Search: “arxiv-cs-cl”

Reasoning-Token Spikes Under Prompted Untruthful Responding in Large Language Models

arXiv:2610.10405v1 Announce Type: cross Abstract: Monitoring the chain-of-thought of reasoning artificial intelligence (AI) models remains a key approach to detecting deception and other forms of misb

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

RELATE: An Evaluation Framework for measuring Relational Orientation of Large Language Models

DGX agent

arXiv:2610.09569v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for emotional support, raising concern that sustained use may draw users away from their real-world r

safetyarxiv-cs-cl

8 Oct 2026

Rephrase Before You Act: Characterizing and Mitigating Language Sensitivity in Vision-Language-Action Models

arXiv:2610.10526v1 Announce Type: cross Abstract: Vision-language-action models (VLAs) are strikingly sensitive to instruction phrasing and do not inherit the language robustness of the vision-languag

Right Number, Wrong State? Measuring Cross-Jurisdiction Substitution in LLM Recall of State Policy

arXiv:2610.09458v1 Announce Type: new Abstract: When an LLM answers a state-specific policy question wrongly, it may be hallucinating, or it may be returning a real value that holds in another state.

Route-Verify-Vote: Procedure-Conditioned Self-Consistency for Mixed-Domain Reasoning

arXiv:2610.08814v1 Announce Type: cross Abstract: Compositional generalization remains challenging when language models must combine familiar reasoning operations in unfamiliar ways. The Scenario-Base

Rubric Spans are Label Representations: Joint LLM Encoding for Short Answer Scoring

arXiv:2610.09660v1 Announce Type: new Abstract: Automatic Short Answer Scoring (ASAS) requires models that can score student responses against question-specific criteria while remaining efficient and

RunningTab: Direct Workspace Interaction with Environment-Side Tabs

arXiv:2610.10444v1 Announce Type: cross Abstract: Much knowledge work produces new deliverables from files a workspace already holds, and LLM agents are beginning to take such work over. Through direc

Same Text, Different Prediction: Serving-Context Nondeterminism in Text Classifiers

arXiv:2610.09111v1 Announce Type: new Abstract: Deterministic inference is essential for reliable and trustworthy machine learning. Prior studies of text generation have shown that changing factors su

SAPD: Step-Aligned Privileged Distillation

arXiv:2610.09665v1 Announce Type: new Abstract: On-policy post-training can improve large language models by learning from their own trajectories, but requires costly rollout generation. We ask whethe

SEER: Self-Enhancing Chain-of-Thought Compression for Reasoning Models

arXiv:2509.14093v3 Announce Type: replace-cross Abstract: Chain-of-Thought (CoT) prompting can substantially improve the reasoning ability of large language models (LLMs), but it often comes with high

SemanticFold: Latent Sequence Compression SeparatesLanguage Modeling, Decodability, and Reasoning

arXiv:2610.10304v1 Announce Type: cross Abstract: We study whether latent sequence compression of prompt prefixes preserves the capabilities that large language models rely on during inference. We int

Sensitive-Topic Leakage Through LLM Routing Metadata: Measurement and Mitigation

arXiv:2610.09981v1 Announce Type: cross Abstract: LLM routers pick a cheap or expensive model per request by its content, and many gateways and some cloud platforms can log that choice with content lo

SeOPD: Self-Evolving LLMs via Online Policy Distillation from Self-Generated Chain-of-Thought

arXiv:2609.33181v2 Announce Type: replace-cross Abstract: Recent advances in online policy self-distillation (OPSD) have demonstrated that large language models (LLMs) can improve their capabilities b

Shaer: Controlled Arabic Poetry Generation with Meter Subform and Semantic Conditioning

arXiv:2610.09756v1 Announce Type: new Abstract: Classical Arabic poetry generation requires simultaneously satisfying semantic, linguistic, and fine-grained prosodic constraints. Existing systems typi

sk-bench: A Native-First Benchmark for Evaluating Large Language Models in Slovak

arXiv:2610.09152v1 Announce Type: new Abstract: Multilingual LLM benchmarks omit Slovak, a morphologically rich West Slavic language of five million speakers, or cover it only by machine translation.

