Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit history

Commits on Oct 21, 2024

• Only add warning for deprecated scaling vpred loss function

  catboxanon

  committed
  e1b63c2

  Copy full SHA for e1b63c2

• Fix training for V-pred and ztSNR

  Show description for 8fc30f8
  catboxanon

  committed
  8fc30f8

  Copy full SHA for 8fc30f8

Commits on Sep 29, 2024

• adjust min/max bucket reso divisible by reso steps #1632

  kohya-ss

  committed
  fe2aa32

  Copy full SHA for fe2aa32

Commits on May 19, 2024

• simplify and update alpha mask to work with various cases

  kohya-ss

  committed
  da6fea3

  Copy full SHA for da6fea3

• 画像のアルファチャンネルをlossのマスクとして使用するオプションを追加 (#1223)

  Show description for db67529
  u-haru

  authored
  db67529

  Copy full SHA for db67529

• add `--log_config` option to enable/disable output training config

  kohya-ss

  committed
  c68baae

  Copy full SHA for c68baae

Commits on Apr 20, 2024

• passing filtered hyperparameters to accelerate

  ccharest93

  committed
  2c9db5d

  Copy full SHA for 2c9db5d

Commits on Apr 7, 2024

• Add option to use Scheduled Huber Loss in all training pipelines to improve resilience to data corruption (#1228)

  Show description for 90b1879
  

  kabachuha

  and

  kohya-ss

  authored
  90b1879

  Copy full SHA for 90b1879

Commits on Apr 4, 2024

• verify command line args if wandb is enabled

  kohya-ss

  committed
  cd587ce

  Copy full SHA for cd587ce

Commits on Mar 26, 2024

• Merge branch 'dev' into masked-loss

  kohya-ss

  committed
  ab1e389

  Copy full SHA for ab1e389

Commits on Mar 25, 2024

• make each script consistent, fix to work w/o DeepSpeed

  kohya-ss

  committed
  a2b8531

  Copy full SHA for a2b8531

Commits on Mar 20, 2024

• Merge branch 'dev' into deep-speed

  kohya-ss

  committed
  fbb98f1

  Copy full SHA for fbb98f1

• Merge branch 'dev' into masked-loss

  kohya-ss

  committed
  9b6b39f

  Copy full SHA for 9b6b39f

Commits on Mar 17, 2024

• common masked loss func, apply to all training script

  kohya-ss

  committed
  3419c3d

  Copy full SHA for 3419c3d

Commits on Mar 10, 2024

• save state on train end

  gesen2egee

  committed
  095b803

  Copy full SHA for 095b803

Commits on Feb 27, 2024

• make deepspeed_utils

  kohya-ss

  committed
  e3ccf8f

  Copy full SHA for e3ccf8f

• Merge branch 'deep-speed' into deepspeed

  kohya-ss

  authored
  eefb3cc

  Copy full SHA for eefb3cc

Commits on Feb 24, 2024

• fix to work with cpu_count() == 1 closes #1134

  kohya-ss

  committed
  f413201

  Copy full SHA for f413201

Commits on Feb 22, 2024

• refactored codes, some function moved into train_utils.py

  BootsofLagrangian

  committed
  4d5186d

  Copy full SHA for 4d5186d

Commits on Feb 17, 2024

• Merge branch 'dev' into dev_device_support

  kohya-ss

  committed
  baa0e97

  Copy full SHA for baa0e97

Commits on Feb 12, 2024

• fix to work `--console_log_xxx` options

  kohya-ss

  committed
  93bed60

  Copy full SHA for 93bed60

• Merge branch 'dev' into dev_device_support

  kohya-ss

  committed
  358ca20

  Copy full SHA for 358ca20

• add clean_memory_on_device and use it from training

  kohya-ss

  committed
  e24d960

  Copy full SHA for e24d960

Commits on Feb 9, 2024

• the reason not working grad accum steps found. it was becasue of my accelerate settings

  BootsofLagrangian

  committed
  03f0816

  Copy full SHA for 03f0816

Commits on Feb 8, 2024

• add logging args for training scripts

  kohya-ss

  committed
  055f02e

  Copy full SHA for 055f02e

Commits on Feb 7, 2024

• fix full_fp16 compatible and train_step

  BootsofLagrangian

  committed
  6255661

  Copy full SHA for 6255661

Commits on Feb 5, 2024

• apply offloading method runable for all trainer

  BootsofLagrangian

  committed
  7d2a926

  Copy full SHA for 7d2a926

• fix all trainer about vae

  BootsofLagrangian

  committed
  4295f91

  Copy full SHA for 4295f91

Commits on Feb 4, 2024

• Replace print with logger if they are logs (#905)

  Show description for 5f6bf29
  

  shirayu

  and

  kohya-ss

  authored
  5f6bf29

  Copy full SHA for 5f6bf29

Commits on Feb 3, 2024

• support deepspeed

  BootsofLagrangian

  committed
  dfe08f3

  Copy full SHA for dfe08f3

Commits on Jan 31, 2024

• Fix IPEX support and add XPU device to device_utils

  Disty0

  committed
  a6a2b5a

  Copy full SHA for a6a2b5a

Commits on Jan 23, 2024

• Refactor memory cleaning into a single function

  akx

  committed
  afc3870

  Copy full SHA for afc3870

Commits on Jan 19, 2024

• Deduplicate ipex initialization code

  akx

  committed
  6f3f701

  Copy full SHA for 6f3f701

Commits on Jan 14, 2024

• Add wandb_run_name parameter to init_kwargs #1032

  kohya-ss

  committed
  32b759a

  Copy full SHA for 32b759a

Commits on Dec 7, 2023

• fix duplicated sample gen for every epoch ref #907

  kohya-ss

  committed
  912dca8

  Copy full SHA for 912dca8