Free open-source community edition of DevCleaner - a fast development cache cleaner for macOS

Tile-Based Runtime for Ultra-Low-Latency LLM Inference

Systematic benchmark study of DeepSeek-V4-Flash inference on 4× NVIDIA RTX PRO 6000 Blackwell (TP=4, FP8 KV, MTP=2, 1M context). Sustained decode matrix + Estonia long-context profile.

Reverse engineering notes. Personal reference only. Everything here is a best-guess reconstruction.

A collection of tricks and tools to speed up transformer models

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

How good are LLMs at generating Lean code

Nim CPS runtime with http1.1, http2, http3, ws, sse, webtransport, irc, dns and a React-like DSL, http server DSL, and wasm compilation

Wayland Compositor in Minecraft

(at least a useful portion of) Temporal Logic of Actions, a.k.a. TLA in Lean 4

Minim(al/ized) LRU cache

Dynamic Memory Management for Serving LLMs without PagedAttention

A SillyTavern fork with Bun as the backend, along with UI/UX improvements

Miles is an enterprise-facing reinforcement learning framework for LLM and VLM post-training, forked from and co-evolving with slime.

A pure-Python implementation of the Nvidia CuTe layout algebra intended to be approachable and easy to learn.

henka is a bindings generator for Nim 👑

An ultra-fast, distributed Safetensors loader

High-performance safetensors model loader

Hub for ongoing Qwen inference benchmarks on NVIDIA Blackwell. Indexes all studies, hosts the rolling SOTA leaderboard, points to the toolchain.

Fast Tokens

Docker images for LLM inference (SGLang + vLLM) on NVIDIA Blackwell GPUs (SM120, CUDA 13.2)

A multimodal database engine written in Nim

Library for Nim.

Shaders in Nim language

Constraint solving library

Remove large amounts of unwanted applications quickly.

A modern system cleaner built in Go with a TUI and CLI.

A high performance Python graph library implemented in Rust.