This release adds an `Example` extractor to fill the templates with examples from the dataset. Now, you can define prompts like the following one, without specifying examples by hand in the template:
Write a legal document in German. Here you have some examples:
- Example 1: {example_1}\n
- Example 2: {example_2}\n
Document:
or combine it with another extractors, e.g.:
Write a legal document in German. Here you have some examples:\n
- Example 1: {example_1}\n
- Example 2: {example_2}\n
You must use the following entities: {entities}.
Document: