Research Notes

This page collects longer-form notes and essays. The focus is on research ideas, technical observations, paper notes, and practical lessons from building intelligent systems.

A Survey of Workloads for Large Language Model Inference Serving: Description, Public Datasets, and Evaluation Practice featured image

A Survey of Workloads for Large Language Model Inference Serving: Description, Public Datasets, and Evaluation Practice

A five-aspect framework for describing LLM inference-serving workloads, with a survey of public traces, load generators, and evaluation practice.

avatar
Jeff XING
•