Update pipeline.py

pipeline.py

@@ -1,45 +1,66 @@
-# Implementation of StableDiffusionPAGPipeline
+# Implementation of StableDiffusionXLPAGPipeline
 
 import inspect
-from typing import Any, Callable, Dict, List, Optional, Union
+from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import torch
 import torch.nn.functional as F
 from packaging import version
-from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
-from diffusers.configuration_utils import FrozenDict
+from transformers import (
+    CLIPImageProcessor,
+    CLIPTextModel,
+    CLIPTextModelWithProjection,
+    CLIPTokenizer,
+    CLIPVisionModelWithProjection,
+)
+
 from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
-from diffusers.loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from diffusers.loaders import (
+    FromSingleFileMixin,
+    IPAdapterMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    TextualInversionLoaderMixin,
+)
 from diffusers.models import AutoencoderKL, ImageProjection, UNet2DConditionModel
-from diffusers.models.attention_processor import FusedAttnProcessor2_0
+from diffusers.models.attention_processor import (
+    AttnProcessor2_0,
+    FusedAttnProcessor2_0,
+    LoRAAttnProcessor2_0,
+    LoRAXFormersAttnProcessor,
+    XFormersAttnProcessor,
+)
 from diffusers.models.lora import adjust_lora_scale_text_encoder
 from diffusers.schedulers import KarrasDiffusionSchedulers
 from diffusers.utils import (
     USE_PEFT_BACKEND,
     deprecate,
+    is_invisible_watermark_available,
+    is_torch_xla_available,
     logging,
     replace_example_docstring,
     scale_lora_layers,
     unscale_lora_layers,
 )
 from diffusers.utils.torch_utils import randn_tensor
-from diffusers.pipelines.pipeline_utils import DiffusionPipeline
-from diffusers.pipelines.stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
-from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from diffusers.pipelines.pipeline_utils import DiffusionPipeline, StableDiffusionMixin
+from diffusers.pipelines.stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput
 
 from diffusers.models.attention_processor import Attention, AttnProcessor2_0
 
-
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 EXAMPLE_DOC_STRING = """
     Examples:
         ```py
         >>> import torch
-        >>> from diffusers import StableDiffusionPipeline
-        >>> pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+        >>> from diffusers import StableDiffusionXLPipeline
+
+        >>> pipe = StableDiffusionXLPipeline.from_pretrained(
+        ...     "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+        ... )
         >>> pipe = pipe.to("cuda")
+
         >>> prompt = "a photo of an astronaut riding a horse on mars"
         >>> image = pipe(prompt).images[0]
         ```
@@ -271,6 +292,17 @@ class PAGCFGIdentitySelfAttnProcessor:
 
         return hidden_states
 
+if is_invisible_watermark_available():
+    from .watermark import StableDiffusionXLWatermarker
+
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
 def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
     """
@@ -296,18 +328,20 @@ def retrieve_timesteps(
     """
     Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
     custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
+
     Args:
         scheduler (`SchedulerMixin`):
             The scheduler to get timesteps from.
         num_inference_steps (`int`):
-            The number of diffusion steps used when generating samples with a pre-trained model. If used,
-            `timesteps` must be `None`.
+            The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps`
+            must be `None`.
         device (`str` or `torch.device`, *optional*):
             The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
         timesteps (`List[int]`, *optional*):
                 Custom timesteps used to support arbitrary spacing between timesteps. If `None`, then the default
                 timestep spacing strategy of the scheduler is used. If `timesteps` is passed, `num_inference_steps`
                 must be `None`.
+
     Returns:
         `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the
         second element is the number of inference steps.
@@ -327,215 +361,143 @@ def retrieve_timesteps(
         timesteps = scheduler.timesteps
     return timesteps, num_inference_steps
 
-
-class StableDiffusionPipeline(
-    DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin, IPAdapterMixin, FromSingleFileMixin
+class StableDiffusionXLPipeline(
+    DiffusionPipeline,
+    StableDiffusionMixin,
+    FromSingleFileMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    TextualInversionLoaderMixin,
+    IPAdapterMixin,
 ):
     r"""
-    Pipeline for text-to-image generation using Stable Diffusion.
-    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
-    implemented for all pipelines (downloading, saving, running on a particular device, etc.).
+    Pipeline for text-to-image generation using Stable Diffusion XL.
+
+    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the
+    library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
+
     The pipeline also inherits the following loading methods:
         - [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
-        - [`~loaders.LoraLoaderMixin.load_lora_weights`] for loading LoRA weights
-        - [`~loaders.LoraLoaderMixin.save_lora_weights`] for saving LoRA weights
         - [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
+        - [`~loaders.StableDiffusionXLLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
+        - [`~loaders.StableDiffusionXLLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
         - [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
+
     Args:
         vae ([`AutoencoderKL`]):
-            Variational Auto-Encoder (VAE) model to encode and decode images to and from latent representations.
-        text_encoder ([`~transformers.CLIPTextModel`]):
-            Frozen text-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).
-        tokenizer ([`~transformers.CLIPTokenizer`]):
-            A `CLIPTokenizer` to tokenize text.
-        unet ([`UNet2DConditionModel`]):
-            A `UNet2DConditionModel` to denoise the encoded image latents.
+            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
+        text_encoder ([`CLIPTextModel`]):
+            Frozen text-encoder. Stable Diffusion XL uses the text portion of
+            [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically
+            the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant.
+        text_encoder_2 ([` CLIPTextModelWithProjection`]):
+            Second frozen text-encoder. Stable Diffusion XL uses the text and pool portion of
+            [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModelWithProjection),
+            specifically the
+            [laion/CLIP-ViT-bigG-14-laion2B-39B-b160k](https://huggingface.co/laion/CLIP-ViT-bigG-14-laion2B-39B-b160k)
+            variant.
+        tokenizer (`CLIPTokenizer`):
+            Tokenizer of class
+            [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer).
+        tokenizer_2 (`CLIPTokenizer`):
+            Second Tokenizer of class
+            [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer).
+        unet ([`UNet2DConditionModel`]): Conditional U-Net architecture to denoise the encoded image latents.
         scheduler ([`SchedulerMixin`]):
             A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
-        safety_checker ([`StableDiffusionSafetyChecker`]):
-            Classification module that estimates whether generated images could be considered offensive or harmful.
-            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
-            about a model's potential harms.
-        feature_extractor ([`~transformers.CLIPImageProcessor`]):
-            A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
+        force_zeros_for_empty_prompt (`bool`, *optional*, defaults to `"True"`):
+            Whether the negative prompt embeddings shall be forced to always be set to 0. Also see the config of
+            `stabilityai/stable-diffusion-xl-base-1-0`.
+        add_watermarker (`bool`, *optional*):
+            Whether to use the [invisible_watermark library](https://github.com/ShieldMnt/invisible-watermark/) to
+            watermark output images. If not defined, it will default to True if the package is installed, otherwise no
+            watermarker will be used.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
-    _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
-    _exclude_from_cpu_offload = ["safety_checker"]
-    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->unet->vae"
+    _optional_components = [
+        "tokenizer",
+        "tokenizer_2",
+        "text_encoder",
+        "text_encoder_2",
+        "image_encoder",
+        "feature_extractor",
+    ]
+    _callback_tensor_inputs = [
+        "latents",
+        "prompt_embeds",
+        "negative_prompt_embeds",
+        "add_text_embeds",
+        "add_time_ids",
+        "negative_pooled_prompt_embeds",
+        "negative_add_time_ids",
+    ]
 
     def __init__(
         self,
         vae: AutoencoderKL,
         text_encoder: CLIPTextModel,
+        text_encoder_2: CLIPTextModelWithProjection,
         tokenizer: CLIPTokenizer,
+        tokenizer_2: CLIPTokenizer,
         unet: UNet2DConditionModel,
         scheduler: KarrasDiffusionSchedulers,
-        safety_checker: StableDiffusionSafetyChecker,
-        feature_extractor: CLIPImageProcessor,
         image_encoder: CLIPVisionModelWithProjection = None,
-        requires_safety_checker: bool = True,
+        feature_extractor: CLIPImageProcessor = None,
+        force_zeros_for_empty_prompt: bool = True,
+        add_watermarker: Optional[bool] = None,
     ):
         super().__init__()
 
