arxiv arXiv cs.CL · 2d ago

Dual-Track Framework for Template-Constrained LaTeX Conversion

A new Dual-Track Framework decouples template formatting from document processing by using an offline track to extract template constraints into a reusable manifest and an online track with a hybrid pipeline. It limits LLM use to reasoning tasks like metadata and bibliographic handling, while applying rule-based engines for deterministic operations, improving structural fidelity, layout compliance, and compilation success over baseline methods.

arxiv arXiv cs.CL · 2d ago

Self-Evolution of Tool-Calling Agents via Divergence-Point Preference Learning

ToolGraph enhances multi-turn tool-using agents by integrating schema topology, transition weights, and history-aware controls. Training with DPO on 161 divergence-point preference pairs improves performance: ToolGraph+DPO achieves a 16.8% relative reward gain over baseline, especially in airline and retail tasks, with reward positivity emerging as the key diagnostic signal.

arxiv arXiv cs.CL · 2d ago

PRIDE: Privileged Information-enhanced Distillation for Empathetic Dialogue Generation

PRIDE introduces a knowledge distillation method that transfers empathetic reasoning from large models to smaller ones using privileged information available only during training. It achieves competitive or superior performance on empathy-related tasks by leveraging structured prompts, multi-source attention, and dual-alignment loss.

media Hugging Face Forums · 2d ago

Coolest Theoretical AI Topics with Realistic AI System Basis

The discussion explores theoretical AI topics that have mathematical foundations and plausible implementation in current AI systems, such as large language models. Topics include reasoning chains, knowledge graphs, and probabilistic reasoning, all of which are grounded in formal math and show potential for real-world AI applications.

arxiv arXiv cs.CL · 2d ago

Language shapes historical credit in large language models

A study of 11 large language models across 21 disputed inventions shows that query language systematically influences which inventor is credited. Lower-status claimants appear more frequently when questions are phrased in their native language, while dominant Anglophone figures remain consistent. The findings suggest language acts as a switch that activates distinct national versions of history, indicating that LLMs function as systems of cultural memory.

arxiv arXiv cs.CL · 2d ago

DART: Training-Free Routing for Adaptive Thinking Budgets

DART enables hybrid reasoning models to route queries between direct answering and extended thinking without training data. It uses two no-think drafts to decide response mode and estimates thinking budget from draft disagreement. DART improves accuracy by up to 9.0 points in math and 22.-5 points in code reasoning while reducing thinking tokens by 15-69% and 51-63% respectively.

arxiv arXiv cs.CL · 2d ago

Memory Contagion: Bias Propagation in Agent Memory

Researchers identify Memory Contagion, a phenomenon where evaluator bias propagates across time in agent memory. Even with perfect memory consolidation, bias spreads to future agents retrieving from the same memory store, with contamination detected as low as p=0.2. The effect varies by bias type: length bias is attenuated, while authority bias is amplified, indicating a bias-dependent interaction.

arxiv arXiv cs.CL · 2d ago

Task-Sensitive Analysis of Intrinsic Self-Correction

A study examines when intrinsic self-correction works by analyzing its performance across different task structures. The research finds that self-correction yields consistent gains only when the task supports explicit constraint verification, complex reasoning revision, or strategy evaluation. The results show SC is effective only in specific task contexts, not universally.

arxiv arXiv cs.CL · 2d ago

CFPO: Counterfactual Policy Optimization for Multimodal Reasoning

CFPO introduces a cross-modal counterfactual enhancement mechanism to improve causal consistency between visual perception and textual reasoning in vision-language models. It achieves 3.17%-6.25% gains over standard RL baselines and 1.32%-2.13% over PAPO, without requiring external rewards or supervision.

arxiv arXiv cs.CL · 2d ago

Judgment-Grounded Expansion for Peer Review Generation

A new human-AI collaboration method called judgment-grounded expansion enables accountable peer review generation. The approach involves a reviewer providing an evaluative claim, which the system expands into review comment candidates through a structured generate-check-refine process. The study addresses scalable evaluation and candidate set curation, showing conformal prediction effectively balances candidate size and coverage.

