InvokeAI/invokeai/app/invocations/sdxl.py

727 lines
33 KiB
Python
Raw Normal View History

import torch
import inspect
from tqdm import tqdm
from typing import List, Literal, Optional, Union
from pydantic import Field, validator
2023-07-31 20:18:02 +00:00
from ...backend.model_management import ModelType, SubModelType, ModelPatcher
2023-07-20 15:45:54 +00:00
from invokeai.app.util.step_callback import stable_diffusion_xl_step_callback
2023-07-27 14:54:01 +00:00
from .baseinvocation import BaseInvocation, BaseInvocationOutput, InvocationConfig, InvocationContext
from .model import UNetField, ClipField, VaeField, MainModelField, ModelInfo
from .compel import ConditioningField
from .latent import LatentsField, SAMPLER_NAME_VALUES, LatentsOutput, get_scheduler, build_latents_output
2023-07-27 14:54:01 +00:00
class SDXLModelLoaderOutput(BaseInvocationOutput):
"""SDXL base model loader output"""
# fmt: off
type: Literal["sdxl_model_loader_output"] = "sdxl_model_loader_output"
unet: UNetField = Field(default=None, description="UNet submodel")
clip: ClipField = Field(default=None, description="Tokenizer and text_encoder submodels")
2023-07-16 16:36:38 +00:00
clip2: ClipField = Field(default=None, description="Tokenizer and text_encoder submodels")
vae: VaeField = Field(default=None, description="Vae submodel")
# fmt: on
2023-07-27 14:54:01 +00:00
2023-07-16 16:36:38 +00:00
class SDXLRefinerModelLoaderOutput(BaseInvocationOutput):
"""SDXL refiner model loader output"""
2023-07-27 14:54:01 +00:00
# fmt: off
type: Literal["sdxl_refiner_model_loader_output"] = "sdxl_refiner_model_loader_output"
2023-07-16 16:36:38 +00:00
unet: UNetField = Field(default=None, description="UNet submodel")
clip2: ClipField = Field(default=None, description="Tokenizer and text_encoder submodels")
vae: VaeField = Field(default=None, description="Vae submodel")
# fmt: on
2023-07-27 14:54:01 +00:00
# fmt: on
class SDXLModelLoaderInvocation(BaseInvocation):
"""Loads an sdxl base model, outputting its submodels."""
type: Literal["sdxl_model_loader"] = "sdxl_model_loader"
model: MainModelField = Field(description="The model to load")
# TODO: precision?
# Schema customisation
class Config(InvocationConfig):
schema_extra = {
"ui": {
"title": "SDXL Model Loader",
"tags": ["model", "loader", "sdxl"],
"type_hints": {"model": "model"},
},
}
def invoke(self, context: InvocationContext) -> SDXLModelLoaderOutput:
base_model = self.model.base_model
model_name = self.model.model_name
model_type = ModelType.Main
# TODO: not found exceptions
if not context.services.model_manager.model_exists(
model_name=model_name,
base_model=base_model,
model_type=model_type,
):
raise Exception(f"Unknown {base_model} {model_type} model: {model_name}")
2023-07-16 16:36:38 +00:00
return SDXLModelLoaderOutput(
unet=UNetField(
unet=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.UNet,
),
scheduler=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Scheduler,
),
loras=[],
),
clip=ClipField(
tokenizer=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Tokenizer,
),
text_encoder=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.TextEncoder,
),
loras=[],
skipped_layers=0,
),
clip2=ClipField(
tokenizer=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Tokenizer2,
),
text_encoder=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.TextEncoder2,
),
loras=[],
skipped_layers=0,
),
vae=VaeField(
vae=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Vae,
),
),
)
2023-07-27 14:54:01 +00:00
2023-07-16 16:36:38 +00:00
class SDXLRefinerModelLoaderInvocation(BaseInvocation):
"""Loads an sdxl refiner model, outputting its submodels."""
2023-07-27 14:54:01 +00:00
type: Literal["sdxl_refiner_model_loader"] = "sdxl_refiner_model_loader"
2023-07-16 16:38:04 +00:00
model: MainModelField = Field(description="The model to load")
# TODO: precision?
# Schema customisation
class Config(InvocationConfig):
schema_extra = {
"ui": {
"title": "SDXL Refiner Model Loader",
"tags": ["model", "loader", "sdxl_refiner"],
"type_hints": {"model": "refiner_model"},
},
}
2023-07-16 16:36:38 +00:00
def invoke(self, context: InvocationContext) -> SDXLRefinerModelLoaderOutput:
base_model = self.model.base_model
model_name = self.model.model_name
model_type = ModelType.Main
2023-07-16 16:36:38 +00:00
# TODO: not found exceptions
if not context.services.model_manager.model_exists(
model_name=model_name,
base_model=base_model,
model_type=model_type,
):
raise Exception(f"Unknown {base_model} {model_type} model: {model_name}")
return SDXLRefinerModelLoaderOutput(
unet=UNetField(
unet=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.UNet,
),
scheduler=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Scheduler,
),
loras=[],
),
clip2=ClipField(
tokenizer=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Tokenizer2,
),
text_encoder=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.TextEncoder2,
),
loras=[],
skipped_layers=0,
),
vae=VaeField(
vae=ModelInfo(
model_name=model_name,
base_model=base_model,
model_type=model_type,
submodel=SubModelType.Vae,
),
),
)
2023-07-27 14:54:01 +00:00
# Text to image
class SDXLTextToLatentsInvocation(BaseInvocation):
"""Generates latents from conditionings."""
type: Literal["t2l_sdxl"] = "t2l_sdxl"
# Inputs
# fmt: off
positive_conditioning: Optional[ConditioningField] = Field(description="Positive conditioning for generation")
negative_conditioning: Optional[ConditioningField] = Field(description="Negative conditioning for generation")
noise: Optional[LatentsField] = Field(description="The noise to use")
steps: int = Field(default=10, gt=0, description="The number of steps to use to generate the image")
cfg_scale: Union[float, List[float]] = Field(default=7.5, ge=1, description="The Classifier-Free Guidance, higher values may result in a result closer to the prompt", )
scheduler: SAMPLER_NAME_VALUES = Field(default="euler", description="The scheduler to use" )
unet: UNetField = Field(default=None, description="UNet submodel")
denoising_end: float = Field(default=1.0, gt=0, le=1, description="")
2023-07-27 14:54:01 +00:00
# control: Union[ControlField, list[ControlField]] = Field(default=None, description="The control to use")
# seamless: bool = Field(default=False, description="Whether or not to generate an image that can tile without seams", )
# seamless_axes: str = Field(default="", description="The axes to tile the image on, 'x' and/or 'y'")
# fmt: on
@validator("cfg_scale")
def ge_one(cls, v):
"""validate that all cfg_scale values are >= 1"""
if isinstance(v, list):
for i in v:
if i < 1:
2023-07-27 14:54:01 +00:00
raise ValueError("cfg_scale must be greater than 1")
else:
if v < 1:
2023-07-27 14:54:01 +00:00
raise ValueError("cfg_scale must be greater than 1")
return v
# Schema customisation
class Config(InvocationConfig):
schema_extra = {
"ui": {
2023-07-18 14:26:45 +00:00
"title": "SDXL Text To Latents",
"tags": ["latents"],
"type_hints": {
2023-07-27 14:54:01 +00:00
"model": "model",
# "cfg_scale": "float",
"cfg_scale": "number",
},
},
}
2023-07-20 15:45:54 +00:00
def dispatch_progress(
self,
context: InvocationContext,
source_node_id: str,
sample,
step,
total_steps,
) -> None:
stable_diffusion_xl_step_callback(
context=context,
node=self.dict(),
source_node_id=source_node_id,
sample=sample,
step=step,
total_steps=total_steps,
)
# based on
# https://github.com/huggingface/diffusers/blob/3ebbaf7c96801271f9e6c21400033b6aa5ffcf29/src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion.py#L375
@torch.no_grad()
def invoke(self, context: InvocationContext) -> LatentsOutput:
2023-07-27 14:54:01 +00:00
graph_execution_state = context.services.graph_execution_manager.get(context.graph_execution_state_id)
2023-07-20 15:45:54 +00:00
source_node_id = graph_execution_state.prepared_source_mapping[self.id]
latents = context.services.latents.get(self.noise.latents_name)
