Deploying large language models (LLMs) for structured data extraction in domains such as financial compliance reporting, legal document analytics, and multilingual knowledge base construction is often impractical for smaller teams due to the high cost of running large architectures and the difficulty of preparing large, high-quality datasets. Most recent instruction-tuning studies focus on seven-billion-parameter or larger models, leaving limited evidence on whether much smaller models can work reliably under low-resource, multi-task conditions. This work presents ETLCH, a billion-parameter LLaMA-based model fine-tuned with low-rank adaptation on only a few hundred to one thousand samples per task for JSON extraction, knowledge graph extraction, and named entity recognition. Despite its small scale, ETLCH outperforms strong baselines across most evaluation metrics, with substantial gains observed even at the lowest data scale. These findings demonstrate that well-tuned small models can deliver stable and accurate structured outputs at a fraction of the computational cost, enabling cost-effective and reliable information extraction pipelines in resource-constrained environments.

Modern Large Language Models (LLMs) are excellent at generating synthetic data. However, their performance in sensitive domains such as text detoxification has not received proper attention from the scientific community. This paper explores the possibility of using LLM-generated synthetic toxic data as an alternative to human-generated data for training models for detoxification. Using Llama 3 and Qwen activation-patched models, we generated synthetic toxic counterparts for neutral texts from ParaDetox and SST-2 datasets. Our experiments show that models fine-tuned on synthetic data consistently perform worse than those trained on human data, with a drop in performance of up to 30% in joint metrics. The root cause is identified as a critical lexical diversity gap: LLMs generate toxic content using a small, repetitive vocabulary of insults that fails to capture the nuances and variety of human toxicity. These findings highlight the limitations of current LLMs in this domain and emphasize the continued importance of diverse, human-annotated data for building robust detoxification systems.

Mixture-of-Experts (MoE) has emerged as a promising architecture for modern large language models (LLMs). However, massive parameters impose heavy GPU memory (i.e., VRAM) demands, hindering the widespread adoption of MoE LLMs. Offloading the expert parameters to CPU RAM offers an effective way to alleviate the VRAM requirements for MoE inference. Existing approaches typically cache a small subset of experts in VRAM and dynamically prefetch experts from RAM during inference, leading to significant degradation in inference speed due to the poor cache hit rate and substantial expert loading latency. In this work, we propose MoEpic, an efficient MoE inference system with a novel expert split mechanism. Specifically, each expert is vertically divided into two segments: top and bottom. MoEpic caches the top segment of hot experts, so that more experts will be stored under the limited VRAM budget, thereby improving the cache hit rate. During each layer's inference, MoEpic predicts and prefetches the activated experts for the next layer. Since the top segments of cached experts are exempt from fetching, the loading time is reduced, which allows efficient transfer-computation overlap. Nevertheless, the performance of MoEpic critically depends on the cache configuration (i.e., each layer's VRAM budget and expert split ratio). To this end, we propose a divide-and-conquer algorithm based on fixed-point iteration for adaptive cache configuration. Extensive experiments on popular MoE LLMs demonstrate that MoEpic can save about half of the GPU cost, while lowering the inference latency by about 37.51%-65.73% compared to the baselines.

Accurate and early diagnosis of malignant melanoma is critical for improving patient outcomes. While convolutional neural networks (CNNs) have shown promise in dermoscopic image analysis, they often neglect clinical metadata and require extensive preprocessing. Vision-language models (VLMs) offer a multimodal alternative but struggle to capture clinical specificity when trained on general-domain data. To address this, we propose a retrieval-augmented VLM framework that incorporates semantically similar patient cases into the diagnostic prompt. Our method enables informed predictions without fine-tuning and significantly improves classification accuracy and error correction over conventional baselines. These results demonstrate that retrieval-augmented prompting provides a robust strategy for clinical decision support.

Existing key-value (KV) cache compression methods typically rely on heuristics, such as uniform cache allocation across layers or static eviction policies, however, they ignore the critical interplays among layer-specific feature patterns and task performance, which can lead to degraded generalization. In this paper, we propose EvolKV, an adaptive framework for layer-wise, task-driven KV cache compression that jointly optimizes the memory efficiency and task performance. By reformulating cache allocation as a multi-objective optimization problem, EvolKV leverages evolutionary search to dynamically configure layer budgets while directly maximizing downstream performance. Extensive experiments on 11 tasks demonstrate that our approach outperforms all baseline methods across a wide range of KV cache budgets on long-context tasks and surpasses heuristic baselines by up to 7 percentage points on GSM8K. Notably, EvolKV achieves superior performance over the full KV cache setting on code completion while utilizing only 1.5% of the original budget, suggesting the untapped potential in learned compression strategies for KV cache budget allocation.

