| { |
| "best_metric": 0.5938104448742747, |
| "best_model_checkpoint": "/mnt/chenzhi/dialogzoo/finetune/txt2sql_picard_spider/checkpoint-6144", |
| "epoch": 2325.292857142857, |
| "global_step": 6976, |
| "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": 9.0732, |
| "step": 1 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 0.0001, |
| "loss": 4.8824, |
| "step": 4 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 0.0001, |
| "loss": 2.1843, |
| "step": 8 |
| }, |
| { |
| "epoch": 3.88, |
| "learning_rate": 0.0001, |
| "loss": 1.2402, |
| "step": 12 |
| }, |
| { |
| "epoch": 5.29, |
| "learning_rate": 0.0001, |
| "loss": 0.8986, |
| "step": 16 |
| }, |
| { |
| "epoch": 6.59, |
| "learning_rate": 0.0001, |
| "loss": 0.6313, |
| "step": 20 |
| }, |
| { |
| "epoch": 7.88, |
| "learning_rate": 0.0001, |
| "loss": 0.5105, |
| "step": 24 |
| }, |
| { |
| "epoch": 9.29, |
| "learning_rate": 0.0001, |
| "loss": 0.4653, |
| "step": 28 |
| }, |
| { |
| "epoch": 10.59, |
| "learning_rate": 0.0001, |
| "loss": 0.3667, |
| "step": 32 |
| }, |
| { |
| "epoch": 11.88, |
| "learning_rate": 0.0001, |
| "loss": 0.3282, |
| "step": 36 |
| }, |
| { |
| "epoch": 13.29, |
| "learning_rate": 0.0001, |
| "loss": 0.3399, |
| "step": 40 |
| }, |
| { |
| "epoch": 14.59, |
| "learning_rate": 0.0001, |
| "loss": 0.2716, |
| "step": 44 |
| }, |
| { |
| "epoch": 15.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2606, |
| "step": 48 |
| }, |
| { |
| "epoch": 17.29, |
| "learning_rate": 0.0001, |
| "loss": 0.254, |
| "step": 52 |
| }, |
| { |
| "epoch": 18.59, |
| "learning_rate": 0.0001, |
| "loss": 0.2197, |
| "step": 56 |
| }, |
| { |
| "epoch": 19.88, |
| "learning_rate": 0.0001, |
| "loss": 0.2168, |
| "step": 60 |
| }, |
| { |
| "epoch": 21.29, |
| "learning_rate": 0.0001, |
| "loss": 0.2316, |
| "step": 64 |
| }, |
| { |
| "epoch": 21.29, |
| "eval_exact_match": 0.4071566731141199, |
| "eval_exec": 0.4400386847195358, |
| "eval_loss": 0.26413747668266296, |
| "eval_runtime": 110.5638, |
| "eval_samples_per_second": 9.352, |
| "step": 64 |
| }, |
| { |
| "epoch": 22.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1937, |
| "step": 68 |
| }, |
| { |
| "epoch": 23.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1791, |
| "step": 72 |
| }, |
| { |
| "epoch": 25.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1859, |
| "step": 76 |
| }, |
| { |
| "epoch": 26.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1742, |
| "step": 80 |
| }, |
| { |
| "epoch": 27.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1688, |
| "step": 84 |
| }, |
| { |
| "epoch": 29.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1793, |
| "step": 88 |
| }, |
| { |
| "epoch": 30.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1483, |
| "step": 92 |
| }, |
| { |
| "epoch": 31.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1396, |
| "step": 96 |
| }, |
| { |
| "epoch": 33.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1527, |
| "step": 100 |
| }, |
| { |
| "epoch": 34.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1404, |
| "step": 104 |
| }, |
| { |
| "epoch": 35.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1259, |
| "step": 108 |
| }, |
| { |
| "epoch": 37.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1322, |
| "step": 112 |
| }, |
| { |
| "epoch": 38.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1217, |
| "step": 116 |
| }, |
| { |
| "epoch": 39.88, |
| "learning_rate": 0.0001, |
| "loss": 0.115, |
| "step": 120 |
| }, |
| { |
| "epoch": 41.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1268, |
| "step": 124 |
| }, |
| { |
| "epoch": 42.59, |
| "learning_rate": 0.0001, |
| "loss": 0.1137, |
| "step": 128 |
| }, |
| { |
| "epoch": 42.59, |
| "eval_exact_match": 0.45938104448742745, |
| "eval_exec": 0.49032882011605416, |
| "eval_loss": 0.23041628301143646, |
| "eval_runtime": 152.3182, |
| "eval_samples_per_second": 6.788, |
| "step": 128 |
| }, |
| { |
| "epoch": 43.88, |
| "learning_rate": 0.0001, |
| "loss": 0.1038, |
| "step": 132 |
| }, |
| { |
| "epoch": 45.29, |
| "learning_rate": 0.0001, |
| "loss": 0.1112, |
| "step": 136 |
| }, |
| { |
| "epoch": 46.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0989, |
| "step": 140 |
| }, |
| { |
| "epoch": 47.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0955, |
| "step": 144 |
| }, |
| { |
| "epoch": 49.29, |
| "learning_rate": 0.0001, |
| "loss": 0.103, |
| "step": 148 |
| }, |
| { |
| "epoch": 50.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0954, |
| "step": 152 |
| }, |
| { |
| "epoch": 51.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0921, |
| "step": 156 |
| }, |
| { |
| "epoch": 53.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0941, |
| "step": 160 |
| }, |
| { |
| "epoch": 54.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0853, |
| "step": 164 |
| }, |
| { |
| "epoch": 55.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0838, |
| "step": 168 |
| }, |
| { |
| "epoch": 57.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0834, |
| "step": 172 |
| }, |
| { |
| "epoch": 58.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0759, |
| "step": 176 |
| }, |
| { |
| "epoch": 59.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0732, |
| "step": 180 |
| }, |
| { |
| "epoch": 61.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0819, |
| "step": 184 |
| }, |
| { |
| "epoch": 62.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0787, |
| "step": 188 |
| }, |
| { |
| "epoch": 63.88, |
| "learning_rate": 0.0001, |
| "loss": 0.066, |
| "step": 192 |
| }, |
| { |
| "epoch": 63.88, |
| "eval_exact_match": 0.4990328820116054, |
| "eval_exec": 0.5203094777562862, |
| "eval_loss": 0.2356981337070465, |
| "eval_runtime": 148.5976, |
| "eval_samples_per_second": 6.958, |
| "step": 192 |
| }, |
| { |
| "epoch": 65.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0701, |
| "step": 196 |
| }, |
| { |
| "epoch": 66.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0669, |
| "step": 200 |
| }, |
| { |
| "epoch": 67.88, |
| "learning_rate": 0.0001, |
| "loss": 0.066, |
| "step": 204 |
| }, |
| { |
| "epoch": 69.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0705, |
| "step": 208 |
| }, |
| { |
| "epoch": 70.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0581, |
| "step": 212 |
| }, |
| { |
| "epoch": 71.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0596, |
| "step": 216 |
| }, |
| { |
| "epoch": 73.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0621, |
| "step": 220 |
| }, |
| { |
| "epoch": 74.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0575, |
| "step": 224 |
| }, |
| { |
| "epoch": 75.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0571, |
| "step": 228 |
| }, |
| { |
| "epoch": 77.29, |
| "learning_rate": 0.0001, |
| "loss": 0.058, |
| "step": 232 |
| }, |
| { |
| "epoch": 78.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0516, |
| "step": 236 |
| }, |
| { |
| "epoch": 79.88, |
| "learning_rate": 0.0001, |
| "loss": 0.056, |
| "step": 240 |
| }, |
| { |
| "epoch": 81.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0545, |
| "step": 244 |
| }, |
| { |
| "epoch": 82.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0494, |
| "step": 248 |
| }, |
| { |
| "epoch": 83.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0494, |
| "step": 252 |
| }, |
| { |
| "epoch": 85.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0517, |
| "step": 256 |
| }, |
| { |
| "epoch": 85.29, |
| "eval_exact_match": 0.5029013539651838, |
| "eval_exec": 0.5174081237911026, |
| "eval_loss": 0.25896430015563965, |
| "eval_runtime": 156.1392, |
| "eval_samples_per_second": 6.622, |
| "step": 256 |
| }, |
| { |
| "epoch": 86.59, |
| "learning_rate": 0.0001, |
| "loss": 0.044, |
| "step": 260 |
| }, |
| { |
| "epoch": 87.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0451, |
| "step": 264 |
| }, |
| { |
| "epoch": 89.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0465, |
| "step": 268 |
| }, |
| { |
| "epoch": 90.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0428, |
| "step": 272 |
| }, |
| { |
| "epoch": 91.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0408, |
| "step": 276 |
| }, |
| { |
| "epoch": 93.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0423, |
| "step": 280 |
| }, |
| { |
| "epoch": 94.59, |
| "learning_rate": 0.0001, |
| "loss": 0.04, |
| "step": 284 |
| }, |
| { |
| "epoch": 95.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0399, |
| "step": 288 |
| }, |
| { |
| "epoch": 97.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0418, |
| "step": 292 |
| }, |
| { |
| "epoch": 98.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0368, |
| "step": 296 |
| }, |
| { |
| "epoch": 99.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0349, |
| "step": 300 |
| }, |
| { |
| "epoch": 101.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0366, |
| "step": 304 |
| }, |
| { |
| "epoch": 102.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0361, |
| "step": 308 |
| }, |
| { |
| "epoch": 103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0377, |
| "step": 312 |
| }, |
| { |
| "epoch": 105.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0345, |
| "step": 316 |
| }, |
| { |
| "epoch": 106.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0302, |
| "step": 320 |
| }, |
| { |
| "epoch": 106.59, |
| "eval_exact_match": 0.5135396518375241, |
| "eval_exec": 0.5299806576402321, |
| "eval_loss": 0.2772854268550873, |
| "eval_runtime": 151.1172, |
| "eval_samples_per_second": 6.842, |
| "step": 320 |
| }, |
| { |
| "epoch": 107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0303, |
| "step": 324 |
| }, |
| { |
| "epoch": 109.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0326, |
| "step": 328 |
| }, |
| { |
| "epoch": 110.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0331, |
| "step": 332 |
| }, |
| { |
| "epoch": 111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0298, |
| "step": 336 |
| }, |
| { |
| "epoch": 113.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0302, |
| "step": 340 |
| }, |
| { |
| "epoch": 114.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0276, |
| "step": 344 |
| }, |
| { |
| "epoch": 115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0307, |
| "step": 348 |
| }, |
| { |
| "epoch": 117.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0283, |
| "step": 352 |
| }, |
| { |
| "epoch": 118.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0267, |
| "step": 356 |
| }, |
| { |
| "epoch": 119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0313, |
| "step": 360 |
| }, |
| { |
| "epoch": 121.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0274, |
| "step": 364 |
| }, |
| { |
| "epoch": 122.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0259, |
| "step": 368 |
| }, |
| { |
| "epoch": 123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0264, |
| "step": 372 |
| }, |
| { |
| "epoch": 125.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0277, |
| "step": 376 |
| }, |
| { |
| "epoch": 126.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0225, |
| "step": 380 |
| }, |
| { |
| "epoch": 127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0229, |
| "step": 384 |
| }, |
| { |
| "epoch": 127.88, |
| "eval_exact_match": 0.5309477756286267, |
| "eval_exec": 0.5406189555125726, |
| "eval_loss": 0.28500038385391235, |
| "eval_runtime": 149.4099, |
| "eval_samples_per_second": 6.921, |
| "step": 384 |
| }, |
| { |
| "epoch": 129.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0242, |
| "step": 388 |
| }, |
| { |
| "epoch": 130.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0212, |
| "step": 392 |
| }, |
| { |
| "epoch": 131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0219, |
| "step": 396 |
| }, |
| { |
| "epoch": 133.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0237, |
| "step": 400 |
| }, |
| { |
| "epoch": 134.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0214, |
| "step": 404 |
| }, |
| { |
| "epoch": 135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0198, |
| "step": 408 |
| }, |
| { |
| "epoch": 137.29, |
| "learning_rate": 0.0001, |
| "loss": 0.02, |
| "step": 412 |
| }, |
| { |
| "epoch": 138.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0176, |
| "step": 416 |
| }, |
| { |
| "epoch": 139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0183, |
| "step": 420 |
| }, |
| { |
| "epoch": 141.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0221, |
| "step": 424 |
| }, |
| { |
| "epoch": 142.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0203, |
| "step": 428 |
| }, |
| { |
| "epoch": 143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0182, |
| "step": 432 |
| }, |
| { |
| "epoch": 145.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0214, |
| "step": 436 |
| }, |
| { |
| "epoch": 146.59, |
| "learning_rate": 0.0001, |
| "loss": 0.017, |
| "step": 440 |
| }, |
| { |
| "epoch": 147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0157, |
| "step": 444 |
| }, |
| { |
| "epoch": 149.29, |
| "learning_rate": 0.0001, |
| "loss": 0.017, |
| "step": 448 |
| }, |
| { |
| "epoch": 149.29, |
| "eval_exact_match": 0.5338491295938105, |
| "eval_exec": 0.5454545454545454, |
| "eval_loss": 0.3071914315223694, |
| "eval_runtime": 146.7029, |
| "eval_samples_per_second": 7.048, |
| "step": 448 |
| }, |
| { |
| "epoch": 150.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0164, |
| "step": 452 |
| }, |
| { |
| "epoch": 151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0161, |
| "step": 456 |
| }, |
| { |
| "epoch": 153.29, |
| "learning_rate": 0.0001, |
| "loss": 0.016, |
| "step": 460 |
| }, |
| { |
| "epoch": 154.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0159, |
| "step": 464 |
| }, |
| { |
| "epoch": 155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0144, |
| "step": 468 |
| }, |
| { |
| "epoch": 157.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0153, |
| "step": 472 |
| }, |
| { |
| "epoch": 158.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0147, |
| "step": 476 |
| }, |
| { |
| "epoch": 159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 480 |
| }, |
| { |
| "epoch": 161.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0147, |
| "step": 484 |
| }, |
| { |
| "epoch": 162.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0143, |
| "step": 488 |
| }, |
| { |
| "epoch": 163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0135, |
| "step": 492 |
| }, |
| { |
| "epoch": 165.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0134, |
| "step": 496 |
| }, |
| { |
| "epoch": 166.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0125, |
| "step": 500 |
| }, |
| { |
| "epoch": 167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0137, |
| "step": 504 |
| }, |
| { |
| "epoch": 169.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0128, |
| "step": 508 |
| }, |
| { |
| "epoch": 170.59, |
| "learning_rate": 0.0001, |
| "loss": 0.012, |
| "step": 512 |
| }, |
| { |
| "epoch": 170.59, |
| "eval_exact_match": 0.5464216634429401, |
| "eval_exec": 0.5551257253384912, |
| "eval_loss": 0.3156057894229889, |
| "eval_runtime": 155.4486, |
| "eval_samples_per_second": 6.652, |
| "step": 512 |
| }, |
| { |
| "epoch": 171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0113, |
| "step": 516 |
| }, |
| { |
| "epoch": 173.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0121, |
| "step": 520 |
| }, |
| { |
| "epoch": 174.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0112, |
| "step": 524 |
| }, |
| { |
| "epoch": 175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0118, |
| "step": 528 |
| }, |
| { |
| "epoch": 177.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0144, |
| "step": 532 |
| }, |
| { |
| "epoch": 178.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0135, |
| "step": 536 |
| }, |
| { |
| "epoch": 179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 540 |
| }, |
| { |
| "epoch": 181.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 544 |
| }, |
| { |
| "epoch": 182.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0102, |
| "step": 548 |
| }, |
| { |
| "epoch": 183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0105, |
| "step": 552 |
| }, |
| { |
| "epoch": 185.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0111, |
| "step": 556 |
| }, |
| { |
| "epoch": 186.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0097, |
| "step": 560 |
| }, |
| { |
| "epoch": 187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 564 |
| }, |
| { |
| "epoch": 189.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0119, |
| "step": 568 |
| }, |
| { |
| "epoch": 190.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 572 |
| }, |
| { |
| "epoch": 191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0093, |
| "step": 576 |
| }, |
| { |
| "epoch": 191.88, |
| "eval_exact_match": 0.5512572533849129, |
| "eval_exec": 0.5589941972920697, |
| "eval_loss": 0.3371215760707855, |
| "eval_runtime": 145.9635, |
| "eval_samples_per_second": 7.084, |
| "step": 576 |
| }, |
| { |
| "epoch": 193.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0098, |
| "step": 580 |
| }, |
| { |
| "epoch": 194.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 584 |
| }, |
| { |
| "epoch": 195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.008, |
| "step": 588 |
| }, |
| { |
| "epoch": 197.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0089, |
| "step": 592 |
| }, |
| { |
| "epoch": 198.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 596 |
| }, |
| { |
| "epoch": 199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 600 |
| }, |
| { |
| "epoch": 201.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0085, |
| "step": 604 |
| }, |
| { |
| "epoch": 202.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0081, |
| "step": 608 |
| }, |
| { |
| "epoch": 203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0084, |
| "step": 612 |
| }, |
| { |
| "epoch": 205.29, |
| "learning_rate": 0.0001, |
| "loss": 0.009, |
| "step": 616 |
| }, |
| { |
| "epoch": 206.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0079, |
| "step": 620 |
| }, |
| { |
| "epoch": 207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0074, |
| "step": 624 |
| }, |
| { |
| "epoch": 209.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0077, |
| "step": 628 |
| }, |
| { |
| "epoch": 210.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 632 |
| }, |
| { |
| "epoch": 211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0069, |
| "step": 636 |
| }, |
| { |
| "epoch": 213.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0073, |
| "step": 640 |
| }, |
| { |
| "epoch": 213.29, |
| "eval_exact_match": 0.5696324951644101, |
| "eval_exec": 0.5812379110251451, |
| "eval_loss": 0.3502403795719147, |
| "eval_runtime": 151.5831, |
| "eval_samples_per_second": 6.821, |
| "step": 640 |
| }, |
| { |
| "epoch": 214.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0065, |
| "step": 644 |
| }, |
| { |
| "epoch": 215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0064, |
| "step": 648 |
| }, |
| { |
| "epoch": 217.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0069, |
| "step": 652 |
| }, |
| { |
| "epoch": 218.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0067, |
| "step": 656 |
| }, |
| { |
| "epoch": 219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 660 |
| }, |
| { |
| "epoch": 221.29, |
| "learning_rate": 0.0001, |
| "loss": 0.007, |
| "step": 664 |
| }, |
| { |
| "epoch": 222.59, |
| "learning_rate": 0.0001, |
| "loss": 0.006, |
| "step": 668 |
| }, |
| { |
| "epoch": 223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 672 |
| }, |
| { |
| "epoch": 225.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0062, |
| "step": 676 |
| }, |
| { |
| "epoch": 226.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 680 |
| }, |
| { |
| "epoch": 227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 684 |
| }, |
| { |
| "epoch": 229.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0063, |
| "step": 688 |
| }, |
| { |
| "epoch": 230.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0135, |
| "step": 692 |
| }, |
| { |
| "epoch": 231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0068, |
| "step": 696 |
| }, |
| { |
| "epoch": 233.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0059, |
| "step": 700 |
| }, |
| { |
| "epoch": 234.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0056, |
| "step": 704 |
| }, |
| { |
| "epoch": 234.59, |
| "eval_exact_match": 0.5647969052224371, |
| "eval_exec": 0.5735009671179884, |
| "eval_loss": 0.3497712314128876, |
| "eval_runtime": 145.6537, |
| "eval_samples_per_second": 7.099, |
| "step": 704 |
| }, |
| { |
| "epoch": 235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 708 |
| }, |
| { |
| "epoch": 237.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0055, |
| "step": 712 |
| }, |
| { |
| "epoch": 238.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0047, |
| "step": 716 |
| }, |
| { |
| "epoch": 239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0053, |
| "step": 720 |
| }, |
| { |
| "epoch": 241.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0114, |
| "step": 724 |
| }, |
| { |
| "epoch": 242.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 728 |
| }, |
| { |
| "epoch": 243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0049, |
| "step": 732 |
| }, |
| { |
| "epoch": 245.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 736 |
| }, |
| { |
| "epoch": 246.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 740 |
| }, |
| { |
| "epoch": 247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 744 |
| }, |
| { |
| "epoch": 249.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 748 |
| }, |
| { |
| "epoch": 250.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0048, |
| "step": 752 |
| }, |
| { |
| "epoch": 251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 756 |
| }, |
| { |
| "epoch": 253.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0086, |
| "step": 760 |
| }, |
| { |
| "epoch": 254.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0046, |
| "step": 764 |
| }, |
| { |
| "epoch": 255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0042, |
| "step": 768 |
| }, |
| { |
| "epoch": 255.88, |
| "eval_exact_match": 0.5560928433268859, |
| "eval_exec": 0.574468085106383, |
| "eval_loss": 0.35794734954833984, |
| "eval_runtime": 154.8692, |
| "eval_samples_per_second": 6.677, |
| "step": 768 |
| }, |
| { |
| "epoch": 257.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 772 |
| }, |
| { |
| "epoch": 258.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 776 |
| }, |
| { |
| "epoch": 259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 780 |
| }, |
| { |
| "epoch": 261.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0044, |
| "step": 784 |
| }, |
| { |
| "epoch": 262.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 788 |
| }, |
| { |
| "epoch": 263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 792 |
| }, |
| { |
| "epoch": 265.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 796 |
| }, |
| { |
| "epoch": 266.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 800 |
| }, |
| { |
| "epoch": 267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 804 |
| }, |
| { |
| "epoch": 269.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0041, |
| "step": 808 |
| }, |
| { |
| "epoch": 270.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0036, |
| "step": 812 |
| }, |
| { |
| "epoch": 271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 816 |
| }, |
| { |
| "epoch": 273.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0039, |
| "step": 820 |
| }, |
| { |
| "epoch": 274.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 824 |
| }, |
| { |
| "epoch": 275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 828 |
| }, |
| { |
| "epoch": 277.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 832 |
| }, |
| { |
| "epoch": 277.29, |
| "eval_exact_match": 0.5618955512572534, |
| "eval_exec": 0.5696324951644101, |
| "eval_loss": 0.3841744661331177, |
| "eval_runtime": 147.3603, |
| "eval_samples_per_second": 7.017, |
| "step": 832 |
| }, |
| { |
| "epoch": 278.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 836 |
| }, |
| { |
| "epoch": 279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 840 |
| }, |
| { |
| "epoch": 281.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 844 |
| }, |
| { |
| "epoch": 282.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0037, |
| "step": 848 |
| }, |
| { |
| "epoch": 283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0034, |
| "step": 852 |
| }, |
| { |
| "epoch": 285.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0038, |
| "step": 856 |
| }, |
| { |
| "epoch": 286.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0057, |
| "step": 860 |
| }, |
| { |
| "epoch": 287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 864 |
| }, |
| { |
| "epoch": 289.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 868 |
| }, |
| { |
| "epoch": 290.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0035, |
| "step": 872 |
| }, |
| { |
| "epoch": 291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 876 |
| }, |
| { |
| "epoch": 293.29, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 880 |
| }, |
| { |
| "epoch": 294.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0033, |
| "step": 884 |
| }, |
| { |
| "epoch": 295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 888 |
| }, |
| { |
| "epoch": 297.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0032, |
| "step": 892 |
| }, |
| { |
| "epoch": 298.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 896 |
| }, |
| { |
| "epoch": 298.59, |
| "eval_exact_match": 0.5676982591876208, |
| "eval_exec": 0.5802707930367504, |
| "eval_loss": 0.3835436701774597, |
| "eval_runtime": 146.0079, |
| "eval_samples_per_second": 7.082, |
| "step": 896 |
| }, |
| { |
| "epoch": 299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 900 |
| }, |
| { |
| "epoch": 301.29, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 904 |
| }, |
| { |
| "epoch": 302.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 908 |
| }, |
| { |
| "epoch": 303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 912 |
| }, |
| { |
| "epoch": 305.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 916 |
| }, |
| { |
| "epoch": 306.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 920 |
| }, |
| { |
| "epoch": 307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 924 |
| }, |
| { |
| "epoch": 309.29, |
| "learning_rate": 0.0001, |
| "loss": 0.003, |
| "step": 928 |
| }, |
| { |
| "epoch": 310.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 932 |
| }, |
| { |
| "epoch": 311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 936 |
| }, |
| { |
| "epoch": 313.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0027, |
| "step": 940 |
| }, |
| { |
| "epoch": 314.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 944 |
| }, |
| { |
| "epoch": 315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 948 |
| }, |
| { |
| "epoch": 317.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 952 |
| }, |
| { |
| "epoch": 318.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 956 |
| }, |
| { |
| "epoch": 319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 960 |
| }, |
| { |
| "epoch": 319.88, |
| "eval_exact_match": 0.5705996131528046, |
| "eval_exec": 0.5851063829787234, |
| "eval_loss": 0.40229931473731995, |
| "eval_runtime": 148.3785, |
| "eval_samples_per_second": 6.969, |
| "step": 960 |
| }, |
| { |
| "epoch": 321.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0028, |
| "step": 964 |
| }, |
| { |
| "epoch": 322.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 968 |
| }, |
| { |
| "epoch": 323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 972 |
| }, |
| { |
| "epoch": 325.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 976 |
| }, |
| { |
| "epoch": 326.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 980 |
| }, |
| { |
| "epoch": 327.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 984 |
| }, |
| { |
| "epoch": 329.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 988 |
| }, |
| { |
| "epoch": 330.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 992 |
| }, |
| { |
| "epoch": 331.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 996 |
| }, |
| { |
| "epoch": 333.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 1000 |
| }, |
| { |
| "epoch": 334.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 1004 |
| }, |
| { |
| "epoch": 335.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 1008 |
| }, |
| { |
| "epoch": 337.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0025, |
| "step": 1012 |
| }, |
| { |
| "epoch": 338.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1016 |
| }, |
| { |
| "epoch": 339.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 1020 |
| }, |
| { |
| "epoch": 341.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 1024 |
| }, |
| { |
| "epoch": 341.29, |
| "eval_exact_match": 0.562862669245648, |
| "eval_exec": 0.5812379110251451, |
| "eval_loss": 0.39506930112838745, |
| "eval_runtime": 149.0359, |
| "eval_samples_per_second": 6.938, |
| "step": 1024 |
| }, |
| { |
| "epoch": 342.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0026, |
| "step": 1028 |
| }, |
| { |
| "epoch": 343.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1032 |
| }, |
| { |
| "epoch": 345.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1036 |
| }, |
| { |
| "epoch": 346.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 1040 |
| }, |
| { |
| "epoch": 347.88, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 1044 |
| }, |
| { |
| "epoch": 349.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 1048 |
| }, |
| { |
| "epoch": 350.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1052 |
| }, |
| { |
| "epoch": 351.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1056 |
| }, |
| { |
| "epoch": 353.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1060 |
| }, |
| { |
| "epoch": 354.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1064 |
| }, |
| { |
| "epoch": 355.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0024, |
| "step": 1068 |
| }, |
| { |
| "epoch": 357.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0023, |
| "step": 1072 |
| }, |
| { |
| "epoch": 358.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1076 |
| }, |
| { |
| "epoch": 359.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1080 |
| }, |
| { |
| "epoch": 361.29, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 1084 |
| }, |
| { |
| "epoch": 362.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1088 |
| }, |
| { |
| "epoch": 362.59, |
| "eval_exact_match": 0.5589941972920697, |
| "eval_exec": 0.5715667311411993, |
| "eval_loss": 0.40029463171958923, |
| "eval_runtime": 149.9391, |
| "eval_samples_per_second": 6.896, |
| "step": 1088 |
| }, |
| { |
| "epoch": 363.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 1092 |
| }, |
| { |
| "epoch": 365.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0022, |
| "step": 1096 |
| }, |
| { |
| "epoch": 366.59, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 1100 |
| }, |
| { |
| "epoch": 367.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1104 |
| }, |
| { |
| "epoch": 369.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1108 |
| }, |
| { |
| "epoch": 370.59, |
| "learning_rate": 0.0001, |
| "loss": 0.002, |
| "step": 1112 |
| }, |
| { |
| "epoch": 371.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1116 |
| }, |
| { |
| "epoch": 373.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1120 |
| }, |
| { |
| "epoch": 374.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1124 |
| }, |
| { |
| "epoch": 375.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1128 |
| }, |
| { |
| "epoch": 377.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 1132 |
| }, |
| { |
| "epoch": 378.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1136 |
| }, |
| { |
| "epoch": 379.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1140 |
| }, |
| { |
| "epoch": 381.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 1144 |
| }, |
| { |
| "epoch": 382.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1148 |
| }, |
| { |
| "epoch": 383.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1152 |
| }, |
| { |
| "epoch": 383.88, |
| "eval_exact_match": 0.558027079303675, |
| "eval_exec": 0.574468085106383, |
| "eval_loss": 0.4003511965274811, |
| "eval_runtime": 145.6547, |
| "eval_samples_per_second": 7.099, |
| "step": 1152 |
| }, |
| { |
| "epoch": 385.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1156 |
| }, |
| { |
| "epoch": 386.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1160 |
| }, |
| { |
| "epoch": 387.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1164 |
| }, |
| { |
| "epoch": 389.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1168 |
| }, |
| { |
| "epoch": 390.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1172 |
| }, |
| { |
| "epoch": 391.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1176 |
| }, |
| { |
| "epoch": 393.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1180 |
| }, |
| { |
| "epoch": 394.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 1184 |
| }, |
| { |
| "epoch": 395.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1188 |
| }, |
| { |
| "epoch": 397.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0018, |
| "step": 1192 |
| }, |
| { |
| "epoch": 398.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1196 |
| }, |
| { |
| "epoch": 399.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1200 |
| }, |
| { |
| "epoch": 401.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1204 |
| }, |
| { |
| "epoch": 402.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 1208 |
| }, |
| { |
| "epoch": 403.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1212 |
| }, |
| { |
| "epoch": 405.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1216 |
| }, |
| { |
| "epoch": 405.29, |
| "eval_exact_match": 0.5667311411992263, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.4143199026584625, |
| "eval_runtime": 149.8872, |
| "eval_samples_per_second": 6.899, |
| "step": 1216 |
| }, |
| { |
| "epoch": 406.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1220 |
| }, |
| { |
| "epoch": 407.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1224 |
| }, |
| { |
| "epoch": 409.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0017, |
| "step": 1228 |
| }, |
| { |
| "epoch": 410.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1232 |
| }, |
| { |
| "epoch": 411.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1236 |
| }, |
| { |
| "epoch": 413.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1240 |
| }, |
| { |
| "epoch": 414.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1244 |
| }, |
| { |
| "epoch": 415.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 1248 |
| }, |
| { |
| "epoch": 417.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 1252 |
| }, |
| { |
| "epoch": 418.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 1256 |
| }, |
| { |
| "epoch": 419.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1260 |
| }, |
| { |
| "epoch": 421.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1264 |
| }, |
| { |
| "epoch": 422.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1268 |
| }, |
| { |
| "epoch": 423.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1272 |
| }, |
| { |
| "epoch": 425.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1276 |
| }, |
| { |
| "epoch": 426.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1280 |
| }, |
| { |
| "epoch": 426.59, |
| "eval_exact_match": 0.5618955512572534, |
| "eval_exec": 0.5783365570599613, |
| "eval_loss": 0.42446255683898926, |
| "eval_runtime": 143.8874, |
| "eval_samples_per_second": 7.186, |
| "step": 1280 |
| }, |
| { |
| "epoch": 427.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1284 |
| }, |
| { |
| "epoch": 429.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1288 |
| }, |
| { |
| "epoch": 430.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1292 |
| }, |
| { |
| "epoch": 431.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1296 |
| }, |
| { |
| "epoch": 433.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1300 |
| }, |
| { |
| "epoch": 434.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1304 |
| }, |
| { |
| "epoch": 435.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1308 |
| }, |
| { |
| "epoch": 437.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1312 |
| }, |
| { |
| "epoch": 438.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1316 |
| }, |
| { |
| "epoch": 439.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1320 |
| }, |
| { |
| "epoch": 441.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1324 |
| }, |
| { |
| "epoch": 442.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1328 |
| }, |
| { |
| "epoch": 443.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1332 |
| }, |
| { |
| "epoch": 445.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1336 |
| }, |
| { |
| "epoch": 446.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1340 |
| }, |
| { |
| "epoch": 447.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1344 |
| }, |
| { |
| "epoch": 447.88, |
| "eval_exact_match": 0.558027079303675, |
| "eval_exec": 0.5754352030947776, |
| "eval_loss": 0.42454418540000916, |
| "eval_runtime": 148.0659, |
| "eval_samples_per_second": 6.983, |
| "step": 1344 |
| }, |
| { |
| "epoch": 449.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1348 |
| }, |
| { |
| "epoch": 450.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1352 |
| }, |
| { |
| "epoch": 451.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1356 |
| }, |
| { |
| "epoch": 453.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1360 |
| }, |
| { |
| "epoch": 454.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1364 |
| }, |
| { |
| "epoch": 455.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1368 |
| }, |
| { |
| "epoch": 457.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1372 |
| }, |
| { |
| "epoch": 458.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1376 |
| }, |
| { |
| "epoch": 459.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1380 |
| }, |
| { |
| "epoch": 461.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1384 |
| }, |
| { |
| "epoch": 462.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1388 |
| }, |
| { |
| "epoch": 463.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1392 |
| }, |
| { |
| "epoch": 465.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1396 |
| }, |
| { |
| "epoch": 466.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1400 |
| }, |
| { |
| "epoch": 467.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1404 |
| }, |
| { |
| "epoch": 469.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1408 |
| }, |
| { |
| "epoch": 469.29, |
| "eval_exact_match": 0.5618955512572534, |
| "eval_exec": 0.5802707930367504, |
| "eval_loss": 0.4586906433105469, |
| "eval_runtime": 149.7041, |
| "eval_samples_per_second": 6.907, |
| "step": 1408 |
| }, |
| { |
| "epoch": 470.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1412 |
| }, |
| { |
| "epoch": 471.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1416 |
| }, |
| { |
| "epoch": 473.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0015, |
| "step": 1420 |
| }, |
| { |
| "epoch": 474.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1424 |
| }, |
| { |
| "epoch": 475.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1428 |
| }, |
| { |
| "epoch": 477.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1432 |
| }, |
| { |
| "epoch": 478.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1436 |
| }, |
| { |
| "epoch": 479.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1440 |
| }, |
| { |
| "epoch": 481.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1444 |
| }, |
| { |
| "epoch": 482.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1448 |
| }, |
| { |
| "epoch": 483.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1452 |
| }, |
| { |
| "epoch": 485.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1456 |
| }, |
| { |
| "epoch": 486.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1460 |
| }, |
| { |
| "epoch": 487.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1464 |
| }, |
| { |
| "epoch": 489.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1468 |
| }, |
| { |
| "epoch": 490.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1472 |
| }, |
| { |
| "epoch": 490.59, |
| "eval_exact_match": 0.558027079303675, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.4234982132911682, |
| "eval_runtime": 147.8428, |
| "eval_samples_per_second": 6.994, |
| "step": 1472 |
| }, |
| { |
| "epoch": 491.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1476 |
| }, |
| { |
| "epoch": 493.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1480 |
| }, |
| { |
| "epoch": 494.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1484 |
| }, |
| { |
| "epoch": 495.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1488 |
| }, |
| { |
| "epoch": 497.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1492 |
| }, |
| { |
| "epoch": 498.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1496 |
| }, |
| { |
| "epoch": 499.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1500 |
| }, |
| { |
| "epoch": 501.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1504 |
| }, |
| { |
| "epoch": 502.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1508 |
| }, |
| { |
| "epoch": 503.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1512 |
| }, |
| { |
| "epoch": 505.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1516 |
| }, |
| { |
| "epoch": 506.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1520 |
| }, |
| { |
| "epoch": 507.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1524 |
| }, |
| { |
| "epoch": 509.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1528 |
| }, |
| { |
| "epoch": 510.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1532 |
| }, |
| { |
| "epoch": 511.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1536 |
| }, |
| { |
| "epoch": 511.88, |
| "eval_exact_match": 0.5725338491295938, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.4380647838115692, |
| "eval_runtime": 146.4649, |
| "eval_samples_per_second": 7.06, |
| "step": 1536 |
| }, |
| { |
| "epoch": 513.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1540 |
| }, |
| { |
| "epoch": 514.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1544 |
| }, |
| { |
| "epoch": 515.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1548 |
| }, |
| { |
| "epoch": 517.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1552 |
| }, |
| { |
| "epoch": 518.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 1556 |
| }, |
| { |
| "epoch": 519.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 1560 |
| }, |
| { |
| "epoch": 521.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 1564 |
| }, |
| { |
| "epoch": 522.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1568 |
| }, |
| { |
| "epoch": 523.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1572 |
| }, |
| { |
| "epoch": 525.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1576 |
| }, |
| { |
| "epoch": 526.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1580 |
| }, |
| { |
| "epoch": 527.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1584 |
| }, |
| { |
| "epoch": 529.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1588 |
| }, |
| { |
| "epoch": 530.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1592 |
| }, |
| { |
| "epoch": 531.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1596 |
| }, |
| { |
| "epoch": 533.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1600 |
| }, |
| { |
| "epoch": 533.29, |
| "eval_exact_match": 0.5531914893617021, |
| "eval_exec": 0.5705996131528046, |
| "eval_loss": 0.44653236865997314, |
| "eval_runtime": 152.4236, |
| "eval_samples_per_second": 6.784, |
| "step": 1600 |
| }, |
| { |
| "epoch": 534.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1604 |
| }, |
| { |
| "epoch": 535.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1608 |
| }, |
| { |
| "epoch": 537.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1612 |
| }, |
| { |
| "epoch": 538.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1616 |
| }, |
| { |
| "epoch": 539.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1620 |
| }, |
| { |
| "epoch": 541.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1624 |
| }, |
| { |
| "epoch": 542.59, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1628 |
| }, |
| { |
| "epoch": 543.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1632 |
| }, |
| { |
| "epoch": 545.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1636 |
| }, |
| { |
| "epoch": 546.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1640 |
| }, |
| { |
| "epoch": 547.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1644 |
| }, |
| { |
| "epoch": 549.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1648 |
| }, |
| { |
| "epoch": 550.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1652 |
| }, |
| { |
| "epoch": 551.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1656 |
| }, |
| { |
| "epoch": 553.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 1660 |
| }, |
| { |
| "epoch": 554.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1664 |
| }, |
| { |
| "epoch": 554.59, |
| "eval_exact_match": 0.5783365570599613, |
| "eval_exec": 0.59284332688588, |
| "eval_loss": 0.43879660964012146, |
| "eval_runtime": 145.1998, |
| "eval_samples_per_second": 7.121, |
| "step": 1664 |
| }, |
| { |
| "epoch": 555.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1668 |
| }, |
| { |
| "epoch": 557.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1672 |
| }, |
| { |
| "epoch": 558.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1676 |
| }, |
| { |
| "epoch": 559.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1680 |
| }, |
| { |
| "epoch": 561.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1684 |
| }, |
| { |
| "epoch": 562.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1688 |
| }, |
| { |
| "epoch": 563.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 1692 |
| }, |
| { |
| "epoch": 565.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1696 |
| }, |
| { |
| "epoch": 566.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1700 |
| }, |
| { |
| "epoch": 567.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1704 |
| }, |
| { |
| "epoch": 569.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1708 |
| }, |
| { |
| "epoch": 570.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1712 |
| }, |
| { |
| "epoch": 571.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1716 |
| }, |
| { |
| "epoch": 573.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1720 |
| }, |
| { |
| "epoch": 574.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1724 |
| }, |
| { |
| "epoch": 575.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1728 |
| }, |
| { |
| "epoch": 575.88, |
| "eval_exact_match": 0.562862669245648, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.4501326382160187, |
| "eval_runtime": 146.0158, |
| "eval_samples_per_second": 7.081, |
| "step": 1728 |
| }, |
| { |
| "epoch": 577.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1732 |
| }, |
| { |
| "epoch": 578.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1736 |
| }, |
| { |
| "epoch": 579.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1740 |
| }, |
| { |
| "epoch": 581.29, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1744 |
| }, |
| { |
| "epoch": 582.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1748 |
| }, |
| { |
| "epoch": 583.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1752 |
| }, |
| { |
| "epoch": 585.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1756 |
| }, |
| { |
| "epoch": 586.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1760 |
| }, |
| { |
| "epoch": 587.88, |
| "learning_rate": 0.0001, |
| "loss": 0.001, |
| "step": 1764 |
| }, |
| { |
| "epoch": 589.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 1768 |
| }, |
| { |
| "epoch": 590.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1772 |
| }, |
| { |
| "epoch": 591.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 1776 |
| }, |
| { |
| "epoch": 593.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1780 |
| }, |
| { |
| "epoch": 594.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 1784 |
| }, |
| { |
| "epoch": 595.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0014, |
| "step": 1788 |
| }, |
| { |
| "epoch": 597.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1792 |
| }, |
| { |
| "epoch": 597.29, |
| "eval_exact_match": 0.5589941972920697, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.4375011622905731, |
| "eval_runtime": 148.7299, |
| "eval_samples_per_second": 6.952, |
| "step": 1792 |
| }, |
| { |
| "epoch": 598.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1796 |
| }, |
| { |
| "epoch": 599.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1800 |
| }, |
| { |
| "epoch": 601.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1804 |
| }, |
| { |
| "epoch": 602.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1808 |
| }, |
| { |
| "epoch": 603.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1812 |
| }, |
| { |
| "epoch": 605.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1816 |
| }, |
| { |
| "epoch": 606.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1820 |
| }, |
| { |
| "epoch": 607.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1824 |
| }, |
| { |
| "epoch": 609.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1828 |
| }, |
| { |
| "epoch": 610.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1832 |
| }, |
| { |
| "epoch": 611.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0016, |
| "step": 1836 |
| }, |
| { |
| "epoch": 613.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0071, |
| "step": 1840 |
| }, |
| { |
| "epoch": 614.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1844 |
| }, |
| { |
| "epoch": 615.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1848 |
| }, |
| { |
| "epoch": 617.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1852 |
| }, |
| { |
| "epoch": 618.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1856 |
| }, |
| { |
| "epoch": 618.59, |
| "eval_exact_match": 0.574468085106383, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.44161295890808105, |
| "eval_runtime": 150.3696, |
| "eval_samples_per_second": 6.876, |
| "step": 1856 |
| }, |
| { |
| "epoch": 619.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1860 |
| }, |
| { |
| "epoch": 621.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1864 |
| }, |
| { |
| "epoch": 622.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1868 |
| }, |
| { |
| "epoch": 623.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1872 |
| }, |
| { |
| "epoch": 625.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1876 |
| }, |
| { |
| "epoch": 626.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1880 |
| }, |
| { |
| "epoch": 627.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1884 |
| }, |
| { |
| "epoch": 629.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1888 |
| }, |
| { |
| "epoch": 630.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1892 |
| }, |
| { |
| "epoch": 631.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1896 |
| }, |
| { |
| "epoch": 633.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1900 |
| }, |
| { |
| "epoch": 634.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1904 |
| }, |
| { |
| "epoch": 635.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1908 |
| }, |
| { |
| "epoch": 637.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1912 |
| }, |
| { |
| "epoch": 638.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1916 |
| }, |
| { |
| "epoch": 639.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1920 |
| }, |
| { |
| "epoch": 639.88, |
| "eval_exact_match": 0.5715667311411993, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.4707731306552887, |
| "eval_runtime": 143.7262, |
| "eval_samples_per_second": 7.194, |
| "step": 1920 |
| }, |
| { |
| "epoch": 641.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1924 |
| }, |
| { |
| "epoch": 642.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1928 |
| }, |
| { |
| "epoch": 643.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1932 |
| }, |
| { |
| "epoch": 645.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 1936 |
| }, |
| { |
| "epoch": 646.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1940 |
| }, |
| { |
| "epoch": 647.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1944 |
| }, |
| { |
| "epoch": 649.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1948 |
| }, |
| { |
| "epoch": 650.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1952 |
| }, |
| { |
| "epoch": 651.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1956 |
| }, |
| { |
| "epoch": 653.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1960 |
| }, |
| { |
| "epoch": 654.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1964 |
| }, |
| { |
| "epoch": 655.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 1968 |
| }, |
| { |
| "epoch": 657.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1972 |
| }, |
| { |
| "epoch": 658.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 1976 |
| }, |
| { |
| "epoch": 659.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1980 |
| }, |
| { |
| "epoch": 661.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 1984 |
| }, |
| { |
| "epoch": 661.29, |
| "eval_exact_match": 0.5667311411992263, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.4614003896713257, |
| "eval_runtime": 146.4514, |
| "eval_samples_per_second": 7.06, |
| "step": 1984 |
| }, |
| { |
| "epoch": 662.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1988 |
| }, |
| { |
| "epoch": 663.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1992 |
| }, |
| { |
| "epoch": 665.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 1996 |
| }, |
| { |
| "epoch": 666.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2000 |
| }, |
| { |
| "epoch": 667.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2004 |
| }, |
| { |
| "epoch": 669.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2008 |
| }, |
| { |
| "epoch": 670.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2012 |
| }, |
| { |
| "epoch": 671.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2016 |
| }, |
| { |
| "epoch": 673.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2020 |
| }, |
| { |
| "epoch": 674.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2024 |
| }, |
| { |
| "epoch": 675.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2028 |
| }, |
| { |
| "epoch": 677.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2032 |
| }, |
| { |
| "epoch": 678.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2036 |
| }, |
| { |
| "epoch": 679.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2040 |
| }, |
| { |
| "epoch": 681.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2044 |
| }, |
| { |
| "epoch": 682.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2048 |
| }, |
| { |
| "epoch": 682.59, |
| "eval_exact_match": 0.5638297872340425, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.4639793932437897, |
| "eval_runtime": 146.7707, |
| "eval_samples_per_second": 7.045, |
| "step": 2048 |
| }, |
| { |
| "epoch": 683.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2052 |
| }, |
| { |
| "epoch": 685.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2056 |
| }, |
| { |
| "epoch": 686.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2060 |
| }, |
| { |
| "epoch": 687.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2064 |
| }, |
| { |
| "epoch": 689.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2068 |
| }, |
| { |
| "epoch": 690.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2072 |
| }, |
| { |
| "epoch": 691.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2076 |
| }, |
| { |
| "epoch": 693.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2080 |
| }, |
| { |
| "epoch": 694.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2084 |
| }, |
| { |
| "epoch": 695.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2088 |
| }, |
| { |
| "epoch": 697.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2092 |
| }, |
| { |
| "epoch": 698.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2096 |
| }, |
| { |
| "epoch": 699.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2100 |
| }, |
| { |
| "epoch": 701.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2104 |
| }, |
| { |
| "epoch": 702.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2108 |
| }, |
| { |
| "epoch": 703.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2112 |
| }, |
| { |
| "epoch": 703.88, |
| "eval_exact_match": 0.5676982591876208, |
| "eval_exec": 0.5841392649903289, |
| "eval_loss": 0.46682995557785034, |
| "eval_runtime": 142.6808, |
| "eval_samples_per_second": 7.247, |
| "step": 2112 |
| }, |
| { |
| "epoch": 705.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2116 |
| }, |
| { |
| "epoch": 706.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2120 |
| }, |
| { |
| "epoch": 707.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2124 |
| }, |
| { |
| "epoch": 709.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2128 |
| }, |
| { |
| "epoch": 710.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2132 |
| }, |
| { |
| "epoch": 711.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2136 |
| }, |
| { |
| "epoch": 713.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2140 |
| }, |
| { |
| "epoch": 714.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2144 |
| }, |
| { |
| "epoch": 715.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2148 |
| }, |
| { |
| "epoch": 717.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2152 |
| }, |
| { |
| "epoch": 718.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2156 |
| }, |
| { |
| "epoch": 719.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2160 |
| }, |
| { |
| "epoch": 721.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2164 |
| }, |
| { |
| "epoch": 722.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2168 |
| }, |
| { |
| "epoch": 723.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2172 |
| }, |
| { |
| "epoch": 725.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2176 |
| }, |
| { |
| "epoch": 725.29, |
| "eval_exact_match": 0.5618955512572534, |
| "eval_exec": 0.5773694390715667, |
| "eval_loss": 0.4697101414203644, |
| "eval_runtime": 152.233, |
| "eval_samples_per_second": 6.792, |
| "step": 2176 |
| }, |
| { |
| "epoch": 726.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2180 |
| }, |
| { |
| "epoch": 727.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2184 |
| }, |
| { |
| "epoch": 729.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2188 |
| }, |
| { |
| "epoch": 730.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2192 |
| }, |
| { |
| "epoch": 731.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2196 |
| }, |
| { |
| "epoch": 733.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2200 |
| }, |
| { |
| "epoch": 734.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2204 |
| }, |
| { |
| "epoch": 735.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2208 |
| }, |
| { |
| "epoch": 737.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2212 |
| }, |
| { |
| "epoch": 738.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2216 |
| }, |
| { |
| "epoch": 739.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2220 |
| }, |
| { |
| "epoch": 741.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2224 |
| }, |
| { |
| "epoch": 742.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2228 |
| }, |
| { |
| "epoch": 743.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2232 |
| }, |
| { |
| "epoch": 745.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2236 |
| }, |
| { |
| "epoch": 746.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2240 |
| }, |
| { |
| "epoch": 746.59, |
| "eval_exact_match": 0.5725338491295938, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.4622710347175598, |
| "eval_runtime": 156.3741, |
| "eval_samples_per_second": 6.612, |
| "step": 2240 |
| }, |
| { |
| "epoch": 747.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2244 |
| }, |
| { |
| "epoch": 749.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2248 |
| }, |
| { |
| "epoch": 750.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2252 |
| }, |
| { |
| "epoch": 751.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2256 |
| }, |
| { |
| "epoch": 753.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2260 |
| }, |
| { |
| "epoch": 754.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2264 |
| }, |
| { |
| "epoch": 755.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2268 |
| }, |
| { |
| "epoch": 757.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2272 |
| }, |
| { |
| "epoch": 758.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2276 |
| }, |
| { |
| "epoch": 759.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2280 |
| }, |
| { |
| "epoch": 761.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2284 |
| }, |
| { |
| "epoch": 762.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2288 |
| }, |
| { |
| "epoch": 763.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2292 |
| }, |
| { |
| "epoch": 765.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 2296 |
| }, |
| { |
| "epoch": 766.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2300 |
| }, |
| { |
| "epoch": 767.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2304 |
| }, |
| { |
| "epoch": 767.88, |
| "eval_exact_match": 0.5676982591876208, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.4578416049480438, |
| "eval_runtime": 148.569, |
| "eval_samples_per_second": 6.96, |
| "step": 2304 |
| }, |
| { |
| "epoch": 769.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2308 |
| }, |
| { |
| "epoch": 770.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2312 |
| }, |
| { |
| "epoch": 771.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2316 |
| }, |
| { |
| "epoch": 773.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2320 |
| }, |
| { |
| "epoch": 774.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2324 |
| }, |
| { |
| "epoch": 775.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2328 |
| }, |
| { |
| "epoch": 777.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2332 |
| }, |
| { |
| "epoch": 778.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2336 |
| }, |
| { |
| "epoch": 779.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2340 |
| }, |
| { |
| "epoch": 781.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2344 |
| }, |
| { |
| "epoch": 782.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2348 |
| }, |
| { |
| "epoch": 783.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 2352 |
| }, |
| { |
| "epoch": 785.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2356 |
| }, |
| { |
| "epoch": 786.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2360 |
| }, |
| { |
| "epoch": 787.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2364 |
| }, |
| { |
| "epoch": 789.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2368 |
| }, |
| { |
| "epoch": 789.29, |
| "eval_exact_match": 0.5599613152804642, |
| "eval_exec": 0.5725338491295938, |
| "eval_loss": 0.47208261489868164, |
| "eval_runtime": 145.1522, |
| "eval_samples_per_second": 7.124, |
| "step": 2368 |
| }, |
| { |
| "epoch": 790.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2372 |
| }, |
| { |
| "epoch": 791.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2376 |
| }, |
| { |
| "epoch": 793.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2380 |
| }, |
| { |
| "epoch": 794.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2384 |
| }, |
| { |
| "epoch": 795.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2388 |
| }, |
| { |
| "epoch": 797.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2392 |
| }, |
| { |
| "epoch": 798.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2396 |
| }, |
| { |
| "epoch": 799.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2400 |
| }, |
| { |
| "epoch": 801.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2404 |
| }, |
| { |
| "epoch": 802.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2408 |
| }, |
| { |
| "epoch": 803.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2412 |
| }, |
| { |
| "epoch": 805.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2416 |
| }, |
| { |
| "epoch": 806.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2420 |
| }, |
| { |
| "epoch": 807.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2424 |
| }, |
| { |
| "epoch": 809.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2428 |
| }, |
| { |
| "epoch": 810.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2432 |
| }, |
| { |
| "epoch": 810.59, |
| "eval_exact_match": 0.5609284332688588, |
| "eval_exec": 0.5754352030947776, |
| "eval_loss": 0.4775034785270691, |
| "eval_runtime": 140.5187, |
| "eval_samples_per_second": 7.358, |
| "step": 2432 |
| }, |
| { |
| "epoch": 811.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2436 |
| }, |
| { |
| "epoch": 813.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2440 |
| }, |
| { |
| "epoch": 814.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2444 |
| }, |
| { |
| "epoch": 815.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2448 |
| }, |
| { |
| "epoch": 817.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0008, |
| "step": 2452 |
| }, |
| { |
| "epoch": 818.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0029, |
| "step": 2456 |
| }, |
| { |
| "epoch": 819.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0019, |
| "step": 2460 |
| }, |
| { |
| "epoch": 821.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2464 |
| }, |
| { |
| "epoch": 822.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2468 |
| }, |
| { |
| "epoch": 823.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2472 |
| }, |
| { |
| "epoch": 825.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2476 |
| }, |
| { |
| "epoch": 826.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2480 |
| }, |
| { |
| "epoch": 827.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2484 |
| }, |
| { |
| "epoch": 829.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2488 |
| }, |
| { |
| "epoch": 830.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2492 |
| }, |
| { |
| "epoch": 831.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2496 |
| }, |
| { |
| "epoch": 831.88, |
| "eval_exact_match": 0.5676982591876208, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.46634814143180847, |
| "eval_runtime": 147.2976, |
| "eval_samples_per_second": 7.02, |
| "step": 2496 |
| }, |
| { |
| "epoch": 833.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2500 |
| }, |
| { |
| "epoch": 834.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2504 |
| }, |
| { |
| "epoch": 835.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2508 |
| }, |
| { |
| "epoch": 837.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2512 |
| }, |
| { |
| "epoch": 838.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2516 |
| }, |
| { |
| "epoch": 839.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2520 |
| }, |
| { |
| "epoch": 841.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2524 |
| }, |
| { |
| "epoch": 842.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2528 |
| }, |
| { |
| "epoch": 843.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2532 |
| }, |
| { |
| "epoch": 845.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2536 |
| }, |
| { |
| "epoch": 846.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2540 |
| }, |
| { |
| "epoch": 847.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2544 |
| }, |
| { |
| "epoch": 849.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2548 |
| }, |
| { |
| "epoch": 850.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2552 |
| }, |
| { |
| "epoch": 851.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2556 |
| }, |
| { |
| "epoch": 853.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2560 |
| }, |
| { |
| "epoch": 853.29, |
| "eval_exact_match": 0.562862669245648, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.47793546319007874, |
| "eval_runtime": 144.726, |
| "eval_samples_per_second": 7.145, |
| "step": 2560 |
| }, |
| { |
| "epoch": 854.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2564 |
| }, |
| { |
| "epoch": 855.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2568 |
| }, |
| { |
| "epoch": 857.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2572 |
| }, |
| { |
| "epoch": 858.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2576 |
| }, |
| { |
| "epoch": 859.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2580 |
| }, |
| { |
| "epoch": 861.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2584 |
| }, |
| { |
| "epoch": 862.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2588 |
| }, |
| { |
| "epoch": 863.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2592 |
| }, |
| { |
| "epoch": 865.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2596 |
| }, |
| { |
| "epoch": 866.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2600 |
| }, |
| { |
| "epoch": 867.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2604 |
| }, |
| { |
| "epoch": 869.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2608 |
| }, |
| { |
| "epoch": 870.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2612 |
| }, |
| { |
| "epoch": 871.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 2616 |
| }, |
| { |
| "epoch": 873.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 2620 |
| }, |
| { |
| "epoch": 874.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2624 |
| }, |
| { |
| "epoch": 874.59, |
| "eval_exact_match": 0.5657640232108317, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.46636903285980225, |
| "eval_runtime": 145.0172, |
| "eval_samples_per_second": 7.13, |
| "step": 2624 |
| }, |
| { |
| "epoch": 875.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2628 |
| }, |
| { |
| "epoch": 877.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 2632 |
| }, |
| { |
| "epoch": 878.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2636 |
| }, |
| { |
| "epoch": 879.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2640 |
| }, |
| { |
| "epoch": 881.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2644 |
| }, |
| { |
| "epoch": 882.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2648 |
| }, |
| { |
| "epoch": 883.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2652 |
| }, |
| { |
| "epoch": 885.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2656 |
| }, |
| { |
| "epoch": 886.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2660 |
| }, |
| { |
| "epoch": 887.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2664 |
| }, |
| { |
| "epoch": 889.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2668 |
| }, |
| { |
| "epoch": 890.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2672 |
| }, |
| { |
| "epoch": 891.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2676 |
| }, |
| { |
| "epoch": 893.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2680 |
| }, |
| { |
| "epoch": 894.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2684 |
| }, |
| { |
| "epoch": 895.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2688 |
| }, |
| { |
| "epoch": 895.88, |
| "eval_exact_match": 0.5735009671179884, |
| "eval_exec": 0.5889748549323017, |
| "eval_loss": 0.48156842589378357, |
| "eval_runtime": 142.6086, |
| "eval_samples_per_second": 7.251, |
| "step": 2688 |
| }, |
| { |
| "epoch": 897.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2692 |
| }, |
| { |
| "epoch": 898.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2696 |
| }, |
| { |
| "epoch": 899.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2700 |
| }, |
| { |
| "epoch": 901.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2704 |
| }, |
| { |
| "epoch": 902.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2708 |
| }, |
| { |
| "epoch": 903.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2712 |
| }, |
| { |
| "epoch": 905.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2716 |
| }, |
| { |
| "epoch": 906.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2720 |
| }, |
| { |
| "epoch": 907.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2724 |
| }, |
| { |
| "epoch": 909.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2728 |
| }, |
| { |
| "epoch": 910.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2732 |
| }, |
| { |
| "epoch": 911.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2736 |
| }, |
| { |
| "epoch": 913.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2740 |
| }, |
| { |
| "epoch": 914.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 2744 |
| }, |
| { |
| "epoch": 915.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2748 |
| }, |
| { |
| "epoch": 917.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2752 |
| }, |
| { |
| "epoch": 917.29, |
| "eval_exact_match": 0.5686653771760155, |
| "eval_exec": 0.5802707930367504, |
| "eval_loss": 0.4817686080932617, |
| "eval_runtime": 145.4094, |
| "eval_samples_per_second": 7.111, |
| "step": 2752 |
| }, |
| { |
| "epoch": 918.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2756 |
| }, |
| { |
| "epoch": 919.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2760 |
| }, |
| { |
| "epoch": 921.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2764 |
| }, |
| { |
| "epoch": 922.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2768 |
| }, |
| { |
| "epoch": 923.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2772 |
| }, |
| { |
| "epoch": 925.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0012, |
| "step": 2776 |
| }, |
| { |
| "epoch": 926.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0083, |
| "step": 2780 |
| }, |
| { |
| "epoch": 927.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2784 |
| }, |
| { |
| "epoch": 929.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2788 |
| }, |
| { |
| "epoch": 930.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2792 |
| }, |
| { |
| "epoch": 931.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2796 |
| }, |
| { |
| "epoch": 933.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2800 |
| }, |
| { |
| "epoch": 934.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2804 |
| }, |
| { |
| "epoch": 935.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2808 |
| }, |
| { |
| "epoch": 937.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2812 |
| }, |
| { |
| "epoch": 938.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2816 |
| }, |
| { |
| "epoch": 938.59, |
| "eval_exact_match": 0.5686653771760155, |
| "eval_exec": 0.5909090909090909, |
| "eval_loss": 0.4858631193637848, |
| "eval_runtime": 145.1632, |
| "eval_samples_per_second": 7.123, |
| "step": 2816 |
| }, |
| { |
| "epoch": 939.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2820 |
| }, |
| { |
| "epoch": 941.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2824 |
| }, |
| { |
| "epoch": 942.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2828 |
| }, |
| { |
| "epoch": 943.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2832 |
| }, |
| { |
| "epoch": 945.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 2836 |
| }, |
| { |
| "epoch": 946.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2840 |
| }, |
| { |
| "epoch": 947.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2844 |
| }, |
| { |
| "epoch": 949.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2848 |
| }, |
| { |
| "epoch": 950.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2852 |
| }, |
| { |
| "epoch": 951.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 2856 |
| }, |
| { |
| "epoch": 953.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2860 |
| }, |
| { |
| "epoch": 954.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2864 |
| }, |
| { |
| "epoch": 955.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 2868 |
| }, |
| { |
| "epoch": 957.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2872 |
| }, |
| { |
| "epoch": 958.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2876 |
| }, |
| { |
| "epoch": 959.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2880 |
| }, |
| { |
| "epoch": 959.88, |
| "eval_exact_match": 0.5880077369439072, |
| "eval_exec": 0.6015473887814313, |
| "eval_loss": 0.4901029169559479, |
| "eval_runtime": 142.6992, |
| "eval_samples_per_second": 7.246, |
| "step": 2880 |
| }, |
| { |
| "epoch": 961.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2884 |
| }, |
| { |
| "epoch": 962.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2888 |
| }, |
| { |
| "epoch": 963.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 2892 |
| }, |
| { |
| "epoch": 965.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2896 |
| }, |
| { |
| "epoch": 966.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2900 |
| }, |
| { |
| "epoch": 967.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2904 |
| }, |
| { |
| "epoch": 969.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2908 |
| }, |
| { |
| "epoch": 970.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2912 |
| }, |
| { |
| "epoch": 971.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2916 |
| }, |
| { |
| "epoch": 973.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2920 |
| }, |
| { |
| "epoch": 974.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2924 |
| }, |
| { |
| "epoch": 975.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2928 |
| }, |
| { |
| "epoch": 977.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2932 |
| }, |
| { |
| "epoch": 978.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2936 |
| }, |
| { |
| "epoch": 979.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2940 |
| }, |
| { |
| "epoch": 981.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2944 |
| }, |
| { |
| "epoch": 981.29, |
| "eval_exact_match": 0.5705996131528046, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.48461684584617615, |
| "eval_runtime": 145.6186, |
| "eval_samples_per_second": 7.101, |
| "step": 2944 |
| }, |
| { |
| "epoch": 982.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2948 |
| }, |
| { |
| "epoch": 983.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2952 |
| }, |
| { |
| "epoch": 985.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2956 |
| }, |
| { |
| "epoch": 986.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2960 |
| }, |
| { |
| "epoch": 987.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2964 |
| }, |
| { |
| "epoch": 989.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2968 |
| }, |
| { |
| "epoch": 990.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2972 |
| }, |
| { |
| "epoch": 991.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2976 |
| }, |
| { |
| "epoch": 993.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2980 |
| }, |
| { |
| "epoch": 994.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2984 |
| }, |
| { |
| "epoch": 995.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2988 |
| }, |
| { |
| "epoch": 997.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 2992 |
| }, |
| { |
| "epoch": 998.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 2996 |
| }, |
| { |
| "epoch": 999.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1001.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3004 |
| }, |
| { |
| "epoch": 1002.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3008 |
| }, |
| { |
| "epoch": 1002.59, |
| "eval_exact_match": 0.5667311411992263, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.49284225702285767, |
| "eval_runtime": 144.5341, |
| "eval_samples_per_second": 7.154, |
| "step": 3008 |
| }, |
| { |
| "epoch": 1003.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3012 |
| }, |
| { |
| "epoch": 1005.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3016 |
| }, |
| { |
| "epoch": 1006.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3020 |
| }, |
| { |
| "epoch": 1007.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3024 |
| }, |
| { |
| "epoch": 1009.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3028 |
| }, |
| { |
| "epoch": 1010.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3032 |
| }, |
| { |
| "epoch": 1011.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3036 |
| }, |
| { |
| "epoch": 1013.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3040 |
| }, |
| { |
| "epoch": 1014.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3044 |
| }, |
| { |
| "epoch": 1015.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3048 |
| }, |
| { |
| "epoch": 1017.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3052 |
| }, |
| { |
| "epoch": 1018.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3056 |
| }, |
| { |
| "epoch": 1019.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3060 |
| }, |
| { |
| "epoch": 1021.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3064 |
| }, |
| { |
| "epoch": 1022.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3068 |
| }, |
| { |
| "epoch": 1023.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3072 |
| }, |
| { |
| "epoch": 1023.88, |
| "eval_exact_match": 0.5764023210831721, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.49118873476982117, |
| "eval_runtime": 154.6804, |
| "eval_samples_per_second": 6.685, |
| "step": 3072 |
| }, |
| { |
| "epoch": 1025.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3076 |
| }, |
| { |
| "epoch": 1026.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3080 |
| }, |
| { |
| "epoch": 1027.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3084 |
| }, |
| { |
| "epoch": 1029.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3088 |
| }, |
| { |
| "epoch": 1030.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3092 |
| }, |
| { |
| "epoch": 1031.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3096 |
| }, |
| { |
| "epoch": 1033.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3100 |
| }, |
| { |
| "epoch": 1034.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3104 |
| }, |
| { |
| "epoch": 1035.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3108 |
| }, |
| { |
| "epoch": 1037.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3112 |
| }, |
| { |
| "epoch": 1038.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3116 |
| }, |
| { |
| "epoch": 1039.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3120 |
| }, |
| { |
| "epoch": 1041.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3124 |
| }, |
| { |
| "epoch": 1042.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3128 |
| }, |
| { |
| "epoch": 1043.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3132 |
| }, |
| { |
| "epoch": 1045.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3136 |
| }, |
| { |
| "epoch": 1045.29, |
| "eval_exact_match": 0.5754352030947776, |
| "eval_exec": 0.5851063829787234, |
| "eval_loss": 0.49213075637817383, |
| "eval_runtime": 146.5831, |
| "eval_samples_per_second": 7.054, |
| "step": 3136 |
| }, |
| { |
| "epoch": 1046.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3140 |
| }, |
| { |
| "epoch": 1047.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3144 |
| }, |
| { |
| "epoch": 1049.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3148 |
| }, |
| { |
| "epoch": 1050.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3152 |
| }, |
| { |
| "epoch": 1051.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3156 |
| }, |
| { |
| "epoch": 1053.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3160 |
| }, |
| { |
| "epoch": 1054.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3164 |
| }, |
| { |
| "epoch": 1055.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3168 |
| }, |
| { |
| "epoch": 1057.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3172 |
| }, |
| { |
| "epoch": 1058.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3176 |
| }, |
| { |
| "epoch": 1059.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3180 |
| }, |
| { |
| "epoch": 1061.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3184 |
| }, |
| { |
| "epoch": 1062.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3188 |
| }, |
| { |
| "epoch": 1063.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3192 |
| }, |
| { |
| "epoch": 1065.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3196 |
| }, |
| { |
| "epoch": 1066.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1066.59, |
| "eval_exact_match": 0.5793036750483559, |
| "eval_exec": 0.5880077369439072, |
| "eval_loss": 0.505233645439148, |
| "eval_runtime": 144.9781, |
| "eval_samples_per_second": 7.132, |
| "step": 3200 |
| }, |
| { |
| "epoch": 1067.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3204 |
| }, |
| { |
| "epoch": 1069.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3208 |
| }, |
| { |
| "epoch": 1070.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3212 |
| }, |
| { |
| "epoch": 1071.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 3216 |
| }, |
| { |
| "epoch": 1073.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3220 |
| }, |
| { |
| "epoch": 1074.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3224 |
| }, |
| { |
| "epoch": 1075.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3228 |
| }, |
| { |
| "epoch": 1077.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3232 |
| }, |
| { |
| "epoch": 1078.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3236 |
| }, |
| { |
| "epoch": 1079.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3240 |
| }, |
| { |
| "epoch": 1081.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3244 |
| }, |
| { |
| "epoch": 1082.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3248 |
| }, |
| { |
| "epoch": 1083.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3252 |
| }, |
| { |
| "epoch": 1085.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3256 |
| }, |
| { |
| "epoch": 1086.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3260 |
| }, |
| { |
| "epoch": 1087.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3264 |
| }, |
| { |
| "epoch": 1087.88, |
| "eval_exact_match": 0.5812379110251451, |
| "eval_exec": 0.5947775628626693, |
| "eval_loss": 0.5012010931968689, |
| "eval_runtime": 154.9379, |
| "eval_samples_per_second": 6.674, |
| "step": 3264 |
| }, |
| { |
| "epoch": 1089.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3268 |
| }, |
| { |
| "epoch": 1090.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3272 |
| }, |
| { |
| "epoch": 1091.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3276 |
| }, |
| { |
| "epoch": 1093.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3280 |
| }, |
| { |
| "epoch": 1094.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3284 |
| }, |
| { |
| "epoch": 1095.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3288 |
| }, |
| { |
| "epoch": 1097.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3292 |
| }, |
| { |
| "epoch": 1098.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3296 |
| }, |
| { |
| "epoch": 1099.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3300 |
| }, |
| { |
| "epoch": 1101.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3304 |
| }, |
| { |
| "epoch": 1102.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3308 |
| }, |
| { |
| "epoch": 1103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3312 |
| }, |
| { |
| "epoch": 1105.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3316 |
| }, |
| { |
| "epoch": 1106.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3320 |
| }, |
| { |
| "epoch": 1107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3324 |
| }, |
| { |
| "epoch": 1109.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3328 |
| }, |
| { |
| "epoch": 1109.29, |
| "eval_exact_match": 0.5696324951644101, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.4975697994232178, |
| "eval_runtime": 147.4955, |
| "eval_samples_per_second": 7.01, |
| "step": 3328 |
| }, |
| { |
| "epoch": 1110.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3332 |
| }, |
| { |
| "epoch": 1111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3336 |
| }, |
| { |
| "epoch": 1113.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3340 |
| }, |
| { |
| "epoch": 1114.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3344 |
| }, |
| { |
| "epoch": 1115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3348 |
| }, |
| { |
| "epoch": 1117.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3352 |
| }, |
| { |
| "epoch": 1118.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3356 |
| }, |
| { |
| "epoch": 1119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3360 |
| }, |
| { |
| "epoch": 1121.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3364 |
| }, |
| { |
| "epoch": 1122.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3368 |
| }, |
| { |
| "epoch": 1123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3372 |
| }, |
| { |
| "epoch": 1125.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3376 |
| }, |
| { |
| "epoch": 1126.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3380 |
| }, |
| { |
| "epoch": 1127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3384 |
| }, |
| { |
| "epoch": 1129.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3388 |
| }, |
| { |
| "epoch": 1130.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3392 |
| }, |
| { |
| "epoch": 1130.59, |
| "eval_exact_match": 0.5909090909090909, |
| "eval_exec": 0.6015473887814313, |
| "eval_loss": 0.49673762917518616, |
| "eval_runtime": 148.9555, |
| "eval_samples_per_second": 6.942, |
| "step": 3392 |
| }, |
| { |
| "epoch": 1131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3396 |
| }, |
| { |
| "epoch": 1133.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3400 |
| }, |
| { |
| "epoch": 1134.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3404 |
| }, |
| { |
| "epoch": 1135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3408 |
| }, |
| { |
| "epoch": 1137.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3412 |
| }, |
| { |
| "epoch": 1138.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3416 |
| }, |
| { |
| "epoch": 1139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3420 |
| }, |
| { |
| "epoch": 1141.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3424 |
| }, |
| { |
| "epoch": 1142.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3428 |
| }, |
| { |
| "epoch": 1143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3432 |
| }, |
| { |
| "epoch": 1145.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3436 |
| }, |
| { |
| "epoch": 1146.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3440 |
| }, |
| { |
| "epoch": 1147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3444 |
| }, |
| { |
| "epoch": 1149.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3448 |
| }, |
| { |
| "epoch": 1150.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3452 |
| }, |
| { |
| "epoch": 1151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3456 |
| }, |
| { |
| "epoch": 1151.88, |
| "eval_exact_match": 0.5773694390715667, |
| "eval_exec": 0.5909090909090909, |
| "eval_loss": 0.500177800655365, |
| "eval_runtime": 144.7205, |
| "eval_samples_per_second": 7.145, |
| "step": 3456 |
| }, |
| { |
| "epoch": 1153.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3460 |
| }, |
| { |
| "epoch": 1154.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3464 |
| }, |
| { |
| "epoch": 1155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3468 |
| }, |
| { |
| "epoch": 1157.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3472 |
| }, |
| { |
| "epoch": 1158.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3476 |
| }, |
| { |
| "epoch": 1159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 3480 |
| }, |
| { |
| "epoch": 1161.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3484 |
| }, |
| { |
| "epoch": 1162.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3488 |
| }, |
| { |
| "epoch": 1163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3492 |
| }, |
| { |
| "epoch": 1165.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3496 |
| }, |
| { |
| "epoch": 1166.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3504 |
| }, |
| { |
| "epoch": 1169.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3508 |
| }, |
| { |
| "epoch": 1170.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3512 |
| }, |
| { |
| "epoch": 1171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3516 |
| }, |
| { |
| "epoch": 1173.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3520 |
| }, |
| { |
| "epoch": 1173.29, |
| "eval_exact_match": 0.5696324951644101, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.5047540068626404, |
| "eval_runtime": 146.9779, |
| "eval_samples_per_second": 7.035, |
| "step": 3520 |
| }, |
| { |
| "epoch": 1174.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3524 |
| }, |
| { |
| "epoch": 1175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3528 |
| }, |
| { |
| "epoch": 1177.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3532 |
| }, |
| { |
| "epoch": 1178.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3536 |
| }, |
| { |
| "epoch": 1179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3540 |
| }, |
| { |
| "epoch": 1181.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3544 |
| }, |
| { |
| "epoch": 1182.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3548 |
| }, |
| { |
| "epoch": 1183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3552 |
| }, |
| { |
| "epoch": 1185.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3556 |
| }, |
| { |
| "epoch": 1186.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3560 |
| }, |
| { |
| "epoch": 1187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3564 |
| }, |
| { |
| "epoch": 1189.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3568 |
| }, |
| { |
| "epoch": 1190.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3572 |
| }, |
| { |
| "epoch": 1191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3576 |
| }, |
| { |
| "epoch": 1193.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3580 |
| }, |
| { |
| "epoch": 1194.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3584 |
| }, |
| { |
| "epoch": 1194.59, |
| "eval_exact_match": 0.5570599613152805, |
| "eval_exec": 0.5764023210831721, |
| "eval_loss": 0.48414555191993713, |
| "eval_runtime": 150.4526, |
| "eval_samples_per_second": 6.873, |
| "step": 3584 |
| }, |
| { |
| "epoch": 1195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3588 |
| }, |
| { |
| "epoch": 1197.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3592 |
| }, |
| { |
| "epoch": 1198.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3596 |
| }, |
| { |
| "epoch": 1199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3600 |
| }, |
| { |
| "epoch": 1201.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3604 |
| }, |
| { |
| "epoch": 1202.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3608 |
| }, |
| { |
| "epoch": 1203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3612 |
| }, |
| { |
| "epoch": 1205.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3616 |
| }, |
| { |
| "epoch": 1206.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3620 |
| }, |
| { |
| "epoch": 1207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3624 |
| }, |
| { |
| "epoch": 1209.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3628 |
| }, |
| { |
| "epoch": 1210.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3632 |
| }, |
| { |
| "epoch": 1211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3636 |
| }, |
| { |
| "epoch": 1213.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3640 |
| }, |
| { |
| "epoch": 1214.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3644 |
| }, |
| { |
| "epoch": 1215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3648 |
| }, |
| { |
| "epoch": 1215.88, |
| "eval_exact_match": 0.5657640232108317, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.5146532654762268, |
| "eval_runtime": 150.3937, |
| "eval_samples_per_second": 6.875, |
| "step": 3648 |
| }, |
| { |
| "epoch": 1217.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3652 |
| }, |
| { |
| "epoch": 1218.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 3656 |
| }, |
| { |
| "epoch": 1219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3660 |
| }, |
| { |
| "epoch": 1221.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3664 |
| }, |
| { |
| "epoch": 1222.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3668 |
| }, |
| { |
| "epoch": 1223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3672 |
| }, |
| { |
| "epoch": 1225.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3676 |
| }, |
| { |
| "epoch": 1226.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3680 |
| }, |
| { |
| "epoch": 1227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3684 |
| }, |
| { |
| "epoch": 1229.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3688 |
| }, |
| { |
| "epoch": 1230.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3692 |
| }, |
| { |
| "epoch": 1231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3696 |
| }, |
| { |
| "epoch": 1233.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3700 |
| }, |
| { |
| "epoch": 1234.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3704 |
| }, |
| { |
| "epoch": 1235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3708 |
| }, |
| { |
| "epoch": 1237.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3712 |
| }, |
| { |
| "epoch": 1237.29, |
| "eval_exact_match": 0.5783365570599613, |
| "eval_exec": 0.586073500967118, |
| "eval_loss": 0.5017855763435364, |
| "eval_runtime": 143.4607, |
| "eval_samples_per_second": 7.208, |
| "step": 3712 |
| }, |
| { |
| "epoch": 1238.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3716 |
| }, |
| { |
| "epoch": 1239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3720 |
| }, |
| { |
| "epoch": 1241.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3724 |
| }, |
| { |
| "epoch": 1242.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3728 |
| }, |
| { |
| "epoch": 1243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3732 |
| }, |
| { |
| "epoch": 1245.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3736 |
| }, |
| { |
| "epoch": 1246.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3740 |
| }, |
| { |
| "epoch": 1247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3744 |
| }, |
| { |
| "epoch": 1249.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3748 |
| }, |
| { |
| "epoch": 1250.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3752 |
| }, |
| { |
| "epoch": 1251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3756 |
| }, |
| { |
| "epoch": 1253.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3760 |
| }, |
| { |
| "epoch": 1254.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3764 |
| }, |
| { |
| "epoch": 1255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3768 |
| }, |
| { |
| "epoch": 1257.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3772 |
| }, |
| { |
| "epoch": 1258.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3776 |
| }, |
| { |
| "epoch": 1258.59, |
| "eval_exact_match": 0.5725338491295938, |
| "eval_exec": 0.5812379110251451, |
| "eval_loss": 0.5150123834609985, |
| "eval_runtime": 145.4002, |
| "eval_samples_per_second": 7.111, |
| "step": 3776 |
| }, |
| { |
| "epoch": 1259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3780 |
| }, |
| { |
| "epoch": 1261.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3784 |
| }, |
| { |
| "epoch": 1262.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3788 |
| }, |
| { |
| "epoch": 1263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3792 |
| }, |
| { |
| "epoch": 1265.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3796 |
| }, |
| { |
| "epoch": 1266.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 3800 |
| }, |
| { |
| "epoch": 1267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3804 |
| }, |
| { |
| "epoch": 1269.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3808 |
| }, |
| { |
| "epoch": 1270.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3812 |
| }, |
| { |
| "epoch": 1271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3816 |
| }, |
| { |
| "epoch": 1273.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3820 |
| }, |
| { |
| "epoch": 1274.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3824 |
| }, |
| { |
| "epoch": 1275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3828 |
| }, |
| { |
| "epoch": 1277.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3832 |
| }, |
| { |
| "epoch": 1278.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3836 |
| }, |
| { |
| "epoch": 1279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3840 |
| }, |
| { |
| "epoch": 1279.88, |
| "eval_exact_match": 0.562862669245648, |
| "eval_exec": 0.574468085106383, |
| "eval_loss": 0.5197062492370605, |
| "eval_runtime": 154.0184, |
| "eval_samples_per_second": 6.713, |
| "step": 3840 |
| }, |
| { |
| "epoch": 1281.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3844 |
| }, |
| { |
| "epoch": 1282.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3848 |
| }, |
| { |
| "epoch": 1283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3852 |
| }, |
| { |
| "epoch": 1285.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3856 |
| }, |
| { |
| "epoch": 1286.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3860 |
| }, |
| { |
| "epoch": 1287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3864 |
| }, |
| { |
| "epoch": 1289.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3868 |
| }, |
| { |
| "epoch": 1290.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3872 |
| }, |
| { |
| "epoch": 1291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3876 |
| }, |
| { |
| "epoch": 1293.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 3880 |
| }, |
| { |
| "epoch": 1294.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3884 |
| }, |
| { |
| "epoch": 1295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3888 |
| }, |
| { |
| "epoch": 1297.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3892 |
| }, |
| { |
| "epoch": 1298.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3896 |
| }, |
| { |
| "epoch": 1299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3900 |
| }, |
| { |
| "epoch": 1301.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3904 |
| }, |
| { |
| "epoch": 1301.29, |
| "eval_exact_match": 0.5705996131528046, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5007818341255188, |
| "eval_runtime": 156.3152, |
| "eval_samples_per_second": 6.615, |
| "step": 3904 |
| }, |
| { |
| "epoch": 1302.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3908 |
| }, |
| { |
| "epoch": 1303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3912 |
| }, |
| { |
| "epoch": 1305.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 3916 |
| }, |
| { |
| "epoch": 1306.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3920 |
| }, |
| { |
| "epoch": 1307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3924 |
| }, |
| { |
| "epoch": 1309.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3928 |
| }, |
| { |
| "epoch": 1310.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3932 |
| }, |
| { |
| "epoch": 1311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3936 |
| }, |
| { |
| "epoch": 1313.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3940 |
| }, |
| { |
| "epoch": 1314.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3944 |
| }, |
| { |
| "epoch": 1315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3948 |
| }, |
| { |
| "epoch": 1317.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3952 |
| }, |
| { |
| "epoch": 1318.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3956 |
| }, |
| { |
| "epoch": 1319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3960 |
| }, |
| { |
| "epoch": 1321.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 3964 |
| }, |
| { |
| "epoch": 1322.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3968 |
| }, |
| { |
| "epoch": 1322.59, |
| "eval_exact_match": 0.5725338491295938, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.505159318447113, |
| "eval_runtime": 151.0798, |
| "eval_samples_per_second": 6.844, |
| "step": 3968 |
| }, |
| { |
| "epoch": 1323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3972 |
| }, |
| { |
| "epoch": 1325.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3976 |
| }, |
| { |
| "epoch": 1326.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 3980 |
| }, |
| { |
| "epoch": 1327.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3984 |
| }, |
| { |
| "epoch": 1329.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 3988 |
| }, |
| { |
| "epoch": 1330.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3992 |
| }, |
| { |
| "epoch": 1331.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 3996 |
| }, |
| { |
| "epoch": 1333.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1334.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 4004 |
| }, |
| { |
| "epoch": 1335.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4008 |
| }, |
| { |
| "epoch": 1337.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4012 |
| }, |
| { |
| "epoch": 1338.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4016 |
| }, |
| { |
| "epoch": 1339.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4020 |
| }, |
| { |
| "epoch": 1341.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4024 |
| }, |
| { |
| "epoch": 1342.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4028 |
| }, |
| { |
| "epoch": 1343.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4032 |
| }, |
| { |
| "epoch": 1343.88, |
| "eval_exact_match": 0.5812379110251451, |
| "eval_exec": 0.5986460348162476, |
| "eval_loss": 0.49928486347198486, |
| "eval_runtime": 145.8647, |
| "eval_samples_per_second": 7.089, |
| "step": 4032 |
| }, |
| { |
| "epoch": 1345.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4036 |
| }, |
| { |
| "epoch": 1346.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4040 |
| }, |
| { |
| "epoch": 1347.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4044 |
| }, |
| { |
| "epoch": 1349.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4048 |
| }, |
| { |
| "epoch": 1350.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4052 |
| }, |
| { |
| "epoch": 1351.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4056 |
| }, |
| { |
| "epoch": 1353.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4060 |
| }, |
| { |
| "epoch": 1354.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4064 |
| }, |
| { |
| "epoch": 1355.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4068 |
| }, |
| { |
| "epoch": 1357.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4072 |
| }, |
| { |
| "epoch": 1358.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4076 |
| }, |
| { |
| "epoch": 1359.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4080 |
| }, |
| { |
| "epoch": 1361.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4084 |
| }, |
| { |
| "epoch": 1362.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4088 |
| }, |
| { |
| "epoch": 1363.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4092 |
| }, |
| { |
| "epoch": 1365.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4096 |
| }, |
| { |
| "epoch": 1365.29, |
| "eval_exact_match": 0.5754352030947776, |
| "eval_exec": 0.586073500967118, |
| "eval_loss": 0.5093615055084229, |
| "eval_runtime": 148.9495, |
| "eval_samples_per_second": 6.942, |
| "step": 4096 |
| }, |
| { |
| "epoch": 1366.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4100 |
| }, |
| { |
| "epoch": 1367.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4104 |
| }, |
| { |
| "epoch": 1369.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4108 |
| }, |
| { |
| "epoch": 1370.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4112 |
| }, |
| { |
| "epoch": 1371.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4116 |
| }, |
| { |
| "epoch": 1373.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4120 |
| }, |
| { |
| "epoch": 1374.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4124 |
| }, |
| { |
| "epoch": 1375.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4128 |
| }, |
| { |
| "epoch": 1377.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4132 |
| }, |
| { |
| "epoch": 1378.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4136 |
| }, |
| { |
| "epoch": 1379.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4140 |
| }, |
| { |
| "epoch": 1381.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4144 |
| }, |
| { |
| "epoch": 1382.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4148 |
| }, |
| { |
| "epoch": 1383.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4152 |
| }, |
| { |
| "epoch": 1385.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4156 |
| }, |
| { |
| "epoch": 1386.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4160 |
| }, |
| { |
| "epoch": 1386.59, |
| "eval_exact_match": 0.5831721470019342, |
| "eval_exec": 0.5957446808510638, |
| "eval_loss": 0.5088846683502197, |
| "eval_runtime": 151.6057, |
| "eval_samples_per_second": 6.82, |
| "step": 4160 |
| }, |
| { |
| "epoch": 1387.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4164 |
| }, |
| { |
| "epoch": 1389.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4168 |
| }, |
| { |
| "epoch": 1390.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4172 |
| }, |
| { |
| "epoch": 1391.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4176 |
| }, |
| { |
| "epoch": 1393.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4180 |
| }, |
| { |
| "epoch": 1394.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4184 |
| }, |
| { |
| "epoch": 1395.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 4188 |
| }, |
| { |
| "epoch": 1397.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4192 |
| }, |
| { |
| "epoch": 1398.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4196 |
| }, |
| { |
| "epoch": 1399.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4200 |
| }, |
| { |
| "epoch": 1401.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4204 |
| }, |
| { |
| "epoch": 1402.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4208 |
| }, |
| { |
| "epoch": 1403.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4212 |
| }, |
| { |
| "epoch": 1405.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4216 |
| }, |
| { |
| "epoch": 1406.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4220 |
| }, |
| { |
| "epoch": 1407.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4224 |
| }, |
| { |
| "epoch": 1407.88, |
| "eval_exact_match": 0.574468085106383, |
| "eval_exec": 0.5851063829787234, |
| "eval_loss": 0.5246360301971436, |
| "eval_runtime": 155.5696, |
| "eval_samples_per_second": 6.647, |
| "step": 4224 |
| }, |
| { |
| "epoch": 1409.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4228 |
| }, |
| { |
| "epoch": 1410.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4232 |
| }, |
| { |
| "epoch": 1411.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4236 |
| }, |
| { |
| "epoch": 1413.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4240 |
| }, |
| { |
| "epoch": 1414.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4244 |
| }, |
| { |
| "epoch": 1415.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4248 |
| }, |
| { |
| "epoch": 1417.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4252 |
| }, |
| { |
| "epoch": 1418.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4256 |
| }, |
| { |
| "epoch": 1419.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4260 |
| }, |
| { |
| "epoch": 1421.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 4264 |
| }, |
| { |
| "epoch": 1422.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 4268 |
| }, |
| { |
| "epoch": 1423.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4272 |
| }, |
| { |
| "epoch": 1425.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4276 |
| }, |
| { |
| "epoch": 1426.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4280 |
| }, |
| { |
| "epoch": 1427.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4284 |
| }, |
| { |
| "epoch": 1429.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4288 |
| }, |
| { |
| "epoch": 1429.29, |
| "eval_exact_match": 0.5715667311411993, |
| "eval_exec": 0.5831721470019342, |
| "eval_loss": 0.5079379081726074, |
| "eval_runtime": 150.2141, |
| "eval_samples_per_second": 6.884, |
| "step": 4288 |
| }, |
| { |
| "epoch": 1430.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4292 |
| }, |
| { |
| "epoch": 1431.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4296 |
| }, |
| { |
| "epoch": 1433.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 4300 |
| }, |
| { |
| "epoch": 1434.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0011, |
| "step": 4304 |
| }, |
| { |
| "epoch": 1435.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4308 |
| }, |
| { |
| "epoch": 1437.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4312 |
| }, |
| { |
| "epoch": 1438.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4316 |
| }, |
| { |
| "epoch": 1439.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4320 |
| }, |
| { |
| "epoch": 1441.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0031, |
| "step": 4324 |
| }, |
| { |
| "epoch": 1442.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 4328 |
| }, |
| { |
| "epoch": 1443.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4332 |
| }, |
| { |
| "epoch": 1445.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4336 |
| }, |
| { |
| "epoch": 1446.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4340 |
| }, |
| { |
| "epoch": 1447.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4344 |
| }, |
| { |
| "epoch": 1449.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4348 |
| }, |
| { |
| "epoch": 1450.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4352 |
| }, |
| { |
| "epoch": 1450.59, |
| "eval_exact_match": 0.5783365570599613, |
| "eval_exec": 0.5957446808510638, |
| "eval_loss": 0.500064492225647, |
| "eval_runtime": 143.1681, |
| "eval_samples_per_second": 7.222, |
| "step": 4352 |
| }, |
| { |
| "epoch": 1451.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4356 |
| }, |
| { |
| "epoch": 1453.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4360 |
| }, |
| { |
| "epoch": 1454.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4364 |
| }, |
| { |
| "epoch": 1455.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4368 |
| }, |
| { |
| "epoch": 1457.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4372 |
| }, |
| { |
| "epoch": 1458.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4376 |
| }, |
| { |
| "epoch": 1459.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4380 |
| }, |
| { |
| "epoch": 1461.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4384 |
| }, |
| { |
| "epoch": 1462.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4388 |
| }, |
| { |
| "epoch": 1463.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4392 |
| }, |
| { |
| "epoch": 1465.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4396 |
| }, |
| { |
| "epoch": 1466.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4400 |
| }, |
| { |
| "epoch": 1467.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4404 |
| }, |
| { |
| "epoch": 1469.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4408 |
| }, |
| { |
| "epoch": 1470.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4412 |
| }, |
| { |
| "epoch": 1471.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4416 |
| }, |
| { |
| "epoch": 1471.88, |
| "eval_exact_match": 0.5705996131528046, |
| "eval_exec": 0.5889748549323017, |
| "eval_loss": 0.509338915348053, |
| "eval_runtime": 146.7739, |
| "eval_samples_per_second": 7.045, |
| "step": 4416 |
| }, |
| { |
| "epoch": 1473.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4420 |
| }, |
| { |
| "epoch": 1474.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4424 |
| }, |
| { |
| "epoch": 1475.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4428 |
| }, |
| { |
| "epoch": 1477.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4432 |
| }, |
| { |
| "epoch": 1478.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4436 |
| }, |
| { |
| "epoch": 1479.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4440 |
| }, |
| { |
| "epoch": 1481.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4444 |
| }, |
| { |
| "epoch": 1482.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4448 |
| }, |
| { |
| "epoch": 1483.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4452 |
| }, |
| { |
| "epoch": 1485.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4456 |
| }, |
| { |
| "epoch": 1486.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 4460 |
| }, |
| { |
| "epoch": 1487.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4464 |
| }, |
| { |
| "epoch": 1489.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4468 |
| }, |
| { |
| "epoch": 1490.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4472 |
| }, |
| { |
| "epoch": 1491.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4476 |
| }, |
| { |
| "epoch": 1493.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4480 |
| }, |
| { |
| "epoch": 1493.29, |
| "eval_exact_match": 0.5783365570599613, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.5201787352561951, |
| "eval_runtime": 153.0254, |
| "eval_samples_per_second": 6.757, |
| "step": 4480 |
| }, |
| { |
| "epoch": 1494.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4484 |
| }, |
| { |
| "epoch": 1495.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4488 |
| }, |
| { |
| "epoch": 1497.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4492 |
| }, |
| { |
| "epoch": 1498.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4496 |
| }, |
| { |
| "epoch": 1499.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1501.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4504 |
| }, |
| { |
| "epoch": 1502.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4508 |
| }, |
| { |
| "epoch": 1503.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0006, |
| "step": 4512 |
| }, |
| { |
| "epoch": 1505.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0013, |
| "step": 4516 |
| }, |
| { |
| "epoch": 1506.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4520 |
| }, |
| { |
| "epoch": 1507.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4524 |
| }, |
| { |
| "epoch": 1509.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4528 |
| }, |
| { |
| "epoch": 1510.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4532 |
| }, |
| { |
| "epoch": 1511.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4536 |
| }, |
| { |
| "epoch": 1513.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4540 |
| }, |
| { |
| "epoch": 1514.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4544 |
| }, |
| { |
| "epoch": 1514.59, |
| "eval_exact_match": 0.5793036750483559, |
| "eval_exec": 0.5899419729206963, |
| "eval_loss": 0.514903724193573, |
| "eval_runtime": 150.0947, |
| "eval_samples_per_second": 6.889, |
| "step": 4544 |
| }, |
| { |
| "epoch": 1515.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4548 |
| }, |
| { |
| "epoch": 1517.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4552 |
| }, |
| { |
| "epoch": 1518.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4556 |
| }, |
| { |
| "epoch": 1519.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4560 |
| }, |
| { |
| "epoch": 1521.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4564 |
| }, |
| { |
| "epoch": 1522.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4568 |
| }, |
| { |
| "epoch": 1523.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4572 |
| }, |
| { |
| "epoch": 1525.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4576 |
| }, |
| { |
| "epoch": 1526.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4580 |
| }, |
| { |
| "epoch": 1527.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4584 |
| }, |
| { |
| "epoch": 1529.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4588 |
| }, |
| { |
| "epoch": 1530.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4592 |
| }, |
| { |
| "epoch": 1531.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4596 |
| }, |
| { |
| "epoch": 1533.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4600 |
| }, |
| { |
| "epoch": 1534.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4604 |
| }, |
| { |
| "epoch": 1535.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4608 |
| }, |
| { |
| "epoch": 1535.88, |
| "eval_exact_match": 0.5802707930367504, |
| "eval_exec": 0.59284332688588, |
| "eval_loss": 0.5181902647018433, |
| "eval_runtime": 156.4624, |
| "eval_samples_per_second": 6.609, |
| "step": 4608 |
| }, |
| { |
| "epoch": 1537.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4612 |
| }, |
| { |
| "epoch": 1538.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4616 |
| }, |
| { |
| "epoch": 1539.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4620 |
| }, |
| { |
| "epoch": 1541.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4624 |
| }, |
| { |
| "epoch": 1542.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4628 |
| }, |
| { |
| "epoch": 1543.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4632 |
| }, |
| { |
| "epoch": 1545.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4636 |
| }, |
| { |
| "epoch": 1546.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4640 |
| }, |
| { |
| "epoch": 1547.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4644 |
| }, |
| { |
| "epoch": 1549.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4648 |
| }, |
| { |
| "epoch": 1550.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4652 |
| }, |
| { |
| "epoch": 1551.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4656 |
| }, |
| { |
| "epoch": 1553.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4660 |
| }, |
| { |
| "epoch": 1554.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4664 |
| }, |
| { |
| "epoch": 1555.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4668 |
| }, |
| { |
| "epoch": 1557.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4672 |
| }, |
| { |
| "epoch": 1557.29, |
| "eval_exact_match": 0.5715667311411993, |
| "eval_exec": 0.5802707930367504, |
| "eval_loss": 0.5268357396125793, |
| "eval_runtime": 159.3416, |
| "eval_samples_per_second": 6.489, |
| "step": 4672 |
| }, |
| { |
| "epoch": 1558.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4676 |
| }, |
| { |
| "epoch": 1559.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4680 |
| }, |
| { |
| "epoch": 1561.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4684 |
| }, |
| { |
| "epoch": 1562.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4688 |
| }, |
| { |
| "epoch": 1563.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4692 |
| }, |
| { |
| "epoch": 1565.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4696 |
| }, |
| { |
| "epoch": 1566.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4700 |
| }, |
| { |
| "epoch": 1567.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4704 |
| }, |
| { |
| "epoch": 1569.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4708 |
| }, |
| { |
| "epoch": 1570.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4712 |
| }, |
| { |
| "epoch": 1571.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4716 |
| }, |
| { |
| "epoch": 1573.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4720 |
| }, |
| { |
| "epoch": 1574.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4724 |
| }, |
| { |
| "epoch": 1575.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4728 |
| }, |
| { |
| "epoch": 1577.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4732 |
| }, |
| { |
| "epoch": 1578.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4736 |
| }, |
| { |
| "epoch": 1578.59, |
| "eval_exact_match": 0.5735009671179884, |
| "eval_exec": 0.5812379110251451, |
| "eval_loss": 0.5166209936141968, |
| "eval_runtime": 155.0334, |
| "eval_samples_per_second": 6.67, |
| "step": 4736 |
| }, |
| { |
| "epoch": 1579.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4740 |
| }, |
| { |
| "epoch": 1581.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4744 |
| }, |
| { |
| "epoch": 1582.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4748 |
| }, |
| { |
| "epoch": 1583.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4752 |
| }, |
| { |
| "epoch": 1585.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4756 |
| }, |
| { |
| "epoch": 1586.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4760 |
| }, |
| { |
| "epoch": 1587.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4764 |
| }, |
| { |
| "epoch": 1589.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4768 |
| }, |
| { |
| "epoch": 1590.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4772 |
| }, |
| { |
| "epoch": 1591.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4776 |
| }, |
| { |
| "epoch": 1593.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4780 |
| }, |
| { |
| "epoch": 1594.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4784 |
| }, |
| { |
| "epoch": 1595.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4788 |
| }, |
| { |
| "epoch": 1597.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4792 |
| }, |
| { |
| "epoch": 1598.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4796 |
| }, |
| { |
| "epoch": 1599.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1599.88, |
| "eval_exact_match": 0.562862669245648, |
| "eval_exec": 0.5705996131528046, |
| "eval_loss": 0.521124541759491, |
| "eval_runtime": 157.5476, |
| "eval_samples_per_second": 6.563, |
| "step": 4800 |
| }, |
| { |
| "epoch": 1601.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4804 |
| }, |
| { |
| "epoch": 1602.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4808 |
| }, |
| { |
| "epoch": 1603.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4812 |
| }, |
| { |
| "epoch": 1605.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4816 |
| }, |
| { |
| "epoch": 1606.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4820 |
| }, |
| { |
| "epoch": 1607.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4824 |
| }, |
| { |
| "epoch": 1609.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4828 |
| }, |
| { |
| "epoch": 1610.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4832 |
| }, |
| { |
| "epoch": 1611.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4836 |
| }, |
| { |
| "epoch": 1613.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4840 |
| }, |
| { |
| "epoch": 1614.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4844 |
| }, |
| { |
| "epoch": 1615.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4848 |
| }, |
| { |
| "epoch": 1617.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4852 |
| }, |
| { |
| "epoch": 1618.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4856 |
| }, |
| { |
| "epoch": 1619.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4860 |
| }, |
| { |
| "epoch": 1621.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4864 |
| }, |
| { |
| "epoch": 1621.29, |
| "eval_exact_match": 0.5638297872340425, |
| "eval_exec": 0.5754352030947776, |
| "eval_loss": 0.5117584466934204, |
| "eval_runtime": 158.6575, |
| "eval_samples_per_second": 6.517, |
| "step": 4864 |
| }, |
| { |
| "epoch": 1622.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4868 |
| }, |
| { |
| "epoch": 1623.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4872 |
| }, |
| { |
| "epoch": 1625.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4876 |
| }, |
| { |
| "epoch": 1626.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4880 |
| }, |
| { |
| "epoch": 1627.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4884 |
| }, |
| { |
| "epoch": 1629.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4888 |
| }, |
| { |
| "epoch": 1630.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4892 |
| }, |
| { |
| "epoch": 1631.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4896 |
| }, |
| { |
| "epoch": 1633.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4900 |
| }, |
| { |
| "epoch": 1634.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4904 |
| }, |
| { |
| "epoch": 1635.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4908 |
| }, |
| { |
| "epoch": 1637.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4912 |
| }, |
| { |
| "epoch": 1638.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4916 |
| }, |
| { |
| "epoch": 1639.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4920 |
| }, |
| { |
| "epoch": 1641.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 4924 |
| }, |
| { |
| "epoch": 1642.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4928 |
| }, |
| { |
| "epoch": 1642.59, |
| "eval_exact_match": 0.5667311411992263, |
| "eval_exec": 0.5754352030947776, |
| "eval_loss": 0.5151154398918152, |
| "eval_runtime": 158.184, |
| "eval_samples_per_second": 6.537, |
| "step": 4928 |
| }, |
| { |
| "epoch": 1643.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4932 |
| }, |
| { |
| "epoch": 1645.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4936 |
| }, |
| { |
| "epoch": 1646.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4940 |
| }, |
| { |
| "epoch": 1647.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4944 |
| }, |
| { |
| "epoch": 1649.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4948 |
| }, |
| { |
| "epoch": 1650.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4952 |
| }, |
| { |
| "epoch": 1651.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4956 |
| }, |
| { |
| "epoch": 1653.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4960 |
| }, |
| { |
| "epoch": 1654.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4964 |
| }, |
| { |
| "epoch": 1655.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4968 |
| }, |
| { |
| "epoch": 1657.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 4972 |
| }, |
| { |
| "epoch": 1658.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4976 |
| }, |
| { |
| "epoch": 1659.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4980 |
| }, |
| { |
| "epoch": 1661.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4984 |
| }, |
| { |
| "epoch": 1662.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4988 |
| }, |
| { |
| "epoch": 1663.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 4992 |
| }, |
| { |
| "epoch": 1663.88, |
| "eval_exact_match": 0.5754352030947776, |
| "eval_exec": 0.5851063829787234, |
| "eval_loss": 0.5139929056167603, |
| "eval_runtime": 160.5278, |
| "eval_samples_per_second": 6.441, |
| "step": 4992 |
| }, |
| { |
| "epoch": 1665.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 4996 |
| }, |
| { |
| "epoch": 1666.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5000 |
| }, |
| { |
| "epoch": 1667.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5004 |
| }, |
| { |
| "epoch": 1669.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5008 |
| }, |
| { |
| "epoch": 1670.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5012 |
| }, |
| { |
| "epoch": 1671.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5016 |
| }, |
| { |
| "epoch": 1673.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5020 |
| }, |
| { |
| "epoch": 1674.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5024 |
| }, |
| { |
| "epoch": 1675.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5028 |
| }, |
| { |
| "epoch": 1677.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5032 |
| }, |
| { |
| "epoch": 1678.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5036 |
| }, |
| { |
| "epoch": 1679.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5040 |
| }, |
| { |
| "epoch": 1681.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5044 |
| }, |
| { |
| "epoch": 1682.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5048 |
| }, |
| { |
| "epoch": 1683.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5052 |
| }, |
| { |
| "epoch": 1685.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5056 |
| }, |
| { |
| "epoch": 1685.29, |
| "eval_exact_match": 0.5793036750483559, |
| "eval_exec": 0.5880077369439072, |
| "eval_loss": 0.5246556401252747, |
| "eval_runtime": 156.3008, |
| "eval_samples_per_second": 6.615, |
| "step": 5056 |
| }, |
| { |
| "epoch": 1686.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5060 |
| }, |
| { |
| "epoch": 1687.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5064 |
| }, |
| { |
| "epoch": 1689.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5068 |
| }, |
| { |
| "epoch": 1690.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5072 |
| }, |
| { |
| "epoch": 1691.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5076 |
| }, |
| { |
| "epoch": 1693.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5080 |
| }, |
| { |
| "epoch": 1694.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5084 |
| }, |
| { |
| "epoch": 1695.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0004, |
| "step": 5088 |
| }, |
| { |
| "epoch": 1697.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5092 |
| }, |
| { |
| "epoch": 1698.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5096 |
| }, |
| { |
| "epoch": 1699.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5100 |
| }, |
| { |
| "epoch": 1701.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5104 |
| }, |
| { |
| "epoch": 1702.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5108 |
| }, |
| { |
| "epoch": 1703.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5112 |
| }, |
| { |
| "epoch": 1705.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5116 |
| }, |
| { |
| "epoch": 1706.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5120 |
| }, |
| { |
| "epoch": 1706.59, |
| "eval_exact_match": 0.5705996131528046, |
| "eval_exec": 0.5773694390715667, |
| "eval_loss": 0.5314697027206421, |
| "eval_runtime": 157.2907, |
| "eval_samples_per_second": 6.574, |
| "step": 5120 |
| }, |
| { |
| "epoch": 1707.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5124 |
| }, |
| { |
| "epoch": 1709.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5128 |
| }, |
| { |
| "epoch": 1710.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5132 |
| }, |
| { |
| "epoch": 1711.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5136 |
| }, |
| { |
| "epoch": 1713.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5140 |
| }, |
| { |
| "epoch": 1714.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5144 |
| }, |
| { |
| "epoch": 1715.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5148 |
| }, |
| { |
| "epoch": 1717.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5152 |
| }, |
| { |
| "epoch": 1718.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5156 |
| }, |
| { |
| "epoch": 1719.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5160 |
| }, |
| { |
| "epoch": 1721.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5164 |
| }, |
| { |
| "epoch": 1722.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5168 |
| }, |
| { |
| "epoch": 1723.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5172 |
| }, |
| { |
| "epoch": 1725.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5176 |
| }, |
| { |
| "epoch": 1726.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5180 |
| }, |
| { |
| "epoch": 1727.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5184 |
| }, |
| { |
| "epoch": 1727.88, |
| "eval_exact_match": 0.5764023210831721, |
| "eval_exec": 0.5802707930367504, |
| "eval_loss": 0.5283452272415161, |
| "eval_runtime": 154.7301, |
| "eval_samples_per_second": 6.683, |
| "step": 5184 |
| }, |
| { |
| "epoch": 1729.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5188 |
| }, |
| { |
| "epoch": 1730.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5192 |
| }, |
| { |
| "epoch": 1731.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5196 |
| }, |
| { |
| "epoch": 1733.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5200 |
| }, |
| { |
| "epoch": 1734.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5204 |
| }, |
| { |
| "epoch": 1735.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5208 |
| }, |
| { |
| "epoch": 1737.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5212 |
| }, |
| { |
| "epoch": 1738.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5216 |
| }, |
| { |
| "epoch": 1739.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5220 |
| }, |
| { |
| "epoch": 1741.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5224 |
| }, |
| { |
| "epoch": 1742.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5228 |
| }, |
| { |
| "epoch": 1743.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5232 |
| }, |
| { |
| "epoch": 1745.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5236 |
| }, |
| { |
| "epoch": 1746.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5240 |
| }, |
| { |
| "epoch": 1747.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5244 |
| }, |
| { |
| "epoch": 1749.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5248 |
| }, |
| { |
| "epoch": 1749.29, |
| "eval_exact_match": 0.5735009671179884, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.5236985087394714, |
| "eval_runtime": 159.5661, |
| "eval_samples_per_second": 6.48, |
| "step": 5248 |
| }, |
| { |
| "epoch": 1750.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5252 |
| }, |
| { |
| "epoch": 1751.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5256 |
| }, |
| { |
| "epoch": 1753.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5260 |
| }, |
| { |
| "epoch": 1754.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5264 |
| }, |
| { |
| "epoch": 1755.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5268 |
| }, |
| { |
| "epoch": 1757.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5272 |
| }, |
| { |
| "epoch": 1758.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5276 |
| }, |
| { |
| "epoch": 1759.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5280 |
| }, |
| { |
| "epoch": 1761.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5284 |
| }, |
| { |
| "epoch": 1762.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5288 |
| }, |
| { |
| "epoch": 1763.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5292 |
| }, |
| { |
| "epoch": 1765.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5296 |
| }, |
| { |
| "epoch": 1766.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5300 |
| }, |
| { |
| "epoch": 1767.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5304 |
| }, |
| { |
| "epoch": 1769.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5308 |
| }, |
| { |
| "epoch": 1770.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5312 |
| }, |
| { |
| "epoch": 1770.59, |
| "eval_exact_match": 0.5667311411992263, |
| "eval_exec": 0.5667311411992263, |
| "eval_loss": 0.5358487963676453, |
| "eval_runtime": 155.837, |
| "eval_samples_per_second": 6.635, |
| "step": 5312 |
| }, |
| { |
| "epoch": 1771.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5316 |
| }, |
| { |
| "epoch": 1773.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5320 |
| }, |
| { |
| "epoch": 1774.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5324 |
| }, |
| { |
| "epoch": 1775.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5328 |
| }, |
| { |
| "epoch": 1777.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5332 |
| }, |
| { |
| "epoch": 1778.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5336 |
| }, |
| { |
| "epoch": 1779.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5340 |
| }, |
| { |
| "epoch": 1781.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5344 |
| }, |
| { |
| "epoch": 1782.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5348 |
| }, |
| { |
| "epoch": 1783.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5352 |
| }, |
| { |
| "epoch": 1785.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5356 |
| }, |
| { |
| "epoch": 1786.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5360 |
| }, |
| { |
| "epoch": 1787.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5364 |
| }, |
| { |
| "epoch": 1789.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5368 |
| }, |
| { |
| "epoch": 1790.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5372 |
| }, |
| { |
| "epoch": 1791.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5376 |
| }, |
| { |
| "epoch": 1791.88, |
| "eval_exact_match": 0.5754352030947776, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5336489081382751, |
| "eval_runtime": 158.8612, |
| "eval_samples_per_second": 6.509, |
| "step": 5376 |
| }, |
| { |
| "epoch": 1793.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 5380 |
| }, |
| { |
| "epoch": 1794.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5384 |
| }, |
| { |
| "epoch": 1795.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5388 |
| }, |
| { |
| "epoch": 1797.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5392 |
| }, |
| { |
| "epoch": 1798.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5396 |
| }, |
| { |
| "epoch": 1799.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5400 |
| }, |
| { |
| "epoch": 1801.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5404 |
| }, |
| { |
| "epoch": 1802.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5408 |
| }, |
| { |
| "epoch": 1803.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5412 |
| }, |
| { |
| "epoch": 1805.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5416 |
| }, |
| { |
| "epoch": 1806.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5420 |
| }, |
| { |
| "epoch": 1807.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5424 |
| }, |
| { |
| "epoch": 1809.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5428 |
| }, |
| { |
| "epoch": 1810.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5432 |
| }, |
| { |
| "epoch": 1811.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5436 |
| }, |
| { |
| "epoch": 1813.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5440 |
| }, |
| { |
| "epoch": 1813.29, |
| "eval_exact_match": 0.5638297872340425, |
| "eval_exec": 0.5773694390715667, |
| "eval_loss": 0.5105124711990356, |
| "eval_runtime": 165.7881, |
| "eval_samples_per_second": 6.237, |
| "step": 5440 |
| }, |
| { |
| "epoch": 1814.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5444 |
| }, |
| { |
| "epoch": 1815.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5448 |
| }, |
| { |
| "epoch": 1817.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5452 |
| }, |
| { |
| "epoch": 1818.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5456 |
| }, |
| { |
| "epoch": 1819.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5460 |
| }, |
| { |
| "epoch": 1821.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5464 |
| }, |
| { |
| "epoch": 1822.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5468 |
| }, |
| { |
| "epoch": 1823.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5472 |
| }, |
| { |
| "epoch": 1825.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5476 |
| }, |
| { |
| "epoch": 1826.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5480 |
| }, |
| { |
| "epoch": 1827.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5484 |
| }, |
| { |
| "epoch": 1829.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5488 |
| }, |
| { |
| "epoch": 1830.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5492 |
| }, |
| { |
| "epoch": 1831.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5496 |
| }, |
| { |
| "epoch": 1833.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5500 |
| }, |
| { |
| "epoch": 1834.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5504 |
| }, |
| { |
| "epoch": 1834.59, |
| "eval_exact_match": 0.574468085106383, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5365678668022156, |
| "eval_runtime": 152.3461, |
| "eval_samples_per_second": 6.787, |
| "step": 5504 |
| }, |
| { |
| "epoch": 1835.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5508 |
| }, |
| { |
| "epoch": 1837.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5512 |
| }, |
| { |
| "epoch": 1838.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5516 |
| }, |
| { |
| "epoch": 1839.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5520 |
| }, |
| { |
| "epoch": 1841.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5524 |
| }, |
| { |
| "epoch": 1842.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5528 |
| }, |
| { |
| "epoch": 1843.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5532 |
| }, |
| { |
| "epoch": 1845.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5536 |
| }, |
| { |
| "epoch": 1846.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5540 |
| }, |
| { |
| "epoch": 1847.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5544 |
| }, |
| { |
| "epoch": 1849.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5548 |
| }, |
| { |
| "epoch": 1850.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5552 |
| }, |
| { |
| "epoch": 1851.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5556 |
| }, |
| { |
| "epoch": 1853.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5560 |
| }, |
| { |
| "epoch": 1854.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5564 |
| }, |
| { |
| "epoch": 1855.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5568 |
| }, |
| { |
| "epoch": 1855.88, |
| "eval_exact_match": 0.59284332688588, |
| "eval_exec": 0.5996131528046421, |
| "eval_loss": 0.5280882120132446, |
| "eval_runtime": 153.4328, |
| "eval_samples_per_second": 6.739, |
| "step": 5568 |
| }, |
| { |
| "epoch": 1857.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5572 |
| }, |
| { |
| "epoch": 1858.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5576 |
| }, |
| { |
| "epoch": 1859.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5580 |
| }, |
| { |
| "epoch": 1861.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5584 |
| }, |
| { |
| "epoch": 1862.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5588 |
| }, |
| { |
| "epoch": 1863.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5592 |
| }, |
| { |
| "epoch": 1865.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5596 |
| }, |
| { |
| "epoch": 1866.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5600 |
| }, |
| { |
| "epoch": 1867.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5604 |
| }, |
| { |
| "epoch": 1869.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5608 |
| }, |
| { |
| "epoch": 1870.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5612 |
| }, |
| { |
| "epoch": 1871.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5616 |
| }, |
| { |
| "epoch": 1873.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5620 |
| }, |
| { |
| "epoch": 1874.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5624 |
| }, |
| { |
| "epoch": 1875.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5628 |
| }, |
| { |
| "epoch": 1877.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5632 |
| }, |
| { |
| "epoch": 1877.29, |
| "eval_exact_match": 0.5851063829787234, |
| "eval_exec": 0.5889748549323017, |
| "eval_loss": 0.5338191390037537, |
| "eval_runtime": 150.8255, |
| "eval_samples_per_second": 6.856, |
| "step": 5632 |
| }, |
| { |
| "epoch": 1878.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5636 |
| }, |
| { |
| "epoch": 1879.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5640 |
| }, |
| { |
| "epoch": 1881.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5644 |
| }, |
| { |
| "epoch": 1882.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5648 |
| }, |
| { |
| "epoch": 1883.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5652 |
| }, |
| { |
| "epoch": 1885.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5656 |
| }, |
| { |
| "epoch": 1886.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5660 |
| }, |
| { |
| "epoch": 1887.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5664 |
| }, |
| { |
| "epoch": 1889.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5668 |
| }, |
| { |
| "epoch": 1890.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5672 |
| }, |
| { |
| "epoch": 1891.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5676 |
| }, |
| { |
| "epoch": 1893.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5680 |
| }, |
| { |
| "epoch": 1894.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5684 |
| }, |
| { |
| "epoch": 1895.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5688 |
| }, |
| { |
| "epoch": 1897.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5692 |
| }, |
| { |
| "epoch": 1898.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5696 |
| }, |
| { |
| "epoch": 1898.59, |
| "eval_exact_match": 0.5764023210831721, |
| "eval_exec": 0.5841392649903289, |
| "eval_loss": 0.5366023182868958, |
| "eval_runtime": 155.027, |
| "eval_samples_per_second": 6.67, |
| "step": 5696 |
| }, |
| { |
| "epoch": 1899.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5700 |
| }, |
| { |
| "epoch": 1901.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5704 |
| }, |
| { |
| "epoch": 1902.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5708 |
| }, |
| { |
| "epoch": 1903.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5712 |
| }, |
| { |
| "epoch": 1905.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5716 |
| }, |
| { |
| "epoch": 1906.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5720 |
| }, |
| { |
| "epoch": 1907.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5724 |
| }, |
| { |
| "epoch": 1909.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0009, |
| "step": 5728 |
| }, |
| { |
| "epoch": 1910.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5732 |
| }, |
| { |
| "epoch": 1911.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5736 |
| }, |
| { |
| "epoch": 1913.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5740 |
| }, |
| { |
| "epoch": 1914.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5744 |
| }, |
| { |
| "epoch": 1915.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5748 |
| }, |
| { |
| "epoch": 1917.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5752 |
| }, |
| { |
| "epoch": 1918.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5756 |
| }, |
| { |
| "epoch": 1919.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5760 |
| }, |
| { |
| "epoch": 1919.88, |
| "eval_exact_match": 0.5676982591876208, |
| "eval_exec": 0.5773694390715667, |
| "eval_loss": 0.5160409212112427, |
| "eval_runtime": 161.7593, |
| "eval_samples_per_second": 6.392, |
| "step": 5760 |
| }, |
| { |
| "epoch": 1921.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5764 |
| }, |
| { |
| "epoch": 1922.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5768 |
| }, |
| { |
| "epoch": 1923.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5772 |
| }, |
| { |
| "epoch": 1925.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5776 |
| }, |
| { |
| "epoch": 1926.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5780 |
| }, |
| { |
| "epoch": 1927.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5784 |
| }, |
| { |
| "epoch": 1929.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5788 |
| }, |
| { |
| "epoch": 1930.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5792 |
| }, |
| { |
| "epoch": 1931.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5796 |
| }, |
| { |
| "epoch": 1933.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5800 |
| }, |
| { |
| "epoch": 1934.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5804 |
| }, |
| { |
| "epoch": 1935.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5808 |
| }, |
| { |
| "epoch": 1937.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5812 |
| }, |
| { |
| "epoch": 1938.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5816 |
| }, |
| { |
| "epoch": 1939.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5820 |
| }, |
| { |
| "epoch": 1941.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5824 |
| }, |
| { |
| "epoch": 1941.29, |
| "eval_exact_match": 0.5831721470019342, |
| "eval_exec": 0.5947775628626693, |
| "eval_loss": 0.5400940775871277, |
| "eval_runtime": 153.8305, |
| "eval_samples_per_second": 6.722, |
| "step": 5824 |
| }, |
| { |
| "epoch": 1942.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5828 |
| }, |
| { |
| "epoch": 1943.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5832 |
| }, |
| { |
| "epoch": 1945.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5836 |
| }, |
| { |
| "epoch": 1946.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5840 |
| }, |
| { |
| "epoch": 1947.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5844 |
| }, |
| { |
| "epoch": 1949.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5848 |
| }, |
| { |
| "epoch": 1950.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5852 |
| }, |
| { |
| "epoch": 1951.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5856 |
| }, |
| { |
| "epoch": 1953.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5860 |
| }, |
| { |
| "epoch": 1954.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5864 |
| }, |
| { |
| "epoch": 1955.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5868 |
| }, |
| { |
| "epoch": 1957.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5872 |
| }, |
| { |
| "epoch": 1958.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5876 |
| }, |
| { |
| "epoch": 1959.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5880 |
| }, |
| { |
| "epoch": 1961.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5884 |
| }, |
| { |
| "epoch": 1962.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5888 |
| }, |
| { |
| "epoch": 1962.59, |
| "eval_exact_match": 0.5638297872340425, |
| "eval_exec": 0.5676982591876208, |
| "eval_loss": 0.533243715763092, |
| "eval_runtime": 154.5568, |
| "eval_samples_per_second": 6.69, |
| "step": 5888 |
| }, |
| { |
| "epoch": 1963.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5892 |
| }, |
| { |
| "epoch": 1965.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5896 |
| }, |
| { |
| "epoch": 1966.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5900 |
| }, |
| { |
| "epoch": 1967.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5904 |
| }, |
| { |
| "epoch": 1969.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5908 |
| }, |
| { |
| "epoch": 1970.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5912 |
| }, |
| { |
| "epoch": 1971.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5916 |
| }, |
| { |
| "epoch": 1973.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5920 |
| }, |
| { |
| "epoch": 1974.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0021, |
| "step": 5924 |
| }, |
| { |
| "epoch": 1975.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5928 |
| }, |
| { |
| "epoch": 1977.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5932 |
| }, |
| { |
| "epoch": 1978.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5936 |
| }, |
| { |
| "epoch": 1979.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5940 |
| }, |
| { |
| "epoch": 1981.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5944 |
| }, |
| { |
| "epoch": 1982.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5948 |
| }, |
| { |
| "epoch": 1983.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5952 |
| }, |
| { |
| "epoch": 1983.88, |
| "eval_exact_match": 0.574468085106383, |
| "eval_exec": 0.5841392649903289, |
| "eval_loss": 0.5079652070999146, |
| "eval_runtime": 153.9904, |
| "eval_samples_per_second": 6.715, |
| "step": 5952 |
| }, |
| { |
| "epoch": 1985.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5956 |
| }, |
| { |
| "epoch": 1986.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5960 |
| }, |
| { |
| "epoch": 1987.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5964 |
| }, |
| { |
| "epoch": 1989.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5968 |
| }, |
| { |
| "epoch": 1990.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5972 |
| }, |
| { |
| "epoch": 1991.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5976 |
| }, |
| { |
| "epoch": 1993.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 5980 |
| }, |
| { |
| "epoch": 1994.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5984 |
| }, |
| { |
| "epoch": 1995.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5988 |
| }, |
| { |
| "epoch": 1997.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 5992 |
| }, |
| { |
| "epoch": 1998.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 5996 |
| }, |
| { |
| "epoch": 1999.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6000 |
| }, |
| { |
| "epoch": 2001.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6004 |
| }, |
| { |
| "epoch": 2002.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6008 |
| }, |
| { |
| "epoch": 2003.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6012 |
| }, |
| { |
| "epoch": 2005.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6016 |
| }, |
| { |
| "epoch": 2005.29, |
| "eval_exact_match": 0.5754352030947776, |
| "eval_exec": 0.5841392649903289, |
| "eval_loss": 0.5421436429023743, |
| "eval_runtime": 149.7153, |
| "eval_samples_per_second": 6.906, |
| "step": 6016 |
| }, |
| { |
| "epoch": 2006.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6020 |
| }, |
| { |
| "epoch": 2007.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6024 |
| }, |
| { |
| "epoch": 2009.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6028 |
| }, |
| { |
| "epoch": 2010.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6032 |
| }, |
| { |
| "epoch": 2011.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6036 |
| }, |
| { |
| "epoch": 2013.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6040 |
| }, |
| { |
| "epoch": 2014.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6044 |
| }, |
| { |
| "epoch": 2015.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6048 |
| }, |
| { |
| "epoch": 2017.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6052 |
| }, |
| { |
| "epoch": 2018.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6056 |
| }, |
| { |
| "epoch": 2019.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6060 |
| }, |
| { |
| "epoch": 2021.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6064 |
| }, |
| { |
| "epoch": 2022.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6068 |
| }, |
| { |
| "epoch": 2023.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6072 |
| }, |
| { |
| "epoch": 2025.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6076 |
| }, |
| { |
| "epoch": 2026.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6080 |
| }, |
| { |
| "epoch": 2026.59, |
| "eval_exact_match": 0.5725338491295938, |
| "eval_exec": 0.5793036750483559, |
| "eval_loss": 0.5351167917251587, |
| "eval_runtime": 157.4046, |
| "eval_samples_per_second": 6.569, |
| "step": 6080 |
| }, |
| { |
| "epoch": 2027.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6084 |
| }, |
| { |
| "epoch": 2029.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6088 |
| }, |
| { |
| "epoch": 2030.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6092 |
| }, |
| { |
| "epoch": 2031.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6096 |
| }, |
| { |
| "epoch": 2033.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6100 |
| }, |
| { |
| "epoch": 2034.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6104 |
