1
0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-27 17:22:53 +03:00
Commit Graph

974 Commits

Author SHA1 Message Date
patil-suraj
661ca4677e do two forward passes for prior preservation 2022-09-26 22:41:47 +02:00
patil-suraj
87bc75231a add option for 8bit adam 2022-09-26 17:02:02 +02:00
patil-suraj
16ecc089e3 add grad ckpt 2022-09-26 15:10:54 +02:00
patil-suraj
2894a92f92 Merge branch 'main' of https://github.com/huggingface/diffusers into dreambooth-example 2022-09-26 15:08:12 +02:00
patil-suraj
c66cf4dc1a concat batch in collate fn 2022-09-26 15:02:13 +02:00
patil-suraj
ef01331146 fix collate fn 2022-09-26 14:45:37 +02:00
patil-suraj
8e92d69ca2 fix dataset 2022-09-26 14:44:23 +02:00
patil-suraj
f4a91a640c fix key name 2022-09-26 14:37:32 +02:00
patil-suraj
a3d604e99a fix collate fn 2022-09-26 14:36:39 +02:00
patil-suraj
d6c88f4ef6 fix collate_fn 2022-09-26 14:35:40 +02:00
patil-suraj
392fbf3c1d fix collate fun 2022-09-26 14:33:02 +02:00
patil-suraj
6f99f29f71 improve dataset 2022-09-26 14:31:30 +02:00
patil-suraj
eafc0002d4 fix saving 2022-09-26 14:09:03 +02:00
patil-suraj
509e4e3f96 actuallu fix saving 2022-09-26 14:06:18 +02:00
patil-suraj
f1c3c8e5a4 fix saving 2022-09-26 14:01:54 +02:00
patil-suraj
627cc49447 improve sampling more 2022-09-26 13:58:07 +02:00
patil-suraj
1acc6786e5 fix autocast 2022-09-26 13:49:21 +02:00
patil-suraj
195cd463a8 improgve sampling 2022-09-26 13:45:54 +02:00
Suraj Patil
2eeabe7adc Apply suggestions from code review 2022-09-26 10:34:27 +02:00
Anton Lozhkov
57b70c599c [CI] Fix onnxruntime installation order (#633) 2022-09-24 18:32:03 +02:00
Grigory Sizov
35e9209601 Fix formula for noise levels in Karras scheduler and tests (#627)
fix formula for noise levels in karras scheduler and tests
2022-09-24 18:24:08 +02:00
Ryan Russell
d0aa899f0e docs: src/diffusers readability improvements (#629)
* docs: `src/diffusers` readability improvements

Signed-off-by: Ryan Russell <git@ryanrussell.org>

* docs: `make style` lint

Signed-off-by: Ryan Russell <git@ryanrussell.org>

Signed-off-by: Ryan Russell <git@ryanrussell.org>
2022-09-24 16:21:28 +02:00
Pedro Cuenca
1e152030bd Fix breaking error: "ort is not defined" (#626)
Fix "ort is not defined" issue.
2022-09-23 17:02:03 +02:00
cloudhan
8211b62227 Allow passing session_options for ORT backend (#620) 2022-09-23 15:28:31 +02:00
Ryan Russell
ce31f83d8c refactor: pipelines readability improvements (#622)
* refactor: pipelines readability improvements

Signed-off-by: Ryan Russell <git@ryanrussell.org>

* docs: remove todo comment from flax pipeline

Signed-off-by: Ryan Russell <git@ryanrussell.org>

Signed-off-by: Ryan Russell <git@ryanrussell.org>
2022-09-23 15:02:12 +02:00
Abdullah Alfaraj
b00382e2a7 fix docs: change sample to images (#613)
the result of running the pipeline is stored in StableDiffusionPipelineOutput.images
2022-09-23 14:27:29 +02:00
Younes Belkada
8b0be93596 Flax documentation (#589)
* documenting `attention_flax.py` file

* documenting `embeddings_flax.py`

* documenting `unet_blocks_flax.py`

* Add new objs to doc page

* document `vae_flax.py`

* Apply suggestions from code review

* modify `unet_2d_condition_flax.py`

* make style

* Apply suggestions from code review

* make style

* Apply suggestions from code review

* fix indent

* fix typo

* fix indent unet

* Update src/diffusers/models/vae_flax.py

* Apply suggestions from code review

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

Co-authored-by: Mishig Davaadorj <dmishig@gmail.com>
Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
2022-09-23 13:24:16 +02:00
Ryan Russell
df80ccf7de docs: .md readability fixups (#619)
Signed-off-by: Ryan Russell <git@ryanrussell.org>
2022-09-23 12:02:27 +02:00
Zhenhuan Liu
faffe23627 Update for multi-gpu trianing. 2022-09-23 04:31:15 -04:00
Zhenhuan Liu
5bb534b0a5 Reformatting code with black. 2022-09-22 10:23:53 -07:00
Zhenhuan Liu
88ab347d3c Update readme and default hyperparameters. 2022-09-22 10:19:45 -07:00
Zhenhuan Liu
51340f9921 Fix bugs. 2022-09-22 10:19:03 -07:00
Jonathan Whitaker
91db81894b Adding pred_original_sample to SchedulerOutput for some samplers (#614)
* Adding pred_original_sample to SchedulerOutput of DDPMScheduler, DDIMScheduler, LMSDiscreteScheduler, KarrasVeScheduler step methods so we can access the predicted denoised outputs

