kanshan/diffusers
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[Styling] stylify using ruff (#5841)

Browse Source 
* ruff format

* not need to use doc-builder's black styling as the doc is styled in ruff

* make fix-copies

* comment

* use run_ruff
This commit is contained in:
Kashif Rasul 2023-11-20 11:48:34 +01:00 committed by GitHub
parent 9c7f7fc475
commit 6b04d61cf6
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
134 changed files with 430 additions and 297 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
.github/workflows/pr_quality.yml vendored
View File

@ -27,9 +27,8 @@ jobs:
pip install .[quality]
- name: Check quality
run: |
black --check examples tests src utils scripts
ruff examples tests src utils scripts
doc-builder style src/diffusers docs/source --max_len 119 --check_only --path_to_docs docs/source
ruff check examples tests src utils scripts
ruff format examples tests src utils scripts --check
check_repository_consistency:
runs-on: ubuntu-latest

2
CONTRIBUTING.md
View File

@ -410,7 +410,7 @@ Diffusers has grown a lot. Here is the command for it:
$ make test
```
🧨 Diffusers relies on `black` and `isort` to format its source code
🧨 Diffusers relies on `ruff` and `isort` to format its source code
consistently. After you make changes, apply automatic style corrections and code verifications
that can't be automated in one go with:

14
Makefile
View File

@ -9,8 +9,8 @@ modified_only_fixup:
$(eval modified_py_files := $(shell python utils/get_modified_files.py $(check_dirs)))
@if test -n "$(modified_py_files)"; then \
echo "Checking/fixing $(modified_py_files)"; \
black $(modified_py_files); \
ruff $(modified_py_files); \
ruff check $(modified_py_files) --fix; \
ruff format $(modified_py_files);\
else \
echo "No library .py files were modified"; \
fi
@ -40,23 +40,21 @@ repo-consistency:
# this target runs checks on all files
quality:
black --check $(check_dirs)
ruff $(check_dirs)
doc-builder style src/diffusers docs/source --max_len 119 --check_only --path_to_docs docs/source
ruff check $(check_dirs) setup.py
ruff format --check $(check_dirs) setup.py
python utils/check_doc_toc.py
# Format source code automatically and check is there are any problems left that need manual fixing
extra_style_checks:
python utils/custom_init_isort.py
doc-builder style src/diffusers docs/source --max_len 119 --path_to_docs docs/source
python utils/check_doc_toc.py --fix_and_overwrite
# this target runs checks on all files and potentially modifies some of them
style:
black $(check_dirs)
ruff $(check_dirs) --fix
ruff check $(check_dirs) setup.py --fix
ruff format $(check_dirs) setup.py
${MAKE} autogenerate_code
${MAKE} extra_style_checks

1
examples/community/composable_stable_diffusion.py
View File

@ -65,6 +65,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(

4
examples/community/latent_consistency_img2img.py
View File

@ -564,9 +564,7 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
self.betas = torch.linspace(beta_start, beta_end, num_train_timesteps, dtype=torch.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
self.betas = (
torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
)
self.betas = torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
self.betas = betas_for_alpha_bar(num_train_timesteps)

4
examples/community/latent_consistency_txt2img.py
View File

@ -469,9 +469,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
self.betas = torch.linspace(beta_start, beta_end, num_train_timesteps, dtype=torch.float32)
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
self.betas = (
torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
)
self.betas = torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
self.betas = betas_for_alpha_bar(num_train_timesteps)

10
examples/community/lpw_stable_diffusion.py
View File

@ -56,10 +56,10 @@ def parse_prompt_attention(text):
(abc) - increases attention to abc by a multiplier of 1.1
(abc:3.12) - increases attention to abc by a multiplier of 3.12
[abc] - decreases attention to abc by a multiplier of 1.1
\( - literal character '('
\[ - literal character '['
\) - literal character ')'
\] - literal character ']'
\\( - literal character '('
\\[ - literal character '['
\\) - literal character ')'
\\] - literal character ']'
\\ - literal character '\'
anything else - just text
>>> parse_prompt_attention('normal text')
@ -68,7 +68,7 @@ def parse_prompt_attention(text):
[['an ', 1.0], ['important', 1.1], [' word', 1.0]]
>>> parse_prompt_attention('(unbalanced')
[['unbalanced', 1.1]]
>>> parse_prompt_attention('\(literal\]')
>>> parse_prompt_attention('\\(literal\\]')
[['(literal]', 1.0]]
>>> parse_prompt_attention('(unnecessary)(parens)')
[['unnecessaryparens', 1.1]]

11
examples/community/lpw_stable_diffusion_onnx.py
View File

@ -82,10 +82,10 @@ def parse_prompt_attention(text):
(abc) - increases attention to abc by a multiplier of 1.1
(abc:3.12) - increases attention to abc by a multiplier of 3.12
[abc] - decreases attention to abc by a multiplier of 1.1
\( - literal character '('
\[ - literal character '['
\) - literal character ')'
\] - literal character ']'
\\( - literal character '('
\\[ - literal character '['
\\) - literal character ')'
\\] - literal character ']'
\\ - literal character '\'
anything else - just text
>>> parse_prompt_attention('normal text')
@ -94,7 +94,7 @@ def parse_prompt_attention(text):
[['an ', 1.0], ['important', 1.1], [' word', 1.0]]
>>> parse_prompt_attention('(unbalanced')
[['unbalanced', 1.1]]
>>> parse_prompt_attention('\(literal\]')
>>> parse_prompt_attention('\\(literal\\]')
[['(literal]', 1.0]]
>>> parse_prompt_attention('(unnecessary)(parens)')
[['unnecessaryparens', 1.1]]
@ -433,6 +433,7 @@ class OnnxStableDiffusionLongPromptWeightingPipeline(OnnxStableDiffusionPipeline
This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the
library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
"""
if version.parse(version.parse(diffusers.__version__).base_version) >= version.parse("0.9.0"):
def __init__(

10
examples/community/lpw_stable_diffusion_xl.py
View File

@ -46,10 +46,10 @@ def parse_prompt_attention(text):
(abc) - increases attention to abc by a multiplier of 1.1
(abc:3.12) - increases attention to abc by a multiplier of 3.12
[abc] - decreases attention to abc by a multiplier of 1.1
\( - literal character '('
\[ - literal character '['
\) - literal character ')'
\] - literal character ']'
\\( - literal character '('
\\[ - literal character '['
\\) - literal character ')'
\\] - literal character ']'
\\ - literal character '\'
anything else - just text
@ -59,7 +59,7 @@ def parse_prompt_attention(text):
[['an ', 1.0], ['important', 1.1], [' word', 1.0]]
>>> parse_prompt_attention('(unbalanced')
[['unbalanced', 1.1]]
>>> parse_prompt_attention('\(literal\]')
>>> parse_prompt_attention('\\(literal\\]')
[['(literal]', 1.0]]
>>> parse_prompt_attention('(unnecessary)(parens)')
[['unnecessaryparens', 1.1]]

6
examples/community/magic_mix.py
View File

@ -127,9 +127,9 @@ class MagicMixPipeline(DiffusionPipeline):
timesteps=t,
)
input = (mix_factor * latents) + (
1 - mix_factor
) * orig_latents # interpolating between layout noise and conditionally generated noise to preserve layout sematics
input = (
(mix_factor * latents) + (1 - mix_factor) * orig_latents
) # interpolating between layout noise and conditionally generated noise to preserve layout sematics
input = torch.cat([input] * 2)
else: # content generation phase

4
examples/community/mixture_canvas.py
View File

@ -453,9 +453,7 @@ class StableDiffusionCanvasPipeline(DiffusionPipeline):
:,
region.latent_row_init : region.latent_row_end,
region.latent_col_init : region.latent_col_end,
] += (
noise_pred_region * mask_weights_region
)
] += noise_pred_region * mask_weights_region
contributors[
:,
:,

1
examples/community/pipeline_prompt2prompt.py
View File

@ -65,6 +65,7 @@ class Prompt2PromptPipeline(StableDiffusionPipeline):
feature_extractor ([`CLIPFeatureExtractor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
@torch.no_grad()

4
examples/community/pipeline_zero1to3.py
View File

@ -94,6 +94,7 @@ class Zero1to3StableDiffusionPipeline(DiffusionPipeline):
cc_projection ([`CCProjection`]):
Projection layer to project the concated CLIP features and pose embeddings to the original CLIP feature size.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(
@ -658,7 +659,8 @@ class Zero1to3StableDiffusionPipeline(DiffusionPipeline):
if isinstance(generator, list):
init_latents = [
self.vae.encode(image[i : i + 1]).latent_dist.mode(generator[i]) for i in range(batch_size) # sample
self.vae.encode(image[i : i + 1]).latent_dist.mode(generator[i])
for i in range(batch_size) # sample
]
init_latents = torch.cat(init_latents, dim=0)
else:

7
examples/community/run_onnx_controlnet.py
View File

@ -651,9 +651,10 @@ class OnnxStableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = num_controlnet
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

7
examples/community/run_tensorrt_controlnet.py
View File

@ -755,9 +755,10 @@ class TensorRTStableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline):
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = num_controlnet
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

1
examples/community/sd_text2img_k_diffusion.py
View File

@ -68,6 +68,7 @@ class StableDiffusionPipeline(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(

1
examples/community/stable_diffusion_ipex.py
View File

@ -89,6 +89,7 @@ class StableDiffusionIPEXPipeline(DiffusionPipeline, TextualInversionLoaderMixin
feature_extractor ([`CLIPFeatureExtractor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(

1
examples/community/stable_diffusion_mega.py
View File

@ -50,6 +50,7 @@ class StableDiffusionMegaPipeline(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(

1
examples/community/stable_diffusion_repaint.py
View File

@ -170,6 +170,7 @@ class StableDiffusionRepaintPipeline(DiffusionPipeline, TextualInversionLoaderMi
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
def __init__(

4
examples/research_projects/colossalai/train_dreambooth_colossalai.py
View File

@ -464,9 +464,7 @@ def main(args):
unet = gemini_zero_dpp(unet, args.placement)
# config optimizer for colossalai zero
optimizer = GeminiAdamOptimizer(
unet, lr=args.learning_rate, initial_scale=2**5, clipping_norm=args.max_grad_norm
)
optimizer = GeminiAdamOptimizer(unet, lr=args.learning_rate, initial_scale=2**5, clipping_norm=args.max_grad_norm)
# load noise_scheduler
noise_scheduler = DDPMScheduler.from_pretrained(args.pretrained_model_name_or_path, subfolder="scheduler")

19
pyproject.toml
View File

@ -1,10 +1,6 @@
[tool.black]
line-length = 119
target-version = ['py37']
[tool.ruff]
