DeepFloyd IF

Overview

DeepFloyd IF is a novel state-of-the-art open-source text-to-image model with a high degree of photorealism and language understanding. The model is a modular composed of a frozen text encoder and three cascaded pixel diffusion modules:

• Stage 1: a base model that generates 64x64 px image based on text prompt,
• Stage 2: a 64x64 px => 256x256 px super-resolution model, and
• Stage 3: a 256x256 px => 1024x1024 px super-resolution model

Stage 1 and Stage 2 utilize a frozen text encoder based on the T5 transformer to extract text embeddings, which are then fed into a UNet architecture enhanced with cross-attention and attention pooling. Stage 3 is Stability AI's x4 Upscaling model. The result is a highly efficient model that outperforms current state-of-the-art models, achieving a zero-shot FID score of 6.66 on the COCO dataset. Our work underscores the potential of larger UNet architectures in the first stage of cascaded diffusion models and depicts a promising future for text-to-image synthesis.

Usage

Before you can use IF, you need to accept its usage conditions. To do so: 1. Make sure to have a Hugging Face account and be logged in. 2. Accept the license on the model card of DeepFloyd/IF-I-XL-v1.0. Accepting the license on the stage I model card will auto accept for the other IF models. 3. Make sure to login locally. Install huggingface_hub:

pip install huggingface_hub --upgrade

run the login function in a Python shell:

from huggingface_hub import login

login()

and enter your Hugging Face Hub access token.

The following sections give more in-detail examples of how to use IF. Specifically:

• Text-to-Image Generation
• Image-to-Image Generation
• Inpainting
• Reusing model weights
• Speed optimization

Available checkpoints - Stage-1 - DeepFloyd/IF-I-XL-v1.0 - DeepFloyd/IF-I-L-v1.0 - DeepFloyd/IF-I-M-v1.0

• Stage-2
• DeepFloyd/IF-II-L-v1.0

• DeepFloyd/IF-II-M-v1.0

• Stage-3

• stabilityai/stable-diffusion-x4-upscaler

Text-to-Image Generation

from mindone.diffusers import DiffusionPipeline
from mindone.diffusers.utils import pt_to_pil, make_image_grid
import mindspore as ms
import numpy as np

# stage 1
stage_1 = DiffusionPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", mindspore_dtype=ms.float16)

# stage 2
stage_2 = DiffusionPipeline.from_pretrained(
    "DeepFloyd/IF-II-L-v1.0", text_encoder=None, variant="fp16", mindspore_dtype=ms.float16
)

# stage 3
safety_modules = {
    "feature_extractor": stage_1.feature_extractor,
    "safety_checker": stage_1.safety_checker,
    "watermarker": stage_1.watermarker,
}
stage_3 = DiffusionPipeline.from_pretrained(
    "stabilityai/stable-diffusion-x4-upscaler", **safety_modules, mindspore_dtype=ms.float16
)

prompt = 'a photo of a kangaroo wearing an orange hoodie and blue sunglasses standing in front of the eiffel tower holding a sign that says "very deep learning"'
generator = np.random.Generator(np.random.PCG64(1))

# text embeds
prompt_embeds, negative_embeds = stage_1.encode_prompt(prompt)

# stage 1
stage_1_output = stage_1(
    prompt_embeds=prompt_embeds, negative_prompt_embeds=negative_embeds, generator=generator, output_type="ms"
)[0]
#pt_to_pil(stage_1_output)[0].save("./if_stage_I.png")

# stage 2
stage_2_output = stage_2(
    image=stage_1_output,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    generator=generator,
    output_type="ms",
)[0]
#pt_to_pil(stage_2_output)[0].save("./if_stage_II.png")

# stage 3
stage_3_output = stage_3(prompt=prompt, image=stage_2_output, noise_level=100, generator=generator)[0]
#stage_3_output[0].save("./if_stage_III.png")
make_image_grid([pt_to_pil(stage_1_output)[0], pt_to_pil(stage_2_output)[0], stage_3_output[0]], cols=1, rows=3)

Text Guided Image-to-Image Generation

The same IF model weights can be used for text-guided image-to-image translation or image variation. In this case just make sure to load the weights using the IFImg2ImgPipeline and IFImg2ImgSuperResolutionPipeline pipelines.

Note: You can also directly move the weights of the text-to-image pipelines to the image-to-image pipelines without loading them twice by making use of the diffusionpipeline.components argument as explained here.

from mindone.diffusers import IFImg2ImgPipeline, IFImg2ImgSuperResolutionPipeline, DiffusionPipeline
from mindone.diffusers.utils import pt_to_pil, load_image, make_image_grid
import mindspore as ms
import numpy as np

# download image
url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
original_image = load_image(url)
original_image = original_image.resize((768, 512))

# stage 1
stage_1 = IFImg2ImgPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", mindspore_dtype=ms.float16)

# stage 2
stage_2 = IFImg2ImgSuperResolutionPipeline.from_pretrained(
    "DeepFloyd/IF-II-L-v1.0", text_encoder=None, variant="fp16", mindspore_dtype=ms.float16
)

# stage 3
safety_modules = {
    "feature_extractor": stage_1.feature_extractor,
    "safety_checker": stage_1.safety_checker,
    "watermarker": stage_1.watermarker,
}
stage_3 = DiffusionPipeline.from_pretrained(
    "stabilityai/stable-diffusion-x4-upscaler", **safety_modules, mindspore_dtype=ms.float16
)

prompt = "A fantasy landscape in style minecraft"
generator = np.random.Generator(np.random.PCG64(1))

# text embeds
prompt_embeds, negative_embeds = stage_1.encode_prompt(prompt)

# stage 1
stage_1_output = stage_1(
    image=original_image,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    generator=generator,
    output_type="ms",
)[0]
#pt_to_pil(stage_1_output)[0].save("./if_stage_I.png")

# stage 2
stage_2_output = stage_2(
    image=stage_1_output,
    original_image=original_image,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    generator=generator,
    output_type="ms",
)[0]
#pt_to_pil(stage_2_output)[0].save("./if_stage_II.png")

# stage 3
stage_3_output = stage_3(prompt=prompt, image=stage_2_output, generator=generator, noise_level=100)[0]
#stage_3_output[0].save("./if_stage_III.png")
make_image_grid([original_image, pt_to_pil(stage_1_output)[0], pt_to_pil(stage_2_output)[0], stage_3_output[0]], cols=1, rows=4)

Text Guided Inpainting Generation

The same IF model weights can be used for text-guided image-to-image translation or image variation. In this case just make sure to load the weights using the IFInpaintingPipeline and IFInpaintingSuperResolutionPipeline pipelines.

Note: You can also directly move the weights of the text-to-image pipelines to the image-to-image pipelines without loading them twice by making use of the DiffusionPipeline.components() function as explained here.

from mindone.diffusers import IFInpaintingPipeline, IFInpaintingSuperResolutionPipeline, DiffusionPipeline
from mindone.diffusers.utils import pt_to_pil, load_image, make_image_grid
import mindspore as ms
import numpy as np

# download image
url = "https://huggingface.co/datasets/diffusers/docs-images/resolve/main/if/person.png"
original_image = load_image(url)

# download mask
url = "https://huggingface.co/datasets/diffusers/docs-images/resolve/main/if/glasses_mask.png"
mask_image = load_image(url)

# stage 1
stage_1 = IFInpaintingPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", mindspore_dtype=ms.float16)

# stage 2
stage_2 = IFInpaintingSuperResolutionPipeline.from_pretrained(
    "DeepFloyd/IF-II-L-v1.0", text_encoder=None, variant="fp16", mindspore_dtype=ms.float16
)

# stage 3
safety_modules = {
    "feature_extractor": stage_1.feature_extractor,
    "safety_checker": stage_1.safety_checker,
    "watermarker": stage_1.watermarker,
}
stage_3 = DiffusionPipeline.from_pretrained(
    "stabilityai/stable-diffusion-x4-upscaler", **safety_modules, mindspore_dtype=ms.float16
)

prompt = "blue sunglasses"
generator = np.random.Generator(np.random.PCG64(1))

# text embeds
prompt_embeds, negative_embeds = stage_1.encode_prompt(prompt)

# stage 1
stage_1_output = stage_1(
    image=original_image,
    mask_image=mask_image,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    generator=generator,
    output_type="ms",
)[0]
#pt_to_pil(stage_1_output)[0].save("./if_stage_I.png")

# stage 2
stage_2_output = stage_2(
    image=stage_1_output,
    original_image=original_image,
    mask_image=mask_image,
    prompt_embeds=prompt_embeds,
    negative_prompt_embeds=negative_embeds,
    generator=generator,
    output_type="ms",
)[0]
#pt_to_pil(stage_1_output)[0].save("./if_stage_II.png")

# stage 3
stage_3_output = stage_3(prompt=prompt, image=stage_2_output, generator=generator, noise_level=100)[0]
#stage_3_output[0].save("./if_stage_III.png")
make_image_grid([original_image, mask_image, pt_to_pil(stage_1_output)[0], pt_to_pil(stage_2_output)[0], stage_3_output[0]], cols=1, rows=5)

Converting between different pipelines

In addition to being loaded with from_pretrained, Pipelines can also be loaded directly from each other.

from mindone.diffusers import IFPipeline, IFSuperResolutionPipeline

pipe_1 = IFPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0")
pipe_2 = IFSuperResolutionPipeline.from_pretrained("DeepFloyd/IF-II-L-v1.0")


from mindone.diffusers import IFImg2ImgPipeline, IFImg2ImgSuperResolutionPipeline

pipe_1 = IFImg2ImgPipeline(**pipe_1.components)
pipe_2 = IFImg2ImgSuperResolutionPipeline(**pipe_2.components)


from mindone.diffusers import IFInpaintingPipeline, IFInpaintingSuperResolutionPipeline

pipe_1 = IFInpaintingPipeline(**pipe_1.components)
pipe_2 = IFInpaintingSuperResolutionPipeline(**pipe_2.components)

Optimizing for speed

The simplest optimization to run IF faster is to move all model components to the GPU.

import mindspore as ms
from mindone.diffusers import DiffusionPipeline

pipe = DiffusionPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", mindspore_dtype=ms.float16)

You can also run the diffusion process for a shorter number of timesteps.

