v22.4.0 #1815
Merged
v22.4.0 #1815
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…training Accept sampler designation in sampling of training
Add option --sample_at_first
Add PagedAdamW
Add attention processor
Update IPEX hijacks
* delete DDP wrapper * fix train_db vae and train_network * fix train_db vae and train_network unwrap * network grad sync --------- Co-authored-by: Kohya S <[email protected]>
IPEX support for Torch 2.1 and fix dtype erros
If max_size is too large relative to max_reso, it will calculate a height of zero for some buckets. This causes a crash later when it divides the width by the height. This change also simplifies some math and consolidates the redundant "size" variable into "width".
IPEX fix SDPA
Fix zero height buckets
Update 3 documents' url to the Chinese version.
Update README_中文教程.md
speed up latents nan replace
merge dev to main
Fix convert_diffusers20_original_sd.py and add --variant option for loading
Fix convert_diffusers20_original_sd.py and add metadata & variant options
|
Amazing PR. So when doing multi GPU training always include --ddp_gradient_as_bucket_view and --ddp_bucket_view ? both of them? by the way they do not exists in the code? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
tools/convert_diffusers20_original_sd.py. Thanks to Disty0! PR #1016--ddp_gradient_as_bucket_viewand--ddp_bucket_viewoptions are added tosdxl_train.py. Please specify these options for multi-GPU training.min_bucket_reso. Thanks to Cauldrath! PR #1008--sample_at_firstoption is added to each training script. This option is useful to generate images at the first step, before training. Thanks to shirayu! PR #907--ssoption is added to the sampling prompt in training. You can specify the scheduler for the sampling like--ss euler_a. Thanks to shirayu! PR #906keep_tokens_separatoris added to the dataset config. This option is useful to keep (prevent from shuffling) the tokens in the captions. See #975 for details. Thanks to Linaqruf!--keep_tokens_separator "|||"or withkeep_tokens_separator: "|||"in.toml. The tokens before|||are not shuffled.PagedAdamWis added. Thanks to xzuyn! PR #955finetune/make_captions.py. Thanks to CjangCjengh! PR #986