archive
Every paper Pith has read. Search by title, abstract, or pith.
85880 papers indexed · page 137
-
Noise schedule produces linear Wilson action decay on Lie groups
Noise scheduling and linear dynamics in diffusion models on Lie groups
-
New system converges cross-sector threat response in 12-20 seconds
Federated Stream-Processing and Latency-Gated Response for Cross-Sector Threat Detection and Collaborative Containment
-
Clarification boosts LLM prompt injection success from 2% to 35%
ASPI: Seeking Ambiguity Clarification Amplifies Prompt Injection Vulnerability in LLM Agents
-
Nonuniform periodic wavelet frames constructed on non-Archimedean fields
Nonuniform Periodic Wavelet Frames on Non-Archimedean Fields
-
Masses predicted for 1F-wave heavy baryons
Investigating the mass spectra of $1F$-wave singly heavy $\Sigma_{Q}$, $\Xi^{\prime}_{Q}$, and $\Omega_{Q}$ baryons
-
Higher magnetic fields suppress zonal flows by shifting turbulence to high frequencies
Role of Magnetic Field in the Redistribution of Turbulence from Large-Scale Structures to Small-Scale Fluctuations
-
TClone forks live GUI workspaces at low latency for agents
TClone: Low-Latency Forking of Live GUI Environments for Computer-Use Agents
-
Flexible nozzles boost jet impulse by 62% via wave timing
Elastic wave propagation governs impulse enhancement in pulsed jets through flexible nozzles
-
LLM agent runs ROOT physics analyses from plain English prompts
RooAgent: An LLM Agent for Root-Based High Energy Physics Analysis
-
Right holders may sue cloud copyright cases from home
Jurisdiction over Ubiquitous Copyright Infringements: Should Right-Holders Be Allowed to Sue at Home?
-
Hypergraphs recover group patterns missed by graphs in missing sensor data
Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural Refinement
-
Countable prime products factor every nonunit in this almost Dedekind ring
Factorization in almost Dedekind domain
-
Mismatched wrong drafts boost GRPO math performance
Weak-to-Strong Elicitation via Mismatched Wrong Drafts
-
Limited doses reduce to none or classic interventions in disease models
Dose-limited interventions in an epidemiological model
-
Triplet data lets diffusion models stylize videos without flicker
VISTA: Triplet-Supervised Video Style Transfer with Diffusion Transformers
-
Spectral-semantic fusion spots AI videos from latest generators
SpecSem-Net: Integrating Spectral and Semantic Features for Robust AI-generated Video Detection
-
Attention hijacking makes one image force target answers across queries
Attention Hijacking: Response Manipulation Across Queries in Vision-Language Models
-
StyleText dataset trains better scene text inpainting models
StyleText: A Large-Scale Dataset and Benchmark for Stylized Scene Text Inpainting
-
Four reasoning stages improve ECG diagnosis accuracy
Reasoning Before Diagnosis: Physician-Inspired Structured Thinking for ECG Classification
-
RL portfolio strategies compete with buy-and-hold in Euro Stoxx 50
Deep Reinforcement Learning Framework for Diversified Portfolio Management Across Global Equity Markets
-
Inexact proximal steps enable sampling from nonsmooth posteriors
Speeding Up Nonsmooth Bayesian MCMC Sampling via Inexact Proximal Unadjusted Langevin Algorithm
-
Control loop raises LLM self-correction accuracy by 6.2 points
CyberCorrect: A Cybernetic Framework for Closed-Loop Self-Correction in Large Language Models
-
Context Codec verifies which commitments survive LLM context compression
Compress the Context, Keep the Commitments: A Formal Framework for Verifiable LLM Context Compression
-
Overlapping chunks link local 3D reconstructions into long coherent sequences
LongDPM: Overlap-Aware 4D Reconstruction from Long Monocular Videos
-
Single reference yields consistent anime characters in Stable Diffusion
