Search: “apple-ml-research”

Understanding how transformer components operate in LLMs is important, as it is at the core of recent technological advances in artificial intelligence. In this work, we revisit the challenges associa

Multi-agent LLM systems are increasingly deployed as autonomous collaborators, where agents interact freely rather than execute fixed, pre-specified workflows. In such settings, effective coordination

Reinforcement learning (RL) finetuning has become a key technique for enhancing large language models (LLMs) on reasoning-intensive tasks, motivating its extension to vision language models (VLMs). Wh

Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to purely autoregressive language models because they can decode multiple tokens in parallel. However, state-of-the-art

When annotators disagree on a label, the disagreement itself carries signal—and the number of annotators needed to capture it depends on the evaluation metric. We fine-tune NLI models on label distrib

Our next generation of Apple Intelligence is centered around our users, integrated deeply into our operating systems, and powered by a bold new architecture with privacy at its core. At the heart of t

CVPR 2026 is a major international conference on computer vision and pattern recognition organized by IEEE and the Computer Vision Foundation, where Apple presents its latest machine learning research

Streaming vision-language models (VLMs) continuously generate responses given an instruction prompt and an online stream of input frames. This is a core mechanism for real-time visual assistants. Exis

At Apple, we believe privacy is a fundamental human right. As AI capabilities increase and become more integrated into people’s daily lives, advancing research in privacy-preserving techniques is incr

Current critic-less RLHF methods aggregate multi-objective rewards via an arithmetic mean, leaving them vulnerable to constraint neglect: high-magnitude success in one objective can numerically offset

We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and appearance; compressive, aiding in downstream efficiency; and

One of the major differentiators unlocked by learned codecs relative to their hard-coded traditional counterparts is their ability to be optimized directly to appeal to the human visual system. Despit

True spatial intelligence for multimodal agents transcends low-level geometric perception, evolving from knowing where things are to understanding what they are for. While existing benchmarks, such as

Mixture-of-Experts (MoE) models enable sparse expert activation, meaning that only a subset of the model’s parameters is used during each inference. However, to translate this sparsity into practical

AI-driven sign language interpretation is limited by a lack of high-quality annotated data. New datasets including ASL STEM Wiki and FLEURS-ASL contain professional interpreters and 100s of hours of d

ICASSP 2026 is a major international conference focused on acoustics, speech, and signal processing research and applications. Apple's machine learning research team is participating in the conference

Normalizing flows (NFs) are end-to-end likelihood-based generative models for continuous data, and have recently regained attention with encouraging progress on image generation. Yet in the video gene

Recent advances in large language models (LLMs) test-time computing have introduced the capability to perform intermediate chain-of-thought (CoT) reasoning (thinking) before generating answers. While

Generative models are often deployed to make decisions on behalf of users, such as vision-language models (VLMs) identifying which person in a room is a doctor to help visually impaired individuals. Y

Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM’s autoregressive decoding may limit the ability to revisit and refine earlier t

Conditional diffusion models appear capable of compositional generalization, i.e., generating convincing samples for out-of-distribution combinations of conditioners, but the mechanisms underlying thi

We present StereoFoley, a video-to-audio generation framework that produces semantically aligned, temporally synchronized, and spatially accurate stereo sound at 48 kHz. While recent generative video-

Understanding and predicting motion is a fundamental component of visual intelligence. Although modern video models exhibit strong comprehension of scene dynamics, exploring multiple possible futures

Recurrent Neural Networks (RNNs) are naturally suited to efficient inference, requiring far less memory and compute than attention-based architectures, but the sequential nature of their computation h

Apple is advancing AI and ML with fundamental research, much of which is shared through publications and engagement at conferences in order to accelerate progress in this important field and support t

Understanding context is key to understanding human language, an ability which Large Language Models (LLMs) have been increasingly seen to demonstrate to an impressive extent. However, though the eval

ICLR 2026 is a major international conference on machine learning and deep learning where Apple's research team will present their latest work and innovations in representation learning. The conferenc

This paper was accepted at the Workshop on Navigating and Addressing Data Problems for Foundation Models (NADPFM) at ICLR 2026. Principled domain reweighting can substantially improve sample efficienc

Apple is presenting new research at the ACM CHI Conference on Human Factors in Computing Systems, held in person in Barcelona, Spain, from April 13–17, 2026, and is a sponsor of the event. Research...

This paper provides a theoretical framework for interpreting acoustic neighbor embeddings, which are representations of the phonetic content of variable-width audio or text in a fixed-dimensional embe

This paper was accepted at the Workshop on Memory for LLM-Based Agentic Systems at ICLR. Language models have consistently grown to compress more world knowledge into their parameters, but the knowled

Enterprise multi-agent AI systems produce thousands of inter-agent interactions per hour, yet existing observability tools capture these dependencies without enforcing anything. OpenTelemetry and Lang