| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 1200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9975e-05, | |
| "loss": 14.5662, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.995e-05, | |
| "loss": 7.1236, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.992500000000001e-05, | |
| "loss": 5.3202, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.99e-05, | |
| "loss": 3.9452, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9875000000000006e-05, | |
| "loss": 3.4958, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9850000000000006e-05, | |
| "loss": 2.8169, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9825000000000005e-05, | |
| "loss": 1.9955, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9800000000000004e-05, | |
| "loss": 1.5664, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9775000000000004e-05, | |
| "loss": 1.3016, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.975e-05, | |
| "loss": 0.8614, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9725e-05, | |
| "loss": 1.0415, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.97e-05, | |
| "loss": 0.5895, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.967500000000001e-05, | |
| "loss": 0.4986, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.965e-05, | |
| "loss": 0.8056, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.962500000000001e-05, | |
| "loss": 0.8067, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.96e-05, | |
| "loss": 0.5058, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9575000000000006e-05, | |
| "loss": 0.6796, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9550000000000005e-05, | |
| "loss": 0.8734, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9525000000000004e-05, | |
| "loss": 0.9313, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9500000000000004e-05, | |
| "loss": 0.5948, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9475e-05, | |
| "loss": 0.6952, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.945e-05, | |
| "loss": 0.562, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9425e-05, | |
| "loss": 0.7806, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.94e-05, | |
| "loss": 0.5262, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.937500000000001e-05, | |
| "loss": 0.7553, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.935e-05, | |
| "loss": 0.7249, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9325000000000006e-05, | |
| "loss": 0.7524, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.93e-05, | |
| "loss": 0.6196, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9275000000000005e-05, | |
| "loss": 0.7026, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9250000000000004e-05, | |
| "loss": 0.7122, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9225000000000004e-05, | |
| "loss": 0.7, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.92e-05, | |
| "loss": 0.6362, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9175e-05, | |
| "loss": 0.5532, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.915e-05, | |
| "loss": 0.735, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9125e-05, | |
| "loss": 0.6036, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.91e-05, | |
| "loss": 0.583, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.907500000000001e-05, | |
| "loss": 0.5929, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.905e-05, | |
| "loss": 0.7786, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9025000000000006e-05, | |
| "loss": 0.6554, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.6304, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8975000000000005e-05, | |
| "loss": 0.6701, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8950000000000004e-05, | |
| "loss": 0.6078, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8925e-05, | |
| "loss": 0.6914, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.89e-05, | |
| "loss": 0.6281, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.8875e-05, | |
| "loss": 0.6858, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.885e-05, | |
| "loss": 0.5675, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.8825e-05, | |
| "loss": 0.7266, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.88e-05, | |
| "loss": 0.9016, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8775000000000007e-05, | |
| "loss": 0.5519, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.875e-05, | |
| "loss": 0.5509, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.8725000000000005e-05, | |
| "loss": 1.0457, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.87e-05, | |
| "loss": 0.8419, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.8675000000000004e-05, | |
| "loss": 0.6735, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.8650000000000003e-05, | |
| "loss": 0.5402, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8625e-05, | |
| "loss": 0.5331, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.86e-05, | |
| "loss": 0.5752, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8575e-05, | |
| "loss": 0.5226, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.855e-05, | |
| "loss": 0.6734, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.8525e-05, | |
| "loss": 0.4719, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.85e-05, | |
| "loss": 0.5542, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.8475000000000006e-05, | |
| "loss": 0.7252, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.845e-05, | |
| "loss": 0.5549, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.8425000000000005e-05, | |
| "loss": 0.4797, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.8400000000000004e-05, | |
| "loss": 0.6463, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.8375000000000004e-05, | |
| "loss": 0.6462, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.835e-05, | |
| "loss": 0.5555, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8325e-05, | |
| "loss": 0.7847, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.83e-05, | |
| "loss": 0.6703, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8275e-05, | |
| "loss": 0.5331, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.825e-05, | |
| "loss": 0.5694, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.822500000000001e-05, | |
| "loss": 0.6958, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.82e-05, | |
| "loss": 0.6098, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.8175000000000005e-05, | |
| "loss": 0.5491, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.815e-05, | |
| "loss": 0.6715, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.8125000000000004e-05, | |
