Skip to content

fix masking formula bug and only calculate loss on masked tokens #16

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
EmaadKhwaja wants to merge 3 commits into
base: main
Choose a base branch
from
Open

fix masking formula bug and only calculate loss on masked tokens #16

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f858226
fix masking formula
EmaadKhwaja Sep 3, 2023
66cefbc
only calculate loss on masked tokens
EmaadKhwaja Sep 3, 2023
06541e1
bug fix
EmaadKhwaja Sep 3, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions transformer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,15 +68,15 @@ def forward(self, x):
# mask = mask.round().to(dtype=torch.int64)
# masked_indices = torch.zeros_like(z_indices)
masked_indices = self.mask_token_id * torch.ones_like(z_indices, device=z_indices.device)
a_indices = mask * z_indices + (~mask) * masked_indices
a_indices = ~mask * z_indices + (mask) * masked_indices

a_indices = torch.cat((sos_tokens, a_indices), dim=1)

target = torch.cat((sos_tokens, z_indices), dim=1)

logits = self.transformer(a_indices)

return logits, target
return logits[~mask], target[~mask]

def top_k_logits(self, logits, k):
v, ix = torch.topk(logits, k)
Expand Down