Text-Davinci-3 API returning unwanted results not found in playground

Prompting
1

Hi, for the following prompt:

Interpret each keyword in the search and decide if it fits within the following categories: job title, company, location, industry 

if it does then output the categories in the following format:
JOBTITLE=job titles*
COMPANY=companies* 
LOCATION=locations*
INDUSTRY=industries*

Don't infer any categories not explicitly mentioned in the keywords.
If a category doesn't have any keywords leave it blank with "*" after the "=" sign, otherwise separate multiple results in a category by ", " and add "*" after the last result of that category

If you do not understand the question only return -1 for the entire response and don't add the RESPONSE= line at the end

Otherwise at the end of the output using the following format: RESPONSE=response * summarize in a sentence the search starting with the words "Here are the results for" and add "Did you find what you were looking for?" to the end of the sentence output

search:
find me everyone in retail

I’m getting the response:

JOBTITLE=*
COMPANY=*
LOCATION=*
INDUSTRY=Retail*
RESPONSE=Here are the results for everyone in the Retail industry. Did you find what you were looking for?

However when I make the same call via API I’m getting “Chicago” in the “LOCATION” output that was never mentioned in the search and doesn’t appear in the playground output:

image
image1450×232 22.8 KB

Here are my settings in node.js:

    const openai = new OpenAIApi(configuration)
    const response = await openai.createCompletion({
        model: 'text-davinci-003',
        prompt,
        max_tokens: 2022,
        temperature: 0,
        top_p: 0,
        frequency_penalty: 0,
        presence_penalty: 0,
    })

Is this to do with my settings and is there a way to fix this issue? Thank you very much!