SilentPatch for GTA III, Vice City, and San Andreas

Mixture-of-experts (MoE) training megakernel for NVL72s

MSLK (Meta Superintelligence Labs Kernels) is a collection of PyTorch GPU operator libraries that are designed and optimized for GenAI training and inference, such as FP8 row-wise quantization and …

Arm Performix: https://developer.arm.com/servers-and-cloud-computing/arm-performix

Reference implementation and examples of the CuTe Layout representation and algebra.

MoonEP: A Perfectly Balanced Expert Parallelism Library via Dynamic Redundant Experts

Flash-Flash KDA: H100-optimized Flash Kimi Delta Attention kernels

Language model tokenization at GB/s

A feed-forward 3D foundation model for reconstructing scenes from streaming data

SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible.

Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

A web engine for pencil-style rendering of 3D scenes to SVG — exact silhouettes, hidden-line ghosting, hatching. Great for abstract and technical scenes.

Cross-platform instrumentation and introspection library written in C

minimal cross-platform standalone C headers

Cross-Platform HW accelerated CRC32c and CRC32 with fallback to efficient SW implementations. C interface with language bindings for each of our SDKs

Random stuff about lower level iOS

Standalone C++/GGML runtime for ThinkSound text->sound-effect generation

Box3D is a 3D physics engine for games

A comprehensive collection of IQA papers

slime is an LLM post-training framework for RL Scaling.

High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static bin…

Low-latency Rust thread pool with parallel iterators

C++ implementation of a fast hash map and hash set using robin hood hashing

Real-time 3D full-body reconstruction from a single camera, Multiperson BVH output, Pure C++ runtime, ONNX + ggml, 70-joint skeleton with hands.

FlashKDA: high-performance Kimi Delta Attention kernels

justine's optimization for musl qsort smoothsort

DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

TokenSpeed is a speed-of-light LLM inference engine.