The Llama 4 Herd: Architecture, Training, Evaluation, and Deployment Notes

This document consolidates publicly reported technical details about Metas Llama 4 model family. It summarizes (i) released variants (Scout and Maverick) and the broader herd context including the previewed Behemoth teacher model, (ii) architectural characteristics beyond a high-level MoE description covering routed/shared-expert structure, early-fusion multimodality, and long-context design elements reported for Scout (iRoPE and length generalization strategies), (iii) training disclosures spanning pre-training, mid-training for long-context extension, and post-training methodology (lightweight SFT, online RL, and lightweight DPO) as described in release materials, (iv) developer-reported benchmark results for both base and instruction-tuned checkpoints, and (v) practical deployment constraints observed across major serving environments, including provider-specific context limits and quantization packaging. The manuscript also summarizes licensing obligations relevant to redistribution and derivative naming, and reviews publicly described safeguards and evaluation practices. The goal is to provide a compact technical reference for researchers and practitioners who need precise, source-backed facts about Llama 4.

Toward Large-Scale Photonics-Empowered AI Systems: From Physical Design Automation to System-Algorithm Co-Exploration

In this work, we identify three considerations that are essential for realizing practical photonic AI systems at scale: (1) dynamic tensor operation support for modern models rather than only weight-static kernels, especially for attention/Transformer-style workloads; (2) systematic management of conversion, control, and data-movement overheads, where multiplexing and dataflow must amortize electronic costs instead of letting ADC/DAC and I/O dominate; and (3) robustness under hardware non-idealities that become more severe as integration density grows. To study these coupled tradeoffs quantitatively, and to ensure they remain meaningful under real implementation constraints, we build a cross-layer toolchain that supports photonic AI design from early exploration to physical realization. SimPhony provides implementation-aware modeling and rapid cross-layer evaluation, translating physical costs into system-level metrics so architectural decisions are grounded in realistic assumptions. ADEPT and ADEPT-Z enable end-to-end circuit and topology exploration, connecting system objectives to feasible photonic fabrics under practical device and circuit constraints. Finally, Apollo and LiDAR provide scalable photonic physical design automation, turning candidate circuits into manufacturable layouts while accounting for routing, thermal, and crosstalk constraints.

Division-based Receiver-agnostic RFF Identification in WiFi Systems

In physical-layer security schemes, radio frequency fingerprint (RFF) identification of WiFi devices is susceptible to receiver differences, which can significantly degrade classification performance when a model is trained on one receiver but tested on another. In this paper, we propose a division-based receiver-agnostic RFF extraction method for WiFi systems, which removes the receivers' effects by dividing different preambles in the frequency domain. The proposed method requires only a single receiver for training and does not rely on additional calibration or stacking processes. First, for flat fading channel scenarios, the legacy short training field (L-STF) and legacy long training field (L-LTF) of the unknown device are divided by those of the reference device in the frequency domain. The receiver-dependent effects can be eliminated with the requirement of only a single receiver for training, and the higher-dimensional RFF features can be extracted. Second, for frequency-selective fading channel scenarios, the high-throughput long training field (HT-LTF) is divided by the L-LTF in the frequency domain. Only a single receiver is required for training and the higher-dimensional RFF features that are both channel-invariant and receiver-agnostic are extracted. Finally, simulation and experimental results demonstrate that the proposed method effectively mitigate the impacts of channel variations and receiver differences. The classification results show that, even when training on a single receiver and testing on a different one, the proposed method achieves classification accuracy improvements of 15.5% and 28.45% over the state-of-the-art approach in flat fading and frequency-selective fading channel scenarios, respectively.