-        if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1:
-            deprecation_message = (
-                f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`"
-                f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure "
-                "to update the config accordingly as leaving `steps_offset` might led to incorrect results"
-                " in future versions. If you have downloaded this checkpoint from the Hugging Face Hub,"
-                " it would be very nice if you could open a Pull request for the `scheduler/scheduler_config.json`"
-                " file"
-            )
-            deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False)
-            new_config = dict(scheduler.config)
-            new_config["steps_offset"] = 1
-            scheduler._internal_dict = FrozenDict(new_config)
-
-        if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample is True:
-            deprecation_message = (
-                f"The configuration file of this scheduler: {scheduler} has not set the configuration `clip_sample`."
-                " `clip_sample` should be set to False in the configuration file. Please make sure to update the"
-                " config accordingly as not setting `clip_sample` in the config might lead to incorrect results in"
-                " future versions. If you have downloaded this checkpoint from the Hugging Face Hub, it would be very"
-                " nice if you could open a Pull request for the `scheduler/scheduler_config.json` file"
-            )
-            deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False)
-            new_config = dict(scheduler.config)
-            new_config["clip_sample"] = False
-            scheduler._internal_dict = FrozenDict(new_config)
-
-        if safety_checker is None and requires_safety_checker:
-            logger.warning(
-                f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
-                " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered"
-                " results in services or applications open to the public. Both the diffusers team and Hugging Face"
-                " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling"
-                " it only for use-cases that involve analyzing network behavior or auditing its results. For more"
-                " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
-            )
-
-        if safety_checker is not None and feature_extractor is None:
-            raise ValueError(
-                "Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety"
-                " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
-            )
-
-        is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
-            version.parse(unet.config._diffusers_version).base_version
-        ) < version.parse("0.9.0.dev0")
-        is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
-        if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
-            deprecation_message = (
-                "The configuration file of the unet has set the default `sample_size` to smaller than"
-                " 64 which seems highly unlikely. If your checkpoint is a fine-tuned version of any of the"
-                " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
-                " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
-                " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
-                " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
-                " in the config might lead to incorrect results in future versions. If you have downloaded this"
-                " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
-                " the `unet/config.json` file"
-            )
-            deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
-            new_config = dict(unet.config)
-            new_config["sample_size"] = 64
-            unet._internal_dict = FrozenDict(new_config)
-
         self.register_modules(
             vae=vae,
             text_encoder=text_encoder,
+            text_encoder_2=text_encoder_2,
             tokenizer=tokenizer,
+            tokenizer_2=tokenizer_2,
             unet=unet,
             scheduler=scheduler,
-            safety_checker=safety_checker,
-            feature_extractor=feature_extractor,
             image_encoder=image_encoder,
+            feature_extractor=feature_extractor,
         )
+        self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
-        self.register_to_config(requires_safety_checker=requires_safety_checker)
 
-    def enable_vae_slicing(self):
-        r"""
-        Enable sliced VAE decoding. When this option is enabled, the VAE will split the input tensor in slices to
-        compute decoding in several steps. This is useful to save some memory and allow larger batch sizes.
-        """
-        self.vae.enable_slicing()
+        self.default_sample_size = self.unet.config.sample_size
 
-    def disable_vae_slicing(self):
-        r"""
-        Disable sliced VAE decoding. If `enable_vae_slicing` was previously enabled, this method will go back to
-        computing decoding in one step.
-        """
-        self.vae.disable_slicing()
+        add_watermarker = add_watermarker if add_watermarker is not None else is_invisible_watermark_available()
 
-    def enable_vae_tiling(self):
-        r"""
-        Enable tiled VAE decoding. When this option is enabled, the VAE will split the input tensor into tiles to
-        compute decoding and encoding in several steps. This is useful for saving a large amount of memory and to allow
-        processing larger images.
-        """
-        self.vae.enable_tiling()
-
-    def disable_vae_tiling(self):
-        r"""
-        Disable tiled VAE decoding. If `enable_vae_tiling` was previously enabled, this method will go back to
-        computing decoding in one step.
-        """
-        self.vae.disable_tiling()
-
-    def _encode_prompt(
-        self,
-        prompt,
-        device,
-        num_images_per_prompt,
-        do_classifier_free_guidance,
-        negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        lora_scale: Optional[float] = None,
-        **kwargs,
-    ):
-        deprecation_message = "`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()` instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
-        deprecate("_encode_prompt()", "1.0.0", deprecation_message, standard_warn=False)
-
-        prompt_embeds_tuple = self.encode_prompt(
-            prompt=prompt,
-            device=device,
-            num_images_per_prompt=num_images_per_prompt,
-            do_classifier_free_guidance=do_classifier_free_guidance,
-            negative_prompt=negative_prompt,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            lora_scale=lora_scale,
-            **kwargs,
-        )
-
-        # concatenate for backwards comp
-        prompt_embeds = torch.cat([prompt_embeds_tuple[1], prompt_embeds_tuple[0]])
-
-        return prompt_embeds
+        if add_watermarker:
+            self.watermark = StableDiffusionXLWatermarker()
+        else:
+            self.watermark = None
 
     def encode_prompt(
         self,
-        prompt,
-        device,
-        num_images_per_prompt,
-        do_classifier_free_guidance,
-        negative_prompt=None,
+        prompt: str,
+        prompt_2: Optional[str] = None,
+        device: Optional[torch.device] = None,
+        num_images_per_prompt: int = 1,
+        do_classifier_free_guidance: bool = True,
+        negative_prompt: Optional[str] = None,
+        negative_prompt_2: Optional[str] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         lora_scale: Optional[float] = None,
         clip_skip: Optional[int] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
+
         Args:
             prompt (`str` or `List[str]`, *optional*):
                 prompt to be encoded
+            prompt_2 (`str` or `List[str]`, *optional*):
+                The prompt or prompts to be sent to the `tokenizer_2` and `text_encoder_2`. If not defined, `prompt` is
+                used in both text-encoders
             device: (`torch.device`):
                 torch device
             num_images_per_prompt (`int`):
@@ -546,6 +508,9 @@ class StableDiffusionPipeline(
                 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`).
+            negative_prompt_2 (`str` or `List[str]`, *optional*):
+                The prompt or prompts not to guide the image generation to be sent to `tokenizer_2` and
+                `text_encoder_2`. If not defined, `negative_prompt` is used in both text-encoders
             prompt_embeds (`torch.FloatTensor`, *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.
@@ -553,104 +518,118 @@ class StableDiffusionPipeline(
                 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.
+            pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
+                If not provided, pooled text embeddings will be generated from `prompt` input argument.
+            negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
+                weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_prompt`
+                input argument.
             lora_scale (`float`, *optional*):
-                A LoRA scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.
+                A lora scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.
             clip_skip (`int`, *optional*):
                 Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
                 the output of the pre-final layer will be used for computing the prompt embeddings.
         """
+        device = device or self._execution_device
+
         # set lora scale so that monkey patched LoRA
         # function of text encoder can correctly access it
-        if lora_scale is not None and isinstance(self, LoraLoaderMixin):
+        if lora_scale is not None and isinstance(self, StableDiffusionXLLoraLoaderMixin):
             self._lora_scale = lora_scale
 
             # dynamically adjust the LoRA scale
-            if not USE_PEFT_BACKEND:
-                adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)
-            else:
-                scale_lora_layers(self.text_encoder, lora_scale)
+            if self.text_encoder is not None:
+                if not USE_PEFT_BACKEND:
+                    adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)
+                else:
+                    scale_lora_layers(self.text_encoder, lora_scale)
 
-        if prompt is not None and isinstance(prompt, str):
-            batch_size = 1
-        elif prompt is not None and isinstance(prompt, list):
+            if self.text_encoder_2 is not None:
+                if not USE_PEFT_BACKEND:
+                    adjust_lora_scale_text_encoder(self.text_encoder_2, lora_scale)
+                else:
+                    scale_lora_layers(self.text_encoder_2, lora_scale)
+
+        prompt = [prompt] if isinstance(prompt, str) else prompt
+
+        if prompt is not None:
             batch_size = len(prompt)
         else:
             batch_size = prompt_embeds.shape[0]
 
+        # Define tokenizers and text encoders
+        tokenizers = [self.tokenizer, self.tokenizer_2] if self.tokenizer is not None else [self.tokenizer_2]
+        text_encoders = (
+            [self.text_encoder, self.text_encoder_2] if self.text_encoder is not None else [self.text_encoder_2]
+        )
+
         if prompt_embeds is None:
-            # textual inversion: process multi-vector tokens if necessary
-            if isinstance(self, TextualInversionLoaderMixin):
-                prompt = self.maybe_convert_prompt(prompt, self.tokenizer)
-
-            text_inputs = self.tokenizer(
-                prompt,
-                padding="max_length",
-                max_length=self.tokenizer.model_max_length,
-                truncation=True,
-                return_tensors="pt",
-            )
-            text_input_ids = text_inputs.input_ids
-            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
+            prompt_2 = prompt_2 or prompt
+            prompt_2 = [prompt_2] if isinstance(prompt_2, str) else prompt_2
 
