Visual Question Answering on TableVQA-Bench v1 (test)
82VTabFactQwen2.5-VL-7B-Instruct
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen2.5-VL-7B-Instruct2025.09 | 82 | 96.4 | 68.53 | 74 | |
| GPT-4o2025.09 | 78 | 96.8 | 72.8 | 82.4 | |
| LLaVA-Next-34B2025.09 | 71.2 | — | 36.4 | 38 | |
| Qwen2.5-VL-7B-InstructFine-tuned on=Visual-TableQA2025.09 | 70.6 | 97.2 | 61.5 | 69.6 | |
| LLaVA-Next-Llama3-8BFine-tuned on=Visual-TableQA2025.09 | 52 | 56.8 | 33.2 | 33.6 | |
| LLaVA-Next-Llama3-8B2025.09 | 37.2 | 52.4 | 21.5 | 24.8 |