0 0 1 Python

Let gradient checkpointing skip layers with every_n_layers (#48200)

* Let gradient checkpointing skip layers with every_n_layers

* Read every_n_layers from gradient_checkpointing_kwargs instead of a new argument

* Document every_n_layers in the field help text too

---------

Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

Quentin Gallouédec committed 7d ago

2b4e2b3eea90a02aa0c754335dc51b6d57ad4c45

Parent: fe1dfb3

Committed by GitHub <noreply@github.com> on 8/25/2026, 2:38:56 PM