }, { "epoch": 195.34267912772586, "grad_norm": 1.3125, "learning_rate": 6.843988112611721e-08, "loss": 0.127, "step": 31450 }, { "epoch": 195.34890965732086, "grad_norm": 1.890625, "learning_rate": 6.825782207077191e-08, "loss": 0.1602, "step": 31451 }, { "epoch": 195.3551401869159, "grad_norm": 1.015625, "learning_rate": 6.807600515749135e-08, "loss": 0.0871, "step": 31452 }, { "epoch": 195.3613707165109, "grad_norm": 1.53125, "learning_rate": 6.7894430388038e-08, "loss": 0.0795, "step": 31453 }, { "epoch": 195.36760124610592, "grad_norm": 2.359375, "learning_rate": 6.771309776418267e-08, "loss": 0.1278, "step": 31454 }, { "epoch": 195.37383177570092, "grad_norm": 1.2578125, "learning_rate": 6.753200728767673e-08, "loss": 0.1081, "step": 31455 }, { "epoch": 195.38006230529595, "grad_norm": 3.484375, "learning_rate": 6.735115896028543e-08, "loss": 0.3112, "step": 31456 }, { "epoch": 195.38629283489095, "grad_norm": 1.53125, "learning_rate": 6.717055278376294e-08, "loss": 0.1235, "step": 31457 }, { "epoch": 195.39252336448598, "grad_norm": 2.453125, "learning_rate": 6.699018875986618e-08, "loss": 0.1773, "step": 31458 }, { "epoch": 195.39875389408098, "grad_norm": 1.6328125, "learning_rate": 6.681006689034098e-08, "loss": 0.1029, "step": 31459 }, { "epoch": 195.404984423676, "grad_norm": 2.96875, "learning_rate": 6.663018717694424e-08, "loss": 0.296, "step": 31460 }, { "epoch": 195.41121495327104, "grad_norm": 1.1875, "learning_rate": 6.645054962141906e-08, "loss": 0.1078, "step": 31461 }, { "epoch": 195.41744548286604, "grad_norm": 1.28125, "learning_rate": 6.627115422550844e-08, "loss": 0.1132, "step": 31462 }, { "epoch": 195.42367601246107, "grad_norm": 1.34375, "learning_rate": 6.609200099095547e-08, "loss": 0.1193, "step": 31463 }, { "epoch": 195.42990654205607, "grad_norm": 1.2109375, "learning_rate": 6.591308991950318e-08, "loss": 0.1097, "step": 31464 }, { "epoch": 195.4361370716511, "grad_norm": 1.65625, "learning_rate": 6.57344210128863e-08, "loss": 0.1266, "step": 31465 }, { "epoch": 195.4423676012461, "grad_norm": 2.25, "learning_rate": 6.555599427284231e-08, "loss": 0.2383, "step": 31466 }, { "epoch": 195.44859813084113, "grad_norm": 2.390625, "learning_rate": 6.537780970110319e-08, "loss": 0.1947, "step": 31467 }, { "epoch": 195.45482866043614, "grad_norm": 1.6640625, "learning_rate": 6.519986729939808e-08, "loss": 0.129, "step": 31468 }, { "epoch": 195.46105919003116, "grad_norm": 1.3515625, "learning_rate": 6.502216706945618e-08, "loss": 0.1153, "step": 31469 }, { "epoch": 195.46728971962617, "grad_norm": 2.8125, "learning_rate": 6.484470901300111e-08, "loss": 0.2225, "step": 31470 }, { "epoch": 195.4735202492212, "grad_norm": 2.0625, "learning_rate": 6.466749313176202e-08, "loss": 0.1987, "step": 31471 }, { "epoch": 195.4797507788162, "grad_norm": 1.59375, "learning_rate": 6.44905194274542e-08, "loss": 0.1968, "step": 31472 }, { "epoch": 195.48598130841123, "grad_norm": 2.9375, "learning_rate": 6.431378790180131e-08, "loss": 0.2296, "step": 31473 }, { "epoch": 195.49221183800623, "grad_norm": 2.078125, "learning_rate": 6.413729855651585e-08, "loss": 0.1484, "step": 31474 }, { "epoch": 195.49844236760126, "grad_norm": 3.359375, "learning_rate": 6.39610513933131e-08, "loss": 0.2986, "step": 31475 }, { "epoch": 195.50467289719626, "grad_norm": 1.15625, "learning_rate": 6.378504641390836e-08, "loss": 0.1069, "step": 31476 }, { "epoch": 195.5109034267913, "grad_norm": 1.640625, "learning_rate": 6.360928362000585e-08, "loss": 0.1548, "step": 31477 }, { "epoch": 195.5171339563863, "grad_norm": 2.546875, "learning_rate": 6.34337630133125e-08, "loss": 0.1514, "step": 31478 }, { "epoch": 195.52336448598132, "grad_norm": 2.875, "learning_rate": 6.325848459553807e-08, "loss": 0.2265, "step": 31479 }, { "epoch": 195.52959501557632, "grad_norm": 1.46875, "learning_rate": 6.30834483683812e-08, "loss": 0.1174, "step": 31480 }, { "epoch": 195.53582554517135, "grad_norm": 1.8828125, "learning_rate": 6.290865433354331e-08, "loss": 0.1392, "step": 31481 }, { "epoch": 195.54205607476635, "grad_norm": 2.859375, "learning_rate": 6.273410249272027e-08, "loss": 0.1683, "step": 31482 }, { "epoch": 195.54828660436138, "grad_norm": 1.9453125, "learning_rate": 6.255979284760794e-08, "loss": 0.1113, "step": 31483 }, { "epoch": 195.55451713395638, "grad_norm": 0.96484375, "learning_rate": 6.238572539990217e-08, "loss": 0.1081, "step": 31484 }, { "epoch": 195.5607476635514, "grad_norm": 2.609375, "learning_rate": 6.221190015129052e-08, "loss": 0.2314, "step": 31485 }, { "epoch": 195.5669781931464, "grad_norm": 2.140625, "learning_rate": 6.203831710346054e-08, "loss": 0.223, "step": 31486 }, { "epoch": 195.57320872274144, "grad_norm": 1.171875, "learning_rate": 6.186497625809972e-08, "loss": 0.1179, "step": 31487 }, { "epoch": 195.57943925233644, "grad_norm": 2.109375, "learning_rate": 6.169187761689011e-08, "loss": 0.1104, "step": 31488 }, { "epoch": 195.58566978193147, "grad_norm": 2.140625, "learning_rate": 6.151902118151643e-08, "loss": 0.1745, "step": 31489 }, { "epoch": 195.59190031152647, "grad_norm": 1.1171875, "learning_rate": 6.134640695365513e-08, "loss": 0.1227, "step": 31490 }, { "epoch": 195.5981308411215, "grad_norm": 1.875, "learning_rate": 6.117403493497987e-08, "loss": 0.0986, "step": 31491 }, { "epoch": 195.6043613707165, "grad_norm": 2.625, "learning_rate": 6.100190512716985e-08, "loss": 0.1812, "step": 31492 }, { "epoch": 195.61059190031153, "grad_norm": 1.6171875, "learning_rate": 6.083001753189599e-08, "loss": 0.132, "step": 31493 }, { "epoch": 195.61682242990653, "grad_norm": 1.25, "learning_rate": 6.06583721508236e-08, "loss": 0.1084, "step": 31494 }, { "epoch": 195.62305295950156, "grad_norm": 2.734375, "learning_rate": 6.048696898562356e-08, "loss": 0.1835, "step": 31495 }, { "epoch": 195.62928348909657, "grad_norm": 2.046875, "learning_rate": 6.031580803795845e-08, "loss": 0.1218, "step": 31496 }, { "epoch": 195.6355140186916, "grad_norm": 2.859375, "learning_rate": 6.01448893094908e-08, "loss": 0.275, "step": 31497 }, { "epoch": 195.6417445482866, "grad_norm": 2.484375, "learning_rate": 5.99742128018832e-08, "loss": 0.1591, "step": 31498 }, { "epoch": 195.64797507788163, "grad_norm": 2.046875, "learning_rate": 5.980377851678986e-08, "loss": 0.1006, "step": 31499 }, { "epoch": 195.65420560747663, "grad_norm": 2.21875, "learning_rate": 5.963358645587059e-08, "loss": 0.2141, "step": 31500 }, { "epoch": 195.66043613707166, "grad_norm": 3.5625, "learning_rate": 5.9463636620774035e-08, "loss": 0.4202, "step": 31501 }, { "epoch": 195.66666666666666, "grad_norm": 3.453125, "learning_rate": 5.92939290131489e-08, "loss": 0.2743, "step": 31502 }, { "epoch": 195.6728971962617, "grad_norm": 2.46875, "learning_rate": 5.912446363464941e-08, "loss": 0.1465, "step": 31503 }, { "epoch": 195.6791277258567, "grad_norm": 2.21875, "learning_rate": 5.8955240486921474e-08, "loss": 0.1814, "step": 31504 }, { "epoch": 195.68535825545172, "grad_norm": 1.5859375, "learning_rate": 5.878625957160266e-08, "loss": 0.1751, "step": 31505 }, { "epoch": 195.69158878504672, "grad_norm": 1.875, "learning_rate": 5.861752089033612e-08, "loss": 0.1343, "step": 31506 }, { "epoch": 195.69781931464175, "grad_norm": 2.5, "learning_rate": 5.844902444476497e-08, "loss": 0.2331, "step": 31507 }, { "epoch": 195.70404984423675, "grad_norm": 1.296875, "learning_rate": 5.8280770236518456e-08, "loss": 0.1255, "step": 31508 }, { "epoch": 195.71028037383178, "grad_norm": 0.8046875, "learning_rate": 5.8112758267239717e-08, "loss": 0.1003, "step": 31509 }, { "epoch": 195.71651090342678, "grad_norm": 2.6875, "learning_rate": 5.794498853855246e-08, "loss": 0.2187, "step": 31510 }, { "epoch": 195.7227414330218, "grad_norm": 2.265625, "learning_rate": 5.7777461052091474e-08, "loss": 0.1823, "step": 31511 }, { "epoch": 195.7289719626168, "grad_norm": 1.9375, "learning_rate": 5.7610175809480473e-08, "loss": 0.1498, "step": 31512 }, { "epoch": 195.73520249221184, "grad_norm": 1.265625, "learning_rate": 5.7443132812345925e-08, "loss": 0.1092, "step": 31513 }, { "epoch": 195.74143302180684, "grad_norm": 1.53125, "learning_rate": 5.727633206230876e-08, "loss": 0.1101, "step": 31514 }, { "epoch": 195.74766355140187, "grad_norm": 2.65625, "learning_rate": 5.71097735609899e-08, "loss": 0.2079, "step": 31515 }, { "epoch": 195.75389408099687, "grad_norm": 2.015625, "learning_rate": 5.69434573100075e-08, "loss": 0.1352, "step": 31516 }, { "epoch": 195.7601246105919, "grad_norm": 1.4921875, "learning_rate": 5.677738331097415e-08, "loss": 0.1228, "step": 31517 }, { "epoch": 195.7663551401869, "grad_norm": 2.203125, "learning_rate": 5.661155156550801e-08, "loss": 0.2562, "step": 31518 }, { "epoch": 195.77258566978193, "grad_norm": 1.8359375, "learning_rate": 5.644596207521335e-08, "loss": 0.1423, "step": 31519 }, { "epoch": 195.77881619937693, "grad_norm": 0.9921875, "learning_rate": 5.628061484170555e-08, "loss": 0.0879, "step": 31520 }, { "epoch": 195.78504672897196, "grad_norm": 2.75, "learning_rate": 5.611550986658332e-08, "loss": 0.2157, "step": 31521 }, { "epoch": 195.79127725856696, "grad_norm": 1.0859375, "learning_rate": 5.5950647151453705e-08, "loss": 0.0901, "step": 31522 }, { "epoch": 195.797507788162, "grad_norm": 2.875, "learning_rate": 5.578602669791821e-08, "loss": 0.2569, "step": 31523 }, { "epoch": 195.80373831775702, "grad_norm": 1.7578125, "learning_rate": 5.562164850757556e-08, "loss": 0.1773, "step": 31524 }, { "epoch": 195.80996884735202, "grad_norm": 1.9921875, "learning_rate": 5.54575125820217e-08, "loss": 0.1768, "step": 31525 }, { "epoch": 195.81619937694705, "grad_norm": 3.53125, "learning_rate": 5.529361892284979e-08, "loss": 0.2673, "step": 31526 }, { "epoch": 195.82242990654206, "grad_norm": 1.796875, "learning_rate": 5.512996753165578e-08, "loss": 0.1285, "step": 31527 }, { "epoch": 195.82866043613708, "grad_norm": 2.453125, "learning_rate": 5.496655841002174e-08, "loss": 0.2106, "step": 31528 }, { "epoch": 195.8348909657321, "grad_norm": 1.9296875, "learning_rate": 5.480339155954084e-08, "loss": 0.1859, "step": 31529 }, { "epoch": 195.84112149532712, "grad_norm": 1.09375, "learning_rate": 5.464046698179515e-08, "loss": 0.1032, "step": 31530 }, { "epoch": 195.84735202492212, "grad_norm": 1.9140625, "learning_rate": 5.44777846783695e-08, "loss": 0.1647, "step": 31531 }, { "epoch": 195.85358255451715, "grad_norm": 2.234375, "learning_rate": 5.431534465084043e-08, "loss": 0.1572, "step": 31532 }, { "epoch": 195.85981308411215, "grad_norm": 1.4375, "learning_rate": 5.415314690078721e-08, "loss": 0.0965, "step": 31533 }, { "epoch": 195.86604361370718, "grad_norm": 2.578125, "learning_rate": 5.399119142978637e-08, "loss": 0.2633, "step": 31534 }, { "epoch": 195.87227414330218, "grad_norm": 2.4375, "learning_rate": 5.382947823940887e-08, "loss": 0.2045, "step": 31535 }, { "epoch": 195.8785046728972, "grad_norm": 2.40625, "learning_rate": 5.3668007331225676e-08, "loss": 0.1699, "step": 31536 }, { "epoch": 195.8847352024922, "grad_norm": 1.1171875, "learning_rate": 5.350677870680776e-08, "loss": 0.108, "step": 31537 }, { "epoch": 195.89096573208724, "grad_norm": 1.4296875, "learning_rate": 5.334579236771775e-08, "loss": 0.0768, "step": 31538 }, { "epoch": 195.89719626168224, "grad_norm": 2.3125, "learning_rate": 5.3185048315518295e-08, "loss": 0.1869, "step": 31539 }, { "epoch": 195.90342679127727, "grad_norm": 1.78125, "learning_rate": 5.3024546551772026e-08, "loss": 0.0958, "step": 31540 }, { "epoch": 195.90965732087227, "grad_norm": 0.93359375, "learning_rate": 5.2864287078041584e-08, "loss": 0.1007, "step": 31541 }, { "epoch": 195.9158878504673, "grad_norm": 2.375, "learning_rate": 5.270426989587851e-08, "loss": 0.2019, "step": 31542 }, { "epoch": 195.9221183800623, "grad_norm": 1.0859375, "learning_rate": 5.254449500683711e-08, "loss": 0.1118, "step": 31543 }, { "epoch": 195.92834890965733, "grad_norm": 