Text-based Task Completion on ScienceWorld
32.43Mean Normalised ScoreDORA
Evaluation Results
| Method | Links | |
|---|---|---|
| DORAModel=Mistral Small 22B2026.04 | 32.43 | |
| ReActModel=Mistral Small 22B2026.04 | 26.7 | |
| Zero-shotModel=Mistral Small 22B2026.04 | 26.47 | |
| Prompt ExploreModel=Mistral Small 22B2026.04 | 21.48 | |
| DORAModel=Qwen-2.5 7B2026.04 | 19.01 | |
| DORAModel=Llama-3.1 8B2026.04 | 16.85 | |
| ReActModel=Llama-3.1 8B2026.04 | 15.94 | |
| Zero-shotModel=Llama-3.1 8B2026.04 | 15.53 | |
| Prompt ExploreModel=Llama-3.1 8B2026.04 | 13.7 | |
| Zero-shotModel=Qwen-2.5 7B2026.04 | 13.49 | |
| Prompt ExploreModel=Qwen-2.5 7B2026.04 | 13.21 | |
| Tree of ThoughtModel=Mistral Small 22B2026.04 | 10.88 | |
| Tree of ThoughtModel=Qwen-2.5 7B2026.04 | 10.1 | |
| ReActModel=Qwen-2.5 7B2026.04 | 9.92 | |
| Chain of ThoughtModel=Qwen-2.5 7B2026.04 | 9.81 | |
| Chain of ThoughtModel=Mistral Small 22B2026.04 | 9.63 | |
| Tree of ThoughtModel=Llama-3.1 8B2026.04 | 8.64 | |
| Chain of ThoughtModel=Llama-3.1 8B2026.04 | 7.81 |