Success Rate on ACRE
93Success RateGPT-4o (Standard One-Stage)
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o (Standard One-Stage)Pipeline=One-stage, Model=GPT-4o2025.12 | 93 | |
| Two-stage pipeline (GPT-4o + LLaVA-1.5)Pipeline=Two-stage, Perception=GPT-4o, Reasoning=LLaVA-1.52025.12 | 82.5 | |
| Two-stage pipeline (LLaVA-1.5 + LLaVA-1.5)Pipeline=Two-stage, Perception=LLaVA-1.5, Reasoning=LLaVA-1.52025.12 | 34.5 | |
| LLaVA-1.5 (Standard One-Stage)Pipeline=One-stage, Model=LLaVA-1.52025.12 | 22 |