Sign in to view source links and access this dataset
Description
800 Java exception-diagnosis question-answer pairs distilled from Stack Overflow posts and normalized for benchmark use. The dataset was created by author HTJ008 and last updated on 2026-07-21. It is split into 639 training, 80 validation, and 81 test records, with difficulty ratings of easy (324), medium (388), and hard (88).
Use Cases
Training question-answering models based on Java exception descriptions.
Benchmarking AI assistants for software development based on curated difficulty levels.
Evaluating code understanding models on their ability to diagnose errors from Stack Overflow posts.