Large language models (LLMs) are increasingly used in educational settings where decisions require factual, pedagogical, and value judgments. Existing benchmarks assess general value alignment or educational capability, but rarely evaluate whether model decisions are consistent with a specified value framework in situated educational contexts. We present EduValueBench, an expert-led benchmark of 1,696 decision-oriented items with expert reference decisions and rationales, covering 12 core values in China, seven educational scenario families, and three challenge mechanisms. We evaluated 15 LLMs from China-based providers using a multi-LLM evaluator panel. Target LLMs received only the educational question, without value labels or expert references. Against the expert-authored reference decisions, model scores range from 24.7% to 54.1%, while value- and scenario-level profiles differentiate similarly ranked models. Case analyses show that responses may be fluent and seemingly reasonable yet still optimize a different educational objective or overlook ethical boundaries. Among responses with jointly valid labels, pairwise evaluator agreement ranges from 88.2% to 92.0%. EduValueBench provides a culturally situated diagnostic benchmark for comparing unguided model decisions with expert-authored educational references. Code and data are available at https://github.com/tjunlp-lab/EduValueBench.

Current value alignment evaluations for large language models (LLMs) heavily rely on output compliance, overlooking whether LLMs maintain value consistency when alternating between generator and evaluator roles. Such consistency is increasingly important because modern LLM agents and alignment pipelines often rely on the same LLM to both generate actions and evaluate their acceptability. To mitigate this limitation, we propose VALCON, a role-driven data construction framework that operationalizes value consistency into three benchmark suites (VAB-Consistency, VAB-Boundary, VAB-Challenge) via dual-role variation. By systematically pairing an LLM’s generative behavior with its evaluative judgments under identical principles, VALCON synthesizes target evaluation data that exposes hidden value discrepancies. Experiments on 28 LLMs reveal pervasive cross-role inconsistency: models with highly compliant outputs frequently fail to align their judgments with their own generations. To understand and address this gap, we perform mechanistic probing that uncovers a role-dependent representational decoupling associated with this behavioral asymmetry. Guided by these insights, we propose a representation-directed intervention that effectively mitigates consistency failures. Our work establishes value consistency as a distinct alignment dimension and provides a complete evaluation-analysis-mitigation pipeline. Code and data are available at https://github.com/tjunlp-lab/VALCON.

As large language models (LLMs) are increasingly deployed globally, aligning them with pluralistic human values remains a challenge, as current efforts are predominantly Western-focused. To bridge this gap, we introduce SEAPlur, a dataset designed to benchmark and improve LLM alignment with the pluralistic values of four Southeast Asian (SEA) countries: Indonesia, Laos, Myanmar, and Vietnam. SEAPlur consists of independent datasets for each country and is constructed in their respective native, non-English languages. Sourced from local news and structured around the Schwartz’s theory of basic human values, SEAPlur captures local human agreement distributions across complex social dilemmas. In total, the dataset comprises 4,000 curated scenarios. For evaluation, we split 250 scenarios for each country resulting in SEAPlur-Test. The remaining data with "Agree" and "Disagree" annotations is transformed into a SEAPlur-Train 7,460 instruction-response pairs. Our evaluations using SEAPlur-Test reveal that general API LLMs struggle with SEA values, achieving peak overall accuracy (OA) scores of only 0.50. In contrast, region-specific LLMs perform better, reaching OA scores up to 0.58. Furthermore, supervised fine-tuning (SFT) using SEAPlur-Train significantly improves alignment. However, this process reveals a critical safety tension: enhancing regional alignment in certain LLMs may trigger unsafe behaviors. The dataset is publicly available at https://github.com/tjunlp-lab/SEAPlur.

