r""" Save the LoRA parameters corresponding to the UNet and text encoder. Arguments: save_directory (`str` or `os.PathLik

Unloads the LoRA parameters. Examples: ```python >>> # Assuming `pipeline` is already loaded with the LoRA paramete

r""" Download and cache a PyTorch diffusion pipeline from pretrained pipeline weights. Parameters: pretrained_model_name

r""" Enable memory efficient attention from [xFormers](https://facebookresearch.github.io/xformers/). When this option is enabled, yo

r""" Load Textual Inversion embeddings into the text encoder of [`StableDiffusionPipeline`] (both 🤗 Diffusers and Automatic1111 format

r""" Prepare the attention mask for the attention computation. Args: attention_mask (`torch.Tensor`): The

Convert a PIL image or a list of PIL images to NumPy arrays.

r""" Normalize the encoder hidden states. Requires `self.norm_cross` to be specified when constructing the `Attention` class.

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

r""" Instantiate a [`DiffusionPipeline`] from pretrained pipeline weights saved in the `.ckpt` or `.safetensors` format. The pipeline

Predict the sample from the previous timestep by reversing the SDE. This function propagates the sample with the linear multistep met

Predict the sample from the previous timestep by reversing the SDE. This function propagates the sample with the Runge-Kutta method.

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

r""" Reshape the tensor from `[batch_size, seq_len, dim]` to `[batch_size // heads, seq_len, dim * heads]`. `heads` is the number of h

Args: Compares the current Transformers version to a given reference with an operation. operation (`str`): A string repre

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

r""" Reverses the effect of [`pipe.fuse_lora()`](https://huggingface.co/docs/diffusers/main/en/api/loaders#diffusers.loaders.LoraLoade

This matches the implementation in Denoising Diffusion Probabilistic Models: Create sinusoidal timestep embeddings. :param timesteps: a 1-D

Checking for safetensors compatibility: - By default, all models are saved with the default pytorch serialization, so we use the list of defa

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take

r""" Compute the attention scores. Args: query (`torch.Tensor`): The query tensor. key (`torch.Tensor`): The

Check whether the module was compiled with torch.compile()

r""" Maybe convert a prompt into a "multi vector"-compatible prompt. If the prompt includes a token that corresponds to a multi-vector

r""" Get the attention processor in use. Args: return_deprecated_lora (`bool`, *optional*, defaults to `False`):

Updates paths inside resnets to the new naming scheme (local renaming)

Updates paths inside resnets to the new naming scheme (local renaming)

Updates paths inside resnets to the new naming scheme (local renaming)

r""" Returns: `torch.device`: The torch device on which the pipeline is located.

Convert a numpy image or a batch of images to a PIL image.

r""" Enable sliced attention computation. When this option is enabled, the attention module splits the input tensor in slices to comp