r""" A custom 3D causal convolution layer with feature caching support. This layer extends the standard Conv3D layer by ensuring causality in

MLP as used in Vision Transformer, MLP-Mixer and related networks

Visualizer that draws data about detection/segmentation on images. It contains methods like `draw_{text,box,circle,line,binary_mask,polygon}`

r""" A custom residual block module. Args: in_dim (int): Number of input channels. out_dim (int): Number of output channels.

r""" A cross attention layer. Parameters: query_dim (`int`): The number of channels in the query. cross_attention

r""" Default processor for performing attention-related computations.

r""" Processor for implementing scaled dot-product attention (enabled by default if you're using PyTorch 2.0).

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

Projects caption embeddings. Also handles dropout for classifier-free guidance. Adapted from https://github.com/PixArt-alpha/PixArt-alpha/bl

r""" A custom RMS normalization layer. Args: dim (int): The number of dimensions to normalize over. channel_first (bool, opti

DPT Head for dense prediction tasks. This implementation follows the architecture described in "Vision Transformers for Dense Prediction"

Sample one batch from a selected dataset with given probability. Compatible with datasets at different resolution

r""" A custom resampling module for 2D and 3D data. Args: dim (int): The number of input/output channels. mode (str): The res

CameraHead predicts camera parameters from token representations using iterative refinement. It applies a series of transformer blocks (the

Create a simple end-to-end predictor with the given config that runs on single device for a single input image. Compared to using the mod

Attribute: polygons (list[ndarray]): list[ndarray]: polygons for this mask. Each ndarray has format [x, y, x, y, ...]

Very simple multi-layer perceptron (also called FFN)

MLP as used in Vision Transformer, MLP-Mixer and related networks

This is a more standard version of the position embedding, very similar to the one used by the Attention is all you need paper, generalized t

The Aggregator applies alternating-attention over input frames, as described in VGGT: Visual Geometry Grounded Transformer. Args:

Wrapper class to combine multiple :class:`DatasetEvaluator` instances. This class dispatches every evaluation call to all of its :class:

MLP as used in Vision Transformer, MLP-Mixer and related networks

MLP as used in Vision Transformer, MLP-Mixer and related networks

2D image to patch embedding: (B,C,H,W) -> (B,N,D) Args: img_size: Image size. patch_size: Patch token size. in_chans

Generates and caches 2D spatial positions for patches in a grid. This class efficiently manages the generation of spatial coordinates for patches

Residual convolution module.

2D Rotary Position Embedding implementation. This module applies rotary position embeddings to input tokens based on their 2D spatial positio

r""" Causal self-attention with a single head. Args: dim (int): The number of channels in the input tensor.

Middle block for WanVAE encoder and decoder. Args: dim (int): Number of input/output channels. dropout (float): Dropout rate

r""" Perform upsampling while ensuring the output tensor has the same data type as the input. Args: x (torch.Tensor): Input tensor to

The Aggregator applies alternating-attention over input frames, as described in VGGT: Visual Geometry Grounded Transformer. Args:

A predictor that runs the model asynchronously, possibly on >1 GPUs. Because rendering the visualization takes considerably amount of time,

r""" Processor for performing attention-related computations with extra learnable key and value matrices for the text encoder.

r""" Processor for implementing flash attention using torch_npu. Torch_npu supports only fp16 and bf16 data types. If fp32 is used, F.scaled_d

A basic Swin Transformer layer for one stage. Args: dim (int): Number of feature channels depth (int): Depths of this stage.

Modified version of COCOeval for evaluating AP with a custom maxDets (by default for COCO, maxDets is 100)

Modified version of COCOeval for evaluating AP with a custom maxDets (by default for COCO, maxDets is 100)

r""" Processor for implementing scaled dot-product attention for the CogVideoX model. It applies a rotary embedding on query and key vectors,

r""" Transformer block used in [CogVideoX](https://github.com/THUDM/CogVideo) model. Parameters: dim (`int`): The number

r""" Processor for implementing memory efficient attention using xFormers for the Custom Diffusion method. Args: train_kv (`bool`, defaul