Search: “arxiv-cs-cl”

Which Skill to Distill? SGUID: Selecting a Compact Skill Bank for Model-Skill Co-Evolution

arXiv:2610.12367v1 Announce Type: new Abstract: Skills, reusable procedural guidance added at inference, can substantially improve LLM downstream performance (Li et al., 2026). Prior work retrieves sk

9 Oct 2026

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Why On-Policy Distillation Sometimes Fails: Vanishing Learning Signals

DGX agent

arXiv:2610.11247v1 Announce Type: cross Abstract: On-policy distillation (OPD) enables effective capability transfer between language models, yet the mechanisms underlying its failures are not fully u

model-releasesarxiv-cs-cl

9 Oct 2026

Wieszcz-XIX: A 3.1-Billion-Word Corpus of Pre-1918 Polish and Temporally Bounded Language Models Trained From Scratch

arXiv:2610.10592v1 Announce Type: new Abstract: Historical Polish is well documented as a language but annotated in machine-readable form only to about a million words for the period this paper covers

9 Oct 2026

WOVEN: Weaving Visual World Modeling into Multimodal LLMs

arXiv:2610.12417v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) struggle with spatial, embodied, physical, and temporal reasoning. We hypothesize that these failures reflect

9 Oct 2026

A Comparative Study of Evaluation Metrics for Long-Document Financial Narrative Summarization with Transformers

arXiv:2610.09529v1 Announce Type: new Abstract: There are more than 2,000 listed companies on the UK's London Stock Exchange, divided into 11 sectors who are required to communicate their financial re

A conceptual framework for ideology in online discourse beyond the left and right

arXiv:2603.18945v2 Announce Type: replace-cross Abstract: Computational social science (CSS) has largely operationalized ideology along a single left/right partisan axis when studying online discourse

A Deafening Silence: Catastrophic Forgetting Lives in the Output Embeddings of Tokens the Data Never Speaks

arXiv:2610.09835v1 Announce Type: new Abstract: Continual pre-training and fine-tuning in Large Language Models (LLMs) inevitably induce catastrophic forgetting, typically mitigated by replay using of

Advancing LLM-based phoneme-to-grapheme for multilingual speech recognition

arXiv:2603.29217v3 Announce Type: replace-cross Abstract: Phoneme-based ASR factorizes recognition into speech-to-phoneme (S2P) and phoneme-to-grapheme (P2G), enabling cross-lingual acoustic sharing w

Adversarial Images Hijack Web Agents from Visual Grounding to Browser Execution

arXiv:2610.09240v1 Announce Type: cross Abstract: Modern web agents built on large vision-language models process webpages, select relevant UI elements, and translate model outputs into browser action

Alice: A Large-Scale German Benchmark for Rubric-Based Multi-Dimensional Automatic Short Answer Scoring

arXiv:2610.09661v1 Announce Type: new Abstract: Automatic Short Answer Scoring (ASAS) is central to NLP for Education. However, openly available benchmarks remain scarce, and existing datasets largely

APE: Selective Fine-tuning with Acceptance Criteria for Language Model Adaptation

arXiv:2505.19912v3 Announce Type: replace Abstract: We present Adjacent Possible Exploration (APE), a selective fine-tuning method for adapting large language models that systematically explores param

ARCS: Towards Precise Text-to-SQL via Structured Disambiguation

arXiv:2610.09396v1 Announce Type: new Abstract: As text-to-SQL systems move beyond demonstrations toward real-world deployment, ambiguity in user questions becomes a primary source of errors. Such amb

Arctic Questions, Missing Answers: A Dataset and Benchmark for LLM Abstention in Arctic Science

arXiv:2610.09446v1 Announce Type: new Abstract: Large language models (LLMs) should abstain from scientific multiple-choice questions when no option is valid, but frequent abstention alone does not de

Attention-Mass Condensation for Sparse Decoding

arXiv:2602.06317v3 Announce Type: replace-cross Abstract: Attention-mass concentration creates an opportunity for sparse decoding, but retained mass alone does not guarantee a stable greedy decision:

BanglaRhet: Benchmarking Classical and Transformer Models for Rhetorical and Persuasion Detection in Bangla Political Speech

arXiv:2610.09464v1 Announce Type: new Abstract: Political discourse often uses rhetorical and persuasive language to frame narratives, influence public opinion, and mobilize audiences. While Bangla na

