Stars
TPU inference for vLLM, with unified JAX and PyTorch support.
A simple, performant, and scalable Jax LLM!
A batteries-included framework for building web apps
Instant, Concurrent, Secure & Lightweight Sandbox for AI Agents.
🌟 [My TouchBar My rules]. The Touch Bar Customisation App for your MacBook Pro
Zmeu — a systems programming language with its own syntax, compile-time ownership & borrow checking, structured concurrency, and an MLIR-to-LLVM native backend.
DeepSpec: a full-stack codebase for training and evaluating speculative decoding algorithms
SGLang is a high-performance serving framework for large language models and multimodal models.
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer
Automated alignment adjustment for LLMs — direct steering, LoRA, and MoE expert-granular abliteration, optimized via multi-objective Optuna TPE.
A PyTorch native library for training speculative decoding models
Fully automatic censorship removal for language models
Blazingly fast pusher drop-in replacement written in rust
Compare command performance using Linux hardware counters (perf_event_open) — Linux-only CLI
Full Transformer into a custom chip. microGPT in RTL, generating names on a Virtex-5 FPGA at ~56k tokens/second.
A high-performance ASCII video rendering engine featuring real-time WebSocket binary streaming and an isolated compiler for serverless static generation. Built for low-latency 30 FPS playback on HT…
Automatically generates Rust FFI bindings to C (and some C++) libraries.
An extremely fast Python type checker and language server, written in Rust.
The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
Simple, fast, safe, compiled language for developing maintainable software. Compiles itself in <1s with zero library dependencies. Supports automatic C => V translation. https://vlang.io
Unofficial ChatGPT desktop app for Linux (formerly the Codex app), built locally from OpenAI’s official macOS app. Includes Chat, Work, and Codex. Packages for Debian/Ubuntu (.deb), Fedora/openSUSE…
Fast stackful fibers with a NUMA-aware work-stealing scheduler