The Text Matcher evaluator examines the raw text content from your model’s response exactly as generated. No preprocessing or modification occurs—you’re testing against the actual output users receive.

​
Available Response Criteria

Configure your text validation using these operators:

  • equals: Response must match your text exactly
  • starts-with: Response must begin with your specified text
  • ends-with: Response must conclude with your specified text
  • contains-any: Response must include at least one item from your array of strings
  • contains-all: Response must include every item from your array of strings
  • not-contains-any: Response must not include any item from your array of strings
  • regex: Response must match your regular expression pattern

​
Setting Up Text Matcher

1

Select the Text Matcher evaluation from the Add evaluator action menu.

2

Link the dataset, select the desired response using the dropdown, and add the texts.

3

Run the evaluation.

​
Examples

Here are some examples of how you can effectively use Text Matching as an evaluator.

  1. Finding SEO oriented keywords
  2. Ensuring the beginning of the response starts with a phrase
  3. Ensuring the LLM response doesn’t contain personal information