Reverb/MedLLaMA-3

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:May 27, 2024License:cc-by-nc-nd-4.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

MedLLaMA-3 is an 8 billion parameter language model developed by Basel Anaya, designed for medical question answering and related tasks. It demonstrates strong performance across various medical benchmarks, including MMLU medical subtasks and MedQA. This model is optimized for accuracy in medical contexts, making it suitable for applications requiring specialized healthcare knowledge.

Loading preview...

MedLLaMA-3 Overview

MedLLaMA-3 is an 8 billion parameter language model developed by Basel Anaya, specifically tailored for medical applications. It is built to address complex medical questions and provide accurate information within healthcare contexts. The model's architecture and training focus on achieving high performance on specialized medical benchmarks.

Key Capabilities

  • Medical Question Answering: Excels at answering questions across various medical domains.
  • Benchmark Performance: Achieves notable scores on medical evaluation tasks, including:
    • medmcqa: 0.6118 accuracy
    • medqa_4options: 0.6143 accuracy
    • MMLU medical subtasks:
      • anatomy: 0.7185 accuracy
      • clinical_knowledge: 0.7811 accuracy
      • college_biology: 0.8264 accuracy
      • college_medicine: 0.7110 accuracy
      • medical_genetics: 0.8300 accuracy
      • professional_medicine: 0.7868 accuracy
    • pubmedqa: 0.7420 accuracy

Good For

  • Healthcare AI Applications: Ideal for integrating into systems that require robust medical knowledge.
  • Research and Development: Useful for researchers exploring specialized language models in the medical field.
  • Educational Tools: Can support the development of tools for medical students and professionals.

Finetunes
1 models