On the Use of Modality-Specific Large-Scale Pre-Trained Encoders for Multimodal Sentiment Analysis

Open in new window