diff --git "a/ICM-LLaVA-v1.5-13B/trainer_state.json" "b/ICM-LLaVA-v1.5-13B/trainer_state.json" new file mode 100644--- /dev/null +++ "b/ICM-LLaVA-v1.5-13B/trainer_state.json" @@ -0,0 +1,15060 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 500, + "global_step": 2505, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 2.6315789473684213e-07, + "loss": 1.3667, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 5.263157894736843e-07, + "loss": 1.4669, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 7.894736842105263e-07, + "loss": 1.4032, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 1.0526315789473685e-06, + "loss": 1.2917, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 1.3157894736842106e-06, + "loss": 1.395, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 1.5789473684210526e-06, + "loss": 1.3943, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 1.8421052631578948e-06, + "loss": 1.3845, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 2.105263157894737e-06, + "loss": 1.2892, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 2.368421052631579e-06, + "loss": 1.2771, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 2.631578947368421e-06, + "loss": 1.2156, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 2.8947368421052634e-06, + "loss": 1.215, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 3.157894736842105e-06, + "loss": 1.1859, + "step": 12 + }, + { + "epoch": 0.01, + "learning_rate": 3.421052631578948e-06, + "loss": 1.0249, + "step": 13 + }, + { + "epoch": 0.01, + "learning_rate": 3.6842105263157896e-06, + "loss": 1.0523, + "step": 14 + }, + { + "epoch": 0.01, + "learning_rate": 3.947368421052632e-06, + "loss": 0.9712, + "step": 15 + }, + { + "epoch": 0.01, + "learning_rate": 4.210526315789474e-06, + "loss": 1.042, + "step": 16 + }, + { + "epoch": 0.01, + "learning_rate": 4.473684210526316e-06, + "loss": 1.022, + "step": 17 + }, + { + "epoch": 0.01, + "learning_rate": 4.736842105263158e-06, + "loss": 0.8822, + "step": 18 + }, + { + "epoch": 0.01, + "learning_rate": 5e-06, + "loss": 0.9929, + "step": 19 + }, + { + "epoch": 0.01, + "learning_rate": 5.263157894736842e-06, + "loss": 1.0309, + "step": 20 + }, + { + "epoch": 0.01, + "learning_rate": 5.526315789473685e-06, + "loss": 0.9417, + "step": 21 + }, + { + "epoch": 0.01, + "learning_rate": 5.789473684210527e-06, + "loss": 0.9038, + "step": 22 + }, + { + "epoch": 0.01, + "learning_rate": 6.0526315789473685e-06, + "loss": 0.97, + "step": 23 + }, + { + "epoch": 0.01, + "learning_rate": 6.31578947368421e-06, + "loss": 0.8542, + "step": 24 + }, + { + "epoch": 0.01, + "learning_rate": 6.578947368421054e-06, + "loss": 0.851, + "step": 25 + }, + { + "epoch": 0.01, + "learning_rate": 6.842105263157896e-06, + "loss": 0.9293, + "step": 26 + }, + { + "epoch": 0.01, + "learning_rate": 7.1052631578947375e-06, + "loss": 0.9345, + "step": 27 + }, + { + "epoch": 0.01, + "learning_rate": 7.368421052631579e-06, + "loss": 0.8899, + "step": 28 + }, + { + "epoch": 0.01, + "learning_rate": 7.631578947368423e-06, + "loss": 0.9293, + "step": 29 + }, + { + "epoch": 0.01, + "learning_rate": 7.894736842105265e-06, + "loss": 0.9103, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 8.157894736842106e-06, + "loss": 0.9227, + "step": 31 + }, + { + "epoch": 0.01, + "learning_rate": 8.421052631578948e-06, + "loss": 0.8916, + "step": 32 + }, + { + "epoch": 0.01, + "learning_rate": 8.68421052631579e-06, + "loss": 0.8897, + "step": 33 + }, + { + "epoch": 0.01, + "learning_rate": 8.947368421052632e-06, + "loss": 0.8868, + "step": 34 + }, + { + "epoch": 0.01, + "learning_rate": 9.210526315789474e-06, + "loss": 0.8503, + "step": 35 + }, + { + "epoch": 0.01, + "learning_rate": 9.473684210526315e-06, + "loss": 0.8485, + "step": 36 + }, + { + "epoch": 0.01, + "learning_rate": 9.736842105263159e-06, + "loss": 0.6967, + "step": 37 + }, + { + "epoch": 0.02, + "learning_rate": 1e-05, + "loss": 0.8237, + "step": 38 + }, + { + "epoch": 0.02, + "learning_rate": 1.0263157894736844e-05, + "loss": 0.8865, + "step": 39 + }, + { + "epoch": 0.02, + "learning_rate": 1.0526315789473684e-05, + "loss": 0.8603, + "step": 40 + }, + { + "epoch": 0.02, + "learning_rate": 1.0789473684210528e-05, + "loss": 0.7898, + "step": 41 + }, + { + "epoch": 0.02, + "learning_rate": 1.105263157894737e-05, + "loss": 0.8426, + "step": 42 + }, + { + "epoch": 0.02, + "learning_rate": 1.1315789473684212e-05, + "loss": 0.8137, + "step": 43 + }, + { + "epoch": 0.02, + "learning_rate": 1.1578947368421053e-05, + "loss": 0.7421, + "step": 44 + }, + { + "epoch": 0.02, + "learning_rate": 1.1842105263157895e-05, + "loss": 0.7519, + "step": 45 + }, + { + "epoch": 0.02, + "learning_rate": 1.2105263157894737e-05, + "loss": 0.8385, + "step": 46 + }, + { + "epoch": 0.02, + "learning_rate": 1.236842105263158e-05, + "loss": 0.8534, + "step": 47 + }, + { + "epoch": 0.02, + "learning_rate": 1.263157894736842e-05, + "loss": 0.8638, + "step": 48 + }, + { + "epoch": 0.02, + "learning_rate": 1.2894736842105264e-05, + "loss": 0.8566, + "step": 49 + }, + { + "epoch": 0.02, + "learning_rate": 1.3157894736842108e-05, + "loss": 0.8702, + "step": 50 + }, + { + "epoch": 0.02, + "learning_rate": 1.3421052631578948e-05, + "loss": 0.843, + "step": 51 + }, + { + "epoch": 0.02, + "learning_rate": 1.3684210526315791e-05, + "loss": 0.7185, + "step": 52 + }, + { + "epoch": 0.02, + "learning_rate": 1.3947368421052631e-05, + "loss": 0.8169, + "step": 53 + }, + { + "epoch": 0.02, + "learning_rate": 1.4210526315789475e-05, + "loss": 0.8247, + "step": 54 + }, + { + "epoch": 0.02, + "learning_rate": 1.4473684210526317e-05, + "loss": 0.7503, + "step": 55 + }, + { + "epoch": 0.02, + "learning_rate": 1.4736842105263159e-05, + "loss": 0.7951, + "step": 56 + }, + { + "epoch": 0.02, + "learning_rate": 1.5000000000000002e-05, + "loss": 0.8168, + "step": 57 + }, + { + "epoch": 0.02, + "learning_rate": 1.5263157894736846e-05, + "loss": 0.8103, + "step": 58 + }, + { + "epoch": 0.02, + "learning_rate": 1.5526315789473686e-05, + "loss": 0.8028, + "step": 59 + }, + { + "epoch": 0.02, + "learning_rate": 1.578947368421053e-05, + "loss": 0.8101, + "step": 60 + }, + { + "epoch": 0.02, + "learning_rate": 1.605263157894737e-05, + "loss": 0.8169, + "step": 61 + }, + { + "epoch": 0.02, + "learning_rate": 1.6315789473684213e-05, + "loss": 0.8336, + "step": 62 + }, + { + "epoch": 0.03, + "learning_rate": 1.6578947368421053e-05, + "loss": 0.7442, + "step": 63 + }, + { + "epoch": 0.03, + "learning_rate": 1.6842105263157896e-05, + "loss": 0.7358, + "step": 64 + }, + { + "epoch": 0.03, + "learning_rate": 1.7105263157894737e-05, + "loss": 0.8073, + "step": 65 + }, + { + "epoch": 0.03, + "learning_rate": 1.736842105263158e-05, + "loss": 0.8087, + "step": 66 + }, + { + "epoch": 0.03, + "learning_rate": 1.763157894736842e-05, + "loss": 0.7804, + "step": 67 + }, + { + "epoch": 0.03, + "learning_rate": 1.7894736842105264e-05, + "loss": 0.8268, + "step": 68 + }, + { + "epoch": 0.03, + "learning_rate": 1.8157894736842107e-05, + "loss": 0.8519, + "step": 69 + }, + { + "epoch": 0.03, + "learning_rate": 1.8421052631578947e-05, + "loss": 0.7764, + "step": 70 + }, + { + "epoch": 0.03, + "learning_rate": 1.868421052631579e-05, + "loss": 0.8059, + "step": 71 + }, + { + "epoch": 0.03, + "learning_rate": 1.894736842105263e-05, + "loss": 0.8386, + "step": 72 + }, + { + "epoch": 0.03, + "learning_rate": 1.9210526315789474e-05, + "loss": 0.7393, + "step": 73 + }, + { + "epoch": 0.03, + "learning_rate": 1.9473684210526318e-05, + "loss": 0.7383, + "step": 74 + }, + { + "epoch": 0.03, + "learning_rate": 1.9736842105263158e-05, + "loss": 0.8369, + "step": 75 + }, + { + "epoch": 0.03, + "learning_rate": 2e-05, + "loss": 0.8179, + "step": 76 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999991635987763e-05, + "loss": 0.7915, + "step": 77 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999966543965042e-05, + "loss": 0.8061, + "step": 78 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999924723973812e-05, + "loss": 0.7748, + "step": 79 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999866176084026e-05, + "loss": 0.7575, + "step": 80 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999790900393628e-05, + "loss": 0.7751, + "step": 81 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999698897028537e-05, + "loss": 0.7951, + "step": 82 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999590166142656e-05, + "loss": 0.7897, + "step": 83 + }, + { + "epoch": 0.03, + "learning_rate": 1.999946470791787e-05, + "loss": 0.8516, + "step": 84 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999322522564048e-05, + "loss": 0.8401, + "step": 85 + }, + { + "epoch": 0.03, + "learning_rate": 1.9999163610319035e-05, + "loss": 0.7802, + "step": 86 + }, + { + "epoch": 0.03, + "learning_rate": 1.9998987971448664e-05, + "loss": 0.7852, + "step": 87 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998795606246738e-05, + "loss": 0.7662, + "step": 88 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998586515035053e-05, + "loss": 0.8532, + "step": 89 + }, + { + "epoch": 0.04, + "learning_rate": 1.9998360698163375e-05, + "loss": 0.7588, + "step": 90 + }, + { + "epoch": 0.04, + "learning_rate": 1.999811815600945e-05, + "loss": 0.7607, + "step": 91 + }, + { + "epoch": 0.04, + "learning_rate": 1.9997858888978997e-05, + "loss": 0.7544, + "step": 92 + }, + { + "epoch": 0.04, + "learning_rate": 1.999758289750573e-05, + "loss": 0.8162, + "step": 93 + }, + { + "epoch": 0.04, + "learning_rate": 1.999729018205132e-05, + "loss": 0.7875, + "step": 94 + }, + { + "epoch": 0.04, + "learning_rate": 1.9996980743105427e-05, + "loss": 0.8886, + "step": 95 + }, + { + "epoch": 0.04, + "learning_rate": 1.999665458118568e-05, + "loss": 0.835, + "step": 96 + }, + { + "epoch": 0.04, + "learning_rate": 1.999631169683768e-05, + "loss": 0.7994, + "step": 97 + }, + { + "epoch": 0.04, + "learning_rate": 1.9995952090635007e-05, + "loss": 0.8543, + "step": 98 + }, + { + "epoch": 0.04, + "learning_rate": 1.9995575763179213e-05, + "loss": 0.777, + "step": 99 + }, + { + "epoch": 0.04, + "learning_rate": 1.999518271509982e-05, + "loss": 0.7948, + "step": 100 + }, + { + "epoch": 0.04, + "learning_rate": 1.999477294705431e-05, + "loss": 0.8141, + "step": 101 + }, + { + "epoch": 0.04, + "learning_rate": 1.999434645972816e-05, + "loss": 0.7906, + "step": 102 + }, + { + "epoch": 0.04, + "learning_rate": 1.999390325383479e-05, + "loss": 0.8265, + "step": 103 + }, + { + "epoch": 0.04, + "learning_rate": 1.9993443330115592e-05, + "loss": 0.8766, + "step": 104 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992966689339936e-05, + "loss": 0.7439, + "step": 105 + }, + { + "epoch": 0.04, + "learning_rate": 1.9992473332305145e-05, + "loss": 0.8201, + "step": 106 + }, + { + "epoch": 0.04, + "learning_rate": 1.9991963259836504e-05, + "loss": 0.7336, + "step": 107 + }, + { + "epoch": 0.04, + "learning_rate": 1.9991436472787267e-05, + "loss": 0.8224, + "step": 108 + }, + { + "epoch": 0.04, + "learning_rate": 1.9990892972038646e-05, + "loss": 0.7889, + "step": 109 + }, + { + "epoch": 0.04, + "learning_rate": 1.9990332758499805e-05, + "loss": 0.8431, + "step": 110 + }, + { + "epoch": 0.04, + "learning_rate": 1.9989755833107875e-05, + "loss": 0.7188, + "step": 111 + }, + { + "epoch": 0.04, + "learning_rate": 1.998916219682794e-05, + "loss": 0.8327, + "step": 112 + }, + { + "epoch": 0.05, + "learning_rate": 1.998855185065303e-05, + "loss": 0.7889, + "step": 113 + }, + { + "epoch": 0.05, + "learning_rate": 1.998792479560414e-05, + "loss": 0.8218, + "step": 114 + }, + { + "epoch": 0.05, + "learning_rate": 1.9987281032730206e-05, + "loss": 0.7737, + "step": 115 + }, + { + "epoch": 0.05, + "learning_rate": 1.9986620563108117e-05, + "loss": 0.7578, + "step": 116 + }, + { + "epoch": 0.05, + "learning_rate": 1.9985943387842704e-05, + "loss": 0.8361, + "step": 117 + }, + { + "epoch": 0.05, + "learning_rate": 1.9985249508066754e-05, + "loss": 0.8202, + "step": 118 + }, + { + "epoch": 0.05, + "learning_rate": 1.9984538924940987e-05, + "loss": 0.7849, + "step": 119 + }, + { + "epoch": 0.05, + "learning_rate": 1.998381163965407e-05, + "loss": 0.8505, + "step": 120 + }, + { + "epoch": 0.05, + "learning_rate": 1.9983067653422603e-05, + "loss": 0.8631, + "step": 121 + }, + { + "epoch": 0.05, + "learning_rate": 1.9982306967491136e-05, + "loss": 0.7784, + "step": 122 + }, + { + "epoch": 0.05, + "learning_rate": 1.998152958313214e-05, + "loss": 0.7732, + "step": 123 + }, + { + "epoch": 0.05, + "learning_rate": 1.9980735501646026e-05, + "loss": 0.7643, + "step": 124 + }, + { + "epoch": 0.05, + "learning_rate": 1.997992472436114e-05, + "loss": 0.8322, + "step": 125 + }, + { + "epoch": 0.05, + "learning_rate": 1.9979097252633748e-05, + "loss": 0.779, + "step": 126 + }, + { + "epoch": 0.05, + "learning_rate": 1.9978253087848046e-05, + "loss": 0.8101, + "step": 127 + }, + { + "epoch": 0.05, + "learning_rate": 1.9977392231416158e-05, + "loss": 0.809, + "step": 128 + }, + { + "epoch": 0.05, + "learning_rate": 1.9976514684778124e-05, + "loss": 0.8142, + "step": 129 + }, + { + "epoch": 0.05, + "learning_rate": 1.9975620449401913e-05, + "loss": 0.7599, + "step": 130 + }, + { + "epoch": 0.05, + "learning_rate": 1.997470952678339e-05, + "loss": 0.7742, + "step": 131 + }, + { + "epoch": 0.05, + "learning_rate": 1.9973781918446363e-05, + "loss": 0.8018, + "step": 132 + }, + { + "epoch": 0.05, + "learning_rate": 1.9972837625942533e-05, + "loss": 0.7887, + "step": 133 + }, + { + "epoch": 0.05, + "learning_rate": 1.997187665085151e-05, + "loss": 0.7791, + "step": 134 + }, + { + "epoch": 0.05, + "learning_rate": 1.997089899478082e-05, + "loss": 0.7995, + "step": 135 + }, + { + "epoch": 0.05, + "learning_rate": 1.9969904659365887e-05, + "loss": 0.7661, + "step": 136 + }, + { + "epoch": 0.05, + "learning_rate": 1.9968893646270037e-05, + "loss": 0.8295, + "step": 137 + }, + { + "epoch": 0.06, + "learning_rate": 1.99678659571845e-05, + "loss": 0.8073, + "step": 138 + }, + { + "epoch": 0.06, + "learning_rate": 1.9966821593828393e-05, + "loss": 0.8292, + "step": 139 + }, + { + "epoch": 0.06, + "learning_rate": 1.996576055794873e-05, + "loss": 0.7534, + "step": 140 + }, + { + "epoch": 0.06, + "learning_rate": 1.996468285132041e-05, + "loss": 0.7811, + "step": 141 + }, + { + "epoch": 0.06, + "learning_rate": 1.9963588475746233e-05, + "loss": 0.8424, + "step": 142 + }, + { + "epoch": 0.06, + "learning_rate": 1.9962477433056864e-05, + "loss": 0.7954, + "step": 143 + }, + { + "epoch": 0.06, + "learning_rate": 1.996134972511086e-05, + "loss": 0.7963, + "step": 144 + }, + { + "epoch": 0.06, + "learning_rate": 1.996020535379466e-05, + "loss": 0.7631, + "step": 145 + }, + { + "epoch": 0.06, + "learning_rate": 1.9959044321022563e-05, + "loss": 0.8193, + "step": 146 + }, + { + "epoch": 0.06, + "learning_rate": 1.9957866628736756e-05, + "loss": 0.7911, + "step": 147 + }, + { + "epoch": 0.06, + "learning_rate": 1.9956672278907273e-05, + "loss": 0.8448, + "step": 148 + }, + { + "epoch": 0.06, + "learning_rate": 1.9955461273532037e-05, + "loss": 0.7566, + "step": 149 + }, + { + "epoch": 0.06, + "learning_rate": 1.9954233614636817e-05, + "loss": 0.8123, + "step": 150 + }, + { + "epoch": 0.06, + "learning_rate": 1.995298930427524e-05, + "loss": 0.7874, + "step": 151 + }, + { + "epoch": 0.06, + "learning_rate": 1.99517283445288e-05, + "loss": 0.768, + "step": 152 + }, + { + "epoch": 0.06, + "learning_rate": 1.9950450737506825e-05, + "loss": 0.7528, + "step": 153 + }, + { + "epoch": 0.06, + "learning_rate": 1.9949156485346502e-05, + "loss": 0.8067, + "step": 154 + }, + { + "epoch": 0.06, + "learning_rate": 1.994784559021286e-05, + "loss": 0.8375, + "step": 155 + }, + { + "epoch": 0.06, + "learning_rate": 1.9946518054298768e-05, + "loss": 0.8129, + "step": 156 + }, + { + "epoch": 0.06, + "learning_rate": 1.994517387982493e-05, + "loss": 0.7765, + "step": 157 + }, + { + "epoch": 0.06, + "learning_rate": 1.9943813069039883e-05, + "loss": 0.8144, + "step": 158 + }, + { + "epoch": 0.06, + "learning_rate": 1.994243562422e-05, + "loss": 0.7728, + "step": 159 + }, + { + "epoch": 0.06, + "learning_rate": 1.9941041547669467e-05, + "loss": 0.8061, + "step": 160 + }, + { + "epoch": 0.06, + "learning_rate": 1.9939630841720302e-05, + "loss": 0.7803, + "step": 161 + }, + { + "epoch": 0.06, + "learning_rate": 1.993820350873234e-05, + "loss": 0.7117, + "step": 162 + }, + { + "epoch": 0.07, + "learning_rate": 1.993675955109322e-05, + "loss": 0.7394, + "step": 163 + }, + { + "epoch": 0.07, + "learning_rate": 1.993529897121841e-05, + "loss": 0.7994, + "step": 164 + }, + { + "epoch": 0.07, + "learning_rate": 1.993382177155116e-05, + "loss": 0.7255, + "step": 165 + }, + { + "epoch": 0.07, + "learning_rate": 1.993232795456254e-05, + "loss": 0.7427, + "step": 166 + }, + { + "epoch": 0.07, + "learning_rate": 1.9930817522751403e-05, + "loss": 0.7559, + "step": 167 + }, + { + "epoch": 0.07, + "learning_rate": 1.9929290478644412e-05, + "loss": 0.7892, + "step": 168 + }, + { + "epoch": 0.07, + "learning_rate": 1.9927746824796003e-05, + "loss": 0.7877, + "step": 169 + }, + { + "epoch": 0.07, + "learning_rate": 1.992618656378841e-05, + "loss": 0.826, + "step": 170 + }, + { + "epoch": 0.07, + "learning_rate": 1.9924609698231637e-05, + "loss": 0.7785, + "step": 171 + }, + { + "epoch": 0.07, + "learning_rate": 1.992301623076347e-05, + "loss": 0.7688, + "step": 172 + }, + { + "epoch": 0.07, + "learning_rate": 1.9921406164049467e-05, + "loss": 0.7849, + "step": 173 + }, + { + "epoch": 0.07, + "learning_rate": 1.991977950078295e-05, + "loss": 0.7868, + "step": 174 + }, + { + "epoch": 0.07, + "learning_rate": 1.9918136243685003e-05, + "loss": 0.7601, + "step": 175 + }, + { + "epoch": 0.07, + "learning_rate": 1.9916476395504473e-05, + "loss": 0.757, + "step": 176 + }, + { + "epoch": 0.07, + "learning_rate": 1.991479995901796e-05, + "loss": 0.7602, + "step": 177 + }, + { + "epoch": 0.07, + "learning_rate": 1.991310693702981e-05, + "loss": 0.8142, + "step": 178 + }, + { + "epoch": 0.07, + "learning_rate": 1.991139733237211e-05, + "loss": 0.7908, + "step": 179 + }, + { + "epoch": 0.07, + "learning_rate": 1.9909671147904693e-05, + "loss": 0.7681, + "step": 180 + }, + { + "epoch": 0.07, + "learning_rate": 1.9907928386515126e-05, + "loss": 0.8557, + "step": 181 + }, + { + "epoch": 0.07, + "learning_rate": 1.990616905111871e-05, + "loss": 0.8016, + "step": 182 + }, + { + "epoch": 0.07, + "learning_rate": 1.9904393144658455e-05, + "loss": 0.7459, + "step": 183 + }, + { + "epoch": 0.07, + "learning_rate": 1.9902600670105107e-05, + "loss": 0.7695, + "step": 184 + }, + { + "epoch": 0.07, + "learning_rate": 1.9900791630457122e-05, + "loss": 0.7501, + "step": 185 + }, + { + "epoch": 0.07, + "learning_rate": 1.9898966028740662e-05, + "loss": 0.8202, + "step": 186 + }, + { + "epoch": 0.07, + "learning_rate": 1.9897123868009606e-05, + "loss": 0.7029, + "step": 187 + }, + { + "epoch": 0.08, + "learning_rate": 1.9895265151345516e-05, + "loss": 0.8524, + "step": 188 + }, + { + "epoch": 0.08, + "learning_rate": 1.9893389881857665e-05, + "loss": 0.8094, + "step": 189 + }, + { + "epoch": 0.08, + "learning_rate": 1.9891498062683005e-05, + "loss": 0.768, + "step": 190 + }, + { + "epoch": 0.08, + "learning_rate": 1.9889589696986173e-05, + "loss": 0.8333, + "step": 191 + }, + { + "epoch": 0.08, + "learning_rate": 1.9887664787959495e-05, + "loss": 0.7184, + "step": 192 + }, + { + "epoch": 0.08, + "learning_rate": 1.988572333882296e-05, + "loss": 0.7137, + "step": 193 + }, + { + "epoch": 0.08, + "learning_rate": 1.9883765352824224e-05, + "loss": 0.7936, + "step": 194 + }, + { + "epoch": 0.08, + "learning_rate": 1.9881790833238615e-05, + "loss": 0.7713, + "step": 195 + }, + { + "epoch": 0.08, + "learning_rate": 1.9879799783369117e-05, + "loss": 0.7762, + "step": 196 + }, + { + "epoch": 0.08, + "learning_rate": 1.987779220654636e-05, + "loss": 0.7531, + "step": 197 + }, + { + "epoch": 0.08, + "learning_rate": 1.9875768106128627e-05, + "loss": 0.7284, + "step": 198 + }, + { + "epoch": 0.08, + "learning_rate": 1.987372748550183e-05, + "loss": 0.7663, + "step": 199 + }, + { + "epoch": 0.08, + "learning_rate": 1.987167034807953e-05, + "loss": 0.7214, + "step": 200 + }, + { + "epoch": 0.08, + "learning_rate": 1.9869596697302912e-05, + "loss": 0.7447, + "step": 201 + }, + { + "epoch": 0.08, + "learning_rate": 1.9867506536640782e-05, + "loss": 0.7819, + "step": 202 + }, + { + "epoch": 0.08, + "learning_rate": 1.9865399869589565e-05, + "loss": 0.685, + "step": 203 + }, + { + "epoch": 0.08, + "learning_rate": 1.9863276699673305e-05, + "loss": 0.7767, + "step": 204 + }, + { + "epoch": 0.08, + "learning_rate": 1.9861137030443638e-05, + "loss": 0.7879, + "step": 205 + }, + { + "epoch": 0.08, + "learning_rate": 1.985898086547981e-05, + "loss": 0.7345, + "step": 206 + }, + { + "epoch": 0.08, + "learning_rate": 1.9856808208388663e-05, + "loss": 0.8273, + "step": 207 + }, + { + "epoch": 0.08, + "learning_rate": 1.985461906280462e-05, + "loss": 0.7524, + "step": 208 + }, + { + "epoch": 0.08, + "learning_rate": 1.9852413432389685e-05, + "loss": 0.759, + "step": 209 + }, + { + "epoch": 0.08, + "learning_rate": 1.985019132083345e-05, + "loss": 0.793, + "step": 210 + }, + { + "epoch": 0.08, + "learning_rate": 1.9847952731853067e-05, + "loss": 0.8179, + "step": 211 + }, + { + "epoch": 0.08, + "learning_rate": 1.9845697669193248e-05, + "loss": 0.7947, + "step": 212 + }, + { + "epoch": 0.09, + "learning_rate": 1.984342613662627e-05, + "loss": 0.7698, + "step": 213 + }, + { + "epoch": 0.09, + "learning_rate": 1.984113813795196e-05, + "loss": 0.7553, + "step": 214 + }, + { + "epoch": 0.09, + "learning_rate": 1.9838833676997687e-05, + "loss": 0.811, + "step": 215 + }, + { + "epoch": 0.09, + "learning_rate": 1.9836512757618355e-05, + "loss": 0.738, + "step": 216 + }, + { + "epoch": 0.09, + "learning_rate": 1.983417538369641e-05, + "loss": 0.738, + "step": 217 + }, + { + "epoch": 0.09, + "learning_rate": 1.9831821559141817e-05, + "loss": 0.8366, + "step": 218 + }, + { + "epoch": 0.09, + "learning_rate": 1.9829451287892048e-05, + "loss": 0.7353, + "step": 219 + }, + { + "epoch": 0.09, + "learning_rate": 1.9827064573912113e-05, + "loss": 0.833, + "step": 220 + }, + { + "epoch": 0.09, + "learning_rate": 1.9824661421194506e-05, + "loss": 0.7176, + "step": 221 + }, + { + "epoch": 0.09, + "learning_rate": 1.9822241833759226e-05, + "loss": 0.763, + "step": 222 + }, + { + "epoch": 0.09, + "learning_rate": 1.9819805815653768e-05, + "loss": 0.7626, + "step": 223 + }, + { + "epoch": 0.09, + "learning_rate": 1.9817353370953105e-05, + "loss": 0.8586, + "step": 224 + }, + { + "epoch": 0.09, + "learning_rate": 1.9814884503759698e-05, + "loss": 0.8328, + "step": 225 + }, + { + "epoch": 0.09, + "learning_rate": 1.9812399218203467e-05, + "loss": 0.8083, + "step": 226 + }, + { + "epoch": 0.09, + "learning_rate": 1.9809897518441813e-05, + "loss": 0.7878, + "step": 227 + }, + { + "epoch": 0.09, + "learning_rate": 1.980737940865958e-05, + "loss": 0.7842, + "step": 228 + }, + { + "epoch": 0.09, + "learning_rate": 1.9804844893069063e-05, + "loss": 0.831, + "step": 229 + }, + { + "epoch": 0.09, + "learning_rate": 1.9802293975910016e-05, + "loss": 0.7448, + "step": 230 + }, + { + "epoch": 0.09, + "learning_rate": 1.979972666144961e-05, + "loss": 0.8183, + "step": 231 + }, + { + "epoch": 0.09, + "learning_rate": 1.9797142953982463e-05, + "loss": 0.755, + "step": 232 + }, + { + "epoch": 0.09, + "learning_rate": 1.9794542857830605e-05, + "loss": 0.7549, + "step": 233 + }, + { + "epoch": 0.09, + "learning_rate": 1.979192637734348e-05, + "loss": 0.7699, + "step": 234 + }, + { + "epoch": 0.09, + "learning_rate": 1.9789293516897947e-05, + "loss": 0.7959, + "step": 235 + }, + { + "epoch": 0.09, + "learning_rate": 1.9786644280898258e-05, + "loss": 0.788, + "step": 236 + }, + { + "epoch": 0.09, + "learning_rate": 1.9783978673776067e-05, + "loss": 0.7227, + "step": 237 + }, + { + "epoch": 0.1, + "learning_rate": 1.9781296699990398e-05, + "loss": 0.7571, + "step": 238 + }, + { + "epoch": 0.1, + "learning_rate": 1.977859836402767e-05, + "loss": 0.7468, + "step": 239 + }, + { + "epoch": 0.1, + "learning_rate": 1.9775883670401667e-05, + "loss": 0.7511, + "step": 240 + }, + { + "epoch": 0.1, + "learning_rate": 1.9773152623653535e-05, + "loss": 0.7805, + "step": 241 + }, + { + "epoch": 0.1, + "learning_rate": 1.977040522835177e-05, + "loss": 0.8516, + "step": 242 + }, + { + "epoch": 0.1, + "learning_rate": 1.9767641489092232e-05, + "loss": 0.7001, + "step": 243 + }, + { + "epoch": 0.1, + "learning_rate": 1.97648614104981e-05, + "loss": 0.7487, + "step": 244 + }, + { + "epoch": 0.1, + "learning_rate": 1.97620649972199e-05, + "loss": 0.7447, + "step": 245 + }, + { + "epoch": 0.1, + "learning_rate": 1.9759252253935485e-05, + "loss": 0.7712, + "step": 246 + }, + { + "epoch": 0.1, + "learning_rate": 1.975642318535001e-05, + "loss": 0.7241, + "step": 247 + }, + { + "epoch": 0.1, + "learning_rate": 1.975357779619595e-05, + "loss": 0.8578, + "step": 248 + }, + { + "epoch": 0.1, + "learning_rate": 1.9750716091233083e-05, + "loss": 0.7485, + "step": 249 + }, + { + "epoch": 0.1, + "learning_rate": 1.974783807524847e-05, + "loss": 0.8312, + "step": 250 + }, + { + "epoch": 0.1, + "learning_rate": 1.974494375305647e-05, + "loss": 0.7677, + "step": 251 + }, + { + "epoch": 0.1, + "learning_rate": 1.974203312949871e-05, + "loss": 0.7512, + "step": 252 + }, + { + "epoch": 0.1, + "learning_rate": 1.9739106209444087e-05, + "loss": 0.7521, + "step": 253 + }, + { + "epoch": 0.1, + "learning_rate": 1.9736162997788757e-05, + "loss": 0.7333, + "step": 254 + }, + { + "epoch": 0.1, + "learning_rate": 1.9733203499456136e-05, + "loss": 0.7389, + "step": 255 + }, + { + "epoch": 0.1, + "learning_rate": 1.9730227719396883e-05, + "loss": 0.7762, + "step": 256 + }, + { + "epoch": 0.1, + "learning_rate": 1.9727235662588882e-05, + "loss": 0.7512, + "step": 257 + }, + { + "epoch": 0.1, + "learning_rate": 1.9724227334037255e-05, + "loss": 0.7057, + "step": 258 + }, + { + "epoch": 0.1, + "learning_rate": 1.9721202738774346e-05, + "loss": 0.7984, + "step": 259 + }, + { + "epoch": 0.1, + "learning_rate": 1.9718161881859703e-05, + "loss": 0.822, + "step": 260 + }, + { + "epoch": 0.1, + "learning_rate": 1.9715104768380078e-05, + "loss": 0.762, + "step": 261 + }, + { + "epoch": 0.1, + "learning_rate": 1.971203140344942e-05, + "loss": 0.7968, + "step": 262 + }, + { + "epoch": 0.1, + "learning_rate": 1.9708941792208857e-05, + "loss": 0.7697, + "step": 263 + }, + { + "epoch": 0.11, + "learning_rate": 1.9705835939826704e-05, + "loss": 0.7772, + "step": 264 + }, + { + "epoch": 0.11, + "learning_rate": 1.9702713851498436e-05, + "loss": 0.7581, + "step": 265 + }, + { + "epoch": 0.11, + "learning_rate": 1.9699575532446688e-05, + "loss": 0.7487, + "step": 266 + }, + { + "epoch": 0.11, + "learning_rate": 1.969642098792125e-05, + "loss": 0.6632, + "step": 267 + }, + { + "epoch": 0.11, + "learning_rate": 1.9693250223199053e-05, + "loss": 0.7478, + "step": 268 + }, + { + "epoch": 0.11, + "learning_rate": 1.9690063243584153e-05, + "loss": 0.8053, + "step": 269 + }, + { + "epoch": 0.11, + "learning_rate": 1.9686860054407745e-05, + "loss": 0.7582, + "step": 270 + }, + { + "epoch": 0.11, + "learning_rate": 1.968364066102813e-05, + "loss": 0.752, + "step": 271 + }, + { + "epoch": 0.11, + "learning_rate": 1.9680405068830716e-05, + "loss": 0.8279, + "step": 272 + }, + { + "epoch": 0.11, + "learning_rate": 1.967715328322801e-05, + "loss": 0.7099, + "step": 273 + }, + { + "epoch": 0.11, + "learning_rate": 1.967388530965961e-05, + "loss": 0.7939, + "step": 274 + }, + { + "epoch": 0.11, + "learning_rate": 1.967060115359218e-05, + "loss": 0.8033, + "step": 275 + }, + { + "epoch": 0.11, + "learning_rate": 1.9667300820519478e-05, + "loss": 0.8189, + "step": 276 + }, + { + "epoch": 0.11, + "learning_rate": 1.96639843159623e-05, + "loss": 0.7836, + "step": 277 + }, + { + "epoch": 0.11, + "learning_rate": 1.9660651645468504e-05, + "loss": 0.7983, + "step": 278 + }, + { + "epoch": 0.11, + "learning_rate": 1.965730281461299e-05, + "loss": 0.7193, + "step": 279 + }, + { + "epoch": 0.11, + "learning_rate": 1.9653937828997694e-05, + "loss": 0.7305, + "step": 280 + }, + { + "epoch": 0.11, + "learning_rate": 1.965055669425157e-05, + "loss": 0.782, + "step": 281 + }, + { + "epoch": 0.11, + "learning_rate": 1.9647159416030588e-05, + "loss": 0.8005, + "step": 282 + }, + { + "epoch": 0.11, + "learning_rate": 1.9643746000017722e-05, + "loss": 0.7896, + "step": 283 + }, + { + "epoch": 0.11, + "learning_rate": 1.9640316451922943e-05, + "loss": 0.7741, + "step": 284 + }, + { + "epoch": 0.11, + "learning_rate": 1.9636870777483207e-05, + "loss": 0.7602, + "step": 285 + }, + { + "epoch": 0.11, + "learning_rate": 1.9633408982462453e-05, + "loss": 0.7443, + "step": 286 + }, + { + "epoch": 0.11, + "learning_rate": 1.9629931072651572e-05, + "loss": 0.7595, + "step": 287 + }, + { + "epoch": 0.11, + "learning_rate": 1.9626437053868423e-05, + "loss": 0.788, + "step": 288 + }, + { + "epoch": 0.12, + "learning_rate": 1.962292693195781e-05, + "loss": 0.753, + "step": 289 + }, + { + "epoch": 0.12, + "learning_rate": 1.9619400712791473e-05, + "loss": 0.7471, + "step": 290 + }, + { + "epoch": 0.12, + "learning_rate": 1.961585840226808e-05, + "loss": 0.7739, + "step": 291 + }, + { + "epoch": 0.12, + "learning_rate": 1.9612300006313216e-05, + "loss": 0.7573, + "step": 292 + }, + { + "epoch": 0.12, + "learning_rate": 1.9608725530879376e-05, + "loss": 0.7548, + "step": 293 + }, + { + "epoch": 0.12, + "learning_rate": 1.960513498194595e-05, + "loss": 0.7171, + "step": 294 + }, + { + "epoch": 0.12, + "learning_rate": 1.9601528365519218e-05, + "loss": 0.7282, + "step": 295 + }, + { + "epoch": 0.12, + "learning_rate": 1.9597905687632336e-05, + "loss": 0.7544, + "step": 296 + }, + { + "epoch": 0.12, + "learning_rate": 1.959426695434533e-05, + "loss": 0.7981, + "step": 297 + }, + { + "epoch": 0.12, + "learning_rate": 1.9590612171745076e-05, + "loss": 0.8163, + "step": 298 + }, + { + "epoch": 0.12, + "learning_rate": 1.958694134594531e-05, + "loss": 0.7772, + "step": 299 + }, + { + "epoch": 0.12, + "learning_rate": 1.95832544830866e-05, + "loss": 0.7725, + "step": 300 + }, + { + "epoch": 0.12, + "learning_rate": 1.9579551589336333e-05, + "loss": 0.7704, + "step": 301 + }, + { + "epoch": 0.12, + "learning_rate": 1.9575832670888723e-05, + "loss": 0.7451, + "step": 302 + }, + { + "epoch": 0.12, + "learning_rate": 1.957209773396478e-05, + "loss": 0.8181, + "step": 303 + }, + { + "epoch": 0.12, + "learning_rate": 1.956834678481232e-05, + "loss": 0.7521, + "step": 304 + }, + { + "epoch": 0.12, + "learning_rate": 1.9564579829705945e-05, + "loss": 0.8028, + "step": 305 + }, + { + "epoch": 0.12, + "learning_rate": 1.9560796874947016e-05, + "loss": 0.6878, + "step": 306 + }, + { + "epoch": 0.12, + "learning_rate": 1.9556997926863676e-05, + "loss": 0.7599, + "step": 307 + }, + { + "epoch": 0.12, + "learning_rate": 1.9553182991810812e-05, + "loss": 0.6992, + "step": 308 + }, + { + "epoch": 0.12, + "learning_rate": 1.954935207617006e-05, + "loss": 0.7747, + "step": 309 + }, + { + "epoch": 0.12, + "learning_rate": 1.954550518634978e-05, + "loss": 0.7928, + "step": 310 + }, + { + "epoch": 0.12, + "learning_rate": 1.9541642328785065e-05, + "loss": 0.7757, + "step": 311 + }, + { + "epoch": 0.12, + "learning_rate": 1.9537763509937707e-05, + "loss": 0.7845, + "step": 312 + }, + { + "epoch": 0.12, + "learning_rate": 1.9533868736296205e-05, + "loss": 0.7239, + "step": 313 + }, + { + "epoch": 0.13, + "learning_rate": 1.9529958014375748e-05, + "loss": 0.7212, + "step": 314 + }, + { + "epoch": 0.13, + "learning_rate": 1.95260313507182e-05, + "loss": 0.7527, + "step": 315 + }, + { + "epoch": 0.13, + "learning_rate": 1.952208875189209e-05, + "loss": 0.7617, + "step": 316 + }, + { + "epoch": 0.13, + "learning_rate": 1.9518130224492612e-05, + "loss": 0.7371, + "step": 317 + }, + { + "epoch": 0.13, + "learning_rate": 1.9514155775141602e-05, + "loss": 0.766, + "step": 318 + }, + { + "epoch": 0.13, + "learning_rate": 1.9510165410487524e-05, + "loss": 0.7363, + "step": 319 + }, + { + "epoch": 0.13, + "learning_rate": 1.950615913720547e-05, + "loss": 0.7198, + "step": 320 + }, + { + "epoch": 0.13, + "learning_rate": 1.9502136961997144e-05, + "loss": 0.7836, + "step": 321 + }, + { + "epoch": 0.13, + "learning_rate": 1.9498098891590855e-05, + "loss": 0.7969, + "step": 322 + }, + { + "epoch": 0.13, + "learning_rate": 1.9494044932741488e-05, + "loss": 0.7675, + "step": 323 + }, + { + "epoch": 0.13, + "learning_rate": 1.948997509223052e-05, + "loss": 0.7349, + "step": 324 + }, + { + "epoch": 0.13, + "learning_rate": 1.9485889376865998e-05, + "loss": 0.6925, + "step": 325 + }, + { + "epoch": 0.13, + "learning_rate": 1.9481787793482503e-05, + "loss": 0.7556, + "step": 326 + }, + { + "epoch": 0.13, + "learning_rate": 1.9477670348941182e-05, + "loss": 0.7757, + "step": 327 + }, + { + "epoch": 0.13, + "learning_rate": 1.9473537050129704e-05, + "loss": 0.8299, + "step": 328 + }, + { + "epoch": 0.13, + "learning_rate": 1.9469387903962263e-05, + "loss": 0.7939, + "step": 329 + }, + { + "epoch": 0.13, + "learning_rate": 1.946522291737956e-05, + "loss": 0.6782, + "step": 330 + }, + { + "epoch": 0.13, + "learning_rate": 1.9461042097348796e-05, + "loss": 0.7904, + "step": 331 + }, + { + "epoch": 0.13, + "learning_rate": 1.9456845450863654e-05, + "loss": 0.7802, + "step": 332 + }, + { + "epoch": 0.13, + "learning_rate": 1.9452632984944295e-05, + "loss": 0.7697, + "step": 333 + }, + { + "epoch": 0.13, + "learning_rate": 1.9448404706637344e-05, + "loss": 0.7577, + "step": 334 + }, + { + "epoch": 0.13, + "learning_rate": 1.9444160623015877e-05, + "loss": 0.7912, + "step": 335 + }, + { + "epoch": 0.13, + "learning_rate": 1.94399007411794e-05, + "loss": 0.835, + "step": 336 + }, + { + "epoch": 0.13, + "learning_rate": 1.9435625068253863e-05, + "loss": 0.7906, + "step": 337 + }, + { + "epoch": 0.13, + "learning_rate": 1.9431333611391615e-05, + "loss": 0.7534, + "step": 338 + }, + { + "epoch": 0.14, + "learning_rate": 1.942702637777142e-05, + "loss": 0.8155, + "step": 339 + }, + { + "epoch": 0.14, + "learning_rate": 1.9422703374598425e-05, + "loss": 0.7933, + "step": 340 + }, + { + "epoch": 0.14, + "learning_rate": 1.9418364609104166e-05, + "loss": 0.7518, + "step": 341 + }, + { + "epoch": 0.14, + "learning_rate": 1.9414010088546535e-05, + "loss": 0.7631, + "step": 342 + }, + { + "epoch": 0.14, + "learning_rate": 1.940963982020979e-05, + "loss": 0.7871, + "step": 343 + }, + { + "epoch": 0.14, + "learning_rate": 1.9405253811404522e-05, + "loss": 0.8125, + "step": 344 + }, + { + "epoch": 0.14, + "learning_rate": 1.9400852069467657e-05, + "loss": 0.7913, + "step": 345 + }, + { + "epoch": 0.14, + "learning_rate": 1.9396434601762444e-05, + "loss": 0.7237, + "step": 346 + }, + { + "epoch": 0.14, + "learning_rate": 1.939200141567843e-05, + "loss": 0.7541, + "step": 347 + }, + { + "epoch": 0.14, + "learning_rate": 1.938755251863146e-05, + "loss": 0.7267, + "step": 348 + }, + { + "epoch": 0.14, + "learning_rate": 1.9383087918063662e-05, + "loss": 0.7884, + "step": 349 + }, + { + "epoch": 0.14, + "learning_rate": 1.937860762144343e-05, + "loss": 0.6844, + "step": 350 + }, + { + "epoch": 0.14, + "learning_rate": 1.9374111636265412e-05, + "loss": 0.6843, + "step": 351 + }, + { + "epoch": 0.14, + "learning_rate": 1.9369599970050506e-05, + "loss": 0.7885, + "step": 352 + }, + { + "epoch": 0.14, + "learning_rate": 1.9365072630345838e-05, + "loss": 0.7332, + "step": 353 + }, + { + "epoch": 0.14, + "learning_rate": 1.9360529624724754e-05, + "loss": 0.7237, + "step": 354 + }, + { + "epoch": 0.14, + "learning_rate": 1.9355970960786803e-05, + "loss": 0.7703, + "step": 355 + }, + { + "epoch": 0.14, + "learning_rate": 1.935139664615773e-05, + "loss": 0.7239, + "step": 356 + }, + { + "epoch": 0.14, + "learning_rate": 1.934680668848946e-05, + "loss": 0.8067, + "step": 357 + }, + { + "epoch": 0.14, + "learning_rate": 1.9342201095460083e-05, + "loss": 0.7498, + "step": 358 + }, + { + "epoch": 0.14, + "learning_rate": 1.933757987477385e-05, + "loss": 0.7512, + "step": 359 + }, + { + "epoch": 0.14, + "learning_rate": 1.933294303416115e-05, + "loss": 0.8152, + "step": 360 + }, + { + "epoch": 0.14, + "learning_rate": 1.9328290581378497e-05, + "loss": 0.7584, + "step": 361 + }, + { + "epoch": 0.14, + "learning_rate": 1.932362252420853e-05, + "loss": 0.7165, + "step": 362 + }, + { + "epoch": 0.14, + "learning_rate": 1.9318938870459984e-05, + "loss": 0.7321, + "step": 363 + }, + { + "epoch": 0.15, + "learning_rate": 1.9314239627967688e-05, + "loss": 0.8046, + "step": 364 + }, + { + "epoch": 0.15, + "learning_rate": 1.9309524804592545e-05, + "loss": 0.773, + "step": 365 + }, + { + "epoch": 0.15, + "learning_rate": 1.9304794408221524e-05, + "loss": 0.6992, + "step": 366 + }, + { + "epoch": 0.15, + "learning_rate": 1.9300048446767643e-05, + "loss": 0.75, + "step": 367 + }, + { + "epoch": 0.15, + "learning_rate": 1.929528692816996e-05, + "loss": 0.7391, + "step": 368 + }, + { + "epoch": 0.15, + "learning_rate": 1.929050986039355e-05, + "loss": 0.7818, + "step": 369 + }, + { + "epoch": 0.15, + "learning_rate": 1.9285717251429507e-05, + "loss": 0.7756, + "step": 370 + }, + { + "epoch": 0.15, + "learning_rate": 1.928090910929492e-05, + "loss": 0.7725, + "step": 371 + }, + { + "epoch": 0.15, + "learning_rate": 1.927608544203286e-05, + "loss": 0.7654, + "step": 372 + }, + { + "epoch": 0.15, + "learning_rate": 1.9271246257712368e-05, + "loss": 0.7185, + "step": 373 + }, + { + "epoch": 0.15, + "learning_rate": 1.926639156442845e-05, + "loss": 0.7626, + "step": 374 + }, + { + "epoch": 0.15, + "learning_rate": 1.9261521370302035e-05, + "loss": 0.7868, + "step": 375 + }, + { + "epoch": 0.15, + "learning_rate": 1.925663568348001e-05, + "loss": 0.7442, + "step": 376 + }, + { + "epoch": 0.15, + "learning_rate": 1.9251734512135155e-05, + "loss": 0.7183, + "step": 377 + }, + { + "epoch": 0.15, + "learning_rate": 1.9246817864466166e-05, + "loss": 0.8302, + "step": 378 + }, + { + "epoch": 0.15, + "learning_rate": 1.9241885748697618e-05, + "loss": 0.7701, + "step": 379 + }, + { + "epoch": 0.15, + "learning_rate": 1.9236938173079974e-05, + "loss": 0.7761, + "step": 380 + }, + { + "epoch": 0.15, + "learning_rate": 1.9231975145889543e-05, + "loss": 0.8344, + "step": 381 + }, + { + "epoch": 0.15, + "learning_rate": 1.9226996675428497e-05, + "loss": 0.8058, + "step": 382 + }, + { + "epoch": 0.15, + "learning_rate": 1.9222002770024827e-05, + "loss": 0.7717, + "step": 383 + }, + { + "epoch": 0.15, + "learning_rate": 1.921699343803235e-05, + "loss": 0.7319, + "step": 384 + }, + { + "epoch": 0.15, + "learning_rate": 1.9211968687830693e-05, + "loss": 0.6841, + "step": 385 + }, + { + "epoch": 0.15, + "learning_rate": 1.9206928527825268e-05, + "loss": 0.7784, + "step": 386 + }, + { + "epoch": 0.15, + "learning_rate": 1.920187296644727e-05, + "loss": 0.749, + "step": 387 + }, + { + "epoch": 0.15, + "learning_rate": 1.919680201215365e-05, + "loss": 0.7448, + "step": 388 + }, + { + "epoch": 0.16, + "learning_rate": 1.9191715673427113e-05, + "loss": 0.6588, + "step": 389 + }, + { + "epoch": 0.16, + "learning_rate": 1.91866139587761e-05, + "loss": 0.7127, + "step": 390 + }, + { + "epoch": 0.16, + "learning_rate": 1.918149687673478e-05, + "loss": 0.772, + "step": 391 + }, + { + "epoch": 0.16, + "learning_rate": 1.9176364435863005e-05, + "loss": 0.7335, + "step": 392 + }, + { + "epoch": 0.16, + "learning_rate": 1.9171216644746345e-05, + "loss": 0.7289, + "step": 393 + }, + { + "epoch": 0.16, + "learning_rate": 1.9166053511996035e-05, + "loss": 0.6897, + "step": 394 + }, + { + "epoch": 0.16, + "learning_rate": 1.9160875046248974e-05, + "loss": 0.6907, + "step": 395 + }, + { + "epoch": 0.16, + "learning_rate": 1.915568125616772e-05, + "loss": 0.6756, + "step": 396 + }, + { + "epoch": 0.16, + "learning_rate": 1.9150472150440443e-05, + "loss": 0.7353, + "step": 397 + }, + { + "epoch": 0.16, + "learning_rate": 1.9145247737780963e-05, + "loss": 0.6975, + "step": 398 + }, + { + "epoch": 0.16, + "learning_rate": 1.9140008026928683e-05, + "loss": 0.7623, + "step": 399 + }, + { + "epoch": 0.16, + "learning_rate": 1.9134753026648605e-05, + "loss": 0.835, + "step": 400 + }, + { + "epoch": 0.16, + "learning_rate": 1.9129482745731308e-05, + "loss": 0.8391, + "step": 401 + }, + { + "epoch": 0.16, + "learning_rate": 1.9124197192992927e-05, + "loss": 0.7046, + "step": 402 + }, + { + "epoch": 0.16, + "learning_rate": 1.911889637727515e-05, + "loss": 0.788, + "step": 403 + }, + { + "epoch": 0.16, + "learning_rate": 1.9113580307445196e-05, + "loss": 0.7555, + "step": 404 + }, + { + "epoch": 0.16, + "learning_rate": 1.9108248992395797e-05, + "loss": 0.7918, + "step": 405 + }, + { + "epoch": 0.16, + "learning_rate": 1.910290244104519e-05, + "loss": 0.7241, + "step": 406 + }, + { + "epoch": 0.16, + "learning_rate": 1.9097540662337102e-05, + "loss": 0.7386, + "step": 407 + }, + { + "epoch": 0.16, + "learning_rate": 1.9092163665240726e-05, + "loss": 0.7517, + "step": 408 + }, + { + "epoch": 0.16, + "learning_rate": 1.9086771458750716e-05, + "loss": 0.7325, + "step": 409 + }, + { + "epoch": 0.16, + "learning_rate": 1.9081364051887173e-05, + "loss": 0.8032, + "step": 410 + }, + { + "epoch": 0.16, + "learning_rate": 1.9075941453695617e-05, + "loss": 0.7595, + "step": 411 + }, + { + "epoch": 0.16, + "learning_rate": 1.9070503673246982e-05, + "loss": 0.7593, + "step": 412 + }, + { + "epoch": 0.16, + "learning_rate": 1.9065050719637604e-05, + "loss": 0.7957, + "step": 413 + }, + { + "epoch": 0.17, + "learning_rate": 1.9059582601989196e-05, + "loss": 0.7264, + "step": 414 + }, + { + "epoch": 0.17, + "learning_rate": 1.9054099329448835e-05, + "loss": 0.7653, + "step": 415 + }, + { + "epoch": 0.17, + "learning_rate": 1.904860091118896e-05, + "loss": 0.737, + "step": 416 + }, + { + "epoch": 0.17, + "learning_rate": 1.904308735640733e-05, + "loss": 0.7475, + "step": 417 + }, + { + "epoch": 0.17, + "learning_rate": 1.9037558674327036e-05, + "loss": 0.7614, + "step": 418 + }, + { + "epoch": 0.17, + "learning_rate": 1.9032014874196476e-05, + "loss": 0.6836, + "step": 419 + }, + { + "epoch": 0.17, + "learning_rate": 1.9026455965289326e-05, + "loss": 0.7826, + "step": 420 + }, + { + "epoch": 0.17, + "learning_rate": 1.9020881956904543e-05, + "loss": 0.7697, + "step": 421 + }, + { + "epoch": 0.17, + "learning_rate": 1.901529285836635e-05, + "loss": 0.6963, + "step": 422 + }, + { + "epoch": 0.17, + "learning_rate": 1.900968867902419e-05, + "loss": 0.7663, + "step": 423 + }, + { + "epoch": 0.17, + "learning_rate": 1.9004069428252765e-05, + "loss": 0.7662, + "step": 424 + }, + { + "epoch": 0.17, + "learning_rate": 1.899843511545196e-05, + "loss": 0.7335, + "step": 425 + }, + { + "epoch": 0.17, + "learning_rate": 1.8992785750046866e-05, + "loss": 0.8018, + "step": 426 + }, + { + "epoch": 0.17, + "learning_rate": 1.898712134148776e-05, + "loss": 0.7281, + "step": 427 + }, + { + "epoch": 0.17, + "learning_rate": 1.8981441899250082e-05, + "loss": 0.8408, + "step": 428 + }, + { + "epoch": 0.17, + "learning_rate": 1.897574743283441e-05, + "loss": 0.7273, + "step": 429 + }, + { + "epoch": 0.17, + "learning_rate": 1.8970037951766468e-05, + "loss": 0.7767, + "step": 430 + }, + { + "epoch": 0.17, + "learning_rate": 1.896431346559708e-05, + "loss": 0.7087, + "step": 431 + }, + { + "epoch": 0.17, + "learning_rate": 1.895857398390219e-05, + "loss": 0.7702, + "step": 432 + }, + { + "epoch": 0.17, + "learning_rate": 1.895281951628281e-05, + "loss": 0.8321, + "step": 433 + }, + { + "epoch": 0.17, + "learning_rate": 1.8947050072365033e-05, + "loss": 0.7266, + "step": 434 + }, + { + "epoch": 0.17, + "learning_rate": 1.89412656618e-05, + "loss": 0.768, + "step": 435 + }, + { + "epoch": 0.17, + "learning_rate": 1.8935466294263876e-05, + "loss": 0.7135, + "step": 436 + }, + { + "epoch": 0.17, + "learning_rate": 1.8929651979457874e-05, + "loss": 0.7761, + "step": 437 + }, + { + "epoch": 0.17, + "learning_rate": 1.892382272710818e-05, + "loss": 0.804, + "step": 438 + }, + { + "epoch": 0.18, + "learning_rate": 1.891797854696599e-05, + "loss": 0.7114, + "step": 439 + }, + { + "epoch": 0.18, + "learning_rate": 1.891211944880746e-05, + "loss": 0.701, + "step": 440 + }, + { + "epoch": 0.18, + "learning_rate": 1.89062454424337e-05, + "loss": 0.7672, + "step": 441 + }, + { + "epoch": 0.18, + "learning_rate": 1.8900356537670768e-05, + "loss": 0.7597, + "step": 442 + }, + { + "epoch": 0.18, + "learning_rate": 1.8894452744369632e-05, + "loss": 0.7473, + "step": 443 + }, + { + "epoch": 0.18, + "learning_rate": 1.8888534072406182e-05, + "loss": 0.6375, + "step": 444 + }, + { + "epoch": 0.18, + "learning_rate": 1.888260053168118e-05, + "loss": 0.764, + "step": 445 + }, + { + "epoch": 0.18, + "learning_rate": 1.8876652132120267e-05, + "loss": 0.778, + "step": 446 + }, + { + "epoch": 0.18, + "learning_rate": 1.887068888367394e-05, + "loss": 0.7661, + "step": 447 + }, + { + "epoch": 0.18, + "learning_rate": 1.8864710796317537e-05, + "loss": 0.7281, + "step": 448 + }, + { + "epoch": 0.18, + "learning_rate": 1.885871788005122e-05, + "loss": 0.738, + "step": 449 + }, + { + "epoch": 0.18, + "learning_rate": 1.8852710144899948e-05, + "loss": 0.7972, + "step": 450 + }, + { + "epoch": 0.18, + "learning_rate": 1.884668760091348e-05, + "loss": 0.7988, + "step": 451 + }, + { + "epoch": 0.18, + "learning_rate": 1.8840650258166343e-05, + "loss": 0.7064, + "step": 452 + }, + { + "epoch": 0.18, + "learning_rate": 1.8834598126757812e-05, + "loss": 0.7231, + "step": 453 + }, + { + "epoch": 0.18, + "learning_rate": 1.8828531216811912e-05, + "loss": 0.6869, + "step": 454 + }, + { + "epoch": 0.18, + "learning_rate": 1.882244953847739e-05, + "loss": 0.811, + "step": 455 + }, + { + "epoch": 0.18, + "learning_rate": 1.8816353101927686e-05, + "loss": 0.7511, + "step": 456 + }, + { + "epoch": 0.18, + "learning_rate": 1.881024191736093e-05, + "loss": 0.7783, + "step": 457 + }, + { + "epoch": 0.18, + "learning_rate": 1.8804115994999933e-05, + "loss": 0.7196, + "step": 458 + }, + { + "epoch": 0.18, + "learning_rate": 1.8797975345092153e-05, + "loss": 0.7194, + "step": 459 + }, + { + "epoch": 0.18, + "learning_rate": 1.8791819977909682e-05, + "loss": 0.778, + "step": 460 + }, + { + "epoch": 0.18, + "learning_rate": 1.8785649903749236e-05, + "loss": 0.7429, + "step": 461 + }, + { + "epoch": 0.18, + "learning_rate": 1.8779465132932124e-05, + "loss": 0.7562, + "step": 462 + }, + { + "epoch": 0.18, + "learning_rate": 1.877326567580425e-05, + "loss": 0.6943, + "step": 463 + }, + { + "epoch": 0.19, + "learning_rate": 1.8767051542736084e-05, + "loss": 0.7429, + "step": 464 + }, + { + "epoch": 0.19, + "learning_rate": 1.8760822744122637e-05, + "loss": 0.7951, + "step": 465 + }, + { + "epoch": 0.19, + "learning_rate": 1.8754579290383465e-05, + "loss": 0.7621, + "step": 466 + }, + { + "epoch": 0.19, + "learning_rate": 1.8748321191962625e-05, + "loss": 0.7538, + "step": 467 + }, + { + "epoch": 0.19, + "learning_rate": 1.8742048459328684e-05, + "loss": 0.6886, + "step": 468 + }, + { + "epoch": 0.19, + "learning_rate": 1.8735761102974684e-05, + "loss": 0.8096, + "step": 469 + }, + { + "epoch": 0.19, + "learning_rate": 1.872945913341813e-05, + "loss": 0.7334, + "step": 470 + }, + { + "epoch": 0.19, + "learning_rate": 1.8723142561200973e-05, + "loss": 0.7219, + "step": 471 + }, + { + "epoch": 0.19, + "learning_rate": 1.871681139688959e-05, + "loss": 0.7755, + "step": 472 + }, + { + "epoch": 0.19, + "learning_rate": 1.8710465651074766e-05, + "loss": 0.7438, + "step": 473 + }, + { + "epoch": 0.19, + "learning_rate": 1.8704105334371683e-05, + "loss": 0.7262, + "step": 474 + }, + { + "epoch": 0.19, + "learning_rate": 1.8697730457419893e-05, + "loss": 0.7319, + "step": 475 + }, + { + "epoch": 0.19, + "learning_rate": 1.8691341030883306e-05, + "loss": 0.7693, + "step": 476 + }, + { + "epoch": 0.19, + "learning_rate": 1.868493706545017e-05, + "loss": 0.7069, + "step": 477 + }, + { + "epoch": 0.19, + "learning_rate": 1.8678518571833054e-05, + "loss": 0.7373, + "step": 478 + }, + { + "epoch": 0.19, + "learning_rate": 1.867208556076883e-05, + "loss": 0.7819, + "step": 479 + }, + { + "epoch": 0.19, + "learning_rate": 1.866563804301866e-05, + "loss": 0.7415, + "step": 480 + }, + { + "epoch": 0.19, + "learning_rate": 1.8659176029367957e-05, + "loss": 0.7775, + "step": 481 + }, + { + "epoch": 0.19, + "learning_rate": 1.8652699530626398e-05, + "loss": 0.769, + "step": 482 + }, + { + "epoch": 0.19, + "learning_rate": 1.8646208557627888e-05, + "loss": 0.7343, + "step": 483 + }, + { + "epoch": 0.19, + "learning_rate": 1.8639703121230542e-05, + "loss": 0.8211, + "step": 484 + }, + { + "epoch": 0.19, + "learning_rate": 1.8633183232316668e-05, + "loss": 0.7205, + "step": 485 + }, + { + "epoch": 0.19, + "learning_rate": 1.8626648901792753e-05, + "loss": 0.7409, + "step": 486 + }, + { + "epoch": 0.19, + "learning_rate": 1.8620100140589443e-05, + "loss": 0.747, + "step": 487 + }, + { + "epoch": 0.19, + "learning_rate": 1.8613536959661518e-05, + "loss": 0.7558, + "step": 488 + }, + { + "epoch": 0.2, + "learning_rate": 1.8606959369987885e-05, + "loss": 0.7578, + "step": 489 + }, + { + "epoch": 0.2, + "learning_rate": 1.860036738257155e-05, + "loss": 0.7971, + "step": 490 + }, + { + "epoch": 0.2, + "learning_rate": 1.8593761008439614e-05, + "loss": 0.7264, + "step": 491 + }, + { + "epoch": 0.2, + "learning_rate": 1.8587140258643225e-05, + "loss": 0.7748, + "step": 492 + }, + { + "epoch": 0.2, + "learning_rate": 1.8580505144257595e-05, + "loss": 0.7242, + "step": 493 + }, + { + "epoch": 0.2, + "learning_rate": 1.857385567638196e-05, + "loss": 0.7837, + "step": 494 + }, + { + "epoch": 0.2, + "learning_rate": 1.8567191866139564e-05, + "loss": 0.7684, + "step": 495 + }, + { + "epoch": 0.2, + "learning_rate": 1.8560513724677644e-05, + "loss": 0.7384, + "step": 496 + }, + { + "epoch": 0.2, + "learning_rate": 1.8553821263167417e-05, + "loss": 0.7723, + "step": 497 + }, + { + "epoch": 0.2, + "learning_rate": 1.8547114492804042e-05, + "loss": 0.7529, + "step": 498 + }, + { + "epoch": 0.2, + "learning_rate": 1.8540393424806627e-05, + "loss": 0.7458, + "step": 499 + }, + { + "epoch": 0.2, + "learning_rate": 1.8533658070418186e-05, + "loss": 0.7551, + "step": 500 + }, + { + "epoch": 0.2, + "learning_rate": 1.852690844090564e-05, + "loss": 0.7839, + "step": 501 + }, + { + "epoch": 0.2, + "learning_rate": 1.8520144547559783e-05, + "loss": 0.8113, + "step": 502 + }, + { + "epoch": 0.2, + "learning_rate": 1.8513366401695277e-05, + "loss": 0.7797, + "step": 503 + }, + { + "epoch": 0.2, + "learning_rate": 1.8506574014650613e-05, + "loss": 0.6893, + "step": 504 + }, + { + "epoch": 0.2, + "learning_rate": 1.849976739778812e-05, + "loss": 0.7333, + "step": 505 + }, + { + "epoch": 0.2, + "learning_rate": 1.849294656249392e-05, + "loss": 0.8052, + "step": 506 + }, + { + "epoch": 0.2, + "learning_rate": 1.8486111520177925e-05, + "loss": 0.7489, + "step": 507 + }, + { + "epoch": 0.2, + "learning_rate": 1.8479262282273805e-05, + "loss": 0.7933, + "step": 508 + }, + { + "epoch": 0.2, + "learning_rate": 1.8472398860238988e-05, + "loss": 0.8037, + "step": 509 + }, + { + "epoch": 0.2, + "learning_rate": 1.8465521265554623e-05, + "loss": 0.6913, + "step": 510 + }, + { + "epoch": 0.2, + "learning_rate": 1.8458629509725565e-05, + "loss": 0.8008, + "step": 511 + }, + { + "epoch": 0.2, + "learning_rate": 1.845172360428036e-05, + "loss": 0.7656, + "step": 512 + }, + { + "epoch": 0.2, + "learning_rate": 1.8444803560771222e-05, + "loss": 0.7666, + "step": 513 + }, + { + "epoch": 0.21, + "learning_rate": 1.8437869390774022e-05, + "loss": 0.722, + "step": 514 + }, + { + "epoch": 0.21, + "learning_rate": 1.8430921105888254e-05, + "loss": 0.7097, + "step": 515 + }, + { + "epoch": 0.21, + "learning_rate": 1.842395871773702e-05, + "loss": 0.7689, + "step": 516 + }, + { + "epoch": 0.21, + "learning_rate": 1.841698223796703e-05, + "loss": 0.7445, + "step": 517 + }, + { + "epoch": 0.21, + "learning_rate": 1.8409991678248548e-05, + "loss": 0.6917, + "step": 518 + }, + { + "epoch": 0.21, + "learning_rate": 1.8402987050275408e-05, + "loss": 0.6971, + "step": 519 + }, + { + "epoch": 0.21, + "learning_rate": 1.839596836576496e-05, + "loss": 0.7321, + "step": 520 + }, + { + "epoch": 0.21, + "learning_rate": 1.838893563645808e-05, + "loss": 0.7042, + "step": 521 + }, + { + "epoch": 0.21, + "learning_rate": 1.8381888874119135e-05, + "loss": 0.7571, + "step": 522 + }, + { + "epoch": 0.21, + "learning_rate": 1.837482809053597e-05, + "loss": 0.7878, + "step": 523 + }, + { + "epoch": 0.21, + "learning_rate": 1.8367753297519874e-05, + "loss": 0.7135, + "step": 524 + }, + { + "epoch": 0.21, + "learning_rate": 1.8360664506905587e-05, + "loss": 0.7129, + "step": 525 + }, + { + "epoch": 0.21, + "learning_rate": 1.8353561730551243e-05, + "loss": 0.7638, + "step": 526 + }, + { + "epoch": 0.21, + "learning_rate": 1.8346444980338394e-05, + "loss": 0.7766, + "step": 527 + }, + { + "epoch": 0.21, + "learning_rate": 1.8339314268171953e-05, + "loss": 0.7465, + "step": 528 + }, + { + "epoch": 0.21, + "learning_rate": 1.8332169605980193e-05, + "loss": 0.7516, + "step": 529 + }, + { + "epoch": 0.21, + "learning_rate": 1.8325011005714726e-05, + "loss": 0.6397, + "step": 530 + }, + { + "epoch": 0.21, + "learning_rate": 1.8317838479350473e-05, + "loss": 0.7531, + "step": 531 + }, + { + "epoch": 0.21, + "learning_rate": 1.831065203888565e-05, + "loss": 0.7786, + "step": 532 + }, + { + "epoch": 0.21, + "learning_rate": 1.830345169634176e-05, + "loss": 0.6881, + "step": 533 + }, + { + "epoch": 0.21, + "learning_rate": 1.8296237463763543e-05, + "loss": 0.715, + "step": 534 + }, + { + "epoch": 0.21, + "learning_rate": 1.8289009353218995e-05, + "loss": 0.6861, + "step": 535 + }, + { + "epoch": 0.21, + "learning_rate": 1.8281767376799315e-05, + "loss": 0.7402, + "step": 536 + }, + { + "epoch": 0.21, + "learning_rate": 1.82745115466189e-05, + "loss": 0.7246, + "step": 537 + }, + { + "epoch": 0.21, + "learning_rate": 1.8267241874815312e-05, + "loss": 0.7718, + "step": 538 + }, + { + "epoch": 0.22, + "learning_rate": 1.8259958373549285e-05, + "loss": 0.7599, + "step": 539 + }, + { + "epoch": 0.22, + "learning_rate": 1.8252661055004672e-05, + "loss": 0.7761, + "step": 540 + }, + { + "epoch": 0.22, + "learning_rate": 1.824534993138845e-05, + "loss": 0.6704, + "step": 541 + }, + { + "epoch": 0.22, + "learning_rate": 1.823802501493068e-05, + "loss": 0.7119, + "step": 542 + }, + { + "epoch": 0.22, + "learning_rate": 1.8230686317884502e-05, + "loss": 0.7753, + "step": 543 + }, + { + "epoch": 0.22, + "learning_rate": 1.8223333852526107e-05, + "loss": 0.7905, + "step": 544 + }, + { + "epoch": 0.22, + "learning_rate": 1.8215967631154718e-05, + "loss": 0.6533, + "step": 545 + }, + { + "epoch": 0.22, + "learning_rate": 1.8208587666092566e-05, + "loss": 0.7377, + "step": 546 + }, + { + "epoch": 0.22, + "learning_rate": 1.8201193969684875e-05, + "loss": 0.6425, + "step": 547 + }, + { + "epoch": 0.22, + "learning_rate": 1.819378655429984e-05, + "loss": 0.7681, + "step": 548 + }, + { + "epoch": 0.22, + "learning_rate": 1.8186365432328605e-05, + "loss": 0.7723, + "step": 549 + }, + { + "epoch": 0.22, + "learning_rate": 1.8178930616185236e-05, + "loss": 0.6642, + "step": 550 + }, + { + "epoch": 0.22, + "learning_rate": 1.8171482118306712e-05, + "loss": 0.7154, + "step": 551 + }, + { + "epoch": 0.22, + "learning_rate": 1.8164019951152903e-05, + "loss": 0.739, + "step": 552 + }, + { + "epoch": 0.22, + "learning_rate": 1.8156544127206536e-05, + "loss": 0.6675, + "step": 553 + }, + { + "epoch": 0.22, + "learning_rate": 1.8149054658973194e-05, + "loss": 0.7325, + "step": 554 + }, + { + "epoch": 0.22, + "learning_rate": 1.814155155898127e-05, + "loss": 0.7725, + "step": 555 + }, + { + "epoch": 0.22, + "learning_rate": 1.8134034839781972e-05, + "loss": 0.7817, + "step": 556 + }, + { + "epoch": 0.22, + "learning_rate": 1.8126504513949287e-05, + "loss": 0.7206, + "step": 557 + }, + { + "epoch": 0.22, + "learning_rate": 1.811896059407996e-05, + "loss": 0.7656, + "step": 558 + }, + { + "epoch": 0.22, + "learning_rate": 1.811140309279348e-05, + "loss": 0.7385, + "step": 559 + }, + { + "epoch": 0.22, + "learning_rate": 1.8103832022732056e-05, + "loss": 0.7615, + "step": 560 + }, + { + "epoch": 0.22, + "learning_rate": 1.8096247396560585e-05, + "loss": 0.7303, + "step": 561 + }, + { + "epoch": 0.22, + "learning_rate": 1.8088649226966658e-05, + "loss": 0.7372, + "step": 562 + }, + { + "epoch": 0.22, + "learning_rate": 1.8081037526660505e-05, + "loss": 0.7408, + "step": 563 + }, + { + "epoch": 0.23, + "learning_rate": 1.8073412308375e-05, + "loss": 0.7297, + "step": 564 + }, + { + "epoch": 0.23, + "learning_rate": 1.8065773584865623e-05, + "loss": 0.8023, + "step": 565 + }, + { + "epoch": 0.23, + "learning_rate": 1.805812136891046e-05, + "loss": 0.7318, + "step": 566 + }, + { + "epoch": 0.23, + "learning_rate": 1.8050455673310142e-05, + "loss": 0.7345, + "step": 567 + }, + { + "epoch": 0.23, + "learning_rate": 1.804277651088787e-05, + "loss": 0.795, + "step": 568 + }, + { + "epoch": 0.23, + "learning_rate": 1.8035083894489364e-05, + "loss": 0.7147, + "step": 569 + }, + { + "epoch": 0.23, + "learning_rate": 1.8027377836982857e-05, + "loss": 0.7494, + "step": 570 + }, + { + "epoch": 0.23, + "learning_rate": 1.8019658351259056e-05, + "loss": 0.7274, + "step": 571 + }, + { + "epoch": 0.23, + "learning_rate": 1.8011925450231135e-05, + "loss": 0.7522, + "step": 572 + }, + { + "epoch": 0.23, + "learning_rate": 1.8004179146834712e-05, + "loss": 0.7966, + "step": 573 + }, + { + "epoch": 0.23, + "learning_rate": 1.7996419454027828e-05, + "loss": 0.7288, + "step": 574 + }, + { + "epoch": 0.23, + "learning_rate": 1.79886463847909e-05, + "loss": 0.7354, + "step": 575 + }, + { + "epoch": 0.23, + "learning_rate": 1.7980859952126755e-05, + "loss": 0.7446, + "step": 576 + }, + { + "epoch": 0.23, + "learning_rate": 1.7973060169060546e-05, + "loss": 0.7817, + "step": 577 + }, + { + "epoch": 0.23, + "learning_rate": 1.796524704863977e-05, + "loss": 0.7171, + "step": 578 + }, + { + "epoch": 0.23, + "learning_rate": 1.795742060393424e-05, + "loss": 0.7084, + "step": 579 + }, + { + "epoch": 0.23, + "learning_rate": 1.7949580848036045e-05, + "loss": 0.6413, + "step": 580 + }, + { + "epoch": 0.23, + "learning_rate": 1.7941727794059553e-05, + "loss": 0.7867, + "step": 581 + }, + { + "epoch": 0.23, + "learning_rate": 1.7933861455141365e-05, + "loss": 0.7014, + "step": 582 + }, + { + "epoch": 0.23, + "learning_rate": 1.792598184444032e-05, + "loss": 0.6806, + "step": 583 + }, + { + "epoch": 0.23, + "learning_rate": 1.7918088975137445e-05, + "loss": 0.765, + "step": 584 + }, + { + "epoch": 0.23, + "learning_rate": 1.7910182860435954e-05, + "loss": 0.7644, + "step": 585 + }, + { + "epoch": 0.23, + "learning_rate": 1.7902263513561214e-05, + "loss": 0.7475, + "step": 586 + }, + { + "epoch": 0.23, + "learning_rate": 1.7894330947760728e-05, + "loss": 0.8067, + "step": 587 + }, + { + "epoch": 0.23, + "learning_rate": 1.788638517630411e-05, + "loss": 0.7561, + "step": 588 + }, + { + "epoch": 0.24, + "learning_rate": 1.7878426212483066e-05, + "loss": 0.7512, + "step": 589 + }, + { + "epoch": 0.24, + "learning_rate": 1.7870454069611375e-05, + "loss": 0.8004, + "step": 590 + }, + { + "epoch": 0.24, + "learning_rate": 1.786246876102485e-05, + "loss": 0.6689, + "step": 591 + }, + { + "epoch": 0.24, + "learning_rate": 1.7854470300081338e-05, + "loss": 0.7269, + "step": 592 + }, + { + "epoch": 0.24, + "learning_rate": 1.7846458700160685e-05, + "loss": 0.6416, + "step": 593 + }, + { + "epoch": 0.24, + "learning_rate": 1.7838433974664714e-05, + "loss": 0.7942, + "step": 594 + }, + { + "epoch": 0.24, + "learning_rate": 1.7830396137017202e-05, + "loss": 0.7733, + "step": 595 + }, + { + "epoch": 0.24, + "learning_rate": 1.7822345200663874e-05, + "loss": 0.7458, + "step": 596 + }, + { + "epoch": 0.24, + "learning_rate": 1.7814281179072345e-05, + "loss": 0.7177, + "step": 597 + }, + { + "epoch": 0.24, + "learning_rate": 1.780620408573213e-05, + "loss": 0.6977, + "step": 598 + }, + { + "epoch": 0.24, + "learning_rate": 1.7798113934154623e-05, + "loss": 0.7716, + "step": 599 + }, + { + "epoch": 0.24, + "learning_rate": 1.7790010737873035e-05, + "loss": 0.7018, + "step": 600 + }, + { + "epoch": 0.24, + "learning_rate": 1.778189451044242e-05, + "loss": 0.7459, + "step": 601 + }, + { + "epoch": 0.24, + "learning_rate": 1.777376526543962e-05, + "loss": 0.7979, + "step": 602 + }, + { + "epoch": 0.24, + "learning_rate": 1.776562301646326e-05, + "loss": 0.7275, + "step": 603 + }, + { + "epoch": 0.24, + "learning_rate": 1.775746777713371e-05, + "loss": 0.7528, + "step": 604 + }, + { + "epoch": 0.24, + "learning_rate": 1.7749299561093076e-05, + "loss": 0.7484, + "step": 605 + }, + { + "epoch": 0.24, + "learning_rate": 1.7741118382005172e-05, + "loss": 0.7497, + "step": 606 + }, + { + "epoch": 0.24, + "learning_rate": 1.7732924253555487e-05, + "loss": 0.6986, + "step": 607 + }, + { + "epoch": 0.24, + "learning_rate": 1.772471718945119e-05, + "loss": 0.698, + "step": 608 + }, + { + "epoch": 0.24, + "learning_rate": 1.7716497203421072e-05, + "loss": 0.7101, + "step": 609 + }, + { + "epoch": 0.24, + "learning_rate": 1.7708264309215543e-05, + "loss": 0.6875, + "step": 610 + }, + { + "epoch": 0.24, + "learning_rate": 1.770001852060661e-05, + "loss": 0.7388, + "step": 611 + }, + { + "epoch": 0.24, + "learning_rate": 1.7691759851387852e-05, + "loss": 0.775, + "step": 612 + }, + { + "epoch": 0.24, + "learning_rate": 1.7683488315374386e-05, + "loss": 0.6943, + "step": 613 + }, + { + "epoch": 0.25, + "learning_rate": 1.7675203926402863e-05, + "loss": 0.7513, + "step": 614 + }, + { + "epoch": 0.25, + "learning_rate": 1.7666906698331426e-05, + "loss": 0.769, + "step": 615 + }, + { + "epoch": 0.25, + "learning_rate": 1.76585966450397e-05, + "loss": 0.6769, + "step": 616 + }, + { + "epoch": 0.25, + "learning_rate": 1.765027378042876e-05, + "loss": 0.6702, + "step": 617 + }, + { + "epoch": 0.25, + "learning_rate": 1.764193811842112e-05, + "loss": 0.7519, + "step": 618 + }, + { + "epoch": 0.25, + "learning_rate": 1.763358967296069e-05, + "loss": 0.7719, + "step": 619 + }, + { + "epoch": 0.25, + "learning_rate": 1.762522845801277e-05, + "loss": 0.7817, + "step": 620 + }, + { + "epoch": 0.25, + "learning_rate": 1.7616854487564024e-05, + "loss": 0.7633, + "step": 621 + }, + { + "epoch": 0.25, + "learning_rate": 1.7608467775622448e-05, + "loss": 0.6792, + "step": 622 + }, + { + "epoch": 0.25, + "learning_rate": 1.7600068336217355e-05, + "loss": 0.7423, + "step": 623 + }, + { + "epoch": 0.25, + "learning_rate": 1.759165618339935e-05, + "loss": 0.7373, + "step": 624 + }, + { + "epoch": 0.25, + "learning_rate": 1.7583231331240297e-05, + "loss": 0.7439, + "step": 625 + }, + { + "epoch": 0.25, + "learning_rate": 1.7574793793833312e-05, + "loss": 0.7567, + "step": 626 + }, + { + "epoch": 0.25, + "learning_rate": 1.7566343585292732e-05, + "loss": 0.7741, + "step": 627 + }, + { + "epoch": 0.25, + "learning_rate": 1.7557880719754082e-05, + "loss": 0.6776, + "step": 628 + }, + { + "epoch": 0.25, + "learning_rate": 1.7549405211374072e-05, + "loss": 0.7383, + "step": 629 + }, + { + "epoch": 0.25, + "learning_rate": 1.7540917074330543e-05, + "loss": 0.7796, + "step": 630 + }, + { + "epoch": 0.25, + "learning_rate": 1.7532416322822473e-05, + "loss": 0.7773, + "step": 631 + }, + { + "epoch": 0.25, + "learning_rate": 1.7523902971069944e-05, + "loss": 0.6947, + "step": 632 + }, + { + "epoch": 0.25, + "learning_rate": 1.7515377033314113e-05, + "loss": 0.7188, + "step": 633 + }, + { + "epoch": 0.25, + "learning_rate": 1.7506838523817183e-05, + "loss": 0.8095, + "step": 634 + }, + { + "epoch": 0.25, + "learning_rate": 1.7498287456862398e-05, + "loss": 0.7519, + "step": 635 + }, + { + "epoch": 0.25, + "learning_rate": 1.7489723846754e-05, + "loss": 0.6933, + "step": 636 + }, + { + "epoch": 0.25, + "learning_rate": 1.7481147707817227e-05, + "loss": 0.68, + "step": 637 + }, + { + "epoch": 0.25, + "learning_rate": 1.7472559054398256e-05, + "loss": 0.7605, + "step": 638 + }, + { + "epoch": 0.26, + "learning_rate": 1.746395790086421e-05, + "loss": 0.7019, + "step": 639 + }, + { + "epoch": 0.26, + "learning_rate": 1.7455344261603116e-05, + "loss": 0.6643, + "step": 640 + }, + { + "epoch": 0.26, + "learning_rate": 1.7446718151023894e-05, + "loss": 0.7476, + "step": 641 + }, + { + "epoch": 0.26, + "learning_rate": 1.7438079583556326e-05, + "loss": 0.71, + "step": 642 + }, + { + "epoch": 0.26, + "learning_rate": 1.7429428573651023e-05, + "loss": 0.6693, + "step": 643 + }, + { + "epoch": 0.26, + "learning_rate": 1.742076513577942e-05, + "loss": 0.7234, + "step": 644 + }, + { + "epoch": 0.26, + "learning_rate": 1.7412089284433734e-05, + "loss": 0.7251, + "step": 645 + }, + { + "epoch": 0.26, + "learning_rate": 1.7403401034126954e-05, + "loss": 0.7391, + "step": 646 + }, + { + "epoch": 0.26, + "learning_rate": 1.7394700399392804e-05, + "loss": 0.7672, + "step": 647 + }, + { + "epoch": 0.26, + "learning_rate": 1.7385987394785723e-05, + "loss": 0.7922, + "step": 648 + }, + { + "epoch": 0.26, + "learning_rate": 1.7377262034880854e-05, + "loss": 0.7401, + "step": 649 + }, + { + "epoch": 0.26, + "learning_rate": 1.7368524334273998e-05, + "loss": 0.7217, + "step": 650 + }, + { + "epoch": 0.26, + "learning_rate": 1.7359774307581602e-05, + "loss": 0.726, + "step": 651 + }, + { + "epoch": 0.26, + "learning_rate": 1.7351011969440728e-05, + "loss": 0.7817, + "step": 652 + }, + { + "epoch": 0.26, + "learning_rate": 1.734223733450904e-05, + "loss": 0.7589, + "step": 653 + }, + { + "epoch": 0.26, + "learning_rate": 1.733345041746477e-05, + "loss": 0.7265, + "step": 654 + }, + { + "epoch": 0.26, + "learning_rate": 1.732465123300669e-05, + "loss": 0.6829, + "step": 655 + }, + { + "epoch": 0.26, + "learning_rate": 1.7315839795854103e-05, + "loss": 0.7399, + "step": 656 + }, + { + "epoch": 0.26, + "learning_rate": 1.73070161207468e-05, + "loss": 0.7467, + "step": 657 + }, + { + "epoch": 0.26, + "learning_rate": 1.7298180222445043e-05, + "loss": 0.7532, + "step": 658 + }, + { + "epoch": 0.26, + "learning_rate": 1.7289332115729554e-05, + "loss": 0.7153, + "step": 659 + }, + { + "epoch": 0.26, + "learning_rate": 1.7280471815401453e-05, + "loss": 0.695, + "step": 660 + }, + { + "epoch": 0.26, + "learning_rate": 1.7271599336282285e-05, + "loss": 0.7498, + "step": 661 + }, + { + "epoch": 0.26, + "learning_rate": 1.7262714693213948e-05, + "loss": 0.705, + "step": 662 + }, + { + "epoch": 0.26, + "learning_rate": 1.7253817901058698e-05, + "loss": 0.7947, + "step": 663 + }, + { + "epoch": 0.27, + "learning_rate": 1.7244908974699112e-05, + "loss": 0.6685, + "step": 664 + }, + { + "epoch": 0.27, + "learning_rate": 1.7235987929038056e-05, + "loss": 0.7129, + "step": 665 + }, + { + "epoch": 0.27, + "learning_rate": 1.7227054778998682e-05, + "loss": 0.7036, + "step": 666 + }, + { + "epoch": 0.27, + "learning_rate": 1.721810953952439e-05, + "loss": 0.7373, + "step": 667 + }, + { + "epoch": 0.27, + "learning_rate": 1.720915222557879e-05, + "loss": 0.7118, + "step": 668 + }, + { + "epoch": 0.27, + "learning_rate": 1.7200182852145705e-05, + "loss": 0.6344, + "step": 669 + }, + { + "epoch": 0.27, + "learning_rate": 1.7191201434229118e-05, + "loss": 0.7264, + "step": 670 + }, + { + "epoch": 0.27, + "learning_rate": 1.7182207986853176e-05, + "loss": 0.7638, + "step": 671 + }, + { + "epoch": 0.27, + "learning_rate": 1.7173202525062133e-05, + "loss": 0.7011, + "step": 672 + }, + { + "epoch": 0.27, + "learning_rate": 1.716418506392035e-05, + "loss": 0.7662, + "step": 673 + }, + { + "epoch": 0.27, + "learning_rate": 1.7155155618512255e-05, + "loss": 0.7271, + "step": 674 + }, + { + "epoch": 0.27, + "learning_rate": 1.7146114203942332e-05, + "loss": 0.7562, + "step": 675 + }, + { + "epoch": 0.27, + "learning_rate": 1.7137060835335076e-05, + "loss": 0.7919, + "step": 676 + }, + { + "epoch": 0.27, + "learning_rate": 1.712799552783499e-05, + "loss": 0.7463, + "step": 677 + }, + { + "epoch": 0.27, + "learning_rate": 1.7118918296606536e-05, + "loss": 0.7298, + "step": 678 + }, + { + "epoch": 0.27, + "learning_rate": 1.7109829156834133e-05, + "loss": 0.6717, + "step": 679 + }, + { + "epoch": 0.27, + "learning_rate": 1.7100728123722115e-05, + "loss": 0.7353, + "step": 680 + }, + { + "epoch": 0.27, + "learning_rate": 1.7091615212494714e-05, + "loss": 0.7525, + "step": 681 + }, + { + "epoch": 0.27, + "learning_rate": 1.7082490438396027e-05, + "loss": 0.7923, + "step": 682 + }, + { + "epoch": 0.27, + "learning_rate": 1.707335381669e-05, + "loss": 0.7602, + "step": 683 + }, + { + "epoch": 0.27, + "learning_rate": 1.7064205362660397e-05, + "loss": 0.7548, + "step": 684 + }, + { + "epoch": 0.27, + "learning_rate": 1.7055045091610772e-05, + "loss": 0.7965, + "step": 685 + }, + { + "epoch": 0.27, + "learning_rate": 1.7045873018864452e-05, + "loss": 0.6743, + "step": 686 + }, + { + "epoch": 0.27, + "learning_rate": 1.70366891597645e-05, + "loss": 0.7396, + "step": 687 + }, + { + "epoch": 0.27, + "learning_rate": 1.70274935296737e-05, + "loss": 0.7722, + "step": 688 + }, + { + "epoch": 0.28, + "learning_rate": 1.701828614397452e-05, + "loss": 0.7542, + "step": 689 + }, + { + "epoch": 0.28, + "learning_rate": 1.7009067018069106e-05, + "loss": 0.6523, + "step": 690 + }, + { + "epoch": 0.28, + "learning_rate": 1.6999836167379228e-05, + "loss": 0.7061, + "step": 691 + }, + { + "epoch": 0.28, + "learning_rate": 1.6990593607346276e-05, + "loss": 0.7394, + "step": 692 + }, + { + "epoch": 0.28, + "learning_rate": 1.698133935343123e-05, + "loss": 0.7396, + "step": 693 + }, + { + "epoch": 0.28, + "learning_rate": 1.6972073421114626e-05, + "loss": 0.7186, + "step": 694 + }, + { + "epoch": 0.28, + "learning_rate": 1.6962795825896537e-05, + "loss": 0.7011, + "step": 695 + }, + { + "epoch": 0.28, + "learning_rate": 1.695350658329655e-05, + "loss": 0.7761, + "step": 696 + }, + { + "epoch": 0.28, + "learning_rate": 1.6944205708853735e-05, + "loss": 0.7179, + "step": 697 + }, + { + "epoch": 0.28, + "learning_rate": 1.6934893218126607e-05, + "loss": 0.7114, + "step": 698 + }, + { + "epoch": 0.28, + "learning_rate": 1.6925569126693135e-05, + "loss": 0.7096, + "step": 699 + }, + { + "epoch": 0.28, + "learning_rate": 1.6916233450150675e-05, + "loss": 0.7638, + "step": 700 + }, + { + "epoch": 0.28, + "learning_rate": 1.6906886204115974e-05, + "loss": 0.7486, + "step": 701 + }, + { + "epoch": 0.28, + "learning_rate": 1.6897527404225128e-05, + "loss": 0.7269, + "step": 702 + }, + { + "epoch": 0.28, + "learning_rate": 1.6888157066133555e-05, + "loss": 0.789, + "step": 703 + }, + { + "epoch": 0.28, + "learning_rate": 1.6878775205515983e-05, + "loss": 0.7078, + "step": 704 + }, + { + "epoch": 0.28, + "learning_rate": 1.6869381838066412e-05, + "loss": 0.7141, + "step": 705 + }, + { + "epoch": 0.28, + "learning_rate": 1.685997697949809e-05, + "loss": 0.7488, + "step": 706 + }, + { + "epoch": 0.28, + "learning_rate": 1.6850560645543485e-05, + "loss": 0.7192, + "step": 707 + }, + { + "epoch": 0.28, + "learning_rate": 1.684113285195427e-05, + "loss": 0.7769, + "step": 708 + }, + { + "epoch": 0.28, + "learning_rate": 1.6831693614501272e-05, + "loss": 0.6811, + "step": 709 + }, + { + "epoch": 0.28, + "learning_rate": 1.6822242948974475e-05, + "loss": 0.7344, + "step": 710 + }, + { + "epoch": 0.28, + "learning_rate": 1.6812780871182976e-05, + "loss": 0.6152, + "step": 711 + }, + { + "epoch": 0.28, + "learning_rate": 1.6803307396954962e-05, + "loss": 0.7213, + "step": 712 + }, + { + "epoch": 0.28, + "learning_rate": 1.6793822542137683e-05, + "loss": 0.7073, + "step": 713 + }, + { + "epoch": 0.29, + "learning_rate": 1.6784326322597427e-05, + "loss": 0.6996, + "step": 714 + }, + { + "epoch": 0.29, + "learning_rate": 1.6774818754219495e-05, + "loss": 0.7209, + "step": 715 + }, + { + "epoch": 0.29, + "learning_rate": 1.676529985290817e-05, + "loss": 0.6318, + "step": 716 + }, + { + "epoch": 0.29, + "learning_rate": 1.675576963458669e-05, + "loss": 0.7636, + "step": 717 + }, + { + "epoch": 0.29, + "learning_rate": 1.674622811519723e-05, + "loss": 0.7564, + "step": 718 + }, + { + "epoch": 0.29, + "learning_rate": 1.673667531070087e-05, + "loss": 0.8221, + "step": 719 + }, + { + "epoch": 0.29, + "learning_rate": 1.672711123707756e-05, + "loss": 0.7961, + "step": 720 + }, + { + "epoch": 0.29, + "learning_rate": 1.6717535910326108e-05, + "loss": 0.7032, + "step": 721 + }, + { + "epoch": 0.29, + "learning_rate": 1.6707949346464147e-05, + "loss": 0.7394, + "step": 722 + }, + { + "epoch": 0.29, + "learning_rate": 1.66983515615281e-05, + "loss": 0.7069, + "step": 723 + }, + { + "epoch": 0.29, + "learning_rate": 1.6688742571573167e-05, + "loss": 0.7781, + "step": 724 + }, + { + "epoch": 0.29, + "learning_rate": 1.6679122392673284e-05, + "loss": 0.689, + "step": 725 + }, + { + "epoch": 0.29, + "learning_rate": 1.6669491040921123e-05, + "loss": 0.6727, + "step": 726 + }, + { + "epoch": 0.29, + "learning_rate": 1.665984853242802e-05, + "loss": 0.7625, + "step": 727 + }, + { + "epoch": 0.29, + "learning_rate": 1.6650194883323995e-05, + "loss": 0.672, + "step": 728 + }, + { + "epoch": 0.29, + "learning_rate": 1.664053010975769e-05, + "loss": 0.8106, + "step": 729 + }, + { + "epoch": 0.29, + "learning_rate": 1.6630854227896364e-05, + "loss": 0.7444, + "step": 730 + }, + { + "epoch": 0.29, + "learning_rate": 1.6621167253925857e-05, + "loss": 0.7207, + "step": 731 + }, + { + "epoch": 0.29, + "learning_rate": 1.6611469204050566e-05, + "loss": 0.735, + "step": 732 + }, + { + "epoch": 0.29, + "learning_rate": 1.6601760094493408e-05, + "loss": 0.7443, + "step": 733 + }, + { + "epoch": 0.29, + "learning_rate": 1.6592039941495803e-05, + "loss": 0.719, + "step": 734 + }, + { + "epoch": 0.29, + "learning_rate": 1.6582308761317652e-05, + "loss": 0.757, + "step": 735 + }, + { + "epoch": 0.29, + "learning_rate": 1.6572566570237293e-05, + "loss": 0.697, + "step": 736 + }, + { + "epoch": 0.29, + "learning_rate": 1.656281338455149e-05, + "loss": 0.7055, + "step": 737 + }, + { + "epoch": 0.29, + "learning_rate": 1.6553049220575392e-05, + "loss": 0.7871, + "step": 738 + }, + { + "epoch": 0.3, + "learning_rate": 1.6543274094642522e-05, + "loss": 0.7265, + "step": 739 + }, + { + "epoch": 0.3, + "learning_rate": 1.6533488023104732e-05, + "loss": 0.8002, + "step": 740 + }, + { + "epoch": 0.3, + "learning_rate": 1.6523691022332184e-05, + "loss": 0.6925, + "step": 741 + }, + { + "epoch": 0.3, + "learning_rate": 1.6513883108713326e-05, + "loss": 0.811, + "step": 742 + }, + { + "epoch": 0.3, + "learning_rate": 1.6504064298654864e-05, + "loss": 0.748, + "step": 743 + }, + { + "epoch": 0.3, + "learning_rate": 1.6494234608581722e-05, + "loss": 0.7385, + "step": 744 + }, + { + "epoch": 0.3, + "learning_rate": 1.6484394054937033e-05, + "loss": 0.6935, + "step": 745 + }, + { + "epoch": 0.3, + "learning_rate": 1.64745426541821e-05, + "loss": 0.7673, + "step": 746 + }, + { + "epoch": 0.3, + "learning_rate": 1.6464680422796367e-05, + "loss": 0.6886, + "step": 747 + }, + { + "epoch": 0.3, + "learning_rate": 1.64548073772774e-05, + "loss": 0.7274, + "step": 748 + }, + { + "epoch": 0.3, + "learning_rate": 1.6444923534140856e-05, + "loss": 0.6874, + "step": 749 + }, + { + "epoch": 0.3, + "learning_rate": 1.643502890992045e-05, + "loss": 0.8254, + "step": 750 + }, + { + "epoch": 0.3, + "learning_rate": 1.642512352116793e-05, + "loss": 0.7353, + "step": 751 + }, + { + "epoch": 0.3, + "learning_rate": 1.641520738445306e-05, + "loss": 0.7479, + "step": 752 + }, + { + "epoch": 0.3, + "learning_rate": 1.6405280516363576e-05, + "loss": 0.7013, + "step": 753 + }, + { + "epoch": 0.3, + "learning_rate": 1.639534293350517e-05, + "loss": 0.7278, + "step": 754 + }, + { + "epoch": 0.3, + "learning_rate": 1.6385394652501444e-05, + "loss": 0.713, + "step": 755 + }, + { + "epoch": 0.3, + "learning_rate": 1.637543568999392e-05, + "loss": 0.7211, + "step": 756 + }, + { + "epoch": 0.3, + "learning_rate": 1.6365466062641974e-05, + "loss": 0.7149, + "step": 757 + }, + { + "epoch": 0.3, + "learning_rate": 1.6355485787122814e-05, + "loss": 0.8077, + "step": 758 + }, + { + "epoch": 0.3, + "learning_rate": 1.6345494880131476e-05, + "loss": 0.6856, + "step": 759 + }, + { + "epoch": 0.3, + "learning_rate": 1.633549335838077e-05, + "loss": 0.7151, + "step": 760 + }, + { + "epoch": 0.3, + "learning_rate": 1.632548123860127e-05, + "loss": 0.6993, + "step": 761 + }, + { + "epoch": 0.3, + "learning_rate": 1.6315458537541272e-05, + "loss": 0.6942, + "step": 762 + }, + { + "epoch": 0.3, + "learning_rate": 1.6305425271966772e-05, + "loss": 0.7544, + "step": 763 + }, + { + "epoch": 0.3, + "learning_rate": 1.6295381458661447e-05, + "loss": 0.7041, + "step": 764 + }, + { + "epoch": 0.31, + "learning_rate": 1.6285327114426608e-05, + "loss": 0.7259, + "step": 765 + }, + { + "epoch": 0.31, + "learning_rate": 1.627526225608119e-05, + "loss": 0.7307, + "step": 766 + }, + { + "epoch": 0.31, + "learning_rate": 1.6265186900461715e-05, + "loss": 0.6805, + "step": 767 + }, + { + "epoch": 0.31, + "learning_rate": 1.6255101064422252e-05, + "loss": 0.6151, + "step": 768 + }, + { + "epoch": 0.31, + "learning_rate": 1.6245004764834423e-05, + "loss": 0.6578, + "step": 769 + }, + { + "epoch": 0.31, + "learning_rate": 1.6234898018587336e-05, + "loss": 0.7607, + "step": 770 + }, + { + "epoch": 0.31, + "learning_rate": 1.6224780842587586e-05, + "loss": 0.7222, + "step": 771 + }, + { + "epoch": 0.31, + "learning_rate": 1.6214653253759206e-05, + "loss": 0.7214, + "step": 772 + }, + { + "epoch": 0.31, + "learning_rate": 1.6204515269043653e-05, + "loss": 0.795, + "step": 773 + }, + { + "epoch": 0.31, + "learning_rate": 1.619436690539977e-05, + "loss": 0.7407, + "step": 774 + }, + { + "epoch": 0.31, + "learning_rate": 1.6184208179803772e-05, + "loss": 0.7442, + "step": 775 + }, + { + "epoch": 0.31, + "learning_rate": 1.617403910924919e-05, + "loss": 0.6876, + "step": 776 + }, + { + "epoch": 0.31, + "learning_rate": 1.6163859710746878e-05, + "loss": 0.7289, + "step": 777 + }, + { + "epoch": 0.31, + "learning_rate": 1.6153670001324955e-05, + "loss": 0.6998, + "step": 778 + }, + { + "epoch": 0.31, + "learning_rate": 1.6143469998028792e-05, + "loss": 0.7277, + "step": 779 + }, + { + "epoch": 0.31, + "learning_rate": 1.613325971792098e-05, + "loss": 0.6848, + "step": 780 + }, + { + "epoch": 0.31, + "learning_rate": 1.6123039178081303e-05, + "loss": 0.7825, + "step": 781 + }, + { + "epoch": 0.31, + "learning_rate": 1.61128083956067e-05, + "loss": 0.7389, + "step": 782 + }, + { + "epoch": 0.31, + "learning_rate": 1.610256738761125e-05, + "loss": 0.7472, + "step": 783 + }, + { + "epoch": 0.31, + "learning_rate": 1.609231617122614e-05, + "loss": 0.7298, + "step": 784 + }, + { + "epoch": 0.31, + "learning_rate": 1.6082054763599627e-05, + "loss": 0.7048, + "step": 785 + }, + { + "epoch": 0.31, + "learning_rate": 1.607178318189702e-05, + "loss": 0.76, + "step": 786 + }, + { + "epoch": 0.31, + "learning_rate": 1.6061501443300644e-05, + "loss": 0.7616, + "step": 787 + }, + { + "epoch": 0.31, + "learning_rate": 1.605120956500982e-05, + "loss": 0.6793, + "step": 788 + }, + { + "epoch": 0.31, + "learning_rate": 1.6040907564240823e-05, + "loss": 0.7988, + "step": 789 + }, + { + "epoch": 0.32, + "learning_rate": 1.6030595458226872e-05, + "loss": 0.6935, + "step": 790 + }, + { + "epoch": 0.32, + "learning_rate": 1.6020273264218076e-05, + "loss": 0.6739, + "step": 791 + }, + { + "epoch": 0.32, + "learning_rate": 1.600994099948143e-05, + "loss": 0.7007, + "step": 792 + }, + { + "epoch": 0.32, + "learning_rate": 1.5999598681300766e-05, + "loss": 0.7535, + "step": 793 + }, + { + "epoch": 0.32, + "learning_rate": 1.5989246326976746e-05, + "loss": 0.6999, + "step": 794 + }, + { + "epoch": 0.32, + "learning_rate": 1.5978883953826814e-05, + "loss": 0.7164, + "step": 795 + }, + { + "epoch": 0.32, + "learning_rate": 1.596851157918517e-05, + "loss": 0.732, + "step": 796 + }, + { + "epoch": 0.32, + "learning_rate": 1.5958129220402745e-05, + "loss": 0.7284, + "step": 797 + }, + { + "epoch": 0.32, + "learning_rate": 1.594773689484718e-05, + "loss": 0.7066, + "step": 798 + }, + { + "epoch": 0.32, + "learning_rate": 1.593733461990278e-05, + "loss": 0.777, + "step": 799 + }, + { + "epoch": 0.32, + "learning_rate": 1.5926922412970495e-05, + "loss": 0.6643, + "step": 800 + }, + { + "epoch": 0.32, + "learning_rate": 1.5916500291467892e-05, + "loss": 0.7571, + "step": 801 + }, + { + "epoch": 0.32, + "learning_rate": 1.5906068272829122e-05, + "loss": 0.7561, + "step": 802 + }, + { + "epoch": 0.32, + "learning_rate": 1.589562637450489e-05, + "loss": 0.6655, + "step": 803 + }, + { + "epoch": 0.32, + "learning_rate": 1.5885174613962427e-05, + "loss": 0.8119, + "step": 804 + }, + { + "epoch": 0.32, + "learning_rate": 1.5874713008685464e-05, + "loss": 0.765, + "step": 805 + }, + { + "epoch": 0.32, + "learning_rate": 1.5864241576174205e-05, + "loss": 0.6736, + "step": 806 + }, + { + "epoch": 0.32, + "learning_rate": 1.5853760333945282e-05, + "loss": 0.676, + "step": 807 + }, + { + "epoch": 0.32, + "learning_rate": 1.5843269299531748e-05, + "loss": 0.6909, + "step": 808 + }, + { + "epoch": 0.32, + "learning_rate": 1.583276849048303e-05, + "loss": 0.7342, + "step": 809 + }, + { + "epoch": 0.32, + "learning_rate": 1.5822257924364902e-05, + "loss": 0.6885, + "step": 810 + }, + { + "epoch": 0.32, + "learning_rate": 1.581173761875947e-05, + "loss": 0.6646, + "step": 811 + }, + { + "epoch": 0.32, + "learning_rate": 1.580120759126512e-05, + "loss": 0.7565, + "step": 812 + }, + { + "epoch": 0.32, + "learning_rate": 1.579066785949652e-05, + "loss": 0.6838, + "step": 813 + }, + { + "epoch": 0.32, + "learning_rate": 1.5780118441084544e-05, + "loss": 0.7319, + "step": 814 + }, + { + "epoch": 0.33, + "learning_rate": 1.5769559353676297e-05, + "loss": 0.7276, + "step": 815 + }, + { + "epoch": 0.33, + "learning_rate": 1.5758990614935042e-05, + "loss": 0.6894, + "step": 816 + }, + { + "epoch": 0.33, + "learning_rate": 1.5748412242540193e-05, + "loss": 0.7782, + "step": 817 + }, + { + "epoch": 0.33, + "learning_rate": 1.5737824254187275e-05, + "loss": 0.7232, + "step": 818 + }, + { + "epoch": 0.33, + "learning_rate": 1.5727226667587898e-05, + "loss": 0.7293, + "step": 819 + }, + { + "epoch": 0.33, + "learning_rate": 1.5716619500469735e-05, + "loss": 0.7244, + "step": 820 + }, + { + "epoch": 0.33, + "learning_rate": 1.5706002770576478e-05, + "loss": 0.7549, + "step": 821 + }, + { + "epoch": 0.33, + "learning_rate": 1.5695376495667828e-05, + "loss": 0.7259, + "step": 822 + }, + { + "epoch": 0.33, + "learning_rate": 1.568474069351943e-05, + "loss": 0.7685, + "step": 823 + }, + { + "epoch": 0.33, + "learning_rate": 1.5674095381922886e-05, + "loss": 0.7859, + "step": 824 + }, + { + "epoch": 0.33, + "learning_rate": 1.5663440578685706e-05, + "loss": 0.7323, + "step": 825 + }, + { + "epoch": 0.33, + "learning_rate": 1.565277630163126e-05, + "loss": 0.7022, + "step": 826 + }, + { + "epoch": 0.33, + "learning_rate": 1.564210256859878e-05, + "loss": 0.7688, + "step": 827 + }, + { + "epoch": 0.33, + "learning_rate": 1.563141939744332e-05, + "loss": 0.6538, + "step": 828 + }, + { + "epoch": 0.33, + "learning_rate": 1.56207268060357e-05, + "loss": 0.7676, + "step": 829 + }, + { + "epoch": 0.33, + "learning_rate": 1.5610024812262526e-05, + "loss": 0.6682, + "step": 830 + }, + { + "epoch": 0.33, + "learning_rate": 1.559931343402612e-05, + "loss": 0.7629, + "step": 831 + }, + { + "epoch": 0.33, + "learning_rate": 1.5588592689244492e-05, + "loss": 0.8038, + "step": 832 + }, + { + "epoch": 0.33, + "learning_rate": 1.5577862595851335e-05, + "loss": 0.7049, + "step": 833 + }, + { + "epoch": 0.33, + "learning_rate": 1.556712317179598e-05, + "loss": 0.7101, + "step": 834 + }, + { + "epoch": 0.33, + "learning_rate": 1.5556374435043353e-05, + "loss": 0.7148, + "step": 835 + }, + { + "epoch": 0.33, + "learning_rate": 1.5545616403573977e-05, + "loss": 0.7397, + "step": 836 + }, + { + "epoch": 0.33, + "learning_rate": 1.5534849095383902e-05, + "loss": 0.7384, + "step": 837 + }, + { + "epoch": 0.33, + "learning_rate": 1.5524072528484716e-05, + "loss": 0.7111, + "step": 838 + }, + { + "epoch": 0.33, + "learning_rate": 1.5513286720903488e-05, + "loss": 0.6912, + "step": 839 + }, + { + "epoch": 0.34, + "learning_rate": 1.5502491690682732e-05, + "loss": 0.7212, + "step": 840 + }, + { + "epoch": 0.34, + "learning_rate": 1.5491687455880414e-05, + "loss": 0.6764, + "step": 841 + }, + { + "epoch": 0.34, + "learning_rate": 1.5480874034569876e-05, + "loss": 0.7154, + "step": 842 + }, + { + "epoch": 0.34, + "learning_rate": 1.547005144483984e-05, + "loss": 0.6909, + "step": 843 + }, + { + "epoch": 0.34, + "learning_rate": 1.5459219704794358e-05, + "loss": 0.6112, + "step": 844 + }, + { + "epoch": 0.34, + "learning_rate": 1.5448378832552796e-05, + "loss": 0.7361, + "step": 845 + }, + { + "epoch": 0.34, + "learning_rate": 1.5437528846249783e-05, + "loss": 0.7308, + "step": 846 + }, + { + "epoch": 0.34, + "learning_rate": 1.542666976403521e-05, + "loss": 0.6926, + "step": 847 + }, + { + "epoch": 0.34, + "learning_rate": 1.541580160407417e-05, + "loss": 0.7326, + "step": 848 + }, + { + "epoch": 0.34, + "learning_rate": 1.5404924384546955e-05, + "loss": 0.7873, + "step": 849 + }, + { + "epoch": 0.34, + "learning_rate": 1.5394038123649e-05, + "loss": 0.649, + "step": 850 + }, + { + "epoch": 0.34, + "learning_rate": 1.5383142839590872e-05, + "loss": 0.7764, + "step": 851 + }, + { + "epoch": 0.34, + "learning_rate": 1.5372238550598225e-05, + "loss": 0.6581, + "step": 852 + }, + { + "epoch": 0.34, + "learning_rate": 1.536132527491178e-05, + "loss": 0.6894, + "step": 853 + }, + { + "epoch": 0.34, + "learning_rate": 1.53504030307873e-05, + "loss": 0.7471, + "step": 854 + }, + { + "epoch": 0.34, + "learning_rate": 1.5339471836495524e-05, + "loss": 0.7491, + "step": 855 + }, + { + "epoch": 0.34, + "learning_rate": 1.5328531710322198e-05, + "loss": 0.7239, + "step": 856 + }, + { + "epoch": 0.34, + "learning_rate": 1.5317582670567986e-05, + "loss": 0.74, + "step": 857 + }, + { + "epoch": 0.34, + "learning_rate": 1.5306624735548465e-05, + "loss": 0.7205, + "step": 858 + }, + { + "epoch": 0.34, + "learning_rate": 1.52956579235941e-05, + "loss": 0.7507, + "step": 859 + }, + { + "epoch": 0.34, + "learning_rate": 1.52846822530502e-05, + "loss": 0.7315, + "step": 860 + }, + { + "epoch": 0.34, + "learning_rate": 1.527369774227689e-05, + "loss": 0.7803, + "step": 861 + }, + { + "epoch": 0.34, + "learning_rate": 1.526270440964909e-05, + "loss": 0.7233, + "step": 862 + }, + { + "epoch": 0.34, + "learning_rate": 1.5251702273556475e-05, + "loss": 0.7377, + "step": 863 + }, + { + "epoch": 0.34, + "learning_rate": 1.5240691352403441e-05, + "loss": 0.7494, + "step": 864 + }, + { + "epoch": 0.35, + "learning_rate": 1.5229671664609088e-05, + "loss": 0.7653, + "step": 865 + }, + { + "epoch": 0.35, + "learning_rate": 1.521864322860717e-05, + "loss": 0.696, + "step": 866 + }, + { + "epoch": 0.35, + "learning_rate": 1.5207606062846092e-05, + "loss": 0.7974, + "step": 867 + }, + { + "epoch": 0.35, + "learning_rate": 1.5196560185788842e-05, + "loss": 0.7025, + "step": 868 + }, + { + "epoch": 0.35, + "learning_rate": 1.5185505615912998e-05, + "loss": 0.7249, + "step": 869 + }, + { + "epoch": 0.35, + "learning_rate": 1.517444237171067e-05, + "loss": 0.7724, + "step": 870 + }, + { + "epoch": 0.35, + "learning_rate": 1.5163370471688471e-05, + "loss": 0.7446, + "step": 871 + }, + { + "epoch": 0.35, + "learning_rate": 1.5152289934367516e-05, + "loss": 0.6557, + "step": 872 + }, + { + "epoch": 0.35, + "learning_rate": 1.5141200778283345e-05, + "loss": 0.6653, + "step": 873 + }, + { + "epoch": 0.35, + "learning_rate": 1.5130103021985929e-05, + "loss": 0.7967, + "step": 874 + }, + { + "epoch": 0.35, + "learning_rate": 1.5118996684039621e-05, + "loss": 0.7232, + "step": 875 + }, + { + "epoch": 0.35, + "learning_rate": 1.5107881783023138e-05, + "loss": 0.686, + "step": 876 + }, + { + "epoch": 0.35, + "learning_rate": 1.50967583375295e-05, + "loss": 0.7378, + "step": 877 + }, + { + "epoch": 0.35, + "learning_rate": 1.5085626366166039e-05, + "loss": 0.7622, + "step": 878 + }, + { + "epoch": 0.35, + "learning_rate": 1.5074485887554351e-05, + "loss": 0.6785, + "step": 879 + }, + { + "epoch": 0.35, + "learning_rate": 1.5063336920330244e-05, + "loss": 0.7237, + "step": 880 + }, + { + "epoch": 0.35, + "learning_rate": 1.5052179483143752e-05, + "loss": 0.7566, + "step": 881 + }, + { + "epoch": 0.35, + "learning_rate": 1.5041013594659051e-05, + "loss": 0.7706, + "step": 882 + }, + { + "epoch": 0.35, + "learning_rate": 1.5029839273554469e-05, + "loss": 0.7093, + "step": 883 + }, + { + "epoch": 0.35, + "learning_rate": 1.5018656538522442e-05, + "loss": 0.7402, + "step": 884 + }, + { + "epoch": 0.35, + "learning_rate": 1.5007465408269477e-05, + "loss": 0.6275, + "step": 885 + }, + { + "epoch": 0.35, + "learning_rate": 1.4996265901516116e-05, + "loss": 0.688, + "step": 886 + }, + { + "epoch": 0.35, + "learning_rate": 1.4985058036996929e-05, + "loss": 0.7567, + "step": 887 + }, + { + "epoch": 0.35, + "learning_rate": 1.4973841833460457e-05, + "loss": 0.7034, + "step": 888 + }, + { + "epoch": 0.35, + "learning_rate": 1.4962617309669192e-05, + "loss": 0.7125, + "step": 889 + }, + { + "epoch": 0.36, + "learning_rate": 1.4951384484399542e-05, + "loss": 0.6969, + "step": 890 + }, + { + "epoch": 0.36, + "learning_rate": 1.4940143376441812e-05, + "loss": 0.7094, + "step": 891 + }, + { + "epoch": 0.36, + "learning_rate": 1.492889400460015e-05, + "loss": 0.6931, + "step": 892 + }, + { + "epoch": 0.36, + "learning_rate": 1.491763638769253e-05, + "loss": 0.7134, + "step": 893 + }, + { + "epoch": 0.36, + "learning_rate": 1.490637054455073e-05, + "loss": 0.7068, + "step": 894 + }, + { + "epoch": 0.36, + "learning_rate": 1.4895096494020274e-05, + "loss": 0.7059, + "step": 895 + }, + { + "epoch": 0.36, + "learning_rate": 1.4883814254960418e-05, + "loss": 0.7223, + "step": 896 + }, + { + "epoch": 0.36, + "learning_rate": 1.4872523846244127e-05, + "loss": 0.7097, + "step": 897 + }, + { + "epoch": 0.36, + "learning_rate": 1.4861225286758018e-05, + "loss": 0.7931, + "step": 898 + }, + { + "epoch": 0.36, + "learning_rate": 1.4849918595402353e-05, + "loss": 0.7084, + "step": 899 + }, + { + "epoch": 0.36, + "learning_rate": 1.4838603791090989e-05, + "loss": 0.7286, + "step": 900 + }, + { + "epoch": 0.36, + "learning_rate": 1.4827280892751364e-05, + "loss": 0.7926, + "step": 901 + }, + { + "epoch": 0.36, + "learning_rate": 1.4815949919324443e-05, + "loss": 0.7257, + "step": 902 + }, + { + "epoch": 0.36, + "learning_rate": 1.4804610889764712e-05, + "loss": 0.7326, + "step": 903 + }, + { + "epoch": 0.36, + "learning_rate": 1.4793263823040127e-05, + "loss": 0.7804, + "step": 904 + }, + { + "epoch": 0.36, + "learning_rate": 1.4781908738132085e-05, + "loss": 0.7004, + "step": 905 + }, + { + "epoch": 0.36, + "learning_rate": 1.4770545654035403e-05, + "loss": 0.7194, + "step": 906 + }, + { + "epoch": 0.36, + "learning_rate": 1.4759174589758272e-05, + "loss": 0.6807, + "step": 907 + }, + { + "epoch": 0.36, + "learning_rate": 1.4747795564322243e-05, + "loss": 0.7616, + "step": 908 + }, + { + "epoch": 0.36, + "learning_rate": 1.473640859676217e-05, + "loss": 0.6851, + "step": 909 + }, + { + "epoch": 0.36, + "learning_rate": 1.4725013706126207e-05, + "loss": 0.7324, + "step": 910 + }, + { + "epoch": 0.36, + "learning_rate": 1.471361091147575e-05, + "loss": 0.7519, + "step": 911 + }, + { + "epoch": 0.36, + "learning_rate": 1.4702200231885425e-05, + "loss": 0.7597, + "step": 912 + }, + { + "epoch": 0.36, + "learning_rate": 1.4690781686443044e-05, + "loss": 0.6799, + "step": 913 + }, + { + "epoch": 0.36, + "learning_rate": 1.4679355294249575e-05, + "loss": 0.7063, + "step": 914 + }, + { + "epoch": 0.37, + "learning_rate": 1.4667921074419118e-05, + "loss": 0.9109, + "step": 915 + }, + { + "epoch": 0.37, + "learning_rate": 1.4656479046078862e-05, + "loss": 0.6954, + "step": 916 + }, + { + "epoch": 0.37, + "learning_rate": 1.4645029228369062e-05, + "loss": 0.7546, + "step": 917 + }, + { + "epoch": 0.37, + "learning_rate": 1.4633571640442998e-05, + "loss": 0.7376, + "step": 918 + }, + { + "epoch": 0.37, + "learning_rate": 1.4622106301466956e-05, + "loss": 0.7129, + "step": 919 + }, + { + "epoch": 0.37, + "learning_rate": 1.4610633230620179e-05, + "loss": 0.7309, + "step": 920 + }, + { + "epoch": 0.37, + "learning_rate": 1.4599152447094846e-05, + "loss": 0.6998, + "step": 921 + }, + { + "epoch": 0.37, + "learning_rate": 1.4587663970096048e-05, + "loss": 0.6711, + "step": 922 + }, + { + "epoch": 0.37, + "learning_rate": 1.4576167818841732e-05, + "loss": 0.8417, + "step": 923 + }, + { + "epoch": 0.37, + "learning_rate": 1.4564664012562686e-05, + "loss": 0.7627, + "step": 924 + }, + { + "epoch": 0.37, + "learning_rate": 1.4553152570502506e-05, + "loss": 0.7172, + "step": 925 + }, + { + "epoch": 0.37, + "learning_rate": 1.4541633511917567e-05, + "loss": 0.6909, + "step": 926 + }, + { + "epoch": 0.37, + "learning_rate": 1.453010685607697e-05, + "loss": 0.7444, + "step": 927 + }, + { + "epoch": 0.37, + "learning_rate": 1.451857262226254e-05, + "loss": 0.7009, + "step": 928 + }, + { + "epoch": 0.37, + "learning_rate": 1.4507030829768765e-05, + "loss": 0.7559, + "step": 929 + }, + { + "epoch": 0.37, + "learning_rate": 1.4495481497902789e-05, + "loss": 0.7428, + "step": 930 + }, + { + "epoch": 0.37, + "learning_rate": 1.4483924645984359e-05, + "loss": 0.7324, + "step": 931 + }, + { + "epoch": 0.37, + "learning_rate": 1.4472360293345807e-05, + "loss": 0.7708, + "step": 932 + }, + { + "epoch": 0.37, + "learning_rate": 1.4460788459332011e-05, + "loss": 0.6982, + "step": 933 + }, + { + "epoch": 0.37, + "learning_rate": 1.444920916330036e-05, + "loss": 0.6758, + "step": 934 + }, + { + "epoch": 0.37, + "learning_rate": 1.4437622424620733e-05, + "loss": 0.6988, + "step": 935 + }, + { + "epoch": 0.37, + "learning_rate": 1.4426028262675451e-05, + "loss": 0.7192, + "step": 936 + }, + { + "epoch": 0.37, + "learning_rate": 1.441442669685926e-05, + "loss": 0.7863, + "step": 937 + }, + { + "epoch": 0.37, + "learning_rate": 1.440281774657929e-05, + "loss": 0.6634, + "step": 938 + }, + { + "epoch": 0.37, + "learning_rate": 1.4391201431255014e-05, + "loss": 0.7147, + "step": 939 + }, + { + "epoch": 0.38, + "learning_rate": 1.4379577770318234e-05, + "loss": 0.6696, + "step": 940 + }, + { + "epoch": 0.38, + "learning_rate": 1.436794678321304e-05, + "loss": 0.7312, + "step": 941 + }, + { + "epoch": 0.38, + "learning_rate": 1.4356308489395783e-05, + "loss": 0.7132, + "step": 942 + }, + { + "epoch": 0.38, + "learning_rate": 1.4344662908335018e-05, + "loss": 0.7166, + "step": 943 + }, + { + "epoch": 0.38, + "learning_rate": 1.4333010059511505e-05, + "loss": 0.7151, + "step": 944 + }, + { + "epoch": 