MS MARCO
A large Microsoft dataset of real Bing search queries with passages judged relevant, widely used to train and test retrieval and reranking models.
In more detail
Many open rerankers, such as the ms-marco-MiniLM cross-encoders, are trained on it. Its web-search questions differ from support tickets, so measure such models on your own labelled queries.
Related terms
- cross-encoder — A model that reads a query and a passage together and outputs one relevance score, instead of embedding each separately.
- reranker — A model that reorders a shortlist of retrieved passages by reading each one together with the query.
Taught in
- Retrieval That Finds The Right Thing — AI Engineering: LLM Apps, RAG & Agents
References
- Hugging Face: cross-encoder/ms-marco-MiniLM-L6-v2 model card
- MS MARCO: A Human Generated MAchine Reading COmprehension Dataset