Zeyuan Yang

5 posts

Zeyuan Yang

@miiche_yang

UMass PhD | Current Intern @ Samsung | Previous @ THU

miicheyang.github.io

Joined May 2022

RepliesReplies RepostsReposts MediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone

or

Log in with username or email

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.

Zeyuan Yang
@miiche_yang
Jun 27, 2025
Thanks @_akhaliq for sharing our work! Step into the Mirage and explore multimodal reasoning!
AK
@_akhaliq
Jun 27, 2025
Machine Mental Imagery Empower Multimodal Reasoning with Latent Visual Tokens
00:00
Zeyuan Yang
@miiche_yang
Jun 27, 2025
VLMs can think visually without generating pixels! I am thrilled to announce Mirage, a multimodal reasoning framework that interleaves latent visual representations among text tokens!
Chuang Gan
@gan_chuang
Jun 27, 2025
VLM can think visually without generating pixels! VLM can think visually without generating pixels! VLM can think visually without generating pixels! 📢 We introduce Machine Mental Imagery (Mirage): a new framework that enables VLM to imagine using latent visual

Advertisement

Advertisement