UltraEdit/diffusers/examples/community/instaflow_one_step.py

# Copyright 2024 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import inspect
from typing import Any, Callable, Dict, List, Optional, Union

import torch
from packaging import version
from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer

from diffusers.configuration_utils import FrozenDict
from diffusers.image_processor import VaeImageProcessor
from diffusers.loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
from diffusers.models import AutoencoderKL, UNet2DConditionModel
from diffusers.models.lora import adjust_lora_scale_text_encoder
from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput
from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
from diffusers.schedulers import KarrasDiffusionSchedulers
from diffusers.utils import (
    deprecate,
    logging,
)
from diffusers.utils.torch_utils import randn_tensor


logger = logging.get_logger(__name__)  # pylint: disable=invalid-name


def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
    """

    Rescale `noise_cfg` according to `guidance_rescale`. Based on findings of [Common Diffusion Noise Schedules and

    Sample Steps are Flawed](https://arxiv.org/pdf/2305.08891.pdf). See Section 3.4

    """
    std_text = noise_pred_text.std(dim=list(range(1, noise_pred_text.ndim)), keepdim=True)
    std_cfg = noise_cfg.std(dim=list(range(1, noise_cfg.ndim)), keepdim=True)
    # rescale the results from guidance (fixes overexposure)
    noise_pred_rescaled = noise_cfg * (std_text / std_cfg)
    # mix with the original results from guidance by factor guidance_rescale to avoid "plain looking" images
    noise_cfg = guidance_rescale * noise_pred_rescaled + (1 - guidance_rescale) * noise_cfg
    return noise_cfg


class InstaFlowPipeline(
    DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, LoraLoaderMixin, FromSingleFileMixin
):
    r"""

    Pipeline for text-to-image generation using Rectified Flow and Euler discretization.

    This customized pipeline is based on StableDiffusionPipeline from the official Diffusers library (0.21.4)



    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods

    implemented for all pipelines (downloading, saving, running on a particular device, etc.).



    The pipeline also inherits the following loading methods:

        - [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings

        - [`~loaders.LoraLoaderMixin.load_lora_weights`] for loading LoRA weights

        - [`~loaders.LoraLoaderMixin.save_lora_weights`] for saving LoRA weights

        - [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files



    Args:

        vae ([`AutoencoderKL`]):

            Variational Auto-Encoder (VAE) model to encode and decode images to and from latent representations.

        text_encoder ([`~transformers.CLIPTextModel`]):

            Frozen text-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).

        tokenizer ([`~transformers.CLIPTokenizer`]):

            A `CLIPTokenizer` to tokenize text.

        unet ([`UNet2DConditionModel`]):

            A `UNet2DConditionModel` to denoise the encoded image latents.

        scheduler ([`SchedulerMixin`]):

            A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of

            [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].

        safety_checker ([`StableDiffusionSafetyChecker`]):

            Classification module that estimates whether generated images could be considered offensive or harmful.

            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details

            about a model's potential harms.

        feature_extractor ([`~transformers.CLIPImageProcessor`]):

            A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.

    """

    model_cpu_offload_seq = "text_encoder->unet->vae"
    _optional_components = ["safety_checker", "feature_extractor"]
    _exclude_from_cpu_offload = ["safety_checker"]

    def __init__(

        self,

        vae: AutoencoderKL,

        text_encoder: CLIPTextModel,

        tokenizer: CLIPTokenizer,

        unet: UNet2DConditionModel,

        scheduler: KarrasDiffusionSchedulers,

        safety_checker: StableDiffusionSafetyChecker,

        feature_extractor: CLIPImageProcessor,

        requires_safety_checker: bool = True,

    ):
        super().__init__()

        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1:
            deprecation_message = (
                f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`"
                f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure "
                "to update the config accordingly as leaving `steps_offset` might led to incorrect results"
                " in future versions. If you have downloaded this checkpoint from the Hugging Face Hub,"
                " it would be very nice if you could open a Pull request for the `scheduler/scheduler_config.json`"
                " file"
            )
            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
            new_config = dict(scheduler.config)
            new_config["steps_offset"] = 1
            scheduler._internal_dict = FrozenDict(new_config)

        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample is True:
            deprecation_message = (
                f"The configuration file of this scheduler: {scheduler} has not set the configuration `clip_sample`."
                " `clip_sample` should be set to False in the configuration file. Please make sure to update the"
                " config accordingly as not setting `clip_sample` in the config might lead to incorrect results in"
                " future versions. If you have downloaded this checkpoint from the Hugging Face Hub, it would be very"
                " nice if you could open a Pull request for the `scheduler/scheduler_config.json` file"
            )
            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
            new_config = dict(scheduler.config)
            new_config["clip_sample"] = False
            scheduler._internal_dict = FrozenDict(new_config)

        if safety_checker is None and requires_safety_checker:
            logger.warning(
                f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
                " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered"
                " results in services or applications open to the public. Both the diffusers team and Hugging Face"
                " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling"
                " it only for use-cases that involve analyzing network behavior or auditing its results. For more"
                " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
            )

        if safety_checker is not None and feature_extractor is None:
            raise ValueError(
                "Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety"
                " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
            )

        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
            version.parse(unet.config._diffusers_version).base_version
        ) < version.parse("0.9.0.dev0")
        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
        if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
            deprecation_message = (
                "The configuration file of the unet has set the default `sample_size` to smaller than"
                " 64 which seems highly unlikely. If your checkpoint is a fine-tuned version of any of the"
                " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
                " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
                " in the config might lead to incorrect results in future versions. If you have downloaded this"
                " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
                " the `unet/config.json` file"
            )
            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
            new_config = dict(unet.config)
            new_config["sample_size"] = 64
            unet._internal_dict = FrozenDict(new_config)

        self.register_modules(
            vae=vae,
            text_encoder=text_encoder,
            tokenizer=tokenizer,
            unet=unet,
            scheduler=scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
        )
        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    def _encode_prompt(

        self,

        prompt,

        device,

        num_images_per_prompt,

        do_classifier_free_guidance,

        negative_prompt=None,

        prompt_embeds: Optional[torch.Tensor] = None,

        negative_prompt_embeds: Optional[torch.Tensor] = None,

        lora_scale: Optional[float] = None,

    ):
        deprecation_message = "`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()` instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
        deprecate("_encode_prompt()", "1.0.0", deprecation_message, standard_warn=False)

        prompt_embeds_tuple = self.encode_prompt(
            prompt=prompt,
            device=device,
            num_images_per_prompt=num_images_per_prompt,
            do_classifier_free_guidance=do_classifier_free_guidance,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            lora_scale=lora_scale,
        )

        # concatenate for backwards comp
        prompt_embeds = torch.cat([prompt_embeds_tuple[1], prompt_embeds_tuple[0]])

        return prompt_embeds

    def encode_prompt(

        self,

        prompt,

        device,

        num_images_per_prompt,

        do_classifier_free_guidance,

        negative_prompt=None,

        prompt_embeds: Optional[torch.Tensor] = None,

        negative_prompt_embeds: Optional[torch.Tensor] = None,

        lora_scale: Optional[float] = None,

    ):
        r"""

        Encodes the prompt into text encoder hidden states.



        Args:

            prompt (`str` or `List[str]`, *optional*):

                prompt to be encoded

            device: (`torch.device`):

                torch device

            num_images_per_prompt (`int`):

                number of images that should be generated per prompt

            do_classifier_free_guidance (`bool`):

                whether to use classifier free guidance or not

            negative_prompt (`str` or `List[str]`, *optional*):

                The prompt or prompts not to guide the image generation. If not defined, one has to pass

                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is

                less than `1`).

            prompt_embeds (`torch.Tensor`, *optional*):

                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not

                provided, text embeddings will be generated from `prompt` input argument.

            negative_prompt_embeds (`torch.Tensor`, *optional*):

                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt

                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input

                argument.

            lora_scale (`float`, *optional*):

                A lora scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.

