azure / openai-samples

Samples for working with Azure OpenAI Service

ℹ️	This repository has been migrated to a new location. Click to redirect.

openai-samples's Issues

I want to fine tune curie model on chatbot data so that model should already know context of the user.

Pls guide the way to do it

Python 3.7 is close to end of life - we should update to Python 3.9 or 3.10

I want to prompt tune GPT-35-Turbo with my own data so that chatgpt could give results in context to my data

I can't find how to do that so can you provide some solution.
My data will be about some user and its projects.

How do I use managed identity instead of access key?

I can't find any documentation that states how to use managed identity auth to OpenAI instead of using the access key, but I can find a reference in Microsoft docs saying it is supported.

Can't create azure OpenAI Service

I have been onboarded to the Azure OpenAI Service, but I can't create it in azure portal.

My Account: [email protected]
My Subscription ID: 8c3e12cf-16b0-4c55-9ef2-d2fdd475e158

Azure OpenAI service containers for local deployment?

Is there a roadmap for if / when the Azure OpenAI service will be available for local deployments via containers?

Local deployment seems to be available for most of the other Cognitive Services. I was wondering if this is also planned for the OpenAI service.

Embeddings - limit output

Embeddings are a 1536 vector. We should limit the output to something like [:20]

Capitalize service in README.MD H1

Official name of the service is: "Azure OpenAI Service". On first use full name is required in Microsoft content, on subsequent uses Azure OpenAI without "Service" is allowed.

Deployment steps

Is there a script where all the service are getting deployed on Azure and we can test the app.

Tried setting up each service separately and run the solution locally but no luck

Sample Terraform Scripts for OpenAI Services

Hi,

These samples are great, however they assume you have manually setup OpenAI services in Azure, would it be possible to provide a sample Terraform script to build the infrastructure?

Thank you,
Peter

Invalid Request Error: Resource not found

Hi,

I'm facing an issue, I'm able to use in Python the code from the Azure OpenAI sandbox, so my keys and rest of configuration is ok, but when I try to use your samples I got a Invalid Request Error: Resource not found,

this is the trace:

InvalidRequestError Traceback (most recent call last)
Cell In[121], line 3
1 max_response_tokens = 500
----> 3 response = send_message(messages, chatgpt_model_name, max_response_tokens)
4 messages.append({"role": "assistant", "content": response})
6 print_conversation(messages)

Cell In[117], line 4, in send_message(messages, model_name, max_response_tokens)
3 def send_message(messages, model_name, max_response_tokens=500):
----> 4 response = openai.ChatCompletion.create(
5 engine=chatgpt_model_name,
6 messages=messages,
7 temperature=0.5,
8 max_tokens=max_response_tokens,
9 top_p=0.9,
10 frequency_penalty=0,
11 presence_penalty=0,
12 )
13 return response['choices'][0]['message']['content']

File /opt/homebrew/lib/python3.10/site-packages/openai/api_resources/chat_completion.py:25, in ChatCompletion.create(cls, *args, **kwargs)
23 while True:
24 try:
---> 25 return super().create(*args, **kwargs)
...
684 rbody, rcode, resp.data, rheaders, stream_error=stream_error
685 )
686 return resp

InvalidRequestError: Resource not found

Regards.

Add branch protection rules

Use OpenAI standard environment variables

See here for reference on EVAR names, as well as pattern for using shell environment variables or inline Python os.environ settings. https://github.com/hwchase17/langchain/blob/master/docs/modules/llms/integrations/azure_openai_example.ipynb

Specifically:

OPENAI_API_KEY
OPENAI_API_TYPE
OPENAI_API_VERSION
OPENAI_API_BASE