Tensward
github.com/tensward/tensward- Category
- Developer Tools
- Rank
- No. 1584Tools index
- Pricing
- Open Source
- Type
- TOOL
- GitHub
- 18 stars
- Latest release
- v0.4.2
- Date
About
Tensward profiles how an LLM is served on your own GPU. It starts its own copy of a vLLM or llama.cpp engine, runs a realistic workload, reads engine and GPU metrics, and proposes one flag change at a time. A change is kept only if it is faster and answers pass a comparison check. It does not stop or reconfigure a running server.
Tags
llminferencevllmllama-cppgpuprofiling
Tech Stack
Python
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.