Build streaming text applications with RWKV.
Guides developers in implementing RWKV streaming text generation, long-context processing, and fine-tuning.
When to use it
Use for long-context or streaming applications that need constant memory usage or memory-constrained deployment.
It changes how the agent implements RWKV inference and training, including preserving recurrent state between inference calls.
What you provide
The CUDA examples use an NVIDIA GPU; the hardware requirements specify CUDA 11.6+ and also permit CPU execution.
This skill
torch
Installs or upgrades PyTorch
pytorch-lightning
Installs PyTorch Lightning
deepspeed
Installs or upgrades DeepSpeed