Where it lives

  • Go to Datasets → Generate dataset. Two generation options exist; this section covers the simpler one, "Examples". 1:56

What "Examples" mode does

  • Generates a set of plausible inputs only from a description of your task scenario. 1:56

Steps demonstrated

  1. Describe the task — the demo uses a health and fitness chatbot. 1:56
  2. Provide an example input, e.g. "How much water should I drink daily?", request 10 more questions like it, and name the dataset (e.g. "QA"). 2:15
  3. Click "next" and choose the API key, then generate. 2:15

Result

  • The generated dataset contains similar realistic questions such as "how often should I work out?" and "how do I create a balanced meal plan?"; inputs can be edited and diversified further. 2:36
  • These are inputs only — the intended use is to run them through your LLM system to get responses. 3:06