# FullyShardedDataParallel

```{eval-rst}
.. automodule:: torch.distributed.fsdp
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.FullyShardedDataParallel
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.BackwardPrefetch
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.ShardingStrategy
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.MixedPrecision
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.CPUOffload
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.StateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.FullStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.ShardedStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.LocalStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.OptimStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.FullOptimStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.ShardedOptimStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.LocalOptimStateDictConfig
  :members:
```

```{eval-rst}
.. autoclass:: torch.distributed.fsdp.StateDictSettings
  :members:
```

```{eval-rst}
.. currentmodule:: torch.distributed.fsdp.wrap

.. autofunction:: always_wrap_policy

.. autofunction:: enable_wrap

.. autofunction:: lambda_auto_wrap_policy

.. autofunction:: size_based_auto_wrap_policy

.. autofunction:: transformer_auto_wrap_policy

.. autofunction:: wrap
```