Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.

0 0 104 Python

Fix ModelParallelStrategy fails with non-distributed checkpoint. (#21384)

* Add regression test for ModelParallel single-file checkpoint

* Fix ModelParallel single-file checkpoint with compiled modules

littlebullGit committed 8mo ago

28f18dcc4e806f42c0191671a55a1da4fcb57ca2

Parent: 9e1b038

Committed by GitHub <noreply@github.com> on 12/19/2025, 2:23:25 PM