Reasoning on GPQA (Accuracy (Multi-choice))
32.89Accuracy (Multi-choice)Full (100%)
Evaluation Results
| Method | Links | |
|---|---|---|
| Full (100%)Backbone=Qwen2.5-7B2026.05 | 32.89 | |
| TMAPBackbone=Qwen2.5-7B2026.05 | 32.74 | |
| Data-WhispererBackbone=Qwen2.5-7B2026.05 | 32.55 | |
| BaseBackbone=Qwen2.5-7B2026.05 | 32.39 | |
| IFDBackbone=Qwen2.5-7B2026.05 | 32.38 | |
| AlpagasusBackbone=Qwen2.5-7B2026.05 | 32.21 | |
| DEITABackbone=Qwen2.5-7B2026.05 | 31.88 | |
| LongestBackbone=Qwen2.5-7B2026.05 | 31.71 | |
| RandomBackbone=Qwen2.5-7B2026.05 | 31.55 | |
| Full (100%)Backbone=Llama3-8B2026.05 | 31.46 | |
| BaseBackbone=Llama3-8B2026.05 | 31.29 | |
| TMAPBackbone=Llama3-8B2026.05 | 30.73 | |
| DEITABackbone=Llama3-8B2026.05 | 30.63 | |
| Data-WhispererBackbone=Llama3-8B2026.05 | 30.2 | |
| IFDBackbone=Llama3-8B2026.05 | 30.17 | |
| RandomBackbone=Llama3-8B2026.05 | 30.12 | |
| AlpagasusBackbone=Llama3-8B2026.05 | 30.08 | |
| LongestBackbone=Llama3-8B2026.05 | 30.03 | |
| BaseBackbone=Mistral-v0.2-7B2026.05 | 28.86 | |
| Full (100%)Backbone=Mistral-v0.2-7B2026.05 | 28.78 | |
| TMAPBackbone=Mistral-v0.2-7B2026.05 | 27.94 | |
| Data-WhispererBackbone=Mistral-v0.2-7B2026.05 | 27.77 | |
| RandomBackbone=Mistral-v0.2-7B2026.05 | 27.68 | |
| LongestBackbone=Mistral-v0.2-7B2026.05 | 27.68 | |
| IFDBackbone=Mistral-v0.2-7B2026.05 | 27.52 | |
| AlpagasusBackbone=Mistral-v0.2-7B2026.05 | 27.36 | |
| DEITABackbone=Mistral-v0.2-7B2026.05 | 27.18 |