Select Prompt Template Scorers
Choose the scorers you want to apply to your test cases. Each scorer evaluates a different dimension of your prompt template's output and returns a pass or fail result.
Required Editions
| Available in: Lightning Experience |
| Available in: Enterprise, Performance, Unlimited, and Developer Editions. Required add-on licenses vary by agent type. |
| User Permissions Needed | |
|---|---|
| To create tests in the Testing Center: | Manage Agentforce Grids AND Manage Agentforce Testing |
-
Select the scorers you want to apply to your test cases.
For information about how to view scores and LLM judge reasoning after a run, see Use Test Results.
Scorer Description Coherence Returns a pass or fail based on whether the response is logically structured, easy to follow, and free of grammatical errors. A pass means the response communicates clearly; a fail indicates the response may be confusing, disjointed, or difficult to read. Conciseness Returns a pass or fail based on whether the response is appropriately brief without sacrificing accuracy. A pass means the response avoids unnecessary length; a fail indicates the response may be verbose or padded with irrelevant content. Completeness Returns a pass or fail based on whether the response fully addresses the query by including all necessary information. A pass means the response covers what was asked; a fail indicates the response omits key information or only partially addresses the input. Factuality Returns a pass or fail based on whether the information in the response is factually correct. A pass means the response contains accurate information; a fail indicates the response contains errors or unsupported claims. Instruction Following Returns a score of 1, 3, or 5 based on how closely the response follows the instructions defined in the prompt template. A score of 5 (High) means the response fully followed the instructions. A score of 3 (Uncertain) means adherence can't be conclusively determined. A score of 1 (Low) means the response deviated from the instructions. Response Match Returns a score based on how closely the prompt template's response matches the expected output you provided. Uses a judge LLM to compare the response against your Expected Output column. A higher score indicates a closer match. To run this scorer, you must include an Expected Output column in your CSV. -
Click Save.
The test suite is created and appears in the grid view. To run it, open the test suite and click Run Test.
此文章是否解決您的問題？
請讓我們知道，以便我們改進！
