Azure OpenAI Embeddings Connector Error in RAG Workflow

Hi Knime Community,

I am attempting to create a Retrieval-Augmented Generation (RAG) workflow using Azure OpenAI, where the workflow simply takes one PDF as input, allowing for Q&A through the GUI.

I have successfully created the workflow using OpenAI Connectors.
However, I am encountering errors with the Azure OpenAI connectors. Specifically, the “Azure OpenAI Embeddings Connector” returns the following error message:

Configure failed (Py4JException): An exception was raised by the Python Proxy. Return Message: Traceback (most recent call last): 
File "C:\KNIME\bundling\envs\org_knime_python_llm\Lib\site-packages\py4j\clientserver.py", line 617, in _call_proxy return_value = getattr(self.pool[obj_id], method)(*params)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
File "C:\KNIME\plugins\org.knime.python3.nodes_5.3.0.v202407031038\src\main\python\_node_backend_launcher.py", line 984, in configure java_outputs = self.postprocess_configure_outputs(java_config_context, outputs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
File "C:\KNIME\plugins\org.knime.python3.nodes_5.3.0.v202407031038\src\main\python\_node_backend_launcher.py", line 1013, in postprocess_configure_outputs self._port_type_registry.spec_from_python(
File "C:\KNIME\plugins\org.knime.python3.nodes_5.3.0.v202407031038\src\main\python\_node_backend_launcher.py", line 549, in spec_from_python data = {"id": port.type.id, "data": spec.serialize()}
^^^^^^^^^^^^^^^^ 
File "C:\KNIME\plugins\org.knime.python.llm_5.3.0.v202407041639\src\main\python\src\models\openai.py", line 828, in serialize "dimensions": self._dimensions,
^^^^^^^^^^^^^^^^ AttributeError: 'AzureOpenAIEmbeddingsPortObjectSpec' object has no attribute '_dimensions'

I am also attaching an image for your reference.

Configuration for Azure OpenAI nodes, I am using

  1. Azure OpenAI Chat Model Connector → LLM-gpt-35-turbo
  2. Azure OpenAI Embedding Connector → text-embedding-ada-002

Could someone please help me resolve this issue?

Thank you in advance for your assistance!

1 Like

Hello @OMKAR_SHINDE, thank you for reporting the issue, and welcome to the forum and community!

We’ve already fixed this issue and will include the update in the next bugfix release for version 5.3, which will be available in the coming days.

Best,
Alex

3 Likes

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.