GitLab

* fuse attention mask

* lint

* use 0 beta when no attention mask re: @Birch-san

* fix PosixPath is not JSON serializable

* use PosixPath

* forgot elif like a dummy

* make scaling factor cnfig arg of vae

* fix

* make flake happy

* fix ldm

* fix upscaler

* qualirty

* Apply suggestions from code review
Co-authored-by: Anton Lozhkov <anton@huggingface.co>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* solve conflicts, addres some comments

* examples

* examples min version

* doc

* fix type

* typo

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

* remove duplicate line

* Apply suggestions from code review
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Anton Lozhkov <anton@huggingface.co>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Allow `UNet2DModel` to use arbitrary class embeddings.

We can currently use class conditioning in `UNet2DConditionModel`, but
not in `UNet2DModel`. However, `UNet2DConditionModel` requires text
conditioning too, which is unrelated to other types of conditioning.
This commit makes it possible for `UNet2DModel` to be conditioned on
entities other than timesteps. This is useful for training /
research purposes. We can currently train models to perform
unconditional image generation or text-to-image generation, but it's not
straightforward to train a model to perform class-conditioned image
generation, if text conditioning is not required.

We could potentiall use `UNet2DConditionModel` for class-conditioning
without text embeddings by using down/up blocks without
cross-conditioning. However:
- The mid block currently requires cross attention.
- We are required to provide `encoder_hidden_states` to `forward`.

* Style

* Align class conditioning, add docstring for `num_class_embeds`.

* Copy docstring to versatile_diffusion UNetFlatConditionModel

* [textual inversion] Allow validation images.

* Change key to `validation`

* Specify format instead of transposing.

As discussed with @sayakpaul.

* Style
Co-authored-by: isamu-isozaki <isamu.website@gmail.com>

check the dtype before preparing model

* [Bump version] 0.13

* Bump model up

* up

* make tests deterministic

* run slow tests

* prepare for testing

* finish

* refactor

* add print statements

* finish more

* correct some test failures

* more fixes

* set up to correct tests

* more corrections

* up

* fix more

* more prints

* add

* up

* up

* up

* uP

* uP

* more fixes

* uP

* up

* up

* up

* up

* fix more

* up

* up

* clean tests

* up

* up

* up

* more fixes

* Apply suggestions from code review
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* make

* correct

* finish

* finish
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* add text embeds to sd

* add text embeds to sd

* finish tests

* finish

* finish

* make style

* fix tests

* make style

* make style

* up

* better docs

* fix

* fix

* new try

* up

* up

* finish

* add: a doc on LoRA support in diffusers.

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

* apply PR suggestions.

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

* remove visually incoherent elements.
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>

* Add `lora` tag to the model tags

For lora training

* uP
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

update example for pix2pix

* [docs] [dreambooth] note random crop

documenting default random crop behavior

* allow passing op to xFormers attention

original code by @patil-suraj
huggingface/diffusers@ae0cc0b71f28c0f2c5c27026b18f1bea98b505f1

* correct style by `make style`

* add attention_op arg documents

* add usage example to docstring
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* add usage example to docstring
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* code style correction by `make style`

* Update docstring code to a valid python example
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* Update docstring code to a valid python example
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* style correction by `make style`

* Update code exmaple to fully functional
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* allow a local model directory to be used for merging

* moved checkpoint merge bugfix into main for testing

* possibly fix local variable "config_dict" referenced before assignment

* fix deprecation warning

* debugging...

* debugging

* allow safetensors

* safetensors try again

* fix syntax error

* further debugging

* fix logic error when checkpoint 2 is none

* more debugging...

* more debuging...

* more debugging...

* more debugging...

* debugging

* clean up status reporting

* skip the requires_safety_checker boolean

* moved checkpoint merge bugfix into main for testing

* possibly fix local variable "config_dict" referenced before assignment

* fix deprecation warning

* allow safetensors

* fix logic error when checkpoint 2 is none

* clean up status reporting

* undo hack to use private repo for community pipelines

* allow a local model directory to be used for merging

* allow safetensors

* clean up status reporting

* reformatted with black

* sort imported modules correctly

* Update examples/community/checkpoint_merger.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Update examples/community/checkpoint_merger.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Update examples/community/checkpoint_merger.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* fix import style error
Co-authored-by: Lincoln Stein <lstein@gmail.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

unwrap model on multi gpu

* [lora] Log images when using tensorboard.

* Specify image format instead of transposing.

As discussed with @sayakpaul.

* Style

* Fix resuming state when using gradient checkpointing.

Also, allow --resume_from_checkpoint to be used when the checkpoint does
not yet exist (a normal training run will be started).

* style

* [Paint by example] Fix paint by example

* fix more

* final fix

* Dreambooth: use `optimizer.zero_grad(set_to_none=True)` to reduce VRAM usage

* Allow the user to control `optimizer.zero_grad(set_to_none=True)` with --set_grads_to_none

* Update Dreambooth readme

* Fix link in readme

* Fix header size in readme

add --dataloader_num_workers argument

* Safetensors loading in "convert_diffusers_to_original_stable_diffusion"

Adds diffusers format saftetensors loading support

* Fix import sort order: convert_diffusers_to_original_stable_diffusion.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* example on fine-tuning with LoRA.

* apply make quality.

* fix: pipeline loading.

* Apply suggestions from code review
Co-authored-by: Suraj Patil <surajp815@gmail.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* apply suggestions for PR review.
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* apply make style and make quality.

* chore: remove mention of dreambooth from text2image.

* add: weight path and wandb run link.

* Apply suggestions from code review

* apply make style.

* make style
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* add community pipeline: StableUnCLIPPipeline

* reformt stable_unclip.py with isort and black

* Correct Pix2Pix example

- no advertisement of revision -> it'll be deprecated soon
- by default safety checker should be used

* Update docs/source/en/api/pipelines/stable_diffusion/pix2pix.mdx

* up

* [From pretrained] Don't download .safetensors files if safetensors is not available

* tests

* tests

* up

* use cpu generator in slow tests

* ifx get_inputs

* convert __main__ to a function call and call it

* add missing type hint

* make style check pass

* move loading to src/diffusers
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>