This can either be done with the num_inference_steps argument:

pipe("<prompt>", num_inference_steps=30)

Or with the timesteps argument:

from mindone.diffusers.pipelines.deepfloyd_if import fast27_timesteps

pipe("<prompt>", timesteps=fast27_timesteps)

When doing image variation or inpainting, you can also decrease the number of timesteps with the strength argument. The strength argument is the amount of noise to add to the input image which also determines how many steps to run in the denoising process. A smaller number will vary the image less but run faster.

pipe = IFImg2ImgPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", variant="fp16", mindspore_dtype=ms.float16)

image = pipe(image=image, prompt="<prompt>", strength=0.3).images

Available Pipelines:

Pipeline	Tasks
pipeline_if.py	Text-to-Image Generation
pipeline_if_superresolution.py	Text-to-Image Generation
pipeline_if_img2img.py	Image-to-Image Generation
pipeline_if_img2img_superresolution.py	Image-to-Image Generation
pipeline_if_inpainting.py	Image-to-Image Generation
pipeline_if_inpainting_superresolution.py	Image-to-Image Generation

mindone.diffusers.IFPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if.py

class IFPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

    _optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]
    model_cpu_offload_seq = "text_encoder->unet"

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
    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,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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_intermediate_images(self, batch_size, num_channels, height, width, dtype, generator):
        shape = (batch_size, num_channels, height, width)
        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."
            )

        intermediate_images = randn_tensor(shape, generator=generator, dtype=dtype)

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

    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        num_inference_steps: int = 100,
        timesteps: List[int] = None,
        guidance_scale: float = 7.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        height: Optional[int] = None,
        width: Optional[int] = None,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        clean_caption: bool = True,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            num_inference_steps (`int`, *optional*, defaults to 100):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 7.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            height (`int`, *optional*, defaults to self.unet.config.sample_size):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to self.unet.config.sample_size):
                The width in pixels of the generated image.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct
        self.check_inputs(prompt, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds)

        # 2. Define call parameters
        height = height or self.unet.config.sample_size
        width = width or self.unet.config.sample_size

        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]

        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(0)

        # 5. Prepare intermediate images
        intermediate_images = self.prepare_intermediate_images(
            batch_size * num_images_per_prompt,
            self.unet.config.in_channels,
            height,
            width,
            prompt_embeds.dtype,
            generator,
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 7. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 10. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 11. Apply watermark
            if self.watermarker is not None:
                image = self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None
        else:
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFPipeline.__call__(prompt=None, num_inference_steps=100, timesteps=None, guidance_scale=7.0, negative_prompt=None, num_images_per_prompt=1, height=None, width=None, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, clean_caption=True, cross_attention_kwargs=None)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 100 DEFAULT: 100
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional* DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 7.0 DEFAULT: 7.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
height	The height in pixels of the generated image. TYPE: `int`, *optional*, defaults to self.unet.config.sample_size DEFAULT: None
width	The width in pixels of the generated image. TYPE: `int`, *optional*, defaults to self.unet.config.sample_size DEFAULT: None
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    num_inference_steps: int = 100,
    timesteps: List[int] = None,
    guidance_scale: float = 7.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    height: Optional[int] = None,
    width: Optional[int] = None,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    clean_caption: bool = True,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        num_inference_steps (`int`, *optional*, defaults to 100):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        timesteps (`List[int]`, *optional*):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 7.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            The number of images to generate per prompt.
        height (`int`, *optional*, defaults to self.unet.config.sample_size):
            The height in pixels of the generated image.
        width (`int`, *optional*, defaults to self.unet.config.sample_size):
            The width in pixels of the generated image.
        eta (`float`, *optional*, defaults to 0.0):
            Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct
    self.check_inputs(prompt, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds)

    # 2. Define call parameters
    height = height or self.unet.config.sample_size
    width = width or self.unet.config.sample_size

    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]

    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    if hasattr(self.scheduler, "set_begin_index"):
        self.scheduler.set_begin_index(0)

    # 5. Prepare intermediate images
    intermediate_images = self.prepare_intermediate_images(
        batch_size * num_images_per_prompt,
        self.unet.config.in_channels,
        height,
        width,
        prompt_embeds.dtype,
        generator,
    )

    # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 7. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 10. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 11. Apply watermark
        if self.watermarker is not None:
            image = self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None
    else:
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.IFSuperResolutionPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_superresolution.py

class IFSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler
    image_noising_scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

    _optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]
    model_cpu_offload_seq = "text_encoder->unet"

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        image_noising_scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        if unet.config.in_channels != 6:
            logger.warning(
                f"It seems like you have loaded a checkpoint that shall not be used for super resolution from {unet.config._name_or_path} as it accepts {unet.config.in_channels} input channels instead of 6. Please make sure to pass a super resolution checkpoint as the `'unet'`: IFSuperResolutionPipeline.from_pretrained(unet=super_resolution_unet, ...)`."  # noqa E501
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            image_noising_scheduler=image_noising_scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.run_safety_checker
    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_extra_step_kwargs
    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,
        image,
        batch_size,
        noise_level,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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}."
                )

        if noise_level < 0 or noise_level >= self.image_noising_scheduler.config.num_train_timesteps:
            raise ValueError(
                f"`noise_level`: {noise_level} must be a valid timestep in `self.noising_scheduler`, [0, {self.image_noising_scheduler.config.num_train_timesteps})"  # noqa E501
            )

        if isinstance(image, list):
            check_image_type = image[0]
        else:
            check_image_type = image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(image, list):
            image_batch_size = len(image)
        elif isinstance(image, ms.Tensor):
            image_batch_size = image.shape[0]
        elif isinstance(image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(image, np.ndarray):
            image_batch_size = image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(f"image batch size: {image_batch_size} must be same as prompt batch size {batch_size}")

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_intermediate_images
    def prepare_intermediate_images(self, batch_size, num_channels, height, width, dtype, generator):
        shape = (batch_size, num_channels, height, width)
        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."
            )

        intermediate_images = randn_tensor(shape, generator=generator, dtype=dtype)

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

    def preprocess_image(self, image, num_images_per_prompt):
        if not isinstance(image, ms.Tensor) and not isinstance(image, list):
            image = [image]

        if isinstance(image[0], PIL.Image.Image):
            image = [np.array(i).astype(np.float32) / 127.5 - 1.0 for i in image]

            image = np.stack(image, axis=0)  # to np
            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image[0], np.ndarray):
            image = np.stack(image, axis=0)  # to np
            if image.ndim == 5:
                image = image[0]

            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image, list) and isinstance(image[0], ms.Tensor):
            dims = image[0].ndim

            if dims == 3:
                image = ops.stack(image, axis=0)
            elif dims == 4:
                image = ops.concat(image, axis=0)
            else:
                raise ValueError(f"Image must have 3 or 4 dimensions, instead got {dims}")

        image = image.to(dtype=self.unet.dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)

        return image

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        height: int = None,
        width: int = None,
        image: Union[PIL.Image.Image, np.ndarray, ms.Tensor] = None,
        num_inference_steps: int = 50,
        timesteps: List[int] = None,
        guidance_scale: float = 4.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
        noise_level: int = 250,
        clean_caption: bool = True,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            height (`int`, *optional*, defaults to None):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to None):
                The width in pixels of the generated image.
            image (`PIL.Image.Image`, `np.ndarray`, `ms.Tensor`):
                The image to be upscaled.
            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.
            timesteps (`List[int]`, *optional*, defaults to None):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 4.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            noise_level (`int`, *optional*, defaults to 250):
                The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct

        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]

        self.check_inputs(
            prompt,
            image,
            batch_size,
            noise_level,
            callback_steps,
            negative_prompt,
            prompt_embeds,
            negative_prompt_embeds,
        )

        # 2. Define call parameters

        height = height or self.unet.config.sample_size
        width = width or self.unet.config.sample_size

        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(0)

        # 5. Prepare intermediate images
        num_channels = self.unet.config.in_channels // 2
        intermediate_images = self.prepare_intermediate_images(
            batch_size * num_images_per_prompt,
            num_channels,
            height,
            width,
            prompt_embeds.dtype,
            generator,
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # 7. Prepare upscaled image and noise level
        image = self.preprocess_image(image, num_images_per_prompt)
        upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

        noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
        noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
        upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

        if do_classifier_free_guidance:
            noise_level = ops.cat([noise_level] * 2)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 8. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images, upscaled], axis=1)

                model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    class_labels=noise_level,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 9. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 10. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 11. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 12. Apply watermark
            if self.watermarker is not None:
                self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None
        else:
            # 9. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 10. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFSuperResolutionPipeline.__call__(prompt=None, height=None, width=None, image=None, num_inference_steps=50, timesteps=None, guidance_scale=4.0, negative_prompt=None, num_images_per_prompt=1, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, cross_attention_kwargs=None, noise_level=250, clean_caption=True)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
height	The height in pixels of the generated image. TYPE: `int`, *optional*, defaults to None DEFAULT: None
width	The width in pixels of the generated image. TYPE: `int`, *optional*, defaults to None DEFAULT: None
image	The image to be upscaled. TYPE: `PIL.Image.Image`, `np.ndarray`, `ms.Tensor` DEFAULT: None
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 50 DEFAULT: 50
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional*, defaults to None DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 4.0 DEFAULT: 4.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
noise_level	The amount of noise to add to the upscaled image. Must be in the range [0, 1000) TYPE: `int`, *optional*, defaults to 250 DEFAULT: 250
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_superresolution.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    height: int = None,
    width: int = None,
    image: Union[PIL.Image.Image, np.ndarray, ms.Tensor] = None,
    num_inference_steps: int = 50,
    timesteps: List[int] = None,
    guidance_scale: float = 4.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    noise_level: int = 250,
    clean_caption: bool = True,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        height (`int`, *optional*, defaults to None):
            The height in pixels of the generated image.
        width (`int`, *optional*, defaults to None):
            The width in pixels of the generated image.
        image (`PIL.Image.Image`, `np.ndarray`, `ms.Tensor`):
            The image to be upscaled.
        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.
        timesteps (`List[int]`, *optional*, defaults to None):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 4.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `True`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        noise_level (`int`, *optional*, defaults to 250):
            The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct

    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]

    self.check_inputs(
        prompt,
        image,
        batch_size,
        noise_level,
        callback_steps,
        negative_prompt,
        prompt_embeds,
        negative_prompt_embeds,
    )

    # 2. Define call parameters

    height = height or self.unet.config.sample_size
    width = width or self.unet.config.sample_size

    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    if hasattr(self.scheduler, "set_begin_index"):
        self.scheduler.set_begin_index(0)

    # 5. Prepare intermediate images
    num_channels = self.unet.config.in_channels // 2
    intermediate_images = self.prepare_intermediate_images(
        batch_size * num_images_per_prompt,
        num_channels,
        height,
        width,
        prompt_embeds.dtype,
        generator,
    )

    # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # 7. Prepare upscaled image and noise level
    image = self.preprocess_image(image, num_images_per_prompt)
    upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

    noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
    noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
    upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

    if do_classifier_free_guidance:
        noise_level = ops.cat([noise_level] * 2)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 8. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images, upscaled], axis=1)

            model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                class_labels=noise_level,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 9. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 10. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 11. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 12. Apply watermark
        if self.watermarker is not None:
            self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None
    else:
        # 9. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 10. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFSuperResolutionPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_superresolution.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.IFImg2ImgPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py

class IFImg2ImgPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

    _optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]
    model_cpu_offload_seq = "text_encoder->unet"

