CoolFace
Modelpublic

zjudai/flowertune-medical-lora-mistral-7b-instruct-v0.3

sourceHugging Faceupdated 1y agoView on Hugging Face
0likes151downloads
Model Card

FlowerTune Medical LoRA Model

This is a LoRA adapter for mistralai/Mistral-7B-Instruct-v0.3 fine-tuned with Flower federated learning framework on medical NLP datasets.

Model Performance

  • —PubMedQA: 0.5440
  • —MedMCQA: 0.5520
  • —MedQA: 0.5440
  • —CareQA: 0.4980
  • —Average: 0.5345

Training Details

  • —Domain: Medical
  • —Training method: Federated LoRA fine-tuning with FlowerTune
  • —Framework: Flower

This model is a LoRA adapter fine-tuned on mistralai/Mistral-7B-Instruct-v0.3 using the Flower federated learning framework. It was trained on medical NLP datasets through distributed learning. The model achieved scores of PubMedQA: 0.5440, MedMCQA: 0.5520, MedQA: 0.5440, CareQA: 0.4980, with an average score of 0.5345.

Links