Fact → Joke Generator

Humor is one of the most challenging evaluation domains for natural language models because it requires comedic timing, subverted expectations, and contextual premise recognition.

This application takes arbitrary, dry factual statements and constructs a prompt chaining pipeline to extract the core premise, generate misdirected punchlines, and render formatted comedic outputs within a responsive web interface.

From prompt design to cloud deployment:

Video demonstration highlighting real-time prompt generation, temperature adjustment, and comedic output formatting.

Explore document intelligence and RAG syllabus querying.