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
5,291 Commits 740 Branches 94 Tags
78c2fdc52ef4dd79fb710605a4876894cafdb492
Commit Graph

5291 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
hlky
78c2fdc52e SchedulerMixin from_pretrained and ConfigMixin Self type annotation (#11192) 2025-04-02 08:24:02 -10:00
hlky
54dac3a87c Fix enable_sequential_cpu_offload in CogView4Pipeline (#11195) 
* Fix enable_sequential_cpu_offload in CogView4Pipeline

* make fix-copies
 2025-04-02 16:51:23 +01:00
hlky
e5c6027ef8 [docs] torch_dtype map (#11194) 2025-04-02 12:46:28 +01:00
hlky
da857bebb6 Revert save_model in ModelMixin save_pretrained and use safe_serialization=False in test (#11196) 2025-04-02 12:45:36 +01:00
Fanli Lin
52b460feb9 [tests] HunyuanDiTControlNetPipeline inference precision issue on XPU (#11197) 
* add xpu part

* fix more cases

* remove some cases

* no canny

* format fix
 2025-04-02 12:45:02 +01:00
hlky
d8c617ccb0 allow models to run with a user-provided dtype map instead of a single dtype (#10301) 
* allow models to run with a user-provided dtype map instead of a single dtype

* make style

* Add warning, change `_` to `default`

* make style

* add test

* handle shared tensors

* remove warning

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2025-04-02 09:05:46 +01:00
Bruno Magalhaes
fe2b397426 remove unnecessary call to F.pad (#10620) 
* rewrite memory count without implicitly using dimensions by @ic-synth

* replace F.pad by built-in padding in Conv3D

* in-place sums to reduce memory allocations

* fixed trailing whitespace

* file reformatted

* in-place sums

* simpler in-place expressions

* removed in-place sum, may affect backward propagation logic

* removed in-place sum, may affect backward propagation logic

* removed in-place sum, may affect backward propagation logic

* reverted change
 2025-04-02 08:19:51 +01:00
Eliseu Silva
be0b7f55cc fix: for checking mandatory and optional pipeline components (#11189) 
fix: optional componentes verification on load
 2025-04-02 08:07:24 +01:00
jiqing-feng
4d5a96e40a fix autocast (#11190) 
Signed-off-by: jiqing-feng <jiqing.feng@intel.com>
 2025-04-02 07:26:27 +01:00
Yao Matrix
a7f07c1ef5 map BACKEND_RESET_MAX_MEMORY_ALLOCATED to reset_peak_memory_stats on XPU (#11191) 
Signed-off-by: YAO Matrix <matrix.yao@intel.com>
 2025-04-02 07:25:48 +01:00
Dhruv Nair
df1d7b01f1 [WIP] Add Wan Video2Video (#11053) 
* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update

* update
 2025-04-01 17:22:11 +05:30
Fanli Lin
5a6edac087 [tests] no hard-coded cuda (#11186) 
no cuda only
 2025-04-01 12:14:31 +01:00
kakukakujirori
e8fc8b1f81 Bug fix in LTXImageToVideoPipeline.prepare_latents() when latents is already set (#10918) 
* Bug fix in ltx

* Assume packed latents.

---------

Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>
Co-authored-by: YiYi Xu <yixu310@gmail.com>
 2025-03-31 12:15:43 -10:00
hlky
d6f4774c1c Add latents_mean and latents_std to SDXLLongPromptWeightingPipeline (#11034) 2025-03-31 11:32:29 -10:00
Mark
eb50defff2 [Docs] Fix environment variables in installation.md (#11179) 2025-03-31 09:15:25 -07:00
Aryan
2c59af7222 Raise warning and round down if Wan num_frames is not 4k + 1 (#11167) 
* update

* raise warning and round to nearest multiple of scale factor
 2025-03-31 13:33:28 +05:30
hlky
75d7e5cc45 Fix LatteTransformer3DModel dtype mismatch with enable_temporal_attentions (#11139) 2025-03-29 15:52:56 +01:00
Dhruv Nair
617c208bb4 [Docs] Update Wan Docs with memory optimizations (#11089) 
* update

* update
 2025-03-28 19:05:56 +05:30
hlky
5d970a4aa9 WanI2V encode_image (#11164) 
* WanI2V encode_image
 2025-03-28 18:05:34 +05:30
kentdan3msu
de6a88c2d7 Set self._hf_peft_config_loaded to True when LoRA is loaded using load_lora_adapter in PeftAdapterMixin class (#11155) 
set self._hf_peft_config_loaded to True on successful lora load

Sets the `_hf_peft_config_loaded` flag if a LoRA is successfully loaded in `load_lora_adapter`. Fixes bug huggingface/diffusers/issues/11148

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2025-03-26 18:31:18 +01:00
Dhruv Nair
7dc52ea769 [Quantization] dtype fix for GGUF + fix BnB tests (#11159) 
* update

* update

* update

* update
 2025-03-26 22:22:16 +05:30
Junsong Chen
739d6ec731 add a timestep scale for sana-sprint teacher model (#11150) 2025-03-25 08:47:39 -10:00
Aryan
1ddf3f3a19 Improve information about group offloading and layerwise casting (#11101) 
* update

* Update docs/source/en/optimization/memory.md

* Apply suggestions from code review

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

* apply review suggestions

* update

---------

Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com>
 2025-03-24 23:25:59 +05:30
Jun Yeop Na
7aac77affa [doc] Fix Korean Controlnet Train doc (#11141) 
* remove typo from korean controlnet train doc

* removed more paragraphs to remain in sync with the english document
 2025-03-24 09:38:21 -07:00
Aryan
8907a70a36 New HunyuanVideo-I2V (#11066) 
* update

* update

* update

* add tests

* update docs

* raise value error

* warning for true cfg and guidance scale

* fix test
 2025-03-24 21:18:40 +05:30
Junsong Chen
5dbe4f5de6 [fix SANA-Sprint] (#11142) 
* fix bug in sana conversion script;

* add more model paths;

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2025-03-23 23:38:14 -10:00
Yuxuan Zhang
1d37f42055 Modify the implementation of retrieve_timesteps in CogView4-Control. (#11125) 
* 1

* change to channel 1

* cogview4 control training

* add CacheMixin

* 1

* remove initial_input_channels change for val

* 1

* update

* use 3.5

* new loss

* 1

* use imagetoken

* for megatron convert

* 1

* train con and uc

* 2

* remove guidance_scale

* Update pipeline_cogview4_control.py

* fix

* use cogview4 pipeline with timestep

* update shift_factor

* remove the uncond

* add max length

* change convert and use GLMModel instead of GLMForCasualLM

* fix

* [cogview4] Add attention mask support to transformer model

* [fix] Add attention mask for padded token

* update

* remove padding type

* Update train_control_cogview4.py

* resolve conflicts with #10981

* add control convert

* use control format

* fix

* add missing import

* update with cogview4 formate

* make style

* Update pipeline_cogview4_control.py

* Update pipeline_cogview4_control.py

* remove

* Update pipeline_cogview4_control.py

* put back

* Apply style fixes

---------

Co-authored-by: OleehyO <leehy0357@gmail.com>
Co-authored-by: yiyixuxu <yixu310@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2025-03-23 21:17:14 +05:30
Tolga Cangöz
0213179ba8 Update README and example code for AnyText usage (#11028) 
* [Documentation] Update README and example code with additional usage instructions for AnyText

* [Documentation] Update README for AnyTextPipeline and improve logging in code

* Remove wget command for font file from example docstring in anytext.py
 2025-03-23 21:15:57 +05:30
hlky
a7d53a5939 Don't override torch_dtype and don't use when quantization_config is set (#11039) 
* Don't use `torch_dtype` when `quantization_config` is set

* up

* djkajka

* Apply suggestions from code review

---------

Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
 2025-03-21 21:58:38 +05:30
YiYi Xu
8a63aa5e4f add sana-sprint (#11074) 
* add sana-sprint




---------

Co-authored-by: Junsong Chen <cjs1020440147@icloud.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: Aryan <aryan@huggingface.co>
 2025-03-21 06:21:18 -10:00
Aryan
844221ae4e [core] FasterCache (#10163) 
* init

* update

* update

* update

* make style

* update

* fix

* make it work with guidance distilled models

* update

* make fix-copies

* add tests

* update

* apply_faster_cache -> apply_fastercache

* fix

* reorder

* update

* refactor

* update docs

* add fastercache to CacheMixin

* update tests

* Apply suggestions from code review

* make style

* try to fix partial import error

* Apply style fixes

* raise warning

* update

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
 2025-03-21 09:35:04 +05:30
CyberVy
9b2c0a7dbe fix _callback_tensor_inputs of sd controlnet inpaint pipeline missing some elements (#11073) 
* Update pipeline_controlnet_inpaint.py

