Project Details

Description

This project focuses on fine-tuning the Phi3 model, developed by Microsoft, to enhance its performance in medical question and answering tasks. The QLoRA (Quantized Low Rank Approximation) technique is employed to optimize the model for this specific use case. The fine-tuned model is then saved and made available on the HuggingFace hub.

Key Features

  • Model: Microsoft/Phi3
  • Techniques: Fine-tuning, Quantization, QLoRA
  • Dataset: MEDQuaD-dataset
  • Category: Generative AI

Keywords

  • Microsoft/Phi3
  • Fine-tuning
  • Quantization
  • QLoRA
  • MEDQuaD-dataset
  • Generative AI

Usage

You can access and use the fine-tuned model from the HuggingFace hub using the following link: AchrafABBAOUI/MedQuad_Phi3_QLoRA. For detailed instructions on how to utilize the model, please refer to the documentation provided in the repository.

Acknowledgements

This project leverages the latest advancements by Microsoft in model development and optimization techniques. Special thanks to the developers and contributors who made this project possible.