WangResearchLab

/

Yarn-Llama-2-13b-128k

like 0

Follow

WangLab 6

This PR updates and expands the model card for the "Predicting Task Performance with Context-aware Scaling Laws" model.

Key improvements include:

• Adding pipeline_tag: text-generation to improve discoverability on the Hub.
• Specifying the license: apache-2.0, aligning with the base Llama-2 models.
• Including relevant tags like scaling-laws, long-context, and llama.
• Providing a comprehensive model description based on the paper's abstract.
• Linking directly to the paper (https://huggingface.co/papers/2510.14919) and the GitHub repository (https://github.com/wang-research-lab/context-scaling).
• Populating model details such as developers, type, language, and base models.
• Including a detailed "How to Get Started" section with installation and evaluation commands directly from the GitHub README.
• Filling in details about training data, procedure, and evaluation from the paper and repository.
• Adding the BibTeX citation.

Please review and merge this PR if everything looks good.