한국어EN
Inference Infrastructure & ServingUC Berkeley

UC Berkeley releases vLLM + PagedAttention

vLLM + PagedAttention released

Why it matters

Manages the KV cache like OS paging, cutting memory waste to under 4%; the starting point of the open-source serving standard

Source — Primary sources first; secondary reports are labelled “Press”

Officialvllm.aihttps://vllm.ai/blog/2023-06-20-vllm

Added in Week 35, 2026 · See that week

Same week

Hugging Face explains the MMLU score mismatch on the Open LLM LeaderboardOfficialTextbooks Are All You NeedPaper

Nearby on the same axis

AWQPaperllama.cpp community introduces the GGUF formatPressGeorgi Gerganov creates the llama.cpp repositoryOfficialNVIDIA releases TensorRT-LLMOfficial