Soham Daga
Computer science
Hometown: Bangalore, Karnataka, IN
Graduation date: Spring 2027
FURI | Fall 2026, Needs Review
Reasoning by Sight: Can Pictures Replace Tokens?
Most AI models read a problem as tokens, short pieces of text. Recent work showed a vision model can instead solve visual puzzles as pictures. This research asks whether that idea can grow into a foundation model for reasoning: one vision model that handles many kinds of puzzles, learns new kinds from a few examples, and improves with size. The researcher trains small vision models on puzzles from math, logic, algorithms, patterns and games, each shown as a picture, with a matching token version as a reference. The goal is one model that can read any problem as a picture.
Mentor: Huan Liu
It’s hip to be square.
Students presenting projects at the Fulton Forge Student Research Expo are encouraged to download this personal QR code and include it within your poster. This allows expo attendees to explore more about your project and about you in the future.
Right click the image to save it to your computer.