1.3515625, "learning_rate": 5.2384962412471703e-08, "loss": 0.1132, "step": 31544 }, { "epoch": 195.93457943925233, "grad_norm": 1.8671875, "learning_rate": 5.2225672114328274e-08, "loss": 0.1419, "step": 31545 }, { "epoch": 195.94080996884736, "grad_norm": 2.890625, "learning_rate": 5.206662411395835e-08, "loss": 0.1615, "step": 31546 }, { "epoch": 195.94704049844236, "grad_norm": 2.671875, "learning_rate": 5.190781841290515e-08, "loss": 0.2629, "step": 31547 }, { "epoch": 195.9532710280374, "grad_norm": 2.28125, "learning_rate": 5.174925501270911e-08, "loss": 0.1247, "step": 31548 }, { "epoch": 195.9595015576324, "grad_norm": 1.453125, "learning_rate": 5.159093391491066e-08, "loss": 0.1254, "step": 31549 }, { "epoch": 195.96573208722742, "grad_norm": 1.0625, "learning_rate": 5.1432855121050224e-08, "loss": 0.1001, "step": 31550 }, { "epoch": 195.97196261682242, "grad_norm": 2.875, "learning_rate": 5.1275018632657154e-08, "loss": 0.2026, "step": 31551 }, { "epoch": 195.97819314641745, "grad_norm": 0.828125, "learning_rate": 5.111742445127188e-08, "loss": 0.0981, "step": 31552 }, { "epoch": 195.98442367601245, "grad_norm": 1.6953125, "learning_rate": 5.096007257842095e-08, "loss": 0.1157, "step": 31553 }, { "epoch": 195.99065420560748, "grad_norm": 1.5390625, "learning_rate": 5.080296301563092e-08, "loss": 0.1237, "step": 31554 }, { "epoch": 195.99688473520249, "grad_norm": 1.4296875, "learning_rate": 5.064609576443113e-08, "loss": 0.1156, "step": 31555 }, { "epoch": 196.0, "grad_norm": 1.78125, "learning_rate": 5.0489470826345366e-08, "loss": 0.1306, "step": 31556 }, { "epoch": 196.00623052959503, "grad_norm": 0.85546875, "learning_rate": 5.033308820289184e-08, "loss": 0.1042, "step": 31557 }, { "epoch": 196.01246105919003, "grad_norm": 2.546875, "learning_rate": 5.01769478955888e-08, "loss": 0.2812, "step": 31558 }, { "epoch": 196.01869158878506, "grad_norm": 2.96875, "learning_rate": 5.002104990595724e-08, "loss": 0.2364, "step": 31559 }, { "epoch": 196.02492211838006, "grad_norm": 1.84375, "learning_rate": 4.9865394235507065e-08, "loss": 0.1645, "step": 31560 }, { "epoch": 196.0311526479751, "grad_norm": 2.265625, "learning_rate": 4.970998088575096e-08, "loss": 0.1518, "step": 31561 }, { "epoch": 196.0373831775701, "grad_norm": 2.734375, "learning_rate": 4.95548098582016e-08, "loss": 0.1529, "step": 31562 }, { "epoch": 196.04361370716512, "grad_norm": 2.625, "learning_rate": 4.939988115436056e-08, "loss": 0.199, "step": 31563 }, { "epoch": 196.04984423676012, "grad_norm": 2.5, "learning_rate": 4.9245194775737745e-08, "loss": 0.2546, "step": 31564 }, { "epoch": 196.05607476635515, "grad_norm": 1.5, "learning_rate": 4.909075072383196e-08, "loss": 0.0981, "step": 31565 }, { "epoch": 196.06230529595015, "grad_norm": 2.96875, "learning_rate": 4.893654900014477e-08, "loss": 0.1634, "step": 31566 }, { "epoch": 196.06853582554518, "grad_norm": 2.359375, "learning_rate": 4.878258960617221e-08, "loss": 0.1698, "step": 31567 }, { "epoch": 196.07476635514018, "grad_norm": 1.2421875, "learning_rate": 4.8628872543413084e-08, "loss": 0.1144, "step": 31568 }, { "epoch": 196.0809968847352, "grad_norm": 1.875, "learning_rate": 4.847539781335786e-08, "loss": 0.1649, "step": 31569 }, { "epoch": 196.08722741433021, "grad_norm": 1.5078125, "learning_rate": 4.832216541749424e-08, "loss": 0.1248, "step": 31570 }, { "epoch": 196.09345794392524, "grad_norm": 2.484375, "learning_rate": 4.816917535731547e-08, "loss": 0.1845, "step": 31571 }, { "epoch": 196.09968847352025, "grad_norm": 2.796875, "learning_rate": 4.801642763430647e-08, "loss": 0.2573, "step": 31572 }, { "epoch": 196.10591900311528, "grad_norm": 2.40625, "learning_rate": 4.7863922249946624e-08, "loss": 0.1902, "step": 31573 }, { "epoch": 196.11214953271028, "grad_norm": 3.453125, "learning_rate": 4.771165920572362e-08, "loss": 0.2358, "step": 31574 }, { "epoch": 196.1183800623053, "grad_norm": 2.828125, "learning_rate": 4.7559638503111294e-08, "loss": 0.2261, "step": 31575 }, { "epoch": 196.1246105919003, "grad_norm": 2.0, "learning_rate": 4.7407860143589e-08, "loss": 0.1147, "step": 31576 }, { "epoch": 196.13084112149534, "grad_norm": 1.984375, "learning_rate": 4.72563241286278e-08, "loss": 0.1577, "step": 31577 }, { "epoch": 196.13707165109034, "grad_norm": 2.21875, "learning_rate": 4.7105030459701496e-08, "loss": 0.1778, "step": 31578 }, { "epoch": 196.14330218068537, "grad_norm": 2.125, "learning_rate": 4.695397913827837e-08, "loss": 0.1605, "step": 31579 }, { "epoch": 196.14953271028037, "grad_norm": 1.421875, "learning_rate": 4.680317016582669e-08, "loss": 0.1257, "step": 31580 }, { "epoch": 196.1557632398754, "grad_norm": 2.4375, "learning_rate": 4.665260354381196e-08, "loss": 0.2076, "step": 31581 }, { "epoch": 196.1619937694704, "grad_norm": 1.4609375, "learning_rate": 4.650227927369411e-08, "loss": 0.171, "step": 31582 }, { "epoch": 196.16822429906543, "grad_norm": 2.0, "learning_rate": 4.6352197356935856e-08, "loss": 0.1582, "step": 31583 }, { "epoch": 196.17445482866043, "grad_norm": 2.046875, "learning_rate": 4.6202357794991604e-08, "loss": 0.1425, "step": 31584 }, { "epoch": 196.18068535825546, "grad_norm": 2.8125, "learning_rate": 4.605276058931851e-08, "loss": 0.2564, "step": 31585 }, { "epoch": 196.18691588785046, "grad_norm": 2.8125, "learning_rate": 4.59034057413682e-08, "loss": 0.1716, "step": 31586 }, { "epoch": 196.1931464174455, "grad_norm": 1.71875, "learning_rate": 4.5754293252595056e-08, "loss": 0.1146, "step": 31587 }, { "epoch": 196.1993769470405, "grad_norm": 2.359375, "learning_rate": 4.5605423124442384e-08, "loss": 0.19, "step": 31588 }, { "epoch": 196.20560747663552, "grad_norm": 2.40625, "learning_rate": 4.545679535835623e-08, "loss": 0.1679, "step": 31589 }, { "epoch": 196.21183800623052, "grad_norm": 1.6484375, "learning_rate": 4.5308409955785445e-08, "loss": 0.1655, "step": 31590 }, { "epoch": 196.21806853582555, "grad_norm": 2.546875, "learning_rate": 4.5160266918164995e-08, "loss": 0.1652, "step": 31591 }, { "epoch": 196.22429906542055, "grad_norm": 2.125, "learning_rate": 4.5012366246938166e-08, "loss": 0.119, "step": 31592 }, { "epoch": 196.23052959501558, "grad_norm": 2.765625, "learning_rate": 4.486470794353992e-08, "loss": 0.2128, "step": 31593 }, { "epoch": 196.23676012461058, "grad_norm": 1.7578125, "learning_rate": 4.471729200940522e-08, "loss": 0.1043, "step": 31594 }, { "epoch": 196.2429906542056, "grad_norm": 1.203125, "learning_rate": 4.457011844596071e-08, "loss": 0.1127, "step": 31595 }, { "epoch": 196.2492211838006, "grad_norm": 1.5625, "learning_rate": 4.442318725464134e-08, "loss": 0.1329, "step": 31596 }, { "epoch": 196.25545171339564, "grad_norm": 3.0, "learning_rate": 4.427649843687376e-08, "loss": 0.2176, "step": 31597 }, { "epoch": 196.26168224299064, "grad_norm": 2.6875, "learning_rate": 4.413005199408182e-08, "loss": 0.1767, "step": 31598 }, { "epoch": 196.26791277258567, "grad_norm": 2.765625, "learning_rate": 4.3983847927683844e-08, "loss": 0.2682, "step": 31599 }, { "epoch": 196.27414330218068, "grad_norm": 1.1796875, "learning_rate": 4.383788623910645e-08, "loss": 0.1087, "step": 31600 }, { "epoch": 196.2803738317757, "grad_norm": 1.3828125, "learning_rate": 4.369216692976519e-08, "loss": 0.11, "step": 31601 }, { "epoch": 196.2866043613707, "grad_norm": 2.671875, "learning_rate": 4.354669000107281e-08, "loss": 0.2949, "step": 31602 }, { "epoch": 196.29283489096574, "grad_norm": 1.0625, "learning_rate": 4.340145545444207e-08, "loss": 0.099, "step": 31603 }, { "epoch": 196.29906542056074, "grad_norm": 2.078125, "learning_rate": 4.325646329128574e-08, "loss": 0.1553, "step": 31604 }, { "epoch": 196.30529595015577, "grad_norm": 2.109375, "learning_rate": 4.3111713513013797e-08, "loss": 0.1623, "step": 31605 }, { "epoch": 196.31152647975077, "grad_norm": 1.0859375, "learning_rate": 4.296720612102789e-08, "loss": 0.1037, "step": 31606 }, { "epoch": 196.3177570093458, "grad_norm": 1.703125, "learning_rate": 4.282294111673246e-08, "loss": 0.1122, "step": 31607 }, { "epoch": 196.3239875389408, "grad_norm": 3.328125, "learning_rate": 4.267891850152916e-08, "loss": 0.3053, "step": 31608 }, { "epoch": 196.33021806853583, "grad_norm": 1.5234375, "learning_rate": 4.253513827681965e-08, "loss": 0.118, "step": 31609 }, { "epoch": 196.33644859813083, "grad_norm": 1.703125, "learning_rate": 4.2391600443994486e-08, "loss": 0.1443, "step": 31610 }, { "epoch": 196.34267912772586, "grad_norm": 1.0625, "learning_rate": 4.224830500445254e-08, "loss": 0.1172, "step": 31611 }, { "epoch": 196.34890965732086, "grad_norm": 1.9375, "learning_rate": 4.210525195958437e-08, "loss": 0.1435, "step": 31612 }, { "epoch": 196.3551401869159, "grad_norm": 3.390625, "learning_rate": 4.196244131078053e-08, "loss": 0.3523, "step": 31613 }, { "epoch": 196.3613707165109, "grad_norm": 2.59375, "learning_rate": 4.181987305942603e-08, "loss": 0.233, "step": 31614 }, { "epoch": 196.36760124610592, "grad_norm": 1.8671875, "learning_rate": 4.167754720690309e-08, "loss": 0.1195, "step": 31615 }, { "epoch": 196.37383177570092, "grad_norm": 1.4296875, "learning_rate": 4.153546375459949e-08, "loss": 0.129, "step": 31616 }, { "epoch": 196.38006230529595, "grad_norm": 2.03125, "learning_rate": 4.139362270389191e-08, "loss": 0.1936, "step": 31617 }, { "epoch": 196.38629283489095, "grad_norm": 1.1171875, "learning_rate": 4.1252024056157026e-08, "loss": 0.0973, "step": 31618 }, { "epoch": 196.39252336448598, "grad_norm": 3.296875, "learning_rate": 4.111066781277428e-08, "loss": 0.3392, "step": 31619 }, { "epoch": 196.39875389408098, "grad_norm": 1.6484375, "learning_rate": 4.096955397511204e-08, "loss": 0.1332, "step": 31620 }, { "epoch": 196.404984423676, "grad_norm": 1.265625, "learning_rate": 4.082868254454419e-08, "loss": 0.1138, "step": 31621 }, { "epoch": 196.41121495327104, "grad_norm": 2.703125, "learning_rate": 4.0688053522436323e-08, "loss": 0.2342, "step": 31622 }, { "epoch": 196.41744548286604, "grad_norm": 1.7109375, "learning_rate": 4.0547666910153994e-08, "loss": 0.1171, "step": 31623 }, { "epoch": 196.42367601246107, "grad_norm": 3.375, "learning_rate": 4.040752270906278e-08, "loss": 0.3092, "step": 31624 }, { "epoch": 196.42990654205607, "grad_norm": 0.77734375, "learning_rate": 4.0267620920525495e-08, "loss": 0.0821, "step": 31625 }, { "epoch": 196.4361370716511, "grad_norm": 2.34375, "learning_rate": 4.01279615458966e-08, "loss": 0.1639, "step": 31626 }, { "epoch": 196.4423676012461, "grad_norm": 1.96875, "learning_rate": 3.998854458653334e-08, "loss": 0.1523, "step": 31627 }, { "epoch": 196.44859813084113, "grad_norm": 2.234375, "learning_rate": 3.984937004379297e-08, "loss": 0.1883, "step": 31628 }, { "epoch": 196.45482866043614, "grad_norm": 3.28125, "learning_rate": 3.97104379190244e-08, "loss": 0.371, "step": 31629 }, { "epoch": 196.46105919003116, "grad_norm": 2.625, "learning_rate": 3.957174821357656e-08, "loss": 0.1976, "step": 31630 }, { "epoch": 196.46728971962617, "grad_norm": 1.9453125, "learning_rate": 3.943330092879838e-08, "loss": 0.1309, "step": 31631 }, { "epoch": 196.4735202492212, "grad_norm": 2.109375, "learning_rate": 3.929509606603321e-08, "loss": 0.1584, "step": 31632 }, { "epoch": 196.4797507788162, "grad_norm": 1.625, "learning_rate": 3.915713362662166e-08, "loss": 0.1657, "step": 31633 }, { "epoch": 196.48598130841123, "grad_norm": 2.03125, "learning_rate": 3.9019413611907086e-08, "loss": 0.1214, "step": 31634 }, { "epoch": 196.49221183800623, "grad_norm": 2.3125, "learning_rate": 3.8881936023227314e-08, "loss": 0.1532, "step": 31635 }, { "epoch": 196.49844236760126, "grad_norm": 3.375, "learning_rate": 3.874470086191184e-08, "loss": 0.2143, "step": 31636 }, { "epoch": 196.50467289719626, "grad_norm": 2.125, "learning_rate": 3.860770812930126e-08, "loss": 0.1378, "step": 31637 }, { "epoch": 196.5109034267913, "grad_norm": 3.21875, "learning_rate": 3.847095782672228e-08, "loss": 0.2111, "step": 31638 }, { "epoch": 196.5171339563863, "grad_norm": 2.3125, "learning_rate": 3.8334449955501616e-08, "loss": 0.1937, "step": 31639 }, { "epoch": 196.52336448598132, "grad_norm": 0.83984375, "learning_rate": 3.8198184516965994e-08, "loss": 0.1001, "step": 31640 }, { "epoch": 196.52959501557632, "grad_norm": 2.4375, "learning_rate": 3.806216151243935e-08, "loss": 0.1898, "step": 31641 }, { "epoch": 196.53582554517135, "grad_norm": 1.4296875, "learning_rate": 3.7926380943242854e-08, "loss": 0.1203, "step": 31642 }, { "epoch": 196.54205607476635, "grad_norm": 1.96875, "learning_rate": 3.7790842810694896e-08, "loss": 0.1822, "step": 31643 }, { "epoch": 196.54828660436138, "grad_norm": 2.5625, "learning_rate": 3.765554711611385e-08, "loss": 0.2516, "step": 31644 }, { "epoch": 196.55451713395638, "grad_norm": 2.53125, "learning_rate": 3.7520493860812575e-08, "loss": 0.2619, "step": 31645 }, { "epoch": 196.5607476635514, "grad_norm": 1.8515625, "learning_rate": 3.7385683046098355e-08, "loss": 0.1323, "step": 31646 }, { "epoch": 196.5669781931464, "grad_norm": 1.671875, "learning_rate": 3.7251114673286793e-08, "loss": 0.1116, "step": 31647 }, { "epoch": 196.57320872274144, "grad_norm": 3.8125, "learning_rate": 3.711678874368241e-08, "loss": 0.2209, "step": 31648 }, { "epoch": 196.57943925233644, "grad_norm": 1.0625, "learning_rate": 3.698270525858971e-08, "loss": 0.1089, "step": 31649 }, { "epoch": 196.58566978193147, "grad_norm": 1.921875, "learning_rate": 3.684886421931044e-08, "loss": 0.1862, "step": 31650 }, { "epoch": 196.59190031152647, "grad_norm": 1.4375, "learning_rate": 3.6715265627146334e-08, "loss": 0.1025, "step": 31651 }, { "epoch": 196.5981308411215, "grad_norm": 1.3671875, "learning_rate": 3.6581909483393574e-08, "loss": 0.1018, "step": 31652 }, { "epoch": 196.6043613707165, "grad_norm": 1.546875, "learning_rate": 3.644879578934557e-08, "loss": 0.1423, "step": 31653 }, { "epoch": 196.61059190031153, "grad_norm": 3.09375, "learning_rate": 3.6315924546298506e-08, "loss": 0.2662, "step": 31654 }, { "epoch": 196.61682242990653, "grad_norm": 2.84375, "learning_rate": 3.6183295755540245e-08, "loss": 0.2657, "step": 31655 }, { "epoch": 196.62305295950156, "grad_norm": 2.265625, "learning_rate": 3.605090941836142e-08, "loss": 0.264, "step": 31656 }, { "epoch": 196.62928348909657, "grad_norm": 2.609375, "learning_rate": 3.5918765536044344e-08, "loss": 0.1845, "step": 31657 }, { "epoch": 196.6355140186916, "grad_norm": 1.6796875, "learning_rate": 3.578686410987686e-08, "loss": 0.1747, "step": 31658 }, { "epoch": 196.6417445482866, "grad_norm": 1.046875, "learning_rate": 3.565520514113574e-08, "loss": 0.0998, "step": 31659 }, { "epoch": 196.64797507788163, "grad_norm": 2.65625, "learning_rate": 3.5523788631100505e-08, "loss": 0.2247, "step": 31660 }, { "epoch": 196.65420560747663, "grad_norm": 2.171875, "learning_rate": 3.53926145810507e-08, "loss": 0.1789, "step": 31661 }, { "epoch": 196.66043613707166, "grad_norm": 2.15625, "learning_rate": 3.526168299225751e-08, "loss": 0.1586, "step": 31662 }, { "epoch": 196.66666666666666, "grad_norm": 1.421875, "learning_rate": 3.5130993865992165e-08, "loss": 0.133, "step": 31663 }, { "epoch": 196.6728971962617, "grad_norm": 1.3359375, "learning_rate": 3.500054720352863e-08, "loss": 0.1294, "step": 31664 }, { "epoch": 196.6791277258567, "grad_norm": 1.0625, "learning_rate": 3.4870343006127014e-08, "loss": 0.1125, "step": 31665 }, { "epoch": 196.68535825545172, "grad_norm": 1.9375, "learning_rate": 3.474038127505574e-08, "loss": 0.139, "step": 31666 }, { "epoch": 196.69158878504672, "grad_norm": 2.234375, "learning_rate": 3.4610662011574926e-08, "loss": 0.1912, "step": 31667 }, { "epoch": 196.69781931464175, "grad_norm": 1.21875, "learning_rate": 3.448118521694743e-08, "loss": 0.1244, "step": 31668 }, { "epoch": 196.70404984423675, "grad_norm": 2.15625, "learning_rate": 3.435195089242782e-08, "loss": 0.2319, "step": 31669 }, { "epoch": 196.71028037383178, "grad_norm": 1.6328125, "learning_rate": 3.4222959039276195e-08, "loss": 0.1133, "step": 31670 }, { "epoch": 196.71651090342678, "grad_norm": 3.0625, "learning_rate": 3.409420965873877e-08, "loss": 0.278, "step": 31671 }, { "epoch": 196.7227414330218, "grad_norm": 2.4375, "learning_rate": 3.396570275207012e-08, "loss": 0.1671, "step": 31672 }, { "epoch": 196.7289719626168, "grad_norm": 2.4375, "learning_rate": 3.383743832051367e-08, "loss": 0.1774, "step": 31673 }, { "epoch": 196.73520249221184, "grad_norm": 2.65625, "learning_rate": 3.370941636532121e-08, "loss": 0.2026, "step": 31674 }, { "epoch": 196.74143302180684, "grad_norm": 2.3125, "learning_rate": 3.358163688773619e-08, "loss": 0.2084, "step": 31675 }, { "epoch": 196.74766355140187, "grad_norm": 2.03125, "learning_rate": 3.345409988899373e-08, "loss": 0.2336, "step": 31676 }, { "epoch": 196.75389408099687, "grad_norm": 2.671875, "learning_rate": 3.3326805370337276e-08, "loss": 0.1485, "step": 31677 }, { "epoch": 196.7601246105919, "grad_norm": 2.078125, "learning_rate": 3.319975333299918e-08, "loss": 0.1092, "step": 31678 }, { "epoch": 196.7663551401869, "grad_norm": 1.8203125, "learning_rate": 3.307294377822012e-08, "loss": 0.1509, "step": 31679 }, { "epoch": 196.77258566978193, "grad_norm": 2.84375, "learning_rate": 3.2946376707224114e-08, "loss": 0.2203, "step": 31680 }, { "epoch": 196.77881619937693, "grad_norm": 2.40625, "learning_rate": 3.2820052121243504e-08, "loss": 0.1974, "step": 31681 }, { "epoch": 196.78504672897196, "grad_norm": 2.15625, "learning_rate": 3.269397002150787e-08, "loss": 0.2031, "step": 31682 }, { "epoch": 196.79127725856696, "grad_norm": 1.8359375, "learning_rate": 3.256813040923845e-08, "loss": 0.1255, "step": 31683 }, { "epoch": 196.797507788162, "grad_norm": 1.8671875, "learning_rate": 3.2442533285656496e-08, "loss": 0.1793, "step": 31684 }, { "epoch": 196.80373831775702, "grad_norm": 2.203125, "learning_rate": 3.231717865198325e-08, "loss": 0.1072, "step": 31685 }, { "epoch": 196.80996884735202, "grad_norm": 1.4296875, "learning_rate": 3.2192066509439955e-08, "loss": 0.1158, "step": 31686 }, { "epoch": 196.81619937694705, "grad_norm": 1.828125, "learning_rate": 3.206719685923676e-08, "loss": 0.1669, "step": 31687 }, { "epoch": 196.82242990654206, "grad_norm": 1.78125, "learning_rate": 3.194256970258658e-08, "loss": 0.1185, "step": 31688 }, { "epoch": 196.82866043613708, "grad_norm": 2.984375, "learning_rate": 3.181818504070233e-08, "loss": 0.2102, "step": 31689 }, { "epoch": 196.8348909657321, "grad_norm": 1.40625, "learning_rate": 3.169404287479139e-08, "loss": 0.1221, "step": 31690 }, { "epoch": 196.84112149532712, "grad_norm": 2.03125, "learning_rate": 3.1570143206058334e-08, "loss": 0.1523, "step": 31691 }, { "epoch": 196.84735202492212, "grad_norm": 2.09375, "learning_rate": 3.144648603570499e-08, "loss": 0.1509, "step": 31692 }, { "epoch": 196.85358255451715, "grad_norm": 2.015625, "learning_rate": 3.1323071364935954e-08, "loss": 0.1188, "step": 31693 }, { "epoch": 196.85981308411215, "grad_norm": 2.453125, "learning_rate": 3.1199899194950257e-08, "loss": 0.225, "step": 31694 }, { "epoch": 196.86604361370718, "grad_norm": 3.828125, "learning_rate": 3.107696952694139e-08, "loss": 0.4054, "step": 31695 }, { "epoch": 196.87227414330218, "grad_norm": 2.671875, "learning_rate": 3.095428236210285e-08, "loss": 0.2466, "step": 31696 }, { "epoch": 196.8785046728972, "grad_norm": 3.46875, "learning_rate": 3.083183770162812e-08, "loss": 0.2573, "step": 31697 }, { "epoch": 196.8847352024922, "grad_norm": 1.8515625, "learning_rate": 3.0709635546707914e-08, "loss": 0.1316, "step": 31698 }, { "epoch": 196.89096573208724, "grad_norm": 1.8359375, "learning_rate": 3.0587675898524624e-08, "loss": 0.0972, "step": 31699 }, { "epoch": 196.89719626168224, "grad_norm": 1.7890625, "learning_rate": 3.046595875826619e-08, "loss": 0.1525, "step": 31700 }, { "epoch": 196.90342679127727, "grad_norm": 1.6875, "learning_rate": 3.0344484127112215e-08, "loss": 0.1565, "step": 31701 }, { "epoch": 196.90965732087227, "grad_norm": 0.76171875, "learning_rate": 3.0223252006245094e-08, "loss": 0.0879, "step": 31702 }, { "epoch": 196.9158878504673, "grad_norm": 2.921875, "learning_rate": 3.0102262396841664e-08, "loss": 0.2426, "step": 31703 }, { "epoch": 196.9221183800623, "grad_norm": 1.8515625, "learning_rate": 2.998151530007598e-08, "loss": 0.1437, "step": 31704 }, { "epoch": 196.92834890965733, "grad_norm": 3.15625, "learning_rate": 2.986101071711933e-08, "loss": 0.2973, "step": 31705 }, { "epoch": 196.93457943925233, "grad_norm": 2.25, "learning_rate": 2.974074864914578e-08, "loss": 0.1751, "step": 31706 }, { "epoch": 196.94080996884736, "grad_norm": 1.9765625, "learning_rate": 2.9620729097321052e-08, "loss": 0.1519, "step": 31707 }, { "epoch": 196.94704049844236, "grad_norm": 2.046875, "learning_rate": 2.950095206281367e-08, "loss": 0.1476, "step": 31708 }, { "epoch": 196.9532710280374, "grad_norm": 1.3125, "learning_rate": 2.938141754678103e-08, "loss": 0.1181, "step": 31709 }, { "epoch": 196.9595015576324, "grad_norm": 0.9609375, "learning_rate": 2.9262125550388875e-08, "loss": 0.0868, "step": 31710 }, { "epoch": 196.96573208722742, "grad_norm": 1.9375, "learning_rate": 2.9143076074794607e-08, "loss": 0.1373, "step": 31711 }, { "epoch": 196.97196261682242, "grad_norm": 1.9609375, "learning_rate": 2.9024269121155634e-08, "loss": 0.1725, "step": 31712 }, { "epoch": 196.97819314641745, "grad_norm": 1.9609375, "learning_rate": 2.8905704690623813e-08, "loss": 0.138, "step": 31713 }, { "epoch": 196.98442367601245, "grad_norm": 3.109375, "learning_rate": 2.8787382784351e-08, "loss": 0.234, "step": 31714 }, { "epoch": 196.99065420560748, "grad_norm": 0.98828125, "learning_rate": 2.866930340348628e-08, "loss": 0.0878, "step": 31715 }, { "epoch": 196.99688473520249, "grad_norm": 2.796875, "learning_rate": 2.8551466549175954e-08, "loss": 0.2724, "step": 31716 }, { "epoch": 197.0, "grad_norm": 1.0859375, "learning_rate": 2.84338722225691e-08, "loss": 0.0539, "step": 31717 }, { "epoch": 197.00623052959503, "grad_norm": 2.140625, "learning_rate": 2.831652042480093e-08, "loss": 0.1949, "step": 31718 }, { "epoch": 197.01246105919003, "grad_norm": 2.25, "learning_rate": 2.8199411157014966e-08, "loss": 0.1024, "step": 31719 }, { "epoch": 197.01869158878506, "grad_norm": 1.03125, "learning_rate": 2.8082544420349187e-08, "loss": 0.0958, "step": 31720 }, { "epoch": 197.02492211838006, "grad_norm": 2.890625, "learning_rate": 2.7965920215936026e-08, "loss": 0.2311, "step": 31721 }, { "epoch": 197.0311526479751, "grad_norm": 2.03125, "learning_rate": 2.784953854491068e-08, "loss": 0.1813, "step": 31722 }, { "epoch": 197.0373831775701, "grad_norm": 2.203125, "learning_rate": 2.7733399408402804e-08, "loss": 0.185, "step": 31723 }, { "epoch": 197.04361370716512, "grad_norm": 2.0, "learning_rate": 2.7617502807539275e-08, "loss": 0.2499, "step": 31724 }, { "epoch": 197.04984423676012, "grad_norm": 2.375, "learning_rate": 2.7501848743446967e-08, "loss": 0.1834, "step": 31725 }, { "epoch": 197.05607476635515, "grad_norm": 2.34375, "learning_rate": 2.7386437217249983e-08, "loss": 0.3129, "step": 31726 }, { "epoch": 197.06230529595015, "grad_norm": 2.46875, "learning_rate": 2.727126823006687e-08, "loss": 0.2337, "step": 31727 }, { "epoch": 197.06853582554518, "grad_norm": 3.46875, "learning_rate": 2.715634178301618e-08, "loss": 0.3811, "step": 31728 }, { "epoch": 197.07476635514018, "grad_norm": 2.890625, "learning_rate": 2.704165787721369e-08, "loss": 0.1068, "step": 31729 }, { "epoch": 197.0809968847352, "grad_norm": 1.96875, "learning_rate": 2.6927216513777942e-08, "loss": 0.1801, "step": 31730 }, { "epoch": 197.08722741433021, "grad_norm": 2.03125, "learning_rate": 2.6813017693813613e-08, "loss": 0.1011, "step": 31731 }, { "epoch": 197.09345794392524, "grad_norm": 1.8515625, "learning_rate": 2.669906141843648e-08, "loss": 0.1591, "step": 31732 }, { "epoch": 