AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-27 17:22:53 +03:00
Code Activity
1,752 Commits 740 Branches 94 Tags
d43972ae71ffbc94dee7045ecfe1e5f7c6ac329e
Commit Graph

1752 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Jorge C. Gomes
d43972ae71 Fixes prompt input checks in StableDiffusion img2img pipeline (#2206) 
* Fixes prompt input checks in img2img

Allows providing prompt_embeds instead of the prompt, which is not currently possible as the first check fails.
This becomes the same as the function found in 8267c78445/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py (L393)

* Continues the fix

This also needs to be fixed. Becomes consistent with 8267c78445/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py (L558)

I've now tested this implementation, and it produces the expected results.
 2023-02-07 09:10:24 +01:00
Fazzie-Maqianli
ffed2420c4 fix distributed init twice (#2252) 
fix colossalai dreambooth
 2023-02-07 08:55:39 +01:00
Pedro Cuenca
8178c840f2 Mention training problems with xFormers 0.0.16 (#2254) 2023-02-06 11:19:26 +01:00
nickkolok
3a0d3da66f Fix a typo: bfloa16 -> bfloat16 (#2243) 2023-02-06 09:14:08 +01:00
psychedelicious
22c1ba56c2 Fix k_dpm_2 & k_dpm_2_a on MPS (#2241) 
Needed to convert `timesteps` to `float32` a bit sooner.

Fixes #1537
 2023-02-05 23:45:15 +01:00
Pedro Cuenca
7386e7730c Show error when loading safety_checker from_flax (#2187) 
* Show error when loading safety_checker `from_flax`

* fix style
 2023-02-04 20:55:11 +01:00
Pedro Cuenca
154a7865fc [Flax DDPM] Make key optional so default pipelines don't fail (#2176) 
Make `key` optional so default pipelines don't fail.
 2023-02-04 20:45:20 +01:00
Robin Hutmacher
9baa29e9c0 Fix typo in StableDiffusionInpaintPipeline (#2197) 
* Fix typo in StableDiffusionInpaintPipeline

* Add embedded prompt handling

---------

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
 2023-02-03 19:03:15 +01:00
Jorge C. Gomes
58c416ab0c Fixes LoRAXFormersCrossAttnProcessor (#2207) 
Related to #2124 
The current implementation is throwing a shape mismatch error. Which makes sense, as this line is obviously missing, comparing to XFormersCrossAttnProcessor and LoRACrossAttnProcessor.

I don't have formal tests, but I compared `LoRACrossAttnProcessor` and `LoRAXFormersCrossAttnProcessor` ad-hoc, and they produce the same results with this fix.
 2023-02-03 18:10:48 +01:00
Isamu Isozaki
d46d78c584 Hotfix textual inv logging (#2183) 2023-02-03 18:08:46 +01:00
Patrick von Platen
05168e5d83 make style 2023-02-03 19:03:13 +02:00
Justin Merrell
948022e1e8 fix: flagged_images implementation (#1947) 
Flagged images would be set to the blank image instead of the original image that contained the NSF concept for optional viewing.

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
 2023-02-03 18:02:56 +01:00
Patrick von Platen
2f9a70aa85 [LoRA] Make sure validation works in multi GPU setup (#2172) 
* [LoRA] Make sure validation works in multi GPU setup

* more fixes

* up
 2023-02-03 16:50:10 +01:00
Sayak Paul
e43e206dc7 removes ~s in favor of full-fledged links. (#2229) 
remove ~ in favor of full-fledged links.
 2023-02-03 20:18:39 +05:30
Will Berman
99c39b4012 [nit] negative_prompt typo (#2227) 
* negative_prompt typo

* fix
 2023-02-03 14:05:50 +01:00
dymil
7547f9b475 Fix timestep dtype in legacy inpaint (#2120) 
* Fix timestep dtype in legacy inpaint

This matches the structure in the text2img, img2img, and inpaint ONNX pipelines

* Fix style in dtype patch
 2023-02-03 13:04:21 +01:00
Prathik Rao
a87e87fcbe refactor onnxruntime integration (#2042) 
* refactor onnxruntime integration

* fix requirements.txt bug

* make style

* add support for textual_inversion

* make style

* add readme

* cleanup README files

* 1/27/2023 update to training scripts

* make style

* 1/30 update to train_unconditional

* style with black-22.8.0

---------

Co-authored-by: Prathik Rao <prathikrao@microsoft.com>
Co-authored-by: anton- <anton@huggingface.co>
 2023-02-03 12:04:59 +01:00
Dudu Moshe
ecadcdefe1 [Bug] scheduling_ddpm: fix variance in the case of learned_range type. (#2090) 
scheduling_ddpm: fix variance in the case of learned_range type.

In the case of learned_range variance type, there are missing logs
and exponent comparing to the theory (see "Improved Denoising Diffusion
Probabilistic Models" section 3.1 equation 15:
https://arxiv.org/pdf/2102.09672.pdf).
 2023-02-03 09:42:42 +01:00
Pedro Cuenca
2bbd532990 Docs: short section on changing the scheduler in Flax (#2181) 
* Short doc on changing the scheduler in Flax.

