Vector Institute

An incredible start to DiscoveryX—a powerful showcase of Ontario’s innovation ecosystem in action. Congratulations to Dr. Graham Taylor of University of Guelph, the first University Winner of the OCI Educator Innovation & Entrepreneurship Award, recognized for exceptional leadership and innovation in AI. A strong reminder of the talent, research excellence, and real-world impact being driven across our universities in partnership with Ontario Centre of Innovation. #DiscoveryX #AI #Innovation #Ontario @OCInnovation @uofg @UofGuelphNews @VectorInst

Orthus: an RNA foundation model pretrained via contrastive learning based on splicing isoforms and orthologous transcripts. @phil_fradkin @ianshi3 @Taykhoom_Dalal @BoWang87 @quaidmorris @UofT @VectorInst @MSKCancerCenter nature.com/articles/s4159…

🚨More time to submit your work to the Workshop on Weight-Space Symmetries at #ICML2026! 🗓️Deadline extended to May 8, 23:59 AoE 📜Submission: openreview.net/group?id=ICML.… We welcome papers on model merging, symmetry-aware optimization, weight-space learning, and more!

Our 15 April webinar on the use of #AI in the public sector has been postponed to 30 April. Register below and don't miss precious insights from experts in the field! 👇 us02web.zoom.us/webinar/regist… Thrilled to have partner @VectorInst from Canada on board! #HorizonEurope #XAI

This work was conducted by Vector Institute and the National Research Council of Canada (@NRC_CNRC) on behalf of the Canadian AI Safety Institute (CAISI) @ISED_CA. By making these tools openly available, we aim to support responsible AI development and strengthen trust in AI systems. Explore the code: MLflow: github.com/VectorInstitut… Podman: github.com/VectorInstitut…

These tools make it easier to assess how AI systems perform in practice, supporting more consistent, transparent, and reliable testing. The MLflow extension helps users track and compare results across different tests, while the Podman extension allows evaluations to run in secure, self-contained environments that can be easily reused and shared. Together, these extensions address key challenges in AI evaluation. They help users understand how systems actually behave under different conditions, beyond basic performance scores.

We’re pleased to announce two new open-source extensions to the Inspect framework for AI evaluation: the MLflow extension and the Podman extension 🔧

These papers represent more than research output; they're the building blocks for AI systems that are more capable, more trustworthy, and more beneficial for society. Want the full details? Discover all 48 accepted papers across our Vector Faculty Member, Faculty Affiliate, and postdoc community: vectorinstitute.ai/vector-researc…

🚀 AI's expanding frontiers Steven Waslander’s DrivingGen: Comprehensive benchmarking for generative driving world models Wenhu Chen’s VisCoder2: Multi-language visualization coding agents, and YuE: Scaling open foundation models for long-form music generation Renjie Liao’s StreamSplat: Online dynamic 3D reconstruction from video

Today marks the start of the world’s premier deep learning conference in Rio de Janeiro. Here’s some of the work Vector’s research community is bringing to #ICLR2026 👇🧵

Featuring Mehdy Dousty, who is building RadioVision to sense how people move & interact in everyday environments using radio-frequency signals. As RadioVision has moved from research to startup, the mission is clear to build technology that performs reliably in the real world🚀

Excited to share that the code and weights for our relighting (UniRelight) and lighting estimation (LuxDiT) works are now public. 📢 1/n Code 🧑‍💻 💡 github.com/nv-tlabs/UniRe… 📷 github.com/nv-tlabs/LuxDiT Weights 🏋️‍♀️ 💡 huggingface.co/nvidia/UniReli… 📷 huggingface.co/nvidia/LuxDiT

Applications open now: 📅 Fall deadline: May 19, 2026 📅 Winter deadline: June 22, 2026 Apply now: vectorinstitute.ai/research-talen…

📚 Professional development Access targeted career development events and resources, and gain hands-on experience designed to accelerate your future in AI research

Ready to work on AI research that bridges academic theory and real-world application? Applications for Vector's Research Internships are now open for Fall 2026 and Winter 2027 at vectorinstitute.ai/research-talen… What awaits you: 🧵