Optical Character Recognition on TextOCR VL-ICL Bench
52.2Average AccuracyLLaVA-OneVision-72B
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-OneVision-72BVL-Data=10.4M, Params trained=73.2B, TTA=ICL2025.06 | 52.2 | |
| Qwen2-VL-7B-InstructVL-Data=-NA-, Params trained=8B, TTA=ICL2025.06 | 50.5 | |
| LLaVA-OneVision-7BVL-Data=10.4M, Params trained=8B, TTA=FT≤302025.06 | 45.5 | |
| LLaVA-OneVision-7BVL-Data=10.4M, Params trained=8B, TTA=ICL2025.06 | 42.3 | |
| Qwen2.5-VL-7B-InstructVL-Data=-NA-, Params trained=8B, TTA=FT≤302025.06 | 41.1 | |
| Qwen2.5-VL-7B-InstructVL-Data=-NA-, Params trained=8B, TTA=ICL2025.06 | 36.9 | |
| LAQ-7B + MAPDVL-Data=1.3M, Params trained=24M, TTA=FT≤302025.06 | 26.4 | |
| LAQ-7B + In-ContextPDVL-Data=1.3M, Params trained=24M, TTA=ICL2025.06 | 23.8 | |
| LAQ-7B + In-ContextPDVL-Data=1.3M, Params trained=24M, TTA=FT≤302025.06 | 18.9 | |
| LLaVA v1.5-7BVL-Data=1.2M, Params trained=7B, TTA=FT≤302025.06 | 15.6 | |
| LLaVA-Next-7BVL-Data=1.3M, Params trained=7.06B, TTA=FT≤302025.06 | 7.8 | |
| LAQ-7B + MAPDVL-Data=1.3M, Params trained=24M, TTA=ICL2025.06 | 7.3 | |
| LLaVA v1.5-7BVL-Data=1.2M, Params trained=7B, TTA=ICL2025.06 | 4.4 | |
| LLaVA-Next-7BVL-Data=1.3M, Params trained=7.06B, TTA=ICL2025.06 | 0.4 |