-            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
-                text_input_ids, untruncated_ids
-            ):
-                removed_text = self.tokenizer.batch_decode(
-                    untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1]
-                )
-                logger.warning(
-                    "The following part of your input was truncated because CLIP can only handle sequences up to"
-                    f" {self.tokenizer.model_max_length} tokens: {removed_text}"
+            # textual inversion: process multi-vector tokens if necessary
+            prompt_embeds_list = []
+            prompts = [prompt, prompt_2]
+            for prompt, tokenizer, text_encoder in zip(prompts, tokenizers, text_encoders):
+                if isinstance(self, TextualInversionLoaderMixin):
+                    prompt = self.maybe_convert_prompt(prompt, tokenizer)
+
+                text_inputs = tokenizer(
+                    prompt,
+                    padding="max_length",
+                    max_length=tokenizer.model_max_length,
+                    truncation=True,
+                    return_tensors="pt",
                 )
 
-            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
-                attention_mask = text_inputs.attention_mask.to(device)
-            else:
-                attention_mask = None
+                text_input_ids = text_inputs.input_ids
+                untruncated_ids = tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
 
-            if clip_skip is None:
-                prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=attention_mask)
-                prompt_embeds = prompt_embeds[0]
-            else:
-                prompt_embeds = self.text_encoder(
-                    text_input_ids.to(device), attention_mask=attention_mask, output_hidden_states=True
-                )
-                # Access the `hidden_states` first, that contains a tuple of
-                # all the hidden states from the encoder layers. Then index into
-                # the tuple to access the hidden states from the desired layer.
-                prompt_embeds = prompt_embeds[-1][-(clip_skip + 1)]
-                # We also need to apply the final LayerNorm here to not mess with the
-                # representations. The `last_hidden_states` that we typically use for
-                # obtaining the final prompt representations passes through the LayerNorm
-                # layer.
-                prompt_embeds = self.text_encoder.text_model.final_layer_norm(prompt_embeds)
+                if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
+                    text_input_ids, untruncated_ids
+                ):
+                    removed_text = tokenizer.batch_decode(untruncated_ids[:, tokenizer.model_max_length - 1 : -1])
+                    logger.warning(
+                        "The following part of your input was truncated because CLIP can only handle sequences up to"
+                        f" {tokenizer.model_max_length} tokens: {removed_text}"
+                    )
 
-        if self.text_encoder is not None:
-            prompt_embeds_dtype = self.text_encoder.dtype
-        elif self.unet is not None:
-            prompt_embeds_dtype = self.unet.dtype
-        else:
-            prompt_embeds_dtype = prompt_embeds.dtype
+                prompt_embeds = text_encoder(text_input_ids.to(device), output_hidden_states=True)
 
-        prompt_embeds = prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)
+                # We are only ALWAYS interested in the pooled output of the final text encoder
+                pooled_prompt_embeds = prompt_embeds[0]
+                if clip_skip is None:
+                    prompt_embeds = prompt_embeds.hidden_states[-2]
+                else:
+                    # "2" because SDXL always indexes from the penultimate layer.
+                    prompt_embeds = prompt_embeds.hidden_states[-(clip_skip + 2)]
 
-        bs_embed, seq_len, _ = prompt_embeds.shape
-        # duplicate text embeddings for each generation per prompt, using mps friendly method
-        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
-        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
+                prompt_embeds_list.append(prompt_embeds)
+
+            prompt_embeds = torch.concat(prompt_embeds_list, dim=-1)
 
         # get unconditional embeddings for classifier free guidance
-        if do_classifier_free_guidance and negative_prompt_embeds is None:
+        zero_out_negative_prompt = negative_prompt is None and self.config.force_zeros_for_empty_prompt
+        if do_classifier_free_guidance and negative_prompt_embeds is None and zero_out_negative_prompt:
+            negative_prompt_embeds = torch.zeros_like(prompt_embeds)
+            negative_pooled_prompt_embeds = torch.zeros_like(pooled_prompt_embeds)
+        elif do_classifier_free_guidance and negative_prompt_embeds is None:
+            negative_prompt = negative_prompt or ""
+            negative_prompt_2 = negative_prompt_2 or negative_prompt
+
+            # normalize str to list
+            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt
+            negative_prompt_2 = (
+                batch_size * [negative_prompt_2] if isinstance(negative_prompt_2, str) else negative_prompt_2
+            )
+
             uncond_tokens: List[str]
-            if negative_prompt is None:
-                uncond_tokens = [""] * batch_size
-            elif prompt is not None and type(prompt) is not type(negative_prompt):
+            if prompt is not None and type(prompt) is not type(negative_prompt):
                 raise TypeError(
                     f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
                     f" {type(prompt)}."
                 )
-            elif isinstance(negative_prompt, str):
-                uncond_tokens = [negative_prompt]
             elif batch_size != len(negative_prompt):
                 raise ValueError(
                     f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
@@ -658,47 +637,77 @@ class StableDiffusionPipeline(
                     " the batch size of `prompt`."
                 )
             else:
-                uncond_tokens = negative_prompt
+                uncond_tokens = [negative_prompt, negative_prompt_2]
+
+            negative_prompt_embeds_list = []
+            for negative_prompt, tokenizer, text_encoder in zip(uncond_tokens, tokenizers, text_encoders):
+                if isinstance(self, TextualInversionLoaderMixin):
+                    negative_prompt = self.maybe_convert_prompt(negative_prompt, tokenizer)
+
+                max_length = prompt_embeds.shape[1]
+                uncond_input = tokenizer(
+                    negative_prompt,
+                    padding="max_length",
+                    max_length=max_length,
+                    truncation=True,
+                    return_tensors="pt",
+                )
 
-            # textual inversion: process multi-vector tokens if necessary
-            if isinstance(self, TextualInversionLoaderMixin):
-                uncond_tokens = self.maybe_convert_prompt(uncond_tokens, self.tokenizer)
-
-            max_length = prompt_embeds.shape[1]
-            uncond_input = self.tokenizer(
-                uncond_tokens,
-                padding="max_length",
-                max_length=max_length,
-                truncation=True,
-                return_tensors="pt",
-            )
+                negative_prompt_embeds = text_encoder(
+                    uncond_input.input_ids.to(device),
+                    output_hidden_states=True,
+                )
+                # We are only ALWAYS interested in the pooled output of the final text encoder
+                negative_pooled_prompt_embeds = negative_prompt_embeds[0]
+                negative_prompt_embeds = negative_prompt_embeds.hidden_states[-2]
 
-            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
-                attention_mask = uncond_input.attention_mask.to(device)
-            else:
-                attention_mask = None
+                negative_prompt_embeds_list.append(negative_prompt_embeds)
 
-            negative_prompt_embeds = self.text_encoder(
-                uncond_input.input_ids.to(device),
-                attention_mask=attention_mask,
-            )
-            negative_prompt_embeds = negative_prompt_embeds[0]
+            negative_prompt_embeds = torch.concat(negative_prompt_embeds_list, dim=-1)
+
+        if self.text_encoder_2 is not None:
+            prompt_embeds = prompt_embeds.to(dtype=self.text_encoder_2.dtype, device=device)
+        else:
+            prompt_embeds = prompt_embeds.to(dtype=self.unet.dtype, device=device)
+
+        bs_embed, seq_len, _ = prompt_embeds.shape
+        # duplicate text embeddings for each generation per prompt, using mps friendly method
+        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
 
         if do_classifier_free_guidance:
             # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
             seq_len = negative_prompt_embeds.shape[1]
 
-            negative_prompt_embeds = negative_prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)
+            if self.text_encoder_2 is not None:
+                negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.text_encoder_2.dtype, device=device)
+            else:
+                negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.unet.dtype, device=device)
 
             negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
             negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        if isinstance(self, LoraLoaderMixin) and USE_PEFT_BACKEND:
-            # Retrieve the original scale by scaling back the LoRA layers
-            unscale_lora_layers(self.text_encoder, lora_scale)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt).view(
+            bs_embed * num_images_per_prompt, -1
+        )
+        if do_classifier_free_guidance:
+            negative_pooled_prompt_embeds = negative_pooled_prompt_embeds.repeat(1, num_images_per_prompt).view(
+                bs_embed * num_images_per_prompt, -1
+            )
 
-        return prompt_embeds, negative_prompt_embeds
+        if self.text_encoder is not None:
+            if isinstance(self, StableDiffusionXLLoraLoaderMixin) and USE_PEFT_BACKEND:
+                # Retrieve the original scale by scaling back the LoRA layers
+                unscale_lora_layers(self.text_encoder, lora_scale)
 
+        if self.text_encoder_2 is not None:
+            if isinstance(self, StableDiffusionXLLoraLoaderMixin) and USE_PEFT_BACKEND:
+                # Retrieve the original scale by scaling back the LoRA layers
+                unscale_lora_layers(self.text_encoder_2, lora_scale)
+
+        return prompt_embeds, negative_prompt_embeds, pooled_prompt_embeds, negative_pooled_prompt_embeds
+
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
     def encode_image(self, image, device, num_images_per_prompt, output_hidden_states=None):
         dtype = next(self.image_encoder.parameters()).dtype
 
