Diffusers Bot
commited on
Upload folder using huggingface_hub
Browse files- main/fresco_v2v.py +5 -5
- main/gluegen.py +4 -4
- main/instaflow_one_step.py +9 -5
- main/ip_adapter_face_id.py +11 -6
- main/kohya_hires_fix.py +2 -2
- main/latent_consistency_interpolate.py +10 -6
- main/llm_grounded_diffusion.py +9 -4
- main/lpw_stable_diffusion.py +6 -2
- main/lpw_stable_diffusion_xl.py +10 -5
- main/pipeline_animatediff_controlnet.py +10 -6
- main/pipeline_animatediff_img2video.py +10 -6
- main/pipeline_demofusion_sdxl.py +7 -3
- main/pipeline_fabric.py +2 -2
- main/pipeline_prompt2prompt.py +6 -6
- main/pipeline_stable_diffusion_boxdiff.py +11 -6
- main/pipeline_stable_diffusion_pag.py +11 -6
- main/pipeline_stable_diffusion_upscale_ldm3d.py +6 -6
- main/pipeline_stable_diffusion_xl_controlnet_adapter_inpaint.py +2 -2
- main/pipeline_stable_diffusion_xl_differential_img2img.py +2 -2
- main/sde_drag.py +2 -2
- main/stable_diffusion_ipex.py +2 -2
- main/stable_diffusion_reference.py +11 -6
- main/stable_diffusion_repaint.py +4 -4
main/fresco_v2v.py
CHANGED
@@ -26,7 +26,7 @@ from gmflow.gmflow import GMFlow
|
|
26 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
27 |
|
28 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
29 |
-
from diffusers.loaders import
|
30 |
from diffusers.models import AutoencoderKL, ControlNetModel, ImageProjection, UNet2DConditionModel
|
31 |
from diffusers.models.attention_processor import AttnProcessor2_0
|
32 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
@@ -1252,8 +1252,8 @@ class FrescoV2VPipeline(StableDiffusionControlNetImg2ImgPipeline):
|
|
1252 |
|
1253 |
The pipeline also inherits the following loading methods:
|
1254 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
1255 |
-
- [`~loaders.
|
1256 |
-
- [`~loaders.
|
1257 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
1258 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
1259 |
|
@@ -1456,7 +1456,7 @@ class FrescoV2VPipeline(StableDiffusionControlNetImg2ImgPipeline):
|
|
1456 |
"""
|
1457 |
# set lora scale so that monkey patched LoRA
|
1458 |
# function of text encoder can correctly access it
|
1459 |
-
if lora_scale is not None and isinstance(self,
|
1460 |
self._lora_scale = lora_scale
|
1461 |
|
1462 |
# dynamically adjust the LoRA scale
|
@@ -1588,7 +1588,7 @@ class FrescoV2VPipeline(StableDiffusionControlNetImg2ImgPipeline):
|
|
1588 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
1589 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
1590 |
|
1591 |
-
if isinstance(self,
|
1592 |
# Retrieve the original scale by scaling back the LoRA layers
|
1593 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
1594 |
|
|
|
26 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
27 |
|
28 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
29 |
+
from diffusers.loaders import StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
30 |
from diffusers.models import AutoencoderKL, ControlNetModel, ImageProjection, UNet2DConditionModel
|
31 |
from diffusers.models.attention_processor import AttnProcessor2_0
|
32 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
|
|
1252 |
|
1253 |
The pipeline also inherits the following loading methods:
|
1254 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
1255 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
1256 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
1257 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
1258 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
1259 |
|
|
|
1456 |
"""
|
1457 |
# set lora scale so that monkey patched LoRA
|
1458 |
# function of text encoder can correctly access it
|
1459 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
1460 |
self._lora_scale = lora_scale
|
1461 |
|
1462 |
# dynamically adjust the LoRA scale
|
|
|
1588 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
1589 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
1590 |
|
1591 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
1592 |
# Retrieve the original scale by scaling back the LoRA layers
|
1593 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
1594 |
|
main/gluegen.py
CHANGED
@@ -7,7 +7,7 @@ from transformers import AutoModel, AutoTokenizer, CLIPImageProcessor
|
|
7 |
|
8 |
from diffusers import DiffusionPipeline
|
9 |
from diffusers.image_processor import VaeImageProcessor
|
10 |
-
from diffusers.loaders import
|
11 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
12 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
13 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
@@ -194,7 +194,7 @@ def retrieve_timesteps(
|
|
194 |
return timesteps, num_inference_steps
|
195 |
|
196 |
|
197 |
-
class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin,
|
198 |
def __init__(
|
199 |
self,
|
200 |
vae: AutoencoderKL,
|
@@ -290,7 +290,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
|
|
290 |
"""
|
291 |
# set lora scale so that monkey patched LoRA
|
292 |
# function of text encoder can correctly access it
|
293 |
-
if lora_scale is not None and isinstance(self,
|
294 |
self._lora_scale = lora_scale
|
295 |
|
296 |
# dynamically adjust the LoRA scale
|
@@ -424,7 +424,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
|
|
424 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
425 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
426 |
|
427 |
-
if isinstance(self,
|
428 |
# Retrieve the original scale by scaling back the LoRA layers
|
429 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
430 |
|
|
|
7 |
|
8 |
from diffusers import DiffusionPipeline
|
9 |
from diffusers.image_processor import VaeImageProcessor
|
10 |
+
from diffusers.loaders import StableDiffusionLoraLoaderMixin
|
11 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
12 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
13 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
|
|
194 |
return timesteps, num_inference_steps
|
195 |
|
196 |
|
197 |
+
class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, StableDiffusionLoraLoaderMixin):
|
198 |
def __init__(
|
199 |
self,
|
200 |
vae: AutoencoderKL,
|
|
|
290 |
"""
|
291 |
# set lora scale so that monkey patched LoRA
|
292 |
# function of text encoder can correctly access it
|
293 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
294 |
self._lora_scale = lora_scale
|
295 |
|
296 |
# dynamically adjust the LoRA scale
|
|
|
424 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
425 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
426 |
|
427 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
428 |
# Retrieve the original scale by scaling back the LoRA layers
|
429 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
430 |
|
main/instaflow_one_step.py
CHANGED
@@ -21,7 +21,7 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
|
21 |
|
22 |
from diffusers.configuration_utils import FrozenDict
|
23 |
from diffusers.image_processor import VaeImageProcessor
|
24 |
-
from diffusers.loaders import FromSingleFileMixin,
|
25 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
26 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
27 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
@@ -53,7 +53,11 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
|
|
53 |
|
54 |
|
55 |
class InstaFlowPipeline(
|
56 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
57 |
):
|
58 |
r"""
|
59 |
Pipeline for text-to-image generation using Rectified Flow and Euler discretization.
|
@@ -64,8 +68,8 @@ class InstaFlowPipeline(
|
|
64 |
|
65 |
The pipeline also inherits the following loading methods:
|
66 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
67 |
-
- [`~loaders.
|
68 |
-
- [`~loaders.
|
69 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
70 |
|
71 |
Args:
|
@@ -251,7 +255,7 @@ class InstaFlowPipeline(
|
|
251 |
"""
|
252 |
# set lora scale so that monkey patched LoRA
|
253 |
# function of text encoder can correctly access it
|
254 |
-
if lora_scale is not None and isinstance(self,
|
255 |
self._lora_scale = lora_scale
|
256 |
|
257 |
# dynamically adjust the LoRA scale
|
|
|
21 |
|
22 |
from diffusers.configuration_utils import FrozenDict
|
23 |
from diffusers.image_processor import VaeImageProcessor
|
24 |
+
from diffusers.loaders import FromSingleFileMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
25 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
26 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
27 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
|
|
53 |
|
54 |
|
55 |
class InstaFlowPipeline(
|
56 |
+
DiffusionPipeline,
|
57 |
+
StableDiffusionMixin,
|
58 |
+
TextualInversionLoaderMixin,
|
59 |
+
StableDiffusionLoraLoaderMixin,
|
60 |
+
FromSingleFileMixin,
|
61 |
):
|
62 |
r"""
|
63 |
Pipeline for text-to-image generation using Rectified Flow and Euler discretization.
|
|
|
68 |
|
69 |
The pipeline also inherits the following loading methods:
|
70 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
71 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
72 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
73 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
74 |
|
75 |
Args:
|
|
|
255 |
"""
|
256 |
# set lora scale so that monkey patched LoRA
|
257 |
# function of text encoder can correctly access it
|
258 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
259 |
self._lora_scale = lora_scale
|
260 |
|
261 |
# dynamically adjust the LoRA scale
|
main/ip_adapter_face_id.py
CHANGED
@@ -24,7 +24,12 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPV
|
|
24 |
|
25 |
from diffusers.configuration_utils import FrozenDict
|
26 |
from diffusers.image_processor import VaeImageProcessor
|
27 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
28 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
29 |
from diffusers.models.attention_processor import (
|
30 |
AttnProcessor,
|
@@ -130,7 +135,7 @@ class IPAdapterFaceIDStableDiffusionPipeline(
|
|
130 |
DiffusionPipeline,
|
131 |
StableDiffusionMixin,
|
132 |
TextualInversionLoaderMixin,
|
133 |
-
|
134 |
IPAdapterMixin,
|
135 |
FromSingleFileMixin,
|
136 |
):
|
@@ -142,8 +147,8 @@ class IPAdapterFaceIDStableDiffusionPipeline(
|
|
142 |
|
143 |
The pipeline also inherits the following loading methods:
|
144 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
145 |
-
- [`~loaders.
|
146 |
-
- [`~loaders.
|
147 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
148 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
149 |
|
@@ -518,7 +523,7 @@ class IPAdapterFaceIDStableDiffusionPipeline(
|
|
518 |
"""
|
519 |
# set lora scale so that monkey patched LoRA
|
520 |
# function of text encoder can correctly access it
|
521 |
-
if lora_scale is not None and isinstance(self,
|
522 |
self._lora_scale = lora_scale
|
523 |
|
524 |
# dynamically adjust the LoRA scale
|
@@ -650,7 +655,7 @@ class IPAdapterFaceIDStableDiffusionPipeline(
|
|
650 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
651 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
652 |
|
653 |
-
if isinstance(self,
|
654 |
# Retrieve the original scale by scaling back the LoRA layers
|
655 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
656 |
|
|
|
24 |
|
25 |
from diffusers.configuration_utils import FrozenDict
|
26 |
from diffusers.image_processor import VaeImageProcessor
|
27 |
+
from diffusers.loaders import (
|
28 |
+
FromSingleFileMixin,
|
29 |
+
IPAdapterMixin,
|
30 |
+
StableDiffusionLoraLoaderMixin,
|
31 |
+
TextualInversionLoaderMixin,
|
32 |
+
)
|
33 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
34 |
from diffusers.models.attention_processor import (
|
35 |
AttnProcessor,
|
|
|
135 |
DiffusionPipeline,
|
136 |
StableDiffusionMixin,
|
137 |
TextualInversionLoaderMixin,
|
138 |
+
StableDiffusionLoraLoaderMixin,
|
139 |
IPAdapterMixin,
|
140 |
FromSingleFileMixin,
|
141 |
):
|
|
|
147 |
|
148 |
The pipeline also inherits the following loading methods:
|
149 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
150 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
151 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
152 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
153 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
154 |
|
|
|
523 |
"""
|
524 |
# set lora scale so that monkey patched LoRA
|
525 |
# function of text encoder can correctly access it
|
526 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
527 |
self._lora_scale = lora_scale
|
528 |
|
529 |
# dynamically adjust the LoRA scale
|
|
|
655 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
656 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
657 |
|
658 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
659 |
# Retrieve the original scale by scaling back the LoRA layers
|
660 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
661 |
|
main/kohya_hires_fix.py
CHANGED
@@ -395,8 +395,8 @@ class StableDiffusionHighResFixPipeline(StableDiffusionPipeline):
|
|
395 |
|
396 |
The pipeline also inherits the following loading methods:
|
397 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
398 |
-
- [`~loaders.
|
399 |
-
- [`~loaders.
|
400 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
401 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
402 |
|
|
|
395 |
|
396 |
The pipeline also inherits the following loading methods:
|
397 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
398 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
399 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
400 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
401 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
402 |
|
main/latent_consistency_interpolate.py
CHANGED
@@ -6,7 +6,7 @@ import torch
|
|
6 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
7 |
|
8 |
from diffusers.image_processor import VaeImageProcessor
|
9 |
-
from diffusers.loaders import FromSingleFileMixin,
|
10 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
11 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
12 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
@@ -190,7 +190,11 @@ def slerp(
|
|
190 |
|
191 |
|
192 |
class LatentConsistencyModelWalkPipeline(
|
193 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
194 |
):
|
195 |
r"""
|
196 |
Pipeline for text-to-image generation using a latent consistency model.
|
@@ -200,8 +204,8 @@ class LatentConsistencyModelWalkPipeline(
|
|
200 |
|
201 |
The pipeline also inherits the following loading methods:
|
202 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
203 |
-
- [`~loaders.
|
204 |
-
- [`~loaders.
|
205 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
206 |
|
207 |
Args:
|
@@ -317,7 +321,7 @@ class LatentConsistencyModelWalkPipeline(
|
|
317 |
"""
|
318 |
# set lora scale so that monkey patched LoRA
|
319 |
# function of text encoder can correctly access it
|
320 |
-
if lora_scale is not None and isinstance(self,
|
321 |
self._lora_scale = lora_scale
|
322 |
|
323 |
# dynamically adjust the LoRA scale
|
@@ -449,7 +453,7 @@ class LatentConsistencyModelWalkPipeline(
|
|
449 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
450 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
451 |
|
452 |
-
if isinstance(self,
|
453 |
# Retrieve the original scale by scaling back the LoRA layers
|
454 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
455 |
|
|
|
6 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
7 |
|
8 |
from diffusers.image_processor import VaeImageProcessor
|
9 |
+
from diffusers.loaders import FromSingleFileMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
10 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
11 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
12 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
|
|
190 |
|
191 |
|
192 |
class LatentConsistencyModelWalkPipeline(
|
193 |
+
DiffusionPipeline,
|
194 |
+
StableDiffusionMixin,
|
195 |
+
TextualInversionLoaderMixin,
|
196 |
+
StableDiffusionLoraLoaderMixin,
|
197 |
+
FromSingleFileMixin,
|
198 |
):
|
199 |
r"""
|
200 |
Pipeline for text-to-image generation using a latent consistency model.
