# tensorplay.distributed.fsdp API Source: https://www.tensorplay.cn/docs/api/tensorplay.distributed.fsdp.html ## Functions 3 [#](#api-tensorplay.distributed.fsdp.fully_shard) ### fully_shard function[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.fully_shard.html) ```python tensorplay.distributed.fsdp.fully_shard(module: Any, *, mesh: Any = None, reshard_after_forward: bool | int | None = None, shard_placement_fn: Callable[[Any], Any] | None = None, mp_policy: MixedPrecisionPolicy | None = None, offload_policy: OffloadPolicy | None = None, ignored_params: set[Any] | None = None, dp_mesh_dims: DataParallelMeshDims | None = None) → Any ``` [#](#api-tensorplay.distributed.fsdp.register_fsdp_forward_method) ### register_fsdp_forward_method function[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.register_fsdp_forward_method.html) ```python tensorplay.distributed.fsdp.register_fsdp_forward_method(module: Any, method_name: str) → None ``` [#](#api-tensorplay.distributed.fsdp.share_comm_ctx) ### share_comm_ctx function[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.share_comm_ctx.html) ```python tensorplay.distributed.fsdp.share_comm_ctx(modules: list[FSDPModule]) → None ``` ## Classes 22 [#](#api-tensorplay.distributed.fsdp.BackwardPrefetch) ### BackwardPrefetch class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.BackwardPrefetch.html) ```python class tensorplay.distributed.fsdp.BackwardPrefetch(*values) ``` [#](#api-tensorplay.distributed.fsdp.CPUOffload) ### CPUOffload class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.CPUOffload.html) ```python class tensorplay.distributed.fsdp.CPUOffload(offload_params: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.CPUOffloadPolicy) ### CPUOffloadPolicy class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.CPUOffloadPolicy.html) ```python class tensorplay.distributed.fsdp.CPUOffloadPolicy(pin_memory: bool = True) ``` [#](#api-tensorplay.distributed.fsdp.DataParallelMeshDims) ### DataParallelMeshDims class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.DataParallelMeshDims.html) ```python class tensorplay.distributed.fsdp.DataParallelMeshDims(shard: str | tuple[str, ...] | None = None, replicate: str | tuple[str, ...] | None = None) ``` [#](#api-tensorplay.distributed.fsdp.FSDPModule) ### FSDPModule class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.FSDPModule.html) ```python class tensorplay.distributed.fsdp.FSDPModule(*args: Any, **kwargs: Any) ``` Methods mixed into modules managed by [fully_shard()](/docs/generated/tensorplay.distributed.fsdp.fully_shard.html#tensorplay.distributed.fsdp.fully_shard). [#](#api-tensorplay.distributed.fsdp.FullOptimStateDictConfig) ### FullOptimStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.FullOptimStateDictConfig.html) ```python class tensorplay.distributed.fsdp.FullOptimStateDictConfig(offload_to_cpu: bool = True, rank0_only: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.FullStateDictConfig) ### FullStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.FullStateDictConfig.html) ```python class tensorplay.distributed.fsdp.FullStateDictConfig(offload_to_cpu: bool = False, rank0_only: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.FullyShardedDataParallel) ### FullyShardedDataParallel class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.FullyShardedDataParallel.html) ```python class tensorplay.distributed.fsdp.FullyShardedDataParallel(module: Module, process_group: Any = None, sharding_strategy: ShardingStrategy | None = None, cpu_offload: CPUOffload | None = None, auto_wrap_policy: Any = None, backward_prefetch: BackwardPrefetch | None = BackwardPrefetch.BACKWARD_PRE, mixed_precision: MixedPrecision | None = None, ignored_modules: Iterable[Module] | None = None, param_init_fn: Any = None, device_id: Any = None, sync_module_states: bool = False, forward_prefetch: bool = False, limit_all_gathers: bool = True, use_orig_params: bool = False, ignored_states: Iterable[Any] | None = None, device_mesh: Any = None) ``` Wrap a module and manage its parameter shards around each forward. ```python add_module(name: str, module: Module | None) → None ``` Add a child module to the current module. The module can be accessed as an attribute using the given name. Parameters: - name ([str](https://docs.python.org/3/builtins/stdtypes.html#str)) – name of the child module. The child module can be accessed from this module using the given name - module (Module) – child module to be added to the module. ```python bfloat16() → Self ``` Casts all floating point parameters and buffers to bfloat16 datatype. > **Note** > > This method modifies the module in-place. Returns: self Return type: Module ```python buffers(recurse: bool = True) → Iterator[Tensor] ``` Return an iterator over module buffers. Parameters: recurse ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – if True, then yields buffers of this module and all submodules. Otherwise, yields only buffers that are direct members of this module. Yields: tensorplay.Tensor – module buffer Example: ``` >>> # xdoctest: +SKIP("undefined vars") >>> for buf in model.buffers(): >>> print(type(buf), buf.size()) (20L,) (20L, 1L, 5L, 5L) ``` ```python children() → Iterator[Module] ``` Return an iterator over immediate children modules. Yields: Module – a child module ```python compile(*args, **kwargs) ``` Compile this Module’s forward using tensorplay.compile(). This Module’s __call__ method is compiled and all arguments are passed as-is to tensorplay.compile(). See tensorplay.compile() for details on the arguments for this function. ```python cpu() → Self ``` Move all model parameters and buffers to the CPU. > **Note** > > This method modifies the module in-place. Returns: self Return type: Module ```python cuda(device: int | device | None = None) → Self ``` Move all model parameters and buffers to the GPU. This also makes associated parameters and buffers different objects. So it should be called before constructing the optimizer if the module will live on GPU while being optimized. > **Note** > > This method modifies the module in-place. Parameters: device ([int](https://docs.python.org/3/builtins/functions.html#int), optional) – if specified, all parameters will be copied to that device Returns: self Return type: Module ```python double() → Self ``` Casts all floating point parameters and buffers to double datatype. > **Note** > > This method modifies the module in-place. Returns: self Return type: Module ```python eval() → Self ``` Set the module in evaluation mode. This has an effect only on certain modules. See the documentation of particular modules for details of their behaviors in training/evaluation mode, i.e. whether they are affected, e.g. Dropout, BatchNorm, etc. This is equivalent with self.train(False). See [Locally disabling gradient computation](/docs/notes/autograd.html#locally-disable-grad-doc) for a comparison between .eval() and several similar mechanisms that may be confused with it. Returns: self Return type: Module ```python extra_repr() → str ``` Return the extra representation of the module. To print customized extra information, you should re-implement this method in your own modules. Both single-line and multi-line strings are acceptable. ```python float() → Self ``` Casts all floating point parameters and buffers to float datatype. > **Note** > > This method modifies the module in-place. Returns: self Return type: Module ```python get_buffer(target: str) → Tensor ``` Return the buffer given by target if it exists, otherwise throw an error. See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target. Parameters: target – The fully-qualified string name of the buffer to look for. (See get_submodule for how to specify a fully-qualified string.) Returns: The buffer referenced by target Return type: [tensorplay.Tensor](/docs/generated/tensorplay.Tensor.html#tensorplay.Tensor) Raises: [AttributeError](https://docs.python.org/3/builtins/exceptions.html#AttributeError) – If the target string references an invalid path or resolves to something that is not a buffer ```python get_extra_state() → Any ``` Return any extra state to include in the module’s state_dict. Implement this and a corresponding [set_extra_state()](#tensorplay.distributed.fsdp.FullyShardedDataParallel.set_extra_state) for your module if you need to store extra state. This function is called when building the module’s state_dict(). Note that extra state should be picklable to ensure working serialization of the state_dict. We only provide backwards compatibility guarantees for serializing Tensors; other objects may break backwards compatibility if their serialized pickled form changes. Returns: Any extra state to store in the module’s state_dict Return type: [object](https://docs.python.org/3/builtins/functions.html#object) ```python get_parameter(target: str) → Parameter ``` Return the parameter given by target if it exists, otherwise throw an error. See the docstring for get_submodule for a more detailed explanation of this method’s functionality as well as how to correctly specify target. Parameters: target – The fully-qualified string name of the Parameter to look for. (See get_submodule for how to specify a fully-qualified string.) Returns: The Parameter referenced by target Return type: tensorplay.nn.Parameter Raises: [AttributeError](https://docs.python.org/3/builtins/exceptions.html#AttributeError) – If the target string references an invalid path or resolves to something that is not an nn.Parameter ```python get_submodule(target: str) → Module ``` Return the submodule given by target if it exists, otherwise throw an error. For example, let’s say you have an nn.Module A that looks like this: ``` A( (net_b): Module( (net_c): Module( (conv): Conv2d(16, 33, kernel_size=(3, 3), stride=(2, 2)) ) (linear): Linear(in_features=100, out_features=200, bias=True) ) ) ``` (The diagram shows an nn.Module A. A which has a nested submodule net_b, which itself has two submodules net_c and linear. net_c then has a submodule conv.) To check whether or not we have the linear submodule, we would call get_submodule("net_b.linear"). To check whether we have the conv submodule, we would call get_submodule("net_b.net_c.conv"). The runtime of get_submodule is bounded by the degree of module nesting in target. A query against named_modules achieves the same result, but it is O(N) in the number of transitive modules. So, for a simple check to see if some submodule exists, get_submodule should always be used. Parameters: target – The fully-qualified string name of the submodule to look for. (See above example for how to specify a fully-qualified string.) Returns: The submodule referenced by target Return type: tensorplay.nn.Module Raises: [AttributeError](https://docs.python.org/3/builtins/exceptions.html#AttributeError) – If at any point along the path resulting from the target string the (sub)path resolves to a non-existent attribute name or an object that is not an instance of nn.Module. ```python half() → Self ``` Casts all floating point parameters and buffers to half datatype. > **Note** > > This method modifies the module in-place. Returns: self Return type: Module ```python modules() → Iterator[Module] ``` Return an iterator over all modules in the network. Yields: Module – a module in the network > **Note** > > Duplicate modules are returned only once. In the following example, l will be returned only once. Example: ``` >>> l = nn.Linear(2, 2) >>> net = nn.Sequential(l, l) >>> for idx, m in enumerate(net.modules()): ... print(idx, '->', m) 0 -> Sequential( (0): Linear(in_features=2, out_features=2, bias=True) (1): Linear(in_features=2, out_features=2, bias=True) ) 1 -> Linear(in_features=2, out_features=2, bias=True) ``` ```python named_children() → Iterator[tuple[str, Module]] ``` Return an iterator over immediate children modules, yielding both the name of the module as well as the module itself. Yields: (str, Module) – Tuple containing a name and child module Example: ``` >>> # xdoctest: +SKIP("undefined vars") >>> for name, module in model.named_children(): >>> if name in ['conv4', 'conv5']: >>> print(module) ``` ```python named_modules(memo: set[Module] | None = None, prefix: str = '', remove_duplicate: bool = True) ``` Return an iterator over all modules in the network, yielding both the name of the module as well as the module itself. Parameters: - memo – a memo to store the set of modules already added to the result - prefix – a prefix that will be added to the name of the module - remove_duplicate – whether to remove the duplicated module instances in the result or not Yields: (str, Module) – Tuple of name and module > **Note** > > Duplicate modules are returned only once. In the following example, l will be returned only once. Example: ``` >>> l = nn.Linear(2, 2) >>> net = nn.Sequential(l, l) >>> for idx, m in enumerate(net.named_modules()): ... print(idx, '->', m) 0 -> ('', Sequential( (0): Linear(in_features=2, out_features=2, bias=True) (1): Linear(in_features=2, out_features=2, bias=True) )) 1 -> ('0', Linear(in_features=2, out_features=2, bias=True)) ``` ```python parameters(recurse: bool = True) → Iterator[Parameter] ``` Return an iterator over module parameters. This is typically passed to an optimizer. Parameters: recurse ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – if True, then yields parameters of this module and all submodules. Otherwise, yields only parameters that are direct members of this