Browse trusted datasets for evaluation, enrichment, and production use.
Xiezhi is a comprehensive assessment suite designed to evaluate broad domain knowledge. It comprises 516 different disciplines, offering multiple‑choice questions across 13 topics, totaling 249,587 items, plus two sub‑sets (Xiezhi‑Specialty and Xiezhi‑Interdiscipline) each containing 15,000 questions.
TeleQnA is a comprehensive dataset designed to evaluate large language models' knowledge in the telecommunications domain. It comprises 10,000 multiple‑choice questions divided into five categories: Lexicon (500), Research Overview (2,000), Research Publications (4,500), Standards Overview (1,000) and Standards Specifications (2,000). Each question is represented in JSON format with five fields: question, options, answer, explanation, and category. Experimental code is provided to assess the performance of OpenAI models such as GPT‑3.5.