SkillSandbox: Skill Verification via Dynamic Scenario Synthesis

arXiv:2610.10088v1 Announce Type: cross Abstract: Self-evolving agents distill task-solving experience into skills for future reuse, but these skills can encode incorrect procedures or non-transferabl

Spend Bytes on Breadth: Precision-Count Trade-offs for Decode-Time KV Compression in Long Chain-of-Thought Reasoning

arXiv:2610.05685v2 Announce Type: replace Abstract: Reasoning models write most of their KV cache while decoding long chains of thought (CoT), so the cache has to be compressed online under a fixed me

SpikingVLA: Asynchronous Spiking Vision-Language-Action Models

arXiv:2610.09710v1 Announce Type: new Abstract: ANN-to-SNN conversion offers a practical route toward energy-efficient spiking Vision-Language-Action (VLA) models by bypassing the substantial cost of

Steering Follows Geometry, Not Labels: Emotion Directions in a Full-Duplex Speech Model

arXiv:2610.08887v1 Announce Type: new Abstract: Full-duplex voice agents need to modulate emotion and delivery during real-time conversations, when de-escalating a complaint, carrying urgency in dispa

Talking with Language Models

arXiv:2610.09064v1 Announce Type: cross Abstract: When we interact with large language models (LLMs), are we having a conversation? They are designed to invite us to treat them as intelligent interloc

Agents

The Confidence Game: Strategic Miscalibration in Human-AI Delegation

arXiv:2610.09371v1 Announce Type: cross Abstract: Calibrated uncertainty quantification is essential to ensuring AI agents are trustworthy and reliable. However, when agents seek to maximize user enga

agentsarxiv-cs-cl

The IroyinSpeech Text Corpus: 24,905 Curated Yoruba Sentences for Speech and Language Technology

arXiv:2610.05366v2 Announce Type: replace Abstract: IroyinSpeech is a 42-hour, 80-speaker Yoruba read-speech corpus whose audio has been distributed by ELRA since 2024. This paper describes the releas

The Long Road to the Same Answer: Cognitive Bias Under Escalating Reasoning Budgets in Large Language Models

arXiv:2610.10049v1 Announce Type: new Abstract: Reasoning models allocate extra computation at inference time and present their answers as the product of deliberate thought. If this deliberation works

Local Ai

The Persona Hierarchy Model: Understanding Contextual Generalization in Fine-Tuning LLMs

arXiv:2610.09384v1 Announce Type: new Abstract: Language models are routinely fine-tuned under a fixed context, such as a generic system prompt, persona or domain-specific instruction, yet the learned

local-aiarxiv-cs-cl

Tiny-Scale Chinese BERT Pretraining: A Controlled Comparison of MLM, WWM, and MacBERT Strategies

arXiv:2610.08879v1 Announce Type: new Abstract: Pretraining strategies significantly impact the quality of language models, yet existing comparisons of Masked Language Modeling (MLM), Whole Word Maski

Tokka-Bench: Evaluating Tokenizers Across 100 Natural and 20 Programming Languages

arXiv:2610.08794v1 Announce Type: new Abstract: Large language models rely on subword tokenizers whose quality varies across languages, yet no standardized multi-metric framework exists for broad comp

ToolRACER: A Robust Agentic Conversation Emulation Resource for Agent Training and Evaluation

arXiv:2610.09163v1 Announce Type: new Abstract: Task-oriented conversational agents remain fragile under real world conversation scenarios as they rarely follow a predictable script, especially when u

TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence Reasoning

arXiv:2610.09360v1 Announce Type: cross Abstract: Real-world documents distribute evidence across text, tables, figures, and captions within complex page layouts. Answering complex questions over such

Towards Explaining Query Expansion Performance in Information Retrieval

arXiv:2610.09724v1 Announce Type: cross Abstract: Query Expansion (QE) techniques have long been widely used in Information Retrieval (IR) to address the vocabulary mismatch problem. They remain relev

Training Advisors for LLM Agents from Task Outcomes

arXiv:2610.09858v1 Announce Type: new Abstract: Large language model agents tackle multi-step tasks by interleaving reasoning and tool calls with observations from the environment. Prior work has show

Local Ai

Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds

arXiv:2610.10411v1 Announce Type: cross Abstract: Speculative decoding accelerates large language model inference by using a low-cost draft model to propose tokens that the full-size target model veri

local-aiarxiv-cs-cl

Agents

Trajectory Abstraction for the Science of Language Agent Behavior

arXiv:2610.09237v1 Announce Type: cross Abstract: Scientific studies of language agents need behavioral variables that support hypotheses across tasks and models. We formulate this research problem as

agentsarxiv-cs-cl

U-Space: Uncovering When and Why Uncertainty Arises in Language Models

arXiv:2610.09087v1 Announce Type: new Abstract: Large language models are informing decisions with ever-higher stakes. As the consequences of their errors grow, a central question becomes harder to ig

Validity Without Ground Truth: What Stated-Preference Economics Offers the Evaluation of Language Models

arXiv:2610.10506v1 Announce Type: cross Abstract: Many of the questions now put to large language models have no correct answer to score against: what a policy is worth, which option a user should cho

Applications

Vectorizing the Trie: Efficient Constrained Decoding for LLM-based Generative Retrieval on Accelerators

arXiv:2602.22647v3 Announce Type: replace-cross Abstract: Generative retrieval has emerged as a powerful paradigm for LLM-based recommendation. However, industrial recommender systems often benefit fr

applicationsarxiv-cs-cl

When Forgetting Looks Like Improvement: Metric Masking in Streaming Diarizer Adaptation and the Price of Rehearsal

arXiv:2610.08828v1 Announce Type: new Abstract: Small-data adaptation can improve speech detection while degrading speaker attribution. We study this discrepancy in a released streaming diarizer adapt

When Rank Rises as LLMs Degrade

arXiv:2610.09647v1 Announce Type: cross Abstract: Post-training adapts language models in non-stationary environments. Practitioners monitor representation health with RankMe and related spectral stat

Which Language Should a Skeleton Speak? Language Choices in Multilingual Reasoning

arXiv:2610.09607v1 Announce Type: new Abstract: Skeleton-based reasoning prompting is a promising training-free approach for structuring LLM reasoning, but prior work largely assumes an English-centri

Which Rollout Taught It That? BehaviorTrace and the Limits of Training-Data Attribution in Online RL

arXiv:2610.10422v1 Announce Type: cross Abstract: When reinforcement learning teaches a language model a new behavior, can we find the training rollouts that taught it? And when an attribution method

Hardware

YANchor-4B: Effective Long-Horizon Reasoning in O(N) Time with O(1) Memory

arXiv:2610.10118v1 Announce Type: cross Abstract: Long-horizon reasoning demands access to earlier information at a manageable generation cost. Full-history attention incurs growing storage and comput

hardwarearxiv-cs-cl

Your Prompt Should Do More: Effects of Retrieval Instructions in Embedding Models

arXiv:2610.10508v1 Announce Type: new Abstract: Prompted embedding models have recently received increasing attention, particularly for retrieval, where detailed retrieval instructions are provided as

A Broader Look at Model Merging: Rethinking Implicit Regularization Induced by Task Arithmetic

arXiv:2610.07990v1 Announce Type: cross Abstract: Model merging aims to build a multi-task model cheaply by combining the weights of individual task-specific models. To perform well across multiple ta

A Novel Sentence Stress Detection Framework Leveraging Auxiliary Word-Stress Modeling and Loss Optimization

arXiv:2610.07626v1 Announce Type: cross Abstract: Prosodic stress is a crucial aspect of automatic pronunciation assessment (APA), encompassing both sentence stress detection (SSD) and word stress det

A Systematic Study of Semantic ID Spaces for Generative Information Retrieval

arXiv:2610.08732v1 Announce Type: cross Abstract: Generative Information Retrieval (GIR) has emerged as a transformative paradigm, shifting document retrieval from a traditional 'retrieve-and-rank' wo

A Systematic Study of Small Language Models on Abstract Reasoning Tasks

arXiv:2610.08680v1 Announce Type: cross Abstract: Endpoint accuracy on abstract-reasoning benchmarks does not reveal whether a language model has acquired a transferable rule or fit distribution-speci

A theory of platonic representations in language models

arXiv:2610.07168v1 Announce Type: cross Abstract: Representations of translated sentences are similar in the inner layers of multilingual language models -- an observation connected to the platonic re

AccentCL: Robust Accent Classification with Incremental Expansion

arXiv:2610.07426v1 Announce Type: new Abstract: Accent classifiers are typically trained with a fixed label inventory and cannot accommodate new accent categories as new data becomes available. Moreov

Agents

Agentic AutoRAG: RAG Pipeline Optimization through Reasoning-Driven Agents

arXiv:2610.08452v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) is a widely used approach for grounding large language models (LLMs) in external knowledge. However, configuring a

agentsarxiv-cs-cl