All Articles
16980 articles total
Buzz Aldrin sells famous felt-tip pen that helped launch Apollo from the Moon
Buzz Aldrin sold a historic felt-tip pen used during the Apollo missions from the Moon.
The TIME Machine: On The Power of Motion for Efficient Perception
Researchers propose TIME, a novel video representation learning approach using motion as the central modality to reduce training data needs and improve temporal understanding.
Correcting Visual Blur Induced by Attention Distraction to Reduce Hallucinations: Algorithm and Theory
The AFIP approach corrects attention distraction in multimodal LLMs to reduce hallucinations by enhancing cross-head attention and visual grounding.
dMX: Differentiable Mixed-Precision Assignment for Low-Precision Floating-Point Formats
dMX introduces a differentiable mixed-precision quantization framework for learnable floating-point bit-width assignment in LLMs, optimizing for the MXFP standard.
The Score Hamiltonian: Mapping Diffusion Models to Adiabatic Transport
This research establishes a mathematical correspondence between score-based diffusion models and adiabatic transport, providing new bounds on density reconstruction.
Does Topic Sentiment Cause Perceived Ideology? Comparing Human and LLM Annotations in Political News Articles
A study comparing human and LLM annotations of political news reveals that zero-shot LLMs often inflate the perceived effect of topic sentiment on political ideology.
Show HN: E-- – A language you dial between English and Python
A new programming language named E-- is introduced, positioned as a bridge between English and Python.
Xray-Visual Models: Scaling Vision models on Industry Scale Data
Researchers present Xray-Visual, a unified vision model architecture trained on massive industry-scale social media data for image and video understanding.
1D-Bench: A Benchmark for Iterative UI Code Generation with Visual Feedback in Real-World
The 1D-Bench introduces a benchmark for evaluating iterative UI code generation from designs, focusing on real-world e-commerce workflows.
Understanding Sources of Demographic Predictability in Brain MRI via Disentangling Anatomy and Contrast
A study investigates demographic predictability in brain MRI, finding that anatomical variation is the primary driver of demographic signals rather than acquisition characteristics.
TADPO: Reinforcement Learning Goes Off-road
TADPO introduces a novel policy gradient formulation for reinforcement learning to enable high-speed, end-to-end off-road autonomous driving.
Research Novelty in Information Systems Journals After ChatGPT: Differences Across Institutional Language Contexts
An analysis of research novelty in Information Systems journals reveals a decline in semantic novelty post-ChatGPT, particularly in non-English-dominant regions.
Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory
A new method using Signal Detection Theory is applied to evaluate the metacognitive efficiency of LLMs, separating accuracy from confidence calibration.
Navigating the Mirage: A Dual-Path Agentic Framework for Robust Misleading Chart Question Answering
The ChartCynics framework uses an agentic dual-path approach to improve the robustness of Vision-Language Models against misleading charts.
Filtered Reasoning Score: Evaluating Reasoning Quality on a Model's Most-Confident Traces
Researchers propose the Filtered Reasoning Score (FRS) to assess the quality of LLM reasoning traces rather than just outcome-based accuracy.
Neuro-Symbolic ODE Discovery with Latent Grammar Flow
Latent Grammar Flow (LGF) is a neuro-symbolic generative framework designed for discovering ordinary differential equations from data.
SQLite should have (Rust-style) editions
A discussion on whether SQLite should adopt an 'editions' system similar to Rust to manage breaking changes and evolution.
The Anti-Mac User Interface (1996)
A retrospective look at the 'Anti-Mac User Interface' from 1996, exploring unconventional UI design.
Metal-Organic Frameworks, Chemistry's New Miracle Materials
An overview of Metal-Organic Frameworks (MOFs), highlighting their potential as revolutionary materials in chemistry.
LLM Networking with MikroTik
Discussion regarding the implementation and networking of Large Language Models (LLMs) using MikroTik hardware.