Support for Azure OpenAI

[vansen]

Hello, thanks for this really helpful project! I am using Azure OpenAI over OpenAI itself and I would like to know if there is already a way to do that (Usually the baseurl needs to be set) or if that support is planned to be added?
Thank you!

[llxxxll]

Perhaps this is possible? The code is untested. CrawAI can use APIs based on langchain models.

import os

from langchain.schema import HumanMessage
from langchain_openai import AzureChatOpenAI

os.environ["AZURE_OPENAI_API_KEY"] = "..."
os.environ["AZURE_OPENAI_ENDPOINT"] = "https://<your-endpoint>.openai.azure.com/"

Azure_model = AzureChatOpenAI(
    openai_api_version="2023-05-15",
    azure_deployment="your-deployment-name",
)


researcher = Agent(
  role='Researcher',
  goal='Discover new insights',
  backstory="...",
  verbose=True,
  llm = Azure_model

[younes-io]

I've taken this example from the wiki and adjusted it for Azure as well. I tested it and it works..

import os
from crewai import Agent, Task, Crew, Process
from langchain.agents import load_tools

from langchain.chat_models.azure_openai import AzureChatOpenAI

llm = AzureChatOpenAI(
    model=model_name,
    api_version=openai_api_version,
    api_key=openai_api_key,
    # azure_endpoint=open_ai_base,    
    base_url=open_ai_base,    
)

# Loading Human Tools
human_tools = load_tools(["human"])

# Define your agents with roles and goals
researcher = Agent(
  role='Senior Research Analyst',
  goal='Uncover cutting-edge developments in AI and data science in',
  backstory="""You are a Senior Research Analyst at a leading tech think tank.
  Your expertise lies in identifying emerging trends and technologies in AI and
  data science. You have a knack for dissecting complex data and presenting
  actionable insights.""",
  verbose=True,
  allow_delegation=False,
  # Passing human tools to the agent
  llm=llm,
  tools=human_tools
)
# the rest of the code...

[sorin-costea]

Azure changed to regional endpoints so might be that's why the above stopped working?
Now it's like https://your_region.api.cognitive.microsoft.com/openai/deployments/your_deployment/chat/completions?api-version=some_version_string

[BodapatiNirupamasai]

I am trying to run the below code but somehow I am getting " Error code: 401 - {'error': {'message': 'Incorrect API key provided"
Could anyone help me with this?

from crewai import Agent
from textwrap import dedent
from tools import yt_tool
from load_dotenv import load_dotenv
from langchain_openai import AzureChatOpenAI
import os

load_dotenv()

llm = AzureChatOpenAI(
azure_deployment="gpt-4",
api_version="2023-06-01-preview",
api_key=os.environ.get("key"),
azure_endpoint=os.environ.get("endpoint"),
)

researchAgent = Agent(
role='Research Analsyt for Youtube Videos',
goal='Search popular youtube videos to extract actionable insights for the topic {topic}',
tools=[yt_tool],
llm=llm,
verbose= True,
memory=True,
allow_delegation=True,
backstory=dedent("""You're a Research Analsyt at a large company.
list 5 popular youtube videos on the {topic} as you're responsible for analyzing content in the youtube videos,
providing insights and suggestions to the business."""),
)

[BodapatiNirupamasai]

sorin
I am not able to connect to crew Ai using Azure Open AI key.

[sorin-costea]

Could be because of the same Azure endpoint format upgrade I have? See discussion here (LiteLLM is the underlying connection): BerriAI/litellm#5995

…

On Thu, 3 Oct 2024, 21:02 Nirupama Bodapati, ***@***.***> wrote: sorin I am not able to connect to crew Ai using Azure Open AI key. — Reply to this email directly, view it on GitHub <#101 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA2KEWDRGAZFYSYPPFOXDYTZZWBETAVCNFSM6AAAAABPEDU3NKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJSGAYTCNZRGU> . You are receiving this because you commented.Message ID: ***@***.***>

[amishkhatri]

I am trying to run the below code but somehow I am getting " Error code: 401 - {'error': {'message': 'Incorrect API key provided" Could anyone help me with this?

from crewai import Agent from textwrap import dedent from tools import yt_tool from load_dotenv import load_dotenv from langchain_openai import AzureChatOpenAI import os

load_dotenv()

llm = AzureChatOpenAI( azure_deployment="gpt-4", api_version="2023-06-01-preview", api_key=os.environ.get("key"), azure_endpoint=os.environ.get("endpoint"), )

researchAgent = Agent( role='Research Analsyt for Youtube Videos', goal='Search popular youtube videos to extract actionable insights for the topic {topic}', tools=[yt_tool], llm=llm, verbose= True, memory=True, allow_delegation=True, backstory=dedent("""You're a Research Analsyt at a large company. list 5 popular youtube videos on the {topic} as you're responsible for analyzing content in the youtube videos, providing insights and suggestions to the business."""), )

[amishkhatri]

You should setup environment variables in the system as AZURE_OPENAI_API_KEY
AZURE_OPENAI_ENDPOINT with the valid key and endpoint
Further, initialize the object as
llm = AzureChatOpenAI(
azure_deployment="azure/<deployment_name>",
api_version="<api_version>",
temperature=<temprature 0-1 in numbers>
)
You dont need to mention key and endpoint in the code anymore.
Remember you should have deployment as "azure/<dep. name>" otherwise it wont work.

[JoseGeorges8]

Hello I'm having some issues with this.

Whenever i try to run my script it keeps on failing saying that I need to set the OPENAI_API_KEY environment variable. Any idea on how to get past this and why is it not using the Azure environment variables?

I also tried to set them up directly with the model but it keeps on throwing that error. If I try to set an arbitrary OPENAI_API_KEY then I get a 401

[jayanti-prasad]

Hello I'm having some issues with this.

Whenever i try to run my script it keeps on failing saying that I need to set the OPENAI_API_KEY environment variable. Any idea on how to get past this and why is it not using the Azure environment variables?

I also tried to set them up directly with the model but it keeps on throwing that error. If I try to set an arbitrary OPENAI_API_KEY then I get a 401

Having the same issue. I want to use Azure OpenAI but it seems that it is hardcoded to OpenAI.

[thl456]

I have this same issue. Tried to change os.environ with codes below but doesn't work either. When I publish my app to Streamlit, the llm seems to still be calling OpenAI, I get the error message 'Incorrect API key provided............You can find your API key at https://platform.openai.com/account/api-keys....'

os.environ["AZURE_OPENAI_API_KEY"] = [my Azure API key]
os.environ["AZURE_OPENAI_ENDPOINT"] =[my Azure endpoint]

llm = AzureChatOpenAI(
model = model,
api_key = api_key,
api_version = api_version,
azure_endpoint = api_base)