llm-jp

/

optimal-sparsity-math-d2048-E32-k16-13.6B-A7.1B

like 0

Follow

LLM-jp 364

This PR significantly improves the model card for the model by adding crucial information and metadata.

Key changes include:

• Metadata:
  • Added license: apache-2.0 (as identified for the project).
  • Added library_name: transformers based on config.json (MixtralForCausalLM architecture and transformers_version), enabling the "Use in Transformers" widget.
  • Added pipeline_tag: text-generation to categorize the model for language generation tasks.
  • Included relevant tags: mixture-of-experts, moe, llm, and reasoning for better discoverability.
• Content:
  • Added the paper's title as the main heading.
  • Included prominent links to the paper (Optimal Sparsity of Mixture-of-Experts Language Models for Reasoning Tasks) and the GitHub repository (https://github.com/rioyokotalab/optimal-sparsity).
  • Incorporated the full abstract to provide detailed context about the model and its research.
  • Preserved the existing citation information.

These updates make the model more informative and accessible to users on the Hugging Face Hub.