# Never enforce `E501` (line length violations).
ignore = ["C901", "E501", "E741", "W605"]
ignore = ["C901", "E501", "E741", "F402", "F823"]
select = ["C", "E", "F", "I", "W"]
line-length = 119
@ -16,3 +12,16 @@ line-length = 119
[tool.ruff.isort]
lines-after-imports = 2
known-first-party = ["diffusers"]
[tool.ruff.format]
# Like Black, use double quotes for strings.
quote-style = "double"
# Like Black, indent with spaces, rather than tabs.
indent-style = "space"
# Like Black, respect magic trailing commas.
skip-magic-trailing-comma = false
# Like Black, automatically detect the appropriate line ending.
line-ending = "auto"

2
scripts/convert_kakao_brain_unclip_to_diffusers.py
View File

@ -11,7 +11,7 @@ from diffusers.pipelines.unclip.text_proj import UnCLIPTextProjModel
from diffusers.schedulers.scheduling_unclip import UnCLIPScheduler
"""
r"""
Example - From the diffusers root directory:
Download weights:

20
setup.cfg
View File

@ -1,20 +0,0 @@
[isort]
default_section = FIRSTPARTY
ensure_newline_before_comments = True
force_grid_wrap = 0
include_trailing_comma = True
known_first_party = accelerate
known_third_party =
numpy
torch
torch_xla
line_length = 119
lines_after_imports = 2
multi_line_output = 3
use_parentheses = True
[flake8]
ignore = E203, E722, E501, E741, W503, W605
max-line-length = 119
per-file-ignores = __init__.py:F401

26
setup.py
View File

@ -44,9 +44,9 @@ To create the package for PyPI.
For the sources, run: "python setup.py sdist"
You should now have a /dist directory with both .whl and .tar.gz source versions.
Long story cut short, you need to run both before you can upload the distribution to the
test PyPI and the actual PyPI servers:
Long story cut short, you need to run both before you can upload the distribution to the
test PyPI and the actual PyPI servers:
python setup.py bdist_wheel && python setup.py sdist
8. Check that everything looks correct by uploading the package to the PyPI test server:
@ -78,9 +78,9 @@ To create the package for PyPI.
you need to go back to main before executing this.
"""
import sys
import os
import re
import sys
from distutils.core import Command
from setuptools import find_packages, setup
@ -93,7 +93,6 @@ _deps = [
"Pillow", # keep the PIL.Image.Resampling deprecation away
"accelerate>=0.11.0",
"compel==0.1.8",
"black~=23.1",
"datasets",
"filelock",
"flax>=0.4.1",
@ -119,7 +118,7 @@ _deps = [
"pytest-timeout",
"pytest-xdist",
"python>=3.8.0",
"ruff==0.0.280",
"ruff>=0.1.5,<=0.2",
"safetensors>=0.3.1",
"sentencepiece>=0.1.91,!=0.1.92",
"scipy",
@ -171,7 +170,11 @@ class DepsTableUpdateCommand(Command):
description = "build runtime dependency table"
user_options = [
# format: (long option, short option, description).
("dep-table-update", None, "updates src/diffusers/dependency_versions_table.py"),
(
"dep-table-update",
None,
"updates src/diffusers/dependency_versions_table.py",
),
]
def initialize_options(self):
@ -197,10 +200,8 @@ class DepsTableUpdateCommand(Command):
f.write("\n".join(content))
extras = {}
extras["quality"] = deps_list("urllib3", "black", "isort", "ruff", "hf-doc-builder")
extras["quality"] = deps_list("urllib3", "isort", "ruff", "hf-doc-builder")
extras["docs"] = deps_list("hf-doc-builder")
extras["training"] = deps_list("accelerate", "datasets", "protobuf", "tensorboard", "Jinja2")
extras["test"] = deps_list(
@ -275,10 +276,7 @@ setup(
"Topic :: Scientific/Engineering :: Artificial Intelligence",
"Programming Language :: Python :: 3",
]
+ [
f"Programming Language :: Python :: 3.{i}"
for i in range(8, version_range_max)
],
+ [f"Programming Language :: Python :: 3.{i}" for i in range(8, version_range_max)],
cmdclass={"deps_table_update": DepsTableUpdateCommand},
)

1
src/diffusers/configuration_utils.py
View File

@ -95,6 +95,7 @@ class ConfigMixin:
should only have a `kwargs` argument if at least one argument is deprecated (should be overridden by
subclass).
"""
config_name = None
ignore_for_config = []
has_compatibles = False

3
src/diffusers/dependency_versions_table.py
View File

@ -5,7 +5,6 @@ deps = {
"Pillow": "Pillow",
"accelerate": "accelerate>=0.11.0",
"compel": "compel==0.1.8",
"black": "black~=23.1",
"datasets": "datasets",
"filelock": "filelock",
"flax": "flax>=0.4.1",
@ -31,7 +30,7 @@ deps = {
"pytest-timeout": "pytest-timeout",
"pytest-xdist": "pytest-xdist",
"python": "python>=3.8.0",
"ruff": "ruff==0.0.280",
"ruff": "ruff>=0.1.5,<=0.2",
"safetensors": "safetensors>=0.3.1",
"sentencepiece": "sentencepiece>=0.1.91,!=0.1.92",
"scipy": "scipy",

1
src/diffusers/loaders/lora.py
View File

@ -71,6 +71,7 @@ class LoraLoaderMixin:
Load LoRA layers into [`UNet2DConditionModel`] and
[`CLIPTextModel`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel).
"""
text_encoder_name = TEXT_ENCODER_NAME
unet_name = UNET_NAME
num_fused_loras = 0

10
src/diffusers/models/attention_flax.py
View File

@ -110,7 +110,10 @@ def jax_memory_efficient_attention(
)
_, res = jax.lax.scan(
f=chunk_scanner, init=0, xs=None, length=math.ceil(num_q / query_chunk_size) # start counter # stop counter
f=chunk_scanner,
init=0,
xs=None,
length=math.ceil(num_q / query_chunk_size), # start counter # stop counter
)
return jnp.concatenate(res, axis=-3) # fuse the chunked result back
@ -138,6 +141,7 @@ class FlaxAttention(nn.Module):
Parameters `dtype`
"""
query_dim: int
heads: int = 8
dim_head: int = 64
@ -262,6 +266,7 @@ class FlaxBasicTransformerBlock(nn.Module):
Whether to split the head dimension into a new axis for the self-attention computation. In most cases,
enabling this flag should speed up the computation for Stable Diffusion 2.x and Stable Diffusion XL.
"""
dim: int
n_heads: int
d_head: int
@ -347,6 +352,7 @@ class FlaxTransformer2DModel(nn.Module):
Whether to split the head dimension into a new axis for the self-attention computation. In most cases,
enabling this flag should speed up the computation for Stable Diffusion 2.x and Stable Diffusion XL.
"""
in_channels: int
n_heads: int
d_head: int
@ -442,6 +448,7 @@ class FlaxFeedForward(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
dim: int
dropout: float = 0.0
dtype: jnp.dtype = jnp.float32
@ -471,6 +478,7 @@ class FlaxGEGLU(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
dim: int
dropout: float = 0.0
dtype: jnp.dtype = jnp.float32

1
src/diffusers/models/autoencoder_tiny.py
View File

@ -91,6 +91,7 @@ class AutoencoderTiny(ModelMixin, ConfigMixin):
`force_upcast` can be set to `False` (see this fp16-friendly
[AutoEncoder](https://huggingface.co/madebyollin/sdxl-vae-fp16-fix)).
"""
_supports_gradient_checkpointing = True
@register_to_config

1
src/diffusers/models/controlnet_flax.py
View File

@ -146,6 +146,7 @@ class FlaxControlNetModel(nn.Module, FlaxModelMixin, ConfigMixin):
conditioning_embedding_out_channels (`tuple`, *optional*, defaults to `(16, 32, 96, 256)`):
The tuple of output channel for each block in the `conditioning_embedding` layer.
"""
sample_size: int = 32
in_channels: int = 4
down_block_types: Tuple[str, ...] = (

2
src/diffusers/models/embeddings_flax.py
View File

@ -65,6 +65,7 @@ class FlaxTimestepEmbedding(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
time_embed_dim: int = 32
dtype: jnp.dtype = jnp.float32
@ -84,6 +85,7 @@ class FlaxTimesteps(nn.Module):
dim (`int`, *optional*, defaults to `32`):
Time step embedding dimension
"""
dim: int = 32
flip_sin_to_cos: bool = False
freq_shift: float = 1

1
src/diffusers/models/modeling_flax_utils.py
View File

@ -52,6 +52,7 @@ class FlaxModelMixin(PushToHubMixin):
- **config_name** ([`str`]) -- Filename to save a model to when calling [`~FlaxModelMixin.save_pretrained`].
"""
config_name = CONFIG_NAME
_automatically_saved_args = ["_diffusers_version", "_class_name", "_name_or_path"]
_flax_internal_args = ["name", "parent", "dtype"]

1
src/diffusers/models/modeling_utils.py
View File

@ -193,6 +193,7 @@ class ModelMixin(torch.nn.Module, PushToHubMixin):
- **config_name** ([`str`]) -- Filename to save a model to when calling [`~models.ModelMixin.save_pretrained`].
"""
config_name = CONFIG_NAME
_automatically_saved_args = ["_diffusers_version", "_class_name", "_name_or_path"]
_supports_gradient_checkpointing = False

5
src/diffusers/models/unet_2d_blocks_flax.py
View File

@ -45,6 +45,7 @@ class FlaxCrossAttnDownBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
dropout: float = 0.0
@ -125,6 +126,7 @@ class FlaxDownBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
dropout: float = 0.0
@ -190,6 +192,7 @@ class FlaxCrossAttnUpBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
prev_output_channel: int
@ -275,6 +278,7 @@ class FlaxUpBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
prev_output_channel: int
@ -339,6 +343,7 @@ class FlaxUNetMidBlock2DCrossAttn(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
dropout: float = 0.0
num_layers: int = 1

1
src/diffusers/models/unet_motion_model.py
View File

@ -174,6 +174,7 @@ class UNetMotionModel(ModelMixin, ConfigMixin, UNet2DConditionLoadersMixin):
This model inherits from [`ModelMixin`]. Check the superclass documentation for it's generic methods implemented
for all models (such as downloading or saving).
