huggingface/diffusers

Pull Requests Commits

[docs] `torch_dtype` map (#11194)

hlky committed 360 days ago

Verified e5c6027e

Revert `save_model` in ModelMixin save_pretrained and use safe_serialization=False in test (#11196)

hlky committed 360 days ago

Verified da857beb

[tests] HunyuanDiTControlNetPipeline inference precision issue on XPU (#11197)

faaany committed 360 days ago

Verified 52b460fe

allow models to run with a user-provided dtype map instead of a single dtype (#10301)

hlky committed 360 days ago

Verified d8c617cc

remove unnecessary call to `F.pad` (#10620)

bm-synth committed 360 days ago

Verified fe2b3974

fix: for checking mandatory and optional pipeline components (#11189)

elismasilva committed 360 days ago

Verified be0b7f55

fix autocast (#11190)

jiqing-feng committed 360 days ago

Verified 4d5a96e4

map BACKEND_RESET_MAX_MEMORY_ALLOCATED to reset_peak_memory_stats on XPU (#11191)

yao-matrix committed 360 days ago

Verified a7f07c1e

Merge branch 'main' into integrations/first-block-cache-2

a-r-r-o-w committed 360 days ago

Verified 315e357a

support flux, ltx i2v, ltx condition

a-r-r-o-w committed 360 days ago

1f33ca27

fix controlnet flux

a-r-r-o-w committed 360 days ago

41b0c473

fix hs residual bug for single return outputs; support ltx

a-r-r-o-w committed 360 days ago

0e232ac8

cache context for different batches of data

a-r-r-o-w committed 360 days ago

2557238b

a-r-r-o-w committed 361 days ago

d71fe558

[WIP] Add Wan Video2Video (#11053)

DN6 committed 361 days ago

Verified df1d7b01

[tests] no hard-coded cuda (#11186)

faaany committed 361 days ago

Verified 5a6edac0

remove debug logs

a-r-r-o-w committed 361 days ago

7ab424a1

modify flux single blocks to make compatible with cache techniques (without too much model-specific intrusion code)

a-r-r-o-w committed 361 days ago

dd69b418

Bug fix in LTXImageToVideoPipeline.prepare_latents() when latents is already set (#10918)

kakukakujirori committed 361 days ago

Verified e8fc8b1f

Add `latents_mean` and `latents_std` to `SDXLLongPromptWeightingPipeline` (#11034)

hlky committed 361 days ago

Verified d6f4774c

[Docs] Fix environment variables in `installation.md` (#11179)

remarkablemark committed 361 days ago

Verified eb50deff

Raise warning and round down if Wan num_frames is not 4k + 1 (#11167)

a-r-r-o-w committed 362 days ago

Verified 2c59af72

a-r-r-o-w committed 362 days ago

406b1062

Fix LatteTransformer3DModel dtype mismatch with enable_temporal_attentions (#11139)

hlky committed 364 days ago

Verified 75d7e5cc

[Docs] Update Wan Docs with memory optimizations (#11089)

DN6 committed 1 year ago

Verified 617c208b

WanI2V encode_image (#11164)

hlky committed 1 year ago

Verified 5d970a4a

Set self._hf_peft_config_loaded to True when LoRA is loaded using `load_lora_adapter` in PeftAdapterMixin class (#11155)

kentdan3msu committed 1 year ago

Verified de6a88c2

[Quantization] dtype fix for GGUF + fix BnB tests (#11159)

DN6 committed 1 year ago

Verified 7dc52ea7

add a timestep scale for sana-sprint teacher model (#11150)

lawrence-cj committed 1 year ago

Verified 739d6ec7

Improve information about group offloading and layerwise casting (#11101)

a-r-r-o-w committed 1 year ago

Verified 1ddf3f3a

Newer Older