Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
arXiv:2602.12430v2 Announce Type: cross Abstract: The transition from monolithic language models to modular, skill-equipped agents marks a defining shift in how large language models (LLMs) are deployed in practice. Rather than encoding all procedural knowledge within model weights, agent skills...
Safe Reinforcement Learning via Recovery-based Shielding with Gaussian Process Dynamics Models
arXiv:2602.12444v1 Announce Type: cross Abstract: Reinforcement learning (RL) is a powerful framework for optimal decision-making and control but often lacks provable guarantees for safety-critical applications. In this paper, we introduce a novel recovery-based shielding framework that enables safe RL with...
Designing RNAs with Language Models
arXiv:2602.12470v1 Announce Type: cross Abstract: RNA design, the task of finding a sequence that folds into a target secondary structure, has broad biological and biomedical impact but remains computationally challenging due to the exponentially large sequence space and exponentially many...
CLASE: A Hybrid Method for Chinese Legalese Stylistic Evaluation
arXiv:2602.12639v1 Announce Type: new Abstract: Legal text generated by large language models (LLMs) can usually achieve reasonable factual accuracy, but it frequently fails to adhere to the specialised stylistic norms and linguistic conventions of legal writing. In order to improve...
Beyond Normalization: Rethinking the Partition Function as a Difficulty Scheduler for RLVR
arXiv:2602.12642v1 Announce Type: new Abstract: Reward-maximizing RL methods enhance the reasoning performance of LLMs, but often reduce the diversity among outputs. Recent works address this issue by adopting GFlowNets, training LLMs to match a target distribution while jointly learning its...
Learning Ordinal Probabilistic Reward from Preferences
arXiv:2602.12660v1 Announce Type: new Abstract: Reward models are crucial for aligning large language models (LLMs) with human values and intentions. Existing approaches follow either Generative (GRMs) or Discriminative (DRMs) paradigms, yet both suffer from limitations: GRMs typically demand costly point-wise...
$\mathcal{X}$-KD: General Experiential Knowledge Distillation for Large Language Models
arXiv:2602.12674v1 Announce Type: new Abstract: Knowledge Distillation (KD) for Large Language Models (LLMs) has become increasingly important as models grow in size and complexity. While existing distillation approaches focus on imitating teacher behavior, they often overlook the original learning environment...
Lamer-SSL: Layer-aware Mixture of LoRA Experts for Continual Multilingual Expansion of Self-supervised Models without Forgetting
arXiv:2602.12746v1 Announce Type: new Abstract: Despite their impressive performance, self-supervised speech models often struggle to generalize to new languages and tend to forget previously acquired knowledge during continual training. To address this, we propose Lamer-SSL, a parameter-efficient framework that integrates...
Aspect-Based Sentiment Analysis for Future Tourism Experiences: A BERT-MoE Framework for Persian User Reviews
arXiv:2602.12778v1 Announce Type: new Abstract: This study advances aspect-based sentiment analysis (ABSA) for Persian-language user reviews in the tourism domain, addressing challenges of low-resource languages. We propose a hybrid BERT-based model with Top-K routing and auxiliary losses to mitigate routing...
RAT-Bench: A Comprehensive Benchmark for Text Anonymization
arXiv:2602.12806v1 Announce Type: new Abstract: Data containing personal information is increasingly used to train, fine-tune, or query Large Language Models (LLMs). Text is typically scrubbed of identifying information prior to use, often with tools such as Microsoft's Presidio or Anthropic's...
AIWizards at MULTIPRIDE: A Hierarchical Approach to Slur Reclamation Detection
arXiv:2602.12818v1 Announce Type: new Abstract: Detecting reclaimed slurs represents a fundamental challenge for hate speech detection systems, as the same lexcal items can function either as abusive expressions or as in-group affirmations depending on social identity and context. In this...
BaziQA-Benchmark: Evaluating Symbolic and Temporally Compositional Reasoning in Large Language Models
arXiv:2602.12889v1 Announce Type: new Abstract: We present BaziQA-Benchmark, a standardized benchmark for evaluating symbolic and temporally compositional reasoning in large language models. The benchmark is derived from 200 professionally curated, multiple-choice problems from the Global Fortune-teller Competition (2021--2025), where each...
Curriculum Learning and Pseudo-Labeling Improve the Generalization of Multi-Label Arabic Dialect Identification Models
arXiv:2602.12937v1 Announce Type: new Abstract: Being modeled as a single-label classification task for a long time, recent work has argued that Arabic Dialect Identification (ADI) should be framed as a multi-label classification task. However, ADI remains constrained by the availability...
ProbeLLM: Automating Principled Diagnosis of LLM Failures
arXiv:2602.12966v1 Announce Type: new Abstract: Understanding how and why large language models (LLMs) fail is becoming a central challenge as models rapidly evolve and static evaluations fall behind. While automated probing has been enabled by dynamic test generation, existing approaches...
Evaluating the Homogeneity of Keyphrase Prediction Models
arXiv:2602.12989v1 Announce Type: new Abstract: Keyphrases which are useful in several NLP and IR applications are either extracted from text or predicted by generative models. Contrarily to keyphrase extraction approaches, keyphrase generation models can predict keyphrases that do not appear...