| "loss": 0.3916, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.8100000000000004e-05, | |
| "loss": 0.6793, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.8075e-05, | |
| "loss": 0.5973, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.805e-05, | |
| "loss": 1.2975, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.8025e-05, | |
| "loss": 0.6328, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.4737, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.7975e-05, | |
| "loss": 0.6352, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.795e-05, | |
| "loss": 0.4901, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.7925000000000006e-05, | |
| "loss": 0.5981, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.79e-05, | |
| "loss": 0.507, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.7875000000000005e-05, | |
| "loss": 0.8278, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.785e-05, | |
| "loss": 0.8347, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.7825000000000004e-05, | |
| "loss": 0.5688, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.78e-05, | |
| "loss": 0.5499, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.7775e-05, | |
| "loss": 0.8471, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.775e-05, | |
| "loss": 0.917, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.7725e-05, | |
| "loss": 0.6619, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.77e-05, | |
| "loss": 0.7828, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.7675e-05, | |
| "loss": 0.5367, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.765e-05, | |
| "loss": 0.4836, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.7625000000000006e-05, | |
| "loss": 0.5622, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.76e-05, | |
| "loss": 0.9229, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.7575000000000004e-05, | |
| "loss": 0.5629, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.755e-05, | |
| "loss": 0.4308, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.7525e-05, | |
| "loss": 0.4076, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.75e-05, | |
| "loss": 0.4774, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.7475e-05, | |
| "loss": 0.5469, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.745e-05, | |
| "loss": 0.5152, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.7425e-05, | |
| "loss": 0.6016, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.74e-05, | |
| "loss": 0.5576, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.7375e-05, | |
| "loss": 0.6042, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.735e-05, | |
| "loss": 0.4603, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.7325000000000005e-05, | |
| "loss": 0.6274, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.73e-05, | |
| "loss": 0.4953, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.7275000000000004e-05, | |
| "loss": 0.5774, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.7249999999999997e-05, | |
| "loss": 0.7247, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7225e-05, | |
| "loss": 0.541, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.72e-05, | |
| "loss": 0.8272, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7175e-05, | |
| "loss": 0.7274, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.715e-05, | |
| "loss": 0.5989, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7125e-05, | |
| "loss": 0.4759, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.71e-05, | |
| "loss": 0.5049, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.7075e-05, | |
| "loss": 0.4231, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.705e-05, | |
| "loss": 0.6353, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.7025000000000005e-05, | |
| "loss": 0.7109, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.4595, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.6975000000000003e-05, | |
| "loss": 0.4064, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.695e-05, | |
| "loss": 0.5207, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.6925e-05, | |
| "loss": 0.5171, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.69e-05, | |
| "loss": 0.5339, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.6875e-05, | |
| "loss": 0.6237, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.685000000000001e-05, | |
| "loss": 0.459, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.6825e-05, | |
| "loss": 0.4735, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.6800000000000006e-05, | |
| "loss": 0.623, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.6775000000000005e-05, | |
| "loss": 0.5452, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.6750000000000005e-05, | |
| "loss": 0.5576, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.6725000000000004e-05, | |
| "loss": 0.4818, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.6700000000000003e-05, | |
| "loss": 0.5624, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.6675e-05, | |
| "loss": 0.6891, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.665e-05, | |
| "loss": 0.5276, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.6625e-05, | |
| "loss": 0.4512, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.660000000000001e-05, | |
| "loss": 0.5978, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.6575e-05, | |
| "loss": 0.5461, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.655000000000001e-05, | |
| "loss": 0.5528, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.6525e-05, | |
| "loss": 0.4676, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.6500000000000005e-05, | |
| "loss": 0.4727, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.6475000000000005e-05, | |
| "loss": 0.4363, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.6450000000000004e-05, | |
| "loss": 0.5155, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.6425000000000004e-05, | |
| "loss": 0.593, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.64e-05, | |
| "loss": 0.4865, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.6375e-05, | |
| "loss": 0.5427, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.635e-05, | |
| "loss": 0.3649, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.6325e-05, | |
| "loss": 0.3859, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.630000000000001e-05, | |
| "loss": 0.4382, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.6275e-05, | |
| "loss": 0.5052, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.6250000000000006e-05, | |
| "loss": 0.7267, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.6225e-05, | |
| "loss": 0.3374, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.6200000000000005e-05, | |