0.38, + "learning_rate": 1.4321349962418163e-05, + "loss": 0.7141, + "step": 945 + }, + { + "epoch": 0.38, + "learning_rate": 1.4309682636560027e-05, + "loss": 0.6664, + "step": 946 + }, + { + "epoch": 0.38, + "learning_rate": 1.4298008101454226e-05, + "loss": 0.7223, + "step": 947 + }, + { + "epoch": 0.38, + "learning_rate": 1.4286326376629956e-05, + "loss": 0.7222, + "step": 948 + }, + { + "epoch": 0.38, + "learning_rate": 1.4274637481628431e-05, + "loss": 0.7376, + "step": 949 + }, + { + "epoch": 0.38, + "learning_rate": 1.4262941436002862e-05, + "loss": 0.7515, + "step": 950 + }, + { + "epoch": 0.38, + "learning_rate": 1.425123825931843e-05, + "loss": 0.7296, + "step": 951 + }, + { + "epoch": 0.38, + "learning_rate": 1.423952797115223e-05, + "loss": 0.7029, + "step": 952 + }, + { + "epoch": 0.38, + "learning_rate": 1.4227810591093262e-05, + "loss": 0.6757, + "step": 953 + }, + { + "epoch": 0.38, + "learning_rate": 1.4216086138742395e-05, + "loss": 0.7553, + "step": 954 + }, + { + "epoch": 0.38, + "learning_rate": 1.4204354633712312e-05, + "loss": 0.6547, + "step": 955 + }, + { + "epoch": 0.38, + "learning_rate": 1.4192616095627506e-05, + "loss": 0.7446, + "step": 956 + }, + { + "epoch": 0.38, + "learning_rate": 1.4180870544124236e-05, + "loss": 0.7203, + "step": 957 + }, + { + "epoch": 0.38, + "learning_rate": 1.416911799885049e-05, + "loss": 0.7809, + "step": 958 + }, + { + "epoch": 0.38, + "learning_rate": 1.4157358479465949e-05, + "loss": 0.7599, + "step": 959 + }, + { + "epoch": 0.38, + "learning_rate": 1.4145592005641967e-05, + "loss": 0.7771, + "step": 960 + }, + { + "epoch": 0.38, + "learning_rate": 1.413381859706153e-05, + "loss": 0.6599, + "step": 961 + }, + { + "epoch": 0.38, + "learning_rate": 1.412203827341923e-05, + "loss": 0.7281, + "step": 962 + }, + { + "epoch": 0.38, + "learning_rate": 1.4110251054421218e-05, + "loss": 0.6748, + "step": 963 + }, + { + "epoch": 0.38, + "learning_rate": 1.4098456959785178e-05, + "loss": 0.7118, + "step": 964 + }, + { + "epoch": 0.39, + "learning_rate": 1.4086656009240308e-05, + "loss": 0.6742, + "step": 965 + }, + { + "epoch": 0.39, + "learning_rate": 1.4074848222527262e-05, + "loss": 0.7773, + "step": 966 + }, + { + "epoch": 0.39, + "learning_rate": 1.4063033619398136e-05, + "loss": 0.7868, + "step": 967 + }, + { + "epoch": 0.39, + "learning_rate": 1.4051212219616427e-05, + "loss": 0.7136, + "step": 968 + }, + { + "epoch": 0.39, + "learning_rate": 1.4039384042957003e-05, + "loss": 0.7151, + "step": 969 + }, + { + "epoch": 0.39, + "learning_rate": 1.4027549109206064e-05, + "loss": 0.7788, + "step": 970 + }, + { + "epoch": 0.39, + "learning_rate": 1.4015707438161122e-05, + "loss": 0.6691, + "step": 971 + }, + { + "epoch": 0.39, + "learning_rate": 1.4003859049630943e-05, + "loss": 0.7017, + "step": 972 + }, + { + "epoch": 0.39, + "learning_rate": 1.399200396343555e-05, + "loss": 0.7083, + "step": 973 + }, + { + "epoch": 0.39, + "learning_rate": 1.3980142199406155e-05, + "loss": 0.6739, + "step": 974 + }, + { + "epoch": 0.39, + "learning_rate": 1.3968273777385146e-05, + "loss": 0.7386, + "step": 975 + }, + { + "epoch": 0.39, + "learning_rate": 1.3956398717226055e-05, + "loss": 0.6874, + "step": 976 + }, + { + "epoch": 0.39, + "learning_rate": 1.39445170387935e-05, + "loss": 0.6918, + "step": 977 + }, + { + "epoch": 0.39, + "learning_rate": 1.3932628761963192e-05, + "loss": 0.7531, + "step": 978 + }, + { + "epoch": 0.39, + "learning_rate": 1.3920733906621861e-05, + "loss": 0.7332, + "step": 979 + }, + { + "epoch": 0.39, + "learning_rate": 1.3908832492667261e-05, + "loss": 0.6303, + "step": 980 + }, + { + "epoch": 0.39, + "learning_rate": 1.3896924540008098e-05, + "loss": 0.7639, + "step": 981 + }, + { + "epoch": 0.39, + "learning_rate": 1.3885010068564026e-05, + "loss": 0.6884, + "step": 982 + }, + { + "epoch": 0.39, + "learning_rate": 1.3873089098265606e-05, + "loss": 0.7703, + "step": 983 + }, + { + "epoch": 0.39, + "learning_rate": 1.386116164905426e-05, + "loss": 0.7685, + "step": 984 + }, + { + "epoch": 0.39, + "learning_rate": 1.3849227740882257e-05, + "loss": 0.6787, + "step": 985 + }, + { + "epoch": 0.39, + "learning_rate": 1.3837287393712666e-05, + "loss": 0.7025, + "step": 986 + }, + { + "epoch": 0.39, + "learning_rate": 1.3825340627519337e-05, + "loss": 0.7497, + "step": 987 + }, + { + "epoch": 0.39, + "learning_rate": 1.3813387462286839e-05, + "loss": 0.7557, + "step": 988 + }, + { + "epoch": 0.39, + "learning_rate": 1.3801427918010461e-05, + "loss": 0.7946, + "step": 989 + }, + { + "epoch": 0.4, + "learning_rate": 1.378946201469616e-05, + "loss": 0.7426, + "step": 990 + }, + { + "epoch": 0.4, + "learning_rate": 1.3777489772360521e-05, + "loss": 0.7485, + "step": 991 + }, + { + "epoch": 0.4, + "learning_rate": 1.3765511211030753e-05, + "loss": 0.7442, + "step": 992 + }, + { + "epoch": 0.4, + "learning_rate": 1.3753526350744611e-05, + "loss": 0.8228, + "step": 993 + }, + { + "epoch": 0.4, + "learning_rate": 1.3741535211550402e-05, + "loss": 0.8018, + "step": 994 + }, + { + "epoch": 0.4, + "learning_rate": 1.3729537813506932e-05, + "loss": 0.664, + "step": 995 + }, + { + "epoch": 0.4, + "learning_rate": 1.3717534176683485e-05, + "loss": 0.6951, + "step": 996 + }, + { + "epoch": 0.4, + "learning_rate": 1.3705524321159764e-05, + "loss": 0.7899, + "step": 997 + }, + { + "epoch": 0.4, + "learning_rate": 1.369350826702589e-05, + "loss": 0.6793, + "step": 998 + }, + { + "epoch": 0.4, + "learning_rate": 1.3681486034382347e-05, + "loss": 0.7321, + "step": 999 + }, + { + "epoch": 0.4, + "learning_rate": 1.3669457643339956e-05, + "loss": 0.6275, + "step": 1000 + }, + { + "epoch": 0.4, + "learning_rate": 1.3657423114019837e-05, + "loss": 0.7294, + "step": 1001 + }, + { + "epoch": 0.4, + "learning_rate": 1.3645382466553381e-05, + "loss": 0.7588, + "step": 1002 + }, + { + "epoch": 0.4, + "learning_rate": 1.3633335721082215e-05, + "loss": 0.6977, + "step": 1003 + }, + { + "epoch": 0.4, + "learning_rate": 1.362128289775816e-05, + "loss": 0.6995, + "step": 1004 + }, + { + "epoch": 0.4, + "learning_rate": 1.3609224016743213e-05, + "loss": 0.7721, + "step": 1005 + }, + { + "epoch": 0.4, + "learning_rate": 1.3597159098209493e-05, + "loss": 0.7281, + "step": 1006 + }, + { + "epoch": 0.4, + "learning_rate": 1.3585088162339233e-05, + "loss": 0.748, + "step": 1007 + }, + { + "epoch": 0.4, + "learning_rate": 1.357301122932472e-05, + "loss": 0.7178, + "step": 1008 + }, + { + "epoch": 0.4, + "learning_rate": 1.3560928319368278e-05, + "loss": 0.6825, + "step": 1009 + }, + { + "epoch": 0.4, + "learning_rate": 1.3548839452682225e-05, + "loss": 0.6973, + "step": 1010 + }, + { + "epoch": 0.4, + "learning_rate": 1.353674464948885e-05, + "loss": 0.701, + "step": 1011 + }, + { + "epoch": 0.4, + "learning_rate": 1.3524643930020374e-05, + "loss": 0.6997, + "step": 1012 + }, + { + "epoch": 0.4, + "learning_rate": 1.35125373145189e-05, + "loss": 0.7767, + "step": 1013 + }, + { + "epoch": 0.4, + "learning_rate": 1.3500424823236413e-05, + "loss": 0.6611, + "step": 1014 + }, + { + "epoch": 0.41, + "learning_rate": 1.3488306476434714e-05, + "loss": 0.7465, + "step": 1015 + }, + { + "epoch": 0.41, + "learning_rate": 1.3476182294385398e-05, + "loss": 0.8132, + "step": 1016 + }, + { + "epoch": 0.41, + "learning_rate": 1.3464052297369838e-05, + "loss": 0.7148, + "step": 1017 + }, + { + "epoch": 0.41, + "learning_rate": 1.3451916505679115e-05, + "loss": 0.6836, + "step": 1018 + }, + { + "epoch": 0.41, + "learning_rate": 1.343977493961401e-05, + "loss": 0.7502, + "step": 1019 + }, + { + "epoch": 0.41, + "learning_rate": 1.3427627619484968e-05, + "loss": 0.7362, + "step": 1020 + }, + { + "epoch": 0.41, + "learning_rate": 1.3415474565612059e-05, + "loss": 0.7326, + "step": 1021 + }, + { + "epoch": 0.41, + "learning_rate": 1.3403315798324935e-05, + "loss": 0.6254, + "step": 1022 + }, + { + "epoch": 0.41, + "learning_rate": 1.3391151337962813e-05, + "loss": 0.6952, + "step": 1023 + }, + { + "epoch": 0.41, + "learning_rate": 1.3378981204874432e-05, + "loss": 0.6705, + "step": 1024 + }, + { + "epoch": 0.41, + "learning_rate": 1.3366805419418026e-05, + "loss": 0.6478, + "step": 1025 + }, + { + "epoch": 0.41, + "learning_rate": 1.3354624001961267e-05, + "loss": 0.6997, + "step": 1026 + }, + { + "epoch": 0.41, + "learning_rate": 1.3342436972881272e-05, + "loss": 0.6544, + "step": 1027 + }, + { + "epoch": 0.41, + "learning_rate": 1.3330244352564527e-05, + "loss": 0.7257, + "step": 1028 + }, + { + "epoch": 0.41, + "learning_rate": 1.3318046161406876e-05, + "loss": 0.6675, + "step": 1029 + }, + { + "epoch": 0.41, + "learning_rate": 1.3305842419813488e-05, + "loss": 0.7425, + "step": 1030 + }, + { + "epoch": 0.41, + "learning_rate": 1.3293633148198812e-05, + "loss": 0.683, + "step": 1031 + }, + { + "epoch": 0.41, + "learning_rate": 1.3281418366986535e-05, + "loss": 0.6478, + "step": 1032 + }, + { + "epoch": 0.41, + "learning_rate": 1.3269198096609592e-05, + "loss": 0.7244, + "step": 1033 + }, + { + "epoch": 0.41, + "learning_rate": 1.3256972357510068e-05, + "loss": 0.763, + "step": 1034 + }, + { + "epoch": 0.41, + "learning_rate": 1.324474117013921e-05, + "loss": 0.5932, + "step": 1035 + }, + { + "epoch": 0.41, + "learning_rate": 1.3232504554957385e-05, + "loss": 0.7454, + "step": 1036 + }, + { + "epoch": 0.41, + "learning_rate": 1.322026253243403e-05, + "loss": 0.7296, + "step": 1037 + }, + { + "epoch": 0.41, + "learning_rate": 1.3208015123047623e-05, + "loss": 0.7184, + "step": 1038 + }, + { + "epoch": 0.41, + "learning_rate": 1.319576234728567e-05, + "loss": 0.6558, + "step": 1039 + }, + { + "epoch": 0.42, + "learning_rate": 1.3183504225644643e-05, + "loss": 0.7235, + "step": 1040 + }, + { + "epoch": 0.42, + "learning_rate": 1.3171240778629952e-05, + "loss": 0.7098, + "step": 1041 + }, + { + "epoch": 0.42, + "learning_rate": 1.3158972026755927e-05, + "loss": 0.651, + "step": 1042 + }, + { + "epoch": 0.42, + "learning_rate": 1.3146697990545762e-05, + "loss": 0.6754, + "step": 1043 + }, + { + "epoch": 0.42, + "learning_rate": 1.3134418690531498e-05, + "loss": 0.7272, + "step": 1044 + }, + { + "epoch": 0.42, + "learning_rate": 1.3122134147253975e-05, + "loss": 0.7759, + "step": 1045 + }, + { + "epoch": 0.42, + "learning_rate": 1.3109844381262816e-05, + "loss": 0.7124, + "step": 1046 + }, + { + "epoch": 0.42, + "learning_rate": 1.3097549413116358e-05, + "loss": 0.7155, + "step": 1047 + }, + { + "epoch": 0.42, + "learning_rate": 1.3085249263381661e-05, + "loss": 0.7626, + "step": 1048 + }, + { + "epoch": 0.42, + "learning_rate": 1.3072943952634446e-05, + "loss": 0.71, + "step": 1049 + }, + { + "epoch": 0.42, + "learning_rate": 1.3060633501459065e-05, + "loss": 0.6449, + "step": 1050 + }, + { + "epoch": 0.42, + "learning_rate": 1.3048317930448472e-05, + "loss": 0.7252, + "step": 1051 + }, + { + "epoch": 0.42, + "learning_rate": 1.3035997260204187e-05, + "loss": 0.8045, + "step": 1052 + }, + { + "epoch": 0.42, + "learning_rate": 1.3023671511336253e-05, + "loss": 0.6678, + "step": 1053 + }, + { + "epoch": 0.42, + "learning_rate": 1.301134070446321e-05, + "loss": 0.729, + "step": 1054 + }, + { + "epoch": 0.42, + "learning_rate": 1.2999004860212071e-05, + "loss": 0.6459, + "step": 1055 + }, + { + "epoch": 0.42, + "learning_rate": 1.2986663999218264e-05, + "loss": 0.7451, + "step": 1056 + }, + { + "epoch": 0.42, + "learning_rate": 1.2974318142125604e-05, + "loss": 0.5985, + "step": 1057 + }, + { + "epoch": 0.42, + "learning_rate": 1.2961967309586279e-05, + "loss": 0.7804, + "step": 1058 + }, + { + "epoch": 0.42, + "learning_rate": 1.2949611522260788e-05, + "loss": 0.729, + "step": 1059 + }, + { + "epoch": 0.42, + "learning_rate": 1.2937250800817922e-05, + "loss": 0.6245, + "step": 1060 + }, + { + "epoch": 0.42, + "learning_rate": 1.2924885165934728e-05, + "loss": 0.6134, + "step": 1061 + }, + { + "epoch": 0.42, + "learning_rate": 1.2912514638296472e-05, + "loss": 0.7461, + "step": 1062 + }, + { + "epoch": 0.42, + "learning_rate": 1.2900139238596598e-05, + "loss": 0.7602, + "step": 1063 + }, + { + "epoch": 0.42, + "learning_rate": 1.288775898753671e-05, + "loss": 0.7697, + "step": 1064 + }, + { + "epoch": 0.43, + "learning_rate": 1.287537390582652e-05, + "loss": 0.6862, + "step": 1065 + }, + { + "epoch": 0.43, + "learning_rate": 1.286298401418382e-05, + "loss": 0.687, + "step": 1066 + }, + { + "epoch": 0.43, + "learning_rate": 1.2850589333334455e-05, + "loss": 0.6116, + "step": 1067 + }, + { + "epoch": 0.43, + "learning_rate": 1.2838189884012275e-05, + "loss": 0.7144, + "step": 1068 + }, + { + "epoch": 0.43, + "learning_rate": 1.2825785686959113e-05, + "loss": 0.7589, + "step": 1069 + }, + { + "epoch": 0.43, + "learning_rate": 1.2813376762924735e-05, + "loss": 0.7187, + "step": 1070 + }, + { + "epoch": 0.43, + "learning_rate": 1.2800963132666825e-05, + "loss": 0.7129, + "step": 1071 + }, + { + "epoch": 0.43, + "learning_rate": 1.278854481695093e-05, + "loss": 0.6735, + "step": 1072 + }, + { + "epoch": 0.43, + "learning_rate": 1.277612183655044e-05, + "loss": 0.688, + "step": 1073 + }, + { + "epoch": 0.43, + "learning_rate": 1.276369421224655e-05, + "loss": 0.7438, + "step": 1074 + }, + { + "epoch": 0.43, + "learning_rate": 1.2751261964828214e-05, + "loss": 0.7739, + "step": 1075 + }, + { + "epoch": 0.43, + "learning_rate": 1.273882511509213e-05, + "loss": 0.5997, + "step": 1076 + }, + { + "epoch": 0.43, + "learning_rate": 1.272638368384269e-05, + "loss": 0.7333, + "step": 1077 + }, + { + "epoch": 0.43, + "learning_rate": 1.2713937691891953e-05, + "loss": 0.7239, + "step": 1078 + }, + { + "epoch": 0.43, + "learning_rate": 1.2701487160059601e-05, + "loss": 0.7352, + "step": 1079 + }, + { + "epoch": 0.43, + "learning_rate": 1.2689032109172914e-05, + "loss": 0.6482, + "step": 1080 + }, + { + "epoch": 0.43, + "learning_rate": 1.2676572560066736e-05, + "loss": 0.7916, + "step": 1081 + }, + { + "epoch": 0.43, + "learning_rate": 1.2664108533583428e-05, + "loss": 0.6773, + "step": 1082 + }, + { + "epoch": 0.43, + "learning_rate": 1.2651640050572845e-05, + "loss": 0.7356, + "step": 1083 + }, + { + "epoch": 0.43, + "learning_rate": 1.2639167131892294e-05, + "loss": 0.7084, + "step": 1084 + }, + { + "epoch": 0.43, + "learning_rate": 1.2626689798406506e-05, + "loss": 0.6952, + "step": 1085 + }, + { + "epoch": 0.43, + "learning_rate": 1.2614208070987592e-05, + "loss": 0.6386, + "step": 1086 + }, + { + "epoch": 0.43, + "learning_rate": 1.2601721970515024e-05, + "loss": 0.7488, + "step": 1087 + }, + { + "epoch": 0.43, + "learning_rate": 1.258923151787557e-05, + "loss": 0.6701, + "step": 1088 + }, + { + "epoch": 0.43, + "learning_rate": 1.25767367339633e-05, + "loss": 0.7588, + "step": 1089 + }, + { + "epoch": 0.44, + "learning_rate": 1.2564237639679512e-05, + "loss": 0.7211, + "step": 1090 + }, + { + "epoch": 0.44, + "learning_rate": 1.2551734255932728e-05, + "loss": 0.6789, + "step": 1091 + }, + { + "epoch": 0.44, + "learning_rate": 1.2539226603638633e-05, + "loss": 0.7169, + "step": 1092 + }, + { + "epoch": 0.44, + "learning_rate": 1.252671470372006e-05, + "loss": 0.6923, + "step": 1093 + }, + { + "epoch": 0.44, + "learning_rate": 1.2514198577106948e-05, + "loss": 0.703, + "step": 1094 + }, + { + "epoch": 0.44, + "learning_rate": 1.25016782447363e-05, + "loss": 0.7048, + "step": 1095 + }, + { + "epoch": 0.44, + "learning_rate": 1.2489153727552164e-05, + "loss": 0.7622, + "step": 1096 + }, + { + "epoch": 0.44, + "learning_rate": 1.247662504650558e-05, + "loss": 0.7554, + "step": 1097 + }, + { + "epoch": 0.44, + "learning_rate": 1.2464092222554554e-05, + "loss": 0.6517, + "step": 1098 + }, + { + "epoch": 0.44, + "learning_rate": 1.2451555276664032e-05, + "loss": 0.6925, + "step": 1099 + }, + { + "epoch": 0.44, + "learning_rate": 1.2439014229805836e-05, + "loss": 0.7254, + "step": 1100 + }, + { + "epoch": 0.44, + "learning_rate": 1.2426469102958671e-05, + "loss": 0.7689, + "step": 1101 + }, + { + "epoch": 0.44, + "learning_rate": 1.2413919917108053e-05, + "loss": 0.6423, + "step": 1102 + }, + { + "epoch": 0.44, + "learning_rate": 1.2401366693246289e-05, + "loss": 0.6574, + "step": 1103 + }, + { + "epoch": 0.44, + "learning_rate": 1.2388809452372442e-05, + "loss": 0.6821, + "step": 1104 + }, + { + "epoch": 0.44, + "learning_rate": 1.2376248215492297e-05, + "loss": 0.7314, + "step": 1105 + }, + { + "epoch": 0.44, + "learning_rate": 1.2363683003618322e-05, + "loss": 0.7361, + "step": 1106 + }, + { + "epoch": 0.44, + "learning_rate": 1.235111383776963e-05, + "loss": 0.6688, + "step": 1107 + }, + { + "epoch": 0.44, + "learning_rate": 1.233854073897196e-05, + "loss": 0.7335, + "step": 1108 + }, + { + "epoch": 0.44, + "learning_rate": 1.2325963728257616e-05, + "loss": 0.7129, + "step": 1109 + }, + { + "epoch": 0.44, + "learning_rate": 1.2313382826665456e-05, + "loss": 0.615, + "step": 1110 + }, + { + "epoch": 0.44, + "learning_rate": 1.2300798055240836e-05, + "loss": 0.6797, + "step": 1111 + }, + { + "epoch": 0.44, + "learning_rate": 1.2288209435035605e-05, + "loss": 0.7145, + "step": 1112 + }, + { + "epoch": 0.44, + "learning_rate": 1.2275616987108028e-05, + "loss": 0.6863, + "step": 1113 + }, + { + "epoch": 0.44, + "learning_rate": 1.2263020732522786e-05, + "loss": 0.7051, + "step": 1114 + }, + { + "epoch": 0.45, + "learning_rate": 1.2250420692350921e-05, + "loss": 0.662, + "step": 1115 + }, + { + "epoch": 0.45, + "learning_rate": 1.2237816887669817e-05, + "loss": 0.7349, + "step": 1116 + }, + { + "epoch": 0.45, + "learning_rate": 1.2225209339563144e-05, + "loss": 0.6257, + "step": 1117 + }, + { + "epoch": 0.45, + "learning_rate": 1.2212598069120845e-05, + "loss": 0.694, + "step": 1118 + }, + { + "epoch": 0.45, + "learning_rate": 1.2199983097439079e-05, + "loss": 0.625, + "step": 1119 + }, + { + "epoch": 0.45, + "learning_rate": 1.2187364445620203e-05, + "loss": 0.6342, + "step": 1120 + }, + { + "epoch": 0.45, + "learning_rate": 1.2174742134772726e-05, + "loss": 0.6658, + "step": 1121 + }, + { + "epoch": 0.45, + "learning_rate": 1.2162116186011285e-05, + "loss": 0.726, + "step": 1122 + }, + { + "epoch": 0.45, + "learning_rate": 1.2149486620456599e-05, + "loss": 0.6691, + "step": 1123 + }, + { + "epoch": 0.45, + "learning_rate": 1.2136853459235427e-05, + "loss": 0.6447, + "step": 1124 + }, + { + "epoch": 0.45, + "learning_rate": 1.2124216723480565e-05, + "loss": 0.7263, + "step": 1125 + }, + { + "epoch": 0.45, + "learning_rate": 1.2111576434330767e-05, + "loss": 0.6962, + "step": 1126 + }, + { + "epoch": 0.45, + "learning_rate": 1.209893261293074e-05, + "loss": 0.6578, + "step": 1127 + }, + { + "epoch": 0.45, + "learning_rate": 1.2086285280431103e-05, + "loss": 0.6731, + "step": 1128 + }, + { + "epoch": 0.45, + "learning_rate": 1.2073634457988342e-05, + "loss": 0.736, + "step": 1129 + }, + { + "epoch": 0.45, + "learning_rate": 1.2060980166764782e-05, + "loss": 0.6777, + "step": 1130 + }, + { + "epoch": 0.45, + "learning_rate": 1.2048322427928556e-05, + "loss": 0.7161, + "step": 1131 + }, + { + "epoch": 0.45, + "learning_rate": 1.2035661262653562e-05, + "loss": 0.6609, + "step": 1132 + }, + { + "epoch": 0.45, + "learning_rate": 1.2022996692119424e-05, + "loss": 0.7662, + "step": 1133 + }, + { + "epoch": 0.45, + "learning_rate": 1.2010328737511466e-05, + "loss": 0.6488, + "step": 1134 + }, + { + "epoch": 0.45, + "learning_rate": 1.1997657420020678e-05, + "loss": 0.7402, + "step": 1135 + }, + { + "epoch": 0.45, + "learning_rate": 1.1984982760843666e-05, + "loss": 0.6878, + "step": 1136 + }, + { + "epoch": 0.45, + "learning_rate": 1.1972304781182636e-05, + "loss": 0.7306, + "step": 1137 + }, + { + "epoch": 0.45, + "learning_rate": 1.195962350224534e-05, + "loss": 0.6714, + "step": 1138 + }, + { + "epoch": 0.45, + "learning_rate": 1.1946938945245052e-05, + "loss": 0.6727, + "step": 1139 + }, + { + "epoch": 0.46, + "learning_rate": 1.1934251131400532e-05, + "loss": 0.678, + "step": 1140 + }, + { + "epoch": 0.46, + "learning_rate": 1.1921560081935986e-05, + "loss": 0.6445, + "step": 1141 + }, + { + "epoch": 0.46, + "learning_rate": 1.1908865818081032e-05, + "loss": 0.7038, + "step": 1142 + }, + { + "epoch": 0.46, + "learning_rate": 1.1896168361070664e-05, + "loss": 0.7862, + "step": 1143 + }, + { + "epoch": 0.46, + "learning_rate": 1.1883467732145224e-05, + "loss": 0.6302, + "step": 1144 + }, + { + "epoch": 0.46, + "learning_rate": 1.1870763952550347e-05, + "loss": 0.7335, + "step": 1145 + }, + { + "epoch": 0.46, + "learning_rate": 1.1858057043536955e-05, + "loss": 0.683, + "step": 1146 + }, + { + "epoch": 0.46, + "learning_rate": 1.1845347026361192e-05, + "loss": 0.7796, + "step": 1147 + }, + { + "epoch": 0.46, + "learning_rate": 1.1832633922284407e-05, + "loss": 0.7235, + "step": 1148 + }, + { + "epoch": 0.46, + "learning_rate": 1.1819917752573109e-05, + "loss": 0.6534, + "step": 1149 + }, + { + "epoch": 0.46, + "learning_rate": 1.1807198538498944e-05, + "loss": 0.7211, + "step": 1150 + }, + { + "epoch": 0.46, + "learning_rate": 1.179447630133864e-05, + "loss": 0.6842, + "step": 1151 + }, + { + "epoch": 0.46, + "learning_rate": 1.1781751062373982e-05, + "loss": 0.6413, + "step": 1152 + }, + { + "epoch": 0.46, + "learning_rate": 1.1769022842891794e-05, + "loss": 0.708, + "step": 1153 + }, + { + "epoch": 0.46, + "learning_rate": 1.1756291664183858e-05, + "loss": 0.7201, + "step": 1154 + }, + { + "epoch": 0.46, + "learning_rate": 1.1743557547546928e-05, + "loss": 0.6646, + "step": 1155 + }, + { + "epoch": 0.46, + "learning_rate": 1.1730820514282669e-05, + "loss": 0.7069, + "step": 1156 + }, + { + "epoch": 0.46, + "learning_rate": 1.1718080585697617e-05, + "loss": 0.7486, + "step": 1157 + }, + { + "epoch": 0.46, + "learning_rate": 1.1705337783103155e-05, + "loss": 0.7431, + "step": 1158 + }, + { + "epoch": 0.46, + "learning_rate": 1.1692592127815475e-05, + "loss": 0.7432, + "step": 1159 + }, + { + "epoch": 0.46, + "learning_rate": 1.1679843641155544e-05, + "loss": 0.703, + "step": 1160 + }, + { + "epoch": 0.46, + "learning_rate": 1.1667092344449053e-05, + "loss": 0.7146, + "step": 1161 + }, + { + "epoch": 0.46, + "learning_rate": 1.1654338259026413e-05, + "loss": 0.6921, + "step": 1162 + }, + { + "epoch": 0.46, + "learning_rate": 1.1641581406222682e-05, + "loss": 0.6552, + "step": 1163 + }, + { + "epoch": 0.46, + "learning_rate": 1.1628821807377557e-05, + "loss": 0.7622, + "step": 1164 + }, + { + "epoch": 0.47, + "learning_rate": 1.1616059483835325e-05, + "loss": 0.6498, + "step": 1165 + }, + { + "epoch": 0.47, + "learning_rate": 1.1603294456944835e-05, + "loss": 0.6836, + "step": 1166 + }, + { + "epoch": 0.47, + "learning_rate": 1.1590526748059453e-05, + "loss": 0.7107, + "step": 1167 + }, + { + "epoch": 0.47, + "learning_rate": 1.1577756378537033e-05, + "loss": 0.6549, + "step": 1168 + }, + { + "epoch": 0.47, + "learning_rate": 1.1564983369739884e-05, + "loss": 0.6989, + "step": 1169 + }, + { + "epoch": 0.47, + "learning_rate": 1.155220774303472e-05, + "loss": 0.5995, + "step": 1170 + }, + { + "epoch": 0.47, + "learning_rate": 1.1539429519792647e-05, + "loss": 0.6574, + "step": 1171 + }, + { + "epoch": 0.47, + "learning_rate": 1.1526648721389105e-05, + "loss": 0.7471, + "step": 1172 + }, + { + "epoch": 0.47, + "learning_rate": 1.1513865369203844e-05, + "loss": 0.6522, + "step": 1173 + }, + { + "epoch": 0.47, + "learning_rate": 1.1501079484620887e-05, + "loss": 0.7344, + "step": 1174 + }, + { + "epoch": 0.47, + "learning_rate": 1.1488291089028501e-05, + "loss": 0.7837, + "step": 1175 + }, + { + "epoch": 0.47, + "learning_rate": 1.1475500203819135e-05, + "loss": 0.7136, + "step": 1176 + }, + { + "epoch": 0.47, + "learning_rate": 1.1462706850389413e-05, + "loss": 0.7004, + "step": 1177 + }, + { + "epoch": 0.47, + "learning_rate": 1.1449911050140097e-05, + "loss": 0.6814, + "step": 1178 + }, + { + "epoch": 0.47, + "learning_rate": 1.1437112824476025e-05, + "loss": 0.7304, + "step": 1179 + }, + { + "epoch": 0.47, + "learning_rate": 1.14243121948061e-05, + "loss": 0.7298, + "step": 1180 + }, + { + "epoch": 0.47, + "learning_rate": 1.1411509182543253e-05, + "loss": 0.7389, + "step": 1181 + }, + { + "epoch": 0.47, + "learning_rate": 1.1398703809104392e-05, + "loss": 0.6989, + "step": 1182 + }, + { + "epoch": 0.47, + "learning_rate": 1.1385896095910374e-05, + "loss": 0.7083, + "step": 1183 + }, + { + "epoch": 0.47, + "learning_rate": 1.1373086064385974e-05, + "loss": 0.7471, + "step": 1184 + }, + { + "epoch": 0.47, + "learning_rate": 1.1360273735959848e-05, + "loss": 0.7214, + "step": 1185 + }, + { + "epoch": 0.47, + "learning_rate": 1.1347459132064485e-05, + "loss": 0.6346, + "step": 1186 + }, + { + "epoch": 0.47, + "learning_rate": 1.1334642274136186e-05, + "loss": 0.7178, + "step": 1187 + }, + { + "epoch": 0.47, + "learning_rate": 1.1321823183615027e-05, + "loss": 0.7022, + "step": 1188 + }, + { + "epoch": 0.47, + "learning_rate": 1.130900188194481e-05, + "loss": 0.7881, + "step": 1189 + }, + { + "epoch": 0.48, + "learning_rate": 1.1296178390573038e-05, + "loss": 0.7496, + "step": 1190 + }, + { + "epoch": 0.48, + "learning_rate": 1.1283352730950886e-05, + "loss": 0.7071, + "step": 1191 + }, + { + "epoch": 0.48, + "learning_rate": 1.1270524924533146e-05, + "loss": 0.6865, + "step": 1192 + }, + { + "epoch": 0.48, + "learning_rate": 1.1257694992778199e-05, + "loss": 0.6757, + "step": 1193 + }, + { + "epoch": 0.48, + "learning_rate": 1.1244862957147993e-05, + "loss": 0.6983, + "step": 1194 + }, + { + "epoch": 0.48, + "learning_rate": 1.1232028839107985e-05, + "loss": 0.6968, + "step": 1195 + }, + { + "epoch": 0.48, + "learning_rate": 1.1219192660127117e-05, + "loss": 0.6897, + "step": 1196 + }, + { + "epoch": 0.48, + "learning_rate": 1.1206354441677782e-05, + "loss": 0.7326, + "step": 1197 + }, + { + "epoch": 0.48, + "learning_rate": 1.1193514205235789e-05, + "loss": 0.6756, + "step": 1198 + }, + { + "epoch": 0.48, + "learning_rate": 1.1180671972280311e-05, + "loss": 0.7253, + "step": 1199 + }, + { + "epoch": 0.48, + "learning_rate": 1.1167827764293867e-05, + "loss": 0.6906, + "step": 1200 + }, + { + "epoch": 0.48, + "learning_rate": 1.115498160276228e-05, + "loss": 0.7203, + "step": 1201 + }, + { + "epoch": 0.48, + "learning_rate": 1.1142133509174644e-05, + "loss": 0.7159, + "step": 1202 + }, + { + "epoch": 0.48, + "learning_rate": 1.1129283505023273e-05, + "loss": 0.6763, + "step": 1203 + }, + { + "epoch": 0.48, + "learning_rate": 1.1116431611803696e-05, + "loss": 0.6315, + "step": 1204 + }, + { + "epoch": 0.48, + "learning_rate": 1.1103577851014587e-05, + "loss": 0.6792, + "step": 1205 + }, + { + "epoch": 0.48, + "learning_rate": 1.109072224415774e-05, + "loss": 0.6316, + "step": 1206 + }, + { + "epoch": 0.48, + "learning_rate": 1.1077864812738061e-05, + "loss": 0.6777, + "step": 1207 + }, + { + "epoch": 0.48, + "learning_rate": 1.1065005578263483e-05, + "loss": 0.7369, + "step": 1208 + }, + { + "epoch": 0.48, + "learning_rate": 1.1052144562244967e-05, + "loss": 0.7302, + "step": 1209 + }, + { + "epoch": 0.48, + "learning_rate": 1.1039281786196453e-05, + "loss": 0.6966, + "step": 1210 + }, + { + "epoch": 0.48, + "learning_rate": 1.1026417271634821e-05, + "loss": 0.733, + "step": 1211 + }, + { + "epoch": 0.48, + "learning_rate": 1.1013551040079866e-05, + "loss": 0.695, + "step": 1212 + }, + { + "epoch": 0.48, + "learning_rate": 1.1000683113054251e-05, + "loss": 0.7301, + "step": 1213 + }, + { + "epoch": 0.48, + "learning_rate": 1.0987813512083476e-05, + "loss": 0.7707, + "step": 1214 + }, + { + "epoch": 0.49, + "learning_rate": 1.0974942258695838e-05, + "loss": 0.6929, + "step": 1215 + }, + { + "epoch": 0.49, + "learning_rate": 1.0962069374422407e-05, + "loss": 0.6828, + "step": 1216 + }, + { + "epoch": 0.49, + "learning_rate": 1.0949194880796967e-05, + "loss": 0.7026, + "step": 1217 + }, + { + "epoch": 0.49, + "learning_rate": 1.0936318799356009e-05, + "loss": 0.702, + "step": 1218 + }, + { + "epoch": 0.49, + "learning_rate": 1.0923441151638676e-05, + "loss": 0.7636, + "step": 1219 + }, + { + "epoch": 0.49, + "learning_rate": 1.091056195918672e-05, + "loss": 0.7483, + "step": 1220 + }, + { + "epoch": 0.49, + "learning_rate": 1.0897681243544492e-05, + "loss": 0.7116, + "step": 1221 + }, + { + "epoch": 0.49, + "learning_rate": 1.0884799026258879e-05, + "loss": 0.7102, + "step": 1222 + }, + { + "epoch": 0.49, + "learning_rate": 1.0871915328879294e-05, + "loss": 0.6821, + "step": 1223 + }, + { + "epoch": 0.49, + "learning_rate": 1.085903017295761e-05, + "loss": 0.706, + "step": 1224 + }, + { + "epoch": 0.49, + "learning_rate": 1.0846143580048152e-05, + "loss": 0.6097, + "step": 1225 + }, + { + "epoch": 0.49, + "learning_rate": 1.083325557170764e-05, + "loss": 0.7672, + "step": 1226 + }, + { + "epoch": 0.49, + "learning_rate": 1.0820366169495171e-05, + "loss": 0.7437, + "step": 1227 + }, + { + "epoch": 0.49, + "learning_rate": 1.0807475394972164e-05, + "loss": 0.7873, + "step": 1228 + }, + { + "epoch": 0.49, + "learning_rate": 1.0794583269702341e-05, + "loss": 0.7397, + "step": 1229 + }, + { + "epoch": 0.49, + "learning_rate": 1.0781689815251682e-05, + "loss": 0.8086, + "step": 1230 + }, + { + "epoch": 0.49, + "learning_rate": 1.0768795053188379e-05, + "loss": 0.6964, + "step": 1231 + }, + { + "epoch": 0.49, + "learning_rate": 1.0755899005082837e-05, + "loss": 0.6802, + "step": 1232 + }, + { + "epoch": 0.49, + "learning_rate": 1.0743001692507586e-05, + "loss": 0.684, + "step": 1233 + }, + { + "epoch": 0.49, + "learning_rate": 1.0730103137037284e-05, + "loss": 0.6481, + "step": 1234 + }, + { + "epoch": 0.49, + "learning_rate": 1.0717203360248669e-05, + "loss": 0.5927, + "step": 1235 + }, + { + "epoch": 0.49, + "learning_rate": 1.070430238372052e-05, + "loss": 0.6751, + "step": 1236 + }, + { + "epoch": 0.49, + "learning_rate": 1.0691400229033617e-05, + "loss": 0.6285, + "step": 1237 + }, + { + "epoch": 0.49, + "learning_rate": 1.067849691777072e-05, + "loss": 0.7115, + "step": 1238 + }, + { + "epoch": 0.49, + "learning_rate": 1.0665592471516518e-05, + "loss": 0.6953, + "step": 1239 + }, + { + "epoch": 0.5, + "learning_rate": 1.0652686911857599e-05, + "loss": 0.7741, + "step": 1240 + }, + { + "epoch": 0.5, + "learning_rate": 1.0639780260382417e-05, + "loss": 0.6922, + "step": 1241 + }, + { + "epoch": 0.5, + "learning_rate": 1.062687253868125e-05, + "loss": 0.7097, + "step": 1242 + }, + { + "epoch": 0.5, + "learning_rate": 1.0613963768346167e-05, + "loss": 0.5847, + "step": 1243 + }, + { + "epoch": 0.5, + "learning_rate": 1.060105397097099e-05, + "loss": 0.7531, + "step": 1244 + }, + { + "epoch": 0.5, + "learning_rate": 1.0588143168151257e-05, + "loss": 0.6995, + "step": 1245 + }, + { + "epoch": 0.5, + "learning_rate": 1.0575231381484195e-05, + "loss": 0.6386, + "step": 1246 + }, + { + "epoch": 0.5, + "learning_rate": 1.0562318632568667e-05, + "loss": 0.7166, + "step": 1247 + }, + { + "epoch": 0.5, + "learning_rate": 1.0549404943005157e-05, + "loss": 0.7139, + "step": 1248 + }, + { + "epoch": 0.5, + "learning_rate": 1.0536490334395713e-05, + "loss": 0.714, + "step": 1249 + }, + { + "epoch": 0.5, + "learning_rate": 1.0523574828343923e-05, + "loss": 0.7358, + "step": 1250 + }, + { + "epoch": 0.5, + "learning_rate": 1.051065844645488e-05, + "loss": 0.6637, + "step": 1251 + }, + { + "epoch": 0.5, + "learning_rate": 1.049774121033514e-05, + "loss": 0.7453, + "step": 1252 + }, + { + "epoch": 0.5, + "learning_rate": 1.0484823141592682e-05, + "loss": 0.6844, + "step": 1253 + }, + { + "epoch": 0.5, + "learning_rate": 1.0471904261836888e-05, + "loss": 0.6509, + "step": 1254 + }, + { + "epoch": 0.5, + "learning_rate": 1.045898459267849e-05, + "loss": 0.7113, + "step": 1255 + }, + { + "epoch": 0.5, + "learning_rate": 1.0446064155729538e-05, + "loss": 0.6989, + "step": 1256 + }, + { + "epoch": 