fix: scheduler division-by-zero, loss_mask bool

- schedule.py: guard warmup_steps/lr_decay_steps against zero
- strategy.py: use ~loss_mask instead of loss_mask==0 on bool tensor
This commit is contained in:
2026-07-03 22:04:55 +08:00
parent 70c0e5de90
commit 27d1921d9c
2 changed files with 10 additions and 4 deletions
+1 -1
View File
@@ -196,7 +196,7 @@ class SFTStrategy(BaseStrategy):
ignore_index = -100
input_mask = make_doc_boundary_mask(position_ids)
target_ids = target_ids.masked_fill(loss_mask == 0, ignore_index)
target_ids = target_ids.masked_fill(~loss_mask, ignore_index)
logits = self.model(
input_ids=input_ids, position_ids=position_ids, input_mask=input_mask
)["logits"]