Stars
Heavily compressed docker images for SWE Bench Verified
Bend 2: a fast language that blocks AI mistakes via proof. Install: curl -fsSL https://bend-lang.com/install.sh | sh
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent…
Open-source replacement for the MECHREVO, XMG and Eluktronics Control Center on Uniwill-based laptops. Controls CPU/GPU power, fans, battery, lighting and more directly through the EC. Native C++ f…
Unofficial description of the CUDA assembly (SASS) instruction sets.
A GPU task manager for Windows, Linux, and WSL: per-process VRAM, GPU activity, and LLM model/KV-cache insights for NVIDIA, AMD, and Intel GPUs (plus Huawei Ascend on Linux).
Use GPT, DeepSeek, GLM, Gemini, Kimi and 400+ models in Claude Desktop's own model picker and in Claude Code, while staying signed in to your Claude subscription (1P, no gateway mode). Auto reasoni…
A VL nuget to interact with Spotify's web API using JohnnyCrazy's SpotifyAPI-NET lib.
Inference engine for Qwen3.x-27B (NVFP4) on one RTX 5090 (Blackwell, SM120), written from scratch in C++/CUDA
LLM Inference Engine for Metal, CUDA and Vulkan.
Local inference engine optimized for NVIDIA Blackwell
A live gauge for your local LLM inference engine in the menu bar: prefill and decode speed, prompt-cache hits, context, RAM and GPU. Works with Strata.
DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm
Distributed AI/LLM for the people. Share compute privately or publicly to power your agents and chat.
nantomarioni / input-leap
Forked from input-leap/input-leapOpen-source KVM software
A local inference engine for Apple silicon, built around the model.
Strata fork: Qwen3.8-Flash-Next 125B MoE in NVFP4 on one RTX 20-50 card (12 GB+) and 64 GB of RAM or more. Setup installs our GPTQ quants (huihui abliterated, OrcaRouter) from Hugging Face. CPU exp…
Blazing fast Apple Silicon inference engine optimized for Qwen & Nemotron. OpenAI + Anthropic API compatible. Built by folks from nace.ai
Agent skill that removes signs of AI-generated writing from text
Soft-MTP: Multi-Token Prediction Done Right. KL gradient recovery for faster speculative decoding and better language model training.
Tips & tricks to tune .NET workloads for better perf (mostly, targeting benchmarks).
Orca lets you run a fleet of coding agents (e.g. Claude Code, Codex) with your own subscription. Available on desktop, mobile and remote runtime.
Matrix-Game 4.0: Streaming Neural Rendering for Playable Games
Reverse engineer anything with agents, from app behavior down to native binaries.
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
Get information about your VL documents without opening them
A high-performance inference engine for AI models
Skills for Real Engineers. Straight from my .agents directory.