0.5, + "learning_rate": 1.0433142972603378e-05, + "loss": 0.6793, + "step": 1257 + }, + { + "epoch": 0.5, + "learning_rate": 1.0420221064914592e-05, + "loss": 0.7064, + "step": 1258 + }, + { + "epoch": 0.5, + "learning_rate": 1.0407298454278983e-05, + "loss": 0.7381, + "step": 1259 + }, + { + "epoch": 0.5, + "learning_rate": 1.0394375162313519e-05, + "loss": 0.6665, + "step": 1260 + }, + { + "epoch": 0.5, + "learning_rate": 1.0381451210636321e-05, + "loss": 0.704, + "step": 1261 + }, + { + "epoch": 0.5, + "learning_rate": 1.0368526620866604e-05, + "loss": 0.6968, + "step": 1262 + }, + { + "epoch": 0.5, + "learning_rate": 1.0355601414624653e-05, + "loss": 0.7186, + "step": 1263 + }, + { + "epoch": 0.5, + "learning_rate": 1.0342675613531785e-05, + "loss": 0.6175, + "step": 1264 + }, + { + "epoch": 0.5, + "learning_rate": 1.0329749239210311e-05, + "loss": 0.7402, + "step": 1265 + }, + { + "epoch": 0.51, + "learning_rate": 1.0316822313283504e-05, + "loss": 0.6249, + "step": 1266 + }, + { + "epoch": 0.51, + "learning_rate": 1.0303894857375557e-05, + "loss": 0.7534, + "step": 1267 + }, + { + "epoch": 0.51, + "learning_rate": 1.0290966893111547e-05, + "loss": 0.6938, + "step": 1268 + }, + { + "epoch": 0.51, + "learning_rate": 1.0278038442117405e-05, + "loss": 0.6359, + "step": 1269 + }, + { + "epoch": 0.51, + "learning_rate": 1.0265109526019878e-05, + "loss": 0.7451, + "step": 1270 + }, + { + "epoch": 0.51, + "learning_rate": 1.025218016644649e-05, + "loss": 0.7267, + "step": 1271 + }, + { + "epoch": 0.51, + "learning_rate": 1.0239250385025496e-05, + "loss": 0.669, + "step": 1272 + }, + { + "epoch": 0.51, + "learning_rate": 1.0226320203385877e-05, + "loss": 0.6472, + "step": 1273 + }, + { + "epoch": 0.51, + "learning_rate": 1.0213389643157268e-05, + "loss": 0.7005, + "step": 1274 + }, + { + "epoch": 0.51, + "learning_rate": 1.020045872596994e-05, + "loss": 0.6913, + "step": 1275 + }, + { + "epoch": 0.51, + "learning_rate": 1.0187527473454764e-05, + "loss": 0.726, + "step": 1276 + }, + { + "epoch": 0.51, + "learning_rate": 1.0174595907243172e-05, + "loss": 0.7062, + "step": 1277 + }, + { + "epoch": 0.51, + "learning_rate": 1.016166404896712e-05, + "loss": 0.5962, + "step": 1278 + }, + { + "epoch": 0.51, + "learning_rate": 1.014873192025905e-05, + "loss": 0.7528, + "step": 1279 + }, + { + "epoch": 0.51, + "learning_rate": 1.0135799542751861e-05, + "loss": 0.7011, + "step": 1280 + }, + { + "epoch": 0.51, + "learning_rate": 1.0122866938078863e-05, + "loss": 0.6372, + "step": 1281 + }, + { + "epoch": 0.51, + "learning_rate": 1.0109934127873751e-05, + "loss": 0.6334, + "step": 1282 + }, + { + "epoch": 0.51, + "learning_rate": 1.0097001133770561e-05, + "loss": 0.7536, + "step": 1283 + }, + { + "epoch": 0.51, + "learning_rate": 1.0084067977403637e-05, + "loss": 0.7381, + "step": 1284 + }, + { + "epoch": 0.51, + "learning_rate": 1.0071134680407593e-05, + "loss": 0.5806, + "step": 1285 + }, + { + "epoch": 0.51, + "learning_rate": 1.0058201264417285e-05, + "loss": 0.7093, + "step": 1286 + }, + { + "epoch": 0.51, + "learning_rate": 1.0045267751067758e-05, + "loss": 0.6829, + "step": 1287 + }, + { + "epoch": 0.51, + "learning_rate": 1.0032334161994225e-05, + "loss": 0.7844, + "step": 1288 + }, + { + "epoch": 0.51, + "learning_rate": 1.0019400518832028e-05, + "loss": 0.6881, + "step": 1289 + }, + { + "epoch": 0.51, + "learning_rate": 1.0006466843216593e-05, + "loss": 0.6956, + "step": 1290 + }, + { + "epoch": 0.52, + "learning_rate": 9.993533156783409e-06, + "loss": 0.6346, + "step": 1291 + }, + { + "epoch": 0.52, + "learning_rate": 9.980599481167977e-06, + "loss": 0.7057, + "step": 1292 + }, + { + "epoch": 0.52, + "learning_rate": 9.967665838005777e-06, + "loss": 0.7215, + "step": 1293 + }, + { + "epoch": 0.52, + "learning_rate": 9.954732248932243e-06, + "loss": 0.6753, + "step": 1294 + }, + { + "epoch": 0.52, + "learning_rate": 9.941798735582718e-06, + "loss": 0.7567, + "step": 1295 + }, + { + "epoch": 0.52, + "learning_rate": 9.92886531959241e-06, + "loss": 0.6844, + "step": 1296 + }, + { + "epoch": 0.52, + "learning_rate": 9.915932022596368e-06, + "loss": 0.6605, + "step": 1297 + }, + { + "epoch": 0.52, + "learning_rate": 9.90299886622944e-06, + "loss": 0.6211, + "step": 1298 + }, + { + "epoch": 0.52, + "learning_rate": 9.890065872126252e-06, + "loss": 0.715, + "step": 1299 + }, + { + "epoch": 0.52, + "learning_rate": 9.877133061921138e-06, + "loss": 0.6978, + "step": 1300 + }, + { + "epoch": 0.52, + "learning_rate": 9.864200457248144e-06, + "loss": 0.6383, + "step": 1301 + }, + { + "epoch": 0.52, + "learning_rate": 9.851268079740951e-06, + "loss": 0.6712, + "step": 1302 + }, + { + "epoch": 0.52, + "learning_rate": 9.838335951032883e-06, + "loss": 0.696, + "step": 1303 + }, + { + "epoch": 0.52, + "learning_rate": 9.82540409275683e-06, + "loss": 0.623, + "step": 1304 + }, + { + "epoch": 0.52, + "learning_rate": 9.81247252654524e-06, + "loss": 0.657, + "step": 1305 + }, + { + "epoch": 0.52, + "learning_rate": 9.799541274030065e-06, + "loss": 0.6868, + "step": 1306 + }, + { + "epoch": 0.52, + "learning_rate": 9.786610356842734e-06, + "loss": 0.645, + "step": 1307 + }, + { + "epoch": 0.52, + "learning_rate": 9.773679796614124e-06, + "loss": 0.6588, + "step": 1308 + }, + { + "epoch": 0.52, + "learning_rate": 9.760749614974506e-06, + "loss": 0.6743, + "step": 1309 + }, + { + "epoch": 0.52, + "learning_rate": 9.747819833553516e-06, + "loss": 0.6128, + "step": 1310 + }, + { + "epoch": 0.52, + "learning_rate": 9.734890473980122e-06, + "loss": 0.6667, + "step": 1311 + }, + { + "epoch": 0.52, + "learning_rate": 9.721961557882597e-06, + "loss": 0.6599, + "step": 1312 + }, + { + "epoch": 0.52, + "learning_rate": 9.709033106888456e-06, + "loss": 0.6981, + "step": 1313 + }, + { + "epoch": 0.52, + "learning_rate": 9.696105142624448e-06, + "loss": 0.7234, + "step": 1314 + }, + { + "epoch": 0.52, + "learning_rate": 9.683177686716501e-06, + "loss": 0.7089, + "step": 1315 + }, + { + "epoch": 0.53, + "learning_rate": 9.670250760789692e-06, + "loss": 0.6594, + "step": 1316 + }, + { + "epoch": 0.53, + "learning_rate": 9.657324386468217e-06, + "loss": 0.7767, + "step": 1317 + }, + { + "epoch": 0.53, + "learning_rate": 9.644398585375352e-06, + "loss": 0.7641, + "step": 1318 + }, + { + "epoch": 0.53, + "learning_rate": 9.6314733791334e-06, + "loss": 0.6762, + "step": 1319 + }, + { + "epoch": 0.53, + "learning_rate": 9.61854878936368e-06, + "loss": 0.7169, + "step": 1320 + }, + { + "epoch": 0.53, + "learning_rate": 9.605624837686483e-06, + "loss": 0.7482, + "step": 1321 + }, + { + "epoch": 0.53, + "learning_rate": 9.592701545721022e-06, + "loss": 0.6473, + "step": 1322 + }, + { + "epoch": 0.53, + "learning_rate": 9.57977893508541e-06, + "loss": 0.6729, + "step": 1323 + }, + { + "epoch": 0.53, + "learning_rate": 9.566857027396627e-06, + "loss": 0.6265, + "step": 1324 + }, + { + "epoch": 0.53, + "learning_rate": 9.553935844270464e-06, + "loss": 0.7421, + "step": 1325 + }, + { + "epoch": 0.53, + "learning_rate": 9.541015407321514e-06, + "loss": 0.7286, + "step": 1326 + }, + { + "epoch": 0.53, + "learning_rate": 9.528095738163117e-06, + "loss": 0.6587, + "step": 1327 + }, + { + "epoch": 0.53, + "learning_rate": 9.515176858407323e-06, + "loss": 0.6696, + "step": 1328 + }, + { + "epoch": 0.53, + "learning_rate": 9.502258789664865e-06, + "loss": 0.6711, + "step": 1329 + }, + { + "epoch": 0.53, + "learning_rate": 9.489341553545121e-06, + "loss": 0.7447, + "step": 1330 + }, + { + "epoch": 0.53, + "learning_rate": 9.47642517165608e-06, + "loss": 0.7053, + "step": 1331 + }, + { + "epoch": 0.53, + "learning_rate": 9.46350966560429e-06, + "loss": 0.6059, + "step": 1332 + }, + { + "epoch": 0.53, + "learning_rate": 9.450595056994848e-06, + "loss": 0.6334, + "step": 1333 + }, + { + "epoch": 0.53, + "learning_rate": 9.437681367431335e-06, + "loss": 0.6278, + "step": 1334 + }, + { + "epoch": 0.53, + "learning_rate": 9.424768618515809e-06, + "loss": 0.7077, + "step": 1335 + }, + { + "epoch": 0.53, + "learning_rate": 9.411856831848745e-06, + "loss": 0.7211, + "step": 1336 + }, + { + "epoch": 0.53, + "learning_rate": 9.398946029029015e-06, + "loss": 0.7562, + "step": 1337 + }, + { + "epoch": 0.53, + "learning_rate": 9.386036231653836e-06, + "loss": 0.7211, + "step": 1338 + }, + { + "epoch": 0.53, + "learning_rate": 9.37312746131875e-06, + "loss": 0.7262, + "step": 1339 + }, + { + "epoch": 0.53, + "learning_rate": 9.360219739617585e-06, + "loss": 0.6379, + "step": 1340 + }, + { + "epoch": 0.54, + "learning_rate": 9.347313088142403e-06, + "loss": 0.5732, + "step": 1341 + }, + { + "epoch": 0.54, + "learning_rate": 9.334407528483487e-06, + "loss": 0.7226, + "step": 1342 + }, + { + "epoch": 0.54, + "learning_rate": 9.321503082229281e-06, + "loss": 0.6958, + "step": 1343 + }, + { + "epoch": 0.54, + "learning_rate": 9.308599770966385e-06, + "loss": 0.6887, + "step": 1344 + }, + { + "epoch": 0.54, + "learning_rate": 9.295697616279483e-06, + "loss": 0.7044, + "step": 1345 + }, + { + "epoch": 0.54, + "learning_rate": 9.282796639751333e-06, + "loss": 0.699, + "step": 1346 + }, + { + "epoch": 0.54, + "learning_rate": 9.269896862962718e-06, + "loss": 0.7058, + "step": 1347 + }, + { + "epoch": 0.54, + "learning_rate": 9.256998307492415e-06, + "loss": 0.7372, + "step": 1348 + }, + { + "epoch": 0.54, + "learning_rate": 9.244100994917167e-06, + "loss": 0.7239, + "step": 1349 + }, + { + "epoch": 0.54, + "learning_rate": 9.231204946811624e-06, + "loss": 0.5787, + "step": 1350 + }, + { + "epoch": 0.54, + "learning_rate": 9.218310184748325e-06, + "loss": 0.6239, + "step": 1351 + }, + { + "epoch": 0.54, + "learning_rate": 9.20541673029766e-06, + "loss": 0.6161, + "step": 1352 + }, + { + "epoch": 0.54, + "learning_rate": 9.192524605027838e-06, + "loss": 0.6972, + "step": 1353 + }, + { + "epoch": 0.54, + "learning_rate": 9.179633830504832e-06, + "loss": 0.754, + "step": 1354 + }, + { + "epoch": 0.54, + "learning_rate": 9.166744428292362e-06, + "loss": 0.7415, + "step": 1355 + }, + { + "epoch": 0.54, + "learning_rate": 9.153856419951854e-06, + "loss": 0.7791, + "step": 1356 + }, + { + "epoch": 0.54, + "learning_rate": 9.140969827042392e-06, + "loss": 0.7447, + "step": 1357 + }, + { + "epoch": 0.54, + "learning_rate": 9.128084671120709e-06, + "loss": 0.6897, + "step": 1358 + }, + { + "epoch": 0.54, + "learning_rate": 9.115200973741123e-06, + "loss": 0.6918, + "step": 1359 + }, + { + "epoch": 0.54, + "learning_rate": 9.102318756455513e-06, + "loss": 0.7557, + "step": 1360 + }, + { + "epoch": 0.54, + "learning_rate": 9.089438040813282e-06, + "loss": 0.6692, + "step": 1361 + }, + { + "epoch": 0.54, + "learning_rate": 9.076558848361328e-06, + "loss": 0.7041, + "step": 1362 + }, + { + "epoch": 0.54, + "learning_rate": 9.063681200643993e-06, + "loss": 0.6908, + "step": 1363 + }, + { + "epoch": 0.54, + "learning_rate": 9.050805119203035e-06, + "loss": 0.7037, + "step": 1364 + }, + { + "epoch": 0.54, + "learning_rate": 9.037930625577598e-06, + "loss": 0.6786, + "step": 1365 + }, + { + "epoch": 0.55, + "learning_rate": 9.025057741304164e-06, + "loss": 0.6348, + "step": 1366 + }, + { + "epoch": 0.55, + "learning_rate": 9.012186487916526e-06, + "loss": 0.7015, + "step": 1367 + }, + { + "epoch": 0.55, + "learning_rate": 8.999316886945752e-06, + "loss": 0.7216, + "step": 1368 + }, + { + "epoch": 0.55, + "learning_rate": 8.986448959920137e-06, + "loss": 0.6494, + "step": 1369 + }, + { + "epoch": 0.55, + "learning_rate": 8.97358272836518e-06, + "loss": 0.6428, + "step": 1370 + }, + { + "epoch": 0.55, + "learning_rate": 8.96071821380355e-06, + "loss": 0.7302, + "step": 1371 + }, + { + "epoch": 0.55, + "learning_rate": 8.947855437755036e-06, + "loss": 0.666, + "step": 1372 + }, + { + "epoch": 0.55, + "learning_rate": 8.934994421736519e-06, + "loss": 0.6802, + "step": 1373 + }, + { + "epoch": 0.55, + "learning_rate": 8.922135187261944e-06, + "loss": 0.7163, + "step": 1374 + }, + { + "epoch": 0.55, + "learning_rate": 8.90927775584226e-06, + "loss": 0.6934, + "step": 1375 + }, + { + "epoch": 0.55, + "learning_rate": 8.896422148985418e-06, + "loss": 0.6987, + "step": 1376 + }, + { + "epoch": 0.55, + "learning_rate": 8.883568388196308e-06, + "loss": 0.6894, + "step": 1377 + }, + { + "epoch": 0.55, + "learning_rate": 8.87071649497673e-06, + "loss": 0.664, + "step": 1378 + }, + { + "epoch": 0.55, + "learning_rate": 8.85786649082536e-06, + "loss": 0.687, + "step": 1379 + }, + { + "epoch": 0.55, + "learning_rate": 8.845018397237722e-06, + "loss": 0.7738, + "step": 1380 + }, + { + "epoch": 0.55, + "learning_rate": 8.832172235706137e-06, + "loss": 0.7605, + "step": 1381 + }, + { + "epoch": 0.55, + "learning_rate": 8.819328027719692e-06, + "loss": 0.6862, + "step": 1382 + }, + { + "epoch": 0.55, + "learning_rate": 8.806485794764215e-06, + "loss": 0.7155, + "step": 1383 + }, + { + "epoch": 0.55, + "learning_rate": 8.793645558322218e-06, + "loss": 0.6947, + "step": 1384 + }, + { + "epoch": 0.55, + "learning_rate": 8.780807339872886e-06, + "loss": 0.7652, + "step": 1385 + }, + { + "epoch": 0.55, + "learning_rate": 8.76797116089202e-06, + "loss": 0.7067, + "step": 1386 + }, + { + "epoch": 0.55, + "learning_rate": 8.755137042852012e-06, + "loss": 0.6682, + "step": 1387 + }, + { + "epoch": 0.55, + "learning_rate": 8.742305007221801e-06, + "loss": 0.6264, + "step": 1388 + }, + { + "epoch": 0.55, + "learning_rate": 8.729475075466856e-06, + "loss": 0.738, + "step": 1389 + }, + { + "epoch": 0.55, + "learning_rate": 8.716647269049115e-06, + "loss": 0.7141, + "step": 1390 + }, + { + "epoch": 0.56, + "learning_rate": 8.703821609426963e-06, + "loss": 0.6528, + "step": 1391 + }, + { + "epoch": 0.56, + "learning_rate": 8.690998118055193e-06, + "loss": 0.7106, + "step": 1392 + }, + { + "epoch": 0.56, + "learning_rate": 8.678176816384975e-06, + "loss": 0.7204, + "step": 1393 + }, + { + "epoch": 0.56, + "learning_rate": 8.665357725863816e-06, + "loss": 0.7355, + "step": 1394 + }, + { + "epoch": 0.56, + "learning_rate": 8.65254086793552e-06, + "loss": 0.6963, + "step": 1395 + }, + { + "epoch": 0.56, + "learning_rate": 8.639726264040159e-06, + "loss": 0.7263, + "step": 1396 + }, + { + "epoch": 0.56, + "learning_rate": 8.626913935614026e-06, + "loss": 0.6098, + "step": 1397 + }, + { + "epoch": 0.56, + "learning_rate": 8.61410390408963e-06, + "loss": 0.7582, + "step": 1398 + }, + { + "epoch": 0.56, + "learning_rate": 8.601296190895611e-06, + "loss": 0.6922, + "step": 1399 + }, + { + "epoch": 0.56, + "learning_rate": 8.588490817456749e-06, + "loss": 0.7528, + "step": 1400 + }, + { + "epoch": 0.56, + "learning_rate": 8.575687805193902e-06, + "loss": 0.7138, + "step": 1401 + }, + { + "epoch": 0.56, + "learning_rate": 8.562887175523977e-06, + "loss": 0.6996, + "step": 1402 + }, + { + "epoch": 0.56, + "learning_rate": 8.550088949859906e-06, + "loss": 0.7408, + "step": 1403 + }, + { + "epoch": 0.56, + "learning_rate": 8.53729314961059e-06, + "loss": 0.5894, + "step": 1404 + }, + { + "epoch": 0.56, + "learning_rate": 8.52449979618087e-06, + "loss": 0.6789, + "step": 1405 + }, + { + "epoch": 0.56, + "learning_rate": 8.511708910971506e-06, + "loss": 0.7069, + "step": 1406 + }, + { + "epoch": 0.56, + "learning_rate": 8.498920515379112e-06, + "loss": 0.7547, + "step": 1407 + }, + { + "epoch": 0.56, + "learning_rate": 8.486134630796157e-06, + "loss": 0.6827, + "step": 1408 + }, + { + "epoch": 0.56, + "learning_rate": 8.4733512786109e-06, + "loss": 0.634, + "step": 1409 + }, + { + "epoch": 0.56, + "learning_rate": 8.460570480207358e-06, + "loss": 0.6724, + "step": 1410 + }, + { + "epoch": 0.56, + "learning_rate": 8.447792256965282e-06, + "loss": 0.6559, + "step": 1411 + }, + { + "epoch": 0.56, + "learning_rate": 8.43501663026012e-06, + "loss": 0.6817, + "step": 1412 + }, + { + "epoch": 0.56, + "learning_rate": 8.42224362146297e-06, + "loss": 0.6884, + "step": 1413 + }, + { + "epoch": 0.56, + "learning_rate": 8.409473251940549e-06, + "loss": 0.6585, + "step": 1414 + }, + { + "epoch": 0.56, + "learning_rate": 8.39670554305517e-06, + "loss": 0.6983, + "step": 1415 + }, + { + "epoch": 0.57, + "learning_rate": 8.383940516164677e-06, + "loss": 0.7004, + "step": 1416 + }, + { + "epoch": 0.57, + "learning_rate": 8.371178192622445e-06, + "loss": 0.7176, + "step": 1417 + }, + { + "epoch": 0.57, + "learning_rate": 8.358418593777322e-06, + "loss": 0.6674, + "step": 1418 + }, + { + "epoch": 0.57, + "learning_rate": 8.345661740973592e-06, + "loss": 0.6984, + "step": 1419 + }, + { + "epoch": 0.57, + "learning_rate": 8.332907655550948e-06, + "loss": 0.6485, + "step": 1420 + }, + { + "epoch": 0.57, + "learning_rate": 8.32015635884446e-06, + "loss": 0.6533, + "step": 1421 + }, + { + "epoch": 0.57, + "learning_rate": 8.307407872184527e-06, + "loss": 0.6748, + "step": 1422 + }, + { + "epoch": 0.57, + "learning_rate": 8.294662216896848e-06, + "loss": 0.6556, + "step": 1423 + }, + { + "epoch": 0.57, + "learning_rate": 8.281919414302388e-06, + "loss": 0.6194, + "step": 1424 + }, + { + "epoch": 0.57, + "learning_rate": 8.269179485717331e-06, + "loss": 0.7139, + "step": 1425 + }, + { + "epoch": 0.57, + "learning_rate": 8.256442452453073e-06, + "loss": 0.6401, + "step": 1426 + }, + { + "epoch": 0.57, + "learning_rate": 8.243708335816145e-06, + "loss": 0.7004, + "step": 1427 + }, + { + "epoch": 0.57, + "learning_rate": 8.230977157108213e-06, + "loss": 0.633, + "step": 1428 + }, + { + "epoch": 0.57, + "learning_rate": 8.218248937626018e-06, + "loss": 0.7269, + "step": 1429 + }, + { + "epoch": 0.57, + "learning_rate": 8.205523698661365e-06, + "loss": 0.7099, + "step": 1430 + }, + { + "epoch": 0.57, + "learning_rate": 8.19280146150106e-06, + "loss": 0.6306, + "step": 1431 + }, + { + "epoch": 0.57, + "learning_rate": 8.180082247426895e-06, + "loss": 0.7135, + "step": 1432 + }, + { + "epoch": 0.57, + "learning_rate": 8.167366077715598e-06, + "loss": 0.6903, + "step": 1433 + }, + { + "epoch": 0.57, + "learning_rate": 8.15465297363881e-06, + "loss": 0.7208, + "step": 1434 + }, + { + "epoch": 0.57, + "learning_rate": 8.141942956463049e-06, + "loss": 0.7039, + "step": 1435 + }, + { + "epoch": 0.57, + "learning_rate": 8.129236047449654e-06, + "loss": 0.677, + "step": 1436 + }, + { + "epoch": 0.57, + "learning_rate": 8.116532267854781e-06, + "loss": 0.6221, + "step": 1437 + }, + { + "epoch": 0.57, + "learning_rate": 8.103831638929337e-06, + "loss": 0.7231, + "step": 1438 + }, + { + "epoch": 0.57, + "learning_rate": 8.091134181918971e-06, + "loss": 0.6373, + "step": 1439 + }, + { + "epoch": 0.57, + "learning_rate": 8.078439918064016e-06, + "loss": 0.7476, + "step": 1440 + }, + { + "epoch": 0.58, + "learning_rate": 8.065748868599471e-06, + "loss": 0.7141, + "step": 1441 + }, + { + "epoch": 0.58, + "learning_rate": 8.053061054754952e-06, + "loss": 0.6323, + "step": 1442 + }, + { + "epoch": 0.58, + "learning_rate": 8.04037649775466e-06, + "loss": 0.7174, + "step": 1443 + }, + { + "epoch": 0.58, + "learning_rate": 8.027695218817366e-06, + "loss": 0.6481, + "step": 1444 + }, + { + "epoch": 0.58, + "learning_rate": 8.015017239156338e-06, + "loss": 0.6869, + "step": 1445 + }, + { + "epoch": 0.58, + "learning_rate": 8.002342579979327e-06, + "loss": 0.6772, + "step": 1446 + }, + { + "epoch": 0.58, + "learning_rate": 7.989671262488534e-06, + "loss": 0.6974, + "step": 1447 + }, + { + "epoch": 0.58, + "learning_rate": 7.977003307880579e-06, + "loss": 0.7023, + "step": 1448 + }, + { + "epoch": 0.58, + "learning_rate": 7.964338737346441e-06, + "loss": 0.7041, + "step": 1449 + }, + { + "epoch": 0.58, + "learning_rate": 7.951677572071446e-06, + "loss": 0.7428, + "step": 1450 + }, + { + "epoch": 0.58, + "learning_rate": 7.939019833235223e-06, + "loss": 0.6636, + "step": 1451 + }, + { + "epoch": 0.58, + "learning_rate": 7.926365542011662e-06, + "loss": 0.7002, + "step": 1452 + }, + { + "epoch": 0.58, + "learning_rate": 7.9137147195689e-06, + "loss": 0.693, + "step": 1453 + }, + { + "epoch": 0.58, + "learning_rate": 7.901067387069264e-06, + "loss": 0.693, + "step": 1454 + }, + { + "epoch": 0.58, + "learning_rate": 7.888423565669236e-06, + "loss": 0.7093, + "step": 1455 + }, + { + "epoch": 0.58, + "learning_rate": 7.875783276519435e-06, + "loss": 0.7258, + "step": 1456 + }, + { + "epoch": 0.58, + "learning_rate": 7.863146540764572e-06, + "loss": 0.7135, + "step": 1457 + }, + { + "epoch": 0.58, + "learning_rate": 7.850513379543405e-06, + "loss": 0.6715, + "step": 1458 + }, + { + "epoch": 0.58, + "learning_rate": 7.837883813988717e-06, + "loss": 0.634, + "step": 1459 + }, + { + "epoch": 0.58, + "learning_rate": 7.825257865227278e-06, + "loss": 0.6738, + "step": 1460 + }, + { + "epoch": 0.58, + "learning_rate": 7.812635554379802e-06, + "loss": 0.74, + "step": 1461 + }, + { + "epoch": 0.58, + "learning_rate": 7.800016902560924e-06, + "loss": 0.7239, + "step": 1462 + }, + { + "epoch": 0.58, + "learning_rate": 7.787401930879159e-06, + "loss": 0.6026, + "step": 1463 + }, + { + "epoch": 0.58, + "learning_rate": 7.774790660436857e-06, + "loss": 0.6539, + "step": 1464 + }, + { + "epoch": 0.58, + "learning_rate": 7.762183112330182e-06, + "loss": 0.6735, + "step": 1465 + }, + { + "epoch": 0.59, + "learning_rate": 7.74957930764908e-06, + "loss": 0.7306, + "step": 1466 + }, + { + "epoch": 0.59, + "learning_rate": 7.736979267477218e-06, + "loss": 0.7623, + "step": 1467 + }, + { + "epoch": 0.59, + "learning_rate": 7.724383012891976e-06, + "loss": 0.6747, + "step": 1468 + }, + { + "epoch": 0.59, + "learning_rate": 7.7117905649644e-06, + "loss": 0.6564, + "step": 1469 + }, + { + "epoch": 0.59, + "learning_rate": 7.699201944759164e-06, + "loss": 0.683, + "step": 1470 + }, + { + "epoch": 0.59, + "learning_rate": 7.686617173334547e-06, + "loss": 0.7416, + "step": 1471 + }, + { + "epoch": 0.59, + "learning_rate": 7.674036271742388e-06, + "loss": 0.6835, + "step": 1472 + }, + { + "epoch": 0.59, + "learning_rate": 7.661459261028045e-06, + "loss": 0.7603, + "step": 1473 + }, + { + "epoch": 0.59, + "learning_rate": 7.648886162230373e-06, + "loss": 0.6504, + "step": 1474 + }, + { + "epoch": 0.59, + "learning_rate": 7.63631699638168e-06, + "loss": 0.6985, + "step": 1475 + }, + { + "epoch": 0.59, + "learning_rate": 7.623751784507707e-06, + "loss": 0.6696, + "step": 1476 + }, + { + "epoch": 0.59, + "learning_rate": 7.6111905476275606e-06, + "loss": 0.7005, + "step": 1477 + }, + { + "epoch": 0.59, + "learning_rate": 7.598633306753715e-06, + "loss": 0.7119, + "step": 1478 + }, + { + "epoch": 0.59, + "learning_rate": 7.586080082891949e-06, + "loss": 0.6827, + "step": 1479 + }, + { + "epoch": 0.59, + "learning_rate": 7.57353089704133e-06, + "loss": 0.6551, + "step": 1480 + }, + { + "epoch": 0.59, + "learning_rate": 7.5609857701941645e-06, + "loss": 0.6936, + "step": 1481 + }, + { + "epoch": 0.59, + "learning_rate": 7.548444723335975e-06, + "loss": 0.6802, + "step": 1482 + }, + { + "epoch": 0.59, + "learning_rate": 7.535907777445449e-06, + "loss": 0.6485, + "step": 1483 + }, + { + "epoch": 0.59, + "learning_rate": 7.523374953494422e-06, + "loss": 0.6586, + "step": 1484 + }, + { + "epoch": 0.59, + "learning_rate": 7.510846272447839e-06, + "loss": 0.6641, + "step": 1485 + }, + { + "epoch": 0.59, + "learning_rate": 7.498321755263704e-06, + "loss": 0.6307, + "step": 1486 + }, + { + "epoch": 0.59, + "learning_rate": 7.485801422893057e-06, + "loss": 0.6755, + "step": 1487 + }, + { + "epoch": 0.59, + "learning_rate": 7.473285296279942e-06, + "loss": 0.6797, + "step": 1488 + }, + { + "epoch": 0.59, + "learning_rate": 7.460773396361371e-06, + "loss": 0.7302, + "step": 1489 + }, + { + "epoch": 0.59, + "learning_rate": 7.448265744067275e-06, + "loss": 0.7323, + "step": 1490 + }, + { + "epoch": 0.6, + "learning_rate": 7.435762360320491e-06, + "loss": 0.6731, + "step": 1491 + }, + { + "epoch": 0.6, + "learning_rate": 7.423263266036705e-06, + "loss": 0.7572, + "step": 1492 + }, + { + "epoch": 0.6, + "learning_rate": 7.410768482124431e-06, + "loss": 0.6739, + "step": 1493 + }, + { + "epoch": 0.6, + "learning_rate": 7.3982780294849795e-06, + "loss": 0.6767, + "step": 1494 + }, + { + "epoch": 0.6, + "learning_rate": 7.38579192901241e-06, + "loss": 0.7141, + "step": 1495 + }, + { + "epoch": 0.6, + "learning_rate": 7.373310201593498e-06, + "loss": 0.7204, + "step": 1496 + }, + { + "epoch": 0.6, + "learning_rate": 7.360832868107708e-06, + "loss": 0.7751, + "step": 1497 + }, + { + "epoch": 0.6, + "learning_rate": 7.348359949427159e-06, + "loss": 0.6852, + "step": 1498 + }, + { + "epoch": 0.6, + "learning_rate": 7.335891466416574e-06, + "loss": 0.56, + "step": 1499 + }, + { + "epoch": 0.6, + "learning_rate": 7.323427439933269e-06, + "loss": 0.6566, + "step": 1500 + }, + { + "epoch": 0.6, + "learning_rate": 7.31096789082709e-06, + "loss": 0.7349, + "step": 1501 + }, + { + "epoch": 0.6, + "learning_rate": 7.2985128399404016e-06, + "loss": 0.715, + "step": 1502 + }, + { + "epoch": 0.6, + "learning_rate": 7.2860623081080506e-06, + "loss": 0.5972, + "step": 1503 + }, + { + "epoch": 0.6, + "learning_rate": 7.273616316157312e-06, + "loss": 0.7061, + "step": 1504 + }, + { + "epoch": 0.6, + "learning_rate": 7.261174884907874e-06, + "loss": 0.6949, + "step": 1505 + }, + { + "epoch": 0.6, + "learning_rate": 7.248738035171787e-06, + "loss": 0.7098, + "step": 1506 + }, + { + "epoch": 0.6, + "learning_rate": 7.236305787753454e-06, + "loss": 0.6458, + "step": 1507 + }, + { + "epoch": 0.6, + "learning_rate": 7.223878163449563e-06, + "loss": 0.6401, + "step": 1508 + }, + { + "epoch": 0.6, + "learning_rate": 7.211455183049072e-06, + "loss": 0.7907, + "step": 1509 + }, + { + "epoch": 0.6, + "learning_rate": 7.199036867333179e-06, + "loss": 0.7432, + "step": 1510 + }, + { + "epoch": 0.6, + "learning_rate": 7.186623237075265e-06, + "loss": 0.6803, + "step": 1511 + }, + { + "epoch": 0.6, + "learning_rate": 7.174214313040888e-06, + "loss": 0.7481, + "step": 1512 + }, + { + "epoch": 0.6, + "learning_rate": 7.161810115987727e-06, + "loss": 0.7145, + "step": 1513 + }, + { + "epoch": 0.6, + "learning_rate": 7.14941066666555e-06, + "loss": 0.6793, + "step": 1514 + }, + { + "epoch": 0.6, + "learning_rate": 7.137015985816183e-06, + "loss": 0.7073, + "step": 1515 + }, + { + "epoch": 0.61, + "learning_rate": 7.124626094173484e-06, + "loss": 0.647, + "step": 1516 + }, + { + "epoch": 0.61, + "learning_rate": 7.1122410124632935e-06, + "loss": 0.5615, + "step": 1517 + }, + { + "epoch": 0.61, + "learning_rate": 7.0998607614034035e-06, + "loss": 0.6673, + "step": 1518 + }, + { + "epoch": 0.61, + "learning_rate": 7.087485361703532e-06, + "loss": 0.665, + "step": 1519 + }, + { + "epoch": 0.61, + "learning_rate": 7.075114834065272e-06, + "loss": 0.7226, + "step": 1520 + }, + { + "epoch": 0.61, + "learning_rate": 7.06274919918208e-06, + "loss": 0.6319, + "step": 1521 + }, + { + "epoch": 0.61, + "learning_rate": 7.0503884777392166e-06, + "loss": 0.7213, + "step": 1522 + }, + { + "epoch": 0.61, + "learning_rate": 7.0380326904137265e-06, + "loss": 0.6607, + "step": 1523 + }, + { + "epoch": 0.61, + "learning_rate": 7.025681857874398e-06, + "loss": 0.6817, + "step": 1524 + }, + { + "epoch": 0.61, + "learning_rate": 7.013336000781739e-06, + "loss": 0.6968, + "step": 1525 + }, + { + "epoch": 0.61, + "learning_rate": 7.000995139787929e-06, + "loss": 0.6864, + "step": 1526 + }, + { + "epoch": 0.61, + "learning_rate": 6.988659295536792e-06, + "loss": 0.6638, + "step": 1527 + }, + { + "epoch": 0.61, + "learning_rate": 6.9763284886637516e-06, + "loss": 0.7239, + "step": 1528 + }, + { + "epoch": 0.61, + "learning_rate": 6.964002739795814e-06, + "loss": 0.6517, + "step": 1529 + }, + { + "epoch": 0.61, + "learning_rate": 6.9516820695515285e-06, + "loss": 0.6417, + "step": 1530 + }, + { + "epoch": 0.61, + "learning_rate": 6.939366498540936e-06, + "loss": 0.7525, + "step": 1531 + }, + { + "epoch": 0.61, + "learning_rate": 6.927056047365558e-06, + "loss": 0.7167, + "step": 1532 + }, + { + "epoch": 0.61, + "learning_rate": 6.914750736618341e-06, + "loss": 0.6819, + "step": 1533 + }, + { + "epoch": 0.61, + "learning_rate": 6.9024505868836465e-06, + "loss": 0.6989, + "step": 1534 + }, + { + "epoch": 0.61, + "learning_rate": 6.890155618737189e-06, + "loss": 0.6683, + "step": 1535 + }, + { + "epoch": 0.61, + "learning_rate": 6.877865852746026e-06, + "loss": 0.6648, + "step": 1536 + }, + { + "epoch": 0.61, + "learning_rate": 6.865581309468506e-06, + "loss": 0.6508, + "step": 1537 + }, + { + "epoch": 0.61, + "learning_rate": 6.853302009454239e-06, + "loss": 0.7361, + "step": 1538 + }, + { + "epoch": 0.61, + "learning_rate": 6.841027973244077e-06, + "loss": 0.671, + "step": 1539 + }, + { + "epoch": 0.61, + "learning_rate": 6.828759221370049e-06, + "loss": 0.664, + "step": 1540 + }, + { + "epoch": 0.62, + "learning_rate": 6.816495774355362e-06, + "loss": 0.6628, + "step": 1541 + }, + { + "epoch": 0.62, + "learning_rate": 6.804237652714334e-06, + "loss": 0.6887, + "step": 1542 + }, + { + "epoch": 0.62, + "learning_rate": 6.7919848769523786e-06, + "loss": 0.6959, + "step": 1543 + }, + { + "epoch": 0.62, + "learning_rate": 6.779737467565974e-06, + "loss": 0.7291, + "step": 1544 + }, + { + "epoch": 0.62, + "learning_rate": 6.767495445042618e-06, + "loss": 0.6379, + "step": 1545 + }, + { + "epoch": 0.62, + "learning_rate": 6.755258829860791e-06, + "loss": 0.6475, + "step": 1546 + }, + { + "epoch": 0.62, + "learning_rate": 6.743027642489935e-06, + "loss": 0.6873, + "step": 1547 + }, + { + "epoch": 0.62, + "learning_rate": 6.730801903390412e-06, + "loss": 0.6846, + "step": 1548 + }, + { + "epoch": 0.62, + "learning_rate": 6.718581633013465e-06, + "loss": 0.671, + "step": 1549 + }, + { + "epoch": 0.62, + "learning_rate": 6.706366851801194e-06, + "loss": 0.6644, + "step": 1550 + }, + { + "epoch": 0.62, + "learning_rate": 6.694157580186516e-06, + "loss": 0.649, + "step": 1551 + }, + { + "epoch": 0.62, + "learning_rate": 6.681953838593125e-06, + "loss": 0.68, + "step": 1552 + }, + { + "epoch": 0.62, + "learning_rate": 6.669755647435474e-06, + "loss": 0.6287, + "step": 1553 + }, + { + "epoch": 0.62, + "learning_rate": 6.657563027118732e-06, + "loss": 0.618, + "step": 1554 + }, + { + "epoch": 0.62, + "learning_rate": 6.645375998038737e-06, + "loss": 0.7425, + "step": 1555 + }, + { + "epoch": 0.62, + "learning_rate": 6.633194580581978e-06, + "loss": 0.6856, + "step": 1556 + }, + { + "epoch": 0.62, + "learning_rate": 6.621018795125569e-06, + "loss": 0.6296, + "step": 1557 + }, + { + "epoch": 0.62, + "learning_rate": 6.6088486620371905e-06, + "loss": 0.7821, + "step": 1558 + }, + { + "epoch": 0.62, + "learning_rate": 6.596684201675069e-06, + "loss": 0.7136, + "step": 1559 + }, + { + "epoch": 0.62, + "learning_rate": 6.584525434387945e-06, + "loss": 0.6857, + "step": 1560 + }, + { + "epoch": 0.62, + "learning_rate": 6.572372380515031e-06, + "loss": 0.7551, + "step": 1561 + }, + { + "epoch": 0.62, + "learning_rate": 6.560225060385992e-06, + "loss": 0.6941, + "step": 1562 + }, + { + "epoch": 0.62, + "learning_rate": 6.54808349432089e-06, + "loss": 0.6764, + "step": 1563 + }, + { + "epoch": 0.62, + "learning_rate": 6.535947702630168e-06, + "loss": 0.727, + "step": 1564 + }, + { + "epoch": 0.62, + "learning_rate": 6.523817705614602e-06, + "loss": 0.5695, + "step": 1565 + }, + { + "epoch": 0.63, + "learning_rate": 6.51169352356529e-06, + "loss": 0.6763, + "step": 1566 + }, + { + "epoch": 0.63, + "learning_rate": 6.499575176763591e-06, + "loss": 0.6184, + "step": 1567 + }, + { + "epoch": 0.63, + "learning_rate": 6.487462685481103e-06, + "loss": 0.6486, + "step": 1568 + }, + { + "epoch": 0.63, + "learning_rate": 6.4753560699796305e-06, + "loss": 0.6916, + "step": 1569 + }, + { + "epoch": 0.63, + "learning_rate": 6.4632553505111484e-06, + "loss": 0.6838, + "step": 1570 + }, + { + "epoch": 0.63, + "learning_rate": 