
  <rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
      <title>Apple Machine Learning Research</title>
      <link>https://machinelearning.apple.com</link>
      <description>Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.</description>
      <language>en</language>
      <lastBuildDate>Mon, 16 Nov 2026 00:00:00 GMT</lastBuildDate>
      <atom:link href="https://machinelearning.apple.com/rss.xml" rel="self" type="application/rss+xml"/>
      
  <item>
    <guid>split-scale-latent-3d</guid>
    <title>Split-and-Scale Latent 3D Representations</title>
    <link>https://machinelearning.apple.com/research/split-scale-latent-3d</link>
    <description>Latent 3D representations have significantly expanded the capabilities of modern shape modeling, enabling compact encoding of 3D scenes and supporting powerful 3D generative models. Despite this progress, current representations struggle to capture the fine details of complex objects, fundamentally limited by the capacity of their latents. However, scaling latent capacity at training time is computationally expensive and often infeasible under memory limits, and naively increasing the number of latents at inference fails to make effective use of the added capacity. In this paper, we address…</description>
    <pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>theoretical-limits-alignment</guid>
    <title>Theoretical Limits of Language Model Alignment</title>
    <link>https://machinelearning.apple.com/research/theoretical-limits-alignment</link>
    <description>Language model (LM) alignment improves model outputs to reflect human preferences while preserving the capabilities of the base model. The most common alignment approaches are (i) reinforcement learning, which maximizes the expected reward under a KL-divergence constraint, and (ii) best-of-NNN alignment, which selects the highest-reward output among NNN independent samples. Despite their widespread use, the fundamental limits of reward improvement under a KL budget remain poorly understood. We characterize the information-theoretic limits of KL-regularized alignment by deriving the maximum…</description>
    <pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>evaluating-evaluators-llm-judges</guid>
    <title>Evaluating the Evaluators: Investigating LLM Judges for Personalized Writing Style Assessment</title>
    <link>https://machinelearning.apple.com/research/evaluating-evaluators-llm-judges</link>
    <description>This paper was accepted at the Can We Trust AI Evaluation? Workshop at NeurIPS 2026.
We evaluate how good large language model (LLM) judges are for personalized writing tasks. This is important when a user’s preferred writing style is known (e.g., “upbeat”) and an LLM judge is used to evaluate whether generated text adheres to this preference. We examine how the writing task, the combination of generator and judge LLM, the evaluation objective, and general commonsense and reasoning ability impact LLM judge performance. To this end, we generate text with four LLMs for three long-form writing…</description>
    <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>gating-memorization-in-context</guid>
    <title>On the Importance of Gating: Memorization vs. In-Context Learning in State Space Models</title>
    <link>https://machinelearning.apple.com/research/gating-memorization-in-context</link>
    <description>State Space Models (SSMs) have emerged as a compelling alternative to Transformers, enabling sequence modeling with constant memory and linear compute. Although SSMs exhibit reasonable performance and favorable computational characteristics, they continue to lag behind Transformers on tasks that require in-context learning and precise retrieval, slowing their adoption for large-scale language modeling. In this work, we demonstrate that both the success and failure of SSMs in these domains can be explained by studying the role of the gating mechanism, a prevalent component in modern recurrent…</description>
    <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>weighted-state-souping</guid>
    <title>Weighted State Souping for Efficient Long-Context Retrieval</title>
    <link>https://machinelearning.apple.com/research/weighted-state-souping</link>
    <description>This paper was accepted at the Workshop on Personalized, Aligned, Long-term Memory for AI Systems (PALM) at NeurIPS 2026.
State space models (SSMs) combine fast inference and fixed state size with competitive language-modeling performance. Recently, state souping has been proposed as a procedure whereby an SSM independently encodes chunks of tokens like documents, tool definitions, or memory records. The resulting state encodings may be cached and recombined to cheaply construct a context for downstream queries. However, both state souping and vanilla SSMs deteriorate on long, crowded…</description>
    <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>private-evolution-clustering</guid>
    <title>Understanding Private Evolution as Learning-Augmented Clustering</title>
    <link>https://machinelearning.apple.com/research/private-evolution-clustering</link>
    <description>Private Evolution (PE) is a differentially private algorithm for synthetic data generation. While it can be viewed as a Wasserstein learning algorithm, it performs much better in practice than worst-case Wasserstein analyses would predict. We recast PE as generative model-augmented Wasserstein learning. We show theoretically that when we take into account the use of a generative model that is able to capture something about the true distribution, then we can obtain much better performance bounds. For example, if the generator gives samples in the same low-dimensional space as the distribution…</description>
    <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>theory-evaluability</guid>
    <title>Toward a Theory of Evaluability</title>
    <link>https://machinelearning.apple.com/research/theory-evaluability</link>
    <description>Machine learning models, particularly large foundation models, are increasingly evaluated along multiple axes that are relevant for their usage. In addition to accuracy measures appropriate for the specific task of interest, measures capturing safety, calibration and fairness properties are often of interest. The classical approach to these problems assumes that the test set has been held out from the training process, an assumption that is increasingly harder to justify for large foundation models. In this work, we formalize a notion of evaluability without this assumption, and ask if it is…</description>
    <pubDate>Tue, 10 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>faster-retrieval</guid>
    <title>Faster Retrieval for Better, More Efficient RAG</title>
    <link>https://machinelearning.apple.com/research/faster-retrieval</link>
    <description>Maximum inner product search (MIPS) sits at the core of recommendation systems, search, and retrieval-augmented generation (RAG). Given a query vector, MIPS systems search within a database of millions of candidate key vectors for the one that best matches the query. The original data point associated with that best-matching key is then used to handle the query: in RAG, the document that is best suited to answer the question is typically prefixed to the query.
As the number of keys grows to millions and vector dimensions reach thousands, the computational cost of running an exhaustive…</description>
    <pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>attention-mamba-distillation</guid>
    <title>Attention to Mamba: A Recipe for Cross-Architecture Distillation</title>
    <link>https://machinelearning.apple.com/research/attention-mamba-distillation</link>
    <description>State Space Models (SSMs) such as Mamba have become a popular alternative to Transformer models, due to their reduced memory consumption and higher throughput at generation compared to their Attention-based counterparts. On the other hand, the community has built up a considerable body of knowledge on how to train Transformers, and many pretrained Transformer models are readily available. To facilitate the adoption of SSMs while leveraging existing pretrained Transformers, we aim to identify an effective recipe to distill an Attention-based model into a Mamba-like architecture. In prior work…</description>
    <pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate>
  </item>

  <item>
    <guid>gharp-head-animation</guid>
    <title>GHARP: Real-Time Gaussian Head Animation from Large-Scale Reconstruction Prior</title>
    <link>https://machinelearning.apple.com/research/gharp-head-animation</link>
    <description>We present GHARP (Real-time Gaussian Head Animation from Large-scale Reconstruction Prior), a method that animates 3D human heads in real time from a few input images of a subject and a driving expression signal. We decouple the problem into an identity stage that builds a representation of the subject’s geometry and appearance offline, and an animation stage that predicts expression-dependent residuals on top of it at runtime. This separation offers a favorable trade-off with respect to fidelity, quality and runtime: the identity stage can be expensive while the animation stage runs a…</description>
    <pubDate>Mon, 26 Oct 2026 00:00:00 GMT</pubDate>
  </item>

    </channel>
  </rss>
