Instruction Hierarchy Adherence on IHEval Conflict Single-turn
52.13AccuracyIHDec
Evaluation Results
| Method | Links | |
|---|---|---|
| IHDecModel=Llama-3.1-8B-Instruct2026.06 | 52.13 | |
| IHDec + PromptingModel=Qwen3-4B-Instruct2026.06 | 51.1 | |
| IHDec + ISEModel=Llama-3.1-8B-Instruct2026.06 | 50.4 | |
| IHDec + VerIHModel=Qwen3-8B-Instruct2026.06 | 49.3 | |
| IHDec + PromptingModel=Qwen3-8B-Instruct2026.06 | 49.1 | |
| PromptingModel=Qwen3-8B-Instruct2026.06 | 48.9 | |
| VerIHModel=Qwen3-8B-Instruct2026.06 | 47.56 | |
| IHDecModel=Qwen3-4B-Instruct2026.06 | 45.7 | |
| IHDecModel=Qwen3-8B-Instruct2026.06 | 43.8 | |
| PromptingModel=Qwen3-4B-Instruct2026.06 | 39.7 | |
| IHDec + PromptingModel=Llama-3.1-8B-Instruct2026.06 | 37.9 | |
| VanillaModel=Qwen3-4B-Instruct2026.06 | 31.3 | |
| VanillaModel=Qwen3-8B-Instruct2026.06 | 27.7 | |
| ISEModel=Llama-3.1-8B-Instruct2026.06 | 26.83 | |
| PromptingModel=Llama-3.1-8B-Instruct2026.06 | 15.7 | |
| VanillaModel=Llama-3.1-8B-Instruct2026.06 | 15.13 |