Implement .with_structured_output() for Ollama models

[aks-456]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

We would like to implement Ollama.with_structured_output() to directly support Ollama models capable of structuring its output. For LLM's that do not support structured output, the method should return an error indicating that structured output is not supported.

Motivation

In Langchain, the OpenAI class has a with_structured_output() method, allowing structured outputs such as JSON from supported models. However, this functionality is currently missing in the Ollama class.

Proposal (If applicable)

If the issue is accepted, my team and I would like to work on it.

[aks-456]

Reason: Streamline use of structured output for Langchain to make it more consistent with other models
Bump: We would like to approach this by harmonizing the current method of using ChatOllama.withStructuredOutput with .with_structured_output to create consistent usage across models

[efriis]

hey! I believe this is already implemented on the base class

[aks-456]

Thank you for your reply! Our apologies, but the documentation in Ollama.ipynb is out of date. Here is the link to the notebook.