Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Dec 25, 2022

• Fixes a major typo (victorca25#74)

  Kim2091

  committedDec 25, 2022
  1d25dcd

Commits on Oct 26, 2021

• add A2N architecture for testing

  victorca25

  committedOct 26, 2021
  12d006f

Commits on Oct 23, 2021

• fix auto norm calculation

  victorca25

  committedOct 23, 2021
  a0f0a22

Commits on Oct 22, 2021

• add gradient clipping options to SR template files

  victorca25

  committedOct 22, 2021
  898db67

• add gradient clipping options. Both 'norm' and 'value' gradient clipping are supported. Initially only available in the SR model, can be added to other models.

  victorca25

  committedOct 22, 2021
  a2c14f3

Commits on Oct 21, 2021

• for issue victorca25#56: add mention of the LR crop size for the super-resolution case in documentation

  victorca25

  committedOct 21, 2021
  c0ea432

• Various fixes (victorca25#58)

  joeyballentine

  committedOct 21, 2021
  b187a1b

Commits on Oct 1, 2021

• add feature extraction-matching for patchgan and multiscale patchgan discriminators

  victorca25

  committedOct 1, 2021
  f332913

• Show current step/epoch when saving models (victorca25#49)

  Sazoji

  committedOct 1, 2021
  ecc357f

Commits on Sep 18, 2021

• restructured the adversarial loss interface, refactored into reusable methods, added the missing formulations for conditional case, preparation for pix2pixHD tests. Add safety check for when layer …

  victorca25

  committedSep 18, 2021
  d09bd21

Commits on Sep 12, 2021

• merged the regular and precise losses loss builder, code refactoring, modified models to use merged refactored class

  victorca25

  committedSep 12, 2021
  59ad9bf

Commits on Sep 7, 2021

• fix error in if block

  victorca25

  committedSep 7, 2021
  7a1179c

• add automatic selector for mid-size ESRGAN network configuration for anime model

  victorca25

  committedSep 7, 2021
  626f281

Commits on Sep 6, 2021

• small doc correction

  victorca25

  committedSep 6, 2021
  1e94858

• add option to use pixel-unshuffle to convert the input images to lower resolution and extended depth, reducing the resources required to train them

  victorca25

  committedSep 6, 2021
  7909a94

Commits on Sep 4, 2021

• fix for virtual batch

  victorca25

  committedSep 4, 2021
  0f20430

Commits on Sep 3, 2021

• combined all depth_to_space/pixel_shuffle/unsqueeze2d and space_to_depth/pixel_unshuffle/squeeze2d functions in the code into reusable functions and classes. Could be replaced by torch.nn.PixelShuf…

  victorca25

  committedSep 3, 2021
  26867a7

Commits on Aug 31, 2021

• added a BatchAugment class to interface with the function, cleaned up models modules with refactoring using the new class

  victorca25

  committedAug 31, 2021
  a69d9f2

Commits on Aug 29, 2021

• models refactoring for reusable components, simplifying and reducing code duplication (sr_model, base_model, ppon_model, pix2pix_model, cyclegan_model, wbc_model, SRFlow_model, others TBD), add Ada…

  victorca25

  committedAug 29, 2021
  d97b95f

Commits on Aug 26, 2021

• add AdaTarget option for adaptive model output generation and initial required changes to use. Small doc updates

  victorca25

  committedAug 26, 2021
  5113291

Commits on Aug 17, 2021

• update ppon and srflow models to refer to renamed sr_model module

  victorca25

  committedAug 17, 2021
  58f57a9

Commits on Aug 16, 2021

• update change to sr_model module, minor documentation updates

  victorca25

  committedAug 16, 2021
  84faf37

• renamed SRRaGAN_model module to sr_model

  victorca25

  committedAug 16, 2021
  1c9d85c

Commits on Aug 15, 2021

• small corrections in docs

  victorca25

  committedAug 15, 2021
  478bfe4

• update SR template files, clean-up

  victorca25

  committedAug 15, 2021
  b5682f1

• update documentation with presets information, extended pipeline options

  victorca25

  committedAug 15, 2021
  8a559f8

• Added a UNet discriminator, with optional flag to use spectral norm

  victorca25

  committedAug 15, 2021
  68f4828

• Added parsing of the preset files to create final options dictionary. Consolidated scaling functions and new scaling class for composable augmentations, which now use the arguments passed from the …

  victorca25

  committedAug 15, 2021
  8ee93a9

Commits on Aug 14, 2021

• updated base preset files, merged misc preset with noise preset, added resize base preset. Added corresponding presets for Real-SR, BSRGAN and Real-ESRGAN (rESRGAN) training strategies.

  victorca25

  committedAug 14, 2021
  b978d4f

• rename base presets files

  victorca25

  committedAug 14, 2021
  4686fef

• updated augmennt

  victorca25

  committedAug 14, 2021
  3011df7

• added even image size correction to camera unprocess pipeline (for mosaic operation), added scale factor for Poisson noise type by extracting the correlated noise from the corrupted image and readd…

  victorca25

  committedAug 14, 2021
  38d8c48

Commits on Jul 24, 2021

• update files to reflect new repo name, from 'BasicSR' to 'traiNNer'

  victorca25

  committedJul 24, 2021
  66ec08a

• add documentation and training template for WBC

  victorca25

  committedJul 24, 2021
  b6c6204

• add WBC model, network architectures and update defaults and options handling to include WBC

  victorca25

  committedJul 24, 2021
  8bbab16

Pagination

PreviousNext