diff --git "a/checkpoint-9900/trainer_state.json" "b/checkpoint-9900/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-9900/trainer_state.json" @@ -0,0 +1,60406 @@ +{ + "best_metric": 16.172945052397182, + "best_model_checkpoint": "wmt19-ende-t5-small/checkpoint-9900", + "epoch": 0.99, + "global_step": 9900, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.4663, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.7682, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.4259, + "step": 3 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.4291, + "step": 4 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.3851, + "step": 5 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.5746, + "step": 6 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.955, + "step": 7 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.4584, + "step": 8 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.1171, + "step": 9 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.2373, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.1546, + "step": 11 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.0485, + "step": 12 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.0734, + "step": 13 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9997, + "step": 14 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.0071, + "step": 15 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.7614, + "step": 16 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.0746, + "step": 17 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.7083, + "step": 18 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9204, + "step": 19 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.1216, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9752, + "step": 21 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7399, + "step": 22 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.855, + "step": 23 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9157, + "step": 24 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.6479, + "step": 25 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.0461, + "step": 26 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7031, + "step": 27 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8675, + "step": 28 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7588, + "step": 29 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8662, + "step": 30 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.6512, + "step": 31 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9068, + "step": 32 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7663, + "step": 33 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8204, + "step": 34 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.1299, + "step": 35 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7475, + "step": 36 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7006, + "step": 37 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.7249, + "step": 38 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.91, + "step": 39 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.6321, + "step": 40 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.934, + "step": 41 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8877, + "step": 42 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9024, + "step": 43 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.9581, + "step": 44 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8292, + "step": 45 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 3.3768, + "step": 46 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8814, + "step": 47 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8935, + "step": 48 + }, + { + "epoch": 0.0, + "learning_rate": 0.0001, + "loss": 2.8217, + "step": 49 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.519, + "step": 50 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8367, + "step": 51 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.0951, + "step": 52 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8405, + "step": 53 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7408, + "step": 54 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6207, + "step": 55 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.548, + "step": 56 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6639, + "step": 57 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9578, + "step": 58 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7555, + "step": 59 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7869, + "step": 60 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7008, + "step": 61 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7733, + "step": 62 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8, + "step": 63 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8631, + "step": 64 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7782, + "step": 65 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7401, + "step": 66 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5941, + "step": 67 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7415, + "step": 68 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6899, + "step": 69 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9748, + "step": 70 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7563, + "step": 71 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7991, + "step": 72 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6486, + "step": 73 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7161, + "step": 74 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7027, + "step": 75 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7595, + "step": 76 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6428, + "step": 77 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.6012, + "step": 78 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7219, + "step": 79 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.5364, + "step": 80 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5702, + "step": 81 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9898, + "step": 82 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6664, + "step": 83 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6907, + "step": 84 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7501, + "step": 85 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6913, + "step": 86 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6477, + "step": 87 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7592, + "step": 88 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7689, + "step": 89 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.729, + "step": 90 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9724, + "step": 91 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7313, + "step": 92 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8927, + "step": 93 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6544, + "step": 94 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9762, + "step": 95 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.716, + "step": 96 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8332, + "step": 97 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7202, + "step": 98 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.789, + "step": 99 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7369, + "step": 100 + }, + { + "epoch": 0.01, + "eval_bleu": 9.085051126443968, + "eval_brevity_penalty": 0.5106766364801893, + "eval_loss": 2.0018303394317627, + "eval_runtime": 5.8258, + "eval_samples_per_second": 514.61, + "eval_steps_per_second": 1.03, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6762, + "step": 101 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7293, + "step": 102 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.4312, + "step": 103 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.3982, + "step": 104 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6178, + "step": 105 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.3215, + "step": 106 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6384, + "step": 107 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.9455, + "step": 108 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6396, + "step": 109 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8566, + "step": 110 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7088, + "step": 111 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8343, + "step": 112 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6486, + "step": 113 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7356, + "step": 114 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6597, + "step": 115 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8338, + "step": 116 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5511, + "step": 117 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6564, + "step": 118 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.4826, + "step": 119 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.445, + "step": 120 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5586, + "step": 121 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.8519, + "step": 122 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5281, + "step": 123 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6049, + "step": 124 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6373, + "step": 125 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7958, + "step": 126 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.639, + "step": 127 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5067, + "step": 128 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6237, + "step": 129 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5472, + "step": 130 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5322, + "step": 131 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6194, + "step": 132 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6129, + "step": 133 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5784, + "step": 134 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.4133, + "step": 135 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6157, + "step": 136 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.4615, + "step": 137 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5247, + "step": 138 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.4891, + "step": 139 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.3907, + "step": 140 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.6543, + "step": 141 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.5707, + "step": 142 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.835, + "step": 143 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 3.0679, + "step": 144 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7197, + "step": 145 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7853, + "step": 146 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.7265, + "step": 147 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5039, + "step": 148 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5992, + "step": 149 + }, + { + "epoch": 0.01, + "learning_rate": 0.0001, + "loss": 2.5673, + "step": 150 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3885, + "step": 151 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5934, + "step": 152 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4589, + "step": 153 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6356, + "step": 154 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.1734, + "step": 155 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5525, + "step": 156 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4886, + "step": 157 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.7129, + "step": 158 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5616, + "step": 159 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6208, + "step": 160 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4783, + "step": 161 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6107, + "step": 162 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5652, + "step": 163 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5955, + "step": 164 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5859, + "step": 165 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.606, + "step": 166 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3434, + "step": 167 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.1438, + "step": 168 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5197, + "step": 169 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.0064, + "step": 170 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3667, + "step": 171 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.8341, + "step": 172 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4456, + "step": 173 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6078, + "step": 174 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5203, + "step": 175 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5444, + "step": 176 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4292, + "step": 177 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6453, + "step": 178 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5085, + "step": 179 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.659, + "step": 180 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5371, + "step": 181 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.8879, + "step": 182 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4631, + "step": 183 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.0762, + "step": 184 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3898, + "step": 185 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.0154, + "step": 186 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4244, + "step": 187 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5178, + "step": 188 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.2668, + "step": 189 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3956, + "step": 190 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.572, + "step": 191 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.7143, + "step": 192 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.2775, + "step": 193 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3652, + "step": 194 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5624, + "step": 195 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5946, + "step": 196 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5783, + "step": 197 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.4453, + "step": 198 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.58, + "step": 199 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.3896, + "step": 200 + }, + { + "epoch": 0.02, + "eval_bleu": 9.997024948761176, + "eval_brevity_penalty": 0.5126926107121719, + "eval_loss": 1.9108339548110962, + "eval_runtime": 6.1602, + "eval_samples_per_second": 486.671, + "eval_steps_per_second": 0.974, + "step": 200 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3912, + "step": 201 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.4515, + "step": 202 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4903, + "step": 203 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.8211, + "step": 204 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3584, + "step": 205 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.9959, + "step": 206 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.365, + "step": 207 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.9612, + "step": 208 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4021, + "step": 209 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4684, + "step": 210 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3937, + "step": 211 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6033, + "step": 212 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4816, + "step": 213 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6983, + "step": 214 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6607, + "step": 215 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.587, + "step": 216 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.542, + "step": 217 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5979, + "step": 218 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5306, + "step": 219 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6165, + "step": 220 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4665, + "step": 221 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6255, + "step": 222 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5402, + "step": 223 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6164, + "step": 224 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4148, + "step": 225 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5271, + "step": 226 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4922, + "step": 227 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6202, + "step": 228 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4151, + "step": 229 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6399, + "step": 230 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5631, + "step": 231 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6184, + "step": 232 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6411, + "step": 233 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.2796, + "step": 234 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5084, + "step": 235 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.0203, + "step": 236 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.4968, + "step": 237 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 3.2446, + "step": 238 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5332, + "step": 239 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5041, + "step": 240 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.5186, + "step": 241 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.8515, + "step": 242 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3801, + "step": 243 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.567, + "step": 244 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.544, + "step": 245 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.7933, + "step": 246 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.6131, + "step": 247 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.7054, + "step": 248 + }, + { + "epoch": 0.02, + "learning_rate": 0.0001, + "loss": 2.3467, + "step": 249 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3588, + "step": 250 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.314, + "step": 251 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4462, + "step": 252 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.511, + "step": 253 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7577, + "step": 254 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4974, + "step": 255 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4135, + "step": 256 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5068, + "step": 257 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7906, + "step": 258 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4432, + "step": 259 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7876, + "step": 260 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3252, + "step": 261 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6486, + "step": 262 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4302, + "step": 263 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5628, + "step": 264 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3921, + "step": 265 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.69, + "step": 266 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4154, + "step": 267 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.3579, + "step": 268 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6235, + "step": 269 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.4851, + "step": 270 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4679, + "step": 271 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.0836, + "step": 272 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3639, + "step": 273 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7544, + "step": 274 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4051, + "step": 275 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3073, + "step": 276 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3895, + "step": 277 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4208, + "step": 278 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5261, + "step": 279 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4003, + "step": 280 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.367, + "step": 281 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6916, + "step": 282 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5555, + "step": 283 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5824, + "step": 284 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4086, + "step": 285 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5559, + "step": 286 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.2112, + "step": 287 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5318, + "step": 288 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4514, + "step": 289 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4761, + "step": 290 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4002, + "step": 291 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5901, + "step": 292 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3735, + "step": 293 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5448, + "step": 294 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4546, + "step": 295 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6278, + "step": 296 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5367, + "step": 297 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.694, + "step": 298 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5222, + "step": 299 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.0442, + "step": 300 + }, + { + "epoch": 0.03, + "eval_bleu": 10.766995828901416, + "eval_brevity_penalty": 0.5244988538375123, + "eval_loss": 1.8626734018325806, + "eval_runtime": 8.0526, + "eval_samples_per_second": 372.3, + "eval_steps_per_second": 0.745, + "step": 300 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4532, + "step": 301 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.9992, + "step": 302 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6407, + "step": 303 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6858, + "step": 304 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5446, + "step": 305 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.8742, + "step": 306 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4087, + "step": 307 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4071, + "step": 308 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4703, + "step": 309 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.2562, + "step": 310 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4937, + "step": 311 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7677, + "step": 312 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4141, + "step": 313 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6928, + "step": 314 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4694, + "step": 315 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6883, + "step": 316 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4326, + "step": 317 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.9255, + "step": 318 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4625, + "step": 319 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.1847, + "step": 320 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4065, + "step": 321 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7093, + "step": 322 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3824, + "step": 323 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5536, + "step": 324 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4423, + "step": 325 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.7707, + "step": 326 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4109, + "step": 327 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.9737, + "step": 328 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3435, + "step": 329 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.8754, + "step": 330 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3455, + "step": 331 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.2262, + "step": 332 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4853, + "step": 333 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5102, + "step": 334 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.6098, + "step": 335 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4253, + "step": 336 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5801, + "step": 337 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4686, + "step": 338 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4407, + "step": 339 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4275, + "step": 340 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3797, + "step": 341 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.2751, + "step": 342 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.3743, + "step": 343 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.0679, + "step": 344 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5376, + "step": 345 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 3.2261, + "step": 346 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4679, + "step": 347 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.5374, + "step": 348 + }, + { + "epoch": 0.03, + "learning_rate": 0.0001, + "loss": 2.4754, + "step": 349 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2934, + "step": 350 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5705, + "step": 351 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4066, + "step": 352 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4326, + "step": 353 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4327, + "step": 354 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5388, + "step": 355 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4361, + "step": 356 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4676, + "step": 357 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5944, + "step": 358 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4739, + "step": 359 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5055, + "step": 360 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3478, + "step": 361 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4458, + "step": 362 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4285, + "step": 363 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4923, + "step": 364 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2729, + "step": 365 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4965, + "step": 366 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5523, + "step": 367 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5488, + "step": 368 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3982, + "step": 369 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5764, + "step": 370 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3626, + "step": 371 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3352, + "step": 372 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4119, + "step": 373 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4271, + "step": 374 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5, + "step": 375 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5336, + "step": 376 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4236, + "step": 377 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 3.0173, + "step": 378 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4699, + "step": 379 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.9103, + "step": 380 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4736, + "step": 381 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6213, + "step": 382 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4647, + "step": 383 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.7151, + "step": 384 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3607, + "step": 385 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.7621, + "step": 386 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4329, + "step": 387 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.8935, + "step": 388 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3115, + "step": 389 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.9991, + "step": 390 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2966, + "step": 391 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6163, + "step": 392 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4309, + "step": 393 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3882, + "step": 394 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4484, + "step": 395 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4868, + "step": 396 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3369, + "step": 397 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5925, + "step": 398 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3441, + "step": 399 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5136, + "step": 400 + }, + { + "epoch": 0.04, + "eval_bleu": 10.928010986851017, + "eval_brevity_penalty": 0.5132499429254788, + "eval_loss": 1.8244277238845825, + "eval_runtime": 5.7343, + "eval_samples_per_second": 522.823, + "eval_steps_per_second": 1.046, + "step": 400 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2932, + "step": 401 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6948, + "step": 402 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2305, + "step": 403 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6933, + "step": 404 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3046, + "step": 405 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.289, + "step": 406 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3428, + "step": 407 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.9014, + "step": 408 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.1936, + "step": 409 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 3.401, + "step": 410 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.401, + "step": 411 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 3.3396, + "step": 412 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4271, + "step": 413 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 3.1966, + "step": 414 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4629, + "step": 415 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6279, + "step": 416 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4949, + "step": 417 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3936, + "step": 418 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.415, + "step": 419 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4102, + "step": 420 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2532, + "step": 421 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2858, + "step": 422 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.1706, + "step": 423 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.6436, + "step": 424 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2926, + "step": 425 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3625, + "step": 426 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2981, + "step": 427 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4862, + "step": 428 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4193, + "step": 429 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4825, + "step": 430 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3121, + "step": 431 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3695, + "step": 432 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2626, + "step": 433 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3344, + "step": 434 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2512, + "step": 435 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5091, + "step": 436 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2487, + "step": 437 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5058, + "step": 438 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2803, + "step": 439 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4901, + "step": 440 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4487, + "step": 441 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4894, + "step": 442 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2707, + "step": 443 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.8379, + "step": 444 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.2862, + "step": 445 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.5145, + "step": 446 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4201, + "step": 447 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.4094, + "step": 448 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3462, + "step": 449 + }, + { + "epoch": 0.04, + "learning_rate": 0.0001, + "loss": 2.3412, + "step": 450 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2349, + "step": 451 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.7062, + "step": 452 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4598, + "step": 453 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.8648, + "step": 454 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3628, + "step": 455 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3817, + "step": 456 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3272, + "step": 457 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6671, + "step": 458 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5271, + "step": 459 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.7152, + "step": 460 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4288, + "step": 461 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6505, + "step": 462 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2696, + "step": 463 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3591, + "step": 464 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4899, + "step": 465 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6766, + "step": 466 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5604, + "step": 467 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.509, + "step": 468 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5654, + "step": 469 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5599, + "step": 470 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3975, + "step": 471 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3798, + "step": 472 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2135, + "step": 473 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.8556, + "step": 474 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3647, + "step": 475 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.2631, + "step": 476 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5691, + "step": 477 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.4263, + "step": 478 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4425, + "step": 479 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6265, + "step": 480 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4327, + "step": 481 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3837, + "step": 482 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5345, + "step": 483 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3607, + "step": 484 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4768, + "step": 485 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4113, + "step": 486 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4618, + "step": 487 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3597, + "step": 488 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2651, + "step": 489 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5689, + "step": 490 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4256, + "step": 491 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5094, + "step": 492 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.407, + "step": 493 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4528, + "step": 494 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4253, + "step": 495 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3766, + "step": 496 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2867, + "step": 497 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3825, + "step": 498 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3579, + "step": 499 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4092, + "step": 500 + }, + { + "epoch": 0.05, + "eval_bleu": 11.471695390862497, + "eval_brevity_penalty": 0.5260133269808146, + "eval_loss": 1.7950547933578491, + "eval_runtime": 7.2316, + "eval_samples_per_second": 414.568, + "eval_steps_per_second": 0.83, + "step": 500 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2321, + "step": 501 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4504, + "step": 502 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.1692, + "step": 503 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.2322, + "step": 504 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3889, + "step": 505 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4397, + "step": 506 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3427, + "step": 507 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3831, + "step": 508 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2671, + "step": 509 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4615, + "step": 510 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4967, + "step": 511 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6878, + "step": 512 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3029, + "step": 513 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.617, + "step": 514 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4312, + "step": 515 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6179, + "step": 516 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3305, + "step": 517 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4688, + "step": 518 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2508, + "step": 519 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4681, + "step": 520 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2915, + "step": 521 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3168, + "step": 522 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3233, + "step": 523 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5918, + "step": 524 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3239, + "step": 525 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.6262, + "step": 526 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3211, + "step": 527 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.7037, + "step": 528 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.38, + "step": 529 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.5135, + "step": 530 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2328, + "step": 531 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4934, + "step": 532 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2141, + "step": 533 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4871, + "step": 534 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.0858, + "step": 535 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.3509, + "step": 536 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.28, + "step": 537 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.7591, + "step": 538 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2199, + "step": 539 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.4031, + "step": 540 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.1467, + "step": 541 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.3178, + "step": 542 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2936, + "step": 543 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.2252, + "step": 544 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.2393, + "step": 545 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 3.292, + "step": 546 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.101, + "step": 547 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.4677, + "step": 548 + }, + { + "epoch": 0.05, + "learning_rate": 0.0001, + "loss": 2.311, + "step": 549 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4436, + "step": 550 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3541, + "step": 551 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3582, + "step": 552 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4034, + "step": 553 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5948, + "step": 554 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4826, + "step": 555 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5553, + "step": 556 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3413, + "step": 557 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6206, + "step": 558 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4258, + "step": 559 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4163, + "step": 560 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5366, + "step": 561 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3312, + "step": 562 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2443, + "step": 563 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4532, + "step": 564 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3679, + "step": 565 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5423, + "step": 566 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3443, + "step": 567 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6959, + "step": 568 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3189, + "step": 569 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5793, + "step": 570 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2787, + "step": 571 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6453, + "step": 572 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2129, + "step": 573 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6325, + "step": 574 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4581, + "step": 575 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3458, + "step": 576 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2431, + "step": 577 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.369, + "step": 578 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2254, + "step": 579 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.631, + "step": 580 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3662, + "step": 581 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.7582, + "step": 582 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4281, + "step": 583 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5432, + "step": 584 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.327, + "step": 585 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4314, + "step": 586 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.304, + "step": 587 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3737, + "step": 588 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4594, + "step": 589 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.7682, + "step": 590 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4153, + "step": 591 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3969, + "step": 592 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3381, + "step": 593 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3482, + "step": 594 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2937, + "step": 595 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4222, + "step": 596 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3173, + "step": 597 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 3.4151, + "step": 598 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2249, + "step": 599 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 3.2441, + "step": 600 + }, + { + "epoch": 0.06, + "eval_bleu": 11.734959027124207, + "eval_brevity_penalty": 0.5196932620946793, + "eval_loss": 1.7736010551452637, + "eval_runtime": 5.6339, + "eval_samples_per_second": 532.135, + "eval_steps_per_second": 1.065, + "step": 600 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.1283, + "step": 601 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 3.2063, + "step": 602 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.1738, + "step": 603 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5168, + "step": 604 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2599, + "step": 605 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2913, + "step": 606 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3451, + "step": 607 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3944, + "step": 608 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2937, + "step": 609 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4604, + "step": 610 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3674, + "step": 611 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4279, + "step": 612 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.1968, + "step": 613 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.403, + "step": 614 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2169, + "step": 615 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5313, + "step": 616 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.256, + "step": 617 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4357, + "step": 618 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.319, + "step": 619 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5224, + "step": 620 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2574, + "step": 621 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2488, + "step": 622 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3314, + "step": 623 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.8449, + "step": 624 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3791, + "step": 625 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3086, + "step": 626 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4106, + "step": 627 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6144, + "step": 628 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3916, + "step": 629 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.6309, + "step": 630 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.21, + "step": 631 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.303, + "step": 632 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2686, + "step": 633 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 1.9969, + "step": 634 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3783, + "step": 635 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.8066, + "step": 636 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3805, + "step": 637 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.5832, + "step": 638 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3221, + "step": 639 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4873, + "step": 640 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2963, + "step": 641 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.0642, + "step": 642 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3271, + "step": 643 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.1974, + "step": 644 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3399, + "step": 645 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.1965, + "step": 646 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.4226, + "step": 647 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.3666, + "step": 648 + }, + { + "epoch": 0.06, + "learning_rate": 0.0001, + "loss": 2.2443, + "step": 649 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4556, + "step": 650 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3391, + "step": 651 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 3.3726, + "step": 652 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.361, + "step": 653 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 3.3049, + "step": 654 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3861, + "step": 655 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.9239, + "step": 656 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.394, + "step": 657 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4177, + "step": 658 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3086, + "step": 659 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3633, + "step": 660 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3531, + "step": 661 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3546, + "step": 662 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2535, + "step": 663 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2137, + "step": 664 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4269, + "step": 665 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3699, + "step": 666 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.394, + "step": 667 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5592, + "step": 668 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4337, + "step": 669 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.6395, + "step": 670 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3903, + "step": 671 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3839, + "step": 672 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3903, + "step": 673 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3151, + "step": 674 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3263, + "step": 675 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.419, + "step": 676 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2558, + "step": 677 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3209, + "step": 678 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2768, + "step": 679 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3882, + "step": 680 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3584, + "step": 681 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2518, + "step": 682 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3491, + "step": 683 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.6933, + "step": 684 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3183, + "step": 685 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5052, + "step": 686 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5502, + "step": 687 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.1001, + "step": 688 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4334, + "step": 689 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5927, + "step": 690 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3407, + "step": 691 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.7748, + "step": 692 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3451, + "step": 693 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2409, + "step": 694 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3831, + "step": 695 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5929, + "step": 696 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3744, + "step": 697 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5647, + "step": 698 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4407, + "step": 699 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.6997, + "step": 700 + }, + { + "epoch": 0.07, + "eval_bleu": 12.074149623991028, + "eval_brevity_penalty": 0.526034650826011, + "eval_loss": 1.7563488483428955, + "eval_runtime": 6.3114, + "eval_samples_per_second": 475.014, + "eval_steps_per_second": 0.951, + "step": 700 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3885, + "step": 701 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2764, + "step": 702 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2856, + "step": 703 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5727, + "step": 704 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3034, + "step": 705 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.7152, + "step": 706 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.257, + "step": 707 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.1402, + "step": 708 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3122, + "step": 709 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2871, + "step": 710 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2722, + "step": 711 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 3.201, + "step": 712 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3085, + "step": 713 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 3.246, + "step": 714 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3009, + "step": 715 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.8952, + "step": 716 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3892, + "step": 717 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.321, + "step": 718 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4976, + "step": 719 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4594, + "step": 720 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2823, + "step": 721 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4433, + "step": 722 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5014, + "step": 723 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3572, + "step": 724 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5388, + "step": 725 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4147, + "step": 726 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4227, + "step": 727 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5094, + "step": 728 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2353, + "step": 729 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4299, + "step": 730 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2465, + "step": 731 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.46, + "step": 732 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3618, + "step": 733 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5224, + "step": 734 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.268, + "step": 735 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5082, + "step": 736 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3626, + "step": 737 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.4607, + "step": 738 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2886, + "step": 739 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.6387, + "step": 740 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3522, + "step": 741 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5154, + "step": 742 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.2623, + "step": 743 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.8481, + "step": 744 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.286, + "step": 745 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.3488, + "step": 746 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 1.9961, + "step": 747 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.127, + "step": 748 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.1968, + "step": 749 + }, + { + "epoch": 0.07, + "learning_rate": 0.0001, + "loss": 2.5122, + "step": 750 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3379, + "step": 751 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.277, + "step": 752 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2914, + "step": 753 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.411, + "step": 754 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2123, + "step": 755 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.6667, + "step": 756 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1862, + "step": 757 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.9795, + "step": 758 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2557, + "step": 759 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 3.103, + "step": 760 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1032, + "step": 761 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3585, + "step": 762 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1021, + "step": 763 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.9251, + "step": 764 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1691, + "step": 765 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3353, + "step": 766 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2232, + "step": 767 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4962, + "step": 768 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2962, + "step": 769 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2977, + "step": 770 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3598, + "step": 771 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.492, + "step": 772 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3684, + "step": 773 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.8352, + "step": 774 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2436, + "step": 775 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 3.2166, + "step": 776 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3281, + "step": 777 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.7624, + "step": 778 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2991, + "step": 779 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.475, + "step": 780 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3507, + "step": 781 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3869, + "step": 782 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.0869, + "step": 783 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4886, + "step": 784 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2923, + "step": 785 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3819, + "step": 786 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1227, + "step": 787 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4993, + "step": 788 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2373, + "step": 789 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.5223, + "step": 790 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.0689, + "step": 791 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4463, + "step": 792 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1329, + "step": 793 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3417, + "step": 794 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2323, + "step": 795 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4385, + "step": 796 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.1994, + "step": 797 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3922, + "step": 798 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4752, + "step": 799 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.5072, + "step": 800 + }, + { + "epoch": 0.08, + "eval_bleu": 12.373490547500005, + "eval_brevity_penalty": 0.5282726042804718, + "eval_loss": 1.7415881156921387, + "eval_runtime": 5.4443, + "eval_samples_per_second": 550.665, + "eval_steps_per_second": 1.102, + "step": 800 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2457, + "step": 801 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.6283, + "step": 802 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2629, + "step": 803 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.8649, + "step": 804 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2016, + "step": 805 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.6424, + "step": 806 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3186, + "step": 807 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2084, + "step": 808 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3998, + "step": 809 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4413, + "step": 810 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4392, + "step": 811 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4743, + "step": 812 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4667, + "step": 813 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3184, + "step": 814 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3833, + "step": 815 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.666, + "step": 816 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.5182, + "step": 817 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3351, + "step": 818 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4401, + "step": 819 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3113, + "step": 820 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.414, + "step": 821 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.6171, + "step": 822 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3828, + "step": 823 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2569, + "step": 824 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3825, + "step": 825 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3934, + "step": 826 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3211, + "step": 827 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2336, + "step": 828 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3015, + "step": 829 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.5692, + "step": 830 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3322, + "step": 831 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 3.2635, + "step": 832 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3125, + "step": 833 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 3.2536, + "step": 834 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2505, + "step": 835 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.8501, + "step": 836 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3141, + "step": 837 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3771, + "step": 838 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3698, + "step": 839 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3452, + "step": 840 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4531, + "step": 841 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.2631, + "step": 842 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3715, + "step": 843 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.4499, + "step": 844 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3454, + "step": 845 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.5086, + "step": 846 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.539, + "step": 847 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.3374, + "step": 848 + }, + { + "epoch": 0.08, + "learning_rate": 0.0001, + "loss": 2.418, + "step": 849 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3137, + "step": 850 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3909, + "step": 851 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2881, + "step": 852 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.0718, + "step": 853 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3164, + "step": 854 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3413, + "step": 855 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4851, + "step": 856 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4389, + "step": 857 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.535, + "step": 858 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3088, + "step": 859 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3439, + "step": 860 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1591, + "step": 861 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.528, + "step": 862 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1607, + "step": 863 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3617, + "step": 864 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3919, + "step": 865 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2358, + "step": 866 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3704, + "step": 867 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5127, + "step": 868 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3698, + "step": 869 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5025, + "step": 870 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4118, + "step": 871 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4799, + "step": 872 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2244, + "step": 873 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4621, + "step": 874 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2368, + "step": 875 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.49, + "step": 876 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3867, + "step": 877 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2712, + "step": 878 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.343, + "step": 879 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3174, + "step": 880 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2267, + "step": 881 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5207, + "step": 882 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4593, + "step": 883 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5235, + "step": 884 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5077, + "step": 885 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 3.2918, + "step": 886 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5418, + "step": 887 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.924, + "step": 888 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.443, + "step": 889 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3784, + "step": 890 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2454, + "step": 891 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3773, + "step": 892 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3788, + "step": 893 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4932, + "step": 894 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4986, + "step": 895 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3933, + "step": 896 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3425, + "step": 897 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4706, + "step": 898 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1267, + "step": 899 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3788, + "step": 900 + }, + { + "epoch": 0.09, + "eval_bleu": 12.428775568298708, + "eval_brevity_penalty": 0.5284643317499238, + "eval_loss": 1.7267242670059204, + "eval_runtime": 5.4201, + "eval_samples_per_second": 553.13, + "eval_steps_per_second": 1.107, + "step": 900 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.0748, + "step": 901 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4139, + "step": 902 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1001, + "step": 903 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3206, + "step": 904 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.221, + "step": 905 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4003, + "step": 906 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1422, + "step": 907 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3238, + "step": 908 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1934, + "step": 909 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3679, + "step": 910 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.328, + "step": 911 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.6308, + "step": 912 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5124, + "step": 913 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5971, + "step": 914 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.2208, + "step": 915 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5342, + "step": 916 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.9647, + "step": 917 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4809, + "step": 918 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.9566, + "step": 919 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.4621, + "step": 920 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.3306, + "step": 921 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5101, + "step": 922 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.804, + "step": 923 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.9224, + "step": 924 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.1176, + "step": 925 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.6711, + "step": 926 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8983, + "step": 927 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5858, + "step": 928 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.9946, + "step": 929 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.6, + "step": 930 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8521, + "step": 931 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5809, + "step": 932 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8507, + "step": 933 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.6611, + "step": 934 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.7742, + "step": 935 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5592, + "step": 936 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8739, + "step": 937 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.5587, + "step": 938 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8838, + "step": 939 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.273, + "step": 940 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.717, + "step": 941 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.133, + "step": 942 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.7449, + "step": 943 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.0413, + "step": 944 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 2.0327, + "step": 945 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 3.0239, + "step": 946 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.7105, + "step": 947 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 3.3089, + "step": 948 + }, + { + "epoch": 0.09, + "learning_rate": 0.0001, + "loss": 1.8148, + "step": 949 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 3.2378, + "step": 950 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6169, + "step": 951 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5341, + "step": 952 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6848, + "step": 953 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5124, + "step": 954 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.823, + "step": 955 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3715, + "step": 956 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7127, + "step": 957 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2094, + "step": 958 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7721, + "step": 959 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2762, + "step": 960 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7777, + "step": 961 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2502, + "step": 962 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6235, + "step": 963 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4264, + "step": 964 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.8469, + "step": 965 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3137, + "step": 966 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7111, + "step": 967 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4434, + "step": 968 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.816, + "step": 969 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.261, + "step": 970 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.1209, + "step": 971 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4882, + "step": 972 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.4565, + "step": 973 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3542, + "step": 974 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6781, + "step": 975 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5831, + "step": 976 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.8083, + "step": 977 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 3.2242, + "step": 978 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6847, + "step": 979 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4984, + "step": 980 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.5997, + "step": 981 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.6068, + "step": 982 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7302, + "step": 983 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.7498, + "step": 984 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.5255, + "step": 985 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.6036, + "step": 986 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.9716, + "step": 987 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.8907, + "step": 988 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6126, + "step": 989 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3794, + "step": 990 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7647, + "step": 991 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3067, + "step": 992 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.8002, + "step": 993 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3896, + "step": 994 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6943, + "step": 995 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3068, + "step": 996 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.8645, + "step": 997 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4318, + "step": 998 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.4698, + "step": 999 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3533, + "step": 1000 + }, + { + "epoch": 0.1, + "eval_bleu": 12.439455304549925, + "eval_brevity_penalty": 0.5249468950825891, + "eval_loss": 1.724652647972107, + "eval_runtime": 5.4769, + "eval_samples_per_second": 547.391, + "eval_steps_per_second": 1.096, + "step": 1000 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.9195, + "step": 1001 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.476, + "step": 1002 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.1259, + "step": 1003 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 3.166, + "step": 1004 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0652, + "step": 1005 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 3.3205, + "step": 1006 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0085, + "step": 1007 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2778, + "step": 1008 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.9196, + "step": 1009 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0339, + "step": 1010 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7178, + "step": 1011 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.1175, + "step": 1012 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.6231, + "step": 1013 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.1926, + "step": 1014 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.9903, + "step": 1015 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2595, + "step": 1016 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2885, + "step": 1017 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2902, + "step": 1018 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0412, + "step": 1019 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4043, + "step": 1020 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.111, + "step": 1021 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2887, + "step": 1022 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0819, + "step": 1023 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2084, + "step": 1024 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0636, + "step": 1025 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3082, + "step": 1026 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.8118, + "step": 1027 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3219, + "step": 1028 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.7898, + "step": 1029 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3674, + "step": 1030 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2353, + "step": 1031 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4563, + "step": 1032 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5813, + "step": 1033 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.6558, + "step": 1034 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4922, + "step": 1035 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.6764, + "step": 1036 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5826, + "step": 1037 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.66, + "step": 1038 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4596, + "step": 1039 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.369, + "step": 1040 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.5105, + "step": 1041 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.581, + "step": 1042 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4148, + "step": 1043 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.4202, + "step": 1044 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0011, + "step": 1045 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.2897, + "step": 1046 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 1.9517, + "step": 1047 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.343, + "step": 1048 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.0976, + "step": 1049 + }, + { + "epoch": 0.1, + "learning_rate": 0.0001, + "loss": 2.3578, + "step": 1050 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.9549, + "step": 1051 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3043, + "step": 1052 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0788, + "step": 1053 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.6131, + "step": 1054 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0787, + "step": 1055 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.7103, + "step": 1056 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.9921, + "step": 1057 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5975, + "step": 1058 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.9221, + "step": 1059 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3886, + "step": 1060 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.8689, + "step": 1061 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.34, + "step": 1062 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.8414, + "step": 1063 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2424, + "step": 1064 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1411, + "step": 1065 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3864, + "step": 1066 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0959, + "step": 1067 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5048, + "step": 1068 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2478, + "step": 1069 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 3.1831, + "step": 1070 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5807, + "step": 1071 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.8399, + "step": 1072 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2706, + "step": 1073 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.456, + "step": 1074 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0891, + "step": 1075 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3004, + "step": 1076 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.948, + "step": 1077 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.313, + "step": 1078 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3539, + "step": 1079 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3412, + "step": 1080 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3311, + "step": 1081 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.4311, + "step": 1082 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3056, + "step": 1083 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.4721, + "step": 1084 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5353, + "step": 1085 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3544, + "step": 1086 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1434, + "step": 1087 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3063, + "step": 1088 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0573, + "step": 1089 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3413, + "step": 1090 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1493, + "step": 1091 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2186, + "step": 1092 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0971, + "step": 1093 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.7698, + "step": 1094 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.9825, + "step": 1095 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5889, + "step": 1096 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1425, + "step": 1097 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5334, + "step": 1098 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2054, + "step": 1099 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2911, + "step": 1100 + }, + { + "epoch": 0.11, + "eval_bleu": 12.388735507445373, + "eval_brevity_penalty": 0.5200566781591875, + "eval_loss": 1.7078295946121216, + "eval_runtime": 7.4337, + "eval_samples_per_second": 403.3, + "eval_steps_per_second": 0.807, + "step": 1100 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2183, + "step": 1101 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.8819, + "step": 1102 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2164, + "step": 1103 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.6593, + "step": 1104 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1611, + "step": 1105 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.4621, + "step": 1106 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1828, + "step": 1107 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.5422, + "step": 1108 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1934, + "step": 1109 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3334, + "step": 1110 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2983, + "step": 1111 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.7019, + "step": 1112 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1626, + "step": 1113 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3944, + "step": 1114 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3142, + "step": 1115 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1814, + "step": 1116 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2915, + "step": 1117 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.6277, + "step": 1118 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1863, + "step": 1119 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 3.024, + "step": 1120 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1706, + "step": 1121 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 3.1399, + "step": 1122 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0705, + "step": 1123 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 3.2452, + "step": 1124 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.35, + "step": 1125 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.9049, + "step": 1126 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2479, + "step": 1127 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2205, + "step": 1128 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1328, + "step": 1129 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2052, + "step": 1130 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.9822, + "step": 1131 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3932, + "step": 1132 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2446, + "step": 1133 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2748, + "step": 1134 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.0319, + "step": 1135 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.38, + "step": 1136 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.8489, + "step": 1137 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3309, + "step": 1138 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.5973, + "step": 1139 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2538, + "step": 1140 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.7805, + "step": 1141 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2941, + "step": 1142 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3309, + "step": 1143 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3666, + "step": 1144 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 1.8275, + "step": 1145 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.3054, + "step": 1146 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.2713, + "step": 1147 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.4066, + "step": 1148 + }, + { + "epoch": 0.11, + "learning_rate": 0.0001, + "loss": 2.1656, + "step": 1149 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3293, + "step": 1150 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.203, + "step": 1151 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5667, + "step": 1152 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.0159, + "step": 1153 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.6668, + "step": 1154 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1931, + "step": 1155 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.9615, + "step": 1156 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3742, + "step": 1157 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4827, + "step": 1158 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.463, + "step": 1159 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4751, + "step": 1160 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.9847, + "step": 1161 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4586, + "step": 1162 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.4073, + "step": 1163 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3036, + "step": 1164 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.8465, + "step": 1165 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5184, + "step": 1166 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.9915, + "step": 1167 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4548, + "step": 1168 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1637, + "step": 1169 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4198, + "step": 1170 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.0655, + "step": 1171 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2014, + "step": 1172 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2536, + "step": 1173 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3222, + "step": 1174 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3765, + "step": 1175 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3604, + "step": 1176 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.9143, + "step": 1177 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3094, + "step": 1178 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.6657, + "step": 1179 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.7923, + "step": 1180 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.5297, + "step": 1181 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.4048, + "step": 1182 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.5935, + "step": 1183 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.8404, + "step": 1184 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.5738, + "step": 1185 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2689, + "step": 1186 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.5635, + "step": 1187 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.0735, + "step": 1188 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.6609, + "step": 1189 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2347, + "step": 1190 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3586, + "step": 1191 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2943, + "step": 1192 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3519, + "step": 1193 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4317, + "step": 1194 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3983, + "step": 1195 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4526, + "step": 1196 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3543, + "step": 1197 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1305, + "step": 1198 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4143, + "step": 1199 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3949, + "step": 1200 + }, + { + "epoch": 0.12, + "eval_bleu": 12.810938898358149, + "eval_brevity_penalty": 0.5288477405297071, + "eval_loss": 1.6996500492095947, + "eval_runtime": 5.4415, + "eval_samples_per_second": 550.953, + "eval_steps_per_second": 1.103, + "step": 1200 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3267, + "step": 1201 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4091, + "step": 1202 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3539, + "step": 1203 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3633, + "step": 1204 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2584, + "step": 1205 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3996, + "step": 1206 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3728, + "step": 1207 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4245, + "step": 1208 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2952, + "step": 1209 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5451, + "step": 1210 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2786, + "step": 1211 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.0057, + "step": 1212 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2326, + "step": 1213 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.398, + "step": 1214 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2189, + "step": 1215 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5898, + "step": 1216 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2212, + "step": 1217 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1426, + "step": 1218 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3251, + "step": 1219 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.59, + "step": 1220 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2714, + "step": 1221 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5785, + "step": 1222 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3521, + "step": 1223 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4318, + "step": 1224 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2975, + "step": 1225 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2948, + "step": 1226 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2829, + "step": 1227 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1645, + "step": 1228 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2881, + "step": 1229 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.6078, + "step": 1230 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3697, + "step": 1231 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3919, + "step": 1232 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4875, + "step": 1233 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3435, + "step": 1234 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 1.7071, + "step": 1235 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.1464, + "step": 1236 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1603, + "step": 1237 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.0544, + "step": 1238 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3125, + "step": 1239 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 3.168, + "step": 1240 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.5252, + "step": 1241 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.7528, + "step": 1242 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1435, + "step": 1243 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2086, + "step": 1244 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3866, + "step": 1245 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.1285, + "step": 1246 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.4474, + "step": 1247 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2076, + "step": 1248 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.3061, + "step": 1249 + }, + { + "epoch": 0.12, + "learning_rate": 0.0001, + "loss": 2.2772, + "step": 1250 