Xunfei Xunhuo large model upgrade V1.5 release: open question and answer breakthrough, multiple rounds of dialogue, logic and mathematics ability upgrade
At the 24th anniversary celebration of iFlytek held yesterday, iFLYTEK Xunhuo Cognitive Model V1.5 was officially released. According to the official, after a lapse of one month, the Xinghuo model has not only achieved continuous improvement in various abilities, but also achieved three major upgrades in comprehensive abilities: a breakthrough in open knowledge quizzes, multiple rounds of dialogue, and further upgrades in logic and mathematics.
HKUST Xunfei said that by combining the language understanding ability and general expression ability of the large model with the search plug-in, the Xinghuo large model effectively solves industry problems such as the difficulty of updating new knowledge and the fact that the question and answer of the fact is easy to "be confused". The original 24% improvement.
With the blessing of long-chain thinking reasoning and mathematical logic reasoning ability, "Xunfei Spark" ushered in the upgrade of logic and mathematics ability. The upgrade of the multi-round dialogue ability allows "Xunfei Spark" to complete specific complex tasks, and to talk to examiners through simulated interview scenarios, providing assistance to college students and job seekers who are about to enter the society. In addition, it can create more inspirational and imaginative content between conversational interactions.
In addition, Xunfei Xinghuo App and Mini Programs were officially released at the meeting. Functions such as pure voice dialogue, multi-modal input, multi-terminal support, and multi-functional assistants have been launched, enabling barrier-free voice communication anytime, anywhere, supporting graphic and text recognition, and mathematical formula recognition, and presenting them in various forms on various terminals .
In terms of application implementation, HKUST Xunfei further promotes the application of Xinghuo’s cognitive large-scale model in education, medical care, industry, office and other fields, empowering Xinghuo Voice Companion App, medical post-diagnosis rehabilitation management platform, Antelope industrial Internet platform, Xunfei Hearing Wisdom Screen and other products, and opened the Xunfei Spark development interface, and worked with developers to build the "Spark" ecosystem.
HKUST Xunfei also cooperated with the first batch of 22 national key universities in the country to launch the Spark Large Model Scene Innovation Competition, build the Xunfei University AI Spark Camp, and cultivate leading talents in the era of general artificial intelligence for the future of the industry.
The author noticed that from the release of the Spark Cognitive Large Model on May 6th to the upgrade to V1.5 on June 9th, according to the key milestones of the large-scale model development previously announced by HKUST Xunfei, the ChatGPT will be benchmarked on October 24th this year: Chinese surpasses, English equals.
