Long-context Understanding on LooGLE
57CR ScoreLongCrafter
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LongCrafterBackbone=LLaMA-3.1-8B, SFT Method=LongCrafter2026.07 | 57 | 34 | 53 | 42 | 46.5 | |
| LongMagpieBackbone=LLaMA-3.1-8B, SFT Method=LongMagpie2026.07 | 54 | 26 | 30 | 41 | 37.75 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B, SFT Method=Instruct (official)2026.07 | 53 | 29 | 53 | 36 | 42.8 | |
| LongCrafterBackbone=Qwen2.5-7B, SFT Method=LongCrafter2026.07 | 53 | 31 | 53 | 37 | 43.5 | |
| LongAlignBackbone=LLaMA-3.1-8B, SFT Method=LongAlign2026.07 | 52 | 29 | 37 | 39 | 39.25 | |
| LongRewardBackbone=LLaMA-3.1-8B, SFT Method=LongReward2026.07 | 52 | 27 | 39 | 37 | 38.75 | |
| LongFaithBackbone=LLaMA-3.1-8B, SFT Method=LongFaith2026.07 | 52 | 21 | 40 | 31 | 36 | |
| LongAlignBackbone=Qwen2.5-7B, SFT Method=LongAlign2026.07 | 51 | 24 | 38 | 27 | 35 | |
| LLaMA-3.1-8B-InstructBackbone=LLaMA-3.1-8B, SFT Method=Instruct (official)2026.07 | 51 | 30 | 46 | 35 | 40.5 | |
| LongMagpieBackbone=Qwen2.5-7B, SFT Method=LongMagpie2026.07 | 47 | 25 | 49 | 32 | 38.3 | |
| LongRewardBackbone=Qwen2.5-7B, SFT Method=LongReward2026.07 | 45 | 24 | 38 | 29 | 34 | |
| LongFaithBackbone=Qwen2.5-7B, SFT Method=LongFaith2026.07 | 45 | 20 | 43 | 27 | 33.8 |