tanhuajie2001

/

Reason-RFT-Spatial-Transformation-Qwen2-VL-2B

like 0

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

• Updating the main title to accurately reflect that the repository contains model checkpoints from the paper "Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning of Vision Language Models".
• Adding the pipeline_tag: image-text-to-text metadata, which categorizes the model's functionality and improves discoverability on the Hugging Face Hub.
• Including library_name: transformers metadata, as evidenced by the config.json and tokenizer_config.json files showing compatibility with the transformers library's Qwen2VLForConditionalGeneration, Qwen2VLProcessor, and Qwen2Tokenizer classes. This enables the automated "How to use" widget on the Hub.
• Adding a prominent link to the official Hugging Face paper page: Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning of Vision Language Models. The existing ArXiv link has been preserved in the link cluster.
• Integrating a detailed "Quick Start Inference" code example directly from the project's GitHub repository, providing users with an immediate way to interact with the model. A clarifying note is added to the snippet regarding the original authors' choice of imports versus standard transformers usage for Hugging Face models.

These improvements aim to provide a more complete, discoverable, and accessible resource for the Reason-RFT models on the Hugging Face Hub.