@@ -723,8 +732,9 @@ class StableDiffusionPipeline(
 
             return image_embeds, uncond_image_embeds
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_ip_adapter_image_embeds
     def prepare_ip_adapter_image_embeds(
-        self, ip_adapter_image, ip_adapter_image_embeds, device, num_images_per_prompt
+        self, ip_adapter_image, ip_adapter_image_embeds, device, num_images_per_prompt, do_classifier_free_guidance
     ):
         if ip_adapter_image_embeds is None:
             if not isinstance(ip_adapter_image, list):
@@ -748,40 +758,33 @@ class StableDiffusionPipeline(
                     [single_negative_image_embeds] * num_images_per_prompt, dim=0
                 )
 
-                if self.do_classifier_free_guidance:
+                if do_classifier_free_guidance:
                     single_image_embeds = torch.cat([single_negative_image_embeds, single_image_embeds])
                     single_image_embeds = single_image_embeds.to(device)
 
                 image_embeds.append(single_image_embeds)
         else:
-            image_embeds = ip_adapter_image_embeds
-        return image_embeds
-
-    def run_safety_checker(self, image, device, dtype):
-        if self.safety_checker is None:
-            has_nsfw_concept = None
-        else:
-            if torch.is_tensor(image):
-                feature_extractor_input = self.image_processor.postprocess(image, output_type="pil")
-            else:
-                feature_extractor_input = self.image_processor.numpy_to_pil(image)
-            safety_checker_input = self.feature_extractor(feature_extractor_input, return_tensors="pt").to(device)
-            image, has_nsfw_concept = self.safety_checker(
-                images=image, clip_input=safety_checker_input.pixel_values.to(dtype)
-            )
-        return image, has_nsfw_concept
-
-    def decode_latents(self, latents):
-        deprecation_message = "The decode_latents method is deprecated and will be removed in 1.0.0. Please use VaeImageProcessor.postprocess(...) instead"
-        deprecate("decode_latents", "1.0.0", deprecation_message, standard_warn=False)
+            repeat_dims = [1]
+            image_embeds = []
+            for single_image_embeds in ip_adapter_image_embeds:
+                if do_classifier_free_guidance:
+                    single_negative_image_embeds, single_image_embeds = single_image_embeds.chunk(2)
+                    single_image_embeds = single_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_image_embeds.shape[1:]))
+                    )
+                    single_negative_image_embeds = single_negative_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_negative_image_embeds.shape[1:]))
+                    )
+                    single_image_embeds = torch.cat([single_negative_image_embeds, single_image_embeds])
+                else:
+                    single_image_embeds = single_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_image_embeds.shape[1:]))
+                    )
+                image_embeds.append(single_image_embeds)
 
-        latents = 1 / self.vae.config.scaling_factor * latents
-        image = self.vae.decode(latents, return_dict=False)[0]
-        image = (image / 2 + 0.5).clamp(0, 1)
-        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
-        image = image.cpu().permute(0, 2, 3, 1).float().numpy()
-        return image
+        return image_embeds
 
+    # 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.
@@ -802,12 +805,16 @@ class StableDiffusionPipeline(
     def check_inputs(
         self,
         prompt,
+        prompt_2,
         height,
         width,
         callback_steps,
         negative_prompt=None,
+        negative_prompt_2=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
+        pooled_prompt_embeds=None,
+        negative_pooled_prompt_embeds=None,
         ip_adapter_image=None,
         ip_adapter_image_embeds=None,
         callback_on_step_end_tensor_inputs=None,
@@ -820,6 +827,7 @@ class StableDiffusionPipeline(
                 f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
                 f" {type(callback_steps)}."
             )
+
         if callback_on_step_end_tensor_inputs is not None and not all(
             k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
         ):
@@ -832,18 +840,30 @@ class StableDiffusionPipeline(
                 f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
                 " only forward one of the two."
             )
+        elif prompt_2 is not None and prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `prompt_2`: {prompt_2} 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)}")
+        elif prompt_2 is not None and (not isinstance(prompt_2, str) and not isinstance(prompt_2, list)):
+            raise ValueError(f"`prompt_2` has to be of type `str` or `list` but is {type(prompt_2)}")
 
         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."
             )
+        elif negative_prompt_2 is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `negative_prompt_2`: {negative_prompt_2} 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:
@@ -853,11 +873,32 @@ class StableDiffusionPipeline(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+        if prompt_embeds is not None and pooled_prompt_embeds is None:
+            raise ValueError(
+                "If `prompt_embeds` are provided, `pooled_prompt_embeds` also have to be passed. Make sure to generate `pooled_prompt_embeds` from the same text encoder that was used to generate `prompt_embeds`."
+            )
+
+        if negative_prompt_embeds is not None and negative_pooled_prompt_embeds is None:
+            raise ValueError(
+                "If `negative_prompt_embeds` are provided, `negative_pooled_prompt_embeds` also have to be passed. Make sure to generate `negative_pooled_prompt_embeds` from the same text encoder that was used to generate `negative_prompt_embeds`."
+            )
+
         if ip_adapter_image is not None and ip_adapter_image_embeds is not None:
             raise ValueError(
                 "Provide either `ip_adapter_image` or `ip_adapter_image_embeds`. Cannot leave both `ip_adapter_image` and `ip_adapter_image_embeds` defined."
             )
 
+        if ip_adapter_image_embeds is not None:
+            if not isinstance(ip_adapter_image_embeds, list):
+                raise ValueError(
+                    f"`ip_adapter_image_embeds` has to be of type `list` but is {type(ip_adapter_image_embeds)}"
+                )
+            elif ip_adapter_image_embeds[0].ndim not in [3, 4]:
+                raise ValueError(
+                    f"`ip_adapter_image_embeds` has to be a list of 3D or 4D tensors but is {ip_adapter_image_embeds[0].ndim}D"
+                )
+
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
     def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
         shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
         if isinstance(generator, list) and len(generator) != batch_size:
@@ -875,94 +916,61 @@ class StableDiffusionPipeline(
         latents = latents * self.scheduler.init_noise_sigma
         return latents
 
-    def enable_freeu(self, s1: float, s2: float, b1: float, b2: float):
-        r"""Enables the FreeU mechanism as in https://arxiv.org/abs/2309.11497.
-        The suffixes after the scaling factors represent the stages where they are being applied.
-        Please refer to the [official repository](https://github.com/ChenyangSi/FreeU) for combinations of the values
-        that are known to work well for different pipelines such as Stable Diffusion v1, v2, and Stable Diffusion XL.
-        Args:
-            s1 (`float`):
-                Scaling factor for stage 1 to attenuate the contributions of the skip features. This is done to
-                mitigate "oversmoothing effect" in the enhanced denoising process.
-            s2 (`float`):
-                Scaling factor for stage 2 to attenuate the contributions of the skip features. This is done to
-                mitigate "oversmoothing effect" in the enhanced denoising process.
-            b1 (`float`): Scaling factor for stage 1 to amplify the contributions of backbone features.
-            b2 (`float`): Scaling factor for stage 2 to amplify the contributions of backbone features.
-        """
-        if not hasattr(self, "unet"):
-            raise ValueError("The pipeline must have `unet` for using FreeU.")
-        self.unet.enable_freeu(s1=s1, s2=s2, b1=b1, b2=b2)
+    def _get_add_time_ids(
+        self, original_size, crops_coords_top_left, target_size, dtype, text_encoder_projection_dim=None
+    ):
+        add_time_ids = list(original_size + crops_coords_top_left + target_size)
 
-    def disable_freeu(self):
-        """Disables the FreeU mechanism if enabled."""
-        self.unet.disable_freeu()
+        passed_add_embed_dim = (
+            self.unet.config.addition_time_embed_dim * len(add_time_ids) + text_encoder_projection_dim
+        )
+        expected_add_embed_dim = self.unet.add_embedding.linear_1.in_features
 
