Overview
LLMSys PaperList is a community-curated collection of papers, code, and tutorials focused on LLM systems—spanning training, inference, serving, benchmarks, and courses—useful for researchers and engineers.
Key features
- Structured table of contents covering training, serving, multi-modal systems, and benchmarks.
- Links to papers, code repositories and reading materials, regularly updated by contributors.
- Serves as a reading list for courses, literature reviews, and system design research.
Use cases
- Quickly find canonical papers and implementations for LLM systems research.
- Use as course reading material or assignment references for MLSys/LLM classes.
- Jumpstart system-level investigations into parallelism, inference optimization and serving.
Technical characteristics
- Simple GitHub Markdown format that allows PR-based contributions and issue tracking.
- Entries often include arXiv/DOI links and associated implementation repos for fast access.
- Lightweight and actively maintained by the community.