Question Answering on 2-Hop QA Full Knowledge
54.91AccThinking States
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Thinking StatesBase Model=Qwen2.5-Base-1.5B2026.02 | 54.91 | 1.19 | |
| CoTBase Model=Qwen2.5-Base-1.5B2026.02 | 54.79 | 1 | |
| CoconutBase Model=Qwen2.5-Base-1.5B2026.02 | 33.71 | 1.14 | |
| No CoTBase Model=Qwen2.5-Base-1.5B2026.02 | 33.47 | 1.89 | |
| iCoTBase Model=Qwen2.5-Base-1.5B2026.02 | 28.84 | 1.59 |