malhajar/Mistral-7B-v0.2-meditron-turkish

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Jan 5, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

malhajar/Mistral-7B-v0.2-meditron-turkish is a 7 billion parameter Mistral-based model, fine-tuned by Mohamad Alhajar using the Freeze technique on a Turkish Meditron dataset. This model specializes in providing medical information in both Turkish and English, making it suitable for healthcare-related natural language processing tasks. It leverages the Mistral-7B-Instruct-v0.2 architecture, optimized for domain-specific medical queries.

Loading preview...

Overview

malhajar/Mistral-7B-v0.2-meditron-turkish is a 7 billion parameter language model developed by Mohamad Alhajar. It is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2, specifically adapted for medical contexts. The model was trained using the Freeze technique on the malhajar/meditron-tr dataset, which focuses on Turkish medical information.

Key Capabilities

  • Medical Information Retrieval: Excels at answering questions and providing information on various explicit medical ideas.
  • Bilingual Support: Capable of generating responses in both Turkish and English, catering to a broader user base in medical domains.
  • Domain-Specific Fine-tuning: Optimized for medical terminology and concepts through specialized training on a Meditron dataset.

Performance

Evaluated on the Open LLM Leaderboard, the model achieved an average score of 63.34. Notable benchmark results include:

  • HellaSwag (10-Shot): 81.79
  • Winogrande (5-shot): 76.24
  • TruthfulQA (0-shot): 66.19
  • MMLU (5-Shot): 60.35

Use Cases

This model is particularly well-suited for applications requiring accurate medical information in Turkish and English, such as:

  • Medical question-answering systems.
  • Healthcare chatbots.
  • Assisting medical professionals or students with information retrieval.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p