NER response is a string, should be list of dicts

[sevenChrisK]

The response for the NER template is a string representation of a list of dictionaries.

Currently this means the user first needs to parse and convert into list of dicts to use the output effectively.

[monk1337]

Hi, Did you try the eval() method on response?

[sevenChrisK]

Hi the eval() method should normally work but sometimes the output string starts with a whitespace character that throws an exception.

Maybe the return should involve a step to strip whitespace?

[monk1337]

We are working on a verifier step to check the response type.

[abushoeb]

@sevenChrisK thanks for reporting this. I encountered the same problem. A quick fix that worked for me was using strip().

import ast
response = nlp_prompter.fit('ner.jinja',
                                domain = 'medical',
                                text_input = text, 
                                labels = None)

print(ast.literal_eval(response['text'].strip()))