tanhuajie2001

/

Reason-RFT-Visual-Counting-Qwen2-VL-7B

like 1

This PR significantly enhances the model card for the Reason-RFT model by:

• Updating the title: The current model card title # 🤗 Reason-RFT CoT Dateset is misleading as the repository contains model checkpoints. This PR updates the main heading to # Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning of Vision Language Models, aligning it with the paper title and the GitHub repository's main heading. A clarifying sentence is added below the title.
• Adding pipeline_tag: image-text-to-text: This tag is added to the metadata to accurately categorize the model's functionality (visual reasoning with VLMs) and improve its discoverability on the Hugging Face Hub.
• Adding library_name: transformers: Evidence from config.json (e.g., Qwen2VLForConditionalGeneration, transformers_version) and tokenizer_config.json (e.g., Qwen2VLProcessor, Qwen2Tokenizer) indicates compatibility with the Hugging Face Transformers library. This tag will enable automated usage snippets on the model page.
• Updating paper links: The links to the paper within the model card content have been updated from the arXiv URL to the official Hugging Face Papers URL (https://huggingface.co/papers/2503.20752) for better integration within the Hub.

No sample usage code has been added, as none was explicitly provided in the GitHub README for direct inference, adhering to the guidelines.