Skip to content
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.

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

Support loss over all tokens to be back-propagated during training #28

Open
wants to merge 5 commits into
base: main
Choose a base branch
from

Conversation

maydixit
Copy link

@maydixit maydixit commented Sep 6, 2024

Support loss over all tokens to be back-propagated during training

Intended to resolve issue: #17

Copy link

google-cla bot commented Sep 6, 2024

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

@iislucas iislucas changed the title Maydixit Support loss over all tokens to be back-propagated during training (Maydixit) Sep 18, 2024
@iislucas
Copy link
Collaborator

As per our discussion, the key things to add to this are:

  • Causal masking so that training doesn't try to use future tokens to predict past ones.
  • A script that validates it "works" (we should see clearly better loss curves using this).

@iislucas iislucas self-requested a review September 18, 2024 10:03
@iislucas iislucas changed the title Support loss over all tokens to be back-propagated during training (Maydixit) Support loss over all tokens to be back-propagated during training Sep 25, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants