iFLYTEK Releases XF Xinghuo V4.0 and Upgrades Education Model
On June 28th, TapTechNews reported that iFLYTEK held the XF Xinghuo V4.0 conference yesterday and released the XF Xinghuo large model V4.0 while upgrading the Xinghuo education large model.
iFLYTEK stated that this upgrade enhanced the capabilities of the large model base and multi-modal capabilities. The graphic recognition effect for complex educational scenarios was further improved, claiming to be better than GPT-4o. Complex handwritten formulas can be recognized with high precision.
Based on the Xinghuo education large model, iFLYTEK released the XF Smart Marking Machine, which can automatically analyze the learning situation of the whole class through marking and give the planning of each child's learning path, realizing software and hardware integration.
Through automated recognition, the accuracy rate of the test question analysis by the XF Smart Marking Machine is greater than 99%. Students can scribble freely and the layout is not limited to any format. The original 90-minute homework marking time can be reduced to 5 minutes, and the 60-minute learning situation statistics time can be shortened to 1 minute.
The iFLYTEK AI learning machine based on the Xinghuo large model has also been upgraded to further realize super human-like Q&A tutoring. The completion rate of children's independent learning has increased from 67% to 90%, and the error-solving rate has increased from 72% to 93%.
According to TapTechNews' previous report, the XF Xinghuo large model V4.0 is trained based on the country's first domestic 10,000-card computing power cluster Feixing One, enhancing the seven core capabilities of the large model base. Among 12 mainstream Chinese and English tests at home and abroad, XF Xinghuo V4.0 ranked first in 8 tests .
