English Español Français Deutsch
← Back to the list
Foundational figures Methodologists Directly upstream of ai100's methodology

Xipeng Qiu

Building an open Chinese-language large language model that the academic community can actually study under the hood — and the tooling around it.

Qiu's group at Fudan released MOSS, one of the early open Chinese-language LLMs that came with enough documentation and training tooling for independent groups to reproduce and modify. The same lab is responsible for FastNLP — one of the more widely used Chinese-side NLP frameworks for sequence modeling — and a research line on in-context demonstration retrieval that connects to current evaluation methodology. For ai100, which audits AI engines including Chinese-language behavior, MOSS sits in the part of the literature that documents what an open Chinese LLM actually looks like inside, and Qiu's group is the most readable entry point.

Worth following when
you need to understand the open Chinese-language LLM landscape from inside its academic origin, rather than from English-language news coverage of it.
Topics
open Chinese-language LLMs (MOSS); the FudanNLP / FastNLP ecosystem; in-context demonstration retrieval methodology.
Key works
MOSS open Chinese LLM (2023, project lead); FastNLP / FudanNLP framework body of work; publications on in-context learning and demonstration retrieval.