Why is padding for target and input_ids performed before the for loop in traineagle3/cnets? #325

Open

opened

on Jan 26, 2026

Won't this cause the loss_mask to be misaligned?

Metadata

Assignees

No one assigned

Labels

No labels

No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests