CodeT5+ 220m for classifying developer questions

This model classifies questions in developer forums (e.g., Stack Overflow) as 'open', 'not a real question', 'off topic', 'not constructive' or 'too localized'.

• Developed by: Fabian C. Peña, Steffen Herbold
• Finetuned from: Salesforce/codet5p-220m
• Replication kit: https://github.com/aieng-lab/senlp-benchmark
• Language: English
• License: MIT

Citation

@misc{pena2025benchmark,
  author    = {Fabian Peña and Steffen Herbold},
  title     = {Evaluating Large Language Models on Non-Code Software Engineering Tasks},
  year      = {2025}
}