HAWK: Head Importance-Aware Visual Token Pruning in Multimodal Models

In multimodal large language models (MLLMs), the surge of visual tokens significantly increases the inference time and computational overhead, making them impractical for real-time or resource-constrained applications. Visual token pruning is a promising strategy for reducing the cost of MLLM inference by removing redundant visual tokens. Existing research usually assumes that all attention heads contribute equally to the visual interpretation. However, our study reveals that different heads may capture distinct visual semantics and inherently play distinct roles in visual processing. In light of this observation, we propose HAWK, a head importance-aware visual token pruning method that perceives the varying importance of attention heads in visual tasks to maximize the retention of crucial tokens. By leveraging head importance weights and text-guided attention to assess visual token significance, HAWK effectively retains task-relevant visual tokens while removing redundant ones. The proposed HAWK is entirely training-free and can be seamlessly applied to various MLLMs. Extensive experiments on multiple mainstream vision-language benchmarks demonstrate that HAWK achieves state-of-the-art accuracy. When applied to Qwen2.5-VL, HAWK retains 96.0% of the original accuracy after pruning 80.2% of the visual tokens. Additionally, it reduces end-to-end latency to 74.4% of the original and further decreases GPU memory usage across the tested models. The code is available at https://github.com/peppery77/HAWK.git.

BATQuant: Outlier-resilient MXFP4 Quantization via Learnable Block-wise Optimization

Microscaling floating-point (MXFP) formats have emerged as a promising standard for deploying Multi-modal Large Language Models (MLLMs) and Large Language Models (LLMs) on modern accelerator architectures. However, existing Post-Training Quantization (PTQ) methods, particularly rotation-based techniques designed for integer formats, suffer from severe performance collapse when applied to MXFP4. Recent studies attribute this failure to a fundamental format mismatch: global orthogonal rotations inadvertently transfer outlier energy across quantization blocks, inducing new outliers that disrupt local block-wise scaling, while often creating bimodal activation distributions that underutilize the limited quantization range. To address these issues, we propose BATQuant (Block-wise Affine Transformation), which restricts transformations to align with MXFP granularity to prevent cross-block outlier propagation, while relaxing orthogonality constraints to optimize distribution shaping. To ensure parameter efficiency, we introduce Global and Private Kronecker (GPK) decomposition to effectively reduces storage and runtime overhead and incorporate Block-wise Learnable Clipping to suppress residual outliers. Extensive experiments on both MLLMs and LLMs demonstrate that BATQuant establishes new state-of-the-art results under aggressive W4A4KV16 configurations, recovering up to 96.43% of full-precision performance on multimodal benchmarks and clearly outperforming existing methods across diverse tasks.

FairFS: Addressing Deep Feature Selection Biases for Recommender System

Large-scale online marketplaces and recommender systems serve as critical technological support for e-commerce development. In industrial recommender systems, features play vital roles as they carry information for downstream models. Accurate feature importance estimation is critical because it helps identify the most useful feature subsets from thousands of feature candidates for online services. Such selection enables improved online performance while reducing computational cost. To address feature selection problems in deep learning, trainable gate-based and sensitivity-based methods have been proposed and proven effective in industrial practice. However, through the analysis of real-world cases, we identified three bias issues that cause feature importance estimation to rely on partial model layers, samples, or gradients, ultimately leading to inaccurate importance estimation. We refer to these as layer bias, baseline bias, and approximation bias. To mitigate these issues, we propose FairFS, a fair and accurate feature selection algorithm. FairFS regularizes feature importance estimated across all nonlinear transformation layers to address layer bias. It also introduces a smooth baseline feature close to the classifier decision boundary and adopts an aggregated approximation method to alleviate baseline and approximation biases. Extensive experiments demonstrate that FairFS effectively mitigates these biases and achieves state-of-the-art feature selection performance.

Unleashing Low-Bit Inference on Ascend NPUs: A Comprehensive Evaluation of HiFloat Formats

As LLMs scale, low-bit floating-point formats like MXFP and NVFP4 offer new opportunities for precision and efficiency. In this work, we evaluate HiFloat (HiF8 and HiF4), a family of formats tailored for Ascend NPUs. Through rigorous comparison across weight-activation and KV-cache tasks, we provide three key insights: (1) INT8 suits narrow-range data, while floating-point formats excel with high-variance data; (2) in 4-bit regimes, HiF4's hierarchical scaling prevents the accuracy collapse seen in integer formats; and (3) HiFloat is fully compatible with state-of-the-art post-training quantization frameworks. Overall, HiFloat provides a solution for high-efficiency LLM inference on NPUs.

MALLOC: Benchmarking the Memory-aware Long Sequence Compression for Large Sequential Recommendation

The scaling law, which indicates that model performance improves with increasing dataset and model capacity, has fueled a growing trend in expanding recommendation models in both industry and academia. However, the advent of large-scale recommenders also brings significantly higher computational costs, particularly under the long-sequence dependencies inherent in the user intent of recommendation systems. Current approaches often rely on pre-storing the intermediate states of the past behavior for each user, thereby reducing the quadratic re-computation cost for the following requests. Despite their effectiveness, these methods often treat memory merely as a medium for acceleration, without adequately considering the space overhead it introduces. This presents a critical challenge in real-world recommendation systems with billions of users, each of whom might initiate thousands of interactions and require massive memory for state storage. Fortunately, there have been several memory management strategies examined for compression in LLM, while most have not been evaluated on the recommendation task. To mitigate this gap, we introduce MALLOC, a comprehensive benchmark for memory-aware long sequence compression. MALLOC presents a comprehensive investigation and systematic classification of memory management techniques applicable to large sequential recommendations. These techniques are integrated into state-of-the-art recommenders, enabling a reproducible and accessible evaluation platform. Through extensive experiments across accuracy, efficiency, and complexity, we demonstrate the holistic reliability of MALLOC in advancing large-scale recommendation. Code is available at https://anonymous.4open.science/r/MALLOC.

Length-Adaptive Interest Network for Balancing Long and Short Sequence Modeling in CTR Prediction

User behavior sequences in modern recommendation systems exhibit significant length heterogeneity, ranging from sparse short-term interactions to rich long-term histories. While longer sequences provide more context, we observe that increasing the maximum input sequence length in existing CTR models paradoxically degrades performance for short-sequence users due to attention polarization and length imbalance in training data. To address this, we propose LAIN(Length-Adaptive Interest Network), a plug-and-play framework that explicitly incorporates sequence length as a conditioning signal to balance long- and short-sequence modeling. LAIN consists of three lightweight components: a Spectral Length Encoder that maps length into continuous representations, Length-Conditioned Prompting that injects global contextual cues into both long- and short-term behavior branches, and Length-Modulated Attention that adaptively adjusts attention sharpness based on sequence length. Extensive experiments on three real-world benchmarks across five strong CTR backbones show that LAIN consistently improves overall performance, achieving up to 1.15% AUC gain and 2.25% log loss reduction. Notably, our method significantly improves accuracy for short-sequence users without sacrificing longsequence effectiveness. Our work offers a general, efficient, and deployable solution to mitigate length-induced bias in sequential recommendation.

Benchmarking Post-Training Quantization of Large Language Models under Microscaling Floating Point Formats

Microscaling Floating-Point (MXFP) has emerged as a promising low-precision format for large language models (LLMs). Despite various post-training quantization (PTQ) algorithms being proposed, they mostly focus on integer quantization, while their applicability and behavior under MXFP formats remain largely unexplored. To address this gap, this work conducts a systematic investigation of PTQ under MXFP formats, encompassing over 7 PTQ algorithms, 15 evaluation benchmarks, and 3 LLM families. The key findings include: 1) MXFP8 consistently achieves near-lossless performance, while MXFP4 introduces substantial accuracy degradation and remains challenging; 2) PTQ effectiveness under MXFP depends strongly on format compatibility, with some algorithmic paradigms being consistently more effective than others; 3) PTQ performance exhibits highly consistent trends across model families and modalities, in particular, quantization sensitivity is dominated by the language model rather than the vision encoder in multimodal LLMs; 4) The scaling factor of quantization is a critical error source in MXFP4, and a simple pre-scale optimization strategy can significantly mitigate its impact. Together, these results provide practical guidance on adapting existing PTQ methods to MXFP quantization.

SwiftMem: Fast Agentic Memory via Query-aware Indexing

Agentic memory systems have become critical for enabling LLM agents to maintain long-term context and retrieve relevant information efficiently. However, existing memory frameworks suffer from a fundamental limitation: they perform exhaustive retrieval across the entire storage layer regardless of query characteristics. This brute-force approach creates severe latency bottlenecks as memory grows, hindering real-time agent interactions. We propose SwiftMem, a query-aware agentic memory system that achieves sub-linear retrieval through specialized indexing over temporal and semantic dimensions. Our temporal index enables logarithmic-time range queries for time-sensitive retrieval, while the semantic DAG-Tag index maps queries to relevant topics through hierarchical tag structures. To address memory fragmentation during growth, we introduce an embedding-tag co-consolidation mechanism that reorganizes storage based on semantic clusters to improve cache locality. Experiments on LoCoMo and LongMemEval benchmarks demonstrate that SwiftMem achieves 47$\times$ faster search compared to state-of-the-art baselines while maintaining competitive accuracy, enabling practical deployment of memory-augmented LLM agents.

What Matters For Safety Alignment?

This paper presents a comprehensive empirical study on the safety alignment capabilities. We evaluate what matters for safety alignment in LLMs and LRMs to provide essential insights for developing more secure and reliable AI systems. We systematically investigate and compare the influence of six critical intrinsic model characteristics and three external attack techniques. Our large-scale evaluation is conducted using 32 recent, popular LLMs and LRMs across thirteen distinct model families, spanning a parameter scale from 3B to 235B. The assessment leverages five established safety datasets and probes model vulnerabilities with 56 jailbreak techniques and four CoT attack strategies, resulting in 4.6M API calls. Our key empirical findings are fourfold. First, we identify the LRMs GPT-OSS-20B, Qwen3-Next-80B-A3B-Thinking, and GPT-OSS-120B as the top-three safest models, which substantiates the significant advantage of integrated reasoning and self-reflection mechanisms for robust safety alignment. Second, post-training and knowledge distillation may lead to a systematic degradation of safety alignment. We thus argue that safety must be treated as an explicit constraint or a core optimization objective during these stages, not merely subordinated to the pursuit of general capability. Third, we reveal a pronounced vulnerability: employing a CoT attack via a response prefix can elevate the attack success rate by 3.34x on average and from 0.6% to 96.3% for Seed-OSS-36B-Instruct. This critical finding underscores the safety risks inherent in text-completion interfaces and features that allow user-defined response prefixes in LLM services, highlighting an urgent need for architectural and deployment safeguards. Fourth, roleplay, prompt injection, and gradient-based search for adversarial prompts are the predominant methodologies for eliciting unaligned behaviors in modern models.

Towards Efficient Agents: A Co-Design of Inference Architecture and System

The rapid development of large language model (LLM)-based agents has unlocked new possibilities for autonomous multi-turn reasoning and tool-augmented decision-making. However, their real-world deployment is hindered by severe inefficiencies that arise not from isolated model inference, but from the systemic latency accumulated across reasoning loops, context growth, and heterogeneous tool interactions. This paper presents AgentInfer, a unified framework for end-to-end agent acceleration that bridges inference optimization and architectural design. We decompose the problem into four synergistic components: AgentCollab, a hierarchical dual-model reasoning framework that balances large- and small-model usage through dynamic role assignment; AgentSched, a cache-aware hybrid scheduler that minimizes latency under heterogeneous request patterns; AgentSAM, a suffix-automaton-based speculative decoding method that reuses multi-session semantic memory to achieve low-overhead inference acceleration; and AgentCompress, a semantic compression mechanism that asynchronously distills and reorganizes agent memory without disrupting ongoing reasoning. Together, these modules form a Self-Evolution Engine capable of sustaining efficiency and cognitive stability throughout long-horizon reasoning tasks. Experiments on the BrowseComp-zh and DeepDiver benchmarks demonstrate that through the synergistic collaboration of these methods, AgentInfer reduces ineffective token consumption by over 50%, achieving an overall 1.8-2.5 times speedup with preserved accuracy. These results underscore that optimizing for agentic task completion-rather than merely per-token throughput-is the key to building scalable, efficient, and self-improving intelligent systems.