197.09968847352025, "grad_norm": 2.515625, "learning_rate": 2.658534768874843e-08, "loss": 0.1313, "step": 31733 }, { "epoch": 197.10591900311528, "grad_norm": 2.671875, "learning_rate": 2.6471876505856918e-08, "loss": 0.1778, "step": 31734 }, { "epoch": 197.11214953271028, "grad_norm": 1.0859375, "learning_rate": 2.6358647870861064e-08, "loss": 0.1283, "step": 31735 }, { "epoch": 197.1183800623053, "grad_norm": 3.46875, "learning_rate": 2.6245661784862764e-08, "loss": 0.3072, "step": 31736 }, { "epoch": 197.1246105919003, "grad_norm": 1.4453125, "learning_rate": 2.613291824895836e-08, "loss": 0.0956, "step": 31737 }, { "epoch": 197.13084112149534, "grad_norm": 1.8125, "learning_rate": 2.60204172642442e-08, "loss": 0.1487, "step": 31738 }, { "epoch": 197.13707165109034, "grad_norm": 2.40625, "learning_rate": 2.590815883181108e-08, "loss": 0.1384, "step": 31739 }, { "epoch": 197.14330218068537, "grad_norm": 2.484375, "learning_rate": 2.579614295274979e-08, "loss": 0.1732, "step": 31740 }, { "epoch": 197.14953271028037, "grad_norm": 2.46875, "learning_rate": 2.5684369628148353e-08, "loss": 0.1311, "step": 31741 }, { "epoch": 197.1557632398754, "grad_norm": 2.09375, "learning_rate": 2.5572838859094784e-08, "loss": 0.2464, "step": 31742 }, { "epoch": 197.1619937694704, "grad_norm": 1.5859375, "learning_rate": 2.5461550646668775e-08, "loss": 0.1161, "step": 31743 }, { "epoch": 197.16822429906543, "grad_norm": 1.703125, "learning_rate": 2.5350504991952794e-08, "loss": 0.147, "step": 31744 }, { "epoch": 197.17445482866043, "grad_norm": 2.5, "learning_rate": 2.5239701896023758e-08, "loss": 0.2251, "step": 31745 }, { "epoch": 197.18068535825546, "grad_norm": 2.515625, "learning_rate": 2.512914135996136e-08, "loss": 0.2398, "step": 31746 }, { "epoch": 197.18691588785046, "grad_norm": 2.609375, "learning_rate": 2.5018823384836966e-08, "loss": 0.2185, "step": 31747 }, { "epoch": 197.1931464174455, "grad_norm": 1.3125, "learning_rate": 2.490874797171916e-08, "loss": 0.1191, "step": 31748 }, { "epoch": 197.1993769470405, "grad_norm": 1.7734375, "learning_rate": 2.4798915121682088e-08, "loss": 0.1646, "step": 31749 }, { "epoch": 197.20560747663552, "grad_norm": 1.234375, "learning_rate": 2.468932483579156e-08, "loss": 0.1035, "step": 31750 }, { "epoch": 197.21183800623052, "grad_norm": 1.171875, "learning_rate": 2.4579977115107845e-08, "loss": 0.0988, "step": 31751 }, { "epoch": 197.21806853582555, "grad_norm": 2.5, "learning_rate": 2.447087196069675e-08, "loss": 0.1669, "step": 31752 }, { "epoch": 197.22429906542055, "grad_norm": 2.453125, "learning_rate": 2.4362009373618543e-08, "loss": 0.2053, "step": 31753 }, { "epoch": 197.23052959501558, "grad_norm": 2.3125, "learning_rate": 2.4253389354927935e-08, "loss": 0.28, "step": 31754 }, { "epoch": 197.23676012461058, "grad_norm": 1.4765625, "learning_rate": 2.4145011905682414e-08, "loss": 0.1233, "step": 31755 }, { "epoch": 197.2429906542056, "grad_norm": 2.5625, "learning_rate": 2.4036877026931137e-08, "loss": 0.2205, "step": 31756 }, { "epoch": 197.2492211838006, "grad_norm": 2.484375, "learning_rate": 2.392898471972882e-08, "loss": 0.136, "step": 31757 }, { "epoch": 197.25545171339564, "grad_norm": 2.765625, "learning_rate": 2.382133498511907e-08, "loss": 0.2628, "step": 31758 }, { "epoch": 197.26168224299064, "grad_norm": 2.015625, "learning_rate": 2.3713927824148274e-08, "loss": 0.1561, "step": 31759 }, { "epoch": 197.26791277258567, "grad_norm": 1.3671875, "learning_rate": 2.3606763237862818e-08, "loss": 0.1351, "step": 31760 }, { "epoch": 197.27414330218068, "grad_norm": 1.6171875, "learning_rate": 2.349984122729798e-08, "loss": 0.1321, "step": 31761 }, { "epoch": 197.2803738317757, "grad_norm": 1.6328125, "learning_rate": 2.3393161793497376e-08, "loss": 0.1275, "step": 31762 }, { "epoch": 197.2866043613707, "grad_norm": 1.0625, "learning_rate": 2.328672493749351e-08, "loss": 0.1098, "step": 31763 }, { "epoch": 197.29283489096574, "grad_norm": 2.234375, "learning_rate": 2.318053066032444e-08, "loss": 0.2102, "step": 31764 }, { "epoch": 197.29906542056074, "grad_norm": 1.15625, "learning_rate": 2.3074578963017123e-08, "loss": 0.1016, "step": 31765 }, { "epoch": 197.30529595015577, "grad_norm": 1.46875, "learning_rate": 2.2968869846604067e-08, "loss": 0.1079, "step": 31766 }, { "epoch": 197.31152647975077, "grad_norm": 1.4765625, "learning_rate": 2.286340331210668e-08, "loss": 0.1437, "step": 31767 }, { "epoch": 197.3177570093458, "grad_norm": 2.3125, "learning_rate": 2.2758179360557463e-08, "loss": 0.1647, "step": 31768 }, { "epoch": 197.3239875389408, "grad_norm": 2.859375, "learning_rate": 2.26531979929695e-08, "loss": 0.2728, "step": 31769 }, { "epoch": 197.33021806853583, "grad_norm": 2.203125, "learning_rate": 2.2548459210366967e-08, "loss": 0.239, "step": 31770 }, { "epoch": 197.33644859813083, "grad_norm": 1.4375, "learning_rate": 2.2443963013768498e-08, "loss": 0.1374, "step": 31771 }, { "epoch": 197.34267912772586, "grad_norm": 0.68359375, "learning_rate": 2.233970940418717e-08, "loss": 0.0781, "step": 31772 }, { "epoch": 197.34890965732086, "grad_norm": 2.25, "learning_rate": 2.223569838263606e-08, "loss": 0.1702, "step": 31773 }, { "epoch": 197.3551401869159, "grad_norm": 2.234375, "learning_rate": 2.21319299501227e-08, "loss": 0.2046, "step": 31774 }, { "epoch": 197.3613707165109, "grad_norm": 1.625, "learning_rate": 2.2028404107657385e-08, "loss": 0.1287, "step": 31775 }, { "epoch": 197.36760124610592, "grad_norm": 1.15625, "learning_rate": 2.1925120856244875e-08, "loss": 0.1116, "step": 31776 }, { "epoch": 197.37383177570092, "grad_norm": 1.3125, "learning_rate": 2.1822080196889914e-08, "loss": 0.1276, "step": 31777 }, { "epoch": 197.38006230529595, "grad_norm": 2.609375, "learning_rate": 2.1719282130591713e-08, "loss": 0.1948, "step": 31778 }, { "epoch": 197.38629283489095, "grad_norm": 2.859375, "learning_rate": 2.1616726658349462e-08, "loss": 0.2205, "step": 31779 }, { "epoch": 197.39252336448598, "grad_norm": 2.78125, "learning_rate": 2.151441378115682e-08, "loss": 0.2596, "step": 31780 }, { "epoch": 197.39875389408098, "grad_norm": 2.65625, "learning_rate": 2.1412343500010203e-08, "loss": 0.3392, "step": 31781 }, { "epoch": 197.404984423676, "grad_norm": 2.359375, "learning_rate": 2.131051581590049e-08, "loss": 0.2135, "step": 31782 }, { "epoch": 197.41121495327104, "grad_norm": 1.859375, "learning_rate": 2.120893072981578e-08, "loss": 0.1283, "step": 31783 }, { "epoch": 197.41744548286604, "grad_norm": 1.8984375, "learning_rate": 2.1107588242744168e-08, "loss": 0.1184, "step": 31784 }, { "epoch": 197.42367601246107, "grad_norm": 2.109375, "learning_rate": 2.1006488355668208e-08, "loss": 0.1769, "step": 31785 }, { "epoch": 197.42990654205607, "grad_norm": 1.5859375, "learning_rate": 2.0905631069567667e-08, "loss": 0.1595, "step": 31786 }, { "epoch": 197.4361370716511, "grad_norm": 1.4140625, "learning_rate": 2.0805016385427865e-08, "loss": 0.114, "step": 31787 }, { "epoch": 197.4423676012461, "grad_norm": 2.703125, "learning_rate": 2.070464430422303e-08, "loss": 0.2355, "step": 31788 }, { "epoch": 197.44859813084113, "grad_norm": 1.8125, "learning_rate": 2.0604514826927378e-08, "loss": 0.1322, "step": 31789 }, { "epoch": 197.45482866043614, "grad_norm": 2.328125, "learning_rate": 2.0504627954512357e-08, "loss": 0.165, "step": 31790 }, { "epoch": 197.46105919003116, "grad_norm": 1.3671875, "learning_rate": 2.0404983687952183e-08, "loss": 0.1379, "step": 31791 }, { "epoch": 197.46728971962617, "grad_norm": 0.95703125, "learning_rate": 2.0305582028209978e-08, "loss": 0.1145, "step": 31792 }, { "epoch": 197.4735202492212, "grad_norm": 1.4765625, "learning_rate": 2.0206422976257188e-08, "loss": 0.1331, "step": 31793 }, { "epoch": 197.4797507788162, "grad_norm": 2.1875, "learning_rate": 2.0107506533048602e-08, "loss": 0.1675, "step": 31794 }, { "epoch": 197.48598130841123, "grad_norm": 1.484375, "learning_rate": 2.0008832699550118e-08, "loss": 0.1181, "step": 31795 }, { "epoch": 197.49221183800623, "grad_norm": 1.6640625, "learning_rate": 1.9910401476719298e-08, "loss": 0.1079, "step": 31796 }, { "epoch": 197.49844236760126, "grad_norm": 1.34375, "learning_rate": 1.9812212865513714e-08, "loss": 0.1099, "step": 31797 }, { "epoch": 197.50467289719626, "grad_norm": 1.3125, "learning_rate": 1.9714266866885377e-08, "loss": 0.1121, "step": 31798 }, { "epoch": 197.5109034267913, "grad_norm": 1.0625, "learning_rate": 1.9616563481783535e-08, "loss": 0.0938, "step": 31799 }, { "epoch": 197.5171339563863, "grad_norm": 2.15625, "learning_rate": 1.9519102711160197e-08, "loss": 0.1765, "step": 31800 }, { "epoch": 197.52336448598132, "grad_norm": 1.171875, "learning_rate": 1.9421884555961834e-08, "loss": 0.1059, "step": 31801 }, { "epoch": 197.52959501557632, "grad_norm": 2.703125, "learning_rate": 1.9324909017129355e-08, "loss": 0.1898, "step": 31802 }, { "epoch": 197.53582554517135, "grad_norm": 2.59375, "learning_rate": 1.922817609560923e-08, "loss": 0.1528, "step": 31803 }, { "epoch": 197.54205607476635, "grad_norm": 1.0859375, "learning_rate": 1.9131685792336818e-08, "loss": 0.0882, "step": 31804 }, { "epoch": 197.54828660436138, "grad_norm": 1.6796875, "learning_rate": 1.9035438108253033e-08, "loss": 0.1092, "step": 31805 }, { "epoch": 197.55451713395638, "grad_norm": 2.015625, "learning_rate": 1.8939433044290467e-08, "loss": 0.1585, "step": 31806 }, { "epoch": 197.5607476635514, "grad_norm": 1.9921875, "learning_rate": 1.8843670601378927e-08, "loss": 0.1983, "step": 31807 }, { "epoch": 197.5669781931464, "grad_norm": 3.125, "learning_rate": 1.8748150780453777e-08, "loss": 0.2749, "step": 31808 }, { "epoch": 197.57320872274144, "grad_norm": 1.9140625, "learning_rate": 1.8652873582439278e-08, "loss": 0.1638, "step": 31809 }, { "epoch": 197.57943925233644, "grad_norm": 1.578125, "learning_rate": 1.8557839008262467e-08, "loss": 0.1194, "step": 31810 }, { "epoch": 197.58566978193147, "grad_norm": 1.125, "learning_rate": 1.8463047058844828e-08, "loss": 0.1068, "step": 31811 }, { "epoch": 197.59190031152647, "grad_norm": 1.25, "learning_rate": 1.836849773510785e-08, "loss": 0.1198, "step": 31812 }, { "epoch": 197.5981308411215, "grad_norm": 1.84375, "learning_rate": 1.8274191037970234e-08, "loss": 0.2263, "step": 31813 }, { "epoch": 197.6043613707165, "grad_norm": 2.34375, "learning_rate": 1.8180126968347923e-08, "loss": 0.1607, "step": 31814 }, { "epoch": 197.61059190031153, "grad_norm": 0.9296875, "learning_rate": 1.8086305527154067e-08, "loss": 0.0853, "step": 31815 }, { "epoch": 197.61682242990653, "grad_norm": 2.625, "learning_rate": 1.7992726715299058e-08, "loss": 0.2294, "step": 31816 }, { "epoch": 197.62305295950156, "grad_norm": 0.8828125, "learning_rate": 1.7899390533690498e-08, "loss": 0.1049, "step": 31817 }, { "epoch": 197.62928348909657, "grad_norm": 1.59375, "learning_rate": 1.780629698323877e-08, "loss": 0.1293, "step": 31818 }, { "epoch": 197.6355140186916, "grad_norm": 1.921875, "learning_rate": 1.771344606484593e-08, "loss": 0.1634, "step": 31819 }, { "epoch": 197.6417445482866, "grad_norm": 1.65625, "learning_rate": 1.7620837779414033e-08, "loss": 0.1842, "step": 31820 }, { "epoch": 197.64797507788163, "grad_norm": 1.578125, "learning_rate": 1.7528472127845143e-08, "loss": 0.1076, "step": 31821 }, { "epoch": 197.65420560747663, "grad_norm": 1.5859375, "learning_rate": 1.7436349111030204e-08, "loss": 0.1602, "step": 31822 }, { "epoch": 197.66043613707166, "grad_norm": 3.671875, "learning_rate": 1.73444687298685e-08, "loss": 0.1425, "step": 31823 }, { "epoch": 197.66666666666666, "grad_norm": 0.9453125, "learning_rate": 1.725283098525099e-08, "loss": 0.0973, "step": 31824 }, { "epoch": 197.6728971962617, "grad_norm": 2.90625, "learning_rate": 1.7161435878068622e-08, "loss": 0.2523, "step": 31825 }, { "epoch": 197.6791277258567, "grad_norm": 2.328125, "learning_rate": 1.70702834092068e-08, "loss": 0.1449, "step": 31826 }, { "epoch": 