content-learning-examples

What problem does it solve?

This Skill helps you create high-quality supervised fine-tuning (SFT) training data by generating question-and-answer pairs either from your own documents or from a domain topic that is answered via web research.

Core Features & Use Cases

• Document-grounded SFT generation: Uses document chunks to extract Q&A pairs so the answers come directly from the source text.
• Topic-driven Q&A with web validation: Decomposes a domain into targeted subtopics and uses web search to label answers when no documents are provided.
• SFT-ready formatting guidance: Produces training examples suitable for preparing datasets for hosted SFT runs, with emphasis on quality filtering and linting before splitting.

Quick Start

Ask your AI to generate an SFT dataset for clinical nutrition by uploading textbook PDFs, running a FileSetSeed + QuestionAndLabel pipeline to produce Q&A pairs, and then formatting the results into messages for training.