* Apply style fixes
 2025-03-20 23:56:12 -03:00
Parag Ekbote
f424b1b062 Notebooks for Community Scripts-8 (#11128) 
Add 4 Notebooks and update the missing links for the
example README.
 2025-03-20 12:24:46 -07:00
YiYi Xu
e9fda3924f remove F.rms_norm for now (#11126) 
up
 2025-03-20 07:55:01 -10:00
Dhruv Nair
2c1ed50fc5 Provide option to reduce CPU RAM usage in Group Offload (#11106) 
* update

* update

* clean up
 2025-03-20 17:01:09 +05:30
Fanli Lin
15ad97f782 [tests] make cuda only tests device-agnostic (#11058) 
* enable bnb on xpu

* add 2 more cases

* add missing change

* add missing change

* add one more

* enable cuda only tests on xpu

* enable big gpu cases
 2025-03-20 10:12:35 +00:00
hlky
9f2d5c9ee9 Flux with Remote Encode (#11091) 
* Flux img2img remote encode

* Flux inpaint

* -copied from
 2025-03-20 09:44:08 +00:00
Junsong Chen
dc62e6931e [fix bug] PixArt inference_steps=1 (#11079) 
* fix bug when pixart-dmd inference with `num_inference_steps=1`

* use return_dict=False and return [1] element for 1-step pixart model, which works for both lcm and dmd
 2025-03-20 07:44:30 +00:00
Fanli Lin
56f740051d [tests] enable bnb tests on xpu (#11001) 
* enable bnb on xpu

* add 2 more cases

* add missing change

* add missing change

* add one more
 2025-03-19 16:33:11 +00:00
Linoy Tsaban
a34d97cef0 [Wan LoRAs] make T2V LoRAs compatible with Wan I2V (#11107) 
* @hlky t2v->i2v

* Apply style fixes

* try with ones to not nullify layers

* fix method name

* revert to zeros

* add check to state_dict keys

* add comment

* copies fix

* Revert "copies fix"

This reverts commit 051f534d18.

* remove copied from

* Update src/diffusers/loaders/lora_pipeline.py

Co-authored-by: hlky <hlky@hlky.ac>

* Update src/diffusers/loaders/lora_pipeline.py

Co-authored-by: hlky <hlky@hlky.ac>

* update

* update

* Update src/diffusers/loaders/lora_pipeline.py

Co-authored-by: hlky <hlky@hlky.ac>

* Apply style fixes

---------

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Linoy <linoy@hf.co>
Co-authored-by: hlky <hlky@hlky.ac>
 2025-03-19 21:44:19 +05:30
Yuqian Hong
fc28791fc8 [BUG] Fix Autoencoderkl train script (#11113) 
* add disc_optimizer step (not fix)

* support syncbatchnorm in discriminator
 2025-03-19 16:49:02 +05:30
Sayak Paul
ae14612673 [CI] uninstall deps properly from pr gpu tests. (#11102) 
uninstall deps properly from pr gpu tests.
 2025-03-19 08:58:36 +05:30
hlky
0ab8fe49bf Quality options in export_to_video (#11090) 
* Quality options in `export_to_video`

* make style
 2025-03-18 10:32:33 -10:00
Aryan
3be6706018 Fix Group offloading behaviour when using streams (#11097) 
* update

* update
 2025-03-18 14:44:10 +05:30
Cheng Jin
cb1b8b21b8 Resolve stride mismatch in UNet's ResNet to support Torch DDP (#11098) 
Modify UNet's ResNet implementation to resolve stride mismatch in Torch's DDP
 2025-03-18 07:38:13 +00:00
Juan Acevedo
27916822b2 update readme instructions. (#11096) 
Co-authored-by: Juan Acevedo <jfacevedo@google.com>
 2025-03-17 20:07:48 -10:00
co63oc
3fe3bc0642 Fix pipeline_flux_controlnet.py (#11095) 
* Fix pipeline_flux_controlnet.py

* Fix style
 2025-03-17 19:52:15 -10:00
Aryan
813d42cc96 Group offloading improvements (#11094) 
update
 2025-03-18 11:18:00 +05:30
Sayak Paul
b4d7e9c632 make PR GPU tests conditioned on styling. (#11099) 2025-03-18 11:15:35 +05:30
Aryan
2e83cbbb6d LTX 0.9.5 (#10968) 
* update


---------

Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: hlky <hlky@hlky.ac>
 2025-03-17 16:43:36 -10:00