MSAF : Multimodal Sentiment Detection via Multi-Scale Adaptive Fusion

This is the repository of MSAF : Multimodal Sentiment Detection via Multi-Scale Adaptive Fusion

Requirements:

numpy==1.24.3
scikit_learn==1.3.2
pytorch==2.0.0

Download:

MVSA: https://mcrlab.net/research/mvsa-sentiment-analysis-on-multi-view-social-data/

HFM: https://github.com/headacheboy/data-of-multimodal-sarcasm-detection

RU-Senti: https://github.com/PhenoixYANG/TOM

Checkpoint of BERT-base: https://huggingface.co/google-bert/bert-base-uncased

Checkpoint of swin_transformer: https://github.com/microsoft/Swin-Transformer

Multimodal Knowledge Graph Completion:

train on MVSA-Single

sh train_single.sh 

train on MVSA-Multiple

sh train_mul.sh 

train on HFM

sh train_hfm.sh 

train on RU-Senti

sh train_ru.sh