GitHub - LostRuins/koboldcpp: A simple one-file way to run various GGML and GGUF models with KoboldAI's UI
🚀 Discover KoboldCpp - an efficient AI text-generation tool! 🧠✨ Run GGML and GGUF models seamlessly with KoboldAI's user-friendly UI. Featuring stable Diffusion image generation, a versatile API endpoint, and more. Available on Windows, Linux, OSX, and even Android! #AI #Tech #GitHub
- Project name: KoboldCpp
- Description: AI text-generation software for GGML and GGUF models, building off llama.cpp.
- Features: Kobold API endpoint, Stable Diffusion image generation, versatile, backward compatibility, UI with persistent stories, editing tools, save formats, memory, world info, etc.
- Usage: Download .exe release or clone repo; rebuild with makefiles and scripts; run KoboldCpp.py for non-Windows systems.
- Performance tips: Use GPU Acceleration (CUDA, CLBlast), GPU Layer Offloading, Increase Context Size, Experiment with settings.
- Platforms: Windows, Linux (Precompiled binary, automated compiler script), OSX (Manual compiling), Android (Termux installation).
- Additional info: Arch Linux AUR packages available; Docker images by community members; support available on Github and KoboldAI Discord.
- Licensing: Original GGML library and llama.cpp under MIT License, Kobold Lite under AGPL v3.0 License.
- Supported models: GGUF models, LLAMA, GPT-2, GPT-NeoX, RWKV, Falcon, and more.
- Github activity: 3.3k stars, 249 forks, 71 releases.