SwissGov-RSD is the first naturalistic cross-lingual document-level benchmark with human token-level semantic difference annotations, on which both LLMs and encoders show a large performance gap relative to simpler settings.
super hub Mixed citations
online" 'onlinestring :=
Mixed citation behavior. Most common role is background (60%).
171
Pith papers citing it
Background
60% of classified citations
hub tools
citation-role summary
background 6
dataset 2
baseline 1
method 1
citation-polarity summary
authors
co-cited works
representative citing papers
Introduces Unlearning Depth Score (UDS) via activation patching to quantify LLM unlearning depth and claims it outperforms 20 other metrics in faithfulness and robustness on 150 models.
Builds a 2M-contribution graph from 230k papers with 12.5M prerequisite links and reports 0.48 MAP on temporal backtesting for predicting enabling technologies.
SkillSafetyBench is a benchmark of 155 cases across 47 tasks and 6 risk domains showing that non-user attacks via skills, artifacts, or environments can consistently induce unsafe agent behavior.
Local attention in fixed-precision transformers introduces a second past operator in linear temporal logic, strictly increasing expressivity over global attention alone, with hybrids being most expressive.
SecGoal is a new expert-annotated benchmark dataset covering 15 protocols for extracting formalizable security goals from natural-language documents, paired with the AIFG framework; fine-tuned Gemma2-9B reaches 66.6% precision and 97.6% recall on held-out protocols.
OCR-Memory encodes agent trajectories as images with visual anchors and retrieves verbatim text via locate-and-transcribe, yielding gains on long-horizon benchmarks under strict context limits.
Cross-cultural survey of 4,641 participants shows LLM emotional support adoption varies widely by country and demographics, with socioeconomic status as strongest predictor of trust and use, and English-speaking nations more accepting than others in Europe.
VLMs reach only 42.1% exact accuracy on counting pushups in videos, with weaker models exploiting modal counts, and 1k-sample fine-tuning transfers gains to MVBench, PerceptionTest, and TVBench.
A controlled formal language task reveals fine-tuning outperforms in-context learning on in-distribution generalization but equals it on out-of-distribution, with ICL showing greater sensitivity to model size and tokenization.
StoryTR is a new benchmark and agentic data pipeline that adds explicit Theory of Mind reasoning chains to train smaller video retrieval models, yielding a 15% relative IoU gain over larger baselines on narrative content.
Language models frequently violate temporal scope stability in multi-turn dialogues by drifting toward present-day assumptions even when they possess the correct facts.
BERAG applies Bayesian ensemble weighting of individual documents via token-by-token posterior updates in retrieval-augmented generation, yielding gains on knowledge-based visual QA tasks.
BiasedTales-ML provides a parallel multilingual corpus of LLM-generated children's stories that reveals substantial cross-lingual differences in narrative attributes not captured by English-centric analyses.
Conjunctive prompt attacks split adversarial elements across agents and routing paths in multi-agent LLM systems, evading isolated defenses and succeeding through topology-aware optimization.
VisPCO uses continuous relaxation, straight-through estimators, and budget-aware Pareto-frontier learning to automatically discover optimal visual token pruning configurations that approximate grid-search results across VLMs and benchmarks.
HintPilot synthesizes semantics-preserving compiler hints via retrieval-augmented LLM generation and profiling-guided refinement, delivering up to 6.88x geometric mean speedup over -Ofast on PolyBench and HumanEval-CPP while preserving correctness.
ADAPT augments planners with affordance reasoning to raise task success in environments with unspecified and time-varying object affordances, and a LoRA-finetuned VLM backend beats GPT-4o on the new DynAfford benchmark.
Schema-key wording functions as an implicit instruction channel under constrained decoding, with experiments showing that rephrasing only the keys can substantially change accuracy on math benchmarks while prompt, model, structure, and decoding remain unchanged.
SPAGBias reveals that LLMs form nuanced gender associations with specific urban micro-spaces that exceed real-world distributions and produce failures in planning and descriptive tasks.
CAR is a new retrieval objective that targets the currently active authority set rather than most-similar documents, with theorems on coverage conditions and evaluations showing two-stage methods outperform dense retrieval on authority-governed datasets.
Multimodal ICL lags text-only ICL in few-shot settings due to weak cross-modal reasoning alignment and unreliable task mapping transfer, with an inference-stage method proposed to strengthen transfer.
Tabular QA LLMs are overconfident, but Multi-Format Agreement using Markdown/HTML/JSON/CSV variants improves AUROC to 0.80 and cuts calibration error by 44-63% at lower cost than sampling.