197.68535825545172, "grad_norm": 2.390625, "learning_rate": 1.6979373579553703e-08, "loss": 0.1883, "step": 31827 }, { "epoch": 197.69158878504672, "grad_norm": 1.8828125, "learning_rate": 1.6888706389989185e-08, "loss": 0.1416, "step": 31828 }, { "epoch": 197.69781931464175, "grad_norm": 2.484375, "learning_rate": 1.6798281841398646e-08, "loss": 0.1962, "step": 31829 }, { "epoch": 197.70404984423675, "grad_norm": 2.140625, "learning_rate": 1.670809993465361e-08, "loss": 0.1918, "step": 31830 }, { "epoch": 197.71028037383178, "grad_norm": 2.46875, "learning_rate": 1.6618160670636706e-08, "loss": 0.2249, "step": 31831 }, { "epoch": 197.71651090342678, "grad_norm": 2.53125, "learning_rate": 1.6528464050216686e-08, "loss": 0.1582, "step": 31832 }, { "epoch": 197.7227414330218, "grad_norm": 2.0625, "learning_rate": 1.6439010074265072e-08, "loss": 0.125, "step": 31833 }, { "epoch": 197.7289719626168, "grad_norm": 1.6875, "learning_rate": 1.6349798743656165e-08, "loss": 0.1019, "step": 31834 }, { "epoch": 197.73520249221184, "grad_norm": 1.859375, "learning_rate": 1.626083005924761e-08, "loss": 0.1369, "step": 31835 }, { "epoch": 197.74143302180684, "grad_norm": 1.9609375, "learning_rate": 1.617210402191094e-08, "loss": 0.1884, "step": 31836 }, { "epoch": 197.74766355140187, "grad_norm": 2.15625, "learning_rate": 1.6083620632503793e-08, "loss": 0.1482, "step": 31837 }, { "epoch": 197.75389408099687, "grad_norm": 1.7734375, "learning_rate": 1.5995379891886598e-08, "loss": 0.1351, "step": 31838 }, { "epoch": 197.7601246105919, "grad_norm": 3.0625, "learning_rate": 1.5907381800917e-08, "loss": 0.2923, "step": 31839 }, { "epoch": 197.7663551401869, "grad_norm": 2.546875, "learning_rate": 1.5819626360447094e-08, "loss": 0.1697, "step": 31840 }, { "epoch": 197.77258566978193, "grad_norm": 3.046875, "learning_rate": 1.5732113571334527e-08, "loss": 0.2713, "step": 31841 }, { "epoch": 197.77881619937693, "grad_norm": 2.765625, "learning_rate": 1.5644843434423073e-08, "loss": 0.2757, "step": 31842 }, { "epoch": 197.78504672897196, "grad_norm": 2.140625, "learning_rate": 1.5557815950564825e-08, "loss": 0.1871, "step": 31843 }, { "epoch": 197.79127725856696, "grad_norm": 1.2421875, "learning_rate": 1.5471031120603552e-08, "loss": 0.1053, "step": 31844 }, { "epoch": 197.797507788162, "grad_norm": 2.390625, "learning_rate": 1.538448894538025e-08, "loss": 0.242, "step": 31845 }, { "epoch": 197.80373831775702, "grad_norm": 1.6328125, "learning_rate": 1.5298189425738684e-08, "loss": 0.1242, "step": 31846 }, { "epoch": 197.80996884735202, "grad_norm": 2.703125, "learning_rate": 1.5212132562514302e-08, "loss": 0.2822, "step": 31847 }, { "epoch": 197.81619937694705, "grad_norm": 2.8125, "learning_rate": 1.5126318356545322e-08, "loss": 0.2336, "step": 31848 }, { "epoch": 197.82242990654206, "grad_norm": 2.75, "learning_rate": 1.5040746808664407e-08, "loss": 0.2864, "step": 31849 }, { "epoch": 197.82866043613708, "grad_norm": 1.3046875, "learning_rate": 1.4955417919701453e-08, "loss": 0.115, "step": 31850 }, { "epoch": 197.8348909657321, "grad_norm": 1.2578125, "learning_rate": 1.4870331690483575e-08, "loss": 0.0963, "step": 31851 }, { "epoch": 197.84112149532712, "grad_norm": 1.2734375, "learning_rate": 1.4785488121843438e-08, "loss": 0.0992, "step": 31852 }, { "epoch": 197.84735202492212, "grad_norm": 2.171875, "learning_rate": 1.4700887214597059e-08, "loss": 0.1774, "step": 31853 }, { "epoch": 197.85358255451715, "grad_norm": 2.515625, "learning_rate": 1.4616528969574328e-08, "loss": 0.2506, "step": 31854 }, { "epoch": 197.85981308411215, "grad_norm": 1.765625, "learning_rate": 1.4532413387588484e-08, "loss": 0.2843, "step": 31855 }, { "epoch": 197.86604361370718, "grad_norm": 1.6171875, "learning_rate": 1.4448540469458316e-08, "loss": 0.1164, "step": 31856 }, { "epoch": 197.87227414330218, "grad_norm": 0.9609375, "learning_rate": 1.4364910215997062e-08, "loss": 0.0899, "step": 31857 }, { "epoch": 197.8785046728972, "grad_norm": 2.265625, "learning_rate": 1.4281522628017962e-08, "loss": 0.2122, "step": 31858 }, { "epoch": 197.8847352024922, "grad_norm": 1.296875, "learning_rate": 1.4198377706334253e-08, "loss": 0.1024, "step": 31859 }, { "epoch": 197.89096573208724, "grad_norm": 0.98828125, "learning_rate": 1.4115475451748072e-08, "loss": 0.1083, "step": 31860 }, { "epoch": 197.89719626168224, "grad_norm": 0.80078125, "learning_rate": 1.4032815865067105e-08, "loss": 0.094, "step": 31861 }, { "epoch": 197.90342679127727, "grad_norm": 2.90625, "learning_rate": 1.3950398947096267e-08, "loss": 0.3459, "step": 31862 }, { "epoch": 197.90965732087227, "grad_norm": 2.09375, "learning_rate": 1.3868224698632137e-08, "loss": 0.1065, "step": 31863 }, { "epoch": 197.9158878504673, "grad_norm": 1.203125, "learning_rate": 1.3786293120474081e-08, "loss": 0.1023, "step": 31864 }, { "epoch": 197.9221183800623, "grad_norm": 1.2890625, "learning_rate": 1.3704604213418682e-08, "loss": 0.1259, "step": 31865 }, { "epoch": 197.92834890965733, "grad_norm": 1.6484375, "learning_rate": 1.3623157978256973e-08, "loss": 0.1368, "step": 31866 }, { "epoch": 197.93457943925233, "grad_norm": 1.2265625, "learning_rate": 1.3541954415785541e-08, "loss": 0.1027, "step": 31867 }, { "epoch": 197.94080996884736, "grad_norm": 1.2109375, "learning_rate": 1.3460993526787091e-08, "loss": 0.128, "step": 31868 }, { "epoch": 197.94704049844236, "grad_norm": 0.9921875, "learning_rate": 1.3380275312049884e-08, "loss": 0.0996, "step": 31869 }, { "epoch": 197.9532710280374, "grad_norm": 0.9921875, "learning_rate": 1.32997997723594e-08, "loss": 0.1041, "step": 31870 }, { "epoch": 197.9595015576324, "grad_norm": 2.5, "learning_rate": 1.3219566908495573e-08, "loss": 0.2122, "step": 31871 }, { "epoch": 197.96573208722742, "grad_norm": 1.765625, "learning_rate": 1.3139576721238333e-08, "loss": 0.1725, "step": 31872 }, { "epoch": 197.97196261682242, "grad_norm": 1.7734375, "learning_rate": 1.3059829211362063e-08, "loss": 0.133, "step": 31873 }, { "epoch": 197.97819314641745, "grad_norm": 1.078125, "learning_rate": 1.2980324379646692e-08, "loss": 0.1069, "step": 31874 }, { "epoch": 197.98442367601245, "grad_norm": 2.640625, "learning_rate": 1.2901062226858274e-08, "loss": 0.238, "step": 31875 }, { "epoch": 197.99065420560748, "grad_norm": 2.265625, "learning_rate": 1.2822042753771191e-08, "loss": 0.1706, "step": 31876 }, { "epoch": 197.99688473520249, "grad_norm": 2.40625, "learning_rate": 1.2743265961151496e-08, "loss": 0.2205, "step": 31877 }, { "epoch": 198.0, "grad_norm": 1.9921875, "learning_rate": 1.2664731849765244e-08, "loss": 0.1435, "step": 31878 }, { "epoch": 198.00623052959503, "grad_norm": 2.28125, "learning_rate": 1.2586440420372936e-08, "loss": 0.2053, "step": 31879 }, { "epoch": 198.01246105919003, "grad_norm": 3.171875, "learning_rate": 1.2508391673735076e-08, "loss": 0.175, "step": 31880 }, { "epoch": 198.01869158878506, "grad_norm": 2.515625, "learning_rate": 1.2430585610609391e-08, "loss": 0.1563, "step": 31881 }, { "epoch": 198.02492211838006, "grad_norm": 1.5859375, "learning_rate": 1.2353022231756384e-08, "loss": 0.2315, "step": 31882 }, { "epoch": 198.0311526479751, "grad_norm": 2.03125, "learning_rate": 1.2275701537922679e-08, "loss": 0.1263, "step": 31883 }, { "epoch": 198.0373831775701, "grad_norm": 3.390625, "learning_rate": 1.2198623529863228e-08, "loss": 0.3418, "step": 31884 }, { "epoch": 198.04361370716512, "grad_norm": 1.1015625, "learning_rate": 1.2121788208327434e-08, "loss": 0.1196, "step": 31885 }, { "epoch": 198.04984423676012, "grad_norm": 1.0234375, "learning_rate": 1.2045195574056368e-08, "loss": 0.0958, "step": 31886 }, { "epoch": 198.05607476635515, "grad_norm": 1.171875, "learning_rate": 1.196884562779943e-08, "loss": 0.1214, "step": 31887 }, { "epoch": 198.06230529595015, "grad_norm": 2.625, "learning_rate": 1.1892738370297697e-08, "loss": 0.1454, "step": 31888 }, { "epoch": 198.06853582554518, "grad_norm": 1.90625, "learning_rate": 1.1816873802286688e-08, "loss": 0.1774, "step": 31889 }, { "epoch": 198.07476635514018, "grad_norm": 1.25, "learning_rate": 1.1741251924504704e-08, "loss": 0.1019, "step": 31890 }, { "epoch": 198.0809968847352, "grad_norm": 3.421875, "learning_rate": 1.1665872737687266e-08, "loss": 0.3705, "step": 31891 }, { "epoch": 198.08722741433021, "grad_norm": 2.625, "learning_rate": 1.1590736242567124e-08, "loss": 0.2025, "step": 31892 }, { "epoch": 198.09345794392524, "grad_norm": 1.7109375, "learning_rate": 1.1515842439871472e-08, "loss": 0.1012, "step": 31893 }, { "epoch": 198.09968847352025, "grad_norm": 2.3125, "learning_rate": 1.144119133033028e-08, "loss": 0.1934, "step": 31894 }, { "epoch": 198.10591900311528, "grad_norm": 2.71875, "learning_rate": 1.1366782914665198e-08, "loss": 0.2173, "step": 31895 }, { "epoch": 198.11214953271028, "grad_norm": 2.578125, "learning_rate": 1.1292617193603417e-08, "loss": 0.1043, "step": 31896 }, { "epoch": 198.1183800623053, "grad_norm": 1.5703125, "learning_rate": 1.1218694167861032e-08, "loss": 0.1244, "step": 31897 }, { "epoch": 198.1246105919003, "grad_norm": 2.375, "learning_rate": 1.114501383815969e-08, "loss": 0.2041, "step": 31898 }, { "epoch": 198.13084112149534, "grad_norm": 2.296875, "learning_rate": 1.1071576205212707e-08, "loss": 0.1822, "step": 31899 }, { "epoch": 198.13707165109034, "grad_norm": 1.6953125, "learning_rate": 1.0998381269733405e-08, "loss": 0.1489, "step": 31900 }, { "epoch": 198.14330218068537, "grad_norm": 2.71875, "learning_rate": 1.0925429032432322e-08, "loss": 0.2228, "step": 31901 }, { "epoch": 198.14953271028037, "grad_norm": 1.4296875, "learning_rate": 1.0852719494020003e-08, "loss": 0.1358, "step": 31902 }, { "epoch": 198.1557632398754, "grad_norm": 2.078125, "learning_rate": 1.0780252655198665e-08, "loss": 0.1745, "step": 31903 }, { "epoch": 198.1619937694704, "grad_norm": 3.21875, "learning_rate": 1.0708028516676072e-08, "loss": 0.3555, "step": 31904 }, { "epoch": 198.16822429906543, "grad_norm": 1.359375, "learning_rate": 1.0636047079154443e-08, "loss": 0.1292, "step": 31905 }, { "epoch": 198.17445482866043, "grad_norm": 3.09375, "learning_rate": 1.0564308343327666e-08, "loss": 0.3062, "step": 31906 }, { "epoch": 198.18068535825546, "grad_norm": 1.15625, "learning_rate": 1.0492812309897959e-08, "loss": 0.1357, "step": 31907 }, { "epoch": 198.18691588785046, "grad_norm": 1.6796875, "learning_rate": 1.0421558979556434e-08, "loss": 0.1403, "step": 31908 }, { "epoch": 198.1931464174455, "grad_norm": 1.4921875, "learning_rate": 1.035054835299698e-08, "loss": 0.127, "step": 31909 }, { "epoch": 198.1993769470405, "grad_norm": 2.359375, "learning_rate": 1.0279780430907937e-08, "loss": 0.2572, "step": 31910 }, { "epoch": 198.20560747663552, "grad_norm": 2.34375, "learning_rate": 1.0209255213977643e-08, "loss": 0.1743, "step": 31911 }, { "epoch": 198.21183800623052, "grad_norm": 2.15625, "learning_rate": 1.0138972702888882e-08, "loss": 0.1963, "step": 31912 }, { "epoch": 198.21806853582555, "grad_norm": 2.328125, "learning_rate": 1.0068932898324446e-08, "loss": 0.2808, "step": 31913 }, { "epoch": 198.22429906542055, "grad_norm": 2.28125, "learning_rate": 9.999135800969895e-09, "loss": 0.1051, "step": 31914 }, { "epoch": 198.23052959501558, "grad_norm": 1.9609375, "learning_rate": 9.929581411496913e-09, "loss": 0.1629, "step": 31915 }, { "epoch": 198.23676012461058, "grad_norm": 2.25, "learning_rate": 9.860269730582739e-09, "loss": 0.133, "step": 31916 }, { "epoch": 198.2429906542056, "grad_norm": 1.8515625, "learning_rate": 9.79120075890183e-09, "loss": 0.1713, "step": 31917 }, { "epoch": 198.2492211838006, "grad_norm": 1.3203125, "learning_rate": 9.722374497123098e-09, "loss": 0.0946, "step": 31918 }, { "epoch": 198.25545171339564, "grad_norm": 1.859375, "learning_rate": 9.653790945918228e-09, "loss": 0.1436, "step": 31919 }, { "epoch": 198.26168224299064, "grad_norm": 2.234375, "learning_rate": 9.585450105950578e-09, "loss": 