positive_cond_data = context.services.latents.get(self.positive_conditioning.conditioning_name)
prompt_embeds = positive_cond_data.conditionings[0].embeds
pooled_prompt_embeds = positive_cond_data.conditionings[0].pooled_embeds
add_time_ids = positive_cond_data.conditionings[0].add_time_ids
negative_cond_data = context.services.latents.get(self.negative_conditioning.conditioning_name)
negative_prompt_embeds = negative_cond_data.conditionings[0].embeds
negative_pooled_prompt_embeds = negative_cond_data.conditionings[0].pooled_embeds
add_neg_time_ids = negative_cond_data.conditionings[0].add_time_ids
scheduler = get_scheduler(
context=context,
scheduler_info=self.unet.scheduler,
scheduler_name=self.scheduler,
)
num_inference_steps = self.steps
2023-07-31 20:18:02 +00:00
def _lora_loader():
for lora in self.unet.loras:
lora_info = context.services.model_manager.get_model(
**lora.dict(exclude={"weight"}),
context=context,
)
yield (lora_info.context.model, lora.weight)
del lora_info
return
unet_info = context.services.model_manager.get_model(**self.unet.unet.dict(), context=context)
do_classifier_free_guidance = True
cross_attention_kwargs = None
2023-08-03 14:26:52 +00:00
with ModelPatcher.apply_lora_unet(unet_info.context.model, _lora_loader()), unet_info as unet:
scheduler.set_timesteps(num_inference_steps, device=unet.device)
timesteps = scheduler.timesteps
2023-07-28 14:49:29 +00:00
2023-07-28 14:27:49 +00:00
latents = latents.to(device=unet.device, dtype=unet.dtype) * scheduler.init_noise_sigma
extra_step_kwargs = dict()
if "eta" in set(inspect.signature(scheduler.step).parameters.keys()):
extra_step_kwargs.update(
eta=0.0,
)
if "generator" in set(inspect.signature(scheduler.step).parameters.keys()):
extra_step_kwargs.update(
generator=torch.Generator(device=unet.device).manual_seed(0),
)
num_warmup_steps = len(timesteps) - self.steps * scheduler.order
# apply denoising_end
skipped_final_steps = int(round((1 - self.denoising_end) * self.steps))
num_inference_steps = num_inference_steps - skipped_final_steps
timesteps = timesteps[: num_warmup_steps + scheduler.order * num_inference_steps]
if not context.services.configuration.sequential_guidance:
prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
add_text_embeds = torch.cat([negative_pooled_prompt_embeds, pooled_prompt_embeds], dim=0)
add_time_ids = torch.cat([add_neg_time_ids, add_time_ids], dim=0)
prompt_embeds = prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_text_embeds = add_text_embeds.to(device=unet.device, dtype=unet.dtype)
add_time_ids = add_time_ids.to(device=unet.device, dtype=unet.dtype)
latents = latents.to(device=unet.device, dtype=unet.dtype)
2023-07-18 13:51:16 +00:00
with tqdm(total=num_inference_steps) as progress_bar:
for i, t in enumerate(timesteps):
# expand the latents if we are doing classifier free guidance
latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
latent_model_input = scheduler.scale_model_input(latent_model_input, t)
# predict the noise residual
added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
noise_pred = unet(
latent_model_input,
t,
encoder_hidden_states=prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
# perform guidance
if do_classifier_free_guidance:
noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
noise_pred = noise_pred_uncond + self.cfg_scale * (noise_pred_text - noise_pred_uncond)
2023-07-27 14:54:01 +00:00
# del noise_pred_uncond
# del noise_pred_text
2023-07-27 14:54:01 +00:00
# if do_classifier_free_guidance and guidance_rescale > 0.0:
# # Based on 3.4. in https://arxiv.org/pdf/2305.08891.pdf
# noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=guidance_rescale)
# compute the previous noisy sample x_t -> x_t-1
latents = scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
# call the callback, if provided
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % scheduler.order == 0):
progress_bar.update()
2023-07-20 15:45:54 +00:00
self.dispatch_progress(context, source_node_id, latents, i, num_inference_steps)
2023-07-27 14:54:01 +00:00
# if callback is not None and i % callback_steps == 0:
# callback(i, t, latents)
else:
negative_pooled_prompt_embeds = negative_pooled_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
