chano12

/

update_llama

like 0

This PR significantly enhances the model card for the SHARE model by populating it with detailed information from the associated paper, "SHARE: Shared Memory-Aware Open-Domain Long-Term Dialogue Dataset Constructed from Movie Script".

Key improvements include:

• A comprehensive model description, outlining its purpose and methodology.
• Links to the official paper (https://huggingface.co/papers/2410.20682) and the accompanying GitHub repository (https://github.com/share-dialogue/SHARE), providing users with direct access to research and code.
• A practical sample usage code snippet demonstrating how to load and use this PEFT adapter model with its meta-llama/Meta-Llama-3.1-8B-Instruct base model.
• Details regarding training data (SHARE dataset) and the EPISODE framework.
• Inclusion of crucial metadata tags for better discoverability on the Hub, such as:
  • pipeline_tag: text-generation
  • license: cc-by-4.0
  • datasets: - SHARE
  • tags: - long-form-dialogue - dialogue-generation

This update aims to provide a clear and informative resource for researchers and developers interested in long-term dialogue systems leveraging shared memories.