Interactive Reasoning on ScienceWorld (Seen)
63.91Success RateITPR
Evaluation Results
| Method | Links | |
|---|---|---|
| ITPRBackbone=Llama3.1-8B, Type=Training2026.01 | 63.91 | |
| ITPRBackbone=Qwen2.5-7B, Type=Training2026.01 | 62.58 | |
| ITPRBackbone=Qwen3-8B, Type=Training2026.01 | 61.85 | |
| AECBackbone=Llama3-8B2026.02 | 61.36 | |
| WKMBackbone=Llama3.1-8B, Type=Training2026.01 | 61.34 | |
| IWMBackbone=Qwen2.5-7B, Type=Training2026.01 | 60.82 | |
| WKMBackbone=Qwen3-8B, Type=Training2026.01 | 60.31 | |
| WKMBackbone=Llama3-8B2026.02 | 60.12 | |
| IWMBackbone=Qwen3-8B, Type=Training2026.01 | 59.27 | |
| KnowAgentBackbone=Llama3-8B2026.02 | 58.67 | |
| ETOBackbone=Llama3-8B2026.02 | 57.9 | |
| IWMBackbone=Llama3.1-8B, Type=Training2026.01 | 57.56 | |
| SFTBackbone=Llama3.1-8B, Type=Training2026.01 | 57.21 | |
| SFTBackbone=Qwen3-8B, Type=Training2026.01 | 56.7 | |
| SFTBackbone=Qwen2.5-7B, Type=Training2026.01 | 55.67 | |
| NATBackbone=Llama3-8B2026.02 | 55.24 | |
| WKMBackbone=Qwen2.5-7B, Type=Training2026.01 | 54.12 | |
| ReflexionBackbone=Llama3-8B2026.02 | 27.23 | |
| ReActBackbone=Llama3-8B2026.02 | 24.76 | |
| ITPIBackbone=Qwen3-8B, Type=Prompting2026.01 | 20.61 | |
| ITPIBackbone=Llama3.1-8B, Type=Prompting2026.01 | 19.58 | |
| ITPIBackbone=Qwen2.5-7B, Type=Prompting2026.01 | 16.49 | |
| RAPBackbone=Qwen3-8B, Type=Prompting2026.01 | 15.46 | |
| RAPBackbone=Llama3.1-8B, Type=Prompting2026.01 | 11.34 | |
| RAPBackbone=Qwen2.5-7B, Type=Prompting2026.01 | 10.3 | |
| ReActBackbone=Qwen3-8B, Type=Prompting2026.01 | 9.79 | |
| ReActBackbone=Llama3.1-8B, Type=Prompting2026.01 | 9.27 | |
| ReActBackbone=Qwen2.5-7B, Type=Prompting2026.01 | 8.24 | |
| CoTBackbone=Qwen2.5-7B, Type=Prompting2026.01 | 3.09 | |
| CoTBackbone=Llama3.1-8B, Type=Prompting2026.01 | 3.09 | |
| CoTBackbone=Qwen3-8B, Type=Prompting2026.01 | 2.44 |