| "loss": 0.4778, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.6175000000000004e-05, | |
| "loss": 0.4859, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.6150000000000004e-05, | |
| "loss": 0.279, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.6125e-05, | |
| "loss": 1.4163, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.61e-05, | |
| "loss": 0.3552, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.6075e-05, | |
| "loss": 0.4339, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.605e-05, | |
| "loss": 0.5127, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.6025e-05, | |
| "loss": 0.3658, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.339, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.5975e-05, | |
| "loss": 0.5611, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.5950000000000006e-05, | |
| "loss": 0.493, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.5925e-05, | |
| "loss": 0.5616, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.5900000000000004e-05, | |
| "loss": 0.8039, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.5875000000000004e-05, | |
| "loss": 0.4011, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.585e-05, | |
| "loss": 0.5168, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.5825e-05, | |
| "loss": 0.5843, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.58e-05, | |
| "loss": 0.5999, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.5775e-05, | |
| "loss": 0.5357, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.575e-05, | |
| "loss": 0.3802, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.5725e-05, | |
| "loss": 0.5216, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.5700000000000006e-05, | |
| "loss": 0.4038, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.5675e-05, | |
| "loss": 0.3344, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.5650000000000005e-05, | |
| "loss": 0.5792, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.5625e-05, | |
| "loss": 0.453, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.5600000000000004e-05, | |
| "loss": 0.3997, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.5575e-05, | |
| "loss": 0.3398, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.555e-05, | |
| "loss": 0.3204, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.5525e-05, | |
| "loss": 0.291, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.55e-05, | |
| "loss": 0.3974, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.5475e-05, | |
| "loss": 0.3647, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.545000000000001e-05, | |
| "loss": 0.3516, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.5425e-05, | |
| "loss": 0.3865, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.5400000000000006e-05, | |
| "loss": 0.4976, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.5375e-05, | |
| "loss": 0.3554, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.5350000000000005e-05, | |
| "loss": 0.526, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.5325000000000004e-05, | |
| "loss": 0.5028, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.53e-05, | |
| "loss": 0.2651, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.5275e-05, | |
| "loss": 0.2753, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.525e-05, | |
| "loss": 0.3614, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.5225e-05, | |
| "loss": 0.4194, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.52e-05, | |
| "loss": 0.4073, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5175e-05, | |
| "loss": 0.8908, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5150000000000006e-05, | |
| "loss": 0.5112, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5125e-05, | |
| "loss": 0.4367, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.5100000000000005e-05, | |
| "loss": 0.5748, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.5075e-05, | |
| "loss": 0.4903, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5050000000000004e-05, | |
| "loss": 0.4007, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5025000000000003e-05, | |
| "loss": 0.3218, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.2959, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6915902829595236, | |
| "eval_f1": 0.7120379490516452, | |
| "eval_loss": 0.3397860825061798, | |
| "eval_precision": 0.7520909138849339, | |
| "eval_recall": 0.6915902829595236, | |
| "eval_runtime": 46.5586, | |
| "eval_samples_per_second": 17.183, | |
| "eval_steps_per_second": 2.148, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.4975e-05, | |
| "loss": 0.4021, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.495e-05, | |
| "loss": 0.3147, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.4925e-05, | |
| "loss": 0.7101, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.49e-05, | |
| "loss": 0.4026, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.4875e-05, | |
| "loss": 0.3797, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.4850000000000006e-05, | |
| "loss": 0.6129, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.4825e-05, | |
| "loss": 0.4431, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.4800000000000005e-05, | |
| "loss": 0.381, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.4775e-05, | |
| "loss": 0.2991, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.4750000000000004e-05, | |
| "loss": 0.5143, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.4725e-05, | |
| "loss": 0.3687, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.47e-05, | |
| "loss": 0.3229, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.4675e-05, | |
| "loss": 0.3542, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.465e-05, | |
| "loss": 0.3149, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.4625e-05, | |
| "loss": 0.4093, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.46e-05, | |
| "loss": 0.3791, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.4575e-05, | |
| "loss": 0.2603, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.4550000000000005e-05, | |
| "loss": 0.4405, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.4525e-05, | |
| "loss": 0.4683, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.4500000000000004e-05, | |
| "loss": 0.3554, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.4475e-05, | |
| "loss": 0.5901, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.445e-05, | |
| "loss": 0.4715, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 4.4425e-05, | |
| "loss": 0.2257, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.44e-05, | |
| "loss": 0.3887, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.4375e-05, | |
