Multi-agent Debate on MMLU-Pro Debate
65.7Success Rate (SR)Concise
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ConciseListener=Llama-70B2026.04 | 65.7 | 847 | |
| HANDRAISERListener=Llama-70B2026.04 | 65.7 | 806 | |
| PromptingListener=Llama-70B2026.04 | 65.3 | 1,707 | |
| GenericListener=Llama-70B2026.04 | 64.7 | 1,617 | |
| RandomListener=Llama-70B2026.04 | 62.3 | 1,617 | |
| HANDRAISERListener=Llama-8B2026.04 | 58.3 | 782 | |
| RandomListener=Llama-8B2026.04 | 53.7 | 1,531 | |
| PromptingListener=Llama-8B2026.04 | 53.7 | 1,690 | |
| ConciseListener=Llama-8B2026.04 | 52.3 | 829 | |
| GenericListener=Llama-8B2026.04 | 52 | 1,531 |