Getting spam back from the API

Is it normal to receive spam back as a response from the API. Here is some example text I’m seeing in the response. It has nothing to do with my prompt.

ob Details Location: Anywhere (Burbank) Type: Part-time Salary/Rate: $600/Month Work Schedule: Full-time Contact Name: Formilae Phone: 1-800-555-1212 www.workathometheatres.com {product-name-var label="Your Field Name"} See more jobs at Ingenious Solutions Related 2015 Part Time Jobs Babysitter Jobs Waitress Jobs Housekeeping Jobs Commercial Cleaning Jobs House Manager Jobs Hostess Jobs Server Jobs Teen Jobs See More Part Time Jobs People looking for a data entry\n\n\n\nPay: $ 2637\n\nWorking time: Full-time\n\nDate posted: Sun, 16 Aug 2011 04:26:28 GMT\n\nEn Welestaria!\n\nInformation Technology jobs and earn extra income while still Employed full-time, supply part-time income, existing Contractors/Consultants Search and part Time Occasional Labourers in Other Jobs\n\n\n\nTo learn more information about us and interested, please contact Here: Find a job - Job Bank Workathometheatre\n\nLocation: Anywhere (Burbank)\n\nJob Title: Part Time Jobs, Babysitter Jobs, Waitress Jobs, Housekeeping Jobs, Commercial Cleaning Jobs, House Manager Jobs, Hostess Jobs, Server Jobs, Teen Sunshine Cleaning Part Time Employment Opportunities Started Jan 1, 2010. Earn up to $600/Month in short time.\n\nEarn an average of $600 - $2000/month from just a few hours a day. Make your own hours, weekends and evenings. Join a legitimate work at home business opportunity with a solid foundation in business services and customer"

Yes, it can be “normal” for generative AIs to generate a lot of blah, blah, blah…

If you need further help, please post the text of your prompt, and the details of your API params and I can test for you and provide more detailed feedback in reply.

:slight_smile:

Ok well don’t laugh at what I last asked it because I was frustrated. LOL.

Here is what I have:
// Set up the request body
string requestBody = JsonConvert.SerializeObject(new
{
model = “davinci”,
prompt = “In less than 500 characters, provide a response to the following question/request. (Do not repeat any sentences in the response. Also, do not return spam or any text unrelated to the question. Just answer the following question/request: Compose a sample email following up with a lead that would like to sign up to use our service. She mentioned that she will be using our service, and I’d like to follow up and also let her know that we have a tutorial she can watch to get started.”,
//examples_context = "Q: " + prompt,
max_tokens = 500
});


So this is usually how I ask chat gbt when I do it manually on the web site version, and he usually brings back great responses. Not so great with the api version.
I’m using this endpoint:
https://api.openai.com/v1/completions

Hi @earthsendangered

Here is an “API version” but it uses my custom Rails UI for easy testing:

However, I used text-davinci-003 and not davinci first:

text-davinci-003

davinci

HTH

:slight_smile:

Appendix; Set Up

Thanks for your help. I’m getting good results nows!

1 Like