Papers › Retrieval-Augmented Multimodal Language Modeling

Retrieval-Augmented Multimodal Language Modeling

22 Nov 2022arXiv:2211.12561archive 2025-07-28

Michihiro Yasunaga, Armen Aghajanyan, Weijia Shi, Rich James, Jure Leskovec, Percy Liang, Mike Lewis, Luke Zettlemoyer, Wen-tau Yih

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 learned 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).

PaperPDF

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Caption GenerationImage CaptioningImage GenerationImage to textIn-Context LearningLanguage ModelingLanguage ModellingRetrievalText GenerationText-to-Image Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Captioning COCO (Common Objects in Context) Flamingo (80B; 4-shot) CIDEr 103 #8 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) RA-CM3 (2.7B) CIDEr 89.1 #9 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) Flamingo (3B; 4-shot) CIDEr 85 #10 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) Parti CIDEr 83.9 #11 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) Vanilla CM3 CIDEr 71.9 #13 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) X-LXMERT CIDEr 55.8 #14 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) minDALL-E CIDEr 48 #15 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) ruDALL-E-XL CIDEr 38.7 #16 of 17 Archive leaderboard report
Image Captioning COCO (Common Objects in Context) DALL-E CIDEr 20.2 #17 of 17 Archive leaderboard report
Text-to-Image Generation COCO (Common Objects in Context) Stable Diffusion FID 12.63 #35 of 69 Archive leaderboard report
Text-to-Image Generation COCO (Common Objects in Context) RA-CM3 (2.7B) FID 15.7 #41 of 69 Archive leaderboard report
Text-to-Image Generation COCO (Common Objects in Context) DALL-E (12B) FID 28 #56 of 69 Archive leaderboard report
Text-to-Image Generation COCO (Common Objects in Context) Vanilla CM3 FID 29.5 #58 of 69 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionBASEBPECLIPDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections