Skip to content

Re-execute tutorials after inner_max_depth default change (#68) - #89

Merged
upadhyan merged 1 commit into
mainfrom
docs/68-rerun-tutorials
Oct 16, 2026
Merged

upadhyan merged 1 commit into
mainfrom
docs/68-rerun-tutorials

Conversation

@upadhyan

Copy link
Copy Markdown
Collaborator

Re-runs all tutorial notebooks after #88 changed the inner_max_depth default to None. Printed accuracies and importances move in most notebooks.

Prose updates:

  • tao.ipynb: the before/after TAO paragraph and the lambda_ sweep paragraph now match the outputs (test accuracy is flat after TAO; the sweep peaks at lambda_=0.03). Both paragraphs already quoted stale numbers before this change.
  • shape-functions.ipynb: the knobs list gives the new defaults and explains how inner_max_depth and inner_max_leaf_nodes interact.

Refs #68

The new default lets inner trees grow to 8 bins, so most printed
accuracies and importances move. Update the TAO prose (test accuracy is
now flat after refinement; the lambda_ sweep peaks at 0.03) and describe
the new inner-tree defaults in the shape-functions knobs list.

Refs #68
@upadhyan
upadhyan merged commit 3d42c70 into main Oct 16, 2026
8 checks passed
@upadhyan
upadhyan deleted the docs/68-rerun-tutorials branch October 16, 2026 18:47
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant