Search: “arxiv-cs-cl”

Copy First, Translate Later: Interpreting Translation Dynamics in Multilingual Pretraining

arXiv:2604.17633v1 Announce Type: new Abstract: Large language models exhibit impressive cross-lingual capabilities. However, prior work analyzes this phenomenon through isolated factors and at sparse

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

COSEARCH: Joint Training of Reasoning and Document Ranking via Reinforcement Learning for Agentic Search

DGX agent

arXiv:2604.17555v1 Announce Type: cross Abstract: Agentic search -- the task of training agents that iteratively reason, issue queries, and synthesize retrieved information to answer complex questions

safetyarxiv-cs-cl

21 Apr 2026

Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR

arXiv:2603.07084v2 Announce Type: replace-cross Abstract: Reward hacking is a form of misalignment in which models overoptimize proxy rewards without genuinely solving the underlying task. Precisely m

Creating ConLangs to Probe the Metalinguistic Grammatical Knowledge of LLMs

arXiv:2510.07591v3 Announce Type: replace Abstract: We present a system that uses LLMs as a tool in the development of Constructed Languages -- ConLangs, which we call IASC (Interactive Agentic System

CreditDecoding: Accelerating Parallel Decoding in Diffusion Large Language Models with Trace Credit

arXiv:2510.06133v2 Announce Type: replace Abstract: Diffusion large language models (dLLMs) generate text through iterative denoising. In commonly adopted parallel decoding schemes, each step confirms

CRISP: Compressing Redundancy in Chain-of-Thought via Intrinsic Saliency Pruning

arXiv:2604.17297v1 Announce Type: new Abstract: Long Chain-of-Thought (CoT) reasoning is pivotal for the success of recent reasoning models but suffers from high computational overhead and latency. Wh

CROC: Evaluating and Training T2I Metrics with Pseudo- and Human-Labeled Contrastive Robustness Checks

arXiv:2505.11314v2 Announce Type: replace-cross Abstract: The assessment of evaluation metrics (meta-evaluation) is crucial for determining the suitability of existing metrics in text-to-image (T2I) g

Cross-Family Speculative Decoding for Polish Language Models on Apple~Silicon: An Empirical Evaluation of Bielik~11B with UAG-Extended MLX-LM

arXiv:2604.16368v1 Announce Type: new Abstract: Speculative decoding accelerates LLM inference by using a small draft model to propose k candidate tokens for a target model to verify. While effective

Applications

Crowded in B-Space: Calibrating Shared Directions for LoRA Merging

arXiv:2604.16826v1 Announce Type: new Abstract: Merging separately trained LoRA adapters is a practical alternative to joint multi-task training, but it often hurts performance. Existing methods usual

applicationsarxiv-cs-cl

Applications

CT Open: An Open-Access, Uncontaminated, Live Platform for the Open Challenge of Clinical Trial Outcome Prediction

arXiv:2604.16742v1 Announce Type: cross Abstract: Scientists have long sought to accurately predict outcomes of real-world events before they happen. Can AI systems do so more reliably? We study this

applicationsarxiv-cs-cl

Culinary Crossroads: A RAG Framework for Enhancing Diversity in Cross-Cultural Recipe Adaptation

arXiv:2507.21934v2 Announce Type: replace Abstract: In cross-cultural recipe adaptation, the goal is not only to ensure cultural appropriateness and retain the original dish's essence, but also to pro

Culture-Aware Humorous Captioning: Multimodal Humor Generation across Cultural Contexts

arXiv:2604.18091v1 Announce Type: new Abstract: Recent multimodal large language models have shown promising ability in generating humorous captions for images, yet they still lack stable control over

DART: Mitigating Harm Drift in Difference-Aware LLMs via Distill-Audit-Repair Training

arXiv:2604.16845v1 Announce Type: new Abstract: Large language models (LLMs) tuned for safety often avoid acknowledging demographic differences, even when such acknowledgment is factually correct (e.g

Data Compressibility Quantifies LLM Memorization

arXiv:2507.06056v4 Announce Type: replace Abstract: Large Language Models (LLMs) are known to memorize portions of their training data, sometimes even reproduce content verbatim when prompted appropri

Data Mixing for Large Language Models Pretraining: A Survey and Outlook

arXiv:2604.16380v1 Announce Type: new Abstract: Large language models (LLMs) rely on pretraining on massive and heterogeneous corpora, where training data composition has a decisive impact on training

Decisive: Guiding User Decisions with Optimal Preference Elicitation from Unstructured Documents

arXiv:2604.18122v1 Announce Type: new Abstract: Decision-making is a cognitively intensive task that requires synthesizing relevant information from multiple unstructured sources, weighing competing f

Decoupling the Effect of Chain-of-Thought Reasoning: A Human Label Variation Perspective

arXiv:2601.03154v2 Announce Type: replace Abstract: Reasoning-tuned LLMs utilizing long Chain-of-Thought (CoT) excel at single-answer tasks, yet their ability to model Human Label Variation--which req

Defragmenting Language Models: An Interpretability-based Approach for Vocabulary Expansion

arXiv:2604.16656v1 Announce Type: new Abstract: All languages are equal; when it comes to tokenization, some are more equal than others. Tokens are the hidden currency that dictate the cost and latenc

DeInfer: Efficient Parallel Inferencing for Decomposed Large Language Models

arXiv:2604.17709v1 Announce Type: new Abstract: Existing works on large language model (LLM) decomposition mainly focus on improving performance on downstream tasks, but they ignore the poor parallel

Demystifying the unreasonable effectiveness of online alignment methods

arXiv:2604.17207v1 Announce Type: cross Abstract: Iterative alignment methods based on purely greedy updates are remarkably effective in practice, yet existing theoretical guarantees of (O(log T)) KL-

Depth Registers Unlock W4A4 on SwiGLU: A Reader/Generator Decomposition

arXiv:2604.18128v1 Announce Type: new Abstract: We study post-training W4A4 quantization in a controlled 300M-parameter SwiGLU decoder-only language model trained on 5B tokens of FineWeb-Edu, and ask

Agents

Designing Explainable Conversational Agentic Systems for Guarani Speakers

arXiv:2603.05743v3 Announce Type: replace Abstract: Although artificial intelligence (AI) and Human-Computer Interaction (HCI) systems are often presented as universal solutions, their design remains

agentsarxiv-cs-cl

Detecting Alarming Student Verbal Responses using Text and Audio Classifier

arXiv:2604.16717v1 Announce Type: new Abstract: This paper addresses a critical safety gap in the use Automated Verbal Response Scoring (AVRS). We present a novel hybrid framework for troubled student

Detecting LLM-Generated Spam Reviews by Integrating Language Model Embeddings and Graph Neural Network

arXiv:2510.01801v2 Announce Type: replace Abstract: The rise of large language models (LLMs) has enabled the generation of highly persuasive spam reviews that closely mimic human writing. These review

Diagnosing LLM-based Rerankers in Cold-Start Recommender Systems: Coverage, Exposure and Practical Mitigations

arXiv:2604.16318v1 Announce Type: cross Abstract: Large language models (LLMs) and cross-encoder rerankers have gained attention for improving recommender systems, particularly in cold-start scenarios

DiffCoT: Diffusion-styled Chain-of-Thought Reasoning in LLMs

arXiv:2601.03559v2 Announce Type: replace Abstract: Chain-of-Thought (CoT) reasoning improves multi-step mathematical problem solving in large language models but remains vulnerable to exposure bias a

Local Ai

Different Paths to Harmful Compliance: Behavioral Side Effects and Mechanistic Divergence Across LLM Jailbreaks

arXiv:2604.18510v1 Announce Type: cross Abstract: Open-weight language models can be rendered unsafe through several distinct interventions, but the resulting models may differ substantially in capabi

local-aiarxiv-cs-cl

Agents

Diversity Collapse in Multi-Agent LLM Systems: Structural Coupling and Collective Failure in Open-Ended Idea Generation

arXiv:2604.18005v1 Announce Type: cross Abstract: Multi-agent systems (MAS) are increasingly used for open-ended idea generation, driven by the expectation that collective interaction will broaden the

agentsarxiv-cs-cl

Do LLMs Encode Functional Importance of Reasoning Tokens?

arXiv:2601.03066v2 Announce Type: replace Abstract: Large language models solve complex tasks by generating long reasoning chains, achieving higher accuracy at the cost of increased computational cost

Do LLMs Use Cultural Knowledge Without Being Told? A Multilingual Evaluation of Implicit Pragmatic Adaptation

arXiv:2604.17718v1 Announce Type: new Abstract: Many benchmarks show that large language models can answer direct questions about culture. We study a different question: do they also change how they s

DocQAC: Adaptive Trie-Guided Decoding for Effective In-Document Query Auto-Completion

arXiv:2604.18257v1 Announce Type: cross Abstract: Query auto-completion (QAC) has been widely studied in the context of web search, yet remains underexplored for in-document search, which we term DocQ