| }, |
| { |
| "epoch": 2035.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6108 |
| }, |
| { |
| "epoch": 2037.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6112 |
| }, |
| { |
| "epoch": 2038.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6116 |
| }, |
| { |
| "epoch": 2039.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6120 |
| }, |
| { |
| "epoch": 2041.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6124 |
| }, |
| { |
| "epoch": 2042.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6128 |
| }, |
| { |
| "epoch": 2043.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6132 |
| }, |
| { |
| "epoch": 2045.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6136 |
| }, |
| { |
| "epoch": 2046.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6140 |
| }, |
| { |
| "epoch": 2047.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6144 |
| }, |
| { |
| "epoch": 2047.88, |
| "eval_exact_match": 0.5938104448742747, |
| "eval_exec": 0.5967117988394585, |
| "eval_loss": 0.5208129286766052, |
| "eval_runtime": 150.5875, |
| "eval_samples_per_second": 6.866, |
| "step": 6144 |
| }, |
| { |
| "epoch": 2049.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 6148 |
| }, |
| { |
| "epoch": 2050.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6152 |
| }, |
| { |
| "epoch": 2051.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6156 |
| }, |
| { |
| "epoch": 2053.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6160 |
| }, |
| { |
| "epoch": 2054.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6164 |
| }, |
| { |
| "epoch": 2055.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6168 |
| }, |
| { |
| "epoch": 2057.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6172 |
| }, |
| { |
| "epoch": 2058.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6176 |
| }, |
| { |
| "epoch": 2059.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6180 |
| }, |
| { |
| "epoch": 2061.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6184 |
| }, |
| { |
| "epoch": 2062.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6188 |
| }, |
| { |
| "epoch": 2063.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6192 |
| }, |
| { |
| "epoch": 2065.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6196 |
| }, |
| { |
| "epoch": 2066.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6200 |
| }, |
| { |
| "epoch": 2067.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6204 |
| }, |
| { |
| "epoch": 2069.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6208 |
| }, |
| { |
| "epoch": 2069.29, |
| "eval_exact_match": 0.5793036750483559, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.5256672501564026, |
| "eval_runtime": 143.3402, |
| "eval_samples_per_second": 7.214, |
| "step": 6208 |
| }, |
| { |
| "epoch": 2070.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6212 |
| }, |
| { |
| "epoch": 2071.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6216 |
| }, |
| { |
| "epoch": 2073.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6220 |
| }, |
| { |
| "epoch": 2074.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6224 |
| }, |
| { |
| "epoch": 2075.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6228 |
| }, |
| { |
| "epoch": 2077.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6232 |
| }, |
| { |
| "epoch": 2078.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6236 |
| }, |
| { |
| "epoch": 2079.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6240 |
| }, |
| { |
| "epoch": 2081.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6244 |
| }, |
| { |
| "epoch": 2082.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6248 |
| }, |
| { |
| "epoch": 2083.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6252 |
| }, |
| { |
| "epoch": 2085.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6256 |
| }, |
| { |
| "epoch": 2086.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6260 |
| }, |
| { |
| "epoch": 2087.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6264 |
| }, |
| { |
| "epoch": 2089.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 6268 |
| }, |
| { |
| "epoch": 2090.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6272 |
| }, |
| { |
| "epoch": 2090.59, |
| "eval_exact_match": 0.5773694390715667, |
| "eval_exec": 0.5822050290135397, |
| "eval_loss": 0.5306528806686401, |
| "eval_runtime": 153.5764, |
| "eval_samples_per_second": 6.733, |
| "step": 6272 |
| }, |
| { |
| "epoch": 2091.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6276 |
| }, |
| { |
| "epoch": 2093.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6280 |
| }, |
| { |
| "epoch": 2094.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6284 |
| }, |
| { |
| "epoch": 2095.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6288 |
| }, |
| { |
| "epoch": 2097.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6292 |
| }, |
| { |
| "epoch": 2098.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6296 |
| }, |
| { |
| "epoch": 2099.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6300 |
| }, |
| { |
| "epoch": 2101.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6304 |
| }, |
| { |
| "epoch": 2102.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6308 |
| }, |
| { |
| "epoch": 2103.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6312 |
| }, |
| { |
| "epoch": 2105.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6316 |
| }, |
| { |
| "epoch": 2106.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6320 |
| }, |
| { |
| "epoch": 2107.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6324 |
| }, |
| { |
| "epoch": 2109.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6328 |
| }, |
| { |
| "epoch": 2110.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6332 |
| }, |
| { |
| "epoch": 2111.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6336 |
| }, |
| { |
| "epoch": 2111.88, |
| "eval_exact_match": 0.5773694390715667, |
| "eval_exec": 0.5851063829787234, |
| "eval_loss": 0.5272626280784607, |
| "eval_runtime": 152.3079, |
| "eval_samples_per_second": 6.789, |
| "step": 6336 |
| }, |
| { |
| "epoch": 2113.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6340 |
| }, |
| { |
| "epoch": 2114.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6344 |
| }, |
| { |
| "epoch": 2115.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6348 |
| }, |
| { |
| "epoch": 2117.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6352 |
| }, |
| { |
| "epoch": 2118.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6356 |
| }, |
| { |
| "epoch": 2119.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6360 |
| }, |
| { |
| "epoch": 2121.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6364 |
| }, |
| { |
| "epoch": 2122.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6368 |
| }, |
| { |
| "epoch": 2123.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6372 |
| }, |
| { |
| "epoch": 2125.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6376 |
| }, |
| { |
| "epoch": 2126.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6380 |
| }, |
| { |
| "epoch": 2127.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6384 |
| }, |
| { |
| "epoch": 2129.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6388 |
| }, |
| { |
| "epoch": 2130.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6392 |
| }, |
| { |
| "epoch": 2131.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6396 |
| }, |
| { |
| "epoch": 2133.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6400 |
| }, |
| { |
| "epoch": 2133.29, |
| "eval_exact_match": 0.5831721470019342, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5177983045578003, |
| "eval_runtime": 155.3624, |
| "eval_samples_per_second": 6.655, |
| "step": 6400 |
| }, |
| { |
| "epoch": 2134.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6404 |
| }, |
| { |
| "epoch": 2135.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0007, |
| "step": 6408 |
| }, |
| { |
| "epoch": 2137.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6412 |
| }, |
| { |
| "epoch": 2138.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 6416 |
| }, |
| { |
| "epoch": 2139.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6420 |
| }, |
| { |
| "epoch": 2141.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6424 |
| }, |
| { |
| "epoch": 2142.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6428 |
| }, |
| { |
| "epoch": 2143.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6432 |
| }, |
| { |
| "epoch": 2145.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6436 |
| }, |
| { |
| "epoch": 2146.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6440 |
| }, |
| { |
| "epoch": 2147.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6444 |
| }, |
| { |
| "epoch": 2149.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6448 |
| }, |
| { |
| "epoch": 2150.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6452 |
| }, |
| { |
| "epoch": 2151.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6456 |
| }, |
| { |
| "epoch": 2153.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6460 |
| }, |
| { |
| "epoch": 2154.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6464 |
| }, |
| { |
| "epoch": 2154.59, |
| "eval_exact_match": 0.5851063829787234, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.5292801260948181, |
| "eval_runtime": 158.0702, |
| "eval_samples_per_second": 6.541, |
| "step": 6464 |
| }, |
| { |
| "epoch": 2155.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6468 |
| }, |
| { |
| "epoch": 2157.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6472 |
| }, |
| { |
| "epoch": 2158.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6476 |
| }, |
| { |
| "epoch": 2159.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6480 |
| }, |
| { |
| "epoch": 2161.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6484 |
| }, |
| { |
| "epoch": 2162.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6488 |
| }, |
| { |
| "epoch": 2163.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6492 |
| }, |
| { |
| "epoch": 2165.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6496 |
| }, |
| { |
| "epoch": 2166.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6500 |
| }, |
| { |
| "epoch": 2167.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6504 |
| }, |
| { |
| "epoch": 2169.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6508 |
| }, |
| { |
| "epoch": 2170.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6512 |
| }, |
| { |
| "epoch": 2171.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6516 |
| }, |
| { |
| "epoch": 2173.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6520 |
| }, |
| { |
| "epoch": 2174.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6524 |
| }, |
| { |
| "epoch": 2175.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6528 |
| }, |
| { |
| "epoch": 2175.88, |
| "eval_exact_match": 0.5812379110251451, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5337159633636475, |
| "eval_runtime": 165.0332, |
| "eval_samples_per_second": 6.265, |
| "step": 6528 |
| }, |
| { |
| "epoch": 2177.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6532 |
| }, |
| { |
| "epoch": 2178.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6536 |
| }, |
| { |
| "epoch": 2179.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6540 |
| }, |
| { |
| "epoch": 2181.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6544 |
| }, |
| { |
| "epoch": 2182.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6548 |
| }, |
| { |
| "epoch": 2183.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6552 |
| }, |
| { |
| "epoch": 2185.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6556 |
| }, |
| { |
| "epoch": 2186.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6560 |
| }, |
| { |
| "epoch": 2187.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6564 |
| }, |
| { |
| "epoch": 2189.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6568 |
| }, |
| { |
| "epoch": 2190.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 6572 |
| }, |
| { |
| "epoch": 2191.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6576 |
| }, |
| { |
| "epoch": 2193.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6580 |
| }, |
| { |
| "epoch": 2194.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6584 |
| }, |
| { |
| "epoch": 2195.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6588 |
| }, |
| { |
| "epoch": 2197.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6592 |
| }, |
| { |
| "epoch": 2197.29, |
| "eval_exact_match": 0.5802707930367504, |
| "eval_exec": 0.5909090909090909, |
| "eval_loss": 0.5296819806098938, |
| "eval_runtime": 157.7645, |
| "eval_samples_per_second": 6.554, |
| "step": 6592 |
| }, |
| { |
| "epoch": 2198.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6596 |
| }, |
| { |
| "epoch": 2199.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6600 |
| }, |
| { |
| "epoch": 2201.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6604 |
| }, |
| { |
| "epoch": 2202.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6608 |
| }, |
| { |
| "epoch": 2203.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6612 |
| }, |
| { |
| "epoch": 2205.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6616 |
| }, |
| { |
| "epoch": 2206.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6620 |
| }, |
| { |
| "epoch": 2207.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6624 |
| }, |
| { |
| "epoch": 2209.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6628 |
| }, |
| { |
| "epoch": 2210.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6632 |
| }, |
| { |
| "epoch": 2211.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6636 |
| }, |
| { |
| "epoch": 2213.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6640 |
| }, |
| { |
| "epoch": 2214.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6644 |
| }, |
| { |
| "epoch": 2215.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6648 |
| }, |
| { |
| "epoch": 2217.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6652 |
| }, |
| { |
| "epoch": 2218.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6656 |
| }, |
| { |
| "epoch": 2218.59, |
| "eval_exact_match": 0.5783365570599613, |
| "eval_exec": 0.5889748549323017, |
| "eval_loss": 0.5069222450256348, |
| "eval_runtime": 155.3024, |
| "eval_samples_per_second": 6.658, |
| "step": 6656 |
| }, |
| { |
| "epoch": 2219.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6660 |
| }, |
| { |
| "epoch": 2221.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0003, |
| "step": 6664 |
| }, |
| { |
| "epoch": 2222.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6668 |
| }, |
| { |
| "epoch": 2223.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6672 |
| }, |
| { |
| "epoch": 2225.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6676 |
| }, |
| { |
| "epoch": 2226.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6680 |
| }, |
| { |
| "epoch": 2227.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6684 |
| }, |
| { |
| "epoch": 2229.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6688 |
| }, |
| { |
| "epoch": 2230.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6692 |
| }, |
| { |
| "epoch": 2231.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6696 |
| }, |
| { |
| "epoch": 2233.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6700 |
| }, |
| { |
| "epoch": 2234.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6704 |
| }, |
| { |
| "epoch": 2235.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6708 |
| }, |
| { |
| "epoch": 2237.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6712 |
| }, |
| { |
| "epoch": 2238.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6716 |
| }, |
| { |
| "epoch": 2239.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6720 |
| }, |
| { |
| "epoch": 2239.88, |
| "eval_exact_match": 0.5909090909090909, |
| "eval_exec": 0.5938104448742747, |
| "eval_loss": 0.5312435030937195, |
| "eval_runtime": 163.4814, |
| "eval_samples_per_second": 6.325, |
| "step": 6720 |
| }, |
| { |
| "epoch": 2241.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0005, |
| "step": 6724 |
| }, |
| { |
| "epoch": 2242.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6728 |
| }, |
| { |
| "epoch": 2243.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6732 |
| }, |
| { |
| "epoch": 2245.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6736 |
| }, |
| { |
| "epoch": 2246.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6740 |
| }, |
| { |
| "epoch": 2247.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6744 |
| }, |
| { |
| "epoch": 2249.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6748 |
| }, |
| { |
| "epoch": 2250.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6752 |
| }, |
| { |
| "epoch": 2251.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6756 |
| }, |
| { |
| "epoch": 2253.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6760 |
| }, |
| { |
| "epoch": 2254.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6764 |
| }, |
| { |
| "epoch": 2255.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6768 |
| }, |
| { |
| "epoch": 2257.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6772 |
| }, |
| { |
| "epoch": 2258.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6776 |
| }, |
| { |
| "epoch": 2259.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6780 |
| }, |
| { |
| "epoch": 2261.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6784 |
| }, |
| { |
| "epoch": 2261.29, |
| "eval_exact_match": 0.5764023210831721, |
| "eval_exec": 0.5870406189555126, |
| "eval_loss": 0.5298124551773071, |
| "eval_runtime": 164.9545, |
| "eval_samples_per_second": 6.268, |
| "step": 6784 |
| }, |
| { |
| "epoch": 2262.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6788 |
| }, |
| { |
| "epoch": 2263.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6792 |
| }, |
| { |
| "epoch": 2265.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6796 |
| }, |
| { |
| "epoch": 2266.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6800 |
| }, |
| { |
| "epoch": 2267.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6804 |
| }, |
| { |
| "epoch": 2269.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6808 |
| }, |
| { |
| "epoch": 2270.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6812 |
| }, |
| { |
| "epoch": 2271.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6816 |
| }, |
| { |
| "epoch": 2273.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6820 |
| }, |
| { |
| "epoch": 2274.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6824 |
| }, |
| { |
| "epoch": 2275.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6828 |
| }, |
| { |
| "epoch": 2277.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6832 |
| }, |
| { |
| "epoch": 2278.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6836 |
| }, |
| { |
| "epoch": 2279.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6840 |
| }, |
| { |
| "epoch": 2281.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6844 |
| }, |
| { |
| "epoch": 2282.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6848 |
| }, |
| { |
| "epoch": 2282.59, |
| "eval_exact_match": 0.586073500967118, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.536562979221344, |
| "eval_runtime": 154.5453, |
| "eval_samples_per_second": 6.691, |
| "step": 6848 |
| }, |
| { |
| "epoch": 2283.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6852 |
| }, |
| { |
| "epoch": 2285.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6856 |
| }, |
| { |
| "epoch": 2286.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6860 |
| }, |
| { |
| "epoch": 2287.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6864 |
| }, |
| { |
| "epoch": 2289.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6868 |
| }, |
| { |
| "epoch": 2290.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6872 |
| }, |
| { |
| "epoch": 2291.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6876 |
| }, |
| { |
| "epoch": 2293.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6880 |
| }, |
| { |
| "epoch": 2294.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6884 |
| }, |
| { |
| "epoch": 2295.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6888 |
| }, |
| { |
| "epoch": 2297.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6892 |
| }, |
| { |
| "epoch": 2298.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6896 |
| }, |
| { |
| "epoch": 2299.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6900 |
| }, |
| { |
| "epoch": 2301.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6904 |
| }, |
| { |
| "epoch": 2302.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6908 |
| }, |
| { |
| "epoch": 2303.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6912 |
| }, |
| { |
| "epoch": 2303.88, |
| "eval_exact_match": 0.5802707930367504, |
| "eval_exec": 0.5918762088974855, |
| "eval_loss": 0.5313507914543152, |
| "eval_runtime": 158.3017, |
| "eval_samples_per_second": 6.532, |
| "step": 6912 |
| }, |
| { |
| "epoch": 2305.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6916 |
| }, |
| { |
| "epoch": 2306.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6920 |
| }, |
| { |
| "epoch": 2307.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6924 |
| }, |
| { |
| "epoch": 2309.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6928 |
| }, |
| { |
| "epoch": 2310.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6932 |
| }, |
| { |
| "epoch": 2311.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6936 |
| }, |
| { |
| "epoch": 2313.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6940 |
| }, |
| { |
| "epoch": 2314.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6944 |
| }, |
| { |
| "epoch": 2315.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6948 |
| }, |
| { |
| "epoch": 2317.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0002, |
| "step": 6952 |
| }, |
| { |
| "epoch": 2318.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6956 |
| }, |
| { |
| "epoch": 2319.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6960 |
| }, |
| { |
| "epoch": 2321.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6964 |
| }, |
| { |
| "epoch": 2322.59, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6968 |
| }, |
| { |
| "epoch": 2323.88, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6972 |
| }, |
| { |
| "epoch": 2325.29, |
| "learning_rate": 0.0001, |
| "loss": 0.0001, |
| "step": 6976 |
| }, |
| { |
| "epoch": 2325.29, |
| "eval_exact_match": 0.5938104448742747, |
| "eval_exec": 0.6005802707930368, |
| "eval_loss": 0.5208510160446167, |
| "eval_runtime": 154.6385, |
| "eval_samples_per_second": 6.687, |
| "step": 6976 |
| } |
| ], |
| "max_steps": 9216, |
| "num_train_epochs": 3072, |
| "total_flos": 8.325069030825984e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
