Vision Transformer with support for patch or hybrid CNN input stage

A Transformer model for video-like data. Parameters: num_attention_heads (`int`, *optional*, defaults to 16): The number of head

A Transformer model for video-like data. Parameters: num_attention_heads (`int`, *optional*, defaults to 16): The number of head

r""" Pipeline to generate video from an input image using Stable Video Diffusion. This model inherits from [`DiffusionPipeline`]. Check th

https://arxiv.org/abs/2203.15565 A distributed sparsely updating variant of the FC layer, named Partial FC (PFC). When sample rate less t

The output of [`TransformerTemporalModel`]. Args: sample (`torch.Tensor` of shape `(batch_size x num_frames, num_channels, heigh

The output of [`TransformerTemporalModel`]. Args: sample (`torch.FloatTensor` of shape `(batch_size x num_frames, num_channels, heig

r""" Pipeline to generate video from an input image using Stable Video Diffusion. This model inherits from [`DiffusionPipeline`]. Check th

r""" Output class for mimicmotion pipeline. Args: frames (`[List[List[PIL.Image.Image]]`, `np.ndarray`, `torch.Tensor`]):

Computes and stores the average and current value

Output class for the scheduler's `step` function output. Args: prev_sample (`torch.Tensor` of shape `(batch_size, num_channels,

r""" Output class for mimicmotion pipeline. Args: frames (`[List[List[PIL.Image.Image]]`, `np.ndarray`, `torch.Tensor`]):

r""" Pipeline to generate video from an input image using Stable Video Diffusion. This model inherits from [`DiffusionPipeline`]. Check th

r""" Output class for mimicmotion pipeline. Args: frames (`[List[List[PIL.Image.Image]]`, `np.ndarray`, `torch.Tensor`]):

The output of [`UNetSpatioTemporalConditionModel`]. Args: sample (`torch.FloatTensor` of shape `(batch_size, num_frames, num_channel

r""" Output class for mimicmotion pipeline. Args: frames (`[List[List[PIL.Image.Image]]`, `np.ndarray`, `torch.Tensor`]):

AllGather op with gradient backward

r""" Pipeline to generate video from an input image using Stable Video Diffusion. This model inherits from [`DiffusionPipeline`]. Check the s

ArcFace (https://arxiv.org/pdf/1801.07698v1.pdf):

CrossEntropy loss is calculated in parallel, allreduce denominator into single gpu and calculate softmax. Implemented of ArcFace (https://arx

Euler scheduler. This model inherits from [`SchedulerMixin`] and [`ConfigMixin`]. Check the superclass documentation for the generic

r""" A VAE model with KL loss for encoding images into latents and decoding latent representations into images. This model inherits from [

A Transformer model for video-like data. Parameters: num_attention_heads (`int`, *optional*, defaults to 16): The number of heads to

A Transformer model for video-like data. Parameters: num_attention_heads (`int`, *optional*, defaults to 16): The number of heads to

r""" A conditional Spatio-Temporal UNet model that takes a noisy video frames, conditional state, and a timestep and returns a sample shaped o