"""
_supports_gradient_checkpointing = True
@register_to_config

7
src/diffusers/models/vae_flax.py
View File

@ -214,6 +214,7 @@ class FlaxAttentionBlock(nn.Module):
Parameters `dtype`
"""
channels: int
num_head_channels: int = None
num_groups: int = 32
@ -291,6 +292,7 @@ class FlaxDownEncoderBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
dropout: float = 0.0
@ -347,6 +349,7 @@ class FlaxUpDecoderBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
out_channels: int
dropout: float = 0.0
@ -401,6 +404,7 @@ class FlaxUNetMidBlock2D(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int
dropout: float = 0.0
num_layers: int = 1
@ -488,6 +492,7 @@ class FlaxEncoder(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
Parameters `dtype`
"""
in_channels: int = 3
out_channels: int = 3
down_block_types: Tuple[str] = ("DownEncoderBlock2D",)
@ -600,6 +605,7 @@ class FlaxDecoder(nn.Module):
dtype (:obj:`jnp.dtype`, *optional*, defaults to jnp.float32):
parameters `dtype`
"""
in_channels: int = 3
out_channels: int = 3
up_block_types: Tuple[str] = ("UpDecoderBlock2D",)
@ -767,6 +773,7 @@ class FlaxAutoencoderKL(nn.Module, FlaxModelMixin, ConfigMixin):
dtype (`jnp.dtype`, *optional*, defaults to `jnp.float32`):
The `dtype` of the parameters.
"""
in_channels: int = 3
out_channels: int = 3
down_block_types: Tuple[str] = ("DownEncoderBlock2D",)

19
src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion.py
View File

@ -243,10 +243,7 @@ class AltDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
lora_scale: Optional[float] = None,
**kwargs,
):
deprecation_message = (
"`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()`"
" instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
)
deprecation_message = "`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()` instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
deprecate("_encode_prompt()", "1.0.0", deprecation_message, standard_warn=False)
prompt_embeds_tuple = self.encode_prompt(
@ -462,10 +459,7 @@ class AltDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
return image, has_nsfw_concept
def decode_latents(self, latents):
deprecation_message = (
"The decode_latents method is deprecated and will be removed in 1.0.0. Please use"
" VaeImageProcessor.postprocess(...) instead"
)
deprecation_message = "The decode_latents method is deprecated and will be removed in 1.0.0. Please use VaeImageProcessor.postprocess(...) instead"
deprecate("decode_latents", "1.0.0", deprecation_message, standard_warn=False)
latents = 1 / self.vae.config.scaling_factor * latents
@ -515,8 +509,7 @@ class AltDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
):
raise ValueError(
f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found"
f" {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
)
if prompt is not None and prompt_embeds is not None:
@ -747,15 +740,13 @@ class AltDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
deprecate(
"callback",
"1.0.0",
"Passing `callback` as an input argument to `__call__` is deprecated, consider using"
" `callback_on_step_end`",
"Passing `callback` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
)
if callback_steps is not None:
deprecate(
"callback_steps",
"1.0.0",
"Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using"
" `callback_on_step_end`",
"Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
)
# 0. Default height and width to unet

23
src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion_img2img.py
View File

@ -252,10 +252,7 @@ class AltDiffusionImg2ImgPipeline(
lora_scale: Optional[float] = None,
**kwargs,
):
deprecation_message = (
"`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()`"
" instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
)
deprecation_message = "`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()` instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
deprecate("_encode_prompt()", "1.0.0", deprecation_message, standard_warn=False)
prompt_embeds_tuple = self.encode_prompt(
@ -471,10 +468,7 @@ class AltDiffusionImg2ImgPipeline(
return image, has_nsfw_concept
def decode_latents(self, latents):
deprecation_message = (
"The decode_latents method is deprecated and will be removed in 1.0.0. Please use"
" VaeImageProcessor.postprocess(...) instead"
)
deprecation_message = "The decode_latents method is deprecated and will be removed in 1.0.0. Please use VaeImageProcessor.postprocess(...) instead"
deprecate("decode_latents", "1.0.0", deprecation_message, standard_warn=False)
latents = 1 / self.vae.config.scaling_factor * latents
@ -524,8 +518,7 @@ class AltDiffusionImg2ImgPipeline(
k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
):
raise ValueError(
f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found"
f" {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
)
if prompt is not None and prompt_embeds is not None:
raise ValueError(
@ -578,8 +571,8 @@ class AltDiffusionImg2ImgPipeline(
else:
if isinstance(generator, list) and len(generator) != batch_size:
raise ValueError(
f"You have passed a list of generators of length {len(generator)}, but requested an effective"
f" batch size of {batch_size}. Make sure the batch size matches the length of the generators."
f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
f" size of {batch_size}. Make sure the batch size matches the length of the generators."
)
elif isinstance(generator, list):
@ -798,15 +791,13 @@ class AltDiffusionImg2ImgPipeline(
deprecate(
"callback",
"1.0.0",
"Passing `callback` as an input argument to `__call__` is deprecated, consider use"
" `callback_on_step_end`",
"Passing `callback` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
)
if callback_steps is not None:
deprecate(
"callback_steps",
"1.0.0",
"Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use"
" `callback_on_step_end`",
"Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
)
# 1. Check inputs. Raise error if not correct

1
src/diffusers/pipelines/animatediff/pipeline_animatediff.py
View File

@ -99,6 +99,7 @@ class AnimateDiffPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraLo
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
def __init__(

1
src/diffusers/pipelines/audioldm/pipeline_audioldm.py
View File

@ -72,6 +72,7 @@ class AudioLDMPipeline(DiffusionPipeline):
vocoder ([`~transformers.SpeechT5HifiGan`]):
Vocoder of class `SpeechT5HifiGan`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
def __init__(

3
src/diffusers/pipelines/auto_pipeline.py
View File

@ -181,6 +181,7 @@ class AutoPipelineForText2Image(ConfigMixin):
diffusion pipeline's components.
