Persuasion Evaluation on Anthropic
1.33Persuasion GainDeepSeek-R1
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeek-R1Setting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 1.33 | |
| Claude 3.7 SonnetSetting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 1.13 | |
| GPT-4oSetting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.73 | |
| Mistral-7B-Instruct-v0.3Setting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.6 | |
| Qwen2.5-7B-InstructSetting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.51 | |
| Llama-3.3-70B-InstructSetting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.49 | |
| Llama-3.1-8B-InstructSetting=Dynamic, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.44 | |
| DeepSeek-R1Setting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.29 | |
| Claude 3.7 SonnetSetting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.28 | |
| GPT-4oSetting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.15 | |
| Llama-3.1-8B-InstructSetting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.12 | |
| Mistral-7B-Instruct-v0.3Setting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.11 | |
| Qwen2.5-7B-InstructSetting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.08 | |
| Llama-3.3-70B-InstructSetting=Static, Receiver Model=Llama-3.1-8B-Instruct2025.09 | 0.08 |