README.md

Question Answering on PubMedQA in Reasoning Required Setting

Data

Download data from PubMedQA and following the steps of splitting dataset.

Copy the files pqal_fold0/train_set.json, pqal_fold0/dev_set.json, test_set.json and test_ground_truth.json to ../../data/PubMedQA/raw

Then, you can process the data by:

bash preprocess.sh # for BioGPT

or

bash preprocess_large.sh # for BioGPT-Large

Model Checkpoint

We provide our fine-tuned model on the task. See here

Inference and Evaluating

You can inference and evaluate the model on the test set by:

bash infer.sh # for BioGPT

or

bash infer_large.sh # for BioGPT-Large