GitLab

* Support training SD V2 with Flax

Mostly involves supporting a v_prediction scheduler.

The implementation in #1777 doesn't take into account a recent refactor of `scheduling_utils_flax`, so this should be used instead.

* Add to other top-level files.

* [Deterministic torch randn] Allow tensors to be generated on CPU

* fix more

* up

* fix more

* up

* Update src/diffusers/utils/torch_utils.py
Co-authored-by: Anton Lozhkov <anton@huggingface.co>

* Apply suggestions from code review

* up

* up

* Apply suggestions from code review
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Anton Lozhkov <anton@huggingface.co>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* Add condtional generation

* add fast test for conditional audio generation

* add add_noise to scheduling-sde-ve

* run Black formater

Add accelerate and xformers to diffusers-cli env

* Add UnCLIPImageVariationPipeline to dummy imports

* style

* add a doc page for each pipeline under api/pipelines/stable_diffusion
* add pipeline examples to docstrings
* updated stable_diffusion_2 page
* updated default markdown syntax to list methods based on https://github.com/huggingface/diffusers/pull/1870


* add function decorator
Co-authored-by: yiyixuxu <yixu@Yis-MacBook-Pro.lan>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

Numpy 1.24 had removed the "float" scalar alias as it was depricated in v1.20.
https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations
https://numpy.org/devdocs/release/1.24.0-notes.html#expired-deprecations

* [Attention] Finish refactor attention file

* correct more

* fix

* more fixes

* correct

* up

[Unclip] Make sure text_embeddings & image_embeddings can directly be passed to enable interpolation tasks. (#1858)

* [Unclip] Make sure latents can be reused

* allow one to directly pass embeddings

* up

* make unclip for text work

* finish allowing to pass embeddings

* correct more

* make style

* move files a bit

* more refactors

* fix more

* more fixes

* fix more onnx

* make style

* upload

* fix

* up

* fix more

* up again

* up

* small fix

* Update src/diffusers/__init__.py
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* correct
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* Flax: Add components function

* Flax: Fix img2img and align with other pipeline

* Flax: Fix PRNGKey type

* Refactor strength to start_timestep

* Fix preprocess images

* Fix processed_images dimen

* latents.shape -> latents_shape

* Fix typo

* Remove "static" comment

* Remove unnecessary optional types in _generate

* Apply doc-builder code style.
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* [Dtype] Align automatic dtype

* up

* up

* fix

* re-add accelerate

fix versatile

* unCLIP image variation

* remove prior comment re: @pcuenca

* stable diffusion -> unCLIP re: @pcuenca

* add copy froms re: @patil-suraj

* Width was typod as weight

* Run Black

* Make safety_checker optional in more pipelines.

* Remove inappropriate comment in inpaint pipeline.

* InPaint Test: set feature_extractor to None.

* Remove import

* img2img test: set feature_extractor to None.

* inpaint sd2 test: set feature_extractor to None.
Co-authored-by: Suraj Patil <surajp815@gmail.com>

Fix inconsistencies between code and comments in the function 'preprocess'

Don't initialize Jax on startup.

* first proposal

* rename

* up

* Apply suggestions from code review

* better

* up

* finish

* up

* rename

* correct versatile

* up

* up

* up

* up

* fix

* Apply suggestions from code review

* make style

* Apply suggestions from code review
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* add error message
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* add flax img2img pipeline

* update pipeline

* black format file

* remove argg from get_timesteps

* update get_timesteps

* fix bug: make use of timesteps for for_loop

* black file

* black, isort, flake8

* update docstring

* update readme

* update flax img2img readme

* update sd pipeline init

* Update src/diffusers/pipelines/stable_diffusion/pipeline_flax_stable_diffusion_img2img.py
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* Update src/diffusers/pipelines/stable_diffusion/pipeline_flax_stable_diffusion_img2img.py
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* update inits

* revert change

* update var name to image, typo

* update readme

* return new t_start instead of modified timestep

* black format

* isort files

* update docs

* fix-copies

* update prng_seed typing
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* use repeat_interleave

* fix repeat

* Trigger Build

* don't install accelerate from main

* install released accelrate for mps test

* Remove additional accelerate installation from main.
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* allow model download when no internet

* up

* make style

* [Flax] Stateless schedulers, fixes and refactors

* Remove scheduling_common_flax and some renames

* Update src/diffusers/schedulers/scheduling_pndm_flax.py
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* expose polynomial:power and cosine_with_restarts:num_cycles using get_scheduler func, add it to train_dreambooth.py

* fix formatting

* fix style

* Update src/diffusers/optimization.py
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* only check for xformers when xformers are enabled

* only test for xformers when enabling them

* Fail if there are less images than the effective batch size.

* Remove lr-scheduler arg as it's currently ignored.

* Make guidance_scale work for batch_size > 1.

* Transformers version req for UnCLIP

* add to the list

* Add CPU offloading to UnCLIP

* use fp32 for testing the offload

duplicate maks for num_images_per_prompt