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.fuse_qkv_projections
-    def fuse_qkv_projections(self, unet: bool = True, vae: bool = True):
-        """
-        Enables fused QKV projections. For self-attention modules, all projection matrices (i.e., query,
-        key, value) are fused. For cross-attention modules, key and value projection matrices are fused.
-        <Tip warning={true}>
-        This API is 🧪 experimental.
-        </Tip>
-        Args:
-            unet (`bool`, defaults to `True`): To apply fusion on the UNet.
-            vae (`bool`, defaults to `True`): To apply fusion on the VAE.
-        """
-        self.fusing_unet = False
-        self.fusing_vae = False
-
-        if unet:
-            self.fusing_unet = True
-            self.unet.fuse_qkv_projections()
-            self.unet.set_attn_processor(FusedAttnProcessor2_0())
-
-        if vae:
-            if not isinstance(self.vae, AutoencoderKL):
-                raise ValueError("`fuse_qkv_projections()` is only supported for the VAE of type `AutoencoderKL`.")
-
-            self.fusing_vae = True
-            self.vae.fuse_qkv_projections()
-            self.vae.set_attn_processor(FusedAttnProcessor2_0())
-
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.unfuse_qkv_projections
-    def unfuse_qkv_projections(self, unet: bool = True, vae: bool = True):
-        """Disable QKV projection fusion if enabled.
-        <Tip warning={true}>
-        This API is 🧪 experimental.
-        </Tip>
-        Args:
-            unet (`bool`, defaults to `True`): To apply fusion on the UNet.
-            vae (`bool`, defaults to `True`): To apply fusion on the VAE.
-        """
-        if unet:
-            if not self.fusing_unet:
-                logger.warning("The UNet was not initially fused for QKV projections. Doing nothing.")
-            else:
-                self.unet.unfuse_qkv_projections()
-                self.fusing_unet = False
+        if expected_add_embed_dim != passed_add_embed_dim:
+            raise ValueError(
+                f"Model expects an added time embedding vector of length {expected_add_embed_dim}, but a vector of {passed_add_embed_dim} was created. The model has an incorrect config. Please check `unet.config.time_embedding_type` and `text_encoder_2.config.projection_dim`."
+            )
 
-        if vae:
-            if not self.fusing_vae:
-                logger.warning("The VAE was not initially fused for QKV projections. Doing nothing.")
-            else:
-                self.vae.unfuse_qkv_projections()
-                self.fusing_vae = False
+        add_time_ids = torch.tensor([add_time_ids], dtype=dtype)
+        return add_time_ids
+
+    def upcast_vae(self):
+        dtype = self.vae.dtype
+        self.vae.to(dtype=torch.float32)
+        use_torch_2_0_or_xformers = isinstance(
+            self.vae.decoder.mid_block.attentions[0].processor,
+            (
+                AttnProcessor2_0,
+                XFormersAttnProcessor,
+                LoRAXFormersAttnProcessor,
+                LoRAAttnProcessor2_0,
+                FusedAttnProcessor2_0,
+            ),
+        )
+        # if xformers or torch_2_0 is used attention block does not need
+        # to be in float32 which can save lots of memory
+        if use_torch_2_0_or_xformers:
+            self.vae.post_quant_conv.to(dtype)
+            self.vae.decoder.conv_in.to(dtype)
+            self.vae.decoder.mid_block.to(dtype)
 
     # Copied from diffusers.pipelines.latent_consistency_models.pipeline_latent_consistency_text2img.LatentConsistencyModelPipeline.get_guidance_scale_embedding
-    def get_guidance_scale_embedding(self, w, embedding_dim=512, dtype=torch.float32):
+    def get_guidance_scale_embedding(
+        self, w: torch.Tensor, embedding_dim: int = 512, dtype: torch.dtype = torch.float32
+    ) -> torch.FloatTensor:
         """
         See https://github.com/google-research/vdm/blob/dc27b98a554f65cdc654b800da5aa1846545d41b/model_vdm.py#L298
+
         Args:
-            timesteps (`torch.Tensor`):
-                generate embedding vectors at these timesteps
+            w (`torch.Tensor`):
+                Generate embedding vectors with a specified guidance scale to subsequently enrich timestep embeddings.
             embedding_dim (`int`, *optional*, defaults to 512):
-                dimension of the embeddings to generate
-            dtype:
-                data type of the generated embeddings
+                Dimension of the embeddings to generate.
+            dtype (`torch.dtype`, *optional*, defaults to `torch.float32`):
+                Data type of the generated embeddings.
+
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.FloatTensor`: Embedding vectors with shape `(len(w), embedding_dim)`.
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -976,7 +984,7 @@ class StableDiffusionPipeline(
             emb = torch.nn.functional.pad(emb, (0, 1))
         assert emb.shape == (w.shape[0], embedding_dim)
         return emb
-    
+
     def pred_z0(self, sample, model_output, timestep):
         alpha_prod_t = self.scheduler.alphas_cumprod[timestep].to(sample.device)
 
@@ -998,19 +1006,18 @@ class StableDiffusionPipeline(
         return pred_original_sample
     
     def pred_x0(self, latents, noise_pred, t, generator, device, prompt_embeds, output_type):
-        
         pred_z0 = self.pred_z0(latents, noise_pred, t)
         pred_x0 = self.vae.decode(
             pred_z0 / self.vae.config.scaling_factor,
             return_dict=False,
             generator=generator
         )[0]
-        pred_x0, ____ = self.run_safety_checker(pred_x0, device, prompt_embeds.dtype)
+        #pred_x0, ____ = self.run_safety_checker(pred_x0, device, prompt_embeds.dtype)
         do_denormalize = [True] * pred_x0.shape[0]
         pred_x0 = self.image_processor.postprocess(pred_x0, output_type=output_type, do_denormalize=do_denormalize)
         
         return pred_x0
-
+        
     @property
     def guidance_scale(self):
         return self._guidance_scale
@@ -1034,6 +1041,10 @@ class StableDiffusionPipeline(
     def cross_attention_kwargs(self):
         return self._cross_attention_kwargs
 
