Enhancing Precision Learning: Tongyi Qianwen Launches First AI-Focused Multimodal Education Model

Precision Learning Unveils Innovative Multimodal Education Model

Precision Learning, an AI education technology company, has introduced an advanced educational model named “Flow Knowledge Mirror.” Built upon the open-source Tongyi Qwen large-scale model, this innovative approach integrates text, speech, and emotional analysis to effectively navigate essential textbooks and problem-solving techniques used by top educators. It features real-time, low-latency interactive learning capabilities, delivering a more efficient and enjoyable experience for students.

Set to launch in June, Precision Learning's first AI tutoring device utilizing this model aims to revolutionize the educational landscape. Since its establishment in 2018, the company has rolled out a variety of intelligent learning machines, powered by recommendation engines and knowledge graphs, currently assisting millions of primary and secondary school students in China.

In response to the rise of foundational models last year, Precision Learning identified the need for a more natural interactive learning tool that leverages cutting-edge model technologies. After evaluating numerous open-source and closed-source models, the company determined that Tongyi Qwen (Qwen-72B) excelled in performance, openness, and usability. Its multimodal capabilities, including visual and audio understanding, meet the growing demand for personalized education and engaging learning experiences.

By combining Qwen-72B, Qwen-VL, Qwen-Audio, and a vast repository of educational data gathered over six years, Precision Learning has trained the “Flow Knowledge Mirror” model on Alibaba Cloud. This sophisticated model enhances problem-solving skills and facilitates natural multimodal interactions, such as voice dialogues and emotional sensing, functioning effectively as an AI educator.

Students can directly engage with the AI tutor to address gaps in their knowledge, with the AI generating real-time explanations and visual aids. It can also design unique learning pathways to provide customized support, promoting rapid knowledge acquisition.

Zhang Ning, head of AI technology at Precision Learning, states, “We chose Tongyi Qwen for its versatility and rich modalities, allowing for rapid and effective processing of text, speech, and visual interactions—ideal for complex educational scenarios. With tailored training, this model meets students' individual needs, significantly enhancing learning outcomes. As large model technology advances and learning data expands, these models will reshape the education sector, delivering personalized interactions and deep learning experiences that not only mimic teachers’ tutoring styles but also adapt to each student’s pace and emotional state.”

Looking ahead, Precision Learning aims to further exploit the various dimensions of the Tongyi Qwen models, enhancing student engagement in diverse settings, including edge computing.