NEWS.md

In artichaud1/tidytune: A package for Tidy Hyperparameter Tuning

tidytune 0.0.1.9002

New feature:

• Leveraging furrr package. Processing is done in parallel across the folds, using future_pmap instead of pmap.

tidytune 0.0.1.9001

Major changes:

• Removed all dependencies to modeling functions, replacing train_predict_func parameter with score_func. Whereas train_predict_func returned a list of predictions, score_func must return a score or a list of named scores.
• Removed metrics parameter.
• Added Keras example and added accept_func parameter to accept/reject parameter combinations based on convexity assumption.

artichaud1/tidytune documentation built on May 20, 2019, 9:13 p.m.