Apple Machine Learning Research
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 8
- Apple proposes weighted state souping for long-context retrieval
Apple Machine Learning Research published a paper introducing weighted state souping, a method that learns query-dependent weights for composing cached state-space-model representations. The authors report substantial long-context gains across document question answering, retrieval, variable tracking, and tool use, and describe a sparse variant with near-constant encoding cost relative to context length. The paper was accepted at the PALM workshop at NeurIPS 2026, while its performance claims have not been independently corroborated.
Apple Machine Learning Research · 🔥 0 - GHARP brings real-time Gaussian head animation to mobile
GHARP is a method for animating 3D human heads in real time from a few images of a subject and a driving expression signal. Its authors say it separates an offline identity-reconstruction stage from a lightweight runtime animation stage optimized for mobile devices. The runtime predicts expression-dependent residuals in the semantically structured latent space of a pretrained reconstruction model. The authors also report that ambiguous body appearance can cause blur and temporal flicker, which they address with a body alignment network. They claim state-of-the-art quality on Ava-256, execution up to 13 times faster on an A100 GPU, and eight times fewer Gaussians; the supplied evidence does not independently verify these results.
Apple Machine Learning Research · 🔥 0 - Apple study compares plan-first vs retrieve-first tool discovery for LLM agents
Apple Machine Learning Research published an EMNLP paper by Mingfei Zhang and co-authors comparing two ways tool-using LLM agents find APIs before planning: Retrieve-then-Plan (RtP), which embeds the query and retrieves similar tools, and Plan-then-Retrieve (PtR), in which the model first generates abstract tool stubs that become retrieval queries. On ToolBench's 46,984 API endpoints across 50 categories, RtP beat PtR on single-tool queries while PtR beat RtP on multi-tool queries; the crossover held at retrieval budgets k=1–20, registry sizes from 1k to 47k APIs, and with Gemini 2.5 Flash and gpt-oss-120b. The authors also describe a stub-hallucination failure mode, note that PtR's extra discovery call roughly doubles inference cost, and suggest complexity-adaptive routing using RtP's top-1 retrieval score.
Apple Machine Learning Research · 🔥 0 - Apple researchers unveil first tri-modal masked diffusion models trained from scratch
Apple Machine Learning Research published a NeurIPS paper introducing what the authors describe as the first tri-modal Masked Diffusion Models pretrained from scratch on text, image-text and audio-text data. The study systematically analyzes multimodal scaling laws, modality mixing ratios, noise schedules and batch-size effects, and proposes an SDE-based reparameterization that decouples physical batch size from logical batch size so the optimal batch size no longer has to be tuned. The authors also pretrain a preliminary 3B-parameter model on 6.4T tokens, reporting strong results in text generation, text-to-image and text-to-speech tasks; these are self-reported and not independently measured, and the paper's claim to be the largest systematic open study of multimodal discrete diffusion is author-asserted.
Apple Machine Learning Research · 🔥 0 - Apple researchers detect code-switching directly from speech encoders
A paper from Apple Machine Learning Research proposes a lightweight method that detects code-switching — alternating languages within one utterance — directly from speech, without relying on accurate multilingual transcription. The approach uses frame-level language information in pretrained speech encoders: it identifies language-selective hidden-state dimensions from monolingual recordings and aggregates their frame-level activations into a code-switching prediction. With a small code-switching set used to tune a decision rule, the authors report the Whisper-small encoder reaches a mean F1 of 0.798 across 14 language pairs in CS-FLEURS, versus 0.776 for a CNN and 0.751 for a HuBERT linear baseline trained for monolingual language identification.
Apple Machine Learning Research · 🔥 0 - Apple study: bilingual data mixing beats hyperparameter tuning in data-scarce pre-training
Apple Machine Learning Research published a paper, "Mix, Don't Tune," comparing two remedies for data-constrained pre-training: aggressive hyperparameter tuning such as high weight decay, and mixing in data from a high-resource auxiliary language. Across roughly 1,000 pre-training runs at four model scales from 150M to 1.43B parameters, using Arabic as the low-resource target and English as the auxiliary, mixing improved validation loss and downstream task accuracy more than hyperparameter tuning, with the gap growing with model size. The authors report gains equivalent to 2–3× the unique target data on validation loss and 2–13× on downstream accuracy, and argue that target-language validation loss systematically underestimates the value of mixing.
Apple Machine Learning Research · 🔥 0 - Apple's SelfCompact lets models decide when to compact their context
Apple Machine Learning Research published a paper, "Self-Compacting Language Model Agents," proposing SELFCOMPACT, a scaffold in which the model itself invokes a compaction tool to summarize accumulated context, paired with a lightweight rubric specifying when to fire (a sub-task resolved, or the trajectory converging) and when to suppress (mid-derivation, or stuck). The authors report that both elements are needed — the tool alone is used unevenly across open-weight models, while the rubric alone cannot act — and that across six benchmarks (competitive math and agentic search) and seven models, SELFCOMPACT matches or exceeds fixed-interval summarization, beating a no-summarization baseline by up to 18.1 points on math and 5–9 points on agentic search at 30–70% lower per-question cost, with no fine-tuning or external supervision. Some authors carry a Johns Hopkins University affiliation, and the work is listed as a NeurIPS paper.
Apple Machine Learning Research · 🔥 0 - Apple's RISED uses rubrics for multi-environment agent training
Apple Machine Learning Research published RISED, a method that repurposes rubrics to guide online data selection and policy supervision when training a single LLM agent jointly across diverse interactive environments. An LLM judge tags each rollout with a rubric vocabulary shared across environments; the resulting profiles select data that matches the mixed-environment batch's behavioural composition while limiting overlap, positive rubrics provide privileged context for an on-policy self-distillation teacher's token-level supervision, and negative rubrics steer later rollouts away from recurring failure modes. The paper reports that across model backbones RISED achieves the highest mean pass rate across environments and ranks first or second in every individual environment.
Apple Machine Learning Research · 🔥 0
Experience and discussion from the community
Share my Apple Machine Learning Research experienceAsk about Apple Machine Learning Research
Nobody has shared their experience with Apple Machine Learning Research yet.