This repository is used to store source code for paper "SIAST: A Slot Imbalance-Aware Self-Training Scheme for Semi-Supervised Slot Filling" accepted by ICASSP2023. Welcome to cite our paper when using the model.
-
modify the config file, like
configs/by_count/classicSelfTraining_snips_by_count_10.json
-
model train
python train_main.py /data1/xcc/Projects/SIAST/configs/by_count/classicSelfTraining_snips_by_count_10.json