Edited adding context to LLM-generated descriptions notebook

[validbeck]

Internal Notes for Reviewers

sc-8278

Part 1 for getting ready to expose information about adding context to LLM-generated test descriptions. I updated the llm_descriptions_context.ipynb (renamed: add_context_to_llm_descriptions.ipynb) with the idea that it will eventually be moved into "how_to": https://github.com/validmind/validmind-library/blob/beck/sc-8278/documentation-spike-explore-dynamic-llm-context/notebooks/code_sharing/llm/add_context_to_llm_descriptions.ipynb

I will check in with Kristof and Mehdi to see if we want to expose this notebook to customers and update some of the documentation outside of the notebook to signal this use case.

[juanmleng]

Looks great to me Beck! Thanks for the edits :)

[github-actions]

PR Summary

This pull request enhances the functionality of LLM-generated test descriptions in the ValidMind library by allowing users to add custom context. The changes are made in the Jupyter notebook add_context_to_llm_descriptions.ipynb and include the following key updates:

• Title and Introduction Update: The title of the notebook has been updated to better reflect its purpose. The introduction now clearly explains the importance of adding context to LLM-generated descriptions to align with specific use cases and organizational policies.

• Table of Contents: A new table of contents has been added to improve navigation within the notebook.

• Detailed Steps for Adding Context: The notebook now includes detailed markdown sections and code examples that guide users through the process of:

  • Installing and initializing the ValidMind library.
  • Loading and preprocessing a sample dataset.
  • Initializing ValidMind objects such as datasets and models.
  • Assigning predictions to datasets.
  • Setting custom context for test descriptions, including enabling and disabling use case context and adding test-specific context.

• Test-Specific Context Examples: The notebook provides examples of how to add test-specific context for various tests, such as DatasetDescription, ClassImbalance, HighCardinality, MissingValues, UniqueRows, TooManyZeroValues, IQROutliersTable, DescriptiveStatistics, and PearsonCorrelationMatrix. Each example demonstrates how to tailor the LLM-generated descriptions to reflect business-specific criteria and expectations.

These enhancements aim to make the LLM-generated descriptions more relevant and actionable by incorporating additional context that aligns with the user's specific requirements and organizational standards.

Test Suggestions

• Test the notebook to ensure that the table of contents is generated correctly and links to the appropriate sections.
• Verify that enabling and disabling the use case context works as expected by checking the output of LLM-generated descriptions.
• Test the addition of test-specific context for each example provided in the notebook to ensure that the context is correctly applied and reflected in the descriptions.
• Check that the notebook runs without errors and produces the expected outputs for all code cells.
• Validate that the changes do not affect the existing functionality of the ValidMind library.