v0.9.6

What's Changed

• Ensure metadata is a separate copy from model properties by @Shelnutt2 in #212

Full Changelog: v0.9.5...v0.9.6

v0.9.5

What's Changed

• Store Long Model Previews in Array Metadata by @Shelnutt2 in #211

Full Changelog: v0.9.4...v0.9.5

v0.9.4

What's Changed

• Memory budget config parameter fix by @ktsitsi in #208
• Remove unused SharedObjectLoadingScope in tensorflow_keras by @Shelnutt2 in #209
• Update to support tensorflow >= 2.11 by @Shelnutt2 in #210

New Contributors

• @Shelnutt2 made their first contribution in #209

Full Changelog: v0.9.3...v0.9.4

v0.9.3

What's Changed

• Fixed Incomplete Retries Error Check by @georgeSkoumas in #206
• Adding API docs for TileDB-ML by @ktsitsi in #205
• Improve PyTorch Performance with Multiple Workers and Shuffling by @georgeSkoumas in #207

Full Changelog: v0.9.2...v0.9.3

v0.9.2

What's Changed

• Pass Map Function to Array Params - PyTorch by @georgeSkoumas in #204

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

• Remove update attribute from model save by @georgeSkoumas in #200
• Increase unit test coverage by @georgeSkoumas in #201
• Added Timestamp While Writing Models and Metadata by @georgeSkoumas in #203

Full Changelog: v0.9.0...v0.9.1

v0.9.0

What's Changed

• Updates for Tensorflow 2.9+ & Pytorch 1.13 by @gsakkis in #195
• Update mypy to 0.991 by @gsakkis in #197
• Gsk/ch21380/upgrade ml model schema by @georgeSkoumas in #196
• Moderate model refactoring by @gsakkis in #199

Full Changelog: v0.8.0...v0.9.0

v0.8.0

What's Changed

• Introduce PyTorch DataPipes by @gsakkis in #173
• Refactor models design and add TB as extra array by @ktsitsi in #175
• [Hotfix] Fixed example notebooks by @ktsitsi in #176
• Fix PyTorchTileDBDataLoader to be iterable more than once by @gsakkis in #177
• Bump pytorch version to 1.12 by @gsakkis in #178
• Test _csr_to_coo_collate by @gsakkis in #180
• Default to TensorKind.SPARSE_COO for sparse tensors by @gsakkis in #181
• Add support for slicing on non-key indices [final] by @gsakkis in #184
• Unbatched data loading support for PyTorchTileDBDataLoader by @gsakkis in #183
• Reduce parametric tests by @gsakkis in #185
• Remove SparseData by @gsakkis in #186
• Support pytorch 1.11 by @gsakkis in #188
• Support reading dense tensors from sparse arrays by @gsakkis in #187
• Package _tensor_schema by @gsakkis in #189
• [Enhancement] Update Pull Request Guide by @georgeSkoumas in #182
• Add support for slicing on non key indices [sparse arrays] by @gsakkis in #190
• Add support for slicing on key dimension [sparse arrays] by @gsakkis in #191
• Add ConstrainedPartitionsIntRange by @gsakkis in #192
• Improve DenseTensorSchema.key_range by @gsakkis in #193
• Update precommit-hooks by @gsakkis in #194

Full Changelog: v0.7.0...v0.8.0

v0.7.0

What's Changed

• Fix ArrayParams imports by @georgeSkoumas in #162
• Array params import to readers by @ktsitsi in #163
• Take into account var-sized fields in SparseTensorSchema.max_partition_weight by @gsakkis in #164
• Bump minimum tensorflow to 2.6 and pytorch to 1.10 by @gsakkis in #168
• pass dataloader params in Dataloader, pytorch tests update by @georgeSkoumas in #165
• Add ArrayParams.tensor_kind by @gsakkis in #167
• Refactor parametric test utils by @gsakkis in #169
• Pass Through Arguments Tensorflow by @georgeSkoumas in #170
• Avoid intermediate sparse.COO instances if possible by @gsakkis in #172
• Read from sparse TileDB arrays to ragged tensors by @gsakkis in #171
• CI for notebooks by @gsakkis in #174

Full Changelog: v0.6.0...v0.7.0

v0.6.0

What's Changed

• [Hotfix] Re-run notebook examples for warnings and errors by @ktsitsi in #143
• Refactor reader tests by @gsakkis in #146
• Custom key dimension by @gsakkis in #145
• Adding early stopping callback to model notebooks by @ktsitsi in #144
• Support integer dimensions with non-zero based domain by @gsakkis in #147
• Determine default buffer size for sparse arrays by @gsakkis in #148
• Non-empty domain instead of full domain by @gsakkis in #149
• Refactor TensorSchema by @gsakkis in #150
• Privatization of models/base and PytorchTileDBDataset by @ktsitsi in #151
• [CI] Pin protobuf by @gsakkis in #152
• Generate tensors for dimensions by @gsakkis in #153
• InclusiveRange abstraction by @gsakkis in #154
• Refactor data loaders to use InclusiveRange by @gsakkis in #155
• Remove buffer_bytes parameter by @gsakkis in #156
• Support non-integer key dimension by @gsakkis in #157
• Introducing ArrayParams as TF-PyTorch readers argument by @ktsitsi in #158
• Update precommit hooks by @gsakkis in #160
• Make TensorSchema a dataclass by @gsakkis in #159
• Read any number of arrays by @gsakkis in #161

Full Changelog: v0.5.0...v0.6.0