6.451160547317777e-06, + "loss": 0.7143, + "step": 1571 + }, + { + "epoch": 0.63, + "learning_rate": 6.439071680631725e-06, + "loss": 0.6941, + "step": 1572 + }, + { + "epoch": 0.63, + "learning_rate": 6.4269887706752845e-06, + "loss": 0.729, + "step": 1573 + }, + { + "epoch": 0.63, + "learning_rate": 6.414911837660768e-06, + "loss": 0.7041, + "step": 1574 + }, + { + "epoch": 0.63, + "learning_rate": 6.402840901790508e-06, + "loss": 0.6924, + "step": 1575 + }, + { + "epoch": 0.63, + "learning_rate": 6.39077598325679e-06, + "loss": 0.6804, + "step": 1576 + }, + { + "epoch": 0.63, + "learning_rate": 6.378717102241842e-06, + "loss": 0.6853, + "step": 1577 + }, + { + "epoch": 0.63, + "learning_rate": 6.366664278917789e-06, + "loss": 0.7923, + "step": 1578 + }, + { + "epoch": 0.63, + "learning_rate": 6.3546175334466185e-06, + "loss": 0.7074, + "step": 1579 + }, + { + "epoch": 0.63, + "learning_rate": 6.342576885980165e-06, + "loss": 0.7078, + "step": 1580 + }, + { + "epoch": 0.63, + "learning_rate": 6.330542356660047e-06, + "loss": 0.6143, + "step": 1581 + }, + { + "epoch": 0.63, + "learning_rate": 6.318513965617657e-06, + "loss": 0.6558, + "step": 1582 + }, + { + "epoch": 0.63, + "learning_rate": 6.306491732974111e-06, + "loss": 0.6495, + "step": 1583 + }, + { + "epoch": 0.63, + "learning_rate": 6.2944756788402394e-06, + "loss": 0.6719, + "step": 1584 + }, + { + "epoch": 0.63, + "learning_rate": 6.282465823316519e-06, + "loss": 0.6856, + "step": 1585 + }, + { + "epoch": 0.63, + "learning_rate": 6.270462186493071e-06, + "loss": 0.6369, + "step": 1586 + }, + { + "epoch": 0.63, + "learning_rate": 6.2584647884496034e-06, + "loss": 0.6776, + "step": 1587 + }, + { + "epoch": 0.63, + "learning_rate": 6.246473649255391e-06, + "loss": 0.7309, + "step": 1588 + }, + { + "epoch": 0.63, + "learning_rate": 6.234488788969251e-06, + "loss": 0.667, + "step": 1589 + }, + { + "epoch": 0.63, + "learning_rate": 6.22251022763948e-06, + "loss": 0.7149, + "step": 1590 + }, + { + "epoch": 0.64, + "learning_rate": 6.2105379853038436e-06, + "loss": 0.6952, + "step": 1591 + }, + { + "epoch": 0.64, + "learning_rate": 6.19857208198954e-06, + "loss": 0.6295, + "step": 1592 + }, + { + "epoch": 0.64, + "learning_rate": 6.186612537713164e-06, + "loss": 0.7524, + "step": 1593 + }, + { + "epoch": 0.64, + "learning_rate": 6.1746593724806666e-06, + "loss": 0.6547, + "step": 1594 + }, + { + "epoch": 0.64, + "learning_rate": 6.162712606287335e-06, + "loss": 0.6421, + "step": 1595 + }, + { + "epoch": 0.64, + "learning_rate": 6.150772259117748e-06, + "loss": 0.6875, + "step": 1596 + }, + { + "epoch": 0.64, + "learning_rate": 6.138838350945744e-06, + "loss": 0.6819, + "step": 1597 + }, + { + "epoch": 0.64, + "learning_rate": 6.126910901734397e-06, + "loss": 0.6575, + "step": 1598 + }, + { + "epoch": 0.64, + "learning_rate": 6.114989931435976e-06, + "loss": 0.707, + "step": 1599 + }, + { + "epoch": 0.64, + "learning_rate": 6.103075459991905e-06, + "loss": 0.7111, + "step": 1600 + }, + { + "epoch": 0.64, + "learning_rate": 6.0911675073327385e-06, + "loss": 0.7314, + "step": 1601 + }, + { + "epoch": 0.64, + "learning_rate": 6.079266093378138e-06, + "loss": 0.6706, + "step": 1602 + }, + { + "epoch": 0.64, + "learning_rate": 6.067371238036812e-06, + "loss": 0.7071, + "step": 1603 + }, + { + "epoch": 0.64, + "learning_rate": 6.055482961206503e-06, + "loss": 0.7422, + "step": 1604 + }, + { + "epoch": 0.64, + "learning_rate": 6.043601282773952e-06, + "loss": 0.732, + "step": 1605 + }, + { + "epoch": 0.64, + "learning_rate": 6.031726222614855e-06, + "loss": 0.642, + "step": 1606 + }, + { + "epoch": 0.64, + "learning_rate": 6.019857800593847e-06, + "loss": 0.6497, + "step": 1607 + }, + { + "epoch": 0.64, + "learning_rate": 6.007996036564454e-06, + "loss": 0.6338, + "step": 1608 + }, + { + "epoch": 0.64, + "learning_rate": 5.9961409503690605e-06, + "loss": 0.6868, + "step": 1609 + }, + { + "epoch": 0.64, + "learning_rate": 5.984292561838881e-06, + "loss": 0.7424, + "step": 1610 + }, + { + "epoch": 0.64, + "learning_rate": 5.9724508907939345e-06, + "loss": 0.636, + "step": 1611 + }, + { + "epoch": 0.64, + "learning_rate": 5.960615957042999e-06, + "loss": 0.6338, + "step": 1612 + }, + { + "epoch": 0.64, + "learning_rate": 5.948787780383574e-06, + "loss": 0.6044, + "step": 1613 + }, + { + "epoch": 0.64, + "learning_rate": 5.936966380601867e-06, + "loss": 0.6836, + "step": 1614 + }, + { + "epoch": 0.64, + "learning_rate": 5.92515177747274e-06, + "loss": 0.6446, + "step": 1615 + }, + { + "epoch": 0.65, + "learning_rate": 5.913343990759695e-06, + "loss": 0.7554, + "step": 1616 + }, + { + "epoch": 0.65, + "learning_rate": 5.901543040214823e-06, + "loss": 0.7045, + "step": 1617 + }, + { + "epoch": 0.65, + "learning_rate": 5.889748945578788e-06, + "loss": 0.7306, + "step": 1618 + }, + { + "epoch": 0.65, + "learning_rate": 5.877961726580773e-06, + "loss": 0.6662, + "step": 1619 + }, + { + "epoch": 0.65, + "learning_rate": 5.866181402938469e-06, + "loss": 0.6709, + "step": 1620 + }, + { + "epoch": 0.65, + "learning_rate": 5.854407994358037e-06, + "loss": 0.6846, + "step": 1621 + }, + { + "epoch": 0.65, + "learning_rate": 5.842641520534055e-06, + "loss": 0.7122, + "step": 1622 + }, + { + "epoch": 0.65, + "learning_rate": 5.830882001149517e-06, + "loss": 0.7134, + "step": 1623 + }, + { + "epoch": 0.65, + "learning_rate": 5.819129455875764e-06, + "loss": 0.6846, + "step": 1624 + }, + { + "epoch": 0.65, + "learning_rate": 5.8073839043724935e-06, + "loss": 0.651, + "step": 1625 + }, + { + "epoch": 0.65, + "learning_rate": 5.7956453662876935e-06, + "loss": 0.6689, + "step": 1626 + }, + { + "epoch": 0.65, + "learning_rate": 5.783913861257611e-06, + "loss": 0.6747, + "step": 1627 + }, + { + "epoch": 0.65, + "learning_rate": 5.772189408906739e-06, + "loss": 0.6545, + "step": 1628 + }, + { + "epoch": 0.65, + "learning_rate": 5.760472028847773e-06, + "loss": 0.7081, + "step": 1629 + }, + { + "epoch": 0.65, + "learning_rate": 5.748761740681573e-06, + "loss": 0.6908, + "step": 1630 + }, + { + "epoch": 0.65, + "learning_rate": 5.737058563997138e-06, + "loss": 0.6248, + "step": 1631 + }, + { + "epoch": 0.65, + "learning_rate": 5.725362518371576e-06, + "loss": 0.6172, + "step": 1632 + }, + { + "epoch": 0.65, + "learning_rate": 5.713673623370046e-06, + "loss": 0.7342, + "step": 1633 + }, + { + "epoch": 0.65, + "learning_rate": 5.7019918985457745e-06, + "loss": 0.6993, + "step": 1634 + }, + { + "epoch": 0.65, + "learning_rate": 5.6903173634399785e-06, + "loss": 0.7109, + "step": 1635 + }, + { + "epoch": 0.65, + "learning_rate": 5.678650037581841e-06, + "loss": 0.6115, + "step": 1636 + }, + { + "epoch": 0.65, + "learning_rate": 5.666989940488496e-06, + "loss": 0.6896, + "step": 1637 + }, + { + "epoch": 0.65, + "learning_rate": 5.655337091664985e-06, + "loss": 0.6753, + "step": 1638 + }, + { + "epoch": 0.65, + "learning_rate": 5.643691510604221e-06, + "loss": 0.6297, + "step": 1639 + }, + { + "epoch": 0.65, + "learning_rate": 5.632053216786959e-06, + "loss": 0.651, + "step": 1640 + }, + { + "epoch": 0.66, + "learning_rate": 5.620422229681771e-06, + "loss": 0.6685, + "step": 1641 + }, + { + "epoch": 0.66, + "learning_rate": 5.608798568744989e-06, + "loss": 0.666, + "step": 1642 + }, + { + "epoch": 0.66, + "learning_rate": 5.597182253420712e-06, + "loss": 0.6243, + "step": 1643 + }, + { + "epoch": 0.66, + "learning_rate": 5.585573303140741e-06, + "loss": 0.6522, + "step": 1644 + }, + { + "epoch": 0.66, + "learning_rate": 5.573971737324551e-06, + "loss": 0.6861, + "step": 1645 + }, + { + "epoch": 0.66, + "learning_rate": 5.56237757537927e-06, + "loss": 0.6364, + "step": 1646 + }, + { + "epoch": 0.66, + "learning_rate": 5.550790836699642e-06, + "loss": 0.6029, + "step": 1647 + }, + { + "epoch": 0.66, + "learning_rate": 5.539211540667992e-06, + "loss": 0.6462, + "step": 1648 + }, + { + "epoch": 0.66, + "learning_rate": 5.527639706654195e-06, + "loss": 0.714, + "step": 1649 + }, + { + "epoch": 0.66, + "learning_rate": 5.516075354015646e-06, + "loss": 0.7084, + "step": 1650 + }, + { + "epoch": 0.66, + "learning_rate": 5.504518502097212e-06, + "loss": 0.7171, + "step": 1651 + }, + { + "epoch": 0.66, + "learning_rate": 5.4929691702312395e-06, + "loss": 0.6553, + "step": 1652 + }, + { + "epoch": 0.66, + "learning_rate": 5.481427377737465e-06, + "loss": 0.6455, + "step": 1653 + }, + { + "epoch": 0.66, + "learning_rate": 5.4698931439230314e-06, + "loss": 0.6901, + "step": 1654 + }, + { + "epoch": 0.66, + "learning_rate": 5.458366488082436e-06, + "loss": 0.6949, + "step": 1655 + }, + { + "epoch": 0.66, + "learning_rate": 5.446847429497494e-06, + "loss": 0.6834, + "step": 1656 + }, + { + "epoch": 0.66, + "learning_rate": 5.435335987437316e-06, + "loss": 0.7484, + "step": 1657 + }, + { + "epoch": 0.66, + "learning_rate": 5.423832181158274e-06, + "loss": 0.669, + "step": 1658 + }, + { + "epoch": 0.66, + "learning_rate": 5.412336029903956e-06, + "loss": 0.73, + "step": 1659 + }, + { + "epoch": 0.66, + "learning_rate": 5.400847552905152e-06, + "loss": 0.7043, + "step": 1660 + }, + { + "epoch": 0.66, + "learning_rate": 5.3893667693798245e-06, + "loss": 0.7214, + "step": 1661 + }, + { + "epoch": 0.66, + "learning_rate": 5.377893698533048e-06, + "loss": 0.7089, + "step": 1662 + }, + { + "epoch": 0.66, + "learning_rate": 5.366428359557003e-06, + "loss": 0.6707, + "step": 1663 + }, + { + "epoch": 0.66, + "learning_rate": 5.354970771630945e-06, + "loss": 0.668, + "step": 1664 + }, + { + "epoch": 0.66, + "learning_rate": 5.34352095392114e-06, + "loss": 0.6182, + "step": 1665 + }, + { + "epoch": 0.67, + "learning_rate": 5.332078925580883e-06, + "loss": 0.6481, + "step": 1666 + }, + { + "epoch": 0.67, + "learning_rate": 5.320644705750431e-06, + "loss": 0.7033, + "step": 1667 + }, + { + "epoch": 0.67, + "learning_rate": 5.309218313556961e-06, + "loss": 0.6702, + "step": 1668 + }, + { + "epoch": 0.67, + "learning_rate": 5.297799768114575e-06, + "loss": 0.6675, + "step": 1669 + }, + { + "epoch": 0.67, + "learning_rate": 5.286389088524253e-06, + "loss": 0.6136, + "step": 1670 + }, + { + "epoch": 0.67, + "learning_rate": 5.274986293873796e-06, + "loss": 0.6822, + "step": 1671 + }, + { + "epoch": 0.67, + "learning_rate": 5.263591403237831e-06, + "loss": 0.6133, + "step": 1672 + }, + { + "epoch": 0.67, + "learning_rate": 5.252204435677763e-06, + "loss": 0.6395, + "step": 1673 + }, + { + "epoch": 0.67, + "learning_rate": 5.2408254102417275e-06, + "loss": 0.6992, + "step": 1674 + }, + { + "epoch": 0.67, + "learning_rate": 5.2294543459646e-06, + "loss": 0.6346, + "step": 1675 + }, + { + "epoch": 0.67, + "learning_rate": 5.2180912618679195e-06, + "loss": 0.5956, + "step": 1676 + }, + { + "epoch": 0.67, + "learning_rate": 5.206736176959877e-06, + "loss": 0.6515, + "step": 1677 + }, + { + "epoch": 0.67, + "learning_rate": 5.19538911023529e-06, + "loss": 0.7051, + "step": 1678 + }, + { + "epoch": 0.67, + "learning_rate": 5.184050080675558e-06, + "loss": 0.7159, + "step": 1679 + }, + { + "epoch": 0.67, + "learning_rate": 5.1727191072486385e-06, + "loss": 0.6304, + "step": 1680 + }, + { + "epoch": 0.67, + "learning_rate": 5.161396208909012e-06, + "loss": 0.6531, + "step": 1681 + }, + { + "epoch": 0.67, + "learning_rate": 5.150081404597654e-06, + "loss": 0.6541, + "step": 1682 + }, + { + "epoch": 0.67, + "learning_rate": 5.138774713241983e-06, + "loss": 0.6782, + "step": 1683 + }, + { + "epoch": 0.67, + "learning_rate": 5.127476153755873e-06, + "loss": 0.6293, + "step": 1684 + }, + { + "epoch": 0.67, + "learning_rate": 5.116185745039585e-06, + "loss": 0.6962, + "step": 1685 + }, + { + "epoch": 0.67, + "learning_rate": 5.10490350597973e-06, + "loss": 0.6201, + "step": 1686 + }, + { + "epoch": 0.67, + "learning_rate": 5.093629455449273e-06, + "loss": 0.6579, + "step": 1687 + }, + { + "epoch": 0.67, + "learning_rate": 5.082363612307471e-06, + "loss": 0.7357, + "step": 1688 + }, + { + "epoch": 0.67, + "learning_rate": 5.071105995399852e-06, + "loss": 0.5901, + "step": 1689 + }, + { + "epoch": 0.67, + "learning_rate": 5.059856623558189e-06, + "loss": 0.6975, + "step": 1690 + }, + { + "epoch": 0.68, + "learning_rate": 5.0486155156004615e-06, + "loss": 0.6361, + "step": 1691 + }, + { + "epoch": 0.68, + "learning_rate": 5.0373826903308095e-06, + "loss": 0.6902, + "step": 1692 + }, + { + "epoch": 0.68, + "learning_rate": 5.026158166539548e-06, + "loss": 0.6761, + "step": 1693 + }, + { + "epoch": 0.68, + "learning_rate": 5.014941963003075e-06, + "loss": 0.6557, + "step": 1694 + }, + { + "epoch": 0.68, + "learning_rate": 5.003734098483887e-06, + "loss": 0.682, + "step": 1695 + }, + { + "epoch": 0.68, + "learning_rate": 4.992534591730527e-06, + "loss": 0.6713, + "step": 1696 + }, + { + "epoch": 0.68, + "learning_rate": 4.981343461477559e-06, + "loss": 0.7376, + "step": 1697 + }, + { + "epoch": 0.68, + "learning_rate": 4.970160726445531e-06, + "loss": 0.682, + "step": 1698 + }, + { + "epoch": 0.68, + "learning_rate": 4.9589864053409545e-06, + "loss": 0.6766, + "step": 1699 + }, + { + "epoch": 0.68, + "learning_rate": 4.9478205168562535e-06, + "loss": 0.7562, + "step": 1700 + }, + { + "epoch": 0.68, + "learning_rate": 4.936663079669753e-06, + "loss": 0.6514, + "step": 1701 + }, + { + "epoch": 0.68, + "learning_rate": 4.925514112445653e-06, + "loss": 0.7442, + "step": 1702 + }, + { + "epoch": 0.68, + "learning_rate": 4.914373633833962e-06, + "loss": 0.6804, + "step": 1703 + }, + { + "epoch": 0.68, + "learning_rate": 4.9032416624705036e-06, + "loss": 0.7052, + "step": 1704 + }, + { + "epoch": 0.68, + "learning_rate": 4.892118216976871e-06, + "loss": 0.6387, + "step": 1705 + }, + { + "epoch": 0.68, + "learning_rate": 4.881003315960378e-06, + "loss": 0.7031, + "step": 1706 + }, + { + "epoch": 0.68, + "learning_rate": 4.869896978014071e-06, + "loss": 0.6499, + "step": 1707 + }, + { + "epoch": 0.68, + "learning_rate": 4.858799221716659e-06, + "loss": 0.6845, + "step": 1708 + }, + { + "epoch": 0.68, + "learning_rate": 4.847710065632489e-06, + "loss": 0.6311, + "step": 1709 + }, + { + "epoch": 0.68, + "learning_rate": 4.836629528311527e-06, + "loss": 0.7011, + "step": 1710 + }, + { + "epoch": 0.68, + "learning_rate": 4.8255576282893355e-06, + "loss": 0.6767, + "step": 1711 + }, + { + "epoch": 0.68, + "learning_rate": 4.814494384087004e-06, + "loss": 0.6893, + "step": 1712 + }, + { + "epoch": 0.68, + "learning_rate": 4.803439814211158e-06, + "loss": 0.6652, + "step": 1713 + }, + { + "epoch": 0.68, + "learning_rate": 4.792393937153914e-06, + "loss": 0.6802, + "step": 1714 + }, + { + "epoch": 0.68, + "learning_rate": 4.78135677139283e-06, + "loss": 0.7172, + "step": 1715 + }, + { + "epoch": 0.69, + "learning_rate": 4.770328335390915e-06, + "loss": 0.7511, + "step": 1716 + }, + { + "epoch": 0.69, + "learning_rate": 4.759308647596564e-06, + "loss": 0.6871, + "step": 1717 + }, + { + "epoch": 0.69, + "learning_rate": 4.7482977264435305e-06, + "loss": 0.7501, + "step": 1718 + }, + { + "epoch": 0.69, + "learning_rate": 4.737295590350913e-06, + "loss": 0.6433, + "step": 1719 + }, + { + "epoch": 0.69, + "learning_rate": 4.726302257723113e-06, + "loss": 0.7348, + "step": 1720 + }, + { + "epoch": 0.69, + "learning_rate": 4.7153177469498045e-06, + "loss": 0.7438, + "step": 1721 + }, + { + "epoch": 0.69, + "learning_rate": 4.704342076405902e-06, + "loss": 0.7315, + "step": 1722 + }, + { + "epoch": 0.69, + "learning_rate": 4.6933752644515404e-06, + "loss": 0.7445, + "step": 1723 + }, + { + "epoch": 0.69, + "learning_rate": 4.682417329432016e-06, + "loss": 0.6336, + "step": 1724 + }, + { + "epoch": 0.69, + "learning_rate": 4.671468289677802e-06, + "loss": 0.6899, + "step": 1725 + }, + { + "epoch": 0.69, + "learning_rate": 4.660528163504479e-06, + "loss": 0.7473, + "step": 1726 + }, + { + "epoch": 0.69, + "learning_rate": 4.6495969692127084e-06, + "loss": 0.6591, + "step": 1727 + }, + { + "epoch": 0.69, + "learning_rate": 4.6386747250882226e-06, + "loss": 0.6735, + "step": 1728 + }, + { + "epoch": 0.69, + "learning_rate": 4.6277614494017784e-06, + "loss": 0.666, + "step": 1729 + }, + { + "epoch": 0.69, + "learning_rate": 4.616857160409131e-06, + "loss": 0.6789, + "step": 1730 + }, + { + "epoch": 0.69, + "learning_rate": 4.605961876351001e-06, + "loss": 0.6489, + "step": 1731 + }, + { + "epoch": 0.69, + "learning_rate": 4.595075615453049e-06, + "loss": 0.7276, + "step": 1732 + }, + { + "epoch": 0.69, + "learning_rate": 4.584198395925829e-06, + "loss": 0.7292, + "step": 1733 + }, + { + "epoch": 0.69, + "learning_rate": 4.5733302359647955e-06, + "loss": 0.7242, + "step": 1734 + }, + { + "epoch": 0.69, + "learning_rate": 4.562471153750221e-06, + "loss": 0.6776, + "step": 1735 + }, + { + "epoch": 0.69, + "learning_rate": 4.55162116744721e-06, + "loss": 0.7364, + "step": 1736 + }, + { + "epoch": 0.69, + "learning_rate": 4.5407802952056435e-06, + "loss": 0.7191, + "step": 1737 + }, + { + "epoch": 0.69, + "learning_rate": 4.529948555160162e-06, + "loss": 0.7148, + "step": 1738 + }, + { + "epoch": 0.69, + "learning_rate": 4.519125965430125e-06, + "loss": 0.6632, + "step": 1739 + }, + { + "epoch": 0.69, + "learning_rate": 4.508312544119592e-06, + "loss": 0.7307, + "step": 1740 + }, + { + "epoch": 0.7, + "learning_rate": 4.497508309317271e-06, + "loss": 0.705, + "step": 1741 + }, + { + "epoch": 0.7, + "learning_rate": 4.486713279096515e-06, + "loss": 0.7034, + "step": 1742 + }, + { + "epoch": 0.7, + "learning_rate": 4.475927471515287e-06, + "loss": 0.6672, + "step": 1743 + }, + { + "epoch": 0.7, + "learning_rate": 4.465150904616101e-06, + "loss": 0.614, + "step": 1744 + }, + { + "epoch": 0.7, + "learning_rate": 4.454383596426028e-06, + "loss": 0.6489, + "step": 1745 + }, + { + "epoch": 0.7, + "learning_rate": 4.443625564956648e-06, + "loss": 0.6973, + "step": 1746 + }, + { + "epoch": 0.7, + "learning_rate": 4.432876828204022e-06, + "loss": 0.7163, + "step": 1747 + }, + { + "epoch": 0.7, + "learning_rate": 4.4221374041486654e-06, + "loss": 0.6986, + "step": 1748 + }, + { + "epoch": 0.7, + "learning_rate": 4.411407310755513e-06, + "loss": 0.7546, + "step": 1749 + }, + { + "epoch": 0.7, + "learning_rate": 4.400686565973886e-06, + "loss": 0.6168, + "step": 1750 + }, + { + "epoch": 0.7, + "learning_rate": 4.389975187737472e-06, + "loss": 0.6861, + "step": 1751 + }, + { + "epoch": 0.7, + "learning_rate": 4.3792731939643015e-06, + "loss": 0.6757, + "step": 1752 + }, + { + "epoch": 0.7, + "learning_rate": 4.368580602556687e-06, + "loss": 0.6615, + "step": 1753 + }, + { + "epoch": 0.7, + "learning_rate": 4.3578974314012225e-06, + "loss": 0.6795, + "step": 1754 + }, + { + "epoch": 0.7, + "learning_rate": 4.347223698368748e-06, + "loss": 0.7273, + "step": 1755 + }, + { + "epoch": 0.7, + "learning_rate": 4.336559421314298e-06, + "loss": 0.6269, + "step": 1756 + }, + { + "epoch": 0.7, + "learning_rate": 4.325904618077114e-06, + "loss": 0.6788, + "step": 1757 + }, + { + "epoch": 0.7, + "learning_rate": 4.315259306480575e-06, + "loss": 0.7263, + "step": 1758 + }, + { + "epoch": 0.7, + "learning_rate": 4.304623504332178e-06, + "loss": 0.7434, + "step": 1759 + }, + { + "epoch": 0.7, + "learning_rate": 4.293997229423523e-06, + "loss": 0.6652, + "step": 1760 + }, + { + "epoch": 0.7, + "learning_rate": 4.283380499530269e-06, + "loss": 0.7333, + "step": 1761 + }, + { + "epoch": 0.7, + "learning_rate": 4.2727733324121055e-06, + "loss": 0.6379, + "step": 1762 + }, + { + "epoch": 0.7, + "learning_rate": 4.262175745812729e-06, + "loss": 0.6696, + "step": 1763 + }, + { + "epoch": 0.7, + "learning_rate": 4.251587757459813e-06, + "loss": 0.6714, + "step": 1764 + }, + { + "epoch": 0.7, + "learning_rate": 4.241009385064958e-06, + "loss": 0.7487, + "step": 1765 + }, + { + "epoch": 0.7, + "learning_rate": 4.2304406463237026e-06, + "loss": 0.6666, + "step": 1766 + }, + { + "epoch": 0.71, + "learning_rate": 4.219881558915458e-06, + "loss": 0.7077, + "step": 1767 + }, + { + "epoch": 0.71, + "learning_rate": 4.209332140503486e-06, + "loss": 0.6911, + "step": 1768 + }, + { + "epoch": 0.71, + "learning_rate": 4.198792408734882e-06, + "loss": 0.6486, + "step": 1769 + }, + { + "epoch": 0.71, + "learning_rate": 4.1882623812405345e-06, + "loss": 0.6362, + "step": 1770 + }, + { + "epoch": 0.71, + "learning_rate": 4.1777420756351e-06, + "loss": 0.6537, + "step": 1771 + }, + { + "epoch": 0.71, + "learning_rate": 4.1672315095169725e-06, + "loss": 0.6595, + "step": 1772 + }, + { + "epoch": 0.71, + "learning_rate": 4.1567307004682546e-06, + "loss": 0.6062, + "step": 1773 + }, + { + "epoch": 0.71, + "learning_rate": 4.146239666054716e-06, + "loss": 0.6798, + "step": 1774 + }, + { + "epoch": 0.71, + "learning_rate": 4.135758423825798e-06, + "loss": 0.7217, + "step": 1775 + }, + { + "epoch": 0.71, + "learning_rate": 4.125286991314538e-06, + "loss": 0.6591, + "step": 1776 + }, + { + "epoch": 0.71, + "learning_rate": 4.1148253860375764e-06, + "loss": 0.7172, + "step": 1777 + }, + { + "epoch": 0.71, + "learning_rate": 4.104373625495114e-06, + "loss": 0.6432, + "step": 1778 + }, + { + "epoch": 0.71, + "learning_rate": 4.09393172717088e-06, + "loss": 0.6345, + "step": 1779 + }, + { + "epoch": 0.71, + "learning_rate": 4.083499708532108e-06, + "loss": 0.6872, + "step": 1780 + }, + { + "epoch": 0.71, + "learning_rate": 4.073077587029509e-06, + "loss": 0.6374, + "step": 1781 + }, + { + "epoch": 0.71, + "learning_rate": 4.0626653800972236e-06, + "loss": 0.7046, + "step": 1782 + }, + { + "epoch": 0.71, + "learning_rate": 4.0522631051528205e-06, + "loss": 0.6766, + "step": 1783 + }, + { + "epoch": 0.71, + "learning_rate": 4.0418707795972575e-06, + "loss": 0.6547, + "step": 1784 + }, + { + "epoch": 0.71, + "learning_rate": 4.031488420814834e-06, + "loss": 0.7285, + "step": 1785 + }, + { + "epoch": 0.71, + "learning_rate": 4.021116046173189e-06, + "loss": 0.6674, + "step": 1786 + }, + { + "epoch": 0.71, + "learning_rate": 4.010753673023254e-06, + "loss": 0.5786, + "step": 1787 + }, + { + "epoch": 0.71, + "learning_rate": 4.000401318699236e-06, + "loss": 0.7023, + "step": 1788 + }, + { + "epoch": 0.71, + "learning_rate": 3.9900590005185735e-06, + "loss": 0.67, + "step": 1789 + }, + { + "epoch": 0.71, + "learning_rate": 3.9797267357819294e-06, + "loss": 0.716, + "step": 1790 + }, + { + "epoch": 0.71, + "learning_rate": 3.969404541773132e-06, + "loss": 0.6724, + "step": 1791 + }, + { + "epoch": 0.72, + "learning_rate": 3.959092435759174e-06, + "loss": 0.695, + "step": 1792 + }, + { + "epoch": 0.72, + "learning_rate": 3.9487904349901815e-06, + "loss": 0.6837, + "step": 1793 + }, + { + "epoch": 0.72, + "learning_rate": 3.938498556699357e-06, + "loss": 0.6544, + "step": 1794 + }, + { + "epoch": 0.72, + "learning_rate": 3.928216818102982e-06, + "loss": 0.7146, + "step": 1795 + }, + { + "epoch": 0.72, + "learning_rate": 3.917945236400374e-06, + "loss": 0.6849, + "step": 1796 + }, + { + "epoch": 0.72, + "learning_rate": 3.907683828773861e-06, + "loss": 0.6909, + "step": 1797 + }, + { + "epoch": 0.72, + "learning_rate": 3.897432612388752e-06, + "loss": 0.6958, + "step": 1798 + }, + { + "epoch": 0.72, + "learning_rate": 3.8871916043933065e-06, + "loss": 0.6176, + "step": 1799 + }, + { + "epoch": 0.72, + "learning_rate": 3.876960821918704e-06, + "loss": 0.653, + "step": 1800 + }, + { + "epoch": 0.72, + "learning_rate": 3.866740282079023e-06, + "loss": 0.6818, + "step": 1801 + }, + { + "epoch": 0.72, + "learning_rate": 3.8565300019712115e-06, + "loss": 0.6919, + "step": 1802 + }, + { + "epoch": 0.72, + "learning_rate": 3.846329998675048e-06, + "loss": 0.7064, + "step": 1803 + }, + { + "epoch": 0.72, + "learning_rate": 3.836140289253124e-06, + "loss": 0.592, + "step": 1804 + }, + { + "epoch": 0.72, + "learning_rate": 3.8259608907508105e-06, + "loss": 0.6205, + "step": 1805 + }, + { + "epoch": 0.72, + "learning_rate": 3.81579182019623e-06, + "loss": 0.7462, + "step": 1806 + }, + { + "epoch": 0.72, + "learning_rate": 3.8056330946002283e-06, + "loss": 0.7299, + "step": 1807 + }, + { + "epoch": 0.72, + "learning_rate": 3.7954847309563504e-06, + "loss": 0.7064, + "step": 1808 + }, + { + "epoch": 0.72, + "learning_rate": 3.7853467462407966e-06, + "loss": 0.6426, + "step": 1809 + }, + { + "epoch": 0.72, + "learning_rate": 3.7752191574124176e-06, + "loss": 0.6786, + "step": 1810 + }, + { + "epoch": 0.72, + "learning_rate": 3.7651019814126656e-06, + "loss": 0.6807, + "step": 1811 + }, + { + "epoch": 0.72, + "learning_rate": 3.7549952351655795e-06, + "loss": 0.6862, + "step": 1812 + }, + { + "epoch": 0.72, + "learning_rate": 3.7448989355777497e-06, + "loss": 0.6763, + "step": 1813 + }, + { + "epoch": 0.72, + "learning_rate": 3.734813099538289e-06, + "loss": 0.6925, + "step": 1814 + }, + { + "epoch": 0.72, + "learning_rate": 3.724737743918809e-06, + "loss": 0.6703, + "step": 1815 + }, + { + "epoch": 0.72, + "learning_rate": 3.7146728855733947e-06, + "loss": 0.7065, + "step": 1816 + }, + { + "epoch": 0.73, + "learning_rate": 3.7046185413385573e-06, + "loss": 0.6716, + "step": 1817 + }, + { + "epoch": 0.73, + "learning_rate": 3.694574728033231e-06, + "loss": 0.7259, + "step": 1818 + }, + { + "epoch": 0.73, + "learning_rate": 3.6845414624587326e-06, + "loss": 0.6714, + "step": 1819 + }, + { + "epoch": 0.73, + "learning_rate": 3.674518761398733e-06, + "loss": 0.5993, + "step": 1820 + }, + { + "epoch": 0.73, + "learning_rate": 3.664506641619232e-06, + "loss": 0.6699, + "step": 1821 + }, + { + "epoch": 0.73, + "learning_rate": 3.65450511986853e-06, + "loss": 0.6568, + "step": 1822 + }, + { + "epoch": 0.73, + "learning_rate": 3.644514212877187e-06, + "loss": 0.6742, + "step": 1823 + }, + { + "epoch": 0.73, + "learning_rate": 3.6345339373580267e-06, + "loss": 0.6741, + "step": 1824 + }, + { + "epoch": 0.73, + "learning_rate": 3.62456431000608e-06, + "loss": 0.7003, + "step": 1825 + }, + { + "epoch": 0.73, + "learning_rate": 3.6146053474985565e-06, + "loss": 0.718, + "step": 1826 + }, + { + "epoch": 0.73, + "learning_rate": 3.6046570664948357e-06, + "loss": 0.7227, + "step": 1827 + }, + { + "epoch": 0.73, + "learning_rate": 3.5947194836364264e-06, + "loss": 0.7127, + "step": 1828 + }, + { + "epoch": 0.73, + "learning_rate": 3.5847926155469427e-06, + "loss": 0.713, + "step": 1829 + }, + { + "epoch": 0.73, + "learning_rate": 3.574876478832072e-06, + "loss": 0.585, + "step": 1830 + }, + { + "epoch": 0.73, + "learning_rate": 3.5649710900795564e-06, + "loss": 0.6864, + "step": 1831 + }, + { + "epoch": 0.73, + "learning_rate": 3.5550764658591486e-06, + "loss": 0.7154, + "step": 1832 + }, + { + "epoch": 0.73, + "learning_rate": 3.5451926227225997e-06, + "loss": 0.633, + "step": 1833 + }, + { + "epoch": 0.73, + "learning_rate": 3.5353195772036352e-06, + "loss": 0.6982, + "step": 1834 + }, + { + "epoch": 0.73, + "learning_rate": 3.525457345817902e-06, + "loss": 0.6182, + "step": 1835 + }, + { + "epoch": 0.73, + "learning_rate": 3.5156059450629677e-06, + "loss": 0.697, + "step": 1836 + }, + { + "epoch": 0.73, + "learning_rate": 3.5057653914182787e-06, + "loss": 0.6626, + "step": 1837 + }, + { + "epoch": 0.73, + "learning_rate": 3.4959357013451368e-06, + "loss": 0.6936, + "step": 1838 + }, + { + "epoch": 0.73, + "learning_rate": 3.486116891286674e-06, + "loss": 0.7063, + "step": 1839 + }, + { + "epoch": 0.73, + "learning_rate": 3.4763089776678206e-06, + "loss": 0.6365, + "step": 1840 + }, + { + "epoch": 0.73, + "learning_rate": 3.4665119768952736e-06, + "loss": 0.694, + "step": 1841 + }, + { + "epoch": 0.74, + "learning_rate": 3.456725905357482e-06, + "loss": 0.6458, + "step": 1842 + }, + { + "epoch": 0.74, + "learning_rate": 3.4469507794246105e-06, + "loss": 0.6356, + "step": 1843 + }, + { + "epoch": 0.74, + "learning_rate": 3.4371866154485147e-06, + "loss": 0.6742, + "step": 1844 + }, + { + "epoch": 0.74, + "learning_rate": 3.42743342976271e-06, + "loss": 0.7007, + "step": 1845 + }, + { + "epoch": 0.74, + "learning_rate": 3.417691238682351e-06, + "loss": 0.6199, + "step": 1846 + }, + { + "epoch": 0.74, + "learning_rate": 3.4079600585041993e-06, + "loss": 0.6948, + "step": 1847 + }, + { + "epoch": 0.74, + "learning_rate": 3.3982399055065953e-06, + "loss": 0.6847, + "step": 1848 + }, + { + "epoch": 0.74, + "learning_rate": 3.3885307959494374e-06, + "loss": 0.6517, + "step": 1849 + }, + { + "epoch": 0.74, + "learning_rate": 3.378832746074143e-06, + "loss": 0.6781, + "step": 1850 + }, + { + "epoch": 0.74, + "learning_rate": 3.369145772103637e-06, + "loss": 0.7105, + "step": 1851 + }, + { + "epoch": 0.74, + "learning_rate": 3.359469890242313e-06, + "loss": 0.6411, + "step": 1852 + }, + { + "epoch": 0.74, + "learning_rate": 3.349805116676008e-06, + "loss": 0.6623, + "step": 1853 + }, + { + "epoch": 0.74, + "learning_rate": 3.3401514675719815e-06, + "loss": 0.6063, + "step": 1854 + }, + { + "epoch": 0.74, + "learning_rate": 3.3305089590788796e-06, + "loss": 0.7267, + "step": 1855 + }, + { + "epoch": 0.74, + "learning_rate": 3.3208776073267146e-06, + "loss": 0.6729, + "step": 1856 + }, + { + "epoch": 0.74, + "learning_rate": 3.3112574284268395e-06, + "loss": 0.6448, + "step": 1857 + }, + { + "epoch": 0.74, + "learning_rate": 3.301648438471905e-06, + "loss": 0.6505, + "step": 1858 + }, + { + "epoch": 0.74, + "learning_rate": 3.2920506535358566e-06, + "loss": 0.6564, + "step": 1859 + }, + { + "epoch": 0.74, + "learning_rate": 3.2824640896738935e-06, + "loss": 0.6825, + "step": 1860 + }, + { + "epoch": 0.74, + "learning_rate": 3.2728887629224415e-06, + "loss": 0.7094, + "step": 1861 + }, + { + "epoch": 0.74, + "learning_rate": 3.263324689299132e-06, + "loss": 0.6723, + "step": 1862 + }, + { + "epoch": 0.74, + "learning_rate": 3.2537718848027744e-06, + "loss": 0.6382, + "step": 1863 + }, + { + "epoch": 0.74, + "learning_rate": 3.2442303654133124e-06, + "loss": 0.6168, + "step": 1864 + }, + { + "epoch": 0.74, + "learning_rate": 3.234700147091834e-06, + "loss": 0.6095, + "step": 1865 + }, + { + "epoch": 0.74, + "learning_rate": 3.22518124578051e-06, + "loss": 0.7224, + "step": 1866 + }, + { + "epoch": 0.75, + "learning_rate": 3.2156736774025766e-06, + "loss": 0.713, + "step": 1867 + }, + { + "epoch": 0.75, + "learning_rate": 3.206177457862321e-06, + "loss": 0.7005, + "step": 1868 + }, + { + "epoch": 0.75, + "learning_rate": 3.1966926030450408e-06, + "loss": 0.7298, + "step": 1869 + }, + { + "epoch": 0.75, + "learning_rate": 3.187219128817025e-06, + "loss": 0.6716, + "step": 1870 + }, + { + "epoch": 0.75, + "learning_rate": 3.1777570510255262e-06, + "loss": 0.7283, + "step": 1871 + }, + { + "epoch": 0.75, + "learning_rate": 3.168306385498733e-06, + "loss": 0.6554, + "step": 1872 + }, + { + "epoch": 0.75, + "learning_rate": 3.1588671480457324e-06, + "loss": 0.74, + "step": 1873 + }, + { + "epoch": 0.75, + "learning_rate": 3.149439354456514e-06, + "loss": 0.6014, + "step": 1874 + }, + { + "epoch": 0.75, + "learning_rate": 3.1400230205019124e-06, + "loss": 0.6786, + "step": 1875 + }, + { + "epoch": 0.75, + "learning_rate": 3.1306181619335894e-06, + "loss": 0.7073, + "step": 1876 + }, + { + "epoch": 0.75, + "learning_rate": 3.121224794484019e-06, + "loss": 0.6562, + "step": 1877 + }, + { + "epoch": 0.75, + "learning_rate": 3.1118429338664467e-06, + "loss": 0.6157, + "step": 1878 + }, + { + "epoch": 0.75, + "learning_rate": 3.102472595774875e-06, + "loss": 0.5941, + "step": 1879 + }, + { + "epoch": 0.75, + "learning_rate": 3.093113795884026e-06, + "loss": 0.6177, + "step": 1880 + }, + { + "epoch": 0.75, + "learning_rate": 3.0837665498493273e-06, + "loss": 0.6888, + "step": 1881 + }, + { + "epoch": 0.75, + "learning_rate": 3.074430873306865e-06, + "loss": 0.6168, + "step": 1882 + }, + { + "epoch": 0.75, + "learning_rate": 3.0651067818733948e-06, + "loss": 0.7151, + "step": 1883 + }, + { + "epoch": 0.75, + "learning_rate": 3.05579429114627e-06, + "loss": 