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.run_safety_checker
    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_extra_step_kwargs
    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,
        image,
        batch_size,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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}."
                )

        if isinstance(image, list):
            check_image_type = image[0]
        else:
            check_image_type = image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(image, list):
            image_batch_size = len(image)
        elif isinstance(image, ms.Tensor):
            image_batch_size = image.shape[0]
        elif isinstance(image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(image, np.ndarray):
            image_batch_size = image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(f"image batch size: {image_batch_size} must be same as prompt batch size {batch_size}")

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    def preprocess_image(self, image: PIL.Image.Image) -> ms.Tensor:
        if not isinstance(image, list):
            image = [image]

        def numpy_to_ms(images):
            if images.ndim == 3:
                images = images[..., None]

            images = ms.Tensor.from_numpy(images.transpose(0, 3, 1, 2))
            return images

        if isinstance(image[0], PIL.Image.Image):
            new_image = []

            for image_ in image:
                image_ = image_.convert("RGB")
                image_ = resize(image_, self.unet.config.sample_size)
                image_ = np.array(image_)
                image_ = image_.astype(np.float32)
                image_ = image_ / 127.5 - 1
                new_image.append(image_)

            image = new_image

            image = np.stack(image, axis=0)  # to np
            image = numpy_to_ms(image)  # to pt

        elif isinstance(image[0], np.ndarray):
            image = np.concatenate(image, axis=0) if image[0].ndim == 4 else np.stack(image, axis=0)
            image = numpy_to_ms(image)

        elif isinstance(image[0], ms.Tensor):
            image = ops.cat(image, axis=0) if image[0].ndim == 4 else ops.stack(image, axis=0)

        return image

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
    def get_timesteps(self, num_inference_steps, strength):
        # get the original timestep using init_timestep
        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

        t_start = max(num_inference_steps - init_timestep, 0)
        timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(t_start * self.scheduler.order)

        return timesteps, num_inference_steps - t_start

    def prepare_intermediate_images(self, image, timestep, batch_size, num_images_per_prompt, dtype, generator=None):
        _, channels, height, width = image.shape

        batch_size = batch_size * num_images_per_prompt

        shape = (batch_size, channels, height, width)

        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."
            )

        noise = randn_tensor(shape, generator=generator, dtype=dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)
        image = self.scheduler.add_noise(image, noise, timestep)

        return image

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        strength: float = 0.7,
        num_inference_steps: int = 80,
        timesteps: List[int] = None,
        guidance_scale: float = 10.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        clean_caption: bool = True,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            image (`ms.Tensor` or `PIL.Image.Image`):
                `Image`, or tensor representing an image batch, that will be used as the starting point for the
                process.
            strength (`float`, *optional*, defaults to 0.7):
                Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
                will be used as a starting point, adding more noise to it the larger the `strength`. The number of
                denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
                be maximum and the denoising process will run for the full number of iterations specified in
                `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
            num_inference_steps (`int`, *optional*, defaults to 80):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 10.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct
        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]

        self.check_inputs(
            prompt, image, batch_size, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds
        )

        # 2. Define call parameters
        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        dtype = prompt_embeds.dtype

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

        # 5. Prepare intermediate images
        image = self.preprocess_image(image)
        image = image.to(dtype=dtype)

        noise_timestep = timesteps[0:1]
        noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

        intermediate_images = self.prepare_intermediate_images(
            image, noise_timestep, batch_size, num_images_per_prompt, dtype, generator
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 7. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 10. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 11. Apply watermark
            if self.watermarker is not None:
                self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None
        else:
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFImg2ImgPipeline.__call__(prompt=None, image=None, strength=0.7, num_inference_steps=80, timesteps=None, guidance_scale=10.0, negative_prompt=None, num_images_per_prompt=1, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, clean_caption=True, cross_attention_kwargs=None)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
image	Image, or tensor representing an image batch, that will be used as the starting point for the process. TYPE: `ms.Tensor` or `PIL.Image.Image` DEFAULT: None
strength	Conceptually, indicates how much to transform the reference image. Must be between 0 and 1. image will be used as a starting point, adding more noise to it the larger the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise will be maximum and the denoising process will run for the full number of iterations specified in num_inference_steps. A value of 1, therefore, essentially ignores image. TYPE: `float`, *optional*, defaults to 0.7 DEFAULT: 0.7
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 80 DEFAULT: 80
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional* DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 10.0 DEFAULT: 10.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    strength: float = 0.7,
    num_inference_steps: int = 80,
    timesteps: List[int] = None,
    guidance_scale: float = 10.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    clean_caption: bool = True,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        image (`ms.Tensor` or `PIL.Image.Image`):
            `Image`, or tensor representing an image batch, that will be used as the starting point for the
            process.
        strength (`float`, *optional*, defaults to 0.7):
            Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
            will be used as a starting point, adding more noise to it the larger the `strength`. The number of
            denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
            be maximum and the denoising process will run for the full number of iterations specified in
            `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
        num_inference_steps (`int`, *optional*, defaults to 80):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        timesteps (`List[int]`, *optional*):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 10.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct
    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]

    self.check_inputs(
        prompt, image, batch_size, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds
    )

    # 2. Define call parameters
    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    dtype = prompt_embeds.dtype

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

    # 5. Prepare intermediate images
    image = self.preprocess_image(image)
    image = image.to(dtype=dtype)

    noise_timestep = timesteps[0:1]
    noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

    intermediate_images = self.prepare_intermediate_images(
        image, noise_timestep, batch_size, num_images_per_prompt, dtype, generator
    )

    # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 7. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 10. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 11. Apply watermark
        if self.watermarker is not None:
            self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None
    else:
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFImg2ImgPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.IFImg2ImgSuperResolutionPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img_superresolution.py

class IFImg2ImgSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler
    image_noising_scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

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

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        image_noising_scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        if unet.config.in_channels != 6:
            logger.warning(
                f"It seems like you have loaded a checkpoint that shall not be used for super resolution from {unet.config._name_or_path} as it accepts {unet.config.in_channels} input channels instead of 6. Please make sure to pass a super resolution checkpoint as the `'unet'`: IFSuperResolutionPipeline.from_pretrained(unet=super_resolution_unet, ...)`."  # noqa E501
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            image_noising_scheduler=image_noising_scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.run_safety_checker
    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_extra_step_kwargs
    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,
        image,
        original_image,
        batch_size,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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}."
                )

        # image

        if isinstance(image, list):
            check_image_type = image[0]
        else:
            check_image_type = image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(image, list):
            image_batch_size = len(image)
        elif isinstance(image, ms.Tensor):
            image_batch_size = image.shape[0]
        elif isinstance(image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(image, np.ndarray):
            image_batch_size = image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(f"image batch size: {image_batch_size} must be same as prompt batch size {batch_size}")

        # original_image

        if isinstance(original_image, list):
            check_image_type = original_image[0]
        else:
            check_image_type = original_image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`original_image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(original_image, list):
            image_batch_size = len(original_image)
        elif isinstance(original_image, ms.Tensor):
            image_batch_size = original_image.shape[0]
        elif isinstance(original_image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(original_image, np.ndarray):
            image_batch_size = original_image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(
                f"original_image batch size: {image_batch_size} must be same as prompt batch size {batch_size}"
            )

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_img2img.IFImg2ImgPipeline.preprocess_image with preprocess_image -> preprocess_original_image
    def preprocess_original_image(self, image: PIL.Image.Image) -> ms.Tensor:
        if not isinstance(image, list):
            image = [image]

        def numpy_to_ms(images):
            if images.ndim == 3:
                images = images[..., None]

            images = ms.Tensor.from_numpy(images.transpose(0, 3, 1, 2))
            return images

        if isinstance(image[0], PIL.Image.Image):
            new_image = []

            for image_ in image:
                image_ = image_.convert("RGB")
                image_ = resize(image_, self.unet.config.sample_size)
                image_ = np.array(image_)
                image_ = image_.astype(np.float32)
                image_ = image_ / 127.5 - 1
                new_image.append(image_)

            image = new_image

            image = np.stack(image, axis=0)  # to np
            image = numpy_to_ms(image)  # to pt

        elif isinstance(image[0], np.ndarray):
            image = np.concatenate(image, axis=0) if image[0].ndim == 4 else np.stack(image, axis=0)
            image = numpy_to_ms(image)

        elif isinstance(image[0], ms.Tensor):
            image = ops.cat(image, axis=0) if image[0].ndim == 4 else ops.stack(image, axis=0)

        return image

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_superresolution.IFSuperResolutionPipeline.preprocess_image
    def preprocess_image(self, image, num_images_per_prompt):
        if not isinstance(image, ms.Tensor) and not isinstance(image, list):
            image = [image]

        if isinstance(image[0], PIL.Image.Image):
            image = [np.array(i).astype(np.float32) / 127.5 - 1.0 for i in image]

            image = np.stack(image, axis=0)  # to np
            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image[0], np.ndarray):
            image = np.stack(image, axis=0)  # to np
            if image.ndim == 5:
                image = image[0]

            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image, list) and isinstance(image[0], ms.Tensor):
            dims = image[0].ndim

            if dims == 3:
                image = ops.stack(image, axis=0)
            elif dims == 4:
                image = ops.concat(image, axis=0)
            else:
                raise ValueError(f"Image must have 3 or 4 dimensions, instead got {dims}")

        image = image.to(dtype=self.unet.dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)

        return image

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
    def get_timesteps(self, num_inference_steps, strength):
        # get the original timestep using init_timestep
        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

        t_start = max(num_inference_steps - init_timestep, 0)
        timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(t_start * self.scheduler.order)

        return timesteps, num_inference_steps - t_start

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_img2img.IFImg2ImgPipeline.prepare_intermediate_images
    def prepare_intermediate_images(self, image, timestep, batch_size, num_images_per_prompt, dtype, generator=None):
        _, channels, height, width = image.shape

        batch_size = batch_size * num_images_per_prompt

        shape = (batch_size, channels, height, width)

        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."
            )

        noise = randn_tensor(shape, generator=generator, dtype=dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)
        image = self.scheduler.add_noise(image, noise, timestep)

        return image

    def __call__(
        self,
        image: Union[PIL.Image.Image, np.ndarray, ms.Tensor],
        original_image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        strength: float = 0.8,
        prompt: Union[str, List[str]] = None,
        num_inference_steps: int = 50,
        timesteps: List[int] = None,
        guidance_scale: float = 4.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
        noise_level: int = 250,
        clean_caption: bool = True,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            image (`ms.Tensor` or `PIL.Image.Image`):
                `Image`, or tensor representing an image batch, that will be used as the starting point for the
                process.
            original_image (`ms.Tensor` or `PIL.Image.Image`):
                The original image that `image` was varied from.
            strength (`float`, *optional*, defaults to 0.8):
                Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
                will be used as a starting point, adding more noise to it the larger the `strength`. The number of
                denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
                be maximum and the denoising process will run for the full number of iterations specified in
                `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            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.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 4.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            noise_level (`int`, *optional*, defaults to 250):
                The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct
        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]

        self.check_inputs(
            prompt,
            image,
            original_image,
            batch_size,
            callback_steps,
            negative_prompt,
            prompt_embeds,
            negative_prompt_embeds,
        )

        # 2. Define call parameters

        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        dtype = prompt_embeds.dtype

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

        # 5. prepare original image
        original_image = self.preprocess_original_image(original_image)
        original_image = original_image.to(dtype=dtype)

        # 6. Prepare intermediate images
        noise_timestep = timesteps[0:1]
        noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

        intermediate_images = self.prepare_intermediate_images(
            original_image,
            noise_timestep,
            batch_size,
            num_images_per_prompt,
            dtype,
            generator,
        )

        # 7. Prepare upscaled image and noise level
        _, _, height, width = original_image.shape

        image = self.preprocess_image(image, num_images_per_prompt)

        upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

        noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
        noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
        upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

        if do_classifier_free_guidance:
            noise_level = ops.cat([noise_level] * 2)

        # 8. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 9. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images, upscaled], axis=1)

                model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    class_labels=noise_level,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 10. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 11. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 12. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 13. Apply watermark
            if self.watermarker is not None:
                self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None
        else:
            # 10. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 11. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFImg2ImgSuperResolutionPipeline.__call__(image, original_image=None, strength=0.8, prompt=None, num_inference_steps=50, timesteps=None, guidance_scale=4.0, negative_prompt=None, num_images_per_prompt=1, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, cross_attention_kwargs=None, noise_level=250, clean_caption=True)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
image	Image, or tensor representing an image batch, that will be used as the starting point for the process. TYPE: `ms.Tensor` or `PIL.Image.Image`
original_image	The original image that image was varied from. TYPE: `ms.Tensor` or `PIL.Image.Image` DEFAULT: None
strength	Conceptually, indicates how much to transform the reference image. Must be between 0 and 1. image will be used as a starting point, adding more noise to it the larger the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise will be maximum and the denoising process will run for the full number of iterations specified in num_inference_steps. A value of 1, therefore, essentially ignores image. TYPE: `float`, *optional*, defaults to 0.8 DEFAULT: 0.8
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 50 DEFAULT: 50
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional* DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 4.0 DEFAULT: 4.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
noise_level	The amount of noise to add to the upscaled image. Must be in the range [0, 1000) TYPE: `int`, *optional*, defaults to 250 DEFAULT: 250
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img_superresolution.py

def __call__(
    self,
    image: Union[PIL.Image.Image, np.ndarray, ms.Tensor],
    original_image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    strength: float = 0.8,
    prompt: Union[str, List[str]] = None,
    num_inference_steps: int = 50,
    timesteps: List[int] = None,
    guidance_scale: float = 4.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    noise_level: int = 250,
    clean_caption: bool = True,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        image (`ms.Tensor` or `PIL.Image.Image`):
            `Image`, or tensor representing an image batch, that will be used as the starting point for the
            process.
        original_image (`ms.Tensor` or `PIL.Image.Image`):
            The original image that `image` was varied from.
        strength (`float`, *optional*, defaults to 0.8):
            Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
            will be used as a starting point, adding more noise to it the larger the `strength`. The number of
            denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
            be maximum and the denoising process will run for the full number of iterations specified in
            `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        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.
        timesteps (`List[int]`, *optional*):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 4.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        noise_level (`int`, *optional*, defaults to 250):
            The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct
    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]

    self.check_inputs(
        prompt,
        image,
        original_image,
        batch_size,
        callback_steps,
        negative_prompt,
        prompt_embeds,
        negative_prompt_embeds,
    )

    # 2. Define call parameters

    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    dtype = prompt_embeds.dtype

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

    # 5. prepare original image
    original_image = self.preprocess_original_image(original_image)
    original_image = original_image.to(dtype=dtype)

    # 6. Prepare intermediate images
    noise_timestep = timesteps[0:1]
    noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

    intermediate_images = self.prepare_intermediate_images(
        original_image,
        noise_timestep,
        batch_size,
        num_images_per_prompt,
        dtype,
        generator,
    )

    # 7. Prepare upscaled image and noise level
    _, _, height, width = original_image.shape

    image = self.preprocess_image(image, num_images_per_prompt)

    upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

    noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
    noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
    upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

    if do_classifier_free_guidance:
        noise_level = ops.cat([noise_level] * 2)

    # 8. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 9. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images, upscaled], axis=1)

            model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                class_labels=noise_level,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 10. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 11. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 12. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 13. Apply watermark
        if self.watermarker is not None:
            self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None
    else:
        # 10. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 11. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFImg2ImgSuperResolutionPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img_superresolution.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.IFInpaintingPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting.py

class IFInpaintingPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

    _optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]
    model_cpu_offload_seq = "text_encoder->unet"

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.run_safety_checker
    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_extra_step_kwargs
    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,
        image,
        mask_image,
        batch_size,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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}."
                )

        # image

        if isinstance(image, list):
            check_image_type = image[0]
        else:
            check_image_type = image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(image, list):
            image_batch_size = len(image)
        elif isinstance(image, ms.Tensor):
            image_batch_size = image.shape[0]
        elif isinstance(image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(image, np.ndarray):
            image_batch_size = image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(f"image batch size: {image_batch_size} must be same as prompt batch size {batch_size}")

        # mask_image

        if isinstance(mask_image, list):
            check_image_type = mask_image[0]
        else:
            check_image_type = mask_image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`mask_image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(mask_image, list):
            image_batch_size = len(mask_image)
        elif isinstance(mask_image, ms.Tensor):
            image_batch_size = mask_image.shape[0]
        elif isinstance(mask_image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(mask_image, np.ndarray):
            image_batch_size = mask_image.shape[0]
        else:
            assert False

        if image_batch_size != 1 and batch_size != image_batch_size:
            raise ValueError(
                f"mask_image batch size: {image_batch_size} must be `1` or the same as prompt batch size {batch_size}"
            )

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_img2img.IFImg2ImgPipeline.preprocess_image
    def preprocess_image(self, image: PIL.Image.Image) -> ms.Tensor:
        if not isinstance(image, list):
            image = [image]

        def numpy_to_ms(images):
            if images.ndim == 3:
                images = images[..., None]

            images = ms.Tensor.from_numpy(images.transpose(0, 3, 1, 2))
            return images

        if isinstance(image[0], PIL.Image.Image):
            new_image = []

            for image_ in image:
                image_ = image_.convert("RGB")
                image_ = resize(image_, self.unet.config.sample_size)
                image_ = np.array(image_)
                image_ = image_.astype(np.float32)
                image_ = image_ / 127.5 - 1
                new_image.append(image_)

            image = new_image

            image = np.stack(image, axis=0)  # to np
            image = numpy_to_ms(image)  # to pt

        elif isinstance(image[0], np.ndarray):
            image = np.concatenate(image, axis=0) if image[0].ndim == 4 else np.stack(image, axis=0)
            image = numpy_to_ms(image)

        elif isinstance(image[0], ms.Tensor):
            image = ops.cat(image, axis=0) if image[0].ndim == 4 else ops.stack(image, axis=0)

        return image

    def preprocess_mask_image(self, mask_image) -> ms.Tensor:
        if not isinstance(mask_image, list):
            mask_image = [mask_image]

        if isinstance(mask_image[0], ms.Tensor):
            mask_image = ops.cat(mask_image, axis=0) if mask_image[0].ndim == 4 else ops.stack(mask_image, axis=0)

            if mask_image.ndim == 2:
                # Batch and add channel dim for single mask
                mask_image = mask_image.unsqueeze(0).unsqueeze(0)
            elif mask_image.ndim == 3 and mask_image.shape[0] == 1:
                # Single mask, the 0'th dimension is considered to be
                # the existing batch size of 1
                mask_image = mask_image.unsqueeze(0)
            elif mask_image.ndim == 3 and mask_image.shape[0] != 1:
                # Batch of mask, the 0'th dimension is considered to be
                # the batching dimension
                mask_image = mask_image.unsqueeze(1)

            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1

        elif isinstance(mask_image[0], PIL.Image.Image):
            new_mask_image = []

            for mask_image_ in mask_image:
                mask_image_ = mask_image_.convert("L")
                mask_image_ = resize(mask_image_, self.unet.config.sample_size)
                mask_image_ = np.array(mask_image_)
                mask_image_ = mask_image_[None, None, :]
                new_mask_image.append(mask_image_)

            mask_image = new_mask_image

            mask_image = np.concatenate(mask_image, axis=0)
            mask_image = mask_image.astype(np.float32) / 255.0
            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1
            mask_image = ms.Tensor.from_numpy(mask_image)

        elif isinstance(mask_image[0], np.ndarray):
            mask_image = np.concatenate([m[None, None, :] for m in mask_image], axis=0)

            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1
            mask_image = ms.Tensor.from_numpy(mask_image)

        return mask_image

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
    def get_timesteps(self, num_inference_steps, strength):
        # get the original timestep using init_timestep
        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

        t_start = max(num_inference_steps - init_timestep, 0)
        timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(t_start * self.scheduler.order)

        return timesteps, num_inference_steps - t_start

    def prepare_intermediate_images(
        self, image, timestep, batch_size, num_images_per_prompt, dtype, mask_image, generator=None
    ):
        image_batch_size, channels, height, width = image.shape

        batch_size = batch_size * num_images_per_prompt

        shape = (batch_size, channels, height, width)

        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."
            )

        noise = randn_tensor(shape, generator=generator, dtype=dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)
        noised_image = self.scheduler.add_noise(image, noise, timestep)

        image = (1 - mask_image) * image + mask_image * noised_image

        return image

    def __call__(
        self,
        prompt: Union[str, List[str]] = None,
        image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        mask_image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        strength: float = 1.0,
        num_inference_steps: int = 50,
        timesteps: List[int] = None,
        guidance_scale: float = 7.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        clean_caption: bool = True,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            image (`ms.Tensor` or `PIL.Image.Image`):
                `Image`, or tensor representing an image batch, that will be used as the starting point for the
                process.
            mask_image (`PIL.Image.Image`):
                `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
                repainted, while black pixels will be preserved. If `mask_image` is a PIL image, it will be converted
                to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L)
                instead of 3, so the expected shape would be `(B, H, W, 1)`.
            strength (`float`, *optional*, defaults to 1.0):
                Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
                will be used as a starting point, adding more noise to it the larger the `strength`. The number of
                denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
                be maximum and the denoising process will run for the full number of iterations specified in
                `num_inference_steps`. A value of 1, therefore, essentially ignores `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.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 7.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct
        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]

        self.check_inputs(
            prompt,
            image,
            mask_image,
            batch_size,
            callback_steps,
            negative_prompt,
            prompt_embeds,
            negative_prompt_embeds,
        )

        # 2. Define call parameters

        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        dtype = prompt_embeds.dtype

