No system in this track uses task-specific fine-tuning, and every system converts videos to text before reasoning. Multi-video reasoning has lots of room for improvement.
๐งต The MAGMaR 2026 Shared Task results are out!
We challenged teams to retrieve relevant videos from a corpus of 110K+ multilingual clips AND generate grounded, persona-driven articles from them.
Come see the findings at our workshop at ACL on July 4!
https://t.co/9YNvHbEYKH
๐คWhat we learned abt generation:
Humans and automatic metrics don't agree on what's good here. Annotators rewarded systems that said "I don't have enough evidence to answer that" MiRAGE penalized them for it. Figuring out how to evaluate appropriate refusal is a clear next step
Nanyun (Violet) Peng @VioletNPeng is giving a Keynote entitled "Towards Self-Improving Multimodal Models"
Chenliang Xu's keynote will be "Multi-level Alignment in Audio-Visual Scene Generation and Learning" @ChenliangXu
โฐ Deadline extended! The submission deadline for MAGMaR 2026 is now April 15. Donโt miss your chance to share work on multimodal retrieval & generation at ACL 2026! ๐
๐ https://t.co/I4cEEoWQAC
Shared Tasks (Retrieval, Generation, & RAG) are April 20th! See you in San Diego!
This year's shared task allows you to submit for the retrieval track, generation track, or full RAG track on a challenging new collection of unedited ("raw") videos.
Research Papers (Apr. 1)
Shared Task (Apr. 20)
๐น + ๐ง + ๐ = ๐ฅ First call for MAGMaR 2026, the 2nd workshop on multimodal augmented generation via multimodal retrieval! If #RAG isn't hard enough for you, try multilingually and multimodally. Collocated with @aclmeeting in San Diego in July.
https://t.co/I4cEEoWQAC
24 hours left to submit a paper to the Multimodal RAG workshop at #acl2025! We welcome all papers on related topics. Also be sure to check out our shared task (paper deadline 5/9).
https://t.co/WhjOkkSi4A
๐จ IT'S HERE! ๐จ
The Eval Leaderboard is now LIVE! ๐๐ป
Our video retrieval collection stumps most pre-trained models. See if you can build a better system!
https://t.co/adA7rWffaL
๐จ IT'S HERE! ๐จ
The Eval Leaderboard is now LIVE! ๐๐ป
Our video retrieval collection stumps most pre-trained models. See if you can build a better system!
https://t.co/adA7rWffaL
๐จ Deadline extension! ๐จ
The MAGMaR paper submission date has been extended to May 1, 2025 (archival and non-archival). Show off your multimodal + RAG projects!