0.1678, "step": 31920 }, { "epoch": 198.26791277258567, "grad_norm": 2.75, "learning_rate": 9.517351977883505e-09, "loss": 0.2446, "step": 31921 }, { "epoch": 198.27414330218068, "grad_norm": 2.1875, "learning_rate": 9.449496562380367e-09, "loss": 0.199, "step": 31922 }, { "epoch": 198.2803738317757, "grad_norm": 2.34375, "learning_rate": 9.381883860098971e-09, "loss": 0.2133, "step": 31923 }, { "epoch": 198.2866043613707, "grad_norm": 1.21875, "learning_rate": 9.314513871694353e-09, "loss": 0.1155, "step": 31924 }, { "epoch": 198.29283489096574, "grad_norm": 1.6796875, "learning_rate": 9.247386597824314e-09, "loss": 0.1565, "step": 31925 }, { "epoch": 198.29906542056074, "grad_norm": 1.9609375, "learning_rate": 9.180502039141114e-09, "loss": 0.2549, "step": 31926 }, { "epoch": 198.30529595015577, "grad_norm": 1.2890625, "learning_rate": 9.113860196288682e-09, "loss": 0.1032, "step": 31927 }, { "epoch": 198.31152647975077, "grad_norm": 2.71875, "learning_rate": 9.047461069922048e-09, "loss": 0.1508, "step": 31928 }, { "epoch": 198.3177570093458, "grad_norm": 2.6875, "learning_rate": 8.98130466067959e-09, "loss": 0.1118, "step": 31929 }, { "epoch": 198.3239875389408, "grad_norm": 2.21875, "learning_rate": 8.91539096920524e-09, "loss": 0.1809, "step": 31930 }, { "epoch": 198.33021806853583, "grad_norm": 1.59375, "learning_rate": 8.849719996142925e-09, "loss": 0.1199, "step": 31931 }, { "epoch": 198.33644859813083, "grad_norm": 3.078125, "learning_rate": 8.784291742125473e-09, "loss": 0.1975, "step": 31932 }, { "epoch": 198.34267912772586, "grad_norm": 3.0, "learning_rate": 8.719106207794036e-09, "loss": 0.1947, "step": 31933 }, { "epoch": 198.34890965732086, "grad_norm": 1.9375, "learning_rate": 8.654163393775894e-09, "loss": 0.1521, "step": 31934 }, { "epoch": 198.3551401869159, "grad_norm": 1.6328125, "learning_rate": 8.58946330070387e-09, "loss": 0.1328, "step": 31935 }, { "epoch": 198.3613707165109, "grad_norm": 3.171875, "learning_rate": 8.525005929205243e-09, "loss": 0.2743, "step": 31936 }, { "epoch": 198.36760124610592, "grad_norm": 2.59375, "learning_rate": 8.460791279910063e-09, "loss": 0.183, "step": 31937 }, { "epoch": 198.37383177570092, "grad_norm": 1.1171875, "learning_rate": 8.396819353440056e-09, "loss": 0.1311, "step": 31938 }, { "epoch": 198.38006230529595, "grad_norm": 0.8203125, "learning_rate": 8.333090150414168e-09, "loss": 0.0923, "step": 31939 }, { "epoch": 198.38629283489095, "grad_norm": 2.8125, "learning_rate": 8.269603671451353e-09, "loss": 0.2086, "step": 31940 }, { "epoch": 198.39252336448598, "grad_norm": 2.5, "learning_rate": 8.206359917173335e-09, "loss": 0.2545, "step": 31941 }, { "epoch": 198.39875389408098, "grad_norm": 2.765625, "learning_rate": 8.143358888187958e-09, "loss": 0.2193, "step": 31942 }, { "epoch": 198.404984423676, "grad_norm": 2.5625, "learning_rate": 8.080600585111397e-09, "loss": 0.1906, "step": 31943 }, { "epoch": 198.41121495327104, "grad_norm": 1.5703125, "learning_rate": 8.018085008554277e-09, "loss": 0.2376, "step": 31944 }, { "epoch": 198.41744548286604, "grad_norm": 2.375, "learning_rate": 7.955812159118891e-09, "loss": 0.1501, "step": 31945 }, { "epoch": 198.42367601246107, "grad_norm": 2.34375, "learning_rate": 7.893782037410313e-09, "loss": 0.2768, "step": 31946 }, { "epoch": 198.42990654205607, "grad_norm": 1.65625, "learning_rate": 7.83199464403639e-09, "loss": 0.1242, "step": 31947 }, { "epoch": 198.4361370716511, "grad_norm": 3.96875, "learning_rate": 7.770449979593864e-09, "loss": 0.4207, "step": 31948 }, { "epoch": 198.4423676012461, "grad_norm": 2.78125, "learning_rate": 7.709148044679481e-09, "loss": 0.2456, "step": 31949 }, { "epoch": 198.44859813084113, "grad_norm": 3.453125, "learning_rate": 7.648088839889988e-09, "loss": 0.2058, "step": 31950 }, { "epoch": 198.45482866043614, "grad_norm": 3.25, "learning_rate": 7.587272365816577e-09, "loss": 0.2569, "step": 31951 }, { "epoch": 198.46105919003116, "grad_norm": 3.046875, "learning_rate": 7.52669862305322e-09, "loss": 0.1859, "step": 31952 }, { "epoch": 198.46728971962617, "grad_norm": 2.0625, "learning_rate": 7.466367612185553e-09, "loss": 0.1765, "step": 31953 }, { "epoch": 198.4735202492212, "grad_norm": 2.765625, "learning_rate": 7.406279333802002e-09, "loss": 0.201, "step": 31954 }, { "epoch": 198.4797507788162, "grad_norm": 3.109375, "learning_rate": 7.346433788482654e-09, "loss": 0.2035, "step": 31955 }, { "epoch": 198.48598130841123, "grad_norm": 2.078125, "learning_rate": 7.286830976813152e-09, "loss": 0.1847, "step": 31956 }, { "epoch": 198.49221183800623, "grad_norm": 2.453125, "learning_rate": 7.2274708993680385e-09, "loss": 0.1804, "step": 31957 }, { "epoch": 198.49844236760126, "grad_norm": 1.4296875, "learning_rate": 7.168353556724627e-09, "loss": 0.1185, "step": 31958 }, { "epoch": 198.50467289719626, "grad_norm": 1.078125, "learning_rate": 7.1094789494602355e-09, "loss": 0.1399, "step": 31959 }, { "epoch": 198.5109034267913, "grad_norm": 3.546875, "learning_rate": 7.0508470781438515e-09, "loss": 0.2886, "step": 31960 }, { "epoch": 198.5171339563863, "grad_norm": 1.0, "learning_rate": 6.992457943347241e-09, "loss": 0.0935, "step": 31961 }, { "epoch": 198.52336448598132, "grad_norm": 1.3984375, "learning_rate": 6.9343115456338424e-09, "loss": 0.112, "step": 31962 }, { "epoch": 198.52959501557632, "grad_norm": 1.1328125, "learning_rate": 6.876407885569869e-09, "loss": 0.1198, "step": 31963 }, { "epoch": 198.53582554517135, "grad_norm": 1.8046875, "learning_rate": 6.818746963718758e-09, "loss": 0.1796, "step": 31964 }, { "epoch": 198.54205607476635, "grad_norm": 1.90625, "learning_rate": 6.761328780641174e-09, "loss": 0.1407, "step": 31965 }, { "epoch": 198.54828660436138, "grad_norm": 1.78125, "learning_rate": 6.704153336892227e-09, "loss": 0.1404, "step": 31966 }, { "epoch": 198.55451713395638, "grad_norm": 1.5859375, "learning_rate": 6.647220633029805e-09, "loss": 0.0955, "step": 31967 }, { "epoch": 198.5607476635514, "grad_norm": 1.4765625, "learning_rate": 6.590530669606243e-09, "loss": 0.0857, "step": 31968 }, { "epoch": 198.5669781931464, "grad_norm": 1.1875, "learning_rate": 6.534083447168327e-09, "loss": 0.0987, "step": 31969 }, { "epoch": 198.57320872274144, "grad_norm": 2.421875, "learning_rate": 6.477878966271167e-09, "loss": 0.2325, "step": 31970 }, { "epoch": 198.57943925233644, "grad_norm": 1.765625, "learning_rate": 6.421917227455998e-09, "loss": 0.1532, "step": 31971 }, { "epoch": 198.58566978193147, "grad_norm": 0.98046875, "learning_rate": 6.366198231266829e-09, "loss": 0.0979, "step": 31972 }, { "epoch": 198.59190031152647, "grad_norm": 1.765625, "learning_rate": 6.3107219782448936e-09, "loss": 0.125, "step": 31973 }, { "epoch": 198.5981308411215, "grad_norm": 1.984375, "learning_rate": 6.25548846892865e-09, "loss": 0.221, "step": 31974 }, { "epoch": 198.6043613707165, "grad_norm": 1.875, "learning_rate": 6.200497703856556e-09, "loss": 0.2011, "step": 31975 }, { "epoch": 198.61059190031153, "grad_norm": 2.09375, "learning_rate": 6.145749683561519e-09, "loss": 0.144, "step": 31976 }, { "epoch": 198.61682242990653, "grad_norm": 3.8125, "learning_rate": 6.091244408573671e-09, "loss": 0.2968, "step": 31977 }, { "epoch": 198.62305295950156, "grad_norm": 1.9921875, "learning_rate": 6.036981879425918e-09, "loss": 0.1109, "step": 31978 }, { "epoch": 198.62928348909657, "grad_norm": 2.375, "learning_rate": 5.982962096640066e-09, "loss": 0.2551, "step": 31979 }, { "epoch": 198.6355140186916, "grad_norm": 1.421875, "learning_rate": 5.929185060746245e-09, "loss": 0.1244, "step": 31980 }, { "epoch": 198.6417445482866, "grad_norm": 1.3984375, "learning_rate": 5.8756507722634854e-09, "loss": 0.1157, "step": 31981 }, { "epoch": 198.64797507788163, "grad_norm": 1.640625, "learning_rate": 5.822359231713592e-09, "loss": 0.1376, "step": 31982 }, { "epoch": 198.65420560747663, "grad_norm": 3.375, "learning_rate": 5.769310439612818e-09, "loss": 0.1211, "step": 31983 }, { "epoch": 198.66043613707166, "grad_norm": 2.546875, "learning_rate": 5.7165043964774175e-09, "loss": 0.1932, "step": 31984 }, { "epoch": 198.66666666666666, "grad_norm": 2.390625, "learning_rate": 5.663941102818093e-09, "loss": 0.2064, "step": 31985 }, { "epoch": 198.6728971962617, "grad_norm": 2.53125, "learning_rate": 5.611620559148323e-09, "loss": 0.274, "step": 31986 }, { "epoch": 198.6791277258567, "grad_norm": 1.4765625, "learning_rate": 5.559542765973258e-09, "loss": 0.1208, "step": 31987 }, { "epoch": 198.68535825545172, "grad_norm": 2.84375, "learning_rate": 5.507707723800826e-09, "loss": 0.1979, "step": 31988 }, { "epoch": 198.69158878504672, "grad_norm": 1.8203125, "learning_rate": 5.456115433136178e-09, "loss": 0.2414, "step": 31989 }, { "epoch": 198.69781931464175, "grad_norm": 1.9765625, "learning_rate": 5.404765894476138e-09, "loss": 0.1444, "step": 31990 }, { "epoch": 198.70404984423675, "grad_norm": 2.0, "learning_rate": 5.353659108320308e-09, "loss": 0.1684, "step": 31991 }, { "epoch": 198.71028037383178, "grad_norm": 1.3359375, "learning_rate": 5.302795075165512e-09, "loss": 0.1014, "step": 31992 }, { "epoch": 198.71651090342678, "grad_norm": 3.109375, "learning_rate": 5.252173795508575e-09, "loss": 0.1729, "step": 31993 }, { "epoch": 198.7227414330218, "grad_norm": 2.265625, "learning_rate": 5.201795269837995e-09, "loss": 0.1619, "step": 31994 }, { "epoch": 198.7289719626168, "grad_norm": 2.15625, "learning_rate": 5.151659498642269e-09, "loss": 0.1469, "step": 31995 }, { "epoch": 198.73520249221184, "grad_norm": 2.0, "learning_rate": 5.1017664824126736e-09, "loss": 0.1576, "step": 31996 }, { "epoch": 198.74143302180684, "grad_norm": 2.640625, "learning_rate": 5.052116221629377e-09, "loss": 0.2319, "step": 31997 }, { "epoch": 198.74766355140187, "grad_norm": 2.40625, "learning_rate": 5.002708716778104e-09, "loss": 0.1939, "step": 31998 }, { "epoch": 198.75389408099687, "grad_norm": 3.078125, "learning_rate": 4.953543968336249e-09, "loss": 0.2311, "step": 31999 }, { "epoch": 198.7601246105919, "grad_norm": 2.546875, "learning_rate": 4.90462197678121e-09, "loss": 0.2201, "step": 32000 }, { "epoch": 198.7663551401869, "grad_norm": 2.125, "learning_rate": 4.8559427425903805e-09, "loss": 0.1959, "step": 32001 }, { "epoch": 198.77258566978193, "grad_norm": 3.078125, "learning_rate": 4.807506266232831e-09, "loss": 0.1303, "step": 32002 }, { "epoch": 198.77881619937693, "grad_norm": 2.859375, "learning_rate": 4.7593125481831815e-09, "loss": 0.2379, "step": 32003 }, { "epoch": 198.78504672897196, "grad_norm": 3.1875, "learning_rate": 4.71136158890495e-09, "loss": 0.2434, "step": 32004 }, { "epoch": 198.79127725856696, "grad_norm": 3.109375, "learning_rate": 4.663653388867206e-09, "loss": 0.2491, "step": 32005 }, { "epoch": 198.797507788162, "grad_norm": 1.453125, "learning_rate": 4.616187948533468e-09, "loss": 0.1269, "step": 32006 }, { "epoch": 198.80373831775702, "grad_norm": 1.421875, "learning_rate": 4.568965268364478e-09, "loss": 0.0965, "step": 32007 }, { "epoch": 198.80996884735202, "grad_norm": 2.875, "learning_rate": 4.521985348818203e-09, "loss": 0.3372, "step": 32008 }, { "epoch": 198.81619937694705, "grad_norm": 1.140625, "learning_rate": 4.475248190349835e-09, "loss": 0.106, "step": 32009 }, { "epoch": 198.82242990654206, "grad_norm": 2.078125, "learning_rate": 4.428753793414564e-09, "loss": 0.2241, "step": 32010 }, { "epoch": 198.82866043613708, "grad_norm": 1.359375, "learning_rate": 4.382502158464808e-09, "loss": 0.1156, "step": 32011 }, { "epoch": 198.8348909657321, "grad_norm": 1.921875, "learning_rate": 4.33649328594743e-09, "loss": 0.1453, "step": 32012 }, { "epoch": 198.84112149532712, "grad_norm": 1.6796875, "learning_rate": 4.29072717631207e-09, "loss": 0.1556, "step": 32013 }, { "epoch": 198.84735202492212, "grad_norm": 1.9375, "learning_rate": 4.245203830000044e-09, "loss": 0.1587, "step": 32014 }, { "epoch": 198.85358255451715, "grad_norm": 2.109375, "learning_rate": 4.199923247455439e-09, "loss": 0.1641, "step": 32015 }, { "epoch": 198.85981308411215, "grad_norm": 1.859375, "learning_rate": 4.154885429116795e-09, "loss": 0.1999, "step": 32016 }, { "epoch": 198.86604361370718, "grad_norm": 1.6171875, "learning_rate": 4.1100903754254235e-09, "loss": 0.1337, "step": 32017 }, { "epoch": 198.87227414330218, "grad_norm": 2.359375, "learning_rate": 4.065538086811538e-09, "loss": 0.1795, "step": 32018 }, { "epoch": 198.8785046728972, "grad_norm": 1.7890625, "learning_rate": 4.021228563708124e-09, "loss": 0.1715, "step": 32019 }, { "epoch": 198.8847352024922, "grad_norm": 2.515625, "learning_rate": 3.977161806550944e-09, "loss": 0.1675, "step": 32020 }, { "epoch": 198.89096573208724, "grad_norm": 2.375, "learning_rate": 3.933337815761884e-09, "loss": 0.2871, "step": 32021 }, { "epoch": 198.89719626168224, "grad_norm": 2.859375, "learning_rate": 3.889756591768379e-09, "loss": 0.2535, "step": 32022 }, { "epoch": 198.90342679127727, "grad_norm": 1.5859375, "learning_rate": 3.846418134995089e-09, "loss": 0.129, "step": 32023 }, { "epoch": 198.90965732087227, "grad_norm": 1.65625, "learning_rate": 3.803322445861124e-09, "loss": 0.1167, "step": 32024 }, { "epoch": 198.9158878504673, "grad_norm": 1.796875, "learning_rate": 3.760469524785593e-09, "loss": 0.1064, "step": 32025 }, { "epoch": 198.9221183800623, "grad_norm": 1.5390625, "learning_rate": 3.717859372187604e-09, "loss": 0.1334, "step": 32026 }, { "epoch": 198.92834890965733, "grad_norm": 2.3125, "learning_rate": 3.6754919884779414e-09, "loss": 0.2692, "step": 32027 }, { "epoch": 198.93457943925233, "grad_norm": 1.953125, "learning_rate": 3.6333673740673868e-09, "loss": 0.1826, "step": 32028 }, { "epoch": 198.94080996884736, "grad_norm": 2.03125, "learning_rate": 3.591485529363947e-09, "loss": 0.1636, "step": 32029 }, { "epoch": 198.94704049844236, "grad_norm": 2.21875, "learning_rate": 3.5498464547811807e-09, "loss": 0.1197, "step": 32030 }, { "epoch": 198.9532710280374, "grad_norm": 2.015625, "learning_rate": 3.5084501507159917e-09, "loss": 0.111, "step": 32031 }, { "epoch": 198.9595015576324, "grad_norm": 1.71875, "learning_rate": 3.467296617573612e-09, "loss": 0.1311, "step": 32032 }, { "epoch": 198.96573208722742, "grad_norm": 1.734375, "learning_rate": 3.4263858557564977e-09, "loss": 0.1661, "step": 32033 }, { "epoch": 198.97196261682242, "grad_norm": 1.3984375, "learning_rate": 3.385717865656002e-09, "loss": 0.1299, "step": 32034 }, { "epoch": 198.97819314641745, "grad_norm": 1.7890625, "learning_rate": 3.3452926476718048e-09, "loss": 0.1566, "step": 32035 }, { "epoch": 198.98442367601245, "grad_norm": 0.9140625, "learning_rate": 3.3051102021924853e-09, "loss": 0.1099, "step": 32036 }, { "epoch": 198.99065420560748, "grad_norm": 1.203125, "learning_rate": 3.2651705296121717e-09, "loss": 0.1192, "step": 32037 }, { "epoch": 198.99688473520249, "grad_norm": 2.109375, "learning_rate": 3.2254736303166665e-09, "loss": 0.1748, "step": 32038 }, { "epoch": 199.0, "grad_norm": 2.8125, "learning_rate": 3.1860195046945486e-09, "loss": 0.1494, "step": 32039 }, { "epoch": 199.00623052959503, "grad_norm": 1.390625, "learning_rate": 3.146808153123293e-09, "loss": 0.1298, "step": 32040 }, { "epoch": 199.01246105919003, "grad_norm": 2.265625, "learning_rate": 3.1078395759887023e-09, "loss": 0.2135, "step": 32041 }, { "epoch": 199.01869158878506, "grad_norm": 2.734375, "learning_rate": 3.0691137736682528e-09, "loss": 0.2281, "step": 32042 }, { "epoch": 199.02492211838006, "grad_norm": 0.98828125, "learning_rate": 3.0306307465366445e-09, "loss": 0.1, "step": 32043 }, { "epoch": 199.0311526479751, "grad_norm": 2.078125, "learning_rate": 2.9923904949685776e-09, "loss": 0.1497, "step": 32044 }, { "epoch": 199.0373831775701, "grad_norm": 1.5625, "learning_rate": 2.9543930193332014e-09, "loss": 0.1182, "step": 32045 }, { "epoch": 199.04361370716512, "grad_norm": 3.078125, "learning_rate": 2.916638320005216e-09, "loss": 0.1995, "step": 32046 }, { "epoch": 199.04984423676012, "grad_norm": 1.5, "learning_rate": 2.879126397345444e-09, "loss": 0.1057, "step": 32047 }, { "epoch": 199.05607476635515, "grad_norm": 1.5390625, "learning_rate": 2.841857251720259e-09, "loss": 0.079, "step": 32048 }, { "epoch": 199.06230529595015, "grad_norm": 1.2578125, "learning_rate": 2.8048308834904833e-09, "loss": 0.1283, "step": 32049 }, { "epoch": 199.06853582554518, "grad_norm": 1.3671875, "learning_rate": 2.7680472930197154e-09, "loss": 0.1668, "step": 32050 }, { "epoch": 199.07476635514018, "grad_norm": 3.53125, "learning_rate": 2.7315064806604506e-09, "loss": 0.2762, "step": 32051 }, { "epoch": 199.0809968847352, "grad_norm": 1.0078125, "learning_rate": 2.6952084467707363e-09, "loss": 0.0922, "step": 32052 }, { "epoch": 199.08722741433021, "grad_norm": 1.140625, "learning_rate": 2.6591531917030675e-09, "loss": 0.1216, "step": 32053 }, { "epoch": 199.09345794392524, "grad_norm": 2.296875, "learning_rate": 2.62334071580439e-09, "loss": 0.1864, "step": 32054 }, { "epoch": 199.09968847352025, "grad_norm": 1.453125, "learning_rate": 2.5877710194271986e-09, "loss": 0.1104, "step": 32055 }, { "epoch": 199.10591900311528, "grad_norm": 2.1875, "learning_rate": 2.5524441029128876e-09, "loss": 0.2352, "step": 32056 }, { "epoch": 199.11214953271028, "grad_norm": 1.421875, "learning_rate": 2.5173599666056256e-09, "loss": 0.1509, "step": 32057 }, { "epoch": 199.1183800623053, "grad_norm": 2.609375, "learning_rate": 2.4825186108468068e-09, "loss": 0.2956, "step": 32058 }, { "epoch": 199.1246105919003, "grad_norm": 2.453125, "learning_rate": 2.447920035975049e-09, "loss": 0.1705, "step": 32059 }, { "epoch": 199.13084112149534, "grad_norm": 2.21875, "learning_rate": 2.413564242326194e-09, "loss": 0.1543, "step": 32060 }, { "epoch": 199.13707165109034, "grad_norm": 1.375, "learning_rate": 2.3794512302333096e-09, "loss": 0.1104, "step": 32061 }, { "epoch": 199.14330218068537, "grad_norm": 2.328125, "learning_rate": 2.345581000029462e-09, "loss": 0.2189, "step": 32062 }, { "epoch": 199.14953271028037, "grad_norm": 2.375, "learning_rate": 2.3119535520421674e-09, "loss": 0.171, "step": 32063 }, { "epoch": 199.1557632398754, "grad_norm": 2.09375, "learning_rate": 2.2785688865961664e-09, "loss": 0.1827, "step": 32064 }, { "epoch": 199.1619937694704, "grad_norm": 1.4140625, "learning_rate": 2.245427004018974e-09, "loss": 0.1028, "step": 32065 }, { "epoch": 199.16822429906543, "grad_norm": 2.5625, "learning_rate": 2.2125279046297797e-09, "loss": 0.1142, "step": 32066 }, { "epoch": 199.17445482866043, "grad_norm": 2.125, "learning_rate": 2.1798715887505483e-09, "loss": 0.2072, "step": 32067 }, { "epoch": 199.18068535825546, "grad_norm": 1.8125, "learning_rate": 2.147458056697693e-09, "loss": 0.1843, "step": 32068 }, { "epoch": 199.18691588785046, "grad_norm": 1.234375, "learning_rate": 2.1152873087848525e-09, "loss": 0.0766, "step": 32069 }, { "epoch": 199.1931464174455, "grad_norm": 1.3359375, "learning_rate": 2.083359345325664e-09, "loss": 0.1183, "step": 32070 }, { "epoch": 199.1993769470405, "grad_norm": 1.7109375, "learning_rate": 2.05167416663099e-09, "loss": 0.147, "step": 32071 }, { "epoch": 199.20560747663552, "grad_norm": 1.9375, "learning_rate": 2.0202317730061427e-09, "loss": 0.1755, "step": 32072 }, { "epoch": 199.21183800623052, "grad_norm": 1.6875, "learning_rate": 1.989032164759208e-09, "loss": 0.2252, "step": 32073 }, { "epoch": 199.21806853582555, "grad_norm": 1.6875, "learning_rate": 1.958075342189947e-09, "loss": 0.133, "step": 32074 }, { "epoch": 199.22429906542055, "grad_norm": 2.171875, "learning_rate": 1.9273613056008944e-09, "loss": 0.1418, "step": 32075 }, { "epoch": 199.23052959501558, "grad_norm": 2.84375, "learning_rate": 1.896890055291811e-09, "loss": 0.2051, "step": 32076 }, { "epoch": 199.23676012461058, "grad_norm": 1.515625, "learning_rate": 1.8666615915569062e-09, "loss": 0.099, "step": 32077 }, { "epoch": 199.2429906542056, "grad_norm": 2.21875, "learning_rate": 1.836675914690389e-09, "loss": 0.195, "step": 32078 }, { "epoch": 199.2492211838006, "grad_norm": 2.4375, "learning_rate": 1.806933024980917e-09, "loss": 0.1539, "step": 32079 }, { "epoch": 199.25545171339564, "grad_norm": 1.578125, "learning_rate": 1.7774329227199238e-09, "loss": 0.1397, "step": 32080 }, { "epoch": 199.26168224299064, "grad_norm": 1.953125, "learning_rate": 1.7481756081960677e-09, "loss": 0.1623, "step": 32081 }, { "epoch": 199.26791277258567, "grad_norm": 2.421875, "learning_rate": 1.7191610816896797e-09, "loss": 0.1783, "step": 32082 }, { "epoch": 199.27414330218068, "grad_norm": 1.46875, "learning_rate": 1.690389343483867e-09, "loss": 0.1035, "step": 32083 }, { "epoch": 199.2803738317757, "grad_norm": 1.15625, "learning_rate": 1.6618603938561849e-09, "loss": 0.1076, "step": 32084 }, { "epoch": 199.2866043613707, "grad_norm": 2.40625, "learning_rate": 1.6335742330869653e-09, "loss": 0.1456, "step": 32085 }, { "epoch": 199.29283489096574, "grad_norm": 2.140625, "learning_rate": 1.605530861450988e-09, "loss": 0.2466, "step": 32086 }, { "epoch": 199.29906542056074, "grad_norm": 1.1953125, "learning_rate": 1.5777302792174819e-09, "loss": 0.1291, "step": 32087 }, { "epoch": 199.30529595015577, "grad_norm": 3.0, "learning_rate": 1.550172486658452e-09, "loss": 0.2217, "step": 32088 }, { "epoch": 199.31152647975077, "grad_norm": 3.421875, "learning_rate": 1.5228574840431275e-09, "loss": 0.2636, "step": 32089 }, { "epoch": 199.3177570093458, "grad_norm": 2.6875, "learning_rate": 1.4957852716324106e-09, "loss": 0.1433, "step": 32090 }, { "epoch": 199.3239875389408, "grad_norm": 2.53125, "learning_rate": 1.4689558496927547e-09, "loss": 0.187, "step": 32091 }, { "epoch": 199.33021806853583, "grad_norm": 2.28125, "learning_rate": 1.4423692184822868e-09, "loss": 0.1615, "step": 32092 }, { "epoch": 199.33644859813083, "grad_norm": 3.109375, "learning_rate": 1.4160253782619093e-09, "loss": 0.2166, "step": 32093 }, { "epoch": 199.34267912772586, "grad_norm": 1.6015625, "learning_rate": 1.3899243292869736e-09, "loss": 0.1341, "step": 32094 }, { "epoch": 199.34890965732086, "grad_norm": 1.6796875, "learning_rate": 1.3640660718072795e-09, "loss": 0.1156, "step": 32095 }, { "epoch": 199.3551401869159, "grad_norm": 1.1796875, "learning_rate": 1.3384506060809542e-09, "loss": 0.109, "step": 32096 }, { "epoch": 199.3613707165109, "grad_norm": 1.953125, "learning_rate": 1.3130779323494713e-09, "loss": 0.1652, "step": 32097 }, { "epoch": 199.36760124610592, "grad_norm": 1.5390625, "learning_rate": 1.2879480508654062e-09, "loss": 0.1281, "step": 32098 }, { "epoch": 199.37383177570092, "grad_norm": 2.125, "learning_rate": 1.2630609618674571e-09, "loss": 0.1405, "step": 32099 }, { "epoch": 199.38006230529595, "grad_norm": 2.34375, "learning_rate": 1.2384166655998731e-09, "loss": 0.2108, "step": 32100 }, { "epoch": 199.38629283489095, "grad_norm": 1.7265625, "learning_rate": 1.2140151623041274e-09, "loss": 0.1058, "step": 32101 }, { "epoch": 199.39252336448598, "grad_norm": 1.3671875, "learning_rate": 1.189856452213367e-09, "loss": 0.1362, "step": 32102 }, { "epoch": 199.39875389408098, "grad_norm": 1.9609375, "learning_rate": 1.1659405355635145e-09, "loss": 0.1867, "step": 32103 }, { "epoch": 199.404984423676, "grad_norm": 2.046875, "learning_rate": 1.1422674125877163e-09, "loss": 0.1332, "step": 32104 }, { "epoch": 199.41121495327104, "grad_norm": 1.7109375, "learning_rate": 1.1188370835163442e-09, "loss": 0.137, "step": 32105 }, { "epoch": 199.41744548286604, "grad_norm": 1.609375, "learning_rate": 1.095649548574218e-09, "loss": 0.1368, "step": 32106 }, { "epoch": 199.42367601246107, "grad_norm": 2.640625, "learning_rate": 1.0727048079889334e-09, "loss": 0.1929, "step": 32107 }, { "epoch": 199.42990654205607, "grad_norm": 1.640625, "learning_rate": 1.0500028619853108e-09, "loss": 0.1171, "step": 32108 }, { "epoch": 199.4361370716511, "grad_norm": 1.90625, "learning_rate": 1.027543710777068e-09, "loss": 0.1034, "step": 32109 }, { "epoch": 199.4423676012461, "grad_norm": 2.0625, "learning_rate": 1.0053273545890252e-09, "loss": 0.1706, "step": 32110 }, { "epoch": 199.44859813084113, "grad_norm": 3.25, "learning_rate": 9.833537936349003e-10, "loss": 0.1814, "step": 32111 }, { "epoch": 199.45482866043614, "grad_norm": 2.09375, "learning_rate": 9.616230281256355e-10, "loss": 0.1621, "step": 32112 }, { "epoch": 199.46105919003116, "grad_norm": 1.5078125, "learning_rate": 9.40135058274949e-10, "loss": 0.1646, "step": 32113 }, { "epoch": 199.46728971962617, "grad_norm": 2.046875, "learning_rate": 9.188898842910076e-10, "loss": 0.1667, "step": 32114 }, { "epoch": 199.4735202492212, "grad_norm": 0.8046875, "learning_rate": 8.97887506381978e-10, "loss": 0.0978, "step": 32115 }, { "epoch": 199.4797507788162, "grad_norm": 2.96875, "learning_rate": 8.771279247477004e-10, "loss": 0.1894, "step": 32116 }, { "epoch": 199.48598130841123, "grad_norm": 2.0625, "learning_rate": 8.566111395907906e-10, "loss": 0.1749, "step": 32117 }, { "epoch": 199.49221183800623, "grad_norm": 2.421875, "learning_rate": 8.36337151113864e-10, "loss": 0.2031, "step": 32118 }, { "epoch": 199.49844236760126, "grad_norm": 1.34375, "learning_rate": 8.163059595112099e-10, "loss": 0.125, "step": 32119 }, { "epoch": 199.50467289719626, "grad_norm": 1.7734375, "learning_rate": 7.965175649771173e-10, "loss": 0.1465, "step": 32120 }, { "epoch": 199.5109034267913, "grad_norm": 1.21875, "learning_rate": 7.769719677058751e-10, "loss": 0.1114, "step": 32121 }, { "epoch": 199.5171339563863, "grad_norm": 1.78125, "learning_rate": 7.576691678862214e-10, "loss": 0.1626, "step": 32122 }, { "epoch": 199.52336448598132, "grad_norm": 1.6484375, "learning_rate": 7.386091657041183e-10, "loss": 0.1351, "step": 32123 }, { "epoch": 199.52959501557632, "grad_norm": 1.9296875, "learning_rate": 7.197919613455284e-10, "loss": 0.0978, "step": 32124 }, { "epoch": 199.53582554517135, "grad_norm": 1.984375, "learning_rate": 7.012175549936384e-10, "loss": 0.1318, "step": 32125 }, { "epoch": 199.54205607476635, "grad_norm": 1.125, "learning_rate": 6.828859468288596e-10, "loss": 0.0949, "step": 32126 }, { "epoch": 199.54828660436138, "grad_norm": 1.4609375, "learning_rate": 6.64797137031603e-10, "loss": 0.1243, "step": 32127 }, { "epoch": 199.55451713395638, "grad_norm": 1.34375, "learning_rate": 6.46951125771178e-10, "loss": 0.1411, "step": 32128 }, { "epoch": 199.5607476635514, "grad_norm": 1.1328125, "learning_rate": 6.293479132279956e-10, "loss": 0.1051, "step": 32129 }, { "epoch": 199.5669781931464, "grad_norm": 1.0625, "learning_rate": 6.119874995685892e-10, "loss": 0.0936, "step": 32130 }, { "epoch": 199.57320872274144, "grad_norm": 2.5625, "learning_rate": 5.94869884962268e-10, "loss": 0.1757, "step": 32131 }, { "epoch": 199.57943925233644, "grad_norm": 1.734375, "learning_rate": 5.779950695755654e-10, "loss": 0.118, "step": 32132 }, { "epoch": 199.58566978193147, "grad_norm": 2.34375, "learning_rate": 5.613630535750147e-10, "loss": 0.1516, "step": 32133 }, { "epoch": 199.59190031152647, "grad_norm": 2.609375, "learning_rate": 5.449738371160473e-10, "loss": 0.1606, "step": 32134 }, { "epoch": 199.5981308411215, "grad_norm": 1.9140625, "learning_rate": 5.288274203651966e-10, "loss": 0.1611, "step": 32135 }, { "epoch": 199.6043613707165, "grad_norm": 1.78125, "learning_rate": 5.129238034751183e-10, "loss": 0.1444, "step": 32136 }, { "epoch": 199.61059190031153, "grad_norm": 2.109375, "learning_rate": 4.97262986598468e-10, "loss": 0.1893, "step": 32137 }, { "epoch": 199.61682242990653, "grad_norm": 1.5078125, "learning_rate": 4.818449698906769e-10, "loss": 0.0936, "step": 32138 }, { "epoch": 199.62305295950156, "grad_norm": 2.734375, "learning_rate": 4.666697535016251e-10, "loss": 0.262, "step": 32139 }, { "epoch": 199.62928348909657, "grad_norm": 1.3203125, "learning_rate": 4.517373375784173e-10, "loss": 0.086, "step": 32140 }, { "epoch": 199.6355140186916, "grad_norm": 2.546875, "learning_rate": 4.370477222626068e-10, "loss": 0.2095, "step": 32141 }, { "epoch": 199.6417445482866, "grad_norm": 1.015625, "learning_rate": 4.2260090770129826e-10, "loss": 0.11, "step": 32142 }, { "epoch": 199.64797507788163, "grad_norm": 2.15625, "learning_rate": 4.083968940332694e-10, "loss": 0.2042, "step": 32143 }, { "epoch": 199.65420560747663, "grad_norm": 1.796875, "learning_rate": 3.9443568139452267e-10, "loss": 0.1132, "step": 32144 }, { "epoch": 199.66043613707166, "grad_norm": 2.015625, "learning_rate": 3.8071726992383594e-10, "loss": 0.1724, "step": 32145 }, { "epoch": 199.66666666666666, "grad_norm": 0.79296875, "learning_rate": 3.6724165975166034e-10, "loss": 0.0778, "step": 32146 }, { "epoch": 199.6728971962617, "grad_norm": 2.8125, "learning_rate": 3.540088510112227e-10, "loss": 0.2873, "step": 32147 }, { "epoch": 199.6791277258567, "grad_norm": 2.5625, "learning_rate": 3.4101884383019865e-10, "loss": 0.2552, "step": 32148 }, { "epoch": 199.68535825545172, "grad_norm": 1.5234375, "learning_rate": 3.282716383334883e-10, "loss": 0.1092, "step": 32149 }, { "epoch": 199.69158878504672, "grad_norm": 2.546875, "learning_rate": 3.157672346459917e-10, "loss": 0.2219, "step": 32150 }, { "epoch": 199.69781931464175, "grad_norm": 0.9375, "learning_rate": 3.0350563288983337e-10, "loss": 0.0918, "step": 32151 }, { "epoch": 199.70404984423675, "grad_norm": 1.4140625, "learning_rate": 2.9148683318158674e-10, "loss": 0.1241, "step": 32152 }, { "epoch": 199.71028037383178, "grad_norm": 1.6171875, "learning_rate": 2.7971083564337643e-10, "loss": 0.1487, "step": 32153 }, { "epoch": 199.71651090342678, "grad_norm": 1.046875, "learning_rate": 2.6817764038344907e-10, "loss": 0.0947, "step": 32154 }, { "epoch": 199.7227414330218, "grad_norm": 1.8671875, "learning_rate": 2.5688724751560256e-10, "loss": 0.1229, "step": 32155 }, { "epoch": 199.7289719626168, "grad_norm": 2.421875, "learning_rate": 2.4583965715085924e-10, "loss": 0.2161, "step": 32156 }, { "epoch": 199.73520249221184, "grad_norm": 1.9140625, "learning_rate": 2.3503486939746577e-10, "loss": 0.1686, "step": 32157 }, { "epoch": 199.74143302180684, "grad_norm": 1.109375, "learning_rate": 2.244728843553423e-10, "loss": 0.0995, "step": 32158 }, { "epoch": 199.74766355140187, "grad_norm": 1.6015625, "learning_rate": 2.1415370213273556e-10, "loss": 0.1441, "step": 32159 }, { "epoch": 199.75389408099687, "grad_norm": 2.1875, "learning_rate": 2.040773228295656e-10, "loss": 0.1394, "step": 32160 }, { "epoch": 199.7601246105919, "grad_norm": 2.96875, "learning_rate": 1.9424374654020139e-10, "loss": 0.2682, "step": 32161 }, { "epoch": 199.7663551401869, "grad_norm": 1.7734375, "learning_rate": 1.8465297335901188e-10, "loss": 0.2272, "step": 32162 }, { "epoch": 199.77258566978193, "grad_norm": 1.40625, "learning_rate": 1.7530500338591716e-10, "loss": 0.1115, "step": 32163 }, { "epoch": 199.77881619937693, "grad_norm": 1.7265625, "learning_rate": 1.6619983670418394e-10, "loss": 0.125, "step": 32164 }, { "epoch": 199.78504672897196, "grad_norm": 1.9765625, "learning_rate": 1.5733747340818117e-10, "loss": 0.1828, "step": 32165 }, { "epoch": 199.79127725856696, "grad_norm": 1.2578125, "learning_rate": 1.4871791358117558e-10, "loss": 0.1192, "step": 32166 }, { "epoch": 199.797507788162, "grad_norm": 1.375, "learning_rate": 1.4034115730643394e-10, "loss": 0.0965, "step": 32167 }, { "epoch": 199.80373831775702, "grad_norm": 1.453125, "learning_rate": 1.3220720466722292e-10, "loss": 0.1539, "step": 32168 }, { "epoch": 199.80996884735202, "grad_norm": 1.2890625, "learning_rate": 1.243160557384826e-10, "loss": 0.1153, "step": 32169 }, { "epoch": 199.81619937694705, "grad_norm": 2.0625, "learning_rate": 1.166677106034797e-10, "loss": 0.1865, "step": 32170 }, { "epoch": 199.82242990654206, "grad_norm": 2.15625, "learning_rate": 1.0926216933160316e-10, "loss": 0.2126, "step": 32171 }, { "epoch": 199.82866043613708, "grad_norm": 2.171875, "learning_rate": 1.0209943199501747e-10, "loss": 0.175, "step": 32172 }, { "epoch": 199.8348909657321, "grad_norm": 2.578125, "learning_rate": 9.51794986631116e-11, "loss": 0.2464, "step": 32173 }, { "epoch": 199.84112149532712, "grad_norm": 0.81640625, "learning_rate": 8.850236940527445e-11, "loss": 0.09, "step": 32174 }, { "epoch": 199.84735202492212, "grad_norm": 3.0625, "learning_rate": 8.206804428534386e-11, "loss": 0.307, "step": 32175 }, { "epoch": 199.85358255451715, "grad_norm": 1.4921875, "learning_rate": 7.587652336715768e-11, "loss": 0.1378, "step": 32176 }, { "epoch": 199.85981308411215, "grad_norm": 2.390625, "learning_rate": 6.992780670900256e-11, "loss": 0.2287, "step": 32177 }, { "epoch": 199.86604361370718, "grad_norm": 1.0546875, "learning_rate": 6.422189436916526e-11, "loss": 0.089, "step": 32178 }, { "epoch": 199.87227414330218, "grad_norm": 1.4921875, "learning_rate": 5.875878640038135e-11, "loss": 0.1487, "step": 32179 }, { "epoch": 199.8785046728972, "grad_norm": 1.140625, "learning_rate": 5.353848286093754e-11, "loss": 0.1208, "step": 32180 }, { "epoch": 199.8847352024922, "grad_norm": 2.109375, "learning_rate": 4.8560983798018325e-11, "loss": 0.1628, "step": 32181 }, { "epoch": 199.89096573208724, "grad_norm": 2.4375, "learning_rate": 4.382628926158372e-11, "loss": 0.153, "step": 32182 }, { "epoch": 199.89719626168224, "grad_norm": 2.828125, "learning_rate": 3.9334399296042656e-11, "loss": 0.2207, "step": 32183 }, { "epoch": 199.90342679127727, "grad_norm": 2.46875, "learning_rate": 3.508531394580405e-11, "loss": 0.1677, "step": 32184 }, { "epoch": 199.90965732087227, "grad_norm": 2.375, "learning_rate": 3.107903325250128e-11, "loss": 0.1751, "step": 32185 }, { "epoch": 199.9158878504673, "grad_norm": 2.59375, "learning_rate": 2.731555725499213e-11, "loss": 0.2676, "step": 32186 }, { "epoch": 199.9221183800623, "grad_norm": 2.40625, "learning_rate": 2.379488598935886e-11, "loss": 0.1397, "step": 32187 }, { "epoch": 199.92834890965733, "grad_norm": 2.203125, "learning_rate": 2.051701949168372e-11, "loss": 0.2111, "step": 32188 }, { "epoch": 199.93457943925233, "grad_norm": 2.203125, "learning_rate": 1.748195778972228e-11, "loss": 0.1485, "step": 32189 }, { "epoch": 199.94080996884736, "grad_norm": 2.359375, "learning_rate": 1.4689700916781234e-11, "loss": 0.2072, "step": 32190 }, { "epoch": 199.94704049844236, "grad_norm": 1.5703125, "learning_rate": 1.2140248897840601e-11, "loss": 0.1663, "step": 32191 }, { "epoch": 199.9532710280374, "grad_norm": 1.4765625, "learning_rate": 9.833601757880394e-12, "loss": 0.1332, "step": 32192 }, { "epoch": 199.9595015576324, "grad_norm": 1.0234375, "learning_rate": 7.769759521880638e-12, "loss": 0.1065, "step": 32193 }, { "epoch": 199.96573208722742, "grad_norm": 3.21875, "learning_rate": 5.9487222064946726e-12, "loss": 0.3294, "step": 32194 }, { "epoch": 199.97196261682242, "grad_norm": 2.4375, "learning_rate": 4.3704898283758454e-12, "loss": 0.196, "step": 32195 }, { "epoch": 199.97819314641745, "grad_norm": 2.1875, "learning_rate": 3.0350624097286174e-12, "loss": 0.2144, "step": 32196 }, { "epoch": 199.98442367601245, "grad_norm": 2.375, "learning_rate": 1.942439956104103e-12, "loss": 0.2294, "step": 32197 }, { "epoch": 199.99065420560748, "grad_norm": 0.86328125, "learning_rate": 1.09262248138009e-12, "loss": 0.0762, "step": 32198 }, { "epoch": 199.99688473520249, "grad_norm": 1.3125, "learning_rate": 4.856099938832515e-13, "loss": 0.116, "step": 32199 }, { "epoch": 200.0, "grad_norm": 1.7578125, "learning_rate": 1.2140249916470225e-13, "loss": 0.1189, "step": 32200 }, { "epoch": 200.0, "step": 32200, "total_flos": 0.0, "train_loss": 0.2235217819152319, "train_runtime": 24185.174, "train_samples_per_second": 85.176, "train_steps_per_second": 1.331 } ], "logging_steps": 1, "max_steps": 32200, "num_input_tokens_seen": 0, "num_train_epochs": 200, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }