| { | |
| "base_model": "Qwen/Qwen2.5-0.5B-Instruct", | |
| "output_dir": "models\\club-slm", | |
| "corpus": "J:\\MAJOR PROJECT 2026\\MAIN\\data\\clean_corpus.txt", | |
| "dataset": "data\\python\\train_qa.jsonl", | |
| "qa_pairs": 7821, | |
| "train_rows": 7429, | |
| "eval_rows": 392, | |
| "epochs": 5, | |
| "batch_size": 2, | |
| "grad_accum": 8, | |
| "effective_bs": 16, | |
| "lr": 3e-05, | |
| "max_length": 256, | |
| "precision": "bf16", | |
| "device": "cuda", | |
| "train_time_min": 119.6, | |
| "eval_loss": 0.335335, | |
| "perplexity": 1.3984, | |
| "token_accuracy": 0.32498, | |
| "exact_match": 0.0, | |
| "fuzzy_match": 0.0, | |
| "total_tokens": 99960, | |
| "correct_tokens": 32485, | |
| "wrong_tokens": 67475, | |
| "total_samples": 392, | |
| "exact_matches": 0, | |
| "fuzzy_matches": 0, | |
| "epoch_losses": [ | |
| { | |
| "epoch": 0.02, | |
| "train_loss": 3.673507, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "train_loss": 3.380849, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "train_loss": 2.831332, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "train_loss": 2.20228, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "train_loss": 1.890123, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "train_loss": 1.625928, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "train_loss": 1.608759, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "train_loss": 1.523835, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "train_loss": 1.391224, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "train_loss": 1.276159, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "train_loss": 1.265831, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "train_loss": 1.06008, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "train_loss": 1.173187, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "train_loss": 1.191657, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "train_loss": 1.120185, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "train_loss": 1.093464, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "train_loss": 0.995333, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "train_loss": 1.068814, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "train_loss": 1.006557, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "train_loss": 1.053628, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "train_loss": 0.995194, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "train_loss": 0.957164, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "train_loss": 0.979695, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "train_loss": 0.925564, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "train_loss": 0.888211, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "train_loss": 0.914199, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "train_loss": 0.893902, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "train_loss": 0.881656, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "train_loss": 0.850408, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "train_loss": 0.809787, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "train_loss": 0.853976, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "train_loss": 0.821917, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "train_loss": 0.881752, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "train_loss": 0.770348, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "train_loss": 0.760463, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "train_loss": 0.791171, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "train_loss": 0.717607, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "train_loss": 0.782005, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "train_loss": 0.772476, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "train_loss": 0.727276, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "train_loss": 0.781023, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "train_loss": 0.752274, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "train_loss": 0.717247, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "train_loss": 0.720272, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "train_loss": 0.737959, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "train_loss": 0.697645, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "train_loss": 0.548655, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "train_loss": 0.550381, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "train_loss": 0.569225, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "train_loss": 0.538405, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "train_loss": 0.520008, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "train_loss": 0.525656, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "train_loss": 0.555653, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "train_loss": 0.554701, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "train_loss": 0.56037, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "train_loss": 0.571677, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "train_loss": 0.506848, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "train_loss": 0.503055, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "train_loss": 0.560167, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "train_loss": 0.528751, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "train_loss": 0.475181, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "train_loss": 0.537284, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "train_loss": 0.518218, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "train_loss": 0.528746, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "train_loss": 0.464697, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "train_loss": 0.494603, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "train_loss": 0.502719, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "train_loss": 0.477892, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "train_loss": 0.489148, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "train_loss": 0.478489, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "train_loss": 0.519881, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "train_loss": 0.468913, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "train_loss": 0.530156, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "train_loss": 0.449579, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "train_loss": 0.433122, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "train_loss": 0.466816, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "train_loss": 0.466718, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "train_loss": 0.449744, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "train_loss": 0.429334, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "train_loss": 0.465894, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "train_loss": 0.392776, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "train_loss": 0.470184, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "train_loss": 0.463271, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "train_loss": 0.427761, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "train_loss": 0.421049, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "train_loss": 0.41458, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "train_loss": 0.425634, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "train_loss": 0.412593, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "train_loss": 0.417373, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "train_loss": 0.403305, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "train_loss": 0.411155, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "train_loss": 0.427315, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "train_loss": 0.402502, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "train_loss": 0.304085, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "train_loss": 0.285231, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "train_loss": 0.290745, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "train_loss": 0.270124, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "train_loss": 0.283372, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "train_loss": 0.284532, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "train_loss": 0.270921, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "train_loss": 0.282726, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "train_loss": 0.284413, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "train_loss": 0.2682, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "train_loss": 0.268169, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "train_loss": 0.265762, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "train_loss": 0.274262, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "train_loss": 0.267679, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "train_loss": 0.276804, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "train_loss": 0.276497, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "train_loss": 0.28569, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "train_loss": 0.26311, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "train_loss": 0.25099, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "train_loss": 0.274988, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "train_loss": 0.261295, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "train_loss": 0.245447, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "train_loss": 0.287245, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "train_loss": 0.260615, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "train_loss": 0.270639, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "train_loss": 0.263159, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "train_loss": 0.243453, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "train_loss": 0.242194, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "train_loss": 0.230759, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "train_loss": 0.243978, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "train_loss": 0.257585, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "train_loss": 0.251543, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "train_loss": 0.230343, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "train_loss": 0.274376, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "train_loss": 0.261679, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "train_loss": 0.253968, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "train_loss": 0.239549, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "train_loss": 0.238915, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "train_loss": 0.234489, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "train_loss": 0.232685, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "train_loss": 0.231499, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "train_loss": 0.223252, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "train_loss": 0.244325, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "train_loss": 0.255999, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "train_loss": 0.264762, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "train_loss": 0.246688, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "train_loss": 0.204261, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "train_loss": 0.212866, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "train_loss": 0.18749, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "train_loss": 0.187718, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "train_loss": 0.184144, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "train_loss": 0.189121, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "train_loss": 0.201176, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "train_loss": 0.189117, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "train_loss": 0.186818, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "train_loss": 0.198263, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "train_loss": 0.183168, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "train_loss": 0.181653, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "train_loss": 0.18247, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "train_loss": 0.19287, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "train_loss": 0.203817, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "train_loss": 0.191013, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "train_loss": 0.172727, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "train_loss": 0.185981, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "train_loss": 0.188215, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "train_loss": 0.179959, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "train_loss": 0.18324, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "train_loss": 0.1866, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "train_loss": 0.187446, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "train_loss": 0.187642, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "train_loss": 0.185423, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "train_loss": 0.181782, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "train_loss": 0.189735, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "train_loss": 0.209074, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "train_loss": 0.192625, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "train_loss": 0.188765, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "train_loss": 0.191266, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "train_loss": 0.194638, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "train_loss": 0.182128, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "train_loss": 0.190359, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "train_loss": 0.181668, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "train_loss": 0.177796, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "train_loss": 0.184682, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "train_loss": 0.173, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "train_loss": 0.175888, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "train_loss": 0.172485, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "train_loss": 0.189007, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "train_loss": 0.206134, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "train_loss": 0.18332, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "train_loss": 0.195303, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "train_loss": 0.195526, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "train_loss": 0.188107, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "train_loss": 0.195283, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "train_loss": 0.176317, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "train_loss": 0.170773, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "train_loss": 0.180002, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "train_loss": 0.169515, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "train_loss": 0.182661, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "train_loss": 0.169262, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "train_loss": 0.175039, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "train_loss": 0.17197, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "train_loss": 0.174948, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "train_loss": 0.186722, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "train_loss": 0.183905, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "train_loss": 0.175902, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "train_loss": 0.178009, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "train_loss": 0.189156, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "train_loss": 0.172458, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "train_loss": 0.174846, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "train_loss": 0.171839, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "train_loss": 0.174276, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "train_loss": 0.172959, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "train_loss": 0.176811, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "train_loss": 0.17267, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "train_loss": 0.179714, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "train_loss": 0.179473, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "train_loss": 0.172328, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "train_loss": 0.176313, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "train_loss": 0.177838, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "train_loss": 0.183181, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "train_loss": 0.176651, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "train_loss": 0.169534, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "train_loss": 0.167478, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "train_loss": 0.178978, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "train_loss": 0.176534, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "train_loss": 0.175687, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "train_loss": 0.178287, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "train_loss": 0.178707, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "train_loss": 0.187475, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "train_loss": 0.184983, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "train_loss": 0.17736, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "train_loss": 0.171934, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "train_loss": 0.196926, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "train_loss": 0.175234, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "train_loss": 0.18753, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "train_loss": 0.166468, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "train_loss": 0.170573, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "train_loss": 0.177368, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "train_loss": 0.162907, | |
| "step": 2320 | |
| } | |
| ], | |
| "eval_losses": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.727038 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.457646 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.353403 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.335049 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.335335 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.335335 | |
| } | |
| ], | |
| "sample_log": [ | |
| { | |
| "sample": 1, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "47/255" | |
| }, | |
| { | |
| "sample": 2, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "78/255" | |
| }, | |
| { | |
| "sample": 3, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "109/255" | |
| }, | |
| { | |
| "sample": 4, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Ex", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "75/255" | |
| }, | |
| { | |
| "sample": 5, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "132/255" | |
| }, | |
| { | |
| "sample": 6, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Ex", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "149/255" | |
| }, | |
| { | |
| "sample": 7, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "38/255" | |
| }, | |
| { | |
| "sample": 8, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Wh", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "49/255" | |
| }, | |
| { | |
| "sample": 9, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Ex", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "108/255" | |
| }, | |
| { | |
| "sample": 10, | |
| "reference": "You are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: Ex", | |
| "predicted": " are a small domain model trained on one university corpus.\nAnswer briefly and exactly from the corpus.\n\nQuestion: What ", | |
| "exact": false, | |
| "fuzzy": false, | |
| "tok_acc": "91/255" | |
| } | |
| ] | |
| } |