About
Research & Development
AI Models Learn to Selectively Forget Visual Styles, Not Just Categories
TLDR: A new research paper introduces Approximate Domain Unlearning (ADU), a method for Vision-Language Models (VLMs) to selectively forget images from specific visual domains...
Research & Development
Automating Knowledge Base Creation for Contact Center AI
TLDR: AI Knowledge Assist is a system that automatically builds knowledge bases for conversational AI agents in contact centers. It extracts question-answer pairs from...
Research & Development
Neural Exposure Fields: Achieving Consistent 3D Scenes from Varied Lighting
TLDR: Neural Exposure Fields (NExF) is a novel technique for robustly reconstructing 3D scenes with high quality and 3D-consistent appearance from challenging real-world captures,...
Research & Development
DeepPrune: A New Approach to Efficient Parallel Reasoning in Large Language Models
TLDR: DeepPrune is a novel framework that addresses the significant computational inefficiency in parallel reasoning for large language models (LLMs). It tackles inter-trace redundancy,...
Research & Development
Assessing Advanced Video Reasoning in Multimodal AI for Science
TLDR: SciVideoBench is a new benchmark designed to evaluate advanced video reasoning in scientific contexts for Large Multimodal Models (LMMs). It features 1,000 complex...
Research & Development
Unlocking Latent Reasoning: How Base Language Models Learn to Think
TLDR: A new research paper reveals that advanced 'thinking models' don't acquire entirely new reasoning capabilities but rather learn to effectively deploy pre-existing reasoning...