transformers
feat(granitemoe*): Remove logits upcast when computing loss
#42753

Merged

feat(granitemoe*): Remove logits upcast when computing loss #42753

ArthurZucker merged 3 commits into huggingface:main from gabe-l-hart:GraniteOptionalUpcast-42709

feat: Remove logits upcast when computing loss

1f8d4288

chore: make fix-copies

3dbe486d

Ssukriti approved these changes on 2025-12-10

Merge branch 'main' into GraniteOptionalUpcast-42709

47f810f4

ArthurZucker approved these changes on 2025-12-11

ArthurZucker merged 0af2381f into main 9 days ago

gabe-l-hart deleted the GraniteOptionalUpcast-42709 branch 9 days ago

Reviewers

ArthurZucker

Ssukriti

Assignees

No one assigned

Labels

None yet

Milestone

No milestone