Loading activity…
@georgiagkioxari
Computing + Mathematical Sciences at Caltech 🏛️ ∙ Computer vision enthusiast 🤖 ∙ SAM 3D, PyTorch3D ∙ Ex: FAIR, Google Brain 👩🏻💻∙ From 🇬🇷
AI
A post introducing beignet says its multiresolution Fourier frequency pyramid brings faster convergence and higher precision for those tasks.
AI
A post highlights a late-breaking results session for work on robotics agents or Real2Sim2Real.
AI
An author says a new blog post and paper examine how experts in multimodal mixture-of-experts models organize during pre-training.
AI
A researcher describes a spatially localized 3D reward model designed to show where generated scenes go wrong.
AI
A post announcing DynaTokens says its approach trains a few tokens at test time to enable scene dynamics in world models while preserving camera control.
AI
A user sees large-scale real2sim—bringing the real world into simulation—as within reach, noting how hard it once was to imagine OpenAI putting serious resources into 3D/4D.
AI
A researcher describes work across Gemini, Omni and generative media to test ideas about physical AI, arguing that language models remain indispensable.
AI
A collaborator said the tutorial would cover world model evaluation, including Ziqi’s new findings on how best to assess—and, in turn, develop—these models.
AI
A reply says language models could improve with a closed feedback loop, accurate task checks and safety guardrails—but argues the system needed for that setup is still far off.
AI
Atlas generates videos with exact camera control and reconstructs 3D scenes from input images.
