kaira.losses.text.CosineSimilarityLoss

Inheritance diagram for CosineSimilarityLoss

class kaira.losses.text.CosineSimilarityLoss(margin=0.0)

Bases: BaseLoss

Cosine Similarity Loss Module.

This module calculates loss based on cosine similarity between embeddings.

Methods

__init__	Initialize the CosineSimilarityLoss module.
forward	Forward pass through the CosineSimilarityLoss module.

__init__(margin=0.0)

Initialize the CosineSimilarityLoss module.

Parameters:

margin (float) – Margin for similarity. Default is 0.0.

forward(x: Tensor, target: Tensor) → Tensor

Forward pass through the CosineSimilarityLoss module.

Parameters:

• x (torch.Tensor) – The input embeddings tensor.

• target (torch.Tensor) – The target embeddings tensor.

Returns:

The cosine similarity loss.

Return type:

torch.Tensor