🏢 Meta AI

Temp-G³NTK: a novel temporal graph neural tangent kernel guarantees convergence to graphon NTK, offering superior performance in temporal graph classification and node-level tasks.

Stochastic Optimal Control Matching (SOCM) significantly reduces errors in stochastic optimal control by learning a matching vector field using a novel iterative diffusion optimization technique.

Rainbow Teaming: a novel black-box approach generates diverse adversarial prompts to enhance LLM robustness and safety, achieving over 90% attack success rate across various models.

Passive data collection in complex social systems invalidates standard AI model validation; new methods are needed.

Meta 3D AssetGen: High-quality text-to-mesh generation with realistic PBR materials and lighting, exceeding prior methods in speed and accuracy.

MEGALODON: A new neural architecture for LLMs, enabling unlimited context length with improved efficiency and accuracy.

Large language models surprisingly solve unseen arithmetic tasks; this work reveals how they learn to compose simple skills into complex ones through in-context learning, showing a transition from mem…

Researchers reveal how Chain-of-Thought reasoning emerges in transformers via specialized ‘iteration heads’, improving LLM performance and offering insights into mechanistic interpretability.

AI-powered sequence-to-sequence transformers surpass human and algorithmic abilities in discovering Lyapunov functions for dynamical systems, solving a long-standing open problem in mathematics.

FlowLLM revolutionizes material design by cleverly merging large language models and Riemannian flow matching, yielding a 300% boost in stable material generation!

CALVIN: Instruction tuning boosts contextual video captioning, achieving state-of-the-art results!

LLMs spontaneously encode syntax using a polar coordinate system, representing syntactic relations via relative direction and distance of word embeddings.