From f676a0c2483760df8911bb1a011ece7df3c96257 Mon Sep 17 00:00:00 2001 From: Simon Veitner Date: Fri, 27 Dec 2024 16:54:17 +0100 Subject: [PATCH] update --- _posts/2024-12-27-synthetic-data-intro.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/_posts/2024-12-27-synthetic-data-intro.md b/_posts/2024-12-27-synthetic-data-intro.md index 5bb5a55..bfddc44 100644 --- a/_posts/2024-12-27-synthetic-data-intro.md +++ b/_posts/2024-12-27-synthetic-data-intro.md @@ -174,15 +174,15 @@ The second query feels far more realistic, capturing a specific user type and sc So far, we’ve demonstrated how to **generate a single data entry**—which obviously won’t cover all real-world scenarios. In practice, you’d likely: -1. **Generate a Larger Dataset** +* **Generate a Larger Dataset** Leverage asynchronous calls to the OpenAI API to quickly produce hundreds of synthetic queries. -2. **Review with Domain Experts** +* **Review with Domain Experts** Have specialists check whether each query truly reflects real-world user concerns. -3. **Iterate & Refine** +* **Iterate & Refine** Adjust prompts, user profiles, or system settings as needed—especially if you spot edge cases or off-target questions.