Python · Apache-2.0
huawei-csl/KVarN
KVarN is a native vLLM KV-cache quantization backend for your agents: 3-5x more context, throughput above FP16, and FP16-level accuracy. Calibration-free, one flag.
★ 515
Stars
37 forks · 11 open issues
+2.0/wk
Growth · measured
Stars counted between two of our own snapshots roughly a week apart. The repository page gives the exact span.
14.5
Momentum / 100
see the breakdown below
5mo ago
Created
last push 4mo ago
Star history
How this score was built
| Component | Weight | Earned | Why |
|---|---|---|---|
| Velocity | 55 | 2.1 | +2.0/wk, log-scaled against a 750/day ceiling |
| Acceleration | 15 | 0.9 | last week ran 1.06× the monthly pace |
| Freshness | 15 | 3.7 | last push 4mo ago; decays to zero at 180 days |
| Scale | 15 | 7.9 | 515 stars, log-scaled against 150k — stops a 3-star repo topping the board |
| Momentum | 100 | 14.5 | sum of the above |
Detail
| Topics | agentic-aikv-cachellmllm-inferencelong-contextquantizationvllm |
| Matched on | LLM & Inference (llm-inference) |
| Short window | +2 stars over 7 days = 0.3/day |
| Long window | +7 stars over 26 days = 0.3/day |
| Lifetime average | 3.22 stars/day over 160 days |
| First seen here | 2026-10-11 |
| Last read | 2d ago |