Main changes
* Improvements to data loading during training:
- Batch processing of samples, requiring updates `WaveformDataset` and various transforms. This enables data parallelization for pre-processing, improving efficiency.
- Option (default true) to load waveforms from disk on-the-fly during training. This reduces the memory footprint. Optionally copy waveform dataset to local disk at start of training, typically for training on cluster.
* Better LSC integration (Asimov, CVMFS, OSG support).
* Added documentation for flow matching.
* Minimum Python == 3.10.
* Various bug fixes and improvements.
Complete list of changes
* FEAT: add argument to specify calibration correction direction by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/264
* Fix fmpe bug related to the cache by annalena-k in https://github.com/dingo-gw/dingo/pull/278
* fmpe doc added + pos encoding docstring by jonaswildberger in https://github.com/dingo-gw/dingo/pull/270
* Fix bugs related to early stopping and wandb by annalena-k in https://github.com/dingo-gw/dingo/pull/274
* added github pytest workflow, added pypi and pytest badges by vincentberenz in https://github.com/dingo-gw/dingo/pull/271
* Batching training transforms and wfd decompression by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/267
* Fixing failing tests from batched-wfd PR by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/280
* fixed path issue in test_waveform_dataset.py by vincentberenz in https://github.com/dingo-gw/dingo/pull/279
* forcing scipy to 1.13.1 by vincentberenz in https://github.com/dingo-gw/dingo/pull/281
* Adding a function which strictly generates the right number of waveform polarizations by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/262
* Parallelize post-processing change of spin-conversion-phase by stephengreen in https://github.com/dingo-gw/dingo/pull/213
* Fix train_unconditional_density_estimator, dingo_pipe sampling by stephengreen in https://github.com/dingo-gw/dingo/pull/272
* Included error message for non-zero standard deviation by annalena-k in https://github.com/dingo-gw/dingo/pull/286
* Prevent FlowWrapper from using embedding net when there is no context. by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/284
* Updates to batching by stephengreen in https://github.com/dingo-gw/dingo/pull/290
* asimov updates to handle calibration and bugfix with data folder and … by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/276
* using torch loading with fallback by vincentberenz in https://github.com/dingo-gw/dingo/pull/285
* Loading polarizations from disk by annalena-k in https://github.com/dingo-gw/dingo/pull/289
* Fix to GWSignal ASD dataset access by stephengreen in https://github.com/dingo-gw/dingo/pull/293
* Throw error when trying to standardize fixed parameter by annalena-k in https://github.com/dingo-gw/dingo/pull/294
* dingo pipe model meta data: falling back from 'meta' if not available by vincentberenz in https://github.com/dingo-gw/dingo/pull/295
* Fixing Issue When IS prior is broader than train prior by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/298
* added dingo-ci (running toy example and sending email reports) by vincentberenz in https://github.com/dingo-gw/dingo/pull/296
* Update event_dingo.yaml by sclyne623 in https://github.com/dingo-gw/dingo/pull/302
* Fix bug in recursive_hdf5_load by stephengreen in https://github.com/dingo-gw/dingo/pull/303
* Update minimum Python to 3.10. by stephengreen in https://github.com/dingo-gw/dingo/pull/304
* CVMFS Singularity Support and OSG integration by nihargupte-ph in https://github.com/dingo-gw/dingo/pull/291
* fix bug introduced in cvmfs integration by annalena-k in https://github.com/dingo-gw/dingo/pull/305
New Contributors
* vincentberenz made their first contribution in https://github.com/dingo-gw/dingo/pull/271
* sclyne623 made their first contribution in https://github.com/dingo-gw/dingo/pull/302
**Full Changelog**: https://github.com/dingo-gw/dingo/compare/v0.7.0...v0.8.0