Abstract--The evolution toward Level 4 (L4) Autonomous Networks (AN) represents a strategic inflection point in telecommunications, where networks must transcend reactive automation to achieve genuine cognitive capabilities--fulfilling AN's vision of self-configuring, self-healing, and self-optimizing systems that deliver zero-wait, zero-touch, and zero-fault services. This work bridges the gap between architectural theory and operational reality by implementing Joseph Sifakis's AN Agent reference architecture in a functional cognitive system, deploying coordinated proactive-reactive runtimes driven by hybrid knowledge representation. Specifically, the system demonstrates sub-10 ms real-time control in 5G NR sub-6 GHz environments. Empirical results show a 4% increase in downlink throughput over Outer Loop Link Adaptation (OLLA) algorithms for enhanced mobile broadband (eMBB). Furthermore, for the ultra-reliable low-latency communication (URLLC) scenario, the agent achieves an 85% reduction in Block Error Rate (BLER). These improvements confirm the architecture's viability in overcoming traditional autonomy barriers and advancing critical L4-enabling capabilities toward next-generation objectives. UTONOMOUS Networks (AN), a purpose-specific telecommunications technology pioneered by the TM Forum (TMF) in 2019, target networks with intrinsic self-configuration, self-healing, and self-optimization capabilities--collectively termed the Three-Self Capabilities [1]. These fundamental properties enable the realization of zero-wait, zero-touch, and zero-fault network services, known as the Three-Zero Objectives, which collectively deliver optimal user experiences while maximizing resource utilization throughout the entire network lifecycle. By strategically integrating emerging general-purpose technologies including artificial intelligence (AI), digital twins, and big data analytics, AN not only transforms conventional network operations but fundamentally reorients value creation paradigms from traditional device-centric and management-centric models toward customer-oriented, service-driven, and business-focused frameworks.

Foundation models are revolutionizing autonomous driving perception, transitioning the field from narrow, task-specific deep learning models to versatile, general-purpose architectures trained on vast, diverse datasets. This survey examines how these models address critical challenges in autonomous perception, including limitations in generalization, scalability, and robustness to distributional shifts. The survey introduces a novel taxonomy structured around four essential capabilities for robust performance in dynamic driving environments: generalized knowledge, spatial understanding, multi-sensor robustness, and temporal reasoning. For each capability, the survey elucidates its significance and comprehensively reviews cutting-edge approaches. Diverging from traditional method-centric surveys, our unique framework prioritizes conceptual design principles, providing a capability-driven guide for model development and clearer insights into foundational aspects. We conclude by discussing key challenges, particularly those associated with the integration of these capabilities into real-time, scalable systems, and broader deployment challenges related to computational demands and ensuring model reliability against issues like hallucinations and out-of-distribution failures. The survey also outlines crucial future research directions to enable the safe and effective deployment of foundation models in autonomous driving systems.

Deep learning-based Automatic Modulation Recognition (AMR) model has made significant progress with the support of large-scale labeled data. However, when developing new models or performing hyperparameter tuning, the time and energy consumption associated with repeated training using massive amounts of data are often unbearable. To address the above challenges, we propose \emph{FoQuS}, which approximates the effect of full training by selecting a coreset from the original dataset, thereby significantly reducing training overhead. Specifically, \emph{FoQuS} records the prediction trajectory of each sample during full-dataset training and constructs three importance metrics based on training dynamics. Experiments show that \emph{FoQuS} can maintain high recognition accuracy and good cross-architecture generalization on multiple AMR datasets using only 1\%-30\% of the original data.

Audio and music generation systems have been remarkably developed in the music information retrieval (MIR) research field. The advancement of these technologies raises copyright concerns, as ownership and authorship of AI-generated music (AIGM) remain unclear. Also, it can be difficult to determine whether a piece was generated by AI or composed by humans clearly. To address these challenges, we aim to improve the accuracy of AIGM detection by analyzing the structural patterns of music segments. Specifically, to extract musical features from short audio clips, we integrated various pre-trained models, including self-supervised learning (SSL) models or an audio effect encoder, each within our suggested transformer-based framework. Furthermore, for long audio, we developed a segment transformer that divides music into segments and learns inter-segment relationships. We used the FakeMusicCaps and SONICS datasets, achieving high accuracy in both the short-audio and full-audio detection experiments. These findings suggest that integrating segment-level musical features into long-range temporal analysis can effectively enhance both the performance and robustness of AIGM detection systems.

In recent years, VLMs have captured the imagination of the Artificial Intelligence(AI) community, demonstrating an impressive ability to interpret, reason about, and generate content that covers both text and image handling. From answering questions about visual scenes to engaging in multi-modal dialogue, models such as Flamingo [1], PaLI [25], and BLIP-2 [14] are redefining the frontier of vision intelligence. Y et, as these models are widening their application capabilities, a fundamental question emerges: can they truly reason, or are they sophisticated pattern matchers? To explore this question, we turn to the domain of physics--a field that serves as a universal benchmark for logical thoughts of a human being. Physics problems are an ideal testbed for VLMs, as they are multi-modal, combining textual descriptions, mathematical equations, and often clarifying diagrams. A model that can successfully solve these problems must not only understand language and images but also grasp the underlying relationships and principles that govern the physical realm. The challenge, uptil now, has been the lack of accessible tools for this kind of evaluation. Existing benchmarks for scientific reasoning, such as ARC [7] and ScienceQA [17], are often limited to basic text-only question sets, while those that incorporate visual elements, like MathVista [18], frequently depend on complex physics simulators that are computationally expensive for many researchers to deploy, thereby restricting reproducibility.