JackZebra performs long-horizon route hijacking of vision-based AVs by converting adversarial patches into online-selected steering primitives via closed-loop control from an attacker vehicle.
super hub Mixed citations
librosa/librosa: 0.6.3
Mixed citation behavior. Most common role is background (57%).
193
Pith papers citing it
Background
57% of classified citations
hub tools
citation-role summary
background 16
method 9
other 3
baseline 1
extension 1
citation-polarity summary
co-cited works
representative citing papers
Fetal-Gauge benchmark shows state-of-the-art vision-language models reach only 55% accuracy on fetal ultrasound tasks, well below clinical needs and highlighting the requirement for domain-adapted models.
Derives non-asymptotic 2-norm and infinity-norm error bounds for deterministic and stochastic variants of OPTQ and Qronos PTQ algorithms.
SpinGTP uses Spin-Weighted Spherical Harmonics to complete the Gaunt Tensor Product, achieving full E(3)-equivariance with GTP-like scaling and better handling of chiral and parity-odd cases.
A new 321-patient multi-center breast FNAC WSI dataset with 7398 patch-level C1-C5 annotations is released for AI-assisted classification research.
The cubic sum rule of S(q,ω) is tested as a kinetic energy estimator using PIMC data and dielectric models for the uniform electron gas, confirming consistency with thermodynamics but exposing flaws in semi-classical approximations.
STEMGym benchmark demonstrates that perception pipelines dominate dose efficiency in autonomous STEM over navigation methods across 33 agent setups.
A parton-shower-inspired local subtraction scheme for double-real corrections in color singlet decays is introduced, with finiteness verified for the e+e- to qqbar remainder and phase-space integrals computed analytically and via sector decomposition.
ColumnKeeper provides the first mitigations for ColumnDisturb using per-subarray counters or probabilistic refresh, with low overheads at 1M and 128K thresholds.
FAIR+S extends FAIR with sustainability metrics and is validated via expert survey confirming importance but revealing awareness gaps in green practices.
Proves sharp threshold on mutation parameter χ for (1+1)-EA on Dynamic Binary Value and Uniform weight dynamic linear problems, yielding O(n log n) runtime below threshold and 2^Ω(n) above, plus a second stagnation-distance threshold for the former.
POPSICLE introduces benchmark datasets for cryoET segmentation and localization built from the CryoET Data Portal.
First unified benchmark finds GLR family has only 3x median slowdown over LR(1) on deterministic grammars and is the fastest among generalized parsers.
Introduces ESAS benchmark dataset using LLM-assisted event injection into acoustic scenes, showing significant performance drops in existing ASC models.
Optimal SSB frame origin for LGWA cuts sampling time by 10x and tightens chirp mass and sky position constraints for stellar-mass binaries beyond LVK performance.
Randomized experiment finds AI draft assistance raises feedback provision by teaching assistants 10.8 percentage points without harming quality.
A matched-pair protocol and Accurate Differentiation Rate metric reveal that conventional LLM accuracy on SAT problems is often inflated by over-predicting satisfiability, while cross-representation agreement exceeds 80 percent for most models.
A new greedy rebalancing algorithm for multi-constraint hypergraphs, integrated into Mt-KaHyPar, reduces geometric mean connectivity by 11.5% versus Metis while improving partition balance reliability.
mcp-attested adds attested admission to MCP via signed clearance assertions at a well-known URI, deny-by-default tool allowlists, and gated enforcement that turns warnings into hard denials with tamper-evident logs.
Bayesian optimization identifies cement-salt hydrate composites achieving up to five times higher specific energy than prior cement-based TCES materials, with LiCl-based formulations reaching 458 kJ/kg.
FLDD learns non-Markovian marginal and posterior distributions for the forward process so a factorized reverse process can match the target better and produce higher-quality samples in fewer steps.
Human face perception aligns with neural networks trained on inverse-generative and naturalistic discriminative tasks, as these best predict human dissimilarity judgments on controversial and random face pairs.
An SMT-based active learning algorithm learns minimal nondeterministic weighted automata over arbitrary semirings, with partial correctness proofs, a sufficient termination condition, and experiments showing smaller models and fewer queries than baselines.
Rabi coupling allows a third component to join a self-bound binary quantum droplet in Bose gases, stabilized by finite detuning despite added repulsive forces.
citing papers explorer
-
ContextualJailbreak: Evolutionary Red-Teaming via Simulated Conversational Priming
ContextualJailbreak uses evolutionary search over simulated primed dialogues with novel mutations to reach 90-100% attack success on open LLMs and transfers to some closed frontier models at 15-90% rates.
-
Evaluating In-Context Translation with Synchronous Context-Free Grammar Transduction
LLM in-context translation accuracy falls sharply with larger grammars and longer sentences, and drops further when source and target languages differ in morphology or writing system, with common errors including wrong word recall, hallucinations, and untranslated source words.
-
Fragile Knowledge, Robust Instruction-Following: The Width Pruning Dichotomy in Llama-3.2
Width pruning in Llama-3.2 models reduces parametric knowledge while enhancing instruction-following and preserving reasoning.
-
RWKV: Reinventing RNNs for the Transformer Era
RWKV uses a linear attention mechanism to deliver Transformer-level performance with RNN-style inference efficiency, demonstrated at up to 14 billion parameters.
-
UR-BERT: Scaling Text Encoders for Massively Multilingual TTS Through Universal Romanization and Speech Token Prediction
UR-BERT scales multilingual TTS encoders to 495 languages via Romanization unification and speech token prediction, outperforming baselines with better generalization.
-
Leveraging LLMs for Grammar Adaptation: A Study on Metamodel-Grammar Co-Evolution
LLM prompting achieves 100% grammar adaptation consistency on small test DSLs and reuses adaptations across QVTo evolution steps, outperforming rule-based methods, but drops below 90% on large grammars like EAST-ADL.
-
KoRe: Compact Knowledge Representations for Large Language Models
KoRe encodes 1-hop knowledge graph subgraphs as compact discrete tokens for injection into LLMs, achieving competitive benchmark performance with up to 10x token reduction.
-
CodeT5+: Open Code Large Language Models for Code Understanding and Generation
CodeT5+ is a flexible encoder-decoder LLM family for code pretrained with diverse objectives on multilingual corpora and initialized from existing LLMs, achieving state-of-the-art results on code generation, completion, math programming, and retrieval tasks including new SoTA on HumanEval with the 1
-
BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
BLOOM is a 176B-parameter open-access multilingual language model trained on the ROOTS corpus that achieves competitive performance on benchmarks, with improved results after multitask prompted finetuning.
-
Green Prompting: Characterizing Prompt-driven Energy Costs of LLM Inference
Empirical tests on three LLMs show prompt semantics and task keywords drive inference energy costs more than length, with varying patterns by task.