Hi 👋 this PR adds this model's **[VDEBench](https://huggingface.co/datasets/hongzhuyi/VDEBench)** scores to `.eval_results/`, so they show up on the model page and on the VDEBench leaderboard.
VDEBench evaluates instruction-guided editing of dense English/Chinese document pages (3,506 examples). Each score is the equally weighted mean of IoU, CDM, BLEU-4 and TEDS (0–1, higher is better):
| Task | Score |
|---|---:|
| `local` (edited region, headline) | 0.685 |
| `global` (full page) | 0.673 |
Scores come from the VDEBench paper ([arXiv:2602.00122v4](https://arxiv.org/abs/2602.00122v4), Table 2), and the VDEBench authors confirmed both the values and the evaluated checkpoint ([discussion](https://huggingface.co/datasets/hongzhuyi/VDEBench/discussions/2)).
While this PR is open the scores appear as community-provided; merging it makes them official, and closing it removes them. Docs: https://huggingface.co/docs/hub/eval-results · https://huggingface.co/docs/hub/leaderboard-data-guid…
