Skip to content

About calculate trainable params #11

@CaffreyR

Description

@CaffreyR

Hi Rabeeh, in your code, you calculate the total_params_ratio by using this code.

total_params_ratio = ((total_params-t5_base_params)*8+t5_base_params)/t5_base_params

https://github.com/rabeehk/compacter/blob/main/seq2seq/utils/utils.py#L213

Why do your multiply 8 instead of just divide? Many thanks

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions