search

stefanradev93 / BayesFlow

A Python library for amortized Bayesian workflows using generative neural networks.
https://bayesflow.org/
MIT License
286 stars 45 forks source link

Diagnostics refactor #172

Open jerrymhuang opened 1 month ago

jerrymhuang commented 1 month ago

Refactored all diagnostics modules:

  1. Created helper functions for common procedures shared by SBC-related diagnostics modules;
  2. Created a more flexible plot_distribution_2d() module for pairplots;
  3. Migrated all diagnostics modules and related utilities to experimental;
  4. Made all diagnostics modules more dependent on Keras;
  5. Add data types for all module inputs (except samples; in discussion);

Still to come:

  1. Deeper refactoring of plot_sbc_ecdf() and plot_sbc_histogram();
  2. Make module inputs compatible with nested sample dictionaries (discussion focus for above point 5);
  3. Systematic testing of all refactored modules;
  4. A user-friendly workflow.diagnose() front-end (or the like) that encapsulates the modularized design of the modules.

Any feedback or suggestions are welcome.

review-notebook-app[bot] commented 1 month ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

LarsKue commented 1 month ago

2000+ lines in 2 commits is too much, you might have to redo this one