README.md

metadata

language:
  - en
tags:
  - phi-3
  - finetuned
  - question-optimization
license: mit

Fine-tuned Phi-3 Model for Question Optimization

This is a fine-tuned version of Microsoft's Phi-3 model, trained to improve scientific questions by making them more precise and focused while preserving their core intent.

Training Details

• Base model: microsoft/Phi-3-mini-128k-instruct
• Training method: LoRA fine-tuning
• Task: Question optimization and refinement
• Training data: Scientific questions dataset from BioASQ