* Gave DDPMScheduler, DDIMScheduler and LMSDiscreteScheduler their own output dataclasses so the default SchedulerOutput in scheduling_utils does not need pred_original_sample as an optional extra

* Reordered library imports to follow standard

* didnt get import order quite right apparently

* Forgot to change name of LMSDiscreteSchedulerOutput

* Aha, needed some extra libs for make style to fully work
2022-09-22 18:53:40 +02:00
Ryan Russell
f149d037de docs: fix stochastic_karras_ve ref (#618)
Signed-off-by: Ryan Russell <git@ryanrussell.org>

Signed-off-by: Ryan Russell <git@ryanrussell.org>
2022-09-22 18:36:29 +02:00
Suraj Patil
e7120bae95 [UNet2DConditionModel] add gradient checkpointing (#461)
* add grad ckpt to downsample blocks

* make it work

* don't pass gradient_checkpointing to upsample block

* add tests for UNet2DConditionModel

* add test_gradient_checkpointing

* add gradient_checkpointing for up and down blocks

* add functions to enable and disable grad ckpt

* remove the forward argument

* better naming

* make supports_gradient_checkpointing private
2022-09-22 15:36:47 +02:00
Mishig Davaadorj
534512bedb [flax] 'dtype' should not be part of self._internal_dict (#609) 2022-09-22 11:46:31 +02:00
Mishig Davaadorj
4b8880a306 Make flax from_pretrained work with local subfolder (#608) 2022-09-22 11:44:22 +02:00
Anton Lozhkov
dd350c8afe Handle the PIL.Image.Resampling deprecation (#588)
* Handle the PIL.Image.Resampling deprecation

* style
2022-09-22 00:02:14 +02:00
Ryan Russell
80183ca58b docs: fix Berkeley ref (#611)
Signed-off-by: Ryan Russell <git@ryanrussell.org>

Signed-off-by: Ryan Russell <git@ryanrussell.org>
2022-09-21 22:55:32 +02:00
Anton Lozhkov
6bd005ebbe [ONNX] Collate the external weights, speed up loading from the hub (#610) 2022-09-21 22:26:30 +02:00
Pedro Cuenca
a9fdb3de9e Return Flax scheduler state (#601)
* Optionally return state in from_config.

Useful for Flax schedulers.

* has_state is now a property, make check more strict.

I don't check the class is `SchedulerMixin` to prevent circular
dependencies. It should be enough that the class name starts with "Flax"
the object declares it "has_state" and the "create_state" exists too.

* Use state in pipeline from_pretrained.

* Make style
2022-09-21 22:25:27 +02:00
Anton Lozhkov
e72f1a8a71 Add torchvision to training deps (#607) 2022-09-21 13:54:32 +02:00
Anton Lozhkov
4f1c989ffb Add smoke tests for the training examples (#585)
* Add smoke tests for the training examples

* upd

* use a dummy dataset

* mark as slow

* cleanup

* Update test cases

* naming
2022-09-21 13:36:59 +02:00
Younes Belkada
3fc8ef7297 Replace dropout_prob by dropout in vae (#595)
replace `dropout_prob` by `dropout` in `vae`
2022-09-21 11:43:28 +02:00
Mishig Davaadorj
8685699392 Mv weights name consts to diffusers.utils (#605) 2022-09-21 11:30:14 +02:00
Mishig Davaadorj
f810060006 Fix flax from_pretrained pytorch weight check (#603) 2022-09-21 11:17:15 +02:00
Pedro Cuenca
fb2fbab10b Allow dtype to be specified in Flax pipeline (#600)
* Fix typo in docstring.

* Allow dtype to be overridden on model load.

This may be a temporary solution until #567 is addressed.

* Create latents in float32

The denoising loop always computes the next step in float32, so this
would fail when using `bfloat16`.
2022-09-21 10:57:01 +02:00
Pedro Cuenca
fb03aad8b4 Fix params replication when using the dummy checker (#602)
Fix params replication when sing the dummy checker.
2022-09-21 09:38:10 +02:00
Patrick von Platen
2345481c0e [Flax] Fix unet and ddim scheduler (#594)
* [Flax] Fix unet and ddim scheduler

* correct

* finish
2022-09-20 23:29:09 +02:00
Mishig Davaadorj
d934d3d795 FlaxDiffusionPipeline & FlaxStableDiffusionPipeline (#559)
* WIP: flax FlaxDiffusionPipeline & FlaxStableDiffusionPipeline

* todo comment

* Fix imports

* Fix imports

* add dummies

* Fix empty init

* make pipeline work

* up

* Use Flax schedulers (typing, docstring)

* Wrap model imports inside availability checks.

* more updates

* make sure flax is not broken

* make style

* more fixes

* up

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Pedro Cuenca <pedro@latenitesoft.com>
2022-09-20 21:28:07 +02:00