Skip to content

Latest commit

 

History

History
31 lines (25 loc) · 1.12 KB

File metadata and controls

31 lines (25 loc) · 1.12 KB

lasigeBioTM at MedHop track : Can a Lean RAG-Enhanced Model Compete with MedGemma?

This repository contains the code used for our participation at the BioCreative IX Challenge and Workshop Track 1: MedHopQA

The paper describing our participation is available here.

Systems:

Official Runs

  • MedGemma Baseline: src/medgemma_baseline.py
  • Gemma Baseline: src/gemma_baseline.py
  • MistralWiki: src/wiki_pages.py
  • Mistral Onto: src/mistral_onto_definitions.py

Unofficial Runs

  • MedGemma Onto: src/medgemma_onto_definitions.py
  • MedGemma Wiki: src/medgemma_wiki.py

Project Structure

BioCreativeIX_Track1_MedHopQA/
├── data/		# Contains all data and pre-processed data
│   └── ...
├── NER/		# Named-entity recognition Scripts
├── src/		# Systems scripts
├── unofficial_data/	# Contains all data and pre-processed data used in the unnoficial runs
├── utils/		# Utility scripts and helper functions
└── wikidata/		# Scripts to retrieve info from wikipedia
│ 
├── medhop.yml		# Dependencies