KevinMusgrave/pytorch-adapt

Issues

• Broken data links

  #102 opened 5 months ago by justinkay
  1

• Cannot import name 'zero_accuracy' from 'pytorch_metric_learning.utils.accuracy_calculator' in knn_validator.py

  #101 opened a year ago by hwopark
  2

• How to use the dataset wrappers

  #98 opened 2 years ago by Jio0728
  9

• Make it easy to use a hook with optimization turned off

  #99 opened 2 years ago by KevinMusgrave
  0

• Add domain parameter to CLossHook

  #68 opened 2 years ago by KevinMusgrave
  1

• Use DANN with target labels

  #95 opened 2 years ago by rtaiello
  2

• How do I change a few things in the implementation?

  #81 opened 2 years ago by chiragpr
  20

• Add Arxiv article link to readme

  #94 opened 2 years ago by impredicative
  1

• Specific Architecture

  #93 opened 2 years ago by rtaiello
  2

• Saving and Restoring a Trained Model

  #92 opened 2 years ago by r0f1
  2

• Add JUMBOT loss and hook

  #88 opened 2 years ago by KevinMusgrave
  0

• Use --extend-ignore in linter rule

  #79 opened 2 years ago by KevinMusgrave
  0

• Investigate nbval for testing notebooks

  #84 opened 2 years ago by KevinMusgrave
  0

• No module named pytorch_adapt

  #80 opened 2 years ago by chiragpr
  1

• Simplify load_objects by using latest pytorch-ignite

  #74 opened 2 years ago by KevinMusgrave
  0

• Make it clear that the downloaded OfficeHome dataset is resized

  #75 opened 2 years ago by KevinMusgrave
  0

• Add a Combined Dataset wrapper for semi-supervised domain adaptation

  #73 opened 2 years ago by KevinMusgrave
  0

• Rename CombinedSourceAndTargetDataset to something shorter

  #72 opened 2 years ago by KevinMusgrave
  0

• DiversityLoss: after_softmax needs to be checked

  #69 opened 2 years ago by KevinMusgrave
  1

• Hook outputs/losses should have class name as prefix

  #64 opened 2 years ago by KevinMusgrave
  0

• Add test that checks outputs/losses of all hooks and ensures no key collisions

  #65 opened 2 years ago by KevinMusgrave
  0

• Allow TargetDataset to return data with labels

  #13 opened 2 years ago by KevinMusgrave
  0

• Extension of the TargetDataset class.

  #60 opened 2 years ago by just-eoghan
  10

• Allow unlabeled dataset to be passed into TargetDataset

  #12 opened 2 years ago by KevinMusgrave
  0

• Extending the Lightning class (pytorch_adapt/frameworks/lightning/lightning.py)

  #58 opened 2 years ago by just-eoghan
  1

• Question on DataloaderCreator - How to create test sets

  #55 opened 2 years ago by just-eoghan
  6

• Update documentation to match v0.0.60

  #44 opened 2 years ago by KevinMusgrave
  0

• Store debug messages

  #24 opened 3 years ago by KevinMusgrave
  0

• Add more tests for the lightning wrapper

  #48 opened 3 years ago by KevinMusgrave
  0

• Swap order of losses and inputs into hooks

  #45 opened 3 years ago by KevinMusgrave
  0

• Change validator.score to validator.__call__

  #39 opened 3 years ago by KevinMusgrave
  0

• Adapter inference functions should return dictionaries instead of tuples

  #37 opened 3 years ago by KevinMusgrave
  0

• imgs should not be included in the output of f_utils.collector_step

  #38 opened 3 years ago by KevinMusgrave
  0

• Remove containers as an attribute from BaseAdapter

  #40 opened 3 years ago by KevinMusgrave
  0

• Move utils.saver to frameworks.ignite

  #30 opened 3 years ago by KevinMusgrave
  0

• frameworks.ignite: early_stopper should run at validation_interval

  #33 opened 3 years ago by KevinMusgrave
  0

• Assert validators.keys() and weights.keys() are equal

  #35 opened 3 years ago by KevinMusgrave
  0

• Change SND default batch size to ^2 number

  #36 opened 3 years ago by KevinMusgrave
  0

• Retrieve deeply nested hooks by name

  #43 opened 3 years ago by KevinMusgrave
  0

• Check if StochasticLinear weight initialization makes sense

  #23 opened 3 years ago by KevinMusgrave
  0

• Change default ignore_epoch to None

  #22 opened 3 years ago by KevinMusgrave
  0

• Set check_hash to True for load_state_dict_from_url

  #26 opened 3 years ago by KevinMusgrave
  0

• Allow dataloaders to be passed into Ignite.run, instead of just DataloaderCreator

  #14 opened 3 years ago by KevinMusgrave
  0

• Add pretrained models for Office31 and OfficeHome

  #9 opened 3 years ago by KevinMusgrave
  0

• Add example notebooks that shows hooks/adapters for each paper

  #21 opened 3 years ago by KevinMusgrave
  0

• Pass optimizers into hook __call__ instead of __init__

  #15 opened 3 years ago by KevinMusgrave
  0

• Consider "_preds" suffix

  #11 opened 3 years ago by KevinMusgrave
  0

• Typo in the ATDOC algorithm

  #10 opened 3 years ago by bluelg
  1

• Improve table visibility in documentation dark mode

  #8 opened 3 years ago by KevinMusgrave
  0

• Add functional versions of validators

  #7 opened 3 years ago by KevinMusgrave
  0