方法对比

	多模态句子嵌入	CLIP
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2013–2021	2021
提出者	Frome et al. (DeViSE, 2013); popularized by Radford et al. (CLIP, 2021)	Radford, A.; Kim, J. W.; et al. (OpenAI)
类型	Representation learning model	Contrastive vision-language pretraining model
开创性文献	Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., ... & Sutskever, I. (2021). Learning transferable visual models from natural language supervision. In Proceedings of the 38th International Conference on Machine Learning (ICML), pp. 8748–8763. PMLR.	Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., & Sutskever, I. (2021). Learning Transferable Visual Models From Natural Language Supervision. Proceedings of the 38th International Conference on Machine Learning, PMLR 139, 8748–8763.
别名	multimodal embeddings, cross-modal sentence embeddings, vision-language embeddings, joint image-text embeddings	CLIP, Contrastive Language-Image Pre-training, zero-shot image classifier, visual-language model
相关	1	2
摘要	Multimodal sentence embeddings map text and images (and sometimes audio or video) into a shared continuous vector space, so that semantically related pairs from different modalities land close together. Trained by contrastive objectives on large paired corpora, these representations power cross-modal retrieval, zero-shot classification, and vision-language reasoning.	CLIP (Contrastive Language-Image Pretraining) is a vision-language model introduced by Radford et al. at OpenAI in 2021 that jointly learns aligned image and text representations by training on 400 million internet-sourced image-text pairs using a contrastive objective, enabling zero-shot transfer to image classification tasks without any task-specific fine-tuning.