SciFive HyDE Generator (Cardiology Fine Tuned)

Model Description

This model is a fine-tuned version of razent/SciFive-base-Pubmed_PMC for Hypothetical Document Generation (HyDE) in biomedical Retrieval-Augmented Generation (RAG) systems.

The model generates domain-specific hypothetical documents from user queries to improve dense retrieval performance and reduce hallucinations.

Training Data

  • Dataset: MIRIAD medical QA dataset
  • Specialties:
    • Cardiology
    • Cardiothoracic Surgery
    • Vascular Surgery
  • ~15,000 question鈥揳nswer pairs

Training Objective

Sequence-to-sequence generation:

Downloads last month
3
Safetensors
Model size
0.2B params
Tensor type
F32
Inference Providers NEW
This model isn't deployed by any Inference Provider. 馃檵 Ask for provider support