Merge branch 'main' into fsdp-vs-ddp - SemanticDiff

Commit

14 days ago

Merge branch 'main' into fsdp-vs-ddp

References

#44083 - FSDP2 native support in transformers

#44974 - Refactor core_model_loading to support FSDP shard-on-read loading

#45028 - TP refactor for FSDP + TP integration

Author

3outeille

3outeille

Parents

Loading