+    @property
+    def denoising_end(self):
+        return self._denoising_end
+
     @property
     def num_timesteps(self):
         return self._num_timesteps
@@ -1041,7 +1052,7 @@ class StableDiffusionPipeline(
     @property
     def interrupt(self):
         return self._interrupt
-    
+
     @property
     def pag_scale(self):
         return self._pag_scale
@@ -1069,50 +1080,71 @@ class StableDiffusionPipeline(
     @property
     def pag_applied_layers_index(self):
         return self._pag_applied_layers_index
-
-
+    
     @torch.no_grad()
     @replace_example_docstring(EXAMPLE_DOC_STRING)
     def __call__(
         self,
         prompt: Union[str, List[str]] = None,
+        prompt_2: Optional[Union[str, List[str]]] = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
         num_inference_steps: int = 50,
         timesteps: List[int] = None,
-        guidance_scale: float = 7.5,
+        denoising_end: Optional[float] = None,
+        guidance_scale: float = 5.0,
         pag_scale: float = 0.0,
         pag_adaptive_scaling: float = 0.0,
         pag_drop_rate: float = 0.5,
-        pag_applied_layers: List[str] = ['down'], #['down', 'mid', 'up']
-        pag_applied_layers_index: List[str] = ['d4'], #['d4', 'd5', 'm0']
+        pag_applied_layers: List[str] = ['mid'], #['down', 'mid', 'up']
+        pag_applied_layers_index: List[str] = None, #['d4', 'd5', 'm0']
         negative_prompt: Optional[Union[str, List[str]]] = None,
+        negative_prompt_2: Optional[Union[str, List[str]]] = None,
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
         latents: Optional[torch.FloatTensor] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
         ip_adapter_image_embeds: Optional[List[torch.FloatTensor]] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         guidance_rescale: float = 0.0,
+        original_size: Optional[Tuple[int, int]] = None,
+        crops_coords_top_left: Tuple[int, int] = (0, 0),
+        target_size: Optional[Tuple[int, int]] = None,
+        negative_original_size: Optional[Tuple[int, int]] = None,
+        negative_crops_coords_top_left: Tuple[int, int] = (0, 0),
+        negative_target_size: Optional[Tuple[int, int]] = None,
         clip_skip: Optional[int] = None,
         callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
         callback_on_step_end_tensor_inputs: List[str] = ["latents"],
         **kwargs,
     ):
         r"""
-        The call function to the pipeline for generation.
+        Function invoked when calling the pipeline for generation.
+
         Args:
             prompt (`str` or `List[str]`, *optional*):
-                The prompt or prompts to guide image generation. If not defined, you need to pass `prompt_embeds`.
-            height (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
-                The height in pixels of the generated image.
-            width (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
-                The width in pixels of the generated image.
+                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            prompt_2 (`str` or `List[str]`, *optional*):
+                The prompt or prompts to be sent to the `tokenizer_2` and `text_encoder_2`. If not defined, `prompt` is
+                used in both text-encoders
+            height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The height in pixels of the generated image. This is set to 1024 by default for the best results.
+                Anything below 512 pixels won't work well for
+                [stabilityai/stable-diffusion-xl-base-1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0)
+                and checkpoints that are not specifically fine-tuned on low resolutions.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The width in pixels of the generated image. This is set to 1024 by default for the best results.
+                Anything below 512 pixels won't work well for
+                [stabilityai/stable-diffusion-xl-base-1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0)
+                and checkpoints that are not specifically fine-tuned on low resolutions.
             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.
@@ -1120,49 +1152,102 @@ class StableDiffusionPipeline(
                 Custom timesteps to use for the denoising process with schedulers which support a `timesteps` argument
                 in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
                 passed will be used. Must be in descending order.
-            guidance_scale (`float`, *optional*, defaults to 7.5):
-                A higher guidance scale value encourages the model to generate images closely linked to the text
-                `prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`.
+            denoising_end (`float`, *optional*):
+                When specified, determines the fraction (between 0.0 and 1.0) of the total denoising process to be
+                completed before it is intentionally prematurely terminated. As a result, the returned sample will
+                still retain a substantial amount of noise as determined by the discrete timesteps selected by the
+                scheduler. The denoising_end parameter should ideally be utilized when this pipeline forms a part of a
+                "Mixture of Denoisers" multi-pipeline setup, as elaborated in [**Refining the Image
+                Output**](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/stable_diffusion_xl#refining-the-image-output)
+            guidance_scale (`float`, *optional*, defaults to 5.0):
+                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 to guide what to not include in image generation. If not defined, you need to
-                pass `negative_prompt_embeds` instead. Ignored when not using guidance (`guidance_scale < 1`).
+                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`).
+            negative_prompt_2 (`str` or `List[str]`, *optional*):
+                The prompt or prompts not to guide the image generation to be sent to `tokenizer_2` and
+                `text_encoder_2`. If not defined, `negative_prompt` is used in both text-encoders
             num_images_per_prompt (`int`, *optional*, defaults to 1):
                 The number of images to generate per prompt.
             eta (`float`, *optional*, defaults to 0.0):
-                Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies
-                to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
+                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 (`torch.Generator` or `List[torch.Generator]`, *optional*):
-                A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
-                generation deterministic.
+                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
+                to make generation deterministic.
             latents (`torch.FloatTensor`, *optional*):
-                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
+                Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
-                tensor is generated by sampling using the supplied random `generator`.
+                tensor will ge generated by sampling using the supplied random `generator`.
             prompt_embeds (`torch.FloatTensor`, *optional*):
-                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
-                provided, text embeddings are generated from the `prompt` input argument.
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
             negative_prompt_embeds (`torch.FloatTensor`, *optional*):
-                Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
-                not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
+                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.
+            pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
+                If not provided, pooled text embeddings will be generated from `prompt` input argument.
+            negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
+                weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_prompt`
+                input argument.
             ip_adapter_image: (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.
             ip_adapter_image_embeds (`List[torch.FloatTensor]`, *optional*):
-                Pre-generated image embeddings for IP-Adapter. If not
+                Pre-generated image embeddings for IP-Adapter. It should be a list of length same as number of
+                IP-adapters. Each element should be a tensor of shape `(batch_size, num_images, emb_dim)`. It should
+                contain the negative image embedding if `do_classifier_free_guidance` is set to `True`. If not
                 provided, embeddings are computed from the `ip_adapter_image` input argument.
             output_type (`str`, *optional*, defaults to `"pil"`):
-                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
+                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.StableDiffusionPipelineOutput`] instead of a
-                plain tuple.
+                Whether or not to return a [`~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput`] instead
+                of a plain tuple.
             cross_attention_kwargs (`dict`, *optional*):
-                A kwargs dictionary that if specified is passed along to the [`AttentionProcessor`] as defined in
-                [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
+                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).
             guidance_rescale (`float`, *optional*, defaults to 0.0):
-                Guidance rescale factor from [Common Diffusion Noise Schedules and Sample Steps are
-                Flawed](https://arxiv.org/pdf/2305.08891.pdf). Guidance rescale factor should fix overexposure when
-                using zero terminal SNR.
-            clip_skip (`int`, *optional*):
-                Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
-                the output of the pre-final layer will be used for computing the prompt embeddings.
+                Guidance rescale factor proposed by [Common Diffusion Noise Schedules and Sample Steps are
+                Flawed](https://arxiv.org/pdf/2305.08891.pdf) `guidance_scale` is defined as `φ` in equation 16. of
+                [Common Diffusion Noise Schedules and Sample Steps are Flawed](https://arxiv.org/pdf/2305.08891.pdf).
+                Guidance rescale factor should fix overexposure when using zero terminal SNR.
+            original_size (`Tuple[int]`, *optional*, defaults to (1024, 1024)):
+                If `original_size` is not the same as `target_size` the image will appear to be down- or upsampled.
+                `original_size` defaults to `(height, width)` if not specified. Part of SDXL's micro-conditioning as
+                explained in section 2.2 of
+                [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952).
+            crops_coords_top_left (`Tuple[int]`, *optional*, defaults to (0, 0)):
+                `crops_coords_top_left` can be used to generate an image that appears to be "cropped" from the position
+                `crops_coords_top_left` downwards. Favorable, well-centered images are usually achieved by setting
+                `crops_coords_top_left` to (0, 0). Part of SDXL's micro-conditioning as explained in section 2.2 of
+                [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952).
+            target_size (`Tuple[int]`, *optional*, defaults to (1024, 1024)):
+                For most cases, `target_size` should be set to the desired height and width of the generated image. If
+                not specified it will default to `(height, width)`. Part of SDXL's micro-conditioning as explained in
+                section 2.2 of [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952).
+            negative_original_size (`Tuple[int]`, *optional*, defaults to (1024, 1024)):
+                To negatively condition the generation process based on a specific image resolution. Part of SDXL's
+                micro-conditioning as explained in section 2.2 of
+                [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952). For more
+                information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
+            negative_crops_coords_top_left (`Tuple[int]`, *optional*, defaults to (0, 0)):
+                To negatively condition the generation process based on a specific crop coordinates. Part of SDXL's
+                micro-conditioning as explained in section 2.2 of
+                [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952). For more
+                information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
+            negative_target_size (`Tuple[int]`, *optional*, defaults to (1024, 1024)):
+                To negatively condition the generation process based on a target image resolution. It should be as same
+                as the `target_size` for most cases. Part of SDXL's micro-conditioning as explained in section 2.2 of
+                [https://huggingface.co/papers/2307.01952](https://huggingface.co/papers/2307.01952). For more
+                information, refer to this issue thread: https://github.com/huggingface/diffusers/issues/4208.
             callback_on_step_end (`Callable`, *optional*):
                 A function that calls at the end of each denoising steps during the inference. The function is called
                 with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
@@ -1172,13 +1257,13 @@ class StableDiffusionPipeline(
                 The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                 will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                 `._callback_tensor_inputs` attribute of your pipeline class.
+
         Examples:
+
         Returns:
-            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`:
-                If `return_dict` is `True`, [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] is returned,
-                otherwise a `tuple` is returned where the first element is a list with the generated images and the
-                second element is a list of `bool`s indicating whether the corresponding generated image contains
-                "not-safe-for-work" (nsfw) content.
+            [`~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput`] or `tuple`:
+            [`~pipelines.stable_diffusion_xl.StableDiffusionXLPipelineOutput`] if `return_dict` is True, otherwise a
+            `tuple`. When returning a tuple, the first element is a list with the generated images.
         """
 
         callback = kwargs.pop("callback", None)
@@ -1188,29 +1273,35 @@ class StableDiffusionPipeline(
             deprecate(
                 "callback",
                 "1.0.0",
-                "Passing `callback` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
+                "Passing `callback` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
             )
         if callback_steps is not None:
             deprecate(
                 "callback_steps",
                 "1.0.0",
-                "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`",
+                "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
             )
 
         # 0. Default height and width to unet
-        height = height or self.unet.config.sample_size * self.vae_scale_factor
-        width = width or self.unet.config.sample_size * self.vae_scale_factor
-        # to deal with lora scaling and other possible forward hooks
+        height = height or self.default_sample_size * self.vae_scale_factor
+        width = width or self.default_sample_size * self.vae_scale_factor
+
+        original_size = original_size or (height, width)
+        target_size = target_size or (height, width)
 
         # 1. Check inputs. Raise error if not correct
         self.check_inputs(
             prompt,
+            prompt_2,
             height,
             width,
             callback_steps,
             negative_prompt,
+            negative_prompt_2,
             prompt_embeds,
             negative_prompt_embeds,
+            pooled_prompt_embeds,
+            negative_pooled_prompt_embeds,
             ip_adapter_image,
             ip_adapter_image_embeds,
             callback_on_step_end_tensor_inputs,
@@ -1220,14 +1311,15 @@ class StableDiffusionPipeline(
         self._guidance_rescale = guidance_rescale
         self._clip_skip = clip_skip
         self._cross_attention_kwargs = cross_attention_kwargs
+        self._denoising_end = denoising_end
         self._interrupt = False
-        
+
         self._pag_scale = pag_scale
         self._pag_adaptive_scaling = pag_adaptive_scaling
         self._pag_drop_rate = pag_drop_rate
         self._pag_applied_layers = pag_applied_layers
         self._pag_applied_layers_index = pag_applied_layers_index
-
+        
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
@@ -1243,37 +1335,27 @@ class StableDiffusionPipeline(
             self.cross_attention_kwargs.get("scale", None) if self.cross_attention_kwargs is not None else None
         )
 
-        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
-            prompt,
-            device,
-            num_images_per_prompt,
-            self.do_classifier_free_guidance,
-            negative_prompt,
+        (
+            prompt_embeds,
+            negative_prompt_embeds,
+            pooled_prompt_embeds,
+            negative_pooled_prompt_embeds,
+        ) = self.encode_prompt(
+            prompt=prompt,
+            prompt_2=prompt_2,
+            device=device,
+            num_images_per_prompt=num_images_per_prompt,
+            do_classifier_free_guidance=self.do_classifier_free_guidance,
+            negative_prompt=negative_prompt,
+            negative_prompt_2=negative_prompt_2,
             prompt_embeds=prompt_embeds,
             negative_prompt_embeds=negative_prompt_embeds,
+            pooled_prompt_embeds=pooled_prompt_embeds,
+            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
             lora_scale=lora_scale,
             clip_skip=self.clip_skip,
         )
 
-        # 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
-        
-        #cfg
-        if self.do_classifier_free_guidance and not self.do_adversarial_guidance:
-            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])
-        #pag
-        elif not self.do_classifier_free_guidance and self.do_adversarial_guidance:
-            prompt_embeds = torch.cat([prompt_embeds, prompt_embeds])
-        #both
-        elif self.do_classifier_free_guidance and self.do_adversarial_guidance:
-            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds, prompt_embeds])
-
-        if ip_adapter_image is not None or ip_adapter_image_embeds is not None:
-            image_embeds = self.prepare_ip_adapter_image_embeds(
-                ip_adapter_image, ip_adapter_image_embeds, device, batch_size * num_images_per_prompt
-            )
-
         # 4. Prepare timesteps
         timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
 
@@ -1293,14 +1375,80 @@ class StableDiffusionPipeline(
         # 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)
 
-        # 6.1 Add image embeds for IP-Adapter
-        added_cond_kwargs = (
-            {"image_embeds": image_embeds}
-            if (ip_adapter_image is not None or ip_adapter_image_embeds is not None)
-            else None
+        # 7. Prepare added time ids & embeddings
+        add_text_embeds = pooled_prompt_embeds
+        if self.text_encoder_2 is None:
+            text_encoder_projection_dim = int(pooled_prompt_embeds.shape[-1])
+        else:
+            text_encoder_projection_dim = self.text_encoder_2.config.projection_dim
+
+        add_time_ids = self._get_add_time_ids(
+            original_size,
+            crops_coords_top_left,
+            target_size,
+            dtype=prompt_embeds.dtype,
+            text_encoder_projection_dim=text_encoder_projection_dim,
         )
+        if negative_original_size is not None and negative_target_size is not None:
+            negative_add_time_ids = self._get_add_time_ids(
+                negative_original_size,
+                negative_crops_coords_top_left,
+                negative_target_size,
+                dtype=prompt_embeds.dtype,
+                text_encoder_projection_dim=text_encoder_projection_dim,
+            )
+        else:
+            negative_add_time_ids = add_time_ids
+
+        #cfg
+        if self.do_classifier_free_guidance and not self.do_adversarial_guidance:
+            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
+            add_text_embeds = torch.cat([negative_pooled_prompt_embeds, add_text_embeds], dim=0)
+            add_time_ids = torch.cat([negative_add_time_ids, add_time_ids], dim=0)
+        #pag
+        elif not self.do_classifier_free_guidance and self.do_adversarial_guidance:
+            prompt_embeds = torch.cat([prompt_embeds, prompt_embeds], dim=0)
+            add_text_embeds = torch.cat([add_text_embeds, add_text_embeds], dim=0)
+            add_time_ids = torch.cat([add_time_ids, add_time_ids], dim=0)
+        #both
+        elif self.do_classifier_free_guidance and self.do_adversarial_guidance:
+            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds, prompt_embeds], dim=0)
+            add_text_embeds = torch.cat([negative_pooled_prompt_embeds, add_text_embeds, add_text_embeds], dim=0)
+            add_time_ids = torch.cat([negative_add_time_ids, add_time_ids, add_time_ids], dim=0)
+            
+        prompt_embeds = prompt_embeds.to(device)
+        add_text_embeds = add_text_embeds.to(device)
+        add_time_ids = add_time_ids.to(device).repeat(batch_size * num_images_per_prompt, 1)
+
+        if ip_adapter_image is not None or ip_adapter_image_embeds is not None:
+            image_embeds = self.prepare_ip_adapter_image_embeds(
+                ip_adapter_image,
+                ip_adapter_image_embeds,
+                device,
+                batch_size * num_images_per_prompt,
+                self.do_classifier_free_guidance,
+            )
+
+        # 8. Denoising loop
+        num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
+
+        # 8.1 Apply denoising_end
+        if (
+            self.denoising_end is not None
+            and isinstance(self.denoising_end, float)
+            and self.denoising_end > 0
+            and self.denoising_end < 1
+        ):
+            discrete_timestep_cutoff = int(
+                round(
+                    self.scheduler.config.num_train_timesteps
+                    - (self.denoising_end * self.scheduler.config.num_train_timesteps)
+                )
+            )
+            num_inference_steps = len(list(filter(lambda ts: ts >= discrete_timestep_cutoff, timesteps)))
+            timesteps = timesteps[:num_inference_steps]
 
-        # 6.2 Optionally get Guidance Scale Embedding
+        # 9. Optionally get Guidance Scale Embedding
         timestep_cond = None
         if self.unet.config.time_cond_proj_dim is not None:
             guidance_scale_tensor = torch.tensor(self.guidance_scale - 1).repeat(batch_size * num_images_per_prompt)
@@ -1308,7 +1456,7 @@ class StableDiffusionPipeline(
                 guidance_scale_tensor, embedding_dim=self.unet.config.time_cond_proj_dim
             ).to(device=device, dtype=latents.dtype)
 
-        # 7. Denoising loop
+        # 10. Create down mid and up layer lists
         if self.do_adversarial_guidance:
             down_layers = []
             mid_layers = []
@@ -1324,14 +1472,13 @@ class StableDiffusionPipeline(
                         up_layers.append(module)
                     else:
                         raise ValueError(f"Invalid layer type: {layer_type}")
-        
-        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
+            
         self._num_timesteps = len(timesteps)
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 if self.interrupt:
                     continue
-                
+
                 #cfg
                 if self.do_classifier_free_guidance and not self.do_adversarial_guidance:
                     latent_model_input = torch.cat([latents] * 2)
@@ -1344,7 +1491,7 @@ class StableDiffusionPipeline(
                 #no
                 else:
                     latent_model_input = latents
-                
+                    
                 # change attention layer in UNet if use PAG
                 if self.do_adversarial_guidance:
                     
@@ -1352,26 +1499,51 @@ class StableDiffusionPipeline(
                         replace_processor = PAGCFGIdentitySelfAttnProcessor()
                     else:
                         replace_processor = PAGIdentitySelfAttnProcessor()
-                    
-                    drop_layers = self.pag_applied_layers_index
-                    for drop_layer in drop_layers:
-                        try:
-                            if drop_layer[0] == 'd':
-                                down_layers[int(drop_layer[1])].processor = replace_processor
-                            elif drop_layer[0] == 'm':
-                                mid_layers[int(drop_layer[1])].processor = replace_processor
-                            elif drop_layer[0] == 'u':
-                                up_layers[int(drop_layer[1])].processor = replace_processor
-                            else:
-                                raise ValueError(f"Invalid layer type: {drop_layer[0]}")
-                        except IndexError:
-                            raise ValueError(
-                                f"Invalid layer index: {drop_layer}. Available layers: {len(down_layers)} down layers, {len(mid_layers)} mid layers, {len(up_layers)} up layers."
-                            )
-                
+
+                    if(self.pag_applied_layers_index):
+                        drop_layers = self.pag_applied_layers_index
+                        for drop_layer in drop_layers:
+                            layer_number = int(drop_layer[1:])
+                            try:
+                                if drop_layer[0] == 'd':
+                                    down_layers[layer_number].processor = replace_processor
+                                elif drop_layer[0] == 'm':
+                                    mid_layers[layer_number].processor = replace_processor
+                                elif drop_layer[0] == 'u':
+                                    up_layers[layer_number].processor = replace_processor
+                                else:
+                                    raise ValueError(f"Invalid layer type: {drop_layer[0]}")
+                            except IndexError:
+                                raise ValueError(
+                                    f"Invalid layer index: {drop_layer}. Available layers: {len(down_layers)} down layers, {len(mid_layers)} mid layers, {len(up_layers)} up layers."
+                                )
+                    elif(self.pag_applied_layers):
+                        drop_full_layers = self.pag_applied_layers
+                        for drop_full_layer in drop_full_layers:
+                            try:
+                                if drop_full_layer == "down":
+                                    for down_layer in down_layers:
+                                        down_layer.processor = replace_processor
+                                elif drop_full_layer == "mid":
+                                    for mid_layer in mid_layers:
+                                        mid_layer.processor = replace_processor
+                                elif drop_full_layer == "up":
+                                    for up_layer in up_layers:
+                                        up_layer.processor = replace_processor
+                                else:
+                                    raise ValueError(f"Invalid layer type: {drop_full_layer}")
+                            except IndexError:
+                                raise ValueError(
+                                    f"Invalid layer index: {drop_full_layer}. Available layers are: down, mid and up. If you need to specify each layer index, you can use `pag_applied_layers_index`"
+                                )
+                                
                 latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
-                
+
                 # predict the noise residual
+                added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
+                if ip_adapter_image is not None or ip_adapter_image_embeds is not None:
+                    added_cond_kwargs["image_embeds"] = image_embeds
+
                 noise_pred = self.unet(
                     latent_model_input,
                     t,
@@ -1381,20 +1553,13 @@ class StableDiffusionPipeline(
                     added_cond_kwargs=added_cond_kwargs,
                     return_dict=False,
                 )[0]
-                
+
                 # perform guidance
-                
-                # cfg
                 if self.do_classifier_free_guidance and not self.do_adversarial_guidance:
-                    
                     noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
-                    
-                    delta = noise_pred_text - noise_pred_uncond
-                    noise_pred = noise_pred_uncond + self.guidance_scale * delta
-                    
+                    noise_pred = noise_pred_uncond + self.guidance_scale * (noise_pred_text - noise_pred_uncond)
                 # pag
                 elif not self.do_classifier_free_guidance and self.do_adversarial_guidance:
-                    
                     noise_pred_original, noise_pred_perturb = noise_pred.chunk(2)
                     
                     signal_scale = self.pag_scale
@@ -1423,7 +1588,12 @@ class StableDiffusionPipeline(
                     noise_pred = rescale_noise_cfg(noise_pred, noise_pred_text, guidance_rescale=self.guidance_rescale)
 
                 # compute the previous noisy sample x_t -> x_t-1
+                latents_dtype = latents.dtype
                 latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
+                if latents.dtype != latents_dtype:
+                    if torch.backends.mps.is_available():
+                        # some platforms (eg. apple mps) misbehave due to a pytorch bug: https://github.com/pytorch/pytorch/pull/99272
+                        latents = latents.to(latents_dtype)
 
                 if callback_on_step_end is not None:
                     callback_kwargs = {}
@@ -1434,6 +1604,12 @@ class StableDiffusionPipeline(
                     latents = callback_outputs.pop("latents", latents)
                     prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
                     negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
+                    add_text_embeds = callback_outputs.pop("add_text_embeds", add_text_embeds)
+                    negative_pooled_prompt_embeds = callback_outputs.pop(
+                        "negative_pooled_prompt_embeds", negative_pooled_prompt_embeds
+                    )
+                    add_time_ids = callback_outputs.pop("add_time_ids", add_time_ids)
+                    negative_add_time_ids = callback_outputs.pop("negative_add_time_ids", negative_add_time_ids)
 
                 # call the callback, if provided
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
@@ -1442,44 +1618,93 @@ class StableDiffusionPipeline(
                         step_idx = i // getattr(self.scheduler, "order", 1)
                         callback(step_idx, t, latents)
 
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
         if not output_type == "latent":
-            image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False, generator=generator)[
-                0
-            ]
-            image, has_nsfw_concept = self.run_safety_checker(image, device, prompt_embeds.dtype)
+            # make sure the VAE is in float32 mode, as it overflows in float16
+            needs_upcasting = self.vae.dtype == torch.float16 and self.vae.config.force_upcast
+
+            if needs_upcasting:
+                self.upcast_vae()
+                latents = latents.to(next(iter(self.vae.post_quant_conv.parameters())).dtype)
+            elif latents.dtype != self.vae.dtype:
+                if torch.backends.mps.is_available():
+                    # some platforms (eg. apple mps) misbehave due to a pytorch bug: https://github.com/pytorch/pytorch/pull/99272
+                    self.vae = self.vae.to(latents.dtype)
+
+            # unscale/denormalize the latents
+            # denormalize with the mean and std if available and not None
+            has_latents_mean = hasattr(self.vae.config, "latents_mean") and self.vae.config.latents_mean is not None
+            has_latents_std = hasattr(self.vae.config, "latents_std") and self.vae.config.latents_std is not None
+            if has_latents_mean and has_latents_std:
+                latents_mean = (
+                    torch.tensor(self.vae.config.latents_mean).view(1, 4, 1, 1).to(latents.device, latents.dtype)
+                )
+                latents_std = (
+                    torch.tensor(self.vae.config.latents_std).view(1, 4, 1, 1).to(latents.device, latents.dtype)
+                )
+                latents = latents * latents_std / self.vae.config.scaling_factor + latents_mean
+            else:
+                latents = latents / self.vae.config.scaling_factor
+
+            image = self.vae.decode(latents, return_dict=False)[0]
+
+            # cast back to fp16 if needed
+            if needs_upcasting:
+                self.vae.to(dtype=torch.float16)
         else:
             image = latents
-            has_nsfw_concept = None
 
-        if has_nsfw_concept is None:
-            do_denormalize = [True] * image.shape[0]
-        else:
-            do_denormalize = [not has_nsfw for has_nsfw in has_nsfw_concept]
+        if not output_type == "latent":
+            # apply watermark if available
+            if self.watermark is not None:
+                image = self.watermark.apply_watermark(image)
 
-        image = self.image_processor.postprocess(image, output_type=output_type, do_denormalize=do_denormalize)
+            image = self.image_processor.postprocess(image, output_type=output_type)
 
         # Offload all models
         self.maybe_free_model_hooks()
 
         if not return_dict:
-            return (image, has_nsfw_concept)
-            
-        # change attention layer in UNet if use PAG
-        if self.do_adversarial_guidance:
-            drop_layers = self.pag_applied_layers_index
-            for drop_layer in drop_layers:
-                try:
-                    if drop_layer[0] == 'd':
-                        down_layers[int(drop_layer[1])].processor = AttnProcessor2_0()
-                    elif drop_layer[0] == 'm':
-                        mid_layers[int(drop_layer[1])].processor = AttnProcessor2_0()
-                    elif drop_layer[0] == 'u':
-                        up_layers[int(drop_layer[1])].processor = AttnProcessor2_0()
-                    else:
-                        raise ValueError(f"Invalid layer type: {drop_layer[0]}")
-                except IndexError:
-                    raise ValueError(
-                        f"Invalid layer index: {drop_layer}. Available layers: {len(down_layers)} down layers, {len(mid_layers)} mid layers, {len(up_layers)} up layers."
-                    )
+            return (image,)
 
-        return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept)
+        #Change the attention layers back to original ones after PAG was applied
+        if self.do_adversarial_guidance:
+            if(self.pag_applied_layers_index):
+                drop_layers = self.pag_applied_layers_index
+                for drop_layer in drop_layers:
+                    layer_number = int(drop_layer[1:])
+                    try:
+                        if drop_layer[0] == 'd':
+                            down_layers[layer_number].processor = AttnProcessor2_0()
+                        elif drop_layer[0] == 'm':
+                            mid_layers[layer_number].processor = AttnProcessor2_0()
+                        elif drop_layer[0] == 'u':
+                            up_layers[layer_number].processor = AttnProcessor2_0()
+                        else:
+                            raise ValueError(f"Invalid layer type: {drop_layer[0]}")
+                    except IndexError:
+                        raise ValueError(
+                            f"Invalid layer index: {drop_layer}. Available layers: {len(down_layers)} down layers, {len(mid_layers)} mid layers, {len(up_layers)} up layers."
+                        )
+            elif(self.pag_applied_layers):
+                            drop_full_layers = self.pag_applied_layers
+                            for drop_full_layer in drop_full_layers:
+                                try:
+                                    if drop_full_layer == "down":
+                                        for down_layer in down_layers:
+                                            down_layer.processor = AttnProcessor2_0()
+                                    elif drop_full_layer == "mid":
+                                        for mid_layer in mid_layers:
+                                            mid_layer.processor = AttnProcessor2_0()
+                                    elif drop_full_layer == "up":
+                                        for up_layer in up_layers:
+                                            up_layer.processor = AttnProcessor2_0()
+                                    else:
+                                        raise ValueError(f"Invalid layer type: {drop_full_layer}")
+                                except IndexError:
+                                    raise ValueError(
+                                        f"Invalid layer index: {drop_full_layer}. Available layers are: down, mid and up. If you need to specify each layer index, you can use `pag_applied_layers_index`"
+                                    )
+        return StableDiffusionXLPipelineOutput(images=image)