AnimeAdapter: Fine-grained and Consistent Zero-shot Anime Character Generation
-
Surface extraction reduces 3D robot navigation space by over 80%
Beyond Geometry: Efficient Topologically-Grounded Navigation in Complex 3D Environments
-
ConflictRAG resolves document conflicts to raise RAG accuracy
ConflictRAG: Detecting and Resolving Knowledge Conflicts in Retrieval Augmented Generation
-
Quadruped pairs cooperate on manipulation by decoupling control layers
HCLM: A Hierarchical Framework for Cooperative Loco-Manipulation with Dual Quadrupeds
-
Asymmetric entry and exit still let GBM reach steady state
Geometric Brownian motion with intermittent entries and exits
-
Holomorphic disk moduli correspond for GIT quotient potentials
Holomorphic disks and GIT quotients
-
Element-wise convergence yields accurate cell-free rate estimates
SERE: A Stabilized Element-Wise Method for Downlink Rate Estimation in Clustered Cell-Free Networks
-
Orthonormal Gaussian zeros repel for k=1 and attract for k=2
Zero correlations and averaged fields of orthonormal Gaussian functions
-
Offline sampling freezes partition function before LLM-RL policy updates
DISA: Offline Importance Sampling for Distribution-Matching LLM-RL
-
HierEdit refines only edited regions for fast 4K edits
HierEdit: Region-Aware Hierarchical Diffusion for Efficient High-Resolution Editing
-
Off-center grasps raise collaborative robot capability by 5.86%
Task Capability Improvement Algorithm for Collaborative Manipulators
-
Metacognitive agents assess limits to boost accuracy
MetaCogAgent: A Metacognitive Multi-Agent LLM Framework with Self-Aware Task Delegation
-
Graph condensation requires full original data to create smaller versions
Position: Graph Condensation Needs a Reset -- Move Beyond Full-dataset Training and Model-Dependence
-
Gradient matching for graph condensation requires full original dataset
Position: Graph Condensation Needs a Reset -- Move Beyond Full-dataset Training and Model-Dependence
-
Step-wise rewards improve LLM math accuracy over aggregated rubrics
Step-wise Rubric Rewards for LLM Reasoning
-
Block-level slicing triples LLM bug finds in 19K-line processor
Debug Like a Human: Scaling LLM-based Fault Localization to Processor Design via Block-Level Instruction-Oriented Slicing
-
End-to-end masks raise accuracy in pruned LLMs
LEAP: Learnable End-to-End Adaptive Pruning of Large Language Models
-
Attacks turn LLM cascades into inefficient error-prone systems
When Efficiency Backfires: Cascading LLMs Trigger Cascade Failure under Adversarial Attack
-
Stable frequency signals boost driver gaze tracking amid lighting shifts
LISA: Language-guided Interference-aware Spatial-Frequency Attention for Driver Gaze Estimation
-
First hyperspectral backbone adapts to any sensor
HyperVision: A Channel-Adaptive Ground-Based Hyperspectral Vision Pre-trained Backbone
-
Explainer finds subgraphs that flip node neighbors in embeddings
UNR-Explainer: Counterfactual Explanations for Unsupervised Node Representation Learning Models
-
CLAP cuts rare driving errors by 24% via latent prompts
CLAP: Contrastive Latent-space Prompt Optimization for End-to-end Autonomous Driving
-
Agentic training loop lifts Lean prover to record Pass@32 scores
OProver: A Unified Framework for Agentic Formal Theorem Proving
-
Error-growth slope estimates Lyapunov exponent from scalar series
FEG-Pro: Forecast-Error Growth Profiling for Finite-Horizon Instability Analysis of Nonlinear Time Series
-
No LLM clears 80 percent on observation contract compliance
ContractBench: Can LLM Agents Preserve Observation Contracts?
-
Trans-relativistic shocks explain FBOT radio diversity
Radio Emission from Fast Blue Optical Transients Powered by Trans-relativistic Shocks in Confined Circumstellar Material