OPPO announces the new development of AndesGPT model: knowledge and Encyclopedia is second only to GPT4.

Thank CTOnews.com netizens soft media Xinyou 1933769 for the clue delivery! CTOnews.com, September 28, OPPO Xiaobu's assistant announced the latest development of the Andean Model (AndesGPT) released by OPPO, which made it into the September ranking of the SuperCLUE benchmark.

According to the introduction below, ▲ Tu Yuan Xiao Bu official account, AndesGPT is a generative big language model of OPPO self-training. In early August, OPPO Intelligent Assistant Xiao Bu launched an external invitation test, aiming to upgrade the traditional voice assistant at the kernel level. In the ranking of the ten basic abilities of SuperCLUE, AndesGPT scored 98.33 in the knowledge and Encyclopedia competency test, ranking first in China only after GPT4.

OPPO said that through the introduction of high-quality knowledge graph, QA question and answer numbers, and the optimization of pre-training and instruction fine tuning techniques, the coverage and accuracy of AndesGPT large language model knowledge and encyclopedia capabilities have been significantly improved.

OPPO also revealed that the technical route of the AndesGPT model is the dialogue-enhanced big language model, which is highly related to the core application scenario of Xiaobu's intelligent assistant, focusing on the improvement of dialogue ability. In the future, the model will continue to enhance the AI capabilities of OPPO intelligent assistant Xiao Bu, and gradually apply it to more OPPO core products.

As previously reported by CTOnews.com, OPPO announced last month that a new AndesGPT-based Brookie assistant is about to launch a large-scale experience. It is said that the upgraded Xiaobu assistant will have the ability of AI large model, have stronger semantic understanding and dialogue ability, can write the content that users need according to the copywriting they need, and AI capabilities such as induction and summary will also be greatly enhanced.