Long-context Understanding on LongBench (Task Performance Metrics)
32.03Single-Doc QA ScoreET
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ETTarget Context Window=64K, Training Corpus=1B tokens2026.05 | 32.03 | 30.81 | 26.04 | 68.04 | 4.54 | 66.48 | 38.3 | |
| Full FTTarget Context Window=64K, Training Corpus=1B tokens2026.05 | 28.86 | 29.51 | 24.88 | 62.81 | 6.31 | 47.86 | 35.63 | |
| LCEGTarget Context Window=64K, Training Corpus=1B tokens2026.05 | 27.86 | 28.51 | 23.88 | 61.81 | 5.31 | 46.86 | 36.61 | |
| LongLoRATarget Context Window=64K, Training Corpus=1B tokens2026.05 | 26.86 | 27.51 | 22.88 | 60.81 | 4.31 | 45.86 | 36.84 |