EPFLiGHT/Meditron3-8B

TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jul 11, 2024License:llama3.1Architecture:Transformer0.0K Gated Featherless Exclusive Cold

EPFLiGHT/Meditron3-8B is an 8 billion parameter large language model, part of the Meditron suite, specialized in clinical medicine. Developed by LiGHT, it is based on the Llama-3.1 architecture and trained on a diverse mixture of expert-curated medical data, including clinical guidelines and peer-reviewed publications. This model is designed for research purposes to evaluate LLMs in enhancing clinical decision-making and access to evidence-based medical information, particularly emphasizing equitable representation and contextual diversity.

Loading preview...

Meditron3-8B: A Clinical Medicine LLM for Research

EPFLiGHT/Meditron3-8B is an 8 billion parameter model from the Meditron suite, developed by LiGHT and built upon the Llama-3.1 base architecture. This model is specifically specialized in clinical medicine, with a strong emphasis on equitable representation, contextual diversity, and actionable real-world evidence-based guidelines, including those relevant to limited-resource and humanitarian settings.

Key Characteristics & Training:

  • Specialized Domain: Focused on clinical medicine, co-designed with expert clinicians and humanitarian practitioners.
  • Diverse Training Data: Trained on a unique mixture of expert-curated public data, including:
    • International clinical practice guidelines.
    • Full-text peer-reviewed medical publications.
    • Synthetic conversation data for differential diagnoses.
    • LLM-enhanced medical multiple-choice questions.
  • Research-Only Focus: Intended solely for research to study and evaluate LLMs in clinical decision-making; not validated for direct medical use.
  • Foundation Model: Released as a foundation model, not instruction-tuned, but adaptable for downstream tasks like Q&A through in-context learning or fine-tuning.

Important Considerations:

  • Disclaimer: This model is provided "as is" and is not intended for clinical decision-making, diagnosis, or treatment without appropriate validation and regulatory approval. Users must exercise professional judgment.
  • Evaluation: Evaluation results for this specific 8B model are pending. The Meditron suite is evaluated on medical multiple-choice questions, but developers acknowledge the need to go beyond MCQs to assess real-world utility, empathy, and alignment with local guidelines.