Visual Mathematical Reasoning on CLEVR-Math
79AccuracyOctoTools
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| OctoToolsBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Mathematical, Visual Understanding=true, Numerical Calculation=true, Knowledge Retrieval=false, Multi-step Reasoning=false2025.02 | 79 | 14.5 | 3.8 | |
| CoTBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Mathematical, Visual Understanding=true, Numerical Calculation=true, Knowledge Retrieval=false, Multi-step Reasoning=false2025.02 | 75.2 | — | — | |
| OctoToolsbaseBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Mathematical, Visual Understanding=true, Numerical Calculation=true, Knowledge Retrieval=false, Multi-step Reasoning=false2025.02 | 68.8 | — | — | |
| 0-shotBackbone=gpt-4o-2024-08-06, Modality=Vision, Domain=Mathematical, Visual Understanding=true, Numerical Calculation=true, Knowledge Retrieval=false, Multi-step Reasoning=false2025.02 | 64.5 | — | — |