Peng Zhang

Jing Yao, Xiaoyuan Yi, Shitong Duan et al.

As Large Language Models (LLMs) achieve remarkable breakthroughs, aligning their values with humans has become imperative for their responsible development and customized applications. However, there still lack evaluations of LLMs values that fulfill three desirable goals. (1) Value Clarification: We expect to clarify the underlying values of LLMs precisely and comprehensively, while current evaluations focus narrowly on safety risks such as bias and toxicity. (2) Evaluation Validity: Existing static, open-source benchmarks are prone to data contamination and quickly become obsolete as LLMs evolve. Additionally, these discriminative evaluations uncover LLMs' knowledge about values, rather than valid assessments of LLMs' behavioral conformity to values. (3) Value Pluralism: The pluralistic nature of human values across individuals and cultures is largely ignored in measuring LLMs value alignment. To address these challenges, we presents the Value Compass Benchmarks, with three correspondingly designed modules. It (i) grounds the evaluation on motivationally distinct \textit{basic values to clarify LLMs' underlying values from a holistic view; (ii) applies a \textit{generative evolving evaluation framework with adaptive test items for evolving LLMs and direct value recognition from behaviors in realistic scenarios; (iii) propose a metric that quantifies LLMs alignment with a specific value as a weighted sum over multiple dimensions, with weights determined by pluralistic values.

Ethan Li, Anders Boesen Lindbo Larsen, Chen Zhang et al. · apple-ml, cmu

We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon through architectural innovations such as KV-cache sharing and 2-bit quantization-aware training; and ii a scalable server model built on a novel Parallel-Track Mixture-of-Experts PT-MoE transformer that combines track parallelism, mixture-of-experts sparse computation, and interleaved global-local attention to deliver high quality with competitive cost on Apple's Private Cloud Compute platform. Both models are trained on large-scale multilingual and multimodal datasets sourced via responsible web crawling, licensed corpora, and high-quality synthetic data, then further refined with supervised fine-tuning and reinforcement learning on a new asynchronous platform. The resulting models support several additional languages while understanding images and executing tool calls. In public benchmarks and human evaluations, both the server model and the on-device model match or surpass comparably sized open baselines. A new Swift-centric Foundation Models framework exposes guided generation, constrained tool calling, and LoRA adapter fine-tuning, allowing developers to integrate these capabilities with a few lines of code. The latest advancements in Apple Intelligence models are grounded in our Responsible AI approach with safeguards like content filtering and locale-specific evaluation, as well as our commitment to protecting our users' privacy with innovations like Private Cloud Compute.

Asif Iqbal, John Verboncoeur, Peng Zhang

Multipactor is a nonlinear electron avalanche phenomenon that can severely impair the performance of high-power radio frequency (RF) devices and accelerator systems. Accurate prediction of multipactor susceptibility across different materials and operational regimes remains a critical yet computationally intensive challenge in accelerator component design and RF engineering. This study presents the first application of supervised machine learning (ML) for predicting multipactor susceptibility in two-surface planar geometries. A simulation-derived dataset spanning six distinct secondary electron yield (SEY) material profiles is used to train regression models - including Random Forest (RF), Extra Trees (ET), Extreme Gradient Boosting (XGBoost), and funnel-structured Multilayer Perceptrons (MLPs) - to predict the time-averaged electron growth rate, $δ_{avg}$. Performance is evaluated using Intersection over Union (IoU), Structural Similarity Index (SSIM), and Pearson correlation coefficient. Tree-based models consistently outperform MLPs in generalizing across disjoint material domains. MLPs trained using a scalarized objective function that combines IoU and SSIM during Bayesian hyperparameter optimization with 5-fold cross-validation outperform those trained with single-objective loss functions. Principal Component Analysis reveals that performance degradation for certain materials stems from disjoint feature-space distributions, underscoring the need for broader dataset coverage. This study demonstrates both the promise and limitations of ML-based multipactor prediction and lays the groundwork for accelerated, data-driven modeling in advanced RF and accelerator system design.