All news

Tests whether frontier agents can automate ML for science in the wild.

Timestep-aware quantization for diffusion LMs.

Memory-centric continual learning for AI agents.

WorldStream streams interactive 3D worlds with persistent state.

Controlled study of learned unmasking policies in diffusion LMs.

Studies compositionality in RAG.

Causal evidence of hidden latent communication between diffusion LM agents.

Looped uniform diffusion language models.

HyperDrone quadrotor simulator at millions of frames per second.

Recursive self-improvement for forecasting in open worlds.

TraceForge distills with the student own reasoning traces.

Position on recursive self-improvement in physical AI.

Generalist dexterous bimanual tracking from human demos.

Explainable artifact detection in AI-generated videos.

Bridged self-distillation for on-policy learning.

Agentic system for 3D virtual film cinematography.

Foundation4D unified perception and reconstruction for autonomous vehicles.

Continuous theory-of-mind tokens for multi-agent coordination.

StreamEval evaluates interactive streaming video generation.

Stage-aware robustness evaluation of VLA policies.

Resource-linear budget control for feed-forward 3DGS.

Geometry-aware feature lifting for RGB-D object pose.

AgentVQA benchmark for agentic visual understanding.