The value alignment of large language models (LLMs) is crucial for ensuring responses align with human intention and value preferences. However, most evaluations of value alignment focus on Western or universal values, while assessments grounded in the value systems of specific countries remain scarce. In this paper, we introduce Pancasila-Dilemmas, an evaluation dataset of 1,834 questions derived from Indonesian news, classified by 5 values of Pancasila: Religion, Humanity, Unity, Democracy, and Social Justice. This dataset reflects daily life in Indonesia, making it suitable for measuring the value alignment of LLMs deployed for Indonesia. To ensure a more rigorous evaluation, we choose scenarios containing dilemmas. The dataset adopts a multiple-choice format consisting of a scenario, a question, and 4 different options without right/wrong decisions. The dataset is proofread by native speakers and answered by 5 diverse Indonesian citizens. We evaluate 50 closed- and open-source LLMs on our dataset. Results reveal that all evaluated LLMs achieves less than 0.5 Probability Match Score (PMS) and 0.72 Max-Vote Agreement Score (MVAS). Compared by each values, LLMs mostly struggle in Religion and Unity dilemma cases. This highlights a significant gap in capturing Indonesian values. The dataset is publicly available at https://github.com/tjunlp-lab/Pancasila-Dilemmas.

With the widespread industrial deployment of Large Language Models (LLMs), DeepSearch has emerged as the dominant paradigm for resolving complex user queries. It implements an iterative closed-loop workflow consisting of planning and reflection, information retrieval, and answer generation. However, existing reward models (RMs) are largely designed for static single-turn tasks, failing to capture the full-pipeline complexity of DeepSearch workflows. To address this limitation, we propose F²DR, a fine-grained full-pipeline reward framework. F²DR evaluates DeepSearch workflows across three dimensions: Content, Trajectory, and Answer, enabling comprehensive process-level assessment. We further construct DeepSearch RM-Bench, a dedicated benchmark for evaluating RMs in DeepSearch scenarios. Extensive experiments demonstrate that F²DR achieves significantly higher evaluation consistency than existing baselines, while DeepSearch RM-Bench exhibits strong discriminative capability across existing open-source RMs. We will publicly release the complete DeepSearch RM-Bench dataset.

Large Language Model (LLM) agents have demonstrated considerable potential for social simulation, yet struggle to accurately model individual value systems. Most existing methods mechanically stitch survey responses into prompts, which suffer from semantic fragmentation, failing to capture the internal coherence of human value systems. The value systems of LLMs are typically assessed using static multiple-choice questions, which fail to evaluate the value orientation in real-world dialogue interactions. To address these issues, we propose ExpertIVS, a framework employing 14 Sociological Expert Agents to interpret World Values Survey (WVS) responses through structured professional perspectives, rather than direct responses concatenation. These expert agents perform deep semantic reconstruction to generate robust and internally consistent individual profiles. To evaluate the consistency between LLMs and individual value systems during dynamic interactions, we further introduce a multi-agent debate mechanism. Extensive experiments across 480 individuals from 12 countries demonstrate that ExpertIVS achieves 90.78% value restoration fidelity and significantly outperforms baselines in value generalization (+5.3%). Moreover, ExpertIVS exhibits strong personality discriminability and behavioral consistency, enabling a shift from mere response concatenation to genuine sociological role-playing.

Despite recent advances in safety alignment, large language models (LLMs) remain highly susceptible to adversarial attacks, while the internal mechanisms behind such vulnerabilities are still poorly understood. Existing gradient-based attribution methods offer valuable interpretability for analyzing information storage and processing in LLMs. However, they are inapplicable to adversarial attacks, which typically occur in open-ended generation settings without fixed ground-truth outputs. To address these challenges, we propose a novel similarity-based gradient attribution method to identify key neurons sensitive to adversarial behaviors in open-ended generation tasks. The detected neurons, termed targeted neurons, play a critical role in safety training. Building on this neuron-level perspective, we uncover two key neuronal patterns: (i) universal neurons that are consistently exploited across multiple attack strategies, and (ii) interference neurons that hinder safety improvements when fine-tuned indiscriminately, providing mechanistic insights into the interpretability of adversarial vulnerabilities. Inspired by these findings, we propose a neuron-level defense strategy, Targeted Neuron Tuning (TNT), which selectively fine-tunes the identified targeted neurons for specific attacks. Experimental evaluations across multiple LLM architectures and scales demonstrate that TNT substantially improves model robustness against a wide range of jailbreak attacks, achieving safe rates exceeding 90% and even approaching 100%, while preserving general task performance, enabling precise and robust safety interventions. Warning: This paper contains example data that may be harmful.

