🏢 Meta AI

AI models learn intuitive physics from self-supervised video pretraining.

Boosting language model reasoning: A novel hybrid approach using latent tokens drastically shortens reasoning traces, improving model performance and efficiency.

VideoJAM enhances video generation by jointly learning appearance and motion representations, achieving state-of-the-art motion coherence.

Zero-shot image safety judgment is achieved using MLLMs and a novel method called CLUE, objectifying safety rules, and significantly reducing the need for human labeling.

PartGen generates compositional 3D objects with meaningful parts from text, images, or unstructured 3D data using multi-view diffusion models, enabling flexible 3D part editing.

LLMs are trained to reason using language, but COCONUT lets them reason directly in a continuous latent space, boosting performance on logical tasks requiring complex planning.

EfficientTAMs achieve comparable video object segmentation accuracy to SAM 2 with ~2x speedup using lightweight ViTs and efficient cross-attention.

LLMs can dynamically adjust decoding temperature using Adaptive Decoding and Latent Preference Optimization, improving performance across creative and factual tasks.

Adaptive Caching (AdaCache) dramatically speeds up video generation with diffusion transformers by cleverly caching and reusing computations, tailoring the process to each video’s complexity and motio…