Time
1 year
Location
Denmark
Sector
Education
Description
The client aimed to enhance the efficiency of creating assignments for a mobile app focused on improving children’s reading comprehension. The objective was to harness the capabilities of large-scale language models to automatically generate quizzes and simple tasks based on various children’s books.
Solution
We developed a comprehensive framework that incorporated various components, including the automation of model fine-tuning and optimization using a merged dataset. We also implemented a large-scale architecture for batch text generation. Multiple output validators were employed to ensure the quality and safety of the generated materials.
Results
The implementation of this solution enabled the client to process assignments for hundreds of books, a substantial improvement from the previous manual, single-book workflow. The automation not only led to a superior app but also played a pivotal role in securing additional funding for the further development of the application. This achievement represented a significant step forward in the client’s mission to improve children’s reading comprehension through innovative technology.