|
|
|
204 |
|
205 |
The pipeline also inherits the following loading methods:
|
206 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
207 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
208 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
209 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
210 |
|
211 |
Args:
|
|
|
321 |
"""
|
322 |
# set lora scale so that monkey patched LoRA
|
323 |
# function of text encoder can correctly access it
|
324 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
325 |
self._lora_scale = lora_scale
|
326 |
|
327 |
# dynamically adjust the LoRA scale
|
|
|
453 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
454 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
455 |
|
456 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
457 |
# Retrieve the original scale by scaling back the LoRA layers
|
458 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
459 |
|
main/llm_grounded_diffusion.py
CHANGED
@@ -29,7 +29,12 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPV
|
|
29 |
|
30 |
from diffusers.configuration_utils import FrozenDict
|
31 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
32 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
33 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
34 |
from diffusers.models.attention import Attention, GatedSelfAttentionDense
|
35 |
from diffusers.models.attention_processor import AttnProcessor2_0
|
@@ -271,7 +276,7 @@ class LLMGroundedDiffusionPipeline(
|
|
271 |
DiffusionPipeline,
|
272 |
StableDiffusionMixin,
|
273 |
TextualInversionLoaderMixin,
|
274 |
-
|
275 |
IPAdapterMixin,
|
276 |
FromSingleFileMixin,
|
277 |
):
|
@@ -1263,7 +1268,7 @@ class LLMGroundedDiffusionPipeline(
|
|
1263 |
"""
|
1264 |
# set lora scale so that monkey patched LoRA
|
1265 |
# function of text encoder can correctly access it
|
1266 |
-
if lora_scale is not None and isinstance(self,
|
1267 |
self._lora_scale = lora_scale
|
1268 |
|
1269 |
# dynamically adjust the LoRA scale
|
@@ -1397,7 +1402,7 @@ class LLMGroundedDiffusionPipeline(
|
|
1397 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
1398 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
1399 |
|
1400 |
-
if isinstance(self,
|
1401 |
# Retrieve the original scale by scaling back the LoRA layers
|
1402 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
1403 |
|
|
|
29 |
|
30 |
from diffusers.configuration_utils import FrozenDict
|
31 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
32 |
+
from diffusers.loaders import (
|
33 |
+
FromSingleFileMixin,
|
34 |
+
IPAdapterMixin,
|
35 |
+
StableDiffusionLoraLoaderMixin,
|
36 |
+
TextualInversionLoaderMixin,
|
37 |
+
)
|
38 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
39 |
from diffusers.models.attention import Attention, GatedSelfAttentionDense
|
40 |
from diffusers.models.attention_processor import AttnProcessor2_0
|
|
|
276 |
DiffusionPipeline,
|
277 |
StableDiffusionMixin,
|
278 |
TextualInversionLoaderMixin,
|
279 |
+
StableDiffusionLoraLoaderMixin,
|
280 |
IPAdapterMixin,
|
281 |
FromSingleFileMixin,
|
282 |
):
|
|
|
1268 |
"""
|
1269 |
# set lora scale so that monkey patched LoRA
|
1270 |
# function of text encoder can correctly access it
|
1271 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
1272 |
self._lora_scale = lora_scale
|
1273 |
|
1274 |
# dynamically adjust the LoRA scale
|
|
|
1402 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
1403 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
1404 |
|
1405 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
1406 |
# Retrieve the original scale by scaling back the LoRA layers
|
1407 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
1408 |
|
main/lpw_stable_diffusion.py
CHANGED
@@ -11,7 +11,7 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
|
11 |
from diffusers import DiffusionPipeline
|
12 |
from diffusers.configuration_utils import FrozenDict
|
13 |
from diffusers.image_processor import VaeImageProcessor
|
14 |
-
from diffusers.loaders import FromSingleFileMixin,
|
15 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
16 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
17 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker
|
@@ -409,7 +409,11 @@ def preprocess_mask(mask, batch_size, scale_factor=8):
|
|
409 |
|
410 |
|
411 |
class StableDiffusionLongPromptWeightingPipeline(
|
412 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
413 |
):
|
414 |
r"""
|
415 |
Pipeline for text-to-image generation using Stable Diffusion without tokens length limit, and support parsing
|
|
|
11 |
from diffusers import DiffusionPipeline
|
12 |
from diffusers.configuration_utils import FrozenDict
|
13 |
from diffusers.image_processor import VaeImageProcessor
|
14 |
+
from diffusers.loaders import FromSingleFileMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
15 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
16 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
17 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker
|
|
|
409 |
|
410 |
|
411 |
class StableDiffusionLongPromptWeightingPipeline(
|
412 |
+
DiffusionPipeline,
|
413 |
+
StableDiffusionMixin,
|
414 |
+
TextualInversionLoaderMixin,
|
415 |
+
StableDiffusionLoraLoaderMixin,
|
416 |
+
FromSingleFileMixin,
|
417 |
):
|
418 |
r"""
|
419 |
Pipeline for text-to-image generation using Stable Diffusion without tokens length limit, and support parsing
|
main/lpw_stable_diffusion_xl.py
CHANGED
@@ -22,7 +22,12 @@ from transformers import (
|
|
22 |
|
23 |
from diffusers import DiffusionPipeline, StableDiffusionXLPipeline
|
24 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
25 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
26 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
27 |
from diffusers.models.attention_processor import AttnProcessor2_0, XFormersAttnProcessor
|
28 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
@@ -544,7 +549,7 @@ class SDXLLongPromptWeightingPipeline(
|
|
544 |
StableDiffusionMixin,
|
545 |
FromSingleFileMixin,
|
546 |
IPAdapterMixin,
|
547 |
-
|
548 |
TextualInversionLoaderMixin,
|
549 |
):
|
550 |
r"""
|
@@ -556,8 +561,8 @@ class SDXLLongPromptWeightingPipeline(
|
|
556 |
The pipeline also inherits the following loading methods:
|
557 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
558 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
559 |
-
- [`~loaders.
|
560 |
-
- [`~loaders.