"""
config_name = "model_index.json"
def __init__(self, *args, **kwargs):
@ -451,6 +452,7 @@ class AutoPipelineForImage2Image(ConfigMixin):
diffusion pipeline's components.
"""
config_name = "model_index.json"
def __init__(self, *args, **kwargs):
@ -726,6 +728,7 @@ class AutoPipelineForInpainting(ConfigMixin):
diffusion pipeline's components.
"""
config_name = "model_index.json"
def __init__(self, *args, **kwargs):

1
src/diffusers/pipelines/consistency_models/pipeline_consistency_models.py
View File

@ -74,6 +74,7 @@ class ConsistencyModelPipeline(DiffusionPipeline):
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Currently only
compatible with [`CMStochasticIterativeScheduler`].
"""
model_cpu_offload_seq = "unet"
def __init__(self, unet: UNet2DModel, scheduler: CMStochasticIterativeScheduler) -> None:

8
src/diffusers/pipelines/controlnet/pipeline_controlnet.py
View File

@ -126,6 +126,7 @@ class StableDiffusionControlNetPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]
@ -886,9 +887,10 @@ class StableDiffusionControlNetPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

8
src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py
View File

@ -160,6 +160,7 @@ class StableDiffusionControlNetImg2ImgPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]
@ -933,9 +934,10 @@ class StableDiffusionControlNetImg2ImgPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

8
src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py
View File

@ -282,6 +282,7 @@ class StableDiffusionControlNetInpaintPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]
@ -1142,9 +1143,10 @@ class StableDiffusionControlNetInpaintPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

15
src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py
View File

@ -164,6 +164,7 @@ class StableDiffusionXLControlNetInpaintPipeline(
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
@ -1198,9 +1199,10 @@ class StableDiffusionXLControlNetInpaintPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# # 0.0 Default height and width to unet
# height = height or self.unet.config.sample_size * self.vae_scale_factor
@ -1213,9 +1215,10 @@ class StableDiffusionXLControlNetInpaintPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs
self.check_inputs(

8
src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py
View File

@ -139,6 +139,7 @@ class StableDiffusionXLControlNetPipeline(
watermark output images. If not defined, it defaults to `True` if the package is installed; otherwise no
watermarker is used.
"""
# leave controlnet out on purpose because it iterates with unet
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
@ -1005,9 +1006,10 @@ class StableDiffusionXLControlNetPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

8
src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py
View File

@ -192,6 +192,7 @@ class StableDiffusionXLControlNetImg2ImgPipeline(
watermark output images. If not defined, it will default to True if the package is installed, otherwise no
watermarker will be used.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
@ -1154,9 +1155,10 @@ class StableDiffusionXLControlNetImg2ImgPipeline(
control_guidance_end = len(control_guidance_start) * [control_guidance_end]
elif not isinstance(control_guidance_start, list) and not isinstance(control_guidance_end, list):
mult = len(controlnet.nets) if isinstance(controlnet, MultiControlNetModel) else 1
control_guidance_start, control_guidance_end = mult * [control_guidance_start], mult * [
control_guidance_end
]
control_guidance_start, control_guidance_end = (
mult * [control_guidance_start],
mult * [control_guidance_end],
)
# 1. Check inputs. Raise error if not correct
self.check_inputs(

1
src/diffusers/pipelines/dance_diffusion/pipeline_dance_diffusion.py
View File

@ -39,6 +39,7 @@ class DanceDiffusionPipeline(DiffusionPipeline):
A scheduler to be used in combination with `unet` to denoise the encoded audio latents. Can be one of
[`IPNDMScheduler`].
"""
model_cpu_offload_seq = "unet"
def __init__(self, unet, scheduler):

1
src/diffusers/pipelines/ddim/pipeline_ddim.py
View File

@ -35,6 +35,7 @@ class DDIMPipeline(DiffusionPipeline):
A scheduler to be used in combination with `unet` to denoise the encoded image. Can be one of
[`DDPMScheduler`], or [`DDIMScheduler`].
"""
model_cpu_offload_seq = "unet"
def __init__(self, unet, scheduler):

1
src/diffusers/pipelines/ddpm/pipeline_ddpm.py
View File

@ -35,6 +35,7 @@ class DDPMPipeline(DiffusionPipeline):
A scheduler to be used in combination with `unet` to denoise the encoded image. Can be one of
[`DDPMScheduler`], or [`DDIMScheduler`].