0.6962, + "step": 1884 + }, + { + "epoch": 0.75, + "learning_rate": 3.046493416703452e-06, + "loss": 0.6787, + "step": 1885 + }, + { + "epoch": 0.75, + "learning_rate": 3.0372041741034642e-06, + "loss": 0.603, + "step": 1886 + }, + { + "epoch": 0.75, + "learning_rate": 3.0279265788853763e-06, + "loss": 0.6678, + "step": 1887 + }, + { + "epoch": 0.75, + "learning_rate": 3.0186606465687705e-06, + "loss": 0.6468, + "step": 1888 + }, + { + "epoch": 0.75, + "learning_rate": 3.0094063926537233e-06, + "loss": 0.698, + "step": 1889 + }, + { + "epoch": 0.75, + "learning_rate": 3.000163832620775e-06, + "loss": 0.6156, + "step": 1890 + }, + { + "epoch": 0.75, + "learning_rate": 2.990932981930893e-06, + "loss": 0.588, + "step": 1891 + }, + { + "epoch": 0.76, + "learning_rate": 2.98171385602548e-06, + "loss": 0.621, + "step": 1892 + }, + { + "epoch": 0.76, + "learning_rate": 2.972506470326303e-06, + "loss": 0.5919, + "step": 1893 + }, + { + "epoch": 0.76, + "learning_rate": 2.9633108402355027e-06, + "loss": 0.6834, + "step": 1894 + }, + { + "epoch": 0.76, + "learning_rate": 2.95412698113555e-06, + "loss": 0.7228, + "step": 1895 + }, + { + "epoch": 0.76, + "learning_rate": 2.944954908389229e-06, + "loss": 0.7329, + "step": 1896 + }, + { + "epoch": 0.76, + "learning_rate": 2.935794637339605e-06, + "loss": 0.6899, + "step": 1897 + }, + { + "epoch": 0.76, + "learning_rate": 2.9266461833100047e-06, + "loss": 0.7258, + "step": 1898 + }, + { + "epoch": 0.76, + "learning_rate": 2.917509561603977e-06, + "loss": 0.7335, + "step": 1899 + }, + { + "epoch": 0.76, + "learning_rate": 2.9083847875052894e-06, + "loss": 0.6796, + "step": 1900 + }, + { + "epoch": 0.76, + "learning_rate": 2.899271876277887e-06, + "loss": 0.6627, + "step": 1901 + }, + { + "epoch": 0.76, + "learning_rate": 2.890170843165868e-06, + "loss": 0.6368, + "step": 1902 + }, + { + "epoch": 0.76, + "learning_rate": 2.881081703393466e-06, + "loss": 0.6972, + "step": 1903 + }, + { + "epoch": 0.76, + "learning_rate": 2.8720044721650155e-06, + "loss": 0.6917, + "step": 1904 + }, + { + "epoch": 0.76, + "learning_rate": 2.862939164664924e-06, + "loss": 0.7007, + "step": 1905 + }, + { + "epoch": 0.76, + "learning_rate": 2.8538857960576694e-06, + "loss": 0.6081, + "step": 1906 + }, + { + "epoch": 0.76, + "learning_rate": 2.844844381487748e-06, + "loss": 0.6639, + "step": 1907 + }, + { + "epoch": 0.76, + "learning_rate": 2.835814936079655e-06, + "loss": 0.7297, + "step": 1908 + }, + { + "epoch": 0.76, + "learning_rate": 2.8267974749378714e-06, + "loss": 0.587, + "step": 1909 + }, + { + "epoch": 0.76, + "learning_rate": 2.8177920131468273e-06, + "loss": 0.6426, + "step": 1910 + }, + { + "epoch": 0.76, + "learning_rate": 2.808798565770883e-06, + "loss": 0.6864, + "step": 1911 + }, + { + "epoch": 0.76, + "learning_rate": 2.799817147854299e-06, + "loss": 0.6438, + "step": 1912 + }, + { + "epoch": 0.76, + "learning_rate": 2.790847774421215e-06, + "loss": 0.6162, + "step": 1913 + }, + { + "epoch": 0.76, + "learning_rate": 2.7818904604756125e-06, + "loss": 0.6957, + "step": 1914 + }, + { + "epoch": 0.76, + "learning_rate": 2.7729452210013173e-06, + "loss": 0.6697, + "step": 1915 + }, + { + "epoch": 0.76, + "learning_rate": 2.764012070961948e-06, + "loss": 0.6727, + "step": 1916 + }, + { + "epoch": 0.77, + "learning_rate": 2.7550910253008933e-06, + "loss": 0.6875, + "step": 1917 + }, + { + "epoch": 0.77, + "learning_rate": 2.7461820989413036e-06, + "loss": 0.6681, + "step": 1918 + }, + { + "epoch": 0.77, + "learning_rate": 2.737285306786053e-06, + "loss": 0.7507, + "step": 1919 + }, + { + "epoch": 0.77, + "learning_rate": 2.7284006637177164e-06, + "loss": 0.6969, + "step": 1920 + }, + { + "epoch": 0.77, + "learning_rate": 2.7195281845985467e-06, + "loss": 0.6512, + "step": 1921 + }, + { + "epoch": 0.77, + "learning_rate": 2.7106678842704514e-06, + "loss": 0.6819, + "step": 1922 + }, + { + "epoch": 0.77, + "learning_rate": 2.701819777554956e-06, + "loss": 0.604, + "step": 1923 + }, + { + "epoch": 0.77, + "learning_rate": 2.6929838792532035e-06, + "loss": 0.6695, + "step": 1924 + }, + { + "epoch": 0.77, + "learning_rate": 2.684160204145899e-06, + "loss": 0.6242, + "step": 1925 + }, + { + "epoch": 0.77, + "learning_rate": 2.675348766993311e-06, + "loss": 0.6749, + "step": 1926 + }, + { + "epoch": 0.77, + "learning_rate": 2.666549582535233e-06, + "loss": 0.6846, + "step": 1927 + }, + { + "epoch": 0.77, + "learning_rate": 2.6577626654909617e-06, + "loss": 0.6683, + "step": 1928 + }, + { + "epoch": 0.77, + "learning_rate": 2.648988030559274e-06, + "loss": 0.6779, + "step": 1929 + }, + { + "epoch": 0.77, + "learning_rate": 2.640225692418401e-06, + "loss": 0.7126, + "step": 1930 + }, + { + "epoch": 0.77, + "learning_rate": 2.6314756657260053e-06, + "loss": 0.6897, + "step": 1931 + }, + { + "epoch": 0.77, + "learning_rate": 2.622737965119145e-06, + "loss": 0.6699, + "step": 1932 + }, + { + "epoch": 0.77, + "learning_rate": 2.6140126052142788e-06, + "loss": 0.7532, + "step": 1933 + }, + { + "epoch": 0.77, + "learning_rate": 2.6052996006072008e-06, + "loss": 0.5257, + "step": 1934 + }, + { + "epoch": 0.77, + "learning_rate": 2.5965989658730485e-06, + "loss": 0.6552, + "step": 1935 + }, + { + "epoch": 0.77, + "learning_rate": 2.5879107155662676e-06, + "loss": 0.6543, + "step": 1936 + }, + { + "epoch": 0.77, + "learning_rate": 2.5792348642205823e-06, + "loss": 0.6137, + "step": 1937 + }, + { + "epoch": 0.77, + "learning_rate": 2.570571426348978e-06, + "loss": 0.6618, + "step": 1938 + }, + { + "epoch": 0.77, + "learning_rate": 2.5619204164436796e-06, + "loss": 0.6431, + "step": 1939 + }, + { + "epoch": 0.77, + "learning_rate": 2.553281848976109e-06, + "loss": 0.7341, + "step": 1940 + }, + { + "epoch": 0.77, + "learning_rate": 2.544655738396885e-06, + "loss": 0.6282, + "step": 1941 + }, + { + "epoch": 0.78, + "learning_rate": 2.536042099135795e-06, + "loss": 0.6754, + "step": 1942 + }, + { + "epoch": 0.78, + "learning_rate": 2.527440945601747e-06, + "loss": 0.683, + "step": 1943 + }, + { + "epoch": 0.78, + "learning_rate": 2.518852292182774e-06, + "loss": 0.5835, + "step": 1944 + }, + { + "epoch": 0.78, + "learning_rate": 2.5102761532460005e-06, + "loss": 0.6472, + "step": 1945 + }, + { + "epoch": 0.78, + "learning_rate": 2.501712543137602e-06, + "loss": 0.699, + "step": 1946 + }, + { + "epoch": 0.78, + "learning_rate": 2.4931614761828182e-06, + "loss": 0.7149, + "step": 1947 + }, + { + "epoch": 0.78, + "learning_rate": 2.484622966685892e-06, + "loss": 0.6452, + "step": 1948 + }, + { + "epoch": 0.78, + "learning_rate": 2.476097028930058e-06, + "loss": 0.6099, + "step": 1949 + }, + { + "epoch": 0.78, + "learning_rate": 2.4675836771775265e-06, + "loss": 0.6269, + "step": 1950 + }, + { + "epoch": 0.78, + "learning_rate": 2.4590829256694615e-06, + "loss": 0.6662, + "step": 1951 + }, + { + "epoch": 0.78, + "learning_rate": 2.450594788625932e-06, + "loss": 0.6949, + "step": 1952 + }, + { + "epoch": 0.78, + "learning_rate": 2.442119280245917e-06, + "loss": 0.6451, + "step": 1953 + }, + { + "epoch": 0.78, + "learning_rate": 2.433656414707272e-06, + "loss": 0.6177, + "step": 1954 + }, + { + "epoch": 0.78, + "learning_rate": 2.425206206166688e-06, + "loss": 0.7183, + "step": 1955 + }, + { + "epoch": 0.78, + "learning_rate": 2.416768668759705e-06, + "loss": 0.6342, + "step": 1956 + }, + { + "epoch": 0.78, + "learning_rate": 2.408343816600657e-06, + "loss": 0.6453, + "step": 1957 + }, + { + "epoch": 0.78, + "learning_rate": 2.3999316637826486e-06, + "loss": 0.7295, + "step": 1958 + }, + { + "epoch": 0.78, + "learning_rate": 2.3915322243775564e-06, + "loss": 0.6968, + "step": 1959 + }, + { + "epoch": 0.78, + "learning_rate": 2.3831455124359793e-06, + "loss": 0.6339, + "step": 1960 + }, + { + "epoch": 0.78, + "learning_rate": 2.374771541987232e-06, + "loss": 0.7278, + "step": 1961 + }, + { + "epoch": 0.78, + "learning_rate": 2.3664103270393123e-06, + "loss": 0.6902, + "step": 1962 + }, + { + "epoch": 0.78, + "learning_rate": 2.358061881578885e-06, + "loss": 0.6355, + "step": 1963 + }, + { + "epoch": 0.78, + "learning_rate": 2.349726219571239e-06, + "loss": 0.621, + "step": 1964 + }, + { + "epoch": 0.78, + "learning_rate": 2.3414033549603022e-06, + "loss": 0.6438, + "step": 1965 + }, + { + "epoch": 0.78, + "learning_rate": 2.3330933016685753e-06, + "loss": 0.6355, + "step": 1966 + }, + { + "epoch": 0.79, + "learning_rate": 2.3247960735971385e-06, + "loss": 0.618, + "step": 1967 + }, + { + "epoch": 0.79, + "learning_rate": 2.3165116846256154e-06, + "loss": 0.6635, + "step": 1968 + }, + { + "epoch": 0.79, + "learning_rate": 2.308240148612152e-06, + "loss": 0.6272, + "step": 1969 + }, + { + "epoch": 0.79, + "learning_rate": 2.299981479393393e-06, + "loss": 0.7038, + "step": 1970 + }, + { + "epoch": 0.79, + "learning_rate": 2.291735690784461e-06, + "loss": 0.6614, + "step": 1971 + }, + { + "epoch": 0.79, + "learning_rate": 2.2835027965789345e-06, + "loss": 0.6224, + "step": 1972 + }, + { + "epoch": 0.79, + "learning_rate": 2.275282810548811e-06, + "loss": 0.6706, + "step": 1973 + }, + { + "epoch": 0.79, + "learning_rate": 2.2670757464445146e-06, + "loss": 0.5771, + "step": 1974 + }, + { + "epoch": 0.79, + "learning_rate": 2.2588816179948324e-06, + "loss": 0.6321, + "step": 1975 + }, + { + "epoch": 0.79, + "learning_rate": 2.2507004389069267e-06, + "loss": 0.6861, + "step": 1976 + }, + { + "epoch": 0.79, + "learning_rate": 2.2425322228662914e-06, + "loss": 0.6133, + "step": 1977 + }, + { + "epoch": 0.79, + "learning_rate": 2.2343769835367425e-06, + "loss": 0.6687, + "step": 1978 + }, + { + "epoch": 0.79, + "learning_rate": 2.226234734560381e-06, + "loss": 0.7167, + "step": 1979 + }, + { + "epoch": 0.79, + "learning_rate": 2.2181054895575847e-06, + "loss": 0.7118, + "step": 1980 + }, + { + "epoch": 0.79, + "learning_rate": 2.2099892621269705e-06, + "loss": 0.68, + "step": 1981 + }, + { + "epoch": 0.79, + "learning_rate": 2.20188606584538e-06, + "loss": 0.5915, + "step": 1982 + }, + { + "epoch": 0.79, + "learning_rate": 2.193795914267871e-06, + "loss": 0.6398, + "step": 1983 + }, + { + "epoch": 0.79, + "learning_rate": 2.1857188209276605e-06, + "loss": 0.7034, + "step": 1984 + }, + { + "epoch": 0.79, + "learning_rate": 2.1776547993361306e-06, + "loss": 0.7103, + "step": 1985 + }, + { + "epoch": 0.79, + "learning_rate": 2.169603862982801e-06, + "loss": 0.6161, + "step": 1986 + }, + { + "epoch": 0.79, + "learning_rate": 2.1615660253352888e-06, + "loss": 0.6937, + "step": 1987 + }, + { + "epoch": 0.79, + "learning_rate": 2.153541299839318e-06, + "loss": 0.7493, + "step": 1988 + }, + { + "epoch": 0.79, + "learning_rate": 2.145529699918666e-06, + "loss": 0.6235, + "step": 1989 + }, + { + "epoch": 0.79, + "learning_rate": 2.1375312389751546e-06, + "loss": 0.6494, + "step": 1990 + }, + { + "epoch": 0.79, + "learning_rate": 2.129545930388627e-06, + "loss": 0.6255, + "step": 1991 + }, + { + "epoch": 0.8, + "learning_rate": 2.121573787516936e-06, + "loss": 0.6712, + "step": 1992 + }, + { + "epoch": 0.8, + "learning_rate": 2.113614823695892e-06, + "loss": 0.5916, + "step": 1993 + }, + { + "epoch": 0.8, + "learning_rate": 2.105669052239274e-06, + "loss": 0.6827, + "step": 1994 + }, + { + "epoch": 0.8, + "learning_rate": 2.0977364864387896e-06, + "loss": 0.7326, + "step": 1995 + }, + { + "epoch": 0.8, + "learning_rate": 2.0898171395640463e-06, + "loss": 0.7246, + "step": 1996 + }, + { + "epoch": 0.8, + "learning_rate": 2.081911024862554e-06, + "loss": 0.6045, + "step": 1997 + }, + { + "epoch": 0.8, + "learning_rate": 2.074018155559683e-06, + "loss": 0.6076, + "step": 1998 + }, + { + "epoch": 0.8, + "learning_rate": 2.066138544858637e-06, + "loss": 0.6543, + "step": 1999 + }, + { + "epoch": 0.8, + "learning_rate": 2.0582722059404515e-06, + "loss": 0.6783, + "step": 2000 + }, + { + "epoch": 0.8, + "learning_rate": 2.050419151963957e-06, + "loss": 0.7195, + "step": 2001 + }, + { + "epoch": 0.8, + "learning_rate": 2.0425793960657623e-06, + "loss": 0.6747, + "step": 2002 + }, + { + "epoch": 0.8, + "learning_rate": 2.0347529513602305e-06, + "loss": 0.6142, + "step": 2003 + }, + { + "epoch": 0.8, + "learning_rate": 2.0269398309394585e-06, + "loss": 0.6626, + "step": 2004 + }, + { + "epoch": 0.8, + "learning_rate": 2.019140047873246e-06, + "loss": 0.664, + "step": 2005 + }, + { + "epoch": 0.8, + "learning_rate": 2.0113536152091006e-06, + "loss": 0.6426, + "step": 2006 + }, + { + "epoch": 0.8, + "learning_rate": 2.0035805459721768e-06, + "loss": 0.6046, + "step": 2007 + }, + { + "epoch": 0.8, + "learning_rate": 1.9958208531652876e-06, + "loss": 0.6491, + "step": 2008 + }, + { + "epoch": 0.8, + "learning_rate": 1.9880745497688658e-06, + "loss": 0.7318, + "step": 2009 + }, + { + "epoch": 0.8, + "learning_rate": 1.9803416487409465e-06, + "loss": 0.7052, + "step": 2010 + }, + { + "epoch": 0.8, + "learning_rate": 1.972622163017145e-06, + "loss": 0.6415, + "step": 2011 + }, + { + "epoch": 0.8, + "learning_rate": 1.964916105510637e-06, + "loss": 0.622, + "step": 2012 + }, + { + "epoch": 0.8, + "learning_rate": 1.9572234891121354e-06, + "loss": 0.6553, + "step": 2013 + }, + { + "epoch": 0.8, + "learning_rate": 1.9495443266898615e-06, + "loss": 0.7111, + "step": 2014 + }, + { + "epoch": 0.8, + "learning_rate": 1.9418786310895467e-06, + "loss": 0.6808, + "step": 2015 + }, + { + "epoch": 0.8, + "learning_rate": 1.934226415134377e-06, + "loss": 0.6634, + "step": 2016 + }, + { + "epoch": 0.81, + "learning_rate": 1.9265876916250026e-06, + "loss": 0.7151, + "step": 2017 + }, + { + "epoch": 0.81, + "learning_rate": 1.918962473339496e-06, + "loss": 0.623, + "step": 2018 + }, + { + "epoch": 0.81, + "learning_rate": 1.9113507730333435e-06, + "loss": 0.6589, + "step": 2019 + }, + { + "epoch": 0.81, + "learning_rate": 1.9037526034394149e-06, + "loss": 0.6568, + "step": 2020 + }, + { + "epoch": 0.81, + "learning_rate": 1.8961679772679486e-06, + "loss": 0.7222, + "step": 2021 + }, + { + "epoch": 0.81, + "learning_rate": 1.8885969072065224e-06, + "loss": 0.6626, + "step": 2022 + }, + { + "epoch": 0.81, + "learning_rate": 1.88103940592004e-06, + "loss": 0.7087, + "step": 2023 + }, + { + "epoch": 0.81, + "learning_rate": 1.873495486050716e-06, + "loss": 0.5933, + "step": 2024 + }, + { + "epoch": 0.81, + "learning_rate": 1.8659651602180295e-06, + "loss": 0.673, + "step": 2025 + }, + { + "epoch": 0.81, + "learning_rate": 1.858448441018732e-06, + "loss": 0.7387, + "step": 2026 + }, + { + "epoch": 0.81, + "learning_rate": 1.8509453410268086e-06, + "loss": 0.6716, + "step": 2027 + }, + { + "epoch": 0.81, + "learning_rate": 1.8434558727934636e-06, + "loss": 0.7268, + "step": 2028 + }, + { + "epoch": 0.81, + "learning_rate": 1.835980048847098e-06, + "loss": 0.6437, + "step": 2029 + }, + { + "epoch": 0.81, + "learning_rate": 1.8285178816932913e-06, + "loss": 0.6885, + "step": 2030 + }, + { + "epoch": 0.81, + "learning_rate": 1.8210693838147686e-06, + "loss": 0.6504, + "step": 2031 + }, + { + "epoch": 0.81, + "learning_rate": 1.8136345676713962e-06, + "loss": 0.6063, + "step": 2032 + }, + { + "epoch": 0.81, + "learning_rate": 1.8062134457001612e-06, + "loss": 0.6962, + "step": 2033 + }, + { + "epoch": 0.81, + "learning_rate": 1.7988060303151256e-06, + "loss": 0.641, + "step": 2034 + }, + { + "epoch": 0.81, + "learning_rate": 1.7914123339074351e-06, + "loss": 0.6483, + "step": 2035 + }, + { + "epoch": 0.81, + "learning_rate": 1.7840323688452833e-06, + "loss": 0.7294, + "step": 2036 + }, + { + "epoch": 0.81, + "learning_rate": 1.7766661474738933e-06, + "loss": 0.6718, + "step": 2037 + }, + { + "epoch": 0.81, + "learning_rate": 1.7693136821154989e-06, + "loss": 0.6823, + "step": 2038 + }, + { + "epoch": 0.81, + "learning_rate": 1.7619749850693235e-06, + "loss": 0.6257, + "step": 2039 + }, + { + "epoch": 0.81, + "learning_rate": 1.7546500686115542e-06, + "loss": 0.7181, + "step": 2040 + }, + { + "epoch": 0.81, + "learning_rate": 1.7473389449953304e-06, + "loss": 0.6958, + "step": 2041 + }, + { + "epoch": 0.82, + "learning_rate": 1.7400416264507192e-06, + "loss": 0.6977, + "step": 2042 + }, + { + "epoch": 0.82, + "learning_rate": 1.7327581251846902e-06, + "loss": 0.669, + "step": 2043 + }, + { + "epoch": 0.82, + "learning_rate": 1.7254884533811034e-06, + "loss": 0.6546, + "step": 2044 + }, + { + "epoch": 0.82, + "learning_rate": 1.7182326232006875e-06, + "loss": 0.6974, + "step": 2045 + }, + { + "epoch": 0.82, + "learning_rate": 1.7109906467810033e-06, + "loss": 0.7047, + "step": 2046 + }, + { + "epoch": 0.82, + "learning_rate": 1.7037625362364585e-06, + "loss": 0.642, + "step": 2047 + }, + { + "epoch": 0.82, + "learning_rate": 1.6965483036582454e-06, + "loss": 0.6709, + "step": 2048 + }, + { + "epoch": 0.82, + "learning_rate": 1.6893479611143527e-06, + "loss": 0.6953, + "step": 2049 + }, + { + "epoch": 0.82, + "learning_rate": 1.6821615206495313e-06, + "loss": 0.7387, + "step": 2050 + }, + { + "epoch": 0.82, + "learning_rate": 1.674988994285276e-06, + "loss": 0.6685, + "step": 2051 + }, + { + "epoch": 0.82, + "learning_rate": 1.667830394019807e-06, + "loss": 0.6557, + "step": 2052 + }, + { + "epoch": 0.82, + "learning_rate": 1.6606857318280478e-06, + "loss": 0.709, + "step": 2053 + }, + { + "epoch": 0.82, + "learning_rate": 1.6535550196616102e-06, + "loss": 0.6547, + "step": 2054 + }, + { + "epoch": 0.82, + "learning_rate": 1.6464382694487579e-06, + "loss": 0.5857, + "step": 2055 + }, + { + "epoch": 0.82, + "learning_rate": 1.6393354930944194e-06, + "loss": 0.7401, + "step": 2056 + }, + { + "epoch": 0.82, + "learning_rate": 1.6322467024801282e-06, + "loss": 0.6954, + "step": 2057 + }, + { + "epoch": 0.82, + "learning_rate": 1.6251719094640328e-06, + "loss": 0.6253, + "step": 2058 + }, + { + "epoch": 0.82, + "learning_rate": 1.6181111258808658e-06, + "loss": 0.6421, + "step": 2059 + }, + { + "epoch": 0.82, + "learning_rate": 1.611064363541922e-06, + "loss": 0.6626, + "step": 2060 + }, + { + "epoch": 0.82, + "learning_rate": 1.6040316342350425e-06, + "loss": 0.6445, + "step": 2061 + }, + { + "epoch": 0.82, + "learning_rate": 1.597012949724598e-06, + "loss": 0.5191, + "step": 2062 + }, + { + "epoch": 0.82, + "learning_rate": 1.590008321751454e-06, + "loss": 0.7103, + "step": 2063 + }, + { + "epoch": 0.82, + "learning_rate": 1.5830177620329712e-06, + "loss": 0.6725, + "step": 2064 + }, + { + "epoch": 0.82, + "learning_rate": 1.5760412822629822e-06, + "loss": 0.684, + "step": 2065 + }, + { + "epoch": 0.82, + "learning_rate": 1.5690788941117508e-06, + "loss": 0.6522, + "step": 2066 + }, + { + "epoch": 0.83, + "learning_rate": 1.5621306092259804e-06, + "loss": 0.6748, + "step": 2067 + }, + { + "epoch": 0.83, + "learning_rate": 1.5551964392287788e-06, + "loss": 0.6418, + "step": 2068 + }, + { + "epoch": 0.83, + "learning_rate": 1.5482763957196423e-06, + "loss": 0.6313, + "step": 2069 + }, + { + "epoch": 0.83, + "learning_rate": 1.5413704902744364e-06, + "loss": 0.6608, + "step": 2070 + }, + { + "epoch": 0.83, + "learning_rate": 1.5344787344453803e-06, + "loss": 0.7146, + "step": 2071 + }, + { + "epoch": 0.83, + "learning_rate": 1.5276011397610136e-06, + "loss": 0.6825, + "step": 2072 + }, + { + "epoch": 0.83, + "learning_rate": 1.520737717726195e-06, + "loss": 0.5956, + "step": 2073 + }, + { + "epoch": 0.83, + "learning_rate": 1.5138884798220798e-06, + "loss": 0.6534, + "step": 2074 + }, + { + "epoch": 0.83, + "learning_rate": 1.507053437506083e-06, + "loss": 0.713, + "step": 2075 + }, + { + "epoch": 0.83, + "learning_rate": 1.500232602211883e-06, + "loss": 0.6835, + "step": 2076 + }, + { + "epoch": 0.83, + "learning_rate": 1.4934259853493883e-06, + "loss": 0.7259, + "step": 2077 + }, + { + "epoch": 0.83, + "learning_rate": 1.4866335983047264e-06, + "loss": 0.7011, + "step": 2078 + }, + { + "epoch": 0.83, + "learning_rate": 1.4798554524402176e-06, + "loss": 0.6075, + "step": 2079 + }, + { + "epoch": 0.83, + "learning_rate": 1.4730915590943628e-06, + "loss": 0.7029, + "step": 2080 + }, + { + "epoch": 0.83, + "learning_rate": 1.466341929581816e-06, + "loss": 0.6197, + "step": 2081 + }, + { + "epoch": 0.83, + "learning_rate": 1.4596065751933763e-06, + "loss": 0.6749, + "step": 2082 + }, + { + "epoch": 0.83, + "learning_rate": 1.452885507195959e-06, + "loss": 0.6907, + "step": 2083 + }, + { + "epoch": 0.83, + "learning_rate": 1.4461787368325863e-06, + "loss": 0.6944, + "step": 2084 + }, + { + "epoch": 0.83, + "learning_rate": 1.439486275322357e-06, + "loss": 0.6913, + "step": 2085 + }, + { + "epoch": 0.83, + "learning_rate": 1.4328081338604383e-06, + "loss": 0.7284, + "step": 2086 + }, + { + "epoch": 0.83, + "learning_rate": 1.4261443236180418e-06, + "loss": 0.7109, + "step": 2087 + }, + { + "epoch": 0.83, + "learning_rate": 1.4194948557424083e-06, + "loss": 0.6825, + "step": 2088 + }, + { + "epoch": 0.83, + "learning_rate": 1.412859741356778e-06, + "loss": 0.6814, + "step": 2089 + }, + { + "epoch": 0.83, + "learning_rate": 1.4062389915603903e-06, + "loss": 0.6101, + "step": 2090 + }, + { + "epoch": 0.83, + "learning_rate": 1.3996326174284502e-06, + "loss": 0.6181, + "step": 2091 + }, + { + "epoch": 0.84, + "learning_rate": 1.393040630012118e-06, + "loss": 0.6933, + "step": 2092 + }, + { + "epoch": 0.84, + "learning_rate": 1.386463040338485e-06, + "loss": 0.6694, + "step": 2093 + }, + { + "epoch": 0.84, + "learning_rate": 1.3798998594105606e-06, + "loss": 0.7168, + "step": 2094 + }, + { + "epoch": 0.84, + "learning_rate": 1.3733510982072485e-06, + "loss": 0.6924, + "step": 2095 + }, + { + "epoch": 0.84, + "learning_rate": 1.3668167676833332e-06, + "loss": 0.6372, + "step": 2096 + }, + { + "epoch": 0.84, + "learning_rate": 1.3602968787694615e-06, + "loss": 0.6818, + "step": 2097 + }, + { + "epoch": 0.84, + "learning_rate": 1.3537914423721144e-06, + "loss": 0.7029, + "step": 2098 + }, + { + "epoch": 0.84, + "learning_rate": 1.3473004693736037e-06, + "loss": 0.5992, + "step": 2099 + }, + { + "epoch": 0.84, + "learning_rate": 1.3408239706320459e-06, + "loss": 0.7204, + "step": 2100 + }, + { + "epoch": 0.84, + "learning_rate": 1.3343619569813437e-06, + "loss": 0.692, + "step": 2101 + }, + { + "epoch": 0.84, + "learning_rate": 1.3279144392311693e-06, + "loss": 0.6072, + "step": 2102 + }, + { + "epoch": 0.84, + "learning_rate": 1.3214814281669486e-06, + "loss": 0.5775, + "step": 2103 + }, + { + "epoch": 0.84, + "learning_rate": 1.31506293454983e-06, + "loss": 0.6609, + "step": 2104 + }, + { + "epoch": 0.84, + "learning_rate": 1.308658969116694e-06, + "loss": 0.6395, + "step": 2105 + }, + { + "epoch": 0.84, + "learning_rate": 1.302269542580109e-06, + "loss": 0.6793, + "step": 2106 + }, + { + "epoch": 0.84, + "learning_rate": 1.2958946656283188e-06, + "loss": 0.6859, + "step": 2107 + }, + { + "epoch": 0.84, + "learning_rate": 1.2895343489252355e-06, + "loss": 0.6091, + "step": 2108 + }, + { + "epoch": 0.84, + "learning_rate": 1.2831886031104123e-06, + "loss": 0.6474, + "step": 2109 + }, + { + "epoch": 0.84, + "learning_rate": 1.2768574387990284e-06, + "loss": 0.6831, + "step": 2110 + }, + { + "epoch": 0.84, + "learning_rate": 1.2705408665818707e-06, + "loss": 0.6464, + "step": 2111 + }, + { + "epoch": 0.84, + "learning_rate": 1.2642388970253194e-06, + "loss": 0.6392, + "step": 2112 + }, + { + "epoch": 0.84, + "learning_rate": 1.2579515406713194e-06, + "loss": 0.6657, + "step": 2113 + }, + { + "epoch": 0.84, + "learning_rate": 1.2516788080373766e-06, + "loss": 0.6294, + "step": 2114 + }, + { + "epoch": 0.84, + "learning_rate": 1.2454207096165405e-06, + "loss": 0.6117, + "step": 2115 + }, + { + "epoch": 0.84, + "learning_rate": 1.239177255877365e-06, + "loss": 0.7068, + "step": 2116 + }, + { + "epoch": 0.85, + "learning_rate": 1.2329484572639183e-06, + "loss": 0.6672, + "step": 2117 + }, + { + "epoch": 0.85, + "learning_rate": 1.2267343241957508e-06, + "loss": 0.7194, + "step": 2118 + }, + { + "epoch": 0.85, + "learning_rate": 1.2205348670678774e-06, + "loss": 0.6514, + "step": 2119 + }, + { + "epoch": 0.85, + "learning_rate": 1.214350096250767e-06, + "loss": 0.6784, + "step": 2120 + }, + { + "epoch": 0.85, + "learning_rate": 1.2081800220903207e-06, + "loss": 0.7043, + "step": 2121 + }, + { + "epoch": 0.85, + "learning_rate": 1.2020246549078497e-06, + "loss": 0.716, + "step": 2122 + }, + { + "epoch": 0.85, + "learning_rate": 1.1958840050000686e-06, + "loss": 0.6255, + "step": 2123 + }, + { + "epoch": 0.85, + "learning_rate": 1.189758082639072e-06, + "loss": 0.6564, + "step": 2124 + }, + { + "epoch": 0.85, + "learning_rate": 1.183646898072318e-06, + "loss": 0.7081, + "step": 2125 + }, + { + "epoch": 0.85, + "learning_rate": 1.1775504615226118e-06, + "loss": 0.6784, + "step": 2126 + }, + { + "epoch": 0.85, + "learning_rate": 1.1714687831880867e-06, + "loss": 0.7031, + "step": 2127 + }, + { + "epoch": 0.85, + "learning_rate": 1.165401873242188e-06, + "loss": 0.6548, + "step": 2128 + }, + { + "epoch": 0.85, + "learning_rate": 1.1593497418336619e-06, + "loss": 0.662, + "step": 2129 + }, + { + "epoch": 0.85, + "learning_rate": 1.1533123990865225e-06, + "loss": 0.6395, + "step": 2130 + }, + { + "epoch": 0.85, + "learning_rate": 1.1472898551000544e-06, + "loss": 0.6303, + "step": 2131 + }, + { + "epoch": 0.85, + "learning_rate": 1.1412821199487845e-06, + "loss": 0.6664, + "step": 2132 + }, + { + "epoch": 0.85, + "learning_rate": 1.135289203682466e-06, + "loss": 0.6883, + "step": 2133 + }, + { + "epoch": 0.85, + "learning_rate": 1.1293111163260639e-06, + "loss": 0.614, + "step": 2134 + }, + { + "epoch": 0.85, + "learning_rate": 1.123347867879737e-06, + "loss": 0.7076, + "step": 2135 + }, + { + "epoch": 0.85, + "learning_rate": 1.117399468318824e-06, + "loss": 0.675, + "step": 2136 + }, + { + "epoch": 0.85, + "learning_rate": 1.1114659275938188e-06, + "loss": 0.6843, + "step": 2137 + }, + { + "epoch": 0.85, + "learning_rate": 1.105547255630368e-06, + "loss": 0.6603, + "step": 2138 + }, + { + "epoch": 0.85, + "learning_rate": 1.099643462329235e-06, + "loss": 0.6983, + "step": 2139 + }, + { + "epoch": 0.85, + "learning_rate": 1.0937545575663023e-06, + "loss": 0.7402, + "step": 2140 + }, + { + "epoch": 0.85, + "learning_rate": 1.0878805511925438e-06, + "loss": 0.5793, + "step": 2141 + }, + { + "epoch": 0.86, + "learning_rate": 1.0820214530340124e-06, + "loss": 0.6841, + "step": 2142 + }, + { + "epoch": 0.86, + "learning_rate": 1.0761772728918208e-06, + "loss": 0.6077, + "step": 2143 + }, + { + "epoch": 0.86, + "learning_rate": 1.0703480205421302e-06, + "loss": 0.6417, + "step": 2144 + }, + { + "epoch": 0.86, + "learning_rate": 1.0645337057361215e-06, + "loss": 0.632, + "step": 2145 + }, + { + "epoch": 0.86, + "learning_rate": 1.0587343382000027e-06, + "loss": 0.6747, + "step": 2146 + }, + { + "epoch": 0.86, + "learning_rate": 1.0529499276349686e-06, + "loss": 0.5994, + "step": 2147 + }, + { + "epoch": 0.86, + "learning_rate": 1.0471804837171916e-06, + "loss": 0.5687, + "step": 2148 + }, + { + "epoch": 0.86, + "learning_rate": 1.0414260160978141e-06, + "loss": 0.6707, + "step": 2149 + }, + { + "epoch": 0.86, + "learning_rate": 1.0356865344029231e-06, + "loss": 0.4886, + "step": 2150 + }, + { + "epoch": 0.86, + "learning_rate": 1.029962048233537e-06, + "loss": 0.6509, + "step": 2151 + }, + { + "epoch": 0.86, + "learning_rate": 1.0242525671655912e-06, + "loss": 0.6144, + "step": 2152 + }, + { + "epoch": 0.86, + "learning_rate": 1.0185581007499212e-06, + "loss": 0.6814, + "step": 2153 + }, + { + "epoch": 0.86, + "learning_rate": 1.0128786585122385e-06, + "loss": 0.6658, + "step": 2154 + }, + { + "epoch": 0.86, + "learning_rate": 1.0072142499531346e-06, + "loss": 0.676, + "step": 2155 + }, + { + "epoch": 0.86, + "learning_rate": 1.0015648845480453e-06, + "loss": 0.5966, + "step": 2156 + }, + { + "epoch": 0.86, + "learning_rate": 9.95930571747239e-07, + "loss": 0.6368, + "step": 2157 + }, + { + "epoch": 0.86, + "learning_rate": 9.903113209758098e-07, + "loss": 0.6959, + "step": 2158 + }, + { + "epoch": 0.86, + "learning_rate": 9.84707141633654e-07, + "loss": 0.6308, + "step": 2159 + }, + { + "epoch": 0.86, + "learning_rate": 9.791180430954562e-07, + "loss": 0.6793, + "step": 2160 + }, + { + "epoch": 0.86, + "learning_rate": 9.735440347106762e-07, + "loss": 0.6606, + "step": 2161 + }, + { + "epoch": 0.86, + "learning_rate": 9.679851258035277e-07, + "loss": 0.6803, + "step": 2162 + }, + { + "epoch": 0.86, + "learning_rate": 9.624413256729636e-07, + "loss": 0.5774, + "step": 2163 + }, + { + "epoch": 0.86, + "learning_rate": 9.569126435926735e-07, + "loss": 0.7197, + "step": 2164 + }, + { + "epoch": 0.86, + "learning_rate": 9.513990888110447e-07, + "loss": 0.6423, + "step": 2165 + }, + { + "epoch": 0.86, + "learning_rate": 9.459006705511664e-07, + "loss": 0.6318, + "step": 2166 + }, + { + "epoch": 0.87, + "learning_rate": 9.404173980108067e-07, + "loss": 0.6008, + "step": 2167 + }, + { + "epoch": 0.87, + "learning_rate": 9.349492803623972e-07, + "loss": 0.7172, + "step": 2168 + }, + { + "epoch": 0.87, + "learning_rate": 9.294963267530177e-07, + "loss": 0.7605, + "step": 2169 + }, + { + "epoch": 0.87, + "learning_rate": 9.240585463043849e-07, + "loss": 0.6705, + "step": 2170 + }, + { + "epoch": 0.87, + "learning_rate": 9.186359481128282e-07, + "loss": 0.619, + "step": 2171 + }, + { + "epoch": 0.87, + "learning_rate": 9.132285412492825e-07, + "loss": 0.6655, + "step": 2172 + }, + { + "epoch": 0.87, + "learning_rate": 9.078363347592767e-07, + "loss": 0.6528, + "step": 2173 + }, + { + "epoch": 0.87, + "learning_rate": 9.024593376629009e-07, + "loss": 0.6338, + "step": 2174 + }, + { + "epoch": 0.87, + "learning_rate": 8.970975589548114e-07, + "loss": 0.7132, + "step": 2175 + }, + { + "epoch": 0.87, + "learning_rate": 8.917510076042058e-07, + "loss": 0.6475, + "step": 2176 + }, + { + "epoch": 0.87, + "learning_rate": 8.864196925548063e-07, + "loss": 0.5703, + "step": 2177 + }, + { + "epoch": 0.87, + "learning_rate": 8.811036227248515e-07, + "loss": 0.6849, + "step": 2178 + }, + { + "epoch": 0.87, + "learning_rate": 8.758028070070768e-07, + "loss": 0.7121, + "step": 2179 + }, + { + "epoch": 0.87, + "learning_rate": 8.705172542686968e-07, + "loss": 0.6272, + "step": 2180 + }, + { + "epoch": 0.87, + "learning_rate": 8.652469733513958e-07, + "loss": 0.7059, + "step": 2181 + }, + { + "epoch": 0.87, + "learning_rate": 8.599919730713191e-07, + "loss": 0.6933, + "step": 2182 + }, + { + "epoch": 0.87, + "learning_rate": 8.547522622190385e-07, + "loss": 0.5546, + "step": 2183 + }, + { + "epoch": 0.87, + "learning_rate": 8.495278495595572e-07, + "loss": 0.6194, + "step": 2184 + }, + { + "epoch": 0.87, + "learning_rate": 8.443187438322864e-07, + "loss": 0.6409, + "step": 2185 + }, + { + "epoch": 0.87, + "learning_rate": 8.391249537510271e-07, + "loss": 0.6351, + "step": 2186 + }, + { + "epoch": 0.87, + "learning_rate": 8.339464880039671e-07, + "loss": 0.6684, + "step": 2187 + }, + { + "epoch": 0.87, + "learning_rate": 8.287833552536584e-07, + "loss": 0.6821, + "step": 2188 + }, + { + "epoch": 0.87, + "learning_rate": 8.236355641369975e-07, + "loss": 0.7171, + "step": 2189 + }, + { + "epoch": 0.87, + "learning_rate": 8.185031232652252e-07, + "loss": 0.6063, + "step": 2190 + }, + { + "epoch": 0.87, + "learning_rate": 8.133860412238992e-07, + "loss": 0.5808, + "step": 2191 + }, + { + "epoch": 0.88, + "learning_rate": 8.082843265728879e-07, + "loss": 0.6641, + "step": 2192 + }, + { + "epoch": 0.88, + "learning_rate": 8.031979878463525e-07, + "loss": 0.752, + "step": 2193 + }, + { + "epoch": 0.88, + "learning_rate": 7.981270335527347e-07, + "loss": 0.655, + "step": 2194 + }, + { + "epoch": 0.88, + "learning_rate": 7.930714721747323e-07, + "loss": 0.6117, + "step": 2195 + }, + { + "epoch": 0.88, + "learning_rate": 7.880313121693073e-07, + "loss": 0.6563, + "step": 2196 + }, + { + "epoch": 0.88, + "learning_rate": 7.830065619676519e-07, + "loss": 0.6065, + "step": 2197 + }, + { + "epoch": 0.88, + "learning_rate": 7.77997229975177e-07, + "loss": 0.5963, + "step": 2198 + }, + { + "epoch": 0.88, + "learning_rate": 7.730033245715063e-07, + "loss": 0.6669, + "step": 2199 + }, + { + "epoch": 0.88, + "learning_rate": 7.680248541104574e-07, + "loss": 0.7173, + "step": 2200 + }, + { + "epoch": 0.88, + "learning_rate": 7.630618269200285e-07, + "loss": 0.698, + "step": 2201 + }, + { + "epoch": 0.88, + "learning_rate": 7.58114251302382e-07, + "loss": 0.6352, + "step": 2202 + }, + { + "epoch": 0.88, + "learning_rate": 7.531821355338386e-07, + "loss": 0.61, + "step": 2203 + }, + { + "epoch": 0.88, + "learning_rate": 7.482654878648465e-07, + "loss": 0.6725, + "step": 2204 + }, + { + "epoch": 0.88, + "learning_rate": 7.433643165199933e-07, + "loss": 0.6275, + "step": 2205 + }, + { + "epoch": 0.88, + "learning_rate": 7.384786296979662e-07, + "loss": 0.6146, + "step": 2206 + }, + { + "epoch": 0.88, + "learning_rate": 7.336084355715555e-07, + "loss": 0.6334, + "step": 2207 + }, + { + "epoch": 0.88, + "learning_rate": 7.287537422876323e-07, + "loss": 0.596, + "step": 2208 + }, + { + "epoch": 0.88, + "learning_rate": 7.239145579671414e-07, + "loss": 0.7014, + "step": 2209 + }, + { + "epoch": 0.88, + "learning_rate": 7.190908907050808e-07, + "loss": 0.6938, + "step": 2210 + }, + { + "epoch": 0.88, + "learning_rate": 7.142827485704951e-07, + "loss": 0.6408, + "step": 2211 + }, + { + "epoch": 0.88, + "learning_rate": 7.094901396064535e-07, + "loss": 0.6939, + "step": 2212 + }, + { + "epoch": 0.88, + "learning_rate": 7.047130718300433e-07, + "loss": 0.6949, + "step": 2213 + }, + { + "epoch": 0.88, + "learning_rate": 6.999515532323598e-07, + "loss": 0.5715, + "step": 2214 + }, + { + "epoch": 0.88, + "learning_rate": 6.952055917784783e-07, + "loss": 0.6716, + "step": 2215 + }, + { + "epoch": 0.88, + "learning_rate": 6.904751954074574e-07, + "loss": 0.7574, + "step": 2216 + }, + { + "epoch": 0.89, + "learning_rate": 6.857603720323136e-07, + "loss": 0.6852, + "step": 2217 + }, + { + "epoch": 0.89, + "learning_rate": 6.810611295400171e-07, + "loss": 0.7108, + "step": 2218 + }, + { + "epoch": 0.89, + "learning_rate": 6.76377475791471e-07, + "loss": 0.6669, + "step": 2219 + }, + { + "epoch": 0.89, + "learning_rate": 6.717094186215056e-07, + "loss": 0.6764, + "step": 2220 + }, + { + "epoch": 0.89, + "learning_rate": 6.670569658388537e-07, + "loss": 0.7655, + "step": 2221 + }, + { + "epoch": 0.89, + "learning_rate": 6.624201252261497e-07, + "loss": 0.6326, + "step": 2222 + }, + { + "epoch": 0.89, + "learning_rate": 6.57798904539918e-07, + "loss": 0.6373, + "step": 2223 + }, + { + "epoch": 0.89, + "learning_rate": 6.531933115105427e-07, + "loss": 0.6847, + "step": 2224 + }, + { + "epoch": 0.89, + "learning_rate": 6.48603353842272e-07, + "loss": 0.5997, + "step": 2225 + }, + { + "epoch": 0.89, + "learning_rate": 6.440290392131998e-07, + "loss": 0.694, + "step": 2226 + }, + { + "epoch": 0.89, + "learning_rate": 6.394703752752474e-07, + "loss": 0.6807, + "step": 2227 + }, + { + "epoch": 0.89, + "learning_rate": 6.349273696541625e-07, + "loss": 0.6281, + "step": 2228 + }, + { + "epoch": 0.89, + "learning_rate": 6.304000299494961e-07, + "loss": 0.621, + "step": 2229 + }, + { + "epoch": 0.89, + "learning_rate": 6.258883637345914e-07, + "loss": 0.6529, + "step": 2230 + }, + { + "epoch": 0.89, + "learning_rate": 6.213923785565723e-07, + "loss": 0.6779, + "step": 2231 + }, + { + "epoch": 0.89, + "learning_rate": 6.169120819363406e-07, + "loss": 0.6859, + "step": 2232 + }, + { + "epoch": 0.89, + "learning_rate": 6.124474813685422e-07, + "loss": 0.6241, + "step": 2233 + }, + { + "epoch": 0.89, + "learning_rate": 6.079985843215719e-07, + "loss": 0.6923, + "step": 2234 + }, + { + "epoch": 0.89, + "learning_rate": 6.0356539823756e-07, + "loss": 0.6782, + "step": 2235 + }, + { + "epoch": 0.89, + "learning_rate": 5.991479305323433e-07, + "loss": 0.6957, + "step": 2236 + }, + { + "epoch": 0.89, + "learning_rate": 5.947461885954798e-07, + "loss": 0.641, + "step": 2237 + }, + { + "epoch": 0.89, + "learning_rate": 5.903601797902126e-07, + "loss": 0.7299, + "step": 2238 + }, + { + "epoch": 0.89, + "learning_rate": 5.859899114534662e-07, + "loss": 0.6611, + "step": 2239 + }, + { + "epoch": 0.89, + "learning_rate": 5.816353908958361e-07, + "loss": 0.6881, + "step": 2240 + }, + { + "epoch": 0.89, + "learning_rate": 5.772966254015755e-07, + "loss": 0.6499, + "step": 2241 + }, + { + "epoch": 0.9, + "learning_rate": 5.729736222285832e-07, + "loss": 0.7044, + "step": 2242 + }, + { + "epoch": 0.9, + "learning_rate": 5.686663886083877e-07, + "loss": 0.6453, + "step": 2243 + }, + { + "epoch": 0.9, + "learning_rate": 5.643749317461422e-07, + "loss": 0.7034, + "step": 2244 + }, + { + "epoch": 0.9, + "learning_rate": 5.600992588206011e-07, + "loss": 0.6614, + "step": 2245 + }, + { + "epoch": 0.9, + "learning_rate": 5.558393769841286e-07, + "loss": 0.6215, + "step": 2246 + }, + { + "epoch": 0.9, + "learning_rate": 5.51595293362659e-07, + "loss": 0.7524, + "step": 2247 + }, + { + "epoch": 0.9, + "learning_rate": 5.473670150557076e-07, + "loss": 0.6393, + "step": 2248 + }, + { + "epoch": 0.9, + "learning_rate": 5.431545491363488e-07, + "loss": 0.6916, + "step": 2249 + }, + { + "epoch": 0.9, + "learning_rate": 5.389579026512059e-07, + "loss": 0.6528, + "step": 2250 + }, + { + "epoch": 0.9, + "learning_rate": 5.347770826204401e-07, + "loss": 0.7439, + "step": 2251 + }, + { + "epoch": 0.9, + "learning_rate": 5.306120960377381e-07, + "loss": 0.6712, + "step": 2252 + }, + { + "epoch": 0.9, + "learning_rate": 5.264629498702966e-07, + "loss": 0.6405, + "step": 2253 + }, + { + "epoch": 0.9, + "learning_rate": 5.223296510588182e-07, + "loss": 0.605, + "step": 2254 + }, + { + "epoch": 0.9, + "learning_rate": 5.182122065174988e-07, + "loss": 0.6287, + "step": 2255 + }, + { + "epoch": 0.9, + "learning_rate": 5.141106231340065e-07, + "loss": 0.6222, + "step": 2256 + }, + { + "epoch": 0.9, + "learning_rate": 5.100249077694797e-07, + "loss": 0.6635, + "step": 2257 + }, + { + "epoch": 0.9, + "learning_rate": 5.059550672585145e-07, + "loss": 0.6284, + "step": 2258 + }, + { + "epoch": 0.9, + "learning_rate": 5.019011084091496e-07, + "loss": 0.666, + "step": 2259 + }, + { + "epoch": 0.9, + "learning_rate": 4.978630380028582e-07, + "loss": 0.6597, + "step": 2260 + }, + { + "epoch": 0.9, + "learning_rate": 4.938408627945335e-07, + "loss": 0.6329, + "step": 2261 + }, + { + "epoch": 0.9, + "learning_rate": 4.898345895124801e-07, + "loss": 0.7093, + "step": 2262 + }, + { + "epoch": 0.9, + "learning_rate": 4.858442248583995e-07, + "loss": 0.6425, + "step": 2263 + }, + { + "epoch": 0.9, + "learning_rate": 4.818697755073876e-07, + "loss": 0.6582, + "step": 2264 + }, + { + "epoch": 0.9, + "learning_rate": 4.779112481079107e-07, + "loss": 0.5798, + "step": 2265 + }, + { + "epoch": 0.9, + "learning_rate": 4.7396864928180296e-07, + "loss": 0.624, + "step": 2266 + }, + { + "epoch": 0.9, + "learning_rate": 4.700419856242555e-07, + "loss": 0.6741, + "step": 2267 + }, + { + "epoch": 0.91, + "learning_rate": 4.6613126370379627e-07, + "loss": 0.7305, + "step": 2268 + }, + { + "epoch": 0.91, + "learning_rate": 4.6223649006229553e-07, + "loss": 0.7157, + "step": 2269 + }, + { + "epoch": 0.91, + "learning_rate": 4.583576712149385e-07, + "loss": 0.5922, + "step": 2270 + }, + { + "epoch": 0.91, + "learning_rate": 4.5449481365022143e-07, + "loss": 0.6156, + "step": 2271 + }, + { + "epoch": 0.91, + "learning_rate": 4.5064792382994216e-07, + "loss": 0.6635, + "step": 2272 + }, + { + "epoch": 0.91, + "learning_rate": 4.468170081891898e-07, + "loss": 0.6872, + "step": 2273 + }, + { + "epoch": 0.91, + "learning_rate": 4.4300207313632713e-07, + "loss": 0.6408, + "step": 2274 + }, + { + "epoch": 0.91, + "learning_rate": 4.392031250529871e-07, + "loss": 0.6066, + "step": 2275 + }, + { + "epoch": 0.91, + "learning_rate": 4.3542017029406083e-07, + "loss": 0.6139, + "step": 2276 + }, + { + "epoch": 0.91, + "learning_rate": 4.316532151876807e-07, + "loss": 0.7126, + "step": 2277 + }, + { + "epoch": 0.91, + "learning_rate": 4.279022660352228e-07, + "loss": 0.7134, + "step": 2278 + }, + { + "epoch": 0.91, + "learning_rate": 4.2416732911128247e-07, + "loss": 0.7205, + "step": 2279 + }, + { + "epoch": 0.91, + "learning_rate": 4.2044841066367084e-07, + "loss": 0.6782, + "step": 2280 + }, + { + "epoch": 0.91, + "learning_rate": 4.167455169134027e-07, + "loss": 0.694, + "step": 2281 + }, + { + "epoch": 0.91, + "learning_rate": 4.130586540546899e-07, + "loss": 0.7133, + "step": 2282 + }, + { + "epoch": 0.91, + "learning_rate": 4.093878282549246e-07, + "loss": 0.6632, + "step": 2283 + }, + { + "epoch": 0.91, + "learning_rate": 4.057330456546737e-07, + "loss": 0.7287, + "step": 2284 + }, + { + "epoch": 0.91, + "learning_rate": 4.0209431236766793e-07, + "loss": 0.6384, + "step": 2285 + }, + { + "epoch": 0.91, + "learning_rate": 3.9847163448078373e-07, + "loss": 0.6906, + "step": 2286 + }, + { + "epoch": 0.91, + "learning_rate": 3.9486501805405253e-07, + "loss": 0.664, + "step": 2287 + }, + { + "epoch": 0.91, + "learning_rate": 3.9127446912062606e-07, + "loss": 0.6093, + "step": 2288 + }, + { + "epoch": 0.91, + "learning_rate": 3.876999936867865e-07, + "loss": 0.6288, + "step": 2289 + }, + { + "epoch": 0.91, + "learning_rate": 3.8414159773192294e-07, + "loss": 0.6363, + "step": 2290 + }, + { + "epoch": 0.91, + "learning_rate": 3.8059928720852957e-07, + "loss": 0.6301, + "step": 2291 + }, + { + "epoch": 0.91, + "learning_rate": 3.770730680421919e-07, + "loss": 0.6459, + "step": 2292 + }, + { + "epoch": 0.92, + "learning_rate": 3.735629461315804e-07, + "loss": 0.6054, + "step": 2293 + }, + { + "epoch": 0.92, + "learning_rate": 3.7006892734843145e-07, + "loss": 0.6695, + "step": 2294 + }, + { + "epoch": 0.92, + "learning_rate": 3.6659101753754975e-07, + "loss": 0.6569, + "step": 2295 + }, + { + "epoch": 0.92, + "learning_rate": 3.6312922251679373e-07, + "loss": 0.6454, + "step": 2296 + }, + { + "epoch": 0.92, + "learning_rate": 3.5968354807705994e-07, + "loss": 0.6888, + "step": 2297 + }, + { + "epoch": 0.92, + "learning_rate": 3.5625399998228116e-07, + "loss": 0.6349, + "step": 2298 + }, + { + "epoch": 0.92, + "learning_rate": 3.528405839694149e-07, + "loss": 0.6356, + "step": 2299 + }, + { + "epoch": 0.92, + "learning_rate": 3.4944330574843144e-07, + "loss": 0.6753, + "step": 2300 + }, + { + "epoch": 0.92, + "learning_rate": 3.46062171002306e-07, + "loss": 0.7087, + "step": 2301 + }, + { + "epoch": 0.92, + "learning_rate": 3.426971853870109e-07, + "loss": 0.5928, + "step": 2302 + }, + { + "epoch": 0.92, + "learning_rate": 3.3934835453149797e-07, + "loss": 0.6814, + "step": 2303 + }, + { + "epoch": 0.92, + "learning_rate": 3.360156840377027e-07, + "loss": 0.6582, + "step": 2304 + }, + { + "epoch": 0.92, + "learning_rate": 3.3269917948052454e-07, + "loss": 0.7064, + "step": 2305 + }, + { + "epoch": 0.92, + "learning_rate": 3.293988464078202e-07, + "loss": 0.6736, + "step": 2306 + }, + { + "epoch": 0.92, + "learning_rate": 3.2611469034039334e-07, + "loss": 0.6719, + "step": 2307 + }, + { + "epoch": 0.92, + "learning_rate": 3.2284671677199065e-07, + "loss": 0.6654, + "step": 2308 + }, + { + "epoch": 0.92, + "learning_rate": 3.1959493116928473e-07, + "loss": 0.5991, + "step": 2309 + }, + { + "epoch": 0.92, + "learning_rate": 3.163593389718711e-07, + "loss": 0.6861, + "step": 2310 + }, + { + "epoch": 0.92, + "learning_rate": 3.131399455922568e-07, + "loss": 0.7172, + "step": 2311 + }, + { + "epoch": 0.92, + "learning_rate": 3.099367564158484e-07, + "loss": 0.57, + "step": 2312 + }, + { + "epoch": 0.92, + "learning_rate": 3.0674977680095086e-07, + "loss": 0.6722, + "step": 2313 + }, + { + "epoch": 0.92, + "learning_rate": 3.03579012078753e-07, + "loss": 0.6652, + "step": 2314 + }, + { + "epoch": 0.92, + "learning_rate": 3.004244675533141e-07, + "loss": 0.6412, + "step": 2315 + }, + { + "epoch": 0.92, + "learning_rate": 2.972861485015666e-07, + "loss": 0.6895, + "step": 2316 + }, + { + "epoch": 0.92, + "learning_rate": 2.941640601732976e-07, + "loss": 0.609, + "step": 2317 + }, + { + "epoch": 0.93, + "learning_rate": 2.910582077911428e-07, + "loss": 0.5773, + "step": 2318 + }, + { + "epoch": 0.93, + "learning_rate": 2.8796859655058184e-07, + "loss": 0.7484, + "step": 2319 + }, + { + "epoch": 0.93, + "learning_rate": 2.8489523161992385e-07, + "loss": 0.6316, + "step": 2320 + }, + { + "epoch": 0.93, + "learning_rate": 2.818381181402985e-07, + "loss": 0.6377, + "step": 2321 + }, + { + "epoch": 0.93, + "learning_rate": 2.787972612256551e-07, + "loss": 0.6906, + "step": 2322 + }, + { + "epoch": 0.93, + "learning_rate": 2.7577266596274577e-07, + "loss": 0.6478, + "step": 2323 + }, + { + "epoch": 0.93, + "learning_rate": 2.727643374111222e-07, + "loss": 0.6306, + "step": 2324 + }, + { + "epoch": 0.93, + "learning_rate": 2.697722806031211e-07, + "loss": 0.5978, + "step": 2325 + }, + { + "epoch": 0.93, + "learning_rate": 2.667965005438644e-07, + "loss": 0.6709, + "step": 2326 + }, + { + "epoch": 0.93, + "learning_rate": 2.6383700221124355e-07, + "loss": 0.6289, + "step": 2327 + }, + { + "epoch": 0.93, + "learning_rate": 2.6089379055591613e-07, + "loss": 0.6379, + "step": 2328 + }, + { + "epoch": 0.93, + "learning_rate": 2.579668705012917e-07, + "loss": 0.6622, + "step": 2329 + }, + { + "epoch": 0.93, + "learning_rate": 2.5505624694353027e-07, + "loss": 0.6402, + "step": 2330 + }, + { + "epoch": 0.93, + "learning_rate": 2.5216192475152945e-07, + "loss": 0.6741, + "step": 2331 + }, + { + "epoch": 0.93, + "learning_rate": 2.492839087669197e-07, + "loss": 0.6897, + "step": 2332 + }, + { + "epoch": 0.93, + "learning_rate": 2.464222038040509e-07, + "loss": 0.7218, + "step": 2333 + }, + { + "epoch": 0.93, + "learning_rate": 2.435768146499939e-07, + "loss": 0.6188, + "step": 2334 + }, + { + "epoch": 0.93, + "learning_rate": 2.407477460645191e-07, + "loss": 0.6869, + "step": 2335 + }, + { + "epoch": 0.93, + "learning_rate": 2.3793500278009974e-07, + "loss": 0.6749, + "step": 2336 + }, + { + "epoch": 0.93, + "learning_rate": 2.3513858950190206e-07, + "loss": 0.6618, + "step": 2337 + }, + { + "epoch": 0.93, + "learning_rate": 2.3235851090777084e-07, + "loss": 0.6795, + "step": 2338 + }, + { + "epoch": 0.93, + "learning_rate": 2.2959477164822942e-07, + "loss": 0.5928, + "step": 2339 + }, + { + "epoch": 0.93, + "learning_rate": 2.2684737634646737e-07, + "loss": 0.707, + "step": 2340 + }, + { + "epoch": 0.93, + "learning_rate": 2.2411632959833285e-07, + "loss": 0.6543, + "step": 2341 + }, + { + "epoch": 0.93, + "learning_rate": 2.2140163597233033e-07, + "loss": 0.6898, + "step": 2342 + }, + { + "epoch": 0.94, + "learning_rate": 2.1870330000960504e-07, + "loss": 0.7114, + "step": 2343 + }, + { + "epoch": 0.94, + "learning_rate": 2.1602132622393745e-07, + "loss": 0.7062, + "step": 2344 + }, + { + "epoch": 0.94, + "learning_rate": 2.1335571910174214e-07, + "loss": 0.6606, + "step": 2345 + }, + { + "epoch": 0.94, + "learning_rate": 2.1070648310205443e-07, + "loss": 0.7115, + "step": 2346 + }, + { + "epoch": 0.94, + "learning_rate": 2.0807362265651943e-07, + "loss": 0.6885, + "step": 2347 + }, + { + "epoch": 0.94, + "learning_rate": 2.0545714216939515e-07, + "loss": 0.6694, + "step": 2348 + }, + { + "epoch": 0.94, + "learning_rate": 2.0285704601753608e-07, + "loss": 0.6348, + "step": 2349 + }, + { + "epoch": 0.94, + "learning_rate": 2.0027333855038967e-07, + "loss": 0.686, + "step": 2350 + }, + { + "epoch": 0.94, + "learning_rate": 1.9770602408998642e-07, + "loss": 0.6286, + "step": 2351 + }, + { + "epoch": 0.94, + "learning_rate": 1.9515510693093875e-07, + "loss": 0.767, + "step": 2352 + }, + { + "epoch": 0.94, + "learning_rate": 1.9262059134042443e-07, + "loss": 0.7519, + "step": 2353 + }, + { + "epoch": 0.94, + "learning_rate": 1.9010248155818755e-07, + "loss": 0.6677, + "step": 2354 + }, + { + "epoch": 0.94, + "learning_rate": 1.8760078179653196e-07, + "loss": 0.5676, + "step": 2355 + }, + { + "epoch": 0.94, + "learning_rate": 1.8511549624030346e-07, + "loss": 0.6652, + "step": 2356 + }, + { + "epoch": 0.94, + "learning_rate": 1.8264662904689533e-07, + "loss": 0.6257, + "step": 2357 + }, + { + "epoch": 0.94, + "learning_rate": 1.8019418434623405e-07, + "loss": 0.6646, + "step": 2358 + }, + { + "epoch": 0.94, + "learning_rate": 1.7775816624077458e-07, + "loss": 0.6901, + "step": 2359 + }, + { + "epoch": 0.94, + "learning_rate": 1.753385788054962e-07, + "loss": 0.6192, + "step": 2360 + }, + { + "epoch": 0.94, + "learning_rate": 1.7293542608788905e-07, + "loss": 0.648, + "step": 2361 + }, + { + "epoch": 0.94, + "learning_rate": 1.7054871210795188e-07, + "loss": 0.6997, + "step": 2362 + }, + { + "epoch": 0.94, + "learning_rate": 1.6817844085818768e-07, + "loss": 0.6115, + "step": 2363 + }, + { + "epoch": 0.94, + "learning_rate": 1.6582461630359149e-07, + "loss": 0.6642, + "step": 2364 + }, + { + "epoch": 0.94, + "learning_rate": 1.6348724238164583e-07, + "loss": 0.6405, + "step": 2365 + }, + { + "epoch": 0.94, + "learning_rate": 1.6116632300231638e-07, + "loss": 0.6862, + "step": 2366 + }, + { + "epoch": 0.94, + "learning_rate": 1.58861862048042e-07, + "loss": 0.6917, + "step": 2367 + }, + { + "epoch": 0.95, + "learning_rate": 1.5657386337373237e-07, + "loss": 0.6955, + "step": 2368 + }, + { + "epoch": 0.95, + "learning_rate": 1.5430233080675593e-07, + "loss": 0.7348, + "step": 2369 + }, + { + "epoch": 0.95, + "learning_rate": 1.520472681469376e-07, + "loss": 0.6571, + "step": 2370 + }, + { + "epoch": 0.95, + "learning_rate": 1.4980867916655205e-07, + "loss": 0.5744, + "step": 2371 + }, + { + "epoch": 0.95, + "learning_rate": 1.475865676103161e-07, + "loss": 0.6859, + "step": 2372 + }, + { + "epoch": 0.95, + "learning_rate": 1.4538093719538404e-07, + "loss": 0.691, + "step": 2373 + }, + { + "epoch": 0.95, + "learning_rate": 1.431917916113401e-07, + "loss": 0.6932, + "step": 2374 + }, + { + "epoch": 0.95, + "learning_rate": 1.4101913452019277e-07, + "loss": 0.6029, + "step": 2375 + }, + { + "epoch": 0.95, + "learning_rate": 1.3886296955636481e-07, + "loss": 0.7028, + "step": 2376 + }, + { + "epoch": 0.95, + "learning_rate": 1.3672330032669767e-07, + "loss": 0.6845, + "step": 2377 + }, + { + "epoch": 0.95, + "learning_rate": 1.3460013041043606e-07, + "loss": 0.5391, + "step": 2378 + }, + { + "epoch": 0.95, + "learning_rate": 1.324934633592201e-07, + "loss": 0.6586, + "step": 2379 + }, + { + "epoch": 0.95, + "learning_rate": 1.304033026970908e-07, + "loss": 0.6749, + "step": 2380 + }, + { + "epoch": 0.95, + "learning_rate": 1.283296519204713e-07, + "loss": 0.665, + "step": 2381 + }, + { + "epoch": 0.95, + "learning_rate": 1.2627251449817247e-07, + "loss": 0.6648, + "step": 2382 + }, + { + "epoch": 0.95, + "learning_rate": 1.2423189387137713e-07, + "loss": 0.6161, + "step": 2383 + }, + { + "epoch": 0.95, + "learning_rate": 1.2220779345364143e-07, + "loss": 0.6739, + "step": 2384 + }, + { + "epoch": 0.95, + "learning_rate": 1.2020021663088244e-07, + "loss": 0.7221, + "step": 2385 + }, + { + "epoch": 0.95, + "learning_rate": 1.1820916676138384e-07, + "loss": 0.613, + "step": 2386 + }, + { + "epoch": 0.95, + "learning_rate": 1.1623464717577804e-07, + "loss": 0.6879, + "step": 2387 + }, + { + "epoch": 0.95, + "learning_rate": 1.1427666117704406e-07, + "loss": 0.7211, + "step": 2388 + }, + { + "epoch": 0.95, + "learning_rate": 1.1233521204050634e-07, + "loss": 0.7459, + "step": 2389 + }, + { + "epoch": 0.95, + "learning_rate": 1.1041030301382705e-07, + "loss": 0.6619, + "step": 2390 + }, + { + "epoch": 0.95, + "learning_rate": 1.085019373169971e-07, + "loss": 0.6651, + "step": 2391 + }, + { + "epoch": 0.95, + "learning_rate": 1.0661011814233624e-07, + "loss": 0.6713, + "step": 2392 + }, + { + "epoch": 0.96, + "learning_rate": 1.0473484865448524e-07, + "loss": 0.6655, + "step": 2393 + }, + { + "epoch": 0.96, + "learning_rate": 1.0287613199039592e-07, + "loss": 0.697, + "step": 2394 + }, + { + "epoch": 0.96, + "learning_rate": 1.0103397125933778e-07, + "loss": 0.6834, + "step": 2395 + }, + { + "epoch": 0.96, + "learning_rate": 9.920836954288137e-08, + "loss": 0.6264, + "step": 2396 + }, + { + "epoch": 0.96, + "learning_rate": 9.739932989489498e-08, + "loss": 0.5328, + "step": 2397 + }, + { + "epoch": 0.96, + "learning_rate": 9.56068553415479e-08, + "loss": 0.6832, + "step": 2398 + }, + { + "epoch": 0.96, + "learning_rate": 9.383094888129274e-08, + "loss": 0.6567, + "step": 2399 + }, + { + "epoch": 0.96, + "learning_rate": 9.207161348487315e-08, + "loss": 0.6521, + "step": 2400 + }, + { + "epoch": 0.96, + "learning_rate": 9.032885209530717e-08, + "loss": 0.6031, + "step": 2401 + }, + { + "epoch": 0.96, + "learning_rate": 8.860266762789283e-08, + "loss": 0.5634, + "step": 2402 + }, + { + "epoch": 0.96, + "learning_rate": 8.689306297019362e-08, + "loss": 0.6906, + "step": 2403 + }, + { + "epoch": 0.96, + "learning_rate": 8.520004098204193e-08, + "loss": 0.5702, + "step": 2404 + }, + { + "epoch": 0.96, + "learning_rate": 8.352360449552787e-08, + "loss": 0.6309, + "step": 2405 + }, + { + "epoch": 0.96, + "learning_rate": 8.186375631499821e-08, + "loss": 0.6625, + "step": 2406 + }, + { + "epoch": 0.96, + "learning_rate": 8.0220499217053e-08, + "loss": 0.7019, + "step": 2407 + }, + { + "epoch": 0.96, + "learning_rate": 7.859383595053449e-08, + "loss": 0.6384, + "step": 2408 + }, + { + "epoch": 0.96, + "learning_rate": 7.69837692365305e-08, + "loss": 0.6347, + "step": 2409 + }, + { + "epoch": 0.96, + "learning_rate": 7.539030176836549e-08, + "loss": 0.6441, + "step": 2410 + }, + { + "epoch": 0.96, + "learning_rate": 7.381343621159275e-08, + "loss": 0.6489, + "step": 2411 + }, + { + "epoch": 0.96, + "learning_rate": 7.2253175203999e-08, + "loss": 0.6471, + "step": 2412 + }, + { + "epoch": 0.96, + "learning_rate": 7.070952135559195e-08, + "loss": 0.5654, + "step": 2413 + }, + { + "epoch": 0.96, + "learning_rate": 6.918247724859939e-08, + "loss": 0.6293, + "step": 2414 + }, + { + "epoch": 0.96, + "learning_rate": 6.767204543746463e-08, + "loss": 0.6501, + "step": 2415 + }, + { + "epoch": 0.96, + "learning_rate": 6.617822844884325e-08, + "loss": 0.6355, + "step": 2416 + }, + { + "epoch": 0.96, + "learning_rate": 6.470102878159301e-08, + "loss": 0.6607, + "step": 2417 + }, + { + "epoch": 0.97, + "learning_rate": 6.324044890677839e-08, + "loss": 0.6547, + "step": 2418 + }, + { + "epoch": 0.97, + "learning_rate": 6.179649126766274e-08, + "loss": 0.609, + "step": 2419 + }, + { + "epoch": 0.97, + "learning_rate": 6.036915827969947e-08, + "loss": 0.6605, + "step": 2420 + }, + { + "epoch": 0.97, + "learning_rate": 5.895845233053643e-08, + "loss": 0.6665, + "step": 2421 + }, + { + "epoch": 0.97, + "learning_rate": 5.7564375780004843e-08, + "loss": 0.6591, + "step": 2422 + }, + { + "epoch": 0.97, + "learning_rate": 5.618693096011929e-08, + "loss": 0.6818, + "step": 2423 + }, + { + "epoch": 0.97, + "learning_rate": 5.482612017507327e-08, + "loss": 0.5735, + "step": 2424 + }, + { + "epoch": 0.97, + "learning_rate": 5.348194570123588e-08, + "loss": 0.7178, + "step": 2425 + }, + { + "epoch": 0.97, + "learning_rate": 5.2154409787141815e-08, + "loss": 0.6061, + "step": 2426 + }, + { + "epoch": 0.97, + "learning_rate": 5.084351465350024e-08, + "loss": 0.6811, + "step": 2427 + }, + { + "epoch": 0.97, + "learning_rate": 4.954926249317815e-08, + "loss": 0.6903, + "step": 2428 + }, + { + "epoch": 0.97, + "learning_rate": 4.8271655471202605e-08, + "loss": 0.7069, + "step": 2429 + }, + { + "epoch": 0.97, + "learning_rate": 4.701069572475958e-08, + "loss": 0.6505, + "step": 2430 + }, + { + "epoch": 0.97, + "learning_rate": 4.5766385363184006e-08, + "loss": 0.7024, + "step": 2431 + }, + { + "epoch": 0.97, + "learning_rate": 4.453872646796309e-08, + "loss": 0.6336, + "step": 2432 + }, + { + "epoch": 0.97, + "learning_rate": 4.3327721092726314e-08, + "loss": 0.7148, + "step": 2433 + }, + { + "epoch": 0.97, + "learning_rate": 4.213337126324768e-08, + "loss": 0.6285, + "step": 2434 + }, + { + "epoch": 0.97, + "learning_rate": 4.0955678977436796e-08, + "loss": 0.5676, + "step": 2435 + }, + { + "epoch": 0.97, + "learning_rate": 3.979464620534113e-08, + "loss": 0.619, + "step": 2436 + }, + { + "epoch": 0.97, + "learning_rate": 3.8650274889139306e-08, + "loss": 0.6693, + "step": 2437 + }, + { + "epoch": 0.97, + "learning_rate": 3.752256694313783e-08, + "loss": 0.605, + "step": 2438 + }, + { + "epoch": 0.97, + "learning_rate": 3.641152425376992e-08, + "loss": 0.6462, + "step": 2439 + }, + { + "epoch": 0.97, + "learning_rate": 3.531714867959113e-08, + "loss": 0.6499, + "step": 2440 + }, + { + "epoch": 0.97, + "learning_rate": 3.423944205127372e-08, + "loss": 0.5979, + "step": 2441 + }, + { + "epoch": 0.97, + "learning_rate": 3.3178406171608946e-08, + "loss": 0.6062, + "step": 2442 + }, + { + "epoch": 0.98, + "learning_rate": 3.213404281550148e-08, + "loss": 0.6851, + "step": 2443 + }, + { + "epoch": 0.98, + "learning_rate": 3.1106353729962734e-08, + "loss": 0.686, + "step": 2444 + }, + { + "epoch": 0.98, + "learning_rate": 3.009534063411534e-08, + "loss": 0.6795, + "step": 2445 + }, + { + "epoch": 0.98, + "learning_rate": 2.910100521918313e-08, + "loss": 0.6696, + "step": 2446 + }, + { + "epoch": 0.98, + "learning_rate": 2.812334914849335e-08, + "loss": 0.671, + "step": 2447 + }, + { + "epoch": 0.98, + "learning_rate": 2.716237405747113e-08, + "loss": 0.6142, + "step": 2448 + }, + { + "epoch": 0.98, + "learning_rate": 2.6218081553638363e-08, + "loss": 0.6367, + "step": 2449 + }, + { + "epoch": 0.98, + "learning_rate": 2.5290473216610378e-08, + "loss": 0.6343, + "step": 2450 + }, + { + "epoch": 0.98, + "learning_rate": 2.4379550598092604e-08, + "loss": 0.6494, + "step": 2451 + }, + { + "epoch": 0.98, + "learning_rate": 2.3485315221877246e-08, + "loss": 0.6759, + "step": 2452 + }, + { + "epoch": 0.98, + "learning_rate": 2.2607768583843282e-08, + "loss": 0.6137, + "step": 2453 + }, + { + "epoch": 0.98, + "learning_rate": 2.1746912151955346e-08, + "loss": 0.7235, + "step": 2454 + }, + { + "epoch": 0.98, + "learning_rate": 2.090274736625486e-08, + "loss": 0.6912, + "step": 2455 + }, + { + "epoch": 0.98, + "learning_rate": 2.0075275638862247e-08, + "loss": 0.6558, + "step": 2456 + }, + { + "epoch": 0.98, + "learning_rate": 1.9264498353974704e-08, + "loss": 0.6461, + "step": 2457 + }, + { + "epoch": 0.98, + "learning_rate": 1.8470416867861775e-08, + "loss": 0.6604, + "step": 2458 + }, + { + "epoch": 0.98, + "learning_rate": 1.769303250886534e-08, + "loss": 0.6274, + "step": 2459 + }, + { + "epoch": 0.98, + "learning_rate": 1.6932346577397396e-08, + "loss": 0.6836, + "step": 2460 + }, + { + "epoch": 0.98, + "learning_rate": 1.6188360345932297e-08, + "loss": 0.6153, + "step": 2461 + }, + { + "epoch": 0.98, + "learning_rate": 1.546107505901451e-08, + "loss": 0.6751, + "step": 2462 + }, + { + "epoch": 0.98, + "learning_rate": 1.4750491933247513e-08, + "loss": 0.7112, + "step": 2463 + }, + { + "epoch": 0.98, + "learning_rate": 1.4056612157297145e-08, + "loss": 0.6779, + "step": 2464 + }, + { + "epoch": 0.98, + "learning_rate": 1.3379436891886034e-08, + "loss": 0.6113, + "step": 2465 + }, + { + "epoch": 0.98, + "learning_rate": 1.2718967269795823e-08, + "loss": 0.5939, + "step": 2466 + }, + { + "epoch": 0.98, + "learning_rate": 1.2075204395860518e-08, + "loss": 0.7174, + "step": 2467 + }, + { + "epoch": 0.99, + "learning_rate": 1.1448149346969806e-08, + "loss": 0.6953, + "step": 2468 + }, + { + "epoch": 0.99, + "learning_rate": 1.083780317206129e-08, + "loss": 0.6655, + "step": 2469 + }, + { + "epoch": 0.99, + "learning_rate": 1.0244166892124929e-08, + "loss": 0.6732, + "step": 2470 + }, + { + "epoch": 0.99, + "learning_rate": 9.667241500196378e-09, + "loss": 0.6759, + "step": 2471 + }, + { + "epoch": 0.99, + "learning_rate": 9.107027961356985e-09, + "loss": 0.6691, + "step": 2472 + }, + { + "epoch": 0.99, + "learning_rate": 8.563527212734902e-09, + "loss": 0.6781, + "step": 2473 + }, + { + "epoch": 0.99, + "learning_rate": 8.036740163498425e-09, + "loss": 0.6768, + "step": 2474 + }, + { + "epoch": 0.99, + "learning_rate": 7.526667694858214e-09, + "loss": 0.6693, + "step": 2475 + }, + { + "epoch": 0.99, + "learning_rate": 7.033310660065074e-09, + "loss": 0.6575, + "step": 2476 + }, + { + "epoch": 0.99, + "learning_rate": 6.556669884408839e-09, + "loss": 0.5874, + "step": 2477 + }, + { + "epoch": 0.99, + "learning_rate": 6.09674616521283e-09, + "loss": 0.6953, + "step": 2478 + }, + { + "epoch": 0.99, + "learning_rate": 5.653540271841618e-09, + "loss": 0.6688, + "step": 2479 + }, + { + "epoch": 0.99, + "learning_rate": 5.2270529456888155e-09, + "loss": 0.7315, + "step": 2480 + }, + { + "epoch": 0.99, + "learning_rate": 4.817284900183738e-09, + "loss": 0.6875, + "step": 2481 + }, + { + "epoch": 0.99, + "learning_rate": 4.424236820789185e-09, + "loss": 0.6984, + "step": 2482 + }, + { + "epoch": 0.99, + "learning_rate": 4.047909364994773e-09, + "loss": 0.7342, + "step": 2483 + }, + { + "epoch": 0.99, + "learning_rate": 3.688303162322493e-09, + "loss": 0.7396, + "step": 2484 + }, + { + "epoch": 0.99, + "learning_rate": 3.3454188143233758e-09, + "loss": 0.6467, + "step": 2485 + }, + { + "epoch": 0.99, + "learning_rate": 3.019256894575273e-09, + "loss": 0.6484, + "step": 2486 + }, + { + "epoch": 0.99, + "learning_rate": 2.7098179486806376e-09, + "loss": 0.6755, + "step": 2487 + }, + { + "epoch": 0.99, + "learning_rate": 2.4171024942720723e-09, + "loss": 0.652, + "step": 2488 + }, + { + "epoch": 0.99, + "learning_rate": 2.14111102100345e-09, + "loss": 0.621, + "step": 2489 + }, + { + "epoch": 0.99, + "learning_rate": 1.881843990554355e-09, + "loss": 0.7173, + "step": 2490 + }, + { + "epoch": 0.99, + "learning_rate": 1.6393018366278601e-09, + "loss": 0.6266, + "step": 2491 + }, + { + "epoch": 0.99, + "learning_rate": 1.4134849649471982e-09, + "loss": 0.7034, + "step": 2492 + }, + { + "epoch": 1.0, + "learning_rate": 1.2043937532613126e-09, + "loss": 0.68, + "step": 2493 + }, + { + "epoch": 1.0, + "learning_rate": 1.0120285513381957e-09, + "loss": 0.6299, + "step": 2494 + }, + { + "epoch": 1.0, + "learning_rate": 8.363896809659989e-10, + "loss": 0.7378, + "step": 2495 + }, + { + "epoch": 1.0, + "learning_rate": 6.774774359541436e-10, + "loss": 0.6409, + "step": 2496 + }, + { + "epoch": 1.0, + "learning_rate": 5.352920821311002e-10, + "loss": 0.6633, + "step": 2497 + }, + { + "epoch": 1.0, + "learning_rate": 4.0983385734660875e-10, + "loss": 0.6526, + "step": 2498 + }, + { + "epoch": 1.0, + "learning_rate": 3.011029714650171e-10, + "loss": 0.6352, + "step": 2499 + }, + { + "epoch": 1.0, + "learning_rate": 2.090996063741635e-10, + "loss": 0.6635, + "step": 2500 + }, + { + "epoch": 1.0, + "learning_rate": 1.3382391597538403e-10, + "loss": 0.5883, + "step": 2501 + }, + { + "epoch": 1.0, + "learning_rate": 7.527602619017415e-11, + "loss": 0.6822, + "step": 2502 + }, + { + "epoch": 1.0, + "learning_rate": 3.3456034959078456e-11, + "loss": 0.5954, + "step": 2503 + }, + { + "epoch": 1.0, + "learning_rate": 8.364012237249698e-12, + "loss": 0.5953, + "step": 2504 + }, + { + "epoch": 1.0, + "learning_rate": 0.0, + "loss": 0.6442, + "step": 2505 + }, + { + "epoch": 1.0, + "step": 2505, + "total_flos": 452205235666944.0, + "train_loss": 0.7111096807820593, + "train_runtime": 10580.2761, + "train_samples_per_second": 30.299, + "train_steps_per_second": 0.237 + } + ], + "logging_steps": 1.0, + "max_steps": 2505, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 50000, + "total_flos": 452205235666944.0, + "train_batch_size": 16, + "trial_name": null, + "trial_params": null +}