The official source code of EMNLP'21 paper "CSAGN: Conversational Structure Aware Graph Network for Conversational Semantic Role Labeling".
You can find the CSRL data at CSRL_dataset. Download the data and put them
into model_data
.
python train.py MODEL_NAME -batch_size 128 -bert_version BERT_MODEL -intra_loss -inter_loss
python evaluator.py TEST_DATASET MODEL_NAME