Know More, Know Clearer: A Meta-Cognitive Framework for Knowledge Augmentation in Large Language Models
arXiv:2602.12996v1 Announce Type: new Abstract: Knowledge augmentation has significantly enhanced the performance of Large Language Models (LLMs) in knowledge-intensive tasks. However, existing methods typically operate on the simplistic premise that model performance equates with internal knowledge, overlooking the knowledge-confidence gaps...
Can we trust AI to detect healthy multilingual English speakers among the cognitively impaired cohort in the UK? An investigation using real-world conversational speech
arXiv:2602.13047v1 Announce Type: new Abstract: Conversational speech often reveals early signs of cognitive decline, such as dementia and MCI. In the UK, one in four people belongs to an ethnic minority, and dementia prevalence is expected to rise most rapidly...
Exploring a New Competency Modeling Process with Large Language Models
arXiv:2602.13084v1 Announce Type: new Abstract: Competency modeling is widely used in human resource management to select, develop, and evaluate talent. However, traditional expert-driven approaches rely heavily on manual analysis of large volumes of interview transcripts, making them costly and prone...
Towards interpretable models for language proficiency assessment: Predicting the CEFR level of Estonian learner texts
arXiv:2602.13102v1 Announce Type: new Abstract: Using NLP to analyze authentic learner language helps to build automated assessment and feedback tools. It also offers new and extensive insights into the development of second language production. However, there is a lack of...
Semantic Chunking and the Entropy of Natural Language
arXiv:2602.13194v1 Announce Type: new Abstract: The entropy rate of printed English is famously estimated to be about one bit per character, a benchmark that modern large language models (LLMs) have only recently approached. This entropy rate implies that English contains...
Alignment or Integration? Rethinking Multimodal Fusion in DNA-language Foundation Models
arXiv:2602.12286v1 Announce Type: cross Abstract: Fusing DNA foundation models with large language models (LLMs) for DNA-language reasoning raises a fundamental question: at what level should genomic sequences and natural language interact? Most existing approaches encode DNA sequences and text separately...
Beyond Musical Descriptors: Extracting Preference-Bearing Intent in Music Queries
arXiv:2602.12301v1 Announce Type: cross Abstract: Although annotated music descriptor datasets for user queries are increasingly common, few consider the user's intent behind these descriptors, which is essential for effectively meeting their needs. We introduce MusicRecoIntent, a manually annotated corpus of...
Sparse Autoencoders are Capable LLM Jailbreak Mitigators
arXiv:2602.12418v1 Announce Type: cross Abstract: Jailbreak attacks remain a persistent threat to large language model safety. We propose Context-Conditioned Delta Steering (CC-Delta), an SAE-based defense that identifies jailbreak-relevant sparse features by comparing token-level representations of the same harmful request with...
Constraint-Rectified Training for Efficient Chain-of-Thought
arXiv:2602.12526v1 Announce Type: cross Abstract: Chain-of-Thought (CoT) has significantly enhanced the reasoning capabilities of Large Language Models (LLMs), especially when combined with reinforcement learning (RL) based post-training methods. While longer reasoning traces can improve answer quality and unlock abilities such...
DiffuRank: Effective Document Reranking with Diffusion Language Models
arXiv:2602.12528v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have inspired new paradigms for document reranking. While this paradigm better exploits the reasoning and contextual understanding capabilities of LLMs, most existing LLM-based rerankers rely on autoregressive generation,...
HyperMLP: An Integrated Perspective for Sequence Modeling
arXiv:2602.12601v1 Announce Type: cross Abstract: Self-attention is often viewed as probabilistic query-key lookup, motivating designs that preserve normalized attention scores and fixed positional semantics. We advocate a simpler and more unified perspective: an autoregressive attention head can be viewed as...
VimRAG: Navigating Massive Visual Context in Retrieval-Augmented Generation via Multimodal Memory Graph
arXiv:2602.12735v1 Announce Type: cross Abstract: Effectively retrieving, reasoning, and understanding multimodal information remains a critical challenge for agentic systems. Traditional Retrieval-augmented Generation (RAG) methods rely on linear interaction histories, which struggle to handle long-context tasks, especially those involving information-sparse yet...
Abstractive Red-Teaming of Language Model Character
arXiv:2602.12318v1 Announce Type: new Abstract: We want language model assistants to conform to a character specification, which asserts how the model should act across diverse user interactions. While models typically follow these character specifications, they can occasionally violate them in...
The Appeal and Reality of Recycling LoRAs with Adaptive Merging
arXiv:2602.12323v1 Announce Type: new Abstract: The widespread availability of fine-tuned LoRA modules for open pre-trained models has led to an interest in methods that can adaptively merge LoRAs to improve performance. These methods typically include some way of selecting LoRAs...
Wireless TokenCom: RL-Based Tokenizer Agreement for Multi-User Wireless Token Communications
arXiv:2602.12338v1 Announce Type: new Abstract: Token Communications (TokenCom) has recently emerged as an effective new paradigm, where tokens are the unified units of multimodal communications and computations, enabling efficient digital semantic- and goal-oriented communications in future wireless networks. To establish...