Conflict-resolution quality evaluation on Conflict-resolution tasks
1.6Negotiation RoundsMulti-Agent
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Multi-AgentDecoding strategy=greedy, Comparison Opponent=Base2026.03 | 1.6 | 57.1 | |
| Multi-AgentDecoding strategy=greedy, Comparison Opponent=Single-Agent2026.03 | 1.6 | 67.7 | |
| Multi-AgentDecoding strategy=sampling, Comparison Opponent=Base2026.03 | 1.64 | 63 | |
| Multi-AgentDecoding strategy=sampling, Comparison Opponent=Single-Agent2026.03 | 1.64 | 72.8 | |
| Single-AgentDecoding strategy=sampling, Comparison Opponent=Base2026.03 | 1.67 | 41.8 | |
| Single-AgentDecoding strategy=sampling, Comparison Opponent=Multi-Agent2026.03 | 1.67 | — | |
| Single-AgentDecoding strategy=greedy, Comparison Opponent=Base2026.03 | 1.73 | 39.8 | |
| Single-AgentDecoding strategy=greedy, Comparison Opponent=Multi-Agent2026.03 | 1.73 | — | |
| BaseDecoding strategy=sampling2026.03 | 2.07 | — | |
| BaseDecoding strategy=greedy2026.03 | 2.13 | — |