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4201, + "step": 1251 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3113, + "step": 1252 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.6342, + "step": 1253 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3001, + "step": 1254 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.0244, + "step": 1255 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4112, + "step": 1256 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2765, + "step": 1257 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2627, + "step": 1258 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.474, + "step": 1259 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.38, + "step": 1260 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9815, + "step": 1261 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.357, + "step": 1262 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1904, + "step": 1263 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4791, + "step": 1264 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3359, + "step": 1265 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.7244, + "step": 1266 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2744, + "step": 1267 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.6637, + "step": 1268 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9718, + "step": 1269 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.5773, + "step": 1270 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.414, + "step": 1271 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2305, + "step": 1272 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.0077, + "step": 1273 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4169, + "step": 1274 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.273, + "step": 1275 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.6132, + "step": 1276 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3865, + "step": 1277 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4602, + "step": 1278 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1414, + "step": 1279 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4306, + "step": 1280 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4063, + "step": 1281 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.504, + "step": 1282 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.5859, + "step": 1283 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.5071, + "step": 1284 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.138, + "step": 1285 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.7095, + "step": 1286 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2157, + "step": 1287 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.9186, + "step": 1288 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9502, + "step": 1289 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4772, + "step": 1290 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3394, + "step": 1291 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.309, + "step": 1292 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3871, + "step": 1293 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.8427, + "step": 1294 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3861, + "step": 1295 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.245, + "step": 1296 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3635, + "step": 1297 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.9144, + "step": 1298 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3973, + "step": 1299 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2343, + "step": 1300 + }, + { + "epoch": 0.13, + "eval_bleu": 12.821349861122833, + "eval_brevity_penalty": 0.5282513002792236, + "eval_loss": 1.693037986755371, + "eval_runtime": 5.0877, + "eval_samples_per_second": 589.268, + "eval_steps_per_second": 1.179, + "step": 1300 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3481, + "step": 1301 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1439, + "step": 1302 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2647, + "step": 1303 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.157, + "step": 1304 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3565, + "step": 1305 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4126, + "step": 1306 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.378, + "step": 1307 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2024, + "step": 1308 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3493, + "step": 1309 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2949, + "step": 1310 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.318, + "step": 1311 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.374, + "step": 1312 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2561, + "step": 1313 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3616, + "step": 1314 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2905, + "step": 1315 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4127, + "step": 1316 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3849, + "step": 1317 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.6028, + "step": 1318 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2446, + "step": 1319 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.7279, + "step": 1320 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2477, + "step": 1321 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.9186, + "step": 1322 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4325, + "step": 1323 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4436, + "step": 1324 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2402, + "step": 1325 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3647, + "step": 1326 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.0203, + "step": 1327 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.9713, + "step": 1328 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 1.896, + "step": 1329 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4327, + "step": 1330 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1072, + "step": 1331 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4697, + "step": 1332 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.39, + "step": 1333 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4793, + "step": 1334 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1625, + "step": 1335 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4382, + "step": 1336 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3195, + "step": 1337 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4033, + "step": 1338 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.4975, + "step": 1339 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.241, + "step": 1340 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1543, + "step": 1341 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.196, + "step": 1342 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2628, + "step": 1343 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.6883, + "step": 1344 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3316, + "step": 1345 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.2555, + "step": 1346 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.3857, + "step": 1347 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 3.1206, + "step": 1348 + }, + { + "epoch": 0.13, + "learning_rate": 0.0001, + "loss": 2.1071, + "step": 1349 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9233, + "step": 1350 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.9116, + "step": 1351 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3684, + "step": 1352 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.9915, + "step": 1353 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.216, + "step": 1354 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2679, + "step": 1355 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2666, + "step": 1356 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.9683, + "step": 1357 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1825, + "step": 1358 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.409, + "step": 1359 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2863, + "step": 1360 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3242, + "step": 1361 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3259, + "step": 1362 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3118, + "step": 1363 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4307, + "step": 1364 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2677, + "step": 1365 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.323, + "step": 1366 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.6913, + "step": 1367 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2322, + "step": 1368 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1889, + "step": 1369 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3392, + "step": 1370 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3419, + "step": 1371 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3288, + "step": 1372 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2819, + "step": 1373 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1566, + "step": 1374 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4267, + "step": 1375 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4844, + "step": 1376 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3337, + "step": 1377 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.6521, + "step": 1378 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2529, + "step": 1379 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3843, + "step": 1380 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2747, + "step": 1381 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.6082, + "step": 1382 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2512, + "step": 1383 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4116, + "step": 1384 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3204, + "step": 1385 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5113, + "step": 1386 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2721, + "step": 1387 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4399, + "step": 1388 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2733, + "step": 1389 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5359, + "step": 1390 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3062, + "step": 1391 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.9296, + "step": 1392 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.48, + "step": 1393 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4837, + "step": 1394 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4671, + "step": 1395 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3773, + "step": 1396 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1427, + "step": 1397 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3014, + "step": 1398 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 1.9637, + "step": 1399 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2525, + "step": 1400 + }, + { + "epoch": 0.14, + "eval_bleu": 13.122124801275081, + "eval_brevity_penalty": 0.5284643317499238, + "eval_loss": 1.6850625276565552, + "eval_runtime": 5.688, + "eval_samples_per_second": 527.076, + "eval_steps_per_second": 1.055, + "step": 1400 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2051, + "step": 1401 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.0105, + "step": 1402 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3558, + "step": 1403 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 3.1321, + "step": 1404 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5141, + "step": 1405 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.422, + "step": 1406 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2307, + "step": 1407 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4011, + "step": 1408 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4779, + "step": 1409 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.0778, + "step": 1410 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2118, + "step": 1411 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2057, + "step": 1412 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1026, + "step": 1413 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3122, + "step": 1414 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3653, + "step": 1415 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4716, + "step": 1416 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3171, + "step": 1417 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4631, + "step": 1418 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2185, + "step": 1419 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3317, + "step": 1420 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4, + "step": 1421 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3495, + "step": 1422 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2844, + "step": 1423 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2871, + "step": 1424 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1792, + "step": 1425 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2598, + "step": 1426 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3015, + "step": 1427 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5605, + "step": 1428 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2824, + "step": 1429 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3898, + "step": 1430 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2862, + "step": 1431 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.8169, + "step": 1432 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2751, + "step": 1433 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.7714, + "step": 1434 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2494, + "step": 1435 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4783, + "step": 1436 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.1862, + "step": 1437 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.5798, + "step": 1438 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2342, + "step": 1439 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3026, + "step": 1440 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3137, + "step": 1441 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.6029, + "step": 1442 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2393, + "step": 1443 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4593, + "step": 1444 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2768, + "step": 1445 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.233, + "step": 1446 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.2019, + "step": 1447 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.4377, + "step": 1448 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.3925, + "step": 1449 + }, + { + "epoch": 0.14, + "learning_rate": 0.0001, + "loss": 2.7156, + "step": 1450 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2111, + "step": 1451 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2285, + "step": 1452 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3191, + "step": 1453 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.104, + "step": 1454 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4971, + "step": 1455 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.9447, + "step": 1456 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4008, + "step": 1457 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.0206, + "step": 1458 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3625, + "step": 1459 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.8023, + "step": 1460 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3815, + "step": 1461 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4779, + "step": 1462 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3115, + "step": 1463 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0602, + "step": 1464 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1378, + "step": 1465 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.292, + "step": 1466 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3113, + "step": 1467 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.345, + "step": 1468 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2786, + "step": 1469 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4047, + "step": 1470 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3548, + "step": 1471 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3627, + "step": 1472 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3001, + "step": 1473 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3864, + "step": 1474 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.123, + "step": 1475 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3891, + "step": 1476 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4565, + "step": 1477 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3217, + "step": 1478 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2509, + "step": 1479 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.506, + "step": 1480 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.273, + "step": 1481 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5284, + "step": 1482 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1994, + "step": 1483 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3921, + "step": 1484 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2446, + "step": 1485 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2366, + "step": 1486 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1982, + "step": 1487 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3032, + "step": 1488 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.27, + "step": 1489 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4754, + "step": 1490 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3364, + "step": 1491 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0807, + "step": 1492 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3056, + "step": 1493 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2336, + "step": 1494 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1388, + "step": 1495 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4331, + "step": 1496 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2771, + "step": 1497 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.447, + "step": 1498 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.137, + "step": 1499 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2604, + "step": 1500 + }, + { + "epoch": 0.15, + "eval_bleu": 13.089581997774658, + "eval_brevity_penalty": 0.5260772979517508, + "eval_loss": 1.6795377731323242, + "eval_runtime": 10.5108, + "eval_samples_per_second": 285.23, + "eval_steps_per_second": 0.571, + "step": 1500 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.7507, + "step": 1501 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3301, + "step": 1502 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.7004, + "step": 1503 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5295, + "step": 1504 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.59, + "step": 1505 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5035, + "step": 1506 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8578, + "step": 1507 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3276, + "step": 1508 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3104, + "step": 1509 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3486, + "step": 1510 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2841, + "step": 1511 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2586, + "step": 1512 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1943, + "step": 1513 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.3873, + "step": 1514 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2714, + "step": 1515 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 3.1162, + "step": 1516 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 1517 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5094, + "step": 1518 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2078, + "step": 1519 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3368, + "step": 1520 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1575, + "step": 1521 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.253, + "step": 1522 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0712, + "step": 1523 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1935, + "step": 1524 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.052, + "step": 1525 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3766, + "step": 1526 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0525, + "step": 1527 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3333, + "step": 1528 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1322, + "step": 1529 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2595, + "step": 1530 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8494, + "step": 1531 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3845, + "step": 1532 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.5983, + "step": 1533 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3524, + "step": 1534 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.5724, + "step": 1535 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.4353, + "step": 1536 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.8922, + "step": 1537 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0796, + "step": 1538 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2952, + "step": 1539 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3691, + "step": 1540 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.2813, + "step": 1541 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1179, + "step": 1542 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.3239, + "step": 1543 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5748, + "step": 1544 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 1.9708, + "step": 1545 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.5207, + "step": 1546 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.0504, + "step": 1547 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.7563, + "step": 1548 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.1242, + "step": 1549 + }, + { + "epoch": 0.15, + "learning_rate": 0.0001, + "loss": 2.8278, + "step": 1550 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0773, + "step": 1551 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3102, + "step": 1552 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1525, + "step": 1553 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2786, + "step": 1554 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0309, + "step": 1555 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2535, + "step": 1556 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.9593, + "step": 1557 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2841, + "step": 1558 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0243, + "step": 1559 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2255, + "step": 1560 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0109, + "step": 1561 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4279, + "step": 1562 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.038, + "step": 1563 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3123, + "step": 1564 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7844, + "step": 1565 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1469, + "step": 1566 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.789, + "step": 1567 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.8071, + "step": 1568 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7626, + "step": 1569 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.1409, + "step": 1570 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8629, + "step": 1571 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3392, + "step": 1572 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2888, + "step": 1573 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2558, + "step": 1574 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.326, + "step": 1575 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2108, + "step": 1576 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.065, + "step": 1577 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3782, + "step": 1578 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2339, + "step": 1579 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4144, + "step": 1580 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7118, + "step": 1581 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3764, + "step": 1582 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.9825, + "step": 1583 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3525, + "step": 1584 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1458, + "step": 1585 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3356, + "step": 1586 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.859, + "step": 1587 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4162, + "step": 1588 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.085, + "step": 1589 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.554, + "step": 1590 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1681, + "step": 1591 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4662, + "step": 1592 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8363, + "step": 1593 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.8329, + "step": 1594 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0522, + "step": 1595 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3527, + "step": 1596 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8222, + "step": 1597 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3181, + "step": 1598 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7684, + "step": 1599 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3146, + "step": 1600 + }, + { + "epoch": 0.16, + "eval_bleu": 13.174059590154677, + "eval_brevity_penalty": 0.5291459047633211, + "eval_loss": 1.6722744703292847, + "eval_runtime": 7.2064, + "eval_samples_per_second": 416.017, + "eval_steps_per_second": 0.833, + "step": 1600 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0517, + "step": 1601 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3379, + "step": 1602 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.6361, + "step": 1603 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.5712, + "step": 1604 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8108, + "step": 1605 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3283, + "step": 1606 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8312, + "step": 1607 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4274, + "step": 1608 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0293, + "step": 1609 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.6424, + "step": 1610 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3794, + "step": 1611 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2597, + "step": 1612 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3599, + "step": 1613 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3997, + "step": 1614 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7015, + "step": 1615 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1599, + "step": 1616 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3622, + "step": 1617 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.5666, + "step": 1618 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7425, + "step": 1619 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7968, + "step": 1620 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0274, + "step": 1621 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0608, + "step": 1622 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0173, + "step": 1623 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1143, + "step": 1624 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7783, + "step": 1625 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.0255, + "step": 1626 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8492, + "step": 1627 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 3.0559, + "step": 1628 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.9018, + "step": 1629 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.5365, + "step": 1630 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1107, + "step": 1631 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3816, + "step": 1632 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.6453, + "step": 1633 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1625, + "step": 1634 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.9727, + "step": 1635 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2136, + "step": 1636 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.612, + "step": 1637 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2496, + "step": 1638 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.0701, + "step": 1639 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.3915, + "step": 1640 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.6642, + "step": 1641 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.5277, + "step": 1642 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.9184, + "step": 1643 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.4415, + "step": 1644 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.7851, + "step": 1645 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.2326, + "step": 1646 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.6811, + "step": 1647 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 2.1249, + "step": 1648 + }, + { + "epoch": 0.16, + "learning_rate": 0.0001, + "loss": 1.8879, + "step": 1649 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3889, + "step": 1650 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0206, + "step": 1651 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.2517, + "step": 1652 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1483, + "step": 1653 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.6152, + "step": 1654 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.8828, + "step": 1655 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 3.0982, + "step": 1656 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1998, + "step": 1657 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3487, + "step": 1658 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1755, + "step": 1659 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5269, + "step": 1660 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0017, + "step": 1661 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4217, + "step": 1662 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.187, + "step": 1663 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5145, + "step": 1664 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1668, + "step": 1665 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5232, + "step": 1666 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0338, + "step": 1667 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4937, + "step": 1668 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1369, + "step": 1669 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5306, + "step": 1670 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0749, + "step": 1671 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.7382, + "step": 1672 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0576, + "step": 1673 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.451, + "step": 1674 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1789, + "step": 1675 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1433, + "step": 1676 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1457, + "step": 1677 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4583, + "step": 1678 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.2087, + "step": 1679 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5126, + "step": 1680 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1424, + "step": 1681 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.8745, + "step": 1682 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0345, + "step": 1683 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.9762, + "step": 1684 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.8298, + "step": 1685 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 3.1777, + "step": 1686 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0676, + "step": 1687 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.7419, + "step": 1688 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.7612, + "step": 1689 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4785, + "step": 1690 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.8107, + "step": 1691 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4424, + "step": 1692 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1103, + "step": 1693 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5491, + "step": 1694 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1934, + "step": 1695 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0545, + "step": 1696 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.9022, + "step": 1697 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.7309, + "step": 1698 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1244, + "step": 1699 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5767, + "step": 1700 + }, + { + "epoch": 0.17, + "eval_bleu": 13.422444799010147, + "eval_brevity_penalty": 0.5248402260881412, + "eval_loss": 1.6596146821975708, + "eval_runtime": 5.0659, + "eval_samples_per_second": 591.798, + "eval_steps_per_second": 1.184, + "step": 1700 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.9027, + "step": 1701 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4975, + "step": 1702 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0076, + "step": 1703 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.6277, + "step": 1704 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0974, + "step": 1705 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.478, + "step": 1706 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.1143, + "step": 1707 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4853, + "step": 1708 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.759, + "step": 1709 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.2783, + "step": 1710 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.6923, + "step": 1711 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4872, + "step": 1712 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.8402, + "step": 1713 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.8729, + "step": 1714 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.4061, + "step": 1715 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4555, + "step": 1716 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.3399, + "step": 1717 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3308, + "step": 1718 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.1995, + "step": 1719 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5109, + "step": 1720 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.1123, + "step": 1721 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.634, + "step": 1722 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.096, + "step": 1723 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3102, + "step": 1724 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.0158, + "step": 1725 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3126, + "step": 1726 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.0176, + "step": 1727 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.317, + "step": 1728 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0239, + "step": 1729 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.7487, + "step": 1730 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.4511, + "step": 1731 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 3.1149, + "step": 1732 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.3681, + "step": 1733 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.6333, + "step": 1734 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 1.9175, + "step": 1735 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4311, + "step": 1736 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0029, + "step": 1737 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3209, + "step": 1738 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0458, + "step": 1739 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.8896, + "step": 1740 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0779, + "step": 1741 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.3915, + "step": 1742 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.103, + "step": 1743 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.4541, + "step": 1744 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.2041, + "step": 1745 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5548, + "step": 1746 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0486, + "step": 1747 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5651, + "step": 1748 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.0905, + "step": 1749 + }, + { + "epoch": 0.17, + "learning_rate": 0.0001, + "loss": 2.5884, + "step": 1750 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2006, + "step": 1751 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.3282, + "step": 1752 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1474, + "step": 1753 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.3927, + "step": 1754 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9531, + "step": 1755 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.6477, + "step": 1756 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0737, + "step": 1757 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2479, + "step": 1758 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0845, + "step": 1759 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0817, + "step": 1760 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1203, + "step": 1761 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5202, + "step": 1762 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0297, + "step": 1763 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.7376, + "step": 1764 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9814, + "step": 1765 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4339, + "step": 1766 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.196, + "step": 1767 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2025, + "step": 1768 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9983, + "step": 1769 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1544, + "step": 1770 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.04, + "step": 1771 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.2189, + "step": 1772 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1581, + "step": 1773 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.0427, + "step": 1774 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.073, + "step": 1775 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.913, + "step": 1776 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.03, + "step": 1777 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.3984, + "step": 1778 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0619, + "step": 1779 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4747, + "step": 1780 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.102, + "step": 1781 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1936, + "step": 1782 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9057, + "step": 1783 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.1244, + "step": 1784 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0762, + "step": 1785 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.513, + "step": 1786 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.7162, + "step": 1787 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.617, + "step": 1788 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0406, + "step": 1789 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5033, + "step": 1790 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.7971, + "step": 1791 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5193, + "step": 1792 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9955, + "step": 1793 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4435, + "step": 1794 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9901, + "step": 1795 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4559, + "step": 1796 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1179, + "step": 1797 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2254, + "step": 1798 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2581, + "step": 1799 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.698, + "step": 1800 + }, + { + "epoch": 0.18, + "eval_bleu": 13.673285131612609, + "eval_brevity_penalty": 0.533380042444439, + "eval_loss": 1.657647728919983, + "eval_runtime": 6.0523, + "eval_samples_per_second": 495.346, + "eval_steps_per_second": 0.991, + "step": 1800 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0205, + "step": 1801 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5482, + "step": 1802 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9181, + "step": 1803 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5964, + "step": 1804 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1806, + "step": 1805 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.53, + "step": 1806 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0982, + "step": 1807 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.7862, + "step": 1808 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2309, + "step": 1809 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4982, + "step": 1810 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0701, + "step": 1811 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.3709, + "step": 1812 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9273, + "step": 1813 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2866, + "step": 1814 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1256, + "step": 1815 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1146, + "step": 1816 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9796, + "step": 1817 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4482, + "step": 1818 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0187, + "step": 1819 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2938, + "step": 1820 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2546, + "step": 1821 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.0367, + "step": 1822 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9858, + "step": 1823 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.1858, + "step": 1824 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1712, + "step": 1825 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.2652, + "step": 1826 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0884, + "step": 1827 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4717, + "step": 1828 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.259, + "step": 1829 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.3576, + "step": 1830 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9726, + "step": 1831 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 3.1703, + "step": 1832 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0497, + "step": 1833 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.681, + "step": 1834 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 1.9945, + "step": 1835 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4124, + "step": 1836 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0155, + "step": 1837 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.6883, + "step": 1838 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1295, + "step": 1839 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.6226, + "step": 1840 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0281, + "step": 1841 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4342, + "step": 1842 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.122, + "step": 1843 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5977, + "step": 1844 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1051, + "step": 1845 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4001, + "step": 1846 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.0587, + "step": 1847 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.4311, + "step": 1848 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.1301, + "step": 1849 + }, + { + "epoch": 0.18, + "learning_rate": 0.0001, + "loss": 2.5184, + "step": 1850 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0609, + "step": 1851 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3678, + "step": 1852 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0797, + "step": 1853 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.7276, + "step": 1854 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.7974, + "step": 1855 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5311, + "step": 1856 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.8618, + "step": 1857 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2646, + "step": 1858 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0301, + "step": 1859 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4227, + "step": 1860 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.8822, + "step": 1861 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2249, + "step": 1862 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0405, + "step": 1863 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3921, + "step": 1864 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0752, + "step": 1865 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.837, + "step": 1866 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.7055, + "step": 1867 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.9377, + "step": 1868 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0347, + "step": 1869 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3927, + "step": 1870 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0292, + "step": 1871 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5525, + "step": 1872 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.875, + "step": 1873 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3101, + "step": 1874 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0509, + "step": 1875 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 3.1085, + "step": 1876 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.1779, + "step": 1877 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5845, + "step": 1878 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0583, + "step": 1879 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5485, + "step": 1880 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9814, + "step": 1881 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6321, + "step": 1882 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9992, + "step": 1883 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6193, + "step": 1884 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9589, + "step": 1885 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.524, + "step": 1886 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.8935, + "step": 1887 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4737, + "step": 1888 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9771, + "step": 1889 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2488, + "step": 1890 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9629, + "step": 1891 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6117, + "step": 1892 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2219, + "step": 1893 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0398, + "step": 1894 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3055, + "step": 1895 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.8886, + "step": 1896 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9921, + "step": 1897 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3959, + "step": 1898 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.949, + "step": 1899 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6416, + "step": 1900 + }, + { + "epoch": 0.19, + "eval_bleu": 13.71841714641878, + "eval_brevity_penalty": 0.5349738433199753, + "eval_loss": 1.6513681411743164, + "eval_runtime": 6.857, + "eval_samples_per_second": 437.217, + "eval_steps_per_second": 0.875, + "step": 1900 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.5808, + "step": 1901 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4436, + "step": 1902 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.5648, + "step": 1903 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3309, + "step": 1904 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.7378, + "step": 1905 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0577, + "step": 1906 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.951, + "step": 1907 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2226, + "step": 1908 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.7613, + "step": 1909 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 3.0224, + "step": 1910 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.8879, + "step": 1911 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 3.1739, + "step": 1912 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9702, + "step": 1913 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.8656, + "step": 1914 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9176, + "step": 1915 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3081, + "step": 1916 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0936, + "step": 1917 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2659, + "step": 1918 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0544, + "step": 1919 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6321, + "step": 1920 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.24, + "step": 1921 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.6017, + "step": 1922 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9797, + "step": 1923 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3914, + "step": 1924 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0241, + "step": 1925 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5585, + "step": 1926 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.1368, + "step": 1927 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.5836, + "step": 1928 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0804, + "step": 1929 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3625, + "step": 1930 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9956, + "step": 1931 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2452, + "step": 1932 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.1414, + "step": 1933 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4697, + "step": 1934 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0288, + "step": 1935 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4525, + "step": 1936 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2356, + "step": 1937 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4717, + "step": 1938 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0773, + "step": 1939 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.9834, + "step": 1940 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 1.9619, + "step": 1941 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2906, + "step": 1942 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2381, + "step": 1943 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.3482, + "step": 1944 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.0615, + "step": 1945 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.2454, + "step": 1946 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.1189, + "step": 1947 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.4647, + "step": 1948 + }, + { + "epoch": 0.19, + "learning_rate": 0.0001, + "loss": 2.1637, + "step": 1949 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9209, + "step": 1950 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.9342, + "step": 1951 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 3.1304, + "step": 1952 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2501, + "step": 1953 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3074, + "step": 1954 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.779, + "step": 1955 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1393, + "step": 1956 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.744, + "step": 1957 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5929, + "step": 1958 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.6436, + "step": 1959 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9589, + "step": 1960 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.6353, + "step": 1961 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3283, + "step": 1962 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.5964, + "step": 1963 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6092, + "step": 1964 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.877, + "step": 1965 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5192, + "step": 1966 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1746, + "step": 1967 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4535, + "step": 1968 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3272, + "step": 1969 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4372, + "step": 1970 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3292, + "step": 1971 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.176, + "step": 1972 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2812, + "step": 1973 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.728, + "step": 1974 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2512, + "step": 1975 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6565, + "step": 1976 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2261, + "step": 1977 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.297, + "step": 1978 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0683, + "step": 1979 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.412, + "step": 1980 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.201, + "step": 1981 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6852, + "step": 1982 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0591, + "step": 1983 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6434, + "step": 1984 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0852, + "step": 1985 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2886, + "step": 1986 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0884, + "step": 1987 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2915, + "step": 1988 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1541, + "step": 1989 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 3.0207, + "step": 1990 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2687, + "step": 1991 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.482, + "step": 1992 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1499, + "step": 1993 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4285, + "step": 1994 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2875, + "step": 1995 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2625, + "step": 1996 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3213, + "step": 1997 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9836, + "step": 1998 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2933, + "step": 1999 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 3.0841, + "step": 2000 + }, + { + "epoch": 0.2, + "eval_bleu": 13.9079201924629, + "eval_brevity_penalty": 0.5356960066586677, + "eval_loss": 1.6447670459747314, + "eval_runtime": 5.4234, + "eval_samples_per_second": 552.791, + "eval_steps_per_second": 1.106, + "step": 2000 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.084, + "step": 2001 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6445, + "step": 2002 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0764, + "step": 2003 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6381, + "step": 2004 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.8278, + "step": 2005 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5366, + "step": 2006 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1208, + "step": 2007 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4679, + "step": 2008 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2557, + "step": 2009 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3286, + "step": 2010 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1822, + "step": 2011 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.529, + "step": 2012 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.071, + "step": 2013 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5231, + "step": 2014 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0767, + "step": 2015 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4711, + "step": 2016 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.6752, + "step": 2017 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9278, + "step": 2018 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.7438, + "step": 2019 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4985, + "step": 2020 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0269, + "step": 2021 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4633, + "step": 2022 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0663, + "step": 2023 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.8942, + "step": 2024 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0164, + "step": 2025 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5145, + "step": 2026 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2139, + "step": 2027 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2919, + "step": 2028 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.2794, + "step": 2029 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.7741, + "step": 2030 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.108, + "step": 2031 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9594, + "step": 2032 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0346, + "step": 2033 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3887, + "step": 2034 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.9337, + "step": 2035 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.3245, + "step": 2036 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.9111, + "step": 2037 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.9307, + "step": 2038 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.8024, + "step": 2039 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 3.1231, + "step": 2040 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 2041 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6512, + "step": 2042 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0636, + "step": 2043 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.4526, + "step": 2044 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 1.9305, + "step": 2045 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5892, + "step": 2046 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.0315, + "step": 2047 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.6338, + "step": 2048 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.1789, + "step": 2049 + }, + { + "epoch": 0.2, + "learning_rate": 0.0001, + "loss": 2.5609, + "step": 2050 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0511, + "step": 2051 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6477, + "step": 2052 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0161, + "step": 2053 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2268, + "step": 2054 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1057, + "step": 2055 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3275, + "step": 2056 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9784, + "step": 2057 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6003, + "step": 2058 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9885, + "step": 2059 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2839, + "step": 2060 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.7673, + "step": 2061 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6087, + "step": 2062 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.793, + "step": 2063 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6998, + "step": 2064 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1678, + "step": 2065 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2135, + "step": 2066 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2044, + "step": 2067 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.679, + "step": 2068 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0461, + "step": 2069 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 3.1231, + "step": 2070 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1817, + "step": 2071 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 3.1237, + "step": 2072 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1229, + "step": 2073 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3493, + "step": 2074 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1126, + "step": 2075 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4495, + "step": 2076 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1125, + "step": 2077 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.7287, + "step": 2078 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9123, + "step": 2079 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.801, + "step": 2080 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.5213, + "step": 2081 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4072, + "step": 2082 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0354, + "step": 2083 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6459, + "step": 2084 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.405, + "step": 2085 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5915, + "step": 2086 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1513, + "step": 2087 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4389, + "step": 2088 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2242, + "step": 2089 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4133, + "step": 2090 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9604, + "step": 2091 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.7853, + "step": 2092 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.8753, + "step": 2093 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.8069, + "step": 2094 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5957, + "step": 2095 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4573, + "step": 2096 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9987, + "step": 2097 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6196, + "step": 2098 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1931, + "step": 2099 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5039, + "step": 2100 + }, + { + "epoch": 0.21, + "eval_bleu": 13.986028611473095, + "eval_brevity_penalty": 0.5360994704770286, + "eval_loss": 1.637455940246582, + "eval_runtime": 6.6425, + "eval_samples_per_second": 451.339, + "eval_steps_per_second": 0.903, + "step": 2100 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.6654, + "step": 2101 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.8878, + "step": 2102 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1108, + "step": 2103 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3929, + "step": 2104 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1267, + "step": 2105 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3067, + "step": 2106 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1771, + "step": 2107 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2528, + "step": 2108 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.001, + "step": 2109 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.968, + "step": 2110 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.227, + "step": 2111 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.9283, + "step": 2112 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3086, + "step": 2113 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5088, + "step": 2114 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0569, + "step": 2115 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2373, + "step": 2116 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.7884, + "step": 2117 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3002, + "step": 2118 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.7294, + "step": 2119 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.644, + "step": 2120 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1149, + "step": 2121 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.9015, + "step": 2122 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.3278, + "step": 2123 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5804, + "step": 2124 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2985, + "step": 2125 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.4217, + "step": 2126 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2745, + "step": 2127 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6096, + "step": 2128 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1201, + "step": 2129 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.6177, + "step": 2130 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.01, + "step": 2131 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5765, + "step": 2132 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.11, + "step": 2133 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5819, + "step": 2134 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0432, + "step": 2135 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.427, + "step": 2136 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9057, + "step": 2137 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2591, + "step": 2138 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9422, + "step": 2139 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 3.0553, + "step": 2140 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0617, + "step": 2141 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2313, + "step": 2142 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.1306, + "step": 2143 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5609, + "step": 2144 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 1.9644, + "step": 2145 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.7927, + "step": 2146 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.0253, + "step": 2147 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.5285, + "step": 2148 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.074, + "step": 2149 + }, + { + "epoch": 0.21, + "learning_rate": 0.0001, + "loss": 2.2732, + "step": 2150 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0654, + "step": 2151 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4434, + "step": 2152 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0561, + "step": 2153 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.9022, + "step": 2154 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0305, + "step": 2155 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 3.0255, + "step": 2156 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0225, + "step": 2157 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3558, + "step": 2158 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1055, + "step": 2159 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3182, + "step": 2160 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.825, + "step": 2161 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3365, + "step": 2162 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9746, + "step": 2163 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 3.0585, + "step": 2164 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9979, + "step": 2165 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.7067, + "step": 2166 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9637, + "step": 2167 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5444, + "step": 2168 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3223, + "step": 2169 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.6382, + "step": 2170 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.2491, + "step": 2171 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4611, + "step": 2172 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1684, + "step": 2173 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5828, + "step": 2174 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.3404, + "step": 2175 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5219, + "step": 2176 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.6025, + "step": 2177 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0808, + "step": 2178 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1452, + "step": 2179 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4322, + "step": 2180 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1344, + "step": 2181 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.8988, + "step": 2182 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.8891, + "step": 2183 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5275, + "step": 2184 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9548, + "step": 2185 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5674, + "step": 2186 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.7474, + "step": 2187 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5413, + "step": 2188 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.7601, + "step": 2189 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0745, + "step": 2190 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.6033, + "step": 2191 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4452, + "step": 2192 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.5399, + "step": 2193 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.9056, + "step": 2194 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.474, + "step": 2195 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 3.0875, + "step": 2196 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.4256, + "step": 2197 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3505, + "step": 2198 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.3581, + "step": 2199 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5829, + "step": 2200 + }, + { + "epoch": 0.22, + "eval_bleu": 13.924649734478159, + "eval_brevity_penalty": 0.532806007694745, + "eval_loss": 1.6365728378295898, + "eval_runtime": 6.1399, + "eval_samples_per_second": 488.281, + "eval_steps_per_second": 0.977, + "step": 2200 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.3348, + "step": 2201 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0071, + "step": 2202 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.2881, + "step": 2203 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.8093, + "step": 2204 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.2399, + "step": 2205 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4817, + "step": 2206 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.7618, + "step": 2207 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.6559, + "step": 2208 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.7643, + "step": 2209 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.3446, + "step": 2210 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.6149, + "step": 2211 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4574, + "step": 2212 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.949, + "step": 2213 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4295, + "step": 2214 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9288, + "step": 2215 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.7597, + "step": 2216 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0423, + "step": 2217 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5617, + "step": 2218 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1625, + "step": 2219 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4737, + "step": 2220 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9781, + "step": 2221 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4428, + "step": 2222 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0445, + "step": 2223 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5865, + "step": 2224 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1466, + "step": 2225 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5135, + "step": 2226 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0184, + "step": 2227 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1986, + "step": 2228 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.1216, + "step": 2229 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.9057, + "step": 2230 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0116, + "step": 2231 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 3.2311, + "step": 2232 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9226, + "step": 2233 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 3.0897, + "step": 2234 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0186, + "step": 2235 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4382, + "step": 2236 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9257, + "step": 2237 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.2639, + "step": 2238 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.7595, + "step": 2239 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4569, + "step": 2240 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9801, + "step": 2241 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.8257, + "step": 2242 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 1.9299, + "step": 2243 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5296, + "step": 2244 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.089, + "step": 2245 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.5745, + "step": 2246 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.0778, + "step": 2247 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.4061, + "step": 2248 + }, + { + "epoch": 0.22, + "learning_rate": 0.0001, + "loss": 2.081, + "step": 2249 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4887, + "step": 2250 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1479, + "step": 2251 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3835, + "step": 2252 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0427, + "step": 2253 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3226, + "step": 2254 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1209, + "step": 2255 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2462, + "step": 2256 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1904, + "step": 2257 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2906, + "step": 2258 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1406, + "step": 2259 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.284, + "step": 2260 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0883, + "step": 2261 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.387, + "step": 2262 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0724, + "step": 2263 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.794, + "step": 2264 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0575, + "step": 2265 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3596, + "step": 2266 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1311, + "step": 2267 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3198, + "step": 2268 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9855, + "step": 2269 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.6837, + "step": 2270 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0809, + "step": 2271 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.9933, + "step": 2272 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.7858, + "step": 2273 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2913, + "step": 2274 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.3922, + "step": 2275 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1448, + "step": 2276 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.3834, + "step": 2277 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2189, + "step": 2278 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.3028, + "step": 2279 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.452, + "step": 2280 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.1899, + "step": 2281 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 3.0994, + "step": 2282 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.1284, + "step": 2283 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3804, + "step": 2284 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.0546, + "step": 2285 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3606, + "step": 2286 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.0335, + "step": 2287 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.5263, + "step": 2288 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.5269, + "step": 2289 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4331, + "step": 2290 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0004, + "step": 2291 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4954, + "step": 2292 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.096, + "step": 2293 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1813, + "step": 2294 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.154, + "step": 2295 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2195, + "step": 2296 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.13, + "step": 2297 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4095, + "step": 2298 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0211, + "step": 2299 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.5332, + "step": 2300 + }, + { + "epoch": 0.23, + "eval_bleu": 13.489546359784956, + "eval_brevity_penalty": 0.520868827646254, + "eval_loss": 1.6348429918289185, + "eval_runtime": 5.8183, + "eval_samples_per_second": 515.267, + "eval_steps_per_second": 1.031, + "step": 2300 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0837, + "step": 2301 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3768, + "step": 2302 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.8989, + "step": 2303 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4197, + "step": 2304 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9258, + "step": 2305 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3005, + "step": 2306 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.8998, + "step": 2307 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.831, + "step": 2308 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9371, + "step": 2309 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4332, + "step": 2310 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9669, + "step": 2311 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1264, + "step": 2312 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9052, + "step": 2313 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3297, + "step": 2314 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1266, + "step": 2315 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 3.0055, + "step": 2316 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0026, + "step": 2317 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 3.0012, + "step": 2318 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0471, + "step": 2319 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 3.0234, + "step": 2320 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0149, + "step": 2321 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.6065, + "step": 2322 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1605, + "step": 2323 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.2703, + "step": 2324 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4153, + "step": 2325 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4487, + "step": 2326 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1383, + "step": 2327 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.9053, + "step": 2328 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1885, + "step": 2329 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3803, + "step": 2330 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.0497, + "step": 2331 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3989, + "step": 2332 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9176, + "step": 2333 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.5286, + "step": 2334 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9114, + "step": 2335 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4964, + "step": 2336 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.9689, + "step": 2337 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.5082, + "step": 2338 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.5019, + "step": 2339 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.1547, + "step": 2340 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.4401, + "step": 2341 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4255, + "step": 2342 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.3428, + "step": 2343 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.3741, + "step": 2344 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.3194, + "step": 2345 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.6231, + "step": 2346 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.1138, + "step": 2347 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.4212, + "step": 2348 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 1.1462, + "step": 2349 + }, + { + "epoch": 0.23, + "learning_rate": 0.0001, + "loss": 2.7539, + "step": 2350 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.0947, + "step": 2351 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5944, + "step": 2352 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 0.9864, + "step": 2353 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2325, + "step": 2354 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.1414, + "step": 2355 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2054, + "step": 2356 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.1222, + "step": 2357 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.8514, + "step": 2358 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.2037, + "step": 2359 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.7621, + "step": 2360 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.1475, + "step": 2361 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.174, + "step": 2362 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.0719, + "step": 2363 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4873, + "step": 2364 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.1517, + "step": 2365 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.8882, + "step": 2366 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.8768, + "step": 2367 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4915, + "step": 2368 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0076, + "step": 2369 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.629, + "step": 2370 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.9998, + "step": 2371 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5562, + "step": 2372 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2547, + "step": 2373 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.3835, + "step": 2374 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0993, + "step": 2375 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5786, + "step": 2376 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0592, + "step": 2377 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.3443, + "step": 2378 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0515, + "step": 2379 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2135, + "step": 2380 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0902, + "step": 2381 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.6395, + "step": 2382 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1742, + "step": 2383 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.7753, + "step": 2384 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.099, + "step": 2385 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.7109, + "step": 2386 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0528, + "step": 2387 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.569, + "step": 2388 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1519, + "step": 2389 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2717, + "step": 2390 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1531, + "step": 2391 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.7535, + "step": 2392 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0613, + "step": 2393 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 3.0399, + "step": 2394 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.9586, + "step": 2395 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4514, + "step": 2396 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1571, + "step": 2397 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1388, + "step": 2398 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1531, + "step": 2399 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5832, + "step": 2400 + }, + { + "epoch": 0.24, + "eval_bleu": 14.044457952427795, + "eval_brevity_penalty": 0.5348676238691165, + "eval_loss": 1.6240384578704834, + "eval_runtime": 5.2787, + "eval_samples_per_second": 567.945, + "eval_steps_per_second": 1.137, + "step": 2400 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.9627, + "step": 2401 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.6513, + "step": 2402 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1546, + "step": 2403 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.539, + "step": 2404 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1351, + "step": 2405 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.44, + "step": 2406 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1903, + "step": 2407 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4477, + "step": 2408 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1509, + "step": 2409 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.611, + "step": 2410 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0998, + "step": 2411 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4432, + "step": 2412 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1492, + "step": 2413 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2965, + "step": 2414 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0721, + "step": 2415 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2211, + "step": 2416 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1462, + "step": 2417 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.571, + "step": 2418 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.128, + "step": 2419 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.6005, + "step": 2420 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1735, + "step": 2421 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.8024, + "step": 2422 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2356, + "step": 2423 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4688, + "step": 2424 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0325, + "step": 2425 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.4913, + "step": 2426 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.104, + "step": 2427 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.3916, + "step": 2428 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.8985, + "step": 2429 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.3043, + "step": 2430 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0083, + "step": 2431 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.6772, + "step": 2432 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.8516, + "step": 2433 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 3.0339, + "step": 2434 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.85, + "step": 2435 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.3871, + "step": 2436 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.8473, + "step": 2437 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1253, + "step": 2438 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.9445, + "step": 2439 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.2776, + "step": 2440 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.9883, + "step": 2441 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.8153, + "step": 2442 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.1677, + "step": 2443 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.9909, + "step": 2444 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.0742, + "step": 2445 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.478, + "step": 2446 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.6453, + "step": 2447 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5122, + "step": 2448 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 1.5785, + "step": 2449 + }, + { + "epoch": 0.24, + "learning_rate": 0.0001, + "loss": 2.5415, + "step": 2450 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.3236, + "step": 2451 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5348, + "step": 2452 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0732, + "step": 2453 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4636, + "step": 2454 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8886, + "step": 2455 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1104, + "step": 2456 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.6689, + "step": 2457 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5577, + "step": 2458 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.7371, + "step": 2459 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6948, + "step": 2460 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.4174, + "step": 2461 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3359, + "step": 2462 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.1854, + "step": 2463 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9032, + "step": 2464 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.38, + "step": 2465 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4358, + "step": 2466 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.4781, + "step": 2467 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7096, + "step": 2468 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.4009, + "step": 2469 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1359, + "step": 2470 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8691, + "step": 2471 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4133, + "step": 2472 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1405, + "step": 2473 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9102, + "step": 2474 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0606, + "step": 2475 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 3.1134, + "step": 2476 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0065, + "step": 2477 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6257, + "step": 2478 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8713, + "step": 2479 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1698, + "step": 2480 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8829, + "step": 2481 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0251, + "step": 2482 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.9422, + "step": 2483 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9666, + "step": 2484 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.9748, + "step": 2485 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.356, + "step": 2486 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8574, + "step": 2487 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3775, + "step": 2488 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1749, + "step": 2489 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4487, + "step": 2490 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.024, + "step": 2491 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5099, + "step": 2492 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.2034, + "step": 2493 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6857, + "step": 2494 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0275, + "step": 2495 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6141, + "step": 2496 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.9297, + "step": 2497 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.2694, + "step": 2498 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.115, + "step": 2499 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.8577, + "step": 2500 + }, + { + "epoch": 0.25, + "eval_bleu": 14.10847399604805, + "eval_brevity_penalty": 0.5344426975198452, + "eval_loss": 1.6181683540344238, + "eval_runtime": 5.2944, + "eval_samples_per_second": 566.26, + "eval_steps_per_second": 1.133, + "step": 2500 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0036, + "step": 2501 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0704, + "step": 2502 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1315, + "step": 2503 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3164, + "step": 2504 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.069, + "step": 2505 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.016, + "step": 2506 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4884, + "step": 2507 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6934, + "step": 2508 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0683, + "step": 2509 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3834, + "step": 2510 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1476, + "step": 2511 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.4745, + "step": 2512 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.8748, + "step": 2513 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 3.1233, + "step": 2514 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.7267, + "step": 2515 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.803, + "step": 2516 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0761, + "step": 2517 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.9422, + "step": 2518 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4192, + "step": 2519 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9633, + "step": 2520 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1868, + "step": 2521 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.475, + "step": 2522 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5886, + "step": 2523 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.9743, + "step": 2524 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5435, + "step": 2525 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.4224, + "step": 2526 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1471, + "step": 2527 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.446, + "step": 2528 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7549, + "step": 2529 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0148, + "step": 2530 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7066, + "step": 2531 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7083, + "step": 2532 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.037, + "step": 2533 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.441, + "step": 2534 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3927, + "step": 2535 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0382, + "step": 2536 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.5545, + "step": 2537 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7736, + "step": 2538 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.119, + "step": 2539 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.3509, + "step": 2540 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.7688, + "step": 2541 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.0109, + "step": 2542 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9141, + "step": 2543 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.87, + "step": 2544 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.1031, + "step": 2545 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.24, + "step": 2546 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.6945, + "step": 2547 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 1.8905, + "step": 2548 + }, + { + "epoch": 0.25, + "learning_rate": 0.0001, + "loss": 2.9896, + "step": 2549 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8462, + "step": 2550 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.6183, + "step": 2551 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5084, + "step": 2552 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1374, + "step": 2553 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.4402, + "step": 2554 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4041, + "step": 2555 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.9671, + "step": 2556 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8869, + "step": 2557 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2506, + "step": 2558 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.9751, + "step": 2559 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.5893, + "step": 2560 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.7593, + "step": 2561 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8931, + "step": 2562 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.7156, + "step": 2563 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5044, + "step": 2564 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2441, + "step": 2565 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.6965, + "step": 2566 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2689, + "step": 2567 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.4462, + "step": 2568 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.4962, + "step": 2569 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.9237, + "step": 2570 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.6079, + "step": 2571 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.0603, + "step": 2572 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.7528, + "step": 2573 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8, + "step": 2574 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.5651, + "step": 2575 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3958, + "step": 2576 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2961, + "step": 2577 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8962, + "step": 2578 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5996, + "step": 2579 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1655, + "step": 2580 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.6986, + "step": 2581 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.556, + "step": 2582 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4741, + "step": 2583 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.0513, + "step": 2584 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4953, + "step": 2585 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5163, + "step": 2586 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.4808, + "step": 2587 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.0963, + "step": 2588 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1315, + "step": 2589 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.9328, + "step": 2590 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4659, + "step": 2591 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.771, + "step": 2592 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5625, + "step": 2593 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5602, + "step": 2594 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.3046, + "step": 2595 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4477, + "step": 2596 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1897, + "step": 2597 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.1604, + "step": 2598 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4516, + "step": 2599 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.9157, + "step": 2600 + }, + { + "epoch": 0.26, + "eval_bleu": 13.79820202886516, + "eval_brevity_penalty": 0.5365240931550531, + "eval_loss": 1.6285388469696045, + "eval_runtime": 5.5617, + "eval_samples_per_second": 539.039, + "eval_steps_per_second": 1.079, + "step": 2600 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.9964, + "step": 2601 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2763, + "step": 2602 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3244, + "step": 2603 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2358, + "step": 2604 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2242, + "step": 2605 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.9254, + "step": 2606 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.8469, + "step": 2607 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4705, + "step": 2608 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4444, + "step": 2609 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.8181, + "step": 2610 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1271, + "step": 2611 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5295, + "step": 2612 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.6146, + "step": 2613 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3393, + "step": 2614 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2111, + "step": 2615 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.6871, + "step": 2616 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.4904, + "step": 2617 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3993, + "step": 2618 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2116, + "step": 2619 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.6565, + "step": 2620 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1154, + "step": 2621 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2176, + "step": 2622 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.7628, + "step": 2623 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.8049, + "step": 2624 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.1055, + "step": 2625 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1843, + "step": 2626 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.6459, + "step": 2627 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.9643, + "step": 2628 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2267, + "step": 2629 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.0309, + "step": 2630 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 3.0506, + "step": 2631 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.1214, + "step": 2632 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4714, + "step": 2633 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.6101, + "step": 2634 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.0896, + "step": 2635 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5411, + "step": 2636 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.244, + "step": 2637 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.0498, + "step": 2638 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.4222, + "step": 2639 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2181, + "step": 2640 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8473, + "step": 2641 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.2569, + "step": 2642 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.6585, + "step": 2643 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8195, + "step": 2644 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3272, + "step": 2645 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.5701, + "step": 2646 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 1.8725, + "step": 2647 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.7784, + "step": 2648 + }, + { + "epoch": 0.26, + "learning_rate": 0.0001, + "loss": 2.3287, + "step": 2649 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8872, + "step": 2650 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.6799, + "step": 2651 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.7076, + "step": 2652 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9064, + "step": 2653 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5602, + "step": 2654 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3012, + "step": 2655 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.0489, + "step": 2656 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8506, + "step": 2657 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5072, + "step": 2658 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8981, + "step": 2659 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.46, + "step": 2660 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5255, + "step": 2661 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9271, + "step": 2662 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5037, + "step": 2663 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3731, + "step": 2664 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8286, + "step": 2665 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2512, + "step": 2666 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3177, + "step": 2667 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.277, + "step": 2668 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.7651, + "step": 2669 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5348, + "step": 2670 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8715, + "step": 2671 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4968, + "step": 2672 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2862, + "step": 2673 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8111, + "step": 2674 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.36, + "step": 2675 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.6746, + "step": 2676 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8331, + "step": 2677 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.023, + "step": 2678 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5534, + "step": 2679 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.7638, + "step": 2680 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1727, + "step": 2681 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.9473, + "step": 2682 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8473, + "step": 2683 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3997, + "step": 2684 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2055, + "step": 2685 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.6871, + "step": 2686 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.533, + "step": 2687 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4627, + "step": 2688 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9156, + "step": 2689 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.513, + "step": 2690 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2192, + "step": 2691 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8338, + "step": 2692 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2722, + "step": 2693 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4858, + "step": 2694 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9102, + "step": 2695 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4738, + "step": 2696 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5676, + "step": 2697 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8674, + "step": 2698 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5927, + "step": 2699 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.6758, + "step": 2700 + }, + { + "epoch": 0.27, + "eval_bleu": 13.863835920370443, + "eval_brevity_penalty": 0.539176210171706, + "eval_loss": 1.624898076057434, + "eval_runtime": 7.4513, + "eval_samples_per_second": 402.349, + "eval_steps_per_second": 0.805, + "step": 2700 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9651, + "step": 2701 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1462, + "step": 2702 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.5656, + "step": 2703 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.7988, + "step": 2704 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.0115, + "step": 2705 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4175, + "step": 2706 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8409, + "step": 2707 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3697, + "step": 2708 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3438, + "step": 2709 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9269, + "step": 2710 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8801, + "step": 2711 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1682, + "step": 2712 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.7794, + "step": 2713 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4518, + "step": 2714 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2913, + "step": 2715 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.868, + "step": 2716 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.6891, + "step": 2717 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.234, + "step": 2718 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8061, + "step": 2719 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3584, + "step": 2720 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4151, + "step": 2721 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1424, + "step": 2722 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.2956, + "step": 2723 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4912, + "step": 2724 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9556, + "step": 2725 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8468, + "step": 2726 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.4299, + "step": 2727 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9777, + "step": 2728 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8494, + "step": 2729 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.0229, + "step": 2730 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.8602, + "step": 2731 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.1653, + "step": 2732 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8164, + "step": 2733 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9626, + "step": 2734 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.169, + "step": 2735 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.0111, + "step": 2736 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1684, + "step": 2737 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 3.8685, + "step": 2738 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.8762, + "step": 2739 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.3602, + "step": 2740 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.0545, + "step": 2741 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.4603, + "step": 2742 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.5223, + "step": 2743 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.6152, + "step": 2744 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1623, + "step": 2745 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.5162, + "step": 2746 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9679, + "step": 2747 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 2.1641, + "step": 2748 + }, + { + "epoch": 0.27, + "learning_rate": 0.0001, + "loss": 1.9505, + "step": 2749 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.534, + "step": 2750 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2786, + "step": 2751 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8431, + "step": 2752 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4399, + "step": 2753 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 2754 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.7584, + "step": 2755 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.5661, + "step": 2756 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3485, + "step": 2757 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9777, + "step": 2758 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1535, + "step": 2759 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1461, + "step": 2760 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.7948, + "step": 2761 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1949, + "step": 2762 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1398, + "step": 2763 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0316, + "step": 2764 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2807, + "step": 2765 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3761, + "step": 2766 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.906, + "step": 2767 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2919, + "step": 2768 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.7088, + "step": 2769 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9165, + "step": 2770 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.7193, + "step": 2771 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3813, + "step": 2772 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8433, + "step": 2773 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1629, + "step": 2774 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.322, + "step": 2775 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0146, + "step": 2776 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4533, + "step": 2777 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3798, + "step": 2778 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.779, + "step": 2779 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1659, + "step": 2780 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2388, + "step": 2781 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8338, + "step": 2782 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.6989, + "step": 2783 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.5327, + "step": 2784 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9068, + "step": 2785 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.5963, + "step": 2786 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2415, + "step": 2787 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8287, + "step": 2788 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2033, + "step": 2789 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4825, + "step": 2790 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2009, + "step": 2791 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 3.2561, + "step": 2792 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3794, + "step": 2793 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2517, + "step": 2794 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.753, + "step": 2795 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1557, + "step": 2796 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8611, + "step": 2797 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2624, + "step": 2798 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.7598, + "step": 2799 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0391, + "step": 2800 + }, + { + "epoch": 0.28, + "eval_bleu": 13.964488248543509, + "eval_brevity_penalty": 0.539642664866601, + "eval_loss": 1.6204622983932495, + "eval_runtime": 6.837, + "eval_samples_per_second": 438.497, + "eval_steps_per_second": 0.878, + "step": 2800 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0224, + "step": 2801 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.6943, + "step": 2802 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.7899, + "step": 2803 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0805, + "step": 2804 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.5406, + "step": 2805 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.5512, + "step": 2806 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3428, + "step": 2807 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1577, + "step": 2808 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9174, + "step": 2809 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1419, + "step": 2810 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1879, + "step": 2811 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0097, + "step": 2812 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.7198, + "step": 2813 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0772, + "step": 2814 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8902, + "step": 2815 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8288, + "step": 2816 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8638, + "step": 2817 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.3581, + "step": 2818 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4262, + "step": 2819 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8949, + "step": 2820 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.4374, + "step": 2821 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1465, + "step": 2822 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4156, + "step": 2823 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9054, + "step": 2824 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3458, + "step": 2825 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1949, + "step": 2826 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0391, + "step": 2827 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.873, + "step": 2828 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2067, + "step": 2829 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8684, + "step": 2830 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.2735, + "step": 2831 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.8247, + "step": 2832 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9563, + "step": 2833 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.7176, + "step": 2834 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1221, + "step": 2835 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.7551, + "step": 2836 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.611, + "step": 2837 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.1525, + "step": 2838 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.5922, + "step": 2839 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.242, + "step": 2840 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.8536, + "step": 2841 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9868, + "step": 2842 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.3515, + "step": 2843 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.6684, + "step": 2844 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9572, + "step": 2845 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.9862, + "step": 2846 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.9761, + "step": 2847 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.0532, + "step": 2848 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 2.4243, + "step": 2849 + }, + { + "epoch": 0.28, + "learning_rate": 0.0001, + "loss": 1.9242, + "step": 2850 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8339, + "step": 2851 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.445, + "step": 2852 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3588, + "step": 2853 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8153, + "step": 2854 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.4385, + "step": 2855 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.4915, + "step": 2856 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8746, + "step": 2857 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9008, + "step": 2858 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0242, + "step": 2859 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8424, + "step": 2860 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.6345, + "step": 2861 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.9644, + "step": 2862 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8455, + "step": 2863 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 3.1621, + "step": 2864 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 3.0499, + "step": 2865 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.927, + "step": 2866 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 3.1277, + "step": 2867 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 3.0445, + "step": 2868 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.4934, + "step": 2869 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.9761, + "step": 2870 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3522, + "step": 2871 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7543, + "step": 2872 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0224, + "step": 2873 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9373, + "step": 2874 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9881, + "step": 2875 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.4796, + "step": 2876 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0401, + "step": 2877 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.943, + "step": 2878 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3951, + "step": 2879 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1096, + "step": 2880 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8659, + "step": 2881 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.404, + "step": 2882 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2184, + "step": 2883 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.913, + "step": 2884 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2753, + "step": 2885 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9299, + "step": 2886 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8664, + "step": 2887 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1, + "step": 2888 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9109, + "step": 2889 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9314, + "step": 2890 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2116, + "step": 2891 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0966, + "step": 2892 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0287, + "step": 2893 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9569, + "step": 2894 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1791, + "step": 2895 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9446, + "step": 2896 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3484, + "step": 2897 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.4544, + "step": 2898 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9695, + "step": 2899 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.8146, + "step": 2900 + }, + { + "epoch": 0.29, + "eval_bleu": 14.282286548496817, + "eval_brevity_penalty": 0.5408719507871882, + "eval_loss": 1.6210488080978394, + "eval_runtime": 5.3025, + "eval_samples_per_second": 565.39, + "eval_steps_per_second": 1.132, + "step": 2900 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.5336, + "step": 2901 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7021, + "step": 2902 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1043, + "step": 2903 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3149, + "step": 2904 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8358, + "step": 2905 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2479, + "step": 2906 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0099, + "step": 2907 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.2255, + "step": 2908 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0495, + "step": 2909 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1768, + "step": 2910 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7654, + "step": 2911 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.5822, + "step": 2912 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1127, + "step": 2913 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1581, + "step": 2914 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 3.0877, + "step": 2915 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2714, + "step": 2916 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8334, + "step": 2917 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2008, + "step": 2918 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7774, + "step": 2919 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8984, + "step": 2920 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0645, + "step": 2921 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9134, + "step": 2922 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7733, + "step": 2923 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8741, + "step": 2924 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.5764, + "step": 2925 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7561, + "step": 2926 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1104, + "step": 2927 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9612, + "step": 2928 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9531, + "step": 2929 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2112, + "step": 2930 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.117, + "step": 2931 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.924, + "step": 2932 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3918, + "step": 2933 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1327, + "step": 2934 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.7926, + "step": 2935 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9439, + "step": 2936 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.741, + "step": 2937 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.93, + "step": 2938 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.9166, + "step": 2939 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.828, + "step": 2940 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.5548, + "step": 2941 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3887, + "step": 2942 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.1728, + "step": 2943 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8735, + "step": 2944 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2865, + "step": 2945 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.2658, + "step": 2946 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.0997, + "step": 2947 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.53, + "step": 2948 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 2.3111, + "step": 2949 + }, + { + "epoch": 0.29, + "learning_rate": 0.0001, + "loss": 1.8799, + "step": 2950 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.8871, + "step": 2951 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9204, + "step": 2952 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.879, + "step": 2953 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9163, + "step": 2954 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.144, + "step": 2955 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8249, + "step": 2956 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.3169, + "step": 2957 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9161, + "step": 2958 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.5429, + "step": 2959 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.5541, + "step": 2960 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2409, + "step": 2961 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9073, + "step": 2962 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4466, + "step": 2963 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1062, + "step": 2964 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8839, + "step": 2965 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2325, + "step": 2966 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4599, + "step": 2967 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8861, + "step": 2968 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.9091, + "step": 2969 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.541, + "step": 2970 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.6889, + "step": 2971 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.8148, + "step": 2972 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0248, + "step": 2973 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8583, + "step": 2974 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1254, + "step": 2975 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0857, + "step": 2976 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8225, + "step": 2977 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.5335, + "step": 2978 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.407, + "step": 2979 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9384, + "step": 2980 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.6746, + "step": 2981 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.52, + "step": 2982 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.3266, + "step": 2983 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4301, + "step": 2984 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4035, + "step": 2985 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2118, + "step": 2986 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2763, + "step": 2987 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1721, + "step": 2988 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.9404, + "step": 2989 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.1498, + "step": 2990 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.2461, + "step": 2991 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7707, + "step": 2992 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.1602, + "step": 2993 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.9513, + "step": 2994 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8886, + "step": 2995 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.1707, + "step": 2996 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.4875, + "step": 2997 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.5309, + "step": 2998 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 3.0531, + "step": 2999 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.6602, + "step": 3000 + }, + { + "epoch": 0.3, + "eval_bleu": 13.966343617349223, + "eval_brevity_penalty": 0.5391338004675864, + "eval_loss": 1.621935486793518, + "eval_runtime": 5.2896, + "eval_samples_per_second": 566.771, + "eval_steps_per_second": 1.134, + "step": 3000 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.781, + "step": 3001 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9147, + "step": 3002 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4238, + "step": 3003 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8441, + "step": 3004 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4221, + "step": 3005 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.922, + "step": 3006 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9579, + "step": 3007 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4396, + "step": 3008 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0171, + "step": 3009 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.4966, + "step": 3010 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4401, + "step": 3011 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7315, + "step": 3012 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.6758, + "step": 3013 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.4604, + "step": 3014 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0772, + "step": 3015 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0404, + "step": 3016 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1652, + "step": 3017 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2105, + "step": 3018 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7803, + "step": 3019 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.6976, + "step": 3020 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7318, + "step": 3021 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8701, + "step": 3022 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.6924, + "step": 3023 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.7862, + "step": 3024 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0266, + "step": 3025 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.5913, + "step": 3026 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1918, + "step": 3027 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0041, + "step": 3028 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.337, + "step": 3029 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.6939, + "step": 3030 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0358, + "step": 3031 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0113, + "step": 3032 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.1461, + "step": 3033 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8363, + "step": 3034 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2556, + "step": 3035 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.5184, + "step": 3036 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.3975, + "step": 3037 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.3284, + "step": 3038 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.5966, + "step": 3039 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.0628, + "step": 3040 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.2123, + "step": 3041 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8589, + "step": 3042 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8939, + "step": 3043 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 2.0587, + "step": 3044 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.694, + "step": 3045 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7951, + "step": 3046 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.7639, + "step": 3047 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.972, + "step": 3048 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.8702, + "step": 3049 + }, + { + "epoch": 0.3, + "learning_rate": 0.0001, + "loss": 1.9436, + "step": 3050 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.094, + "step": 3051 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9248, + "step": 3052 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8406, + "step": 3053 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9875, + "step": 3054 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6449, + "step": 3055 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0384, + "step": 3056 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0411, + "step": 3057 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7478, + "step": 3058 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0531, + "step": 3059 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.5206, + "step": 3060 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8071, + "step": 3061 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0559, + "step": 3062 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.4754, + "step": 3063 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6652, + "step": 3064 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.366, + "step": 3065 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7846, + "step": 3066 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7517, + "step": 3067 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.366, + "step": 3068 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7607, + "step": 3069 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6736, + "step": 3070 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9681, + "step": 3071 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.8564, + "step": 3072 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8916, + "step": 3073 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2665, + "step": 3074 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1734, + "step": 3075 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8517, + "step": 3076 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6786, + "step": 3077 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8184, + "step": 3078 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7077, + "step": 3079 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2883, + "step": 3080 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0839, + "step": 3081 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9882, + "step": 3082 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.3056, + "step": 3083 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.3686, + "step": 3084 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7746, + "step": 3085 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.095, + "step": 3086 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1364, + "step": 3087 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9823, + "step": 3088 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0529, + "step": 3089 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1625, + "step": 3090 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.728, + "step": 3091 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.7287, + "step": 3092 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.6881, + "step": 3093 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6958, + "step": 3094 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.9366, + "step": 3095 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.6567, + "step": 3096 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7122, + "step": 3097 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1002, + "step": 3098 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0909, + "step": 3099 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7745, + "step": 3100 + }, + { + "epoch": 0.31, + "eval_bleu": 14.420611738531571, + "eval_brevity_penalty": 0.5413380574199849, + "eval_loss": 1.6088463068008423, + "eval_runtime": 5.3389, + "eval_samples_per_second": 561.54, + "eval_steps_per_second": 1.124, + "step": 3100 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.3099, + "step": 3101 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1004, + "step": 3102 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9274, + "step": 3103 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.4541, + "step": 3104 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9267, + "step": 3105 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7547, + "step": 3106 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2358, + "step": 3107 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2877, + "step": 3108 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6674, + "step": 3109 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 3.0757, + "step": 3110 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.8901, + "step": 3111 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8073, + "step": 3112 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 3.1601, + "step": 3113 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.9438, + "step": 3114 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7829, + "step": 3115 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 3.0668, + "step": 3116 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.7216, + "step": 3117 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8893, + "step": 3118 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.4179, + "step": 3119 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.5789, + "step": 3120 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8952, + "step": 3121 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.4218, + "step": 3122 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9802, + "step": 3123 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8899, + "step": 3124 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 3.2731, + "step": 3125 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0442, + "step": 3126 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1859, + "step": 3127 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1939, + "step": 3128 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.3014, + "step": 3129 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8066, + "step": 3130 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.6606, + "step": 3131 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.0798, + "step": 3132 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7939, + "step": 3133 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.9354, + "step": 3134 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.609, + "step": 3135 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7722, + "step": 3136 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.4947, + "step": 3137 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2092, + "step": 3138 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.8414, + "step": 3139 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2969, + "step": 3140 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.5864, + "step": 3141 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6678, + "step": 3142 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.6761, + "step": 3143 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.2098, + "step": 3144 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.6817, + "step": 3145 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.372, + "step": 3146 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1396, + "step": 3147 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 1.7407, + "step": 3148 + }, + { + "epoch": 0.31, + "learning_rate": 0.0001, + "loss": 2.1137, + "step": 3149 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8413, + "step": 3150 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8732, + "step": 3151 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.531, + "step": 3152 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3772, + "step": 3153 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8521, + "step": 3154 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2341, + "step": 3155 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1117, + "step": 3156 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.625, + "step": 3157 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2902, + "step": 3158 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2168, + "step": 3159 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7654, + "step": 3160 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3557, + "step": 3161 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1852, + "step": 3162 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8248, + "step": 3163 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7162, + "step": 3164 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.4408, + "step": 3165 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7047, + "step": 3166 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8433, + "step": 3167 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.474, + "step": 3168 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7669, + "step": 3169 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8151, + "step": 3170 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.5472, + "step": 3171 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.743, + "step": 3172 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3841, + "step": 3173 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2409, + "step": 3174 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.751, + "step": 3175 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7991, + "step": 3176 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3546, + "step": 3177 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8421, + "step": 3178 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8574, + "step": 3179 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8601, + "step": 3180 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.6807, + "step": 3181 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.5176, + "step": 3182 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8418, + "step": 3183 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7407, + "step": 3184 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3559, + "step": 3185 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1692, + "step": 3186 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7021, + "step": 3187 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.6226, + "step": 3188 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.0137, + "step": 3189 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8962, + "step": 3190 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7685, + "step": 3191 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8533, + "step": 3192 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7839, + "step": 3193 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1865, + "step": 3194 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3941, + "step": 3195 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7566, + "step": 3196 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3958, + "step": 3197 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2476, + "step": 3198 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7527, + "step": 3199 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3483, + "step": 3200 + }, + { + "epoch": 0.32, + "eval_bleu": 14.620844619546636, + "eval_brevity_penalty": 0.547135110819549, + "eval_loss": 1.604961633682251, + "eval_runtime": 5.4511, + "eval_samples_per_second": 549.985, + "eval_steps_per_second": 1.101, + "step": 3200 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.0575, + "step": 3201 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8593, + "step": 3202 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.4265, + "step": 3203 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.9308, + "step": 3204 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2524, + "step": 3205 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.9792, + "step": 3206 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2482, + "step": 3207 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7104, + "step": 3208 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.923, + "step": 3209 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2723, + "step": 3210 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8003, + "step": 3211 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1596, + "step": 3212 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3167, + "step": 3213 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.9123, + "step": 3214 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3859, + "step": 3215 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.8009, + "step": 3216 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7706, + "step": 3217 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 3.145, + "step": 3218 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 3.1341, + "step": 3219 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8576, + "step": 3220 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 3.1153, + "step": 3221 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.8237, + "step": 3222 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8828, + "step": 3223 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.9288, + "step": 3224 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.5493, + "step": 3225 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.895, + "step": 3226 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2003, + "step": 3227 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8842, + "step": 3228 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8751, + "step": 3229 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.5746, + "step": 3230 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.9632, + "step": 3231 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7495, + "step": 3232 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.0319, + "step": 3233 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.1784, + "step": 3234 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8086, + "step": 3235 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.2804, + "step": 3236 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.0507, + "step": 3237 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8541, + "step": 3238 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.3848, + "step": 3239 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8033, + "step": 3240 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.5414, + "step": 3241 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.0659, + "step": 3242 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.2783, + "step": 3243 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.7855, + "step": 3244 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.8976, + "step": 3245 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.634, + "step": 3246 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 1.812, + "step": 3247 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.7791, + "step": 3248 + }, + { + "epoch": 0.32, + "learning_rate": 0.0001, + "loss": 2.343, + "step": 3249 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8055, + "step": 3250 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.186, + "step": 3251 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.2866, + "step": 3252 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7455, + "step": 3253 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.507, + "step": 3254 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9823, + "step": 3255 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7936, + "step": 3256 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0293, + "step": 3257 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.2979, + "step": 3258 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7703, + "step": 3259 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3216, + "step": 3260 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8889, + "step": 3261 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6865, + "step": 3262 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0859, + "step": 3263 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4383, + "step": 3264 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6498, + "step": 3265 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7754, + "step": 3266 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6976, + "step": 3267 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7743, + "step": 3268 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0811, + "step": 3269 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.337, + "step": 3270 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.961, + "step": 3271 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0543, + "step": 3272 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4788, + "step": 3273 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6638, + "step": 3274 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9575, + "step": 3275 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0609, + "step": 3276 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7696, + "step": 3277 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.2549, + "step": 3278 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6691, + "step": 3279 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7265, + "step": 3280 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1043, + "step": 3281 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.2514, + "step": 3282 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6707, + "step": 3283 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0511, + "step": 3284 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1147, + "step": 3285 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7958, + "step": 3286 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1322, + "step": 3287 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.5813, + "step": 3288 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7871, + "step": 3289 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.5037, + "step": 3290 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.69, + "step": 3291 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6886, + "step": 3292 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.4947, + "step": 3293 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0558, + "step": 3294 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7589, + "step": 3295 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3741, + "step": 3296 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.6349, + "step": 3297 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8582, + "step": 3298 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4344, + "step": 3299 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9911, + "step": 3300 + }, + { + "epoch": 0.33, + "eval_bleu": 14.545786677463147, + "eval_brevity_penalty": 0.5395578620231306, + "eval_loss": 1.6004188060760498, + "eval_runtime": 5.1565, + "eval_samples_per_second": 581.402, + "eval_steps_per_second": 1.164, + "step": 3300 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8383, + "step": 3301 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0777, + "step": 3302 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.2243, + "step": 3303 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7069, + "step": 3304 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3142, + "step": 3305 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.9236, + "step": 3306 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9108, + "step": 3307 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.7582, + "step": 3308 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3181, + "step": 3309 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.034, + "step": 3310 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8313, + "step": 3311 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1222, + "step": 3312 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8981, + "step": 3313 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.6722, + "step": 3314 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.6003, + "step": 3315 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9784, + "step": 3316 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1756, + "step": 3317 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.5584, + "step": 3318 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9952, + "step": 3319 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 3.0296, + "step": 3320 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 3.1736, + "step": 3321 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9346, + "step": 3322 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 3.063, + "step": 3323 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 3.1613, + "step": 3324 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.007, + "step": 3325 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4034, + "step": 3326 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3971, + "step": 3327 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9548, + "step": 3328 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.3899, + "step": 3329 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1184, + "step": 3330 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9454, + "step": 3331 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1171, + "step": 3332 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.294, + "step": 3333 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8842, + "step": 3334 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4096, + "step": 3335 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.6243, + "step": 3336 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.933, + "step": 3337 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.244, + "step": 3338 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.6739, + "step": 3339 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8948, + "step": 3340 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0398, + "step": 3341 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.6603, + "step": 3342 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.8892, + "step": 3343 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.785, + "step": 3344 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.1423, + "step": 3345 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.7432, + "step": 3346 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.4906, + "step": 3347 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 1.9978, + "step": 3348 + }, + { + "epoch": 0.33, + "learning_rate": 0.0001, + "loss": 2.0527, + "step": 3349 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.1473, + "step": 3350 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0178, + "step": 3351 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9633, + "step": 3352 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9112, + "step": 3353 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.3703, + "step": 3354 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8872, + "step": 3355 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2882, + "step": 3356 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2772, + "step": 3357 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8794, + "step": 3358 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5967, + "step": 3359 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7609, + "step": 3360 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9124, + "step": 3361 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8113, + "step": 3362 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6177, + "step": 3363 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7512, + "step": 3364 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8013, + "step": 3365 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9221, + "step": 3366 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6399, + "step": 3367 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.026, + "step": 3368 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6645, + "step": 3369 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7825, + "step": 3370 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2767, + "step": 3371 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0245, + "step": 3372 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8357, + "step": 3373 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8468, + "step": 3374 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0965, + "step": 3375 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7955, + "step": 3376 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5346, + "step": 3377 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.079, + "step": 3378 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6947, + "step": 3379 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0402, + "step": 3380 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0594, + "step": 3381 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7381, + "step": 3382 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0626, + "step": 3383 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.7423, + "step": 3384 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7245, + "step": 3385 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0937, + "step": 3386 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.3572, + "step": 3387 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6983, + "step": 3388 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.1058, + "step": 3389 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0234, + "step": 3390 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7461, + "step": 3391 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5569, + "step": 3392 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 3393 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7732, + "step": 3394 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9792, + "step": 3395 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2825, + "step": 3396 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6514, + "step": 3397 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4967, + "step": 3398 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.8508, + "step": 3399 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8973, + "step": 3400 + }, + { + "epoch": 0.34, + "eval_bleu": 14.538680062474546, + "eval_brevity_penalty": 0.5399818446526522, + "eval_loss": 1.598526954650879, + "eval_runtime": 5.8692, + "eval_samples_per_second": 510.799, + "eval_steps_per_second": 1.022, + "step": 3400 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.7718, + "step": 3401 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.487, + "step": 3402 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6324, + "step": 3403 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.1515, + "step": 3404 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2873, + "step": 3405 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.7582, + "step": 3406 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.652, + "step": 3407 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.347, + "step": 3408 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0248, + "step": 3409 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5255, + "step": 3410 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6962, + "step": 3411 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8429, + "step": 3412 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0814, + "step": 3413 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.6686, + "step": 3414 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.809, + "step": 3415 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 3.079, + "step": 3416 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 3.1711, + "step": 3417 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.851, + "step": 3418 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.9919, + "step": 3419 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.9151, + "step": 3420 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.6729, + "step": 3421 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.3018, + "step": 3422 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4799, + "step": 3423 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8667, + "step": 3424 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4692, + "step": 3425 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.045, + "step": 3426 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8677, + "step": 3427 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0003, + "step": 3428 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0975, + "step": 3429 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9362, + "step": 3430 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2567, + "step": 3431 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2926, + "step": 3432 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8704, + "step": 3433 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5529, + "step": 3434 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4947, + "step": 3435 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.0282, + "step": 3436 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4607, + "step": 3437 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8707, + "step": 3438 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9479, + "step": 3439 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.2549, + "step": 3440 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.2411, + "step": 3441 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9479, + "step": 3442 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.7374, + "step": 3443 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.6703, + "step": 3444 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9919, + "step": 3445 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.5436, + "step": 3446 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.1829, + "step": 3447 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.8274, + "step": 3448 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 2.4872, + "step": 3449 + }, + { + "epoch": 0.34, + "learning_rate": 0.0001, + "loss": 1.9932, + "step": 3450 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9728, + "step": 3451 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9295, + "step": 3452 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5226, + "step": 3453 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9708, + "step": 3454 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.1385, + "step": 3455 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9954, + "step": 3456 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.6094, + "step": 3457 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2207, + "step": 3458 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9985, + "step": 3459 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8881, + "step": 3460 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2793, + "step": 3461 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8886, + "step": 3462 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9401, + "step": 3463 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.3429, + "step": 3464 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.817, + "step": 3465 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8407, + "step": 3466 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9389, + "step": 3467 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.6985, + "step": 3468 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.7173, + "step": 3469 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8217, + "step": 3470 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.5245, + "step": 3471 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0167, + "step": 3472 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1425, + "step": 3473 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.14, + "step": 3474 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.72, + "step": 3475 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0765, + "step": 3476 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.7573, + "step": 3477 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9736, + "step": 3478 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.9366, + "step": 3479 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.7878, + "step": 3480 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9462, + "step": 3481 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0942, + "step": 3482 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5011, + "step": 3483 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8815, + "step": 3484 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.026, + "step": 3485 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2638, + "step": 3486 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.041, + "step": 3487 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1354, + "step": 3488 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5599, + "step": 3489 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.615, + "step": 3490 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.4697, + "step": 3491 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.8915, + "step": 3492 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9561, + "step": 3493 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2147, + "step": 3494 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2307, + "step": 3495 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.6482, + "step": 3496 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.6752, + "step": 3497 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5389, + "step": 3498 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.5577, + "step": 3499 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.6956, + "step": 3500 + }, + { + "epoch": 0.35, + "eval_bleu": 14.748194332756642, + "eval_brevity_penalty": 0.5457824024959471, + "eval_loss": 1.600500464439392, + "eval_runtime": 5.1969, + "eval_samples_per_second": 576.883, + "eval_steps_per_second": 1.155, + "step": 3500 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1874, + "step": 3501 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9882, + "step": 3502 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.395, + "step": 3503 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1759, + "step": 3504 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8925, + "step": 3505 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.4132, + "step": 3506 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5478, + "step": 3507 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8937, + "step": 3508 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.8198, + "step": 3509 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.2109, + "step": 3510 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8953, + "step": 3511 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.6913, + "step": 3512 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9242, + "step": 3513 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8429, + "step": 3514 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.4491, + "step": 3515 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.464, + "step": 3516 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9038, + "step": 3517 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2167, + "step": 3518 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.5103, + "step": 3519 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0341, + "step": 3520 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.0549, + "step": 3521 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.0585, + "step": 3522 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9063, + "step": 3523 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.1099, + "step": 3524 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.8188, + "step": 3525 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9136, + "step": 3526 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 3.0305, + "step": 3527 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.673, + "step": 3528 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9332, + "step": 3529 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.3152, + "step": 3530 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0484, + "step": 3531 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8126, + "step": 3532 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.4517, + "step": 3533 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0016, + "step": 3534 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.615, + "step": 3535 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1265, + "step": 3536 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9904, + "step": 3537 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.8823, + "step": 3538 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.2567, + "step": 3539 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1284, + "step": 3540 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.3035, + "step": 3541 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.1365, + "step": 3542 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.0518, + "step": 3543 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9256, + "step": 3544 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.274, + "step": 3545 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.203, + "step": 3546 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.9653, + "step": 3547 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.4562, + "step": 3548 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 2.6329, + "step": 3549 + }, + { + "epoch": 0.35, + "learning_rate": 0.0001, + "loss": 1.958, + "step": 3550 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.7097, + "step": 3551 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0235, + "step": 3552 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.7812, + "step": 3553 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.179, + "step": 3554 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2965, + "step": 3555 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.828, + "step": 3556 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.7772, + "step": 3557 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2238, + "step": 3558 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9943, + "step": 3559 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0971, + "step": 3560 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3322, + "step": 3561 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9164, + "step": 3562 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2186, + "step": 3563 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0976, + "step": 3564 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9146, + "step": 3565 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.5368, + "step": 3566 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8799, + "step": 3567 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9241, + "step": 3568 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.7109, + "step": 3569 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8561, + "step": 3570 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9013, + "step": 3571 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.997, + "step": 3572 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9612, + "step": 3573 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9741, + "step": 3574 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.6449, + "step": 3575 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.4013, + "step": 3576 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8477, + "step": 3577 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1041, + "step": 3578 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.324, + "step": 3579 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.864, + "step": 3580 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.4374, + "step": 3581 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2655, + "step": 3582 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.7414, + "step": 3583 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.7818, + "step": 3584 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1659, + "step": 3585 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9093, + "step": 3586 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2632, + "step": 3587 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3827, + "step": 3588 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0443, + "step": 3589 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.5061, + "step": 3590 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.8908, + "step": 3591 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0319, + "step": 3592 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.7087, + "step": 3593 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9306, + "step": 3594 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0506, + "step": 3595 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0675, + "step": 3596 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.2971, + "step": 3597 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8948, + "step": 3598 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.4952, + "step": 3599 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.322, + "step": 3600 + }, + { + "epoch": 0.36, + "eval_bleu": 14.732214976907335, + "eval_brevity_penalty": 0.5448307857504349, + "eval_loss": 1.5948790311813354, + "eval_runtime": 5.2434, + "eval_samples_per_second": 571.77, + "eval_steps_per_second": 1.144, + "step": 3600 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8085, + "step": 3601 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3131, + "step": 3602 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3952, + "step": 3603 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.6009, + "step": 3604 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.9484, + "step": 3605 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3038, + "step": 3606 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.6276, + "step": 3607 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3665, + "step": 3608 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8395, + "step": 3609 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.4071, + "step": 3610 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3185, + "step": 3611 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9111, + "step": 3612 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.91, + "step": 3613 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1315, + "step": 3614 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.368, + "step": 3615 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9693, + "step": 3616 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.6541, + "step": 3617 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 3.1294, + "step": 3618 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.3434, + "step": 3619 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 3.1252, + "step": 3620 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.6959, + "step": 3621 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.4405, + "step": 3622 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 3.0634, + "step": 3623 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3014, + "step": 3624 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.2932, + "step": 3625 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3599, + "step": 3626 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.7816, + "step": 3627 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.2136, + "step": 3628 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1579, + "step": 3629 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.5903, + "step": 3630 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.1886, + "step": 3631 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.3039, + "step": 3632 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.6136, + "step": 3633 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.238, + "step": 3634 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.7511, + "step": 3635 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.5028, + "step": 3636 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.2695, + "step": 3637 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.006, + "step": 3638 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 0.764, + "step": 3639 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.3142, + "step": 3640 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.0587, + "step": 3641 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.23, + "step": 3642 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9797, + "step": 3643 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.633, + "step": 3644 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.8759, + "step": 3645 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.8904, + "step": 3646 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1273, + "step": 3647 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.1845, + "step": 3648 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 1.9673, + "step": 3649 + }, + { + "epoch": 0.36, + "learning_rate": 0.0001, + "loss": 2.499, + "step": 3650 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.1962, + "step": 3651 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9269, + "step": 3652 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.015, + "step": 3653 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8662, + "step": 3654 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.2178, + "step": 3655 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3961, + "step": 3656 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.4668, + "step": 3657 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0202, + "step": 3658 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3099, + "step": 3659 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.4358, + "step": 3660 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0094, + "step": 3661 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 3662 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.4824, + "step": 3663 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9572, + "step": 3664 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3709, + "step": 3665 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0941, + "step": 3666 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9159, + "step": 3667 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8252, + "step": 3668 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8607, + "step": 3669 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9351, + "step": 3670 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2617, + "step": 3671 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.975, + "step": 3672 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0099, + "step": 3673 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.5826, + "step": 3674 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8567, + "step": 3675 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3732, + "step": 3676 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8121, + "step": 3677 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0167, + "step": 3678 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0404, + "step": 3679 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.7477, + "step": 3680 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.7277, + "step": 3681 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0234, + "step": 3682 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9001, + "step": 3683 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3445, + "step": 3684 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0663, + "step": 3685 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9121, + "step": 3686 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.4259, + "step": 3687 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.935, + "step": 3688 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.8068, + "step": 3689 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9969, + "step": 3690 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9816, + "step": 3691 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.6511, + "step": 3692 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2229, + "step": 3693 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8795, + "step": 3694 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9122, + "step": 3695 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.1401, + "step": 3696 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0327, + "step": 3697 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0857, + "step": 3698 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.6031, + "step": 3699 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.5147, + "step": 3700 + }, + { + "epoch": 0.37, + "eval_bleu": 14.845635060416546, + "eval_brevity_penalty": 0.543138020221705, + "eval_loss": 1.5966026782989502, + "eval_runtime": 5.2079, + "eval_samples_per_second": 575.659, + "eval_steps_per_second": 1.152, + "step": 3700 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3764, + "step": 3701 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.8694, + "step": 3702 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0202, + "step": 3703 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.9484, + "step": 3704 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.8469, + "step": 3705 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8384, + "step": 3706 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.9109, + "step": 3707 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.978, + "step": 3708 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.5561, + "step": 3709 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3899, + "step": 3710 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0797, + "step": 3711 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9046, + "step": 3712 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3088, + "step": 3713 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0608, + "step": 3714 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.6663, + "step": 3715 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.5286, + "step": 3716 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.895, + "step": 3717 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.2717, + "step": 3718 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2154, + "step": 3719 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0477, + "step": 3720 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.642, + "step": 3721 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 3.0763, + "step": 3722 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.8053, + "step": 3723 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8616, + "step": 3724 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 3.0168, + "step": 3725 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.652, + "step": 3726 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.4295, + "step": 3727 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.9594, + "step": 3728 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.5378, + "step": 3729 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8155, + "step": 3730 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2274, + "step": 3731 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.6688, + "step": 3732 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8926, + "step": 3733 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0176, + "step": 3734 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0746, + "step": 3735 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.824, + "step": 3736 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.3049, + "step": 3737 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.9598, + "step": 3738 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.0255, + "step": 3739 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2435, + "step": 3740 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2041, + "step": 3741 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.8998, + "step": 3742 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2591, + "step": 3743 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2329, + "step": 3744 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.7942, + "step": 3745 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.2326, + "step": 3746 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.1661, + "step": 3747 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 1.6769, + "step": 3748 + }, + { + "epoch": 0.37, + "learning_rate": 0.0001, + "loss": 2.5297, + "step": 3749 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5332, + "step": 3750 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8304, + "step": 3751 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5787, + "step": 3752 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1725, + "step": 3753 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8061, + "step": 3754 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2126, + "step": 3755 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5238, + "step": 3756 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8354, + "step": 3757 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0227, + "step": 3758 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.6649, + "step": 3759 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8807, + "step": 3760 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5565, + "step": 3761 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9235, + "step": 3762 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9877, + "step": 3763 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5491, + "step": 3764 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4823, + "step": 3765 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.853, + "step": 3766 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.201, + "step": 3767 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1985, + "step": 3768 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8485, + "step": 3769 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.7525, + "step": 3770 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9865, + "step": 3771 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8339, + "step": 3772 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.718, + "step": 3773 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2211, + "step": 3774 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8974, + "step": 3775 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1757, + "step": 3776 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5585, + "step": 3777 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9343, + "step": 3778 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0168, + "step": 3779 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.183, + "step": 3780 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8863, + "step": 3781 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.7707, + "step": 3782 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.4545, + "step": 3783 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.892, + "step": 3784 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.681, + "step": 3785 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8666, + "step": 3786 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8404, + "step": 3787 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2926, + "step": 3788 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4214, + "step": 3789 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4014, + "step": 3790 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3801, + "step": 3791 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3247, + "step": 3792 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.943, + "step": 3793 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.7039, + "step": 3794 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0311, + "step": 3795 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0814, + "step": 3796 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.874, + "step": 3797 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4931, + "step": 3798 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.112, + "step": 3799 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0606, + "step": 3800 + }, + { + "epoch": 0.38, + "eval_bleu": 14.626695077194999, + "eval_brevity_penalty": 0.5333162677578793, + "eval_loss": 1.5899326801300049, + "eval_runtime": 5.7659, + "eval_samples_per_second": 519.955, + "eval_steps_per_second": 1.041, + "step": 3800 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.7061, + "step": 3801 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1833, + "step": 3802 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2684, + "step": 3803 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2682, + "step": 3804 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0923, + "step": 3805 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1682, + "step": 3806 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 3.101, + "step": 3807 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0289, + "step": 3808 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.9048, + "step": 3809 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.8236, + "step": 3810 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1407, + "step": 3811 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4578, + "step": 3812 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5375, + "step": 3813 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.954, + "step": 3814 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.556, + "step": 3815 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.7044, + "step": 3816 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.2513, + "step": 3817 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 3.0537, + "step": 3818 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4618, + "step": 3819 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9476, + "step": 3820 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 3.0608, + "step": 3821 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 3.0674, + "step": 3822 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8495, + "step": 3823 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 3.1352, + "step": 3824 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.9834, + "step": 3825 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9102, + "step": 3826 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.2574, + "step": 3827 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.6981, + "step": 3828 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4089, + "step": 3829 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.5559, + "step": 3830 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3479, + "step": 3831 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9999, + "step": 3832 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.1441, + "step": 3833 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.033, + "step": 3834 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9182, + "step": 3835 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3364, + "step": 3836 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3361, + "step": 3837 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.0478, + "step": 3838 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3401, + "step": 3839 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.9105, + "step": 3840 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.97, + "step": 3841 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.8001, + "step": 3842 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.4296, + "step": 3843 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.9205, + "step": 3844 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.3857, + "step": 3845 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.3802, + "step": 3846 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 1.8966, + "step": 3847 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.8024, + "step": 3848 + }, + { + "epoch": 0.38, + "learning_rate": 0.0001, + "loss": 2.7648, + "step": 3849 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4336, + "step": 3850 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3951, + "step": 3851 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.486, + "step": 3852 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9942, + "step": 3853 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4338, + "step": 3854 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.7109, + "step": 3855 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9259, + "step": 3856 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0836, + "step": 3857 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2553, + "step": 3858 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9558, + "step": 3859 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3509, + "step": 3860 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.5669, + "step": 3861 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8059, + "step": 3862 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4004, + "step": 3863 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2513, + "step": 3864 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9221, + "step": 3865 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8469, + "step": 3866 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8616, + "step": 3867 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0499, + "step": 3868 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3161, + "step": 3869 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0923, + "step": 3870 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8845, + "step": 3871 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3124, + "step": 3872 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.885, + "step": 3873 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8348, + "step": 3874 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.7028, + "step": 3875 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3761, + "step": 3876 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.7012, + "step": 3877 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2399, + "step": 3878 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2698, + "step": 3879 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1052, + "step": 3880 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2876, + "step": 3881 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.74, + "step": 3882 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8818, + "step": 3883 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6813, + "step": 3884 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8201, + "step": 3885 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0793, + "step": 3886 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2929, + "step": 3887 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3894, + "step": 3888 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1542, + "step": 3889 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2366, + "step": 3890 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6717, + "step": 3891 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1585, + "step": 3892 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0742, + "step": 3893 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4162, + "step": 3894 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0771, + "step": 3895 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2847, + "step": 3896 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.8506, + "step": 3897 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1371, + "step": 3898 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.0443, + "step": 3899 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.0341, + "step": 3900 + }, + { + "epoch": 0.39, + "eval_bleu": 14.770451757592893, + "eval_brevity_penalty": 0.5414016099766464, + "eval_loss": 1.5841609239578247, + "eval_runtime": 10.9609, + "eval_samples_per_second": 273.518, + "eval_steps_per_second": 0.547, + "step": 3900 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0456, + "step": 3901 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6778, + "step": 3902 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4454, + "step": 3903 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1268, + "step": 3904 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2288, + "step": 3905 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.5674, + "step": 3906 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8039, + "step": 3907 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3922, + "step": 3908 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6594, + "step": 3909 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1631, + "step": 3910 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3058, + "step": 3911 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.3976, + "step": 3912 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9826, + "step": 3913 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.869, + "step": 3914 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.1908, + "step": 3915 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0264, + "step": 3916 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.1021, + "step": 3917 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.2056, + "step": 3918 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9982, + "step": 3919 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.1253, + "step": 3920 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 3.3291, + "step": 3921 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0104, + "step": 3922 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6752, + "step": 3923 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4003, + "step": 3924 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.9038, + "step": 3925 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0581, + "step": 3926 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1465, + "step": 3927 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.7895, + "step": 3928 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2074, + "step": 3929 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0634, + "step": 3930 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.6521, + "step": 3931 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.0052, + "step": 3932 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.8443, + "step": 3933 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.3851, + "step": 3934 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.4448, + "step": 3935 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.2103, + "step": 3936 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.338, + "step": 3937 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.2592, + "step": 3938 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1777, + "step": 3939 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.1226, + "step": 3940 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.7457, + "step": 3941 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.5089, + "step": 3942 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.3341, + "step": 3943 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.6101, + "step": 3944 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.7976, + "step": 3945 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.2037, + "step": 3946 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 2.4248, + "step": 3947 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.4127, + "step": 3948 + }, + { + "epoch": 0.39, + "learning_rate": 0.0001, + "loss": 1.1688, + "step": 3949 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0029, + "step": 3950 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.1507, + "step": 3951 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0482, + "step": 3952 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3944, + "step": 3953 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5544, + "step": 3954 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.103, + "step": 3955 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.5824, + "step": 3956 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.023, + "step": 3957 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0163, + "step": 3958 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.9068, + "step": 3959 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.7832, + "step": 3960 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.9331, + "step": 3961 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1701, + "step": 3962 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.1406, + "step": 3963 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0827, + "step": 3964 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0399, + "step": 3965 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.3141, + "step": 3966 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0252, + "step": 3967 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2015, + "step": 3968 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.4536, + "step": 3969 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.9787, + "step": 3970 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.7735, + "step": 3971 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.3565, + "step": 3972 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.969, + "step": 3973 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.505, + "step": 3974 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.764, + "step": 3975 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 0.9575, + "step": 3976 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1805, + "step": 3977 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.3392, + "step": 3978 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3806, + "step": 3979 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1971, + "step": 3980 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2687, + "step": 3981 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.9001, + "step": 3982 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3895, + "step": 3983 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.4129, + "step": 3984 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8893, + "step": 3985 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0997, + "step": 3986 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2968, + "step": 3987 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8864, + "step": 3988 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.8287, + "step": 3989 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.8796, + "step": 3990 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0072, + "step": 3991 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.4827, + "step": 3992 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3897, + "step": 3993 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.9394, + "step": 3994 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3937, + "step": 3995 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1206, + "step": 3996 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8507, + "step": 3997 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 3.1305, + "step": 3998 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.9623, + "step": 3999 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5069, + "step": 4000 + }, + { + "epoch": 0.4, + "eval_bleu": 14.686050212073425, + "eval_brevity_penalty": 0.5372245496906878, + "eval_loss": 1.5911463499069214, + "eval_runtime": 11.6225, + "eval_samples_per_second": 257.949, + "eval_steps_per_second": 0.516, + "step": 4000 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 3.0285, + "step": 4001 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.9908, + "step": 4002 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5798, + "step": 4003 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.9695, + "step": 4004 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.7355, + "step": 4005 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.6223, + "step": 4006 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2823, + "step": 4007 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0771, + "step": 4008 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.7987, + "step": 4009 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2147, + "step": 4010 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3845, + "step": 4011 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8114, + "step": 4012 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0232, + "step": 4013 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5838, + "step": 4014 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.4164, + "step": 4015 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1992, + "step": 4016 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.3321, + "step": 4017 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.2051, + "step": 4018 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.6004, + "step": 4019 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0063, + "step": 4020 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.6115, + "step": 4021 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.7031, + "step": 4022 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8901, + "step": 4023 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.4298, + "step": 4024 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3845, + "step": 4025 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0913, + "step": 4026 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.6775, + "step": 4027 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.2222, + "step": 4028 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.4184, + "step": 4029 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0102, + "step": 4030 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0911, + "step": 4031 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0438, + "step": 4032 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.0369, + "step": 4033 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.3104, + "step": 4034 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.9678, + "step": 4035 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.2481, + "step": 4036 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1247, + "step": 4037 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5403, + "step": 4038 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.2313, + "step": 4039 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8598, + "step": 4040 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.071, + "step": 4041 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0493, + "step": 4042 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.5472, + "step": 4043 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.4011, + "step": 4044 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.0526, + "step": 4045 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1733, + "step": 4046 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.363, + "step": 4047 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 1.8964, + "step": 4048 + }, + { + "epoch": 0.4, + "learning_rate": 0.0001, + "loss": 2.1896, + "step": 4049 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7369, + "step": 4050 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.3942, + "step": 4051 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7698, + "step": 4052 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6351, + "step": 4053 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9609, + "step": 4054 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4085, + "step": 4055 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.6121, + "step": 4056 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.894, + "step": 4057 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0812, + "step": 4058 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4848, + "step": 4059 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0687, + "step": 4060 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.5588, + "step": 4061 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6978, + "step": 4062 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0462, + "step": 4063 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0926, + "step": 4064 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.7043, + "step": 4065 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9988, + "step": 4066 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2433, + "step": 4067 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2418, + "step": 4068 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7651, + "step": 4069 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.023, + "step": 4070 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.5605, + "step": 4071 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8452, + "step": 4072 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.8942, + "step": 4073 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 3.0399, + "step": 4074 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9686, + "step": 4075 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4664, + "step": 4076 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3024, + "step": 4077 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9931, + "step": 4078 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1566, + "step": 4079 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3998, + "step": 4080 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9648, + "step": 4081 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1695, + "step": 4082 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2773, + "step": 4083 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.5362, + "step": 4084 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 3.0345, + "step": 4085 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.9132, + "step": 4086 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7648, + "step": 4087 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 3.2312, + "step": 4088 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.8208, + "step": 4089 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7697, + "step": 4090 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3736, + "step": 4091 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1384, + "step": 4092 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.5292, + "step": 4093 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.386, + "step": 4094 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6694, + "step": 4095 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.5855, + "step": 4096 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2708, + "step": 4097 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0832, + "step": 4098 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9135, + "step": 4099 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.339, + "step": 4100 + }, + { + "epoch": 0.41, + "eval_bleu": 14.697005586210231, + "eval_brevity_penalty": 0.5481490991310788, + "eval_loss": 1.5948783159255981, + "eval_runtime": 8.6973, + "eval_samples_per_second": 344.706, + "eval_steps_per_second": 0.69, + "step": 4100 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9295, + "step": 4101 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8378, + "step": 4102 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3721, + "step": 4103 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.1377, + "step": 4104 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.678, + "step": 4105 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6697, + "step": 4106 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6095, + "step": 4107 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3532, + "step": 4108 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.7703, + "step": 4109 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.3253, + "step": 4110 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9218, + "step": 4111 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.277, + "step": 4112 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4254, + "step": 4113 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.5988, + "step": 4114 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4541, + "step": 4115 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2412, + "step": 4116 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8044, + "step": 4117 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9049, + "step": 4118 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.7696, + "step": 4119 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.09, + "step": 4120 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.4917, + "step": 4121 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2147, + "step": 4122 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9005, + "step": 4123 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2162, + "step": 4124 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.5243, + "step": 4125 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6065, + "step": 4126 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9945, + "step": 4127 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7898, + "step": 4128 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.0964, + "step": 4129 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7667, + "step": 4130 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7287, + "step": 4131 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9524, + "step": 4132 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9651, + "step": 4133 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2725, + "step": 4134 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8171, + "step": 4135 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1706, + "step": 4136 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8687, + "step": 4137 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.5307, + "step": 4138 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1611, + "step": 4139 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.2539, + "step": 4140 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8746, + "step": 4141 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9606, + "step": 4142 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.6876, + "step": 4143 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.9433, + "step": 4144 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1391, + "step": 4145 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.1712, + "step": 4146 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.8017, + "step": 4147 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.316, + "step": 4148 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 2.068, + "step": 4149 + }, + { + "epoch": 0.41, + "learning_rate": 0.0001, + "loss": 1.7629, + "step": 4150 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1851, + "step": 4151 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2948, + "step": 4152 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8891, + "step": 4153 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.56, + "step": 4154 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2572, + "step": 4155 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8018, + "step": 4156 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.5965, + "step": 4157 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8712, + "step": 4158 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8791, + "step": 4159 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.8031, + "step": 4160 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.3483, + "step": 4161 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.4572, + "step": 4162 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0614, + "step": 4163 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.5765, + "step": 4164 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8952, + "step": 4165 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.9548, + "step": 4166 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.7895, + "step": 4167 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1122, + "step": 4168 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.3291, + "step": 4169 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.368, + "step": 4170 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9481, + "step": 4171 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.5808, + "step": 4172 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9914, + "step": 4173 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8019, + "step": 4174 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.6125, + "step": 4175 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2833, + "step": 4176 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.665, + "step": 4177 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0429, + "step": 4178 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.8801, + "step": 4179 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7369, + "step": 4180 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 3.1061, + "step": 4181 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.8485, + "step": 4182 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7703, + "step": 4183 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.9913, + "step": 4184 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.8957, + "step": 4185 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8096, + "step": 4186 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1947, + "step": 4187 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9841, + "step": 4188 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.4972, + "step": 4189 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1314, + "step": 4190 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9444, + "step": 4191 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8731, + "step": 4192 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.6633, + "step": 4193 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0262, + "step": 4194 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9543, + "step": 4195 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2105, + "step": 4196 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.5317, + "step": 4197 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6831, + "step": 4198 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.6637, + "step": 4199 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.5221, + "step": 4200 + }, + { + "epoch": 0.42, + "eval_bleu": 14.69957320388323, + "eval_brevity_penalty": 0.5403421677509378, + "eval_loss": 1.5870023965835571, + "eval_runtime": 6.2708, + "eval_samples_per_second": 478.09, + "eval_steps_per_second": 0.957, + "step": 4200 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6889, + "step": 4201 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.7953, + "step": 4202 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2534, + "step": 4203 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.4729, + "step": 4204 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1521, + "step": 4205 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.4224, + "step": 4206 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8306, + "step": 4207 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8834, + "step": 4208 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7062, + "step": 4209 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6715, + "step": 4210 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2041, + "step": 4211 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9246, + "step": 4212 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8108, + "step": 4213 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.322, + "step": 4214 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2465, + "step": 4215 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0243, + "step": 4216 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7345, + "step": 4217 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.3938, + "step": 4218 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6417, + "step": 4219 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7489, + "step": 4220 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2272, + "step": 4221 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0412, + "step": 4222 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.6904, + "step": 4223 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9637, + "step": 4224 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6057, + "step": 4225 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9767, + "step": 4226 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.5662, + "step": 4227 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7479, + "step": 4228 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1309, + "step": 4229 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7885, + "step": 4230 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6081, + "step": 4231 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.429, + "step": 4232 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.686, + "step": 4233 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6637, + "step": 4234 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.3776, + "step": 4235 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.1559, + "step": 4236 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.5491, + "step": 4237 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0021, + "step": 4238 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8022, + "step": 4239 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.7623, + "step": 4240 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.0046, + "step": 4241 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.4788, + "step": 4242 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.9274, + "step": 4243 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 3.0067, + "step": 4244 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.6597, + "step": 4245 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6076, + "step": 4246 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 2.2011, + "step": 4247 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.6104, + "step": 4248 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.2956, + "step": 4249 + }, + { + "epoch": 0.42, + "learning_rate": 0.0001, + "loss": 1.8986, + "step": 4250 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9443, + "step": 4251 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5444, + "step": 4252 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.3866, + "step": 4253 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1707, + "step": 4254 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7458, + "step": 4255 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2065, + "step": 4256 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 3.1076, + "step": 4257 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6675, + "step": 4258 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 3.0608, + "step": 4259 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.9824, + "step": 4260 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7955, + "step": 4261 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7297, + "step": 4262 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2258, + "step": 4263 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8046, + "step": 4264 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9581, + "step": 4265 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9437, + "step": 4266 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6448, + "step": 4267 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0561, + "step": 4268 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6148, + "step": 4269 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8713, + "step": 4270 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7042, + "step": 4271 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6743, + "step": 4272 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8843, + "step": 4273 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1283, + "step": 4274 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7036, + "step": 4275 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8858, + "step": 4276 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0549, + "step": 4277 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7141, + "step": 4278 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7775, + "step": 4279 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5696, + "step": 4280 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2853, + "step": 4281 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6613, + "step": 4282 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.678, + "step": 4283 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.3297, + "step": 4284 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7704, + "step": 4285 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.4159, + "step": 4286 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8266, + "step": 4287 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.616, + "step": 4288 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5258, + "step": 4289 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7749, + "step": 4290 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.4923, + "step": 4291 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.5401, + "step": 4292 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7811, + "step": 4293 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.4364, + "step": 4294 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.3743, + "step": 4295 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0923, + "step": 4296 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5455, + "step": 4297 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.764, + "step": 4298 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.2883, + "step": 4299 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6398, + "step": 4300 + }, + { + "epoch": 0.43, + "eval_bleu": 14.882597923134808, + "eval_brevity_penalty": 0.5430533483587424, + "eval_loss": 1.5789804458618164, + "eval_runtime": 5.3032, + "eval_samples_per_second": 565.323, + "eval_steps_per_second": 1.131, + "step": 4300 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9669, + "step": 4301 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9284, + "step": 4302 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6886, + "step": 4303 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1541, + "step": 4304 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.3859, + "step": 4305 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8927, + "step": 4306 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8485, + "step": 4307 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0814, + "step": 4308 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.4173, + "step": 4309 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1881, + "step": 4310 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2901, + "step": 4311 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7995, + "step": 4312 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6051, + "step": 4313 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7343, + "step": 4314 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8448, + "step": 4315 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2462, + "step": 4316 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2984, + "step": 4317 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6395, + "step": 4318 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1995, + "step": 4319 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.1976, + "step": 4320 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7416, + "step": 4321 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0758, + "step": 4322 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.8878, + "step": 4323 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7299, + "step": 4324 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.8851, + "step": 4325 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.0676, + "step": 4326 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.7794, + "step": 4327 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2085, + "step": 4328 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.4608, + "step": 4329 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.8419, + "step": 4330 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.5417, + "step": 4331 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.4192, + "step": 4332 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9693, + "step": 4333 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.5287, + "step": 4334 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7543, + "step": 4335 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5958, + "step": 4336 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.9876, + "step": 4337 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.9649, + "step": 4338 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.5613, + "step": 4339 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 3.0411, + "step": 4340 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7976, + "step": 4341 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.4621, + "step": 4342 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.9986, + "step": 4343 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.605, + "step": 4344 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.6523, + "step": 4345 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.7197, + "step": 4346 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9102, + "step": 4347 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.4497, + "step": 4348 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 2.2293, + "step": 4349 + }, + { + "epoch": 0.43, + "learning_rate": 0.0001, + "loss": 1.9942, + "step": 4350 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4355, + "step": 4351 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2104, + "step": 4352 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.8522, + "step": 4353 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.5173, + "step": 4354 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2836, + "step": 4355 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.3926, + "step": 4356 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7867, + "step": 4357 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1685, + "step": 4358 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4253, + "step": 4359 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6606, + "step": 4360 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.5603, + "step": 4361 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.6427, + "step": 4362 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.5645, + "step": 4363 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.7567, + "step": 4364 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.9473, + "step": 4365 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.8506, + "step": 4366 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.4541, + "step": 4367 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.975, + "step": 4368 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7433, + "step": 4369 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0346, + "step": 4370 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6489, + "step": 4371 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6128, + "step": 4372 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.89, + "step": 4373 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0359, + "step": 4374 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7709, + "step": 4375 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0467, + "step": 4376 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0213, + "step": 4377 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6677, + "step": 4378 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.7462, + "step": 4379 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.8177, + "step": 4380 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6896, + "step": 4381 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6962, + "step": 4382 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.2685, + "step": 4383 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6884, + "step": 4384 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2568, + "step": 4385 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.9428, + "step": 4386 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.6159, + "step": 4387 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1597, + "step": 4388 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.696, + "step": 4389 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.3487, + "step": 4390 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0233, + "step": 4391 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4938, + "step": 4392 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.5979, + "step": 4393 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.3101, + "step": 4394 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.4967, + "step": 4395 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2746, + "step": 4396 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.3047, + "step": 4397 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 3.162, + "step": 4398 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1224, + "step": 4399 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2758, + "step": 4400 + }, + { + "epoch": 0.44, + "eval_bleu": 14.558023937447368, + "eval_brevity_penalty": 0.5374579878424441, + "eval_loss": 1.5817654132843018, + "eval_runtime": 5.2196, + "eval_samples_per_second": 574.377, + "eval_steps_per_second": 1.15, + "step": 4400 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7894, + "step": 4401 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.9862, + "step": 4402 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.6214, + "step": 4403 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.0652, + "step": 4404 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4601, + "step": 4405 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1409, + "step": 4406 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.6566, + "step": 4407 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.2852, + "step": 4408 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1716, + "step": 4409 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7613, + "step": 4410 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4514, + "step": 4411 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.1451, + "step": 4412 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.3696, + "step": 4413 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.364, + "step": 4414 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.9547, + "step": 4415 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2399, + "step": 4416 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.552, + "step": 4417 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.6097, + "step": 4418 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.5784, + "step": 4419 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4824, + "step": 4420 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.8486, + "step": 4421 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.8587, + "step": 4422 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4278, + "step": 4423 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.3712, + "step": 4424 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.5317, + "step": 4425 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.2606, + "step": 4426 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.5983, + "step": 4427 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.6478, + "step": 4428 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.345, + "step": 4429 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.9315, + "step": 4430 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.5268, + "step": 4431 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.9161, + "step": 4432 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.9792, + "step": 4433 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4265, + "step": 4434 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4623, + "step": 4435 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2059, + "step": 4436 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.8556, + "step": 4437 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.3487, + "step": 4438 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.4557, + "step": 4439 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7061, + "step": 4440 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.426, + "step": 4441 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2068, + "step": 4442 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.7323, + "step": 4443 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4513, + "step": 4444 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.2673, + "step": 4445 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 0.9966, + "step": 4446 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 1.4776, + "step": 4447 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.5432, + "step": 4448 + }, + { + "epoch": 0.44, + "learning_rate": 0.0001, + "loss": 2.334, + "step": 4449 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.1982, + "step": 4450 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2917, + "step": 4451 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7973, + "step": 4452 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4335, + "step": 4453 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.3148, + "step": 4454 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6613, + "step": 4455 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4916, + "step": 4456 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7071, + "step": 4457 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.5402, + "step": 4458 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3407, + "step": 4459 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2092, + "step": 4460 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4995, + "step": 4461 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.446, + "step": 4462 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2573, + "step": 4463 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7638, + "step": 4464 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4131, + "step": 4465 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6397, + "step": 4466 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.1639, + "step": 4467 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4966, + "step": 4468 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2211, + "step": 4469 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6243, + "step": 4470 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.2716, + "step": 4471 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9289, + "step": 4472 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.2826, + "step": 4473 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4637, + "step": 4474 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.353, + "step": 4475 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6577, + "step": 4476 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.488, + "step": 4477 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.4625, + "step": 4478 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.957, + "step": 4479 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.24, + "step": 4480 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9506, + "step": 4481 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4445, + "step": 4482 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.8617, + "step": 4483 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.1928, + "step": 4484 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.7327, + "step": 4485 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.8365, + "step": 4486 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.1944, + "step": 4487 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2056, + "step": 4488 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7614, + "step": 4489 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.0428, + "step": 4490 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.0881, + "step": 4491 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7327, + "step": 4492 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.8964, + "step": 4493 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.7394, + "step": 4494 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.8002, + "step": 4495 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.9115, + "step": 4496 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.815, + "step": 4497 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6954, + "step": 4498 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.3292, + "step": 4499 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2622, + "step": 4500 + }, + { + "epoch": 0.45, + "eval_bleu": 15.006208977903546, + "eval_brevity_penalty": 0.5427781427602153, + "eval_loss": 1.582085132598877, + "eval_runtime": 5.0926, + "eval_samples_per_second": 588.702, + "eval_steps_per_second": 1.178, + "step": 4500 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7792, + "step": 4501 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.3763, + "step": 4502 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.197, + "step": 4503 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7943, + "step": 4504 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.6704, + "step": 4505 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 3.0306, + "step": 4506 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4423, + "step": 4507 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 3.2114, + "step": 4508 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.4317, + "step": 4509 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9804, + "step": 4510 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 3.0152, + "step": 4511 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 3.2015, + "step": 4512 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.555, + "step": 4513 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2006, + "step": 4514 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7524, + "step": 4515 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.2958, + "step": 4516 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.0245, + "step": 4517 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.72, + "step": 4518 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4663, + "step": 4519 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.8843, + "step": 4520 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9351, + "step": 4521 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.2445, + "step": 4522 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.172, + "step": 4523 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3015, + "step": 4524 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6308, + "step": 4525 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.517, + "step": 4526 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7909, + "step": 4527 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3039, + "step": 4528 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.6135, + "step": 4529 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.3622, + "step": 4530 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3604, + "step": 4531 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.1989, + "step": 4532 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9642, + "step": 4533 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3626, + "step": 4534 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.3061, + "step": 4535 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.9016, + "step": 4536 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3075, + "step": 4537 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 2.2521, + "step": 4538 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6167, + "step": 4539 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3553, + "step": 4540 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.992, + "step": 4541 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.4737, + "step": 4542 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.245, + "step": 4543 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6733, + "step": 4544 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3676, + "step": 4545 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.3358, + "step": 4546 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.6004, + "step": 4547 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.7864, + "step": 4548 + }, + { + "epoch": 0.45, + "learning_rate": 0.0001, + "loss": 1.365, + "step": 4549 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.0208, + "step": 4550 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3852, + "step": 4551 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6734, + "step": 4552 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.6663, + "step": 4553 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8847, + "step": 4554 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9925, + "step": 4555 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3146, + "step": 4556 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1315, + "step": 4557 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.4721, + "step": 4558 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1532, + "step": 4559 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3626, + "step": 4560 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.847, + "step": 4561 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.5511, + "step": 4562 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.5064, + "step": 4563 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.7349, + "step": 4564 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9356, + "step": 4565 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1835, + "step": 4566 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6313, + "step": 4567 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1123, + "step": 4568 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8933, + "step": 4569 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8202, + "step": 4570 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1223, + "step": 4571 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.4798, + "step": 4572 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.7563, + "step": 4573 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.716, + "step": 4574 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.8929, + "step": 4575 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3977, + "step": 4576 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.6134, + "step": 4577 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1809, + "step": 4578 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6743, + "step": 4579 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.2963, + "step": 4580 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1024, + "step": 4581 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6682, + "step": 4582 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.694, + "step": 4583 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.8981, + "step": 4584 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.5365, + "step": 4585 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 3.077, + "step": 4586 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.5653, + "step": 4587 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.4543, + "step": 4588 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.9472, + "step": 4589 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.9176, + "step": 4590 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.7987, + "step": 4591 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3744, + "step": 4592 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6848, + "step": 4593 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6367, + "step": 4594 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1424, + "step": 4595 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8069, + "step": 4596 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.2468, + "step": 4597 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1954, + "step": 4598 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.5584, + "step": 4599 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3329, + "step": 4600 + }, + { + "epoch": 0.46, + "eval_bleu": 14.76087827884928, + "eval_brevity_penalty": 0.5376914022834152, + "eval_loss": 1.5791656970977783, + "eval_runtime": 5.2354, + "eval_samples_per_second": 572.645, + "eval_steps_per_second": 1.146, + "step": 4600 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8932, + "step": 4601 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.1403, + "step": 4602 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3899, + "step": 4603 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.2551, + "step": 4604 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9446, + "step": 4605 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.1446, + "step": 4606 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.5577, + "step": 4607 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6968, + "step": 4608 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.7084, + "step": 4609 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.0514, + "step": 4610 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3836, + "step": 4611 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3126, + "step": 4612 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9669, + "step": 4613 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.4975, + "step": 4614 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.2914, + "step": 4615 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.0697, + "step": 4616 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9181, + "step": 4617 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.3583, + "step": 4618 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1887, + "step": 4619 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.2415, + "step": 4620 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.451, + "step": 4621 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6746, + "step": 4622 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.047, + "step": 4623 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.764, + "step": 4624 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3238, + "step": 4625 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8873, + "step": 4626 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.7237, + "step": 4627 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.329, + "step": 4628 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.3662, + "step": 4629 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9859, + "step": 4630 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.8576, + "step": 4631 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9595, + "step": 4632 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.696, + "step": 4633 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1455, + "step": 4634 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.5478, + "step": 4635 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6429, + "step": 4636 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.2035, + "step": 4637 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.0798, + "step": 4638 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.9239, + "step": 4639 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1621, + "step": 4640 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.1902, + "step": 4641 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6313, + "step": 4642 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.2029, + "step": 4643 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.0372, + "step": 4644 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.65, + "step": 4645 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.176, + "step": 4646 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.4811, + "step": 4647 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 1.6095, + "step": 4648 + }, + { + "epoch": 0.46, + "learning_rate": 0.0001, + "loss": 2.7938, + "step": 4649 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.478, + "step": 4650 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3812, + "step": 4651 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.6442, + "step": 4652 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7921, + "step": 4653 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.1898, + "step": 4654 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3, + "step": 4655 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.8841, + "step": 4656 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9973, + "step": 4657 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.401, + "step": 4658 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3584, + "step": 4659 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6909, + "step": 4660 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.8287, + "step": 4661 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 3.0629, + "step": 4662 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.4749, + "step": 4663 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.98, + "step": 4664 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.9265, + "step": 4665 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.327, + "step": 4666 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3846, + "step": 4667 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7612, + "step": 4668 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3832, + "step": 4669 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.14, + "step": 4670 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.8185, + "step": 4671 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.2851, + "step": 4672 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.4638, + "step": 4673 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.4811, + "step": 4674 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3778, + "step": 4675 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5863, + "step": 4676 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.4285, + "step": 4677 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3325, + "step": 4678 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.4216, + "step": 4679 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9837, + "step": 4680 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3841, + "step": 4681 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.5215, + "step": 4682 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1514, + "step": 4683 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3253, + "step": 4684 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.8868, + "step": 4685 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.6372, + "step": 4686 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6646, + "step": 4687 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1856, + "step": 4688 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1887, + "step": 4689 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.77, + "step": 4690 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1369, + "step": 4691 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7566, + "step": 4692 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.8033, + "step": 4693 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.47, + "step": 4694 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.2912, + "step": 4695 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7803, + "step": 4696 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3472, + "step": 4697 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3354, + "step": 4698 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7466, + "step": 4699 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7537, + "step": 4700 + }, + { + "epoch": 0.47, + "eval_bleu": 15.10371590564711, + "eval_brevity_penalty": 0.5424605559752034, + "eval_loss": 1.574406385421753, + "eval_runtime": 5.4363, + "eval_samples_per_second": 551.475, + "eval_steps_per_second": 1.104, + "step": 4700 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7524, + "step": 4701 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.8214, + "step": 4702 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6822, + "step": 4703 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7797, + "step": 4704 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6749, + "step": 4705 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1465, + "step": 4706 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9132, + "step": 4707 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6821, + "step": 4708 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5463, + "step": 4709 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.0712, + "step": 4710 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5758, + "step": 4711 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9306, + "step": 4712 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.748, + "step": 4713 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7542, + "step": 4714 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.6038, + "step": 4715 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7014, + "step": 4716 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.435, + "step": 4717 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5774, + "step": 4718 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6125, + "step": 4719 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.4113, + "step": 4720 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.2352, + "step": 4721 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.6639, + "step": 4722 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5465, + "step": 4723 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1939, + "step": 4724 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9026, + "step": 4725 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3218, + "step": 4726 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1309, + "step": 4727 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3667, + "step": 4728 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.3901, + "step": 4729 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.8746, + "step": 4730 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1861, + "step": 4731 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.2649, + "step": 4732 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.1102, + "step": 4733 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.0989, + "step": 4734 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7774, + "step": 4735 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.4386, + "step": 4736 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.3834, + "step": 4737 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.7113, + "step": 4738 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.4814, + "step": 4739 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.9788, + "step": 4740 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.5154, + "step": 4741 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.9291, + "step": 4742 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 3.533, + "step": 4743 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.4578, + "step": 4744 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 3.1813, + "step": 4745 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.6997, + "step": 4746 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9039, + "step": 4747 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.2114, + "step": 4748 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 2.0142, + "step": 4749 + }, + { + "epoch": 0.47, + "learning_rate": 0.0001, + "loss": 1.9924, + "step": 4750 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2481, + "step": 4751 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2736, + "step": 4752 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.8197, + "step": 4753 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1911, + "step": 4754 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7662, + "step": 4755 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9519, + "step": 4756 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9906, + "step": 4757 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.87, + "step": 4758 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3668, + "step": 4759 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6644, + "step": 4760 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.9684, + "step": 4761 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1054, + "step": 4762 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1932, + "step": 4763 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6873, + "step": 4764 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2035, + "step": 4765 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6753, + "step": 4766 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2899, + "step": 4767 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7355, + "step": 4768 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0562, + "step": 4769 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.5779, + "step": 4770 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.5569, + "step": 4771 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.465, + "step": 4772 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7555, + "step": 4773 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.3028, + "step": 4774 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.388, + "step": 4775 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9628, + "step": 4776 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.4384, + "step": 4777 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.8607, + "step": 4778 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.1728, + "step": 4779 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.2747, + "step": 4780 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9902, + "step": 4781 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2728, + "step": 4782 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.075, + "step": 4783 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2882, + "step": 4784 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3538, + "step": 4785 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.357, + "step": 4786 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1269, + "step": 4787 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6356, + "step": 4788 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2182, + "step": 4789 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2632, + "step": 4790 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6558, + "step": 4791 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9679, + "step": 4792 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.735, + "step": 4793 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.315, + "step": 4794 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1134, + "step": 4795 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.6901, + "step": 4796 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1991, + "step": 4797 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1106, + "step": 4798 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2544, + "step": 4799 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5379, + "step": 4800 + }, + { + "epoch": 0.48, + "eval_bleu": 15.268438743236873, + "eval_brevity_penalty": 0.5479167680013121, + "eval_loss": 1.575563669204712, + "eval_runtime": 5.0398, + "eval_samples_per_second": 594.87, + "eval_steps_per_second": 1.191, + "step": 4800 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1634, + "step": 4801 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0894, + "step": 4802 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.6347, + "step": 4803 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1604, + "step": 4804 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3472, + "step": 4805 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2554, + "step": 4806 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.19, + "step": 4807 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.8209, + "step": 4808 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.754, + "step": 4809 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7062, + "step": 4810 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5447, + "step": 4811 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3105, + "step": 4812 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9533, + "step": 4813 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9547, + "step": 4814 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.2408, + "step": 4815 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.6978, + "step": 4816 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5515, + "step": 4817 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 3.0504, + "step": 4818 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9901, + "step": 4819 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 3.0027, + "step": 4820 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.9174, + "step": 4821 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0892, + "step": 4822 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.7147, + "step": 4823 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 3.2398, + "step": 4824 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.8737, + "step": 4825 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.9681, + "step": 4826 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7909, + "step": 4827 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.3136, + "step": 4828 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0131, + "step": 4829 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.3744, + "step": 4830 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0006, + "step": 4831 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0631, + "step": 4832 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7133, + "step": 4833 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.7382, + "step": 4834 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.274, + "step": 4835 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.7414, + "step": 4836 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9715, + "step": 4837 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5984, + "step": 4838 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.1287, + "step": 4839 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.3646, + "step": 4840 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.4278, + "step": 4841 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3718, + "step": 4842 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.9886, + "step": 4843 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0015, + "step": 4844 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 1.861, + "step": 4845 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0226, + "step": 4846 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.3176, + "step": 4847 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5983, + "step": 4848 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.0904, + "step": 4849 + }, + { + "epoch": 0.48, + "learning_rate": 0.0001, + "loss": 2.5271, + "step": 4850 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3395, + "step": 4851 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4786, + "step": 4852 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.5762, + "step": 4853 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3873, + "step": 4854 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0475, + "step": 4855 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1935, + "step": 4856 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1245, + "step": 4857 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.9448, + "step": 4858 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2731, + "step": 4859 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5546, + "step": 4860 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0684, + "step": 4861 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4498, + "step": 4862 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.8225, + "step": 4863 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1602, + "step": 4864 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.7223, + "step": 4865 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2752, + "step": 4866 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0575, + "step": 4867 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2103, + "step": 4868 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5048, + "step": 4869 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.995, + "step": 4870 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2529, + "step": 4871 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2355, + "step": 4872 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.6677, + "step": 4873 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.523, + "step": 4874 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9299, + "step": 4875 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0002, + "step": 4876 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.7793, + "step": 4877 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3673, + "step": 4878 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.189, + "step": 4879 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5364, + "step": 4880 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5855, + "step": 4881 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.8035, + "step": 4882 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3145, + "step": 4883 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.7388, + "step": 4884 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.3985, + "step": 4885 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2435, + "step": 4886 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.6948, + "step": 4887 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.2473, + "step": 4888 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3883, + "step": 4889 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4842, + "step": 4890 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.8767, + "step": 4891 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 3.0252, + "step": 4892 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 3.1796, + "step": 4893 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.9689, + "step": 4894 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9235, + "step": 4895 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9089, + "step": 4896 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1248, + "step": 4897 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9657, + "step": 4898 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9931, + "step": 4899 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1236, + "step": 4900 + }, + { + "epoch": 0.49, + "eval_bleu": 14.822858975076555, + "eval_brevity_penalty": 0.5478111548402327, + "eval_loss": 1.5822086334228516, + "eval_runtime": 5.3309, + "eval_samples_per_second": 562.384, + "eval_steps_per_second": 1.126, + "step": 4900 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.187, + "step": 4901 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5821, + "step": 4902 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1437, + "step": 4903 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0859, + "step": 4904 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.71, + "step": 4905 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.8558, + "step": 4906 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1148, + "step": 4907 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3236, + "step": 4908 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.9586, + "step": 4909 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3312, + "step": 4910 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5211, + "step": 4911 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1447, + "step": 4912 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3666, + "step": 4913 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3326, + "step": 4914 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.4559, + "step": 4915 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.26, + "step": 4916 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0695, + "step": 4917 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2448, + "step": 4918 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2872, + "step": 4919 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.6833, + "step": 4920 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.9034, + "step": 4921 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2824, + "step": 4922 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4273, + "step": 4923 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.947, + "step": 4924 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3845, + "step": 4925 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4405, + "step": 4926 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1853, + "step": 4927 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4093, + "step": 4928 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2506, + "step": 4929 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4237, + "step": 4930 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1958, + "step": 4931 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3527, + "step": 4932 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.815, + "step": 4933 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2884, + "step": 4934 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3902, + "step": 4935 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.1811, + "step": 4936 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3493, + "step": 4937 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.3117, + "step": 4938 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0931, + "step": 4939 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.5235, + "step": 4940 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.7522, + "step": 4941 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.0392, + "step": 4942 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.4076, + "step": 4943 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.582, + "step": 4944 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.014, + "step": 4945 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.2974, + "step": 4946 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 3.2689, + "step": 4947 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 1.9943, + "step": 4948 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 2.9842, + "step": 4949 + }, + { + "epoch": 0.49, + "learning_rate": 0.0001, + "loss": 3.1869, + "step": 4950 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6885, + "step": 4951 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.9038, + "step": 4952 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.2142, + "step": 4953 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.8557, + "step": 4954 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.0308, + "step": 4955 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.0878, + "step": 4956 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2505, + "step": 4957 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.4424, + "step": 4958 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.7832, + "step": 4959 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.7809, + "step": 4960 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.4217, + "step": 4961 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.9786, + "step": 4962 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5094, + "step": 4963 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1351, + "step": 4964 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.903, + "step": 4965 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.9176, + "step": 4966 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.0677, + "step": 4967 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.714, + "step": 4968 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5835, + "step": 4969 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1245, + "step": 4970 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.4303, + "step": 4971 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3453, + "step": 4972 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.155, + "step": 4973 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3973, + "step": 4974 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1372, + "step": 4975 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3916, + "step": 4976 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.716, + "step": 4977 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1413, + "step": 4978 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2668, + "step": 4979 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.9595, + "step": 4980 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.323, + "step": 4981 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1921, + "step": 4982 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5148, + "step": 4983 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.954, + "step": 4984 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1077, + "step": 4985 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.0722, + "step": 4986 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.8244, + "step": 4987 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3751, + "step": 4988 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3564, + "step": 4989 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6349, + "step": 4990 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3569, + "step": 4991 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2679, + "step": 4992 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6646, + "step": 4993 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.5071, + "step": 4994 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.0473, + "step": 4995 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.8507, + "step": 4996 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.968, + "step": 4997 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2182, + "step": 4998 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5196, + "step": 4999 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.9621, + "step": 5000 + }, + { + "epoch": 0.5, + "eval_bleu": 14.994792301373202, + "eval_brevity_penalty": 0.5442596206063441, + "eval_loss": 1.5747015476226807, + "eval_runtime": 5.3071, + "eval_samples_per_second": 564.907, + "eval_steps_per_second": 1.131, + "step": 5000 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.6337, + "step": 5001 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6363, + "step": 5002 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.0493, + "step": 5003 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.0006, + "step": 5004 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3039, + "step": 5005 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.9509, + "step": 5006 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 3.017, + "step": 5007 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.901, + "step": 5008 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.5902, + "step": 5009 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.904, + "step": 5010 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.0719, + "step": 5011 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.4829, + "step": 5012 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6322, + "step": 5013 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.2322, + "step": 5014 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2553, + "step": 5015 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5866, + "step": 5016 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.2918, + "step": 5017 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.1144, + "step": 5018 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5162, + "step": 5019 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.069, + "step": 5020 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.13, + "step": 5021 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5167, + "step": 5022 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 0.9865, + "step": 5023 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3909, + "step": 5024 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.4838, + "step": 5025 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.3732, + "step": 5026 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2289, + "step": 5027 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.7294, + "step": 5028 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.4913, + "step": 5029 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2727, + "step": 5030 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.9082, + "step": 5031 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.8746, + "step": 5032 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.0999, + "step": 5033 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2851, + "step": 5034 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.9469, + "step": 5035 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.219, + "step": 5036 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2356, + "step": 5037 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6213, + "step": 5038 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.343, + "step": 5039 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2588, + "step": 5040 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.6194, + "step": 5041 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3096, + "step": 5042 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.3822, + "step": 5043 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5921, + "step": 5044 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.2096, + "step": 5045 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.0901, + "step": 5046 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 1.5625, + "step": 5047 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.5086, + "step": 5048 + }, + { + "epoch": 0.5, + "learning_rate": 0.0001, + "loss": 2.895, + "step": 5049 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5723, + "step": 5050 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 3.0008, + "step": 5051 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1876, + "step": 5052 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.9867, + "step": 5053 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.8464, + "step": 5054 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.583, + "step": 5055 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.2071, + "step": 5056 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.9738, + "step": 5057 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1403, + "step": 5058 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.9516, + "step": 5059 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.6151, + "step": 5060 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4548, + "step": 5061 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9223, + "step": 5062 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2447, + "step": 5063 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2038, + "step": 5064 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.6637, + "step": 5065 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1492, + "step": 5066 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.963, + "step": 5067 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.6015, + "step": 5068 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1975, + "step": 5069 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.0195, + "step": 5070 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5086, + "step": 5071 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2534, + "step": 5072 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1002, + "step": 5073 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2383, + "step": 5074 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2435, + "step": 5075 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.0157, + "step": 5076 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.331, + "step": 5077 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1406, + "step": 5078 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9478, + "step": 5079 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5662, + "step": 5080 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.103, + "step": 5081 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2571, + "step": 5082 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.3982, + "step": 5083 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3433, + "step": 5084 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.0354, + "step": 5085 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5106, + "step": 5086 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2922, + "step": 5087 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4504, + "step": 5088 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.9921, + "step": 5089 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2807, + "step": 5090 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.2976, + "step": 5091 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.8846, + "step": 5092 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.164, + "step": 5093 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4736, + "step": 5094 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.8847, + "step": 5095 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.24, + "step": 5096 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4114, + "step": 5097 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 0.925, + "step": 5098 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3506, + "step": 5099 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9832, + "step": 5100 + }, + { + "epoch": 0.51, + "eval_bleu": 14.868152071619512, + "eval_brevity_penalty": 0.5467970111947749, + "eval_loss": 1.5838361978530884, + "eval_runtime": 5.0735, + "eval_samples_per_second": 590.918, + "eval_steps_per_second": 1.183, + "step": 5100 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5725, + "step": 5101 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3659, + "step": 5102 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1947, + "step": 5103 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5916, + "step": 5104 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.738, + "step": 5105 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.784, + "step": 5106 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5403, + "step": 5107 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.8957, + "step": 5108 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.6578, + "step": 5109 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5606, + "step": 5110 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.9481, + "step": 5111 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.8407, + "step": 5112 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5732, + "step": 5113 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 3.0137, + "step": 5114 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.7477, + "step": 5115 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5704, + "step": 5116 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2611, + "step": 5117 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9941, + "step": 5118 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5339, + "step": 5119 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3831, + "step": 5120 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.8818, + "step": 5121 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4567, + "step": 5122 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9782, + "step": 5123 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9638, + "step": 5124 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5643, + "step": 5125 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1402, + "step": 5126 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.8697, + "step": 5127 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.5654, + "step": 5128 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3819, + "step": 5129 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9476, + "step": 5130 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4495, + "step": 5131 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1879, + "step": 5132 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.7787, + "step": 5133 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4297, + "step": 5134 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2218, + "step": 5135 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.9274, + "step": 5136 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4064, + "step": 5137 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.1602, + "step": 5138 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.906, + "step": 5139 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.3577, + "step": 5140 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.2639, + "step": 5141 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.8761, + "step": 5142 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4783, + "step": 5143 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.3496, + "step": 5144 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.974, + "step": 5145 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4751, + "step": 5146 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.0377, + "step": 5147 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 2.0181, + "step": 5148 + }, + { + "epoch": 0.51, + "learning_rate": 0.0001, + "loss": 1.4285, + "step": 5149 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.7858, + "step": 5150 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2137, + "step": 5151 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5562, + "step": 5152 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.4887, + "step": 5153 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.0531, + "step": 5154 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.458, + "step": 5155 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.306, + "step": 5156 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2777, + "step": 5157 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5221, + "step": 5158 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 3.0593, + "step": 5159 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.6454, + "step": 5160 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5047, + "step": 5161 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.9793, + "step": 5162 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.6848, + "step": 5163 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5322, + "step": 5164 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 3.0638, + "step": 5165 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.6716, + "step": 5166 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.6275, + "step": 5167 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.8955, + "step": 5168 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9079, + "step": 5169 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5127, + "step": 5170 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.0947, + "step": 5171 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9375, + "step": 5172 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4197, + "step": 5173 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.3153, + "step": 5174 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.7213, + "step": 5175 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.6376, + "step": 5176 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.1793, + "step": 5177 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.8768, + "step": 5178 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5474, + "step": 5179 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.1827, + "step": 5180 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9424, + "step": 5181 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4396, + "step": 5182 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2527, + "step": 5183 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9459, + "step": 5184 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4946, + "step": 5185 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2253, + "step": 5186 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.8987, + "step": 5187 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4359, + "step": 5188 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.0925, + "step": 5189 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.853, + "step": 5190 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.61, + "step": 5191 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.3014, + "step": 5192 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9599, + "step": 5193 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.6067, + "step": 5194 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.119, + "step": 5195 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.933, + "step": 5196 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4577, + "step": 5197 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.1049, + "step": 5198 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.1597, + "step": 5199 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4962, + "step": 5200 + }, + { + "epoch": 0.52, + "eval_bleu": 14.809364152383218, + "eval_brevity_penalty": 0.5397486639800217, + "eval_loss": 1.5836420059204102, + "eval_runtime": 5.1027, + "eval_samples_per_second": 587.536, + "eval_steps_per_second": 1.176, + "step": 5200 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.3456, + "step": 5201 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9497, + "step": 5202 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5604, + "step": 5203 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2069, + "step": 5204 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2547, + "step": 5205 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.434, + "step": 5206 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.419, + "step": 5207 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.5788, + "step": 5208 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.3545, + "step": 5209 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.9322, + "step": 5210 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.5908, + "step": 5211 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.3989, + "step": 5212 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.9039, + "step": 5213 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.6631, + "step": 5214 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4432, + "step": 5215 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.9986, + "step": 5216 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.5699, + "step": 5217 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5282, + "step": 5218 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 3.0056, + "step": 5219 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.7945, + "step": 5220 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.3463, + "step": 5221 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.0005, + "step": 5222 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9698, + "step": 5223 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.3965, + "step": 5224 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.3173, + "step": 5225 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.8042, + "step": 5226 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.6858, + "step": 5227 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2179, + "step": 5228 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.8372, + "step": 5229 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4265, + "step": 5230 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2836, + "step": 5231 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.7363, + "step": 5232 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.466, + "step": 5233 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.188, + "step": 5234 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.7591, + "step": 5235 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.4311, + "step": 5236 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2556, + "step": 5237 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.8553, + "step": 5238 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.3742, + "step": 5239 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2096, + "step": 5240 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.9284, + "step": 5241 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.2855, + "step": 5242 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.2162, + "step": 5243 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.7599, + "step": 5244 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5625, + "step": 5245 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.1552, + "step": 5246 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.822, + "step": 5247 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 1.5118, + "step": 5248 + }, + { + "epoch": 0.52, + "learning_rate": 0.0001, + "loss": 2.4349, + "step": 5249 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.0049, + "step": 5250 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5351, + "step": 5251 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.4813, + "step": 5252 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.9635, + "step": 5253 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4306, + "step": 5254 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.0575, + "step": 5255 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5781, + "step": 5256 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.373, + "step": 5257 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.3186, + "step": 5258 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.3342, + "step": 5259 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4516, + "step": 5260 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.9583, + "step": 5261 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5933, + "step": 5262 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5505, + "step": 5263 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.8852, + "step": 5264 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5443, + "step": 5265 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.411, + "step": 5266 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.9043, + "step": 5267 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5211, + "step": 5268 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4037, + "step": 5269 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.7011, + "step": 5270 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7894, + "step": 5271 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3789, + "step": 5272 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1237, + "step": 5273 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2226, + "step": 5274 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4451, + "step": 5275 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.9316, + "step": 5276 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6307, + "step": 5277 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6109, + "step": 5278 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.0582, + "step": 5279 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.8551, + "step": 5280 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3942, + "step": 5281 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.064, + "step": 5282 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7022, + "step": 5283 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3155, + "step": 5284 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.0534, + "step": 5285 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.8199, + "step": 5286 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3229, + "step": 5287 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1935, + "step": 5288 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.8477, + "step": 5289 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3112, + "step": 5290 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2065, + "step": 5291 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.861, + "step": 5292 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.2147, + "step": 5293 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1475, + "step": 5294 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7418, + "step": 5295 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4295, + "step": 5296 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2537, + "step": 5297 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.8617, + "step": 5298 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4185, + "step": 5299 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.4318, + "step": 5300 + }, + { + "epoch": 0.53, + "eval_bleu": 14.821343792065884, + "eval_brevity_penalty": 0.5422276305171174, + "eval_loss": 1.5826094150543213, + "eval_runtime": 5.1224, + "eval_samples_per_second": 585.269, + "eval_steps_per_second": 1.171, + "step": 5300 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7399, + "step": 5301 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5494, + "step": 5302 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.9009, + "step": 5303 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.305, + "step": 5304 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4016, + "step": 5305 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5143, + "step": 5306 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1481, + "step": 5307 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4486, + "step": 5308 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.235, + "step": 5309 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5488, + "step": 5310 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5003, + "step": 5311 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.8971, + "step": 5312 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.5894, + "step": 5313 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3408, + "step": 5314 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.9603, + "step": 5315 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.6357, + "step": 5316 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3419, + "step": 5317 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 3.0259, + "step": 5318 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.6088, + "step": 5319 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4271, + "step": 5320 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.045, + "step": 5321 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6612, + "step": 5322 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4908, + "step": 5323 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.4053, + "step": 5324 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.5714, + "step": 5325 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.286, + "step": 5326 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.0691, + "step": 5327 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7828, + "step": 5328 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3821, + "step": 5329 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1375, + "step": 5330 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6419, + "step": 5331 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3564, + "step": 5332 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1355, + "step": 5333 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.818, + "step": 5334 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4239, + "step": 5335 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2321, + "step": 5336 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.7235, + "step": 5337 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.397, + "step": 5338 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.3358, + "step": 5339 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6988, + "step": 5340 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.3697, + "step": 5341 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2119, + "step": 5342 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.8357, + "step": 5343 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.4585, + "step": 5344 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2818, + "step": 5345 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.9143, + "step": 5346 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 1.6474, + "step": 5347 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.2483, + "step": 5348 + }, + { + "epoch": 0.53, + "learning_rate": 0.0001, + "loss": 2.1652, + "step": 5349 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3493, + "step": 5350 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.5938, + "step": 5351 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0822, + "step": 5352 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.5564, + "step": 5353 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2814, + "step": 5354 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0839, + "step": 5355 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.492, + "step": 5356 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.127, + "step": 5357 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.668, + "step": 5358 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4436, + "step": 5359 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 3.0961, + "step": 5360 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.4989, + "step": 5361 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3986, + "step": 5362 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.9054, + "step": 5363 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.5592, + "step": 5364 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3907, + "step": 5365 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.8952, + "step": 5366 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.4845, + "step": 5367 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.5658, + "step": 5368 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1634, + "step": 5369 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0318, + "step": 5370 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4348, + "step": 5371 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2113, + "step": 5372 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7534, + "step": 5373 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4491, + "step": 5374 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1687, + "step": 5375 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7121, + "step": 5376 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4304, + "step": 5377 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.135, + "step": 5378 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.8552, + "step": 5379 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.2867, + "step": 5380 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1688, + "step": 5381 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7185, + "step": 5382 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.291, + "step": 5383 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0935, + "step": 5384 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.813, + "step": 5385 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3868, + "step": 5386 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1468, + "step": 5387 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.9392, + "step": 5388 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4229, + "step": 5389 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0967, + "step": 5390 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.953, + "step": 5391 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3114, + "step": 5392 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2175, + "step": 5393 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7075, + "step": 5394 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.2937, + "step": 5395 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2185, + "step": 5396 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.8165, + "step": 5397 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.1817, + "step": 5398 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.4869, + "step": 5399 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.9338, + "step": 5400 + }, + { + "epoch": 0.54, + "eval_bleu": 14.557055931441914, + "eval_brevity_penalty": 0.5402361963006292, + "eval_loss": 1.586909532546997, + "eval_runtime": 5.4197, + "eval_samples_per_second": 553.169, + "eval_steps_per_second": 1.107, + "step": 5400 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4271, + "step": 5401 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1556, + "step": 5402 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.9214, + "step": 5403 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3645, + "step": 5404 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.9714, + "step": 5405 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.3678, + "step": 5406 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3188, + "step": 5407 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 3.0159, + "step": 5408 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.4486, + "step": 5409 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.2402, + "step": 5410 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.815, + "step": 5411 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.4926, + "step": 5412 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3542, + "step": 5413 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 3.0024, + "step": 5414 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0558, + "step": 5415 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3692, + "step": 5416 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.5503, + "step": 5417 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.3855, + "step": 5418 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.5413, + "step": 5419 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2199, + "step": 5420 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.5803, + "step": 5421 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4479, + "step": 5422 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2593, + "step": 5423 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.6578, + "step": 5424 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3976, + "step": 5425 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0397, + "step": 5426 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7099, + "step": 5427 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3992, + "step": 5428 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.0821, + "step": 5429 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7311, + "step": 5430 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4538, + "step": 5431 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.3513, + "step": 5432 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.8663, + "step": 5433 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4322, + "step": 5434 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2012, + "step": 5435 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7821, + "step": 5436 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3647, + "step": 5437 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1151, + "step": 5438 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.8017, + "step": 5439 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.268, + "step": 5440 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1607, + "step": 5441 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.6449, + "step": 5442 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.4243, + "step": 5443 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.1867, + "step": 5444 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.7545, + "step": 5445 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.3754, + "step": 5446 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 2.2498, + "step": 5447 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.8096, + "step": 5448 + }, + { + "epoch": 0.54, + "learning_rate": 0.0001, + "loss": 1.2813, + "step": 5449 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.1471, + "step": 5450 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.9549, + "step": 5451 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3243, + "step": 5452 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.1882, + "step": 5453 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8047, + "step": 5454 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2718, + "step": 5455 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0819, + "step": 5456 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0634, + "step": 5457 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2994, + "step": 5458 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 3.0002, + "step": 5459 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4487, + "step": 5460 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.6434, + "step": 5461 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 3.1232, + "step": 5462 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4896, + "step": 5463 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.478, + "step": 5464 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.8382, + "step": 5465 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4937, + "step": 5466 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.4064, + "step": 5467 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.7887, + "step": 5468 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7981, + "step": 5469 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3023, + "step": 5470 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4751, + "step": 5471 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8764, + "step": 5472 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.426, + "step": 5473 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0546, + "step": 5474 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.719, + "step": 5475 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2484, + "step": 5476 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2337, + "step": 5477 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.688, + "step": 5478 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3234, + "step": 5479 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2851, + "step": 5480 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.6737, + "step": 5481 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.4074, + "step": 5482 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2116, + "step": 5483 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8345, + "step": 5484 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3185, + "step": 5485 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2962, + "step": 5486 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7257, + "step": 5487 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3137, + "step": 5488 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2517, + "step": 5489 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7275, + "step": 5490 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3694, + "step": 5491 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.101, + "step": 5492 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.709, + "step": 5493 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3704, + "step": 5494 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.196, + "step": 5495 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.5058, + "step": 5496 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3916, + "step": 5497 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.5173, + "step": 5498 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8431, + "step": 5499 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.404, + "step": 5500 + }, + { + "epoch": 0.55, + "eval_bleu": 14.510335124689263, + "eval_brevity_penalty": 0.5413592418040077, + "eval_loss": 1.5890583992004395, + "eval_runtime": 5.2061, + "eval_samples_per_second": 575.859, + "eval_steps_per_second": 1.152, + "step": 5500 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.3625, + "step": 5501 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.9952, + "step": 5502 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.38, + "step": 5503 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0403, + "step": 5504 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7011, + "step": 5505 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3993, + "step": 5506 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2984, + "step": 5507 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.88, + "step": 5508 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.4061, + "step": 5509 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0652, + "step": 5510 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.3987, + "step": 5511 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2142, + "step": 5512 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.895, + "step": 5513 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.446, + "step": 5514 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2668, + "step": 5515 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.8401, + "step": 5516 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4628, + "step": 5517 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3424, + "step": 5518 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.9387, + "step": 5519 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.4971, + "step": 5520 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2458, + "step": 5521 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.773, + "step": 5522 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8048, + "step": 5523 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2825, + "step": 5524 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2403, + "step": 5525 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0879, + "step": 5526 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.255, + "step": 5527 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0442, + "step": 5528 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.6163, + "step": 5529 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3309, + "step": 5530 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.1681, + "step": 5531 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7523, + "step": 5532 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3183, + "step": 5533 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.1498, + "step": 5534 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.7295, + "step": 5535 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2333, + "step": 5536 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0953, + "step": 5537 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.6948, + "step": 5538 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2514, + "step": 5539 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.1532, + "step": 5540 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.972, + "step": 5541 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.446, + "step": 5542 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.0927, + "step": 5543 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.8777, + "step": 5544 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.2481, + "step": 5545 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2344, + "step": 5546 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.6529, + "step": 5547 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 1.3656, + "step": 5548 + }, + { + "epoch": 0.55, + "learning_rate": 0.0001, + "loss": 2.2193, + "step": 5549 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.7151, + "step": 5550 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.4062, + "step": 5551 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2119, + "step": 5552 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1169, + "step": 5553 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3518, + "step": 5554 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.144, + "step": 5555 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8118, + "step": 5556 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3026, + "step": 5557 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.9547, + "step": 5558 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8313, + "step": 5559 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3572, + "step": 5560 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.8768, + "step": 5561 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.5374, + "step": 5562 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3058, + "step": 5563 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.9606, + "step": 5564 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.4339, + "step": 5565 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2656, + "step": 5566 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.8871, + "step": 5567 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.4687, + "step": 5568 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2647, + "step": 5569 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.892, + "step": 5570 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8493, + "step": 5571 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.316, + "step": 5572 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.3255, + "step": 5573 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.181, + "step": 5574 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2985, + "step": 5575 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.6866, + "step": 5576 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6095, + "step": 5577 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2664, + "step": 5578 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1768, + "step": 5579 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6695, + "step": 5580 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3753, + "step": 5581 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1328, + "step": 5582 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6951, + "step": 5583 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3748, + "step": 5584 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2358, + "step": 5585 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6053, + "step": 5586 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3316, + "step": 5587 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.3201, + "step": 5588 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8056, + "step": 5589 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3157, + "step": 5590 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1829, + "step": 5591 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.7588, + "step": 5592 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2894, + "step": 5593 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2081, + "step": 5594 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6484, + "step": 5595 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.4097, + "step": 5596 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.112, + "step": 5597 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6265, + "step": 5598 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2778, + "step": 5599 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2803, + "step": 5600 + }, + { + "epoch": 0.56, + "eval_bleu": 14.633832470472846, + "eval_brevity_penalty": 0.5416769837410983, + "eval_loss": 1.586429476737976, + "eval_runtime": 5.1264, + "eval_samples_per_second": 584.821, + "eval_steps_per_second": 1.17, + "step": 5600 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.7923, + "step": 5601 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2302, + "step": 5602 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1173, + "step": 5603 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6603, + "step": 5604 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2246, + "step": 5605 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.0664, + "step": 5606 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8884, + "step": 5607 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3331, + "step": 5608 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1563, + "step": 5609 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.9499, + "step": 5610 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2968, + "step": 5611 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2247, + "step": 5612 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.4494, + "step": 5613 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2443, + "step": 5614 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.8975, + "step": 5615 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.4505, + "step": 5616 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2805, + "step": 5617 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.9337, + "step": 5618 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.5135, + "step": 5619 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.27, + "step": 5620 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.8946, + "step": 5621 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.4685, + "step": 5622 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3861, + "step": 5623 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.5033, + "step": 5624 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6994, + "step": 5625 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2715, + "step": 5626 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.3166, + "step": 5627 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.4813, + "step": 5628 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2873, + "step": 5629 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.0784, + "step": 5630 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.7944, + "step": 5631 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2597, + "step": 5632 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.9695, + "step": 5633 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8115, + "step": 5634 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3764, + "step": 5635 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1247, + "step": 5636 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6688, + "step": 5637 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.3757, + "step": 5638 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.221, + "step": 5639 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8, + "step": 5640 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.4097, + "step": 5641 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.2242, + "step": 5642 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6483, + "step": 5643 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.4883, + "step": 5644 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.1059, + "step": 5645 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.6563, + "step": 5646 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2197, + "step": 5647 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 2.0685, + "step": 5648 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.8414, + "step": 5649 + }, + { + "epoch": 0.56, + "learning_rate": 0.0001, + "loss": 1.2167, + "step": 5650 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0273, + "step": 5651 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0949, + "step": 5652 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.3682, + "step": 5653 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2132, + "step": 5654 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8347, + "step": 5655 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2895, + "step": 5656 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.5292, + "step": 5657 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2024, + "step": 5658 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2466, + "step": 5659 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2487, + "step": 5660 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4678, + "step": 5661 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.1851, + "step": 5662 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.9724, + "step": 5663 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4661, + "step": 5664 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2569, + "step": 5665 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.8184, + "step": 5666 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.454, + "step": 5667 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2281, + "step": 5668 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.9705, + "step": 5669 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.5271, + "step": 5670 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.3811, + "step": 5671 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.494, + "step": 5672 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9845, + "step": 5673 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.3008, + "step": 5674 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0045, + "step": 5675 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.7309, + "step": 5676 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2841, + "step": 5677 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.6065, + "step": 5678 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.6694, + "step": 5679 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.2742, + "step": 5680 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0641, + "step": 5681 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.5614, + "step": 5682 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.311, + "step": 5683 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9613, + "step": 5684 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.6374, + "step": 5685 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8015, + "step": 5686 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2171, + "step": 5687 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4575, + "step": 5688 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8629, + "step": 5689 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0256, + "step": 5690 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2286, + "step": 5691 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.859, + "step": 5692 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.1839, + "step": 5693 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9481, + "step": 5694 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9138, + "step": 5695 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.1762, + "step": 5696 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.6885, + "step": 5697 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4822, + "step": 5698 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0925, + "step": 5699 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.3725, + "step": 5700 + }, + { + "epoch": 0.57, + "eval_bleu": 14.340475539315412, + "eval_brevity_penalty": 0.5384551380849611, + "eval_loss": 1.5892844200134277, + "eval_runtime": 5.0962, + "eval_samples_per_second": 588.277, + "eval_steps_per_second": 1.177, + "step": 5700 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.5788, + "step": 5701 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2024, + "step": 5702 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.241, + "step": 5703 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8107, + "step": 5704 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0554, + "step": 5705 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.3461, + "step": 5706 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9748, + "step": 5707 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.3401, + "step": 5708 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4234, + "step": 5709 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.7798, + "step": 5710 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.5191, + "step": 5711 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9037, + "step": 5712 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.7961, + "step": 5713 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0777, + "step": 5714 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9162, + "step": 5715 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.5864, + "step": 5716 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4761, + "step": 5717 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.1881, + "step": 5718 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8131, + "step": 5719 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9997, + "step": 5720 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.7652, + "step": 5721 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.7246, + "step": 5722 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.6994, + "step": 5723 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.27, + "step": 5724 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.634, + "step": 5725 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4227, + "step": 5726 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9445, + "step": 5727 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.586, + "step": 5728 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.0281, + "step": 5729 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.336, + "step": 5730 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8035, + "step": 5731 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4015, + "step": 5732 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4556, + "step": 5733 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.5577, + "step": 5734 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8556, + "step": 5735 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.6738, + "step": 5736 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8055, + "step": 5737 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.3551, + "step": 5738 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2115, + "step": 5739 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.8664, + "step": 5740 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2556, + "step": 5741 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2233, + "step": 5742 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.4341, + "step": 5743 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2092, + "step": 5744 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.2569, + "step": 5745 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.9095, + "step": 5746 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.6291, + "step": 5747 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.1212, + "step": 5748 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 1.5444, + "step": 5749 + }, + { + "epoch": 0.57, + "learning_rate": 0.0001, + "loss": 2.4493, + "step": 5750 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.0832, + "step": 5751 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.0366, + "step": 5752 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.4379, + "step": 5753 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.1337, + "step": 5754 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.862, + "step": 5755 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.4087, + "step": 5756 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.7378, + "step": 5757 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5101, + "step": 5758 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9284, + "step": 5759 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.8523, + "step": 5760 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1338, + "step": 5761 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.1386, + "step": 5762 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5083, + "step": 5763 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.15, + "step": 5764 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2094, + "step": 5765 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.4436, + "step": 5766 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1492, + "step": 5767 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2489, + "step": 5768 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9249, + "step": 5769 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.2001, + "step": 5770 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.5164, + "step": 5771 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.4906, + "step": 5772 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.0938, + "step": 5773 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.7582, + "step": 5774 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6674, + "step": 5775 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1438, + "step": 5776 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9729, + "step": 5777 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.8934, + "step": 5778 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1355, + "step": 5779 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.1452, + "step": 5780 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5479, + "step": 5781 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1264, + "step": 5782 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.3006, + "step": 5783 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6517, + "step": 5784 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1145, + "step": 5785 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.0312, + "step": 5786 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.4066, + "step": 5787 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.0141, + "step": 5788 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2159, + "step": 5789 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6537, + "step": 5790 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1335, + "step": 5791 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2675, + "step": 5792 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5931, + "step": 5793 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.2073, + "step": 5794 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.3187, + "step": 5795 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5494, + "step": 5796 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1387, + "step": 5797 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2415, + "step": 5798 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.504, + "step": 5799 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1436, + "step": 5800 + }, + { + "epoch": 0.58, + "eval_bleu": 15.33085185758889, + "eval_brevity_penalty": 0.5456766874028416, + "eval_loss": 1.5702519416809082, + "eval_runtime": 5.5515, + "eval_samples_per_second": 540.032, + "eval_steps_per_second": 1.081, + "step": 5800 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.7206, + "step": 5801 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.3996, + "step": 5802 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.0598, + "step": 5803 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2654, + "step": 5804 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.622, + "step": 5805 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.0581, + "step": 5806 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.3293, + "step": 5807 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.4261, + "step": 5808 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.0925, + "step": 5809 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9578, + "step": 5810 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.587, + "step": 5811 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.186, + "step": 5812 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.3601, + "step": 5813 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5671, + "step": 5814 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1129, + "step": 5815 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2148, + "step": 5816 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6024, + "step": 5817 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1473, + "step": 5818 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.207, + "step": 5819 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.256, + "step": 5820 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1493, + "step": 5821 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2543, + "step": 5822 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.4316, + "step": 5823 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1301, + "step": 5824 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2509, + "step": 5825 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.7705, + "step": 5826 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1857, + "step": 5827 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2034, + "step": 5828 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6067, + "step": 5829 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1793, + "step": 5830 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.3533, + "step": 5831 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.072, + "step": 5832 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1776, + "step": 5833 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.5158, + "step": 5834 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5825, + "step": 5835 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1526, + "step": 5836 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9566, + "step": 5837 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.629, + "step": 5838 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1533, + "step": 5839 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2534, + "step": 5840 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6743, + "step": 5841 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.1415, + "step": 5842 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.1893, + "step": 5843 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.6434, + "step": 5844 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.2498, + "step": 5845 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.9968, + "step": 5846 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5475, + "step": 5847 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.2356, + "step": 5848 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 2.2466, + "step": 5849 + }, + { + "epoch": 0.58, + "learning_rate": 0.0001, + "loss": 1.5053, + "step": 5850 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1538, + "step": 5851 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.6591, + "step": 5852 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.4014, + "step": 5853 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.182, + "step": 5854 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.8723, + "step": 5855 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6211, + "step": 5856 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2755, + "step": 5857 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3751, + "step": 5858 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.471, + "step": 5859 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1973, + "step": 5860 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.8108, + "step": 5861 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.4907, + "step": 5862 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1574, + "step": 5863 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3616, + "step": 5864 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6534, + "step": 5865 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1311, + "step": 5866 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2874, + "step": 5867 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2826, + "step": 5868 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2222, + "step": 5869 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.0641, + "step": 5870 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.7292, + "step": 5871 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2192, + "step": 5872 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2798, + "step": 5873 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.3138, + "step": 5874 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2147, + "step": 5875 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3168, + "step": 5876 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.9271, + "step": 5877 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1882, + "step": 5878 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.9335, + "step": 5879 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.4925, + "step": 5880 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1454, + "step": 5881 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2274, + "step": 5882 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.909, + "step": 5883 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.0955, + "step": 5884 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.903, + "step": 5885 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.5418, + "step": 5886 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2612, + "step": 5887 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.4182, + "step": 5888 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.7832, + "step": 5889 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.176, + "step": 5890 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.0745, + "step": 5891 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.8079, + "step": 5892 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1658, + "step": 5893 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2378, + "step": 5894 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.4635, + "step": 5895 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.0156, + "step": 5896 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.162, + "step": 5897 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.8791, + "step": 5898 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1425, + "step": 5899 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1695, + "step": 5900 + }, + { + "epoch": 0.59, + "eval_bleu": 15.35710840938046, + "eval_brevity_penalty": 0.543751795673914, + "eval_loss": 1.5689730644226074, + "eval_runtime": 5.1847, + "eval_samples_per_second": 578.239, + "eval_steps_per_second": 1.157, + "step": 5900 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.712, + "step": 5901 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.187, + "step": 5902 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3778, + "step": 5903 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.9376, + "step": 5904 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1547, + "step": 5905 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1269, + "step": 5906 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.5346, + "step": 5907 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2725, + "step": 5908 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3187, + "step": 5909 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.8485, + "step": 5910 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1249, + "step": 5911 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2132, + "step": 5912 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.7382, + "step": 5913 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1929, + "step": 5914 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.2453, + "step": 5915 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.919, + "step": 5916 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.3278, + "step": 5917 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1046, + "step": 5918 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.5238, + "step": 5919 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1901, + "step": 5920 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.4694, + "step": 5921 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.3734, + "step": 5922 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.3215, + "step": 5923 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.4945, + "step": 5924 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.5566, + "step": 5925 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1799, + "step": 5926 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.49, + "step": 5927 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6565, + "step": 5928 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1623, + "step": 5929 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.3096, + "step": 5930 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6217, + "step": 5931 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.0958, + "step": 5932 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.282, + "step": 5933 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.8229, + "step": 5934 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1454, + "step": 5935 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1518, + "step": 5936 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.5873, + "step": 5937 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1127, + "step": 5938 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.9818, + "step": 5939 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6795, + "step": 5940 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1814, + "step": 5941 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.226, + "step": 5942 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.6229, + "step": 5943 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.1599, + "step": 5944 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1621, + "step": 5945 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.7078, + "step": 5946 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2053, + "step": 5947 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.5945, + "step": 5948 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 2.1657, + "step": 5949 + }, + { + "epoch": 0.59, + "learning_rate": 0.0001, + "loss": 1.2279, + "step": 5950 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2316, + "step": 5951 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6832, + "step": 5952 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1501, + "step": 5953 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3412, + "step": 5954 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5755, + "step": 5955 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1951, + "step": 5956 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1391, + "step": 5957 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5419, + "step": 5958 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1258, + "step": 5959 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.9218, + "step": 5960 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6262, + "step": 5961 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2154, + "step": 5962 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1247, + "step": 5963 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6507, + "step": 5964 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2323, + "step": 5965 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.259, + "step": 5966 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6178, + "step": 5967 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2002, + "step": 5968 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2724, + "step": 5969 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.666, + "step": 5970 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2001, + "step": 5971 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.4113, + "step": 5972 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7359, + "step": 5973 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.3526, + "step": 5974 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1405, + "step": 5975 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6828, + "step": 5976 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2883, + "step": 5977 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3318, + "step": 5978 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.8729, + "step": 5979 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2123, + "step": 5980 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.4895, + "step": 5981 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6788, + "step": 5982 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1042, + "step": 5983 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.8687, + "step": 5984 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5601, + "step": 5985 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1748, + "step": 5986 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3298, + "step": 5987 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5876, + "step": 5988 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1219, + "step": 5989 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.4231, + "step": 5990 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.3715, + "step": 5991 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1346, + "step": 5992 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1442, + "step": 5993 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7358, + "step": 5994 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2098, + "step": 5995 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3959, + "step": 5996 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7539, + "step": 5997 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1874, + "step": 5998 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2711, + "step": 5999 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7295, + "step": 6000 + }, + { + "epoch": 0.6, + "eval_bleu": 15.35468819150106, + "eval_brevity_penalty": 0.5421217473472971, + "eval_loss": 1.5652563571929932, + "eval_runtime": 5.5821, + "eval_samples_per_second": 537.078, + "eval_steps_per_second": 1.075, + "step": 6000 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2441, + "step": 6001 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.8398, + "step": 6002 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7855, + "step": 6003 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1154, + "step": 6004 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1726, + "step": 6005 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.8932, + "step": 6006 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.08, + "step": 6007 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.4717, + "step": 6008 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5364, + "step": 6009 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.3864, + "step": 6010 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.5639, + "step": 6011 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7697, + "step": 6012 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.0885, + "step": 6013 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1087, + "step": 6014 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5171, + "step": 6015 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2487, + "step": 6016 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3691, + "step": 6017 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.4462, + "step": 6018 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1423, + "step": 6019 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3641, + "step": 6020 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.808, + "step": 6021 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1167, + "step": 6022 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.5569, + "step": 6023 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.3585, + "step": 6024 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2366, + "step": 6025 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2915, + "step": 6026 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.4496, + "step": 6027 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1666, + "step": 6028 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.9314, + "step": 6029 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.4713, + "step": 6030 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.0814, + "step": 6031 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.5548, + "step": 6032 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.4132, + "step": 6033 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1386, + "step": 6034 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.111, + "step": 6035 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.9633, + "step": 6036 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.0424, + "step": 6037 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.9528, + "step": 6038 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.5435, + "step": 6039 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.0948, + "step": 6040 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2611, + "step": 6041 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.7677, + "step": 6042 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.1473, + "step": 6043 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.2152, + "step": 6044 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.6824, + "step": 6045 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2103, + "step": 6046 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.0521, + "step": 6047 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.1025, + "step": 6048 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 1.2897, + "step": 6049 + }, + { + "epoch": 0.6, + "learning_rate": 0.0001, + "loss": 2.3245, + "step": 6050 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6615, + "step": 6051 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1153, + "step": 6052 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.5181, + "step": 6053 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.7134, + "step": 6054 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.272, + "step": 6055 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1035, + "step": 6056 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6465, + "step": 6057 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1141, + "step": 6058 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.3012, + "step": 6059 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.4077, + "step": 6060 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2223, + "step": 6061 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2698, + "step": 6062 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.761, + "step": 6063 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2137, + "step": 6064 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.0943, + "step": 6065 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6492, + "step": 6066 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.17, + "step": 6067 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.5998, + "step": 6068 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6704, + "step": 6069 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.0998, + "step": 6070 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.0738, + "step": 6071 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.5981, + "step": 6072 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1776, + "step": 6073 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.0664, + "step": 6074 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.4835, + "step": 6075 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2546, + "step": 6076 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1392, + "step": 6077 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.8693, + "step": 6078 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2169, + "step": 6079 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.3311, + "step": 6080 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6805, + "step": 6081 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1864, + "step": 6082 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1052, + "step": 6083 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6193, + "step": 6084 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2232, + "step": 6085 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1066, + "step": 6086 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.7314, + "step": 6087 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2043, + "step": 6088 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.3809, + "step": 6089 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.7412, + "step": 6090 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2189, + "step": 6091 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1789, + "step": 6092 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9042, + "step": 6093 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2294, + "step": 6094 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2343, + "step": 6095 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.0438, + "step": 6096 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2698, + "step": 6097 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.278, + "step": 6098 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9043, + "step": 6099 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.3033, + "step": 6100 + }, + { + "epoch": 0.61, + "eval_bleu": 15.308371085768934, + "eval_brevity_penalty": 0.5441961487950252, + "eval_loss": 1.564904808998108, + "eval_runtime": 5.7521, + "eval_samples_per_second": 521.203, + "eval_steps_per_second": 1.043, + "step": 6100 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9122, + "step": 6101 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6914, + "step": 6102 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2231, + "step": 6103 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.204, + "step": 6104 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.5148, + "step": 6105 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.3315, + "step": 6106 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2401, + "step": 6107 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.4004, + "step": 6108 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.3158, + "step": 6109 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1822, + "step": 6110 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.7093, + "step": 6111 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.244, + "step": 6112 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9613, + "step": 6113 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.5319, + "step": 6114 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2866, + "step": 6115 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2312, + "step": 6116 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6237, + "step": 6117 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2425, + "step": 6118 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1456, + "step": 6119 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2553, + "step": 6120 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.257, + "step": 6121 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9319, + "step": 6122 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.844, + "step": 6123 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1681, + "step": 6124 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.5628, + "step": 6125 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6655, + "step": 6126 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1998, + "step": 6127 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2664, + "step": 6128 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.8683, + "step": 6129 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1034, + "step": 6130 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.985, + "step": 6131 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.9639, + "step": 6132 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1693, + "step": 6133 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.3518, + "step": 6134 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.8094, + "step": 6135 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1729, + "step": 6136 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1218, + "step": 6137 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.8512, + "step": 6138 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.0651, + "step": 6139 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.1679, + "step": 6140 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.483, + "step": 6141 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 0.9819, + "step": 6142 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2407, + "step": 6143 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.6926, + "step": 6144 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1302, + "step": 6145 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 2.2799, + "step": 6146 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.2691, + "step": 6147 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.1795, + "step": 6148 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.874, + "step": 6149 + }, + { + "epoch": 0.61, + "learning_rate": 0.0001, + "loss": 1.5097, + "step": 6150 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.1237, + "step": 6151 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.3094, + "step": 6152 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.6378, + "step": 6153 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.3252, + "step": 6154 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.378, + "step": 6155 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7802, + "step": 6156 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.2731, + "step": 6157 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9168, + "step": 6158 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.6502, + "step": 6159 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.089, + "step": 6160 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.3493, + "step": 6161 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.5181, + "step": 6162 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.2869, + "step": 6163 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0855, + "step": 6164 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.8204, + "step": 6165 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.1794, + "step": 6166 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.8988, + "step": 6167 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.3283, + "step": 6168 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.2249, + "step": 6169 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0578, + "step": 6170 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7748, + "step": 6171 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.0662, + "step": 6172 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.8784, + "step": 6173 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7048, + "step": 6174 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.0743, + "step": 6175 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1392, + "step": 6176 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9312, + "step": 6177 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.1115, + "step": 6178 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.221, + "step": 6179 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7794, + "step": 6180 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.0951, + "step": 6181 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9864, + "step": 6182 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9082, + "step": 6183 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.0463, + "step": 6184 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.204, + "step": 6185 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7734, + "step": 6186 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.1023, + "step": 6187 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1081, + "step": 6188 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.4065, + "step": 6189 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 0.9695, + "step": 6190 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1854, + "step": 6191 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7236, + "step": 6192 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.0726, + "step": 6193 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0198, + "step": 6194 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9125, + "step": 6195 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.3879, + "step": 6196 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1747, + "step": 6197 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7089, + "step": 6198 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.3434, + "step": 6199 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.396, + "step": 6200 + }, + { + "epoch": 0.62, + "eval_bleu": 15.559377444906215, + "eval_brevity_penalty": 0.5439634034040142, + "eval_loss": 1.5592288970947266, + "eval_runtime": 5.8334, + "eval_samples_per_second": 513.933, + "eval_steps_per_second": 1.029, + "step": 6200 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.449, + "step": 6201 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.7155, + "step": 6202 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0825, + "step": 6203 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2742, + "step": 6204 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.6915, + "step": 6205 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1405, + "step": 6206 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.3088, + "step": 6207 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.8712, + "step": 6208 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.019, + "step": 6209 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.8524, + "step": 6210 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0503, + "step": 6211 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2695, + "step": 6212 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.4529, + "step": 6213 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.6558, + "step": 6214 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9851, + "step": 6215 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2554, + "step": 6216 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.961, + "step": 6217 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1583, + "step": 6218 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2894, + "step": 6219 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0999, + "step": 6220 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2346, + "step": 6221 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.7975, + "step": 6222 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.8449, + "step": 6223 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0733, + "step": 6224 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.7667, + "step": 6225 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.6656, + "step": 6226 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2446, + "step": 6227 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.8872, + "step": 6228 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.014, + "step": 6229 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.4172, + "step": 6230 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.5225, + "step": 6231 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1246, + "step": 6232 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0691, + "step": 6233 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.2956, + "step": 6234 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 1.9631, + "step": 6235 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.194, + "step": 6236 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.6922, + "step": 6237 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.071, + "step": 6238 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1473, + "step": 6239 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.5579, + "step": 6240 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0004, + "step": 6241 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1332, + "step": 6242 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.4151, + "step": 6243 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.1797, + "step": 6244 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.0839, + "step": 6245 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 3.1403, + "step": 6246 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.6387, + "step": 6247 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.5448, + "step": 6248 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.5347, + "step": 6249 + }, + { + "epoch": 0.62, + "learning_rate": 0.0001, + "loss": 2.3065, + "step": 6250 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0694, + "step": 6251 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.3687, + "step": 6252 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1657, + "step": 6253 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1887, + "step": 6254 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.4516, + "step": 6255 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2199, + "step": 6256 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0821, + "step": 6257 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1932, + "step": 6258 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0405, + "step": 6259 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1542, + "step": 6260 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 3.0693, + "step": 6261 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.8572, + "step": 6262 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.3895, + "step": 6263 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2371, + "step": 6264 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.944, + "step": 6265 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2081, + "step": 6266 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.487, + "step": 6267 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.829, + "step": 6268 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2379, + "step": 6269 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2527, + "step": 6270 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1719, + "step": 6271 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1988, + "step": 6272 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.6405, + "step": 6273 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.8269, + "step": 6274 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.5132, + "step": 6275 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.1956, + "step": 6276 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.8429, + "step": 6277 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.9328, + "step": 6278 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3566, + "step": 6279 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.7495, + "step": 6280 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2431, + "step": 6281 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.4921, + "step": 6282 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.7499, + "step": 6283 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.252, + "step": 6284 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2977, + "step": 6285 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.8132, + "step": 6286 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.8993, + "step": 6287 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.5351, + "step": 6288 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.8658, + "step": 6289 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.056, + "step": 6290 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.5471, + "step": 6291 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 0.9842, + "step": 6292 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.3253, + "step": 6293 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.6564, + "step": 6294 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.211, + "step": 6295 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2585, + "step": 6296 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.9456, + "step": 6297 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2523, + "step": 6298 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2077, + "step": 6299 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.133, + "step": 6300 + }, + { + "epoch": 0.63, + "eval_bleu": 15.368929527312208, + "eval_brevity_penalty": 0.5419946809784227, + "eval_loss": 1.5633881092071533, + "eval_runtime": 5.8425, + "eval_samples_per_second": 513.136, + "eval_steps_per_second": 1.027, + "step": 6300 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2596, + "step": 6301 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.2432, + "step": 6302 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.8809, + "step": 6303 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2479, + "step": 6304 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 3.0127, + "step": 6305 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2967, + "step": 6306 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3415, + "step": 6307 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1995, + "step": 6308 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.8026, + "step": 6309 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2245, + "step": 6310 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0921, + "step": 6311 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1635, + "step": 6312 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3807, + "step": 6313 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.5828, + "step": 6314 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.8304, + "step": 6315 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2458, + "step": 6316 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1588, + "step": 6317 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.7829, + "step": 6318 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2138, + "step": 6319 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.9819, + "step": 6320 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.603, + "step": 6321 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.1734, + "step": 6322 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.5103, + "step": 6323 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.7942, + "step": 6324 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3017, + "step": 6325 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.3024, + "step": 6326 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.7436, + "step": 6327 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.4501, + "step": 6328 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.3047, + "step": 6329 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.7834, + "step": 6330 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3576, + "step": 6331 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.5103, + "step": 6332 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.6078, + "step": 6333 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.3383, + "step": 6334 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.407, + "step": 6335 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.579, + "step": 6336 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.1786, + "step": 6337 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.168, + "step": 6338 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0704, + "step": 6339 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.0963, + "step": 6340 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.0459, + "step": 6341 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1929, + "step": 6342 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.0912, + "step": 6343 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1824, + "step": 6344 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.6577, + "step": 6345 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.2559, + "step": 6346 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 2.1346, + "step": 6347 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.736, + "step": 6348 + }, + { + "epoch": 0.63, + "learning_rate": 0.0001, + "loss": 1.0761, + "step": 6349 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.2639, + "step": 6350 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.7232, + "step": 6351 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1406, + "step": 6352 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.4171, + "step": 6353 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6305, + "step": 6354 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3158, + "step": 6355 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.3043, + "step": 6356 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8211, + "step": 6357 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3043, + "step": 6358 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.4474, + "step": 6359 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0698, + "step": 6360 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.326, + "step": 6361 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0622, + "step": 6362 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6303, + "step": 6363 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1214, + "step": 6364 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.3401, + "step": 6365 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3533, + "step": 6366 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.0835, + "step": 6367 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0804, + "step": 6368 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.724, + "step": 6369 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.0963, + "step": 6370 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0523, + "step": 6371 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.7188, + "step": 6372 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.12, + "step": 6373 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.57, + "step": 6374 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.9301, + "step": 6375 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3734, + "step": 6376 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.2955, + "step": 6377 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.898, + "step": 6378 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3507, + "step": 6379 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1643, + "step": 6380 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8377, + "step": 6381 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2496, + "step": 6382 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0565, + "step": 6383 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.4764, + "step": 6384 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2405, + "step": 6385 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6755, + "step": 6386 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8805, + "step": 6387 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2124, + "step": 6388 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0839, + "step": 6389 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.9008, + "step": 6390 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1253, + "step": 6391 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1613, + "step": 6392 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.7682, + "step": 6393 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1404, + "step": 6394 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8439, + "step": 6395 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.5975, + "step": 6396 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.16, + "step": 6397 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.0074, + "step": 6398 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.687, + "step": 6399 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1775, + "step": 6400 + }, + { + "epoch": 0.64, + "eval_bleu": 15.486854600406513, + "eval_brevity_penalty": 0.5388793256951286, + "eval_loss": 1.563927412033081, + "eval_runtime": 5.173, + "eval_samples_per_second": 579.545, + "eval_steps_per_second": 1.16, + "step": 6400 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1131, + "step": 6401 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.7902, + "step": 6402 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2136, + "step": 6403 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8589, + "step": 6404 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6359, + "step": 6405 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1568, + "step": 6406 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.3081, + "step": 6407 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8723, + "step": 6408 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1719, + "step": 6409 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.274, + "step": 6410 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.9142, + "step": 6411 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2461, + "step": 6412 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1381, + "step": 6413 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6199, + "step": 6414 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1629, + "step": 6415 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1181, + "step": 6416 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6163, + "step": 6417 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.0502, + "step": 6418 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.9384, + "step": 6419 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6856, + "step": 6420 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.247, + "step": 6421 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.2737, + "step": 6422 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8576, + "step": 6423 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2227, + "step": 6424 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.2107, + "step": 6425 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6653, + "step": 6426 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.2723, + "step": 6427 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.238, + "step": 6428 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.7399, + "step": 6429 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.108, + "step": 6430 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.2183, + "step": 6431 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.5335, + "step": 6432 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.4325, + "step": 6433 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 2.1317, + "step": 6434 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.4342, + "step": 6435 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1984, + "step": 6436 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.987, + "step": 6437 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.5393, + "step": 6438 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1802, + "step": 6439 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.826, + "step": 6440 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.3336, + "step": 6441 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1841, + "step": 6442 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6315, + "step": 6443 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6452, + "step": 6444 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.0751, + "step": 6445 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.9704, + "step": 6446 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.6605, + "step": 6447 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.1115, + "step": 6448 + }, + { + "epoch": 0.64, + "learning_rate": 0.0001, + "loss": 1.8463, + "step": 6449 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.5377, + "step": 6450 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1391, + "step": 6451 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7022, + "step": 6452 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.6994, + "step": 6453 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1318, + "step": 6454 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.972, + "step": 6455 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7658, + "step": 6456 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1891, + "step": 6457 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.3598, + "step": 6458 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7023, + "step": 6459 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2807, + "step": 6460 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.5117, + "step": 6461 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1322, + "step": 6462 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.3943, + "step": 6463 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.6169, + "step": 6464 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0627, + "step": 6465 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2117, + "step": 6466 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0275, + "step": 6467 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8513, + "step": 6468 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2937, + "step": 6469 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.2551, + "step": 6470 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8422, + "step": 6471 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1214, + "step": 6472 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.6409, + "step": 6473 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9226, + "step": 6474 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2942, + "step": 6475 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9499, + "step": 6476 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.6947, + "step": 6477 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2893, + "step": 6478 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.114, + "step": 6479 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.6009, + "step": 6480 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.0831, + "step": 6481 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8245, + "step": 6482 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.764, + "step": 6483 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1687, + "step": 6484 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.408, + "step": 6485 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.6774, + "step": 6486 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1613, + "step": 6487 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.3027, + "step": 6488 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7566, + "step": 6489 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2936, + "step": 6490 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9734, + "step": 6491 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7441, + "step": 6492 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.4107, + "step": 6493 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.2592, + "step": 6494 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9305, + "step": 6495 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2785, + "step": 6496 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0574, + "step": 6497 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.4929, + "step": 6498 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2123, + "step": 6499 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0793, + "step": 6500 + }, + { + "epoch": 0.65, + "eval_bleu": 15.632011173395197, + "eval_brevity_penalty": 0.5452960714391446, + "eval_loss": 1.5540847778320312, + "eval_runtime": 5.2378, + "eval_samples_per_second": 572.375, + "eval_steps_per_second": 1.146, + "step": 6500 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.862, + "step": 6501 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2356, + "step": 6502 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0774, + "step": 6503 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8481, + "step": 6504 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2657, + "step": 6505 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.2436, + "step": 6506 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.699, + "step": 6507 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2821, + "step": 6508 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1571, + "step": 6509 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9108, + "step": 6510 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1762, + "step": 6511 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1163, + "step": 6512 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.34, + "step": 6513 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.1891, + "step": 6514 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9316, + "step": 6515 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.5223, + "step": 6516 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2843, + "step": 6517 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0564, + "step": 6518 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8008, + "step": 6519 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2981, + "step": 6520 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0763, + "step": 6521 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9776, + "step": 6522 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2266, + "step": 6523 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7991, + "step": 6524 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.2695, + "step": 6525 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.099, + "step": 6526 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8509, + "step": 6527 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7467, + "step": 6528 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2669, + "step": 6529 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.3517, + "step": 6530 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.774, + "step": 6531 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.176, + "step": 6532 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.2594, + "step": 6533 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.5425, + "step": 6534 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.4068, + "step": 6535 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1165, + "step": 6536 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.8026, + "step": 6537 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.2282, + "step": 6538 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.3154, + "step": 6539 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.7924, + "step": 6540 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.5877, + "step": 6541 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1612, + "step": 6542 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.9364, + "step": 6543 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.4314, + "step": 6544 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.244, + "step": 6545 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1196, + "step": 6546 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 1.5999, + "step": 6547 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.1855, + "step": 6548 + }, + { + "epoch": 0.65, + "learning_rate": 0.0001, + "loss": 2.0052, + "step": 6549 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.419, + "step": 6550 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.2414, + "step": 6551 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.4656, + "step": 6552 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.4292, + "step": 6553 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0709, + "step": 6554 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9317, + "step": 6555 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1053, + "step": 6556 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0434, + "step": 6557 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7265, + "step": 6558 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1362, + "step": 6559 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8465, + "step": 6560 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5444, + "step": 6561 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2118, + "step": 6562 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.031, + "step": 6563 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5139, + "step": 6564 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2999, + "step": 6565 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9301, + "step": 6566 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7658, + "step": 6567 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2952, + "step": 6568 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.2764, + "step": 6569 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8253, + "step": 6570 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5199, + "step": 6571 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0626, + "step": 6572 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8564, + "step": 6573 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.4898, + "step": 6574 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.3456, + "step": 6575 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0808, + "step": 6576 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2187, + "step": 6577 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9887, + "step": 6578 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8279, + "step": 6579 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1934, + "step": 6580 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0546, + "step": 6581 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7193, + "step": 6582 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1844, + "step": 6583 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0689, + "step": 6584 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7543, + "step": 6585 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0828, + "step": 6586 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8716, + "step": 6587 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5268, + "step": 6588 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1014, + "step": 6589 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.1724, + "step": 6590 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5525, + "step": 6591 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0698, + "step": 6592 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.898, + "step": 6593 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7589, + "step": 6594 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0774, + "step": 6595 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.2266, + "step": 6596 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.5614, + "step": 6597 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1106, + "step": 6598 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9282, + "step": 6599 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7569, + "step": 6600 + }, + { + "epoch": 0.66, + "eval_bleu": 15.740532965043307, + "eval_brevity_penalty": 0.5429051649170827, + "eval_loss": 1.5587894916534424, + "eval_runtime": 5.3083, + "eval_samples_per_second": 564.776, + "eval_steps_per_second": 1.13, + "step": 6600 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1296, + "step": 6601 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.197, + "step": 6602 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7875, + "step": 6603 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2409, + "step": 6604 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.2597, + "step": 6605 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7015, + "step": 6606 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0889, + "step": 6607 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.2521, + "step": 6608 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0237, + "step": 6609 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1706, + "step": 6610 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9918, + "step": 6611 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2915, + "step": 6612 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.127, + "step": 6613 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.087, + "step": 6614 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.6973, + "step": 6615 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1385, + "step": 6616 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0811, + "step": 6617 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8273, + "step": 6618 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0845, + "step": 6619 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8539, + "step": 6620 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7315, + "step": 6621 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2083, + "step": 6622 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.073, + "step": 6623 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.6901, + "step": 6624 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1993, + "step": 6625 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.0163, + "step": 6626 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7048, + "step": 6627 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1999, + "step": 6628 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.047, + "step": 6629 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.6631, + "step": 6630 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0171, + "step": 6631 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9815, + "step": 6632 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7485, + "step": 6633 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0146, + "step": 6634 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.1968, + "step": 6635 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.121, + "step": 6636 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1323, + "step": 6637 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7591, + "step": 6638 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.6133, + "step": 6639 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0473, + "step": 6640 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.9493, + "step": 6641 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7471, + "step": 6642 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.0596, + "step": 6643 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.1796, + "step": 6644 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.8001, + "step": 6645 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.1844, + "step": 6646 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 2.4851, + "step": 6647 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.7239, + "step": 6648 + }, + { + "epoch": 0.66, + "learning_rate": 0.0001, + "loss": 1.2172, + "step": 6649 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0726, + "step": 6650 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8343, + "step": 6651 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2708, + "step": 6652 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.149, + "step": 6653 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.656, + "step": 6654 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1453, + "step": 6655 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8766, + "step": 6656 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.7609, + "step": 6657 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2442, + "step": 6658 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0943, + "step": 6659 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.6614, + "step": 6660 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1857, + "step": 6661 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.9117, + "step": 6662 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1567, + "step": 6663 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0852, + "step": 6664 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2707, + "step": 6665 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.6627, + "step": 6666 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2258, + "step": 6667 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0371, + "step": 6668 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.5966, + "step": 6669 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0325, + "step": 6670 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2043, + "step": 6671 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.867, + "step": 6672 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0205, + "step": 6673 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.244, + "step": 6674 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8421, + "step": 6675 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1233, + "step": 6676 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.113, + "step": 6677 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.858, + "step": 6678 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1129, + "step": 6679 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1913, + "step": 6680 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.9045, + "step": 6681 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2324, + "step": 6682 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0666, + "step": 6683 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.7598, + "step": 6684 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0452, + "step": 6685 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.3049, + "step": 6686 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.7419, + "step": 6687 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1416, + "step": 6688 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2281, + "step": 6689 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8062, + "step": 6690 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2566, + "step": 6691 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.3316, + "step": 6692 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3175, + "step": 6693 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1464, + "step": 6694 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0798, + "step": 6695 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3686, + "step": 6696 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1039, + "step": 6697 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0326, + "step": 6698 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.6402, + "step": 6699 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1035, + "step": 6700 + }, + { + "epoch": 0.67, + "eval_bleu": 15.701079516734605, + "eval_brevity_penalty": 0.5449788417370042, + "eval_loss": 1.5519760847091675, + "eval_runtime": 5.6522, + "eval_samples_per_second": 530.409, + "eval_steps_per_second": 1.062, + "step": 6700 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0791, + "step": 6701 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.7988, + "step": 6702 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0805, + "step": 6703 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8326, + "step": 6704 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.4522, + "step": 6705 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2028, + "step": 6706 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1767, + "step": 6707 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.736, + "step": 6708 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1026, + "step": 6709 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1664, + "step": 6710 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.822, + "step": 6711 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0773, + "step": 6712 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8091, + "step": 6713 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2822, + "step": 6714 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3367, + "step": 6715 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.141, + "step": 6716 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8346, + "step": 6717 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3659, + "step": 6718 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.9297, + "step": 6719 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1508, + "step": 6720 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1623, + "step": 6721 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0598, + "step": 6722 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.6983, + "step": 6723 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3092, + "step": 6724 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2119, + "step": 6725 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.2259, + "step": 6726 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.063, + "step": 6727 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1547, + "step": 6728 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8086, + "step": 6729 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1424, + "step": 6730 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8652, + "step": 6731 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.5379, + "step": 6732 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.1758, + "step": 6733 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0689, + "step": 6734 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0823, + "step": 6735 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2519, + "step": 6736 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.0818, + "step": 6737 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.9022, + "step": 6738 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.0951, + "step": 6739 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1693, + "step": 6740 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.7069, + "step": 6741 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.3545, + "step": 6742 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1194, + "step": 6743 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.8529, + "step": 6744 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.168, + "step": 6745 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.3615, + "step": 6746 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.4537, + "step": 6747 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 1.2324, + "step": 6748 + }, + { + "epoch": 0.67, + "learning_rate": 0.0001, + "loss": 2.1954, + "step": 6749 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.6767, + "step": 6750 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2095, + "step": 6751 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9998, + "step": 6752 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.7581, + "step": 6753 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1454, + "step": 6754 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0825, + "step": 6755 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.7603, + "step": 6756 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2395, + "step": 6757 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.2974, + "step": 6758 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9609, + "step": 6759 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2231, + "step": 6760 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.1818, + "step": 6761 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8605, + "step": 6762 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1509, + "step": 6763 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.122, + "step": 6764 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.6864, + "step": 6765 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2326, + "step": 6766 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.588, + "step": 6767 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8161, + "step": 6768 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2167, + "step": 6769 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9601, + "step": 6770 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.735, + "step": 6771 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.0236, + "step": 6772 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.2605, + "step": 6773 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.6157, + "step": 6774 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2415, + "step": 6775 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9941, + "step": 6776 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9119, + "step": 6777 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2218, + "step": 6778 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0273, + "step": 6779 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.957, + "step": 6780 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2976, + "step": 6781 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0479, + "step": 6782 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.4902, + "step": 6783 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1272, + "step": 6784 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.1464, + "step": 6785 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.7075, + "step": 6786 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.3754, + "step": 6787 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.4756, + "step": 6788 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9895, + "step": 6789 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2281, + "step": 6790 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9958, + "step": 6791 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.4148, + "step": 6792 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1561, + "step": 6793 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9065, + "step": 6794 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9443, + "step": 6795 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.0304, + "step": 6796 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.2829, + "step": 6797 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.782, + "step": 6798 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1135, + "step": 6799 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.5799, + "step": 6800 + }, + { + "epoch": 0.68, + "eval_bleu": 15.920274182132074, + "eval_brevity_penalty": 0.5489726208746478, + "eval_loss": 1.5517301559448242, + "eval_runtime": 5.6428, + "eval_samples_per_second": 531.298, + "eval_steps_per_second": 1.063, + "step": 6800 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.4818, + "step": 6801 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1406, + "step": 6802 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.15, + "step": 6803 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.6416, + "step": 6804 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.3085, + "step": 6805 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9049, + "step": 6806 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.4979, + "step": 6807 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1114, + "step": 6808 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9569, + "step": 6809 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.7288, + "step": 6810 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1717, + "step": 6811 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.3778, + "step": 6812 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8539, + "step": 6813 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.0382, + "step": 6814 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8878, + "step": 6815 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8852, + "step": 6816 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1404, + "step": 6817 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.2453, + "step": 6818 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0078, + "step": 6819 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2003, + "step": 6820 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0311, + "step": 6821 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.6764, + "step": 6822 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.0985, + "step": 6823 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.1306, + "step": 6824 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.7714, + "step": 6825 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 0.9737, + "step": 6826 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.1855, + "step": 6827 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.5047, + "step": 6828 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2474, + "step": 6829 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.2123, + "step": 6830 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9981, + "step": 6831 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2466, + "step": 6832 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.167, + "step": 6833 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8339, + "step": 6834 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2265, + "step": 6835 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.4573, + "step": 6836 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.3923, + "step": 6837 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.1337, + "step": 6838 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.0777, + "step": 6839 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8939, + "step": 6840 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.2605, + "step": 6841 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.1556, + "step": 6842 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8327, + "step": 6843 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.0147, + "step": 6844 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 2.4946, + "step": 6845 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.8169, + "step": 6846 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.211, + "step": 6847 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9602, + "step": 6848 + }, + { + "epoch": 0.68, + "learning_rate": 0.0001, + "loss": 1.9064, + "step": 6849 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1052, + "step": 6850 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.218, + "step": 6851 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7275, + "step": 6852 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2039, + "step": 6853 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1294, + "step": 6854 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.6771, + "step": 6855 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1184, + "step": 6856 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.2847, + "step": 6857 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8143, + "step": 6858 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2057, + "step": 6859 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9462, + "step": 6860 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.5073, + "step": 6861 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1291, + "step": 6862 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7741, + "step": 6863 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.535, + "step": 6864 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2773, + "step": 6865 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.2127, + "step": 6866 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0327, + "step": 6867 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1679, + "step": 6868 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.4693, + "step": 6869 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.4098, + "step": 6870 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2308, + "step": 6871 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.3566, + "step": 6872 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9776, + "step": 6873 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.0834, + "step": 6874 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8776, + "step": 6875 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9153, + "step": 6876 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.047, + "step": 6877 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1446, + "step": 6878 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.092, + "step": 6879 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1314, + "step": 6880 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9989, + "step": 6881 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8567, + "step": 6882 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2146, + "step": 6883 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0942, + "step": 6884 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7217, + "step": 6885 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1362, + "step": 6886 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9266, + "step": 6887 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9664, + "step": 6888 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2826, + "step": 6889 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0373, + "step": 6890 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.611, + "step": 6891 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1799, + "step": 6892 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.239, + "step": 6893 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7056, + "step": 6894 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1646, + "step": 6895 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8852, + "step": 6896 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9268, + "step": 6897 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1668, + "step": 6898 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1331, + "step": 6899 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7737, + "step": 6900 + }, + { + "epoch": 0.69, + "eval_bleu": 15.899225858342511, + "eval_brevity_penalty": 0.5480223760992211, + "eval_loss": 1.5473320484161377, + "eval_runtime": 6.012, + "eval_samples_per_second": 498.67, + "eval_steps_per_second": 0.998, + "step": 6900 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.248, + "step": 6901 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.2038, + "step": 6902 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9196, + "step": 6903 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.4674, + "step": 6904 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.2261, + "step": 6905 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8571, + "step": 6906 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2282, + "step": 6907 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8166, + "step": 6908 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7521, + "step": 6909 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.3091, + "step": 6910 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.4308, + "step": 6911 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7129, + "step": 6912 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.172, + "step": 6913 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0849, + "step": 6914 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7895, + "step": 6915 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2893, + "step": 6916 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9197, + "step": 6917 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7691, + "step": 6918 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1642, + "step": 6919 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1483, + "step": 6920 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.5924, + "step": 6921 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2993, + "step": 6922 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0893, + "step": 6923 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0782, + "step": 6924 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1596, + "step": 6925 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1149, + "step": 6926 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.5777, + "step": 6927 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.1851, + "step": 6928 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9072, + "step": 6929 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.6495, + "step": 6930 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2091, + "step": 6931 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.3972, + "step": 6932 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7599, + "step": 6933 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.2502, + "step": 6934 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0399, + "step": 6935 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.7, + "step": 6936 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.3077, + "step": 6937 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1765, + "step": 6938 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.6074, + "step": 6939 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.153, + "step": 6940 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.1364, + "step": 6941 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.8512, + "step": 6942 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.3512, + "step": 6943 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9156, + "step": 6944 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 2.0881, + "step": 6945 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.3957, + "step": 6946 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.6571, + "step": 6947 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.824, + "step": 6948 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.0374, + "step": 6949 + }, + { + "epoch": 0.69, + "learning_rate": 0.0001, + "loss": 1.9526, + "step": 6950 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6588, + "step": 6951 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2758, + "step": 6952 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9938, + "step": 6953 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7978, + "step": 6954 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2654, + "step": 6955 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.454, + "step": 6956 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7413, + "step": 6957 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2727, + "step": 6958 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.4307, + "step": 6959 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.794, + "step": 6960 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1647, + "step": 6961 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.3537, + "step": 6962 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.639, + "step": 6963 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.3075, + "step": 6964 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0257, + "step": 6965 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7466, + "step": 6966 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.0362, + "step": 6967 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2872, + "step": 6968 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2055, + "step": 6969 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2801, + "step": 6970 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0131, + "step": 6971 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.8814, + "step": 6972 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2501, + "step": 6973 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2133, + "step": 6974 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7585, + "step": 6975 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1884, + "step": 6976 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.3404, + "step": 6977 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9205, + "step": 6978 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.116, + "step": 6979 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1851, + "step": 6980 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.8494, + "step": 6981 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.3384, + "step": 6982 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2737, + "step": 6983 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.5961, + "step": 6984 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.4314, + "step": 6985 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1154, + "step": 6986 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1479, + "step": 6987 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.3996, + "step": 6988 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.032, + "step": 6989 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6834, + "step": 6990 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1254, + "step": 6991 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2418, + "step": 6992 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.3084, + "step": 6993 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.3763, + "step": 6994 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1365, + "step": 6995 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7586, + "step": 6996 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1858, + "step": 6997 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0614, + "step": 6998 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2313, + "step": 6999 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.3071, + "step": 7000 + }, + { + "epoch": 0.7, + "eval_bleu": 15.714045632541179, + "eval_brevity_penalty": 0.544576952617431, + "eval_loss": 1.5491451025009155, + "eval_runtime": 5.1943, + "eval_samples_per_second": 577.168, + "eval_steps_per_second": 1.155, + "step": 7000 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2596, + "step": 7001 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7031, + "step": 7002 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1513, + "step": 7003 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0934, + "step": 7004 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.8206, + "step": 7005 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.173, + "step": 7006 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1003, + "step": 7007 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7043, + "step": 7008 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2172, + "step": 7009 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9155, + "step": 7010 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6655, + "step": 7011 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2206, + "step": 7012 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7145, + "step": 7013 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6385, + "step": 7014 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1845, + "step": 7015 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.254, + "step": 7016 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6786, + "step": 7017 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1534, + "step": 7018 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.3771, + "step": 7019 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9408, + "step": 7020 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.137, + "step": 7021 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1821, + "step": 7022 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7164, + "step": 7023 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2587, + "step": 7024 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.047, + "step": 7025 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0778, + "step": 7026 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2485, + "step": 7027 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0994, + "step": 7028 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7331, + "step": 7029 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.0807, + "step": 7030 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2009, + "step": 7031 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.8153, + "step": 7032 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2218, + "step": 7033 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2516, + "step": 7034 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6565, + "step": 7035 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 0.9474, + "step": 7036 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.1599, + "step": 7037 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0718, + "step": 7038 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.2115, + "step": 7039 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.0033, + "step": 7040 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.7316, + "step": 7041 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.0182, + "step": 7042 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 2.2423, + "step": 7043 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.5768, + "step": 7044 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1477, + "step": 7045 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9918, + "step": 7046 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9051, + "step": 7047 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.1816, + "step": 7048 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.9808, + "step": 7049 + }, + { + "epoch": 0.7, + "learning_rate": 0.0001, + "loss": 1.6169, + "step": 7050 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2132, + "step": 7051 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.244, + "step": 7052 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0633, + "step": 7053 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.3188, + "step": 7054 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2647, + "step": 7055 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.8397, + "step": 7056 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.3505, + "step": 7057 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.1944, + "step": 7058 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.6457, + "step": 7059 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2385, + "step": 7060 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0833, + "step": 7061 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1573, + "step": 7062 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1273, + "step": 7063 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.3678, + "step": 7064 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.846, + "step": 7065 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.0541, + "step": 7066 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0617, + "step": 7067 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.6812, + "step": 7068 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.0518, + "step": 7069 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0576, + "step": 7070 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.6466, + "step": 7071 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1111, + "step": 7072 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0836, + "step": 7073 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.6273, + "step": 7074 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.0915, + "step": 7075 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0225, + "step": 7076 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.793, + "step": 7077 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.0613, + "step": 7078 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0805, + "step": 7079 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.5966, + "step": 7080 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1133, + "step": 7081 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9411, + "step": 7082 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.5844, + "step": 7083 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1436, + "step": 7084 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9515, + "step": 7085 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.8471, + "step": 7086 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1841, + "step": 7087 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0934, + "step": 7088 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7699, + "step": 7089 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2557, + "step": 7090 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0445, + "step": 7091 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.8631, + "step": 7092 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1769, + "step": 7093 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2915, + "step": 7094 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2865, + "step": 7095 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1987, + "step": 7096 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.3943, + "step": 7097 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7674, + "step": 7098 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.3096, + "step": 7099 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2214, + "step": 7100 + }, + { + "epoch": 0.71, + "eval_bleu": 15.935953969097069, + "eval_brevity_penalty": 0.5478534007121426, + "eval_loss": 1.5459554195404053, + "eval_runtime": 5.735, + "eval_samples_per_second": 522.757, + "eval_steps_per_second": 1.046, + "step": 7100 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7099, + "step": 7101 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1072, + "step": 7102 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.8917, + "step": 7103 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7143, + "step": 7104 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2153, + "step": 7105 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.3925, + "step": 7106 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0861, + "step": 7107 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1152, + "step": 7108 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0352, + "step": 7109 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.4107, + "step": 7110 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2047, + "step": 7111 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.1, + "step": 7112 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.5505, + "step": 7113 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.16, + "step": 7114 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.6962, + "step": 7115 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7171, + "step": 7116 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.0745, + "step": 7117 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.0692, + "step": 7118 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7685, + "step": 7119 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1782, + "step": 7120 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9805, + "step": 7121 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.3049, + "step": 7122 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2317, + "step": 7123 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.3003, + "step": 7124 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9502, + "step": 7125 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1922, + "step": 7126 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.325, + "step": 7127 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.5262, + "step": 7128 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1645, + "step": 7129 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9592, + "step": 7130 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9333, + "step": 7131 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.2054, + "step": 7132 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.1619, + "step": 7133 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.989, + "step": 7134 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1844, + "step": 7135 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2425, + "step": 7136 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.9151, + "step": 7137 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1659, + "step": 7138 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.7275, + "step": 7139 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.1834, + "step": 7140 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1539, + "step": 7141 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.275, + "step": 7142 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.5605, + "step": 7143 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.1527, + "step": 7144 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.1726, + "step": 7145 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.844, + "step": 7146 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.3831, + "step": 7147 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 2.2109, + "step": 7148 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.8947, + "step": 7149 + }, + { + "epoch": 0.71, + "learning_rate": 0.0001, + "loss": 1.3811, + "step": 7150 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.1675, + "step": 7151 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6947, + "step": 7152 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1618, + "step": 7153 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.1511, + "step": 7154 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7307, + "step": 7155 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.0206, + "step": 7156 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.1799, + "step": 7157 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.022, + "step": 7158 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1972, + "step": 7159 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.002, + "step": 7160 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0873, + "step": 7161 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.4614, + "step": 7162 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8709, + "step": 7163 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7246, + "step": 7164 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2509, + "step": 7165 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0622, + "step": 7166 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.912, + "step": 7167 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1691, + "step": 7168 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.4014, + "step": 7169 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0259, + "step": 7170 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.4319, + "step": 7171 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.3158, + "step": 7172 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6876, + "step": 7173 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2987, + "step": 7174 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6019, + "step": 7175 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.792, + "step": 7176 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.3114, + "step": 7177 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.875, + "step": 7178 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0062, + "step": 7179 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.217, + "step": 7180 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8782, + "step": 7181 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9192, + "step": 7182 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2735, + "step": 7183 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0557, + "step": 7184 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8339, + "step": 7185 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.3604, + "step": 7186 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9278, + "step": 7187 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7801, + "step": 7188 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1417, + "step": 7189 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0776, + "step": 7190 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7857, + "step": 7191 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2729, + "step": 7192 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8604, + "step": 7193 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.5119, + "step": 7194 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2094, + "step": 7195 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7739, + "step": 7196 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.3345, + "step": 7197 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1156, + "step": 7198 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8804, + "step": 7199 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7848, + "step": 7200 + }, + { + "epoch": 0.72, + "eval_bleu": 15.933820778619712, + "eval_brevity_penalty": 0.5489515087893914, + "eval_loss": 1.5430914163589478, + "eval_runtime": 5.0283, + "eval_samples_per_second": 596.224, + "eval_steps_per_second": 1.193, + "step": 7200 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2372, + "step": 7201 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.1235, + "step": 7202 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.802, + "step": 7203 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1814, + "step": 7204 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7041, + "step": 7205 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.93, + "step": 7206 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1207, + "step": 7207 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.5014, + "step": 7208 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2049, + "step": 7209 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1662, + "step": 7210 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9381, + "step": 7211 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6201, + "step": 7212 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.0669, + "step": 7213 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0539, + "step": 7214 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8114, + "step": 7215 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.0716, + "step": 7216 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9778, + "step": 7217 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8026, + "step": 7218 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1096, + "step": 7219 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.3245, + "step": 7220 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6902, + "step": 7221 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.0375, + "step": 7222 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0063, + "step": 7223 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7451, + "step": 7224 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.2035, + "step": 7225 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0537, + "step": 7226 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.677, + "step": 7227 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 0.9776, + "step": 7228 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.775, + "step": 7229 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6305, + "step": 7230 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1868, + "step": 7231 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.1415, + "step": 7232 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7722, + "step": 7233 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1691, + "step": 7234 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.2262, + "step": 7235 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.7526, + "step": 7236 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1017, + "step": 7237 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.2463, + "step": 7238 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.6354, + "step": 7239 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.12, + "step": 7240 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0216, + "step": 7241 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.3811, + "step": 7242 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.1357, + "step": 7243 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.0902, + "step": 7244 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.8258, + "step": 7245 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.113, + "step": 7246 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 2.2696, + "step": 7247 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9689, + "step": 7248 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.034, + "step": 7249 + }, + { + "epoch": 0.72, + "learning_rate": 0.0001, + "loss": 1.9277, + "step": 7250 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4936, + "step": 7251 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0105, + "step": 7252 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0078, + "step": 7253 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8969, + "step": 7254 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0354, + "step": 7255 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1335, + "step": 7256 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6495, + "step": 7257 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6153, + "step": 7258 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2556, + "step": 7259 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0662, + "step": 7260 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2755, + "step": 7261 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8004, + "step": 7262 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0197, + "step": 7263 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1315, + "step": 7264 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2727, + "step": 7265 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.658, + "step": 7266 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2062, + "step": 7267 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.5499, + "step": 7268 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8649, + "step": 7269 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1345, + "step": 7270 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.4135, + "step": 7271 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8276, + "step": 7272 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.039, + "step": 7273 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0376, + "step": 7274 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.5154, + "step": 7275 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0464, + "step": 7276 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.927, + "step": 7277 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6049, + "step": 7278 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1816, + "step": 7279 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2629, + "step": 7280 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.5695, + "step": 7281 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0873, + "step": 7282 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.976, + "step": 7283 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.7917, + "step": 7284 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2562, + "step": 7285 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.4804, + "step": 7286 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.7592, + "step": 7287 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2147, + "step": 7288 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1551, + "step": 7289 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.675, + "step": 7290 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0306, + "step": 7291 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2331, + "step": 7292 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8124, + "step": 7293 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0208, + "step": 7294 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8831, + "step": 7295 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6216, + "step": 7296 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1219, + "step": 7297 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1606, + "step": 7298 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2842, + "step": 7299 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1231, + "step": 7300 + }, + { + "epoch": 0.73, + "eval_bleu": 15.877438147567805, + "eval_brevity_penalty": 0.544407714488849, + "eval_loss": 1.5397740602493286, + "eval_runtime": 5.1714, + "eval_samples_per_second": 579.725, + "eval_steps_per_second": 1.16, + "step": 7300 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2499, + "step": 7301 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6823, + "step": 7302 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1427, + "step": 7303 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0593, + "step": 7304 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4681, + "step": 7305 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1326, + "step": 7306 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2512, + "step": 7307 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4857, + "step": 7308 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1838, + "step": 7309 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1129, + "step": 7310 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6235, + "step": 7311 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0241, + "step": 7312 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9087, + "step": 7313 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.7569, + "step": 7314 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2065, + "step": 7315 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9128, + "step": 7316 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2345, + "step": 7317 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4616, + "step": 7318 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8399, + "step": 7319 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9767, + "step": 7320 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.231, + "step": 7321 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.3692, + "step": 7322 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.8206, + "step": 7323 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.3889, + "step": 7324 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1438, + "step": 7325 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9001, + "step": 7326 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2515, + "step": 7327 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.2429, + "step": 7328 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.881, + "step": 7329 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4319, + "step": 7330 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0848, + "step": 7331 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6886, + "step": 7332 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.4155, + "step": 7333 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9133, + "step": 7334 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.5709, + "step": 7335 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1847, + "step": 7336 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9639, + "step": 7337 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6489, + "step": 7338 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.2519, + "step": 7339 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.948, + "step": 7340 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.638, + "step": 7341 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.224, + "step": 7342 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.1844, + "step": 7343 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.6919, + "step": 7344 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.0531, + "step": 7345 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 2.0532, + "step": 7346 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.7942, + "step": 7347 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.1737, + "step": 7348 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.9913, + "step": 7349 + }, + { + "epoch": 0.73, + "learning_rate": 0.0001, + "loss": 1.3636, + "step": 7350 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2622, + "step": 7351 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1174, + "step": 7352 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5746, + "step": 7353 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1928, + "step": 7354 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1364, + "step": 7355 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.8481, + "step": 7356 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2335, + "step": 7357 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.975, + "step": 7358 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.927, + "step": 7359 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2235, + "step": 7360 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.3365, + "step": 7361 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1636, + "step": 7362 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2611, + "step": 7363 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1146, + "step": 7364 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.846, + "step": 7365 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2182, + "step": 7366 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.2709, + "step": 7367 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.6093, + "step": 7368 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1658, + "step": 7369 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1453, + "step": 7370 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5475, + "step": 7371 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0846, + "step": 7372 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0019, + "step": 7373 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9389, + "step": 7374 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0233, + "step": 7375 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.2683, + "step": 7376 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5598, + "step": 7377 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.061, + "step": 7378 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0151, + "step": 7379 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5671, + "step": 7380 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 0.991, + "step": 7381 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0735, + "step": 7382 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.4139, + "step": 7383 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.045, + "step": 7384 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0998, + "step": 7385 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7594, + "step": 7386 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0505, + "step": 7387 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.2442, + "step": 7388 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9029, + "step": 7389 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0027, + "step": 7390 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9436, + "step": 7391 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.6622, + "step": 7392 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0537, + "step": 7393 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.4829, + "step": 7394 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.6709, + "step": 7395 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.114, + "step": 7396 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.3267, + "step": 7397 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5765, + "step": 7398 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1624, + "step": 7399 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7741, + "step": 7400 + }, + { + "epoch": 0.74, + "eval_bleu": 15.972449075779918, + "eval_brevity_penalty": 0.5451057390400432, + "eval_loss": 1.5398656129837036, + "eval_runtime": 5.5017, + "eval_samples_per_second": 544.921, + "eval_steps_per_second": 1.091, + "step": 7400 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.3217, + "step": 7401 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0876, + "step": 7402 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7788, + "step": 7403 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1199, + "step": 7404 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1021, + "step": 7405 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9358, + "step": 7406 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1296, + "step": 7407 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2254, + "step": 7408 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1892, + "step": 7409 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.4755, + "step": 7410 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.169, + "step": 7411 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.8524, + "step": 7412 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7111, + "step": 7413 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0586, + "step": 7414 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9506, + "step": 7415 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7769, + "step": 7416 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2741, + "step": 7417 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1931, + "step": 7418 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5032, + "step": 7419 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1901, + "step": 7420 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1235, + "step": 7421 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.7918, + "step": 7422 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0965, + "step": 7423 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0989, + "step": 7424 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9157, + "step": 7425 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0544, + "step": 7426 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1079, + "step": 7427 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.8803, + "step": 7428 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1608, + "step": 7429 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1029, + "step": 