RNN+Transformer hybrid with O(n) inference. Linear time, infinite context, no KV cache. Train like GPT (parallel), infer like RNN (sequential). Linux Foundation AI project. Production at Windows, Office, NeMo. RWKV-7 (March 2025). Models up to 14B parameters.
8.7
Rating
0
Installs
AI & LLM
Category
Excellent skill covering RWKV architecture with comprehensive task knowledge including installation, dual-mode usage (GPT/RNN), streaming generation, long-context processing, fine-tuning, and clear comparisons to Transformers. The description accurately captures the key value proposition (O(n) inference, infinite context, no KV cache). Structure is clean with quick start, common workflows, troubleshooting, and references to detailed architecture files. High novelty: RWKV's unique capabilities (constant memory, infinite context) would require significant token usage for a CLI agent to discover and implement correctly. Minor improvement possible: slightly more explicit CLI invocation patterns in description, but overall extremely well-executed skill documentation.
davila7
Skill Author
Loading SKILL.md…
Skill Author
