Long-context understanding on MuSiQue
51SubEMLogo-PO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Logo-POBackbone=Qwen2.5-14B-Instruct2025.08 | 51 | 42.04 | |
| LongMabBackbone=Qwen2.5-14B-Instruct2025.08 | 50 | 50.34 | |
| LongMabBackbone=Llama-3.1-8B-Instruct2025.08 | 50 | 52.15 | |
| LongFaith-POBackbone=Qwen2.5-14B-Instruct2025.08 | 49.33 | 51 | |
| LongReward-POBackbone=Qwen2.5-14B-Instruct2025.08 | 48.5 | 43.41 | |
| LongFaith-POBackbone=Qwen-2.5-7B-Instruct2025.08 | 48.5 | 43.38 | |
| Vanilla LLMBackbone=Qwen2.5-14B-Instruct2025.08 | 46.5 | 39.18 | |
| SeaLong-POBackbone=Qwen2.5-14B-Instruct2025.08 | 46.5 | 37.56 | |
| LongMab-SFTBackbone=Qwen2.5-14B-Instruct2025.08 | 45 | 44 | |
| LongFaith-POBackbone=Llama-3.1-8B-Instruct2025.08 | 44 | 49.23 | |
| LongMabBackbone=Qwen-2.5-7B-Instruct2025.08 | 44 | 43.25 | |
| LongAlpacaBackbone=Qwen2.5-14B-Instruct2025.08 | 43.5 | 41.1 | |
| Logo-POBackbone=Llama-3.1-8B-Instruct2025.08 | 43.5 | 46.58 | |
| SeaLong-POBackbone=Qwen-2.5-7B-Instruct2025.08 | 43 | 22.09 | |
| Logo-POBackbone=Qwen-2.5-7B-Instruct2025.08 | 41 | 36.48 | |
| LongReward-POBackbone=Llama-3.1-8B-Instruct2025.08 | 40.5 | 41.41 | |
| LongAlignBackbone=Qwen2.5-14B-Instruct2025.08 | 37.5 | 38.33 | |
| SeaLong-POBackbone=Llama-3.1-8B-Instruct2025.08 | 37.5 | 40.52 | |
| LongReward-POBackbone=Qwen-2.5-7B-Instruct2025.08 | 37.5 | 33.37 | |
| LongAlignBackbone=Llama-3.1-8B-Instruct2025.08 | 36.5 | 37.62 | |
| LongMab-SFTBackbone=Qwen-2.5-7B-Instruct2025.08 | 36.5 | 39.32 | |
| LongMab-SFTBackbone=Llama-3.1-8B-Instruct2025.08 | 35 | 39.6 | |
| LongAlpacaBackbone=Llama-3.1-8B-Instruct2025.08 | 34 | 38.36 | |
| Vanilla LLMBackbone=Llama-3.1-8B-Instruct2025.08 | 33.5 | 36.56 | |
| Vanilla LLMBackbone=Qwen-2.5-7B-Instruct2025.08 | 33.5 | 30.52 | |
| LongAlpacaBackbone=Qwen-2.5-7B-Instruct2025.08 | 32.5 | 33.9 | |
| LongAlignBackbone=Qwen-2.5-7B-Instruct2025.08 | 28.5 | 31.09 |