Tokenization Is the Bottleneck You're Not Measuringranvier.systems1 points·by mindsaspire·2개월 전·1 comments
KV Cache Locality: The Hidden Variable in Your LLM Serving Costranvier.systems3 points·by mindsaspire·2개월 전·0 comments
Show HN: Ranvier – Prefix-aware routing for LLM inferenceranvier.systems1 points·by mindsaspire·4개월 전·0 comments