negative_prompt_embeds = negative_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_neg_time_ids = add_neg_time_ids.to(device=unet.device, dtype=unet.dtype)
pooled_prompt_embeds = pooled_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
prompt_embeds = prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_time_ids = add_time_ids.to(device=unet.device, dtype=unet.dtype)
latents = latents.to(device=unet.device, dtype=unet.dtype)
2023-07-18 13:51:16 +00:00
with tqdm(total=num_inference_steps) as progress_bar:
for i, t in enumerate(timesteps):
# expand the latents if we are doing classifier free guidance
2023-07-27 14:54:01 +00:00
# latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
latent_model_input = scheduler.scale_model_input(latents, t)
2023-07-27 14:54:01 +00:00
# import gc
# gc.collect()
# torch.cuda.empty_cache()
# predict the noise residual
added_cond_kwargs = {"text_embeds": negative_pooled_prompt_embeds, "time_ids": add_neg_time_ids}
noise_pred_uncond = unet(
latent_model_input,
t,
encoder_hidden_states=negative_prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
added_cond_kwargs = {"text_embeds": pooled_prompt_embeds, "time_ids": add_time_ids}
noise_pred_text = unet(
latent_model_input,
t,
encoder_hidden_states=prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
# perform guidance
noise_pred = noise_pred_uncond + self.cfg_scale * (noise_pred_text - noise_pred_uncond)
2023-07-27 14:54:01 +00:00
# del noise_pred_text
# del noise_pred_uncond
# import gc
# gc.collect()
# torch.cuda.empty_cache()
2023-07-27 14:54:01 +00:00
# if do_classifier_free_guidance and guidance_rescale > 0.0:
# # Based on 3.4. in https://arxiv.org/pdf/2305.08891.pdf
# noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=guidance_rescale)
# compute the previous noisy sample x_t -> x_t-1
latents = scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
2023-07-27 14:54:01 +00:00
# del noise_pred
# import gc
# gc.collect()
# torch.cuda.empty_cache()
# call the callback, if provided
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % scheduler.order == 0):
progress_bar.update()
2023-07-20 15:45:54 +00:00
self.dispatch_progress(context, source_node_id, latents, i, num_inference_steps)
2023-07-27 14:54:01 +00:00
# if callback is not None and i % callback_steps == 0:
# callback(i, t, latents)
#################
latents = latents.to("cpu")
torch.cuda.empty_cache()
2023-07-27 14:54:01 +00:00
name = f"{context.graph_execution_state_id}__{self.id}"
context.services.latents.save(name, latents)
return build_latents_output(latents_name=name, latents=latents)
2023-07-27 14:54:01 +00:00
class SDXLLatentsToLatentsInvocation(BaseInvocation):
"""Generates latents from conditionings."""
type: Literal["l2l_sdxl"] = "l2l_sdxl"
# Inputs
# fmt: off
positive_conditioning: Optional[ConditioningField] = Field(description="Positive conditioning for generation")
negative_conditioning: Optional[ConditioningField] = Field(description="Negative conditioning for generation")
noise: Optional[LatentsField] = Field(description="The noise to use")
steps: int = Field(default=10, gt=0, description="The number of steps to use to generate the image")
cfg_scale: Union[float, List[float]] = Field(default=7.5, ge=1, description="The Classifier-Free Guidance, higher values may result in a result closer to the prompt", )
scheduler: SAMPLER_NAME_VALUES = Field(default="euler", description="The scheduler to use" )
unet: UNetField = Field(default=None, description="UNet submodel")
latents: Optional[LatentsField] = Field(description="Initial latents")
denoising_start: float = Field(default=0.0, ge=0, le=1, description="")
denoising_end: float = Field(default=1.0, ge=0, le=1, description="")
2023-07-27 14:54:01 +00:00
# control: Union[ControlField, list[ControlField]] = Field(default=None, description="The control to use")
# seamless: bool = Field(default=False, description="Whether or not to generate an image that can tile without seams", )
# seamless_axes: str = Field(default="", description="The axes to tile the image on, 'x' and/or 'y'")
# fmt: on
@validator("cfg_scale")
def ge_one(cls, v):
"""validate that all cfg_scale values are >= 1"""
if isinstance(v, list):
for i in v:
if i < 1:
