| { |
| "best_metric": 0.5174081237911026, |
| "best_model_checkpoint": "./experiment/train_0308_t5_small_db_content_26001_weight_decay_00001_ls_00_lr_constant_bs_2064/checkpoint-8192", |
| "epoch": 2730.5890410958905, |
| "global_step": 8192, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.29, |
| "learning_rate": 0.0001, |
| "loss": 4.4465, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 0.0001, |
| "loss": 4.1272, |
| "step": 3 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 0.0001, |
| "loss": 4.0649, |
| "step": 6 |
| }, |
| { |
| "epoch": 2.88, |
| "learning_rate": 0.0001, |
| "loss": 3.5316, |
| "step": 9 |
| }, |
| { |
| "epoch": 3.88, |
| "learning_rate": 0.0001, |
| "loss": 3.1042, |
| "step": 12 |
| }, |
| { |
| "epoch": 4.88, |
| "learning_rate": 0.0001, |
| "loss": 2.733, |
| "step": 15 |
| }, |
| { |
| "epoch": 5.88, |
| "learning_rate": 0.0001, |
| "loss": 2.4421, |
| "step": 18 |
| }, |
| { |
| "epoch": 6.88, |
| "learning_rate": 0.0001, |
| "loss": 2.1818, |
| "step": 21 |
| }, |
| { |
| "epoch": 7.88, |
| "learning_rate": 0.0001, |
| "loss": 1.9857, |
| "step": 24 |
| }, |
| { |
| "epoch": 8.88, |
| "learning_rate": 0.0001, |
| "loss": 1.8413, |
| "step": 27 |
| }, |
| { |
| "epoch": 9.88, |
| "learning_rate": 0.0001, |
| "loss": 1.686, |
| "step": 30 |
| }, |
| { |
| "epoch": 10.88, |
| "learning_rate": 0.0001, |
| "loss": 1.5623, |
| "step": 33 |
| }, |
| { |
| "epoch": 11.88, |
| "learning_rate": 0.0001, |
| "loss": 1.4631, |
| "step": 36 |
| }, |
| { |
| "epoch": 12.88, |
| "learning_rate": 0.0001, |
| "loss": 1.3776, |
| "step": 39 |
| }, |
| { |
| "epoch": 13.88, |
| "learning_rate": 0.0001, |
| "loss": 1.3096, |
| "step": 42 |
| }, |
| { |
| "epoch": 14.88, |
| "learning_rate": 0.0001, |
| "loss": 1.2347, |
| "step": 45 |
| }, |
| { |
| "epoch": 15.88, |
| "learning_rate": 0.0001, |
| "loss": 1.1792, |
| "step": 48 |
| }, |
| { |
| "epoch": 16.88, |
| "learning_rate": 0.0001, |
| "loss": 1.1255, |
| "step": 51 |
| }, |
| { |
| "epoch": 17.88, |
| "learning_rate": 0.0001, |
| "loss": 1.0843, |
| "step": 54 |
| }, |
| { |
| "epoch": 18.88, |
| "learning_rate": 0.0001, |
| "loss": 1.0435, |
| "step": 57 |
| }, |
| { |
| "epoch": 19.88, |
| "learning_rate": 0.0001, |
| "loss": 1.0021, |
| "step": 60 |
| }, |
| { |
| "epoch": 20.88, |
| "learning_rate": 0.0001, |
| "loss": 0.971, |
| "step": 63 |
| }, |
| { |
| "epoch": 21.29, |
| "eval_exact_match": 0.04061895551257253, |
| "eval_exec": 0.043520309477756286, |
| "eval_loss": 0.6706878542900085, |
| "eval_runtime": 693.5751, |
| "eval_samples_per_second": 1.491, |
| "step": 64 |
| }, |
| { |
| "epoch": 21.88, |
| "learning_rate": 0.0001, |
| "loss": 0.9412, |
| "step": 66 |
| }, |
| { |
| "epoch": 22.88, |
| "learning_rate": 0.0001, |
| "loss": 0.9117, |
| "step": 69 |
| }, |
| { |
| "epoch": 23.88, |
| "learning_rate": 0.0001, |
| "loss": 0.8753, |
| "step": 72 |
| }, |
| { |
| "epoch": 24.88, |
| "learning_rate": 0.0001, |
| "loss": 0.8514, |
| "step": 75 |
| }, |
| { |
| "epoch": 25.88, |
| "learning_rate": 0.0001, |
| "loss": 0.8317, |
| "step": 78 |
| }, |
| { |
| "epoch": 26.88, |
| "learning_rate": 0.0001, |
| "loss": 0.8019, |
| "step": 81 |
| }, |
| { |
| "epoch": 27.88, |
| "learning_rate": 0.0001, |
| "loss": 0.7857, |
| "step": 84 |
| }, |
| { |
| "epoch": 28.88, |
| "learning_rate": 0.0001, |
| "loss": 0.7623, |
| "step": 87 |
| }, |
| { |
| "epoch": 29.88, |
| "learning_rate": 0.0001, |
| "loss": 0.7418, |
| "step": 90 |
| }, |
| { |
| "epoch": 30.88, |
| "learning_rate": 0.0001, |
| "loss": 0.7291, |
| "step": 93 |
| }, |
| { |
| "epoch": 31.88, |
| "learning_rate": 0.0001, |
| "loss": 0.7086, |
| "step": 96 |
| }, |
| { |
| "epoch": 32.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6875, |
| "step": 99 |
| }, |
| { |
| "epoch": 33.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6781, |
| "step": 102 |
| }, |
| { |
| "epoch": 34.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6609, |
| "step": 105 |
| }, |
| { |
| "epoch": 35.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6443, |
| "step": 108 |
| }, |
| { |
| "epoch": 36.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6356, |
| "step": 111 |
| }, |
| { |
| "epoch": 37.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6093, |
| "step": 114 |
| }, |
| { |
| "epoch": 38.88, |
| "learning_rate": 0.0001, |
| "loss": 0.6128, |
| "step": 117 |
| }, |
| { |
| "epoch": 39.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5916, |
| "step": 120 |
| }, |
| { |
| "epoch": 40.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5842, |
| "step": 123 |
| }, |
| { |
| "epoch": 41.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5757, |
| "step": 126 |
| }, |
| { |
| "epoch": 42.59, |
| "eval_exact_match": 0.21179883945841393, |
| "eval_exec": 0.2205029013539652, |
| "eval_loss": 0.45285192131996155, |
| "eval_runtime": 131.42, |
| "eval_samples_per_second": 7.868, |
| "step": 128 |
| }, |
| { |
| "epoch": 42.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5613, |
| "step": 129 |
| }, |
| { |
| "epoch": 43.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5474, |
| "step": 132 |
| }, |
| { |
| "epoch": 44.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5398, |
| "step": 135 |
| }, |
| { |
| "epoch": 45.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5321, |
| "step": 138 |
| }, |
| { |
| "epoch": 46.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5247, |
| "step": 141 |
| }, |
| { |
| "epoch": 47.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5135, |
| "step": 144 |
| }, |
| { |
| "epoch": 48.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5065, |
| "step": 147 |
| }, |
| { |
| "epoch": 49.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4933, |
| "step": 150 |
| }, |
| { |
| "epoch": 50.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4881, |
| "step": 153 |
| }, |
| { |
| "epoch": 51.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4831, |
| "step": 156 |
| }, |
| { |
| "epoch": 52.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4778, |
| "step": 159 |
| }, |
| { |
| "epoch": 53.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4641, |
| "step": 162 |
| }, |
| { |
| "epoch": 54.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4581, |
| "step": 165 |
| }, |
| { |
| "epoch": 55.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4536, |
| "step": 168 |
| }, |
| { |
| "epoch": 56.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4472, |
| "step": 171 |
| }, |
| { |
| "epoch": 57.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4394, |
| "step": 174 |
| }, |
| { |
| "epoch": 58.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4311, |
| "step": 177 |
| }, |
| { |
| "epoch": 59.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4283, |
| "step": 180 |
| }, |
| { |
| "epoch": 60.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4192, |
| "step": 183 |
| }, |
| { |
| "epoch": 61.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4132, |
| "step": 186 |
| }, |
| { |
| "epoch": 62.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4106, |
| "step": 189 |
| }, |
| { |
| "epoch": 63.88, |
| "learning_rate": 0.0001, |
| "loss": 0.4049, |
| "step": 192 |
| }, |
| { |
| "epoch": 63.88, |
| "eval_exact_match": 0.28820116054158607, |
| "eval_exec": 0.3056092843326886, |
| "eval_loss": 0.37886810302734375, |
| "eval_runtime": 134.3287, |
| "eval_samples_per_second": 7.698, |
| "step": 192 |
| }, |
| { |
| "epoch": 64.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3949, |
| "step": 195 |
| }, |
| { |
| "epoch": 65.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3943, |
| "step": 198 |
| }, |
| { |
| "epoch": 66.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3912, |
| "step": 201 |
| }, |
| { |
| "epoch": 67.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3833, |
| "step": 204 |
| }, |
| { |
| "epoch": 68.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3804, |
| "step": 207 |
| }, |
| { |
| "epoch": 69.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3694, |
| "step": 210 |
| }, |
| { |
| "epoch": 70.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3723, |
| "step": 213 |
| }, |
| { |
| "epoch": 71.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3637, |
| "step": 216 |
| }, |
| { |
| "epoch": 72.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3622, |
| "step": 219 |
| }, |
| { |
| "epoch": 73.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3541, |
| "step": 222 |
| }, |
| { |
| "epoch": 74.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3519, |
| "step": 225 |
| }, |
| { |
| "epoch": 75.88, |
| "learning_rate": 0.0001, |
| "loss": 0.346, |
| "step": 228 |
| }, |
| { |
| "epoch": 76.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3441, |
| "step": 231 |
| }, |
| { |
| "epoch": 77.88, |
| "learning_rate": 0.0001, |
| "loss": 0.339, |
| "step": 234 |
| }, |
| { |
| "epoch": 78.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3334, |
| "step": 237 |
| }, |
| { |
| "epoch": 79.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3317, |
| "step": 240 |
| }, |
| { |
| "epoch": 80.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3247, |
| "step": 243 |
| }, |
| { |
| "epoch": 81.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3225, |
| "step": 246 |
| }, |
| { |
| "epoch": 82.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3173, |
| "step": 249 |
| }, |
| { |
| "epoch": 83.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3148, |
| "step": 252 |
| }, |
| { |
| "epoch": 84.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3123, |
| "step": 255 |
| }, |
| { |
| "epoch": 85.29, |
| "eval_exact_match": 0.3288201160541586, |
| "eval_exec": 0.3471953578336557, |
| "eval_loss": 0.3469275236129761, |
| "eval_runtime": 135.0596, |
| "eval_samples_per_second": 7.656, |
| "step": 256 |
| }, |
| { |
| "epoch": 85.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3067, |
| "step": 258 |
| }, |
| { |
| "epoch": 86.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3062, |
| "step": 261 |
| }, |
| { |
| "epoch": 87.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3044, |
| "step": 264 |
| }, |
| { |
| "epoch": 88.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2976, |
| "step": 267 |
| }, |
| { |
| "epoch": 89.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2978, |
| "step": 270 |
| }, |
| { |
| "epoch": 90.88, |
| "learning_rate": 0.0001, |
| "loss": 0.292, |
| "step": 273 |
| }, |
| { |
| "epoch": 91.88, |
| "learning_rate": 0.0001, |
| "loss": 0.29, |
| "step": 276 |
| }, |
| { |
| "epoch": 92.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2863, |
| "step": 279 |
| }, |
| { |
| "epoch": 93.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2832, |
| "step": 282 |
| }, |
| { |
| "epoch": 94.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2787, |
| "step": 285 |
| }, |
| { |
| "epoch": 95.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2751, |
| "step": 288 |
| }, |
| { |
| "epoch": 96.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2741, |
| "step": 291 |
| }, |
| { |
| "epoch": 97.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2724, |
| "step": 294 |
| }, |
| { |
| "epoch": 98.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2677, |
| "step": 297 |
| }, |
| { |
| "epoch": 99.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2673, |
| "step": 300 |
| }, |
| { |
| "epoch": 100.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2619, |
| "step": 303 |
| }, |
| { |
| "epoch": 101.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2584, |
| "step": 306 |
| }, |
| { |
| "epoch": 102.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2575, |
| "step": 309 |
| }, |
| { |
| "epoch": 103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2525, |
| "step": 312 |
| }, |
| { |
| "epoch": 104.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2535, |
| "step": 315 |
| }, |
| { |
| "epoch": 105.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2479, |
| "step": 318 |
| }, |
| { |
| "epoch": 106.59, |
| "eval_exact_match": 0.35880077369439073, |
| "eval_exec": 0.3820116054158607, |
| "eval_loss": 0.33587977290153503, |
| "eval_runtime": 103.6076, |
| "eval_samples_per_second": 9.98, |
| "step": 320 |
| }, |
| { |
| "epoch": 106.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2464, |
| "step": 321 |
| }, |
| { |
| "epoch": 107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2455, |
| "step": 324 |
| }, |
| { |
| "epoch": 108.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2403, |
| "step": 327 |
| }, |
| { |
| "epoch": 109.88, |
| "learning_rate": 0.0001, |
| "loss": 0.242, |
| "step": 330 |
| }, |
| { |
| "epoch": 110.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2381, |
| "step": 333 |
| }, |
| { |
| "epoch": 111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2344, |
| "step": 336 |
| }, |
| { |
| "epoch": 112.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2343, |
| "step": 339 |
| }, |
| { |
| "epoch": 113.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2328, |
| "step": 342 |
| }, |
| { |
| "epoch": 114.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2297, |
| "step": 345 |
| }, |
| { |
| "epoch": 115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2303, |
| "step": 348 |
| }, |
| { |
| "epoch": 116.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2247, |
| "step": 351 |
| }, |
| { |
| "epoch": 117.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2255, |
| "step": 354 |
| }, |
| { |
| "epoch": 118.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2243, |
| "step": 357 |
| }, |
| { |
| "epoch": 119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2215, |
| "step": 360 |
| }, |
| { |
| "epoch": 120.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2215, |
| "step": 363 |
| }, |
| { |
| "epoch": 121.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2158, |
| "step": 366 |
| }, |
| { |
| "epoch": 122.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2168, |
| "step": 369 |
| }, |
| { |
| "epoch": 123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2135, |
| "step": 372 |
| }, |
| { |
| "epoch": 124.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2138, |
| "step": 375 |
| }, |
| { |
| "epoch": 125.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2113, |
| "step": 378 |
| }, |
| { |
| "epoch": 126.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2095, |
| "step": 381 |
| }, |
| { |
| "epoch": 127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2085, |
| "step": 384 |
| }, |
| { |
| "epoch": 127.88, |
| "eval_exact_match": 0.379110251450677, |
| "eval_exec": 0.40038684719535783, |
| "eval_loss": 0.3285636603832245, |
| "eval_runtime": 143.1024, |
| "eval_samples_per_second": 7.226, |
| "step": 384 |
| }, |
| { |
| "epoch": 128.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2072, |
| "step": 387 |
| }, |
| { |
| "epoch": 129.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2056, |
| "step": 390 |
| }, |
| { |
| "epoch": 130.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2023, |
| "step": 393 |
| }, |
| { |
| "epoch": 131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.201, |
| "step": 396 |
| }, |
| { |
| "epoch": 132.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1982, |
| "step": 399 |
| }, |
| { |
| "epoch": 133.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1977, |
| "step": 402 |
| }, |
| { |
| "epoch": 134.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1976, |
| "step": 405 |
| }, |
| { |
| "epoch": 135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1946, |
| "step": 408 |
| }, |
| { |
| "epoch": 136.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1961, |
| "step": 411 |
| }, |
| { |
| "epoch": 137.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1931, |
| "step": 414 |
| }, |
| { |
| "epoch": 138.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1913, |
| "step": 417 |
| }, |
| { |
| "epoch": 139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1905, |
| "step": 420 |
| }, |
| { |
| "epoch": 140.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1885, |
| "step": 423 |
| }, |
| { |
| "epoch": 141.88, |
| "learning_rate": 0.0001, |
| "loss": 0.187, |
| "step": 426 |
| }, |
| { |
| "epoch": 142.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1846, |
| "step": 429 |
| }, |
| { |
| "epoch": 143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1859, |
| "step": 432 |
| }, |
| { |
| "epoch": 144.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1835, |
| "step": 435 |
| }, |
| { |
| "epoch": 145.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1811, |
| "step": 438 |
| }, |
| { |
| "epoch": 146.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1817, |
| "step": 441 |
| }, |
| { |
| "epoch": 147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1799, |
| "step": 444 |
| }, |
| { |
| "epoch": 148.88, |
| "learning_rate": 0.0001, |
| "loss": 0.179, |
| "step": 447 |
| }, |
| { |
| "epoch": 149.29, |
| "eval_exact_match": 0.40618955512572535, |
| "eval_exec": 0.42263056092843326, |
| "eval_loss": 0.3326466381549835, |
| "eval_runtime": 101.2464, |
| "eval_samples_per_second": 10.213, |
| "step": 448 |
| }, |
| { |
| "epoch": 149.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1773, |
| "step": 450 |
| }, |
| { |
| "epoch": 150.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1763, |
| "step": 453 |
| }, |
| { |
| "epoch": 151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1766, |
| "step": 456 |
| }, |
| { |
| "epoch": 152.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1754, |
| "step": 459 |
| }, |
| { |
| "epoch": 153.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1727, |
| "step": 462 |
| }, |
| { |
| "epoch": 154.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1698, |
| "step": 465 |
| }, |
| { |
| "epoch": 155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1711, |
| "step": 468 |
| }, |
| { |
| "epoch": 156.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1686, |
| "step": 471 |
| }, |
| { |
| "epoch": 157.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1684, |
| "step": 474 |
| }, |
| { |
| "epoch": 158.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1695, |
| "step": 477 |
| }, |
| { |
| "epoch": 159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1649, |
| "step": 480 |
| }, |
| { |
| "epoch": 160.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1644, |
| "step": 483 |
| }, |
| { |
| "epoch": 161.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1644, |
| "step": 486 |
| }, |
| { |
| "epoch": 162.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1636, |
| "step": 489 |
| }, |
| { |
| "epoch": 163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1611, |
| "step": 492 |
| }, |
| { |
| "epoch": 164.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1596, |
| "step": 495 |
| }, |
| { |
| "epoch": 165.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1608, |
| "step": 498 |
| }, |
| { |
| "epoch": 166.88, |
| "learning_rate": 0.0001, |
| "loss": 0.159, |
| "step": 501 |
| }, |
| { |
| "epoch": 167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1583, |
| "step": 504 |
| }, |
| { |
| "epoch": 168.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1572, |
| "step": 507 |
| }, |
| { |
| "epoch": 169.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1564, |
| "step": 510 |
| }, |
| { |
| "epoch": 170.59, |
| "eval_exact_match": 0.41779497098646035, |
| "eval_exec": 0.43036750483558994, |
| "eval_loss": 0.3266688883304596, |
| "eval_runtime": 118.2565, |
| "eval_samples_per_second": 8.744, |
| "step": 512 |
| }, |
| { |
| "epoch": 170.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1515, |
| "step": 513 |
| }, |
| { |
| "epoch": 171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1531, |
| "step": 516 |
| }, |
| { |
| "epoch": 172.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1527, |
| "step": 519 |
| }, |
| { |
| "epoch": 173.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1517, |
| "step": 522 |
| }, |
| { |
| "epoch": 174.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1502, |
| "step": 525 |
| }, |
| { |
| "epoch": 175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1483, |
| "step": 528 |
| }, |
| { |
| "epoch": 176.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1497, |
| "step": 531 |
| }, |
| { |
| "epoch": 177.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1488, |
| "step": 534 |
| }, |
| { |
| "epoch": 178.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1447, |
| "step": 537 |
| }, |
| { |
| "epoch": 179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1467, |
| "step": 540 |
| }, |
| { |
| "epoch": 180.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1453, |
| "step": 543 |
| }, |
| { |
| "epoch": 181.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1444, |
| "step": 546 |
| }, |
| { |
| "epoch": 182.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1441, |
| "step": 549 |
| }, |
| { |
| "epoch": 183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.143, |
| "step": 552 |
| }, |
| { |
| "epoch": 184.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1419, |
| "step": 555 |
| }, |
| { |
| "epoch": 185.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1413, |
| "step": 558 |
| }, |
| { |
| "epoch": 186.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1405, |
| "step": 561 |
| }, |
| { |
| "epoch": 187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1389, |
| "step": 564 |
| }, |
| { |
| "epoch": 188.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1382, |
| "step": 567 |
| }, |
| { |
| "epoch": 189.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1364, |
| "step": 570 |
| }, |
| { |
| "epoch": 190.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1361, |
| "step": 573 |
| }, |
| { |
| "epoch": 191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1371, |
| "step": 576 |
| }, |
| { |
| "epoch": 191.88, |
| "eval_exact_match": 0.41102514506769827, |
| "eval_exec": 0.42359767891682787, |
| "eval_loss": 0.33160606026649475, |
| "eval_runtime": 124.8766, |
| "eval_samples_per_second": 8.28, |
| "step": 576 |
| }, |
| { |
| "epoch": 192.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1353, |
| "step": 579 |
| }, |
| { |
| "epoch": 193.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1336, |
| "step": 582 |
| }, |
| { |
| "epoch": 194.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1336, |
| "step": 585 |
| }, |
| { |
| "epoch": 195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1335, |
| "step": 588 |
| }, |
| { |
| "epoch": 196.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1332, |
| "step": 591 |
| }, |
| { |
| "epoch": 197.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1312, |
| "step": 594 |
| }, |
| { |
| "epoch": 198.88, |
| "learning_rate": 0.0001, |
| "loss": 0.13, |
| "step": 597 |
| }, |
| { |
| "epoch": 199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1278, |
| "step": 600 |
| }, |
| { |
| "epoch": 200.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1314, |
| "step": 603 |
| }, |
| { |
| "epoch": 201.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1273, |
| "step": 606 |
| }, |
| { |
| "epoch": 202.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1258, |
| "step": 609 |
| }, |
| { |
| "epoch": 203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1264, |
| "step": 612 |
| }, |
| { |
| "epoch": 204.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1256, |
| "step": 615 |
| }, |
| { |
| "epoch": 205.88, |
| "learning_rate": 0.0001, |
| "loss": 0.125, |
| "step": 618 |
| }, |
| { |
| "epoch": 206.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1235, |
| "step": 621 |
| }, |
| { |
| "epoch": 207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.124, |
| "step": 624 |
| }, |
| { |
| "epoch": 208.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1236, |
| "step": 627 |
| }, |
| { |
| "epoch": 209.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1224, |
| "step": 630 |
| }, |
| { |
| "epoch": 210.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1212, |
| "step": 633 |
| }, |
| { |
| "epoch": 211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1218, |
| "step": 636 |
| }, |
| { |
| "epoch": 212.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1204, |
| "step": 639 |
| }, |
| { |
| "epoch": 213.29, |
| "eval_exact_match": 0.42940038684719534, |
| "eval_exec": 0.4332688588007737, |
| "eval_loss": 0.33052191138267517, |
| "eval_runtime": 117.5462, |
| "eval_samples_per_second": 8.797, |
| "step": 640 |
| }, |
| { |
| "epoch": 213.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1195, |
| "step": 642 |
| }, |
| { |
| "epoch": 214.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1181, |
| "step": 645 |
| }, |
| { |
| "epoch": 215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1194, |
| "step": 648 |
| }, |
| { |
| "epoch": 216.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1177, |
| "step": 651 |
| }, |
| { |
| "epoch": 217.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1172, |
| "step": 654 |
| }, |
| { |
| "epoch": 218.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1171, |
| "step": 657 |
| }, |
| { |
| "epoch": 219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1163, |
| "step": 660 |
| }, |
| { |
| "epoch": 220.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1146, |
| "step": 663 |
| }, |
| { |
| "epoch": 221.88, |
| "learning_rate": 0.0001, |
| "loss": 0.116, |
| "step": 666 |
| }, |
| { |
| "epoch": 222.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1142, |
| "step": 669 |
| }, |
| { |
| "epoch": 223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1146, |
| "step": 672 |
| }, |
| { |
| "epoch": 224.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1111, |
| "step": 675 |
| }, |
| { |
| "epoch": 225.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1132, |
| "step": 678 |
| }, |
| { |
| "epoch": 226.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1137, |
| "step": 681 |
| }, |
| { |
| "epoch": 227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1102, |
| "step": 684 |
| }, |
| { |
| "epoch": 228.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1103, |
| "step": 687 |
| }, |
| { |
| "epoch": 229.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1122, |
| "step": 690 |
| }, |
| { |
| "epoch": 230.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1091, |
| "step": 693 |
| }, |
| { |
| "epoch": 231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1067, |
| "step": 696 |
| }, |
| { |
| "epoch": 232.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1075, |
| "step": 699 |
| }, |
| { |
| "epoch": 233.88, |
| "learning_rate": 0.0001, |
| "loss": 0.107, |
| "step": 702 |
| }, |
| { |
| "epoch": 234.59, |
| "eval_exact_match": 0.437137330754352, |
| "eval_exec": 0.44197292069632493, |
| "eval_loss": 0.3356037437915802, |
| "eval_runtime": 105.2625, |
| "eval_samples_per_second": 9.823, |
| "step": 704 |
| }, |
| { |
| "epoch": 234.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1088, |
| "step": 705 |
| }, |
| { |
| "epoch": 235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1054, |
| "step": 708 |
| }, |
| { |
| "epoch": 236.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1039, |
| "step": 711 |
| }, |
| { |
| "epoch": 237.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1053, |
| "step": 714 |
| }, |
| { |
| "epoch": 238.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1028, |
| "step": 717 |
| }, |
| { |
| "epoch": 239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1064, |
| "step": 720 |
| }, |
| { |
| "epoch": 240.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1031, |
| "step": 723 |
| }, |
| { |
| "epoch": 241.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1019, |
| "step": 726 |
| }, |
| { |
| "epoch": 242.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1034, |
| "step": 729 |
| }, |
| { |
| "epoch": 243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1021, |
| "step": 732 |
| }, |
| { |
| "epoch": 244.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1026, |
| "step": 735 |
| }, |
| { |
| "epoch": 245.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1, |
| "step": 738 |
| }, |
| { |
| "epoch": 246.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1008, |
| "step": 741 |
| }, |
| { |
| "epoch": 247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0999, |
| "step": 744 |
| }, |
| { |
| "epoch": 248.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1002, |
| "step": 747 |
| }, |
| { |
| "epoch": 249.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0995, |
| "step": 750 |
| }, |
| { |
| "epoch": 250.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0961, |
| "step": 753 |
| }, |
| { |
| "epoch": 251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0997, |
| "step": 756 |
| }, |
| { |
| "epoch": 252.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0982, |
| "step": 759 |
| }, |
| { |
| "epoch": 253.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0975, |
| "step": 762 |
| }, |
| { |
| "epoch": 254.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0972, |
| "step": 765 |
| }, |
| { |
| "epoch": 255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0972, |
| "step": 768 |
| }, |
| { |
| "epoch": 255.88, |
| "eval_exact_match": 0.45454545454545453, |
| "eval_exec": 0.4564796905222437, |
| "eval_loss": 0.3354557156562805, |
| "eval_runtime": 97.9303, |
| "eval_samples_per_second": 10.559, |
| "step": 768 |
| }, |
| { |
| "epoch": 256.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0955, |
| "step": 771 |
| }, |
| { |
| "epoch": 257.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0939, |
| "step": 774 |
| }, |
| { |
| "epoch": 258.88, |
| "learning_rate": 0.0001, |
| "loss": 0.094, |
| "step": 777 |
| }, |
| { |
| "epoch": 259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0947, |
| "step": 780 |
| }, |
| { |
| "epoch": 260.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0927, |
| "step": 783 |
| }, |
| { |
| "epoch": 261.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0918, |
| "step": 786 |
| }, |
| { |
| "epoch": 262.88, |
| "learning_rate": 0.0001, |
| "loss": 0.09, |
| "step": 789 |
| }, |
| { |
| "epoch": 263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.09, |
| "step": 792 |
| }, |
| { |
| "epoch": 264.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0915, |
| "step": 795 |
| }, |
| { |
| "epoch": 265.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0888, |
| "step": 798 |
| }, |
| { |
| "epoch": 266.88, |
| "learning_rate": 0.0001, |
| "loss": 0.09, |
| "step": 801 |
| }, |
| { |
| "epoch": 267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0886, |
| "step": 804 |
| }, |
| { |
| "epoch": 268.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0913, |
| "step": 807 |
| }, |
| { |
| "epoch": 269.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0889, |
| "step": 810 |
| }, |
| { |
| "epoch": 270.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0876, |
| "step": 813 |
| }, |
| { |
| "epoch": 271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0869, |
| "step": 816 |
| }, |
| { |
| "epoch": 272.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0881, |
| "step": 819 |
| }, |
| { |
| "epoch": 273.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0868, |
| "step": 822 |
| }, |
| { |
| "epoch": 274.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0842, |
| "step": 825 |
| }, |
| { |
| "epoch": 275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0858, |
| "step": 828 |
| }, |
| { |
| "epoch": 276.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0857, |
| "step": 831 |
| }, |
| { |
| "epoch": 277.29, |
| "eval_exact_match": 0.4274661508704062, |
| "eval_exec": 0.43036750483558994, |
| "eval_loss": 0.3401657044887543, |
| "eval_runtime": 115.2041, |
| "eval_samples_per_second": 8.975, |
| "step": 832 |
| }, |
| { |
| "epoch": 277.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0847, |
| "step": 834 |
| }, |
| { |
| "epoch": 278.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0854, |
| "step": 837 |
| }, |
| { |
| "epoch": 279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0852, |
| "step": 840 |
| }, |
| { |
| "epoch": 280.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0843, |
| "step": 843 |
| }, |
| { |
| "epoch": 281.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0826, |
| "step": 846 |
| }, |
| { |
| "epoch": 282.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0833, |
| "step": 849 |
| }, |
| { |
| "epoch": 283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0824, |
| "step": 852 |
| }, |
| { |
| "epoch": 284.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0822, |
| "step": 855 |
| }, |
| { |
| "epoch": 285.88, |
| "learning_rate": 0.0001, |
| "loss": 0.081, |
| "step": 858 |
| }, |
| { |
| "epoch": 286.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0826, |
| "step": 861 |
| }, |
| { |
| "epoch": 287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0801, |
| "step": 864 |
| }, |
| { |
| "epoch": 288.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0806, |
| "step": 867 |
| }, |
| { |
| "epoch": 289.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0804, |
| "step": 870 |
| }, |
| { |
| "epoch": 290.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0824, |
| "step": 873 |
| }, |
| { |
| "epoch": 291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0806, |
| "step": 876 |
| }, |
| { |
| "epoch": 292.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0791, |
| "step": 879 |
| }, |
| { |
| "epoch": 293.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0803, |
| "step": 882 |
| }, |
| { |
| "epoch": 294.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0782, |
| "step": 885 |
| }, |
| { |
| "epoch": 295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0803, |
| "step": 888 |
| }, |
| { |
| "epoch": 296.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0764, |
| "step": 891 |
| }, |
| { |
| "epoch": 297.88, |
| "learning_rate": 0.0001, |
| "loss": 0.077, |
| "step": 894 |
| }, |
| { |
| "epoch": 298.59, |
| "eval_exact_match": 0.45938104448742745, |
| "eval_exec": 0.45357833655706, |
| "eval_loss": 0.34239432215690613, |
| "eval_runtime": 103.8684, |
| "eval_samples_per_second": 9.955, |
| "step": 896 |
| }, |
| { |
| "epoch": 298.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0766, |
| "step": 897 |
| }, |
| { |
| "epoch": 299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0773, |
| "step": 900 |
| }, |
| { |
| "epoch": 300.88, |
| "learning_rate": 0.0001, |
| "loss": 0.076, |
| "step": 903 |
| }, |
| { |
| "epoch": 301.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0757, |
| "step": 906 |
| }, |
| { |
| "epoch": 302.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0753, |
| "step": 909 |
| }, |
| { |
| "epoch": 303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0753, |
| "step": 912 |
| }, |
| { |
| "epoch": 304.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0752, |
| "step": 915 |
| }, |
| { |
| "epoch": 305.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0739, |
| "step": 918 |
| }, |
| { |
| "epoch": 306.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0748, |
| "step": 921 |
| }, |
| { |
| "epoch": 307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.074, |
| "step": 924 |
| }, |
| { |
| "epoch": 308.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0736, |
| "step": 927 |
| }, |
| { |
| "epoch": 309.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0734, |
| "step": 930 |
| }, |
| { |
| "epoch": 310.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0732, |
| "step": 933 |