        """
        # set lora scale so that monkey patched LoRA
        # function of text encoder can correctly access it
        if lora_scale is not None and isinstance(self, LoraLoaderMixin):
            self._lora_scale = lora_scale

            # dynamically adjust the LoRA scale
            adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)

        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        if prompt_embeds is None:
            # textual inversion: procecss multi-vector tokens if necessary
            if isinstance(self, TextualInversionLoaderMixin):
                prompt = self.maybe_convert_prompt(prompt, self.tokenizer)

            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=self.tokenizer.model_max_length,
                truncation=True,
                return_tensors="pt",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids

            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
                text_input_ids, untruncated_ids
            ):
                removed_text = self.tokenizer.batch_decode(
                    untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1]
                )
                logger.warning(
                    "The following part of your input was truncated because CLIP can only handle sequences up to"
                    f" {self.tokenizer.model_max_length} tokens: {removed_text}"
                )

            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
                attention_mask = text_inputs.attention_mask.to(device)
            else:
                attention_mask = None

            prompt_embeds = self.text_encoder(
                text_input_ids.to(device),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

        if self.text_encoder is not None:
            prompt_embeds_dtype = self.text_encoder.dtype
        elif self.unet is not None:
            prompt_embeds_dtype = self.unet.dtype
        else:
            prompt_embeds_dtype = prompt_embeds.dtype

        prompt_embeds = prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)

        # get unconditional embeddings for classifier free guidance
        if do_classifier_free_guidance and negative_prompt_embeds is None:
            uncond_tokens: List[str]
            if negative_prompt is None:
                uncond_tokens = [""] * batch_size
            elif prompt is not None and type(prompt) is not type(negative_prompt):
                raise TypeError(
                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                    f" {type(prompt)}."
                )
            elif isinstance(negative_prompt, str):
                uncond_tokens = [negative_prompt]
            elif batch_size != len(negative_prompt):
                raise ValueError(
                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
                    " the batch size of `prompt`."
                )
            else:
                uncond_tokens = negative_prompt

            # textual inversion: procecss multi-vector tokens if necessary
            if isinstance(self, TextualInversionLoaderMixin):
                uncond_tokens = self.maybe_convert_prompt(uncond_tokens, self.tokenizer)

            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_tensors="pt",
            )

            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
                attention_mask = uncond_input.attention_mask.to(device)
            else:
                attention_mask = None

            negative_prompt_embeds = self.text_encoder(
                uncond_input.input_ids.to(device),
                attention_mask=attention_mask,
            )
            negative_prompt_embeds = negative_prompt_embeds[0]

        if do_classifier_free_guidance:
            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
            seq_len = negative_prompt_embeds.shape[1]

            negative_prompt_embeds = negative_prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)

            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)

        return prompt_embeds, negative_prompt_embeds

    def run_safety_checker(self, image, device, dtype):
        if self.safety_checker is None:
            has_nsfw_concept = None
        else:
            if torch.is_tensor(image):
                feature_extractor_input = self.image_processor.postprocess(image, output_type="pil")
            else:
                feature_extractor_input = self.image_processor.numpy_to_pil(image)
            safety_checker_input = self.feature_extractor(feature_extractor_input, return_tensors="pt").to(device)
            image, has_nsfw_concept = self.safety_checker(
                images=image, clip_input=safety_checker_input.pixel_values.to(dtype)
            )
        return image, has_nsfw_concept

    def decode_latents(self, latents):
        deprecation_message = "The decode_latents method is deprecated and will be removed in 1.0.0. Please use VaeImageProcessor.postprocess(...) instead"
        deprecate("decode_latents", "1.0.0", deprecation_message, standard_warn=False)

        latents = 1 / self.vae.config.scaling_factor * latents
        image = self.vae.decode(latents, return_dict=False)[0]
        image = (image / 2 + 0.5).clamp(0, 1)
        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
        image = image.cpu().permute(0, 2, 3, 1).float().numpy()
        return image

    def merge_dW_to_unet(pipe, dW_dict, alpha=1.0):
        _tmp_sd = pipe.unet.state_dict()
        for key in dW_dict.keys():
            _tmp_sd[key] += dW_dict[key] * alpha
        pipe.unet.load_state_dict(_tmp_sd, strict=False)
        return pipe

    def prepare_extra_step_kwargs(self, generator, eta):
        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
        # and should be between [0, 1]

        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
        extra_step_kwargs = {}
        if accepts_eta:
            extra_step_kwargs["eta"] = eta

        # check if the scheduler accepts generator
        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
        if accepts_generator:
            extra_step_kwargs["generator"] = generator
        return extra_step_kwargs

    def check_inputs(

        self,

        prompt,

        height,

        width,

        callback_steps,

        negative_prompt=None,

        prompt_embeds=None,

        negative_prompt_embeds=None,

    ):
        if height % 8 != 0 or width % 8 != 0:
            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")

        if (callback_steps is None) or (
            callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
        ):
            raise ValueError(
                f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
                f" {type(callback_steps)}."
            )

        if prompt is not None and prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                " only forward one of the two."
            )
        elif prompt is None and prompt_embeds is None:
            raise ValueError(
                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
            )
        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")

        if negative_prompt is not None and negative_prompt_embeds is not None:
            raise ValueError(
                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
            )

        if prompt_embeds is not None and negative_prompt_embeds is not None:
            if prompt_embeds.shape != negative_prompt_embeds.shape:
                raise ValueError(
                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
                    f" {negative_prompt_embeds.shape}."
                )

    def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
        shape = (
            batch_size,
            num_channels_latents,
            int(height) // self.vae_scale_factor,
            int(width) // self.vae_scale_factor,
        )
        if isinstance(generator, list) and len(generator) != batch_size:
            raise ValueError(
                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
            )

        if latents is None:
            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
        else:
            latents = latents.to(device)

        # scale the initial noise by the standard deviation required by the scheduler
        latents = latents * self.scheduler.init_noise_sigma
        return latents

    @torch.no_grad()
    def __call__(

        self,

        prompt: Union[str, List[str]] = None,

        height: Optional[int] = None,

        width: Optional[int] = None,

        num_inference_steps: int = 50,

        guidance_scale: float = 7.5,

        negative_prompt: Optional[Union[str, List[str]]] = None,

        num_images_per_prompt: Optional[int] = 1,

        eta: float = 0.0,

        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,

        latents: Optional[torch.Tensor] = None,

        prompt_embeds: Optional[torch.Tensor] = None,

        negative_prompt_embeds: Optional[torch.Tensor] = None,

        output_type: Optional[str] = "pil",

        return_dict: bool = True,

        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,

        callback_steps: int = 1,

        cross_attention_kwargs: Optional[Dict[str, Any]] = None,

        guidance_rescale: float = 0.0,

    ):
        r"""

        The call function to the pipeline for generation.



        Args:

            prompt (`str` or `List[str]`, *optional*):

                The prompt or prompts to guide image generation. If not defined, you need to pass `prompt_embeds`.

            height (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):

                The height in pixels of the generated image.

            width (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):

                The width in pixels of the generated image.

            num_inference_steps (`int`, *optional*, defaults to 50):

                The number of denoising steps. More denoising steps usually lead to a higher quality image at the

                expense of slower inference.

            guidance_scale (`float`, *optional*, defaults to 7.5):

                A higher guidance scale value encourages the model to generate images closely linked to the text

                `prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`.

            negative_prompt (`str` or `List[str]`, *optional*):

                The prompt or prompts to guide what to not include in image generation. If not defined, you need to

                pass `negative_prompt_embeds` instead. Ignored when not using guidance (`guidance_scale < 1`).

            num_images_per_prompt (`int`, *optional*, defaults to 1):

                The number of images to generate per prompt.

            eta (`float`, *optional*, defaults to 0.0):

                Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies

                to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.

            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):

                A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make

                generation deterministic.

            latents (`torch.Tensor`, *optional*):

                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image

                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents

                tensor is generated by sampling using the supplied random `generator`.

            prompt_embeds (`torch.Tensor`, *optional*):

                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not

                provided, text embeddings are generated from the `prompt` input argument.

            negative_prompt_embeds (`torch.Tensor`, *optional*):

                Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If

                not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.

            output_type (`str`, *optional*, defaults to `"pil"`):

                The output format of the generated image. Choose between `PIL.Image` or `np.array`.

            return_dict (`bool`, *optional*, defaults to `True`):

                Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a

                plain tuple.

            callback (`Callable`, *optional*):

