Lifeboat Foundation

Structured light, which encompasses various spatial patterns of light like donuts or flower petals, is crucial for a myriad of applications from precise measurements to communication systems.

The many properties of light allow it to be manipulated and used for applications that range from very sensitive measurements to communications and intelligent ways to interrogate objects. A compelling degree of freedom is the spatial pattern, called structured light, which can resemble shapes such as donuts and flower petals. For instance, patterns with different numbers of petals can represent letters of the alphabet, and when observed on the other side, deliver the message.

Unfortunately, what makes these patterns sensitive for measurements also make them susceptible to unwanted environmental factors such as air turbulence, aberrated optics, stressed fibers, or biological tissues doing their own “patterning” and distorting the structure. Here the distorted pattern can deteriorate to the point that the output pattern looks nothing like the input, rendering them ineffective.

Continue reading “Researchers develop method to manipulate structured light without distortion” »

ByteDance presents ResAdapter Domain Consistent Resolution Adapter for Diffusion Models.

ByteDance presents ResAdapter.

Domain consistent resolution adapter for diffusion models.

Continue reading “ResAdapter: Domain Consistent Resolution Adapter for Diffusion Models” »

MathScale.

Scaling instruction tuning for mathematical reasoning.

Large language models (LLMs) have demonstrated remarkable capabilities in problem-solving.

Continue reading “MathScale: Scaling Instruction Tuning for Mathematical Reasoning” »

Computer graphic simulations can represent natural phenomena such as tornados, underwater, vortices, and liquid foams more accurately thanks to an advancement in creating artificial intelligence (AI) neural networks.

Working with a multi-institutional team of researchers, Georgia Tech Assistant Professor Bo Zhu combined computer graphic simulations with machine learning models to create enhanced simulations of known phenomena. The new benchmark could lead to researchers constructing representations of other phenomena that have yet to be simulated.

Zhu co-authored the paper “Fluid Simulation on Neural Flow Maps.” The Association for Computing Machinery’s Special Interest Group in Computer Graphics and Interactive Technology (SIGGRAPH) gave it a best paper award in December at the SIGGRAPH Asia conference in Sydney, Australia.

Shortgpt layers in large language models are more redundant than you expect.

ShortGPT

Layers in large language models are more redundant than you expect.

Continue reading “Paper page — ShortGPT: Layers in Large Language Models are More Redundant Than You Expect” »

We propose Strongly Supervised pre-training with ScreenShots (S4) — a novel pre-training paradigm for Vision-Language Models using data from large-scale web screenshot rendering.

Join the discussion on this paper page.

A pioneering Large Language Model for Law https://huggingface.co/papers/2403.

In this paper, we introduce SaulLM-7B, a large language model (LLM) tailored for the legal domain.

Join the discussion on this paper page.

GaLore.

Memory-efficient LLM training by gradient low-rank projection.

V/@animaanandkumar.

Continue reading “GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection” »

Apple presents How Far Are We from Intelligent Visual Deductive Reasoning?

Join the discussion on this paper page.