What is the list of GPT-4 named entity categories?

When using a function such as the one below each entity has a category. I need a list of the possible category values to properly handle the response.

ner_gpt_function = [
        {
            "name": "find_ner",
            "description": "Extracts named entities and their categories from the input text.",
            "parameters": {
                "type": "object",
                "properties": {
                    "entities": {
                        "type": "array",
                        "items": {
                            "type": "object",
                            "properties": {
                                "entity":{"type": "string", 
                                          "description": "A Named entity extracted from text."},
                                "catrgory":{"type": "string", 
                                            "description": "Category of the named entity."}
                            }                            
                          }
                        }
                    }
                },
                "required": ["entities"]
            }
        ]

Thank you.

The AI will write any “category” (or “catrgory” in your case) from any word in any language.

You would need to use a string enum type of specification, with a list of the limited categories you want, if you want to constrain what the AI will write.

"mood": {"type": "string", "enum": ["Happy", "Sad"]},

Also note: the AI doesn’t get descriptions from within nested objects, just at the top level, so anything that the description implies would need to be in the property name (or you re-tier the specification).

This does not answer the question. In order to actually use GPT-4 the user needs to know the range of category values that will be returned and if possible the meaning of those category values.

For example, in test cases I find that locations can be returned as “location”, “Location” or “LOC”. Monetary values can be returned as “MONEY” or “Currency”. Numbers may be returned as “Number”, “ORDINAL” or “CARDINAL”. These are from just initial testing. So far I have seen 26 distinct categories but before using GPT-4 I need to know if this is exhaustive or if I should expect to see other categories (and other ways of representing what seem to be the same categories).

The AI doesn’t have an set operation. Outputting “categories” is not some kind of predefined task. AI is a predictor of language.