Multimodal Transformer
A Multimodal Transformer extends the standard Transformer architecture to process and jointly reason over two or more input modalities — most commonly text and images, but also audio, video, or structured data. Cross-modal attention layers allow information from one modality to inform representations in another, enabling tasks such as visual question answering, image captioning, and multimodal sentiment analysis.
Registro de origem
Citações copiadas literalmente do registro de origem do método. Nenhuma verificação em nível de alegação é inferida delas.
- Lu, J., Batra, D., Parikh, D., & Lee, S. (2019). ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks. Advances in Neural Information Processing Systems (NeurIPS), 32. · URL
- Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., ... & Sutskever, I. (2021). Learning Transferable Visual Models From Natural Language Supervision. Proceedings of the 38th International Conference on Machine Learning (ICML), PMLR 139. · URL
Alegações curadas
Alegações persistidas no livro-razão de evidências, cada uma com sua própria avaliação.
Esta visualização não inventa uma avaliação de alegação quando o livro-razão não a possui.
Métodos relacionados
Gerado a partir do grafo de métodos e mostrado como relações sugeridas por máquina — nenhuma alegação de evidência é inferida.