ChatRWKV

Why we included this project

ChatRWKV offers a rare route to a ChatGPT-style assistant that runs on a pure recurrent network rather than a transformer. If you want a self-hosted chatbot that stays light on VRAM and responds quickly on modest GPUs, the repo gets you there fast: ready-to-run chat scripts, Gradio demos, and an API example mean you can go from clone to a live conversation without assembling your own serving stack. The RWKV architecture is a genuine alternative to the usual GPT family, trading some conventional design choices for lower memory use and faster inference. It also pairs with the RWKV-LM training repo, so it works as a practical starting point for anyone who wants to turn these models into an actual conversational product.