GitLab

Revert "[LoRA] introduce LoraBaseMixin to promote reusability. (#8774)"

This reverts commit 527430d0.

* Update TensorRT img2img pipeline
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>

* Update TensorRT version installed
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>

* make style and quality
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>

* Update examples/community/stable_diffusion_tensorrt_img2img.py
Co-authored-by: Tolga Cangöz <46008593+tolgacangoz@users.noreply.github.com>

* Update examples/community/README.md
Co-authored-by: Tolga Cangöz <46008593+tolgacangoz@users.noreply.github.com>

* Apply style and quality using ruff 0.1.5
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>

---------
Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>
Co-authored-by: Tolga Cangöz <46008593+tolgacangoz@users.noreply.github.com>
Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>

* introduce  to promote reusability.

* up

* add more tests

* up

* remove comments.

* fix fuse_nan test

* clarify the scope of fuse_lora and unfuse_lora

* remove space

* rewrite fuse_lora a bit.

* feedback

* copy over load_lora_into_text_encoder.

* address dhruv's feedback.

* fix-copies

* fix issubclass.

* num_fused_loras

* fix

* fix

* remove mapping

* up

* fix

* style

* fix-copies

* change to SD3TransformerLoRALoadersMixin

* Apply suggestions from code review
Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>

* up

* handle wuerstchen

* up

* move lora to lora_pipeline.py

* up

* fix-copies

* fix documentation.

* comment set_adapters().

* fix-copies

* fix set_adapters() at the model level.

* fix?

* fix

---------
Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>

update

Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com>

* SD3 training fixes
Co-authored-by: bghira <59658056+bghira@users.noreply.github.com>

* rewrite noise addition part to respect the eqn.

* styler

* Update examples/dreambooth/README_sd3.md
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>

---------
Co-authored-by: bghira <59658056+bghira@users.noreply.github.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>

* remove resume_download

* fix: _fetch_index_file call.

* remove resume_download from docs.

* Fix multi-gpu case

* Prefer previously created `unwrap_model()` function

For `torch.compile()` generalizability

* `chore: update unwrap_model() function to use accelerator.unwrap_model()`

* add animatediff_ipex community pipeline

* address the 1st round review comments

* Remove unused line


---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Improve trainer model cards

* Update train_dreambooth_sd3.py

* Update train_dreambooth_lora_sd3.py

* add link to adapters loading doc

* Update train_dreambooth_lora_sd3.py

---------
Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com>

* update

* update

* update

* Add vae_roundtrip.py example

* Add cuda support to vae_roundtrip

* Move vae_roundtrip.py into research_projects/vae

* Fix channel scaling in vae roundrip and also support taesd.

* Apply ruff --fix for CI gatekeep check

---------
Co-authored-by: Álvaro Somoza <asomoza@users.noreply.github.com>

* add clip_skip

* style

* smol fix

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* add experimental scripts to train SD3 transformer lora on colab

* add readme

* add colab

* Apply suggestions from code review
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* fix link in the notebook.

---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

Revert "[LoRA] introduce `LoraBaseMixin` to promote reusability. (#8670)"

This reverts commit a2071a18.

* introduce  to promote reusability.

* up

* add more tests

* up

* remove comments.

* fix fuse_nan test

* clarify the scope of fuse_lora and unfuse_lora

* remove space

update

* add a test suite for SD3 DreamBooth

* lora suite

* style

* add checkpointing tests for LoRA

* add test to cover train_text_encoder.

* fix

* fix things.
Co-authored-by: Linoy Tsaban <linoy.tsaban@gmail.com>

* remove patch

* apply suggestions

---------
Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com>
Co-authored-by: sayakpaul <spsayakpaul@gmail.com>
Co-authored-by: Linoy Tsaban <linoy.tsaban@gmail.com>

* update training

* update

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com>

[train_controlnet_sdxl.py] Fix the LR schedulers when num_train_epochs is passed in a distributed training env (#8476)

* Create diffusers.yml

* num_train_epochs

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* new pipeline

* minor changes

* minor changes

* minor changes

* minor changes

* minor changes

* minor changes

* minor changes

* fix

* fix

* aligning with blora script

* aligning with blora script

* aligning with blora script

* aligning with blora script

* aligning with blora script

* remove prints

* style

* default val

* license

* move save_model_card to outside push_to_hub

* Update train_dreambooth_lora_sdxl_advanced.py

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* add clip text-encoder training

* no dora

* text encoder traing fixes

* text encoder traing fixes

* text encoder training fixes

* text encoder training fixes

* text encoder training fixes

* text encoder training fixes

* add text_encoder layers to save_lora

* style

* fix imports

* style

* fix text encoder

* review changes

* review changes

* review changes

* minor change

* add lora tag

* style

* add readme notes

* add tests for clip encoders

* style

* typo

* fixes

* style

* Update tests/lora/test_lora_layers_sd3.py
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Update examples/dreambooth/README_sd3.md
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* minor readme change

---------
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Class methods are supposed to use `cls` conventionally

* `make style && make quality`

* An Empty commit

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Fix typos

* Fix typos & up style

* chore: Update numbers

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Discourage using `revision`

* `make style && make quality`

* Refactor code to use 'variant' instead of 'revision'

* `revision="bf16"` -> `variant="bf16"`

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Trim all the trailing white space in the whole repo

* Remove unnecessary empty places

* make style && make quality

* Trim trailing white space

* trim

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* Fix typos & improve contributing page

* `make style && make quality`

* fix typos

* Fix typo

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

[train_lcm_distill_lora_sdxl.py] Fix the LR schedulers when num_train_epochs is passed in a distributed training env (#8446)

fix num_train_epochs
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

Add extra performance features for EMAModel, torch._foreach operations and better support for non-blocking CPU offloading (#7685)

* Add support for _foreach operations and non-blocking to EMAModel

* default foreach to false

* add non-blocking EMA offloading to SD1.5 T2I example script

* fix whitespace

* move foreach to cli argument

* linting

* Update README.md re: EMA weight training

* correct args.foreach_ema

* add tests for foreach ema

* code quality

* add foreach to from_pretrained

* default foreach false

* fix linting

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: drhead <a@a.a>

* get rid of the legacy lora remnants and make our codebase lighter

* fix depcrecated lora argument

* fix

* empty commit to trigger ci

* remove print

* empty

* Update train_dreambooth_lora_sd3.py

* Update train_dreambooth_lora_sd3.py

* Update train_dreambooth_sd3.py

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* change to logit_normal as the weighting scheme

* sensible default mote

fix the position of param casting when loading them

refactor the density and weighting utilities.

Co-authored-by: bghira <bghira@users.github.com>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>