---
tags:
- pytorch
- image-quality-assessment
- super-resolution
- hallucination-assessment
- arxiv:2610.25770
datasets:
- dhryou/VHA-bench
---

# VHA+

VHA+ is a lightweight hallucination assessor for ground-truth and super-resolved image pairs. It uses a shared DINOv3 ViT-B/16 encoder and predicts VHA scores from image-pair features. Higher scores indicate more hallucination relative to the reference.

- [Paper](https://arxiv.org/abs/2610.25770)
- [Code and installation](https://github.com/dhryougit/Hallucination-Assessment)
- [Project page](https://dhryougit.github.io/Hallucination-Assessment-pages/)
- [VHA-bench](https://huggingface.co/datasets/dhryou/VHA-bench)

## Download and evaluate

This repository provides the original step-20,000 checkpoint as `vhaplus.pt`, without optimizer state. Use it with the accompanying GitHub implementation; it is a custom PyTorch checkpoint.

After installing the GitHub repository and preparing VHA-bench under `data/benchmark/`:

```python
from huggingface_hub import hf_hub_download

checkpoint = hf_hub_download(
    repo_id="dhryou/VHA-plus",
    filename="vhaplus.pt",
    local_dir="data/checkpoints",
)
```

```bash
python scripts/vha+/evaluate.py \
  --checkpoint data/checkpoints/vhaplus.pt \
  --output outputs/vhaplus_evaluation
```

## Checkpoint details

- Shared DINOv3 ViT-B/16 encoder; the last four transformer blocks are trained.
- Final normalized CLS features and the checkpoint's stored reporting scale.
- Input resolution: 512 × 512.
- Training step: 20,000.
- The `config.json` records the model configuration and reporting scale; the GitHub implementation loads `vhaplus.pt` directly.

On VHA-bench, the released model achieves Spearman 0.659 and Pearson 0.675 with human hallucination judgments. These are the paper's fixed benchmark results, not a new evaluation performed during this upload.