Retrieval-Augmented Multimodal Language Modeling

Michihiro Yasunaga, Armen Aghajanyan, Weijia Shi, Richard James, Jure Leskovec, Percy Liang, Mike Lewis, Luke Zettlemoyer, Wen-Tau Yih
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:39755-39769, 2023.

Abstract

Recent multimodal models such as DALL-E and CM3 have achieved remarkable progress in text-to-image and image-to-text generation. However, these models store all their knowledge (e.g., the appearance of the Eiffel Tower) in the model parameters, requiring increasingly larger models and training data to capture more knowledge. To integrate knowledge in a more scalable and modular way, we propose a retrieval-augmented multimodal model, which enables a base multimodal model (generator) to refer to relevant text and images fetched by a retriever from external memory (e.g., documents on the web). Specifically, for the retriever, we use a pretrained CLIP, and for the generator, we train a CM3 Transformer on the LAION dataset. Our resulting model, named Retrieval-Augmented CM3 (RA-CM3), is the first multimodal model that can retrieve and generate both text and images. We show that RA-CM3 significantly outperforms baseline multimodal models such as DALL-E and CM3 on both image and caption generation tasks (12 FID and 17 CIDEr improvements on MS-COCO), while requiring much less compute for training ($<$30% of DALL-E). Moreover, we show that RA-CM3 exhibits novel capabilities such as faithful image generation and multimodal in-context learning (e.g., image generation from demonstrations).

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-yasunaga23a, title = {Retrieval-Augmented Multimodal Language Modeling}, author = {Yasunaga, Michihiro and Aghajanyan, Armen and Shi, Weijia and James, Richard and Leskovec, Jure and Liang, Percy and Lewis, Mike and Zettlemoyer, Luke and Yih, Wen-Tau}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {39755--39769}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/yasunaga23a/yasunaga23a.pdf}, url = {https://proceedings.mlr.press/v202/yasunaga23a.html}, abstract = {Recent multimodal models such as DALL-E and CM3 have achieved remarkable progress in text-to-image and image-to-text generation. However, these models store all their knowledge (e.g., the appearance of the Eiffel Tower) in the model parameters, requiring increasingly larger models and training data to capture more knowledge. To integrate knowledge in a more scalable and modular way, we propose a retrieval-augmented multimodal model, which enables a base multimodal model (generator) to refer to relevant text and images fetched by a retriever from external memory (e.g., documents on the web). Specifically, for the retriever, we use a pretrained CLIP, and for the generator, we train a CM3 Transformer on the LAION dataset. Our resulting model, named Retrieval-Augmented CM3 (RA-CM3), is the first multimodal model that can retrieve and generate both text and images. We show that RA-CM3 significantly outperforms baseline multimodal models such as DALL-E and CM3 on both image and caption generation tasks (12 FID and 17 CIDEr improvements on MS-COCO), while requiring much less compute for training ($<$30% of DALL-E). Moreover, we show that RA-CM3 exhibits novel capabilities such as faithful image generation and multimodal in-context learning (e.g., image generation from demonstrations).} }
Endnote
%0 Conference Paper %T Retrieval-Augmented Multimodal Language Modeling %A Michihiro Yasunaga %A Armen Aghajanyan %A Weijia Shi %A Richard James %A Jure Leskovec %A Percy Liang %A Mike Lewis %A Luke Zettlemoyer %A Wen-Tau Yih %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-yasunaga23a %I PMLR %P 39755--39769 %U https://proceedings.mlr.press/v202/yasunaga23a.html %V 202 %X Recent multimodal models such as DALL-E and CM3 have achieved remarkable progress in text-to-image and image-to-text generation. However, these models store all their knowledge (e.g., the appearance of the Eiffel Tower) in the model parameters, requiring increasingly larger models and training data to capture more knowledge. To integrate knowledge in a more scalable and modular way, we propose a retrieval-augmented multimodal model, which enables a base multimodal model (generator) to refer to relevant text and images fetched by a retriever from external memory (e.g., documents on the web). Specifically, for the retriever, we use a pretrained CLIP, and for the generator, we train a CM3 Transformer on the LAION dataset. Our resulting model, named Retrieval-Augmented CM3 (RA-CM3), is the first multimodal model that can retrieve and generate both text and images. We show that RA-CM3 significantly outperforms baseline multimodal models such as DALL-E and CM3 on both image and caption generation tasks (12 FID and 17 CIDEr improvements on MS-COCO), while requiring much less compute for training ($<$30% of DALL-E). Moreover, we show that RA-CM3 exhibits novel capabilities such as faithful image generation and multimodal in-context learning (e.g., image generation from demonstrations).
APA
Yasunaga, M., Aghajanyan, A., Shi, W., James, R., Leskovec, J., Liang, P., Lewis, M., Zettlemoyer, L. & Yih, W.. (2023). Retrieval-Augmented Multimodal Language Modeling. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:39755-39769 Available from https://proceedings.mlr.press/v202/yasunaga23a.html.

Related Material