espnet2.s2st.losses.attention_loss.S2STAttentionLoss

source

class espnet2.s2st.losses.attention_loss.S2STAttentionLoss(vocab_size: int, padding_idx: int = -1, weight: float = 1.0, smoothing: float = 0.0, normalize_length: str2bool = False, criterion: Module = KLDivLoss())

Bases: AbsS2STLoss

attention-based label smoothing loss for S2ST.

Initializes internal Module state, shared by both nn.Module and ScriptModule.

forward(dense_y: Tensor, token_y: Tensor)

Forward.

Args: