Liu, Shikun and Fan, Linxi and Johns, Edward and Yu, Zhiding and Xiao, Chaowei and Anandkumar, Anima (2023) Prismer: A Vision-Language Model with An Ensemble of Experts. . (Unpublished) https://resolver.caltech.edu/CaltechAUTHORS:20230316-153658096
![]() |
PDF
- Submitted Version
See Usage Policy. 5MB |
Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20230316-153658096
Abstract
Recent vision-language models have shown impressive multi-modal generation capabilities. However, typically they require training huge models on massive datasets. As a more scalable alternative, we introduce Prismer, a data- and parameter-efficient vision-language model that leverages an ensemble of domain experts. Prismer only requires training of a small number of components, with the majority of network weights inherited from readily-available, pre-trained domain experts, and kept frozen during training. By leveraging experts from a wide range of domains, we show that Prismer can efficiently pool this expert knowledge and adapt it to various vision-language reasoning tasks. In our experiments, we show that Prismer achieves fine-tuned and few-shot learning performance which is competitive with current state-of-the-art models, whilst requiring up to two orders of magnitude less training data. Code is available at https://github.com/NVlabs/prismer.
Item Type: | Report or Paper (Discussion Paper) | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Related URLs: |
| ||||||||||
ORCID: |
| ||||||||||
Record Number: | CaltechAUTHORS:20230316-153658096 | ||||||||||
Persistent URL: | https://resolver.caltech.edu/CaltechAUTHORS:20230316-153658096 | ||||||||||
Usage Policy: | No commercial reproduction, distribution, display or performance rights in this work are provided. | ||||||||||
ID Code: | 120079 | ||||||||||
Collection: | CaltechAUTHORS | ||||||||||
Deposited By: | George Porter | ||||||||||
Deposited On: | 16 Mar 2023 22:18 | ||||||||||
Last Modified: | 16 Mar 2023 22:18 |
Repository Staff Only: item control page