espnet2.s2st.losses.guided_attention_loss.S2STGuidedAttentionLoss

source

class espnet2.s2st.losses.guided_attention_loss.S2STGuidedAttentionLoss(weight: float = 1.0, sigma: float = 0.4, alpha: float = 1.0)

Bases: AbsS2STLoss

Tacotron-based loss for S2ST.

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

forward(att_ws: Tensor, ilens: Tensor, olens_in: Tensor)

Forward.

Args:

• Returns: guided attention loss
• Return type: Tensor