Large model retrieval enhancement generation method and device

The embodiment of the invention provides a large model retrieval enhancement generation method and device, and the method comprises the steps: obtaining a first problem, and coding the first problem to obtain a first problem vector; determining a first vector according to the first problem vector; d...

Full description

Saved in:
Bibliographic Details
Main Authors ZHENG BO, LIN YAN, LIU YAN, HE QUANSHENG, DENG MINYI, HU CHUNYING, HUANG XUEWEI, XIE XIANGMEI, LIN HUI, ZHENG ZHENYU
Format Patent
LanguageChinese
English
Published 09.08.2024
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The embodiment of the invention provides a large model retrieval enhancement generation method and device, and the method comprises the steps: obtaining a first problem, and coding the first problem to obtain a first problem vector; determining a first vector according to the first problem vector; determining a second vector according to the first vector; processing the second vector through a large language model to obtain a target answer; when the type of the first vector is a text vector, the first vector represents at least one text block vector formed by combining at least one text block vector and corresponding context information, and the at least one text block vector is a text block vector related to a first problem vector retrieved in a vector database; when the type of the first vector is a problem vector, the first vector comprises one of the following items: at least one second problem vector meeting a similarity condition with the first problem vector in the vector database, at least one third p
Bibliography:Application Number: CN202410509226