Language Models Use Trigonometry to Do Addition

Kavli Affiliate: Max Tegmark | First 5 Authors: Subhash Kantamneni, Max Tegmark, , , | Summary: Mathematical reasoning is an increasingly important indicator of large language model (LLM) capabilities, yet we lack understanding of how LLMs process even simple mathematical tasks. To address this, we reverse engineer how three mid-sized LLMs compute addition. We first […]


Continue.. Language Models Use Trigonometry to Do Addition

Cosmological super-resolution of the 21-cm signal

Kavli Affiliate: George Efstathiou | First 5 Authors: Simon Pochinda, Jiten Dhandha, Anastasia Fialkov, Eloy de Lera Acedo, | Summary: In this study, we train score-based diffusion models to super-resolve gigaparsec-scale cosmological simulations of the 21-cm signal. We examine the impact of network and training dataset size on model performance, demonstrating that a single simulation […]


Continue.. Cosmological super-resolution of the 21-cm signal

Spectral Sufficient Conditions for Graph Factors

Kavli Affiliate: Ke Wang | First 5 Authors: Fengyun Ren, Shumin Zhang, Ke Wang, , | Summary: The ${K_{1,1}, K_{1,2},C_m: mgeq3}$-factor of a graph is a spanning subgraph whose each component is an element of ${K_{1,1}, K_{1,2},C_m: mgeq3}$. In this paper, through the graph spectral methods, we establish the lower bound of the signless Laplacian […]


Continue.. Spectral Sufficient Conditions for Graph Factors

Characterization of the Teledyne COSMOS Camera: A Large Format CMOS Image Sensor for Astronomy

Kavli Affiliate: Gabor Furesz | First 5 Authors: Christopher Layden, Jill Juneau, Gustav Pettersson, Nathan Lourie, Benjamin Schneider | Summary: The Teledyne COSMOS-66 is a next-generation CMOS camera designed for astronomical imaging, featuring a large-format sensor ($8120 times 8120$ pixels, each $10 mu m$), high quantum efficiency, high frame rates, and a correlated multi-sampling mode […]


Continue.. Characterization of the Teledyne COSMOS Camera: A Large Format CMOS Image Sensor for Astronomy

Low-Rank Adapting Models for Sparse Autoencoders

Kavli Affiliate: Max Tegmark | First 5 Authors: Matthew Chen, Joshua Engels, Max Tegmark, , | Summary: Sparse autoencoders (SAEs) decompose language model representations into a sparse set of linear latent vectors. Recent works have improved SAEs using language model gradients, but these techniques require many expensive backward passes during training and still cause a […]


Continue.. Low-Rank Adapting Models for Sparse Autoencoders

A Spectroscopically Calibrated Prescription for Extracting PAH Flux from JWST MIRI Imaging

Kavli Affiliate: Claudio Ricci | First 5 Authors: Grant P. Donnelly, Thomas S. -Y. Lai, Lee Armus, Tanio Díaz-Santos, Kirsten L. Larson | Summary: We introduce a prescription for estimating the flux of the 7.7 micron and 11.3 micron polycyclic aromatic hydrocarbon (PAH) features from broadband JWST/MIRI images. Probing PAH flux with MIRI imaging data […]


Continue.. A Spectroscopically Calibrated Prescription for Extracting PAH Flux from JWST MIRI Imaging

The Local Galactic Transient Survey Applied to an Optical Search for Directed Intelligence

Kavli Affiliate: Michael McDonald | First 5 Authors: Alex Thomas, Natalie LeBaron, Luca Angeleri, Phillip Morgan, Varun Iyer | Summary: We discuss our transient search for directed energy systems in local galaxies, with calculations indicating the ability of modest searches to detect optical Search for Extraterrestrial Intelligence (SETI) sources in the closest galaxies. Our analysis […]


Continue.. The Local Galactic Transient Survey Applied to an Optical Search for Directed Intelligence

The Local Galactic Transient Survey Applied to an Optical Search for Directed Intelligence

Kavli Affiliate: Michael McDonald | First 5 Authors: Alex Thomas, Natalie LeBaron, Luca Angeleri, Phillip Morgan, Varun Iyer | Summary: We discuss our transient search for directed energy systems in local galaxies, with calculations indicating the ability of modest searches to detect optical Search for Extraterrestrial Intelligence (SETI) sources in the closest galaxies. Our analysis […]


Continue.. The Local Galactic Transient Survey Applied to an Optical Search for Directed Intelligence

BounTCHA: A CAPTCHA Utilizing Boundary Identification in Guided Generative AI-extended Videos

Kavli Affiliate: Ke Wang | First 5 Authors: Lehao Lin, Ke Wang, Maha Abdallah, Wei Cai, | Summary: In recent years, the rapid development of artificial intelligence (AI) especially multi-modal Large Language Models (MLLMs), has enabled it to understand text, images, videos, and other multimedia data, allowing AI systems to execute various tasks based on […]


Continue.. BounTCHA: A CAPTCHA Utilizing Boundary Identification in Guided Generative AI-extended Videos

TRIPP: A General Purpose Data Pipeline for Astronomical Image Processing

Kavli Affiliate: Michael McDonald | First 5 Authors: Alex Thomas, Natalie LeBaron, Luca Angeleri, Samuel Whitebook, Rachel Darlinger | Summary: We present the TRansient Image Processing Pipeline (TRIPP), a transient and variable source detection pipeline that employs both difference imaging and light curve analysis techniques for astronomical data. Additionally, we demonstrate TRIPP’s rapid analysis capability […]


Continue.. TRIPP: A General Purpose Data Pipeline for Astronomical Image Processing