47,133 repositories indexed
1–6 of 6 · Most stars
RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.
C-based/Cached/Core Computer Vision Library, A Modern Computer Vision Library
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
Self-hosted artifact registry and pull-through cache in one binary. 17 formats: Docker/OCI, Maven, npm, PyPI, Cargo, Go, Raw, RubyGems, Terraform, Ansible Galaxy, NuGet, Pub, Conan, CPAN, RPM, Debian/APT, Hugging Face. No database, about 15 MB RAM idle, supply-chain curation built in.
Official implementation of paper "Cache Engineering for Agents"
See where it leaks, stop paying for it — a live Claude Code status line that prices every turn, your subagents, cache rebuilds, plus limits, context and git.