Skip to content

Dataset

Step 2 of the API Connector wizard — upload your test data.

Dataset Step

Required Columns

Based on your settings from Step 1, the required dataset columns are displayed at the top. At minimum, the input column is required.

Download Templates

Use the template buttons to download a pre-formatted file:

Format Description
CSV template Comma-separated values with required column headers
JSON template Array of objects with required fields
JSONL template One JSON object per line

Upload

Drag and drop a file or click Click to upload:

  • Supported formats: CSV, JSON, JSONL
  • Maximum size: 50MB

After upload, a preview table shows the first rows and detected columns.

Dataset Format Examples

input,expected_output,context
"What is Python?","A programming language","Python is a high-level..."
"What is RAG?","Retrieval Augmented Generation","RAG combines retrieval..."
[
  {
    "input": "What is Python?",
    "expected_output": "A programming language",
    "context": "Python is a high-level..."
  }
]
{"input": "What is Python?", "expected_output": "A programming language"}
{"input": "What is RAG?", "expected_output": "Retrieval Augmented Generation"}

Tip

Start with a small dataset (5-10 rows) for testing, then scale up once you've verified the pipeline works correctly in Step 3.