| "loss": 0.3492, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.435e-05, | |
| "loss": 0.3245, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.4325e-05, | |
| "loss": 0.6609, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.43e-05, | |
| "loss": 0.1563, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.4275e-05, | |
| "loss": 0.3054, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.4250000000000005e-05, | |
| "loss": 0.3336, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.4225e-05, | |
| "loss": 0.2735, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.4200000000000004e-05, | |
| "loss": 0.5098, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.4174999999999996e-05, | |
| "loss": 0.517, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.415e-05, | |
| "loss": 0.4756, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.4125e-05, | |
| "loss": 0.386, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.41e-05, | |
| "loss": 0.6908, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.4075e-05, | |
| "loss": 0.3154, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.405e-05, | |
| "loss": 0.7584, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.4025e-05, | |
| "loss": 0.3773, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.3767, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.3975e-05, | |
| "loss": 0.3647, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.3950000000000004e-05, | |
| "loss": 0.297, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.3925e-05, | |
| "loss": 0.3088, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.39e-05, | |
| "loss": 0.4385, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.3875e-05, | |
| "loss": 0.2397, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.385e-05, | |
| "loss": 0.3477, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.3825e-05, | |
| "loss": 0.3544, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.38e-05, | |
| "loss": 0.3124, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.3775e-05, | |
| "loss": 0.5746, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.375e-05, | |
| "loss": 0.2082, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.3725000000000006e-05, | |
| "loss": 0.421, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.3700000000000005e-05, | |
| "loss": 0.2457, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.3675000000000005e-05, | |
| "loss": 0.5156, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.3650000000000004e-05, | |
| "loss": 0.3815, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.3625e-05, | |
| "loss": 0.4878, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.36e-05, | |
| "loss": 0.3831, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.3575e-05, | |
| "loss": 0.4634, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.355e-05, | |
| "loss": 0.4368, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.352500000000001e-05, | |
| "loss": 0.3983, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.35e-05, | |
| "loss": 0.4583, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.3475000000000006e-05, | |
| "loss": 0.5595, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.345e-05, | |
| "loss": 0.3677, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.3425000000000005e-05, | |
| "loss": 0.4267, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.3400000000000005e-05, | |
| "loss": 0.3406, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.3375000000000004e-05, | |
| "loss": 0.4084, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.335e-05, | |
| "loss": 0.3459, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.3325e-05, | |
| "loss": 0.4401, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.33e-05, | |
| "loss": 0.6531, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.3275e-05, | |
| "loss": 0.2975, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.325e-05, | |
| "loss": 0.2649, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 4.322500000000001e-05, | |
| "loss": 0.3199, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.32e-05, | |
| "loss": 0.342, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.3175000000000006e-05, | |
| "loss": 0.631, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.315e-05, | |
| "loss": 0.5276, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.3125000000000005e-05, | |
| "loss": 0.2495, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.3100000000000004e-05, | |
| "loss": 0.3855, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.3075000000000003e-05, | |
| "loss": 0.599, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.305e-05, | |
| "loss": 0.4825, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.3025e-05, | |
| "loss": 0.4482, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.456, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.2975e-05, | |
| "loss": 0.299, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 4.295e-05, | |
| "loss": 0.3304, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.2925000000000007e-05, | |
| "loss": 0.3654, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.29e-05, | |
| "loss": 0.1515, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.2875000000000005e-05, | |
| "loss": 0.2824, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.285e-05, | |
| "loss": 0.4656, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.2825000000000004e-05, | |
| "loss": 0.5197, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.2800000000000004e-05, | |
| "loss": 0.5618, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.2775e-05, | |
| "loss": 0.4091, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.275e-05, | |
| "loss": 0.6243, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.2725e-05, | |
| "loss": 0.2485, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.27e-05, | |
| "loss": 0.3898, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.2675e-05, | |
| "loss": 0.373, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 4.265e-05, | |
| "loss": 0.4957, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.2625000000000006e-05, | |
| "loss": 0.3787, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.26e-05, | |
| "loss": 0.3621, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.2575000000000005e-05, | |
| "loss": 0.3356, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.2550000000000004e-05, | |
| "loss": 0.58, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.2525000000000004e-05, | |
| "loss": 0.3739, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.25e-05, | |
| "loss": 0.265, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.2475e-05, | |
| "loss": 0.541, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.245e-05, | |
