Paola Cascante-Bonilla

Do VLMs Need Vision Transformers? A controlled study of Transformer, state space, and hybrid vision backbones for frozen vision-language models. Finding: - Under matched settings, VMamba improves localization while staying competitive on open-ended VQA, making SSMs a practical alternative to ViTs. - Detection or segmentation pretraining generally improves VQA and localization, with the largest gains appearing in backbones that need more spatial inductive bias. - Better classification scores and naive scaling do not consistently predict stronger downstream VLM behavior, especially for grounding-sensitive tasks. - Some dense-objective checkpoints fail sharply in localization, but simple interface and connector adjustments recover much more robust behavior. Paper Title: Do VLMs Need Vision Transformers? Evaluating State Space Models as Project: lab-spell.github.io/vlm-ssm-vision… Link: arxiv.org/abs/2603.19209

📢 Call for Participation: The #CVPR2026 Doctoral Consortium submission portal is now open! Important Dates • Submission deadline: Apr 13, 2026 • Notification: Apr 27, 2026 (estimated) Details: cvpr.thecvf.com/Conferences/20…

@dhillon_p and my work is in the @NewYorker today. Words can’t describe how I feel. Many thanks to Vauhini Vara for this beautiful and thought provoking essay. If you care about AI disclosures, Fair Use & the Future of Creative Labor I recommend reading it :)

🚨New paper on AI and copyright Several authors have sued LLM companies for allegedly using their books without permission for model training. 👩‍⚖️Courts, however, require empirical evidence of harm (e.g., market dilution). Our new pre-registered study addresses exactly this gap. Joint work with Profs @dhillon_p (@umsi) & Jane Ginsburg ( @ColumbiaLaw) (1/n)🧵

Privacy protection is vital — but blanket masking can cripple model performance. Our COLM 2025 paper introduces FiG-Priv, a fine-grained privacy protection framework that masks only high-risk content, keeping useful details for blind & low-vision users. 📄 arxiv.org/abs/2508.09245

Thanks @arankomatsuzaki, @ManuelFaysse, and @_akhaliq for reposting our work! We extended matryoshka to multi-vector embedding and were glad to see that the popular concept, test-time scaling in LLM, also works in retrieval. Work done with @AIatMeta & @vislang!

@TuhinChakr @icmlconf @stonybrooku @sbucompsc @mbodhisattwa @hsanchaita Awesome work! 🔥Congrats!! 🐺

Honored to get the outstanding position paper award at @icmlconf :) Come attend my talk and poster tomorrow on human centered considerations for a safer and better future of work I will be recruiting PhD students at @stonybrooku @sbucompsc coming fall. Please get in touch.

#ICCV2025 is deeply committed to promoting diversity, equity, and inclusion within our community. As part of this commitment, travel support is available to help broaden participation. Applications will be reviewed on a rolling basis until August 20, 2025 (anywhere on Earth).

Now accepted to @COLM_conf 🤩 Super excited for Montreal 🇨🇦🍁 This also marks my third successful collaboration with my good friend @PhilippeLaban

Full paper, code and webpage will be available soon~! 🦙✨

We introduce a multi-agent collaboration system that produces fine-grained information and privacy risk scores to enable usability and utility while preserving privacy. Kudos to all amazing collaborators: Haoran Niu, Suzanne Barber, @haldaume3, @Triiiiiista!

Beyond Blanket Masking: Examining Granularity for Privacy Protection in Images Captured by Blind and Low Vision Users -now accepted to #COLM Last June, Jeffri Murrugarra presented a short version at LXAI and VizWiz Workshops @ CVPR'25 More info soon; see you all in Montreal!✨

We introduce a super simple yet effective strategy to improve video-language alignment (+18%): add hallucination correction in your training objective👌 Excited to share our accepted paper at ACL: Can Hallucination Correction Improve Video-language Alignment

Thinking about model welfare and catastrophic risks without considering longitudinal harms cause by Generative AI ? Check out our @icmlconf Oral paper on why AI Safety should prioritize Future of Work. Had lots of fun writing this #GenAI

Excited to be at #CHI2025. I will be presenting our work ( Best Paper Honorable Mention) tomorrow April 29th ( 11:10-11:22) moderated by @huashen218 Location : G416+ 417 I am also recruiting PhD students and visiting researchers at @sbucompsc . DMs are open. Please say hi !!

Unlike math/code, writing lacks verifiable rewards. So all we get is slop. To solve this we train reward models on expert edits that beat SOTA #LLMs largely on a new Writing Quality benchmark. We also reduce #AI slop by using our RMs at test time boosting alignment with experts.