|
561 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
562 |
|
563 |
Args:
|
@@ -738,7 +743,7 @@ class SDXLLongPromptWeightingPipeline(
|
|
738 |
|
739 |
# set lora scale so that monkey patched LoRA
|
740 |
# function of text encoder can correctly access it
|
741 |
-
if lora_scale is not None and isinstance(self,
|
742 |
self._lora_scale = lora_scale
|
743 |
|
744 |
if prompt is not None and isinstance(prompt, str):
|
|
|
22 |
|
23 |
from diffusers import DiffusionPipeline, StableDiffusionXLPipeline
|
24 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
25 |
+
from diffusers.loaders import (
|
26 |
+
FromSingleFileMixin,
|
27 |
+
IPAdapterMixin,
|
28 |
+
StableDiffusionLoraLoaderMixin,
|
29 |
+
TextualInversionLoaderMixin,
|
30 |
+
)
|
31 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
32 |
from diffusers.models.attention_processor import AttnProcessor2_0, XFormersAttnProcessor
|
33 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
|
|
549 |
StableDiffusionMixin,
|
550 |
FromSingleFileMixin,
|
551 |
IPAdapterMixin,
|
552 |
+
StableDiffusionLoraLoaderMixin,
|
553 |
TextualInversionLoaderMixin,
|
554 |
):
|
555 |
r"""
|
|
|
561 |
The pipeline also inherits the following loading methods:
|
562 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
563 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
564 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
565 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
566 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
567 |
|
568 |
Args:
|
|
|
743 |
|
744 |
# set lora scale so that monkey patched LoRA
|
745 |
# function of text encoder can correctly access it
|
746 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
747 |
self._lora_scale = lora_scale
|
748 |
|
749 |
if prompt is not None and isinstance(prompt, str):
|
main/pipeline_animatediff_controlnet.py
CHANGED
@@ -22,7 +22,7 @@ from PIL import Image
|
|
22 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
23 |
|
24 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
25 |
-
from diffusers.loaders import IPAdapterMixin,
|
26 |
from diffusers.models import AutoencoderKL, ControlNetModel, ImageProjection, UNet2DConditionModel, UNetMotionModel
|
27 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
28 |
from diffusers.models.unets.unet_motion_model import MotionAdapter
|
@@ -114,7 +114,11 @@ def tensor2vid(video: torch.Tensor, processor, output_type="np"):
|
|
114 |
|
115 |
|
116 |
class AnimateDiffControlNetPipeline(
|
117 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
118 |
):
|
119 |
r"""
|
120 |
Pipeline for text-to-video generation.
|
@@ -124,8 +128,8 @@ class AnimateDiffControlNetPipeline(
|
|
124 |
|
125 |
The pipeline also inherits the following loading methods:
|
126 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
127 |
-
- [`~loaders.
|
128 |
-
- [`~loaders.
|
129 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
130 |
|
131 |
Args:
|
@@ -234,7 +238,7 @@ class AnimateDiffControlNetPipeline(
|
|
234 |
"""
|
235 |
# set lora scale so that monkey patched LoRA
|
236 |
# function of text encoder can correctly access it
|
237 |
-
if lora_scale is not None and isinstance(self,
|
238 |
self._lora_scale = lora_scale
|
239 |
|
240 |
# dynamically adjust the LoRA scale
|
@@ -366,7 +370,7 @@ class AnimateDiffControlNetPipeline(
|
|
366 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
367 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
368 |
|
369 |
-
if isinstance(self,
|
370 |
# Retrieve the original scale by scaling back the LoRA layers
|
371 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
372 |
|
|
|
22 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
23 |
|
24 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
25 |
+
from diffusers.loaders import IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
26 |
from diffusers.models import AutoencoderKL, ControlNetModel, ImageProjection, UNet2DConditionModel, UNetMotionModel
|
27 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
28 |
from diffusers.models.unets.unet_motion_model import MotionAdapter
|
|
|
114 |
|
115 |
|
116 |
class AnimateDiffControlNetPipeline(
|
117 |
+
DiffusionPipeline,
|
118 |
+
StableDiffusionMixin,
|
119 |
+
TextualInversionLoaderMixin,
|
120 |
+
IPAdapterMixin,
|
121 |
+
StableDiffusionLoraLoaderMixin,
|
122 |
):
|
123 |
r"""
|
124 |
Pipeline for text-to-video generation.
|
|
|
128 |
|
129 |
The pipeline also inherits the following loading methods:
|
130 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
131 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
132 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
133 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
134 |
|
135 |
Args:
|
|
|
238 |
"""
|
239 |
# set lora scale so that monkey patched LoRA
|
240 |
# function of text encoder can correctly access it
|
241 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
242 |
self._lora_scale = lora_scale
|
243 |
|
244 |
# dynamically adjust the LoRA scale
|
|
|
370 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
371 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
372 |
|
373 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
374 |
# Retrieve the original scale by scaling back the LoRA layers
|
375 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
376 |
|
main/pipeline_animatediff_img2video.py
CHANGED
@@ -27,7 +27,7 @@ import torch
|
|
27 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
28 |
|
29 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
30 |
-
from diffusers.loaders import IPAdapterMixin,
|
31 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel, UNetMotionModel
|
32 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
33 |
from diffusers.models.unet_motion_model import MotionAdapter
|
@@ -240,7 +240,11 @@ def retrieve_timesteps(
|
|
240 |
|
241 |
|
242 |
class AnimateDiffImgToVideoPipeline(
|
243 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
244 |
):
|
245 |
r"""
|
246 |
Pipeline for image-to-video generation.
|
@@ -250,8 +254,8 @@ class AnimateDiffImgToVideoPipeline(
|
|
250 |
|
251 |
The pipeline also inherits the following loading methods:
|
252 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
253 |
-
- [`~loaders.
|
254 |
-
- [`~loaders.
|
255 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
256 |
|
257 |
Args:
|
@@ -351,7 +355,7 @@ class AnimateDiffImgToVideoPipeline(
|
|
351 |
"""
|
352 |
# set lora scale so that monkey patched LoRA
|
353 |
# function of text encoder can correctly access it
|
354 |
-
if lora_scale is not None and isinstance(self,
|
355 |
self._lora_scale = lora_scale
|
356 |
|
357 |
# dynamically adjust the LoRA scale
|
@@ -483,7 +487,7 @@ class AnimateDiffImgToVideoPipeline(
|
|
483 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
484 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
485 |
|
486 |
-
if isinstance(self,
|
487 |
# Retrieve the original scale by scaling back the LoRA layers
|
488 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
489 |
|
|
|
27 |
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
|
28 |
|
29 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
30 |
+
from diffusers.loaders import IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
31 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel, UNetMotionModel
|
32 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
33 |
from diffusers.models.unet_motion_model import MotionAdapter
|
|
|
240 |
|
241 |
|
242 |
class AnimateDiffImgToVideoPipeline(
|
243 |
+
DiffusionPipeline,
|
244 |
+
StableDiffusionMixin,
|
245 |
+
TextualInversionLoaderMixin,
|
246 |
+
IPAdapterMixin,
|
247 |
+
StableDiffusionLoraLoaderMixin,
|
248 |
):
|
249 |
r"""
|
250 |
Pipeline for image-to-video generation.
|
|
|
254 |
|
255 |
The pipeline also inherits the following loading methods:
|
256 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
257 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
258 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
259 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
260 |
|
261 |
Args:
|
|
|
355 |
"""
|
356 |
# set lora scale so that monkey patched LoRA
|
357 |
# function of text encoder can correctly access it
|
358 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
359 |
self._lora_scale = lora_scale
|
360 |
|
361 |
# dynamically adjust the LoRA scale
|
|
|
487 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
488 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
489 |
|
490 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
491 |
# Retrieve the original scale by scaling back the LoRA layers
|
492 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
493 |
|
main/pipeline_demofusion_sdxl.py
CHANGED
@@ -12,7 +12,7 @@ from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokeniz
|
|
12 |
from diffusers.image_processor import VaeImageProcessor
|
13 |
from diffusers.loaders import (
|
14 |
FromSingleFileMixin,
|
15 |
-
|
16 |
TextualInversionLoaderMixin,
|
17 |
)
|
18 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
@@ -89,7 +89,11 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
|
|
89 |
|
90 |
|
91 |
class DemoFusionSDXLPipeline(
|
92 |
-
DiffusionPipeline,
|
|
|
|
|
|
|
|
|
93 |
):
|
94 |
r"""
|
95 |
Pipeline for text-to-image generation using Stable Diffusion XL.