| "loss": 0.6195, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.2425e-05, | |
| "loss": 0.2676, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.24e-05, | |
| "loss": 0.4074, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.237500000000001e-05, | |
| "loss": 0.4149, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 4.235e-05, | |
| "loss": 0.3282, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.2325000000000006e-05, | |
| "loss": 0.3182, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.23e-05, | |
| "loss": 0.2904, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.2275000000000004e-05, | |
| "loss": 0.3478, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.2250000000000004e-05, | |
| "loss": 0.3693, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.2225e-05, | |
| "loss": 0.4881, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.22e-05, | |
| "loss": 0.2592, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.2175e-05, | |
| "loss": 0.4696, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.215e-05, | |
| "loss": 0.2685, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 4.2125e-05, | |
| "loss": 0.3528, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.21e-05, | |
| "loss": 0.3008, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.2075000000000006e-05, | |
| "loss": 0.2616, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.205e-05, | |
| "loss": 0.6541, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 4.2025000000000005e-05, | |
| "loss": 0.2476, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.2073, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.1975000000000004e-05, | |
| "loss": 0.4824, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.195e-05, | |
| "loss": 0.338, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.1925e-05, | |
| "loss": 0.3602, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.19e-05, | |
| "loss": 0.4082, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 4.1875e-05, | |
| "loss": 0.2916, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 4.185e-05, | |
| "loss": 0.2902, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.1825e-05, | |
| "loss": 0.3261, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 4.18e-05, | |
| "loss": 0.2599, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.1775000000000006e-05, | |
| "loss": 0.4284, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 4.175e-05, | |
| "loss": 0.3981, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.1725000000000005e-05, | |
| "loss": 0.3607, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.17e-05, | |
| "loss": 0.3214, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.1675e-05, | |
| "loss": 0.5022, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.165e-05, | |
| "loss": 0.2203, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.1625e-05, | |
| "loss": 0.2736, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.16e-05, | |
| "loss": 0.2597, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.1575e-05, | |
| "loss": 0.3038, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.155e-05, | |
| "loss": 0.3396, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.1525e-05, | |
| "loss": 0.3447, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.15e-05, | |
| "loss": 0.3168, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.1475000000000005e-05, | |
| "loss": 0.3432, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.145e-05, | |
| "loss": 0.3895, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.1425000000000004e-05, | |
| "loss": 0.387, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.14e-05, | |
| "loss": 0.2479, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.1375e-05, | |
| "loss": 0.3109, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.135e-05, | |
| "loss": 0.2703, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.1325e-05, | |
| "loss": 0.3302, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 4.13e-05, | |
| "loss": 0.1983, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.1275e-05, | |
| "loss": 0.293, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.125e-05, | |
| "loss": 0.5641, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 4.1225e-05, | |
| "loss": 0.3746, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 4.12e-05, | |
| "loss": 0.1783, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.1175000000000005e-05, | |
| "loss": 0.2852, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.115e-05, | |
| "loss": 0.3112, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.1125000000000004e-05, | |
| "loss": 0.1936, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 4.11e-05, | |
| "loss": 0.3486, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.1075e-05, | |
| "loss": 0.2528, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.105e-05, | |
| "loss": 0.3082, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 4.1025e-05, | |
| "loss": 0.5328, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.3125, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.0975e-05, | |
| "loss": 0.2036, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.095e-05, | |
| "loss": 0.295, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.0925000000000005e-05, | |
| "loss": 0.3459, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.09e-05, | |
| "loss": 0.3806, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 4.0875000000000004e-05, | |
| "loss": 0.34, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.085e-05, | |
| "loss": 0.4861, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.0825e-05, | |
| "loss": 0.231, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.08e-05, | |
| "loss": 0.3442, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 4.0775e-05, | |
| "loss": 0.5056, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.075e-05, | |
| "loss": 0.468, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 4.0725e-05, | |
| "loss": 0.2309, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.07e-05, | |
| "loss": 0.2692, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 4.0675e-05, | |
| "loss": 0.3229, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 4.065e-05, | |
| "loss": 0.4786, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.0625000000000005e-05, | |
| "loss": 0.3494, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 4.0600000000000004e-05, | |
| "loss": 0.2746, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.0575000000000004e-05, | |
| "loss": 0.4341, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.055e-05, | |
| "loss": 0.2186, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.0525e-05, | |
