Fast, flexible LLM inference

A fast Rust-based tool for running AI language and vision models locally from Hugging Face, with a built-in chat UI, Python API, automatic quantization, and support for agentic tool-calling workflows.

Efficent platform for inference and serving local LLMs including an OpenAI compatible API server.

X-LoRA: Mixture of LoRA Experts

Low rank adaptation (LoRA) for Candle.

Blazingly fast inference of diffusion models.

A faster Arc.

8-bit floating point types for Rust

CLI utility to inspect and explore .safetensors and .gguf files

Sampling techniques for Candle.

An autonomous robot, powered by AI.

Falcon is a powerful, interpreted programming language.

Graph model execution API for Candle

Minimalist ML framework for Rust

Tensor library for Zig

MXFP4-compatible 4-bit floating point types and block formats for Rust.

ML framework featuring compile time checks and accelerated by a JIT compiler.

Simple and safe.

A classifier based on PerceiverIO

A TUI system monitor with support for NVIDIA GPUs (CUDA/NVML) and Apple Silicon GPUs (Metal)

The Kestrel Programming Language.

Make your code fly.

Autodiff from the ground up.

A GPT-like autoregressive, decoder-only model trained on data about various scientific or coding topics.

Automated generation of UQFF models with mistral.rs.

Up-to-date Rust FFI bindings to GGML.

Macros for candle-lora.

An automatic proof checker

Benchmarking for Candle

A LLM-specific dataset for Candle.

Edge(u)cation: Cutting-edge multimodal LLMs on the edge with mistral.rs, using F8Q8

A CLI tool to manage and switch between Git user profiles

No description.

O(n^2) matmul with SIMD.

Fast, flexible, and safe reference counted type.

Initilize and allocate once, deallocate never.

My custom decoder-only, autoregressive, GPT-style model.

An autoregressive, decoder-only transformer.

MLX: An array framework for Apple silicon

DORA (Dataflow-Oriented Robotic Application) is middleware designed to streamline and simplify the creation of AI-based robotic applications. It offers low latency, composable, and distributed dataflow capabilities. Applications are modeled as directed graphs, also referred to as pipelines.

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

Rust implementation of VibeVoice text-to-speech with voice cloning and multi-speaker synthesis.

Graph-Aware Attention for Adaptive Dynamics in Transformers

Example of pyo3_special_method_derive

Range checked types leveraging Rust's type system.

It's a New Kind of Wrapper for Exposing LLVM (Safely)

Make your code fly.

A GPT style transformer trained on scientfic and programming data.

A codebase implementing a simple GPT-like model from scratch based on the Attention is All You Need paper.

The Python programming language

A Python Interpreter written in Rust

MiniJinja is a powerful but minimal dependency template engine for Rust compatible with Jinja/Jinja2

dLLM: Simple Diffusion Language Modeling

Edge(u)cation: Cutting-edge multimodal LLMs on the edge with mistral.rs, using F8Q8

A high efficiency binary format for sequencing data

No description.

No description.

Check markdown link validity.

No description.

Reverse Engineering Gemma 3n: Google's New Edge-Optimized Language Model

Super-fast Structured Outputs

An extension library to Candle that provides PyTorch functions not currently available in Candle

No description.

an automatic differentiation companion notebook

FPL v2 is a programming language similar to Python.

FPL is a programming language based on BASIC.

A high-throughput and memory-efficient inference and serving engine for LLMs

Safe rust wrapper around CUDA toolkit

💥 Fast State-of-the-Art Tokenizers optimized for Research and Production

LlamaIndex is a data framework for your LLM applications

C++ hash map and hash set which preserve the order of insertion

Implementation of Perceiver, General Perception with Iterative Attention, in Pytorch

Fork of std::sync::Arc with lots of utilities useful for FFI

Rust implementation of the Mistral Tekken tokenizer

Rust bindings for the C++ api of PyTorch.

A cross-platform browser ML framework.

Homebrew tap for mistral.rs

No description.

🍻 Default formulae for the missing package manager for macOS (or Linux)

fused MoE kernel in Candle backend

a complex numbers, 2d/3d graphing, arbitrary precision, vector, cli calculator with real-time output

A small Rust library + binary that serves pre-built OpenAI-API compatible Chat UI

Rust Programming Language

A multi-level tensor algebra superoptimizer

Count lines of code quickly.

No description.

No description.

No description.

An ergonomic tracing garbage collector that supports mark 'n sweep garbage collection

JavaScript-WebAssembly interop library for Rust.

Fire segmentation using a convolutional UNet

The DynaGen model allows the prediction of dynamical field data based on microstructure input. The model is exemplified to dynamic fracture problems in brittle materials.

Programming language made for a tutorial

LLVM wrappers for Rust

The Falcon SBC - 2020

Single file header only C++ implementation of BigInteger / BigNum / unlimited size integers and rational numbers

LiquidCrystal Arduino library for the DFRobot I2C LCD displays

Falcon Programming Lanugage - v0.3

65C02 emulator in Python

No description.

Implementation of Imagen, Google's Text-to-Image Neural Network, in Pytorch

The Pepper Programming Language