lightly-ai/lightly

Issues

• DINO fails with PyTorch Fabric + multiple GPU

  #1535 opened 2 months ago by Chrispresso
  2

• Tutorial 3: small typo "strcture" -> "structure"

  #1534 opened 2 months ago by hmf
  1

• Bug in `GatherLayer.backward`

  #1528 opened 2 months ago by ordabayevy
  3

• Tensorboard shut down as of 1 Jan 2024

  #1527 opened 2 months ago by hmf
  2

• Ability to use pre-made augmentations

  #1525 opened 2 months ago by vedranf
  3

• Support latest `pydantic` version

  #1526 opened 2 months ago by ioangatop
  1

• Training the Swav model with a customized dataset reveals that the size of the .ckpt file is the same for each round of saves.

  #1521 opened 3 months ago by ad5014858
  2

• Support for Python 3.12

  #1473 opened 5 months ago by IgorSusmelj
  1

• Folder Structure for VicRegL

  #1504 opened 3 months ago by ramdhan1989
  2

• How to use lightly models on custom datasets

  #1516 opened 3 months ago by fereshtehforghani
  2

• SimCLR model.fit() does not start training with custom dataset loading from RAM

  #1518 opened 3 months ago by schelmi1
  5

• How to Manually Selecting Negative Samples

  #1449 opened 3 months ago by bryanbocao
  3

• [2024-03-07 22:34:06] An unexpected exception occurred: 1 validation error for GetEmbeddingsCsvReadUrlById embedding_id none is not an allowed value (type=type_error.none.not_allowed)

  #1513 opened 4 months ago by gauravkuppa
  2

• Fix examples on Windows

  #1512 opened 4 months ago by guarin
  0

• Add AIM examples and docs

  #1497 opened 4 months ago by guarin
  0

• Allow skipping weight initialisation for Masked Vision Transformers.

  #1511 opened 4 months ago by ersi-lightly
  0

• Unusual loss trend and 'OSError: [WinError 6] The handle is invalid' issue

  #1428 opened 7 months ago by Nutty233
  5

• Feature Request: New SSL algorithm called SparK: Sparse and Hierarchical masKed modeling

  #1462 opened 6 months ago by Djoels
  13

• Update BYOL and MoCo benchmarks

  #1471 opened 4 months ago by guarin
  0

• Use MMCRProjectionHead in Examples

  #1496 opened 4 months ago by guarin
  0

• Get embedding for Transformer Backbone in MAE and PMSN

  #1505 opened 4 months ago by ramdhan1989
  2

• Moco Memory bank example is error

  #1502 opened 4 months ago by ramdhan1989
  3

• Check for code changes in github actions

  #1485 opened 4 months ago by guarin
  0

• Feature Request: New SSL algorithm called Maximum Manifold Capacity Representations (NeurIPS 2023)

  #1423 opened 8 months ago by RylanSchaeffer
  5

• Add MMCR examples

  #1451 opened 5 months ago by guarin
  4

• Add MMCRProjectionHead

  #1481 opened 5 months ago by guarin
  1

• Respect HTTP proxy environment variables

  #1492 opened 5 months ago by benz0li
  2

• Supporting New SSL algorithm

  #1478 opened 5 months ago by opencomvis
  2

• Add function to load model from pretrained checkpoint

  #1475 opened 5 months ago by guarin
  0

• Remove opencv dependency from tests

  #1477 opened 5 months ago by guarin
  0

• W-MSE benchmark

  #1464 opened 6 months ago by guarin
  0

• W-MSE docs and examples

  #1465 opened 6 months ago by guarin
  0

• Intermittent downstream evaluation during pretraining

  #1420 opened 5 months ago by RylanSchaeffer
  5

• CycleCL

  #1447 opened 6 months ago by guarin
  3

• Clarification of x0 and x1 in the simsiam.py example

  #1448 opened 5 months ago by bryanbocao
  2

• Proper code for ddp training in Lightly

  #1458 opened 5 months ago by AricYang
  2

• Reproducibility of SimCLR

  #1419 opened 5 months ago by asparius
  8

• Add TiCoTransform

  #1470 opened 5 months ago by guarin
  0

• Additional field error in DatasourceConfigLOCAL instance

  #1463 opened 6 months ago by kais-bedioui
  3

• Feature Request: W-MSE (ICML 2021)

  #1424 opened 6 months ago by RylanSchaeffer
  1

• Clarification: Why does KNN eval normalize the representations?

  #1425 opened 6 months ago by RylanSchaeffer
  8

• Bug in MultiCropTransform class

  #1453 opened 6 months ago by Djoels
  1

• VICReg Loss De-Means Twice?

  #1445 opened 6 months ago by RylanSchaeffer
  1

• MMCR benchmark

  #1452 opened 6 months ago by guarin
  0

• deactivate_requires_grad does not fix the model!

  #1438 opened 7 months ago by AmirooR
  2

• MSN Loss computation

  #1439 opened 7 months ago by SaeedShurrab
  2

• Set model to eval mode for knn and linear classification

  #1440 opened 7 months ago by guarin
  0

• Update benchmark logs

  #1435 opened 7 months ago by guarin
  0

• Speed up benchmarks by disabling find_unused_parameters

  #1434 opened 7 months ago by guarin
  0

• Feature Request: Return all losses with multi-loss methods e.g., VICReg

  #1422 opened 8 months ago by RylanSchaeffer
  4