SCCL
This repo contains our pytorch code for IEEE TAC accepted paper: "Cluster-Level Contrastive Learning for Emotion Recognition in Conversations". The architecture for our model is as follows:
Preparation:
-
Set up the Python 3.7 environment, and build the dependencies with the following code: pip install -r requirements.txt
-
Download the data from https://drive.google.com/file/d/1b_ihQYKTAsO67I5LULMbMFBrDgat8bQN/view?usp=sharing.
-
Download the released pre-trained adapter model from the K-Adapter paper(https://arxiv.org/abs/2002.01808): https://github.com/microsoft/k-adapter and put the directory "fac-adapter" and "lin-adapter" under the directory ./SCCL/pretrained_models/.
Training:
You can train the model with the following codes:
Run on IEMOCAP with RoBERTa-Large: python main.py --DATASET IEMOCAP --CUDA True --model_checkpoint roberta-large --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on IEMOCAP with FacAdapter: python main.py --DATASET IEMOCAP --CUDA True --model_checkpoint roberta-facadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on IEMOCAP with LinAdapter: python main.py --DATASET IEMOCAP --CUDA True --model_checkpoint roberta-linadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on MELD with LinAdapter: python main.py --DATASET MELD --CUDA True --model_checkpoint roberta-linadapter --alpha 1.0 --NUM_TRAIN_EPOCHS 3 --BATCH_SIZE 4
Run on MELD with FacAdapter: python main.py --DATASET MELD --CUDA True --model_checkpoint roberta-facadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 3 --BATCH_SIZE 4
Run on MELD with RoBERTa-Large: python main.py --DATASET MELD --CUDA True --model_checkpoint roberta-large --alpha 1.0 --NUM_TRAIN_EPOCHS 3 --BATCH_SIZE 4
Run on EmoryNLP with RoBERTa-Large: python main.py --DATASET EmoryNLP --CUDA True --model_checkpoint roberta-large --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on EmoryNLP with FacAdapter: python main.py --DATASET EmoryNLP --CUDA True --model_checkpoint roberta-facadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on EmoryNLP with LinAdapter: python main.py --DATASET EmoryNLP --CUDA True --model_checkpoint roberta-linadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 4
Run on DailyDialog with RoBERTa-Large: python main.py --DATASET DailyDialog --CUDA True --model_checkpoint roberta-large --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 16
Run on DailyDialog with FacAdapter: python main.py --DATASET DailyDialog --CUDA True --model_checkpoint roberta-facadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 16
Run on DailyDialog with LinAdapter: python main.py --DATASET DailyDialog --CUDA True --model_checkpoint roberta-linadapter --alpha 0.8 --NUM_TRAIN_EPOCHS 5 --BATCH_SIZE 16
Citation:
Please cite our paper as follows:
@ARTICLE{10040720, author={Yang, Kailai and Zhang, Tianlin and Alhuzali, Hassan and Ananiadou, Sophia}, journal={IEEE Transactions on Affective Computing}, title={Cluster-Level Contrastive Learning for Emotion Recognition in Conversations}, year={2023}, volume={}, number={}, pages={1-12}, doi={10.1109/TAFFC.2023.3243463}}