| "loss": 0.5763, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.05e-05, | |
| "loss": 0.5783, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.0475e-05, | |
| "loss": 0.3916, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 4.045000000000001e-05, | |
| "loss": 0.3211, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.0425e-05, | |
| "loss": 0.3737, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 0.3801, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.0375e-05, | |
| "loss": 0.3124, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.0350000000000005e-05, | |
| "loss": 0.4814, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.0325000000000004e-05, | |
| "loss": 0.4904, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.0300000000000004e-05, | |
| "loss": 0.2338, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.0275e-05, | |
| "loss": 0.2274, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.025e-05, | |
| "loss": 0.3754, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.0225e-05, | |
| "loss": 0.5782, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.02e-05, | |
| "loss": 0.2568, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 4.0175e-05, | |
| "loss": 0.3817, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 4.015000000000001e-05, | |
| "loss": 0.1244, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.0125e-05, | |
| "loss": 0.3574, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.0100000000000006e-05, | |
| "loss": 0.184, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.0075e-05, | |
| "loss": 0.5445, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.0050000000000004e-05, | |
| "loss": 0.1936, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.0025000000000004e-05, | |
| "loss": 0.4148, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4133, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7303129273927457, | |
| "eval_f1": 0.7240910454808391, | |
| "eval_loss": 0.33349618315696716, | |
| "eval_precision": 0.7247629591832809, | |
| "eval_recall": 0.7303129273927457, | |
| "eval_runtime": 46.5217, | |
| "eval_samples_per_second": 17.196, | |
| "eval_steps_per_second": 2.15, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.9975e-05, | |
| "loss": 0.4279, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.995e-05, | |
| "loss": 0.3595, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.9925e-05, | |
| "loss": 0.2921, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.99e-05, | |
| "loss": 0.167, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.9875e-05, | |
| "loss": 0.2244, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.9850000000000006e-05, | |
| "loss": 0.1951, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.9825e-05, | |
| "loss": 0.2426, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.9800000000000005e-05, | |
| "loss": 0.3611, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 3.9775e-05, | |
| "loss": 0.3127, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 3.9750000000000004e-05, | |
| "loss": 0.2799, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.9725e-05, | |
| "loss": 0.4368, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.97e-05, | |
| "loss": 0.3167, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.9675e-05, | |
| "loss": 0.1099, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.965e-05, | |
| "loss": 0.2974, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.9625e-05, | |
| "loss": 0.2618, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.960000000000001e-05, | |
| "loss": 0.1214, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 3.9575e-05, | |
| "loss": 0.4326, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.9550000000000006e-05, | |
| "loss": 0.1978, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.9525e-05, | |
| "loss": 0.3145, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.9500000000000005e-05, | |
| "loss": 0.3392, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 3.9475000000000004e-05, | |
| "loss": 0.2255, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 3.9450000000000003e-05, | |
| "loss": 0.4223, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.9425e-05, | |
| "loss": 0.2917, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.94e-05, | |
| "loss": 0.3284, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.9375e-05, | |
| "loss": 0.2235, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.935e-05, | |
| "loss": 0.2539, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 3.9325e-05, | |
| "loss": 0.2643, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.9300000000000007e-05, | |
| "loss": 0.1961, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.9275e-05, | |
| "loss": 0.1986, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.9250000000000005e-05, | |
| "loss": 0.0867, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 3.9225e-05, | |
| "loss": 0.2181, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 3.9200000000000004e-05, | |
| "loss": 0.0727, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.9175000000000004e-05, | |
| "loss": 0.2379, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.915e-05, | |
| "loss": 0.2092, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.9125e-05, | |
| "loss": 0.1124, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.91e-05, | |
| "loss": 0.2252, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.9075e-05, | |
| "loss": 0.4025, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.905e-05, | |
| "loss": 0.2782, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.9025e-05, | |
| "loss": 0.5048, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.3301, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.8975e-05, | |
| "loss": 0.3589, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.8950000000000005e-05, | |
| "loss": 0.4942, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.8925e-05, | |
| "loss": 0.4335, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 3.8900000000000004e-05, | |
| "loss": 0.1538, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.8875e-05, | |
| "loss": 0.442, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.885e-05, | |
| "loss": 0.4319, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.8825e-05, | |
| "loss": 0.3939, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 3.88e-05, | |
| "loss": 0.1888, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.8775e-05, | |
| "loss": 0.6402, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.875e-05, | |
| "loss": 0.1708, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.8725e-05, | |
| "loss": 0.3315, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.8700000000000006e-05, | |
| "loss": 0.3908, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.8675e-05, | |
| "loss": 0.2653, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.8650000000000004e-05, | |
| "loss": 0.2681, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.8625e-05, | |
| "loss": 0.1869, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.86e-05, | |
| "loss": 0.2696, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.8575e-05, | |
| "loss": 0.4004, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.855e-05, | |
| "loss": 0.2019, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.8525e-05, | |
| "loss": 0.1742, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.85e-05, | |
| "loss": 0.3658, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.8475e-05, | |
| "loss": 0.1322, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.845e-05, | |
| "loss": 0.1761, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.8425e-05, | |
| "loss": 0.2245, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.8400000000000005e-05, | |
| "loss": 0.3481, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.8375e-05, | |
| "loss": 0.326, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.8350000000000004e-05, | |
| "loss": 0.2885, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.8324999999999996e-05, | |
| "loss": 0.1434, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.83e-05, | |
| "loss": 0.2765, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.8275e-05, | |
| "loss": 0.3264, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.825e-05, | |
| "loss": 0.4146, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.8225e-05, | |
| "loss": 0.2461, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 3.82e-05, | |
| "loss": 0.4797, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.8175e-05, | |
| "loss": 0.3502, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.8150000000000006e-05, | |
| "loss": 0.1427, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.8125e-05, | |
| "loss": 0.164, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.8100000000000005e-05, | |
| "loss": 0.1947, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.8075e-05, | |
| "loss": 0.1943, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 3.805e-05, | |
| "loss": 0.6346, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.8025e-05, | |
| "loss": 0.6651, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.4448, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.7975e-05, | |
| "loss": 0.4063, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.795e-05, | |
| "loss": 0.3017, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.7925e-05, | |
| "loss": 0.1966, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.79e-05, | |
| "loss": 0.3202, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.7875e-05, | |
| "loss": 0.1277, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.7850000000000005e-05, | |
| "loss": 0.2548, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7825e-05, | |
| "loss": 0.1849, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7800000000000004e-05, | |
| "loss": 0.1185, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7775e-05, | |
| "loss": 0.4444, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 3.775e-05, | |
| "loss": 0.2993, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.7725e-05, | |
| "loss": 0.1434, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.77e-05, | |
| "loss": 0.1306, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.7675e-05, | |
| "loss": 0.0721, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.765e-05, | |
| "loss": 0.1188, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.7625e-05, | |
| "loss": 0.128, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.76e-05, | |
| "loss": 0.3053, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.7575e-05, | |
| "loss": 0.293, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.7550000000000005e-05, | |
| "loss": 0.2691, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 3.7525e-05, | |
| "loss": 0.1903, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 3.7500000000000003e-05, | |
| "loss": 0.1, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 3.7475e-05, | |
| "loss": 0.1671, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.745e-05, | |
| "loss": 0.3463, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.7425e-05, | |
| "loss": 0.2526, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.74e-05, | |
| "loss": 0.4419, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.737500000000001e-05, | |
| "loss": 0.5626, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.735e-05, | |
| "loss": 0.2871, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.7325000000000006e-05, | |
| "loss": 0.0462, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.73e-05, | |
| "loss": 0.2153, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.7275000000000005e-05, | |
| "loss": 0.1801, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 3.7250000000000004e-05, | |
| "loss": 0.2419, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.7225000000000004e-05, | |
| "loss": 0.4692, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.72e-05, | |
| "loss": 0.3793, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.7175e-05, | |
| "loss": 0.3421, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 3.715e-05, | |
| "loss": 0.3306, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.7125e-05, | |
| "loss": 0.1118, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.71e-05, | |
| "loss": 0.1968, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.707500000000001e-05, | |
| "loss": 0.2947, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 3.705e-05, | |
| "loss": 0.177, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 3.7025000000000005e-05, | |
| "loss": 0.2388, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.2208, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.6975000000000004e-05, | |
| "loss": 0.0863, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 3.6950000000000004e-05, | |
| "loss": 0.1208, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.6925e-05, | |
| "loss": 0.161, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.69e-05, | |
| "loss": 0.2525, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.6875e-05, | |
| "loss": 0.0939, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.685e-05, | |
| "loss": 0.1857, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.6825e-05, | |
| "loss": 0.4203, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.68e-05, | |
| "loss": 0.1674, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.6775000000000006e-05, | |
