get activation layer Args: act_type (str): the activation type Returns: torch.nn.functional: the activation layer

Decode a batch of images/videos. Args: z (`torch.FloatTensor`): Input batch of latent vectors. return_d

Initialize the Inference pipeline. Args: pretrained_model_path (str or pathlib.Path): The model path, including t2v, tex

AI is creating summary for apply_gate Args: x (torch.Tensor): input tensor. gate (torch.Tensor, optional): gate tensor. Defaults

modulate by shift and scale Args: x (torch.Tensor): input tensor. shift (torch.Tensor, optional): shift tensor. Defaults to None.

Perform QKV self attention. Args: q (torch.Tensor): Query tensor with shape [b, s, a, d], where a is the number of heads. k

Args: batch_encoding (dict): Batch encoding from tokenizer. use_attention_mask (bool): Whether to use attention mask.

Get the normalization layer. Args: norm_layer (str): The type of normalization layer. Returns: norm_layer (nn.Module):

Sets the discrete timesteps used for the diffusion chain (to be run before inference). Args: num_inference_steps (`int`)

Tokenize the input text. Args: text (str or list): Input text.

align hight, width according to alignment Args: value (int): height or width alignment (int): target alignment factor Return

Apply rotary embeddings to input tensors using the given frequency tensor. This function applies rotary embeddings to the given query 'xq' a

Apply text to template. Args: text (str): Input text. template (str or list): Template string or list of cha

r""" Enable tiled VAE decoding. When this option is enabled, the VAE will split the input tensor into tiles to compute decoding and

r""" Encodes the prompt into text encoder hidden states. Args: prompt (`str` or `List[str]`, *optional*):

Predict the image/video from the given text. Args: prompt (str or List[str]): The input text. kwargs:

Reshape frequency tensor for broadcasting it with another tensor. This function reshapes the frequency tensor to have the same shape as the

save videos by video tensor copy from https://github.com/guoyww/AnimateDiff/blob/e92bd5671ba62c0d774a32951453e328018b7c5b/animatediff/utils/uti

r""" Sets the attention processor to use to compute attention. Parameters: processor (`dict` of `AttentionProcessor`

r""" Decode a batch of images/videos using a tiled decoder. Args: z (`torch.FloatTensor`): Input batch of latent vect

r"""Encode a batch of images/videos using a tiled encoder. When this option is enabled, the VAE will split the input tensor into tiles to c

Apply the RMSNorm normalization to the input tensor. Args: x (torch.Tensor): The input tensor. Returns:

Collect the information of the running environments. Returns: dict: The environment information. The following fields are contained.

Encode a batch of images/videos into latents. Args: x (`torch.FloatTensor`): Input batch of images/videos.

Precompute the frequency tensor for complex exponential (cis) with given dimensions. (Note: `cis` means `cos + i * sin`, where i is the imagi

Obtain the build information of PyTorch or Parrots.

Calculate cu_seqlens_q, cu_seqlens_kv using text_mask and img_len Args: text_mask (torch.Tensor): the mask of text img_len (int):

Get n-D meshgrid with start, stop and num. Args: start (int or tuple): If len(args) == 0, start is num; If len(args) == 1, start is

This is a n-d version of precompute_freqs_cis, which is a RoPE for tokens with n-d structure. Args: rope_dim_list (list of int): Dim

Returns True if cuda devices exist.

Check whether the PyTorch is compiled on ROCm.

Load the denoising scheduler for inference.

load hunyuan video model Args: args (dict): model args in_channels (int): input channels number out_channels (int): outpu

the fucntion to load the 3D VAE model Args: vae_type (str): the type of the 3D VAE model. Defaults to "884-16c-hy". vae_precision

Default is E4M3.

Rescale `noise_cfg` according to `guidance_rescale`. Based on findings of [Common Diffusion Noise Schedules and Sample Steps are Flawed](http

Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles custom timesteps. Any kwargs

Predict the sample from the previous timestep by reversing the SDE. This function propagates the diffusion process from the learned m