GitLab

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU computing

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

* Improve the performance and suitable for NPU

---------
Co-authored-by: 蒋硕 <jiangshuo9@h-partners.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* merge 9588

* max_shard_size="5GB" for colab running

* conversion script updates; modeling test; refactor transformer

* make fix-copies

* Update convert_cogview3_to_diffusers.py

* initial pipeline draft

* make style

* fight bugs 🐛

🪳

* add example

* add tests; refactor

* make style

* make fix-copies

* add co-author

YiYi Xu <yixu310@gmail.com>

* remove files

* add docs

* add co-author
Co-Authored-By: YiYi Xu <yixu310@gmail.com>

* fight docs

* address reviews

* make style

* make model work

* remove qkv fusion

* remove qkv fusion tets

* address review comments

* fix make fix-copies error

* remove None and TODO

* for FP16(draft)

* make style

* remove dynamic cfg

* remove pooled_projection_dim as a parameter

* fix tests

---------
Co-authored-by: Aryan <aryan@huggingface.co>
Co-authored-by: YiYi Xu <yixu310@gmail.com>

* Flux - soft inpainting via differential diffusion

* .

* track changes to FluxInpaintPipeline

* make mask arrangement simplier

* make style

---------
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Álvaro Somoza <asomoza@users.noreply.github.com>
Co-authored-by: asomoza <somoza.alvaro@gmail.com>

increase transformers version in test_low_cpu_mem_usage_with_loading

* docs: fix xDiT doc image damage

* doc: move xdit images to hf dataset

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

replace ubuntu version to 22.04.

* Added diff diff support for kolors img2img

* Fized relative imports

* Fized relative imports

* Added diff diff support for Kolors

* Fized import issues

* Added map

* Fized import issues

* Fixed naming issues

* Added diffdiff support for Kolors img2img pipeline

* Removed example docstrings

* Added map input

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

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

* Improved code quality

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

Add files via upload

fix IsADirectoryError when running the training code

* flux controlnet control_guidance_start and control_guidance_end implement

* minor fix - added docstrings, consistent controlnet scale flux and SD3

fix dora test.

* make controlnet support interrupt

* remove white space in controlnet interrupt

* added pag to sd img2img pipeline


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

* allow loras to be loaded with low_cpu_mem_usage.

* add flux support but note https://github.com/huggingface/diffusers/pull/9510\#issuecomment-2378316687



* low_cpu_mem_usage.

* fix-copies

* fix-copies again

* tests

* _LOW_CPU_MEM_USAGE_DEFAULT_LORA

* _peft_version default.

* version checks.

* version check.

* version check.

* version check.

* require peft 0.13.1.

* explicitly specify low_cpu_mem_usage=False.

* docs.

* transformers version 4.45.2.

* update

* fix

* empty

* better name initialize_dummy_state_dict.

* doc todos.

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

* style

* fix-copies

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

Fix the bug that `joint_attention_kwargs` is not passed to the FLUX's transformer attention processors (#9517)

* Update transformer_flux.py

* Fixed local variable noise_pred_text referenced before assignment when using PAG with guidance scale and guidance rescale at the same time.

* Fixed style.

* Made returning text pred noise an argument.

* handle dora.

* print test

* debug

* fix

* fix-copies

* update logits

* add warning in the test.

* make is_dora check consistent.

* fix-copies

* Removed int8 to float32 conversion (`* 2.0 - 1.0`) from `train_transforms` as it caused image overexposure.

Added `_resize_for_rectangle_crop` function to enable video cropping functionality. The cropping mode can be configured via `video_reshape_mode`, supporting options: ['center', 'random', 'none'].

* The number 127.5 may experience precision loss during division operations.

* wandb request pil image Type

* Resizing bug

* del jupyter

* make style

* Update examples/cogvideo/README.md

* make style

---------

Co-authored-by: --unset <--unset>
Co-authored-by: Aryan <aryan@huggingface.co>

* Update distributed_inference.md to include `transformer.device_map`

* Update docs/source/en/training/distributed_inference.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

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

* Fix for use_safetensors parameters, allow use of parameter on loading submodels (#9576)

* refac: docstrings in import_utils.py

* Update import_utils.py

* fix vae dtype when accelerate config using --mixed_precision="fp16"

* Add param for upcast vae

add a note on the versions.

* fix startswith syntax in xlabs lora conversion

* Trigger CI

https://github.com/huggingface/diffusers/pull/9581#issuecomment-2395530360

* check size

* up

Co-authored-by: Aryan <aryan@huggingface.co>

* Support bfloat16 for Upsample2D

* Add test and use is_torch_version

* Resolve comments and add decorator

* Simplify require_torch_version_greater_equal decorator

* Run make style

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

* Add pag to controlnet inpainting pipeline


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

* support kohya flux loras that have tes.

* fix: retain memory utility.

* fix

* quality

* free_memory.

[train_instruct_pix2pix.py]Fix the LR schedulers when `num_train_epochs` is passed in a distributed training env (#9316)

Fixed pix2pix lr scheduler
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* remove conv cache from the layer and pass as arg instead

* make style

* yiyi's cleaner implementation
Co-Authored-By: YiYi Xu <yixu310@gmail.com>

* sayak's compiled implementation
Co-Authored-By: Sayak Paul <spsayakpaul@gmail.com>

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

* fix variant-idenitification.

* fix variant

* fix sharded variant checkpoint loading.

* Apply suggestions from code review

* fixes.

* more fixes.

* remove print.

* fixes

* fixes

* comments

* fixes

* apply suggestions.

* hub_utils.py

* fix test

* updates

* fixes

* fixes

* Apply suggestions from code review
Co-authored-by: YiYi Xu <yixu310@gmail.com>

* updates.

* removep patch file.

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

* clean up the serialization stuff.

* better

* add train flux-controlnet scripts in example.

* fix error

* fix subfolder error

* fix preprocess error

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

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

* fix readme

* fix note error

* add some Tutorial for deepspeed

* fix some Format Error

* add dataset_path example

* remove print, add guidance_scale CLI, readable apply

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

* update,push_to_hub,save_weight_dtype,static method,clear_objs_and_retain_memory,report_to=wandb

* add push to hub in readme

* apply weighting schemes

* add note

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

* make code style and quality

* fix some unnoticed error

* make code style and quality

* add example controlnet in readme

* add test controlnet

* rm Remove duplicate notes

* Fix formatting errors

* add new control image

* add model cpu offload

* update help for adafactor

* make quality & style

* make quality and style

* rename flux_controlnet_model_name_or_path

* fix back src/diffusers/pipelines/flux/pipeline_flux_controlnet.py

* fix dtype error by pre calculate text emb

* rm image save

* quality fix

* fix test

* fix tiny flux train error

* change report to to tensorboard

* fix save name error when test

* Fix shrinking errors

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

* make set_adapters() method more robust.

* remove patch

* better and concise code.

* Update src/diffusers/loaders/lora_base.py
Co-authored-by: YiYi Xu <yixu310@gmail.com>

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

* bugfix: precedence of operations should be slicing -> tiling

* fix typo

* fix another typo

* deprecate current implementation of tiled_encode and use new impl

* Update src/diffusers/models/autoencoders/autoencoder_kl.py
Co-authored-by: YiYi Xu <yixu310@gmail.com>

* Update src/diffusers/models/autoencoders/autoencoder_kl.py

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