AI4BSLR's picture
Update README.md
b57d428 verified
|
raw
history blame
1.49 kB
metadata
tags:
  - biogpt
  - boolean-query
  - biomedical
  - systematic-review
  - pubmed
license: unknown
model-index:
  - name: BioGPT-BQF-TMK-Small
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: CLEF TAR
          type: biomedical
        metrics:
          - name: Precision @100
            type: precision
            value: 0.134
          - name: Recall @1000
            type: recall
            value: 0.2125

BioGPT-BQF-TMK-Small

A fine-tuned BioGPT model for Boolean query formalization in biomedical systematic reviews, incorporating Titles, MeSH Terms, and Keywords to improve PubMed search query generation.

Model Overview

  • Base Model: BioGPT
  • Fine-tuned on: Semi-synthetic generated data
  • Task: Boolean Query Generation for PubMed searches
  • Inputs: Research topic title, MeSH terms, and Keywords
  • Outputs: Optimized PubMed Boolean search query

Usage

from transformers import BioGptForCausalLM, BioGptTokenizer

model_name = "AI4BSLR/BioGPT-BQF-TMK-Small"
model = BioGptForCausalLM.from_pretrained(model_name)
tokenizer = BioGptTokenizer.from_pretrained(model_name)

input_text = "Title: Heterogeneity in Lung Cancer, MeSH: Biomarkers, Tumor, Genetic Heterogeneity, Keywords: Biomarkers, Query: "
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))