arxiv arXiv cs.CL · 2d ago

IMLogic Benchmark and RootMem Framework for Implicit Logical Memory Retrieval

IMLogic is the first high-quality benchmark for evaluating implicit logical memory retrieval in long-dialogue scenarios. RootMem introduces a structured, decision-preserving representation called root memory to distill reusable personalized logic from user histories, and uses an LLM-based router to activate relevant memories, outperforming existing retrieval baselines in accuracy.

arxiv arXiv cs.CL · 2d ago

Energy-Based Transformers Predict Reading Difficulty

Energy-based transformers show robust predictive power for reading times across multiple corpora, outperforming surprisal in all cases. The energy measure captures known object/subject asymmetries in relative clause processing and subsumes both attention entropy and surprisal, suggesting it as a unified predictor of reading difficulty.

arxiv arXiv cs.CL · 2d ago

Self-Stigma Is Not Uniform: LLMs Need Persona-Aware Support

A study of 1,174 Reddit users reveals four distinct self-stigma personas. LLMs trained to recognize these personas outperform generic models in targeted responses, though clinical experts prefer generic empathy over persona-matched support. The research highlights a tension between tailored empathy and holistic user preference in stigma-related AI interventions.

arxiv arXiv cs.CL · 2d ago

ReasoningLens: Hierarchical Visualization for Large Reasoning Models

ReasoningLens presents an open-source framework that visualizes and audits long-chain-of-thought traces in large reasoning models. It structures reasoning into interactive hierarchies, uses an agentic auditor for error detection, and identifies model-specific blind spots through systemic reasoning profiles.

arxiv arXiv cs.CL · 2d ago

UnBias-Plus: Detect, Explain, and Rewrite Bias

UnBias-Plus is an open-source toolkit that enables segment-level bias classification, biased span localization, neutral text rewriting, and decision reasoning. It offers multiple access methods including Python, CLI, REST API, and web interfaces, with all source code, models, datasets, and documentation publicly available.

arxiv arXiv cs.CL · 2d ago

TriggerBench: Evaluating Prospective Memory in LLMs

TriggerBench introduces a benchmark to assess prospective memory in large language models, revealing a precision-recall trade-off and attentional fragility. Prospective memory is found to be significantly harder than retrospective memory and correlates with spare reasoning capacity, indicating that PM reflects underlying cognitive resources beyond token count.

arxiv arXiv cs.CL · 2d ago

SelfCompact: Self-Driving Context Compaction for Language Models

SelfCompact enables language models to autonomously decide when and how to compact accumulated context during reasoning. By combining a model-invoked summarization tool with a lightweight rubric that guides compaction based on trajectory structure, it achieves effective adaptive compaction without fine-tuning. Results show it matches or exceeds fixed-interval methods on math and agentic search benchmarks, improving baselines by up to 18.1 points on math and 5-9 points on search, at 30-70% lower token cost.

arxiv arXiv cs.CL · 2d ago

VeriEvol: Scaling Multimodal Mathematical Reasoning with Verifiable Evolution

VeriEvol introduces a verifiable data-construction framework for visual mathematical reasoning, decoupling prompt difficulty and answer reliability. It evolves image-question prompts using type-aware operators and verifies answers via multi-source counter-evidence falsification. On five benchmarks, scaling from 10K to 250K samples improves mean accuracy from 35.42 to 54.73, with a cumulative +3.88 over baseline, driven by evolved prompts and HTV-Agent verification.

arxiv arXiv cs.CL · 2d ago

LLMs Fail to Reliably Self-Report Adversarial Prefills

No large language models reliably detect when their responses were influenced by adversarial prefill attacks. Introspective signals are strongest in safety-related reasoning, but are probe-dependent and can be amplified by LoRA fine-tuning, which paradoxically increases attack success rates.

arxiv arXiv cs.CL · 2d ago

Randomized YaRN Improves Length Generalization for Long-Context Reasoning

Randomized YaRN enhances long-context reasoning by combining YaRN positional extrapolation with randomized positional encoding and a length curriculum. It outperforms standard fine-tuning on benchmarks like BABILong and MRCR, showing significant gains at far out-of-distribution context lengths.