                A function that calls every `callback_steps` steps during inference. The function is called with the

                following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.

            callback_steps (`int`, *optional*, defaults to 1):

                The frequency at which the `callback` function is called. If not specified, the callback is called at

                every step.

            cross_attention_kwargs (`dict`, *optional*):

                A kwargs dictionary that if specified is passed along to the [`AttentionProcessor`] as defined in

                [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

            guidance_rescale (`float`, *optional*, defaults to 0.7):

                Guidance rescale factor from [Common Diffusion Noise Schedules and Sample Steps are

                Flawed](https://arxiv.org/pdf/2305.08891.pdf). Guidance rescale factor should fix overexposure when

                using zero terminal SNR.



        Examples:



        Returns:

            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`:

                If `return_dict` is `True`, [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] is returned,

                otherwise a `tuple` is returned where the first element is a list with the generated images and the

                second element is a list of `bool`s indicating whether the corresponding generated image contains

                "not-safe-for-work" (nsfw) content.

        """
        # 0. Default height and width to unet
        height = height or self.unet.config.sample_size * self.vae_scale_factor
        width = width or self.unet.config.sample_size * self.vae_scale_factor

        # 1. Check inputs. Raise error if not correct
        self.check_inputs(
            prompt, height, width, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds
        )

        # 2. Define call parameters
        if prompt is not None and isinstance(prompt, str):
            batch_size = 1
        elif prompt is not None and isinstance(prompt, list):
            batch_size = len(prompt)
        else:
            batch_size = prompt_embeds.shape[0]

        device = self._execution_device
        # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
        # corresponds to doing no classifier free guidance.
        do_classifier_free_guidance = guidance_scale > 1.0

        # 3. Encode input prompt
        text_encoder_lora_scale = (
            cross_attention_kwargs.get("scale", None) if cross_attention_kwargs is not None else None
        )
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            device,
            num_images_per_prompt,
            do_classifier_free_guidance,
            negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            lora_scale=text_encoder_lora_scale,
        )
        # For classifier free guidance, we need to do two forward passes.
        # Here we concatenate the unconditional and text embeddings into a single batch
        # to avoid doing two forward passes
        if do_classifier_free_guidance:
            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])

        # 4. Prepare timesteps
        timesteps = [(1.0 - i / num_inference_steps) * 1000.0 for i in range(num_inference_steps)]

        # 5. Prepare latent variables
        num_channels_latents = self.unet.config.in_channels
        latents = self.prepare_latents(
            batch_size * num_images_per_prompt,
            num_channels_latents,
            height,
            width,
            prompt_embeds.dtype,
            device,
            generator,
            latents,
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        dt = 1.0 / num_inference_steps

        # 7. Denoising loop of Euler discretization from t = 0 to t = 1
        with self.progress_bar(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

                vec_t = torch.ones((latent_model_input.shape[0],), device=latents.device) * t

                v_pred = self.unet(latent_model_input, vec_t, encoder_hidden_states=prompt_embeds).sample

                # perform guidance
                if do_classifier_free_guidance:
                    v_pred_neg, v_pred_text = v_pred.chunk(2)
                    v_pred = v_pred_neg + guidance_scale * (v_pred_text - v_pred_neg)

                latents = latents + dt * v_pred

                # call the callback, if provided
                if i == len(timesteps) - 1 or ((i + 1) % self.scheduler.order == 0):
                    progress_bar.update()
                    if callback is not None and i % callback_steps == 0:
                        step_idx = i // getattr(self.scheduler, "order", 1)
                        callback(step_idx, t, latents)

        if not output_type == "latent":
            image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
            image, has_nsfw_concept = self.run_safety_checker(image, device, prompt_embeds.dtype)
        else:
            image = latents
            has_nsfw_concept = None

        if has_nsfw_concept is None:
            do_denormalize = [True] * image.shape[0]
        else:
            do_denormalize = [not has_nsfw for has_nsfw in has_nsfw_concept]

        image = self.image_processor.postprocess(image, output_type=output_type, do_denormalize=do_denormalize)

        # Offload all models
        self.maybe_free_model_hooks()

        if not return_dict:
            return (image, has_nsfw_concept)

        return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept)