Зачем Авито нужна своя Inference-платформа для ML/LLM-сервисов: PaaS vs отдельное решение
English summary
Why a web PaaS fails for LLM inference (GPU sharing, batching gives 5-10x utilisation difference); decision to build a separate inference platform on an open-source core.
Read the original on habr.com
Russian·Web page·opens habr.com