Scientific Reasoning on SciFIBench
78.3AccuracyOctoTools
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| OctoToolsBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Scientific, Visual Understanding=true, Numerical Calculation=false, Knowledge Retrieval=false, Multi-step Reasoning=true2025.02 | 78.3 | 5.8 | 3.3 | |
| OctoToolsbaseBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Scientific, Visual Understanding=true, Numerical Calculation=false, Knowledge Retrieval=false, Multi-step Reasoning=true2025.02 | 77.3 | — | — | |
| CoTBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Scientific, Visual Understanding=true, Numerical Calculation=false, Knowledge Retrieval=false, Multi-step Reasoning=true2025.02 | 75 | — | — | |
| 0-shotBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Scientific, Visual Understanding=true, Numerical Calculation=false, Knowledge Retrieval=false, Multi-step Reasoning=true2025.02 | 72.5 | — | — |