Skip to content

Bias convention #2

Description

@bjornvz

Hello! In line 74 in layers.py you have implemented bias as follows:

    out = self.act_fn(torch.matmul(self.inp, self.weights))
    if self.use_bias:
        out = out + self.bias

Typically, one would have the bias inside the activation function, i.e. f(wx+b) as opposed to f(wx) + b (or, alternatively, wf(x)+b, but that requires additional changes). Is there any particular reason it was done this way? Did you experiment with the standard method?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions