Hierarchical Crafting on TextCraft
88.9ScoreBest-of-15 (ReAct)
Evaluation Results
| Method | Links | |
|---|---|---|
| Best-of-15 (ReAct)Policy Model=GPT-5.12026.06 | 88.9 | |
| AMC (ReAct)Policy Model=GPT-4.1-mini, Value Model=Llama-3.2-11B2026.06 | 85.2 | |
| AMC (ReAct)Policy Model=GPT-5.1, Value Model=Llama-3.2-11B2026.06 | 79 | |
| Best-of-15 (ReAct)Policy Model=GPT-4.1-mini2026.06 | 72.8 | |
| ReActPolicy Model=GPT-5.12026.06 | 69.1 | |
| AMC (ReAct)Policy Model=Llama-3.2-11B, Value Model=Llama-3.2-11B2026.06 | 54.3 | |
| ReActPolicy Model=GPT-4.1-mini2026.06 | 43.2 | |
| Best-of-15 (ReAct)Policy Model=Llama-3.2-11B2026.06 | 29.6 | |
| ReActPolicy Model=Llama-3.2-11B2026.06 | 10.2 |