Intrinsic dimension (ID) is an important tool in modern LLM analysis, informing studies of training dynamics, scaling behavior, and dataset structure, yet its textual determinants remain underexplored. We provide the first comprehensive study grounding ID in interpretable text properties through cross-encoder analysis, linguistic features, and sparse autoencoders (SAEs). In this work, we establish three key findings. First, ID is complementary to entropy-based metrics: after controlling for length, the two are uncorrelated, with ID capturing geometric complexity orthogonal to prediction quality. Second, ID exhibits robust genre stratification: scientific prose shows low ID (~8), encyclopedic content medium ID (~9), and creative/opinion writing high ID (~10.5) across all models tested. This reveals that contemporary LLMs find scientific text "representationally simple" while fiction requires additional degrees of freedom. Third, using SAEs, we identify causal features: scientific signals (formal tone, report templates, statistics) reduce ID; humanized signals (personalization, emotion, narrative) increase it. Steering experiments confirm these effects are causal. Thus, for contemporary models, scientific writing appears comparatively "easy", whereas fiction, opinion, and affect add representational degrees of freedom. Our multi-faceted analysis provides practical guidance for the proper use of ID and the sound interpretation of ID-based results.

Diffusion Large Language Models (dLLMs) are rapidly emerging alongside autoregressive models as a powerful paradigm for complex reasoning, with reinforcement learning increasingly used for downstream alignment. Existing trajectory-based RL methods uniformly allocate policy gradients across denoising steps, implicitly treating all steps as equally important. We challenge this assumption by analyzing trajectories with several step-level metrics: entropy-based uncertainty, Confidence-Margin (CM) uncertainty, and Rate of Entropy Change (RoEC). These reveal structured "zones of confusion": transient spikes in uncertainty and instability that strongly predict final success or failure, while most steps remain stable. We propose Adaptive Trajectory Policy Optimization (ATPO), a lightweight step-selection strategy that dynamically reallocates gradient updates to these high-leverage steps without changing the RL objective, rewards, or compute budget. Using a hybrid RoEC+CM rule, ATPO delivers substantial gains in reasoning accuracy and training stability across benchmarks, showing that exploiting trajectory dynamics is key to advancing dLLM RL.

Large Language Model (LLM)-based agents with function-calling capabilities are increasingly deployed, but remain vulnerable to Indirect Prompt Injection (IPI) attacks that hijack their tool calls. In response, numerous IPI-centric defense frameworks have emerged. However, these defenses are fragmented, lacking a unified taxonomy and comprehensive evaluation. In this Systematization of Knowledge (SoK), we present the first comprehensive analysis of IPI-centric defense frameworks. We introduce a comprehensive taxonomy of these defenses, classifying them along five dimensions. We then thoroughly assess the security and usability of representative defense frameworks. Through analysis of defensive failures in the assessment, we identify six root causes of defense circumvention. Based on these findings, we design three novel adaptive attacks that significantly improve attack success rates targeting specific frameworks, demonstrating the severity of the flaws in these defenses. Our paper provides a foundation and critical insights for the future development of more secure and usable IPI-centric agent defense frameworks.

This paper introduces SOLID (Synergizing Optimization and Large Language Models for Intelligent Decision-Making), a novel framework that integrates mathematical optimization with the contextual capabilities of large language models (LLMs). SOLID facilitates iterative collaboration between optimization and LLMs agents through dual prices and deviation penalties. This interaction improves the quality of the decisions while maintaining modularity and data privacy. The framework retains theoretical convergence guarantees under convexity assumptions, providing insight into the design of LLMs prompt. To evaluate SOLID, we applied it to a stock portfolio investment case with historical prices and financial news as inputs. Empirical results demonstrate convergence under various scenarios and indicate improved annualized returns compared to a baseline optimizer-only method, validating the synergy of the two agents. SOLID offers a promising framework for advancing automated and intelligent decision-making across diverse domains.

Knowledge Tracing (KT) aims to mine students' evolving knowledge states and predict their future question-answering performance. Existing methods based on heterogeneous information networks (HINs) are prone to introducing noises due to manual or random selection of meta-paths and lack necessary quality assessment of meta-path instances. Conversely, recent large language models (LLMs)- based methods ignore the rich information across students, and both paradigms struggle to deliver consistently accurate and evidence-based explanations. To address these issues, we propose an innovative framework, HIN-LLM Synergistic Enhanced Knowledge Tracing (HISE-KT), which seam-lessly integrates HINs with LLMs. HISE-KT first builds a multi-relationship HIN containing diverse node types to capture the structural relations through multiple meta-paths. The LLM is then employed to intelligently score and filter meta-path instances and retain high-quality paths, pioneering automated meta-path quality assessment. Inspired by educational psychology principles, a similar student retrieval mechanism based on meta-paths is designed to provide a more valuable context for prediction. Finally, HISE-KT uses a structured prompt to integrate the target student's history with the retrieved similar trajectories, enabling the LLM to generate not only accurate predictions but also evidence-backed, explainable analysis reports. Experiments on four public datasets show that HISE-KT outperforms existing KT baselines in both prediction performance and interpretability.

