OFA: A Unified Multimodal Pretrained Model

Home

Discover

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and more.Subscribe

CC BY-SA

Source arXiv

Computer Vision

Computation and Language

OFA: A Unified Multimodal Pretrained Model

Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

To a simple sequence-to-sequence learning framework based on the encoder-decoder architecture.

The proposed model performs pretraining and finetuning with task instructions and introduces no extra task-specific layers for finetuning.

Experimental results show that the proposed model achieves new state-of-the-arts on a series of multimodal tasks, including image captioning (test-std acc.

Authors

Peng Wang, An Yang, Rui Men, Junyang Lin, Shuai Bai, Zhikang Li, Jianxin Ma, Chang Zhou, Jingren Zhou, Hongxia Yang

◐ Recommended Follows

◐ Latest Activity