|
@@ -231,7 +235,7 @@ class DemoFusionSDXLPipeline(
|
|
231 |
|
232 |
# set lora scale so that monkey patched LoRA
|
233 |
# function of text encoder can correctly access it
|
234 |
-
if lora_scale is not None and isinstance(self,
|
235 |
self._lora_scale = lora_scale
|
236 |
|
237 |
# dynamically adjust the LoRA scale
|
|
|
12 |
from diffusers.image_processor import VaeImageProcessor
|
13 |
from diffusers.loaders import (
|
14 |
FromSingleFileMixin,
|
15 |
+
StableDiffusionLoraLoaderMixin,
|
16 |
TextualInversionLoaderMixin,
|
17 |
)
|
18 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
|
|
89 |
|
90 |
|
91 |
class DemoFusionSDXLPipeline(
|
92 |
+
DiffusionPipeline,
|
93 |
+
StableDiffusionMixin,
|
94 |
+
FromSingleFileMixin,
|
95 |
+
StableDiffusionLoraLoaderMixin,
|
96 |
+
TextualInversionLoaderMixin,
|
97 |
):
|
98 |
r"""
|
99 |
Pipeline for text-to-image generation using Stable Diffusion XL.
|
|
|
235 |
|
236 |
# set lora scale so that monkey patched LoRA
|
237 |
# function of text encoder can correctly access it
|
238 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
239 |
self._lora_scale = lora_scale
|
240 |
|
241 |
# dynamically adjust the LoRA scale
|
main/pipeline_fabric.py
CHANGED
@@ -21,7 +21,7 @@ from transformers import CLIPTextModel, CLIPTokenizer
|
|
21 |
from diffusers import AutoencoderKL, UNet2DConditionModel
|
22 |
from diffusers.configuration_utils import FrozenDict
|
23 |
from diffusers.image_processor import VaeImageProcessor
|
24 |
-
from diffusers.loaders import
|
25 |
from diffusers.models.attention import BasicTransformerBlock
|
26 |
from diffusers.models.attention_processor import LoRAAttnProcessor
|
27 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline
|
@@ -222,7 +222,7 @@ class FabricPipeline(DiffusionPipeline):
|
|
222 |
"""
|
223 |
# set lora scale so that monkey patched LoRA
|
224 |
# function of text encoder can correctly access it
|
225 |
-
if lora_scale is not None and isinstance(self,
|
226 |
self._lora_scale = lora_scale
|
227 |
|
228 |
if prompt is not None and isinstance(prompt, str):
|
|
|
21 |
from diffusers import AutoencoderKL, UNet2DConditionModel
|
22 |
from diffusers.configuration_utils import FrozenDict
|
23 |
from diffusers.image_processor import VaeImageProcessor
|
24 |
+
from diffusers.loaders import StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
25 |
from diffusers.models.attention import BasicTransformerBlock
|
26 |
from diffusers.models.attention_processor import LoRAAttnProcessor
|
27 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline
|
|
|
222 |
"""
|
223 |
# set lora scale so that monkey patched LoRA
|
224 |
# function of text encoder can correctly access it
|
225 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
226 |
self._lora_scale = lora_scale
|
227 |
|
228 |
if prompt is not None and isinstance(prompt, str):
|
main/pipeline_prompt2prompt.py
CHANGED
@@ -35,7 +35,7 @@ from diffusers.image_processor import VaeImageProcessor
|
|
35 |
from diffusers.loaders import (
|
36 |
FromSingleFileMixin,
|
37 |
IPAdapterMixin,
|
38 |
-
|
39 |
TextualInversionLoaderMixin,
|
40 |
)
|
41 |
from diffusers.models.attention import Attention
|
@@ -75,7 +75,7 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
|
|
75 |
class Prompt2PromptPipeline(
|
76 |
DiffusionPipeline,
|
77 |
TextualInversionLoaderMixin,
|
78 |
-
|
79 |
IPAdapterMixin,
|
80 |
FromSingleFileMixin,
|
81 |
):
|
@@ -87,8 +87,8 @@ class Prompt2PromptPipeline(
|
|
87 |
|
88 |
The pipeline also inherits the following loading methods:
|
89 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
90 |
-
- [`~loaders.
|
91 |
-
- [`~loaders.
|
92 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
93 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
94 |
|
@@ -286,7 +286,7 @@ class Prompt2PromptPipeline(
|
|
286 |
"""
|
287 |
# set lora scale so that monkey patched LoRA
|
288 |
# function of text encoder can correctly access it
|
289 |
-
if lora_scale is not None and isinstance(self,
|
290 |
self._lora_scale = lora_scale
|
291 |
|
292 |
# dynamically adjust the LoRA scale
|
@@ -420,7 +420,7 @@ class Prompt2PromptPipeline(
|
|
420 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
421 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
422 |
|
423 |
-
if isinstance(self,
|
424 |
# Retrieve the original scale by scaling back the LoRA layers
|
425 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
426 |
|
|
|
35 |
from diffusers.loaders import (
|
36 |
FromSingleFileMixin,
|
37 |
IPAdapterMixin,
|
38 |
+
StableDiffusionLoraLoaderMixin,
|
39 |
TextualInversionLoaderMixin,
|
40 |
)
|
41 |
from diffusers.models.attention import Attention
|
|
|
75 |
class Prompt2PromptPipeline(
|
76 |
DiffusionPipeline,
|
77 |
TextualInversionLoaderMixin,
|
78 |
+
StableDiffusionLoraLoaderMixin,
|
79 |
IPAdapterMixin,
|
80 |
FromSingleFileMixin,
|
81 |
):
|
|
|
87 |
|
88 |
The pipeline also inherits the following loading methods:
|
89 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
90 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
91 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
92 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
93 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
94 |
|
|
|
286 |
"""
|
287 |
# set lora scale so that monkey patched LoRA
|
288 |
# function of text encoder can correctly access it
|
289 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
290 |
self._lora_scale = lora_scale
|
291 |
|
292 |
# dynamically adjust the LoRA scale
|
|
|
420 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
421 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
422 |
|
423 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
424 |
# Retrieve the original scale by scaling back the LoRA layers
|
425 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
426 |
|
main/pipeline_stable_diffusion_boxdiff.py
CHANGED
@@ -27,7 +27,12 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPV
|
|
27 |
|
28 |
from diffusers.configuration_utils import FrozenDict
|
29 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
30 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
31 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
32 |
from diffusers.models.attention_processor import Attention, FusedAttnProcessor2_0
|
33 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
@@ -358,7 +363,7 @@ def retrieve_timesteps(
|
|
358 |
|
359 |
|
360 |
class StableDiffusionBoxDiffPipeline(
|
361 |
-
DiffusionPipeline, TextualInversionLoaderMixin,
|
362 |
):
|
363 |
r"""
|
364 |
Pipeline for text-to-image generation using Stable Diffusion with BoxDiff.
|
@@ -368,8 +373,8 @@ class StableDiffusionBoxDiffPipeline(
|
|
368 |
|
369 |
The pipeline also inherits the following loading methods:
|
370 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
371 |
-
- [`~loaders.
|
372 |
-
- [`~loaders.
|
373 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
374 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
375 |
|
@@ -594,7 +599,7 @@ class StableDiffusionBoxDiffPipeline(
|
|
594 |
"""
|
595 |
# set lora scale so that monkey patched LoRA
|
596 |
# function of text encoder can correctly access it
|
597 |
-
if lora_scale is not None and isinstance(self,
|
598 |
self._lora_scale = lora_scale
|
599 |
|
600 |
# dynamically adjust the LoRA scale
|
@@ -726,7 +731,7 @@ class StableDiffusionBoxDiffPipeline(
|
|
726 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
727 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
728 |
|
729 |
-
if isinstance(self,
|
730 |
# Retrieve the original scale by scaling back the LoRA layers
|
731 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
732 |
|
|
|
27 |
|
28 |
from diffusers.configuration_utils import FrozenDict
|
29 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
30 |
+
from diffusers.loaders import (
|
31 |
+
FromSingleFileMixin,
|
32 |
+
IPAdapterMixin,
|
33 |
+
StableDiffusionLoraLoaderMixin,
|
34 |
+
TextualInversionLoaderMixin,
|
35 |
+
)
|
36 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
37 |
from diffusers.models.attention_processor import Attention, FusedAttnProcessor2_0
|
38 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
|
|
363 |
|
364 |
|
365 |
class StableDiffusionBoxDiffPipeline(
|
366 |
+
DiffusionPipeline, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin, IPAdapterMixin, FromSingleFileMixin
|
367 |
):
|
368 |
r"""
|
369 |
Pipeline for text-to-image generation using Stable Diffusion with BoxDiff.
|
|
|
373 |
|
374 |
The pipeline also inherits the following loading methods:
|
375 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
376 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
377 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
378 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
379 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
380 |
|
|
|
599 |
"""
|
600 |
# set lora scale so that monkey patched LoRA
|
601 |
# function of text encoder can correctly access it
|
602 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
603 |
self._lora_scale = lora_scale
|
604 |
|
605 |
# dynamically adjust the LoRA scale
|
|
|
731 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
732 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
733 |
|
734 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
735 |
# Retrieve the original scale by scaling back the LoRA layers
|
736 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
737 |
|
main/pipeline_stable_diffusion_pag.py
CHANGED
@@ -11,7 +11,12 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPV
|
|
11 |
|
12 |
from diffusers.configuration_utils import FrozenDict
|
13 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
14 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
15 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
16 |
from diffusers.models.attention_processor import Attention, AttnProcessor2_0, FusedAttnProcessor2_0
|
17 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
@@ -328,7 +333,7 @@ def retrieve_timesteps(
|
|
328 |
|
329 |
|
330 |
class StableDiffusionPAGPipeline(
|
331 |
-
DiffusionPipeline, TextualInversionLoaderMixin,
|
332 |
):
|
333 |
r"""
|
334 |
Pipeline for text-to-image generation using Stable Diffusion.
|
@@ -336,8 +341,8 @@ class StableDiffusionPAGPipeline(
|
|
336 |
implemented for all pipelines (downloading, saving, running on a particular device, etc.).
|
337 |
The pipeline also inherits the following loading methods:
|
338 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
339 |
-
- [`~loaders.
|
340 |
-
- [`~loaders.
|
341 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
342 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
343 |
Args:
|
@@ -560,7 +565,7 @@ class StableDiffusionPAGPipeline(
|
|
560 |
"""
|
561 |
# set lora scale so that monkey patched LoRA
|
562 |
# function of text encoder can correctly access it
|
563 |
-
if lora_scale is not None and isinstance(self,
|
564 |
self._lora_scale = lora_scale
|
565 |
|
566 |
# dynamically adjust the LoRA scale
|
@@ -692,7 +697,7 @@ class StableDiffusionPAGPipeline(
|
|
692 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
693 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
694 |
|
695 |
-
if isinstance(self,
|
696 |
# Retrieve the original scale by scaling back the LoRA layers
|
697 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
698 |
|
|
|
11 |
|
12 |
from diffusers.configuration_utils import FrozenDict
|
13 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
14 |
+
from diffusers.loaders import (
|
15 |
+
FromSingleFileMixin,
|
16 |
+
IPAdapterMixin,
|
17 |
+
StableDiffusionLoraLoaderMixin,
|
18 |
+
TextualInversionLoaderMixin,
|
19 |
+
)
|
20 |
from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
|
21 |
from diffusers.models.attention_processor import Attention, AttnProcessor2_0, FusedAttnProcessor2_0
|
22 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
|
|
333 |
|
334 |
|
335 |
class StableDiffusionPAGPipeline(
|
336 |
+
DiffusionPipeline, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin, IPAdapterMixin, FromSingleFileMixin
|
337 |
):
|
338 |
r"""
|
339 |
Pipeline for text-to-image generation using Stable Diffusion.
|
|
|
341 |
implemented for all pipelines (downloading, saving, running on a particular device, etc.).
|
342 |
The pipeline also inherits the following loading methods:
|
343 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
344 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
345 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
346 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
347 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
348 |
Args:
|
|
|
565 |
"""
|
566 |
# set lora scale so that monkey patched LoRA
|
567 |
# function of text encoder can correctly access it
|
568 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
569 |
self._lora_scale = lora_scale
|
570 |
|
571 |
# dynamically adjust the LoRA scale
|
|
|
697 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
698 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
699 |
|
700 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
701 |
# Retrieve the original scale by scaling back the LoRA layers
|
702 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
703 |
|
main/pipeline_stable_diffusion_upscale_ldm3d.py
CHANGED
@@ -22,7 +22,7 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
|
22 |
|
23 |
from diffusers import DiffusionPipeline
|
24 |
from diffusers.image_processor import PipelineDepthInput, PipelineImageInput, VaeImageProcessorLDM3D
|
25 |
-
from diffusers.loaders import FromSingleFileMixin,
|
26 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
27 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
28 |
from diffusers.pipelines.stable_diffusion import StableDiffusionSafetyChecker
|
@@ -69,7 +69,7 @@ EXAMPLE_DOC_STRING = """
|
|
69 |
|
70 |
|
71 |
class StableDiffusionUpscaleLDM3DPipeline(
|
72 |
-
DiffusionPipeline, TextualInversionLoaderMixin,
|
73 |
):
|
74 |
r"""
|
75 |
Pipeline for text-to-image and 3D generation using LDM3D.
|
@@ -79,8 +79,8 @@ class StableDiffusionUpscaleLDM3DPipeline(
|
|
79 |
|
80 |
The pipeline also inherits the following loading methods:
|
81 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
82 |
-
- [`~loaders.
|
83 |
-
- [`~loaders.
|
84 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
85 |
|
86 |
Args:
|
@@ -233,7 +233,7 @@ class StableDiffusionUpscaleLDM3DPipeline(
|
|
233 |
"""
|
234 |
# set lora scale so that monkey patched LoRA
|
235 |
# function of text encoder can correctly access it
|
236 |
-
if lora_scale is not None and isinstance(self,
|
237 |
self._lora_scale = lora_scale
|
238 |
|
239 |
# dynamically adjust the LoRA scale
|
@@ -365,7 +365,7 @@ class StableDiffusionUpscaleLDM3DPipeline(
|
|
365 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
366 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
367 |
|
368 |
-
if isinstance(self,
|
369 |
# Retrieve the original scale by scaling back the LoRA layers
|
370 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
371 |
|
|
|
22 |
|
23 |
from diffusers import DiffusionPipeline
|
24 |
from diffusers.image_processor import PipelineDepthInput, PipelineImageInput, VaeImageProcessorLDM3D
|
25 |
+
from diffusers.loaders import FromSingleFileMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
26 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
27 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
28 |
from diffusers.pipelines.stable_diffusion import StableDiffusionSafetyChecker
|
|
|
69 |
|
70 |
|
71 |
class StableDiffusionUpscaleLDM3DPipeline(
|
72 |
+
DiffusionPipeline, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin, FromSingleFileMixin
|
73 |
):
|
74 |
r"""
|
75 |
Pipeline for text-to-image and 3D generation using LDM3D.
|
|
|
79 |
|
80 |
The pipeline also inherits the following loading methods:
|
81 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
82 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
83 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
84 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
85 |
|
86 |
Args:
|
|
|
233 |
"""
|
234 |
# set lora scale so that monkey patched LoRA
|
235 |
# function of text encoder can correctly access it
|
236 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
237 |
self._lora_scale = lora_scale
|
238 |
|
239 |
# dynamically adjust the LoRA scale
|
|
|
365 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
366 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
367 |
|
368 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
369 |
# Retrieve the original scale by scaling back the LoRA layers
|
370 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
371 |
|
main/pipeline_stable_diffusion_xl_controlnet_adapter_inpaint.py
CHANGED
@@ -33,7 +33,7 @@ from diffusers import DiffusionPipeline
|
|
33 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
34 |
from diffusers.loaders import (
|
35 |
FromSingleFileMixin,
|
36 |
-
|
37 |
StableDiffusionXLLoraLoaderMixin,
|
38 |
TextualInversionLoaderMixin,
|
39 |
)
|
@@ -300,7 +300,7 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
|
|
300 |
|
301 |
|
302 |
class StableDiffusionXLControlNetAdapterInpaintPipeline(
|
303 |
-
DiffusionPipeline, StableDiffusionMixin, FromSingleFileMixin,
|
304 |
):
|
305 |
r"""
|
306 |
Pipeline for text-to-image generation using Stable Diffusion augmented with T2I-Adapter
|
|
|
33 |
from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
|
34 |
from diffusers.loaders import (
|
35 |
FromSingleFileMixin,
|
36 |
+
StableDiffusionLoraLoaderMixin,
|
37 |
StableDiffusionXLLoraLoaderMixin,
|
38 |
TextualInversionLoaderMixin,
|
39 |
)
|
|
|
300 |
|
301 |
|
302 |
class StableDiffusionXLControlNetAdapterInpaintPipeline(
|
303 |
+
DiffusionPipeline, StableDiffusionMixin, FromSingleFileMixin, StableDiffusionLoraLoaderMixin
|
304 |
):
|
305 |
r"""
|
306 |
Pipeline for text-to-image generation using Stable Diffusion augmented with T2I-Adapter
|
main/pipeline_stable_diffusion_xl_differential_img2img.py
CHANGED
@@ -178,11 +178,11 @@ class StableDiffusionXLDifferentialImg2ImgPipeline(
|
|
178 |
|
179 |
In addition the pipeline inherits the following loading methods:
|
180 |
- *Textual-Inversion*: [`loaders.TextualInversionLoaderMixin.load_textual_inversion`]
|
181 |
-
- *LoRA*: [`loaders.
|
182 |
- *Ckpt*: [`loaders.FromSingleFileMixin.from_single_file`]
|
183 |
|
184 |
as well as the following saving methods:
|
185 |
-
- *LoRA*: [`loaders.
|
186 |
|
187 |
Args:
|
188 |
vae ([`AutoencoderKL`]):
|
|
|
178 |
|
179 |
In addition the pipeline inherits the following loading methods:
|
180 |
- *Textual-Inversion*: [`loaders.TextualInversionLoaderMixin.load_textual_inversion`]
|
181 |
+
- *LoRA*: [`loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`]
|
182 |
- *Ckpt*: [`loaders.FromSingleFileMixin.from_single_file`]
|
183 |
|
184 |
as well as the following saving methods:
|
185 |
+
- *LoRA*: [`loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`]
|
186 |
|
187 |
Args:
|
188 |
vae ([`AutoencoderKL`]):
|
main/sde_drag.py
CHANGED
@@ -11,7 +11,7 @@ from tqdm.auto import tqdm
|
|
11 |
from transformers import CLIPTextModel, CLIPTokenizer
|
12 |
|
13 |
from diffusers import AutoencoderKL, DiffusionPipeline, DPMSolverMultistepScheduler, UNet2DConditionModel
|
14 |
-
from diffusers.loaders import AttnProcsLayers,
|
15 |
from diffusers.models.attention_processor import (
|
16 |
AttnAddedKVProcessor,
|
17 |
AttnAddedKVProcessor2_0,
|
@@ -321,7 +321,7 @@ class SdeDragPipeline(DiffusionPipeline):
|
|
321 |
optimizer.zero_grad()
|
322 |
|
323 |
with tempfile.TemporaryDirectory() as save_lora_dir:
|
324 |
-
|
325 |
save_directory=save_lora_dir,
|
326 |
unet_lora_layers=unet_lora_layers,
|
327 |
text_encoder_lora_layers=None,
|
|
|
11 |
from transformers import CLIPTextModel, CLIPTokenizer
|
12 |
|
13 |
from diffusers import AutoencoderKL, DiffusionPipeline, DPMSolverMultistepScheduler, UNet2DConditionModel
|
14 |
+
from diffusers.loaders import AttnProcsLayers, StableDiffusionLoraLoaderMixin
|
15 |
from diffusers.models.attention_processor import (
|
16 |
AttnAddedKVProcessor,
|
17 |
AttnAddedKVProcessor2_0,
|
|
|
321 |
optimizer.zero_grad()
|
322 |
|
323 |
with tempfile.TemporaryDirectory() as save_lora_dir:
|
324 |
+
StableDiffusionLoraLoaderMixin.save_lora_weights(
|
325 |
save_directory=save_lora_dir,
|
326 |
unet_lora_layers=unet_lora_layers,
|
327 |
text_encoder_lora_layers=None,
|
main/stable_diffusion_ipex.py
CHANGED
@@ -21,7 +21,7 @@ from packaging import version
|
|
21 |
from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
|
22 |
|
23 |
from diffusers.configuration_utils import FrozenDict
|
24 |
-
from diffusers.loaders import
|
25 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
26 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
27 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput
|
@@ -61,7 +61,7 @@ EXAMPLE_DOC_STRING = """
|
|
61 |
|
62 |
|
63 |
class StableDiffusionIPEXPipeline(
|
64 |
-
DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin,
|
65 |
):
|
66 |
r"""
|
67 |
Pipeline for text-to-image generation using Stable Diffusion on IPEX.
|
|
|
21 |
from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
|
22 |
|
23 |
from diffusers.configuration_utils import FrozenDict
|
24 |
+
from diffusers.loaders import StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
25 |
from diffusers.models import AutoencoderKL, UNet2DConditionModel
|
26 |
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
|
27 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput
|
|
|
61 |
|
62 |
|
63 |
class StableDiffusionIPEXPipeline(
|
64 |
+
DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin
|
65 |
):
|
66 |
r"""
|
67 |
Pipeline for text-to-image generation using Stable Diffusion on IPEX.
|
main/stable_diffusion_reference.py
CHANGED
@@ -11,7 +11,12 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
|
11 |
from diffusers import AutoencoderKL, DiffusionPipeline, UNet2DConditionModel
|
12 |
from diffusers.configuration_utils import FrozenDict, deprecate
|
13 |
from diffusers.image_processor import VaeImageProcessor
|
14 |
-
from diffusers.loaders import
|
|
|
|
|
|
|
|
|
|
|
15 |
from diffusers.models.attention import BasicTransformerBlock
|
16 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
17 |
from diffusers.models.unets.unet_2d_blocks import CrossAttnDownBlock2D, CrossAttnUpBlock2D, DownBlock2D, UpBlock2D
|
@@ -76,7 +81,7 @@ def torch_dfs(model: torch.nn.Module):
|
|
76 |
|
77 |
|
78 |
class StableDiffusionReferencePipeline(
|
79 |
-
DiffusionPipeline, TextualInversionLoaderMixin,
|
80 |
):
|
81 |
r"""
|
82 |
Pipeline for Stable Diffusion Reference.
|
@@ -86,8 +91,8 @@ class StableDiffusionReferencePipeline(
|
|
86 |
|
87 |
The pipeline also inherits the following loading methods:
|
88 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
89 |
-
- [`~loaders.
|
90 |
-
- [`~loaders.
|
91 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
92 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
93 |
|
@@ -443,7 +448,7 @@ class StableDiffusionReferencePipeline(
|
|
443 |
"""
|
444 |
# set lora scale so that monkey patched LoRA
|
445 |
# function of text encoder can correctly access it
|
446 |
-
if lora_scale is not None and isinstance(self,
|
447 |
self._lora_scale = lora_scale
|
448 |
|
449 |
# dynamically adjust the LoRA scale
|
@@ -575,7 +580,7 @@ class StableDiffusionReferencePipeline(
|
|
575 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
576 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
577 |
|
578 |
-
if isinstance(self,
|
579 |
# Retrieve the original scale by scaling back the LoRA layers
|
580 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
581 |
|
|
|
11 |
from diffusers import AutoencoderKL, DiffusionPipeline, UNet2DConditionModel
|
12 |
from diffusers.configuration_utils import FrozenDict, deprecate
|
13 |
from diffusers.image_processor import VaeImageProcessor
|
14 |
+
from diffusers.loaders import (
|
15 |
+
FromSingleFileMixin,
|
16 |
+
IPAdapterMixin,
|
17 |
+
StableDiffusionLoraLoaderMixin,
|
18 |
+
TextualInversionLoaderMixin,
|
19 |
+
)
|
20 |
from diffusers.models.attention import BasicTransformerBlock
|
21 |
from diffusers.models.lora import adjust_lora_scale_text_encoder
|
22 |
from diffusers.models.unets.unet_2d_blocks import CrossAttnDownBlock2D, CrossAttnUpBlock2D, DownBlock2D, UpBlock2D
|
|
|
81 |
|
82 |
|
83 |
class StableDiffusionReferencePipeline(
|
84 |
+
DiffusionPipeline, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin, IPAdapterMixin, FromSingleFileMixin
|
85 |
):
|
86 |
r"""
|
87 |
Pipeline for Stable Diffusion Reference.
|
|
|
91 |
|
92 |
The pipeline also inherits the following loading methods:
|
93 |
- [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
|
94 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
|
95 |
+
- [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
|
96 |
- [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
|
97 |
- [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
|
98 |
|
|
|
448 |
"""
|
449 |
# set lora scale so that monkey patched LoRA
|
450 |
# function of text encoder can correctly access it
|
451 |
+
if lora_scale is not None and isinstance(self, StableDiffusionLoraLoaderMixin):
|
452 |
self._lora_scale = lora_scale
|
453 |
|
454 |
# dynamically adjust the LoRA scale
|
|
|
580 |
negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
|
581 |
negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
|
582 |
|
583 |
+
if isinstance(self, StableDiffusionLoraLoaderMixin) and USE_PEFT_BACKEND:
|
584 |
# Retrieve the original scale by scaling back the LoRA layers
|
585 |
unscale_lora_layers(self.text_encoder, lora_scale)
|
586 |
|
main/stable_diffusion_repaint.py
CHANGED
@@ -23,7 +23,7 @@ from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
|
|
23 |
|
24 |
from diffusers import AutoencoderKL, DiffusionPipeline, UNet2DConditionModel
|
25 |
from diffusers.configuration_utils import FrozenDict, deprecate
|
26 |
-
from diffusers.loaders import
|
27 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
28 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput
|
29 |
from diffusers.pipelines.stable_diffusion.safety_checker import (
|
@@ -140,7 +140,7 @@ def prepare_mask_and_masked_image(image, mask):
|
|
140 |
|
141 |
|
142 |
class StableDiffusionRepaintPipeline(
|
143 |
-
DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin,
|
144 |
):
|
145 |
r"""
|
146 |
Pipeline for text-guided image inpainting using Stable Diffusion. *This is an experimental feature*.
|
@@ -148,9 +148,9 @@ class StableDiffusionRepaintPipeline(
|
|
148 |
library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
|
149 |
In addition the pipeline inherits the following loading methods:
|
150 |
- *Textual-Inversion*: [`loaders.TextualInversionLoaderMixin.load_textual_inversion`]
|
151 |
-
- *LoRA*: [`loaders.
|
152 |
as well as the following saving methods:
|
153 |
-
- *LoRA*: [`loaders.
|
154 |
Args:
|
155 |
vae ([`AutoencoderKL`]):
|
156 |
Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
|
|
|
23 |
|
24 |
from diffusers import AutoencoderKL, DiffusionPipeline, UNet2DConditionModel
|
25 |
from diffusers.configuration_utils import FrozenDict, deprecate
|
26 |
+
from diffusers.loaders import StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
|
27 |
from diffusers.pipelines.pipeline_utils import StableDiffusionMixin
|
28 |
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput
|
29 |
from diffusers.pipelines.stable_diffusion.safety_checker import (
|
|
|
140 |
|
141 |
|
142 |
class StableDiffusionRepaintPipeline(
|
143 |
+
DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, StableDiffusionLoraLoaderMixin
|
144 |
):
|
145 |
r"""
|
146 |
Pipeline for text-guided image inpainting using Stable Diffusion. *This is an experimental feature*.
|
|
|
148 |
library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
|
149 |
In addition the pipeline inherits the following loading methods:
|
150 |
- *Textual-Inversion*: [`loaders.TextualInversionLoaderMixin.load_textual_inversion`]
|
151 |
+
- *LoRA*: [`loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`]
|
152 |
as well as the following saving methods:
|
153 |
+
- *LoRA*: [`loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`]
|
154 |
Args:
|
155 |
vae ([`AutoencoderKL`]):
|
156 |
Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
|