2023-07-27 14:54:01 +00:00
raise ValueError("cfg_scale must be greater than 1")
else:
if v < 1:
2023-07-27 14:54:01 +00:00
raise ValueError("cfg_scale must be greater than 1")
return v
# Schema customisation
class Config(InvocationConfig):
schema_extra = {
"ui": {
2023-07-18 14:26:45 +00:00
"title": "SDXL Latents to Latents",
"tags": ["latents"],
"type_hints": {
2023-07-27 14:54:01 +00:00
"model": "model",
# "cfg_scale": "float",
"cfg_scale": "number",
},
},
}
2023-07-20 15:45:54 +00:00
def dispatch_progress(
self,
context: InvocationContext,
source_node_id: str,
sample,
step,
total_steps,
) -> None:
stable_diffusion_xl_step_callback(
context=context,
node=self.dict(),
source_node_id=source_node_id,
sample=sample,
step=step,
total_steps=total_steps,
)
# based on
# https://github.com/huggingface/diffusers/blob/3ebbaf7c96801271f9e6c21400033b6aa5ffcf29/src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion.py#L375
@torch.no_grad()
def invoke(self, context: InvocationContext) -> LatentsOutput:
2023-07-27 14:54:01 +00:00
graph_execution_state = context.services.graph_execution_manager.get(context.graph_execution_state_id)
2023-07-20 15:45:54 +00:00
source_node_id = graph_execution_state.prepared_source_mapping[self.id]
latents = context.services.latents.get(self.latents.latents_name)
positive_cond_data = context.services.latents.get(self.positive_conditioning.conditioning_name)
prompt_embeds = positive_cond_data.conditionings[0].embeds
pooled_prompt_embeds = positive_cond_data.conditionings[0].pooled_embeds
add_time_ids = positive_cond_data.conditionings[0].add_time_ids
negative_cond_data = context.services.latents.get(self.negative_conditioning.conditioning_name)
negative_prompt_embeds = negative_cond_data.conditionings[0].embeds
negative_pooled_prompt_embeds = negative_cond_data.conditionings[0].pooled_embeds
add_neg_time_ids = negative_cond_data.conditionings[0].add_time_ids
scheduler = get_scheduler(
context=context,
scheduler_info=self.unet.scheduler,
scheduler_name=self.scheduler,
)
unet_info = context.services.model_manager.get_model(
**self.unet.unet.dict(),
context=context,
)
2023-08-03 16:04:44 +00:00
def _lora_loader():
for lora in self.unet.loras:
lora_info = context.services.model_manager.get_model(
**lora.dict(exclude={"weight"}),
context=context,
)
yield (lora_info.context.model, lora.weight)
del lora_info
return
do_classifier_free_guidance = True
cross_attention_kwargs = None
2023-08-03 16:04:44 +00:00
with ModelPatcher.apply_lora_unet(unet_info.context.model, _lora_loader()), unet_info as unet:
# apply denoising_start
num_inference_steps = self.steps
2023-07-28 06:40:23 +00:00
scheduler.set_timesteps(num_inference_steps, device=unet.device)
t_start = int(round(self.denoising_start * num_inference_steps))
timesteps = scheduler.timesteps[t_start * scheduler.order :]
num_inference_steps = num_inference_steps - t_start
# apply noise(if provided)
if self.noise is not None and timesteps.shape[0] > 0:
noise = context.services.latents.get(self.noise.latents_name)
latents = scheduler.add_noise(latents, noise, timesteps[:1])
del noise
# apply scheduler extra args
extra_step_kwargs = dict()
if "eta" in set(inspect.signature(scheduler.step).parameters.keys()):
extra_step_kwargs.update(
eta=0.0,
)
if "generator" in set(inspect.signature(scheduler.step).parameters.keys()):
extra_step_kwargs.update(
generator=torch.Generator(device=unet.device).manual_seed(0),
)
num_warmup_steps = max(len(timesteps) - num_inference_steps * scheduler.order, 0)
# apply denoising_end
skipped_final_steps = int(round((1 - self.denoising_end) * self.steps))
num_inference_steps = num_inference_steps - skipped_final_steps
timesteps = timesteps[: num_warmup_steps + scheduler.order * num_inference_steps]
if not context.services.configuration.sequential_guidance:
prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
add_text_embeds = torch.cat([negative_pooled_prompt_embeds, pooled_prompt_embeds], dim=0)
add_time_ids = torch.cat([add_neg_time_ids, add_time_ids], dim=0)
prompt_embeds = prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_text_embeds = add_text_embeds.to(device=unet.device, dtype=unet.dtype)
add_time_ids = add_time_ids.to(device=unet.device, dtype=unet.dtype)