| }, |
| { |
| "epoch": 311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0732, |
| "step": 936 |
| }, |
| { |
| "epoch": 312.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0729, |
| "step": 939 |
| }, |
| { |
| "epoch": 313.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0708, |
| "step": 942 |
| }, |
| { |
| "epoch": 314.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0714, |
| "step": 945 |
| }, |
| { |
| "epoch": 315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0711, |
| "step": 948 |
| }, |
| { |
| "epoch": 316.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0722, |
| "step": 951 |
| }, |
| { |
| "epoch": 317.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0705, |
| "step": 954 |
| }, |
| { |
| "epoch": 318.88, |
| "learning_rate": 0.0001, |
| "loss": 0.071, |
| "step": 957 |
| }, |
| { |
| "epoch": 319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0704, |
| "step": 960 |
| }, |
| { |
| "epoch": 319.88, |
| "eval_exact_match": 0.45551257253384914, |
| "eval_exec": 0.4497098646034816, |
| "eval_loss": 0.34637653827667236, |
| "eval_runtime": 103.6764, |
| "eval_samples_per_second": 9.973, |
| "step": 960 |
| }, |
| { |
| "epoch": 320.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0687, |
| "step": 963 |
| }, |
| { |
| "epoch": 321.88, |
| "learning_rate": 0.0001, |
| "loss": 0.069, |
| "step": 966 |
| }, |
| { |
| "epoch": 322.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0694, |
| "step": 969 |
| }, |
| { |
| "epoch": 323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0692, |
| "step": 972 |
| }, |
| { |
| "epoch": 324.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0688, |
| "step": 975 |
| }, |
| { |
| "epoch": 325.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0694, |
| "step": 978 |
| }, |
| { |
| "epoch": 326.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0681, |
| "step": 981 |
| }, |
| { |
| "epoch": 327.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0676, |
| "step": 984 |
| }, |
| { |
| "epoch": 328.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0663, |
| "step": 987 |
| }, |
| { |
| "epoch": 329.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0665, |
| "step": 990 |
| }, |
| { |
| "epoch": 330.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0672, |
| "step": 993 |
| }, |
| { |
| "epoch": 331.88, |
| "learning_rate": 0.0001, |
| "loss": 0.066, |
| "step": 996 |
| }, |
| { |
| "epoch": 332.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0651, |
| "step": 999 |
| }, |
| { |
| "epoch": 333.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0659, |
| "step": 1002 |
| }, |
| { |
| "epoch": 334.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0654, |
| "step": 1005 |
| }, |
| { |
| "epoch": 335.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0641, |
| "step": 1008 |
| }, |
| { |
| "epoch": 336.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0648, |
| "step": 1011 |
| }, |
| { |
| "epoch": 337.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0651, |
| "step": 1014 |
| }, |
| { |
| "epoch": 338.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0653, |
| "step": 1017 |
| }, |
| { |
| "epoch": 339.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0645, |
| "step": 1020 |
| }, |
| { |
| "epoch": 340.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0641, |
| "step": 1023 |
| }, |
| { |
| "epoch": 341.29, |
| "eval_exact_match": 0.46034816247582205, |
| "eval_exec": 0.45164410058027077, |
| "eval_loss": 0.35341936349868774, |
| "eval_runtime": 117.5098, |
| "eval_samples_per_second": 8.799, |
| "step": 1024 |
| }, |
| { |
| "epoch": 341.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0634, |
| "step": 1026 |
| }, |
| { |
| "epoch": 342.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0617, |
| "step": 1029 |
| }, |
| { |
| "epoch": 343.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0625, |
| "step": 1032 |
| }, |
| { |
| "epoch": 344.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0623, |
| "step": 1035 |
| }, |
| { |
| "epoch": 345.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0614, |
| "step": 1038 |
| }, |
| { |
| "epoch": 346.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0616, |
| "step": 1041 |
| }, |
| { |
| "epoch": 347.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0613, |
| "step": 1044 |
| }, |
| { |
| "epoch": 348.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0615, |
| "step": 1047 |
| }, |
| { |
| "epoch": 349.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0609, |
| "step": 1050 |
| }, |
| { |
| "epoch": 350.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0609, |
| "step": 1053 |
| }, |
| { |
| "epoch": 351.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0606, |
| "step": 1056 |
| }, |
| { |
| "epoch": 352.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0599, |
| "step": 1059 |
| }, |
| { |
| "epoch": 353.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0604, |
| "step": 1062 |
| }, |
| { |
| "epoch": 354.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0596, |
| "step": 1065 |
| }, |
| { |
| "epoch": 355.88, |
| "learning_rate": 0.0001, |
| "loss": 0.06, |
| "step": 1068 |
| }, |
| { |
| "epoch": 356.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0589, |
| "step": 1071 |
| }, |
| { |
| "epoch": 357.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0606, |
| "step": 1074 |
| }, |
| { |
| "epoch": 358.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0591, |
| "step": 1077 |
| }, |
| { |
| "epoch": 359.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0566, |
| "step": 1080 |
| }, |
| { |
| "epoch": 360.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0572, |
| "step": 1083 |
| }, |
| { |
| "epoch": 361.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0575, |
| "step": 1086 |
| }, |
| { |
| "epoch": 362.59, |
| "eval_exact_match": 0.45164410058027077, |
| "eval_exec": 0.4458413926499033, |
| "eval_loss": 0.35678333044052124, |
| "eval_runtime": 127.3225, |
| "eval_samples_per_second": 8.121, |
| "step": 1088 |
| }, |
| { |
| "epoch": 362.88, |
| "learning_rate": 0.0001, |
| "loss": 0.057, |
| "step": 1089 |
| }, |
| { |
| "epoch": 363.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0571, |
| "step": 1092 |
| }, |
| { |
| "epoch": 364.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0568, |
| "step": 1095 |
| }, |
| { |
| "epoch": 365.88, |
| "learning_rate": 0.0001, |
| "loss": 0.057, |
| "step": 1098 |
| }, |
| { |
| "epoch": 366.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0561, |
| "step": 1101 |
| }, |
| { |
| "epoch": 367.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0555, |
| "step": 1104 |
| }, |
| { |
| "epoch": 368.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0557, |
| "step": 1107 |
| }, |
| { |
| "epoch": 369.88, |
| "learning_rate": 0.0001, |
| "loss": 0.055, |
| "step": 1110 |
| }, |
| { |
| "epoch": 370.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0556, |
| "step": 1113 |
| }, |
| { |
| "epoch": 371.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0554, |
| "step": 1116 |
| }, |
| { |
| "epoch": 372.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0566, |
| "step": 1119 |
| }, |
| { |
| "epoch": 373.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0557, |
| "step": 1122 |
| }, |
| { |
| "epoch": 374.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0552, |
| "step": 1125 |
| }, |
| { |
| "epoch": 375.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0546, |
| "step": 1128 |
| }, |
| { |
| "epoch": 376.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0545, |
| "step": 1131 |
| }, |
| { |
| "epoch": 377.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0545, |
| "step": 1134 |
| }, |
| { |
| "epoch": 378.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0527, |
| "step": 1137 |
| }, |
| { |
| "epoch": 379.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0536, |
| "step": 1140 |
| }, |
| { |
| "epoch": 380.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0536, |
| "step": 1143 |
| }, |
| { |
| "epoch": 381.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0536, |
| "step": 1146 |
| }, |
| { |
| "epoch": 382.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0532, |
| "step": 1149 |
| }, |
| { |
| "epoch": 383.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0521, |
| "step": 1152 |
| }, |
| { |
| "epoch": 383.88, |
| "eval_exact_match": 0.47292069632495165, |
| "eval_exec": 0.4584139264990329, |
| "eval_loss": 0.3651985228061676, |
| "eval_runtime": 107.8989, |
| "eval_samples_per_second": 9.583, |
| "step": 1152 |
| }, |
| { |
| "epoch": 384.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0518, |
| "step": 1155 |
| }, |
| { |
| "epoch": 385.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0523, |
| "step": 1158 |
| }, |
| { |
| "epoch": 386.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0509, |
| "step": 1161 |
| }, |
| { |
| "epoch": 387.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0525, |
| "step": 1164 |
| }, |
| { |
| "epoch": 388.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0517, |
| "step": 1167 |
| }, |
| { |
| "epoch": 389.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0507, |
| "step": 1170 |
| }, |
| { |
| "epoch": 390.88, |
| "learning_rate": 0.0001, |
| "loss": 0.05, |
| "step": 1173 |
| }, |
| { |
| "epoch": 391.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0506, |
| "step": 1176 |
| }, |
| { |
| "epoch": 392.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0505, |
| "step": 1179 |
| }, |
| { |
| "epoch": 393.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0504, |
| "step": 1182 |
| }, |
| { |
| "epoch": 394.88, |
| "learning_rate": 0.0001, |
| "loss": 0.05, |
| "step": 1185 |
| }, |
| { |
| "epoch": 395.88, |
| "learning_rate": 0.0001, |
| "loss": 0.05, |
| "step": 1188 |
| }, |
| { |
| "epoch": 396.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0497, |
| "step": 1191 |
| }, |
| { |
| "epoch": 397.88, |
| "learning_rate": 0.0001, |
| "loss": 0.049, |
| "step": 1194 |
| }, |
| { |
| "epoch": 398.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0496, |
| "step": 1197 |
| }, |
| { |
| "epoch": 399.88, |
| "learning_rate": 0.0001, |
| "loss": 0.049, |
| "step": 1200 |
| }, |
| { |
| "epoch": 400.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0487, |
| "step": 1203 |
| }, |
| { |
| "epoch": 401.88, |
| "learning_rate": 0.0001, |
| "loss": 0.049, |
| "step": 1206 |
| }, |
| { |
| "epoch": 402.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0486, |
| "step": 1209 |
| }, |
| { |
| "epoch": 403.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0484, |
| "step": 1212 |
| }, |
| { |
| "epoch": 404.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0493, |
| "step": 1215 |
| }, |
| { |
| "epoch": 405.29, |
| "eval_exact_match": 0.46808510638297873, |
| "eval_exec": 0.4632495164410058, |
| "eval_loss": 0.36478203535079956, |
| "eval_runtime": 110.1876, |
| "eval_samples_per_second": 9.384, |
| "step": 1216 |
| }, |
| { |
| "epoch": 405.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0477, |
| "step": 1218 |
| }, |
| { |
| "epoch": 406.88, |
| "learning_rate": 0.0001, |
| "loss": 0.047, |
| "step": 1221 |
| }, |
| { |
| "epoch": 407.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0476, |
| "step": 1224 |
| }, |
| { |
| "epoch": 408.88, |
| "learning_rate": 0.0001, |
| "loss": 0.047, |
| "step": 1227 |
| }, |
| { |
| "epoch": 409.88, |
| "learning_rate": 0.0001, |
| "loss": 0.048, |
| "step": 1230 |
| }, |
| { |
| "epoch": 410.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0462, |
| "step": 1233 |
| }, |
| { |
| "epoch": 411.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0468, |
| "step": 1236 |
| }, |
| { |
| "epoch": 412.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0455, |
| "step": 1239 |
| }, |
| { |
| "epoch": 413.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0468, |
| "step": 1242 |
| }, |
| { |
| "epoch": 414.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0458, |
| "step": 1245 |
| }, |
| { |
| "epoch": 415.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0461, |
| "step": 1248 |
| }, |
| { |
| "epoch": 416.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0462, |
| "step": 1251 |
| }, |
| { |
| "epoch": 417.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0447, |
| "step": 1254 |
| }, |
| { |
| "epoch": 418.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0452, |
| "step": 1257 |
| }, |
| { |
| "epoch": 419.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0439, |
| "step": 1260 |
| }, |
| { |
| "epoch": 420.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0444, |
| "step": 1263 |
| }, |
| { |
| "epoch": 421.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0447, |
| "step": 1266 |
| }, |
| { |
| "epoch": 422.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0447, |
| "step": 1269 |
| }, |
| { |
| "epoch": 423.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0448, |
| "step": 1272 |
| }, |
| { |
| "epoch": 424.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0443, |
| "step": 1275 |
| }, |
| { |
| "epoch": 425.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0437, |
| "step": 1278 |
| }, |
| { |
| "epoch": 426.59, |
| "eval_exact_match": 0.47195357833655704, |
| "eval_exec": 0.4632495164410058, |
| "eval_loss": 0.37120428681373596, |
| "eval_runtime": 101.6241, |
| "eval_samples_per_second": 10.175, |
| "step": 1280 |
| }, |
| { |
| "epoch": 426.88, |
| "learning_rate": 0.0001, |
| "loss": 0.042, |
| "step": 1281 |
| }, |
| { |
| "epoch": 427.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0444, |
| "step": 1284 |
| }, |
| { |
| "epoch": 428.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0445, |
| "step": 1287 |
| }, |
| { |
| "epoch": 429.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0427, |
| "step": 1290 |
| }, |
| { |
| "epoch": 430.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0423, |
| "step": 1293 |
| }, |
| { |
| "epoch": 431.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0427, |
| "step": 1296 |
| }, |
| { |
| "epoch": 432.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0426, |
| "step": 1299 |
| }, |
| { |
| "epoch": 433.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0427, |
| "step": 1302 |
| }, |
| { |
| "epoch": 434.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0418, |
| "step": 1305 |
| }, |
| { |
| "epoch": 435.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0422, |
| "step": 1308 |
| }, |
| { |
| "epoch": 436.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0406, |
| "step": 1311 |
| }, |
| { |
| "epoch": 437.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0413, |
| "step": 1314 |
| }, |
| { |
| "epoch": 438.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0403, |
| "step": 1317 |
| }, |
| { |
| "epoch": 439.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0417, |
| "step": 1320 |
| }, |
| { |
| "epoch": 440.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0407, |
| "step": 1323 |
| }, |
| { |
| "epoch": 441.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0416, |
| "step": 1326 |
| }, |
| { |
| "epoch": 442.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0414, |
| "step": 1329 |
| }, |
| { |
| "epoch": 443.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0407, |
| "step": 1332 |
| }, |
| { |
| "epoch": 444.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0407, |
| "step": 1335 |
| }, |
| { |
| "epoch": 445.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0404, |
| "step": 1338 |
| }, |
| { |
| "epoch": 446.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0401, |
| "step": 1341 |
| }, |
| { |
| "epoch": 447.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0399, |
| "step": 1344 |
| }, |
| { |
| "epoch": 447.88, |
| "eval_exact_match": 0.47195357833655704, |
| "eval_exec": 0.4632495164410058, |
| "eval_loss": 0.3796737492084503, |
| "eval_runtime": 116.7328, |
| "eval_samples_per_second": 8.858, |
| "step": 1344 |
| }, |
| { |
| "epoch": 448.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0403, |
| "step": 1347 |
| }, |
| { |
| "epoch": 449.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0397, |
| "step": 1350 |
| }, |
| { |
| "epoch": 450.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0392, |
| "step": 1353 |
| }, |
| { |
| "epoch": 451.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0391, |
| "step": 1356 |
| }, |
| { |
| "epoch": 452.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0405, |
| "step": 1359 |
| }, |
| { |
| "epoch": 453.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0395, |
| "step": 1362 |
| }, |
| { |
| "epoch": 454.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0394, |
| "step": 1365 |
| }, |
| { |
| "epoch": 455.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0386, |
| "step": 1368 |
| }, |
| { |
| "epoch": 456.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0381, |
| "step": 1371 |
| }, |
| { |
| "epoch": 457.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0384, |
| "step": 1374 |
| }, |
| { |
| "epoch": 458.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0382, |
| "step": 1377 |
| }, |
| { |
| "epoch": 459.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0379, |
| "step": 1380 |
| }, |
| { |
| "epoch": 460.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0381, |
| "step": 1383 |
| }, |
| { |
| "epoch": 461.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0377, |
| "step": 1386 |
| }, |
| { |
| "epoch": 462.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0377, |
| "step": 1389 |
| }, |
| { |
| "epoch": 463.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0374, |
| "step": 1392 |
| }, |
| { |
| "epoch": 464.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0369, |
| "step": 1395 |
| }, |
| { |
| "epoch": 465.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0372, |
| "step": 1398 |
| }, |
| { |
| "epoch": 466.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0386, |
| "step": 1401 |
| }, |
| { |
| "epoch": 467.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0367, |
| "step": 1404 |
| }, |
| { |
| "epoch": 468.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0365, |
| "step": 1407 |
| }, |
| { |
| "epoch": 469.29, |
| "eval_exact_match": 0.47195357833655704, |
| "eval_exec": 0.4622823984526112, |
| "eval_loss": 0.38305917382240295, |
| "eval_runtime": 123.8052, |
| "eval_samples_per_second": 8.352, |
| "step": 1408 |
| }, |
| { |
| "epoch": 469.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0363, |
| "step": 1410 |
| }, |
| { |
| "epoch": 470.88, |
| "learning_rate": 0.0001, |
| "loss": 0.037, |
| "step": 1413 |
| }, |
| { |
| "epoch": 471.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0367, |
| "step": 1416 |
| }, |
| { |
| "epoch": 472.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0367, |
| "step": 1419 |
| }, |
| { |
| "epoch": 473.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0362, |
| "step": 1422 |
| }, |
| { |
| "epoch": 474.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0359, |
| "step": 1425 |
| }, |
| { |
| "epoch": 475.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0352, |
| "step": 1428 |
| }, |
| { |
| "epoch": 476.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0355, |
| "step": 1431 |
| }, |
| { |
| "epoch": 477.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0363, |
| "step": 1434 |
| }, |
| { |
| "epoch": 478.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0342, |
| "step": 1437 |
| }, |
| { |
| "epoch": 479.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0353, |
| "step": 1440 |
| }, |
| { |
| "epoch": 480.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0344, |
| "step": 1443 |
| }, |
| { |
| "epoch": 481.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0344, |
| "step": 1446 |
| }, |
| { |
| "epoch": 482.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0342, |
| "step": 1449 |
| }, |
| { |
| "epoch": 483.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0361, |
| "step": 1452 |
| }, |
| { |
| "epoch": 484.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0343, |
| "step": 1455 |
| }, |
| { |
| "epoch": 485.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0341, |
| "step": 1458 |
| }, |
| { |
| "epoch": 486.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0332, |
| "step": 1461 |
| }, |
| { |
| "epoch": 487.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0338, |
| "step": 1464 |
| }, |
| { |
| "epoch": 488.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0344, |
| "step": 1467 |
| }, |
| { |
| "epoch": 489.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0342, |
| "step": 1470 |
| }, |
| { |
| "epoch": 490.59, |
| "eval_exact_match": 0.47292069632495165, |
| "eval_exec": 0.46421663442940037, |
| "eval_loss": 0.3914486765861511, |
| "eval_runtime": 103.8318, |
| "eval_samples_per_second": 9.958, |
| "step": 1472 |
| }, |
| { |
| "epoch": 490.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0325, |
| "step": 1473 |
| }, |
| { |
| "epoch": 491.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0348, |
| "step": 1476 |
| }, |
| { |
| "epoch": 492.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0336, |
| "step": 1479 |
| }, |
| { |
| "epoch": 493.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0337, |
| "step": 1482 |
| }, |
| { |
| "epoch": 494.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0335, |
| "step": 1485 |
| }, |
| { |
| "epoch": 495.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0332, |
| "step": 1488 |
| }, |
| { |
| "epoch": 496.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0331, |
| "step": 1491 |
| }, |
| { |
| "epoch": 497.88, |
| "learning_rate": 0.0001, |
| "loss": 0.032, |
| "step": 1494 |
| }, |
| { |
| "epoch": 498.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0339, |
| "step": 1497 |
| }, |
| { |
| "epoch": 499.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0326, |
| "step": 1500 |
| }, |
| { |
| "epoch": 500.88, |
| "learning_rate": 0.0001, |
| "loss": 0.032, |
| "step": 1503 |
| }, |
| { |
| "epoch": 501.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0324, |
| "step": 1506 |
| }, |
| { |
| "epoch": 502.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0324, |
| "step": 1509 |
| }, |
| { |
| "epoch": 503.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0326, |
| "step": 1512 |
| }, |
| { |
| "epoch": 504.88, |
| "learning_rate": 0.0001, |
| "loss": 0.032, |
| "step": 1515 |
| }, |
| { |
| "epoch": 505.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0307, |
| "step": 1518 |
| }, |
| { |
| "epoch": 506.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0311, |
| "step": 1521 |
| }, |
| { |
| "epoch": 507.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0313, |
| "step": 1524 |
| }, |
| { |
| "epoch": 508.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0314, |
| "step": 1527 |
| }, |
| { |
| "epoch": 509.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0306, |
| "step": 1530 |
| }, |
| { |
| "epoch": 510.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0305, |
| "step": 1533 |
| }, |
| { |
| "epoch": 511.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0319, |
| "step": 1536 |
| }, |
| { |
| "epoch": 511.88, |
| "eval_exact_match": 0.46808510638297873, |
| "eval_exec": 0.45938104448742745, |
| "eval_loss": 0.3986540734767914, |
| "eval_runtime": 109.5821, |
| "eval_samples_per_second": 9.436, |
| "step": 1536 |
| }, |
| { |
| "epoch": 512.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0309, |
| "step": 1539 |
| }, |
| { |
| "epoch": 513.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0316, |
| "step": 1542 |
| }, |
| { |
| "epoch": 514.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0298, |
| "step": 1545 |
| }, |
| { |
| "epoch": 515.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0302, |
| "step": 1548 |
| }, |
| { |
| "epoch": 516.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0289, |
| "step": 1551 |
| }, |
| { |
| "epoch": 517.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0302, |
| "step": 1554 |
| }, |
| { |
| "epoch": 518.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0299, |
| "step": 1557 |
| }, |
| { |
| "epoch": 519.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0298, |
| "step": 1560 |
| }, |
| { |
| "epoch": 520.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0297, |
| "step": 1563 |
| }, |
| { |
| "epoch": 521.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0292, |
| "step": 1566 |
| }, |
| { |
| "epoch": 522.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0294, |
| "step": 1569 |
| }, |
| { |
| "epoch": 523.88, |
| "learning_rate": 0.0001, |
| "loss": 0.03, |
| "step": 1572 |
| }, |
| { |
| "epoch": 524.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0293, |
| "step": 1575 |
| }, |
| { |
| "epoch": 525.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0291, |
| "step": 1578 |
| }, |
| { |
| "epoch": 526.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0286, |
| "step": 1581 |
| }, |
| { |
| "epoch": 527.88, |
| "learning_rate": 0.0001, |
| "loss": 0.029, |
| "step": 1584 |
| }, |
| { |
| "epoch": 528.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0294, |
| "step": 1587 |
| }, |
| { |
| "epoch": 529.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0298, |
| "step": 1590 |
| }, |
| { |
| "epoch": 530.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0286, |
| "step": 1593 |
| }, |
| { |
| "epoch": 531.88, |
| "learning_rate": 0.0001, |
| "loss": 0.028, |
| "step": 1596 |
| }, |
| { |
| "epoch": 532.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0283, |
| "step": 1599 |
| }, |
| { |
| "epoch": 533.29, |
| "eval_exact_match": 0.46421663442940037, |
| "eval_exec": 0.45938104448742745, |
| "eval_loss": 0.4072656035423279, |
| "eval_runtime": 128.9566, |
| "eval_samples_per_second": 8.018, |
| "step": 1600 |
| }, |
| { |
| "epoch": 533.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0292, |
| "step": 1602 |
| }, |
| { |
| "epoch": 534.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0287, |
| "step": 1605 |
| }, |
| { |
| "epoch": 535.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0294, |
| "step": 1608 |
| }, |
| { |
| "epoch": 536.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0291, |
| "step": 1611 |
| }, |
| { |
| "epoch": 537.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0282, |
| "step": 1614 |
| }, |
| { |
| "epoch": 538.88, |
| "learning_rate": 0.0001, |
| "loss": 0.028, |
| "step": 1617 |
| }, |
| { |
| "epoch": 539.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0289, |
| "step": 1620 |
| }, |
| { |
| "epoch": 540.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0272, |
| "step": 1623 |
| }, |
| { |
| "epoch": 541.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0274, |
| "step": 1626 |
| }, |
| { |
| "epoch": 542.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0275, |
| "step": 1629 |
| }, |
| { |
| "epoch": 543.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0278, |
| "step": 1632 |
| }, |
| { |
| "epoch": 544.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0278, |
| "step": 1635 |
| }, |
| { |
| "epoch": 545.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0279, |
| "step": 1638 |
| }, |
| { |
| "epoch": 546.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0272, |
| "step": 1641 |
| }, |
| { |
| "epoch": 547.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0275, |
| "step": 1644 |
| }, |
| { |
| "epoch": 548.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0262, |
| "step": 1647 |
| }, |
| { |
| "epoch": 549.88, |
| "learning_rate": 0.0001, |
| "loss": 0.027, |
| "step": 1650 |
| }, |
| { |
| "epoch": 550.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0266, |
| "step": 1653 |
| }, |
| { |
| "epoch": 551.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0265, |
| "step": 1656 |
| }, |
| { |
| "epoch": 552.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0272, |
| "step": 1659 |
| }, |
| { |
| "epoch": 553.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0265, |
| "step": 1662 |
| }, |
| { |
| "epoch": 554.59, |
| "eval_exact_match": 0.488394584139265, |
| "eval_exec": 0.4816247582205029, |
| "eval_loss": 0.40782487392425537, |
| "eval_runtime": 108.3403, |
| "eval_samples_per_second": 9.544, |
| "step": 1664 |
| }, |
| { |
| "epoch": 554.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0262, |
| "step": 1665 |
| }, |
| { |
| "epoch": 555.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0276, |
| "step": 1668 |
| }, |
| { |
| "epoch": 556.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0264, |
| "step": 1671 |
| }, |
| { |
| "epoch": 557.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0256, |
| "step": 1674 |
| }, |
| { |
| "epoch": 558.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0255, |
| "step": 1677 |
| }, |
| { |
| "epoch": 559.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0259, |
| "step": 1680 |
| }, |
| { |
| "epoch": 560.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0246, |
| "step": 1683 |
| }, |
| { |
| "epoch": 561.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0257, |
| "step": 1686 |
| }, |
| { |
| "epoch": 562.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0267, |
| "step": 1689 |
| }, |
| { |
| "epoch": 563.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0251, |
| "step": 1692 |
| }, |
| { |
| "epoch": 564.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0253, |
| "step": 1695 |
| }, |
| { |
| "epoch": 565.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0248, |
| "step": 1698 |
| }, |
| { |
| "epoch": 566.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0252, |
| "step": 1701 |
| }, |
| { |
| "epoch": 567.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0255, |
| "step": 1704 |
| }, |
| { |
| "epoch": 568.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0249, |
| "step": 1707 |
| }, |
| { |
| "epoch": 569.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0254, |
| "step": 1710 |
| }, |
| { |
| "epoch": 570.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0251, |
| "step": 1713 |
| }, |
| { |
| "epoch": 571.88, |
| "learning_rate": 0.0001, |
| "loss": 0.024, |
| "step": 1716 |
| }, |
| { |
| "epoch": 572.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0239, |
| "step": 1719 |
| }, |
| { |
| "epoch": 573.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0247, |
| "step": 1722 |
| }, |
| { |
| "epoch": 574.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0246, |
| "step": 1725 |
| }, |
| { |
| "epoch": 575.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0249, |
| "step": 1728 |
| }, |
| { |
| "epoch": 575.88, |
| "eval_exact_match": 0.4738878143133462, |
| "eval_exec": 0.4690522243713733, |
| "eval_loss": 0.413275808095932, |
| "eval_runtime": 120.0672, |
| "eval_samples_per_second": 8.612, |
| "step": 1728 |
| }, |
| { |
| "epoch": 576.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0246, |
| "step": 1731 |
| }, |
| { |
| "epoch": 577.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0235, |
| "step": 1734 |
| }, |
| { |
| "epoch": 578.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0239, |
| "step": 1737 |
| }, |
| { |
| "epoch": 579.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0234, |
| "step": 1740 |
| }, |
| { |
| "epoch": 580.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0231, |
| "step": 1743 |
| }, |
| { |
| "epoch": 581.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0233, |
| "step": 1746 |
| }, |
| { |
| "epoch": 582.88, |
| "learning_rate": 0.0001, |
| "loss": 0.023, |
| "step": 1749 |
| }, |
| { |
| "epoch": 583.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0235, |
| "step": 1752 |
| }, |
| { |
| "epoch": 584.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0238, |
| "step": 1755 |
| }, |
| { |
| "epoch": 585.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0254, |
| "step": 1758 |
| }, |
| { |
| "epoch": 586.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0243, |
| "step": 1761 |
| }, |
| { |
| "epoch": 587.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0236, |
| "step": 1764 |
| }, |
| { |
| "epoch": 588.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0222, |
| "step": 1767 |
| }, |
| { |
| "epoch": 589.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0228, |
| "step": 1770 |
| }, |
| { |
| "epoch": 590.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0236, |
| "step": 1773 |
| }, |
| { |
| "epoch": 591.88, |
| "learning_rate": 0.0001, |
| "loss": 0.023, |
| "step": 1776 |
| }, |
| { |
| "epoch": 592.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0233, |
| "step": 1779 |
| }, |
| { |
| "epoch": 593.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0229, |
| "step": 1782 |
| }, |
| { |
| "epoch": 594.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0226, |
| "step": 1785 |
| }, |
| { |
| "epoch": 595.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0221, |
| "step": 1788 |
| }, |
| { |
| "epoch": 596.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0228, |
| "step": 1791 |
| }, |
| { |
| "epoch": 597.29, |
| "eval_exact_match": 0.4671179883945841, |
| "eval_exec": 0.4622823984526112, |
| "eval_loss": 0.421924889087677, |
| "eval_runtime": 121.4411, |
| "eval_samples_per_second": 8.514, |
| "step": 1792 |
| }, |
| { |
| "epoch": 597.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0225, |
| "step": 1794 |
| }, |
| { |
| "epoch": 598.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0221, |
| "step": 1797 |
| }, |
| { |
| "epoch": 599.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0216, |
| "step": 1800 |
| }, |
| { |
| "epoch": 600.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0219, |
| "step": 1803 |
| }, |
| { |
| "epoch": 601.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0214, |
| "step": 1806 |
| }, |
| { |
| "epoch": 602.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0219, |
| "step": 1809 |
| }, |
| { |
| "epoch": 603.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0216, |
| "step": 1812 |
| }, |
| { |
| "epoch": 604.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0213, |
| "step": 1815 |
| }, |
| { |
| "epoch": 605.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0212, |
| "step": 1818 |
| }, |
| { |
| "epoch": 606.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0218, |
| "step": 1821 |
| }, |
| { |
| "epoch": 607.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0218, |
| "step": 1824 |
| }, |
| { |
| "epoch": 608.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0221, |
| "step": 1827 |
| }, |
| { |
| "epoch": 609.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0217, |
| "step": 1830 |
| }, |
| { |
| "epoch": 610.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0224, |
| "step": 1833 |
| }, |
| { |
| "epoch": 611.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0208, |
| "step": 1836 |
| }, |
| { |
| "epoch": 612.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0208, |
| "step": 1839 |
| }, |
| { |
| "epoch": 613.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0205, |
| "step": 1842 |
| }, |
| { |
| "epoch": 614.88, |
| "learning_rate": 0.0001, |
| "loss": 0.021, |
| "step": 1845 |
| }, |
| { |
| "epoch": 615.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0207, |
| "step": 1848 |
| }, |
| { |
| "epoch": 616.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0206, |
| "step": 1851 |
| }, |
| { |
| "epoch": 617.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0211, |
| "step": 1854 |
| }, |
| { |
| "epoch": 618.59, |
| "eval_exact_match": 0.4941972920696325, |
| "eval_exec": 0.48936170212765956, |
| "eval_loss": 0.42522886395454407, |
| "eval_runtime": 108.5837, |
| "eval_samples_per_second": 9.523, |
| "step": 1856 |
| }, |
| { |
| "epoch": 618.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0206, |
| "step": 1857 |
| }, |
| { |
| "epoch": 619.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0209, |
| "step": 1860 |
| }, |
| { |
| "epoch": 620.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0209, |
| "step": 1863 |
| }, |
| { |
| "epoch": 621.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0209, |
| "step": 1866 |
| }, |
| { |
| "epoch": 622.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0206, |
| "step": 1869 |
| }, |
| { |
| "epoch": 623.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0195, |
| "step": 1872 |
| }, |
| { |
| "epoch": 624.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0194, |
| "step": 1875 |
| }, |
| { |
| "epoch": 625.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0201, |
| "step": 1878 |
| }, |
| { |
| "epoch": 626.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0197, |
| "step": 1881 |
| }, |
| { |
| "epoch": 627.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0194, |
| "step": 1884 |
| }, |
| { |
| "epoch": 628.88, |
| "learning_rate": 0.0001, |
| "loss": 0.02, |
| "step": 1887 |
| }, |
| { |
| "epoch": 629.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0198, |
| "step": 1890 |
| }, |
| { |
| "epoch": 630.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0196, |
| "step": 1893 |
| }, |
| { |
| "epoch": 631.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0185, |
| "step": 1896 |
| }, |
| { |
| "epoch": 632.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0203, |
| "step": 1899 |
| }, |
| { |
| "epoch": 633.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0201, |
| "step": 1902 |
| }, |
| { |
| "epoch": 634.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0197, |
| "step": 1905 |
| }, |
| { |
| "epoch": 635.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0196, |
| "step": 1908 |
| }, |
| { |
| "epoch": 636.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0192, |
| "step": 1911 |
| }, |
| { |
| "epoch": 637.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0184, |
| "step": 1914 |
| }, |
| { |
| "epoch": 638.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0188, |
| "step": 1917 |
| }, |
| { |
| "epoch": 639.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0185, |
| "step": 1920 |
| }, |
| { |
| "epoch": 639.88, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.43099677562713623, |
| "eval_runtime": 104.5413, |
| "eval_samples_per_second": 9.891, |
| "step": 1920 |
| }, |
| { |
| "epoch": 640.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0194, |
| "step": 1923 |
| }, |
| { |
| "epoch": 641.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0189, |
| "step": 1926 |
| }, |
| { |
| "epoch": 642.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0201, |
| "step": 1929 |
| }, |
| { |
| "epoch": 643.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0196, |
| "step": 1932 |
| }, |
| { |
| "epoch": 644.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0187, |
| "step": 1935 |
| }, |
| { |
| "epoch": 645.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0192, |
| "step": 1938 |
| }, |
| { |
| "epoch": 646.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0194, |
| "step": 1941 |
| }, |
| { |
| "epoch": 647.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0188, |
| "step": 1944 |
| }, |
| { |
| "epoch": 648.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0185, |
| "step": 1947 |
| }, |
| { |
| "epoch": 649.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0182, |
| "step": 1950 |
| }, |
| { |
| "epoch": 650.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0181, |
| "step": 1953 |
| }, |
| { |
| "epoch": 651.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0182, |
| "step": 1956 |
| }, |
| { |
| "epoch": 652.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0193, |
| "step": 1959 |
| }, |
| { |
| "epoch": 653.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0179, |
| "step": 1962 |
| }, |
| { |
| "epoch": 654.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0177, |
| "step": 1965 |
| }, |
| { |
| "epoch": 655.88, |
| "learning_rate": 0.0001, |
| "loss": 0.018, |
| "step": 1968 |
| }, |
| { |
| "epoch": 656.88, |
| "learning_rate": 0.0001, |
| "loss": 0.018, |
| "step": 1971 |
| }, |
| { |
| "epoch": 657.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0179, |
| "step": 1974 |
| }, |
| { |
| "epoch": 658.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0173, |
| "step": 1977 |
| }, |
| { |
| "epoch": 659.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0179, |
| "step": 1980 |
| }, |
| { |
| "epoch": 660.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0179, |
| "step": 1983 |
| }, |
| { |
| "epoch": 661.29, |
| "eval_exact_match": 0.47678916827852996, |
| "eval_exec": 0.47678916827852996, |
| "eval_loss": 0.43502098321914673, |
| "eval_runtime": 115.2904, |
| "eval_samples_per_second": 8.969, |
| "step": 1984 |
| }, |
| { |
| "epoch": 661.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0174, |
| "step": 1986 |
| }, |
| { |
| "epoch": 662.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0181, |
| "step": 1989 |
| }, |
| { |
| "epoch": 663.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0185, |
| "step": 1992 |
| }, |
| { |
| "epoch": 664.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0183, |
| "step": 1995 |
| }, |
| { |
| "epoch": 665.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0173, |
| "step": 1998 |
| }, |
| { |
| "epoch": 666.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0169, |
| "step": 2001 |
| }, |
| { |
| "epoch": 667.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0169, |
| "step": 2004 |
| }, |
| { |
| "epoch": 668.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0172, |
| "step": 2007 |
| }, |
| { |
| "epoch": 669.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0165, |
| "step": 2010 |
| }, |
| { |
| "epoch": 670.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0168, |
| "step": 2013 |
| }, |
| { |
| "epoch": 671.88, |
| "learning_rate": 0.0001, |
| "loss": 0.017, |
| "step": 2016 |
| }, |
| { |
| "epoch": 672.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0173, |
| "step": 2019 |
| }, |
| { |
| "epoch": 673.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0178, |
| "step": 2022 |
| }, |
| { |
| "epoch": 674.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0164, |
| "step": 2025 |
| }, |
| { |
| "epoch": 675.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0167, |
| "step": 2028 |
| }, |
| { |
| "epoch": 676.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0166, |
| "step": 2031 |
| }, |
| { |
| "epoch": 677.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0163, |
| "step": 2034 |
| }, |
| { |
| "epoch": 678.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0162, |
| "step": 2037 |
| }, |
| { |
| "epoch": 679.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0163, |
| "step": 2040 |
| }, |
| { |
| "epoch": 680.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0163, |
| "step": 2043 |
| }, |
| { |
| "epoch": 681.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0162, |
| "step": 2046 |
| }, |
| { |
| "epoch": 682.59, |
| "eval_exact_match": 0.4825918762088975, |
| "eval_exec": 0.4816247582205029, |
| "eval_loss": 0.4417296051979065, |
| "eval_runtime": 103.7496, |
| "eval_samples_per_second": 9.966, |
| "step": 2048 |
| }, |
| { |
| "epoch": 682.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0173, |
| "step": 2049 |
| }, |
| { |
| "epoch": 683.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0174, |
| "step": 2052 |
| }, |
| { |
| "epoch": 684.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0153, |
| "step": 2055 |
| }, |
| { |
| "epoch": 685.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0155, |
| "step": 2058 |
| }, |
| { |
| "epoch": 686.88, |
| "learning_rate": 0.0001, |
| "loss": 0.016, |
| "step": 2061 |
| }, |
| { |
| "epoch": 687.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0161, |
| "step": 2064 |
| }, |
| { |
| "epoch": 688.88, |
| "learning_rate": 0.0001, |
| "loss": 0.017, |
| "step": 2067 |
| }, |
| { |
| "epoch": 689.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0164, |
| "step": 2070 |
| }, |
| { |
| "epoch": 690.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0165, |
| "step": 2073 |
| }, |
| { |
| "epoch": 691.88, |
| "learning_rate": 0.0001, |
| "loss": 0.016, |
| "step": 2076 |
| }, |
| { |
| "epoch": 692.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0152, |
| "step": 2079 |
| }, |
| { |
| "epoch": 693.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0164, |
| "step": 2082 |
| }, |
| { |
| "epoch": 694.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0161, |
| "step": 2085 |
| }, |
| { |
| "epoch": 695.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0152, |
| "step": 2088 |
| }, |
| { |
| "epoch": 696.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0156, |
| "step": 2091 |
| }, |
| { |
| "epoch": 697.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0151, |
| "step": 2094 |
| }, |
| { |
| "epoch": 698.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0156, |
| "step": 2097 |
| }, |
| { |
| "epoch": 699.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0161, |
| "step": 2100 |
| }, |
| { |
| "epoch": 700.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0161, |
| "step": 2103 |
| }, |
| { |
| "epoch": 701.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0158, |
| "step": 2106 |
| }, |
| { |
| "epoch": 702.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0154, |
| "step": 2109 |
| }, |
| { |
| "epoch": 703.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0154, |
| "step": 2112 |
| }, |
| { |
| "epoch": 703.88, |
| "eval_exact_match": 0.4874274661508704, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.4481361210346222, |
| "eval_runtime": 110.2347, |
| "eval_samples_per_second": 9.38, |
| "step": 2112 |
| }, |
| { |
| "epoch": 704.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0154, |
| "step": 2115 |
| }, |
| { |
| "epoch": 705.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0151, |
| "step": 2118 |
| }, |
| { |
| "epoch": 706.88, |
| "learning_rate": 0.0001, |
| "loss": 0.015, |
| "step": 2121 |
| }, |
| { |
| "epoch": 707.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0152, |
| "step": 2124 |
| }, |
| { |
| "epoch": 708.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0143, |
| "step": 2127 |
| }, |
| { |
| "epoch": 709.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0151, |
| "step": 2130 |
| }, |
| { |
| "epoch": 710.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0146, |
| "step": 2133 |
| }, |
| { |
| "epoch": 711.88, |
| "learning_rate": 0.0001, |
| "loss": 0.015, |
| "step": 2136 |
| }, |
| { |
| "epoch": 712.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0149, |
| "step": 2139 |
| }, |
| { |
| "epoch": 713.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0153, |
| "step": 2142 |
| }, |
| { |
| "epoch": 714.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0147, |
| "step": 2145 |
| }, |
| { |
| "epoch": 715.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0147, |
| "step": 2148 |
| }, |
| { |
| "epoch": 716.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0149, |
| "step": 2151 |
| }, |
| { |
| "epoch": 717.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0151, |
| "step": 2154 |
| }, |
| { |
| "epoch": 718.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0145, |
| "step": 2157 |
| }, |
| { |
| "epoch": 719.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0147, |
| "step": 2160 |
| }, |
| { |
| "epoch": 720.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0145, |
| "step": 2163 |
| }, |
| { |
| "epoch": 721.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0145, |
| "step": 2166 |
| }, |
| { |
| "epoch": 722.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0142, |
| "step": 2169 |
| }, |
| { |
| "epoch": 723.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0142, |
| "step": 2172 |
| }, |
| { |
| "epoch": 724.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0141, |
| "step": 2175 |
| }, |
| { |
| "epoch": 725.29, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.4548819959163666, |
| "eval_runtime": 108.1709, |
| "eval_samples_per_second": 9.559, |
| "step": 2176 |
| }, |
| { |
| "epoch": 725.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0141, |
| "step": 2178 |
| }, |
| { |
| "epoch": 726.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0153, |
| "step": 2181 |
| }, |
| { |
| "epoch": 727.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0144, |
| "step": 2184 |
| }, |
| { |
| "epoch": 728.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0145, |
| "step": 2187 |
| }, |
| { |
| "epoch": 729.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0136, |
| "step": 2190 |
| }, |
| { |
| "epoch": 730.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0136, |
| "step": 2193 |
| }, |
| { |
| "epoch": 731.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0139, |
| "step": 2196 |
| }, |
| { |
| "epoch": 732.88, |
| "learning_rate": 0.0001, |
| "loss": 0.014, |
| "step": 2199 |
| }, |
| { |
| "epoch": 733.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 2202 |
| }, |
| { |
| "epoch": 734.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0143, |
| "step": 2205 |
| }, |
| { |
| "epoch": 735.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0138, |
| "step": 2208 |
| }, |
| { |
| "epoch": 736.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 2211 |
| }, |
| { |
| "epoch": 737.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0142, |
| "step": 2214 |
| }, |
| { |
| "epoch": 738.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 2217 |
| }, |
| { |
| "epoch": 739.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 2220 |
| }, |
| { |
| "epoch": 740.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0133, |
| "step": 2223 |
| }, |
| { |
| "epoch": 741.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 2226 |
| }, |
| { |
| "epoch": 742.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0139, |
| "step": 2229 |
| }, |
| { |
| "epoch": 743.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 2232 |
| }, |
| { |
| "epoch": 744.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 2235 |
| }, |
| { |
| "epoch": 745.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0135, |
| "step": 2238 |
| }, |
| { |
| "epoch": 746.59, |
| "eval_exact_match": 0.4835589941972921, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.4577375054359436, |
| "eval_runtime": 101.9311, |
| "eval_samples_per_second": 10.144, |
| "step": 2240 |
| }, |
| { |
| "epoch": 746.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 2241 |
| }, |
| { |
| "epoch": 747.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0139, |
| "step": 2244 |
| }, |
| { |
| "epoch": 748.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0136, |
| "step": 2247 |
| }, |
| { |
| "epoch": 749.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 2250 |
| }, |
| { |
| "epoch": 750.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0128, |
| "step": 2253 |
| }, |
| { |
| "epoch": 751.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 2256 |
| }, |
| { |
| "epoch": 752.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0132, |
| "step": 2259 |
| }, |
| { |
| "epoch": 753.88, |
| "learning_rate": 0.0001, |
| "loss": 0.013, |
| "step": 2262 |
| }, |
| { |
| "epoch": 754.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0127, |
| "step": 2265 |
| }, |
| { |
| "epoch": 755.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 2268 |
| }, |
| { |
| "epoch": 756.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2271 |
| }, |
| { |
| "epoch": 757.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0136, |
| "step": 2274 |
| }, |
| { |
| "epoch": 758.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 2277 |
| }, |
| { |
| "epoch": 759.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2280 |
| }, |
| { |
| "epoch": 760.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0122, |
| "step": 2283 |
| }, |
| { |
| "epoch": 761.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0123, |
| "step": 2286 |
| }, |
| { |
| "epoch": 762.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 2289 |
| }, |
| { |
| "epoch": 763.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0117, |
| "step": 2292 |
| }, |
| { |
| "epoch": 764.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0121, |
| "step": 2295 |
| }, |
| { |
| "epoch": 765.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0124, |
| "step": 2298 |
| }, |
| { |
| "epoch": 766.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0122, |
| "step": 2301 |
| }, |
| { |
| "epoch": 767.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2304 |
| }, |
| { |
| "epoch": 767.88, |
| "eval_exact_match": 0.48936170212765956, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.46691325306892395, |
| "eval_runtime": 106.1611, |
| "eval_samples_per_second": 9.74, |
| "step": 2304 |
| }, |
| { |
| "epoch": 768.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0123, |
| "step": 2307 |
| }, |
| { |
| "epoch": 769.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2310 |
| }, |
| { |
| "epoch": 770.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0123, |
| "step": 2313 |
| }, |
| { |
| "epoch": 771.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0128, |
| "step": 2316 |
| }, |
| { |
| "epoch": 772.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0124, |
| "step": 2319 |
| }, |
| { |
| "epoch": 773.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 2322 |
| }, |
| { |
| "epoch": 774.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0124, |
| "step": 2325 |
| }, |
| { |
| "epoch": 775.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0117, |
| "step": 2328 |
| }, |
| { |
| "epoch": 776.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2331 |
| }, |
| { |
| "epoch": 777.88, |
| "learning_rate": 0.0001, |
| "loss": 0.012, |
| "step": 2334 |
| }, |
| { |
| "epoch": 778.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2337 |
| }, |
| { |
| "epoch": 779.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0123, |
| "step": 2340 |
| }, |
| { |
| "epoch": 780.88, |
| "learning_rate": 0.0001, |
| "loss": 0.011, |
| "step": 2343 |
| }, |
| { |
| "epoch": 781.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0116, |
| "step": 2346 |
| }, |
| { |
| "epoch": 782.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0117, |
| "step": 2349 |
| }, |
| { |
| "epoch": 783.88, |
| "learning_rate": 0.0001, |
| "loss": 0.012, |
| "step": 2352 |
| }, |
| { |
| "epoch": 784.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2355 |
| }, |
| { |
| "epoch": 785.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0126, |
| "step": 2358 |
| }, |
| { |
| "epoch": 786.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0112, |
| "step": 2361 |
| }, |
| { |
| "epoch": 787.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0116, |
| "step": 2364 |
| }, |
| { |
| "epoch": 788.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2367 |
| }, |
| { |
| "epoch": 789.29, |
| "eval_exact_match": 0.4951644100580271, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.4785517156124115, |
| "eval_runtime": 101.2749, |
| "eval_samples_per_second": 10.21, |
| "step": 2368 |
| }, |
| { |
| "epoch": 789.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0109, |
| "step": 2370 |
| }, |
| { |
| "epoch": 790.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2373 |
| }, |
| { |
| "epoch": 791.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2376 |
| }, |
| { |
| "epoch": 792.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2379 |
| }, |
| { |
| "epoch": 793.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0109, |
| "step": 2382 |
| }, |
| { |
| "epoch": 794.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0116, |
| "step": 2385 |
| }, |
| { |
| "epoch": 795.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0115, |
| "step": 2388 |
| }, |
| { |
| "epoch": 796.88, |
| "learning_rate": 0.0001, |
| "loss": 0.011, |
| "step": 2391 |
| }, |
| { |
| "epoch": 797.88, |
| "learning_rate": 0.0001, |
| "loss": 0.011, |
| "step": 2394 |
| }, |
| { |
| "epoch": 798.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0113, |
| "step": 2397 |
| }, |
| { |
| "epoch": 799.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 2400 |
| }, |
| { |
| "epoch": 800.88, |
| "learning_rate": 0.0001, |
| "loss": 0.011, |
| "step": 2403 |
| }, |
| { |
| "epoch": 801.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0108, |
| "step": 2406 |
| }, |
| { |
| "epoch": 802.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0112, |
| "step": 2409 |
| }, |
| { |
| "epoch": 803.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0106, |
| "step": 2412 |
| }, |
| { |
| "epoch": 804.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0108, |
| "step": 2415 |
| }, |
| { |
| "epoch": 805.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0106, |
| "step": 2418 |
| }, |
| { |
| "epoch": 806.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0109, |
| "step": 2421 |
| }, |
| { |
| "epoch": 807.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0108, |
| "step": 2424 |
| }, |
| { |
| "epoch": 808.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0113, |
| "step": 2427 |
| }, |
| { |
| "epoch": 809.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0105, |
| "step": 2430 |
| }, |
| { |
| "epoch": 810.59, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.47889411449432373, |
| "eval_runtime": 106.781, |
| "eval_samples_per_second": 9.683, |
| "step": 2432 |
| }, |
| { |
| "epoch": 810.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0109, |
| "step": 2433 |
| }, |
| { |
| "epoch": 811.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0105, |
| "step": 2436 |
| }, |
| { |
| "epoch": 812.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0112, |
| "step": 2439 |
| }, |
| { |
| "epoch": 813.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2442 |
| }, |
| { |
| "epoch": 814.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0104, |
| "step": 2445 |
| }, |
| { |
| "epoch": 815.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0105, |
| "step": 2448 |
| }, |
| { |
| "epoch": 816.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2451 |
| }, |
| { |
| "epoch": 817.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0107, |
| "step": 2454 |
| }, |
| { |
| "epoch": 818.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0104, |
| "step": 2457 |
| }, |
| { |
| "epoch": 819.88, |
| "learning_rate": 0.0001, |
| "loss": 0.01, |
| "step": 2460 |
| }, |
| { |
| "epoch": 820.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 2463 |
| }, |
| { |
| "epoch": 821.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2466 |
| }, |
| { |
| "epoch": 822.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0104, |
| "step": 2469 |
| }, |
| { |
| "epoch": 823.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0118, |
| "step": 2472 |
| }, |
| { |
| "epoch": 824.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0111, |
| "step": 2475 |
| }, |
| { |
| "epoch": 825.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2478 |
| }, |
| { |
| "epoch": 826.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0099, |
| "step": 2481 |
| }, |
| { |
| "epoch": 827.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2484 |
| }, |
| { |
| "epoch": 828.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0108, |
| "step": 2487 |
| }, |
| { |
| "epoch": 829.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0109, |
| "step": 2490 |
| }, |
| { |
| "epoch": 830.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2493 |
| }, |
| { |
| "epoch": 831.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0106, |
| "step": 2496 |
| }, |
| { |
| "epoch": 831.88, |
| "eval_exact_match": 0.4941972920696325, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.47770604491233826, |
| "eval_runtime": 109.549, |
| "eval_samples_per_second": 9.439, |
| "step": 2496 |
| }, |
| { |
| "epoch": 832.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 2499 |
| }, |
| { |
| "epoch": 833.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 2502 |
| }, |
| { |
| "epoch": 834.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2505 |
| }, |
| { |
| "epoch": 835.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 2508 |
| }, |
| { |
| "epoch": 836.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 2511 |
| }, |
| { |
| "epoch": 837.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2514 |
| }, |
| { |
| "epoch": 838.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0092, |
| "step": 2517 |
| }, |
| { |
| "epoch": 839.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2520 |
| }, |
| { |
| "epoch": 840.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2523 |
| }, |
| { |
| "epoch": 841.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 2526 |
| }, |
| { |
| "epoch": 842.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2529 |
| }, |
| { |
| "epoch": 843.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 2532 |
| }, |
| { |
| "epoch": 844.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2535 |
| }, |
| { |
| "epoch": 845.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2538 |
| }, |
| { |
| "epoch": 846.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0094, |
| "step": 2541 |
| }, |
| { |
| "epoch": 847.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2544 |
| }, |
| { |
| "epoch": 848.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0101, |
| "step": 2547 |
| }, |
| { |
| "epoch": 849.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0099, |
| "step": 2550 |
| }, |
| { |
| "epoch": 850.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 2553 |
| }, |
| { |
| "epoch": 851.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0107, |
| "step": 2556 |
| }, |
| { |
| "epoch": 852.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0095, |
| "step": 2559 |
| }, |
| { |
| "epoch": 853.29, |
| "eval_exact_match": 0.4922630560928433, |
| "eval_exec": 0.4932301740812379, |
| "eval_loss": 0.4875582158565521, |
| "eval_runtime": 106.2482, |
| "eval_samples_per_second": 9.732, |
| "step": 2560 |
| }, |
| { |
| "epoch": 853.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 2562 |
| }, |
| { |
| "epoch": 854.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 2565 |
| }, |
| { |
| "epoch": 855.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2568 |
| }, |
| { |
| "epoch": 856.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 2571 |
| }, |
| { |
| "epoch": 857.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0094, |
| "step": 2574 |
| }, |
| { |
| "epoch": 858.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0095, |
| "step": 2577 |
| }, |
| { |
| "epoch": 859.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0094, |
| "step": 2580 |
| }, |
| { |
| "epoch": 860.88, |
| "learning_rate": 0.0001, |
| "loss": 0.009, |
| "step": 2583 |
| }, |
| { |
| "epoch": 861.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0096, |
| "step": 2586 |
| }, |
| { |
| "epoch": 862.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0091, |
| "step": 2589 |
| }, |
| { |
| "epoch": 863.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0087, |
| "step": 2592 |
| }, |
| { |
| "epoch": 864.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2595 |
| }, |
| { |
| "epoch": 865.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0089, |
| "step": 2598 |
| }, |
| { |
| "epoch": 866.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 2601 |
| }, |
| { |
| "epoch": 867.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0094, |
| "step": 2604 |
| }, |
| { |
| "epoch": 868.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 2607 |
| }, |
| { |
| "epoch": 869.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0092, |
| "step": 2610 |
| }, |
| { |
| "epoch": 870.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0092, |
| "step": 2613 |
| }, |
| { |
| "epoch": 871.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0087, |
| "step": 2616 |
| }, |
| { |
| "epoch": 872.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0088, |
| "step": 2619 |
| }, |
| { |
| "epoch": 873.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 2622 |
| }, |
| { |
| "epoch": 874.59, |
| "eval_exact_match": 0.48936170212765956, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.4897403120994568, |
| "eval_runtime": 109.6459, |
| "eval_samples_per_second": 9.43, |
| "step": 2624 |
| }, |
| { |
| "epoch": 874.88, |
| "learning_rate": 0.0001, |
| "loss": 0.009, |
| "step": 2625 |
| }, |
| { |
| "epoch": 875.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 2628 |
| }, |
| { |
| "epoch": 876.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2631 |
| }, |
| { |
| "epoch": 877.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 2634 |
| }, |
| { |
| "epoch": 878.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 2637 |
| }, |
| { |
| "epoch": 879.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0091, |
| "step": 2640 |
| }, |
| { |
| "epoch": 880.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2643 |
| }, |
| { |
| "epoch": 881.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2646 |
| }, |
| { |
| "epoch": 882.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0089, |
| "step": 2649 |
| }, |
| { |
| "epoch": 883.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0082, |
| "step": 2652 |
| }, |
| { |
| "epoch": 884.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 2655 |
| }, |
| { |
| "epoch": 885.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0084, |
| "step": 2658 |
| }, |
| { |
| "epoch": 886.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2661 |
| }, |
| { |
| "epoch": 887.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2664 |
| }, |
| { |
| "epoch": 888.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0082, |
| "step": 2667 |
| }, |
| { |
| "epoch": 889.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0082, |
| "step": 2670 |
| }, |
| { |
| "epoch": 890.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0087, |
| "step": 2673 |
| }, |
| { |
| "epoch": 891.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2676 |
| }, |
| { |
| "epoch": 892.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0084, |
| "step": 2679 |
| }, |
| { |
| "epoch": 893.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0087, |
| "step": 2682 |
| }, |
| { |
| "epoch": 894.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 2685 |
| }, |
| { |
| "epoch": 895.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2688 |
| }, |
| { |
| "epoch": 895.88, |
| "eval_exact_match": 0.5, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.4995100796222687, |
| "eval_runtime": 115.5484, |
| "eval_samples_per_second": 8.949, |
| "step": 2688 |
| }, |
| { |
| "epoch": 896.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0081, |
| "step": 2691 |
| }, |
| { |
| "epoch": 897.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0081, |
| "step": 2694 |
| }, |
| { |
| "epoch": 898.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0088, |
| "step": 2697 |
| }, |
| { |
| "epoch": 899.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 2700 |
| }, |
| { |
| "epoch": 900.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2703 |
| }, |
| { |
| "epoch": 901.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0088, |
| "step": 2706 |
| }, |
| { |
| "epoch": 902.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2709 |
| }, |
| { |
| "epoch": 903.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0084, |
| "step": 2712 |
| }, |
| { |
| "epoch": 904.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2715 |
| }, |
| { |
| "epoch": 905.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0084, |
| "step": 2718 |
| }, |
| { |
| "epoch": 906.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2721 |
| }, |
| { |
| "epoch": 907.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0082, |
| "step": 2724 |
| }, |
| { |
| "epoch": 908.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2727 |
| }, |
| { |
| "epoch": 909.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0081, |
| "step": 2730 |
| }, |
| { |
| "epoch": 910.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2733 |
| }, |
| { |
| "epoch": 911.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0078, |
| "step": 2736 |
| }, |
| { |
| "epoch": 912.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2739 |
| }, |
| { |
| "epoch": 913.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2742 |
| }, |
| { |
| "epoch": 914.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 2745 |
| }, |
| { |
| "epoch": 915.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2748 |
| }, |
| { |
| "epoch": 916.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2751 |
| }, |
| { |
| "epoch": 917.29, |
| "eval_exact_match": 0.49709864603481624, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.4965713918209076, |
| "eval_runtime": 103.1798, |
| "eval_samples_per_second": 10.021, |
| "step": 2752 |
| }, |
| { |
| "epoch": 917.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 2754 |
| }, |
| { |
| "epoch": 918.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2757 |
| }, |
| { |
| "epoch": 919.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2760 |
| }, |
| { |
| "epoch": 920.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2763 |
| }, |
| { |
| "epoch": 921.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2766 |
| }, |
| { |
| "epoch": 922.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0073, |
| "step": 2769 |
| }, |
| { |
| "epoch": 923.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0074, |
| "step": 2772 |
| }, |
| { |
| "epoch": 924.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0075, |
| "step": 2775 |
| }, |
| { |
| "epoch": 925.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2778 |
| }, |
| { |
| "epoch": 926.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 2781 |
| }, |
| { |
| "epoch": 927.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2784 |
| }, |
| { |
| "epoch": 928.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0075, |
| "step": 2787 |
| }, |
| { |
| "epoch": 929.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2790 |
| }, |
| { |
| "epoch": 930.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 2793 |
| }, |
| { |
| "epoch": 931.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0078, |
| "step": 2796 |
| }, |
| { |
| "epoch": 932.88, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 2799 |
| }, |
| { |
| "epoch": 933.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 2802 |
| }, |
| { |
| "epoch": 934.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0072, |
| "step": 2805 |
| }, |
| { |
| "epoch": 935.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2808 |
| }, |
| { |
| "epoch": 936.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2811 |
| }, |
| { |
| "epoch": 937.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0075, |
| "step": 2814 |
| }, |
| { |
| "epoch": 938.59, |
| "eval_exact_match": 0.4912959381044487, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.5068889856338501, |
| "eval_runtime": 114.9149, |
| "eval_samples_per_second": 8.998, |
| "step": 2816 |
| }, |
| { |
| "epoch": 938.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 2817 |
| }, |
| { |
| "epoch": 939.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0074, |
| "step": 2820 |
| }, |
| { |
| "epoch": 940.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 2823 |
| }, |
| { |
| "epoch": 941.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 2826 |
| }, |
| { |
| "epoch": 942.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0073, |
| "step": 2829 |
| }, |
| { |
| "epoch": 943.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0073, |
| "step": 2832 |
| }, |
| { |
| "epoch": 944.88, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 2835 |
| }, |
| { |
| "epoch": 945.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2838 |
| }, |
| { |
| "epoch": 946.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 2841 |
| }, |
| { |
| "epoch": 947.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 2844 |
| }, |
| { |
| "epoch": 948.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2847 |
| }, |
| { |
| "epoch": 949.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0078, |
| "step": 2850 |
| }, |
| { |
| "epoch": 950.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0082, |
| "step": 2853 |
| }, |
| { |
| "epoch": 951.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2856 |
| }, |
| { |
| "epoch": 952.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2859 |
| }, |
| { |
| "epoch": 953.88, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 2862 |
| }, |
| { |
| "epoch": 954.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 2865 |
| }, |
| { |
| "epoch": 955.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 2868 |
| }, |
| { |
| "epoch": 956.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0073, |
| "step": 2871 |
| }, |
| { |
| "epoch": 957.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2874 |
| }, |
| { |
| "epoch": 958.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 2877 |
| }, |
| { |
| "epoch": 959.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0076, |
| "step": 2880 |
| }, |
| { |
| "epoch": 959.88, |
| "eval_exact_match": 0.4835589941972921, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.5026101469993591, |
| "eval_runtime": 106.021, |
| "eval_samples_per_second": 9.753, |
| "step": 2880 |
| }, |
| { |
| "epoch": 960.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0072, |
| "step": 2883 |
| }, |
| { |
| "epoch": 961.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 2886 |
| }, |
| { |
| "epoch": 962.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0074, |
| "step": 2889 |
| }, |
| { |
| "epoch": 963.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 2892 |
| }, |
| { |
| "epoch": 964.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0072, |
| "step": 2895 |
| }, |
| { |
| "epoch": 965.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0074, |
| "step": 2898 |
| }, |
| { |
| "epoch": 966.88, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 2901 |
| }, |
| { |
| "epoch": 967.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2904 |
| }, |
| { |
| "epoch": 968.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 2907 |
| }, |
| { |
| "epoch": 969.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0069, |
| "step": 2910 |
| }, |
| { |
| "epoch": 970.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 2913 |
| }, |
| { |
| "epoch": 971.88, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 2916 |
| }, |
| { |
| "epoch": 972.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2919 |
| }, |
| { |
| "epoch": 973.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 2922 |
| }, |
| { |
| "epoch": 974.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 2925 |
| }, |
| { |
| "epoch": 975.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 2928 |
| }, |
| { |
| "epoch": 976.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 2931 |
| }, |
| { |
| "epoch": 977.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2934 |
| }, |
| { |
| "epoch": 978.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 2937 |
| }, |
| { |
| "epoch": 979.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2940 |
| }, |
| { |
| "epoch": 980.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2943 |
| }, |
| { |
| "epoch": 981.29, |
| "eval_exact_match": 0.4796905222437137, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.5093928575515747, |
| "eval_runtime": 102.014, |
| "eval_samples_per_second": 10.136, |
| "step": 2944 |
| }, |
| { |
| "epoch": 981.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 2946 |
| }, |
| { |
| "epoch": 982.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 2949 |
| }, |
| { |
| "epoch": 983.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 2952 |
| }, |
| { |
| "epoch": 984.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 2955 |
| }, |
| { |
| "epoch": 985.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 2958 |
| }, |
| { |
| "epoch": 986.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 2961 |
| }, |
| { |
| "epoch": 987.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0058, |
| "step": 2964 |
| }, |
| { |
| "epoch": 988.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 2967 |
| }, |
| { |
| "epoch": 989.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 2970 |
| }, |
| { |
| "epoch": 990.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 2973 |
| }, |
| { |
| "epoch": 991.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 2976 |
| }, |
| { |
| "epoch": 992.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 2979 |
| }, |
| { |
| "epoch": 993.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 2982 |
| }, |
| { |
| "epoch": 994.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 2985 |
| }, |
| { |
| "epoch": 995.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 2988 |
| }, |
| { |
| "epoch": 996.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 2991 |
| }, |
| { |
| "epoch": 997.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 2994 |
| }, |
| { |
| "epoch": 998.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 2997 |
| }, |
| { |
| "epoch": 999.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1000.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 3003 |
| }, |
| { |
| "epoch": 1001.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3006 |
| }, |
| { |
| "epoch": 1002.59, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.5153567790985107, |
| "eval_runtime": 112.3969, |
| "eval_samples_per_second": 9.2, |
| "step": 3008 |
| }, |
| { |
| "epoch": 1002.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 3009 |
| }, |
| { |
| "epoch": 1003.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 3012 |
| }, |
| { |
| "epoch": 1004.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 3015 |
| }, |
| { |
| "epoch": 1005.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3018 |
| }, |
| { |
| "epoch": 1006.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 3021 |
| }, |
| { |
| "epoch": 1007.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 3024 |
| }, |
| { |
| "epoch": 1008.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 3027 |
| }, |
| { |
| "epoch": 1009.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 3030 |
| }, |
| { |
| "epoch": 1010.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0066, |
| "step": 3033 |
| }, |
| { |
| "epoch": 1011.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 3036 |
| }, |
| { |
| "epoch": 1012.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3039 |
| }, |
| { |
| "epoch": 1013.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 3042 |
| }, |
| { |
| "epoch": 1014.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3045 |
| }, |
| { |
| "epoch": 1015.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 3048 |
| }, |
| { |
| "epoch": 1016.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 3051 |
| }, |
| { |
| "epoch": 1017.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3054 |
| }, |
| { |
| "epoch": 1018.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3057 |
| }, |
| { |
| "epoch": 1019.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3060 |
| }, |
| { |
| "epoch": 1020.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3063 |
| }, |
| { |
| "epoch": 1021.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3066 |
| }, |
| { |
| "epoch": 1022.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0058, |
| "step": 3069 |
| }, |
| { |
| "epoch": 1023.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3072 |
| }, |
| { |
| "epoch": 1023.88, |
| "eval_exact_match": 0.4864603481624758, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.5245723724365234, |
| "eval_runtime": 108.9392, |
| "eval_samples_per_second": 9.492, |
| "step": 3072 |
| }, |
| { |
| "epoch": 1024.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3075 |
| }, |
| { |
| "epoch": 1025.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 3078 |
| }, |
| { |
| "epoch": 1026.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 3081 |
| }, |
| { |
| "epoch": 1027.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3084 |
| }, |
| { |
| "epoch": 1028.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 3087 |
| }, |
| { |
| "epoch": 1029.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3090 |
| }, |
| { |
| "epoch": 1030.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 3093 |
| }, |
| { |
| "epoch": 1031.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3096 |
| }, |
| { |
| "epoch": 1032.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3099 |
| }, |
| { |
| "epoch": 1033.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3102 |
| }, |
| { |
| "epoch": 1034.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3105 |
| }, |
| { |
| "epoch": 1035.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3108 |
| }, |
| { |
| "epoch": 1036.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 3111 |
| }, |
| { |
| "epoch": 1037.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3114 |
| }, |
| { |
| "epoch": 1038.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0061, |
| "step": 3117 |
| }, |
| { |
| "epoch": 1039.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3120 |
| }, |
| { |
| "epoch": 1040.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 3123 |
| }, |
| { |
| "epoch": 1041.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3126 |
| }, |
| { |
| "epoch": 1042.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3129 |
| }, |
| { |
| "epoch": 1043.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3132 |
| }, |
| { |
| "epoch": 1044.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3135 |
| }, |
| { |
| "epoch": 1045.29, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.5266330242156982, |
| "eval_runtime": 100.7584, |
| "eval_samples_per_second": 10.262, |
| "step": 3136 |
| }, |
| { |
| "epoch": 1045.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 3138 |
| }, |
| { |
| "epoch": 1046.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 3141 |
| }, |
| { |
| "epoch": 1047.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3144 |
| }, |
| { |
| "epoch": 1048.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3147 |
| }, |
| { |
| "epoch": 1049.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3150 |
| }, |
| { |
| "epoch": 1050.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3153 |
| }, |
| { |
| "epoch": 1051.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3156 |
| }, |
| { |
| "epoch": 1052.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3159 |
| }, |
| { |
| "epoch": 1053.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3162 |
| }, |
| { |
| "epoch": 1054.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 3165 |
| }, |
| { |
| "epoch": 1055.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3168 |
| }, |
| { |
| "epoch": 1056.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3171 |
| }, |
| { |
| "epoch": 1057.88, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 3174 |
| }, |
| { |
| "epoch": 1058.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3177 |
| }, |
| { |
| "epoch": 1059.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3180 |
| }, |
| { |
| "epoch": 1060.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3183 |
| }, |
| { |
| "epoch": 1061.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3186 |
| }, |
| { |
| "epoch": 1062.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3189 |
| }, |
| { |
| "epoch": 1063.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3192 |
| }, |
| { |
| "epoch": 1064.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3195 |
| }, |
| { |
| "epoch": 1065.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3198 |
| }, |
| { |
| "epoch": 1066.59, |
| "eval_exact_match": 0.4864603481624758, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5267293453216553, |
| "eval_runtime": 104.4326, |
| "eval_samples_per_second": 9.901, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1066.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3201 |
| }, |
| { |
| "epoch": 1067.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0058, |
| "step": 3204 |
| }, |
| { |
| "epoch": 1068.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3207 |
| }, |
| { |
| "epoch": 1069.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3210 |
| }, |
| { |
| "epoch": 1070.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3213 |
| }, |
| { |
| "epoch": 1071.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3216 |
| }, |
| { |
| "epoch": 1072.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3219 |
| }, |
| { |
| "epoch": 1073.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3222 |
| }, |
| { |
| "epoch": 1074.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3225 |
| }, |
| { |
| "epoch": 1075.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3228 |
| }, |
| { |
| "epoch": 1076.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3231 |
| }, |
| { |
| "epoch": 1077.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3234 |
| }, |
| { |
| "epoch": 1078.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3237 |
| }, |
| { |
| "epoch": 1079.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3240 |
| }, |
| { |
| "epoch": 1080.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3243 |
| }, |
| { |
| "epoch": 1081.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 3246 |
| }, |
| { |
| "epoch": 1082.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3249 |
| }, |
| { |
| "epoch": 1083.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3252 |
| }, |
| { |
| "epoch": 1084.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3255 |
| }, |
| { |
| "epoch": 1085.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3258 |
| }, |
| { |
| "epoch": 1086.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 3261 |
| }, |
| { |
| "epoch": 1087.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3264 |
| }, |
| { |
| "epoch": 1087.88, |
| "eval_exact_match": 0.488394584139265, |
| "eval_exec": 0.4835589941972921, |
| "eval_loss": 0.5323160290718079, |
| "eval_runtime": 100.2092, |
| "eval_samples_per_second": 10.318, |
| "step": 3264 |
| }, |
| { |
| "epoch": 1088.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3267 |
| }, |
| { |
| "epoch": 1089.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3270 |
| }, |
| { |
| "epoch": 1090.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3273 |
| }, |
| { |
| "epoch": 1091.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3276 |
| }, |
| { |
| "epoch": 1092.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3279 |
| }, |
| { |
| "epoch": 1093.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 3282 |
| }, |
| { |
| "epoch": 1094.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3285 |
| }, |
| { |
| "epoch": 1095.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3288 |
| }, |
| { |
| "epoch": 1096.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3291 |
| }, |
| { |
| "epoch": 1097.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3294 |
| }, |
| { |
| "epoch": 1098.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3297 |
| }, |
| { |
| "epoch": 1099.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1100.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3303 |
| }, |
| { |
| "epoch": 1101.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3306 |
| }, |
| { |
| "epoch": 1102.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3309 |
| }, |
| { |
| "epoch": 1103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3312 |
| }, |
| { |
| "epoch": 1104.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3315 |
| }, |
| { |
| "epoch": 1105.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0054, |
| "step": 3318 |
| }, |
| { |
| "epoch": 1106.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3321 |
| }, |
| { |
| "epoch": 1107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3324 |
| }, |
| { |
| "epoch": 1108.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3327 |
| }, |
| { |
| "epoch": 1109.29, |
| "eval_exact_match": 0.48452611218568664, |
| "eval_exec": 0.4796905222437137, |
| "eval_loss": 0.5343111157417297, |
| "eval_runtime": 99.4552, |
| "eval_samples_per_second": 10.397, |
| "step": 3328 |
| }, |
| { |
| "epoch": 1109.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 3330 |
| }, |
| { |
| "epoch": 1110.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3333 |
| }, |
| { |
| "epoch": 1111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3336 |
| }, |
| { |
| "epoch": 1112.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3339 |
| }, |
| { |
| "epoch": 1113.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0047, |
| "step": 3342 |
| }, |
| { |
| "epoch": 1114.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3345 |
| }, |
| { |
| "epoch": 1115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3348 |
| }, |
| { |
| "epoch": 1116.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3351 |
| }, |
| { |
| "epoch": 1117.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3354 |
| }, |
| { |
| "epoch": 1118.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0047, |
| "step": 3357 |
| }, |
| { |
| "epoch": 1119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3360 |
| }, |
| { |
| "epoch": 1120.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3363 |
| }, |
| { |
| "epoch": 1121.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3366 |
| }, |
| { |
| "epoch": 1122.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3369 |
| }, |
| { |
| "epoch": 1123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0047, |
| "step": 3372 |
| }, |
| { |
| "epoch": 1124.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3375 |
| }, |
| { |
| "epoch": 1125.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3378 |
| }, |
| { |
| "epoch": 1126.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3381 |
| }, |
| { |
| "epoch": 1127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3384 |
| }, |
| { |
| "epoch": 1128.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3387 |
| }, |
| { |
| "epoch": 1129.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3390 |
| }, |
| { |
| "epoch": 1130.59, |
| "eval_exact_match": 0.488394584139265, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.5409371256828308, |
| "eval_runtime": 104.8508, |
| "eval_samples_per_second": 9.862, |
| "step": 3392 |
| }, |
| { |
| "epoch": 1130.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3393 |
| }, |
| { |
| "epoch": 1131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3396 |
| }, |
| { |
| "epoch": 1132.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3399 |
| }, |
| { |
| "epoch": 1133.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3402 |
| }, |
| { |
| "epoch": 1134.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0047, |
| "step": 3405 |
| }, |
| { |
| "epoch": 1135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 3408 |
| }, |
| { |
| "epoch": 1136.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3411 |
| }, |
| { |
| "epoch": 1137.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3414 |
| }, |
| { |
| "epoch": 1138.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3417 |
| }, |
| { |
| "epoch": 1139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3420 |
| }, |
| { |
| "epoch": 1140.88, |
| "learning_rate": 0.0001, |
| "loss": 0.005, |
| "step": 3423 |
| }, |
| { |
| "epoch": 1141.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3426 |
| }, |
| { |
| "epoch": 1142.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3429 |
| }, |
| { |
| "epoch": 1143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3432 |
| }, |
| { |
| "epoch": 1144.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3435 |
| }, |
| { |
| "epoch": 1145.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3438 |
| }, |
| { |
| "epoch": 1146.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3441 |
| }, |
| { |
| "epoch": 1147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3444 |
| }, |
| { |
| "epoch": 1148.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3447 |
| }, |
| { |
| "epoch": 1149.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3450 |
| }, |
| { |
| "epoch": 1150.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3453 |
| }, |
| { |
| "epoch": 1151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3456 |
| }, |
| { |
| "epoch": 1151.88, |
| "eval_exact_match": 0.4941972920696325, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.5484944581985474, |
| "eval_runtime": 103.3732, |
| "eval_samples_per_second": 10.003, |
| "step": 3456 |
| }, |
| { |
| "epoch": 1152.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0051, |
| "step": 3459 |
| }, |
| { |
| "epoch": 1153.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0052, |
| "step": 3462 |
| }, |
| { |
| "epoch": 1154.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3465 |
| }, |
| { |
| "epoch": 1155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3468 |
| }, |
| { |
| "epoch": 1156.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3471 |
| }, |
| { |
| "epoch": 1157.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3474 |
| }, |
| { |
| "epoch": 1158.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3477 |
| }, |
| { |
| "epoch": 1159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3480 |
| }, |
| { |
| "epoch": 1160.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3483 |
| }, |
| { |
| "epoch": 1161.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3486 |
| }, |
| { |
| "epoch": 1162.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3489 |
| }, |
| { |
| "epoch": 1163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3492 |
| }, |
| { |
| "epoch": 1164.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3495 |
| }, |
| { |
| "epoch": 1165.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3498 |
| }, |
| { |
| "epoch": 1166.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 3501 |
| }, |
| { |
| "epoch": 1167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3504 |
| }, |
| { |
| "epoch": 1168.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3507 |
| }, |
| { |
| "epoch": 1169.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3510 |
| }, |
| { |
| "epoch": 1170.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3513 |
| }, |
| { |
| "epoch": 1171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3516 |
| }, |
| { |
| "epoch": 1172.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3519 |
| }, |
| { |
| "epoch": 1173.29, |
| "eval_exact_match": 0.4990328820116054, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.547235369682312, |
| "eval_runtime": 96.1089, |
| "eval_samples_per_second": 10.759, |
| "step": 3520 |
| }, |
| { |
| "epoch": 1173.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3522 |
| }, |
| { |
| "epoch": 1174.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3525 |
| }, |
| { |
| "epoch": 1175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3528 |
| }, |
| { |
| "epoch": 1176.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3531 |
| }, |
| { |
| "epoch": 1177.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3534 |
| }, |
| { |
| "epoch": 1178.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3537 |
| }, |
| { |
| "epoch": 1179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3540 |
| }, |
| { |
| "epoch": 1180.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3543 |
| }, |
| { |
| "epoch": 1181.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3546 |
| }, |
| { |
| "epoch": 1182.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 3549 |
| }, |
| { |
| "epoch": 1183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3552 |
| }, |
| { |
| "epoch": 1184.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3555 |
| }, |
| { |
| "epoch": 1185.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3558 |
| }, |
| { |
| "epoch": 1186.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3561 |
| }, |
| { |
| "epoch": 1187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3564 |
| }, |
| { |
| "epoch": 1188.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3567 |
| }, |
| { |
| "epoch": 1189.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3570 |
| }, |
| { |
| "epoch": 1190.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3573 |
| }, |
| { |
| "epoch": 1191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3576 |
| }, |
| { |
| "epoch": 1192.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3579 |
| }, |
| { |
| "epoch": 1193.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3582 |
| }, |
| { |
| "epoch": 1194.59, |
| "eval_exact_match": 0.5009671179883946, |
| "eval_exec": 0.49709864603481624, |
| "eval_loss": 0.5491827130317688, |
| "eval_runtime": 100.0201, |
| "eval_samples_per_second": 10.338, |
| "step": 3584 |
| }, |
| { |
| "epoch": 1194.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3585 |
| }, |
| { |
| "epoch": 1195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3588 |
| }, |
| { |
| "epoch": 1196.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3591 |
| }, |
| { |
| "epoch": 1197.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3594 |
| }, |
| { |
| "epoch": 1198.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3597 |
| }, |
| { |
| "epoch": 1199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1200.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3603 |
| }, |
| { |
| "epoch": 1201.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3606 |
| }, |
| { |
| "epoch": 1202.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3609 |
| }, |
| { |
| "epoch": 1203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 3612 |
| }, |
| { |
| "epoch": 1204.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0045, |
| "step": 3615 |
| }, |
| { |
| "epoch": 1205.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3618 |
| }, |
| { |
| "epoch": 1206.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3621 |
| }, |
| { |
| "epoch": 1207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3624 |
| }, |
| { |
| "epoch": 1208.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3627 |
| }, |
| { |
| "epoch": 1209.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3630 |
| }, |
| { |
| "epoch": 1210.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3633 |
| }, |
| { |
| "epoch": 1211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3636 |
| }, |
| { |
| "epoch": 1212.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3639 |
| }, |
| { |
| "epoch": 1213.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3642 |
| }, |
| { |
| "epoch": 1214.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 3645 |
| }, |
| { |
| "epoch": 1215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3648 |
| }, |
| { |
| "epoch": 1215.88, |
| "eval_exact_match": 0.4864603481624758, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5521473288536072, |
| "eval_runtime": 101.5945, |
| "eval_samples_per_second": 10.178, |
| "step": 3648 |
| }, |
| { |
| "epoch": 1216.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3651 |
| }, |
| { |
| "epoch": 1217.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3654 |
| }, |
| { |
| "epoch": 1218.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3657 |
| }, |
| { |
| "epoch": 1219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3660 |
| }, |
| { |
| "epoch": 1220.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3663 |
| }, |
| { |
| "epoch": 1221.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3666 |
| }, |
| { |
| "epoch": 1222.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3669 |
| }, |
| { |
| "epoch": 1223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3672 |
| }, |
| { |
| "epoch": 1224.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3675 |
| }, |
| { |
| "epoch": 1225.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3678 |
| }, |
| { |
| "epoch": 1226.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3681 |
| }, |
| { |
| "epoch": 1227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3684 |
| }, |
| { |
| "epoch": 1228.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3687 |
| }, |
| { |
| "epoch": 1229.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3690 |
| }, |
| { |
| "epoch": 1230.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3693 |
| }, |
| { |
| "epoch": 1231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3696 |
| }, |
| { |
| "epoch": 1232.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3699 |
| }, |
| { |
| "epoch": 1233.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3702 |
| }, |
| { |
| "epoch": 1234.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3705 |
| }, |
| { |
| "epoch": 1235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3708 |
| }, |
| { |
| "epoch": 1236.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3711 |
| }, |
| { |
| "epoch": 1237.29, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5529308319091797, |
| "eval_runtime": 103.2003, |
| "eval_samples_per_second": 10.019, |
| "step": 3712 |
| }, |
| { |
| "epoch": 1237.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3714 |
| }, |
| { |
| "epoch": 1238.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3717 |
| }, |
| { |
| "epoch": 1239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 3720 |
| }, |
| { |
| "epoch": 1240.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3723 |
| }, |
| { |
| "epoch": 1241.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3726 |
| }, |
| { |
| "epoch": 1242.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3729 |
| }, |
| { |
| "epoch": 1243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3732 |
| }, |
| { |
| "epoch": 1244.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3735 |
| }, |
| { |
| "epoch": 1245.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3738 |
| }, |
| { |
| "epoch": 1246.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3741 |
| }, |
| { |
| "epoch": 1247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3744 |
| }, |
| { |
| "epoch": 1248.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3747 |
| }, |
| { |
| "epoch": 1249.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3750 |
| }, |
| { |
| "epoch": 1250.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3753 |
| }, |
| { |
| "epoch": 1251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0043, |
| "step": 3756 |
| }, |
| { |
| "epoch": 1252.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3759 |
| }, |
| { |
| "epoch": 1253.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3762 |
| }, |
| { |
| "epoch": 1254.88, |
| "learning_rate": 0.0001, |
| "loss": 0.004, |
| "step": 3765 |
| }, |
| { |
| "epoch": 1255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3768 |
| }, |
| { |
| "epoch": 1256.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3771 |
| }, |
| { |
| "epoch": 1257.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3774 |
| }, |
| { |
| "epoch": 1258.59, |
| "eval_exact_match": 0.48936170212765956, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.5576500296592712, |
| "eval_runtime": 102.4554, |
| "eval_samples_per_second": 10.092, |
| "step": 3776 |
| }, |
| { |
| "epoch": 1258.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3777 |
| }, |
| { |
| "epoch": 1259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3780 |
| }, |
| { |
| "epoch": 1260.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3783 |
| }, |
| { |
| "epoch": 1261.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3786 |
| }, |
| { |
| "epoch": 1262.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3789 |
| }, |
| { |
| "epoch": 1263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3792 |
| }, |
| { |
| "epoch": 1264.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3795 |
| }, |
| { |
| "epoch": 1265.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3798 |
| }, |
| { |
| "epoch": 1266.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3801 |
| }, |
| { |
| "epoch": 1267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3804 |
| }, |
| { |
| "epoch": 1268.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3807 |
| }, |
| { |
| "epoch": 1269.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3810 |
| }, |
| { |
| "epoch": 1270.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3813 |
| }, |
| { |
| "epoch": 1271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3816 |
| }, |
| { |
| "epoch": 1272.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3819 |
| }, |
| { |
| "epoch": 1273.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3822 |
| }, |
| { |
| "epoch": 1274.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3825 |
| }, |
| { |
| "epoch": 1275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3828 |
| }, |
| { |
| "epoch": 1276.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3831 |
| }, |
| { |
| "epoch": 1277.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3834 |
| }, |
| { |
| "epoch": 1278.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3837 |
| }, |
| { |
| "epoch": 1279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3840 |
| }, |
| { |
| "epoch": 1279.88, |
| "eval_exact_match": 0.4990328820116054, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.5551109910011292, |
| "eval_runtime": 108.2571, |
| "eval_samples_per_second": 9.551, |
| "step": 3840 |
| }, |
| { |
| "epoch": 1280.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3843 |
| }, |
| { |
| "epoch": 1281.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3846 |
| }, |
| { |
| "epoch": 1282.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3849 |
| }, |
| { |
| "epoch": 1283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3852 |
| }, |
| { |
| "epoch": 1284.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3855 |
| }, |
| { |
| "epoch": 1285.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3858 |
| }, |
| { |
| "epoch": 1286.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3861 |
| }, |
| { |
| "epoch": 1287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3864 |
| }, |
| { |
| "epoch": 1288.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3867 |
| }, |
| { |
| "epoch": 1289.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3870 |
| }, |
| { |
| "epoch": 1290.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3873 |
| }, |
| { |
| "epoch": 1291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3876 |
| }, |
| { |
| "epoch": 1292.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3879 |
| }, |
| { |
| "epoch": 1293.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 3882 |
| }, |
| { |
| "epoch": 1294.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3885 |
| }, |
| { |
| "epoch": 1295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 3888 |
| }, |
| { |
| "epoch": 1296.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3891 |
| }, |
| { |
| "epoch": 1297.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3894 |
| }, |
| { |
| "epoch": 1298.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3897 |
| }, |
| { |
| "epoch": 1299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1300.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3903 |
| }, |
| { |
| "epoch": 1301.29, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.5643026232719421, |
| "eval_runtime": 100.8652, |
| "eval_samples_per_second": 10.251, |
| "step": 3904 |
| }, |
| { |
| "epoch": 1301.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3906 |
| }, |
| { |
| "epoch": 1302.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3909 |
| }, |
| { |
| "epoch": 1303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3912 |
| }, |
| { |
| "epoch": 1304.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3915 |
| }, |
| { |
| "epoch": 1305.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3918 |
| }, |
| { |
| "epoch": 1306.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 3921 |
| }, |
| { |
| "epoch": 1307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3924 |
| }, |
| { |
| "epoch": 1308.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3927 |
| }, |
| { |
| "epoch": 1309.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 3930 |
| }, |
| { |
| "epoch": 1310.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3933 |
| }, |
| { |
| "epoch": 1311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3936 |
| }, |
| { |
| "epoch": 1312.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3939 |
| }, |
| { |
| "epoch": 1313.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3942 |
| }, |
| { |
| "epoch": 1314.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3945 |
| }, |
| { |
| "epoch": 1315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3948 |
| }, |
| { |
| "epoch": 1316.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3951 |
| }, |
| { |
| "epoch": 1317.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3954 |
| }, |
| { |
| "epoch": 1318.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 3957 |
| }, |
| { |
| "epoch": 1319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 3960 |
| }, |
| { |
| "epoch": 1320.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3963 |
| }, |
| { |
| "epoch": 1321.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3966 |
| }, |
| { |
| "epoch": 1322.59, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5721480250358582, |
| "eval_runtime": 106.1276, |
| "eval_samples_per_second": 9.743, |
| "step": 3968 |
| }, |
| { |
| "epoch": 1322.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3969 |
| }, |
| { |
| "epoch": 1323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3972 |
| }, |
| { |
| "epoch": 1324.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3975 |
| }, |
| { |
| "epoch": 1325.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3978 |
| }, |
| { |
| "epoch": 1326.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 3981 |
| }, |
| { |
| "epoch": 1327.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 3984 |
| }, |
| { |
| "epoch": 1328.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 3987 |
| }, |
| { |
| "epoch": 1329.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3990 |
| }, |
| { |
| "epoch": 1330.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 3993 |
| }, |
| { |
| "epoch": 1331.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 3996 |
| }, |
| { |
| "epoch": 1332.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 3999 |
| }, |
| { |
| "epoch": 1333.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4002 |
| }, |
| { |
| "epoch": 1334.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 4005 |
| }, |
| { |
| "epoch": 1335.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4008 |
| }, |
| { |
| "epoch": 1336.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4011 |
| }, |
| { |
| "epoch": 1337.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 4014 |
| }, |
| { |
| "epoch": 1338.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4017 |
| }, |
| { |
| "epoch": 1339.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 4020 |
| }, |
| { |
| "epoch": 1340.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4023 |
| }, |
| { |
| "epoch": 1341.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4026 |
| }, |
| { |
| "epoch": 1342.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4029 |
| }, |
| { |
| "epoch": 1343.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4032 |
| }, |
| { |
| "epoch": 1343.88, |
| "eval_exact_match": 0.4912959381044487, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.5649085640907288, |
| "eval_runtime": 100.2657, |
| "eval_samples_per_second": 10.313, |
| "step": 4032 |
| }, |
| { |
| "epoch": 1344.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4035 |
| }, |
| { |
| "epoch": 1345.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4038 |
| }, |
| { |
| "epoch": 1346.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4041 |
| }, |
| { |
| "epoch": 1347.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4044 |
| }, |
| { |
| "epoch": 1348.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4047 |
| }, |
| { |
| "epoch": 1349.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4050 |
| }, |
| { |
| "epoch": 1350.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4053 |
| }, |
| { |
| "epoch": 1351.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4056 |
| }, |
| { |
| "epoch": 1352.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4059 |
| }, |
| { |
| "epoch": 1353.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4062 |
| }, |
| { |
| "epoch": 1354.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4065 |
| }, |
| { |
| "epoch": 1355.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4068 |
| }, |
| { |
| "epoch": 1356.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4071 |
| }, |
| { |
| "epoch": 1357.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4074 |
| }, |
| { |
| "epoch": 1358.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4077 |
| }, |
| { |
| "epoch": 1359.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4080 |
| }, |
| { |
| "epoch": 1360.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4083 |
| }, |
| { |
| "epoch": 1361.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4086 |
| }, |
| { |
| "epoch": 1362.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4089 |
| }, |
| { |
| "epoch": 1363.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4092 |
| }, |
| { |
| "epoch": 1364.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4095 |
| }, |
| { |
| "epoch": 1365.29, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.5747038722038269, |
| "eval_runtime": 108.1776, |
| "eval_samples_per_second": 9.558, |
| "step": 4096 |
| }, |
| { |
| "epoch": 1365.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4098 |
| }, |
| { |
| "epoch": 1366.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4101 |
| }, |
| { |
| "epoch": 1367.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4104 |
| }, |
| { |
| "epoch": 1368.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4107 |
| }, |
| { |
| "epoch": 1369.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4110 |
| }, |
| { |
| "epoch": 1370.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4113 |
| }, |
| { |
| "epoch": 1371.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4116 |
| }, |
| { |
| "epoch": 1372.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4119 |
| }, |
| { |
| "epoch": 1373.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4122 |
| }, |
| { |
| "epoch": 1374.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4125 |
| }, |
| { |
| "epoch": 1375.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4128 |
| }, |
| { |
| "epoch": 1376.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4131 |
| }, |
| { |
| "epoch": 1377.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4134 |
| }, |
| { |
| "epoch": 1378.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4137 |
| }, |
| { |
| "epoch": 1379.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4140 |
| }, |
| { |
| "epoch": 1380.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4143 |
| }, |
| { |
| "epoch": 1381.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4146 |
| }, |
| { |
| "epoch": 1382.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4149 |
| }, |
| { |
| "epoch": 1383.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4152 |
| }, |
| { |
| "epoch": 1384.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4155 |
| }, |
| { |
| "epoch": 1385.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4158 |
| }, |
| { |
| "epoch": 1386.59, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5771034359931946, |
| "eval_runtime": 105.7502, |
| "eval_samples_per_second": 9.778, |
| "step": 4160 |
| }, |
| { |
| "epoch": 1386.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4161 |
| }, |
| { |
| "epoch": 1387.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4164 |
| }, |
| { |
| "epoch": 1388.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4167 |
| }, |
| { |
| "epoch": 1389.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4170 |
| }, |
| { |
| "epoch": 1390.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 4173 |
| }, |
| { |
| "epoch": 1391.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4176 |
| }, |
| { |
| "epoch": 1392.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4179 |
| }, |
| { |
| "epoch": 1393.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4182 |
| }, |
| { |
| "epoch": 1394.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4185 |
| }, |
| { |
| "epoch": 1395.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4188 |
| }, |
| { |
| "epoch": 1396.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4191 |
| }, |
| { |
| "epoch": 1397.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 4194 |
| }, |
| { |
| "epoch": 1398.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4197 |
| }, |
| { |
| "epoch": 1399.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1400.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4203 |
| }, |
| { |
| "epoch": 1401.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4206 |
| }, |
| { |
| "epoch": 1402.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4209 |
| }, |
| { |
| "epoch": 1403.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4212 |
| }, |
| { |
| "epoch": 1404.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4215 |
| }, |
| { |
| "epoch": 1405.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4218 |
| }, |
| { |
| "epoch": 1406.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4221 |
| }, |
| { |
| "epoch": 1407.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4224 |
| }, |
| { |
| "epoch": 1407.88, |
| "eval_exact_match": 0.4922630560928433, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.5785483717918396, |
| "eval_runtime": 117.2042, |
| "eval_samples_per_second": 8.822, |
| "step": 4224 |
| }, |
| { |
| "epoch": 1408.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4227 |
| }, |
| { |
| "epoch": 1409.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4230 |
| }, |
| { |
| "epoch": 1410.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4233 |
| }, |
| { |
| "epoch": 1411.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4236 |
| }, |
| { |
| "epoch": 1412.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4239 |
| }, |
| { |
| "epoch": 1413.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4242 |
| }, |
| { |
| "epoch": 1414.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4245 |
| }, |
| { |
| "epoch": 1415.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4248 |
| }, |
| { |
| "epoch": 1416.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4251 |
| }, |
| { |
| "epoch": 1417.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4254 |
| }, |
| { |
| "epoch": 1418.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4257 |
| }, |
| { |
| "epoch": 1419.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4260 |
| }, |
| { |
| "epoch": 1420.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4263 |
| }, |
| { |
| "epoch": 1421.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4266 |
| }, |
| { |
| "epoch": 1422.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4269 |
| }, |
| { |
| "epoch": 1423.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4272 |
| }, |
| { |
| "epoch": 1424.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4275 |
| }, |
| { |
| "epoch": 1425.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4278 |
| }, |
| { |
| "epoch": 1426.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4281 |
| }, |
| { |
| "epoch": 1427.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4284 |
| }, |
| { |
| "epoch": 1428.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4287 |
| }, |
| { |
| "epoch": 1429.29, |
| "eval_exact_match": 0.4874274661508704, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.5841559171676636, |
| "eval_runtime": 96.7174, |
| "eval_samples_per_second": 10.691, |
| "step": 4288 |
| }, |
| { |
| "epoch": 1429.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4290 |
| }, |
| { |
| "epoch": 1430.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4293 |
| }, |
| { |
| "epoch": 1431.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4296 |
| }, |
| { |
| "epoch": 1432.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4299 |
| }, |
| { |
| "epoch": 1433.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4302 |
| }, |
| { |
| "epoch": 1434.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4305 |
| }, |
| { |
| "epoch": 1435.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4308 |
| }, |
| { |
| "epoch": 1436.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4311 |
| }, |
| { |
| "epoch": 1437.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4314 |
| }, |
| { |
| "epoch": 1438.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4317 |
| }, |
| { |
| "epoch": 1439.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4320 |
| }, |
| { |
| "epoch": 1440.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4323 |
| }, |
| { |
| "epoch": 1441.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4326 |
| }, |
| { |
| "epoch": 1442.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4329 |
| }, |
| { |
| "epoch": 1443.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4332 |
| }, |
| { |
| "epoch": 1444.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4335 |
| }, |
| { |
| "epoch": 1445.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 4338 |
| }, |
| { |
| "epoch": 1446.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4341 |
| }, |
| { |
| "epoch": 1447.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4344 |
| }, |
| { |
| "epoch": 1448.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4347 |
| }, |
| { |
| "epoch": 1449.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4350 |
| }, |
| { |
| "epoch": 1450.59, |
| "eval_exact_match": 0.49709864603481624, |
| "eval_exec": 0.49806576402321084, |
| "eval_loss": 0.5847323536872864, |
| "eval_runtime": 104.9047, |
| "eval_samples_per_second": 9.857, |
| "step": 4352 |
| }, |
| { |
| "epoch": 1450.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4353 |
| }, |
| { |
| "epoch": 1451.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4356 |
| }, |
| { |
| "epoch": 1452.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4359 |
| }, |
| { |
| "epoch": 1453.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4362 |
| }, |
| { |
| "epoch": 1454.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4365 |
| }, |
| { |
| "epoch": 1455.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4368 |
| }, |
| { |
| "epoch": 1456.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4371 |
| }, |
| { |
| "epoch": 1457.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4374 |
| }, |
| { |
| "epoch": 1458.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4377 |
| }, |
| { |
| "epoch": 1459.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4380 |
| }, |
| { |
| "epoch": 1460.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4383 |
| }, |
| { |
| "epoch": 1461.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4386 |
| }, |
| { |
| "epoch": 1462.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4389 |
| }, |
| { |
| "epoch": 1463.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4392 |
| }, |
| { |
| "epoch": 1464.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4395 |
| }, |
| { |
| "epoch": 1465.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4398 |
| }, |
| { |
| "epoch": 1466.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4401 |
| }, |
| { |
| "epoch": 1467.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4404 |
| }, |
| { |
| "epoch": 1468.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4407 |
| }, |
| { |
| "epoch": 1469.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4410 |
| }, |
| { |
| "epoch": 1470.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 4413 |
| }, |
| { |
| "epoch": 1471.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4416 |
| }, |
| { |
| "epoch": 1471.88, |
| "eval_exact_match": 0.48452611218568664, |
| "eval_exec": 0.48452611218568664, |
| "eval_loss": 0.590592622756958, |
| "eval_runtime": 105.9184, |
| "eval_samples_per_second": 9.762, |
| "step": 4416 |
| }, |
| { |
| "epoch": 1472.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4419 |
| }, |
| { |
| "epoch": 1473.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4422 |
| }, |
| { |
| "epoch": 1474.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4425 |
| }, |
| { |
| "epoch": 1475.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4428 |
| }, |
| { |
| "epoch": 1476.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4431 |
| }, |
| { |
| "epoch": 1477.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4434 |
| }, |
| { |
| "epoch": 1478.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4437 |
| }, |
| { |
| "epoch": 1479.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4440 |
| }, |
| { |
| "epoch": 1480.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4443 |
| }, |
| { |
| "epoch": 1481.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4446 |
| }, |
| { |
| "epoch": 1482.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4449 |
| }, |
| { |
| "epoch": 1483.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4452 |
| }, |
| { |
| "epoch": 1484.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4455 |
| }, |
| { |
| "epoch": 1485.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4458 |
| }, |
| { |
| "epoch": 1486.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4461 |
| }, |
| { |
| "epoch": 1487.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4464 |
| }, |
| { |
| "epoch": 1488.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4467 |
| }, |
| { |
| "epoch": 1489.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4470 |
| }, |
| { |
| "epoch": 1490.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4473 |
| }, |
| { |
| "epoch": 1491.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4476 |
| }, |
| { |
| "epoch": 1492.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4479 |
| }, |
| { |
| "epoch": 1493.29, |
| "eval_exact_match": 0.48549323017408125, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.5873758792877197, |
| "eval_runtime": 99.7821, |
| "eval_samples_per_second": 10.363, |
| "step": 4480 |
| }, |
| { |
| "epoch": 1493.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4482 |
| }, |
| { |
| "epoch": 1494.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4485 |
| }, |
| { |
| "epoch": 1495.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4488 |
| }, |
| { |
| "epoch": 1496.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 4491 |
| }, |
| { |
| "epoch": 1497.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4494 |
| }, |
| { |
| "epoch": 1498.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4497 |
| }, |
| { |
| "epoch": 1499.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1500.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4503 |
| }, |
| { |
| "epoch": 1501.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4506 |
| }, |
| { |
| "epoch": 1502.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4509 |
| }, |
| { |
| "epoch": 1503.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4512 |
| }, |
| { |
| "epoch": 1504.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4515 |
| }, |
| { |
| "epoch": 1505.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4518 |
| }, |
| { |
| "epoch": 1506.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4521 |
| }, |
| { |
| "epoch": 1507.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4524 |
| }, |
| { |
| "epoch": 1508.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4527 |
| }, |
| { |
| "epoch": 1509.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4530 |
| }, |
| { |
| "epoch": 1510.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4533 |
| }, |
| { |
| "epoch": 1511.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4536 |
| }, |
| { |
| "epoch": 1512.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4539 |
| }, |
| { |
| "epoch": 1513.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4542 |
| }, |
| { |
| "epoch": 1514.59, |
| "eval_exact_match": 0.4816247582205029, |
| "eval_exec": 0.4796905222437137, |
| "eval_loss": 0.603107750415802, |
| "eval_runtime": 100.2594, |
| "eval_samples_per_second": 10.313, |
| "step": 4544 |
| }, |
| { |
| "epoch": 1514.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4545 |
| }, |
| { |
| "epoch": 1515.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4548 |
| }, |
| { |
| "epoch": 1516.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4551 |
| }, |
| { |
| "epoch": 1517.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4554 |
| }, |
| { |
| "epoch": 1518.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4557 |
| }, |
| { |
| "epoch": 1519.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4560 |
| }, |
| { |
| "epoch": 1520.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4563 |
| }, |
| { |
| "epoch": 1521.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4566 |
| }, |
| { |
| "epoch": 1522.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4569 |
| }, |
| { |
| "epoch": 1523.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4572 |
| }, |
| { |
| "epoch": 1524.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4575 |
| }, |
| { |
| "epoch": 1525.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4578 |
| }, |
| { |
| "epoch": 1526.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4581 |
| }, |
| { |
| "epoch": 1527.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4584 |
| }, |
| { |
| "epoch": 1528.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4587 |
| }, |
| { |
| "epoch": 1529.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 4590 |
| }, |
| { |
| "epoch": 1530.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4593 |
| }, |
| { |
| "epoch": 1531.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4596 |
| }, |
| { |
| "epoch": 1532.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4599 |
| }, |
| { |
| "epoch": 1533.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4602 |
| }, |
| { |
| "epoch": 1534.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4605 |
| }, |
| { |
| "epoch": 1535.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4608 |
| }, |
| { |
| "epoch": 1535.88, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.4932301740812379, |
| "eval_loss": 0.5945437550544739, |
| "eval_runtime": 131.4626, |
| "eval_samples_per_second": 7.865, |
| "step": 4608 |
| }, |
| { |
| "epoch": 1536.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4611 |
| }, |
| { |
| "epoch": 1537.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4614 |
| }, |
| { |
| "epoch": 1538.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4617 |
| }, |
| { |
| "epoch": 1539.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4620 |
| }, |
| { |
| "epoch": 1540.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4623 |
| }, |
| { |
| "epoch": 1541.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 4626 |
| }, |
| { |
| "epoch": 1542.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4629 |
| }, |
| { |
| "epoch": 1543.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4632 |
| }, |
| { |
| "epoch": 1544.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4635 |
| }, |
| { |
| "epoch": 1545.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4638 |
| }, |
| { |
| "epoch": 1546.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4641 |
| }, |
| { |
| "epoch": 1547.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4644 |
| }, |
| { |
| "epoch": 1548.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4647 |
| }, |
| { |
| "epoch": 1549.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4650 |
| }, |
| { |
| "epoch": 1550.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4653 |
| }, |
| { |
| "epoch": 1551.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4656 |
| }, |
| { |
| "epoch": 1552.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4659 |
| }, |
| { |
| "epoch": 1553.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4662 |
| }, |
| { |
| "epoch": 1554.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4665 |
| }, |
| { |
| "epoch": 1555.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4668 |
| }, |
| { |
| "epoch": 1556.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4671 |
| }, |
| { |
| "epoch": 1557.29, |
| "eval_exact_match": 0.4922630560928433, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.5930950045585632, |
| "eval_runtime": 99.7074, |
| "eval_samples_per_second": 10.37, |
| "step": 4672 |
| }, |
| { |
| "epoch": 1557.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4674 |
| }, |
| { |
| "epoch": 1558.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4677 |
| }, |
| { |
| "epoch": 1559.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 4680 |
| }, |
| { |
| "epoch": 1560.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4683 |
| }, |
| { |
| "epoch": 1561.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4686 |
| }, |
| { |
| "epoch": 1562.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4689 |
| }, |
| { |
| "epoch": 1563.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4692 |
| }, |
| { |
| "epoch": 1564.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4695 |
| }, |
| { |
| "epoch": 1565.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4698 |
| }, |
| { |
| "epoch": 1566.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4701 |
| }, |
| { |
| "epoch": 1567.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4704 |
| }, |
| { |
| "epoch": 1568.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4707 |
| }, |
| { |
| "epoch": 1569.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4710 |
| }, |
| { |
| "epoch": 1570.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4713 |
| }, |
| { |
| "epoch": 1571.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4716 |
| }, |
| { |
| "epoch": 1572.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4719 |
| }, |
| { |
| "epoch": 1573.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4722 |
| }, |
| { |
| "epoch": 1574.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4725 |
| }, |
| { |
| "epoch": 1575.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4728 |
| }, |
| { |
| "epoch": 1576.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4731 |
| }, |
| { |
| "epoch": 1577.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4734 |
| }, |
| { |
| "epoch": 1578.59, |
| "eval_exact_match": 0.5, |
| "eval_exec": 0.5009671179883946, |
| "eval_loss": 0.6037657856941223, |
| "eval_runtime": 102.1529, |
| "eval_samples_per_second": 10.122, |
| "step": 4736 |
| }, |
| { |
| "epoch": 1578.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4737 |
| }, |
| { |
| "epoch": 1579.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4740 |
| }, |
| { |
| "epoch": 1580.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4743 |
| }, |
| { |
| "epoch": 1581.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4746 |
| }, |
| { |
| "epoch": 1582.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4749 |
| }, |
| { |
| "epoch": 1583.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4752 |
| }, |
| { |
| "epoch": 1584.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4755 |
| }, |
| { |
| "epoch": 1585.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4758 |
| }, |
| { |
| "epoch": 1586.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4761 |
| }, |
| { |
| "epoch": 1587.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4764 |
| }, |
| { |
| "epoch": 1588.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4767 |
| }, |
| { |
| "epoch": 1589.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4770 |
| }, |
| { |
| "epoch": 1590.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4773 |
| }, |
| { |
| "epoch": 1591.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4776 |
| }, |
| { |
| "epoch": 1592.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4779 |
| }, |
| { |
| "epoch": 1593.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 4782 |
| }, |
| { |
| "epoch": 1594.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4785 |
| }, |
| { |
| "epoch": 1595.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4788 |
| }, |
| { |
| "epoch": 1596.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4791 |
| }, |
| { |
| "epoch": 1597.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4794 |
| }, |
| { |
| "epoch": 1598.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4797 |
| }, |
| { |
| "epoch": 1599.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1599.88, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.5997296571731567, |
| "eval_runtime": 100.5407, |
| "eval_samples_per_second": 10.284, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1600.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4803 |
| }, |
| { |
| "epoch": 1601.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4806 |
| }, |
| { |
| "epoch": 1602.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4809 |
| }, |
| { |
| "epoch": 1603.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4812 |
| }, |
| { |
| "epoch": 1604.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4815 |
| }, |
| { |
| "epoch": 1605.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4818 |
| }, |
| { |
| "epoch": 1606.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4821 |
| }, |
| { |
| "epoch": 1607.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4824 |
| }, |
| { |
| "epoch": 1608.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4827 |
| }, |
| { |
| "epoch": 1609.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4830 |
| }, |
| { |
| "epoch": 1610.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4833 |
| }, |
| { |
| "epoch": 1611.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4836 |
| }, |
| { |
| "epoch": 1612.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4839 |
| }, |
| { |
| "epoch": 1613.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4842 |
| }, |
| { |
| "epoch": 1614.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 4845 |
| }, |
| { |
| "epoch": 1615.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4848 |
| }, |
| { |
| "epoch": 1616.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4851 |
| }, |
| { |
| "epoch": 1617.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4854 |
| }, |
| { |
| "epoch": 1618.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4857 |
| }, |
| { |
| "epoch": 1619.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4860 |
| }, |
| { |
| "epoch": 1620.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4863 |
| }, |
| { |
| "epoch": 1621.29, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.4932301740812379, |
| "eval_loss": 0.6005607843399048, |
| "eval_runtime": 103.9716, |
| "eval_samples_per_second": 9.945, |
| "step": 4864 |
| }, |
| { |
| "epoch": 1621.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4866 |
| }, |
| { |
| "epoch": 1622.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4869 |
| }, |
| { |
| "epoch": 1623.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4872 |
| }, |
| { |
| "epoch": 1624.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 4875 |
| }, |
| { |
| "epoch": 1625.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4878 |
| }, |
| { |
| "epoch": 1626.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4881 |
| }, |
| { |
| "epoch": 1627.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4884 |
| }, |
| { |
| "epoch": 1628.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4887 |
| }, |
| { |
| "epoch": 1629.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4890 |
| }, |
| { |
| "epoch": 1630.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4893 |
| }, |
| { |
| "epoch": 1631.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4896 |
| }, |
| { |
| "epoch": 1632.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4899 |
| }, |
| { |
| "epoch": 1633.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4902 |
| }, |
| { |
| "epoch": 1634.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4905 |
| }, |
| { |
| "epoch": 1635.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4908 |
| }, |
| { |
| "epoch": 1636.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4911 |
| }, |
| { |
| "epoch": 1637.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4914 |
| }, |
| { |
| "epoch": 1638.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4917 |
| }, |
| { |
| "epoch": 1639.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4920 |
| }, |
| { |
| "epoch": 1640.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4923 |
| }, |
| { |
| "epoch": 1641.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4926 |
| }, |
| { |
| "epoch": 1642.59, |
| "eval_exact_match": 0.4961315280464217, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.6110011339187622, |
| "eval_runtime": 100.0167, |
| "eval_samples_per_second": 10.338, |
| "step": 4928 |
| }, |
| { |
| "epoch": 1642.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4929 |
| }, |
| { |
| "epoch": 1643.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4932 |
| }, |
| { |
| "epoch": 1644.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4935 |
| }, |
| { |
| "epoch": 1645.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4938 |
| }, |
| { |
| "epoch": 1646.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 4941 |
| }, |
| { |
| "epoch": 1647.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4944 |
| }, |
| { |
| "epoch": 1648.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4947 |
| }, |
| { |
| "epoch": 1649.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4950 |
| }, |
| { |
| "epoch": 1650.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4953 |
| }, |
| { |
| "epoch": 1651.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4956 |
| }, |
| { |
| "epoch": 1652.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 4959 |
| }, |
| { |
| "epoch": 1653.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 4962 |
| }, |
| { |
| "epoch": 1654.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4965 |
| }, |
| { |
| "epoch": 1655.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4968 |
| }, |
| { |
| "epoch": 1656.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4971 |
| }, |
| { |
| "epoch": 1657.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 4974 |
| }, |
| { |
| "epoch": 1658.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4977 |
| }, |
| { |
| "epoch": 1659.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 4980 |
| }, |
| { |
| "epoch": 1660.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4983 |
| }, |
| { |
| "epoch": 1661.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4986 |
| }, |
| { |
| "epoch": 1662.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 4989 |
| }, |
| { |
| "epoch": 1663.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4992 |
| }, |
| { |
| "epoch": 1663.88, |
| "eval_exact_match": 0.5, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.6110185980796814, |
| "eval_runtime": 102.0442, |
| "eval_samples_per_second": 10.133, |
| "step": 4992 |
| }, |
| { |
| "epoch": 1664.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4995 |
| }, |
| { |
| "epoch": 1665.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 4998 |
| }, |
| { |
| "epoch": 1666.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5001 |
| }, |
| { |
| "epoch": 1667.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5004 |
| }, |
| { |
| "epoch": 1668.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5007 |
| }, |
| { |
| "epoch": 1669.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5010 |
| }, |
| { |
| "epoch": 1670.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5013 |
| }, |
| { |
| "epoch": 1671.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5016 |
| }, |
| { |
| "epoch": 1672.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5019 |
| }, |
| { |
| "epoch": 1673.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5022 |
| }, |
| { |
| "epoch": 1674.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5025 |
| }, |
| { |
| "epoch": 1675.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5028 |
| }, |
| { |
| "epoch": 1676.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 5031 |
| }, |
| { |
| "epoch": 1677.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5034 |
| }, |
| { |
| "epoch": 1678.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5037 |
| }, |
| { |
| "epoch": 1679.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5040 |
| }, |
| { |
| "epoch": 1680.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5043 |
| }, |
| { |
| "epoch": 1681.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5046 |
| }, |
| { |
| "epoch": 1682.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5049 |
| }, |
| { |
| "epoch": 1683.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5052 |
| }, |
| { |
| "epoch": 1684.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5055 |
| }, |
| { |
| "epoch": 1685.29, |
| "eval_exact_match": 0.4825918762088975, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.6093866229057312, |
| "eval_runtime": 103.7246, |
| "eval_samples_per_second": 9.969, |
| "step": 5056 |
| }, |
| { |
| "epoch": 1685.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5058 |
| }, |
| { |
| "epoch": 1686.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5061 |
| }, |
| { |
| "epoch": 1687.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5064 |
| }, |
| { |
| "epoch": 1688.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5067 |
| }, |
| { |
| "epoch": 1689.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5070 |
| }, |
| { |
| "epoch": 1690.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5073 |
| }, |
| { |
| "epoch": 1691.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5076 |
| }, |
| { |
| "epoch": 1692.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5079 |
| }, |
| { |
| "epoch": 1693.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5082 |
| }, |
| { |
| "epoch": 1694.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5085 |
| }, |
| { |
| "epoch": 1695.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5088 |
| }, |
| { |
| "epoch": 1696.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5091 |
| }, |
| { |
| "epoch": 1697.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5094 |
| }, |
| { |
| "epoch": 1698.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5097 |
| }, |
| { |
| "epoch": 1699.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5100 |
| }, |
| { |
| "epoch": 1700.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5103 |
| }, |
| { |
| "epoch": 1701.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5106 |
| }, |
| { |
| "epoch": 1702.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5109 |
| }, |
| { |
| "epoch": 1703.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5112 |
| }, |
| { |
| "epoch": 1704.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5115 |
| }, |
| { |
| "epoch": 1705.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5118 |
| }, |
| { |
| "epoch": 1706.59, |
| "eval_exact_match": 0.488394584139265, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.6172391176223755, |
| "eval_runtime": 100.9408, |
| "eval_samples_per_second": 10.244, |
| "step": 5120 |
| }, |
| { |
| "epoch": 1706.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5121 |
| }, |
| { |
| "epoch": 1707.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5124 |
| }, |
| { |
| "epoch": 1708.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5127 |
| }, |
| { |
| "epoch": 1709.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 5130 |
| }, |
| { |
| "epoch": 1710.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5133 |
| }, |
| { |
| "epoch": 1711.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5136 |
| }, |
| { |
| "epoch": 1712.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5139 |
| }, |
| { |
| "epoch": 1713.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5142 |
| }, |
| { |
| "epoch": 1714.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5145 |
| }, |
| { |
| "epoch": 1715.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5148 |
| }, |
| { |
| "epoch": 1716.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 5151 |
| }, |
| { |
| "epoch": 1717.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5154 |
| }, |
| { |
| "epoch": 1718.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5157 |
| }, |
| { |
| "epoch": 1719.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5160 |
| }, |
| { |
| "epoch": 1720.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5163 |
| }, |
| { |
| "epoch": 1721.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5166 |
| }, |
| { |
| "epoch": 1722.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5169 |
| }, |
| { |
| "epoch": 1723.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5172 |
| }, |
| { |
| "epoch": 1724.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5175 |
| }, |
| { |
| "epoch": 1725.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5178 |
| }, |
| { |
| "epoch": 1726.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5181 |
| }, |
| { |
| "epoch": 1727.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5184 |
| }, |
| { |
| "epoch": 1727.88, |
| "eval_exact_match": 0.49032882011605416, |
| "eval_exec": 0.48452611218568664, |
| "eval_loss": 0.6174488663673401, |
| "eval_runtime": 103.1102, |
| "eval_samples_per_second": 10.028, |
| "step": 5184 |
| }, |
| { |
| "epoch": 1728.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 5187 |
| }, |
| { |
| "epoch": 1729.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5190 |
| }, |
| { |
| "epoch": 1730.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5193 |
| }, |
| { |
| "epoch": 1731.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5196 |
| }, |
| { |
| "epoch": 1732.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5199 |
| }, |
| { |
| "epoch": 1733.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5202 |
| }, |
| { |
| "epoch": 1734.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5205 |
| }, |
| { |
| "epoch": 1735.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5208 |
| }, |
| { |
| "epoch": 1736.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5211 |
| }, |
| { |
| "epoch": 1737.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5214 |
| }, |
| { |
| "epoch": 1738.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5217 |
| }, |
| { |
| "epoch": 1739.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5220 |
| }, |
| { |
| "epoch": 1740.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5223 |
| }, |
| { |
| "epoch": 1741.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5226 |
| }, |
| { |
| "epoch": 1742.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5229 |
| }, |
| { |
| "epoch": 1743.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5232 |
| }, |
| { |
| "epoch": 1744.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5235 |
| }, |
| { |
| "epoch": 1745.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5238 |
| }, |
| { |
| "epoch": 1746.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5241 |
| }, |
| { |
| "epoch": 1747.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5244 |
| }, |
| { |
| "epoch": 1748.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5247 |
| }, |
| { |
| "epoch": 1749.29, |
| "eval_exact_match": 0.4825918762088975, |
| "eval_exec": 0.4816247582205029, |
| "eval_loss": 0.6244661211967468, |
| "eval_runtime": 105.1481, |
| "eval_samples_per_second": 9.834, |
| "step": 5248 |
| }, |
| { |
| "epoch": 1749.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5250 |
| }, |
| { |
| "epoch": 1750.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5253 |
| }, |
| { |
| "epoch": 1751.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5256 |
| }, |
| { |
| "epoch": 1752.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 5259 |
| }, |
| { |
| "epoch": 1753.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5262 |
| }, |
| { |
| "epoch": 1754.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5265 |
| }, |
| { |
| "epoch": 1755.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5268 |
| }, |
| { |
| "epoch": 1756.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5271 |
| }, |
| { |
| "epoch": 1757.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5274 |
| }, |
| { |
| "epoch": 1758.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5277 |
| }, |
| { |
| "epoch": 1759.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5280 |
| }, |
| { |
| "epoch": 1760.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 5283 |
| }, |
| { |
| "epoch": 1761.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5286 |
| }, |
| { |
| "epoch": 1762.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5289 |
| }, |
| { |
| "epoch": 1763.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5292 |
| }, |
| { |
| "epoch": 1764.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5295 |
| }, |
| { |
| "epoch": 1765.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5298 |
| }, |
| { |
| "epoch": 1766.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5301 |
| }, |
| { |
| "epoch": 1767.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5304 |
| }, |
| { |
| "epoch": 1768.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5307 |
| }, |
| { |
| "epoch": 1769.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5310 |
| }, |
| { |
| "epoch": 1770.59, |
| "eval_exact_match": 0.48936170212765956, |
| "eval_exec": 0.4864603481624758, |
| "eval_loss": 0.6090367436408997, |
| "eval_runtime": 104.7443, |
| "eval_samples_per_second": 9.872, |
| "step": 5312 |
| }, |
| { |
| "epoch": 1770.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5313 |
| }, |
| { |
| "epoch": 1771.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5316 |
| }, |
| { |
| "epoch": 1772.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5319 |
| }, |
| { |
| "epoch": 1773.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5322 |
| }, |
| { |
| "epoch": 1774.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5325 |
| }, |
| { |
| "epoch": 1775.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5328 |
| }, |
| { |
| "epoch": 1776.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5331 |
| }, |
| { |
| "epoch": 1777.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5334 |
| }, |
| { |
| "epoch": 1778.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5337 |
| }, |
| { |
| "epoch": 1779.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5340 |
| }, |
| { |
| "epoch": 1780.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5343 |
| }, |
| { |
| "epoch": 1781.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5346 |
| }, |
| { |
| "epoch": 1782.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5349 |
| }, |
| { |
| "epoch": 1783.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5352 |
| }, |
| { |
| "epoch": 1784.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5355 |
| }, |
| { |
| "epoch": 1785.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5358 |
| }, |
| { |
| "epoch": 1786.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5361 |
| }, |
| { |
| "epoch": 1787.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5364 |
| }, |
| { |
| "epoch": 1788.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5367 |
| }, |
| { |
| "epoch": 1789.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5370 |
| }, |
| { |
| "epoch": 1790.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5373 |
| }, |
| { |
| "epoch": 1791.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5376 |
| }, |
| { |
| "epoch": 1791.88, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.6232331395149231, |
| "eval_runtime": 102.2321, |
| "eval_samples_per_second": 10.114, |
| "step": 5376 |
| }, |
| { |
| "epoch": 1792.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5379 |
| }, |
| { |
| "epoch": 1793.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5382 |
| }, |
| { |
| "epoch": 1794.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5385 |
| }, |
| { |
| "epoch": 1795.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5388 |
| }, |
| { |
| "epoch": 1796.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5391 |
| }, |
| { |
| "epoch": 1797.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5394 |
| }, |
| { |
| "epoch": 1798.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 5397 |
| }, |
| { |
| "epoch": 1799.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5400 |
| }, |
| { |
| "epoch": 1800.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5403 |
| }, |
| { |
| "epoch": 1801.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5406 |
| }, |
| { |
| "epoch": 1802.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5409 |
| }, |
| { |
| "epoch": 1803.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5412 |
| }, |
| { |
| "epoch": 1804.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5415 |
| }, |
| { |
| "epoch": 1805.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5418 |
| }, |
| { |
| "epoch": 1806.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5421 |
| }, |
| { |
| "epoch": 1807.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5424 |
| }, |
| { |
| "epoch": 1808.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5427 |
| }, |
| { |
| "epoch": 1809.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5430 |
| }, |
| { |
| "epoch": 1810.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5433 |
| }, |
| { |
| "epoch": 1811.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5436 |
| }, |
| { |
| "epoch": 1812.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5439 |
| }, |
| { |
| "epoch": 1813.29, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.6267974376678467, |
| "eval_runtime": 100.3635, |
| "eval_samples_per_second": 10.303, |
| "step": 5440 |
| }, |
| { |
| "epoch": 1813.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5442 |
| }, |
| { |
| "epoch": 1814.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5445 |
| }, |
| { |
| "epoch": 1815.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5448 |
| }, |
| { |
| "epoch": 1816.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5451 |
| }, |
| { |
| "epoch": 1817.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5454 |
| }, |
| { |
| "epoch": 1818.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5457 |
| }, |
| { |
| "epoch": 1819.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5460 |
| }, |
| { |
| "epoch": 1820.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5463 |
| }, |
| { |
| "epoch": 1821.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5466 |
| }, |
| { |
| "epoch": 1822.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5469 |
| }, |
| { |
| "epoch": 1823.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5472 |
| }, |
| { |
| "epoch": 1824.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5475 |
| }, |
| { |
| "epoch": 1825.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5478 |
| }, |
| { |
| "epoch": 1826.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5481 |
| }, |
| { |
| "epoch": 1827.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5484 |
| }, |
| { |
| "epoch": 1828.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5487 |
| }, |
| { |
| "epoch": 1829.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5490 |
| }, |
| { |
| "epoch": 1830.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5493 |
| }, |
| { |
| "epoch": 1831.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5496 |
| }, |
| { |
| "epoch": 1832.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5499 |
| }, |
| { |
| "epoch": 1833.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5502 |
| }, |
| { |
| "epoch": 1834.59, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.6204285025596619, |
| "eval_runtime": 99.4678, |
| "eval_samples_per_second": 10.395, |
| "step": 5504 |
| }, |
| { |
| "epoch": 1834.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5505 |
| }, |
| { |
| "epoch": 1835.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5508 |
| }, |
| { |
| "epoch": 1836.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5511 |
| }, |
| { |
| "epoch": 1837.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5514 |
| }, |
| { |
| "epoch": 1838.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5517 |
| }, |
| { |
| "epoch": 1839.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5520 |
| }, |
| { |
| "epoch": 1840.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5523 |
| }, |
| { |
| "epoch": 1841.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5526 |
| }, |
| { |
| "epoch": 1842.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5529 |
| }, |
| { |
| "epoch": 1843.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5532 |
| }, |
| { |
| "epoch": 1844.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5535 |
| }, |
| { |
| "epoch": 1845.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5538 |
| }, |
| { |
| "epoch": 1846.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5541 |
| }, |
| { |
| "epoch": 1847.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5544 |
| }, |
| { |
| "epoch": 1848.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5547 |
| }, |
| { |
| "epoch": 1849.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5550 |
| }, |
| { |
| "epoch": 1850.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5553 |
| }, |
| { |
| "epoch": 1851.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5556 |
| }, |
| { |
| "epoch": 1852.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5559 |
| }, |
| { |
| "epoch": 1853.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5562 |
| }, |
| { |
| "epoch": 1854.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5565 |
| }, |
| { |
| "epoch": 1855.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5568 |
| }, |
| { |
| "epoch": 1855.88, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.4951644100580271, |
| "eval_loss": 0.6250240802764893, |
| "eval_runtime": 98.8362, |
| "eval_samples_per_second": 10.462, |
| "step": 5568 |
| }, |
| { |
| "epoch": 1856.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5571 |
| }, |
| { |
| "epoch": 1857.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5574 |
| }, |
| { |
| "epoch": 1858.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5577 |
| }, |
| { |
| "epoch": 1859.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5580 |
| }, |
| { |
| "epoch": 1860.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5583 |
| }, |
| { |
| "epoch": 1861.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5586 |
| }, |
| { |
| "epoch": 1862.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5589 |
| }, |
| { |
| "epoch": 1863.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5592 |
| }, |
| { |
| "epoch": 1864.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5595 |
| }, |
| { |
| "epoch": 1865.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5598 |
| }, |
| { |
| "epoch": 1866.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5601 |
| }, |
| { |
| "epoch": 1867.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5604 |
| }, |
| { |
| "epoch": 1868.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5607 |
| }, |
| { |
| "epoch": 1869.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5610 |
| }, |
| { |
| "epoch": 1870.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5613 |
| }, |
| { |
| "epoch": 1871.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5616 |
| }, |
| { |
| "epoch": 1872.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 5619 |
| }, |
| { |
| "epoch": 1873.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5622 |
| }, |
| { |
| "epoch": 1874.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5625 |
| }, |
| { |
| "epoch": 1875.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5628 |
| }, |
| { |
| "epoch": 1876.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5631 |
| }, |
| { |
| "epoch": 1877.29, |
| "eval_exact_match": 0.504835589941973, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.6261150240898132, |
| "eval_runtime": 102.3617, |
| "eval_samples_per_second": 10.101, |
| "step": 5632 |
| }, |
| { |
| "epoch": 1877.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5634 |
| }, |
| { |
| "epoch": 1878.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5637 |
| }, |
| { |
| "epoch": 1879.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5640 |
| }, |
| { |
| "epoch": 1880.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5643 |
| }, |
| { |
| "epoch": 1881.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5646 |
| }, |
| { |
| "epoch": 1882.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5649 |
| }, |
| { |
| "epoch": 1883.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5652 |
| }, |
| { |
| "epoch": 1884.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5655 |
| }, |
| { |
| "epoch": 1885.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5658 |
| }, |
| { |
| "epoch": 1886.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5661 |
| }, |
| { |
| "epoch": 1887.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5664 |
| }, |
| { |
| "epoch": 1888.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5667 |
| }, |
| { |
| "epoch": 1889.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5670 |
| }, |
| { |
| "epoch": 1890.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5673 |
| }, |
| { |
| "epoch": 1891.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5676 |
| }, |
| { |
| "epoch": 1892.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5679 |
| }, |
| { |
| "epoch": 1893.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5682 |
| }, |
| { |
| "epoch": 1894.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5685 |
| }, |
| { |
| "epoch": 1895.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5688 |
| }, |
| { |
| "epoch": 1896.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5691 |
| }, |
| { |
| "epoch": 1897.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5694 |
| }, |
| { |
| "epoch": 1898.59, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.48549323017408125, |
| "eval_loss": 0.6312097907066345, |
| "eval_runtime": 102.3467, |
| "eval_samples_per_second": 10.103, |
| "step": 5696 |
| }, |
| { |
| "epoch": 1898.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5697 |
| }, |
| { |
| "epoch": 1899.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5700 |
| }, |
| { |
| "epoch": 1900.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5703 |
| }, |
| { |
| "epoch": 1901.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5706 |
| }, |
| { |
| "epoch": 1902.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5709 |
| }, |
| { |
| "epoch": 1903.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5712 |
| }, |
| { |
| "epoch": 1904.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5715 |
| }, |
| { |
| "epoch": 1905.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5718 |
| }, |
| { |
| "epoch": 1906.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5721 |
| }, |
| { |
| "epoch": 1907.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5724 |
| }, |
| { |
| "epoch": 1908.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5727 |
| }, |
| { |
| "epoch": 1909.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5730 |
| }, |
| { |
| "epoch": 1910.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5733 |
| }, |
| { |
| "epoch": 1911.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5736 |
| }, |
| { |
| "epoch": 1912.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5739 |
| }, |
| { |
| "epoch": 1913.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 5742 |
| }, |
| { |
| "epoch": 1914.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5745 |
| }, |
| { |
| "epoch": 1915.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5748 |
| }, |
| { |
| "epoch": 1916.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5751 |
| }, |
| { |
| "epoch": 1917.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5754 |
| }, |
| { |
| "epoch": 1918.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5757 |
| }, |
| { |
| "epoch": 1919.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5760 |
| }, |
| { |
| "epoch": 1919.88, |
| "eval_exact_match": 0.4932301740812379, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.6297084093093872, |
| "eval_runtime": 102.7251, |
| "eval_samples_per_second": 10.066, |
| "step": 5760 |
| }, |
| { |
| "epoch": 1920.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5763 |
| }, |
| { |
| "epoch": 1921.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5766 |
| }, |
| { |
| "epoch": 1922.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5769 |
| }, |
| { |
| "epoch": 1923.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5772 |
| }, |
| { |
| "epoch": 1924.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5775 |
| }, |
| { |
| "epoch": 1925.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5778 |
| }, |
| { |
| "epoch": 1926.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5781 |
| }, |
| { |
| "epoch": 1927.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5784 |
| }, |
| { |
| "epoch": 1928.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5787 |
| }, |
| { |
| "epoch": 1929.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5790 |
| }, |
| { |
| "epoch": 1930.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5793 |
| }, |
| { |
| "epoch": 1931.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5796 |
| }, |
| { |
| "epoch": 1932.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5799 |
| }, |
| { |
| "epoch": 1933.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5802 |
| }, |
| { |
| "epoch": 1934.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5805 |
| }, |
| { |
| "epoch": 1935.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5808 |
| }, |
| { |
| "epoch": 1936.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5811 |
| }, |
| { |
| "epoch": 1937.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5814 |
| }, |
| { |
| "epoch": 1938.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5817 |
| }, |
| { |
| "epoch": 1939.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5820 |
| }, |
| { |
| "epoch": 1940.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5823 |
| }, |
| { |
| "epoch": 1941.29, |
| "eval_exact_match": 0.5029013539651838, |
| "eval_exec": 0.49806576402321084, |
| "eval_loss": 0.6355636715888977, |
| "eval_runtime": 101.7003, |
| "eval_samples_per_second": 10.167, |
| "step": 5824 |
| }, |
| { |
| "epoch": 1941.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5826 |
| }, |
| { |
| "epoch": 1942.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 5829 |
| }, |
| { |
| "epoch": 1943.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5832 |
| }, |
| { |
| "epoch": 1944.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5835 |
| }, |
| { |
| "epoch": 1945.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5838 |
| }, |
| { |
| "epoch": 1946.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5841 |
| }, |
| { |
| "epoch": 1947.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5844 |
| }, |
| { |
| "epoch": 1948.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5847 |
| }, |
| { |
| "epoch": 1949.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5850 |
| }, |
| { |
| "epoch": 1950.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5853 |
| }, |
| { |
| "epoch": 1951.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5856 |
| }, |
| { |
| "epoch": 1952.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5859 |
| }, |
| { |
| "epoch": 1953.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5862 |
| }, |
| { |
| "epoch": 1954.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5865 |
| }, |
| { |
| "epoch": 1955.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5868 |
| }, |
| { |
| "epoch": 1956.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5871 |
| }, |
| { |
| "epoch": 1957.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5874 |
| }, |
| { |
| "epoch": 1958.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5877 |
| }, |
| { |
| "epoch": 1959.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5880 |
| }, |
| { |
| "epoch": 1960.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5883 |
| }, |
| { |
| "epoch": 1961.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5886 |
| }, |
| { |
| "epoch": 1962.59, |
| "eval_exact_match": 0.49709864603481624, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.6439603567123413, |
| "eval_runtime": 98.8258, |
| "eval_samples_per_second": 10.463, |
| "step": 5888 |
| }, |
| { |
| "epoch": 1962.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5889 |
| }, |
| { |
| "epoch": 1963.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5892 |
| }, |
| { |
| "epoch": 1964.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5895 |
| }, |
| { |
| "epoch": 1965.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5898 |
| }, |
| { |
| "epoch": 1966.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 5901 |
| }, |
| { |
| "epoch": 1967.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5904 |
| }, |
| { |
| "epoch": 1968.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5907 |
| }, |
| { |
| "epoch": 1969.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5910 |
| }, |
| { |
| "epoch": 1970.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5913 |
| }, |
| { |
| "epoch": 1971.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5916 |
| }, |
| { |
| "epoch": 1972.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5919 |
| }, |
| { |
| "epoch": 1973.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5922 |
| }, |
| { |
| "epoch": 1974.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5925 |
| }, |
| { |
| "epoch": 1975.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5928 |
| }, |
| { |
| "epoch": 1976.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5931 |
| }, |
| { |
| "epoch": 1977.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5934 |
| }, |
| { |
| "epoch": 1978.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5937 |
| }, |
| { |
| "epoch": 1979.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5940 |
| }, |
| { |
| "epoch": 1980.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5943 |
| }, |
| { |
| "epoch": 1981.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5946 |
| }, |
| { |
| "epoch": 1982.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 5949 |
| }, |
| { |
| "epoch": 1983.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5952 |
| }, |
| { |
| "epoch": 1983.88, |
| "eval_exact_match": 0.5019342359767892, |
| "eval_exec": 0.4932301740812379, |
| "eval_loss": 0.6456257104873657, |
| "eval_runtime": 103.1728, |
| "eval_samples_per_second": 10.022, |
| "step": 5952 |
| }, |
| { |
| "epoch": 1984.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 5955 |
| }, |
| { |
| "epoch": 1985.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5958 |
| }, |
| { |
| "epoch": 1986.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5961 |
| }, |
| { |
| "epoch": 1987.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5964 |
| }, |
| { |
| "epoch": 1988.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5967 |
| }, |
| { |
| "epoch": 1989.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5970 |
| }, |
| { |
| "epoch": 1990.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5973 |
| }, |
| { |
| "epoch": 1991.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5976 |
| }, |
| { |
| "epoch": 1992.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 5979 |
| }, |
| { |
| "epoch": 1993.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5982 |
| }, |
| { |
| "epoch": 1994.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 5985 |
| }, |
| { |
| "epoch": 1995.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5988 |
| }, |
| { |
| "epoch": 1996.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5991 |
| }, |
| { |
| "epoch": 1997.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 5994 |
| }, |
| { |
| "epoch": 1998.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 5997 |
| }, |
| { |
| "epoch": 1999.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6000 |
| }, |
| { |
| "epoch": 2000.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6003 |
| }, |
| { |
| "epoch": 2001.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6006 |
| }, |
| { |
| "epoch": 2002.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6009 |
| }, |
| { |
| "epoch": 2003.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6012 |
| }, |
| { |
| "epoch": 2004.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 6015 |
| }, |
| { |
| "epoch": 2005.29, |
| "eval_exact_match": 0.4990328820116054, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.6428639888763428, |
| "eval_runtime": 103.19, |
| "eval_samples_per_second": 10.02, |
| "step": 6016 |
| }, |
| { |
| "epoch": 2005.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6018 |
| }, |
| { |
| "epoch": 2006.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6021 |
| }, |
| { |
| "epoch": 2007.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6024 |
| }, |
| { |
| "epoch": 2008.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6027 |
| }, |
| { |
| "epoch": 2009.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6030 |
| }, |
| { |
| "epoch": 2010.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6033 |
| }, |
| { |
| "epoch": 2011.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6036 |
| }, |
| { |
| "epoch": 2012.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6039 |
| }, |
| { |
| "epoch": 2013.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6042 |
| }, |
| { |
| "epoch": 2014.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6045 |
| }, |
| { |
| "epoch": 2015.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6048 |
| }, |
| { |
