Skip to main content
. 2023 Nov 17;6:1279794. doi: 10.3389/frai.2023.1279794

Figure 4.

Figure 4

The prompt template for text-davinci-003 Model. The preamble (1) primes the model to generate semantic type predictions. The document_type token (2) is replaced with the document type according to the data set (e.g., “adjudicatory decisions” for BVA). The category_n_name tokens (3) are substituted with the names of the semantic types and the category_n_definition tokens (4) with the corresponding definitions. The text_snippet_n tokens (5) are replaced with the analyzed text snippets.