Azure chat completion model not working as well as the completion model directly from Open AI

Navneet Chauhan 0 Reputation points
2023-07-13T10:41:45.79+00:00

Hello all,

I am using the chat completions api (tried both gpt-35-turbo and gpt-35-turbo-16k) to process some large texts (>1000 words) kind of like a sentiment analyzer. I have set the top_p=0.0001 to make sure I don't get any variance in the output. I am noticing that for the same models and same prompt the Open AI version of the chat completion api gives me better and more reliable results compared to the Azure version on the api (all api params being the same). Is there a reason for this? Is the Azure model in some way fundamentally different from the direct from Open AI model. Is there something I can do to make the Azure model more reliable? Thanks in advance to anyone who can help?

Azure OpenAI Service
Azure OpenAI Service
An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
3,624 questions
Azure AI services
Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
3,098 questions
{count} votes

1 answer

Sort by: Most helpful
  1. Vinodh247 27,381 Reputation points MVP
    2023-07-13T10:51:07.5933333+00:00

    Hi,

    Thanks for reaching out to Microsoft Q&A.

    Did you try customizing the model for your application?

    https://learn.microsoft.com/en-us/azure/cognitive-services/openai/how-to/fine-tuning?pivots=programming-language-studio

    https://www.advancinganalytics.co.uk/blog/2023/4/24/azure-openai-vs-openai-whats-the-difference

    Please Upvote and Accept as answer if the reply was helpful, this will be benefitting the other community members who go through the same issue.

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.