Constructing Invariant and Equivariant Operations by Symmetric Tensor Network

Design of neural networks that incorporate symmetry is crucial for geometric deep learning. Central to this effort is the development of invariant and equivariant operations. This works presents a systematic method for constructing valid invariant and equivariant operations. It can handle inputs and outputs in the form of Cartesian tensors with different rank, as well as spherical tensors with different types. In addition, our method features a graphical representation utilizing the symmetric tensor network, which simplifies both the proofs and constructions related to invariant and equivariant functions. We also apply this approach to design the equivariant interaction message for the geometry graph neural network, and equivariant machine learning model to learn the constitutive law of materials.

Next Edit Prediction: Learning to Predict Code Edits from Context and Interaction History

The rapid advancement of large language models (LLMs) has led to the widespread adoption of AI-powered coding assistants integrated into a development environment. On one hand, low-latency code completion offers completion suggestions but is fundamentally constrained to the cursor's current position. On the other hand, chat-based editing can perform complex modifications, yet forces developers to stop their work, describe the intent in natural language, which causes a context-switch away from the code. This creates a suboptimal user experience, as neither paradigm proactively predicts the developer's next edit in a sequence of related edits. To bridge this gap and provide the seamless code edit suggestion, we introduce the task of Next Edit Prediction, a novel task designed to infer developer intent from recent interaction history to predict both the location and content of the subsequent edit. Specifically, we curate a high-quality supervised fine-tuning dataset and an evaluation benchmark for the Next Edit Prediction task. Then, we conduct supervised fine-tuning on a series of models and performed a comprehensive evaluation of both the fine-tuned models and other baseline models, yielding several novel findings. This work lays the foundation for a new interaction paradigm that proactively collaborate with developers by anticipating their following action, rather than merely reacting to explicit instructions.

A Physics-informed Deep Operator for Real-Time Freeway Traffic State Estimation

Traffic state estimation (TSE) falls methodologically into three categories: model-driven, data-driven, and model-data dual-driven. Model-driven TSE relies on macroscopic traffic flow models originated from hydrodynamics. Data-driven TSE leverages historical sensing data and employs statistical models or machine learning methods to infer traffic state. Model-data dual-driven traffic state estimation attempts to harness the strengths of both aspects to achieve more accurate TSE. From the perspective of mathematical operator theory, TSE can be viewed as a type of operator that maps available measurements of inerested traffic state into unmeasured traffic state variables in real time. For the first time this paper proposes to study real-time freeway TSE in the idea of physics-informed deep operator network (PI-DeepONet), which is an operator-oriented architecture embedding traffic flow models based on deep neural networks. The paper has developed an extended architecture from the original PI-DeepONet. The extended architecture is featured with: (1) the acceptance of 2-D data input so as to support CNN-based computations; (2) the introduction of a nonlinear expansion layer, an attention mechanism, and a MIMO mechanism; (3) dedicated neural network design for adaptive identification of traffic flow model parameters. A traffic state estimator built on the basis of this extended PI-DeepONet architecture was evaluated with respect to a short freeway stretch of NGSIM and a large-scale urban expressway in China, along with other four baseline TSE methods. The evaluation results demonstrated that this novel TSE method outperformed the baseline methods with high-precision estimation results of flow and mean speed.

Pangu Light: Weight Re-Initialization for Pruning and Accelerating LLMs

Large Language Models (LLMs) deliver state-of-the-art capabilities across numerous tasks, but their immense size and inference costs pose significant computational challenges for practical deployment. While structured pruning offers a promising avenue for model compression, existing methods often struggle with the detrimental effects of aggressive, simultaneous width and depth reductions, leading to substantial performance degradation. This paper argues that a critical, often overlooked, aspect in making such aggressive joint pruning viable is the strategic re-initialization and adjustment of remaining weights to improve the model post-pruning training accuracies. We introduce Pangu Light, a framework for LLM acceleration centered around structured pruning coupled with novel weight re-initialization techniques designed to address this ``missing piece''. Our framework systematically targets multiple axes, including model width, depth, attention heads, and RMSNorm, with its effectiveness rooted in novel re-initialization methods like Cross-Layer Attention Pruning (CLAP) and Stabilized LayerNorm Pruning (SLNP) that mitigate performance drops by providing the network a better training starting point. Further enhancing efficiency, Pangu Light incorporates specialized optimizations such as absorbing Post-RMSNorm computations and tailors its strategies to Ascend NPU characteristics. The Pangu Light models consistently exhibit a superior accuracy-efficiency trade-off, outperforming prominent baseline pruning methods like Nemotron and established LLMs like Qwen3 series. For instance, on Ascend NPUs, Pangu Light-32B's 81.6 average score and 2585 tokens/s throughput exceed Qwen3-32B's 80.9 average score and 2225 tokens/s.

Soft Weighted Machine Unlearning

Machine unlearning, as a post-hoc processing technique, has gained widespread adoption in addressing challenges like bias mitigation and robustness enhancement, colloquially, machine unlearning for fairness and robustness. However, existing non-privacy unlearning-based solutions persist in using binary data removal framework designed for privacy-driven motivation, leading to significant information loss, a phenomenon known as over-unlearning. While over-unlearning has been largely described in many studies as primarily causing utility degradation, we investigate its fundamental causes and provide deeper insights in this work through counterfactual leave-one-out analysis. In this paper, we introduce a weighted influence function that assigns tailored weights to each sample by solving a convex quadratic programming problem analytically. Building on this, we propose a soft-weighted framework enabling fine-grained model adjustments to address the over-unlearning challenge. We demonstrate that the proposed soft-weighted scheme is versatile and can be seamlessly integrated into most existing unlearning algorithms. Extensive experiments show that in fairness- and robustness-driven tasks, the soft-weighted scheme significantly outperforms hard-weighted schemes in fairness/robustness metrics and alleviates the decline in utility metric, thereby enhancing machine unlearning algorithm as an effective correction solution.

Biomed-DPT: Dual Modality Prompt Tuning for Biomedical Vision-Language Models

Prompt learning is one of the most effective paradigms for adapting pre-trained vision-language models (VLMs) to the biomedical image classification tasks in few shot scenarios. However, most of the current prompt learning methods only used the text prompts and ignored the particular structures (such as the complex anatomical structures and subtle pathological features) in the biomedical images. In this work, we propose Biomed-DPT, a knowledge-enhanced dual modality prompt tuning technique. In designing the text prompt, Biomed-DPT constructs a dual prompt including the template-driven clinical prompts and the large language model (LLM)-driven domain-adapted prompts, then extracts the clinical knowledge from the domain-adapted prompts through the knowledge distillation technique. In designing the vision prompt, Biomed-DPT introduces the zero vector as a soft prompt to leverage attention re-weighting so that the focus on non-diagnostic regions and the recognition of non-critical pathological features are avoided. Biomed-DPT achieves an average classification accuracy of 66.14\% across 11 biomedical image datasets covering 9 modalities and 10 organs, with performance reaching 78.06\% in base classes and 75.97\% in novel classes, surpassing the Context Optimization (CoOp) method by 6.20\%, 3.78\%, and 8.04\%, respectively. Our code are available at \underline{https://github.com/Kanyooo/Biomed-DPT}.

NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results

This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images. This challenge received a wide range of impressive solutions, which are developed and evaluated using our collected real-world Raindrop Clarity dataset. Unlike existing deraining datasets, our Raindrop Clarity dataset is more diverse and challenging in degradation types and contents, which includes day raindrop-focused, day background-focused, night raindrop-focused, and night background-focused degradations. This dataset is divided into three subsets for competition: 14,139 images for training, 240 images for validation, and 731 images for testing. The primary objective of this challenge is to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions. There are a total of 361 participants in the competition, and 32 teams submitting valid solutions and fact sheets for the final testing phase. These submissions achieved state-of-the-art (SOTA) performance on the Raindrop Clarity dataset. The project can be found at https://lixinustc.github.io/CVPR-NTIRE2025-RainDrop-Competition.github.io/.