7430 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.6503, + "step": 7431 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1777, + "step": 7432 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9928, + "step": 7433 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9986, + "step": 7434 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2099, + "step": 7435 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9462, + "step": 7436 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.4721, + "step": 7437 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.1434, + "step": 7438 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1173, + "step": 7439 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.9135, + "step": 7440 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0701, + "step": 7441 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.3103, + "step": 7442 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.5887, + "step": 7443 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0491, + "step": 7444 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.0589, + "step": 7445 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.4631, + "step": 7446 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.0256, + "step": 7447 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 2.1847, + "step": 7448 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.8428, + "step": 7449 + }, + { + "epoch": 0.74, + "learning_rate": 0.0001, + "loss": 1.2995, + "step": 7450 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0667, + "step": 7451 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.8648, + "step": 7452 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.2922, + "step": 7453 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0168, + "step": 7454 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.8985, + "step": 7455 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1994, + "step": 7456 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.8724, + "step": 7457 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.552, + "step": 7458 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1617, + "step": 7459 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1257, + "step": 7460 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5629, + "step": 7461 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1082, + "step": 7462 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.126, + "step": 7463 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6783, + "step": 7464 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1762, + "step": 7465 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0601, + "step": 7466 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5436, + "step": 7467 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1026, + "step": 7468 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0361, + "step": 7469 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6769, + "step": 7470 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1981, + "step": 7471 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.923, + "step": 7472 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.708, + "step": 7473 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.9817, + "step": 7474 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1378, + "step": 7475 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5992, + "step": 7476 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.2369, + "step": 7477 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.9462, + "step": 7478 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.8089, + "step": 7479 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1965, + "step": 7480 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1275, + "step": 7481 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.4328, + "step": 7482 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.2596, + "step": 7483 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.2065, + "step": 7484 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5953, + "step": 7485 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.2081, + "step": 7486 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.2078, + "step": 7487 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.614, + "step": 7488 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.126, + "step": 7489 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0856, + "step": 7490 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1554, + "step": 7491 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0808, + "step": 7492 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1708, + "step": 7493 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.4659, + "step": 7494 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.21, + "step": 7495 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1664, + "step": 7496 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6931, + "step": 7497 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1627, + "step": 7498 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.076, + "step": 7499 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.7098, + "step": 7500 + }, + { + "epoch": 0.75, + "eval_bleu": 15.909765355386538, + "eval_brevity_penalty": 0.5447461779152011, + "eval_loss": 1.5361204147338867, + "eval_runtime": 5.414, + "eval_samples_per_second": 553.75, + "eval_steps_per_second": 1.108, + "step": 7500 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 0.9309, + "step": 7501 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0905, + "step": 7502 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6929, + "step": 7503 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1751, + "step": 7504 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1093, + "step": 7505 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6918, + "step": 7506 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.2085, + "step": 7507 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0201, + "step": 7508 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.7394, + "step": 7509 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1255, + "step": 7510 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.7249, + "step": 7511 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6968, + "step": 7512 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0368, + "step": 7513 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.2562, + "step": 7514 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.7484, + "step": 7515 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0944, + "step": 7516 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1687, + "step": 7517 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5197, + "step": 7518 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1053, + "step": 7519 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1644, + "step": 7520 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5735, + "step": 7521 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0908, + "step": 7522 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.2171, + "step": 7523 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.6992, + "step": 7524 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0422, + "step": 7525 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1488, + "step": 7526 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.4803, + "step": 7527 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0462, + "step": 7528 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0334, + "step": 7529 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0996, + "step": 7530 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.3684, + "step": 7531 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.9943, + "step": 7532 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5828, + "step": 7533 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.007, + "step": 7534 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.0713, + "step": 7535 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.4095, + "step": 7536 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.0456, + "step": 7537 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.3364, + "step": 7538 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.5769, + "step": 7539 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1181, + "step": 7540 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.9587, + "step": 7541 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.9791, + "step": 7542 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.142, + "step": 7543 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1455, + "step": 7544 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.9775, + "step": 7545 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.1875, + "step": 7546 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 2.1985, + "step": 7547 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.7437, + "step": 7548 + }, + { + "epoch": 0.75, + "learning_rate": 0.0001, + "loss": 1.135, + "step": 7549 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1418, + "step": 7550 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6639, + "step": 7551 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0685, + "step": 7552 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0016, + "step": 7553 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6683, + "step": 7554 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1523, + "step": 7555 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9631, + "step": 7556 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6837, + "step": 7557 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1353, + "step": 7558 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1297, + "step": 7559 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.574, + "step": 7560 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.163, + "step": 7561 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.2072, + "step": 7562 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.734, + "step": 7563 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.056, + "step": 7564 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9791, + "step": 7565 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7309, + "step": 7566 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.2437, + "step": 7567 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.2334, + "step": 7568 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6686, + "step": 7569 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.2831, + "step": 7570 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1733, + "step": 7571 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.5808, + "step": 7572 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.183, + "step": 7573 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1964, + "step": 7574 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6989, + "step": 7575 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.316, + "step": 7576 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0949, + "step": 7577 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9254, + "step": 7578 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1866, + "step": 7579 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0457, + "step": 7580 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7292, + "step": 7581 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1548, + "step": 7582 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0398, + "step": 7583 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.811, + "step": 7584 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0719, + "step": 7585 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.093, + "step": 7586 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7775, + "step": 7587 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0582, + "step": 7588 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.8923, + "step": 7589 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.5233, + "step": 7590 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1271, + "step": 7591 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.2164, + "step": 7592 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.8162, + "step": 7593 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.3286, + "step": 7594 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9669, + "step": 7595 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7146, + "step": 7596 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0767, + "step": 7597 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.2345, + "step": 7598 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6657, + "step": 7599 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0787, + "step": 7600 + }, + { + "epoch": 0.76, + "eval_bleu": 15.978149623194945, + "eval_brevity_penalty": 0.5456555437806875, + "eval_loss": 1.539277195930481, + "eval_runtime": 5.4362, + "eval_samples_per_second": 551.485, + "eval_steps_per_second": 1.104, + "step": 7600 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1774, + "step": 7601 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.5188, + "step": 7602 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1461, + "step": 7603 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.4314, + "step": 7604 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.4075, + "step": 7605 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.3213, + "step": 7606 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0226, + "step": 7607 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.4229, + "step": 7608 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 0.9195, + "step": 7609 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0336, + "step": 7610 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.3679, + "step": 7611 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1323, + "step": 7612 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.8299, + "step": 7613 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.8831, + "step": 7614 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.091, + "step": 7615 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.2038, + "step": 7616 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9071, + "step": 7617 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1831, + "step": 7618 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1502, + "step": 7619 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.5269, + "step": 7620 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0871, + "step": 7621 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1819, + "step": 7622 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6263, + "step": 7623 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1155, + "step": 7624 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7986, + "step": 7625 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.4808, + "step": 7626 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.149, + "step": 7627 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1878, + "step": 7628 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.7077, + "step": 7629 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0276, + "step": 7630 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1116, + "step": 7631 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.51, + "step": 7632 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1208, + "step": 7633 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1405, + "step": 7634 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.5968, + "step": 7635 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0773, + "step": 7636 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.9001, + "step": 7637 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.3564, + "step": 7638 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1058, + "step": 7639 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1167, + "step": 7640 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.6243, + "step": 7641 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.1401, + "step": 7642 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0644, + "step": 7643 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.838, + "step": 7644 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.0898, + "step": 7645 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.992, + "step": 7646 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.0756, + "step": 7647 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 1.163, + "step": 7648 + }, + { + "epoch": 0.76, + "learning_rate": 0.0001, + "loss": 2.1533, + "step": 7649 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8166, + "step": 7650 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0608, + "step": 7651 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7974, + "step": 7652 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8022, + "step": 7653 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0531, + "step": 7654 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8626, + "step": 7655 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6172, + "step": 7656 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1194, + "step": 7657 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.928, + "step": 7658 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4871, + "step": 7659 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0637, + "step": 7660 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9499, + "step": 7661 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4398, + "step": 7662 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1848, + "step": 7663 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.9202, + "step": 7664 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.715, + "step": 7665 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2364, + "step": 7666 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4985, + "step": 7667 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7114, + "step": 7668 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0608, + "step": 7669 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.1413, + "step": 7670 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7163, + "step": 7671 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1866, + "step": 7672 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9772, + "step": 7673 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7986, + "step": 7674 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0746, + "step": 7675 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.89, + "step": 7676 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4965, + "step": 7677 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.3111, + "step": 7678 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.2053, + "step": 7679 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8792, + "step": 7680 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2407, + "step": 7681 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.231, + "step": 7682 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.19, + "step": 7683 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.3022, + "step": 7684 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.2548, + "step": 7685 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6531, + "step": 7686 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2557, + "step": 7687 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.0272, + "step": 7688 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7666, + "step": 7689 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1717, + "step": 7690 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.14, + "step": 7691 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.0458, + "step": 7692 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.3833, + "step": 7693 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8895, + "step": 7694 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9918, + "step": 7695 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.256, + "step": 7696 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.2061, + "step": 7697 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7476, + "step": 7698 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.3233, + "step": 7699 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9856, + "step": 7700 + }, + { + "epoch": 0.77, + "eval_bleu": 15.952095532008354, + "eval_brevity_penalty": 0.5462263509335825, + "eval_loss": 1.5347563028335571, + "eval_runtime": 6.0271, + "eval_samples_per_second": 497.417, + "eval_steps_per_second": 0.995, + "step": 7700 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8306, + "step": 7701 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2223, + "step": 7702 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9817, + "step": 7703 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.5789, + "step": 7704 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2785, + "step": 7705 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.0846, + "step": 7706 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.2398, + "step": 7707 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1414, + "step": 7708 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8011, + "step": 7709 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.5764, + "step": 7710 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1842, + "step": 7711 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.3662, + "step": 7712 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8643, + "step": 7713 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2183, + "step": 7714 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9166, + "step": 7715 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6421, + "step": 7716 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 0.991, + "step": 7717 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9682, + "step": 7718 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7178, + "step": 7719 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1422, + "step": 7720 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.2398, + "step": 7721 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6179, + "step": 7722 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0815, + "step": 7723 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9395, + "step": 7724 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6705, + "step": 7725 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0969, + "step": 7726 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9932, + "step": 7727 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.7292, + "step": 7728 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4116, + "step": 7729 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.163, + "step": 7730 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8848, + "step": 7731 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2381, + "step": 7732 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.1877, + "step": 7733 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6373, + "step": 7734 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.3123, + "step": 7735 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6755, + "step": 7736 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8892, + "step": 7737 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1395, + "step": 7738 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9724, + "step": 7739 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.4324, + "step": 7740 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.1749, + "step": 7741 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.0821, + "step": 7742 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.516, + "step": 7743 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.0437, + "step": 7744 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.9023, + "step": 7745 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.8268, + "step": 7746 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.2546, + "step": 7747 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 2.343, + "step": 7748 + }, + { + "epoch": 0.77, + "learning_rate": 0.0001, + "loss": 1.6341, + "step": 7749 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1656, + "step": 7750 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2009, + "step": 7751 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8688, + "step": 7752 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1054, + "step": 7753 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2114, + "step": 7754 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6402, + "step": 7755 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1895, + "step": 7756 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0334, + "step": 7757 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6717, + "step": 7758 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1863, + "step": 7759 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8513, + "step": 7760 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0842, + "step": 7761 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.2375, + "step": 7762 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.3123, + "step": 7763 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8352, + "step": 7764 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0376, + "step": 7765 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0907, + "step": 7766 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.013, + "step": 7767 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0677, + "step": 7768 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.4993, + "step": 7769 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7963, + "step": 7770 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.243, + "step": 7771 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.9306, + "step": 7772 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.5667, + "step": 7773 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1673, + "step": 7774 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2177, + "step": 7775 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6255, + "step": 7776 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.2126, + "step": 7777 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0275, + "step": 7778 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.5901, + "step": 7779 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.11, + "step": 7780 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.1696, + "step": 7781 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7434, + "step": 7782 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0431, + "step": 7783 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0541, + "step": 7784 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6304, + "step": 7785 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1144, + "step": 7786 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0408, + "step": 7787 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.5602, + "step": 7788 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0378, + "step": 7789 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7917, + "step": 7790 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7549, + "step": 7791 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.12, + "step": 7792 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.3392, + "step": 7793 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0202, + "step": 7794 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.2053, + "step": 7795 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0496, + "step": 7796 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.3255, + "step": 7797 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1013, + "step": 7798 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0833, + "step": 7799 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.1294, + "step": 7800 + }, + { + "epoch": 0.78, + "eval_bleu": 16.004218883438558, + "eval_brevity_penalty": 0.546268627104608, + "eval_loss": 1.5345324277877808, + "eval_runtime": 5.5268, + "eval_samples_per_second": 542.449, + "eval_steps_per_second": 1.086, + "step": 7800 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1547, + "step": 7801 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.5551, + "step": 7802 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8729, + "step": 7803 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.012, + "step": 7804 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.5857, + "step": 7805 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.024, + "step": 7806 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1011, + "step": 7807 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2619, + "step": 7808 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.5345, + "step": 7809 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.2065, + "step": 7810 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.054, + "step": 7811 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6897, + "step": 7812 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1276, + "step": 7813 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2361, + "step": 7814 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8521, + "step": 7815 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0633, + "step": 7816 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.9451, + "step": 7817 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7834, + "step": 7818 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0168, + "step": 7819 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.929, + "step": 7820 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.7118, + "step": 7821 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0686, + "step": 7822 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.9955, + "step": 7823 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6581, + "step": 7824 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1493, + "step": 7825 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.1244, + "step": 7826 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.4771, + "step": 7827 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.2368, + "step": 7828 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.9466, + "step": 7829 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.1301, + "step": 7830 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1457, + "step": 7831 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.224, + "step": 7832 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6777, + "step": 7833 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1528, + "step": 7834 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0797, + "step": 7835 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6872, + "step": 7836 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1112, + "step": 7837 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.2027, + "step": 7838 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.4172, + "step": 7839 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1003, + "step": 7840 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.6168, + "step": 7841 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.4133, + "step": 7842 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.1424, + "step": 7843 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.8292, + "step": 7844 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.5241, + "step": 7845 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0524, + "step": 7846 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 2.0838, + "step": 7847 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.9795, + "step": 7848 + }, + { + "epoch": 0.78, + "learning_rate": 0.0001, + "loss": 1.0734, + "step": 7849 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1225, + "step": 7850 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.5447, + "step": 7851 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1038, + "step": 7852 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0462, + "step": 7853 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.871, + "step": 7854 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1659, + "step": 7855 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.947, + "step": 7856 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9652, + "step": 7857 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1246, + "step": 7858 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0636, + "step": 7859 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0387, + "step": 7860 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.108, + "step": 7861 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0056, + "step": 7862 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7879, + "step": 7863 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.0678, + "step": 7864 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8498, + "step": 7865 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7251, + "step": 7866 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1766, + "step": 7867 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8028, + "step": 7868 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.4541, + "step": 7869 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1359, + "step": 7870 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2264, + "step": 7871 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0916, + "step": 7872 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.0944, + "step": 7873 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.3212, + "step": 7874 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7008, + "step": 7875 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.0675, + "step": 7876 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2123, + "step": 7877 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.826, + "step": 7878 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.101, + "step": 7879 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.3669, + "step": 7880 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8144, + "step": 7881 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.2492, + "step": 7882 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0276, + "step": 7883 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8623, + "step": 7884 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.184, + "step": 7885 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.3841, + "step": 7886 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.4618, + "step": 7887 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.2044, + "step": 7888 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2136, + "step": 7889 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.848, + "step": 7890 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1886, + "step": 7891 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1943, + "step": 7892 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.6822, + "step": 7893 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.14, + "step": 7894 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1341, + "step": 7895 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.6435, + "step": 7896 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1572, + "step": 7897 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.051, + "step": 7898 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8741, + "step": 7899 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1938, + "step": 7900 + }, + { + "epoch": 0.79, + "eval_bleu": 16.05536335789969, + "eval_brevity_penalty": 0.5495003569450424, + "eval_loss": 1.5313706398010254, + "eval_runtime": 5.1531, + "eval_samples_per_second": 581.785, + "eval_steps_per_second": 1.164, + "step": 7900 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2068, + "step": 7901 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7022, + "step": 7902 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1582, + "step": 7903 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2016, + "step": 7904 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7122, + "step": 7905 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.187, + "step": 7906 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0921, + "step": 7907 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.4026, + "step": 7908 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1475, + "step": 7909 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1174, + "step": 7910 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8093, + "step": 7911 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1879, + "step": 7912 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1695, + "step": 7913 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8705, + "step": 7914 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1353, + "step": 7915 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0637, + "step": 7916 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7103, + "step": 7917 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1373, + "step": 7918 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.2845, + "step": 7919 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8903, + "step": 7920 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1838, + "step": 7921 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1658, + "step": 7922 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9545, + "step": 7923 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1374, + "step": 7924 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.1712, + "step": 7925 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.814, + "step": 7926 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1322, + "step": 7927 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9684, + "step": 7928 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.5154, + "step": 7929 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1094, + "step": 7930 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.7876, + "step": 7931 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.4877, + "step": 7932 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1678, + "step": 7933 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9719, + "step": 7934 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0413, + "step": 7935 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.2287, + "step": 7936 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.3469, + "step": 7937 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.4479, + "step": 7938 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.0826, + "step": 7939 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0426, + "step": 7940 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.933, + "step": 7941 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.123, + "step": 7942 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9857, + "step": 7943 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.5564, + "step": 7944 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.1775, + "step": 7945 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.9738, + "step": 7946 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.8228, + "step": 7947 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 1.0271, + "step": 7948 + }, + { + "epoch": 0.79, + "learning_rate": 0.0001, + "loss": 2.0791, + "step": 7949 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.5921, + "step": 7950 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0915, + "step": 7951 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.2198, + "step": 7952 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8655, + "step": 7953 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.2363, + "step": 7954 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9859, + "step": 7955 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6727, + "step": 7956 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0899, + "step": 7957 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9489, + "step": 7958 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7847, + "step": 7959 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.2378, + "step": 7960 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.1677, + "step": 7961 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8718, + "step": 7962 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.2217, + "step": 7963 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9177, + "step": 7964 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6087, + "step": 7965 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1377, + "step": 7966 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.1783, + "step": 7967 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7334, + "step": 7968 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1021, + "step": 7969 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8588, + "step": 7970 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6123, + "step": 7971 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1479, + "step": 7972 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0671, + "step": 7973 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6534, + "step": 7974 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1444, + "step": 7975 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9908, + "step": 7976 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7657, + "step": 7977 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0867, + "step": 7978 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9652, + "step": 7979 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.3945, + "step": 7980 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0896, + "step": 7981 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8629, + "step": 7982 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.5464, + "step": 7983 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1715, + "step": 7984 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7414, + "step": 7985 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6019, + "step": 7986 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.199, + "step": 7987 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0903, + "step": 7988 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8013, + "step": 7989 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0997, + "step": 7990 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.1124, + "step": 7991 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0604, + "step": 7992 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.2079, + "step": 7993 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.1734, + "step": 7994 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8398, + "step": 7995 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1564, + "step": 7996 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.1989, + "step": 7997 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6239, + "step": 7998 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0904, + "step": 7999 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9579, + "step": 8000 + }, + { + "epoch": 0.8, + "eval_bleu": 15.934878579786263, + "eval_brevity_penalty": 0.5482124579111715, + "eval_loss": 1.5307011604309082, + "eval_runtime": 5.5907, + "eval_samples_per_second": 536.252, + "eval_steps_per_second": 1.073, + "step": 8000 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.564, + "step": 8001 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.154, + "step": 8002 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0015, + "step": 8003 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6069, + "step": 8004 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0932, + "step": 8005 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8771, + "step": 8006 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6985, + "step": 8007 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1135, + "step": 8008 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.2274, + "step": 8009 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.5755, + "step": 8010 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1274, + "step": 8011 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9239, + "step": 8012 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8096, + "step": 8013 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0281, + "step": 8014 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8287, + "step": 8015 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7353, + "step": 8016 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1258, + "step": 8017 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0815, + "step": 8018 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.829, + "step": 8019 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.204, + "step": 8020 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.2984, + "step": 8021 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8609, + "step": 8022 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.2129, + "step": 8023 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9234, + "step": 8024 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.7557, + "step": 8025 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.194, + "step": 8026 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0412, + "step": 8027 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8411, + "step": 8028 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0249, + "step": 8029 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9746, + "step": 8030 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.4857, + "step": 8031 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0456, + "step": 8032 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8808, + "step": 8033 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6469, + "step": 8034 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1444, + "step": 8035 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 2.0356, + "step": 8036 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8512, + "step": 8037 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.0093, + "step": 8038 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9313, + "step": 8039 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.6561, + "step": 8040 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1977, + "step": 8041 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8816, + "step": 8042 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.399, + "step": 8043 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1031, + "step": 8044 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8398, + "step": 8045 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.696, + "step": 8046 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.1387, + "step": 8047 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.9464, + "step": 8048 + }, + { + "epoch": 0.8, + "learning_rate": 0.0001, + "loss": 1.8578, + "step": 8049 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.1038, + "step": 8050 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.3431, + "step": 8051 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5008, + "step": 8052 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.021, + "step": 8053 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8587, + "step": 8054 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8598, + "step": 8055 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.1546, + "step": 8056 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1666, + "step": 8057 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.6461, + "step": 8058 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.6324, + "step": 8059 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9996, + "step": 8060 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.4078, + "step": 8061 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.4733, + "step": 8062 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9609, + "step": 8063 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.4945, + "step": 8064 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.3598, + "step": 8065 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9103, + "step": 8066 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0669, + "step": 8067 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.4473, + "step": 8068 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9238, + "step": 8069 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1908, + "step": 8070 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.1453, + "step": 8071 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1482, + "step": 8072 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.021, + "step": 8073 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.4234, + "step": 8074 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0941, + "step": 8075 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9874, + "step": 8076 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5413, + "step": 8077 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0166, + "step": 8078 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1968, + "step": 8079 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.3167, + "step": 8080 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9457, + "step": 8081 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1899, + "step": 8082 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5719, + "step": 8083 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.3246, + "step": 8084 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.7994, + "step": 8085 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5718, + "step": 8086 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1061, + "step": 8087 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.13, + "step": 8088 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.341, + "step": 8089 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1542, + "step": 8090 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.246, + "step": 8091 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.4382, + "step": 8092 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.088, + "step": 8093 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0893, + "step": 8094 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.7068, + "step": 8095 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0882, + "step": 8096 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.4403, + "step": 8097 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5291, + "step": 8098 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.2015, + "step": 8099 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.844, + "step": 8100 + }, + { + "epoch": 0.81, + "eval_bleu": 15.858918806983555, + "eval_brevity_penalty": 0.5448096340925517, + "eval_loss": 1.5284861326217651, + "eval_runtime": 5.8229, + "eval_samples_per_second": 514.86, + "eval_steps_per_second": 1.03, + "step": 8100 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.2769, + "step": 8101 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.99, + "step": 8102 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0639, + "step": 8103 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.6307, + "step": 8104 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.2034, + "step": 8105 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8544, + "step": 8106 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5571, + "step": 8107 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9069, + "step": 8108 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.6371, + "step": 8109 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.116, + "step": 8110 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9711, + "step": 8111 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.5076, + "step": 8112 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9769, + "step": 8113 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.3526, + "step": 8114 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.7255, + "step": 8115 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.7446, + "step": 8116 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9329, + "step": 8117 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1021, + "step": 8118 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0492, + "step": 8119 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8648, + "step": 8120 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.2164, + "step": 8121 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8196, + "step": 8122 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9913, + "step": 8123 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.2077, + "step": 8124 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.2244, + "step": 8125 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0868, + "step": 8126 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0714, + "step": 8127 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.7872, + "step": 8128 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8959, + "step": 8129 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.5506, + "step": 8130 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9898, + "step": 8131 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.4387, + "step": 8132 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1584, + "step": 8133 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.318, + "step": 8134 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.0364, + "step": 8135 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1535, + "step": 8136 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.1841, + "step": 8137 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1673, + "step": 8138 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9083, + "step": 8139 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.3126, + "step": 8140 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.9964, + "step": 8141 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.5136, + "step": 8142 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.3052, + "step": 8143 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.3614, + "step": 8144 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.8899, + "step": 8145 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 1.4075, + "step": 8146 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1477, + "step": 8147 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.103, + "step": 8148 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 0.8412, + "step": 8149 + }, + { + "epoch": 0.81, + "learning_rate": 0.0001, + "loss": 2.1862, + "step": 8150 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6368, + "step": 8151 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.1005, + "step": 8152 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9327, + "step": 8153 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0259, + "step": 8154 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.5964, + "step": 8155 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3454, + "step": 8156 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1522, + "step": 8157 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.2916, + "step": 8158 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3652, + "step": 8159 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8541, + "step": 8160 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6607, + "step": 8161 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2815, + "step": 8162 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3333, + "step": 8163 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.4703, + "step": 8164 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0541, + "step": 8165 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1745, + "step": 8166 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6305, + "step": 8167 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7777, + "step": 8168 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1517, + "step": 8169 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7203, + "step": 8170 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0441, + "step": 8171 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3449, + "step": 8172 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7732, + "step": 8173 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9527, + "step": 8174 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.4914, + "step": 8175 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7959, + "step": 8176 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7572, + "step": 8177 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.6108, + "step": 8178 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8343, + "step": 8179 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2325, + "step": 8180 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.7297, + "step": 8181 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7826, + "step": 8182 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9835, + "step": 8183 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3179, + "step": 8184 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7596, + "step": 8185 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3849, + "step": 8186 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.11, + "step": 8187 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.3721, + "step": 8188 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9695, + "step": 8189 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.4675, + "step": 8190 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.653, + "step": 8191 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2049, + "step": 8192 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.5333, + "step": 8193 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.276, + "step": 8194 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7482, + "step": 8195 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9205, + "step": 8196 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.9653, + "step": 8197 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0237, + "step": 8198 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0739, + "step": 8199 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.1464, + "step": 8200 + }, + { + "epoch": 0.82, + "eval_bleu": 15.921026905479918, + "eval_brevity_penalty": 0.5434766757314575, + "eval_loss": 1.5412899255752563, + "eval_runtime": 5.3722, + "eval_samples_per_second": 558.055, + "eval_steps_per_second": 1.117, + "step": 8200 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0116, + "step": 8201 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6482, + "step": 8202 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.8864, + "step": 8203 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8758, + "step": 8204 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.3673, + "step": 8205 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 0.9382, + "step": 8206 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1909, + "step": 8207 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.3653, + "step": 8208 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6648, + "step": 8209 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0536, + "step": 8210 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9255, + "step": 8211 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.5218, + "step": 8212 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0381, + "step": 8213 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2133, + "step": 8214 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0926, + "step": 8215 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1007, + "step": 8216 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2244, + "step": 8217 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8738, + "step": 8218 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.1317, + "step": 8219 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.514, + "step": 8220 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9483, + "step": 8221 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9889, + "step": 8222 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.4893, + "step": 8223 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.6793, + "step": 8224 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8994, + "step": 8225 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.5247, + "step": 8226 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8271, + "step": 8227 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8726, + "step": 8228 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.6835, + "step": 8229 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8484, + "step": 8230 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0415, + "step": 8231 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.6193, + "step": 8232 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7782, + "step": 8233 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0288, + "step": 8234 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.4113, + "step": 8235 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7874, + "step": 8236 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.754, + "step": 8237 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.6547, + "step": 8238 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.8571, + "step": 8239 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.9043, + "step": 8240 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.4215, + "step": 8241 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0124, + "step": 8242 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0489, + "step": 8243 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.357, + "step": 8244 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.7391, + "step": 8245 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0584, + "step": 8246 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.2945, + "step": 8247 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.2211, + "step": 8248 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 2.0039, + "step": 8249 + }, + { + "epoch": 0.82, + "learning_rate": 0.0001, + "loss": 1.5069, + "step": 8250 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.1352, + "step": 8251 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9534, + "step": 8252 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8783, + "step": 8253 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.0267, + "step": 8254 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1448, + "step": 8255 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9729, + "step": 8256 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8053, + "step": 8257 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.241, + "step": 8258 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.405, + "step": 8259 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7514, + "step": 8260 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8277, + "step": 8261 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7713, + "step": 8262 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9369, + "step": 8263 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8495, + "step": 8264 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.57, + "step": 8265 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8323, + "step": 8266 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1135, + "step": 8267 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4449, + "step": 8268 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8445, + "step": 8269 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2975, + "step": 8270 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.47, + "step": 8271 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9936, + "step": 8272 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8834, + "step": 8273 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.6233, + "step": 8274 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9855, + "step": 8275 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8937, + "step": 8276 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0677, + "step": 8277 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7458, + "step": 8278 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.5101, + "step": 8279 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.5459, + "step": 8280 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.6845, + "step": 8281 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1014, + "step": 8282 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.6046, + "step": 8283 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7752, + "step": 8284 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3479, + "step": 8285 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2252, + "step": 8286 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.6504, + "step": 8287 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9524, + "step": 8288 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4779, + "step": 8289 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0604, + "step": 8290 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2672, + "step": 8291 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4116, + "step": 8292 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8755, + "step": 8293 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2246, + "step": 8294 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1138, + "step": 8295 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8642, + "step": 8296 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.051, + "step": 8297 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4121, + "step": 8298 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7955, + "step": 8299 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2903, + "step": 8300 + }, + { + "epoch": 0.83, + "eval_bleu": 16.016428608765683, + "eval_brevity_penalty": 0.5405117117842276, + "eval_loss": 1.5230488777160645, + "eval_runtime": 5.8035, + "eval_samples_per_second": 516.588, + "eval_steps_per_second": 1.034, + "step": 8300 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4433, + "step": 8301 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.769, + "step": 8302 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4018, + "step": 8303 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3317, + "step": 8304 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.6324, + "step": 8305 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.312, + "step": 8306 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1497, + "step": 8307 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9075, + "step": 8308 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9911, + "step": 8309 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4503, + "step": 8310 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7366, + "step": 8311 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.048, + "step": 8312 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3945, + "step": 8313 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.558, + "step": 8314 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9598, + "step": 8315 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0672, + "step": 8316 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7934, + "step": 8317 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9055, + "step": 8318 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4583, + "step": 8319 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7559, + "step": 8320 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9346, + "step": 8321 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3042, + "step": 8322 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7982, + "step": 8323 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0846, + "step": 8324 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.8648, + "step": 8325 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7698, + "step": 8326 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.415, + "step": 8327 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.5527, + "step": 8328 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7639, + "step": 8329 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9947, + "step": 8330 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.4553, + "step": 8331 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8194, + "step": 8332 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0667, + "step": 8333 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0316, + "step": 8334 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8826, + "step": 8335 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.9165, + "step": 8336 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.9075, + "step": 8337 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7841, + "step": 8338 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.1083, + "step": 8339 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.437, + "step": 8340 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.8308, + "step": 8341 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.0628, + "step": 8342 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.296, + "step": 8343 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.6243, + "step": 8344 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3024, + "step": 8345 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.3841, + "step": 8346 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7645, + "step": 8347 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.2021, + "step": 8348 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 2.409, + "step": 8349 + }, + { + "epoch": 0.83, + "learning_rate": 0.0001, + "loss": 1.7143, + "step": 8350 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0174, + "step": 8351 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1797, + "step": 8352 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6713, + "step": 8353 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9445, + "step": 8354 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.152, + "step": 8355 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6569, + "step": 8356 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9172, + "step": 8357 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.4173, + "step": 8358 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.8347, + "step": 8359 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.5377, + "step": 8360 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9576, + "step": 8361 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7342, + "step": 8362 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.02, + "step": 8363 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7968, + "step": 8364 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.2884, + "step": 8365 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0753, + "step": 8366 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2321, + "step": 8367 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7281, + "step": 8368 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2868, + "step": 8369 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1902, + "step": 8370 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7331, + "step": 8371 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0489, + "step": 8372 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1949, + "step": 8373 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6554, + "step": 8374 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1444, + "step": 8375 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.69, + "step": 8376 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.5406, + "step": 8377 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1016, + "step": 8378 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0686, + "step": 8379 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6577, + "step": 8380 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9733, + "step": 8381 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.4085, + "step": 8382 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6415, + "step": 8383 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.058, + "step": 8384 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2991, + "step": 8385 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7554, + "step": 8386 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1537, + "step": 8387 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1299, + "step": 8388 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7503, + "step": 8389 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0245, + "step": 8390 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9812, + "step": 8391 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7406, + "step": 8392 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9113, + "step": 8393 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.9282, + "step": 8394 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6343, + "step": 8395 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1994, + "step": 8396 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1537, + "step": 8397 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.4901, + "step": 8398 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2737, + "step": 8399 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1489, + "step": 8400 + }, + { + "epoch": 0.84, + "eval_bleu": 15.942309208367908, + "eval_brevity_penalty": 0.5443442468840576, + "eval_loss": 1.5263079404830933, + "eval_runtime": 5.4227, + "eval_samples_per_second": 552.866, + "eval_steps_per_second": 1.106, + "step": 8400 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6933, + "step": 8401 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.8176, + "step": 8402 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2144, + "step": 8403 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6809, + "step": 8404 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0106, + "step": 8405 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.2535, + "step": 8406 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7017, + "step": 8407 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1994, + "step": 8408 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0927, + "step": 8409 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.025, + "step": 8410 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.014, + "step": 8411 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3951, + "step": 8412 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.9318, + "step": 8413 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1159, + "step": 8414 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6985, + "step": 8415 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.8632, + "step": 8416 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.8967, + "step": 8417 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7561, + "step": 8418 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.0816, + "step": 8419 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0992, + "step": 8420 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3941, + "step": 8421 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.8777, + "step": 8422 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0214, + "step": 8423 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.4907, + "step": 8424 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.8717, + "step": 8425 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.879, + "step": 8426 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3842, + "step": 8427 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.0888, + "step": 8428 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.15, + "step": 8429 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.0253, + "step": 8430 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 0.9171, + "step": 8431 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.8442, + "step": 8432 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.6645, + "step": 8433 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.4273, + "step": 8434 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.5238, + "step": 8435 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3512, + "step": 8436 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.4154, + "step": 8437 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.1564, + "step": 8438 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.5423, + "step": 8439 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.1663, + "step": 8440 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.488, + "step": 8441 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3854, + "step": 8442 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.1852, + "step": 8443 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.4409, + "step": 8444 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.561, + "step": 8445 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.2625, + "step": 8446 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.7421, + "step": 8447 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.7337, + "step": 8448 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 1.3749, + "step": 8449 + }, + { + "epoch": 0.84, + "learning_rate": 0.0001, + "loss": 2.9601, + "step": 8450 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.5521, + "step": 8451 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.3502, + "step": 8452 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.595, + "step": 8453 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.8454, + "step": 8454 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.1882, + "step": 8455 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.16, + "step": 8456 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.6388, + "step": 8457 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.4387, + "step": 8458 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1762, + "step": 8459 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.5918, + "step": 8460 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.3131, + "step": 8461 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4559, + "step": 8462 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.7255, + "step": 8463 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.257, + "step": 8464 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2173, + "step": 8465 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.403, + "step": 8466 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.3025, + "step": 8467 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3289, + "step": 8468 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.2966, + "step": 8469 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.3937, + "step": 8470 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1693, + "step": 8471 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.7204, + "step": 8472 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.7855, + "step": 8473 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3241, + "step": 8474 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.8499, + "step": 8475 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.2576, + "step": 8476 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4732, + "step": 8477 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.4361, + "step": 8478 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.7318, + "step": 8479 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1794, + "step": 8480 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.85, + "step": 8481 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.3499, + "step": 8482 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3094, + "step": 8483 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.9468, + "step": 8484 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1796, + "step": 8485 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.375, + "step": 8486 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.445, + "step": 8487 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1086, + "step": 8488 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.5245, + "step": 8489 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4873, + "step": 8490 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.2951, + "step": 8491 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3896, + "step": 8492 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.4409, + "step": 8493 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.8294, + "step": 8494 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2226, + "step": 8495 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.5091, + "step": 8496 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.0251, + "step": 8497 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4661, + "step": 8498 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1136, + "step": 8499 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.8138, + "step": 8500 + }, + { + "epoch": 0.85, + "eval_bleu": 15.82668498492032, + "eval_brevity_penalty": 0.5464377237267338, + "eval_loss": 1.5349748134613037, + "eval_runtime": 5.5422, + "eval_samples_per_second": 540.945, + "eval_steps_per_second": 1.083, + "step": 8500 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3871, + "step": 8501 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4009, + "step": 8502 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4062, + "step": 8503 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3455, + "step": 8504 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.74, + "step": 8505 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.9863, + "step": 8506 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1917, + "step": 8507 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.0898, + "step": 8508 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1747, + "step": 8509 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2541, + "step": 8510 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.8872, + "step": 8511 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3332, + "step": 8512 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.271, + "step": 8513 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.69, + "step": 8514 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4546, + "step": 8515 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1755, + "step": 8516 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.5144, + "step": 8517 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.347, + "step": 8518 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.0934, + "step": 8519 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2073, + "step": 8520 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4801, + "step": 8521 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.271, + "step": 8522 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.6774, + "step": 8523 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3275, + "step": 8524 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2865, + "step": 8525 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.6984, + "step": 8526 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2553, + "step": 8527 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4054, + "step": 8528 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4289, + "step": 8529 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3424, + "step": 8530 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2206, + "step": 8531 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2373, + "step": 8532 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1627, + "step": 8533 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4373, + "step": 8534 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2485, + "step": 8535 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1865, + "step": 8536 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3105, + "step": 8537 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.5453, + "step": 8538 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2115, + "step": 8539 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2532, + "step": 8540 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.5134, + "step": 8541 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.0838, + "step": 8542 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2381, + "step": 8543 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.3026, + "step": 8544 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 1.9871, + "step": 8545 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.0236, + "step": 8546 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.4936, + "step": 8547 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.1483, + "step": 8548 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.2104, + "step": 8549 + }, + { + "epoch": 0.85, + "learning_rate": 0.0001, + "loss": 2.6485, + "step": 8550 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2928, + "step": 8551 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.6325, + "step": 8552 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4897, + "step": 8553 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9732, + "step": 8554 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.759, + "step": 8555 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1446, + "step": 8556 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.323, + "step": 8557 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3807, + "step": 8558 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.328, + "step": 8559 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1602, + "step": 8560 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2694, + "step": 8561 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2491, + "step": 8562 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2108, + "step": 8563 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2068, + "step": 8564 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5984, + "step": 8565 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.0043, + "step": 8566 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4023, + "step": 8567 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9895, + "step": 8568 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1863, + "step": 8569 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1416, + "step": 8570 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5259, + "step": 8571 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2038, + "step": 8572 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4218, + "step": 8573 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.6004, + "step": 8574 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2095, + "step": 8575 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2406, + "step": 8576 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5766, + "step": 8577 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3408, + "step": 8578 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4135, + "step": 8579 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.7863, + "step": 8580 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3097, + "step": 8581 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5288, + "step": 8582 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.6984, + "step": 8583 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4226, + "step": 8584 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2524, + "step": 8585 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1854, + "step": 8586 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2081, + "step": 8587 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.209, + "step": 8588 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2195, + "step": 8589 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3759, + "step": 8590 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2575, + "step": 8591 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4896, + "step": 8592 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.142, + "step": 8593 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5059, + "step": 8594 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5178, + "step": 8595 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9976, + "step": 8596 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4054, + "step": 8597 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3281, + "step": 8598 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3376, + "step": 8599 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4025, + "step": 8600 + }, + { + "epoch": 0.86, + "eval_bleu": 15.849312984242916, + "eval_brevity_penalty": 0.5430321798941383, + "eval_loss": 1.5274958610534668, + "eval_runtime": 5.467, + "eval_samples_per_second": 548.384, + "eval_steps_per_second": 1.097, + "step": 8600 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.5653, + "step": 8601 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9807, + "step": 8602 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3268, + "step": 8603 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4586, + "step": 8604 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.5493, + "step": 8605 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5171, + "step": 8606 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.6062, + "step": 8607 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.6874, + "step": 8608 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3268, + "step": 8609 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3706, + "step": 8610 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.8085, + "step": 8611 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2453, + "step": 8612 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3079, + "step": 8613 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.4097, + "step": 8614 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1988, + "step": 8615 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3198, + "step": 8616 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2249, + "step": 8617 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1946, + "step": 8618 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.4528, + "step": 8619 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2853, + "step": 8620 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2345, + "step": 8621 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.6552, + "step": 8622 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3498, + "step": 8623 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.2449, + "step": 8624 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.8204, + "step": 8625 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3041, + "step": 8626 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.4747, + "step": 8627 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.5021, + "step": 8628 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3116, + "step": 8629 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3642, + "step": 8630 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.5281, + "step": 8631 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2948, + "step": 8632 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1187, + "step": 8633 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.5856, + "step": 8634 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3026, + "step": 8635 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.563, + "step": 8636 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9644, + "step": 8637 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2702, + "step": 8638 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.5707, + "step": 8639 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3509, + "step": 8640 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2774, + "step": 8641 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.6342, + "step": 8642 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.7008, + "step": 8643 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.2699, + "step": 8644 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.0765, + "step": 8645 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.9589, + "step": 8646 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.514, + "step": 8647 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.3245, + "step": 8648 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 2.1164, + "step": 8649 + }, + { + "epoch": 0.86, + "learning_rate": 0.0001, + "loss": 1.3709, + "step": 8650 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.4995, + "step": 8651 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6047, + "step": 8652 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.1871, + "step": 8653 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.3812, + "step": 8654 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.9354, + "step": 8655 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2458, + "step": 8656 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.342, + "step": 8657 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6453, + "step": 8658 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2376, + "step": 8659 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2024, + "step": 8660 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5625, + "step": 8661 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3435, + "step": 8662 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2627, + "step": 8663 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6851, + "step": 8664 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2551, + "step": 8665 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1197, + "step": 8666 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.847, + "step": 8667 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.1743, + "step": 8668 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.126, + "step": 8669 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7545, + "step": 8670 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3411, + "step": 8671 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.3819, + "step": 8672 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5743, + "step": 8673 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2599, + "step": 8674 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.5282, + "step": 8675 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7438, + "step": 8676 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3275, + "step": 8677 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2506, + "step": 8678 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6315, + "step": 8679 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3871, + "step": 8680 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1948, + "step": 8681 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5769, + "step": 8682 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3424, + "step": 8683 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.3003, + "step": 8684 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5092, + "step": 8685 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5052, + "step": 8686 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.0615, + "step": 8687 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6862, + "step": 8688 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2105, + "step": 8689 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.4245, + "step": 8690 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6313, + "step": 8691 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2981, + "step": 8692 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1479, + "step": 8693 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6652, + "step": 8694 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3221, + "step": 8695 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1424, + "step": 8696 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6417, + "step": 8697 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3489, + "step": 8698 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2798, + "step": 8699 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6758, + "step": 8700 + }, + { + "epoch": 0.87, + "eval_bleu": 15.924625301937459, + "eval_brevity_penalty": 0.5463531770282734, + "eval_loss": 1.5206462144851685, + "eval_runtime": 6.003, + "eval_samples_per_second": 499.418, + "eval_steps_per_second": 1.0, + "step": 8700 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5179, + "step": 8701 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.3969, + "step": 8702 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.9867, + "step": 8703 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.4147, + "step": 8704 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2514, + "step": 8705 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7561, + "step": 8706 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5499, + "step": 8707 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.5126, + "step": 8708 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.4618, + "step": 8709 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3309, + "step": 8710 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2693, + "step": 8711 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.8247, + "step": 8712 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.4291, + "step": 8713 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.8075, + "step": 8714 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7518, + "step": 8715 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2282, + "step": 8716 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.0244, + "step": 8717 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7092, + "step": 8718 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3173, + "step": 8719 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1761, + "step": 8720 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2688, + "step": 8721 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.3178, + "step": 8722 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.5488, + "step": 8723 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7766, + "step": 8724 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2843, + "step": 8725 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.9501, + "step": 8726 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7074, + "step": 8727 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2407, + "step": 8728 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2421, + "step": 8729 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6071, + "step": 8730 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2326, + "step": 8731 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.3166, + "step": 8732 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.5485, + "step": 8733 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2626, + "step": 8734 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.4603, + "step": 8735 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6639, + "step": 8736 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.4258, + "step": 8737 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.2138, + "step": 8738 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7372, + "step": 8739 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2393, + "step": 8740 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.5053, + "step": 8741 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.7975, + "step": 8742 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2465, + "step": 8743 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1925, + "step": 8744 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.523, + "step": 8745 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2935, + "step": 8746 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 2.1717, + "step": 8747 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.6546, + "step": 8748 + }, + { + "epoch": 0.87, + "learning_rate": 0.0001, + "loss": 1.2891, + "step": 8749 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.4499, + "step": 8750 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5967, + "step": 8751 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3024, + "step": 8752 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2208, + "step": 8753 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4526, + "step": 8754 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.2209, + "step": 8755 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2583, + "step": 8756 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7778, + "step": 8757 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.2906, + "step": 8758 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3343, + "step": 8759 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.566, + "step": 8760 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3005, + "step": 8761 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.6422, + "step": 8762 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7058, + "step": 8763 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4009, + "step": 8764 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2846, + "step": 8765 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.8276, + "step": 8766 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3913, + "step": 8767 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2903, + "step": 8768 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.2867, + "step": 8769 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3879, + "step": 8770 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.372, + "step": 8771 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5325, + "step": 8772 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3421, + "step": 8773 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.5285, + "step": 8774 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7327, + "step": 8775 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3623, + "step": 8776 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.18, + "step": 8777 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5379, + "step": 8778 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.367, + "step": 8779 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.372, + "step": 8780 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.877, + "step": 8781 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3159, + "step": 8782 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3602, + "step": 8783 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4424, + "step": 8784 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4567, + "step": 8785 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0984, + "step": 8786 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4941, + "step": 8787 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3503, + "step": 8788 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0524, + "step": 8789 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4772, + "step": 8790 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3109, + "step": 8791 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.4761, + "step": 8792 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.4513, + "step": 8793 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3261, + "step": 8794 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.5725, + "step": 8795 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5526, + "step": 8796 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3127, + "step": 8797 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.5953, + "step": 8798 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.867, + "step": 8799 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3671, + "step": 8800 + }, + { + "epoch": 0.88, + "eval_bleu": 15.96621136920216, + "eval_brevity_penalty": 0.5459515361805245, + "eval_loss": 1.5234949588775635, + "eval_runtime": 5.5104, + "eval_samples_per_second": 544.063, + "eval_steps_per_second": 1.089, + "step": 8800 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0128, + "step": 8801 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5832, + "step": 8802 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3233, + "step": 8803 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2254, + "step": 8804 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.6548, + "step": 8805 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.3238, + "step": 8806 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.6365, + "step": 8807 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7313, + "step": 8808 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.5632, + "step": 8809 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3993, + "step": 8810 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.9489, + "step": 8811 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.665, + "step": 8812 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.9853, + "step": 8813 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2469, + "step": 8814 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.633, + "step": 8815 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2725, + "step": 8816 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3336, + "step": 8817 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.6214, + "step": 8818 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0996, + "step": 8819 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.9959, + "step": 8820 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.6304, + "step": 8821 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3457, + "step": 8822 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.3303, + "step": 8823 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0169, + "step": 8824 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.4814, + "step": 8825 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2921, + "step": 8826 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0758, + "step": 8827 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.2619, + "step": 8828 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1169, + "step": 8829 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.813, + "step": 8830 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1989, + "step": 8831 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0118, + "step": 8832 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.6947, + "step": 8833 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.169, + "step": 8834 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.4072, + "step": 8835 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7998, + "step": 8836 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.0422, + "step": 8837 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1541, + "step": 8838 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.8024, + "step": 8839 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.8336, + "step": 8840 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1568, + "step": 8841 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.7663, + "step": 8842 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.4145, + "step": 8843 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1327, + "step": 8844 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.315, + "step": 8845 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.278, + "step": 8846 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.6982, + "step": 8847 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 1.8773, + "step": 8848 + }, + { + "epoch": 0.88, + "learning_rate": 0.0001, + "loss": 2.1818, + "step": 8849 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.5735, + "step": 8850 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.7173, + "step": 8851 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1739, + "step": 8852 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0605, + "step": 8853 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0737, + "step": 8854 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1642, + "step": 8855 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4419, + "step": 8856 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.063, + "step": 8857 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4538, + "step": 8858 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1545, + "step": 8859 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9371, + "step": 8860 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1022, + "step": 8861 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2795, + "step": 8862 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0059, + "step": 8863 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2589, + "step": 8864 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.5532, + "step": 8865 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.959, + "step": 8866 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.303, + "step": 8867 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2803, + "step": 8868 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0004, + "step": 8869 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2233, + "step": 8870 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9596, + "step": 8871 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0053, + "step": 8872 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3954, + "step": 8873 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3724, + "step": 8874 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0536, + "step": 8875 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.398, + "step": 8876 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1451, + "step": 8877 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1065, + "step": 8878 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3309, + "step": 8879 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0416, + "step": 8880 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8966, + "step": 8881 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3448, + "step": 8882 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1247, + "step": 8883 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8287, + "step": 8884 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.255, + "step": 8885 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2884, + "step": 8886 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9276, + "step": 8887 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3291, + "step": 8888 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2134, + "step": 8889 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0361, + "step": 8890 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1145, + "step": 8891 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4545, + "step": 8892 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0586, + "step": 8893 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3097, + "step": 8894 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4134, + "step": 8895 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0644, + "step": 8896 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3508, + "step": 8897 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.5335, + "step": 8898 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8071, + "step": 8899 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3341, + "step": 8900 + }, + { + "epoch": 0.89, + "eval_bleu": 16.04648990345959, + "eval_brevity_penalty": 0.545634399957386, + "eval_loss": 1.5220980644226074, + "eval_runtime": 5.3747, + "eval_samples_per_second": 557.803, + "eval_steps_per_second": 1.116, + "step": 8900 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0774, + "step": 8901 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.841, + "step": 8902 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0868, + "step": 8903 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1805, + "step": 8904 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8224, + "step": 8905 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3719, + "step": 8906 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9427, + "step": 8907 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1323, + "step": 8908 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1978, + "step": 8909 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.285, + "step": 8910 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1429, + "step": 8911 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3293, + "step": 8912 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4923, + "step": 8913 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0972, + "step": 8914 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3141, + "step": 8915 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3738, + "step": 8916 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2304, + "step": 8917 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.6171, + "step": 8918 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3949, + "step": 8919 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9436, + "step": 8920 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2561, + "step": 8921 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.6336, + "step": 8922 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9282, + "step": 8923 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.6825, + "step": 8924 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3013, + "step": 8925 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9382, + "step": 8926 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2263, + "step": 8927 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.224, + "step": 8928 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.9931, + "step": 8929 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.426, + "step": 8930 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3766, + "step": 8931 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1581, + "step": 8932 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4757, + "step": 8933 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.5351, + "step": 8934 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1852, + "step": 8935 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4243, + "step": 8936 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4742, + "step": 8937 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0713, + "step": 8938 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.1693, + "step": 8939 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.3176, + "step": 8940 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4497, + "step": 8941 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2763, + "step": 8942 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2791, + "step": 8943 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8811, + "step": 8944 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.0179, + "step": 8945 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.4799, + "step": 8946 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 1.8137, + "step": 8947 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.6167, + "step": 8948 + }, + { + "epoch": 0.89, + "learning_rate": 0.0001, + "loss": 2.2406, + "step": 8949 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.9364, + "step": 8950 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1011, + "step": 8951 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8436, + "step": 8952 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.861, + "step": 8953 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.2609, + "step": 8954 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1653, + "step": 8955 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.9962, + "step": 8956 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3494, + "step": 8957 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.644, + "step": 8958 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7889, + "step": 8959 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.334, + "step": 8960 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.0646, + "step": 8961 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.817, + "step": 8962 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4199, + "step": 8963 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6696, + "step": 8964 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7321, + "step": 8965 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5398, + "step": 8966 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1907, + "step": 8967 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8839, + "step": 8968 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4765, + "step": 8969 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.9882, + "step": 8970 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.7373, + "step": 8971 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1934, + "step": 8972 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.6436, + "step": 8973 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8975, + "step": 8974 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1619, + "step": 8975 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.2149, + "step": 8976 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.9547, + "step": 8977 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3066, + "step": 8978 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1503, + "step": 8979 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7129, + "step": 8980 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5451, + "step": 8981 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.2902, + "step": 8982 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.537, + "step": 8983 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5226, + "step": 8984 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1535, + "step": 8985 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6467, + "step": 8986 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4395, + "step": 8987 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7508, + "step": 8988 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.9131, + "step": 8989 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4237, + "step": 8990 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1923, + "step": 8991 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3071, + "step": 8992 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.2318, + "step": 8993 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5476, + "step": 8994 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1169, + "step": 8995 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4338, + "step": 8996 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.9868, + "step": 8997 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8637, + "step": 8998 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3691, + "step": 8999 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8405, + "step": 9000 + }, + { + "epoch": 0.9, + "eval_bleu": 16.08336016934143, + "eval_brevity_penalty": 0.54544409649819, + "eval_loss": 1.5201301574707031, + "eval_runtime": 5.3868, + "eval_samples_per_second": 556.544, + "eval_steps_per_second": 1.114, + "step": 9000 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6701, + "step": 9001 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.2622, + "step": 9002 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8406, + "step": 9003 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.5905, + "step": 9004 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1222, + "step": 9005 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.8715, + "step": 9006 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.4575, + "step": 9007 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5198, + "step": 9008 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.4637, + "step": 9009 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.393, + "step": 9010 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3082, + "step": 9011 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6122, + "step": 9012 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3885, + "step": 9013 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3988, + "step": 9014 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7301, + "step": 9015 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3903, + "step": 9016 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5556, + "step": 9017 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7502, + "step": 9018 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3368, + "step": 9019 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.195, + "step": 9020 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6079, + "step": 9021 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.4114, + "step": 9022 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.1846, + "step": 9023 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6189, + "step": 9024 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.4061, + "step": 9025 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3039, + "step": 9026 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7071, + "step": 9027 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3765, + "step": 9028 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.602, + "step": 9029 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.5941, + "step": 9030 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.2477, + "step": 9031 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3442, + "step": 9032 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.5112, + "step": 9033 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3932, + "step": 9034 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.49, + "step": 9035 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.5042, + "step": 9036 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.2224, + "step": 9037 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.4866, + "step": 9038 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7113, + "step": 9039 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.1272, + "step": 9040 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3159, + "step": 9041 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.6089, + "step": 9042 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.1879, + "step": 9043 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.5736, + "step": 9044 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.252, + "step": 9045 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.4169, + "step": 9046 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 2.3096, + "step": 9047 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.7771, + "step": 9048 + }, + { + "epoch": 0.9, + "learning_rate": 0.0001, + "loss": 1.3257, + "step": 9049 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2867, + "step": 9050 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4474, + "step": 9051 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1647, + "step": 9052 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.1695, + "step": 9053 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3477, + "step": 9054 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3473, + "step": 9055 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.6926, + "step": 9056 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.5917, + "step": 9057 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3721, + "step": 9058 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4401, + "step": 9059 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.8392, + "step": 9060 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2771, + "step": 9061 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3072, + "step": 9062 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4324, + "step": 9063 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.0816, + "step": 9064 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.305, + "step": 9065 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4736, + "step": 9066 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4408, + "step": 9067 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.403, + "step": 9068 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3761, + "step": 9069 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3374, + "step": 9070 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.503, + "step": 9071 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1801, + "step": 9072 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2232, + "step": 9073 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3519, + "step": 9074 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3722, + "step": 9075 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2341, + "step": 9076 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.5465, + "step": 9077 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3449, + "step": 9078 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2867, + "step": 9079 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.5527, + "step": 9080 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.7643, + "step": 9081 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2168, + "step": 9082 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.5493, + "step": 9083 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4739, + "step": 9084 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.176, + "step": 9085 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.0211, + "step": 9086 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.5314, + "step": 9087 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4495, + "step": 9088 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.5798, + "step": 9089 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3177, + "step": 9090 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1477, + "step": 9091 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2764, + "step": 9092 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4128, + "step": 9093 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2074, + "step": 9094 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.1835, + "step": 9095 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.283, + "step": 9096 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1779, + "step": 9097 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.1347, + "step": 9098 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.6509, + "step": 9099 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4133, + "step": 9100 + }, + { + "epoch": 0.91, + "eval_bleu": 15.861924043395135, + "eval_brevity_penalty": 0.5441961487950252, + "eval_loss": 1.5249515771865845, + "eval_runtime": 5.1639, + "eval_samples_per_second": 580.571, + "eval_steps_per_second": 1.162, + "step": 9100 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2708, + "step": 9101 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.6952, + "step": 9102 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2689, + "step": 9103 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.6508, + "step": 9104 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.5591, + "step": 9105 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1826, + "step": 9106 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2904, + "step": 9107 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4008, + "step": 9108 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4795, + "step": 9109 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4768, + "step": 9110 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.7252, + "step": 9111 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3441, + "step": 9112 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.1042, + "step": 9113 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.5133, + "step": 9114 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2743, + "step": 9115 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3858, + "step": 9116 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4258, + "step": 9117 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1433, + "step": 9118 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4429, + "step": 9119 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3524, + "step": 9120 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1712, + "step": 9121 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4711, + "step": 9122 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4452, + "step": 9123 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1324, + "step": 9124 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3024, + "step": 9125 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3263, + "step": 9126 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1197, + "step": 9127 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2802, + "step": 9128 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.9718, + "step": 9129 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3806, + "step": 9130 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.409, + "step": 9131 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4436, + "step": 9132 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2977, + "step": 9133 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2439, + "step": 9134 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.4475, + "step": 9135 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2677, + "step": 9136 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3757, + "step": 9137 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.5342, + "step": 9138 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2728, + "step": 9139 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.3782, + "step": 9140 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2113, + "step": 9141 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2071, + "step": 9142 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4811, + "step": 9143 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3686, + "step": 9144 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.2152, + "step": 9145 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.2021, + "step": 9146 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.3752, + "step": 9147 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 1.1147, + "step": 9148 + }, + { + "epoch": 0.91, + "learning_rate": 0.0001, + "loss": 2.4839, + "step": 9149 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3726, + "step": 9150 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.9669, + "step": 9151 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.9443, + "step": 9152 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.0783, + "step": 9153 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4855, + "step": 9154 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1515, + "step": 9155 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.738, + "step": 9156 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3053, + "step": 9157 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1585, + "step": 9158 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.0798, + "step": 9159 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.8555, + "step": 9160 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2873, + "step": 9161 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1843, + "step": 9162 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1096, + "step": 9163 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2429, + "step": 9164 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1934, + "step": 9165 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2229, + "step": 9166 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.047, + "step": 9167 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.831, + "step": 9168 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.5284, + "step": 9169 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.4792, + "step": 9170 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.9481, + "step": 9171 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4607, + "step": 9172 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.5334, + "step": 9173 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.8424, + "step": 9174 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4038, + "step": 9175 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.5435, + "step": 9176 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.6643, + "step": 9177 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4668, + "step": 9178 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2152, + "step": 9179 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.7213, + "step": 9180 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4376, + "step": 9181 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1661, + "step": 9182 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.565, + "step": 9183 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4831, + "step": 9184 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.3835, + "step": 9185 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.6224, + "step": 9186 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.358, + "step": 9187 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2485, + "step": 9188 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.7747, + "step": 9189 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2877, + "step": 9190 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.3824, + "step": 9191 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3451, + "step": 9192 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2647, + "step": 9193 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2445, + "step": 9194 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.5228, + "step": 9195 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2444, + "step": 9196 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.4561, + "step": 9197 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.5578, + "step": 9198 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2776, + "step": 9199 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.4374, + "step": 9200 + }, + { + "epoch": 0.92, + "eval_bleu": 15.817443823029095, + "eval_brevity_penalty": 0.5428839950561254, + "eval_loss": 1.5260621309280396, + "eval_runtime": 5.5974, + "eval_samples_per_second": 535.607, + "eval_steps_per_second": 1.072, + "step": 9200 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.809, + "step": 9201 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2925, + "step": 9202 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.3156, + "step": 9203 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.6551, + "step": 9204 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.5285, + "step": 9205 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.9246, + "step": 9206 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.6342, + "step": 9207 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.1134, + "step": 9208 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.335, + "step": 9209 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.256, + "step": 9210 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2839, + "step": 9211 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2217, + "step": 9212 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3809, + "step": 9213 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.13, + "step": 9214 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2726, + "step": 9215 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.444, + "step": 9216 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.1566, + "step": 9217 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2127, + "step": 9218 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.5088, + "step": 9219 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.0628, + "step": 9220 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.1645, + "step": 9221 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4531, + "step": 9222 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4177, + "step": 9223 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2174, + "step": 9224 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2889, + "step": 9225 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3879, + "step": 9226 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.632, + "step": 9227 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.411, + "step": 9228 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.122, + "step": 9229 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2913, + "step": 9230 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4318, + "step": 9231 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2044, + "step": 9232 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.4545, + "step": 9233 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.7506, + "step": 9234 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2267, + "step": 9235 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.4024, + "step": 9236 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.8117, + "step": 9237 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.227, + "step": 9238 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2714, + "step": 9239 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3318, + "step": 9240 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.2061, + "step": 9241 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.5053, + "step": 9242 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.3717, + "step": 9243 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.1484, + "step": 9244 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.3158, + "step": 9245 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4651, + "step": 9246 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.4742, + "step": 9247 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 2.2509, + "step": 9248 + }, + { + "epoch": 0.92, + "learning_rate": 0.0001, + "loss": 1.519, + "step": 9249 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.1106, + "step": 9250 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2384, + "step": 9251 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.447, + "step": 9252 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.1989, + "step": 9253 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3121, + "step": 9254 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8535, + "step": 9255 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.224, + "step": 9256 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.8708, + "step": 9257 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2676, + "step": 9258 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.1857, + "step": 9259 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3122, + "step": 9260 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7617, + "step": 9261 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.1226, + "step": 9262 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2367, + "step": 9263 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.556, + "step": 9264 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.2078, + "step": 9265 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3887, + "step": 9266 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8788, + "step": 9267 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.34, + "step": 9268 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3199, + "step": 9269 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.6574, + "step": 9270 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.2346, + "step": 9271 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2378, + "step": 9272 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.6491, + "step": 9273 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5946, + "step": 9274 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2918, + "step": 9275 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.5576, + "step": 9276 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.4971, + "step": 9277 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3552, + "step": 9278 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5262, + "step": 9279 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3179, + "step": 9280 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.9583, + "step": 9281 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.4028, + "step": 9282 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5812, + "step": 9283 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2964, + "step": 9284 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8604, + "step": 9285 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.9177, + "step": 9286 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.259, + "step": 9287 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3888, + "step": 9288 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.9717, + "step": 9289 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4766, + "step": 9290 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2374, + "step": 9291 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7745, + "step": 9292 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3488, + "step": 9293 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3445, + "step": 9294 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8573, + "step": 9295 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4031, + "step": 9296 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7141, + "step": 9297 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.4025, + "step": 9298 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4679, + "step": 9299 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3627, + "step": 9300 + }, + { + "epoch": 0.93, + "eval_bleu": 15.754064195794705, + "eval_brevity_penalty": 0.5449999917894122, + "eval_loss": 1.5256609916687012, + "eval_runtime": 5.2003, + "eval_samples_per_second": 576.508, + "eval_steps_per_second": 1.154, + "step": 9300 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.4798, + "step": 9301 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.5882, + "step": 9302 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.0732, + "step": 9303 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7865, + "step": 9304 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4161, + "step": 9305 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2329, + "step": 9306 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.9228, + "step": 9307 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4053, + "step": 9308 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4711, + "step": 9309 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.6342, + "step": 9310 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3826, + "step": 9311 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3949, + "step": 9312 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.0371, + "step": 9313 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.5038, + "step": 9314 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.6972, + "step": 9315 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.334, + "step": 9316 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2874, + "step": 9317 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8783, + "step": 9318 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3818, + "step": 9319 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3685, + "step": 9320 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8844, + "step": 9321 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.762, + "step": 9322 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3772, + "step": 9323 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4248, + "step": 9324 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8563, + "step": 9325 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.965, + "step": 9326 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.4973, + "step": 9327 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.592, + "step": 9328 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3123, + "step": 9329 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.9513, + "step": 9330 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3911, + "step": 9331 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.8406, + "step": 9332 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5769, + "step": 9333 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3757, + "step": 9334 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.1649, + "step": 9335 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7432, + "step": 9336 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.317, + "step": 9337 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.2989, + "step": 9338 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5235, + "step": 9339 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3393, + "step": 9340 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.5481, + "step": 9341 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.5621, + "step": 9342 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3108, + "step": 9343 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.6964, + "step": 9344 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.7813, + "step": 9345 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.0215, + "step": 9346 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 2.3563, + "step": 9347 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.6882, + "step": 9348 + }, + { + "epoch": 0.93, + "learning_rate": 0.0001, + "loss": 1.3188, + "step": 9349 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3428, + "step": 9350 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.616, + "step": 9351 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.3477, + "step": 9352 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2177, + "step": 9353 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.8326, + "step": 9354 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7146, + "step": 9355 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2575, + "step": 9356 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6917, + "step": 9357 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7714, + "step": 9358 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2494, + "step": 9359 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.292, + "step": 9360 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4891, + "step": 9361 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.07, + "step": 9362 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.785, + "step": 9363 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4443, + "step": 9364 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3678, + "step": 9365 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7189, + "step": 9366 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1204, + "step": 9367 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3898, + "step": 9368 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6827, + "step": 9369 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.2162, + "step": 9370 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3282, + "step": 9371 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.59, + "step": 9372 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1303, + "step": 9373 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2857, + "step": 9374 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4216, + "step": 9375 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.2675, + "step": 9376 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3601, + "step": 9377 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5579, + "step": 9378 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4322, + "step": 9379 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.5557, + "step": 9380 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6057, + "step": 9381 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1331, + "step": 9382 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2638, + "step": 9383 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5299, + "step": 9384 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.3389, + "step": 9385 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4598, + "step": 9386 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.524, + "step": 9387 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.3537, + "step": 9388 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4462, + "step": 9389 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.0492, + "step": 9390 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9608, + "step": 9391 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.291, + "step": 9392 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9523, + "step": 9393 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1784, + "step": 9394 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3919, + "step": 9395 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4469, + "step": 9396 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.3295, + "step": 9397 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2552, + "step": 9398 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.0101, + "step": 9399 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5003, + "step": 9400 + }, + { + "epoch": 0.94, + "eval_bleu": 15.910919861930003, + "eval_brevity_penalty": 0.546268627104608, + "eval_loss": 1.5249271392822266, + "eval_runtime": 5.645, + "eval_samples_per_second": 531.088, + "eval_steps_per_second": 1.063, + "step": 9400 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1661, + "step": 9401 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9929, + "step": 9402 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.049, + "step": 9403 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1035, + "step": 9404 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4557, + "step": 9405 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4807, + "step": 9406 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1835, + "step": 9407 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6772, + "step": 9408 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9128, + "step": 9409 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2705, + "step": 9410 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5976, + "step": 9411 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.2557, + "step": 9412 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2584, + "step": 9413 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6698, + "step": 9414 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1139, + "step": 9415 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3364, + "step": 9416 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7316, + "step": 9417 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.479, + "step": 9418 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1113, + "step": 9419 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2073, + "step": 9420 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5209, + "step": 9421 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4204, + "step": 9422 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.6114, + "step": 9423 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.8172, + "step": 9424 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4673, + "step": 9425 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9617, + "step": 9426 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.6158, + "step": 9427 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3928, + "step": 9428 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.35, + "step": 9429 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.1276, + "step": 9430 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.082, + "step": 9431 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7047, + "step": 9432 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.0901, + "step": 9433 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4519, + "step": 9434 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.7824, + "step": 9435 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5066, + "step": 9436 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.3512, + "step": 9437 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.0833, + "step": 9438 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5793, + "step": 9439 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.8313, + "step": 9440 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.0932, + "step": 9441 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.4114, + "step": 9442 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.2212, + "step": 9443 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.8788, + "step": 9444 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.5041, + "step": 9445 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.4106, + "step": 9446 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1779, + "step": 9447 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.8223, + "step": 9448 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 2.1148, + "step": 9449 + }, + { + "epoch": 0.94, + "learning_rate": 0.0001, + "loss": 1.9663, + "step": 9450 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7829, + "step": 9451 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7957, + "step": 9452 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7266, + "step": 9453 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7109, + "step": 9454 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0058, + "step": 9455 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.5543, + "step": 9456 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8044, + "step": 9457 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.6581, + "step": 9458 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.6861, + "step": 9459 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.154, + "step": 9460 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1304, + "step": 9461 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1978, + "step": 9462 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.3888, + "step": 9463 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2064, + "step": 9464 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.6708, + "step": 9465 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.4617, + "step": 9466 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.6214, + "step": 9467 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1724, + "step": 9468 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8775, + "step": 9469 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2635, + "step": 9470 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1989, + "step": 9471 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1388, + "step": 9472 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3957, + "step": 9473 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3341, + "step": 9474 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.885, + "step": 9475 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3342, + "step": 9476 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4837, + "step": 9477 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0266, + "step": 9478 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3875, + "step": 9479 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2006, + "step": 9480 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7585, + "step": 9481 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3507, + "step": 9482 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.258, + "step": 9483 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7991, + "step": 9484 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1589, + "step": 9485 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.5002, + "step": 9486 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.886, + "step": 9487 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2916, + "step": 9488 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1703, + "step": 9489 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8303, + "step": 9490 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1559, + "step": 9491 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3126, + "step": 9492 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8771, + "step": 9493 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3568, + "step": 9494 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.9686, + "step": 9495 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.5521, + "step": 9496 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3267, + "step": 9497 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0135, + "step": 9498 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.5192, + "step": 9499 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2002, + "step": 9500 + }, + { + "epoch": 0.95, + "eval_bleu": 15.833758460290067, + "eval_brevity_penalty": 0.5434343465887842, + "eval_loss": 1.5251795053482056, + "eval_runtime": 5.196, + "eval_samples_per_second": 576.985, + "eval_steps_per_second": 1.155, + "step": 9500 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2488, + "step": 9501 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0876, + "step": 9502 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2625, + "step": 9503 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.617, + "step": 9504 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.5697, + "step": 9505 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2749, + "step": 9506 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.364, + "step": 9507 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7079, + "step": 9508 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2285, + "step": 9509 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.6561, + "step": 9510 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.853, + "step": 9511 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3354, + "step": 9512 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1794, + "step": 9513 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.1779, + "step": 9514 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2028, + "step": 9515 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7423, + "step": 9516 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.2014, + "step": 9517 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2365, + "step": 9518 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.9895, + "step": 9519 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8075, + "step": 9520 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3736, + "step": 9521 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.053, + "step": 9522 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.5937, + "step": 9523 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2598, + "step": 9524 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1394, + "step": 9525 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7247, + "step": 9526 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4294, + "step": 9527 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.957, + "step": 9528 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.8989, + "step": 9529 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4915, + "step": 9530 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.9874, + "step": 9531 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7198, + "step": 9532 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.242, + "step": 9533 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3705, + "step": 9534 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7619, + "step": 9535 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4475, + "step": 9536 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0889, + "step": 9537 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.6789, + "step": 9538 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.25, + "step": 9539 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3403, + "step": 9540 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 1.7202, + "step": 9541 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4281, + "step": 9542 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3795, + "step": 9543 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.1407, + "step": 9544 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3162, + "step": 9545 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3471, + "step": 9546 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.0693, + "step": 9547 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.4423, + "step": 9548 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.3873, + "step": 9549 + }, + { + "epoch": 0.95, + "learning_rate": 0.0001, + "loss": 2.2963, + "step": 9550 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5685, + "step": 9551 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5733, + "step": 9552 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1101, + "step": 9553 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4623, + "step": 9554 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5775, + "step": 9555 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.273, + "step": 9556 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3106, + "step": 9557 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5498, + "step": 9558 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2192, + "step": 9559 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4729, + "step": 9560 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.6769, + "step": 9561 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1563, + "step": 9562 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5196, + "step": 9563 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.7182, + "step": 9564 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0966, + "step": 9565 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2046, + "step": 9566 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5147, + "step": 9567 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1971, + "step": 9568 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4017, + "step": 9569 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5689, + "step": 9570 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1464, + "step": 9571 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.8633, + "step": 9572 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5678, + "step": 9573 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1754, + "step": 9574 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4428, + "step": 9575 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.7739, + "step": 9576 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0592, + "step": 9577 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2356, + "step": 9578 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4808, + "step": 9579 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1321, + "step": 9580 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5913, + "step": 9581 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5555, + "step": 9582 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2446, + "step": 9583 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1829, + "step": 9584 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5872, + "step": 9585 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2028, + "step": 9586 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3476, + "step": 9587 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.8382, + "step": 9588 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.154, + "step": 9589 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5967, + "step": 9590 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3875, + "step": 9591 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.8876, + "step": 9592 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2734, + "step": 9593 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.9889, + "step": 9594 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.5865, + "step": 9595 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3141, + "step": 9596 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4161, + "step": 9597 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1647, + "step": 9598 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4528, + "step": 9599 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3461, + "step": 9600 + }, + { + "epoch": 0.96, + "eval_bleu": 15.919472548742263, + "eval_brevity_penalty": 0.5468604088120174, + "eval_loss": 1.526186227798462, + "eval_runtime": 5.4713, + "eval_samples_per_second": 547.951, + "eval_steps_per_second": 1.097, + "step": 9600 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1147, + "step": 9601 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0739, + "step": 9602 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4169, + "step": 9603 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1094, + "step": 9604 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2198, + "step": 9605 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5691, + "step": 9606 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.9523, + "step": 9607 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2872, + "step": 9608 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3122, + "step": 9609 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1945, + "step": 9610 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5271, + "step": 9611 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.6971, + "step": 9612 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0811, + "step": 9613 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.9313, + "step": 9614 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.6114, + "step": 9615 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1818, + "step": 9616 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1988, + "step": 9617 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5992, + "step": 9618 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1097, + "step": 9619 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0783, + "step": 9620 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.7252, + "step": 9621 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2834, + "step": 9622 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1314, + "step": 9623 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2672, + "step": 9624 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1693, + "step": 9625 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3072, + "step": 9626 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3259, + "step": 9627 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.1627, + "step": 9628 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3317, + "step": 9629 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0862, + "step": 9630 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.8323, + "step": 9631 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3247, + "step": 9632 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.3949, + "step": 9633 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.8521, + "step": 9634 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.2469, + "step": 9635 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.0313, + "step": 9636 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.1601, + "step": 9637 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3391, + "step": 9638 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.6466, + "step": 9639 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.1833, + "step": 9640 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.4466, + "step": 9641 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.526, + "step": 9642 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.2524, + "step": 9643 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.3278, + "step": 9644 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.7444, + "step": 9645 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.3355, + "step": 9646 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.1585, + "step": 9647 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.5757, + "step": 9648 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 1.3454, + "step": 9649 + }, + { + "epoch": 0.96, + "learning_rate": 0.0001, + "loss": 2.5167, + "step": 9650 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.548, + "step": 9651 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2376, + "step": 9652 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3025, + "step": 9653 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.558, + "step": 9654 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2521, + "step": 9655 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.4102, + "step": 9656 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6826, + "step": 9657 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3405, + "step": 9658 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3578, + "step": 9659 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5085, + "step": 9660 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2709, + "step": 9661 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.1705, + "step": 9662 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6484, + "step": 9663 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4294, + "step": 9664 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.5054, + "step": 9665 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.7065, + "step": 9666 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3555, + "step": 9667 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.304, + "step": 9668 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6804, + "step": 9669 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2257, + "step": 9670 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.9658, + "step": 9671 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.741, + "step": 9672 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2982, + "step": 9673 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.0936, + "step": 9674 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6261, + "step": 9675 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2979, + "step": 9676 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3868, + "step": 9677 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5516, + "step": 9678 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2105, + "step": 9679 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3174, + "step": 9680 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3717, + "step": 9681 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.324, + "step": 9682 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2474, + "step": 9683 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.1591, + "step": 9684 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.23, + "step": 9685 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2096, + "step": 9686 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6165, + "step": 9687 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4055, + "step": 9688 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.4413, + "step": 9689 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.7214, + "step": 9690 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.1177, + "step": 9691 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2304, + "step": 9692 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6066, + "step": 9693 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4873, + "step": 9694 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3982, + "step": 9695 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5996, + "step": 9696 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2788, + "step": 9697 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3137, + "step": 9698 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.8849, + "step": 9699 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2607, + "step": 9700 + }, + { + "epoch": 0.97, + "eval_bleu": 15.836983672173979, + "eval_brevity_penalty": 0.5459092539670956, + "eval_loss": 1.5197162628173828, + "eval_runtime": 5.5089, + "eval_samples_per_second": 544.213, + "eval_steps_per_second": 1.089, + "step": 9700 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.447, + "step": 9701 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6558, + "step": 9702 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2602, + "step": 9703 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3589, + "step": 9704 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.0587, + "step": 9705 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3982, + "step": 9706 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3758, + "step": 9707 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.7925, + "step": 9708 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2712, + "step": 9709 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.7347, + "step": 9710 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4515, + "step": 9711 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3235, + "step": 9712 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.1352, + "step": 9713 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5991, + "step": 9714 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.289, + "step": 9715 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.9625, + "step": 9716 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5498, + "step": 9717 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4384, + "step": 9718 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.124, + "step": 9719 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.914, + "step": 9720 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4616, + "step": 9721 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2774, + "step": 9722 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.8963, + "step": 9723 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2507, + "step": 9724 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2223, + "step": 9725 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.8302, + "step": 9726 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.1914, + "step": 9727 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2643, + "step": 9728 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.8137, + "step": 9729 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3143, + "step": 9730 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.4357, + "step": 9731 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.7672, + "step": 9732 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.3143, + "step": 9733 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.4493, + "step": 9734 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6223, + "step": 9735 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.1632, + "step": 9736 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.101, + "step": 9737 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4019, + "step": 9738 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4179, + "step": 9739 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3436, + "step": 9740 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.6876, + "step": 9741 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2967, + "step": 9742 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.3939, + "step": 9743 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.5289, + "step": 9744 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.2839, + "step": 9745 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.2208, + "step": 9746 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4347, + "step": 9747 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.1733, + "step": 9748 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 2.6593, + "step": 9749 + }, + { + "epoch": 0.97, + "learning_rate": 0.0001, + "loss": 1.4743, + "step": 9750 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.1456, + "step": 9751 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.0507, + "step": 9752 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5735, + "step": 9753 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2415, + "step": 9754 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.4773, + "step": 9755 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4599, + "step": 9756 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.3577, + "step": 9757 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1061, + "step": 9758 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4361, + "step": 9759 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2305, + "step": 9760 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3044, + "step": 9761 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7694, + "step": 9762 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2514, + "step": 9763 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.5309, + "step": 9764 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7678, + "step": 9765 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2271, + "step": 9766 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.4576, + "step": 9767 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5849, + "step": 9768 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2324, + "step": 9769 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2414, + "step": 9770 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5305, + "step": 9771 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2789, + "step": 9772 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1767, + "step": 9773 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5866, + "step": 9774 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2645, + "step": 9775 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2299, + "step": 9776 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4879, + "step": 9777 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2674, + "step": 9778 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2694, + "step": 9779 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.6688, + "step": 9780 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.3207, + "step": 9781 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2402, + "step": 9782 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5988, + "step": 9783 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2118, + "step": 9784 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2809, + "step": 9785 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.6885, + "step": 9786 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2765, + "step": 9787 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.4405, + "step": 9788 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.8648, + "step": 9789 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2985, + "step": 9790 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.5889, + "step": 9791 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7744, + "step": 9792 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.3347, + "step": 9793 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.0424, + "step": 9794 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2926, + "step": 9795 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2361, + "step": 9796 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.7319, + "step": 9797 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.6913, + "step": 9798 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2487, + "step": 9799 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3737, + "step": 9800 + }, + { + "epoch": 0.98, + "eval_bleu": 16.057857941125956, + "eval_brevity_penalty": 0.5475365369372638, + "eval_loss": 1.5178272724151611, + "eval_runtime": 5.4086, + "eval_samples_per_second": 554.307, + "eval_steps_per_second": 1.109, + "step": 9800 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4504, + "step": 9801 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2285, + "step": 9802 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2929, + "step": 9803 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7765, + "step": 9804 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2235, + "step": 9805 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.403, + "step": 9806 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7112, + "step": 9807 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2188, + "step": 9808 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.5196, + "step": 9809 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.6869, + "step": 9810 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2873, + "step": 9811 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 3.0431, + "step": 9812 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.7576, + "step": 9813 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.202, + "step": 9814 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2172, + "step": 9815 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5832, + "step": 9816 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2005, + "step": 9817 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1479, + "step": 9818 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4901, + "step": 9819 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.2318, + "step": 9820 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.0853, + "step": 9821 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.0664, + "step": 9822 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.56, + "step": 9823 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2251, + "step": 9824 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.9641, + "step": 9825 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4879, + "step": 9826 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3319, + "step": 9827 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3012, + "step": 9828 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.4898, + "step": 9829 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.4914, + "step": 9830 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.673, + "step": 9831 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.5352, + "step": 9832 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3221, + "step": 9833 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2298, + "step": 9834 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.3424, + "step": 9835 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.9364, + "step": 9836 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1803, + "step": 9837 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1318, + "step": 9838 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1526, + "step": 9839 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2534, + "step": 9840 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.8646, + "step": 9841 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2115, + "step": 9842 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.2685, + "step": 9843 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.9756, + "step": 9844 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1703, + "step": 9845 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.1753, + "step": 9846 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.0351, + "step": 9847 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.3267, + "step": 9848 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 2.546, + "step": 9849 + }, + { + "epoch": 0.98, + "learning_rate": 0.0001, + "loss": 1.9059, + "step": 9850 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.463, + "step": 9851 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.4895, + "step": 9852 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8429, + "step": 9853 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3692, + "step": 9854 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3603, + "step": 9855 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.0734, + "step": 9856 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3022, + "step": 9857 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3923, + "step": 9858 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.9771, + "step": 9859 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.4467, + "step": 9860 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3977, + "step": 9861 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8203, + "step": 9862 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.7944, + "step": 9863 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.2511, + "step": 9864 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8041, + "step": 9865 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.2617, + "step": 9866 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3131, + "step": 9867 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.7265, + "step": 9868 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3028, + "step": 9869 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3249, + "step": 9870 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.9776, + "step": 9871 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3471, + "step": 9872 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3043, + "step": 9873 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.0972, + "step": 9874 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.1355, + "step": 9875 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.357, + "step": 9876 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8122, + "step": 9877 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3553, + "step": 9878 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.9578, + "step": 9879 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8268, + "step": 9880 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.4559, + "step": 9881 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.909, + "step": 9882 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.7364, + "step": 9883 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.1856, + "step": 9884 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.1543, + "step": 9885 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.9544, + "step": 9886 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.1748, + "step": 9887 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.2342, + "step": 9888 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8667, + "step": 9889 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3058, + "step": 9890 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.0773, + "step": 9891 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.7742, + "step": 9892 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.538, + "step": 9893 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.2916, + "step": 9894 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.8171, + "step": 9895 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.3911, + "step": 9896 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.5265, + "step": 9897 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.3105, + "step": 9898 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 2.1133, + "step": 9899 + }, + { + "epoch": 0.99, + "learning_rate": 0.0001, + "loss": 1.3968, + "step": 9900 + }, + { + "epoch": 0.99, + "eval_bleu": 16.172945052397182, + "eval_brevity_penalty": 0.5522425300246423, + "eval_loss": 1.513184905052185, + "eval_runtime": 6.7035, + "eval_samples_per_second": 447.23, + "eval_steps_per_second": 0.895, + "step": 9900 + } + ], + "max_steps": 10000, + "num_train_epochs": 9223372036854775807, + "total_flos": 8.36283420031058e+16, + "trial_name": null, + "trial_params": null +}