LLM Notebooks: Prompting, Structured Output, and Function Calling

Short, practical notebooks showing how to:

• Prompt effectively
• Produce strict JSON outputs
• Use function calling (tool invocation) with OpenAI-compatible chat APIs on Databricks

Notebooks

• 01.prompt_engineering.ipynb: Prompt engineering basics, temperature, few-shot, reasoning.
• 02.structured_output.ipynb: Returning strict JSON objects and common extraction patterns.
• 03.function_calling.ipynb: End-to-end tool calling flow with arguments and second call.

Running on Databricks

• Import the notebooks and attach to a cluster.
• The examples read a workspace token via dbutils and call Databricks Model Serving endpoints (base_url set to your workspace).
• Ensure you have an available serving endpoint compatible with the OpenAI Chat Completions API.

Notes

• The function calling example expects you to parse the tool call arguments, execute your function, and pass results back to the model for a final answer.
• The structured output examples request strict JSON; validate responses as needed in production.