v0.4.1

Optimized dataloaders to allow for "spawn" multiprocess context.
Minor fixes and additions to utility functions.
New scripts for training from anndata.

v0.4.0

Refactored methods for training to streamline the end-to-end process.
Switched to a tiledb based framework for data stores instead of a zarr based framework.
New tiledb dataloader.
New training tutorial has been created to describe the workflow.
New example training script.
New scripts to build required post-training data structures.

v0.3.0

New methods for exhaustive queries.
New utility methods for filtering, sorting, and summarizing query results.
Expanded some utility methods and added optimizations to training.

v0.2.0

Model structure refactor and expanded documentation.

v0.1.0

Initial public release.