pytorch-fsdp2
Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.
- Version
- 1.0.0
- License
- MIT
Pinned to revision df088027ff23, so it is the text this page describes rather than whatever the author pushed since.
Files
- skills/pytorch-fsdp2/SKILL.md
- skills/pytorch-fsdp2/references/pytorch_dcp_async_recipe.md
- skills/pytorch-fsdp2/references/pytorch_dcp_overview.md
- skills/pytorch-fsdp2/references/pytorch_dcp_recipe.md
- skills/pytorch-fsdp2/references/pytorch_ddp_notes.md
- skills/pytorch-fsdp2/references/pytorch_device_mesh_tutorial.md
- skills/pytorch-fsdp2/references/pytorch_examples_fsdp2.md
- skills/pytorch-fsdp2/references/pytorch_fsdp1_api.md
- skills/pytorch-fsdp2/references/pytorch_fsdp2_tutorial.md
- skills/pytorch-fsdp2/references/pytorch_fully_shard_api.md
- skills/pytorch-fsdp2/references/pytorch_tp_tutorial.md
- skills/pytorch-fsdp2/references/ray_train_fsdp2_example.md
- skills/pytorch-fsdp2/references/torchtitan_fsdp_notes.md
Every link opens the file at its source, pinned to the revision this page describes.