Graduate-level Factual Reasoning on GPQA (Pass@1, FLOPS)
34.9Pass@1MFS (Ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MFS (Ours)Backbone=LLaMA3.1-8B-Instruct2026.01 | 34.9 | — | |
| ϕ-DecodingBackbone=LLaMA3.1-8B-Instruct2026.01 | 34.6 | — | |
| Tree-of-ThoughtsBackbone=LLaMA3.1-8B-Instruct2026.01 | 31.25 | — | |
| Predictive DecodingBackbone=LLaMA3.1-8B-Instruct2026.01 | 31.03 | — | |
| Guided DecodingBackbone=LLaMA3.1-8B-Instruct2026.01 | 30.58 | — | |
| MFS (Ours)Backbone=Mistral-v0.3-7B-Instruct2026.01 | 30.58 | — | |
| ϕ-DecodingBackbone=Mistral-v0.3-7B-Instruct2026.01 | 29.24 | — | |
| Guided DecodingBackbone=Mistral-v0.3-7B-Instruct2026.01 | 27.46 | — | |
| Auto-Regressive (CoT)Backbone=LLaMA3.1-8B-Instruct2026.01 | 26.56 | — | |
| Tree-of-ThoughtsBackbone=Mistral-v0.3-7B-Instruct2026.01 | 26.34 | — | |
| MCTSBackbone=LLaMA3.1-8B-Instruct2026.01 | 24.11 | — | |
| Auto-Regressive (CoT)Backbone=Mistral-v0.3-7B-Instruct2026.01 | 23.88 | — | |
| MCTSBackbone=Mistral-v0.3-7B-Instruct2026.01 | 22.77 | — | |
| Predictive DecodingBackbone=Mistral-v0.3-7B-Instruct2026.01 | 22.1 | — |