| "epoch": 2016.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6051 |
| }, |
| { |
| "epoch": 2017.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6054 |
| }, |
| { |
| "epoch": 2018.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6057 |
| }, |
| { |
| "epoch": 2019.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6060 |
| }, |
| { |
| "epoch": 2020.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6063 |
| }, |
| { |
| "epoch": 2021.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6066 |
| }, |
| { |
| "epoch": 2022.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6069 |
| }, |
| { |
| "epoch": 2023.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6072 |
| }, |
| { |
| "epoch": 2024.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 6075 |
| }, |
| { |
| "epoch": 2025.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6078 |
| }, |
| { |
| "epoch": 2026.59, |
| "eval_exact_match": 0.5067698259187621, |
| "eval_exec": 0.4990328820116054, |
| "eval_loss": 0.6382647156715393, |
| "eval_runtime": 102.8692, |
| "eval_samples_per_second": 10.052, |
| "step": 6080 |
| }, |
| { |
| "epoch": 2026.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6081 |
| }, |
| { |
| "epoch": 2027.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6084 |
| }, |
| { |
| "epoch": 2028.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6087 |
| }, |
| { |
| "epoch": 2029.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6090 |
| }, |
| { |
| "epoch": 2030.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6093 |
| }, |
| { |
| "epoch": 2031.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6096 |
| }, |
| { |
| "epoch": 2032.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6099 |
| }, |
| { |
| "epoch": 2033.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6102 |
| }, |
| { |
| "epoch": 2034.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6105 |
| }, |
| { |
| "epoch": 2035.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6108 |
| }, |
| { |
| "epoch": 2036.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6111 |
| }, |
| { |
| "epoch": 2037.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6114 |
| }, |
| { |
| "epoch": 2038.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6117 |
| }, |
| { |
| "epoch": 2039.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6120 |
| }, |
| { |
| "epoch": 2040.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6123 |
| }, |
| { |
| "epoch": 2041.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6126 |
| }, |
| { |
| "epoch": 2042.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6129 |
| }, |
| { |
| "epoch": 2043.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6132 |
| }, |
| { |
| "epoch": 2044.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6135 |
| }, |
| { |
| "epoch": 2045.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6138 |
| }, |
| { |
| "epoch": 2046.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6141 |
| }, |
| { |
| "epoch": 2047.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6144 |
| }, |
| { |
| "epoch": 2047.88, |
| "eval_exact_match": 0.5009671179883946, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.6453941464424133, |
| "eval_runtime": 118.2755, |
| "eval_samples_per_second": 8.742, |
| "step": 6144 |
| }, |
| { |
| "epoch": 2048.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6147 |
| }, |
| { |
| "epoch": 2049.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6150 |
| }, |
| { |
| "epoch": 2050.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6153 |
| }, |
| { |
| "epoch": 2051.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6156 |
| }, |
| { |
| "epoch": 2052.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6159 |
| }, |
| { |
| "epoch": 2053.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6162 |
| }, |
| { |
| "epoch": 2054.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6165 |
| }, |
| { |
| "epoch": 2055.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6168 |
| }, |
| { |
| "epoch": 2056.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6171 |
| }, |
| { |
| "epoch": 2057.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6174 |
| }, |
| { |
| "epoch": 2058.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6177 |
| }, |
| { |
| "epoch": 2059.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6180 |
| }, |
| { |
| "epoch": 2060.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6183 |
| }, |
| { |
| "epoch": 2061.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6186 |
| }, |
| { |
| "epoch": 2062.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6189 |
| }, |
| { |
| "epoch": 2063.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6192 |
| }, |
| { |
| "epoch": 2064.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6195 |
| }, |
| { |
| "epoch": 2065.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6198 |
| }, |
| { |
| "epoch": 2066.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6201 |
| }, |
| { |
| "epoch": 2067.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6204 |
| }, |
| { |
| "epoch": 2068.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6207 |
| }, |
| { |
| "epoch": 2069.29, |
| "eval_exact_match": 0.5067698259187621, |
| "eval_exec": 0.49709864603481624, |
| "eval_loss": 0.651779055595398, |
| "eval_runtime": 105.2474, |
| "eval_samples_per_second": 9.824, |
| "step": 6208 |
| }, |
| { |
| "epoch": 2069.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6210 |
| }, |
| { |
| "epoch": 2070.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6213 |
| }, |
| { |
| "epoch": 2071.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6216 |
| }, |
| { |
| "epoch": 2072.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6219 |
| }, |
| { |
| "epoch": 2073.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6222 |
| }, |
| { |
| "epoch": 2074.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6225 |
| }, |
| { |
| "epoch": 2075.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6228 |
| }, |
| { |
| "epoch": 2076.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6231 |
| }, |
| { |
| "epoch": 2077.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6234 |
| }, |
| { |
| "epoch": 2078.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6237 |
| }, |
| { |
| "epoch": 2079.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6240 |
| }, |
| { |
| "epoch": 2080.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6243 |
| }, |
| { |
| "epoch": 2081.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6246 |
| }, |
| { |
| "epoch": 2082.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6249 |
| }, |
| { |
| "epoch": 2083.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6252 |
| }, |
| { |
| "epoch": 2084.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6255 |
| }, |
| { |
| "epoch": 2085.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6258 |
| }, |
| { |
| "epoch": 2086.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6261 |
| }, |
| { |
| "epoch": 2087.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6264 |
| }, |
| { |
| "epoch": 2088.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6267 |
| }, |
| { |
| "epoch": 2089.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6270 |
| }, |
| { |
| "epoch": 2090.59, |
| "eval_exact_match": 0.504835589941973, |
| "eval_exec": 0.49709864603481624, |
| "eval_loss": 0.6524325013160706, |
| "eval_runtime": 103.384, |
| "eval_samples_per_second": 10.002, |
| "step": 6272 |
| }, |
| { |
| "epoch": 2090.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6273 |
| }, |
| { |
| "epoch": 2091.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6276 |
| }, |
| { |
| "epoch": 2092.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6279 |
| }, |
| { |
| "epoch": 2093.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6282 |
| }, |
| { |
| "epoch": 2094.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6285 |
| }, |
| { |
| "epoch": 2095.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6288 |
| }, |
| { |
| "epoch": 2096.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6291 |
| }, |
| { |
| "epoch": 2097.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6294 |
| }, |
| { |
| "epoch": 2098.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6297 |
| }, |
| { |
| "epoch": 2099.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6300 |
| }, |
| { |
| "epoch": 2100.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6303 |
| }, |
| { |
| "epoch": 2101.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6306 |
| }, |
| { |
| "epoch": 2102.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6309 |
| }, |
| { |
| "epoch": 2103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6312 |
| }, |
| { |
| "epoch": 2104.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6315 |
| }, |
| { |
| "epoch": 2105.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6318 |
| }, |
| { |
| "epoch": 2106.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6321 |
| }, |
| { |
| "epoch": 2107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6324 |
| }, |
| { |
| "epoch": 2108.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6327 |
| }, |
| { |
| "epoch": 2109.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6330 |
| }, |
| { |
| "epoch": 2110.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6333 |
| }, |
| { |
| "epoch": 2111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6336 |
| }, |
| { |
| "epoch": 2111.88, |
| "eval_exact_match": 0.5067698259187621, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.657222330570221, |
| "eval_runtime": 105.2034, |
| "eval_samples_per_second": 9.829, |
| "step": 6336 |
| }, |
| { |
| "epoch": 2112.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6339 |
| }, |
| { |
| "epoch": 2113.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6342 |
| }, |
| { |
| "epoch": 2114.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6345 |
| }, |
| { |
| "epoch": 2115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6348 |
| }, |
| { |
| "epoch": 2116.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6351 |
| }, |
| { |
| "epoch": 2117.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6354 |
| }, |
| { |
| "epoch": 2118.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6357 |
| }, |
| { |
| "epoch": 2119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6360 |
| }, |
| { |
| "epoch": 2120.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6363 |
| }, |
| { |
| "epoch": 2121.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6366 |
| }, |
| { |
| "epoch": 2122.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6369 |
| }, |
| { |
| "epoch": 2123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6372 |
| }, |
| { |
| "epoch": 2124.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6375 |
| }, |
| { |
| "epoch": 2125.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6378 |
| }, |
| { |
| "epoch": 2126.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6381 |
| }, |
| { |
| "epoch": 2127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6384 |
| }, |
| { |
| "epoch": 2128.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6387 |
| }, |
| { |
| "epoch": 2129.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6390 |
| }, |
| { |
| "epoch": 2130.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6393 |
| }, |
| { |
| "epoch": 2131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6396 |
| }, |
| { |
| "epoch": 2132.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6399 |
| }, |
| { |
| "epoch": 2133.29, |
| "eval_exact_match": 0.5029013539651838, |
| "eval_exec": 0.4961315280464217, |
| "eval_loss": 0.6573466062545776, |
| "eval_runtime": 102.6054, |
| "eval_samples_per_second": 10.077, |
| "step": 6400 |
| }, |
| { |
| "epoch": 2133.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6402 |
| }, |
| { |
| "epoch": 2134.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6405 |
| }, |
| { |
| "epoch": 2135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6408 |
| }, |
| { |
| "epoch": 2136.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6411 |
| }, |
| { |
| "epoch": 2137.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6414 |
| }, |
| { |
| "epoch": 2138.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6417 |
| }, |
| { |
| "epoch": 2139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6420 |
| }, |
| { |
| "epoch": 2140.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6423 |
| }, |
| { |
| "epoch": 2141.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6426 |
| }, |
| { |
| "epoch": 2142.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6429 |
| }, |
| { |
| "epoch": 2143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6432 |
| }, |
| { |
| "epoch": 2144.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6435 |
| }, |
| { |
| "epoch": 2145.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6438 |
| }, |
| { |
| "epoch": 2146.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6441 |
| }, |
| { |
| "epoch": 2147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6444 |
| }, |
| { |
| "epoch": 2148.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6447 |
| }, |
| { |
| "epoch": 2149.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6450 |
| }, |
| { |
| "epoch": 2150.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6453 |
| }, |
| { |
| "epoch": 2151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6456 |
| }, |
| { |
| "epoch": 2152.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6459 |
| }, |
| { |
| "epoch": 2153.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6462 |
| }, |
| { |
| "epoch": 2154.59, |
| "eval_exact_match": 0.5125725338491296, |
| "eval_exec": 0.4990328820116054, |
| "eval_loss": 0.6498933434486389, |
| "eval_runtime": 103.9046, |
| "eval_samples_per_second": 9.951, |
| "step": 6464 |
| }, |
| { |
| "epoch": 2154.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6465 |
| }, |
| { |
| "epoch": 2155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6468 |
| }, |
| { |
| "epoch": 2156.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6471 |
| }, |
| { |
| "epoch": 2157.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6474 |
| }, |
| { |
| "epoch": 2158.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6477 |
| }, |
| { |
| "epoch": 2159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6480 |
| }, |
| { |
| "epoch": 2160.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6483 |
| }, |
| { |
| "epoch": 2161.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6486 |
| }, |
| { |
| "epoch": 2162.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6489 |
| }, |
| { |
| "epoch": 2163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6492 |
| }, |
| { |
| "epoch": 2164.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6495 |
| }, |
| { |
| "epoch": 2165.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6498 |
| }, |
| { |
| "epoch": 2166.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6501 |
| }, |
| { |
| "epoch": 2167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6504 |
| }, |
| { |
| "epoch": 2168.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 6507 |
| }, |
| { |
| "epoch": 2169.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6510 |
| }, |
| { |
| "epoch": 2170.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6513 |
| }, |
| { |
| "epoch": 2171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6516 |
| }, |
| { |
| "epoch": 2172.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6519 |
| }, |
| { |
| "epoch": 2173.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6522 |
| }, |
| { |
| "epoch": 2174.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6525 |
| }, |
| { |
| "epoch": 2175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6528 |
| }, |
| { |
| "epoch": 2175.88, |
| "eval_exact_match": 0.5096711798839458, |
| "eval_exec": 0.5, |
| "eval_loss": 0.6579024195671082, |
| "eval_runtime": 103.7204, |
| "eval_samples_per_second": 9.969, |
| "step": 6528 |
| }, |
| { |
| "epoch": 2176.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6531 |
| }, |
| { |
| "epoch": 2177.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6534 |
| }, |
| { |
| "epoch": 2178.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6537 |
| }, |
| { |
| "epoch": 2179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6540 |
| }, |
| { |
| "epoch": 2180.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6543 |
| }, |
| { |
| "epoch": 2181.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6546 |
| }, |
| { |
| "epoch": 2182.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6549 |
| }, |
| { |
| "epoch": 2183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6552 |
| }, |
| { |
| "epoch": 2184.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6555 |
| }, |
| { |
| "epoch": 2185.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6558 |
| }, |
| { |
| "epoch": 2186.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6561 |
| }, |
| { |
| "epoch": 2187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6564 |
| }, |
| { |
| "epoch": 2188.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6567 |
| }, |
| { |
| "epoch": 2189.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6570 |
| }, |
| { |
| "epoch": 2190.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6573 |
| }, |
| { |
| "epoch": 2191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6576 |
| }, |
| { |
| "epoch": 2192.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6579 |
| }, |
| { |
| "epoch": 2193.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6582 |
| }, |
| { |
| "epoch": 2194.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6585 |
| }, |
| { |
| "epoch": 2195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6588 |
| }, |
| { |
| "epoch": 2196.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6591 |
| }, |
| { |
| "epoch": 2197.29, |
| "eval_exact_match": 0.5067698259187621, |
| "eval_exec": 0.4990328820116054, |
| "eval_loss": 0.6636134386062622, |
| "eval_runtime": 104.9592, |
| "eval_samples_per_second": 9.851, |
| "step": 6592 |
| }, |
| { |
| "epoch": 2197.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6594 |
| }, |
| { |
| "epoch": 2198.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6597 |
| }, |
| { |
| "epoch": 2199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6600 |
| }, |
| { |
| "epoch": 2200.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6603 |
| }, |
| { |
| "epoch": 2201.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6606 |
| }, |
| { |
| "epoch": 2202.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6609 |
| }, |
| { |
| "epoch": 2203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6612 |
| }, |
| { |
| "epoch": 2204.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6615 |
| }, |
| { |
| "epoch": 2205.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6618 |
| }, |
| { |
| "epoch": 2206.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6621 |
| }, |
| { |
| "epoch": 2207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6624 |
| }, |
| { |
| "epoch": 2208.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6627 |
| }, |
| { |
| "epoch": 2209.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6630 |
| }, |
| { |
| "epoch": 2210.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6633 |
| }, |
| { |
| "epoch": 2211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6636 |
| }, |
| { |
| "epoch": 2212.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6639 |
| }, |
| { |
| "epoch": 2213.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6642 |
| }, |
| { |
| "epoch": 2214.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6645 |
| }, |
| { |
| "epoch": 2215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6648 |
| }, |
| { |
| "epoch": 2216.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6651 |
| }, |
| { |
| "epoch": 2217.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6654 |
| }, |
| { |
| "epoch": 2218.59, |
| "eval_exact_match": 0.5029013539651838, |
| "eval_exec": 0.4951644100580271, |
| "eval_loss": 0.6537144780158997, |
| "eval_runtime": 101.4907, |
| "eval_samples_per_second": 10.188, |
| "step": 6656 |
| }, |
| { |
| "epoch": 2218.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6657 |
| }, |
| { |
| "epoch": 2219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6660 |
| }, |
| { |
| "epoch": 2220.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6663 |
| }, |
| { |
| "epoch": 2221.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6666 |
| }, |
| { |
| "epoch": 2222.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6669 |
| }, |
| { |
| "epoch": 2223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6672 |
| }, |
| { |
| "epoch": 2224.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6675 |
| }, |
| { |
| "epoch": 2225.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6678 |
| }, |
| { |
| "epoch": 2226.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6681 |
| }, |
| { |
| "epoch": 2227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6684 |
| }, |
| { |
| "epoch": 2228.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6687 |
| }, |
| { |
| "epoch": 2229.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6690 |
| }, |
| { |
| "epoch": 2230.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6693 |
| }, |
| { |
| "epoch": 2231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6696 |
| }, |
| { |
| "epoch": 2232.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6699 |
| }, |
| { |
| "epoch": 2233.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6702 |
| }, |
| { |
| "epoch": 2234.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6705 |
| }, |
| { |
| "epoch": 2235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6708 |
| }, |
| { |
| "epoch": 2236.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 6711 |
| }, |
| { |
| "epoch": 2237.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6714 |
| }, |
| { |
| "epoch": 2238.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6717 |
| }, |
| { |
| "epoch": 2239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6720 |
| }, |
| { |
| "epoch": 2239.88, |
| "eval_exact_match": 0.511605415860735, |
| "eval_exec": 0.5019342359767892, |
| "eval_loss": 0.6580149531364441, |
| "eval_runtime": 101.3481, |
| "eval_samples_per_second": 10.202, |
| "step": 6720 |
| }, |
| { |
| "epoch": 2240.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6723 |
| }, |
| { |
| "epoch": 2241.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6726 |
| }, |
| { |
| "epoch": 2242.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6729 |
| }, |
| { |
| "epoch": 2243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6732 |
| }, |
| { |
| "epoch": 2244.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6735 |
| }, |
| { |
| "epoch": 2245.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6738 |
| }, |
| { |
| "epoch": 2246.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6741 |
| }, |
| { |
| "epoch": 2247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6744 |
| }, |
| { |
| "epoch": 2248.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6747 |
| }, |
| { |
| "epoch": 2249.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6750 |
| }, |
| { |
| "epoch": 2250.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6753 |
| }, |
| { |
| "epoch": 2251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6756 |
| }, |
| { |
| "epoch": 2252.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6759 |
| }, |
| { |
| "epoch": 2253.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6762 |
| }, |
| { |
| "epoch": 2254.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6765 |
| }, |
| { |
| "epoch": 2255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6768 |
| }, |
| { |
| "epoch": 2256.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6771 |
| }, |
| { |
| "epoch": 2257.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6774 |
| }, |
| { |
| "epoch": 2258.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6777 |
| }, |
| { |
| "epoch": 2259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6780 |
| }, |
| { |
| "epoch": 2260.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6783 |
| }, |
| { |
| "epoch": 2261.29, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.656693696975708, |
| "eval_runtime": 103.6205, |
| "eval_samples_per_second": 9.979, |
| "step": 6784 |
| }, |
| { |
| "epoch": 2261.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6786 |
| }, |
| { |
| "epoch": 2262.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6789 |
| }, |
| { |
| "epoch": 2263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6792 |
| }, |
| { |
| "epoch": 2264.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6795 |
| }, |
| { |
| "epoch": 2265.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6798 |
| }, |
| { |
| "epoch": 2266.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6801 |
| }, |
| { |
| "epoch": 2267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6804 |
| }, |
| { |
| "epoch": 2268.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6807 |
| }, |
| { |
| "epoch": 2269.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6810 |
| }, |
| { |
| "epoch": 2270.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6813 |
| }, |
| { |
| "epoch": 2271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6816 |
| }, |
| { |
| "epoch": 2272.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6819 |
| }, |
| { |
| "epoch": 2273.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6822 |
| }, |
| { |
| "epoch": 2274.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6825 |
| }, |
| { |
| "epoch": 2275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6828 |
| }, |
| { |
| "epoch": 2276.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6831 |
| }, |
| { |
| "epoch": 2277.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6834 |
| }, |
| { |
| "epoch": 2278.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6837 |
| }, |
| { |
| "epoch": 2279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6840 |
| }, |
| { |
| "epoch": 2280.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6843 |
| }, |
| { |
| "epoch": 2281.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6846 |
| }, |
| { |
| "epoch": 2282.59, |
| "eval_exact_match": 0.49709864603481624, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.6702226996421814, |
| "eval_runtime": 100.3881, |
| "eval_samples_per_second": 10.3, |
| "step": 6848 |
| }, |
| { |
| "epoch": 2282.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6849 |
| }, |
| { |
| "epoch": 2283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6852 |
| }, |
| { |
| "epoch": 2284.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6855 |
| }, |
| { |
| "epoch": 2285.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6858 |
| }, |
| { |
| "epoch": 2286.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6861 |
| }, |
| { |
| "epoch": 2287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6864 |
| }, |
| { |
| "epoch": 2288.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6867 |
| }, |
| { |
| "epoch": 2289.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6870 |
| }, |
| { |
| "epoch": 2290.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6873 |
| }, |
| { |
| "epoch": 2291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 6876 |
| }, |
| { |
| "epoch": 2292.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6879 |
| }, |
| { |
| "epoch": 2293.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6882 |
| }, |
| { |
| "epoch": 2294.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6885 |
| }, |
| { |
| "epoch": 2295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6888 |
| }, |
| { |
| "epoch": 2296.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6891 |
| }, |
| { |
| "epoch": 2297.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6894 |
| }, |
| { |
| "epoch": 2298.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6897 |
| }, |
| { |
| "epoch": 2299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6900 |
| }, |
| { |
| "epoch": 2300.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6903 |
| }, |
| { |
| "epoch": 2301.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6906 |
| }, |
| { |
| "epoch": 2302.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6909 |
| }, |
| { |
| "epoch": 2303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6912 |
| }, |
| { |
| "epoch": 2303.88, |
| "eval_exact_match": 0.49806576402321084, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.6657952666282654, |
| "eval_runtime": 102.7872, |
| "eval_samples_per_second": 10.06, |
| "step": 6912 |
| }, |
| { |
| "epoch": 2304.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6915 |
| }, |
| { |
| "epoch": 2305.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6918 |
| }, |
| { |
| "epoch": 2306.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6921 |
| }, |
| { |
| "epoch": 2307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6924 |
| }, |
| { |
| "epoch": 2308.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6927 |
| }, |
| { |
| "epoch": 2309.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 6930 |
| }, |
| { |
| "epoch": 2310.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6933 |
| }, |
| { |
| "epoch": 2311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6936 |
| }, |
| { |
| "epoch": 2312.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6939 |
| }, |
| { |
| "epoch": 2313.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6942 |
| }, |
| { |
| "epoch": 2314.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6945 |
| }, |
| { |
| "epoch": 2315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6948 |
| }, |
| { |
| "epoch": 2316.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6951 |
| }, |
| { |
| "epoch": 2317.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6954 |
| }, |
| { |
| "epoch": 2318.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6957 |
| }, |
| { |
| "epoch": 2319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6960 |
| }, |
| { |
| "epoch": 2320.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6963 |
| }, |
| { |
| "epoch": 2321.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 6966 |
| }, |
| { |
| "epoch": 2322.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6969 |
| }, |
| { |
| "epoch": 2323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6972 |
| }, |
| { |
| "epoch": 2324.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6975 |
| }, |
| { |
| "epoch": 2325.29, |
| "eval_exact_match": 0.4951644100580271, |
| "eval_exec": 0.4951644100580271, |
| "eval_loss": 0.6729605197906494, |
| "eval_runtime": 101.8227, |
| "eval_samples_per_second": 10.155, |
| "step": 6976 |
| }, |
| { |
| "epoch": 2325.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 6978 |
| }, |
| { |
| "epoch": 2326.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6981 |
| }, |
| { |
| "epoch": 2327.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6984 |
| }, |
| { |
| "epoch": 2328.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6987 |
| }, |
| { |
| "epoch": 2329.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 6990 |
| }, |
| { |
| "epoch": 2330.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 6993 |
| }, |
| { |
| "epoch": 2331.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 6996 |
| }, |
| { |
| "epoch": 2332.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 6999 |
| }, |
| { |
| "epoch": 2333.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7002 |
| }, |
| { |
| "epoch": 2334.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7005 |
| }, |
| { |
| "epoch": 2335.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7008 |
| }, |
| { |
| "epoch": 2336.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7011 |
| }, |
| { |
| "epoch": 2337.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7014 |
| }, |
| { |
| "epoch": 2338.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7017 |
| }, |
| { |
| "epoch": 2339.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7020 |
| }, |
| { |
| "epoch": 2340.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7023 |
| }, |
| { |
| "epoch": 2341.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7026 |
| }, |
| { |
| "epoch": 2342.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7029 |
| }, |
| { |
| "epoch": 2343.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7032 |
| }, |
| { |
| "epoch": 2344.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7035 |
| }, |
| { |
| "epoch": 2345.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7038 |
| }, |
| { |
| "epoch": 2346.59, |
| "eval_exact_match": 0.5038684719535783, |
| "eval_exec": 0.5, |
| "eval_loss": 0.6706193685531616, |
| "eval_runtime": 107.1601, |
| "eval_samples_per_second": 9.649, |
| "step": 7040 |
| }, |
| { |
| "epoch": 2346.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7041 |
| }, |
| { |
| "epoch": 2347.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7044 |
| }, |
| { |
| "epoch": 2348.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7047 |
| }, |
| { |
| "epoch": 2349.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7050 |
| }, |
| { |
| "epoch": 2350.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7053 |
| }, |
| { |
| "epoch": 2351.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7056 |
| }, |
| { |
| "epoch": 2352.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7059 |
| }, |
| { |
| "epoch": 2353.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7062 |
| }, |
| { |
| "epoch": 2354.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7065 |
| }, |
| { |
| "epoch": 2355.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7068 |
| }, |
| { |
| "epoch": 2356.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7071 |
| }, |
| { |
| "epoch": 2357.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7074 |
| }, |
| { |
| "epoch": 2358.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7077 |
| }, |
| { |
| "epoch": 2359.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7080 |
| }, |
| { |
| "epoch": 2360.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7083 |
| }, |
| { |
| "epoch": 2361.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7086 |
| }, |
| { |
| "epoch": 2362.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7089 |
| }, |
| { |
| "epoch": 2363.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7092 |
| }, |
| { |
| "epoch": 2364.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7095 |
| }, |
| { |
| "epoch": 2365.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7098 |
| }, |
| { |
| "epoch": 2366.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7101 |
| }, |
| { |
| "epoch": 2367.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7104 |
| }, |
| { |
| "epoch": 2367.88, |
| "eval_exact_match": 0.5058027079303675, |
| "eval_exec": 0.49709864603481624, |
| "eval_loss": 0.6719130277633667, |
| "eval_runtime": 103.7435, |
| "eval_samples_per_second": 9.967, |
| "step": 7104 |
| }, |
| { |
| "epoch": 2368.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7107 |
| }, |
| { |
| "epoch": 2369.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7110 |
| }, |
| { |
| "epoch": 2370.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7113 |
| }, |
| { |
| "epoch": 2371.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7116 |
| }, |
| { |
| "epoch": 2372.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7119 |
| }, |
| { |
| "epoch": 2373.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7122 |
| }, |
| { |
