Blackjack on Blackjack 900 games
4,090Win RateReAct
Evaluation Results
| Method | Links | |
|---|---|---|
| ReActBased Model=GPT42024.02 | 4,090 | |
| ReflexionBased Model=GPT42024.02 | 4,080 | |
| Agent-Pro w/o LearningBased Model=GPT42024.02 | 4,060 | |
| Agent-ProBased Model=GPT42024.02 | 4,040 | |
| Agent-ProBased Model=GPT3.52024.02 | 3,820 | |
| Agent-Pro w/o LearningBased Model=GPT3.52024.02 | 3,740 | |
| ReflexionBased Model=GPT3.52024.02 | 3,670 | |
| ReActBased Model=GPT3.52024.02 | 3,660 | |
| Agent-ProBased Model=Qwen-72B2024.02 | 3,620 | |
| Agent-Pro w/o LearningBased Model=Qwen-72B2024.02 | 3,410 | |
| Vanilla LLMBased Model=GPT42024.02 | 3,400 | |
| ReflexionBased Model=Qwen-72B2024.02 | 3,230 | |
| ReActBased Model=Qwen-72B2024.02 | 3,090 | |
| Vanilla LLMBased Model=GPT3.52024.02 | 2,790 | |
| Agent-ProBased Model=Llama2-70B2024.02 | 2,310 | |
| ReflexionBased Model=Llama2-70B2024.02 | 1,210 | |
| ReActBased Model=Llama2-70B2024.02 | 1,180 | |
| Radical LLMBased Model=GPT42024.02 | 1,150 | |
| Agent-Pro w/o LearningBased Model=Llama2-70B2024.02 | 800 | |
| Radical LLMBased Model=GPT3.52024.02 | 180 | |
| Radical LLMBased Model=Qwen-72B2024.02 | 60 | |
| Vanilla LLMBased Model=Qwen-72B2024.02 | 50 | |
| Radical LLMBased Model=Llama2-70B2024.02 | 40 | |
| Vanilla LLMBased Model=Llama2-70B2024.02 | 30 |