aitherkvcache(tool, public)
Near-optimal KV cache quantization for LLM inference — sub-byte compression.
Name
aitherkvcache — Near-optimal KV cache quantization for LLM inference — sub-byte compression.
The problem it solves
KV cache, not weights, is what runs a GPU out of room during real serving, and it grows with every token of every concurrent request.
Instead of trusting / you check
- instead of trusting
- a vendor's quantisation defaults
- you check
- sub-byte KV cache kernels you can benchmark yourself
Adopt it alone
Quantize the cache on one model you already serve and measure the headroom.
Install
pip install aither-kvcache
Where it sits in the path
- Not on the path yet — reference only.
See also
Links
Generated 2026-09-14 00:48Z from the ecosystem registry. A field here that is wrong is wrong in the registry — fix it there and this page follows.