Search Results for "io gnu/linux"

Next AI Draw.io is an AI-enhanced diagramming application that integrates generative intelligence into the familiar draw.io-style visual workflow. The project aims to help users create diagrams, flowcharts, and structured visual content more efficiently by leveraging AI-assisted generation and editing capabilities. It combines modern web technologies with diagram automation features to reduce the manual effort typically required in visual design tools. The system is intended for developers,...

FlashAttention is a high-performance deep learning optimization library that reimplements the attention mechanism used in transformer models to be significantly faster and more memory-efficient than standard implementations. It achieves this by using IO-aware algorithms that minimize memory reads and writes, reducing the quadratic memory overhead typically associated with attention operations. The project provides implementations of FlashAttention, FlashAttention-2, and newer iterations...

Mosec is a high-performance and flexible model-serving framework for building ML model-enabled backend and microservices. It bridges the gap between any machine learning models you just trained and the efficient online service API.

NVIDIA Earth2Studio is an open-source Python package and framework designed to accelerate the development and deployment of AI-driven weather and climate science workflows. It provides a unified API that lets researchers, data scientists, and engineers build complex forecasting and analysis pipelines by combining modular prognostic and diagnostic AI models with a diverse range of real-world data sources such as global forecast systems, reanalysis datasets, and satellite feeds. The toolkit...

Monoio is a Rust asynchronous runtime designed for high-performance I/O-bound servers and applications, built around native OS async I/O primitives (e.g. io_uring on Linux, epoll / kqueue on other Unix-like systems), rather than layering atop an existing runtime. Its design philosophy centers on a “thread-per-core” model where each core runs its own event loop, minimizing cross-thread synchronization needs, avoiding the overhead and complexity of task scheduling, and letting developers write...