Document-as-Image Representations Fall Short for Scientific Retrieval

arXiv:2604.18508v1 Announce Type: cross Abstract: Many recent document embedding models are trained on document-as-image representations, embedding rendered pages as images rather than the underlying

Does Welsh media need a review? Detecting bias in Nation.Cymru's political reporting

arXiv:2604.17628v1 Announce Type: new Abstract: Wales' political landscape has been marked by growing accusations of bias in Welsh media. This paper takes the first computational step toward testing t

Domain-oriented RAG Assessment (DoRA): Synthetic Benchmarking for RAG-based Question Answering on Defense Documents

arXiv:2604.17943v1 Announce Type: new Abstract: Open-domain RAG benchmarks over public corpora can overestimate deployment performance due to pretraining overlap and weak attribution requirements. We

Agents

Don't Adapt Small Language Models for Tools; Adapt Tool Schemas to the Models

arXiv:2510.07248v3 Announce Type: replace Abstract: Small language models (SLMs) enable scalable tool-augmented multi-agent systems where multiple SLMs handle subtasks orchestrated by a powerful coord

agentsarxiv-cs-cl

DORA Explorer: Improving the Exploration Ability of LLMs Without Training

arXiv:2604.17244v1 Announce Type: new Abstract: Despite the rapid progress, LLMs for sequential decision-making (i.e., LLM agents) still struggle to produce diverse outputs. This leads to insufficient

DOSE: Data Selection for Multi-Modal LLMs via Off-the-Shelf Models

arXiv:2604.16979v1 Announce Type: cross Abstract: High-quality and diverse multimodal data are essential for improving vision-language models (VLMs), yet existing datasets often contain noisy, redunda

Dual Alignment Between Language Model Layers and Human Sentence Processing

arXiv:2604.18563v1 Announce Type: new Abstract: A recent study (Kuribayashi et al., 2025) has shown that human sentence processing behavior, typically measured on syntactically unchallenging construct

DualToken: Towards Unifying Visual Understanding and Generation with Dual Visual Vocabularies

arXiv:2503.14324v3 Announce Type: replace-cross Abstract: The differing representation spaces required for visual understanding and generation pose a challenge in unifying them within the autoregressi

DuConTE: Dual-Granularity Text Encoder with Topology-Constrained Attention for Text-attributed Graphs

arXiv:2604.17411v1 Announce Type: new Abstract: Text-attributed graphs integrate semantic information of node texts with topological structure, offering significant value in various applications such

DuQuant++: Fine-grained Rotation Enhances Microscaling FP4 Quantization

arXiv:2604.17789v1 Announce Type: cross Abstract: The MXFP4 microscaling format, which partitions tensors into blocks of 32 elements sharing an E8M0 scaling factor, has emerged as a promising substrat

Dynamic Emotion and Personality Profiling for Multimodal Deception Detection

arXiv:2604.17037v1 Announce Type: new Abstract: Deception detection is of great significance for ensuring information security and conducting public opinion analysis, with personality factors and emot

DynaWeb: Model-Based Reinforcement Learning of Web Agents

arXiv:2601.22149v2 Announce Type: replace Abstract: The development of autonomous web agents, powered by Large Language Models (LLMs) and reinforcement learning (RL), represents a significant step tow

E2E-GMNER: End-to-End Generative Grounded Multimodal Named Entity Recognition

arXiv:2604.17319v1 Announce Type: cross Abstract: Grounded Multimodal Named Entity Recognition (GMNER) aims to jointly identify named entity mentions in text, predict their semantic types, and ground

EchoChain: A Full-Duplex Benchmark for State-Update Reasoning Under Interruptions

arXiv:2604.16456v1 Announce Type: new Abstract: Real-time voice assistants must revise task state when users interrupt mid-response, but existing spoken-dialog benchmarks largely evaluate turn-based i

Tutorials

EduRABSA: An Education Review Dataset for Aspect-based Sentiment Analysis Tasks

arXiv:2508.17008v2 Announce Type: replace Abstract: Every year, most educational institutions seek and receive an enormous volume of text feedback from students on courses, teaching, and overall exper

tutorialsarxiv-cs-cl

Efficient Low-Resource Language Adaptation via Multi-Source Dynamic Logit Fusion

arXiv:2604.18106v1 Announce Type: new Abstract: Adapting large language models (LLMs) to low-resource languages (LRLs) is constrained by the scarcity of task data and computational resources. Although

Efficient Task Adaptation in Large Language Models via Selective Parameter Optimization

arXiv:2604.17051v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated excellent performance in general language understanding, generation and other tasks. However, when fine-t