Reinforcement learning with verifiable rewards (RLVR) has emerged as a prominent paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, the entropy of LLMs usually collapses during RLVR training, leading to premature convergence to suboptimal local minima and hindering further performance improvement. Although various approaches have been proposed to mitigate entropy collapse, a comprehensive study of entropy in RLVR remains lacking. To bridge this gap, we conduct extensive experiments to investigate the entropy dynamics of LLMs trained with RLVR and analyze how model entropy correlates with response diversity, calibration, and performance across various benchmarks. Our results identify three key factors that influence entropy: the clipping thresholds in the optimization objective, the number of off-policy updates, and the diversity of the training data. Furthermore, through both theoretical analysis and empirical validation, we demonstrate that tokens with positive advantages are the primary drivers of entropy collapse. Motivated by this insight, we propose Positive-Advantage Reweighting, a simple yet effective approach that regulates model entropy by adjusting the loss weights assigned to tokens with positive advantages during RLVR training, while maintaining competitive performance.

Ensuring the alignment of Large Language Models (LLMs) with diverse cultural values is critical for their safe global deployment. However, existing alignment benchmarks are predominantly English-centric or confined to simple discriminative tasks. Moreover, they often fail to effectively evaluate model robustness against complex, adversarial inductions within specific cultural contexts. To address these limitations, we propose ABench, a fine-grained, handcrafted benchmark specifically tailored to taxonomized values in Chinese. Grounded in a rigorous hierarchical taxonomy comprising 158 derived value subcategories, ABench consists of 2,895 open-ended questions. These questions employ techniques such as role-playing and logical traps to probe the stability of models’ stances in adversarially induced contexts. Furthermore, we propose a weighted automated evaluation framework that utilizes question-specific scoring criteria for nuanced assessment. Extensive experiments on 16 mainstream LLMs reveal that most models exhibit significant vulnerabilities when subjected to sophisticated adversary attacks on values (scoring below 60). This underscores the need for more robust, culturally specific alignment strategies. Our data will be publicly released to promote further research on taxonomized value alignment.

Currently, most agentic systems are driven by mainstream closed-source large language models (LLMs), such as GPT-5 and Gemini 3, resulting in high operational costs. As a result, deploying customized small language models (SLMs) to replace LLMs in agent components has attracted increasing attention. However, the limited capabilities of SLMs often render them inadequate for agent tasks. To address this, we propose Edit4Agency, a novel framework that utilizes model editing to efficiently and effectively enhance agent sub-capabilities of SLMs. By formalizing tasks as a sequence generation problem triggered by specific conditions, the framework embeds task triggers into system prompts and performs model editing through localized parameter updates, eliminating the need for large-scale retraining. Extensive experiments across multiple datasets and case studies in agent systems demonstrate that model editing, with only 50 data samples, enables a base model without instruction tuning to acquire new skills such as tool calling while maintaining its general capabilities. The results show that Edit4Agency outperforms instruction-tuned SLMs, constrained decoding, and SLMs trained with large-scale supervised fine-tuning or reinforcement learning methods.

Current Large Language Models (LLMs) typically rely on coarse-grained national labels for pluralistic value alignment. However, such macro-level supervision often obscures intra-country value heterogeneity, yielding a loose alignment.We argue that resolving this limitation requires shifting from national labels to multi-dimensional demographic constraints, which can identify groups with predictable, high-consensus value preference. To this end, we propose DVMap (High-Consensus Demographic-Value Mapping), a framework for fine-grained pluralistic value alignment. In this framework, we first present a demographic archetype extraction strategy to construct a high-quality value alignment corpus of 56,152 samples from the World Values Survey (WVS) by strictly retaining respondents with consistent value preferences under identical demographics. Over this corpus, we introduce a Structured Chain-of-Thought (CoT) mechanism that explicitly guides LLMs to reason about demographic-value correlations. Subsequently, we employ Group Relative Policy Optimization (GRPO) to achieve adaptive anchoring of value distributions. To rigorously evaluate generalization, we further establish a triple-generalization benchmark (spanning cross-demographic, cross-country, and cross-value) comprising 21,553 samples. Experimental results demonstrate that DVMap effectively learns the manifold mapping from demographics to values, exhibiting strong generalization and robustness. On cross-demographic tests, Qwen3-8B-DVMap achieves 48.6% accuracy, surpassing the advanced open-source LLM DeepSeek-v3.2 (45.1%). The source code and dataset are available at https://github.com/EnlightenedAI/DVMap.

Large Language Models (LLMs) are increasingly deployed in contexts requiring complex moral reasoning and value trade-offs. However, existing evaluations typically rely on item-level behavioral metrics, which fail to capture how models structurally prioritize competing values as a cohesive system. To address this, we propose a symmetric human-LLM evaluation framework, grounded in Q methodology, to measure value-structure alignment. Under our protocol, humans and models sort an identical 140-item moral statement set into a shared nine-column forced distribution; for LLMs, we elicit strict rankings and deterministically map them to Q-sort buckets. Using a human reference sample (N=35), we establish a stable three-factor reference geometry specific to this instrument and sample. We evaluate 12 LLMs across four model families via 240 replicated Q-sorts at two temperature settings, quantifying structural alignment via Procrustes similarity (𝜙) and RSA-based Spearman correlation (𝜌). Our results reveal significant cross-family heterogeneity, model-specific sensitivity to generation stochasticity and localized misalignment, which demonstrate that favorable global scores can obscure underlying regional distortions. While rank- and bucket-based analyses remain highly consistent, prompt phrasing introduces notable variance. Ultimately, assessing value-structure alignment provides a crucial structural complement to traditional itemwise moral benchmarks.

Large language models (LLMs), have shown strong potential in scientific discovery, yet existing methods still face substantial challenges in the design of research workflows and multi-role collaboration mechanisms. To mitigate these issues, we propose EvoSci, a multi-agent scientific collaboration framework, which integrates bio-inspired evolution with knowledge graph modeling. To iteratively generate, evaluate, and refine research ideas, EvoSci incorporates multiple role-based agents, including mentor, researcher, and reviewer. By combining collaborative reasoning, shared memory, and evolutionary feedback, EvoSci significantly enhances the coherence and creativity of scientific exploration. Experiments on real-world research topics demonstrate that EvoSci significantly outperforms strong baselines in LLM-based structured peer-review and comparative ranking evaluations, achieving the highest overall peer-review score (ICLR 4.90) and top ranking (Top-10 = 54). These results suggest its superiority in both scientific idea generation and continuous discovery.

Large language models (LLMs) have achieved remarkable success across a wide range of natural language processing tasks, yet their performance remains heavily biased toward high-resource languages. Tibetan, despite its cultural significance and large speaker population, is still substantially underrepresented. In this work, we present a comprehensive pipeline for advancing Tibetan language modeling through large-scale data curation and continual pre-training. We construct a 72 GB high-quality Tibetan corpus, the largest to date, and adapt Qwen2.5-7B through balanced multilingual continual pre-training with Tibetan, Chinese, and English, followed by multilingual instruction tuning. To further scale capacity efficiently, we extend the dense model to a 50B-A10B Mixture-of-Experts architecture. Due to the absence of standardized Tibetan benchmarks, we build multiple evaluation datasets via high-quality translation and human verification. Experimental results show that both dense and MoE models consistently outperform existing open-source and Tibetan-focused models of similar scale across diverse tasks. Our work advances Tibetan-centric LLM research and provides transferable insights for extending LLMs to other low-resource languages. We will release the model weights, evaluation benchmarks, and detailed data processing documentation in the follow-up.