VisionXLab

/

FIRM-Edit-8B

like 0

Follow

SJTU VisionXLab 12

Hi! I'm Niels, part of the community science team at Hugging Face.

I've updated your model card to document this artifact as FIRM-Edit-8B, the robust reward model for image editing introduced in your recent paper.

Changes include:

• Added pipeline_tag: image-text-to-text to the metadata for better discoverability.
• Linked the model to the ArXiv paper, project page, and GitHub repository.
• Added a model description explaining its role as a critic in the FIRM framework.
• Included the citation for your research.
• Retained your original training logs and hyperparameter information.

This will help researchers and practitioners find and use your reward model for reinforcement learning and evaluation in image editing.