Grounding Language Models to Images for Multimodal Inputs and Outputs

Jing Yu Koh, Ruslan Salakhutdinov, Daniel Fried
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:17283-17300, 2023.

Abstract

We propose an efficient method to ground pretrained text-only language models to the visual domain, enabling them to process arbitrarily interleaved image-and-text data, and generate text interleaved with retrieved images. Our method leverages the abilities of language models learnt from large scale text-only pretraining, such as in-context learning and free-form text generation. We keep the language model frozen, and finetune input and output linear layers to enable cross-modality interactions. This allows our model to process arbitrarily interleaved image-and-text inputs, and generate free-form text interleaved with retrieved images. We achieve strong zero-shot performance on grounded tasks such as contextual image retrieval and multimodal dialogue, and showcase compelling interactive abilities. Our approach works with any off-the-shelf language model and paves the way towards an effective, general solution for leveraging pretrained language models in visually grounded settings.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-koh23a, title = {Grounding Language Models to Images for Multimodal Inputs and Outputs}, author = {Koh, Jing Yu and Salakhutdinov, Ruslan and Fried, Daniel}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {17283--17300}, 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/koh23a/koh23a.pdf}, url = {https://proceedings.mlr.press/v202/koh23a.html}, abstract = {We propose an efficient method to ground pretrained text-only language models to the visual domain, enabling them to process arbitrarily interleaved image-and-text data, and generate text interleaved with retrieved images. Our method leverages the abilities of language models learnt from large scale text-only pretraining, such as in-context learning and free-form text generation. We keep the language model frozen, and finetune input and output linear layers to enable cross-modality interactions. This allows our model to process arbitrarily interleaved image-and-text inputs, and generate free-form text interleaved with retrieved images. We achieve strong zero-shot performance on grounded tasks such as contextual image retrieval and multimodal dialogue, and showcase compelling interactive abilities. Our approach works with any off-the-shelf language model and paves the way towards an effective, general solution for leveraging pretrained language models in visually grounded settings.} }
Endnote
%0 Conference Paper %T Grounding Language Models to Images for Multimodal Inputs and Outputs %A Jing Yu Koh %A Ruslan Salakhutdinov %A Daniel Fried %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-koh23a %I PMLR %P 17283--17300 %U https://proceedings.mlr.press/v202/koh23a.html %V 202 %X We propose an efficient method to ground pretrained text-only language models to the visual domain, enabling them to process arbitrarily interleaved image-and-text data, and generate text interleaved with retrieved images. Our method leverages the abilities of language models learnt from large scale text-only pretraining, such as in-context learning and free-form text generation. We keep the language model frozen, and finetune input and output linear layers to enable cross-modality interactions. This allows our model to process arbitrarily interleaved image-and-text inputs, and generate free-form text interleaved with retrieved images. We achieve strong zero-shot performance on grounded tasks such as contextual image retrieval and multimodal dialogue, and showcase compelling interactive abilities. Our approach works with any off-the-shelf language model and paves the way towards an effective, general solution for leveraging pretrained language models in visually grounded settings.
APA
Koh, J.Y., Salakhutdinov, R. & Fried, D.. (2023). Grounding Language Models to Images for Multimodal Inputs and Outputs. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:17283-17300 Available from https://proceedings.mlr.press/v202/koh23a.html.

Related Material