perf: change minimal `kv_chunk_size` back to 128 - Githubissues

flashinfer-ai / flashinfer

FlashInfer: Kernel Library for LLM Serving

https://flashinfer.ai

Apache License 2.0

760 stars 64 forks source link

perf: change minimal `kv_chunk_size` back to 128 #329

Closed yzh119 closed 1 week ago