Test & Preview¶
Step 3 — verify your full pipeline (API call + response extraction) before running a batch evaluation.
How to Test¶
- Click Send Test Request
- The system sends one request using the first row of your dataset
- The API response is displayed with extracted field values
- Verify that all fields are correctly mapped
What to Check¶
After the test request completes, verify:
| Check | What to Look For |
|---|---|
| API Response | Did the API return a valid response (200 OK)? |
| actual_output | Is the model's text response correctly extracted? |
| retrieval_context | Are context chunks present (if using RAG metrics)? |
| tools_called | Are tool calls captured (if using Agent metrics)? |
| token_usage | Is the token count available (for cost tracking)? |
Troubleshooting¶
| Problem | Solution |
|---|---|
| Request timeout | Increase timeout in Settings |
| 401 Unauthorized | Check Authorization header in Connection |
| Empty actual_output | Verify JSONPath in Response mapping |
| Missing fields | Ensure dataset has required columns (Columns) |
Warning
Always run a test before executing a full evaluation. A misconfigured pipeline will waste API credits on every row in your dataset.