Skip to content

Using padded tokens when creating averaged sentence embeddings #10

@AndrewLim1990

Description

@AndrewLim1990

When calculating the similarity loss between two sentences, it looks like we are using the averaged word embeddings per sentence. Within models.SDR.similarity_modeling.SimilarityModeling we have the following:

...
non_masked_outputs = self.roberta(
    non_masked_input_ids,
    attention_mask=attention_mask,
    token_type_ids=token_type_ids,
    position_ids=position_ids,
    head_mask=head_mask,
    inputs_embeds=inputs_embeds,
    output_hidden_states=output_hidden_states,
    return_dict=return_dict,
)
non_masked_seq_out = non_masked_outputs[0]

meaned_sentences = non_masked_seq_out.mean(1)
miner_output = list(self.miner_func(meaned_sentences, sample_labels))

sim_loss = self.similarity_loss_func(meaned_sentences, sample_labels, miner_output)
...

It appears using the embeddings for the padded tokens since we aren't taking into account any sentence lengths. Was this done by design perhaps?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions