pytorch/docs/source/fsdp.rst

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

47 lines
1.1 KiB
ReStructuredText
Raw Normal View History

FullyShardedDataParallel
========================
.. automodule:: torch.distributed.fsdp
.. autoclass:: torch.distributed.fsdp.FullyShardedDataParallel
:members:
.. autoclass:: torch.distributed.fsdp.BackwardPrefetch
:members:
.. autoclass:: torch.distributed.fsdp.ShardingStrategy
:members:
.. autoclass:: torch.distributed.fsdp.MixedPrecision
:members:
.. autoclass:: torch.distributed.fsdp.CPUOffload
:members:
.. autoclass:: torch.distributed.fsdp.StateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.FullStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.ShardedStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.LocalStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.OptimStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.FullOptimStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.ShardedOptimStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.LocalOptimStateDictConfig
:members:
.. autoclass:: torch.distributed.fsdp.StateDictSettings
:members: