Skip to main content

🏢 Mila - Quebec AI Institute

When is an Embedding Model More Promising than Another?
·4115 words·20 mins· loading · loading
AI Theory Representation Learning 🏢 Mila - Quebec AI Institute
This paper introduces a novel, task-agnostic method for ranking embedding models using information sufficiency, a concept derived from communication theory and statistical experiments comparison, demo…
VisMin: Visual Minimal-Change Understanding
·2710 words·13 mins· loading · loading
Multimodal Learning Vision-Language Models 🏢 Mila - Quebec AI Institute
VisMin benchmark evaluates visual-language models’ fine-grained understanding by identifying minimal image-text differences (object, attribute, count, spatial relation). Current VLMs struggle with sp…