Ultimatum Negotiation on NegotiationArena Ultimatum
35.71Self ProfitAMPO
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AMPOBackbone=Qwen2.5-7B-Instruct2025.05 | 35.71 | 79.23 | 28.4 | |
| AMPOBackbone=Llama3.1-8B-Instruct2025.05 | 34.91 | 76.42 | 20.5 | |
| GRPOBackbone=Qwen2.5-7B-Instruct2025.05 | 34.65 | 76.88 | 25.1 | |
| GRPOBackbone=Llama3.1-8B-Instruct2025.05 | 31.64 | 69.68 | 18.4 | |
| Llama3.1-8B-InstructTraining Mode=Vanilla2025.05 | 14.43 | 22.41 | 15.9 | |
| Qwen2.5-7B-InstructTraining Mode=Vanilla2025.05 | 14.23 | 32.68 | 8.1 |