Multitask Learning for Biomedical Named Entity Recognition with Cross-Sharing Structure

This repository is a Biomedical Named Entity Recognition model. The code is based on XuezheMax/NeuroNLP2.

Requirements

Python 3+, PyTorch < 1.0, Gensim >= 0.12.0

Running the experiments

In the root of the repository, first make the tmp directory:

mkdir tmp

To train a Baseline Single-task Model (STM) model,

./run_ner_crf.sh

To train a Fully-shared Multi-task Model (FS-MTM) model,

./run_fullyshare.sh

To train a Shared-private Multi-task Model (SP-MTM) model, (Specify adv_loss_coef = 0 and diff_loss_coef = 0 in config file.)

./run_adversarial.sh 

To train a Adversarial Multi-task Model (ADV-MTM) model,

./run_adversarial.sh

To train a Cross-sharing Multi-task Model (CS-MTM) model,

./run_crossshare.sh

To make a grid search,

python3 grid_search.py