Skip to content

weird behivior/implementation error? #4

Description

@nevakrien

i took the code for BitLinearOptimized and added a small thing so I can run it standalone

super(BitLinearOptimized, self).__init__(in_features, out_features, bias,dtype=torch.bfloat16) 
#just added the right dtype

runing the folowing

w=BitLinearOptimized(1,1)
x=torch.ones(1,dtype=torch.bfloat16)
y=w(x)
print(list(w.parameters()))

gives

[Parameter containing:
tensor([0.0703], dtype=torch.bfloat16, requires_grad=True)]

meaning that there is only the weight. is this an intended behivior?
because I saw in ur training u use model.parameters() so it seems like that would be an issue.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

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