"""
model_cpu_offload_seq = "unet"
def __init__(self, unet, scheduler):

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if.py
View File

@ -98,7 +98,19 @@ class IFPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py
View File

@ -122,7 +122,19 @@ class IFImg2ImgPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img_superresolution.py
View File

@ -126,7 +126,19 @@ class IFImg2ImgSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor"]

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting.py
View File

@ -125,7 +125,19 @@ class IFInpaintingPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting_superresolution.py
View File

@ -128,7 +128,19 @@ class IFInpaintingSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
model_cpu_offload_seq = "text_encoder->unet"

14
src/diffusers/pipelines/deepfloyd_if/pipeline_if_superresolution.py
View File

@ -84,7 +84,19 @@ class IFSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
watermarker: Optional[IFWatermarker]
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]

1
src/diffusers/pipelines/dit/pipeline_dit.py
View File

@ -43,6 +43,7 @@ class DiTPipeline(DiffusionPipeline):
scheduler ([`DDIMScheduler`]):
A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
"""
model_cpu_offload_seq = "transformer->vae"
def __init__(

1
src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py
View File

@ -115,6 +115,7 @@ class LatentConsistencyModelImg2ImgPipeline(
requires_safety_checker (`bool`, *optional*, defaults to `True`):
Whether the pipeline requires a safety checker component.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py
View File

@ -97,6 +97,7 @@ class LatentConsistencyModelPipeline(
requires_safety_checker (`bool`, *optional*, defaults to `True`):
Whether the pipeline requires a safety checker component.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/latent_diffusion/pipeline_latent_diffusion.py
View File

@ -49,6 +49,7 @@ class LDMTextToImagePipeline(DiffusionPipeline):
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "bert->unet->vqvae"
def __init__(

1
src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py
View File

@ -177,6 +177,7 @@ class PaintByExamplePipeline(DiffusionPipeline):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
# TODO: feature_extractor is required to encode initial images (if they are in PIL format),
# we should give a descriptive message if the pipeline doesn't have one.

1
src/diffusers/pipelines/pipeline_flax_utils.py
View File

@ -112,6 +112,7 @@ class FlaxDiffusionPipeline(ConfigMixin, PushToHubMixin):
- **config_name** ([`str`]) -- The configuration filename that stores the class and module names of all the
diffusion pipeline's components.
"""
config_name = "model_index.json"
def register_modules(self, **kwargs):

1
src/diffusers/pipelines/pipeline_utils.py
View File

@ -542,6 +542,7 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
- **_optional_components** (`List[str]`) -- List of all optional components that don't have to be passed to the
pipeline to function (should be overridden by subclasses).
"""
config_name = "model_index.json"
model_cpu_offload_seq = None
_optional_components = []

15
src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py
View File

@ -120,8 +120,21 @@ class PixArtAlphaPipeline(DiffusionPipeline):
scheduler ([`SchedulerMixin`]):
A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
"""
bad_punct_regex = re.compile(
r"[" + "#®•©™&@·º½¾¿¡§~" + "\)" + "\(" + "\]" + "\[" + "\}" + "\{" + "\|" + "\\" + "\/" + "\*" + r"]{1,}"
r"["
+ "#®•©™&@·º½¾¿¡§~"
+ r"\)"
+ r"\("
+ r"\]"
+ r"\["
+ r"\}"
+ r"\{"
+ r"\|"
+ "\\"
+ r"\/"
+ r"\*"
+ r"]{1,}"
) # noqa
_optional_components = ["tokenizer", "text_encoder"]

1
src/diffusers/pipelines/score_sde_ve/pipeline_score_sde_ve.py
View File

@ -35,6 +35,7 @@ class ScoreSdeVePipeline(DiffusionPipeline):
scheduler ([`ScoreSdeVeScheduler`]):
A `ScoreSdeVeScheduler` to be used in combination with `unet` to denoise the encoded image.
"""
unet: UNet2DModel
scheduler: ScoreSdeVeScheduler

1
src/diffusers/pipelines/spectrogram_diffusion/pipeline_spectrogram_diffusion.py
View File

@ -54,6 +54,7 @@ class SpectrogramDiffusionPipeline(DiffusionPipeline):
A scheduler to be used in combination with `decoder` to denoise the encoded audio latents.
melgan ([`OnnxRuntimeModel`]):
"""
_optional_components = ["melgan"]
def __init__(

1
src/diffusers/pipelines/stable_diffusion/pipeline_cycle_diffusion.py
View File

@ -148,6 +148,7 @@ class CycleDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lor
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]

6
src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_img2img.py
View File

@ -33,10 +33,7 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.preprocess with 8->64
def preprocess(image):
deprecation_message = (
"The preprocess method is deprecated and will be removed in diffusers 1.0.0. Please use"
" VaeImageProcessor.preprocess(...) instead"
)
deprecation_message = "The preprocess method is deprecated and will be removed in diffusers 1.0.0. Please use VaeImageProcessor.preprocess(...) instead"
deprecate("preprocess", "1.0.0", deprecation_message, standard_warn=False)
if isinstance(image, torch.Tensor):
return image
@ -85,6 +82,7 @@ class OnnxStableDiffusionImg2ImgPipeline(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
vae_encoder: OnnxRuntimeModel
vae_decoder: OnnxRuntimeModel
text_encoder: OnnxRuntimeModel

1
src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_inpaint.py
View File

@ -80,6 +80,7 @@ class OnnxStableDiffusionInpaintPipeline(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
vae_encoder: OnnxRuntimeModel
vae_decoder: OnnxRuntimeModel
text_encoder: OnnxRuntimeModel

1
src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_inpaint_legacy.py
View File

@ -66,6 +66,7 @@ class OnnxStableDiffusionInpaintPipelineLegacy(DiffusionPipeline):
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
_is_onnx = True

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py
View File

@ -102,6 +102,7 @@ class StableDiffusionPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_attend_and_excite.py
View File

@ -196,6 +196,7 @@ class StableDiffusionAttendAndExcitePipeline(DiffusionPipeline, TextualInversion
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py
View File

@ -95,6 +95,7 @@ class StableDiffusionDepth2ImgPipeline(DiffusionPipeline, TextualInversionLoader
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds", "depth_mask"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_diffedit.py
View File

@ -273,6 +273,7 @@ class StableDiffusionDiffEditPipeline(DiffusionPipeline, TextualInversionLoaderM
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor", "inverse_scheduler"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_gligen.py
View File

@ -125,6 +125,7 @@ class StableDiffusionGLIGENPipeline(DiffusionPipeline):
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
_optional_components = ["safety_checker", "feature_extractor"]
model_cpu_offload_seq = "text_encoder->unet->vae"
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_gligen_text_image.py
View File

@ -177,6 +177,7 @@ class StableDiffusionGLIGENTextImagePipeline(DiffusionPipeline):
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_image_variation.py
View File

@ -62,6 +62,7 @@ class StableDiffusionImageVariationPipeline(DiffusionPipeline):
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
# TODO: feature_extractor is required to encode images (if they are in PIL format),
# we should give a descriptive message if the pipeline doesn't have one.
_optional_components = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py
View File

@ -139,6 +139,7 @@ class StableDiffusionImg2ImgPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py
View File

@ -202,6 +202,7 @@ class StableDiffusionInpaintPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py
View File

@ -115,6 +115,7 @@ class StableDiffusionInpaintPipelineLegacy(
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py
View File

@ -89,6 +89,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_k_diffusion.py
View File

@ -80,6 +80,7 @@ class StableDiffusionKDiffusionPipeline(DiffusionPipeline, TextualInversionLoade
feature_extractor ([`CLIPImageProcessor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py
View File

@ -79,6 +79,7 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, FromSingleFileMixi
scheduler ([`SchedulerMixin`]):
A [`EulerDiscreteScheduler`] to be used in combination with `unet` to denoise the encoded image latents.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
def __init__(

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_ldm3d.py
View File

@ -115,6 +115,7 @@ class StableDiffusionLDM3DPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_model_editing.py
View File

@ -66,6 +66,7 @@ class StableDiffusionModelEditingPipeline(DiffusionPipeline, TextualInversionLoa
with_augs ([`list`]):
Textual augmentations to apply while editing the text-to-image model. Set to `[]` for no augmentations.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_panorama.py
View File

@ -85,6 +85,7 @@ class StableDiffusionPanoramaPipeline(DiffusionPipeline, TextualInversionLoaderM
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_paradigms.py
View File

@ -96,6 +96,7 @@ class StableDiffusionParadigmsPipeline(
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py
View File

@ -310,6 +310,7 @@ class StableDiffusionPix2PixZeroPipeline(DiffusionPipeline):
Whether the pipeline requires a safety checker. We recommend setting it to True if you're using the
pipeline publicly.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = [
"safety_checker",

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_sag.py
View File

@ -124,6 +124,7 @@ class StableDiffusionSAGPipeline(DiffusionPipeline, TextualInversionLoaderMixin)
feature_extractor ([`~transformers.CLIPImageProcessor`]):
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_upscale.py
View File

@ -92,6 +92,7 @@ class StableDiffusionUpscalePipeline(
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
_optional_components = ["watermarker", "safety_checker", "feature_extractor"]
_exclude_from_cpu_offload = ["safety_checker"]

1
src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py
View File

@ -140,6 +140,7 @@ class StableDiffusionXLPipeline(
watermark output images. If not defined, it will default to True if the package is installed, otherwise no
watermarker will be used.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
_callback_tensor_inputs = [

1
src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py
View File

@ -153,6 +153,7 @@ class StableDiffusionXLImg2ImgPipeline(
watermark output images. If not defined, it will default to True if the package is installed, otherwise no
watermarker will be used.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
_callback_tensor_inputs = [

1
src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py
View File

@ -298,6 +298,7 @@ class StableDiffusionXLInpaintPipeline(
watermark output images. If not defined, it will default to True if the package is installed, otherwise no
watermarker will be used.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]

1
src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py
View File

@ -151,6 +151,7 @@ class StableDiffusionXLInstructPix2PixPipeline(
watermark output images. If not defined, it will default to True if the package is installed, otherwise no
watermarker will be used.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]

1
src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py
View File

@ -152,6 +152,7 @@ class StableDiffusionAdapterPipeline(DiffusionPipeline):
feature_extractor ([`CLIPFeatureExtractor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->adapter->unet->vae"
_optional_components = ["safety_checker", "feature_extractor"]

1
src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py
View File

@ -159,6 +159,7 @@ class StableDiffusionXLAdapterPipeline(
feature_extractor ([`CLIPFeatureExtractor`]):
Model that extracts features from generated images to be used as inputs for the `safety_checker`.
"""
model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
_optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]

1
src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth.py
View File

@ -96,6 +96,7 @@ class TextToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lora
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
def __init__(

1
src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py
View File

@ -158,6 +158,7 @@ class VideoToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lor
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
"""
model_cpu_offload_seq = "text_encoder->unet->vae"
def __init__(

1
src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero.py
View File

@ -183,6 +183,7 @@ class TextToVideoPipelineOutput(BaseOutput):
List indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content or
`None` if safety checking could not be performed.
"""
images: Union[List[PIL.Image.Image], np.ndarray]
nsfw_content_detected: Optional[List[bool]]

Some files were not shown because too many files have changed in this diff Show More