k-tahiro introduced a way to construct recommenders in a typed & homogeneous way using pydantic.
0.1.14
* Fix a bug in `split_dataframe_partial_user_holdout` when test_size = 0 * Refactor autopilot's way to communicate with child process
0.1.13
* Add `autopilot` function implementation. * Batch retrieval for known user ids.
0.1.12
0.1.11
* Fixed a bug for optimizer class (which occurs when a trial is pruned) * Added a batch retrieval function of recommendation result for IDMappedRecommender.
0.1.10
* Made two definitions of recall available. * Options to mask interactions other than the training matrix.