RWKV (pronounced RwaKuv) is an RNN with GPT-level LLM performance, and can also be directly trained like a GPT transformer (parallelizable). We are at RWKV v6.
So it's combining the best of RNN and transformer - great performance, fast inference, fast training, saves VRAM, "infinite" ctxlen, and free text embedding. Moreover it's 100% attention-free, and a LFAI project.
Training RWKV
RWKV GUI with one-click install and API
Official RWKV pip package
Finetuning RWKV (9GB VRAM can finetune 7B)
Fast WebGPU inference (NVIDIA/AMD/Intel), nf4/int8/fp16
Fast CPU/cuBLAS/CLBlast inference, int4/int8/fp16/fp32
All latest RWKV weights
HuggingFace-compatible RWKV weights
Community wiki (with guide and FAQ)
RWKV v6 illustrated (download preview checkpts):