About
Research & Development
DiPO: A Fine-Grained Method for LLM Unlearning
TLDR: Distribution Preference Optimization (DiPO) is a new algorithm for LLM unlearning that operates at the distribution level, directly targeting next-token probabilities. It overcomes...
Research & Development
Robot Swarms Adapt On-the-Fly with AI-Generated Code and Self-Organizing Nervous Systems
TLDR: A new research paper introduces the Self-organizing Nervous System (SoNS) which enables robot swarms to automatically request and execute new code generated by...
Research & Development
Unlocking Faster Robotic Control: HyperVLA’s Approach to Efficient AI
TLDR: HyperVLA is a new architecture for Vision-Language-Action (VLA) models that drastically reduces inference costs for robots. By using hypernetworks, it trains a large...
Research & Development
Leveraging Latent Expertise in Diffusion Language Models for Enhanced Reasoning
TLDR: Diffusion-based Large Language Models (dLLMs) implicitly learn multiple "semi-autoregressive experts" during training. A new training-free inference method called HEX (Hidden semi-autoregressive EXperts) leverages...
Research & Development
HybridFlow: A Unified Approach to Quantifying Model and Data Uncertainty in Machine Learning
TLDR: HybridFlow is a new machine learning framework that combines normalizing flows and probabilistic predictors to accurately quantify both aleatoric (data noise) and epistemic...
Research & Development
Precise Image Editing: Introducing SAEdit’s Token-Level Control
TLDR: SAEdit is a new method for image editing that uses Sparse Autoencoders (SAEs) to achieve highly precise and continuous control over image attributes....