We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 3415d15 commit d364c25Copy full SHA for d364c25
modelopt/torch/speculative/plugins/transformers.py
@@ -833,7 +833,7 @@ def _eagle_loss(
833
eagle_logits
834
)
835
classification_loss = -torch.sum(torch.sum(loss_mask * classification_loss, 2)) / (
836
- loss_mask.sum() + 1e-5
+ loss_mask.sum() + 1e-6
837
838
# Compute accuracy
839
base_predict_tok = base_model_logits.clone().detach().argmax(dim=-1)
0 commit comments