BEACON-SP: Ontology-Grounded GraphRAG Framework for Clinical Suicide Risk Assessment

arXiv:2610.09026v1 Announce Type: cross Abstract: We present BEACON-SP, an ontology-grounded Graph Retrieval-Augmented Generation (GraphRAG) framework for clinician-facing decision support in behavior

Agents

Beyond Outcome Rewards: Constructing and Assigning Retrieval Credit for Search Agents

arXiv:2610.10179v1 Announce Type: new Abstract: Search agents enable Large Language Models (LLMs) to iteratively retrieve and use information for complex multi-hop questions. Reinforcement Learning wi

agentsarxiv-cs-cl

Beyond Risk Prediction: Evidence Grounding and Psychosocial Factor Verification for Explainable Suicide Risk Assessment

arXiv:2610.08842v1 Announce Type: new Abstract: Identifying suicide risk from social networking services (SNS) posts is important for detecting suicide-related signals in online environments. However,

Beyond the Sycophancy Score: How Task, Model, and Pressure Shape LLM Yielding

arXiv:2610.08840v1 Announce Type: new Abstract: Large language models (LLMs) often abandon a correct answer, or endorse a user's position, once the user pushes back. This behavior, called sycophancy,

Bookkeeping, Composition, or Unreachable Gold? Reading MemoryAgentBench's Conflict-Resolution Scores Against a Frozen Last-Write Resolver

arXiv:2610.09193v1 Announce Type: cross Abstract: MemoryAgentBench's Conflict Resolution split is read as measuring 'selective forgetting'. We execute the benchmark's own rule - the newest statement a

Boundary-Free Contextual Biasing: Depth-Adaptive Gating and Reading-Space Matching for Unsegmented Languages

arXiv:2610.09467v1 Announce Type: cross Abstract: Contextual biasing supplies an ASR system with a list of expected words at inference time, but existing methods rely on word boundaries that Japanese

Bridge Routing Heads: Where Multilingual Multi-hop Reasoning Lives in LLMs

arXiv:2610.09733v1 Announce Type: new Abstract: Multilingual LLMs answer the same multi-hop reasoning question across languages, but we lack a mechanistic account of whether they share an internal cir

Cache the Encoder Within:Compact, Reusable Memory across LLM Queries

arXiv:2610.10058v1 Announce Type: new Abstract: Repeated queries over shared documents incur redundant encoding, while caching model states introduces persistent storage costs. Building on CoMem's int

CARE: Certifying Acceleration for Vision-Language-Action Inference

arXiv:2610.08917v1 Announce Type: new Abstract: While vision-language-action (VLA) models have advanced rapidly, running them at every control step remains expensive. Prior work accelerates VLA infere

Certified by Abstention: Distribution-Free Guarantees for Chain-of-Thought Verifiers at Small Calibration Budgets

arXiv:2610.09541v1 Announce Type: cross Abstract: Signals that predict whether a chain-of-thought (CoT) trace is correct are compared by AUC, but deploying one requires a threshold with a guarantee. W

CHASE: Channel-Aligned Structure Exploitation for Geometry-Aware Model Engineering

arXiv:2610.09476v1 Announce Type: cross Abstract: Geometric and Spectral Alignment (GSA) characterizes trained networks through spectral concentration, physical-channel alignment, support structure, a

Child ASR Adaptation with Adult Retention: An Empirical Study

arXiv:2610.08827v1 Announce Type: new Abstract: Automatic Speech Recognition (ASR) systems often underperform for children and non-native speakers, while adapting adult ASR models to child speech can

Coding-Agent Benchmarks Should Match Their Users' Task Flows

arXiv:2610.09633v1 Announce Type: cross Abstract: The evaluation of coding agents generally strives to be as realistic as possible. In our study, we collect 4,782 agent sessions of real software engin

CoDR: Training-Free Confidence-Drift Remasking for Diffusion Language Models

arXiv:2610.08833v1 Announce Type: new Abstract: Masked diffusion language models (MDLMs) decode by repeatedly committing tokens to masked positions, but these commitments are usually irreversible. A t

Collaborative Reasoning Distillation via Cross-Feedback and Coherent Curation

arXiv:2610.09587v1 Announce Type: cross Abstract: Reasoning capabilities are critical for advancing Large Language Models, yet current approaches either require massive computational budgets or strugg

Collective Behavior of AI Agents: the Case of Moltbook

arXiv:2602.09270v2 Announce Type: replace-cross Abstract: We present a large scale data analysis of Moltbook, a Reddit-style social media platform exclusively populated by AI agents. Analyzing over 4

Constitution-Guided Watermarking

arXiv:2610.09552v1 Announce Type: cross Abstract: Watermarking enables language model providers to identify text generated by their models. However, its desired properties can conflict (ie~stronger wa

Constrained-Action AI Remediation for SIEM/XDR via a NeMo-Guardrails Proxy

arXiv:2610.09906v1 Announce Type: cross Abstract: Security Operations Centers (SOCs) for information technology and operational technology share one incident-response problem: a flood of correlated al

Agents

Constraint Tree Exploration for Learning from Language Feedback

arXiv:2610.09107v1 Announce Type: cross Abstract: Natural-language feedback in interactive learning often explains why an action failed by pointing to violated requirements. Misinterpreting this feedb

agentsarxiv-cs-cl

CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution

arXiv:2610.10426v1 Announce Type: new Abstract: Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles error recovery. Existi

Decoupling Exploration from Optimization in RLVR

arXiv:2610.10536v1 Announce Type: cross Abstract: Modern language models undergo reinforcement learning with verifiable rewards (RLVR) on top of already-trained checkpoints. A key promise of RLVR is t

Decoupling Logic from Persona: Structural Immunity of Edge LLM Agents to Context Pollution

arXiv:2610.09772v1 Announce Type: new Abstract: Small language-model agents on edge devices must hold a persona and reason correctly at once, inside one context window that fills with conversational h

Despite Instructions: Frontier Agents Improvise Covert Channels at Test Time

arXiv:2609.32701v2 Announce Type: replace-cross Abstract: In security-sensitive applications, language-model agents are often required to coordinate without disclosing confidential information. Yet re

Dialect-Robust Speech Language Models with Synthetic Pseudo-Dialect Augmentation

arXiv:2610.09321v1 Announce Type: new Abstract: Speech Language Model (SLM) performance often degrades on dialects due to data scarcity. Conventional text-to-speech (TTS) augmentation struggles to cov

Document-Level Text Simplification in Estonian Using Large Language Models

arXiv:2610.10378v1 Announce Type: new Abstract: Document-level text simplification involves transformations that go beyond sentence-internal edits, addressing discourse coherence, anaphora resolution,

EASE: Entropy-Adaptive Distribution Shaping for Evading AI-generated Text Detectors

arXiv:2610.09976v1 Announce Type: new Abstract: AI-generated text (AIGT) detection can be sensitive to the decoding choices of the source large language model (LLM). We observe that perturbing next-to

Emo-Jev: Probabilistic Reasoning for Emotion Classification with Jev

arXiv:2610.08829v1 Announce Type: new Abstract: Jev offers an alternative interface for language understanding: given an input and predefined questions, it returns probabilistic decisions rather than

Local Ai

EmphTTS: an emphasis-control TTS with reinforcement learning

arXiv:2609.27599v2 Announce Type: cross Abstract: Generating controllable and human-like emphasis remains an open challenge in text-to-speech, even when explicit emphasis control signals are provided

local-aiarxiv-cs-cl

EngramEdit: Decoupled Knowledge Updates in LLMs through Conditional Memory

arXiv:2610.10533v1 Announce Type: new Abstract: Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language model

Epistemic Policy Divergence in Multi-Turn LLM Contamination: A Protocol-Gradient Investigation

arXiv:2609.35308v2 Announce Type: replace Abstract: Large language models treat conversation history as unverified context, so false premises injected into prior turns can be adopted as fact, a failur

Agents

ExperienceIndex: Artifact-Grounded Memory

arXiv:2610.10091v1 Announce Type: new Abstract: Knowledge-intensive tasks require answering many questions by reasoning about a shared corpus of artifacts (e.g., court cases, or scientific literature)

agentsarxiv-cs-cl

Hardware

Expert Coupling in MoE Pretraining: Reducing All-to-All Overhead with Correlated Placement and Token Shuffling

arXiv:2610.09372v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) layers replace the feed-forward block of a Transformer with E expert networks, and each token is routed to k of these experts.

hardwarearxiv-cs-cl

Local Ai

Few Bits, One Law: Toward W2A4KV2

arXiv:2610.09202v1 Announce Type: cross Abstract: Extreme low-bit LLM compression is most challenging when weights, activations, and KV caches are quantized together: their distributions differ, and q

local-aiarxiv-cs-cl