Data Viewer
Purpose:
Comprehensive Data Inspection: View, filter, and sort all logged data from your AI interactions, including prompts, input data, and performance metrics.
Export and Share: Easily export data in CSV format for further analysis or sharing with team members.
Debugging and Analysis: Quickly identify issues or trends in your logged data to inform prompt optimization decisions.
Example: Suppose you are building a document summarization tool for PDFs. You can use the Data Viewer to inspect logs of how different prompts performed on various PDF documents. For instance, if you notice that the prompts used on scientific PDFs are underperforming, you can filter the logs by document type and identify which prompts consistently receive low evaluation scores. This insight helps you decide whether to adjust those prompts or create a specialized prompt family for scientific articles.
Last updated