Skip to main content

Evaluation Datasets

Status

Owner: TBD · Status: draft · Canonical source: flezi.app.frontend.main/src/pages/Datasets.tsx

Who is this for?​

AI engineers curating test datasets for evaluation.

What can you do?​

  • Manage standardized datasets containing input prompts and expected outputs.
  • Upload test suites from CSV or JSONL files.
  • Run continuous regression testing against established benchmark suites.

Steps​

  1. Select Datasets and click Create Dataset.
  2. Provide a descriptive title (e.g. 100 HR Policy Golden Test Prompts).
  3. Click Import File and upload a CSV/JSONL file containing input and expected_output columns.
  4. Review the parsed table and click Save Dataset.