| "loss": 0.2935, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.675e-05, | |
| "loss": 0.4283, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.6725000000000005e-05, | |
| "loss": 0.0551, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.6700000000000004e-05, | |
| "loss": 0.2137, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.6675000000000004e-05, | |
| "loss": 0.0585, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.665e-05, | |
| "loss": 0.5005, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 3.6625e-05, | |
| "loss": 0.2685, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.66e-05, | |
| "loss": 0.2799, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.6575e-05, | |
| "loss": 0.4154, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.655e-05, | |
| "loss": 0.3866, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.652500000000001e-05, | |
| "loss": 0.275, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 3.65e-05, | |
| "loss": 0.0749, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 3.6475000000000006e-05, | |
| "loss": 0.4298, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 3.645e-05, | |
| "loss": 0.3015, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 3.6425000000000004e-05, | |
| "loss": 0.4693, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.6400000000000004e-05, | |
| "loss": 0.5553, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.6375e-05, | |
| "loss": 0.2838, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.635e-05, | |
| "loss": 0.2485, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.6325e-05, | |
| "loss": 0.1477, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.63e-05, | |
| "loss": 0.1499, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.6275e-05, | |
| "loss": 0.5901, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.625e-05, | |
| "loss": 0.3647, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.6225000000000006e-05, | |
| "loss": 0.29, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.62e-05, | |
| "loss": 0.2172, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.6175000000000005e-05, | |
| "loss": 0.2973, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.615e-05, | |
| "loss": 0.3948, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.6125000000000004e-05, | |
| "loss": 0.134, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.61e-05, | |
| "loss": 0.3913, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.6075e-05, | |
| "loss": 0.167, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.605e-05, | |
| "loss": 0.3009, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.6025e-05, | |
| "loss": 0.2548, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.1343, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.5975e-05, | |
| "loss": 0.4397, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.595e-05, | |
| "loss": 0.2612, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.5925000000000006e-05, | |
| "loss": 0.3434, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.59e-05, | |
| "loss": 0.7125, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.5875000000000005e-05, | |
| "loss": 0.4386, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.585e-05, | |
| "loss": 0.4036, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 3.5825000000000003e-05, | |
| "loss": 0.0801, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.58e-05, | |
| "loss": 0.1766, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.5775e-05, | |
| "loss": 0.2278, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.575e-05, | |
| "loss": 0.2785, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.5725e-05, | |
| "loss": 0.1986, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.57e-05, | |
| "loss": 0.3308, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.5675e-05, | |
| "loss": 0.1467, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.565e-05, | |
| "loss": 0.2535, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.5625000000000005e-05, | |
| "loss": 0.2237, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.56e-05, | |
| "loss": 0.1238, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 3.5575000000000004e-05, | |
| "loss": 0.241, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.555e-05, | |
| "loss": 0.2651, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.5525e-05, | |
| "loss": 0.3775, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 3.55e-05, | |
| "loss": 0.3317, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.5475e-05, | |
| "loss": 0.2678, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 3.545e-05, | |
| "loss": 0.1845, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.5425e-05, | |
| "loss": 0.1676, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.54e-05, | |
| "loss": 0.205, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.5375e-05, | |
| "loss": 0.3689, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 3.535e-05, | |
| "loss": 0.0595, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.5325000000000005e-05, | |
| "loss": 0.1113, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.53e-05, | |
| "loss": 0.5781, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.5275000000000004e-05, | |
| "loss": 0.1274, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.525e-05, | |
| "loss": 0.0737, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.5225e-05, | |
| "loss": 0.169, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.52e-05, | |
| "loss": 0.1855, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 3.5175e-05, | |
| "loss": 0.1207, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.515e-05, | |
| "loss": 0.2846, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.5125e-05, | |
| "loss": 0.2958, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.51e-05, | |
| "loss": 0.3242, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.5075000000000006e-05, | |
| "loss": 0.2893, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.505e-05, | |
| "loss": 0.2842, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.5025000000000004e-05, | |
| "loss": 0.2714, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.3058, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7556953969672984, | |
| "eval_f1": 0.7496529757688725, | |
| "eval_loss": 0.4001426696777344, | |
| "eval_precision": 0.7470013420094054, | |
| "eval_recall": 0.7556953969672984, | |
| "eval_runtime": 46.5671, | |
| "eval_samples_per_second": 17.18, | |
| "eval_steps_per_second": 2.147, | |
| "step": 1200 | |
| } | |
| ], | |
| "max_steps": 4000, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.077633511424e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |