diffusers/pr_13803/en/api/models/autoencoder_kl_kvae_video.md

AutoencoderKLKVAEVideo

The 3D variational autoencoder (VAE) model with KL loss.

The model can be loaded with the following code snippet.

import torch
from diffusers import AutoencoderKLKVAEVideo

vae = AutoencoderKLKVAEVideo.from_pretrained("kandinskylab/KVAE-3D-1.0", subfolder="diffusers", torch_dtype=torch.float16)

AutoencoderKLKVAEVideo[[diffusers.AutoencoderKLKVAEVideo]]

diffusers.AutoencoderKLKVAEVideo[[diffusers.AutoencoderKLKVAEVideo]]

Source

A VAE model with KL loss for encoding videos into latents and decoding latent representations into videos. Used in KVAE.

This model inherits from ModelMixin. Check the superclass documentation for its generic methods implemented for all models (such as downloading or saving).

decodediffusers.AutoencoderKLKVAEVideo.decodehttps://github.com/huggingface/diffusers/blob/vr_13803/src/diffusers/models/autoencoders/autoencoder_kl_kvae_video.py#L915[{"name": "z", "val": ": Tensor"}, {"name": "return_dict", "val": ": bool = True"}]- z (torch.Tensor) -- Input batch of latent vectors with shape (B, C, T, H, W).

• return_dict (bool, optional, defaults to True) -- Whether to return a ~models.vae.DecoderOutput instead of a plain tuple.0~models.vae.DecoderOutput or tupleDecoded video.

Decode a batch of videos.

Parameters:

ch (int, optional, defaults to 128) : Base channel count.

ch_mult (Tuple[int], optional, defaults to (1, 2, 4, 8)) : Channel multipliers per level.

num_res_blocks (int, optional, defaults to 2) : Number of residual blocks per level.

in_channels (int, optional, defaults to 3) : Number of input channels.

out_ch (int, optional, defaults to 3) : Number of output channels.

z_channels (int, optional, defaults to 16) : Number of latent channels.

temporal_compress_times (int, optional, defaults to 4) : Temporal compression factor.

Returns:

~models.vae.DecoderOutput` or `tuple

Decoded video.

disable_slicing[[diffusers.AutoencoderKLKVAEVideo.disable_slicing]]

Source

Disable sliced VAE decoding.

enable_slicing[[diffusers.AutoencoderKLKVAEVideo.enable_slicing]]

Source

Enable sliced VAE decoding.

encode[[diffusers.AutoencoderKLKVAEVideo.encode]]

Source

Encode a batch of videos into latents.

Parameters:

x (torch.Tensor) : Input batch of videos with shape (B, C, T, H, W).

return_dict (bool, optional, defaults to True) : Whether to return a ~models.autoencoder_kl.AutoencoderKLOutput instead of a plain tuple.

Returns:

The latent representations of the encoded videos.

forward[[diffusers.AutoencoderKLKVAEVideo.forward]]

Source

Parameters:

sample (torch.Tensor) : Input sample.

sample_posterior (bool, optional, defaults to False) : Whether to sample from the posterior.

return_dict (bool, optional, defaults to True) : Whether or not to return a DecoderOutput instead of a plain tuple.

generator (torch.Generator, optional) : A torch.Generator to make sampling deterministic.