chore(deps): update dependency lightly to v1.5.25

[renovate]

This PR contains the following updates:

Package	Change	Age	Confidence
lightly	==1.5.22 → ==1.5.25

---

Release Notes

lightly-ai/lightly (lightly)

v1.5.25

Compare Source

What's Changed

• Add LeJEPA ViT-S/16 benchmark results by @​gabrielfruet in #​1935
• Fix: ibotpluplus loss normalisation by @​IgorSusmelj in #​1937
• fix(loss): raise clear error for single-view DINOLoss instead of silent NaN by @​devteamaegis in #​1938
• docs: make LeJEPA public-facing (TRN-2158) by @​gabrielfruet in #​1940
• docs: add docstring to SimCLRViewTransform by @​saud5150 in #​1943
• fix: return plain Python float from cosine_schedule by @​saud5150 in #​1941
• Release: bump version to 1.5.25 by @​gabrielfruet in #​1944

New Contributors

• @​devteamaegis made their first contribution in #​1938
• @​saud5150 made their first contribution in #​1943

Full Changelog: lightly-ai/lightly@v1.5.24...v1.5.25

v1.5.24

Compare Source

Implement LeJEPA!

What's Changed

• isort and black for python pre-commit hook by @​gabrielfruet in #​1909
• [DOCS] Fix errors in CONTRIBUTING.md by @​RecreationalMath in #​1911
• [MNT-style]: Migrate from black, isort, and pylint to ruff by @​RecreationalMath in #​1914
• Main, knn_classifier and simclr modifications by @​Quentincami in #​1901
• fix override errors in masked causal vision transformer by @​gabrielfruet in #​1913
• Add more Lightly in Research papers by @​gabrielfruet in #​1921
• [MNT-style]: Add ruff pydocstyle for Google-style docstrings by @​RecreationalMath in #​1918
• Fix SIGReg distributed gradient under DDP by @​RecreationalMath in #​1923
• [ENH] Add LeJEPA core building blocks by @​RecreationalMath in #​1916
• refactor: simplify LeJEPA loss and module surface by @​gabrielfruet in #​1926
• [MNT] SIGReg cleanup: dedup tests, backfill docstring, document weights by @​RecreationalMath in #​1925
• Add LEJEPA examples by @​gabrielfruet in #​1927
• feat: Implement DSE (Dense representation Structure Estimator) by @​IgorSusmelj in #​1929
• docs: consolidate imagenet benchmark readme by @​gabrielfruet in #​1931
• feat: implement iBOT++ loss by @​IgorSusmelj in #​1930
• Release LightlySSL v1.5.24 by @​gabrielfruet in #​1933
• feat: add LeJEPA benchmark and metrics by @​gabrielfruet in #​1932

New Contributors

• @​RecreationalMath made their first contribution in #​1911
• @​Quentincami made their first contribution in #​1901

Full Changelog: lightly-ai/lightly@v1.5.23...v1.5.24

v1.5.23

Compare Source

What's Changed

• Fix deprecation warning totensor (#​1864) by @​KylevdLangemheen in #​1866
• Move functional transforms API to compatibility layer (#​1867) by @​KylevdLangemheen in #​1870
• Fix the banner and tab name of LightlySSL Docs by @​yutong-xiang-97 in #​1872
• Add DINOv2 to Readme (#​1873) by @​KylevdLangemheen in #​1875
• Fixed SMoG examples by @​thomasmarchioro3 in #​1876
• Add DirectCLR (#​781) by @​KylevdLangemheen in #​1874
• Update README badge links by @​Olexandr88 in #​1880
• Add link to social network twitter by @​Olexandr88 in #​1883
• Update packages for CI by @​IgorSusmelj in #​1896
• Contributing Guide: Add instructions on Imports by @​liopeer in #​1895
• Add LeJEPA SIGReg Loss implementation by @​gabrielfruet in #​1893
• Speed up momentum update by @​vsey in #​1899
• Add LightlyStudio reference to README by @​michal-lightly in #​1903
• Add DINOv2-3D: Self-Supervised 3D Vision Transformer Pretraining to README by @​guarin in #​1890
• Add torchvision 0.26 support by @​guarin in #​1904

New Contributors

• @​thomasmarchioro3 made their first contribution in #​1876
• @​Olexandr88 made their first contribution in #​1880
• @​gabrielfruet made their first contribution in #​1893
• @​vsey made their first contribution in #​1899

Full Changelog: lightly-ai/lightly@v1.5.22...v1.5.23

Many thanks to our contributors!

Models

• AIM: Scalable Pre-training of Large Autoregressive Image Models
• Barlow Twins: Self-Supervised Learning via Redundancy Reduction, 2021
• Bootstrap your own latent: A new approach to self-supervised Learning, 2020
• DCL: Decoupled Contrastive Learning, 2021
• DenseCL: Dense Contrastive Learning for Self-Supervised Visual Pre-Training, 2021
• DINO: Emerging Properties in Self-Supervised Vision Transformers, 2021
• DINOv2: Learning Robust Visual Features without Supervision, 2023
• FastSiam: Resource-Efficient Self-supervised Learning on a Single GPU, 2022
• iBOT: Image BERT Pre-Training with Online Tokenizer, 2021
• I-JEPA: Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture, 2023
• MAE: Masked Autoencoders Are Scalable Vision Learners, 2021
• MSN: Masked Siamese Networks for Label-Efficient Learning, 2022
• MoCo: Momentum Contrast for Unsupervised Visual Representation Learning, 2019
• NNCLR: Nearest-Neighbor Contrastive Learning of Visual Representations, 2021
• PMSN: Prior Matching for Siamese Networks, 2022
• SimCLR: A Simple Framework for Contrastive Learning of Visual Representations, 2020
• SimMIM: A Simple Framework for Masked Image Modeling, 2021
• SimSiam: Exploring Simple Siamese Representation Learning, 2020
• SMoG: Unsupervised Visual Representation Learning by Synchronous Momentum Grouping, 2022
• SwAV: Unsupervised Learning of Visual Features by Contrasting Cluster Assignments, M. Caron, 2020
• TiCo: Transformation Invariance and Covariance Contrast for Self-Supervised Visual Representation Learning, 2022
• VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning, Bardes, A. et. al, 2022
• VICRegL: VICRegL: Self-Supervised Learning of Local Visual Features, 2022

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.