ivis: dimensionality reduction in very large datasets using Siamese Networks

What's Changed

• GitHub actions by @Szubie in #120
• Update doc requirements by @Szubie in #121
• GitHub pages by @Szubie in #122

Full Changelog: 2.09...2.10

ivis: dimensionality reduction in very large datasets using Siamese Networks

Full Changelog: 2.08...2.09

ivis: dimensionality reduction in very large datasets using Siamese Networks

2.08

version bump for minor release

ivis: dimensionality reduction in very large datasets using Siamese Networks

• Added ability to save/load ivis models that have not been trained. This also fixes an issue when using GridSearchCV in conjunction with ivis
• Bugfix for triplet generator when used in conjunction with a dataset exposing the custom get_triplet_data method

ivis: dimensionality reduction in very large datasets using Siamese Networks

New features:

• ivis models are now serializable via pickle/dill/joblib. Thanks to @imatheussm for his contributions toward this.
• The save_model method now accepts an optional "save_format" argument. Setting it to "tfs" will export ivis models in the TensorFlow SavedModel format, which integrates well with other TensorFlow libraries.

ivis: dimensionality reduction in very large datasets using Siamese Networks

• Knn retrieval made more efficient by switching from multi-processing to multi-threading. Memory savings depend on OS and core count.
• Fixed issue where saved ivis models would attempt to load the index at the path they were saved with - this can't be relied on when the index is temporary and deleted after use.
• Fixed issue where Annoy Index metric parameter was not passed to an index that was loaded from disk.
• A few other things changed, including better error handling, cleaner code, and allowing for saving AnnoyKnnMatrix via pickle

ivis: dimensionality reduction in very large datasets using Siamese Networks

Highlights:

• Improved training speed for numpy arrray inputs thanks to a faster triplet generator.
• Batched retrieval capabilities that makes ivis much faster when training on out-of-memory data that is retrieved in parallel.
• Improved performance when using Ivis with precompute=False option by using multi-threading when retrieving batches of KNN on-demand.
• Added deprecation notices for minor upcoming changes to API for consistency and adherence to sklearn API.

ivis: dimensionality reduction in very large datasets using Siamese Networks

• improved memory utilization during KNN retrieval
• AnnoyIndex is now removed from disk after running Ivis

ivis: dimensionality reduction in very large datasets using Siamese Networks

Minor release

• Fixes zero chunk error #90

ivis: dimensionality reduction in very large datasets using Siamese Networks

Minor release addressing:

• Tensorflow 2.4 model save compatibility (#82)
• Training/Inference batch size concordance