Large Language Models (LLMs) have transformed artificial intelligence, offering profound opportunities for educational applications. However, their ability to provide fine-grained educational feedback for K-12 English writing remains underexplored. In this paper, we challenge the error analysis and pedagogical skills of LLMs by introducing the problem of Fine-grained Error Analysis for English Learners and present the Fine-grained Error ANalysis for English Learners (FEANEL) Benchmark. The benchmark comprises 1,000 essays written by elementary and secondary school students, and a well-developed English writing error taxonomy. Each error is annotated by language education experts and categorized by type, severity, and explanatory feedback, using a part-of-speech-based taxonomy they co-developed. We evaluate state-of-the-art LLMs on the FEANEL Benchmark to explore their error analysis and pedagogical abilities. Experimental results reveal significant gaps in current LLMs' ability to perform fine-grained error analysis, highlighting the need for advancements in particular methods for educational applications.

Low-Rank Adaptation (LoRA) has become the de facto method for parameter-efficient fine-tuning of large language models (LLMs), enabling rapid adaptation to diverse domains. In production, LoRA-based models are served at scale, creating multi-tenant environments with hundreds of adapters sharing a base model. However, state-of-the-art serving systems co-batch heterogeneous adapters without accounting for rank (size) variability, leading to severe performance skew, which ultimately requires adding more GPUs to satisfy service-level objectives (SLOs). Existing optimizations, focused on loading, caching, and kernel execution, ignore this heterogeneity, leaving GPU resources underutilized. We present LoRAServe, a workload-aware dynamic adapter placement and routing framework designed to tame rank diversity in LoRA serving. By dynamically rebalancing adapters across GPUs and leveraging GPU Direct RDMA for remote access, LoRAServe maximizes throughput and minimizes tail latency under real-world workload drift. Evaluations on production traces from Company X show that LoRAServe elicits up to 2$\times$ higher throughput, up to 9$\times$ lower TTFT, while using up to 50% fewer GPUs under SLO constraints compared to state-of-the-art systems.

We introduce JBE-QA, a Japanese Bar Exam Question-Answering dataset to evaluate large language models' legal knowledge. Derived from the multiple-choice (tanto-shiki) section of the Japanese bar exam (2015-2024), JBE-QA provides the first comprehensive benchmark for Japanese legal-domain evaluation of LLMs. It covers the Civil Code, the Penal Code, and the Constitution, extending beyond the Civil Code focus of prior Japanese resources. Each question is decomposed into independent true/false judgments with structured contextual fields. The dataset contains 3,464 items with balanced labels. We evaluate 26 LLMs, including proprietary, open-weight, Japanese-specialised, and reasoning models. Our results show that proprietary models with reasoning enabled perform best, and the Constitution questions are generally easier than the Civil Code or the Penal Code questions.

This study discusses the essential components that a Retrieval-Augmented Generation (RAG)-based LLM system should possess in order to support Japanese medical litigation procedures complying with legal norms. In litigation, expert commissioners, such as physicians, architects, accountants, and engineers, provide specialized knowledge to help judges clarify points of dispute. When considering the substitution of these expert roles with a RAG-based LLM system, the constraint of strict adherence to legal norms is imposed. Specifically, three requirements arise: (1) the retrieval module must retrieve appropriate external knowledge relevant to the disputed issues in accordance with the principle prohibiting the use of private knowledge, (2) the responses generated must originate from the context provided by the RAG and remain faithful to that context, and (3) the retrieval module must reference external knowledge with appropriate timestamps corresponding to the issues at hand. This paper discusses the design of a RAG-based LLM system that satisfies these requirements.

When people reason about cause and effect, they often consider many competing "what if" scenarios before deciding which explanation fits best. Analogously, advanced language models capable of causal inference can consider multiple interventions and counterfactuals to judge the validity of causal claims. Crucially, this type of reasoning is less like a single calculation and more like an internal dialogue between alternative hypotheses. In this paper, we make this dialogue explicit through a dual-agent debate framework where one model provides a structured causal inference, and the other critically examines this reasoning for logical flaws. When disagreements arise, agents attempt to persuade each other, challenging each other's logic and revising their conclusions until they converge on a mutually agreed answer. To take advantage of this deliberative process, we specifically use reasoning language models, whose strengths in both causal inference and adversarial debate remain under-explored relative to standard large language models. We evaluate our approach on the CLadder dataset, a benchmark linking natural language questions to formally defined causal graphs across all three rungs of Pearl's ladder of causation. With Qwen3 and DeepSeek-R1 as debater agents, we demonstrate that multi-agent debate improves DeepSeek-R1's overall accuracy in causal inference from 78.03% to 87.45%, with the counterfactual category specifically improving from 67.94% to 80.04% accuracy. Similarly, Qwen3's overall accuracy improves from 84.16% to 89.41%, and counterfactual questions from 71.53% to 80.35%, showing that strong models can still benefit greatly from debate with weaker agents. Our results highlight the potential of reasoning models as building blocks for multi-agent systems in causal inference, and demonstrate the importance of diverse perspectives in causal problem-solving.

Recent advances in sequence modeling have introduced selective SSMs as promising alternatives to Transformer architectures, offering theoretical computational efficiency and sequence processing advantages. A comprehensive understanding of selective SSMs in runtime behavior, resource utilization patterns, and scaling characteristics still remains unexplored, thus obstructing their optimal deployment and further architectural improvements. This paper presents a thorough empirical study of Mamba-1 and Mamba-2, systematically profiled for performance to assess the design principles that contribute to their efficiency in state-space modeling. A detailed analysis of computation patterns, memory access, I/O characteristics, and scaling properties was performed for sequence lengths ranging from 64 to 16384 tokens. Our findings show that the SSM component, a central part of the selective SSM architecture, demands a significant portion of computational resources compared to other components in the Mamba block. Based on these insights, we propose a pruning technique that selectively removes low-activity states within the SSM component, achieving measurable throughput and memory gains while maintaining accuracy within a moderate pruning regime. This approach results in performance improvements across varying sequence lengths, achieving a 1.14x speedup and reducing memory usage by 11.50\%. These results offer valuable guidance for designing more efficient SSM architectures that can be applied to a wide range of real-world applications.

Recent advancements in large language models (LLMs) have shown their potential across both general and domain-specific tasks. However, there is a growing concern regarding their lack of sensitivity, factual incorrectness in responses, inconsistent expressions of empathy, bias, hallucinations, and overall inability to capture the depth and complexity of human understanding, especially in low-resource and sensitive domains such as psychology. To address these challenges, our study employs a mixed-methods approach to evaluate the efficacy of LLMs in psychotherapy. We use LLMs to generate precise summaries of motivational interviewing (MI) dialogues and design a two-stage annotation scheme based on key components of the Motivational Interviewing Treatment Integrity (MITI) framework, namely evocation, collaboration, autonomy, direction, empathy, and a non-judgmental attitude. Using expert-annotated MI dialogues as ground truth, we formulate multi-class classification tasks to assess model performance under progressive prompting techniques, incorporating one-shot and few-shot prompting. Our results offer insights into LLMs' capacity for understanding complex psychological constructs and highlight best practices to mitigate ``semantic drift" in therapeutic settings. Our work contributes not only to the MI community by providing a high-quality annotated dataset to address data scarcity in low-resource domains but also critical insights for using LLMs for precise contextual interpretation in complex behavioral therapy.

While Multimodal Large Language Models (MLLMs) are adept at answering what is in an image-identifying objects and describing scenes-they often lack the ability to understand how an image feels to a human observer. This gap is most evident when considering subjective cognitive properties, such as what makes an image memorable, funny, aesthetically pleasing, or emotionally evocative. To systematically address this challenge, we introduce CogIP-Bench, a comprehensive benchmark for evaluating MLLMs on such image cognitive properties. Our evaluation reveals a significant gap: current models are poorly aligned with human perception of these nuanced properties. We then demonstrate that a post-training phase can effectively bridge this gap, significantly enhancing the model's alignment with human judgments. Furthermore, we show that this learned cognitive alignment is not merely predictive but also transferable to downstream creative tasks. By integrating our cognitively-aligned MLLM into an image generation pipeline, we can guide the synthesis process to produce images that better embody desired traits, such as being more memorable or visually appealing. Our work provides a benchmark to measure this human-like perception, a post-training pipeline to enhance it, and a demonstration that this alignment unlocks more human-centric AI.

Large Language Models (LLMs) demonstrate impressive capabilities in natural language understanding and generation, but incur high communication overhead and privacy risks in cloud deployments, while facing compute and memory constraints when confined to edge devices. Cloud-edge inference has emerged as a promising paradigm for improving privacy in LLM services by retaining sensitive computations on local devices. However, existing cloud-edge inference approaches apply uniform privacy protection without considering input sensitivity, resulting in unnecessary perturbation and degraded utility even for non-sensitive tokens. To address this limitation, we propose Privacy-aware Routing for Inference with Semantic Modulation (PRISM), a context-aware framework that dynamically balances privacy and inference quality. PRISM executes in four stages: (1) the edge device profiles entity-level sensitivity; (2) a soft gating module, also on the edge, selects an execution mode -cloud, edge, or collaboration; (3) for collaborative paths, the edge applies adaptive two-layer local differential privacy based on entity risks; and (4) the cloud LLM generates a semantic sketch from the perturbed prompt, which is then refined by the edge-side small language model (SLM) using local context. Our results show that PRISM consistently achieves superior privacy-utility trade-offs in various scenarios, reducing energy consumption and latency to 40-50% of baseline methods such as Uniform and Selective LDP, while maintaining high output quality under strong privacy constraints. These findings are validated through comprehensive evaluations involving realistic prompts, actual energy measurements, and heterogeneous cloud-edge model deployments.

The past decade has seen a massive rise in the popularity of AI systems, mainly owing to the developments in Gen AI, which has revolutionized numerous industries and applications. However, this progress comes at a considerable cost to the environment as training and deploying these models consume significant computational resources and energy and are responsible for large carbon footprints in the atmosphere. In this paper, we study the amount of carbon dioxide released by models across different domains over varying time periods. By examining parameters such as model size, repository activity (e.g., commits and repository age), task type, and organizational affiliation, we identify key factors influencing the environmental impact of AI development. Our findings reveal that model size and versioning frequency are strongly correlated with higher emissions, while domain-specific trends show that NLP models tend to have lower carbon footprints compared to audio-based systems. Organizational context also plays a significant role, with university-driven projects exhibiting the highest emissions, followed by non-profits and companies, while community-driven projects show a reduction in emissions. These results highlight the critical need for green AI practices, including the adoption of energy-efficient architectures, optimizing development workflows, and leveraging renewable energy sources. We also discuss a few practices that can lead to a more sustainable future with AI, and we end this paper with some future research directions that could be motivated by our work. This work not only provides actionable insights to mitigate the environmental impact of AI but also poses new research questions for the community to explore. By emphasizing the interplay between sustainability and innovation, our study aims to guide future efforts toward building a more ecologically responsible AI ecosystem.