Sparse tensor with support for both torchsparse and spconv backends. Parameters: - feats (torch.Tensor): Features of the sparse tensor.

Embeds spatial positions into vector representations.

A 3D subdivide block that can subdivide the sparse tensor. Args: channels: channels in the inputs and outputs. out_channels:

Embeds scalar timesteps into vector representations.

Renderer for the Voxel representation. Args: rendering_options (dict): Rendering options.

A GroupNorm layer that converts to float32 before the forward pass.

Renderer for the Mesh representation. Args: rendering_options (dict): Rendering options. glctx (nvdiffrast.torch.RasterizeGL

Sparse Transformer cross-attention block (MSA + MCA + FFN) with adaptive layer norm conditioning.

Transformer cross-attention block (MSA + MCA + FFN) with adaptive layer norm conditioning.

A base class for pipelines.

Sparse Transformer block (MSA + FFN).

Pipeline for inferring Trellis image-to-3D models. Args: models (dict[str, nn.Module]): The models to use in the pipeline. s

Resample a variable-length token sequence to a fixed target length.

A mixin class for samplers that apply classifier-free guidance.

Generate samples from a flow-matching model using Euler sampling with classifier-free guidance.

Generate samples from a flow-matching model using Euler sampling with classifier-free guidance and interval.

Generate samples from a flow-matching model using Euler sampling. Args: sigma_min: The minimum scale of noise in flow.

A mixin class for samplers that apply classifier-free guidance with interval.

Sparse Transformer block (MSA + FFN) with adaptive layer norm conditioning.

Transformer block (MSA + FFN) with adaptive layer norm conditioning.

A base class for samplers.

Downsample a sparse tensor by a factor of `factor`. Implemented as average pooling.

A GroupNorm layer that converts to float32 before the forward pass.

A LayerNorm layer that converts to float32 before the forward pass.

Decoder for Sparse Structure (\mathcal{D}_S in the paper Sec. 3.3). Args: out_channels (int): Channels of the output. latent

Encoder for Sparse Structure (\mathcal{E}_S in the paper Sec. 3.3). Args: in_channels (int): Channels of the input. latent_c

Upsample a sparse tensor by a factor of `factor`. Implemented as nearest neighbor interpolation.

Sparse Transformer without output layers. Serve as the base class for encoder and decoder.

Sparse Transformer cross-attention block (MSA + MCA + FFN).

Upsample a sparse tensor by a factor of `factor`. Implemented as nearest neighbor interpolation.

Transformer block (MSA + FFN).

Transformer cross-attention block (MSA + MCA + FFN).