Researchers release large dataset of transactional prompts from GitHub

Researchers have compiled a dataset of 57.5K unique transactional prompts from GitHub, focusing on reproducible natural language instructions integrated into software. They developed a structured ontology to analyze these prompts, revealing diverse usage patterns across languages, domains, tasks, and modalities, with a typical Zipf-like distribution. The dataset and an exploration interface are being released to facilitate empirical study of prompts, with a comprehensive error analysis conducted to ensure annotation quality. AI

IMPACT This dataset could enable deeper understanding and optimization of prompt engineering for LLMs.

RANK_REASON The item is an academic paper detailing a new dataset and methodology. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• GitHub
• Gotit.pub
• Hugging Face
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →