EduTrack
Educational Data & Notes at Fingertips
The problem EduTrack solves
Our solution addresses key challenges in educational institutions by enhancing data management. It automates tracking tuition fees, alerting administrators when payments exceed a certain threshold, ensuring timely interventions. Lecturers gain instant access to student information, streamlining academic oversight and support. Additionally, the system fosters collaboration by using large language models (LLMs) to automatically share drive links to notes on new subjects from other lecturers. This enhances curriculum development and ensures educators can quickly adapt to new topics, improving overall operational efficiency and educational quality.
Challenges we ran into
One significant challenge we encountered was identifying the appropriate model for our specific tasks. After extensive research, we selected LLaMA 2 and proceeded to fine-tune it on Google Colab. During this process, we faced issues related to choosing the optimal floating point precision. Balancing precision and computational efficiency proved to be a complex task. Using higher precision, such as float64, increased the model's accuracy but also significantly slowed down the processing time and required more memory. Conversely, lower precision, like float16, enhanced performance speed but sometimes led to a loss in accuracy, impacting the model's reliability. Finding the right compromise between these factors was essential for ensuring the model's effectiveness and operational efficiency.