All Articles
17507 articles total
Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions
CodeTracer is a forensic framework designed to trace backdoored code completions in LLMs to the specific malicious fine-tuning data that caused them.
Provably Optimal Learning Algorithms for Assistance Games
This research provides the first provably efficient learning algorithms for repeated assistance games between informed and uninformed agents, achieving optimal regret rates.
Can We Trust LLM's Logic? Quantifying Uncertainty, Coherence, and Robustness via a Graph-Based Framework
GRAPHEVAL is a graph-based framework that quantifies LLM reasoning uncertainty and introduces Graph Self-Consistency (GSC) to improve reasoning fidelity over simple majority voting.
APIVOT: Adaptive Planning with Interleaved Vision-Language Thoughts
APIVOT is a VLM-based planner for long-horizon robot tasks that interleaves language for semantic reasoning and visual thoughts for geometric verification.
Structured Pruning of Large Language Models via Power Transformation and Sign-Preserving Score Aggregation with Adaptive Feature Retention
A new structured pruning method for LLMs uses power transformation and sign-preserving aggregation to achieve inference speedups while maintaining accuracy.
DKDNet: Dual Knowledge and Data-Driven Network for Cross-Domain Automatic Modulation Classification
DKDNet is a dual knowledge and data-driven network designed to improve automatic modulation classification across different communication domains using signal prior knowledge.
In Emacs, Everything Looks Like a Service
A discussion on the architectural pattern of treating components as services within the Emacs ecosystem.
Tiny Tapeout Explorer: WASM FET-level circuit SIM&vis
An exploration of the Tiny Tapeout Explorer, which provides FET-level circuit simulation and visualization using WebAssembly.
Efficient Safety Alignment of Language Models via Latent Personality Traits
Introduction of Latent Personality Alignment (LPA), a lightweight safety alignment method for LLMs that uses personality-anchored representations to resist jailbreaks.
Adversarial Decoys: Misdirecting Attention-Based Defenses in ViT
Research on adversarial decoys in Vision Transformers (ViTs) that misdirect attention-based defenses to maintain attack effectiveness.
path_boost: A Python Package for Interpretable Graph-Level Prediction using Path-Based Gradient Boosting
Presentation of path_boost, a Python package for interpretable graph-level prediction using path-based gradient boosting.
Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing
A comparative study of linear attention architectures (DeltaNet, Gated DeltaNet, etc.) focusing on memory decay, throughput, and cross-layer routing.
Beyond Thermal Imaging: Inferring Thermophysical Properties from Time-Resolved Thermal Observations
Introduction of ThermoField, a framework for inferring thermophysical properties and reconstructing 3D thermal scenes using differentiable heat-transfer simulation.
A Multi-cluster Boundary Learning Method for Out-of-Scope Intent Detection via MiniLM Embedding
A multi-cluster boundary learning method using MiniLM embeddings for more efficient out-of-scope intent detection in human-machine interaction.
When Implausible Tokens Get Reinforced: Tail-Aware Credit Calibration for LLM Reinforcement Learning
Introduction of TACO (Tail-Aware Credit calibratiOn), a method to improve LLM reinforcement learning by preventing the reinforcement of implausible 'tail' tokens.
3100 Opinions on Code Review in an AI World: Building Causal Theory from Practitioner Discourse
A large-scale analysis of practitioner discourse to build a causal theory on how AI coding agents impact the software code review process.
DreamCharacter-1: From 3D Generative Foundation Models to Product-Ready Character Generation
Introduces DreamCharacter-1, a lightweight framework for enhancing 3D foundation models to generate production-ready 3D characters with improved geometry and textures.
From Triggers to Emotions: A CPM-Grounded Appraisal Multi-Agent for Dynamic Emotional Evolution in Persona-Based Dialogue
Proposes CPM-MultiAgent, a framework that uses psychological theory to model dynamic emotional evolution in persona-based dialogue agents.
Shift & Drift: A Zero-Shot Benchmark for Generalizable and Robust Autonomous Driving Motion Planning
Presents Shift & Drift, a zero-shot benchmark to test the robustness and generalization of autonomous driving motion planners against semantic and state-distribution shifts.
Kime-Representation Formulations of Three Open Problems in the Foundations of Classical Mechanics: Uncertainty, Invariant Entropy, and Directional Degrees of Freedom
Provides mathematical formulations in the complex-time (kime) representation to address open problems in classical mechanics regarding uncertainty and entropy.