LLM/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-02-03 17:35:10 +08:00
Code Issues Packages Projects Releases Wiki Activity
6,204 Commits 746 Branches 94 Tags
ec6b2bcccb650ca94a0a1d98fb9e96941d5d188f
Commit Graph

6204 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Mikko Lauri
ec6b2bcccb Fix aiter availability check (#13059) 
Update import_utils.py
 2026-01-30 19:24:05 +05:30
Jared Wen
6a1904eb06 [bug fix] GLM-Image fit new get_image_features API (#13052) 
change get_image_features API

Signed-off-by: JaredforReal <w13431838023@gmail.com>
Co-authored-by: YiYi Xu <yixu310@gmail.com>
 2026-01-29 16:16:42 -10:00
Sayak Paul
f5b6b6625a [wan] fix wan 2.2 when either of the transformers isn't present. (#13055) 
fix wan 2.2 when either of the transformers isn't present.
 2026-01-29 08:45:24 -10:00
Olexandr88
1be2f7e8c5 docs: fix grammar in fp16_safetensors CLI warning (#13040) 
* docs: fix grammar in fp16_safetensors CLI warning

* Apply style fixes

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-29 21:33:09 +05:30
Sayak Paul
314cfddf3a [ci] uniform run times and wheels for pytorch cuda. (#13047) 
* uniform run times and wheels for pytorch cuda.

* 12.9

* change to 24.04.

* change to 24.04.
 2026-01-29 19:22:30 +05:30
Sayak Paul
e7de7d8449 [wan] fix layerwise upcasting tests on CPU (#13039) 
up
 2026-01-29 13:16:57 +05:30
Vinh H. Pham
a2ea45a5da LTX2 distilled checkpoint support (#12934) 
* add constants for distill sigmas values and allow ltx pipeline to pass in sigmas

* add time conditioning conversion and token packing for latents

* make style & quality

* remove prenorm

* add sigma param to ltx2 i2v

* fix copies and add pack latents to i2v

* Apply suggestions from code review

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* Infer latent dims if latents/audio_latents is supplied

* add note for predefined sigmas

* run make style and quality

* revert distill timesteps & set original_state_dict_repo_idd to default None

* add latent normalize

* add create noised state, delete last sigmas

* remove normalize step in latent upsample pipeline and move it to ltx2 pipeline

* add create noise latent to i2v pipeline

* fix copies

* parse none value in weight conversion script

* explicit shape handling

* Apply suggestions from code review

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* make style

* add two stage inference tests

* add ltx2 documentation

* update i2v expected_audio_slice

* Apply suggestions from code review

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* Apply suggestion from @dg845

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* Update ltx2.md to remove one-stage example

Removed one-stage generation example code and added comments for noise scale in two-stage generation.

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
Co-authored-by: Daniel Gu <dgu8957@gmail.com>
 2026-01-28 19:35:43 -08:00
Jayce
a58d0b9bec Fix Wan/WanI2V patchification (#13038) 
* Fix Wan/WanI2V patchification

* Apply style fixes

* Apply suggestions from code review

I agree with you for the idea of using `patch_size` instead. Thanks!😊

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* Fix logger warning

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
 2026-01-28 18:06:58 -08:00
David El Malih
0ab2124958 docs: improve docstring scheduling_dpm_cogvideox.py (#13044) 2026-01-28 10:40:08 -08:00
Vasiliy Kuznetsov
74a0f0b694 remove torchao autoquant from diffusers docs (#13048) 
Summary:

Context: https://github.com/pytorch/ao/issues/3739

Test Plan: CI, since this does not change any Python code
 2026-01-28 21:10:22 +05:30
Sayak Paul
2c669e8480 change to CUDA 12.9. (#13045) 
* change to CUDA 12.9.

* up

* change runtime base

* FROM
 2026-01-28 17:22:27 +05:30
Ita Zaporozhets
2ac39ba664 fast tok update (#13036) 
* v5 tok update

* ruff

* keep pre v5 slow code path

* Apply style fixes

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-28 17:13:04 +05:30
Sayak Paul
ef913010d4 [QwenImage] fix prompt isolation tests (#13042) 
* up

* up

* up

* fix
 2026-01-28 15:44:12 +05:30
YiYi Xu
53d8a1e310 [modular]support klein (#13002) 
* support klein

* style

* copies

* Apply suggestions from code review

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

* Update src/diffusers/modular_pipelines/flux2/encoders.py

* a few fix: unpack latents before decoder etc

* style

* remove guidannce to its own block

* style

* flux2-dev work in modular setting

* up

* up up

* add tests

---------

Co-authored-by: yiyi@huggingface.co <yiyi@ip-26-0-160-103.ec2.internal>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: Álvaro Somoza <asomoza@users.noreply.github.com>
 2026-01-27 15:43:14 -10:00
Kashif Rasul
d54669a73e [Qwen] avoid creating attention masks when there is no padding (#12987) 
* avoid creating attention masks when there is no padding

* make fix-copies

* torch compile tests

* set all ones mask to none

* fix positional encoding from becoming > 4096

* fix from review

* slice freqs_cis to match the input sequence length

* keep only attenton masking change

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2026-01-27 12:42:48 -10:00
Jared Wen
22ac6fae24 [GLM-Image] Add batch support for GlmImagePipeline (#13007) 
* init

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* change from right padding to left padding

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* try i2i batch

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix: revert i2i prior_token_image_ids to original 1D tensor format

* refactor KVCache for per prompt batching

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix KVCache

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix shape error

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* refactor pipeline

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix for left padding

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* insert seed to AR model

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* delete generator, use torch manual_seed

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* add batch processing unit tests for GlmImagePipeline

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* simplify normalize images method

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix grids_per_sample

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* fix t2i

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* delete comments, simplify condition statement

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* chage generate_prior_tokens outputs

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* simplify if logic

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* support user provided prior_token_ids directly

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* remove blank lines

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* align with transformers

Signed-off-by: JaredforReal <w13431838023@gmail.com>

* Apply style fixes

---------

Signed-off-by: JaredforReal <w13431838023@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-27 12:22:02 -10:00
Aditya Borate
71a865b742 Fix: Cosmos2.5 Video2World frame extraction and add default negative prompt (#13018) 
* fix: Extract last frames for conditioning in Cosmos Video2World

* Added default negative prompt

* Apply style fixes

* Added default negative prompt in cosmos2 text2image pipeline

---------

Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-27 12:20:44 -10:00
Sam Edwards
53279ef017 [From Single File] support from_single_file method for WanAnimateTransformer3DModel (#12691) 
* Add `WanAnimateTransformer3DModel` to `SINGLE_FILE_LOADABLE_CLASSES`

* Fixed dtype mismatch when loading a single file

* Fixed a bug that results in white noise for generation

* Update dtype check for time embedder - caused white noise output

* Improve code readability

* Optimize dtype handling

Removed unnecessary dtype conversions for timestep and weight.

* Apply style fixes

* Refactor time embedding dtype handling

Adjust time embedding type conversion for compatibility.

* Apply style fixes

* Modify comment for WanTimeTextImageEmbedding class

---------

Co-authored-by: Sam Edwards <sam.edwards1976@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-27 11:57:21 +05:30
Salman Chishti
d9959bd53b Upgrade GitHub Actions to latest versions (#12866) 
* Upgrade GitHub Actions to latest versions

Signed-off-by: Salman Muin Kayser Chishti <13schishti@gmail.com>

* fix: Correct GitHub Actions upgrade (fix branch refs and version formats)

* fix: Correct GitHub Actions upgrade (fix branch refs and version formats)

* fix: Correct GitHub Actions upgrade (fix branch refs and version formats)

---------

Signed-off-by: Salman Muin Kayser Chishti <13schishti@gmail.com>
 2026-01-27 11:52:50 +05:30
YiYi Xu
b1c77f67ac [modular] add auto_docstring & more doc related refactors (#12958) 
* up

* up up

* update outputs

* style

* add modular_auto_docstring!

* more auto docstring

* style

* up up up

* more more

* up

* address feedbacks

* add TODO in the description for empty docstring

* refactor based on dhruv's feedback: remove the class method

* add template method

* up

* up up up

* apply auto docstring

* make style

* rmove space in make docstring

* Apply suggestions from code review

* revert change in z

* fix

* Apply style fixes

* include auto-docstring check in the modular ci. (#13004)

* Run ruff format after auto docstring generation

* up

* upup

* upup

* style

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2026-01-26 17:12:13 -10:00
David El Malih
956bdcc3ea Flag Flax schedulers as deprecated (#13031) 
flag flax schedulers as deprecated
 2026-01-26 09:41:48 -08:00
Hameer Abbasi
2af7baa040 Remove *pooled_* mentions from Chroma inpaint (#13026) 
Remove `*pooled_*` mentions from Chroma as it has just one TE.
 2026-01-26 10:18:29 -03:00
David El Malih
a7cb14efbe Improve docstrings and type hints in scheduling_ddpm_parallel.py (#13027) 
* docs: improve docstring scheduling_ddpm_parallel.py

* Update scheduling_ddpm_parallel.py

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

---------

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
 2026-01-25 10:43:43 -08:00
David El Malih
e8e88ff2ce Improve docstrings and type hints in scheduling_ddpm_flax.py (#13024) 
docs: improve docstring scheduling_ddpm_flax.py
 2026-01-23 11:51:47 -08:00
David El Malih
6e24cd842c Improve docstrings and type hints in scheduling_ddim_parallel.py (#13023) 
* docs: improve docstring scheduling_ddim_parallel.py

* docs: improve docstring scheduling_ddim_parallel.py

* Update src/diffusers/schedulers/scheduling_ddim_parallel.py

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

* Update src/diffusers/schedulers/scheduling_ddim_parallel.py

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

* Update src/diffusers/schedulers/scheduling_ddim_parallel.py

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

* Update src/diffusers/schedulers/scheduling_ddim_parallel.py

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

* fix style

---------

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
 2026-01-23 10:00:32 -08:00
Garry Ling
981eb802c6 feat: add qkv projection fuse for longcat transformers (#13021) 
feat: add qkv fuse for longcat transformers

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2026-01-23 23:02:03 +05:30
jiqing-feng
1eb40c6dbd Resnet only use contiguous in training mode. (#12977) 
* fix contiguous

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* update tol

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* bigger tol

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* fix tests

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

* update tol

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

---------

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2026-01-23 18:40:10 +05:30
Sayak Paul
bff672f47f fix Dockerfiles for cuda and xformers. (#13022) 2026-01-23 16:45:14 +05:30
David El Malih
d4f97d1921 Improve docstrings and type hints in scheduling_ddim_inverse.py (#13020) 
docs: improve docstring scheduling_ddim_inverse.py
 2026-01-22 15:42:45 -08:00
David El Malih
1d32b19ad4 Improve docstrings and type hints in scheduling_ddim_flax.py (#13010) 
* docs: improve docstring scheduling_ddim_flax.py

* docs: improve docstring scheduling_ddim_flax.py

* docs: improve docstring scheduling_ddim_flax.py
 2026-01-22 09:11:14 -08:00
Garry Ling
699297f647 feat: accelerate longcat-image with regional compile (#13019) 2026-01-22 20:21:45 +05:30
Aryan V S
7a02fadad3 [scheduler] Support custom sigmas in UniPCMultistepScheduler (#12109) 
* update

* fix tests

* Apply suggestions from code review

* Revert default flow sigmas change so that tests relying on UniPC multistep still pass

* Remove custom timesteps for UniPC multistep set_timesteps

---------

Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Daniel Gu <dgu8957@gmail.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
 2026-01-21 17:18:59 -08:00
David El Malih
ec37629371 Improve docstrings and type hints in scheduling_ddim_cogvideox.py (#12992) 
docs: improve docstring scheduling_ddim_cogvideox.py
 2026-01-20 12:33:50 -08:00
Guillaume Besson
4b843c8430 Fix variable name in docstring for PeftAdapterMixin.set_adapters (#13003) 
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2026-01-20 15:43:59 +05:30
Gal Davidi
d7a1c31f4f Fibo edit pipeline (#12930) 
* Feature: Add BriaFiboEditPipeline to diffusers

* Introduced BriaFiboEditPipeline class with necessary backend requirements.
* Updated import structures in relevant modules to include BriaFiboEditPipeline.
* Ensured compatibility with existing pipelines and type checking.

* Feature: Introduce Bria Fibo Edit Pipeline

* Added BriaFiboEditPipeline class for structured JSON-native image editing.
* Created documentation for the new pipeline in bria_fibo_edit.md.
* Updated import structures to include the new pipeline and its components.
* Added unit tests for the BriaFiboEditPipeline to ensure functionality and correctness.

* Enhancement: Update Bria Fibo Edit Pipeline and Documentation

* Refined the Bria Fibo Edit model description for clarity and detail.
* Added usage instructions for model authentication and login.
* Implemented mask handling functions in the BriaFiboEditPipeline for improved image editing capabilities.
* Updated unit tests to cover new mask functionalities and ensure input validation.
* Adjusted example code in documentation to reflect changes in the pipeline's usage.

* Update Bria Fibo Edit documentation with corrected Hugging Face page link

* add dreambooth training script

* style and quality

* Delete temp.py

* Enhancement: Improve JSON caption validation in DreamBoothDataset

* Updated the clean_json_caption function to handle both string and dictionary inputs for captions.
* Added error handling to raise a ValueError for invalid caption types, ensuring better input validation.

* Add datasets dependency to requirements_fibo_edit.txt

* Add bria_fibo_edit to docs table of contents

* Fix dummy objects ordering

* Fix BriaFiboEditPipeline to use passed generator parameter

The pipeline was ignoring the generator parameter and only using
the seed parameter. This caused non-deterministic outputs in tests
that pass a seeded generator.

* Remove fibo_edit training script and related files

---------

Co-authored-by: kfirbria <kfir@bria.ai>
 2026-01-19 22:09:53 +05:30
Sayak Paul
29b15f41c7 [chore] make style to push new changes. (#12998) 
make style to push new changes.
 2026-01-19 16:02:13 +05:30
sayakpaul
75edff93a0 Revert "make style && make quality" 
This reverts commit 76f51a5e92.
 2026-01-19 15:35:20 +05:30
sayakpaul
76f51a5e92 make style && make quality 2026-01-19 15:34:29 +05:30
David El Malih
3996788b60 [Docs] Replace root CONTRIBUTING.md with symlink to source docs (#12986) 
Chore: Replace CONTRIBUTING.md with a symlink to documentation
 2026-01-16 12:36:50 -08:00
David El Malih
9fedfe58b7 Improve docstrings and type hints in scheduling_cosine_dpmsolver_multistep.py (#12936) 
* docs: improve docstring scheduling_cosine_dpmsolver_multistep.py

* Update src/diffusers/schedulers/scheduling_cosine_dpmsolver_multistep.py

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

* Update src/diffusers/schedulers/scheduling_cosine_dpmsolver_multistep.py

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

* fix

---------

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
 2026-01-16 12:23:49 -08:00
Sayak Paul
ebf891a254 [core] gracefully error out when attn-backend x cp combo isn't supported. (#12832) 
* gracefully error out when attn-backend x cp combo isn't supported.

* Revert "gracefully error out when attn-backend x cp combo isn't supported."

This reverts commit c8abb5d7c0.

* gracefully error out when attn-backend x cp combo isn't supported.

* up

* address PR feedback.

* up

* Update src/diffusers/models/modeling_utils.py

Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>

* dot.

---------

Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>
 2026-01-16 21:29:42 +05:30
dg845
8af8e86bc7 LTX 2 Single File Support (#12983) 
* LTX 2 transformer single file support

* LTX 2 video VAE single file support

* LTX 2 audio VAE single file support

* Make it easier to distinguish LTX 1 and 2 models
 2026-01-15 22:46:42 -08:00
Sayak Paul
74654df203 add klein docs. (#12984) 2026-01-16 10:12:42 +05:30
YiYi Xu
f112eab97e [modular] fix a bug in mellon param & improve docstrings (#12980) 
* update mellonparams docstring to incude the acutal param definition render in mellon

* style

---------

Co-authored-by: yiyi@huggingface.co <yiyi@ip-26-0-160-103.ec2.internal>
 2026-01-15 10:42:42 -10:00
YiYi Xu
61f175660a Flux2 klein (#12982) 
* flux2-klein

* Apply suggestions from code review

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

* Klein tests (#2)

* tests

* up

* tests

* up

* support step-distilled

* Apply suggestions from code review

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* Apply suggestions from code review

Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>

* doc string etc

* style

* more

* copies

* klein lora training scripts (#3)

* initial commit

* initial commit

* remove remote text encoder

* initial commit

* initial commit

* initial commit

* revert

* img2img fix

* text encoder + tokenizer

* text encoder + tokenizer

* update readme

* guidance

* guidance

* guidance

* test

* test

* revert changes not needed for the non klein model

* Update examples/dreambooth/train_dreambooth_lora_flux2_klein.py

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

* fix guidance

* fix validation

* fix validation

* fix validation

* fix path

* space

---------

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

* style

* Update src/diffusers/pipelines/flux2/pipeline_flux2_klein.py

* Apply style fixes

* auto pipeline

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2026-01-15 09:10:54 -10:00
DefTruth
7f43cb1d79 fix Qwen-Image series context parallel (#12970) 
* fix qwen-image cp

* relax attn_mask limit for cp

* CP plan compatible with zero_cond_t

* move modulate_index plan to top level
 2026-01-15 15:40:24 +05:30
Hameer Abbasi
5efb81fa71 Add ChromaInpaintPipeline (#12848) 
* Add `ChromaInpaintPipeline`

* Set `attention_mask` to `dtype=torch.bool` for `ChromaInpaintPipeline`.

* Revert `.gitignore`.
 2026-01-15 12:58:50 +05:30
Yahweasel
b351be2379 LongCat Image pipeline: Allow offloading/quantization of text_encoder component (#12963) 
* Don't attempt to move the text_encoder. Just move the generated_ids.

* The inputs to the text_encoder should be on its device
 2026-01-14 21:10:57 -10:00
YiYi Xu
d8f4dd295f [Modular] mellon utils (#12978) 
* up

* style

---------

Co-authored-by: yiyi@huggingface.co <yiyi@ip-26-0-160-103.ec2.internal>
 2026-01-14 19:03:41 -10:00
hlky
1ecfbfe12b disable_mmap in pipeline from_pretrained (#12854) 
* update

* `disable_mmap` in `from_pretrained`

---------

Co-authored-by: DN6 <dhruv.nair@gmail.com>
 2026-01-14 21:29:36 +05:30