With nearly 1.5 billion people and more than 120 major languages, India represents one of the most diverse regions in the world. As multilingual Vision-Language Models (VLMs) gain prominence, robust evaluation methodologies are essential to drive progress toward equitable AI for low-resource languages. Current multilingual VLM evaluations suffer from four major limitations: reliance on unverified auto-translations, narrow task/domain coverage, limited sample sizes, and lack of cultural and natively sourced Question-Answering (QA). To address these gaps, we present a scalable framework to evaluate VLMs in Indian languages and compare it with performance in English. Using the framework, we generate HinTel-AlignBench, a benchmark that draws from diverse sources in Hindi and Telugu with English-aligned samples. Our contributions are threefold: (1) a semi-automated dataset creation framework combining back-translation, filtering, and human verification; (2) the most comprehensive vision-language benchmark for Hindi and and Telugu, including adapted English datasets (VQAv2, RealWorldQA, CLEVR-Math) and native novel Indic datasets (JEE for STEM, VAANI for cultural grounding) with approximately 4,000 QA pairs per language; and (3) a detailed performance analysis of various State-of-the-Art (SOTA) open-weight and closed-source VLMs. We find a regression in performance for tasks in English versus in Indian languages for 4 out of 5 tasks across all the models, with an average regression of 8.3 points in Hindi and 5.5 points for Telugu. We categorize common failure modes to highlight concrete areas of improvement in multilingual multimodal understanding.

Large Language Models (LLMs) are increasingly integrated into intelligent tutoring systems to provide human-like and adaptive instruction. However, most existing approaches fail to capture how students' knowledge evolves dynamically across their proficiencies, conceptual gaps, and forgetting patterns. This challenge is particularly acute in mathematics tutoring, where effective instruction requires fine-grained scaffolding precisely calibrated to each student's mastery level and cognitive retention. To address this issue, we propose T ASA (Teaching According to Students' Aptitude), a student-aware tutoring framework that integrates persona, memory, and forgetting dynamics for personalized mathematics learning. Specifically, T ASA maintains a structured student persona capturing proficiency profiles and an event memory recording prior learning interactions. By incorporating a continuous forgetting curve with knowledge tracing, T ASA dynamically updates each student's mastery state and generates contextually appropriate, difficulty-calibrated questions and explanations. Empirical results demonstrate that T ASA achieves superior learning outcomes and more adaptive tutoring behavior compared to representative baselines, underscoring the importance of modeling temporal forgetting and learner profiles in LLM-based tutoring systems.

Recently, large language models (LLMs) have been widely used as recommender systems, owing to their strong reasoning capability and their effectiveness in handling cold-start items. To better adapt LLMs for recommendation, retrieval-augmented generation (RAG) has been incorporated. Most existing RAG methods are user-based, retrieving purchase patterns of users similar to the target user and providing them to the LLM. In this work, we propose ItemRAG, an item-based RAG method for LLM-based recommendation that retrieves relevant items (rather than users) from item-item co-purchase histories. ItemRAG helps LLMs capture co-purchase patterns among items, which are beneficial for recommendations. Especially, our retrieval strategy incorporates semantically similar items to better handle cold-start items and uses co-purchase frequencies to improve the relevance of the retrieved items. Through extensive experiments, we demonstrate that ItemRAG consistently (1) improves the zero-shot LLM-based recommender by up to 43% in Hit-Ratio-1 and (2) outperforms user-based RAG baselines under both standard and cold-start item recommendation settings.

The reasoning capabilities of large language models (LLMs) have been significantly improved through reinforcement learning (RL). Nevertheless, LLMs still struggle to consistently verify their own reasoning traces. This raises the research question of how to enhance the self-verification ability of LLMs and whether such an ability can further improve reasoning performance. In this work, we propose GRPO-Verif, an algorithm that jointly optimizes solution generation and self-verification within a unified loss function, with an adjustable hyperparameter controlling the weight of the verification signal. Experimental results demonstrate that our method enhances self-verification capability while maintaining comparable performance in reasoning.

Abstract--Membership inference attacks (MIAs) on code completion models offer an effective way to assess privacy risks by inferring whether a given code snippet was part of the training data. Existing black-and gray-box MIAs rely on expensive surrogate models or manually crafted heuristic rules, which limit their ability to capture the nuanced memorization patterns exhibited by over-parameterized code language models. T o address these challenges, we propose AdvPrompt-MIA, a method specifically designed for code completion models, combining code-specific adversarial perturbations with deep learning. The core novelty of our method lies in designing a series of adversarial prompts that induce variations in the victim code model's output. By comparing these outputs with the ground-truth completion, we construct feature vectors to train a classifier that automatically distinguishes member from non-member samples. This design allows our method to capture richer memorization patterns and accurately infer training set membership. We conduct comprehensive evaluations on widely adopted models, such as Code Llama 7B, over the APPS and HumanEval benchmarks. The results show that our approach consistently outperforms state-of-the-art baselines, with AUC gains of up to 102%. In addition, our method exhibits strong transferability across different models and datasets, underscoring its practical utility and generalizability. Large language models (LLMs) have shown remarkable success in natural language processing by learning complex semantic and syntactic patterns from large-scale text corpora [1], [2]. This success has extended to the domain of source code, where code-specific LLMs (code LLMs) trained on billions of lines of code [3] now support tasks such as code completion [4], code summarization [5], [6], and vulnerability detection [7], and are integrated into tools like GitHub Copilot [8] and A WS CodeWhisperer [9]. Despite their impressive capabilities, code LLMs remain vulnerable to a variety of security and privacy threats, including adversarial perturbations [10], data poisoning [11], [12], and privacy leakage [13]-[15]. Among these, privacy leakage is particularly concerning due to its implications for sensitive information exposure and potential legal violations, often stemming from the memorization behavior of code LLMs [16]-[18].