This function is used to re-order the :obj:`past_key_values` cache if :meth:`~transformers.PreTrainedModel.beam_search` or :meth:`~tr

This function is used to re-order the :obj:`past_key_values` cache if :meth:`~transformers.PreTrainedModel.beam_search` or :meth:`~tr

r""" Generates sequences of token ids for models with a language modeling head using **multinomial sampling** and can be used for text

Forward pass of the function. Applies the function to the input elementwise. Snake ∶= x + 1/a * sin^2 (xa)

Forward pass of the function. Applies the function to the input elementwise. SnakeBeta := x + 1/b * sin^2 (xa)

Args: x(torch.Tensor): shape (b, t, c) attn_mask(torch.Tensor): shape (b, t, t)

Args: x: shape (b, t, c) mask: shape (b, t, 1)

Args x: shape (b, t, c) c: shape (b, 1, c) attn_mask: shape (b, t, t), bool type attention mask conv_m

Args: x: shape (b, c, t) c: aux condition, shape (b, c, t) t: shape (b,) attn_mask: (b, t, t)

Args: query (torch.Tensor): shape (b, t1, c). key (torch.Tensor): shape (b, t2, c). value (torch.Tensor):

Add positional encoding. Args: x (torch.Tensor): Input tensor (batch, time, `*`). Returns: torch.Tensor: Enc

inputs: (batch_size, seq_len, channels)

Forward function. Args: xs: input tensor (B, L, D) Returns: output tensor, (B, L, D)

Input x. Args: x (torch.Tensor): Input tensor (#batch, time, idim). x_mask (torch.Tensor): Input mask (#batch, 1, time

Args: x: shape (b, t, c) mask: self-attention padded mask, shape (b, 1, t) or (b, t, t) pos_emb: relative

Args: xs: shape (b, t, c) xs_lens: shape (b,) attn_mask1: (token level) shape (b, t, t) attn_