AttributeError: 'OpenAI' object has no attribute 'kwargs'

[rohanprasadsap]

I am trying to use langchain's implementation of OpenAI instead of dspy's

llm = OpenAI(model_name=model_name)
#llm = dspy.OpenAI(model=model_name, max_tokens=250)

dspy.settings.configure(lm=llm)

I was getting my hands on with the Minimal work example. I have set up keys and other requirements. I have followed all the steps mentioned here.
My openai version is openai 1.30.4

When I compile and evaluate the model. I get this error:

from dspy.teleprompt import BootstrapFewShot

# Set up the optimizer: we want to "bootstrap" (i.e., self-generate) 4-shot examples of our CoT program.
config = dict(max_bootstrapped_demos=4, max_labeled_demos=4)

# Optimize! Use the `gsm8k_metric` here. In general, the metric is going to tell the optimizer how well it's doing.
teleprompter = BootstrapFewShot(metric=gsm8k_metric, **config)
optimized_cot = teleprompter.compile(CoT(), trainset=gsm8k_trainset)

Error:
AttributeError: 'OpenAI' object has no attribute 'kwargs'

What am I missing while defining the llm? How can I resolve this?

[tom-doerr]

What's the drawback of using DSPy's OpenAI client?

[arnavsinghvi11]

Hi @rohanprasadsap ,

Langchain’s OpenAI LM is currently not supported as a DSPy LM and will not inherit all the related properties needed to run DSPy pipelines with it. Feel free to share any particular use cases for using Langchain’s OpenAI LM over the current dspy.OpenAI LM and we can explore if any changes are needed!

[venkatganesh96]

for me it was working you have to use dspy to initialize the llm like the below
turbo = dspy.AzureOpenAI(deployment_id=model_name,
api_key=openai_api_key,
api_base=azure_endpoint,
model_type="chat",
api_version=openai_api_version,
max_tokens=4000,
)
dspy.settings.configure(lm=turbo)