Open-source Drupal 7 API dataset with 7,952 question-answer examples for fine-tuning, evaluation, and RAG. GPL-2.0-or-later, ready to download from Hugging Face.

A hands-on account of running Qwen2.5-Coder-1.5B as the workhorse of a real data pipeline, with the numbers to back it up.

A field report on building a Drupal-specific instruction dataset from scratch — pipeline design, quality gates, and the real numbers behind a working system.