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

        # 5. Prepare intermediate images
        image = self.preprocess_image(image)
        image = image.to(dtype=dtype)

        mask_image = self.preprocess_mask_image(mask_image)
        mask_image = mask_image.to(dtype=dtype)

        if mask_image.shape[0] == 1:
            mask_image = mask_image.repeat_interleave(batch_size * num_images_per_prompt, dim=0)
        else:
            mask_image = mask_image.repeat_interleave(num_images_per_prompt, dim=0)

        noise_timestep = timesteps[0:1]
        noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

        intermediate_images = self.prepare_intermediate_images(
            image, noise_timestep, batch_size, num_images_per_prompt, dtype, mask_image, generator
        )

        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 7. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                prev_intermediate_images = intermediate_images

                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

                intermediate_images = (1 - mask_image) * prev_intermediate_images + mask_image * intermediate_images

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 10. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 11. Apply watermark
            if self.watermarker is not None:
                self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None
        else:
            # 8. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 9. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFInpaintingPipeline.__call__(prompt=None, image=None, mask_image=None, strength=1.0, num_inference_steps=50, timesteps=None, guidance_scale=7.0, negative_prompt=None, num_images_per_prompt=1, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, clean_caption=True, cross_attention_kwargs=None)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
image	Image, or tensor representing an image batch, that will be used as the starting point for the process. TYPE: `ms.Tensor` or `PIL.Image.Image` DEFAULT: None
mask_image	Image, or tensor representing an image batch, to mask image. White pixels in the mask will be repainted, while black pixels will be preserved. If mask_image is a PIL image, it will be converted to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L) instead of 3, so the expected shape would be (B, H, W, 1). TYPE: `PIL.Image.Image` DEFAULT: None
strength	Conceptually, indicates how much to transform the reference image. Must be between 0 and 1. image will be used as a starting point, adding more noise to it the larger the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise will be maximum and the denoising process will run for the full number of iterations specified in num_inference_steps. A value of 1, therefore, essentially ignores image. TYPE: `float`, *optional*, defaults to 1.0 DEFAULT: 1.0
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 50 DEFAULT: 50
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional* DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 7.0 DEFAULT: 7.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting.py

def __call__(
    self,
    prompt: Union[str, List[str]] = None,
    image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    mask_image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    strength: float = 1.0,
    num_inference_steps: int = 50,
    timesteps: List[int] = None,
    guidance_scale: float = 7.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    clean_caption: bool = True,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        image (`ms.Tensor` or `PIL.Image.Image`):
            `Image`, or tensor representing an image batch, that will be used as the starting point for the
            process.
        mask_image (`PIL.Image.Image`):
            `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
            repainted, while black pixels will be preserved. If `mask_image` is a PIL image, it will be converted
            to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L)
            instead of 3, so the expected shape would be `(B, H, W, 1)`.
        strength (`float`, *optional*, defaults to 1.0):
            Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
            will be used as a starting point, adding more noise to it the larger the `strength`. The number of
            denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
            be maximum and the denoising process will run for the full number of iterations specified in
            `num_inference_steps`. A value of 1, therefore, essentially ignores `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.
        timesteps (`List[int]`, *optional*):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 7.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct
    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]

    self.check_inputs(
        prompt,
        image,
        mask_image,
        batch_size,
        callback_steps,
        negative_prompt,
        prompt_embeds,
        negative_prompt_embeds,
    )

    # 2. Define call parameters

    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    dtype = prompt_embeds.dtype

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

    # 5. Prepare intermediate images
    image = self.preprocess_image(image)
    image = image.to(dtype=dtype)

    mask_image = self.preprocess_mask_image(mask_image)
    mask_image = mask_image.to(dtype=dtype)

    if mask_image.shape[0] == 1:
        mask_image = mask_image.repeat_interleave(batch_size * num_images_per_prompt, dim=0)
    else:
        mask_image = mask_image.repeat_interleave(num_images_per_prompt, dim=0)

    noise_timestep = timesteps[0:1]
    noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

    intermediate_images = self.prepare_intermediate_images(
        image, noise_timestep, batch_size, num_images_per_prompt, dtype, mask_image, generator
    )

    # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 7. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images] * 2) if do_classifier_free_guidance else intermediate_images
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1], axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1], axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(model_input.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            prev_intermediate_images = intermediate_images

            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

            intermediate_images = (1 - mask_image) * prev_intermediate_images + mask_image * intermediate_images

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 10. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 11. Apply watermark
        if self.watermarker is not None:
            self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None
    else:
        # 8. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 9. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFInpaintingPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds

mindone.diffusers.IFInpaintingSuperResolutionPipeline

Bases: DiffusionPipeline, LoraLoaderMixin

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting_superresolution.py

class IFInpaintingSuperResolutionPipeline(DiffusionPipeline, LoraLoaderMixin):
    tokenizer: T5Tokenizer
    text_encoder: T5EncoderModel

    unet: UNet2DConditionModel
    scheduler: DDPMScheduler
    image_noising_scheduler: DDPMScheduler

    feature_extractor: Optional[CLIPImageProcessor]
    safety_checker: Optional[IFSafetyChecker]

    watermarker: Optional[IFWatermarker]

    bad_punct_regex = re.compile(
        r"["
        + "#®•©™&@·º½¾¿¡§~"
        + r"\)"
        + r"\("
        + r"\]"
        + r"\["
        + r"\}"
        + r"\{"
        + r"\|"
        + "\\"
        + r"\/"
        + r"\*"
        + r"]{1,}"
    )  # noqa

    model_cpu_offload_seq = "text_encoder->unet"
    _optional_components = ["tokenizer", "text_encoder", "safety_checker", "feature_extractor", "watermarker"]

    def __init__(
        self,
        tokenizer: T5Tokenizer,
        text_encoder: T5EncoderModel,
        unet: UNet2DConditionModel,
        scheduler: DDPMScheduler,
        image_noising_scheduler: DDPMScheduler,
        safety_checker: Optional[IFSafetyChecker],
        feature_extractor: Optional[CLIPImageProcessor],
        watermarker: Optional[IFWatermarker],
        requires_safety_checker: bool = True,
    ):
        super().__init__()

        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 IF 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."
            )

        if unet.config.in_channels != 6:
            logger.warning(
                f"It seems like you have loaded a checkpoint that shall not be used for super resolution from {unet.config._name_or_path} as it accepts {unet.config.in_channels} input channels instead of 6. Please make sure to pass a super resolution checkpoint as the `'unet'`: IFSuperResolutionPipeline.from_pretrained(unet=super_resolution_unet, ...)`."  # noqa E501
            )

        self.register_modules(
            tokenizer=tokenizer,
            text_encoder=text_encoder,
            unet=unet,
            scheduler=scheduler,
            image_noising_scheduler=image_noising_scheduler,
            safety_checker=safety_checker,
            feature_extractor=feature_extractor,
            watermarker=watermarker,
        )
        self.register_to_config(requires_safety_checker=requires_safety_checker)

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
    def _text_preprocessing(self, text, clean_caption=False):
        if clean_caption and not is_bs4_available():
            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if clean_caption and not is_ftfy_available():
            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
            logger.warning("Setting `clean_caption` to False...")
            clean_caption = False

        if not isinstance(text, (tuple, list)):
            text = [text]

        def process(text: str):
            if clean_caption:
                text = self._clean_caption(text)
                text = self._clean_caption(text)
            else:
                text = text.lower().strip()
            return text

        return [process(t) for t in text]

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
    def _clean_caption(self, caption):
        caption = str(caption)
        caption = ul.unquote_plus(caption)
        caption = caption.strip().lower()
        caption = re.sub("<person>", "person", caption)
        # urls:
        caption = re.sub(
            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        caption = re.sub(
            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
            "",
            caption,
        )  # regex for urls
        # html:
        caption = BeautifulSoup(caption, features="html.parser").text

        # @<nickname>
        caption = re.sub(r"@[\w\d]+\b", "", caption)

        # 31C0—31EF CJK Strokes
        # 31F0—31FF Katakana Phonetic Extensions
        # 3200—32FF Enclosed CJK Letters and Months
        # 3300—33FF CJK Compatibility
        # 3400—4DBF CJK Unified Ideographs Extension A
        # 4DC0—4DFF Yijing Hexagram Symbols
        # 4E00—9FFF CJK Unified Ideographs
        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
        #######################################################

        # все виды тире / all types of dash --> "-"
        caption = re.sub(
            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
            "-",
            caption,
        )

        # кавычки к одному стандарту
        caption = re.sub(r"[`´«»“”¨]", '"', caption)
        caption = re.sub(r"[‘’]", "'", caption)

        # &quot;
        caption = re.sub(r"&quot;?", "", caption)
        # &amp
        caption = re.sub(r"&amp", "", caption)

        # ip adresses:
        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)

        # article ids:
        caption = re.sub(r"\d:\d\d\s+$", "", caption)

        # \n
        caption = re.sub(r"\\n", " ", caption)

        # "#123"
        caption = re.sub(r"#\d{1,3}\b", "", caption)
        # "#12345.."
        caption = re.sub(r"#\d{5,}\b", "", caption)
        # "123456.."
        caption = re.sub(r"\b\d{6,}\b", "", caption)
        # filenames:
        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)

        #
        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""

        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "

        # this-is-my-cute-cat / this_is_my_cute_cat
        regex2 = re.compile(r"(?:\-|\_)")
        if len(re.findall(regex2, caption)) > 3:
            caption = re.sub(regex2, " ", caption)

        caption = ftfy.fix_text(caption)
        caption = html.unescape(html.unescape(caption))

        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231

        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
        caption = re.sub(r"\bpage\s+\d+\b", "", caption)

        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...

        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)

        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
        caption = re.sub(r"\s+", " ", caption)

        caption.strip()

        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
        caption = re.sub(r"^\.\S+$", "", caption)

        return caption.strip()

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.encode_prompt
    def encode_prompt(
        self,
        prompt: Union[str, List[str]],
        do_classifier_free_guidance: bool = True,
        num_images_per_prompt: int = 1,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        clean_caption: bool = False,
    ):
        r"""
        Encodes the prompt into text encoder hidden states.

