Small button with single emoji as text, fits inside gradio forms

Same as gr.Row but fits inside gradio forms

Causal self-attention with a single head.

RMSNorm with the reference ``(1 + scale)`` weight convention (scale stored zero-centered).

Dataclass container for most / all conditioning info. TODO: serialization methods aren't used, keeping them around for convenience

Defines the Unet submodule with skip connection. X -------------------identity---------------------- |-- downsampling -- |submodule|

Positionwise feed forward layer. FeedForward are appied on each position of the sequence. The output dim is same with the input dim. Arg

r""" A basic Transformer block. Parameters: dim (`int`): The number of channels in the input and output. num_attention_heads

A noise sampler backed by a torchsde.BrownianTree. Args: x (Tensor): The tensor whose shape, device and dtype to use to generate

SDXL-style residual block: GN -> SiLU -> Conv + condition (scale/shift) GN -> SiLU -> Dropout -> Conv + skip connect

r""" A feed-forward layer. Parameters: dim (`int`): The number of channels in the input. dim_out (`int`, *optional*): The num

Subclass torch's LayerNorm to handle fp16.

Reduce feature maps into a single edge map