latents = latents.to(device=unet.device, dtype=unet.dtype)
with tqdm(total=num_inference_steps) as progress_bar:
for i, t in enumerate(timesteps):
# expand the latents if we are doing classifier free guidance
latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
latent_model_input = scheduler.scale_model_input(latent_model_input, t)
# predict the noise residual
added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
noise_pred = unet(
latent_model_input,
t,
encoder_hidden_states=prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
# perform guidance
if do_classifier_free_guidance:
noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
noise_pred = noise_pred_uncond + self.cfg_scale * (noise_pred_text - noise_pred_uncond)
2023-07-27 14:54:01 +00:00
# del noise_pred_uncond
# del noise_pred_text
2023-07-27 14:54:01 +00:00
# if do_classifier_free_guidance and guidance_rescale > 0.0:
# # Based on 3.4. in https://arxiv.org/pdf/2305.08891.pdf
# noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=guidance_rescale)
# compute the previous noisy sample x_t -> x_t-1
latents = scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
# call the callback, if provided
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % scheduler.order == 0):
progress_bar.update()
2023-07-20 15:45:54 +00:00
self.dispatch_progress(context, source_node_id, latents, i, num_inference_steps)
2023-07-27 14:54:01 +00:00
# if callback is not None and i % callback_steps == 0:
# callback(i, t, latents)
else:
negative_pooled_prompt_embeds = negative_pooled_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
negative_prompt_embeds = negative_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_neg_time_ids = add_neg_time_ids.to(device=unet.device, dtype=unet.dtype)
pooled_prompt_embeds = pooled_prompt_embeds.to(device=unet.device, dtype=unet.dtype)
prompt_embeds = prompt_embeds.to(device=unet.device, dtype=unet.dtype)
add_time_ids = add_time_ids.to(device=unet.device, dtype=unet.dtype)
latents = latents.to(device=unet.device, dtype=unet.dtype)
with tqdm(total=num_inference_steps) as progress_bar:
for i, t in enumerate(timesteps):
# expand the latents if we are doing classifier free guidance
2023-07-27 14:54:01 +00:00
# latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
latent_model_input = scheduler.scale_model_input(latents, t)
2023-07-27 14:54:01 +00:00
# import gc
# gc.collect()
# torch.cuda.empty_cache()
# predict the noise residual
added_cond_kwargs = {"text_embeds": negative_pooled_prompt_embeds, "time_ids": add_time_ids}
noise_pred_uncond = unet(
latent_model_input,
t,
encoder_hidden_states=negative_prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
added_cond_kwargs = {"text_embeds": pooled_prompt_embeds, "time_ids": add_time_ids}
noise_pred_text = unet(
latent_model_input,
t,
encoder_hidden_states=prompt_embeds,
cross_attention_kwargs=cross_attention_kwargs,
added_cond_kwargs=added_cond_kwargs,
return_dict=False,
)[0]
# perform guidance
noise_pred = noise_pred_uncond + self.cfg_scale * (noise_pred_text - noise_pred_uncond)
2023-07-27 14:54:01 +00:00
# del noise_pred_text
# del noise_pred_uncond
# import gc
# gc.collect()
# torch.cuda.empty_cache()
2023-07-27 14:54:01 +00:00
# if do_classifier_free_guidance and guidance_rescale > 0.0:
# # Based on 3.4. in https://arxiv.org/pdf/2305.08891.pdf
# noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=guidance_rescale)
# compute the previous noisy sample x_t -> x_t-1
latents = scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
2023-07-27 14:54:01 +00:00
# del noise_pred
# import gc
# gc.collect()
# torch.cuda.empty_cache()
# call the callback, if provided
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % scheduler.order == 0):
progress_bar.update()
2023-07-20 15:45:54 +00:00
self.dispatch_progress(context, source_node_id, latents, i, num_inference_steps)
2023-07-27 14:54:01 +00:00
# if callback is not None and i % callback_steps == 0:
# callback(i, t, latents)
#################
latents = latents.to("cpu")
torch.cuda.empty_cache()
2023-07-27 14:54:01 +00:00
name = f"{context.graph_execution_state_id}__{self.id}"
context.services.latents.save(name, latents)
return build_latents_output(latents_name=name, latents=latents)