EgoEsportsQA is a new egocentric video QA benchmark from esports matches that shows state-of-the-art Video-LLMs reach only 71.58% accuracy and struggle more with tactical reasoning than basic perception.
citing papers explorer
-
SwissGov-RSD: A Human-annotated, Cross-lingual Benchmark for Token-level Recognition of Semantic Differences Between Related Documents
SwissGov-RSD is the first naturalistic cross-lingual document-level benchmark with human token-level semantic difference annotations, on which both LLMs and encoders show a large performance gap relative to simpler settings.
-
Measuring the Depth of LLM Unlearning via Activation Patching
Introduces Unlearning Depth Score (UDS) via activation patching to quantify LLM unlearning depth and claims it outperforms 20 other metrics in faithfulness and robustness on 150 models.
-
The Scientific Contribution Graph: Automated Literature-based Technological Roadmapping at Scale
Builds a 2M-contribution graph from 230k papers with 12.5M prerequisite links and reports 0.48 MAP on temporal backtesting for predicting enabling technologies.
-
SkillSafetyBench: Evaluating Agent Safety under Skill-Facing Attack Surfaces
SkillSafetyBench is a benchmark of 155 cases across 47 tasks and 6 risk domains showing that non-user attacks via skills, artifacts, or environments can consistently induce unsafe agent behavior.
-
Characterizing the Expressivity of Local Attention in Transformers
Local attention in fixed-precision transformers introduces a second past operator in linear temporal logic, strictly increasing expressivity over global attention alone, with hybrids being most expressive.
-
SecGoal: A Benchmark for Extracting Formalizable Security Goals from Protocol Documents
SecGoal is a new expert-annotated benchmark dataset covering 15 protocols for extracting formalizable security goals from natural-language documents, paired with the AIFG framework; fine-tuned Gemma2-9B reaches 66.6% precision and 97.6% recall on held-out protocols.
-
OCR-Memory: Optical Context Retrieval for Long-Horizon Agent Memory
OCR-Memory encodes agent trajectories as images with visual anchors and retrieves verbatim text via locate-and-transcribe, yielding gains on long-horizon benchmarks under strict context limits.
-
From Chatbots to Confidants: A Cross-Cultural Study of LLM Adoption for Emotional Support
Cross-cultural survey of 4,641 participants shows LLM emotional support adoption varies widely by country and demographics, with socioeconomic status as strongest predictor of trust and use, and English-speaking nations more accepting than others in Europe.
-
PushupBench: Your VLM is not good at counting pushups
VLMs reach only 42.1% exact accuracy on counting pushups in videos, with weaker models exploiting modal counts, and 1k-sample fine-tuning transfers gains to MVBench, PerceptionTest, and TVBench.
-
Fine-tuning vs. In-context Learning in Large Language Models: A Formal Language Learning Perspective
A controlled formal language task reveals fine-tuning outperforms in-context learning on in-distribution generalization but equals it on out-of-distribution, with ICL showing greater sensitivity to model size and tokenization.
-
StoryTR: Narrative-Centric Video Temporal Retrieval with Theory of Mind Reasoning
StoryTR is a new benchmark and agentic data pipeline that adds explicit Theory of Mind reasoning chains to train smaller video retrieval models, yielding a 15% relative IoU gain over larger baselines on narrative content.
-
Evaluating Temporal Consistency in Multi-Turn Language Models
Language models frequently violate temporal scope stability in multi-turn dialogues by drifting toward present-day assumptions even when they possess the correct facts.
-
BERAG: Bayesian Ensemble Retrieval-Augmented Generation for Knowledge-based Visual Question Answering
BERAG applies Bayesian ensemble weighting of individual documents via token-by-token posterior updates in retrieval-augmented generation, yielding gains on knowledge-based visual QA tasks.
-
BIASEDTALES-ML: A Multilingual Dataset for Analyzing Narrative Attribute Distributions in LLM-Generated Stories
BiasedTales-ML provides a parallel multilingual corpus of LLM-generated children's stories that reveals substantial cross-lingual differences in narrative attributes not captured by English-centric analyses.
-
Conjunctive Prompt Attacks in Multi-Agent LLM Systems
Conjunctive prompt attacks split adversarial elements across agents and routing paths in multi-agent LLM systems, evading isolated defenses and succeeding through topology-aware optimization.
-
VisPCO: Visual Token Pruning Configuration Optimization via Budget-Aware Pareto-Frontier Learning for Vision-Language Models
VisPCO uses continuous relaxation, straight-through estimators, and budget-aware Pareto-frontier learning to automatically discover optimal visual token pruning configurations that approximate grid-search results across VLMs and benchmarks.
-
HintPilot: LLM-based Compiler Hint Synthesis for Code Optimization
HintPilot synthesizes semantics-preserving compiler hints via retrieval-augmented LLM generation and profiling-guided refinement, delivering up to 6.88x geometric mean speedup over -Ofast on PolyBench and HumanEval-CPP while preserving correctness.
-
ADAPT: Benchmarking Commonsense Planning under Unspecified Affordance Constraints
ADAPT augments planners with affordance reasoning to raise task success in environments with unspecified and time-varying object affordances, and a LoRA-finetuned VLM backend beats GPT-4o on the new DynAfford benchmark.
-
Schema Key Wording as an Instruction Channel in Structured Generation under Constrained Decoding
Schema-key wording functions as an implicit instruction channel under constrained decoding, with experiments showing that rephrasing only the keys can substantially change accuracy on math benchmarks while prompt, model, structure, and decoding remain unchanged.
-
SPAGBias: Uncovering and Tracing Structured Spatial Gender Bias in Large Language Models
SPAGBias reveals that LLMs form nuanced gender associations with specific urban micro-spaces that exceed real-world distributions and produce failures in planning and descriptive tasks.
-
Controlling Authority Retrieval: A Missing Retrieval Objective for Authority-Governed Knowledge
CAR is a new retrieval objective that targets the currently active authority set rather than most-similar documents, with theorems on coverage conditions and evaluations showing two-stage methods outperform dense retrieval on authority-governed datasets.
-
Why Multimodal In-Context Learning Lags Behind? Unveiling the Inner Mechanisms and Bottlenecks
Multimodal ICL lags text-only ICL in few-shot settings due to weak cross-modal reasoning alignment and unreliable task mapping transfer, with an inference-stage method proposed to strengthen transfer.
-
Calibrated Confidence Estimation for Tabular Question Answering
Tabular QA LLMs are overconfident, but Multi-Format Agreement using Markdown/HTML/JSON/CSV variants improves AUROC to 0.80 and cuts calibration error by 44-63% at lower cost than sampling.
-
EgoEsportsQA: An Egocentric Video Benchmark for Perception and Reasoning in Esports
EgoEsportsQA is a new egocentric video QA benchmark from esports matches that shows state-of-the-art Video-LLMs reach only 71.58% accuracy and struggle more with tactical reasoning than basic perception.
-
METRO: Towards Strategy Induction from Expert Dialogue Transcripts for Non-collaborative Dialogues
METRO induces both short-term actions and long-term planning from expert transcripts into a Strategy Forest, outperforming prior methods by 9-10% on two non-collaborative dialogue benchmarks.
-
Too Nice to Tell the Truth: Quantifying Agreeableness-Driven Sycophancy in Role-Playing Language Models
Agreeableness in AI personas reliably predicts sycophantic behavior in 9 of 13 tested language models.
-
Learning and Enforcing Context-Sensitive Control for LLMs
A framework learns context-sensitive constraints automatically from LLM outputs to enforce perfect adherence during generation without manual specification.
-
SPASM: Stable Persona-driven Agent Simulation for Multi-turn Dialogue Generation
SPASM introduces a stability-first framework with Egocentric Context Projection to maintain consistent personas and eliminate echoing in multi-turn LLM agent dialogues.
-
SiMing-Bench: Evaluating Procedural Correctness from Continuous Interactions in Clinical Skill Videos
SiMing-Bench shows current MLLMs have weak agreement with physicians on procedural correctness in clinical videos, with intermediate step judgments remaining poor even when overall scores look acceptable.
-
TaxPraBen: A Scalable Benchmark for Structured Evaluation of LLMs in Chinese Real-World Tax Practice
TaxPraBen is a new benchmark with 14 datasets and a structured evaluation method for measuring LLM performance on Chinese real-world tax tasks and scenarios.
-
Sell More, Play Less: Benchmarking LLM Realistic Selling Skill
SalesLLM provides an automatic evaluation framework for LLM sales dialogues that correlates 0.98 with human experts and shows top models approaching human performance while weaker ones lag.
-
GENFIG1: Visual Summaries of Scholarly Work as a Challenge for Vision-Language Models
GENFIG1 is a new benchmark that tests whether vision-language models can create effective Figure 1 visuals capturing the central scientific idea from paper text.
-
Style Amnesia: Investigating Speaking Style Degradation and Mitigation in Multi-Turn Spoken Language Models
Spoken language models exhibit style amnesia and fail to maintain instructed paralinguistic styles across multi-turn conversations, with explicit recall offering partial mitigation.
-
CricBench: A Multilingual Benchmark for Evaluating LLMs in Cricket Analytics
CricBench is the first multilingual Text-to-SQL benchmark for cricket analytics, showing LLMs achieve over 98% execution accuracy but under 29% semantic correctness with a 37-55 point gap versus general benchmarks like BIRD.
-
MURPHY: Feedback-Aware GRPO with Retrospective Credit Assignment for Multi-Turn Code Generation
MURPHY improves code generation pass rates by up to 6% through retrospective credit assignment on multi-turn feedback trees using max or mean reward propagation.
-
TSVer: A Benchmark for Fact Verification Against Time-Series Evidence
TSVer is a new benchmark dataset for fact verification against time-series evidence, with 304 annotated real-world claims, 400 time series, verdicts, and justifications, plus baseline results showing current models struggle.
-
AudioRole: An Audio Dataset for Character Role-Playing in Large Language Models
AudioRole provides 1M+ character-grounded audio-text dialogues from TV series plus ARP-Eval to train and measure audio role-playing models, with ARP-Model showing 0.31 acoustic and 0.36 content personalization scores.
-
SiDiaC: Sinhala Diachronic Corpus
SiDiaC is a new historical corpus of Sinhala literary works spanning the 5th to 20th centuries, constructed via OCR digitization, orthography modernization, and genre-based annotation.
-
V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Models
V-SEAM combines concept-level visual semantic editing with attention head modulation to identify positive and negative contributors across object, attribute, and relationship levels, then uses this to improve VLM performance on VQA benchmarks.
-
Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models
Evaluations of 53 LLMs on 14 basic math tasks show reasoning models use ~18x more tokens with sometimes lower accuracy, non-monotonic gains from extended budgets, and sharp performance drops under token constraints.
-
FastKV: Decoupling of Context Reduction and KV Cache Compression for Prefill-Decoding Acceleration
FastKV decouples prefill context reduction via Token-Selective Propagation from independent KV cache selection, delivering up to 1.82x prefill and 2.87x decoding speedups while matching decoding-only accuracy.
-
Topic-Based Watermarks for Large Language Models
A topic-guided watermarking scheme partitions the LLM vocabulary into topic-aligned token subsets and green-lists relevant tokens based on the input prompt to embed detectable marks while preserving text quality and improving robustness to attacks.
-
Guarded Repair for Harm-Aware Post-hoc Replacement of LLM Mathematical Reasoning
GuardedRepair uses guarded best-of-N repair with symbolic checks, semantic diagnostics, and conservative policies to selectively replace LLM reasoning traces, raising GSM8K accuracy from 95.60% to 96.89% and ASDiv from 78.40% to 87.60% without breaking correct cases.
-
How Human-Like Are Large Language Models? A Register-Aware Linguistic Evaluation Framework
A new evaluation framework using MMD on Biber features shows LLMs deviate from human linguistic distributions across registers, with closest models varying by register rather than size.
-
AgentAtlas: Beyond Outcome Leaderboards for LLM Agents
AgentAtlas introduces a diagnostic taxonomy and audit protocol to evaluate LLM agent control decisions and trajectories beyond final outcome success.
-
Chain-of-Procedure: Hierarchical Visual-Language Reasoning for Procedural QA
Introduces ProcedureVQA benchmark and Chain-of-Procedure framework that improves VLM next-step prediction in procedures by up to 13% over baselines.
-
Persian MusicGen: A Large-Scale Dataset and Culturally-Aware Generative Model for Persian Music
Introduces the first large-scale Persian music dataset and shows fine-tuned MusicGen produces compositions more aligned with Persian stylistic conventions via tag-based evaluation.
-
GSM-SEM: Benchmark and Framework for Generating Semantically Variant Augmentations
GSM-SEM is a reusable framework for creating semantically variant augmentations of math benchmarks like GSM8K that alter facts but preserve answers and difficulty, with evaluations showing LLM performance drops of up to 28% on the new variants.
-
ScrapMem: A Bio-inspired Framework for On-device Personalized Agent Memory via Optical Forgetting
ScrapMem reports SOTA 51.0% Joint@10 on ATM-Bench with up to 93% memory reduction and 70.3% Recall@10 via optical forgetting and EM-Graph.
-
Debiasing Reward Models via Causally Motivated Inference-Time Intervention
Neuron-level inference-time intervention reduces multiple biases in reward models, enabling 2B and 7B models to match 70B performance on LLM alignment benchmarks without trade-offs.