This project focuses on implementing the query, key, and value computation for knowledge tracing, inspired by the paper “Towards an Appropriate Query, Key, and Value Computation for Knowledge Tracing.” The objective is to enhance the existing knowledge tracing models by improving the computation of queries, keys, and values.

The project utilizes PyTorch, a popular deep learning framework, to implement these enhancements. By incorporating advanced techniques for query, key, and value computation, we aim to improve the accuracy and effectiveness of knowledge tracing systems.

The implemented solution is designed to be applied in the context of educational assessments, specifically targeting the Riiid! Answer Correctness Prediction Kaggle Competition dataset. By leveraging the improved computation of queries, keys, and values, we aim to provide more accurate predictions and better understand the progress and learning patterns of students.

Through this implementation, we contribute to the field of knowledge tracing by exploring novel methods for computing queries, keys, and values, and their impact on improving the performance of knowledge tracing models.