| "epoch": 2374.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7125 |
| }, |
| { |
| "epoch": 2375.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7128 |
| }, |
| { |
| "epoch": 2376.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7131 |
| }, |
| { |
| "epoch": 2377.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7134 |
| }, |
| { |
| "epoch": 2378.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7137 |
| }, |
| { |
| "epoch": 2379.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7140 |
| }, |
| { |
| "epoch": 2380.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7143 |
| }, |
| { |
| "epoch": 2381.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7146 |
| }, |
| { |
| "epoch": 2382.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7149 |
| }, |
| { |
| "epoch": 2383.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7152 |
| }, |
| { |
| "epoch": 2384.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7155 |
| }, |
| { |
| "epoch": 2385.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7158 |
| }, |
| { |
| "epoch": 2386.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7161 |
| }, |
| { |
| "epoch": 2387.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7164 |
| }, |
| { |
| "epoch": 2388.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7167 |
| }, |
| { |
| "epoch": 2389.29, |
| "eval_exact_match": 0.5, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.6656604409217834, |
| "eval_runtime": 110.2501, |
| "eval_samples_per_second": 9.379, |
| "step": 7168 |
| }, |
| { |
| "epoch": 2389.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7170 |
| }, |
| { |
| "epoch": 2390.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7173 |
| }, |
| { |
| "epoch": 2391.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7176 |
| }, |
| { |
| "epoch": 2392.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7179 |
| }, |
| { |
| "epoch": 2393.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7182 |
| }, |
| { |
| "epoch": 2394.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7185 |
| }, |
| { |
| "epoch": 2395.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7188 |
| }, |
| { |
| "epoch": 2396.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7191 |
| }, |
| { |
| "epoch": 2397.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7194 |
| }, |
| { |
| "epoch": 2398.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7197 |
| }, |
| { |
| "epoch": 2399.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7200 |
| }, |
| { |
| "epoch": 2400.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7203 |
| }, |
| { |
| "epoch": 2401.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7206 |
| }, |
| { |
| "epoch": 2402.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7209 |
| }, |
| { |
| "epoch": 2403.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7212 |
| }, |
| { |
| "epoch": 2404.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7215 |
| }, |
| { |
| "epoch": 2405.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7218 |
| }, |
| { |
| "epoch": 2406.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7221 |
| }, |
| { |
| "epoch": 2407.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7224 |
| }, |
| { |
| "epoch": 2408.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7227 |
| }, |
| { |
| "epoch": 2409.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7230 |
| }, |
| { |
| "epoch": 2410.59, |
| "eval_exact_match": 0.5067698259187621, |
| "eval_exec": 0.49806576402321084, |
| "eval_loss": 0.6692262291908264, |
| "eval_runtime": 107.2383, |
| "eval_samples_per_second": 9.642, |
| "step": 7232 |
| }, |
| { |
| "epoch": 2410.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7233 |
| }, |
| { |
| "epoch": 2411.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7236 |
| }, |
| { |
| "epoch": 2412.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7239 |
| }, |
| { |
| "epoch": 2413.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7242 |
| }, |
| { |
| "epoch": 2414.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7245 |
| }, |
| { |
| "epoch": 2415.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7248 |
| }, |
| { |
| "epoch": 2416.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7251 |
| }, |
| { |
| "epoch": 2417.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7254 |
| }, |
| { |
| "epoch": 2418.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7257 |
| }, |
| { |
| "epoch": 2419.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7260 |
| }, |
| { |
| "epoch": 2420.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7263 |
| }, |
| { |
| "epoch": 2421.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7266 |
| }, |
| { |
| "epoch": 2422.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7269 |
| }, |
| { |
| "epoch": 2423.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7272 |
| }, |
| { |
| "epoch": 2424.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7275 |
| }, |
| { |
| "epoch": 2425.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7278 |
| }, |
| { |
| "epoch": 2426.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7281 |
| }, |
| { |
| "epoch": 2427.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7284 |
| }, |
| { |
| "epoch": 2428.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7287 |
| }, |
| { |
| "epoch": 2429.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7290 |
| }, |
| { |
| "epoch": 2430.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7293 |
| }, |
| { |
| "epoch": 2431.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7296 |
| }, |
| { |
| "epoch": 2431.88, |
| "eval_exact_match": 0.5038684719535783, |
| "eval_exec": 0.4912959381044487, |
| "eval_loss": 0.672451376914978, |
| "eval_runtime": 102.0344, |
| "eval_samples_per_second": 10.134, |
| "step": 7296 |
| }, |
| { |
| "epoch": 2432.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7299 |
| }, |
| { |
| "epoch": 2433.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7302 |
| }, |
| { |
| "epoch": 2434.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7305 |
| }, |
| { |
| "epoch": 2435.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7308 |
| }, |
| { |
| "epoch": 2436.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7311 |
| }, |
| { |
| "epoch": 2437.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7314 |
| }, |
| { |
| "epoch": 2438.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7317 |
| }, |
| { |
| "epoch": 2439.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7320 |
| }, |
| { |
| "epoch": 2440.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7323 |
| }, |
| { |
| "epoch": 2441.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7326 |
| }, |
| { |
| "epoch": 2442.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7329 |
| }, |
| { |
| "epoch": 2443.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7332 |
| }, |
| { |
| "epoch": 2444.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7335 |
| }, |
| { |
| "epoch": 2445.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7338 |
| }, |
| { |
| "epoch": 2446.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7341 |
| }, |
| { |
| "epoch": 2447.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7344 |
| }, |
| { |
| "epoch": 2448.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7347 |
| }, |
| { |
| "epoch": 2449.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7350 |
| }, |
| { |
| "epoch": 2450.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7353 |
| }, |
| { |
| "epoch": 2451.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7356 |
| }, |
| { |
| "epoch": 2452.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7359 |
| }, |
| { |
| "epoch": 2453.29, |
| "eval_exact_match": 0.5087040618955513, |
| "eval_exec": 0.4941972920696325, |
| "eval_loss": 0.6790467500686646, |
| "eval_runtime": 105.5575, |
| "eval_samples_per_second": 9.796, |
| "step": 7360 |
| }, |
| { |
| "epoch": 2453.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7362 |
| }, |
| { |
| "epoch": 2454.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7365 |
| }, |
| { |
| "epoch": 2455.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7368 |
| }, |
| { |
| "epoch": 2456.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7371 |
| }, |
| { |
| "epoch": 2457.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7374 |
| }, |
| { |
| "epoch": 2458.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7377 |
| }, |
| { |
| "epoch": 2459.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7380 |
| }, |
| { |
| "epoch": 2460.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7383 |
| }, |
| { |
| "epoch": 2461.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7386 |
| }, |
| { |
| "epoch": 2462.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7389 |
| }, |
| { |
| "epoch": 2463.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7392 |
| }, |
| { |
| "epoch": 2464.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7395 |
| }, |
| { |
| "epoch": 2465.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7398 |
| }, |
| { |
| "epoch": 2466.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7401 |
| }, |
| { |
| "epoch": 2467.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7404 |
| }, |
| { |
| "epoch": 2468.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7407 |
| }, |
| { |
| "epoch": 2469.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7410 |
| }, |
| { |
| "epoch": 2470.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7413 |
| }, |
| { |
| "epoch": 2471.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7416 |
| }, |
| { |
| "epoch": 2472.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7419 |
| }, |
| { |
| "epoch": 2473.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7422 |
| }, |
| { |
| "epoch": 2474.59, |
| "eval_exact_match": 0.5087040618955513, |
| "eval_exec": 0.4990328820116054, |
| "eval_loss": 0.6767649054527283, |
| "eval_runtime": 101.2746, |
| "eval_samples_per_second": 10.21, |
| "step": 7424 |
| }, |
| { |
| "epoch": 2474.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7425 |
| }, |
| { |
| "epoch": 2475.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7428 |
| }, |
| { |
| "epoch": 2476.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7431 |
| }, |
| { |
| "epoch": 2477.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7434 |
| }, |
| { |
| "epoch": 2478.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7437 |
| }, |
| { |
| "epoch": 2479.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7440 |
| }, |
| { |
| "epoch": 2480.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7443 |
| }, |
| { |
| "epoch": 2481.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7446 |
| }, |
| { |
| "epoch": 2482.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7449 |
| }, |
| { |
| "epoch": 2483.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7452 |
| }, |
| { |
| "epoch": 2484.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7455 |
| }, |
| { |
| "epoch": 2485.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7458 |
| }, |
| { |
| "epoch": 2486.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7461 |
| }, |
| { |
| "epoch": 2487.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7464 |
| }, |
| { |
| "epoch": 2488.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7467 |
| }, |
| { |
| "epoch": 2489.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7470 |
| }, |
| { |
| "epoch": 2490.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7473 |
| }, |
| { |
| "epoch": 2491.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7476 |
| }, |
| { |
| "epoch": 2492.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7479 |
| }, |
| { |
| "epoch": 2493.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7482 |
| }, |
| { |
| "epoch": 2494.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7485 |
| }, |
| { |
| "epoch": 2495.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7488 |
| }, |
| { |
| "epoch": 2495.88, |
| "eval_exact_match": 0.5029013539651838, |
| "eval_exec": 0.49806576402321084, |
| "eval_loss": 0.6734866499900818, |
| "eval_runtime": 98.1491, |
| "eval_samples_per_second": 10.535, |
| "step": 7488 |
| }, |
| { |
| "epoch": 2496.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 7491 |
| }, |
| { |
| "epoch": 2497.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7494 |
| }, |
| { |
| "epoch": 2498.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7497 |
| }, |
| { |
| "epoch": 2499.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7500 |
| }, |
| { |
| "epoch": 2500.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7503 |
| }, |
| { |
| "epoch": 2501.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7506 |
| }, |
| { |
| "epoch": 2502.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7509 |
| }, |
| { |
| "epoch": 2503.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7512 |
| }, |
| { |
| "epoch": 2504.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7515 |
| }, |
| { |
| "epoch": 2505.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7518 |
| }, |
| { |
| "epoch": 2506.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7521 |
| }, |
| { |
| "epoch": 2507.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7524 |
| }, |
| { |
| "epoch": 2508.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7527 |
| }, |
| { |
| "epoch": 2509.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7530 |
| }, |
| { |
| "epoch": 2510.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7533 |
| }, |
| { |
| "epoch": 2511.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7536 |
| }, |
| { |
| "epoch": 2512.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7539 |
| }, |
| { |
| "epoch": 2513.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7542 |
| }, |
| { |
| "epoch": 2514.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7545 |
| }, |
| { |
| "epoch": 2515.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7548 |
| }, |
| { |
| "epoch": 2516.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7551 |
| }, |
| { |
| "epoch": 2517.29, |
| "eval_exact_match": 0.5058027079303675, |
| "eval_exec": 0.4951644100580271, |
| "eval_loss": 0.6706563830375671, |
| "eval_runtime": 104.1289, |
| "eval_samples_per_second": 9.93, |
| "step": 7552 |
| }, |
| { |
| "epoch": 2517.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7554 |
| }, |
| { |
| "epoch": 2518.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7557 |
| }, |
| { |
| "epoch": 2519.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7560 |
| }, |
| { |
| "epoch": 2520.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7563 |
| }, |
| { |
| "epoch": 2521.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7566 |
| }, |
| { |
| "epoch": 2522.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7569 |
| }, |
| { |
| "epoch": 2523.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7572 |
| }, |
| { |
| "epoch": 2524.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7575 |
| }, |
| { |
| "epoch": 2525.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7578 |
| }, |
| { |
| "epoch": 2526.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7581 |
| }, |
| { |
| "epoch": 2527.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7584 |
| }, |
| { |
| "epoch": 2528.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7587 |
| }, |
| { |
| "epoch": 2529.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7590 |
| }, |
| { |
| "epoch": 2530.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7593 |
| }, |
| { |
| "epoch": 2531.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7596 |
| }, |
| { |
| "epoch": 2532.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7599 |
| }, |
| { |
| "epoch": 2533.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7602 |
| }, |
| { |
| "epoch": 2534.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7605 |
| }, |
| { |
| "epoch": 2535.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7608 |
| }, |
| { |
| "epoch": 2536.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7611 |
| }, |
| { |
| "epoch": 2537.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7614 |
| }, |
| { |
| "epoch": 2538.59, |
| "eval_exact_match": 0.5135396518375241, |
| "eval_exec": 0.4951644100580271, |
| "eval_loss": 0.6734728813171387, |
| "eval_runtime": 103.6866, |
| "eval_samples_per_second": 9.972, |
| "step": 7616 |
| }, |
| { |
| "epoch": 2538.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7617 |
| }, |
| { |
| "epoch": 2539.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7620 |
| }, |
| { |
| "epoch": 2540.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7623 |
| }, |
| { |
| "epoch": 2541.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7626 |
| }, |
| { |
| "epoch": 2542.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7629 |
| }, |
| { |
| "epoch": 2543.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7632 |
| }, |
| { |
| "epoch": 2544.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7635 |
| }, |
| { |
| "epoch": 2545.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7638 |
| }, |
| { |
| "epoch": 2546.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7641 |
| }, |
| { |
| "epoch": 2547.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7644 |
| }, |
| { |
| "epoch": 2548.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7647 |
| }, |
| { |
| "epoch": 2549.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7650 |
| }, |
| { |
| "epoch": 2550.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7653 |
| }, |
| { |
| "epoch": 2551.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7656 |
| }, |
| { |
| "epoch": 2552.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7659 |
| }, |
| { |
| "epoch": 2553.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7662 |
| }, |
| { |
| "epoch": 2554.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7665 |
| }, |
| { |
| "epoch": 2555.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7668 |
| }, |
| { |
| "epoch": 2556.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7671 |
| }, |
| { |
| "epoch": 2557.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7674 |
| }, |
| { |
| "epoch": 2558.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7677 |
| }, |
| { |
| "epoch": 2559.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7680 |
| }, |
| { |
| "epoch": 2559.88, |
| "eval_exact_match": 0.4951644100580271, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.6788745522499084, |
| "eval_runtime": 100.2149, |
| "eval_samples_per_second": 10.318, |
| "step": 7680 |
| }, |
| { |
| "epoch": 2560.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7683 |
| }, |
| { |
| "epoch": 2561.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7686 |
| }, |
| { |
| "epoch": 2562.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7689 |
| }, |
| { |
| "epoch": 2563.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7692 |
| }, |
| { |
| "epoch": 2564.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7695 |
| }, |
| { |
| "epoch": 2565.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7698 |
| }, |
| { |
| "epoch": 2566.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7701 |
| }, |
| { |
| "epoch": 2567.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7704 |
| }, |
| { |
| "epoch": 2568.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7707 |
| }, |
| { |
| "epoch": 2569.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7710 |
| }, |
| { |
| "epoch": 2570.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7713 |
| }, |
| { |
| "epoch": 2571.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7716 |
| }, |
| { |
| "epoch": 2572.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7719 |
| }, |
| { |
| "epoch": 2573.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7722 |
| }, |
| { |
| "epoch": 2574.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7725 |
| }, |
| { |
| "epoch": 2575.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7728 |
| }, |
| { |
| "epoch": 2576.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7731 |
| }, |
| { |
| "epoch": 2577.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7734 |
| }, |
| { |
| "epoch": 2578.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7737 |
| }, |
| { |
| "epoch": 2579.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7740 |
| }, |
| { |
| "epoch": 2580.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7743 |
| }, |
| { |
| "epoch": 2581.29, |
| "eval_exact_match": 0.5, |
| "eval_exec": 0.488394584139265, |
| "eval_loss": 0.6870007514953613, |
| "eval_runtime": 103.442, |
| "eval_samples_per_second": 9.996, |
| "step": 7744 |
| }, |
| { |
| "epoch": 2581.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7746 |
| }, |
| { |
| "epoch": 2582.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7749 |
| }, |
| { |
| "epoch": 2583.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7752 |
| }, |
| { |
| "epoch": 2584.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7755 |
| }, |
| { |
| "epoch": 2585.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7758 |
| }, |
| { |
| "epoch": 2586.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7761 |
| }, |
| { |
| "epoch": 2587.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7764 |
| }, |
| { |
| "epoch": 2588.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7767 |
| }, |
| { |
| "epoch": 2589.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7770 |
| }, |
| { |
| "epoch": 2590.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7773 |
| }, |
| { |
| "epoch": 2591.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7776 |
| }, |
| { |
| "epoch": 2592.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7779 |
| }, |
| { |
| "epoch": 2593.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7782 |
| }, |
| { |
| "epoch": 2594.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7785 |
| }, |
| { |
| "epoch": 2595.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7788 |
| }, |
| { |
| "epoch": 2596.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7791 |
| }, |
| { |
| "epoch": 2597.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7794 |
| }, |
| { |
| "epoch": 2598.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7797 |
| }, |
| { |
| "epoch": 2599.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7800 |
| }, |
| { |
| "epoch": 2600.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7803 |
| }, |
| { |
| "epoch": 2601.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7806 |
| }, |
| { |
| "epoch": 2602.59, |
| "eval_exact_match": 0.5125725338491296, |
| "eval_exec": 0.49806576402321084, |
| "eval_loss": 0.6841245889663696, |
| "eval_runtime": 100.6823, |
| "eval_samples_per_second": 10.27, |
| "step": 7808 |
| }, |
| { |
| "epoch": 2602.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7809 |
| }, |
| { |
| "epoch": 2603.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7812 |
| }, |
| { |
| "epoch": 2604.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7815 |
| }, |
| { |
| "epoch": 2605.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7818 |
| }, |
| { |
| "epoch": 2606.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7821 |
| }, |
| { |
| "epoch": 2607.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7824 |
| }, |
| { |
| "epoch": 2608.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7827 |
| }, |
| { |
| "epoch": 2609.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7830 |
| }, |
| { |
| "epoch": 2610.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7833 |
| }, |
| { |
| "epoch": 2611.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7836 |
| }, |
| { |
| "epoch": 2612.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 7839 |
| }, |
| { |
| "epoch": 2613.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7842 |
| }, |
| { |
| "epoch": 2614.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7845 |
| }, |
| { |
| "epoch": 2615.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7848 |
| }, |
| { |
| "epoch": 2616.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7851 |
| }, |
| { |
| "epoch": 2617.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7854 |
| }, |
| { |
| "epoch": 2618.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7857 |
| }, |
| { |
| "epoch": 2619.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7860 |
| }, |
| { |
| "epoch": 2620.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7863 |
| }, |
| { |
| "epoch": 2621.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7866 |
| }, |
| { |
| "epoch": 2622.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7869 |
| }, |
| { |
| "epoch": 2623.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7872 |
| }, |
| { |
| "epoch": 2623.88, |
| "eval_exact_match": 0.5058027079303675, |
| "eval_exec": 0.4922630560928433, |
| "eval_loss": 0.6852383017539978, |
| "eval_runtime": 102.8761, |
| "eval_samples_per_second": 10.051, |
| "step": 7872 |
| }, |
| { |
| "epoch": 2624.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7875 |
| }, |
| { |
| "epoch": 2625.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7878 |
| }, |
| { |
| "epoch": 2626.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7881 |
| }, |
| { |
| "epoch": 2627.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7884 |
| }, |
| { |
| "epoch": 2628.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 7887 |
| }, |
| { |
| "epoch": 2629.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 7890 |
| }, |
| { |
| "epoch": 2630.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7893 |
| }, |
| { |
| "epoch": 2631.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7896 |
| }, |
| { |
| "epoch": 2632.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7899 |
| }, |
| { |
| "epoch": 2633.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7902 |
| }, |
| { |
| "epoch": 2634.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7905 |
| }, |
| { |
| "epoch": 2635.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7908 |
| }, |
| { |
| "epoch": 2636.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7911 |
| }, |
| { |
| "epoch": 2637.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7914 |
| }, |
| { |
| "epoch": 2638.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7917 |
| }, |
| { |
| "epoch": 2639.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7920 |
| }, |
| { |
| "epoch": 2640.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7923 |
| }, |
| { |
| "epoch": 2641.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7926 |
| }, |
| { |
| "epoch": 2642.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7929 |
| }, |
| { |
| "epoch": 2643.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7932 |
| }, |
| { |
| "epoch": 2644.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7935 |
| }, |
| { |
| "epoch": 2645.29, |
| "eval_exact_match": 0.5087040618955513, |
| "eval_exec": 0.5, |
| "eval_loss": 0.678929328918457, |
| "eval_runtime": 101.2205, |
| "eval_samples_per_second": 10.215, |
| "step": 7936 |
| }, |
| { |
| "epoch": 2645.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7938 |
| }, |
| { |
| "epoch": 2646.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7941 |
| }, |
| { |
| "epoch": 2647.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7944 |
| }, |
| { |
| "epoch": 2648.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7947 |
| }, |
| { |
| "epoch": 2649.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7950 |
| }, |
| { |
| "epoch": 2650.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7953 |
| }, |
| { |
| "epoch": 2651.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7956 |
| }, |
| { |
| "epoch": 2652.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7959 |
| }, |
| { |
| "epoch": 2653.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7962 |
| }, |
| { |
| "epoch": 2654.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7965 |
| }, |
| { |
| "epoch": 2655.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7968 |
| }, |
| { |
| "epoch": 2656.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7971 |
| }, |
| { |
| "epoch": 2657.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7974 |
| }, |
| { |
| "epoch": 2658.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7977 |
| }, |
| { |
| "epoch": 2659.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7980 |
| }, |
| { |
| "epoch": 2660.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7983 |
| }, |
| { |
| "epoch": 2661.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7986 |
| }, |
| { |
| "epoch": 2662.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7989 |
| }, |
| { |
| "epoch": 2663.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 7992 |
| }, |
| { |
| "epoch": 2664.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 7995 |
| }, |
| { |
| "epoch": 2665.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 7998 |
| }, |
| { |
| "epoch": 2666.59, |
| "eval_exact_match": 0.5125725338491296, |
| "eval_exec": 0.5029013539651838, |
| "eval_loss": 0.6847764849662781, |
| "eval_runtime": 104.2805, |
| "eval_samples_per_second": 9.916, |
| "step": 8000 |
| }, |
| { |
| "epoch": 2666.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8001 |
| }, |
| { |
| "epoch": 2667.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8004 |
| }, |
| { |
| "epoch": 2668.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8007 |
| }, |
| { |
| "epoch": 2669.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8010 |
| }, |
| { |
| "epoch": 2670.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8013 |
| }, |
| { |
| "epoch": 2671.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8016 |
| }, |
| { |
| "epoch": 2672.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8019 |
| }, |
| { |
| "epoch": 2673.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 8022 |
| }, |
| { |
| "epoch": 2674.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8025 |
| }, |
| { |
| "epoch": 2675.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8028 |
| }, |
| { |
| "epoch": 2676.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8031 |
| }, |
| { |
| "epoch": 2677.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8034 |
| }, |
| { |
| "epoch": 2678.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8037 |
| }, |
| { |
| "epoch": 2679.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8040 |
| }, |
| { |
| "epoch": 2680.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8043 |
| }, |
| { |
| "epoch": 2681.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8046 |
| }, |
| { |
| "epoch": 2682.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8049 |
| }, |
| { |
| "epoch": 2683.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8052 |
| }, |
| { |
| "epoch": 2684.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8055 |
| }, |
| { |
| "epoch": 2685.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8058 |
| }, |
| { |
| "epoch": 2686.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8061 |
| }, |
| { |
| "epoch": 2687.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8064 |
| }, |
| { |
| "epoch": 2687.88, |
| "eval_exact_match": 0.5058027079303675, |
| "eval_exec": 0.5, |
| "eval_loss": 0.6950146555900574, |
| "eval_runtime": 103.8342, |
| "eval_samples_per_second": 9.958, |
| "step": 8064 |
| }, |
| { |
| "epoch": 2688.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8067 |
| }, |
| { |
| "epoch": 2689.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8070 |
| }, |
| { |
| "epoch": 2690.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8073 |
| }, |
| { |
| "epoch": 2691.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8076 |
| }, |
| { |
| "epoch": 2692.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8079 |
| }, |
| { |
| "epoch": 2693.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8082 |
| }, |
| { |
| "epoch": 2694.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8085 |
| }, |
| { |
| "epoch": 2695.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 8088 |
| }, |
| { |
| "epoch": 2696.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8091 |
| }, |
| { |
| "epoch": 2697.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8094 |
| }, |
| { |
| "epoch": 2698.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8097 |
| }, |
| { |
| "epoch": 2699.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8100 |
| }, |
| { |
| "epoch": 2700.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8103 |
| }, |
| { |
| "epoch": 2701.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8106 |
| }, |
| { |
| "epoch": 2702.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 8109 |
| }, |
| { |
| "epoch": 2703.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8112 |
| }, |
| { |
| "epoch": 2704.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 8115 |
| }, |
| { |
| "epoch": 2705.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8118 |
| }, |
| { |
| "epoch": 2706.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8121 |
| }, |
| { |
| "epoch": 2707.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8124 |
| }, |
| { |
| "epoch": 2708.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8127 |
| }, |
| { |
| "epoch": 2709.29, |
| "eval_exact_match": 0.4990328820116054, |
| "eval_exec": 0.4874274661508704, |
| "eval_loss": 0.6891441941261292, |
| "eval_runtime": 107.0055, |
| "eval_samples_per_second": 9.663, |
| "step": 8128 |
| }, |
| { |
| "epoch": 2709.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8130 |
| }, |
| { |
| "epoch": 2710.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8133 |
| }, |
| { |
| "epoch": 2711.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8136 |
| }, |
| { |
| "epoch": 2712.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8139 |
| }, |
| { |
| "epoch": 2713.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8142 |
| }, |
| { |
| "epoch": 2714.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8145 |
| }, |
| { |
| "epoch": 2715.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8148 |
| }, |
| { |
| "epoch": 2716.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8151 |
| }, |
| { |
| "epoch": 2717.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8154 |
| }, |
| { |
| "epoch": 2718.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8157 |
| }, |
| { |
| "epoch": 2719.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8160 |
| }, |
| { |
| "epoch": 2720.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8163 |
| }, |
| { |
| "epoch": 2721.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8166 |
| }, |
| { |
| "epoch": 2722.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8169 |
| }, |
| { |
| "epoch": 2723.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8172 |
| }, |
| { |
| "epoch": 2724.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 8175 |
| }, |
| { |
| "epoch": 2725.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8178 |
| }, |
| { |
| "epoch": 2726.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 8181 |
| }, |
| { |
| "epoch": 2727.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8184 |
| }, |
| { |
| "epoch": 2728.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8187 |
| }, |
| { |
| "epoch": 2729.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 8190 |
| }, |
| { |
| "epoch": 2730.59, |
| "eval_exact_match": 0.5174081237911026, |
| "eval_exec": 0.5067698259187621, |
| "eval_loss": 0.6930547952651978, |
| "eval_runtime": 100.8541, |
| "eval_samples_per_second": 10.252, |
| "step": 8192 |
| } |
| ], |
| "max_steps": 9216, |
| "num_train_epochs": 3072, |
| "total_flos": 1.627321267788972e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