* Apply fix from @patil-suraj

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

---------

Co-authored-by: Suraj Patil <surajp815@gmail.com>
 2023-02-02 18:52:21 +01:00
Adalberto
68ef0666e2 Create train_dreambooth_inpaint_lora.py (#2205) 
* Create train_dreambooth_inpaint_lora.py

* Update train_dreambooth_inpaint_lora.py

* Update train_dreambooth_inpaint_lora.py

* Update train_dreambooth_inpaint_lora.py

* Update train_dreambooth_inpaint_lora.py
 2023-02-02 13:15:15 +01:00
Kashif Rasul
7ac95703cd add CITATION.cff (#2211) 
add citation.cff
 2023-02-02 12:46:44 +01:00
Pedro Cuenca
3816c9ad9f Update xFormers docs (#2208) 
Update xFormers docs.
 2023-02-01 19:56:32 +01:00
Patrick von Platen
8267c78445 [Loading] Better error message on missing keys (#2198) 
* up

* finish
 2023-02-01 14:22:39 +01:00
Muyang Li
4fc7084875 Fix a dimension bug in Transform2d (#2144) 
The dimension does not match when `inner_dim` is not equal to `in_channels`.
 2023-02-01 10:11:45 +01:00
Sayak Paul
9213d81bd0 add: guide on kerascv conversion tool. (#2169) 
* add: guide on kerascv conversion tool.

* Apply suggestions from code review

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* address additional suggestions from review.

* change links to documentation-images.

* add separate links for training and inference goodies from diffusers.

* address Patrick's comments.

---------

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>
 2023-02-01 09:41:00 +01:00
Asad Memon
dd3cae3327 Pass LoRA rank to LoRALinearLayer (#2191) 2023-02-01 09:40:02 +01:00
Patrick von Platen
f73d0b6bec [Docs] remove license (#2188) 2023-01-31 22:11:32 +01:00
Patrick von Platen
d0d7ffffbd [Docs] Add components to docs (#2175) 2023-01-31 22:11:14 +01:00
Abhishek Varma
87cf88ed3d Use requests instead of wget in convert_from_ckpt.py (#2168) 
-- This commit adopts `requests` in place of `wget` to fetch config `.yaml`
   files as part of `load_pipeline_from_original_stable_diffusion_ckpt` API.
-- This was done because in Windows PowerShell one needs to explicitly ensure
   that `wget` binary is part of the PATH variable. If not present, this leads
   to the code not being able to download the `.yaml` config file.

Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
Co-authored-by: Abhishek Varma <abhishek@nod-labs.com>
 2023-01-31 14:35:45 +01:00
Patrick von Platen
60d915fbed make style 2023-01-31 11:46:48 +00:00
1lint
d1efefe15e [Breaking change] fix legacy inpaint noise and resize mask tensor (#2147) 
* fix legacy inpaint noise and resize mask tensor

* updated legacy inpaint pipe test expected_slice
 2023-01-31 12:44:35 +01:00
Sayak Paul
7d96b38b70 [examples] Fix CLI argument in the launch script command for text2image with LoRA (#2171) 
* Update README.md

* Update README.md
 2023-01-31 09:47:09 +01:00
Dudu Moshe
cedafb8600 [Bug]: fix DDPM scheduler arbitrary infer steps count. (#2076) 
scheduling_ddpm: fix evaluate with lower timesteps count than train.

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
 2023-01-31 09:13:26 +01:00
Patrick von Platen
69caa96472 fix slow test 2023-01-31 07:39:30 +00:00
hysts
da113364df Add instance prompt to model card of lora dreambooth example (#2112) 2023-01-31 08:14:25 +01:00
Pedro Cuenca
44f6bc81c7 Don't copy when unwrapping model (#2166) 
* Don't copy when unwrapping model.

Otherwise an exception is raised when using fp16.

* Remove unused import
 2023-01-30 20:18:20 +01:00
Pedro Cuenca
164b6e0532 Section on using LoRA alpha / scale (#2139) 
* Section on using LoRA alpha / scale.

* Accept suggestion

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

* Clarify on merge.

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2023-01-30 14:14:46 +01:00
Patrick von Platen
a6610db7a8 [Design philosopy] Create official doc (#2140) 
* finish more

* finish philosophy

* Apply suggestions from code review

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

---------

Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Will Berman <wlbberman@gmail.com>
 2023-01-30 09:27:37 +01:00
Pedro Cuenca
0b68101a13 [diffusers-cli] Fix typo in accelerate and transformers versions (#2154) 
Fix typo in accelerate and transformers versions.
 2023-01-30 09:04:45 +01:00
Ayan Das
125d783076 fix typo in EMAModel's load_state_dict() (#2151) 
Possible typo introduced in 7c82a16fc1
 2023-01-29 13:23:18 +01:00
Pedro Cuenca
fdf70cb54b Fix typo (#2138) 2023-01-27 20:08:56 +01:00
Nicolas Patry
20396e2bd2 Adding some safetensors docs. (#2122) 
* Tmp.

* Adding more docs.

* Doc style.

* Remove the argument `use_safetensors=True`.

* doc-builder
 2023-01-27 18:20:50 +01:00
Will Berman
2cf34e6db4 [from_pretrained] only load config one time (#2131) 2023-01-27 08:23:55 -08:00
Patrick von Platen
04ad948673 make style 2 - sorry 2023-01-27 16:54:40 +02:00
Patrick von Platen
97ef5e0665 make style 2023-01-27 16:52:04 +02:00
Patrick von Platen
31be42209d Don't call the Hub if local_files_only is specifiied (#2119) 
Don't call the Hub if
 2023-01-27 09:42:33 +02:00
RahulBhalley
43c5ac2be7 Typo fix: torwards -> towards (#2134) 2023-01-27 08:20:18 +01:00
Ji soo Kim
c750a82374 Fix typos in loaders.py (#2137) 
Fix typo in loaders.py
 2023-01-27 08:20:07 +01:00
Patrick von Platen
0c39f53cbb Allow lora from pipeline (#2129) 
* [LoRA] All to use in inference with pipeline

* [LoRA] allow cross attention kwargs passed to pipeline

* finish
 2023-01-27 08:19:46 +01:00
Will Berman
0a5948e7f4 remove redundant allow_patterns (#2130) 2023-01-26 13:22:28 -08:00