Using code in LLMGraphTransformer

[quartermaine]

I am trying to pass a script of code as text in the LLMGraphTransformer as in the tutorial .
But I am getting the following error:

BadRequestError: Error code: 400 - {'error': {'message': "Invalid parameter: 'response_format' of type 'json_schema' is not supported with this model. Learn more about supported models at the Structured Outputs guide: https://platform.openai.com/docs/guides/structured-outputs", 'type': 'invalid_request_error', 'param': None, 'code': None}}

Is there any way of passing a script of code as text in order to make a graph?

[dosubot[bot]]

The error you're encountering indicates that the 'response_format' parameter set to 'json_schema' is not supported by the model you are using. To resolve this, you can try the following steps:

1. Check Model Compatibility: Ensure that the language model you are using supports the 'json_schema' response format. You may need to switch to a model that supports this format. Refer to the Structured Outputs guide for more information on supported models.

2. Use Supported Response Formats: If switching models is not an option, consider using a different response format that is supported by your current model. For example, you can use 'content_and_artifact' or 'json_object' as response formats, which are commonly supported ^[1].

3. Modify the Code: Adjust your code to use a supported response format. Here's an example of how you might modify your tool function:

   @tool(response_format="content_and_artifact")
   def tool_informa_o_clima():
       content = "O clima é estável com sol."  # String for the model
       artifact = {"Resposta": "O clima é estável com sol."}  # Additional data
       return content, artifact

By ensuring compatibility with the model and using a supported response format, you should be able to pass a script of code as text to the LLMGraphTransformer without encountering the error.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[quartermaine]

can you give me an example for the second option and 'json_object'

[quartermaine]

I managed to solve the problem by changing the model to gpt-4o.