A generation approach that combines a language model with information retrieved from an external collection.
Sign in to saveRetrieval-Augmented Generation (RAG) brings retrieved documents into a language generation process. The original paper combines a sequence-to-sequence model with a dense index of Wikipedia passages. Retrieval makes external evidence available to the model, but answer quality still depends on the relevance of retrieved material and how it is used.
Research paper first submitted to arXiv on 2020-05-22.
Common in document question answering and knowledge assistants. Inspect retrieval relevance and source attribution rather than assuming retrieval makes every answer correct. Catalog note: this is a source-verified reference entry, not a measured current-trend signal. No search-volume or growth claim is made.
Pro opens the same-day prioritize / watch / skip table, including names that are still too new to publish here.
This name has not appeared in an imported daily report yet.