Emotional Negotiation on CRAD DeepSeek-V3 counterparty
100Success RateIQL+SFT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| IQL+SFTEvaluation=Zero-shot2026.05 | 100 | 73.3 | 73.3 | 7.3 | |
| IQL+SFT+JPOEvaluation=Zero-shot2026.05 | 100 | 81.5 | 81.5 | 9.3 | |
| IQLEvaluation=Zero-shot2026.05 | 90 | 58.2 | 52.4 | 12 | |
| VanillaEvaluation=Zero-shot2026.05 | 50 | 55.6 | 27.8 | 8.2 |