m-a-p

/

CriticLeanGPT-Qwen3-8B-RL

like 3

Follow

Multimodal Art Projection 855

This PR significantly updates the model card for CriticLeanGPT-Qwen3-8B-RL. The previous README incorrectly described the repository as a dataset, while the repository actually contains a pre-trained model.

This update includes:

• A clear description of the CriticLeanGPT-Qwen3-8B-RL model, based on the paper CriticLean: Critic-Guided Reinforcement Learning for Mathematical Formalization.
• Addition of essential metadata: pipeline_tag: text-generation, library_name: transformers, and license: apache-2.0, improving model discoverability and usability on the Hugging Face Hub.
• A practical sample usage example using the transformers library, demonstrating how to use the model for both mathematical formalization and its critic role.
• Links to the official GitHub repository and related datasets (CriticLeanInstruct, CriticLeanBench, FineLeanCorpus) for comprehensive context.