Papers › VLMo: Unified Vision-Language Pre-Training with Mixture-of-Modality-Experts

VLMo: Unified Vision-Language Pre-Training with Mixture-of-Modality-Experts

3 Nov 2021arXiv:2111.02358archive 2025-07-28

Hangbo Bao, Wenhui Wang, Li Dong, Qiang Liu, Owais Khan Mohammed, Kriti Aggarwal, Subhojit Som, Furu Wei

We present a unified Vision-Language pretrained Model (VLMo) that jointly learns a dual encoder and a fusion encoder with a modular Transformer network. Specifically, we introduce Mixture-of-Modality-Experts (MoME) Transformer, where each block contains a pool of modality-specific experts and a shared self-attention layer. Because of the modeling flexibility of MoME, pretrained VLMo can be fine-tuned as a fusion encoder for vision-language classification tasks, or used as a dual encoder for efficient image-text retrieval. Moreover, we propose a stagewise pre-training strategy, which effectively leverages large-scale image-only and text-only data besides image-text pairs. Experimental results show that VLMo achieves state-of-the-art results on various vision-language tasks, including VQA, NLVR2 and image-text retrieval. The code and pretrained models are available at https://aka.ms/vlmo.

PaperPDFCode

In Syntology View this paper on Syntology: its repositories, every harvested function with whether it ran, its licence and the call to fetch it.

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

Code

microsoft/unilm officialpytorch report
ylsung/vl-merging mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

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

Tasks

Image RetrievalImage-text RetrievalRetrievalText RetrievalVisual Question Answering (VQA)Visual Reasoning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Retrieval PhotoChat VLMo R1 11.5 #2 of 5 Archive leaderboard report
Image Retrieval PhotoChat VLMo R@10 39.4 #2 of 5 Archive leaderboard report
Image Retrieval PhotoChat VLMo R@5 30.0 #2 of 5 Archive leaderboard report
Image Retrieval PhotoChat VLMo Sum(R@1,5,10) 83.2 #2 of 5 Archive leaderboard report
Text Retrieval Image-Chat VLMo R@1 46.8 #3 of 3 Archive leaderboard report
Text Retrieval Image-Chat VLMo R@5 67.5 #3 of 3 Archive leaderboard report
Text Retrieval Image-Chat VLMo Sum(R@1,5) 114.3 #3 of 3 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-dev VLMo Accuracy 82.78 #3 of 56 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-std VLMo number 67.26 #5 of 38 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-std VLMo other 72.87 #5 of 38 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-std VLMo overall 81.30 #5 of 38 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-std VLMo yes/no 94.68 #5 of 38 Archive leaderboard report
Visual Reasoning NLVR2 Dev VLMo Accuracy 85.64 #6 of 15 Archive leaderboard report
Visual Reasoning NLVR2 Test VLMo Accuracy 86.86 #6 of 14 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

Introduced by this paper: VLMo

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformerVLMo

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