CARDINAL类

中文数字/数位字符 每个字符除繁简体外还有一个额外的大写字符 e.g. '陆' 和 '陸'

r"""LayerNorm that supports two data formats: channels_last (default) or channels_first. The ordering of the dimensions in the inputs. channels_la

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

DIGIT类

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

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

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

TELEPHONE类

中文数字字符

Encoder layer module. Args: size (int): Input dimension. self_attn (torch.nn.Module): Self-attention module instance.

r"""ConvNeXt Block. There are two equivalent implementations: (1) DwConv -> LayerNorm (channels_first) -> 1x1 Conv -> GELU -> 1x1 Conv; all in (N,

ConvolutionModule in Conformer model.

Attention processor used typically in processing the SD3-like self-attention projections. add rms norm for query and key and apply RoPE

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

DATE类

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

Relative positional encoding module (new implementation). Details can be found in https://github.com/espnet/espnet/pull/2816. See : Appendix

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

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

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

FRACTION类

Linear transform the input without subsampling Args: idim (int): Input dimension. odim (int): Output dimension. dropout_r

A Sana block with global shared adaptive layer norm (adaLN-single) conditioning.

用于中文数字系统的数学符号 (繁/简体), e.g. positive = ['正', '正'] negative = ['负', '負'] point = ['点', '點']

MONEY类

中文数字系统

2D Image to Patch Embedding

PERCENTAGE类

Multi-Head Attention layer with relative position encoding. Paper: https://arxiv.org/abs/1901.02860 Args: n_head (int): The number of

The final layer of Sana.

Dataset for text-to-music generation that processes lyrics and audio files

中文字符 每个字符对应简体和繁体, e.g. 简体 = '负', 繁体 = '負' 转换时可转换为简体或繁体

中文数位字符

Conformer encoder module.

Multi-Head Attention layer. Args: n_head (int): The number of heads. n_feat (int): The number of features. dropout_rate (

Construct an Swish object.