1.0.8

Stable version using newer libraries.

1.0.7

What's Changed

• add dali_device option to the Resize op by @ojss in #311
• CI: precommit by @Borda in #316
• Fix sphinx build by @vturrisi in #312
• Added logo and updated copyright licenses by @vturrisi in #317
• unify requirements by @Borda in #318
• feat(ci): update pip caching in CI by @SauravMaheshkar in #319
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #320
• Added checkpoints and results for MAE by @vturrisi in #321
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #323
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #325
• Support 0<data_fraction<1 for CustomDatasetWithoutLabels by @kaland313 in #328
• Update to pytorch lightning 2.0.2, remove lightning bolts dependency and other fixes by @vturrisi in #344
• Add code highlighting to the BibTeX entry by @bryant1410 in #342
• Use syntax updated with Python 3.8 by @bryant1410 in #343
• Fix tests on Python 3.10 by @bryant1410 in #347
• Temporarily revert back PL version by @vturrisi in #350
• Temporary fix for DALI and bump pytorch lightning version by @vturrisi in #351

New Contributors

• @ojss made their first contribution in #311
• @Borda made their first contribution in #316
• @SauravMaheshkar made their first contribution in #319
• @pre-commit-ci made their first contribution in #320
• @kaland313 made their first contribution in #328
• @bryant1410 made their first contribution in #342

Full Changelog: 1.0.6...1.0.7

1.0.6

What's Changed

• Improve backbones by @vturrisi in #273
• Update main_linear.py by @vturrisi in #274
• Added support for H5 data, improved scripts and data handling by @vturrisi in #275
• Fix num steps scheduler when interval is not step by @DonkeyShot21 in #280
• Fix zoo scripts by @DonkeyShot21 in #281
• Fix main linear by @vturrisi in #282
• Fix simsiam comment by @DonkeyShot21 in #285
• Fix num train steps by @vturrisi in #287
• Implement MAE by @vturrisi in #278
• Remove bias and norm from weight decay by @vturrisi in #289
• Exclude bias and norm from weight decay by @vturrisi in #291
• Pass temperature to mocov3 loss by @vturrisi in #294
• Sphinx actions by @turian in #299
• Add precommit hooks by @turian in #304
• Move from Argparse to Omegaconf by @vturrisi in #300

New Contributors

• @turian made their first contribution in #299

Full Changelog: 1.0.5...1.0.6

1.0.5

What's Changed

• Fix multicrop in readme by @DonkeyShot21 in #260
• link to bib by @DonkeyShot21 in #263
• Add Mocov3 by @vturrisi in #268
• Fix effective batch size and zero grad by @DonkeyShot21 in #261
• Fix how to get the number of devices on pytorch lightning 1.6+ by @vturrisi in #269
• Update issue templates by @DonkeyShot21 in #270

Full Changelog: 1.0.4...1.0.5

1.0.4

What's Changed

• Fixing CONTRIBUTING.md link by @PeppeSaccardi in #202
• Fix/midlevel by @vturrisi in #203
• Better SimCLR + SupCon by @vturrisi in #204
• mocov2plus imagenet by @DonkeyShot21 in #208
• improved tables by @DonkeyShot21 in #209
• barlow imagenet online by @DonkeyShot21 in #210
• Results/barlow by @vturrisi in #212
• Small compatibility fixes by @vturrisi in #220
• Add ConvNeXt by @vturrisi in #223
• Add warmup parameters to linear by @vturrisi in #227
• Update LR Schedulers every epoch by @vturrisi in #228
• Fixes some gathering issues by @vturrisi in #232
• Add MosaicML's ChannelLast optization by @vturrisi in #235
• Channel last by @vturrisi in #236
• Cleans design of methods a bit by @vturrisi in #238
• Provides a way to train with a percentage of data. by @vturrisi in #239
• Added gather to simclr by @vturrisi in #240
• Add check to see if using ddp by @vturrisi in #241
• Data fraction for linear by @vturrisi in #242
• Converts DALI into a DataModule, improves it and cleans stuff a bit. by @vturrisi in #244
• fix momentum tau by @DonkeyShot21 in #247
• Update requirements.txt by @vturrisi in #250
• fixed prototype norm in swav by @DonkeyShot21 in #249
• fix barlow script by @DonkeyShot21 in #254
• fix moco script by @DonkeyShot21 in #255
• lots of changes by @DonkeyShot21 in #251
• Fix backbones in readme by @DonkeyShot21 in #256
• Add Horovod support for multi-node training and improved LARS by @vturrisi in #257
• Fix ddp issue with simclr by @vturrisi in #259

New Contributors

• @PeppeSaccardi made their first contribution in #202

Full Changelog: 1.0.3...1.0.4

1.0.3

What's Changed

• Fix minor issues with NNBYOL and NNSiam by @DonkeyShot21 in #149
• Restructured augmentations by @DonkeyShot21 in #157
• Fix/bash files by @vturrisi in #158
• Fixed knn memory issues by @DonkeyShot21 in #159
• Fix/augmentations by @vturrisi in #164
• Add autoresume functionality by @vturrisi in #169
• Fix online k-NN by @DonkeyShot21 in #171
• Fix/docs by @vturrisi in #172
• Added Detectron by @vturrisi in #170
• Vibcreg by @Froskekongen in #182
• Improvement/jmlr by @vturrisi in #194
• Improvement/umap by @vturrisi in #199
• fix online knn on STL10 by @DonkeyShot21 in #200
• Access to multi-level features by @vturrisi in #201

New Contributors

• @Froskekongen made their first contribution in #182

Full Changelog: 1.0.2...1.0.3

1.0.2

🍕 Added NNSiam, NNBYOL, new tutorials for implementing new methods 1 and 2, more testing and fixed issues with custom data and linear evaluation.

1.0.1

🦘 Added online k-NN evaluation.
🤖 Re-structured ViT and added Swin.

1.0

🤖 Added ViT.
🖊️ Some renaming for better clarity.
📖 Added initial tutorials and improved documentation.
🐗 Extra utilities.

0.9

• Added support for Deepcluster V2
• Different shuffling strategy for DALI and now it can return the correct image indexes if needed
• Naming consistency improvements