        Args:
            prompt (`str` or `List[str]`, *optional*):
                prompt to be encoded
            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
                whether to use classifier free guidance or not
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                number of images that should be generated per prompt
            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. 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 (`ms.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 (`ms.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.
            clean_caption (bool, defaults to `False`):
                If `True`, the function will preprocess and clean the provided caption before encoding.
        """
        if prompt is not None and negative_prompt is not None:
            if 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)}."
                )

        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]

        # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
        max_length = 77

        if prompt_embeds is None:
            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
            text_inputs = self.tokenizer(
                prompt,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            text_input_ids = text_inputs.input_ids
            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

            attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

            prompt_embeds = self.text_encoder(
                ms.tensor(text_input_ids),
                attention_mask=attention_mask,
            )
            prompt_embeds = prompt_embeds[0]

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

        prompt_embeds = prompt_embeds.to(dtype=dtype)

        bs_embed, seq_len, _ = prompt_embeds.shape
        # duplicate text embeddings for each generation per prompt, using mps friendly method
        prompt_embeds = prompt_embeds.tile((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 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

            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
            max_length = prompt_embeds.shape[1]
            uncond_input = self.tokenizer(
                uncond_tokens,
                padding="max_length",
                max_length=max_length,
                truncation=True,
                return_attention_mask=True,
                add_special_tokens=True,
                return_tensors="np",
            )
            attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

            negative_prompt_embeds = self.text_encoder(
                ms.Tensor.from_numpy(uncond_input.input_ids),
                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=dtype)

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

            # 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
        else:
            negative_prompt_embeds = None

        return prompt_embeds, negative_prompt_embeds

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.run_safety_checker
    def run_safety_checker(self, image, dtype):
        if self.safety_checker is not None:
            safety_checker_input = self.feature_extractor(self.numpy_to_pil(image.numpy()), return_tensors="np")
            image, nsfw_detected, watermark_detected = self.safety_checker(
                images=image,
                clip_input=ms.Tensor.from_numpy(safety_checker_input.pixel_values).to(dtype=dtype),
            )
            if ops.any(ops.cat([nsfw_detected[..., None].int(), watermark_detected[..., None].int()], axis=1), axis=1):
                logger.warning(
                    "Potential NSFW or watermarked content was detected in one or more images. A black image will be returned instead."
                    " Try again with a different prompt and/or seed."
                )
        else:
            nsfw_detected = None
            watermark_detected = None

        return image, nsfw_detected, watermark_detected

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline.prepare_extra_step_kwargs
    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,
        image,
        original_image,
        mask_image,
        batch_size,
        callback_steps,
        negative_prompt=None,
        prompt_embeds=None,
        negative_prompt_embeds=None,
    ):
        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}."
                )

        # image

        if isinstance(image, list):
            check_image_type = image[0]
        else:
            check_image_type = image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(image, list):
            image_batch_size = len(image)
        elif isinstance(image, ms.Tensor):
            image_batch_size = image.shape[0]
        elif isinstance(image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(image, np.ndarray):
            image_batch_size = image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(f"image batch size: {image_batch_size} must be same as prompt batch size {batch_size}")

        # original_image

        if isinstance(original_image, list):
            check_image_type = original_image[0]
        else:
            check_image_type = original_image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`original_image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(original_image, list):
            image_batch_size = len(original_image)
        elif isinstance(original_image, ms.Tensor):
            image_batch_size = original_image.shape[0]
        elif isinstance(original_image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(original_image, np.ndarray):
            image_batch_size = original_image.shape[0]
        else:
            assert False

        if batch_size != image_batch_size:
            raise ValueError(
                f"original_image batch size: {image_batch_size} must be same as prompt batch size {batch_size}"
            )

        # mask_image

        if isinstance(mask_image, list):
            check_image_type = mask_image[0]
        else:
            check_image_type = mask_image

        if (
            not isinstance(check_image_type, ms.Tensor)
            and not isinstance(check_image_type, PIL.Image.Image)
            and not isinstance(check_image_type, np.ndarray)
        ):
            raise ValueError(
                "`mask_image` has to be of type `ms.Tensor`, `PIL.Image.Image`, `np.ndarray`, or List[...] but is"
                f" {type(check_image_type)}"
            )

        if isinstance(mask_image, list):
            image_batch_size = len(mask_image)
        elif isinstance(mask_image, ms.Tensor):
            image_batch_size = mask_image.shape[0]
        elif isinstance(mask_image, PIL.Image.Image):
            image_batch_size = 1
        elif isinstance(mask_image, np.ndarray):
            image_batch_size = mask_image.shape[0]
        else:
            assert False

        if image_batch_size != 1 and batch_size != image_batch_size:
            raise ValueError(
                f"mask_image batch size: {image_batch_size} must be `1` or the same as prompt batch size {batch_size}"
            )

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_img2img.IFImg2ImgPipeline.preprocess_image with preprocess_image -> preprocess_original_image
    def preprocess_original_image(self, image: PIL.Image.Image) -> ms.Tensor:
        if not isinstance(image, list):
            image = [image]

        def numpy_to_ms(images):
            if images.ndim == 3:
                images = images[..., None]

            images = ms.Tensor.from_numpy(images.transpose(0, 3, 1, 2))
            return images

        if isinstance(image[0], PIL.Image.Image):
            new_image = []

            for image_ in image:
                image_ = image_.convert("RGB")
                image_ = resize(image_, self.unet.config.sample_size)
                image_ = np.array(image_)
                image_ = image_.astype(np.float32)
                image_ = image_ / 127.5 - 1
                new_image.append(image_)

            image = new_image

            image = np.stack(image, axis=0)  # to np
            image = numpy_to_ms(image)  # to pt

        elif isinstance(image[0], np.ndarray):
            image = np.concatenate(image, axis=0) if image[0].ndim == 4 else np.stack(image, axis=0)
            image = numpy_to_ms(image)

        elif isinstance(image[0], ms.Tensor):
            image = ops.cat(image, axis=0) if image[0].ndim == 4 else ops.stack(image, axis=0)

        return image

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_superresolution.IFSuperResolutionPipeline.preprocess_image
    def preprocess_image(self, image, num_images_per_prompt):
        if not isinstance(image, ms.Tensor) and not isinstance(image, list):
            image = [image]

        if isinstance(image[0], PIL.Image.Image):
            image = [np.array(i).astype(np.float32) / 127.5 - 1.0 for i in image]

            image = np.stack(image, axis=0)  # to np
            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image[0], np.ndarray):
            image = np.stack(image, axis=0)  # to np
            if image.ndim == 5:
                image = image[0]

            image = ms.Tensor.from_numpy(image.transpose(0, 3, 1, 2))
        elif isinstance(image, list) and isinstance(image[0], ms.Tensor):
            dims = image[0].ndim

            if dims == 3:
                image = ops.stack(image, axis=0)
            elif dims == 4:
                image = ops.concat(image, axis=0)
            else:
                raise ValueError(f"Image must have 3 or 4 dimensions, instead got {dims}")

        image = image.to(dtype=self.unet.dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)

        return image

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_inpainting.IFInpaintingPipeline.preprocess_mask_image
    def preprocess_mask_image(self, mask_image) -> ms.Tensor:
        if not isinstance(mask_image, list):
            mask_image = [mask_image]

        if isinstance(mask_image[0], ms.Tensor):
            mask_image = ops.cat(mask_image, axis=0) if mask_image[0].ndim == 4 else ops.stack(mask_image, axis=0)

            if mask_image.ndim == 2:
                # Batch and add channel dim for single mask
                mask_image = mask_image.unsqueeze(0).unsqueeze(0)
            elif mask_image.ndim == 3 and mask_image.shape[0] == 1:
                # Single mask, the 0'th dimension is considered to be
                # the existing batch size of 1
                mask_image = mask_image.unsqueeze(0)
            elif mask_image.ndim == 3 and mask_image.shape[0] != 1:
                # Batch of mask, the 0'th dimension is considered to be
                # the batching dimension
                mask_image = mask_image.unsqueeze(1)

            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1

        elif isinstance(mask_image[0], PIL.Image.Image):
            new_mask_image = []

            for mask_image_ in mask_image:
                mask_image_ = mask_image_.convert("L")
                mask_image_ = resize(mask_image_, self.unet.config.sample_size)
                mask_image_ = np.array(mask_image_)
                mask_image_ = mask_image_[None, None, :]
                new_mask_image.append(mask_image_)

            mask_image = new_mask_image

            mask_image = np.concatenate(mask_image, axis=0)
            mask_image = mask_image.astype(np.float32) / 255.0
            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1
            mask_image = ms.Tensor.from_numpy(mask_image)

        elif isinstance(mask_image[0], np.ndarray):
            mask_image = np.concatenate([m[None, None, :] for m in mask_image], axis=0)

            mask_image[mask_image < 0.5] = 0
            mask_image[mask_image >= 0.5] = 1
            mask_image = ms.Tensor.from_numpy(mask_image)

        return mask_image

    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline.get_timesteps
    def get_timesteps(self, num_inference_steps, strength):
        # get the original timestep using init_timestep
        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

        t_start = max(num_inference_steps - init_timestep, 0)
        timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
        if hasattr(self.scheduler, "set_begin_index"):
            self.scheduler.set_begin_index(t_start * self.scheduler.order)

        return timesteps, num_inference_steps - t_start

    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if_inpainting.IFInpaintingPipeline.prepare_intermediate_images
    def prepare_intermediate_images(
        self, image, timestep, batch_size, num_images_per_prompt, dtype, mask_image, generator=None
    ):
        image_batch_size, channels, height, width = image.shape

        batch_size = batch_size * num_images_per_prompt

        shape = (batch_size, channels, height, width)

        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."
            )

        noise = randn_tensor(shape, generator=generator, dtype=dtype)

        image = image.repeat_interleave(num_images_per_prompt, dim=0)
        noised_image = self.scheduler.add_noise(image, noise, timestep)

        image = (1 - mask_image) * image + mask_image * noised_image

        return image

    def __call__(
        self,
        image: Union[PIL.Image.Image, np.ndarray, ms.Tensor],
        original_image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        mask_image: Union[
            PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
        ] = None,
        strength: float = 0.8,
        prompt: Union[str, List[str]] = None,
        num_inference_steps: int = 100,
        timesteps: List[int] = None,
        guidance_scale: float = 4.0,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        num_images_per_prompt: Optional[int] = 1,
        eta: float = 0.0,
        generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
        prompt_embeds: Optional[ms.Tensor] = None,
        negative_prompt_embeds: Optional[ms.Tensor] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = False,
        callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
        callback_steps: int = 1,
        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
        noise_level: int = 0,
        clean_caption: bool = True,
    ):
        """
        Function invoked when calling the pipeline for generation.

        Args:
            image (`ms.Tensor` or `PIL.Image.Image`):
                `Image`, or tensor representing an image batch, that will be used as the starting point for the
                process.
            original_image (`ms.Tensor` or `PIL.Image.Image`):
                The original image that `image` was varied from.
            mask_image (`PIL.Image.Image`):
                `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
                repainted, while black pixels will be preserved. If `mask_image` is a PIL image, it will be converted
                to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L)
                instead of 3, so the expected shape would be `(B, H, W, 1)`.
            strength (`float`, *optional*, defaults to 0.8):
                Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
                will be used as a starting point, adding more noise to it the larger the `strength`. The number of
                denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
                be maximum and the denoising process will run for the full number of iterations specified in
                `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
            prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                instead.
            num_inference_steps (`int`, *optional*, defaults to 100):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            timesteps (`List[int]`, *optional*):
                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
                timesteps are used. Must be in descending order.
            guidance_scale (`float`, *optional*, defaults to 4.0):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            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`).
            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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            prompt_embeds (`ms.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 (`ms.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.
            output_type (`str`, *optional*, defaults to `"pil"`):
                The output format of the generate image. Choose between
                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.
            cross_attention_kwargs (`dict`, *optional*):
                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                `self.processor` in
                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
            noise_level (`int`, *optional*, defaults to 0):
                The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
            clean_caption (`bool`, *optional*, defaults to `True`):
                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
                be installed. If the dependencies are not installed, the embeddings will be created from the raw
                prompt.

        Examples:

        Returns:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images, and the second element is a list
            of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
            or watermarked content, according to the `safety_checker`.
        """
        # 1. Check inputs. Raise error if not correct
        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]

        self.check_inputs(
            prompt,
            image,
            original_image,
            mask_image,
            batch_size,
            callback_steps,
            negative_prompt,
            prompt_embeds,
            negative_prompt_embeds,
        )

        # 2. Define call parameters

        # 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
        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
            prompt,
            do_classifier_free_guidance,
            num_images_per_prompt=num_images_per_prompt,
            negative_prompt=negative_prompt,
            prompt_embeds=prompt_embeds,
            negative_prompt_embeds=negative_prompt_embeds,
            clean_caption=clean_caption,
        )

        if do_classifier_free_guidance:
            prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

        dtype = prompt_embeds.dtype

        # 4. Prepare timesteps
        if timesteps is not None:
            self.scheduler.set_timesteps(timesteps=timesteps)
            timesteps = self.scheduler.timesteps
            num_inference_steps = len(timesteps)
        else:
            self.scheduler.set_timesteps(num_inference_steps)
            timesteps = self.scheduler.timesteps

        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

        # 5. prepare original image
        original_image = self.preprocess_original_image(original_image)
        original_image = original_image.to(dtype=dtype)

        # 6. prepare mask image
        mask_image = self.preprocess_mask_image(mask_image)
        mask_image = mask_image.to(dtype=dtype)

        if mask_image.shape[0] == 1:
            mask_image = mask_image.repeat_interleave(batch_size * num_images_per_prompt, dim=0)
        else:
            mask_image = mask_image.repeat_interleave(num_images_per_prompt, dim=0)

        # 6. Prepare intermediate images
        noise_timestep = timesteps[0:1]
        noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

        intermediate_images = self.prepare_intermediate_images(
            original_image,
            noise_timestep,
            batch_size,
            num_images_per_prompt,
            dtype,
            mask_image,
            generator,
        )

        # 7. Prepare upscaled image and noise level
        _, _, height, width = original_image.shape

        image = self.preprocess_image(image, num_images_per_prompt)

        upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

        noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
        noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
        upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

        if do_classifier_free_guidance:
            noise_level = ops.cat([noise_level] * 2)

        # 8. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
        # to the unet and will raise RuntimeError.
        lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
        if lora_scale is not None:
            # weight the lora layers by setting `lora_scale` for each PEFT layer
            scale_lora_layers(self.unet, lora_scale)

        # 9. Denoising loop
        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
        with self.progress_bar(total=num_inference_steps) as progress_bar:
            for i, t in enumerate(timesteps):
                model_input = ops.cat([intermediate_images, upscaled], axis=1)

                model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = model_input.dtype
                model_input = self.scheduler.scale_model_input(model_input, t)
                model_input = model_input.to(tmp_dtype)

                # predict the noise residual
                noise_pred = self.unet(
                    model_input,
                    t,
                    encoder_hidden_states=prompt_embeds,
                    class_labels=noise_level,
                    cross_attention_kwargs=cross_attention_kwargs,
                    return_dict=False,
                )[0]

                # perform guidance
                if do_classifier_free_guidance:
                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                    noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                    noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                    noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

                if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                    noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

                # compute the previous noisy sample x_t -> x_t-1
                prev_intermediate_images = intermediate_images

                # TODO: method of scheduler should not change the dtype of input.
                #  Remove the casting after cuiyushi confirm that.
                tmp_dtype = intermediate_images.dtype
                intermediate_images = self.scheduler.step(
                    noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
                )[0]
                intermediate_images = intermediate_images.to(tmp_dtype)

                intermediate_images = (1 - mask_image) * prev_intermediate_images + mask_image * intermediate_images

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

        if lora_scale is not None:
            # remove `lora_scale` from each PEFT layer
            unscale_lora_layers(self.unet, lora_scale)

        image = intermediate_images

        if output_type == "pil":
            # 10. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 11. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

            # 12. Convert to PIL
            image = self.numpy_to_pil(image.numpy())

            # 13. Apply watermark
            if self.watermarker is not None:
                self.watermarker.apply_watermark(image, self.unet.config.sample_size)
        elif output_type == "ms":
            nsfw_detected = None
            watermark_detected = None

        else:
            # 10. Post-processing
            image = (image / 2 + 0.5).clamp(0, 1)
            image = image.permute(0, 2, 3, 1).float()

            # 11. Run safety checker
            image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
            image = image.numpy()

        if not return_dict:
            return (image, nsfw_detected, watermark_detected)

        return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFInpaintingSuperResolutionPipeline.__call__(image, original_image=None, mask_image=None, strength=0.8, prompt=None, num_inference_steps=100, timesteps=None, guidance_scale=4.0, negative_prompt=None, num_images_per_prompt=1, eta=0.0, generator=None, prompt_embeds=None, negative_prompt_embeds=None, output_type='pil', return_dict=False, callback=None, callback_steps=1, cross_attention_kwargs=None, noise_level=0, clean_caption=True)

Function invoked when calling the pipeline for generation.

PARAMETER	DESCRIPTION
image	Image, or tensor representing an image batch, that will be used as the starting point for the process. TYPE: `ms.Tensor` or `PIL.Image.Image`
original_image	The original image that image was varied from. TYPE: `ms.Tensor` or `PIL.Image.Image` DEFAULT: None
mask_image	Image, or tensor representing an image batch, to mask image. White pixels in the mask will be repainted, while black pixels will be preserved. If mask_image is a PIL image, it will be converted to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L) instead of 3, so the expected shape would be (B, H, W, 1). TYPE: `PIL.Image.Image` DEFAULT: None
strength	Conceptually, indicates how much to transform the reference image. Must be between 0 and 1. image will be used as a starting point, adding more noise to it the larger the strength. The number of denoising steps depends on the amount of noise initially added. When strength is 1, added noise will be maximum and the denoising process will run for the full number of iterations specified in num_inference_steps. A value of 1, therefore, essentially ignores image. TYPE: `float`, *optional*, defaults to 0.8 DEFAULT: 0.8
prompt	The prompt or prompts to guide the image generation. If not defined, one has to pass prompt_embeds. instead. TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_inference_steps	The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference. TYPE: `int`, *optional*, defaults to 100 DEFAULT: 100
timesteps	Custom timesteps to use for the denoising process. If not defined, equal spaced num_inference_steps timesteps are used. Must be in descending order. TYPE: `List[int]`, *optional* DEFAULT: None
guidance_scale	Guidance scale as defined in Classifier-Free Diffusion Guidance. guidance_scale is defined as w of equation 2. of Imagen Paper. Guidance scale is enabled by setting guidance_scale > 1. Higher guidance scale encourages to generate images that are closely linked to the text prompt, usually at the expense of lower image quality. TYPE: `float`, *optional*, defaults to 4.0 DEFAULT: 4.0
negative_prompt	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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
num_images_per_prompt	The number of images to generate per prompt. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
eta	Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to [schedulers.DDIMScheduler], will be ignored for others. TYPE: `float`, *optional*, defaults to 0.0 DEFAULT: 0.0
generator	One or a list of torch generator(s) to make generation deterministic. TYPE: `np.random.Generator` or `List[np.random.Generator]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
output_type	The output format of the generate image. Choose between PIL: PIL.Image.Image or np.array. TYPE: `str`, *optional*, defaults to `"pil"` DEFAULT: 'pil'
return_dict	Whether or not to return a [~pipelines.stable_diffusion.IFPipelineOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False
callback	A function that will be called every callback_steps steps during inference. The function will be called with the following arguments: callback(step: int, timestep: int, latents: ms.Tensor). TYPE: `Callable`, *optional* DEFAULT: None
callback_steps	The frequency at which the callback function will be called. If not specified, the callback will be called at every step. TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
cross_attention_kwargs	A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor. TYPE: `dict`, *optional* DEFAULT: None
noise_level	The amount of noise to add to the upscaled image. Must be in the range [0, 1000) TYPE: `int`, *optional*, defaults to 0 DEFAULT: 0
clean_caption	Whether or not to clean the caption before creating embeddings. Requires beautifulsoup4 and ftfy to be installed. If the dependencies are not installed, the embeddings will be created from the raw prompt. TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True

RETURNS	DESCRIPTION
	[~pipelines.stable_diffusion.IFPipelineOutput] or tuple:
	[~pipelines.stable_diffusion.IFPipelineOutput] if return_dict is True, otherwise a `tuple. When
	returning a tuple, the first element is a list with the generated images, and the second element is a list
	of bools denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
	or watermarked content, according to the safety_checker.

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting_superresolution.py

def __call__(
    self,
    image: Union[PIL.Image.Image, np.ndarray, ms.Tensor],
    original_image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    mask_image: Union[
        PIL.Image.Image, ms.Tensor, np.ndarray, List[PIL.Image.Image], List[ms.Tensor], List[np.ndarray]
    ] = None,
    strength: float = 0.8,
    prompt: Union[str, List[str]] = None,
    num_inference_steps: int = 100,
    timesteps: List[int] = None,
    guidance_scale: float = 4.0,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    num_images_per_prompt: Optional[int] = 1,
    eta: float = 0.0,
    generator: Optional[Union[np.random.Generator, List[np.random.Generator]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    output_type: Optional[str] = "pil",
    return_dict: bool = False,
    callback: Optional[Callable[[int, int, ms.Tensor], None]] = None,
    callback_steps: int = 1,
    cross_attention_kwargs: Optional[Dict[str, Any]] = None,
    noise_level: int = 0,
    clean_caption: bool = True,
):
    """
    Function invoked when calling the pipeline for generation.

    Args:
        image (`ms.Tensor` or `PIL.Image.Image`):
            `Image`, or tensor representing an image batch, that will be used as the starting point for the
            process.
        original_image (`ms.Tensor` or `PIL.Image.Image`):
            The original image that `image` was varied from.
        mask_image (`PIL.Image.Image`):
            `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
            repainted, while black pixels will be preserved. If `mask_image` is a PIL image, it will be converted
            to a single channel (luminance) before use. If it's a tensor, it should contain one color channel (L)
            instead of 3, so the expected shape would be `(B, H, W, 1)`.
        strength (`float`, *optional*, defaults to 0.8):
            Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image`
            will be used as a starting point, adding more noise to it the larger the `strength`. The number of
            denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will
            be maximum and the denoising process will run for the full number of iterations specified in
            `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
        prompt (`str` or `List[str]`, *optional*):
            The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
            instead.
        num_inference_steps (`int`, *optional*, defaults to 100):
            The number of denoising steps. More denoising steps usually lead to a higher quality image at the
            expense of slower inference.
        timesteps (`List[int]`, *optional*):
            Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
            timesteps are used. Must be in descending order.
        guidance_scale (`float`, *optional*, defaults to 4.0):
            Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
            `guidance_scale` is defined as `w` of equation 2. of [Imagen
            Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
            1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
            usually at the expense of lower image quality.
        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`).
        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) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
            [`schedulers.DDIMScheduler`], will be ignored for others.
        generator (`np.random.Generator` or `List[np.random.Generator]`, *optional*):
            One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
            to make generation deterministic.
        prompt_embeds (`ms.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 (`ms.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.
        output_type (`str`, *optional*, defaults to `"pil"`):
            The output format of the generate image. Choose between
            [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
        callback (`Callable`, *optional*):
            A function that will be called every `callback_steps` steps during inference. The function will be
            called with the following arguments: `callback(step: int, timestep: int, latents: ms.Tensor)`.
        callback_steps (`int`, *optional*, defaults to 1):
            The frequency at which the `callback` function will be called. If not specified, the callback will be
            called at every step.
        cross_attention_kwargs (`dict`, *optional*):
            A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
            `self.processor` in
            [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
        noise_level (`int`, *optional*, defaults to 0):
            The amount of noise to add to the upscaled image. Must be in the range `[0, 1000)`
        clean_caption (`bool`, *optional*, defaults to `True`):
            Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
            be installed. If the dependencies are not installed, the embeddings will be created from the raw
            prompt.

    Examples:

    Returns:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] or `tuple`:
        [`~pipelines.stable_diffusion.IFPipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
        returning a tuple, the first element is a list with the generated images, and the second element is a list
        of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" (nsfw)
        or watermarked content, according to the `safety_checker`.
    """
    # 1. Check inputs. Raise error if not correct
    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]

    self.check_inputs(
        prompt,
        image,
        original_image,
        mask_image,
        batch_size,
        callback_steps,
        negative_prompt,
        prompt_embeds,
        negative_prompt_embeds,
    )

    # 2. Define call parameters

    # 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
    prompt_embeds, negative_prompt_embeds = self.encode_prompt(
        prompt,
        do_classifier_free_guidance,
        num_images_per_prompt=num_images_per_prompt,
        negative_prompt=negative_prompt,
        prompt_embeds=prompt_embeds,
        negative_prompt_embeds=negative_prompt_embeds,
        clean_caption=clean_caption,
    )

    if do_classifier_free_guidance:
        prompt_embeds = ops.cat([negative_prompt_embeds, prompt_embeds])

    dtype = prompt_embeds.dtype

    # 4. Prepare timesteps
    if timesteps is not None:
        self.scheduler.set_timesteps(timesteps=timesteps)
        timesteps = self.scheduler.timesteps
        num_inference_steps = len(timesteps)
    else:
        self.scheduler.set_timesteps(num_inference_steps)
        timesteps = self.scheduler.timesteps

    timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength)

    # 5. prepare original image
    original_image = self.preprocess_original_image(original_image)
    original_image = original_image.to(dtype=dtype)

    # 6. prepare mask image
    mask_image = self.preprocess_mask_image(mask_image)
    mask_image = mask_image.to(dtype=dtype)

    if mask_image.shape[0] == 1:
        mask_image = mask_image.repeat_interleave(batch_size * num_images_per_prompt, dim=0)
    else:
        mask_image = mask_image.repeat_interleave(num_images_per_prompt, dim=0)

    # 6. Prepare intermediate images
    noise_timestep = timesteps[0:1]
    noise_timestep = noise_timestep.tile((batch_size * num_images_per_prompt,))

    intermediate_images = self.prepare_intermediate_images(
        original_image,
        noise_timestep,
        batch_size,
        num_images_per_prompt,
        dtype,
        mask_image,
        generator,
    )

    # 7. Prepare upscaled image and noise level
    _, _, height, width = original_image.shape

    image = self.preprocess_image(image, num_images_per_prompt)

    upscaled = ops.interpolate(image, (height, width), mode="bilinear", align_corners=True)

    noise_level = ms.Tensor([noise_level] * upscaled.shape[0])
    noise = randn_tensor(upscaled.shape, generator=generator, dtype=upscaled.dtype)
    upscaled = self.image_noising_scheduler.add_noise(upscaled, noise, timesteps=noise_level)

    if do_classifier_free_guidance:
        noise_level = ops.cat([noise_level] * 2)

    # 8. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
    extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)

    # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
    # to the unet and will raise RuntimeError.
    lora_scale = cross_attention_kwargs.pop("scale", None) if cross_attention_kwargs is not None else None
    if lora_scale is not None:
        # weight the lora layers by setting `lora_scale` for each PEFT layer
        scale_lora_layers(self.unet, lora_scale)

    # 9. Denoising loop
    num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
    with self.progress_bar(total=num_inference_steps) as progress_bar:
        for i, t in enumerate(timesteps):
            model_input = ops.cat([intermediate_images, upscaled], axis=1)

            model_input = ops.cat([model_input] * 2) if do_classifier_free_guidance else model_input
            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = model_input.dtype
            model_input = self.scheduler.scale_model_input(model_input, t)
            model_input = model_input.to(tmp_dtype)

            # predict the noise residual
            noise_pred = self.unet(
                model_input,
                t,
                encoder_hidden_states=prompt_embeds,
                class_labels=noise_level,
                cross_attention_kwargs=cross_attention_kwargs,
                return_dict=False,
            )[0]

            # perform guidance
            if do_classifier_free_guidance:
                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
                noise_pred_uncond, _ = noise_pred_uncond.split(model_input.shape[1] // 2, axis=1)
                noise_pred_text, predicted_variance = noise_pred_text.split(model_input.shape[1] // 2, axis=1)
                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
                noise_pred = ops.cat([noise_pred, predicted_variance], axis=1)

            if self.scheduler.config.variance_type not in ["learned", "learned_range"]:
                noise_pred, _ = noise_pred.split(intermediate_images.shape[1], axis=1)

            # compute the previous noisy sample x_t -> x_t-1
            prev_intermediate_images = intermediate_images

            # TODO: method of scheduler should not change the dtype of input.
            #  Remove the casting after cuiyushi confirm that.
            tmp_dtype = intermediate_images.dtype
            intermediate_images = self.scheduler.step(
                noise_pred, t, intermediate_images, **extra_step_kwargs, return_dict=False
            )[0]
            intermediate_images = intermediate_images.to(tmp_dtype)

            intermediate_images = (1 - mask_image) * prev_intermediate_images + mask_image * intermediate_images

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

    if lora_scale is not None:
        # remove `lora_scale` from each PEFT layer
        unscale_lora_layers(self.unet, lora_scale)

    image = intermediate_images

    if output_type == "pil":
        # 10. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 11. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)

        # 12. Convert to PIL
        image = self.numpy_to_pil(image.numpy())

        # 13. Apply watermark
        if self.watermarker is not None:
            self.watermarker.apply_watermark(image, self.unet.config.sample_size)
    elif output_type == "ms":
        nsfw_detected = None
        watermark_detected = None

    else:
        # 10. Post-processing
        image = (image / 2 + 0.5).clamp(0, 1)
        image = image.permute(0, 2, 3, 1).float()

        # 11. Run safety checker
        image, nsfw_detected, watermark_detected = self.run_safety_checker(image, prompt_embeds.dtype)
        image = image.numpy()

    if not return_dict:
        return (image, nsfw_detected, watermark_detected)

    return IFPipelineOutput(images=image, nsfw_detected=nsfw_detected, watermark_detected=watermark_detected)

mindone.diffusers.IFInpaintingSuperResolutionPipeline.encode_prompt(prompt, do_classifier_free_guidance=True, num_images_per_prompt=1, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None, clean_caption=False)

Encodes the prompt into text encoder hidden states.

PARAMETER	DESCRIPTION
prompt	prompt to be encoded TYPE: `str` or `List[str]`, *optional*
do_classifier_free_guidance	whether to use classifier free guidance or not TYPE: `bool`, *optional*, defaults to `True` DEFAULT: True
num_images_per_prompt	number of images that should be generated per prompt TYPE: `int`, *optional*, defaults to 1 DEFAULT: 1
negative_prompt	The prompt or prompts not to guide the image generation. If not defined, one has to pass negative_prompt_embeds. instead. 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). TYPE: `str` or `List[str]`, *optional* DEFAULT: None
prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
negative_prompt_embeds	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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
clean_caption	If True, the function will preprocess and clean the provided caption before encoding. TYPE: bool, defaults to `False` DEFAULT: False

Source code in mindone/diffusers/pipelines/deepfloyd_if/pipeline_if_inpainting_superresolution.py

def encode_prompt(
    self,
    prompt: Union[str, List[str]],
    do_classifier_free_guidance: bool = True,
    num_images_per_prompt: int = 1,
    negative_prompt: Optional[Union[str, List[str]]] = None,
    prompt_embeds: Optional[ms.Tensor] = None,
    negative_prompt_embeds: Optional[ms.Tensor] = None,
    clean_caption: bool = False,
):
    r"""
    Encodes the prompt into text encoder hidden states.

    Args:
        prompt (`str` or `List[str]`, *optional*):
            prompt to be encoded
        do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
            whether to use classifier free guidance or not
        num_images_per_prompt (`int`, *optional*, defaults to 1):
            number of images that should be generated per prompt
        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. 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 (`ms.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 (`ms.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.
        clean_caption (bool, defaults to `False`):
            If `True`, the function will preprocess and clean the provided caption before encoding.
    """
    if prompt is not None and negative_prompt is not None:
        if 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)}."
            )

    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]

    # while T5 can handle much longer input sequences than 77, the text encoder was trained with a max length of 77 for IF
    max_length = 77

    if prompt_embeds is None:
        prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
        text_inputs = self.tokenizer(
            prompt,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        text_input_ids = text_inputs.input_ids
        untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="np").input_ids

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

        attention_mask = ms.Tensor.from_numpy(text_inputs.attention_mask)

        prompt_embeds = self.text_encoder(
            ms.tensor(text_input_ids),
            attention_mask=attention_mask,
        )
        prompt_embeds = prompt_embeds[0]

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

    prompt_embeds = prompt_embeds.to(dtype=dtype)

    bs_embed, seq_len, _ = prompt_embeds.shape
    # duplicate text embeddings for each generation per prompt, using mps friendly method
    prompt_embeds = prompt_embeds.tile((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 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

        uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
        max_length = prompt_embeds.shape[1]
        uncond_input = self.tokenizer(
            uncond_tokens,
            padding="max_length",
            max_length=max_length,
            truncation=True,
            return_attention_mask=True,
            add_special_tokens=True,
            return_tensors="np",
        )
        attention_mask = ms.Tensor.from_numpy(uncond_input.attention_mask)

        negative_prompt_embeds = self.text_encoder(
            ms.Tensor.from_numpy(uncond_input.input_ids),
            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=dtype)

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

        # 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
    else:
        negative_prompt_embeds = None

    return prompt_embeds, negative_prompt_embeds