Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Feb 28, 2024

• Merge branch 'master' into model_part_config

  Show description for d4e382c
  Nerogar

  committed
  d4e382c

  Copy full SHA for d4e382c

Commits on Feb 18, 2024

• support for stable cascade stage C 1B

  Nerogar

  committed
  852c51e

  Copy full SHA for 852c51e

Commits on Feb 17, 2024

• stable cascade sample and train fixes

  Nerogar

  committed
  37018d3

  Copy full SHA for 37018d3

Commits on Feb 14, 2024

• initial stable cascade fine tuning support

  Nerogar

  committed
  ca52fe1

  Copy full SHA for ca52fe1

Commits on Feb 13, 2024

• stable cascade sampling support

  Nerogar

  committed
  448abe9

  Copy full SHA for 448abe9

Commits on Feb 10, 2024

• unify model part config

  Nerogar

  committed
  9d24d89

  Copy full SHA for 9d24d89

Commits on Feb 5, 2024

• change export function to generate single json file

  Nerogar

  committed
  f2626ac

  Copy full SHA for f2626ac

Commits on Feb 4, 2024

• rename Param to Config

  Nerogar

  committed
  411daab

  Copy full SHA for 411daab

Commits on Feb 3, 2024

• unify param and arg classes, remove train by args script

  Nerogar

  committed
  0b0462e

  Copy full SHA for 0b0462e

Commits on Jan 20, 2024

• fix autocast for wuerstechen training

  Nerogar

  committed
  46ebeaa

  Copy full SHA for 46ebeaa

Commits on Jan 11, 2024

• fix other models

  Nerogar

  committed
  6fd0e9a

  Copy full SHA for 6fd0e9a

Commits on Jan 7, 2024

• enable xformers for PixArt and improve caching vram usage

  Nerogar

  committed
  855daf3

  Copy full SHA for 855daf3

Commits on Jan 4, 2024

• fix PixArt text and batch handling

  Nerogar

  committed
  072c049

  Copy full SHA for 072c049

Commits on Jan 3, 2024

• PixArt Alpha fine tuning

  Nerogar

  committed
  0089944

  Copy full SHA for 0089944

Commits on Dec 26, 2023

• fix some debug mode bugs and add prompt output

  Nerogar

  committed
  58bd28d

  Copy full SHA for 58bd28d

Commits on Dec 23, 2023

• fix exception when loading older concept files

  Nerogar

  committed
  3c43a62

  Copy full SHA for 3c43a62

• concept loss weight

  Nerogar

  committed
  11691fb

  Copy full SHA for 11691fb

• add shuffling when aspect ratio bucketing is disabled

  Nerogar

  committed
  92a8779

  Copy full SHA for 92a8779

Commits on Dec 22, 2023

• resolution override without aspect ratio bucketing

  Nerogar

  committed
  e80100a

  Copy full SHA for e80100a

• resolution override with aspect ratio bucketing

  Nerogar

  committed
  4f1e8ad

  Copy full SHA for 4f1e8ad

Commits on Dec 21, 2023

• fixed image augmentations

  Nerogar

  committed
  7c8ae29

  Copy full SHA for 7c8ae29

• update mgds to fix concept name dependency while caching

  Nerogar

  committed
  fe420cf

  Copy full SHA for fe420cf

Commits on Dec 19, 2023

• switch to disable concepts

  Nerogar

  committed
  f6be1b6

  Copy full SHA for f6be1b6

Commits on Dec 18, 2023

• repeats for Wuerstchen and fixes

  Nerogar

  committed
  43cef5b

  Copy full SHA for 43cef5b

Commits on Dec 17, 2023

• repeats for SDXL, optimized caching

  Nerogar

  committed
  e661067

  Copy full SHA for e661067

• initial version of concept variations and repeats

  Nerogar

  committed
  97b9a21

  Copy full SHA for 97b9a21

Commits on Dec 9, 2023

• multi resolution training

  Nerogar

  committed
  53ed75d

  Copy full SHA for 53ed75d

• embedding refactoring

  Nerogar

  committed
  60dca4d

  Copy full SHA for 60dca4d

Commits on Dec 4, 2023

• fix pause between epochs when no caching is needed

  Nerogar

  committed
  7f17c1d

  Copy full SHA for 7f17c1d

• remove kandinsky

  Nerogar

  committed
  32690cc

  Copy full SHA for 32690cc

Commits on Nov 25, 2023

• fix a missing output value in the SDXL data loader

  Nerogar

  committed
  4f144dc

  Copy full SHA for 4f144dc

Commits on Nov 24, 2023

• wuerstchen embedding training

  Nerogar

  committed
  58a1ea1

  Copy full SHA for 58a1ea1

Commits on Nov 16, 2023

• Merge branch 'master' into wuerstchen

  Show description for 725dae0
  Nerogar

  committed
  725dae0

  Copy full SHA for 725dae0

• fixed text encoder output handling for wuerstchen v2

  Nerogar

  committed
  d05b3df

  Copy full SHA for d05b3df

• fixed text encoder output norm when training SD 1.5 without text encoder

  Nerogar

  committed
  614e801

  Copy full SHA for 614e801