v1.8.0

What's Changed

• Fixing errors in JOSS bibliography by @dfm in #509
• IdentityTask Fix by @RasmusOrsoe in #516
• move id extraction into I3FeatureExtractor by @RasmusOrsoe in #520
• Fix unbounded wandb in training example by @MortenHolmRep in #522
• move super() in Dataset by @RasmusOrsoe in #524
• Feature/define tito model by @ArturoLlorente in #500
• fix sim_type by @RasmusOrsoe in #518
• add features to graph, return batch in GraphNeTI3Module by @RasmusOrsoe in #547
• Add NoneType check for truth label assignments in Dataset by @RasmusOrsoe in #550
• Torch pyg pl>=2.0 by @AMHermansen in #548
• Fixed typo in setup py by @AMHermansen in #553
• Removed Python 3.7 from the Usage/Development table by @AMHermansen in #555
• Fixed typos in docstrings by @AMHermansen in #556
• Update acknowledgement on GraphNeT website by @asogaard in #561
• Update torch_gpu.txt by @AMHermansen in #563
• Update suggested CVMFS version in README.md by @RasmusOrsoe in #565
• Add ICRC Proceedings to README by @RasmusOrsoe in #564
• Graph Definition by @RasmusOrsoe in #558
• Minor fixes to graph definition by @RasmusOrsoe in #566
• Changed dtype in config files by @AMHermansen in #569
• Deployment update by @RasmusOrsoe in #567
• Support for python3.11 by @AMHermansen in #571
• Update CONTRIBUTING.md by @AMHermansen in #575
• Add check for torch threads in deployer by @RasmusOrsoe in #578
• Remove unused import statements by @AMHermansen in #577
• Fix type hint by @AMHermansen in #586
• Change recommended Icetray by @AMHermansen in #587
• (Issue #573) Update Random Number Generation to Use NumPy Random Generators by @samadpls in #585
• automatic attribute length adjustment by @RasmusOrsoe in #584
• Update training examples by @AMHermansen in #589
• fix typo in edge definition by @AMHermansen in #590
• Fix #592 by @Aske-Rosted in #593
• docs: Updated _get_all_indices docstring by @samadpls in #595
• Config files refactor, Examples polish, Perturbation in GraphDefinition by @RasmusOrsoe in #603
• Add meta class to handle config saving by @AMHermansen in #591
• Warning once fix by @RasmusOrsoe in #604
• Fix warnonce by @AMHermansen in #602
• Unit tests for Examples by @RasmusOrsoe in #605
• Fix OSError: [Errno 28] No space left on device in IceTray workflow by @RasmusOrsoe in #608
• Workflow Fix by @RasmusOrsoe in #612
• Update gpu requirements.txt by @RasmusOrsoe in #613
• Filter mask by @Aske-Rosted in #597
• RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method by @FredeVHansen in #614
• Feature/sequence bucketing by @ArturoLlorente in #610
• Update coverage reporting by @RasmusOrsoe in #615
• PercentileCluster by @RasmusOrsoe in #616
• Fix tito by @AMHermansen in #607
• Redefinition of Tito model by @ArturoLlorente in #611
• Geometry Tables by @RasmusOrsoe in #623
• Refactored model and standardmodel by @AMHermansen in #624
• Add optional recursive flag to data converter. by @kayla-leonard in #627
• Remove save_model_config decorator by @AMHermansen in #628
• 'ckpt_path' ambiguous in Model class by @ArturoLlorente in #617
• Update awtd and fadc parsing method by @RasmusOrsoe in #631
• cleaning code by @AMHermansen in #633
• enforce gpu arg in .predict_as_dataframe by @RasmusOrsoe in #636
• Add feature to extract and reconstruct track and cascade separately. by @kayla-leonard in #638
• Add improved earlystopping by @AMHermansen in #632
• StandardModel.fit: Automatically load in best-fit weights from early stopping when training is finished. by @RasmusOrsoe in #640
• Make sure warning for no EdgeDefinition is given only once. by @RasmusOrsoe in #641
• Task refactor by @RasmusOrsoe in #639
• Fix save_state_dict by @AMHermansen in #645
• Change argument gnn to architecture by @RasmusOrsoe in #646
• Implemented MinkowskiKNNEdges by @AMHermansen in #630
• Update QUESO energy model and unit test by @RasmusOrsoe in #652
• bump requirement files to torch=2.2 by @RasmusOrsoe in #656
• Log current lr in StandardModel by @RasmusOrsoe in #651
• Adopted pytorch lightning DataModule instead of DataLoaders and Dataset by @samadpls in #642
• Generalized DataConverter by @RasmusOrsoe in #659
• Change import of homophily in models.utils.py by @RasmusOrsoe in #663
• I3 converter bugs by @Aske-Rosted in #662
• Update torch version in gpu requirements to fix mis-match (2.2) by @kayla-leonard in #664
• Update logo and readme by @RasmusOrsoe in #668
• Added extra_repr and __repr__ methods to Model and ModelConfig classes by @samadpls in #665
• Rnn attention by @Aske-Rosted in #653
• Generalize graphnet.deployment to be compatible with other experiments by @RasmusOrsoe in #666
• Icemix kaggle solution by @ArturoLlorente in #660
• add parsing of labels by @Aske-Rosted in #657
• define SWA and EMA model by @ArturoLlorente in #661
• Weight norm per file. by @Aske-Rosted in #672
• Quick fix for proper handling of empty pulsemaps. by @Aske-Rosted in #673
• add_checkpoint_handling by @Aske-Rosted in #679
• predict_as_dataframe bugfix + syntax error in minkowski.py by @RasmusOrsoe in #658
• handling of empty DB when merging by @Aske-Rosted in #675
• Update of Parquet-backend by @RasmusOrsoe in #677
• IceMix models description by @ArturoLlorente in #681
• Integration of #660 and Example Definition by @ArturoLlorente in #680
• combined_extractor by @Aske-Rosted in #678
• Integration of LiquidO by @RasmusOrsoe in #682
• remove pisa-related code by @RasmusOrsoe in #687
• Bugfix to #685 and #683 by @RasmusOrsoe in #686
• Add PrometheusReader and extractors by @RasmusOrsoe in #688
• CuratedDataset - A method for sharing pre-converted datasets by @RasmusOrsoe in #689
• Add all public Prometheus Detector classes by @RasmusOrsoe in #693
• Update DataModule by @RasmusOrsoe in #694
• Update Type hints in GraphNeTDataModule by @RasmusOrsoe in #695
• ISeeCube implentation by @chenlinear in #676
• First Documentation Update by @RasmusOrsoe in #696
• update documentation workflow by @RasmusOrsoe in #697
• Update torch and macos version by @RasmusOrsoe in #700
• Add developer notes to documentation by @RasmusOrsoe in #698
• fix pathing issues for images on landing page in documentation by @RasmusOrsoe in #702
• Add public datasets by @RasmusOrsoe in #701
• fix truth table reference by @RasmusOrsoe in #704
• rename IceCube86 -> IceCube86Prometheus by @RasmusOrsoe in #706
• change pathing for intro.rst by @RasmusOrsoe in #708
• fix references to flowchart figure in landing page by @RasmusOrsoe in #709
• Installation missprint by @ArturoLlorente in #712
• added detector geometry attibutes to Kaggle detector class by @ArturoLlorente in #703
• Fixed cu121 at torch_cu121.txt by @timinar in #710
• requirements update by @Aske-Rosted in #719
• Refactor standard model by @RasmusOrsoe in #707
• RMSE von Mises-Fisher 3d loss by @Aske-Rosted in #716
• add isolated nodes by @Aske-Rosted in #715
• Alter logic of default NodeDefinition in GraphDefinition by @RasmusOrsoe in #724
• add missing abs to Track label by @RasmusOrsoe in #725
• Cast single path to list of paths in DataConverter.merge_files by @RasmusOrsoe in #726
• weights utilities + fix by @Aske-Rosted in #717
• unused params fix by @Aske-Rosted in #727
• Fix unit tests by @RasmusOrsoe in #733
• Updated Examples to GraphNetDataModule by @OscarBarreraGithub in #730
• Remove unused modules by @AlexKurek in #732
• Fix main imports by @RasmusOrsoe in #735
• Particlenet added and all km3net dependencies are deleted by @giogiopg in #740
• exchanged GNN with Model in the checks and in the type hint by @niklasmei in #744
• Improvements to parquet dataloading, sampling, batch sampling by @pweigel in #742
• Jammy flow integration by @RasmusOrsoe in #728
• Changed the import path of SQLiteDataset in the documentation to the currently correct location. by @niklasmei in #749
• [debug]Graph Util function by @KUcyans in #745
• pre-commit autoupdate by @Aske-Rosted in #758
• Revert "pre-commit autoupdate" by @Aske-Rosted in #761
• Pre commit version by @RasmusOrsoe in #760
• [Fix] Correct DynEdge instantiation in documentation by @samadpls in #762
• Train example comments by @Aske-Rosted in #756
• Fix deprecated I3Tray imports by @pweigel in #772
• Improvements for inelasticity reconstructions and other utilities by @pweigel in #763
• Optional replacing of feature mapping with identity by @Aske-Rosted in #769
• Clustering utilities by @Aske-Rosted in #770
• Ensemble from list of paths by @Aske-Rosted in #775
• fix ubuntu version by @Aske-Rosted in #779
• Add GRIT model by @pweigel in #777
• fix readme link by @Aske-Rosted in #785
• add snowstorm_dataset and IceCubehosted class by @sevmag in #783
• Restructure graph definition by @sevmag in #788
• I3Filters by @Aske-Rosted in #792
• f-string fix by @Aske-Rosted in #793
• Converter extractor changes by @Aske-Rosted in #795
• Writer changes by @Aske-Rosted in #794
• I3filtermapextractor by @Aske-Rosted in #801
• I3dictvalueextractor by @Aske-Rosted in #802
• Set model.train(mode=False) by @Aske-Rosted in #804
• add summary feature nodedefinition by @sevmag in #797
• Remove awkward and timer from dependencies by @RasmusOrsoe in #808
• Installation update; expand torch and cuda versions by @RasmusOrsoe in #806
• Fix coverage upload to CodeClimate by @RasmusOrsoe in #809
• Fix Documentation Publication Workflow by @RasmusOrsoe in #810
• Fix paper workflow by @sevmag in #805
• Mc labels by @Aske-Rosted in #807
• Fix gpu installation by @RasmusOrsoe in #811
• Fix typo in installation instruction by @RasmusOrsoe in #815
• Fixing the get_primaries function of I3Extractor by @sevmag in #818
• Fix calorimetry by @sevmag in #819
• Fix I3HighestEparticleExtractor by @sevmag in #822
• fix: eliminate RuntimeWarnings in von Mises-Fisher loss backward pass by @carlosm-silva in #824
• Add ESSnuSB paper to publications list by @kaareendrup in #828
• Improving the data string selection by @sevmag in #827
• Replace CodeClimate with Qlty Cloud for coverage report by @RasmusOrsoe in #838
• Fix CodeClimate by @RasmusOrsoe in #839
• Fix docker by @RasmusOrsoe in #840
• change link to docker action by @RasmusOrsoe in #841
• move docker action by @RasmusOrsoe in #842
• adjust docker path by @RasmusOrsoe in #843
• debug "push" by @RasmusOrsoe in #844
• Code Quality Fix by @RasmusOrsoe in #845
• Bce with logits by @sevmag in #847
• KM3NeT data extractors, example of use and detector classes by @giogiopg in #848
• Add MAGIC Telescope detector class by @astrojarred in #837
• Fix CUDA 12_6 and 12_8 installation commands by @marcjacquart in #850

New Contributors

• @dfm made their first contribution in #509
• @ArturoLlorente made their first contribution in #500
• @samadpls made their first contribution in #585
• @FredeVHansen made their first contribution in #614
• @chenlinear made their first contribution in #676
• @timinar made their first contribution in #710
• @OscarBarreraGithub made their first contribution in #730
• @AlexKurek made their first contribution in #732
• @giogiopg made their first contribution in #740
• @niklasmei made their first contribution in #744
• @pweigel made their first contribution in #742
• @KUcyans made their first contribution in #745
• @sevmag made their first contribution in #783
• @carlosm-silva made their first contribution in #824
• @astrojarred made their first contribution in #837
• @marcjacquart made their first contribution in #850

Full Changelog: v1.0.0...v1.8.0