module. Yields: Parameter – module parameter Example: ``` >>> # xdoctest: +SKIP("undefined vars") >>> for param in model.parameters(): >>> print(type(param), param.size()) (20L,) (20L, 1L, 5L, 5L) ``` ```python register_buffer(name: str, tensor: Tensor | None, persistent: bool = True) → None ``` Add a buffer to the module. This is typically used to register a buffer that should not be considered a model parameter. For example, BatchNorm’s running_mean is not a parameter, but is part of the module’s state. Buffers, by default, are persistent and will be saved alongside parameters. This behavior can be changed by setting persistent to False. The only difference between a persistent buffer and a non-persistent buffer is that the latter will not be a part of this module’s state_dict. Buffers can be accessed as attributes using given names. Parameters: - name ([str](https://docs.python.org/3/builtins/stdtypes.html#str)) – name of the buffer. The buffer can be accessed from this module using the given name - tensor ([Tensor](/docs/generated/tensorplay.Tensor.html#tensorplay.Tensor) or None) – buffer to be registered. If None, then operations that run on buffers, such as [cuda](#tensorplay.distributed.fsdp.FullyShardedDataParallel.cuda), are ignored. If None, the buffer is not included in the module’s state_dict. - persistent ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – whether the buffer is part of this module’s state_dict. Example: ``` >>> # xdoctest: +SKIP("undefined vars") >>> self.register_buffer('running_mean', tensorplay.zeros(num_features)) ``` ```python register_forward_hook(hook: Callable[[T, tuple[Any, ...], Any], Any | None] | Callable[[T, tuple[Any, ...], dict[str, Any], Any], Any | None], *, prepend: bool = False, with_kwargs: bool = False, always_call: bool = False) → RemovableHandle ``` Register a forward hook on the module. The hook will be called every time after forward() has computed an output. If with_kwargs is False or not specified, the input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the output. It can modify the input inplace but it will not have effect on forward since this is called after forward() is called. The hook should have the following signature: ``` hook(module, args, output) -> None or modified output ``` If with_kwargs is True, the forward hook will be passed the kwargs given to the forward function and be expected to return the output possibly modified. The hook should have the following signature: ``` hook(module, args, kwargs, output) -> None or modified output ``` Parameters: - hook (Callable) – The user defined hook to be registered. - prepend ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If True, the provided hook will be fired before all existing forward hooks on this tensorplay.nn.Module. Otherwise, the provided hook will be fired after all existing forward hooks on this tensorplay.nn.Module. Note that global forward hooks registered with register_module_forward_hook() will fire before all hooks registered by this method. Default: False - with_kwargs ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If True, the hook will be passed the kwargs given to the forward function. Default: False - always_call ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If True the hook will be run regardless of whether an exception is raised while calling the Module. Default: False Returns: a handle that can be used to remove the added hook by calling handle.remove() Return type: tensorplay.utils.hooks.RemovableHandle ```python register_forward_pre_hook(hook: Callable[[T, tuple[Any, ...]], Any | None] | Callable[[T, tuple[Any, ...], dict[str, Any]], tuple[Any, dict[str, Any]] | None], *, prepend: bool = False, with_kwargs: bool = False) → RemovableHandle ``` Register a forward pre-hook on the module. The hook will be called every time before forward() is invoked. If with_kwargs is false or not specified, the input contains only the positional arguments given to the module. Keyword arguments won’t be passed to the hooks and only to the forward. The hook can modify the input. User can either return a tuple or a single modified value in the hook. We will wrap the value into a tuple if a single value is returned (unless that value is already a tuple). The hook should have the following signature: ``` hook(module, args) -> None or modified input ``` If with_kwargs is true, the forward pre-hook will be passed the kwargs given to the forward function. And if the hook modifies the input, both the args and kwargs should be returned. The hook should have the following signature: ``` hook(module, args, kwargs) -> None or a tuple of modified input and kwargs ``` Parameters: - hook (Callable) – The user defined hook to be registered. - prepend ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If true, the provided hook will be fired before all existing forward_pre hooks on this tensorplay.nn.Module. Otherwise, the provided hook will be fired after all existing forward_pre hooks on this tensorplay.nn.Module. Note that global forward_pre hooks registered with register_module_forward_pre_hook() will fire before all hooks registered by this method. Default: False - with_kwargs ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If true, the hook will be passed the kwargs given to the forward function. Default: False Returns: a handle that can be used to remove the added hook by calling handle.remove() Return type: tensorplay.utils.hooks.RemovableHandle ```python register_full_backward_hook(hook: Callable[[Module, tuple[Tensor, ...] | Tensor, tuple[Tensor, ...] | Tensor], None | tuple[Tensor, ...] | Tensor], prepend: bool = False) → RemovableHandle ``` Register a backward hook on the module. The hook will be called every time the gradients with respect to a module are computed, and its firing rules are as follows: - Ordinarily, the hook fires when the gradients are computed with respect to the module inputs. - If none of the module inputs require gradients, the hook will fire when the gradients are computed with respect to module outputs. - If none of the module outputs require gradients, then the hooks will not fire. The hook should have the following signature: ``` hook(module, grad_input, grad_output) -> tuple(Tensor) or None ``` The grad_input and grad_output are tuples that contain the gradients with respect to the inputs and outputs respectively. The hook should not modify its arguments, but it can optionally return a new gradient with respect to the input that will be used in place of grad_input in subsequent computations. grad_input will only correspond to the inputs given as positional arguments and all kwarg arguments are ignored. Entries in grad_input and grad_output will be None for all non-Tensor arguments. For technical reasons, when this hook is applied to a Module, its forward function will receive a view of each Tensor passed to the Module. Similarly the caller will receive a view of each Tensor returned by the Module’s forward function. > **Warning** > > Modifying inputs or outputs inplace is not allowed when using backward hooks and will raise an error. Parameters: - hook (Callable) – The user-defined hook to be registered. - prepend ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If true, the provided hook will be fired before all existing backward hooks on this tensorplay.nn.Module. Otherwise, the provided hook will be fired after all existing backward hooks on this tensorplay.nn.Module. Note that global backward hooks registered with register_module_full_backward_hook() will fire before all hooks registered by this method. Returns: a handle that can be used to remove the added hook by calling handle.remove() Return type: tensorplay.utils.hooks.RemovableHandle ```python register_full_backward_pre_hook(hook: Callable[[Module, tuple[Tensor, ...] | Tensor], None | tuple[Tensor, ...] | Tensor], prepend: bool = False) → RemovableHandle ``` Register a backward pre-hook on the module. The hook will be called every time the gradients for the module are computed. The hook should have the following signature: ``` hook(module, grad_output) -> tuple[Tensor] or None ``` The grad_output is a tuple. The hook should not modify its arguments, but it can optionally return a new gradient with respect to the output that will be used in place of grad_output in subsequent computations. Entries in grad_output will be None for all non-Tensor arguments. For technical reasons, when this hook is applied to a Module, its forward function will receive a view of each Tensor passed to the Module. Similarly the caller will receive a view of each Tensor returned by the Module’s forward function. > **Warning** > > Modifying inputs inplace is not allowed when using backward hooks and will raise an error. Parameters: - hook (Callable) – The user-defined hook to be registered. - prepend ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – If true, the provided hook will be fired before all existing backward_pre hooks on this tensorplay.nn.Module. Otherwise, the provided hook will be fired after all existing backward_pre hooks on this tensorplay.nn.Module. Note that global backward_pre hooks registered with register_module_full_backward_pre_hook() will fire before all hooks registered by this method. Returns: a handle that can be used to remove the added hook by calling handle.remove() Return type: tensorplay.utils.hooks.RemovableHandle ```python register_load_state_dict_post_hook(hook) ``` Register a post-hook to be run after module’s load_state_dict() is called. It should have the following signature:: hook(module, incompatible_keys) -> None The module argument is the current module that this hook is registered on, and the incompatible_keys argument is a NamedTuple consisting of attributes missing_keys and unexpected_keys. missing_keys is a list of str containing the missing keys and unexpected_keys is a list of str containing the unexpected keys. The given incompatible_keys can be modified inplace if needed. Note that the checks performed when calling load_state_dict() with strict=True are affected by modifications the hook makes to missing_keys or unexpected_keys, as expected. Additions to either set of keys will result in an error being thrown when strict=True, and clearing out both missing and unexpected keys will avoid an error. Returns: a handle that can be used to remove the added hook by calling handle.remove() Return type: tensorplay.utils.hooks.RemovableHandle ```python register_load_state_dict_pre_hook(hook) ``` Register a pre-hook to be run before module’s load_state_dict() is called. It should have the following signature:: hook(module, state_dict, prefix, local_metadata, strict, missing_keys, unexpected_keys, error_msgs) -> None # noqa: B950 Parameters: hook (Callable) – Callable hook that will be invoked before loading the state dict. ```python register_module(name: str, module: Module | None) → None ``` Alias for [add_module()](#tensorplay.distributed.fsdp.FullyShardedDataParallel.add_module). ```python register_parameter(name: str, param: Parameter | None) → None ``` Add a parameter to the module. The parameter can be accessed as an attribute using given name. Parameters: - name ([str](https://docs.python.org/3/builtins/stdtypes.html#str)) – name of the parameter. The parameter can be accessed from this module using the given name - param (Parameter or None) – parameter to be added to the module. If None, then operations that run on parameters, such as [cuda](#tensorplay.distributed.fsdp.FullyShardedDataParallel.cuda), are ignored. If None, the parameter is not included in the module’s state_dict. ```python register_state_dict_post_hook(hook) ``` Register a post-hook for the state_dict() method. It should have the following signature:: hook(module, state_dict, prefix, local_metadata) -> None The registered hooks can modify the state_dict inplace. ```python register_state_dict_pre_hook(hook) ``` Register a pre-hook for the state_dict() method. It should have the following signature:: hook(module, prefix, keep_vars) -> None The registered hooks can be used to perform pre-processing before the state_dict call is made. ```python requires_grad_(requires_grad: bool = True) → Self ``` Change if autograd should record operations on parameters in this module. This method sets the parameters’ requires_grad attributes in-place. This method is helpful for freezing part of the module for finetuning or training parts of a model individually (e.g., GAN training). See [Locally disabling gradient computation](/docs/notes/autograd.html#locally-disable-grad-doc) for a comparison between .requires_grad_() and several similar mechanisms that may be confused with it. Parameters: requires_grad ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – whether autograd should record operations on parameters in this module. Default: True. Returns: self Return type: Module ```python set_extra_state(state: Any) → None ``` Set extra state contained in the loaded state_dict. This function is called from load_state_dict() to handle any extra state found within the state_dict. Implement this function and a corresponding [get_extra_state()](#tensorplay.distributed.fsdp.FullyShardedDataParallel.get_extra_state) for your module if you need to store extra state within its state_dict. Parameters: state ([dict](https://docs.python.org/3/builtins/stdtypes.html#dict)) – Extra state from the state_dict ```python set_submodule(target: str, module: Module, strict: bool = False) → None ``` Set the submodule given by target if it exists, otherwise throw an error. > **Note** > > If strict is set to False (default), the method will replace an existing submodule or create a new submodule if the parent module exists. If strict is set to True, the method will only attempt to replace an existing submodule and throw an error if the submodule does not exist. For example, let’s say you have an nn.Module A that looks like this: ``` A( (net_b): Module( (net_c): Module( (conv): Conv2d(3, 3, 3) ) (linear): Linear(3, 3) ) ) ``` (The diagram shows an nn.Module A. A has a nested submodule net_b, which itself has two submodules net_c and linear. net_c then has a submodule conv.) To override the Conv2d with a new submodule Linear, you could call set_submodule("net_b.net_c.conv", nn.Linear(1, 1)) where strict could be True or False To add a new submodule Conv2d to the existing net_b module, you would call set_submodule("net_b.conv", nn.Conv2d(1, 1, 1)). In the above if you set strict=True and call set_submodule("net_b.conv", nn.Conv2d(1, 1, 1), strict=True), an AttributeError will be raised because net_b does not have a submodule named conv. Parameters: - target – The fully-qualified string name of the submodule to look for. (See above example for how to specify a fully-qualified string.) - module – The module to set the submodule to. - strict – If False, the method will replace an existing submodule or create a new submodule if the parent module exists. If True, the method will only attempt to replace an existing submodule and throw an error if the submodule doesn’t already exist. Raises: - [ValueError](https://docs.python.org/3/builtins/exceptions.html#ValueError) – If the target string is empty or if module is not an instance of nn.Module. - [AttributeError](https://docs.python.org/3/builtins/exceptions.html#AttributeError) – If at any point along the path resulting from the target string the (sub)path resolves to a non-existent attribute name or an object that is not an instance of nn.Module. ```python share_memory() → Self ``` See tensorplay.Tensor.share_memory_(). ```python to(*args, **kwargs) ``` Move and/or cast the parameters and buffers. This can be called as ```python to(device=None, dtype=None, non_blocking=False) ``` ```python to(dtype, non_blocking=False) ``` ```python to(tensor, non_blocking=False) ``` ```python to(memory_format=tensorplay.channels_last) ``` Its signature is similar to tensorplay.Tensor.to(), but only accepts floating point or complex dtypes. In addition, this method will only cast the floating point or complex parameters and buffers to dtype (if given). The integral parameters and buffers will be moved device, if that is given, but with dtypes unchanged. When non_blocking is set, it tries to convert/move asynchronously with respect to the host if possible, e.g., moving CPU Tensors with pinned memory to CUDA devices. See below for examples. > **Note** > > This method modifies the module in-place. Parameters: - device ([tensorplay.device](/docs/generated/tensorplay.Device.html#tensorplay.Device)) – the desired device of the parameters and buffers in this module - dtype ([tensorplay.dtype](/docs/generated/tensorplay.DType.html#tensorplay.DType)) – the desired floating point or complex dtype of the parameters and buffers in this module - tensor ([tensorplay.Tensor](/docs/generated/tensorplay.Tensor.html#tensorplay.Tensor)) – Tensor whose dtype and device are the desired dtype and device for all parameters and buffers in this module - memory_format ([tensorplay.memory_format](/docs/generated/tensorplay.MemoryFormat.html#tensorplay.MemoryFormat)) – the desired memory format for 4D parameters and buffers in this module (keyword only argument) Returns: self Return type: Module Examples: ``` >>> # xdoctest: +IGNORE_WANT("non-deterministic") >>> linear = nn.Linear(2, 2) >>> linear.weight Parameter containing: tensor([[ 0.1913, -0.3420], [-0.5113, -0.2325]]) >>> linear.to(tensorplay.double) Linear(in_features=2, out_features=2, bias=True) >>> linear.weight Parameter containing: tensor([[ 0.1913, -0.3420], [-0.5113, -0.2325]], dtype=tensorplay.float64) >>> # xdoctest: +REQUIRES(env:TENSORPLAY_DOCTEST_CUDA1) >>> gpu1 = tensorplay.device("cuda:1") >>> linear.to(gpu1, dtype=tensorplay.half, non_blocking=True) Linear(in_features=2, out_features=2, bias=True) >>> linear.weight Parameter containing: tensor([[ 0.1914, -0.3420], [-0.5112, -0.2324]], dtype=tensorplay.float16, device='cuda:1') >>> cpu = tensorplay.device("cpu") >>> linear.to(cpu) Linear(in_features=2, out_features=2, bias=True) >>> linear.weight Parameter containing: tensor([[ 0.1914, -0.3420], [-0.5112, -0.2324]], dtype=tensorplay.float16) >>> linear = nn.Linear(2, 2, bias=None).to(tensorplay.cdouble) >>> linear.weight Parameter containing: tensor([[ 0.3741+0.j, 0.2382+0.j], [ 0.5593+0.j, -0.4443+0.j]], dtype=tensorplay.complex128) >>> linear(tensorplay.ones(3, 2, dtype=tensorplay.cdouble)) tensor([[0.6122+0.j, 0.1150+0.j], [0.6122+0.j, 0.1150+0.j], [0.6122+0.j, 0.1150+0.j]], dtype=tensorplay.complex128) ``` ```python to_empty(*, device: str | device | int | None, recurse: bool = True) → Self ``` Move the parameters and buffers to the specified device without copying storage. Parameters: - device ([tensorplay.device](/docs/generated/tensorplay.Device.html#tensorplay.Device)) – The desired device of the parameters and buffers in this module. - recurse ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – Whether parameters and buffers of submodules should be recursively moved to the specified device. Returns: self Return type: Module ```python train(mode: bool = True) → Self ``` Set the module in training mode. This has an effect only on certain modules. See the documentation of particular modules for details of their behaviors in training/evaluation mode, i.e., whether they are affected, e.g. Dropout, BatchNorm, etc. Parameters: mode ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – whether to set training mode (True) or evaluation mode (False). Default: True. Returns: self Return type: Module ```python type(dst_type: dtype | str) → Self ``` Casts all parameters and buffers to dst_type. > **Note** > > This method modifies the module in-place. Parameters: dst_type ([type](https://docs.python.org/3/builtins/functions.html#type) or string) – the desired type Returns: self Return type: Module ```python zero_grad(set_to_none: bool = True) → None ``` Reset gradients of all model parameters. See similar function under tensorplay.optim.Optimizer for more context. Parameters: set_to_none ([bool](https://docs.python.org/3/builtins/functions.html#bool)) – instead of setting to zero, set the grads to None. See tensorplay.optim.Optimizer.zero_grad() for details. [#](#api-tensorplay.distributed.fsdp.LocalOptimStateDictConfig) ### LocalOptimStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.LocalOptimStateDictConfig.html) ```python class tensorplay.distributed.fsdp.LocalOptimStateDictConfig(offload_to_cpu: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.LocalStateDictConfig) ### LocalStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.LocalStateDictConfig.html) ```python class tensorplay.distributed.fsdp.LocalStateDictConfig(offload_to_cpu: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.MixedPrecision) ### MixedPrecision class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.MixedPrecision.html) ```python class tensorplay.distributed.fsdp.MixedPrecision(param_dtype: Any = None, reduce_dtype: Any = None, buffer_dtype: Any = None, keep_low_precision_grads: bool = False, cast_forward_inputs: bool = False, cast_root_forward_inputs: bool = True, _module_classes_to_ignore: collections.abc.Sequence[type] = (, )) ``` [#](#api-tensorplay.distributed.fsdp.MixedPrecisionPolicy) ### MixedPrecisionPolicy class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.MixedPrecisionPolicy.html) ```python class tensorplay.distributed.fsdp.MixedPrecisionPolicy(param_dtype: Any = None, reduce_dtype: Any = None, output_dtype: Any = None, cast_forward_inputs: bool = True) ``` [#](#api-tensorplay.distributed.fsdp.OffloadPolicy) ### OffloadPolicy class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.OffloadPolicy.html) ```python class tensorplay.distributed.fsdp.OffloadPolicy ``` [#](#api-tensorplay.distributed.fsdp.OptimStateDictConfig) ### OptimStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.OptimStateDictConfig.html) ```python class tensorplay.distributed.fsdp.OptimStateDictConfig(offload_to_cpu: bool = True) ``` [#](#api-tensorplay.distributed.fsdp.OptimStateKeyType) ### OptimStateKeyType class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.OptimStateKeyType.html) ```python class tensorplay.distributed.fsdp.OptimStateKeyType(*values) ``` [#](#api-tensorplay.distributed.fsdp.ShardedOptimStateDictConfig) ### ShardedOptimStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.ShardedOptimStateDictConfig.html) ```python class tensorplay.distributed.fsdp.ShardedOptimStateDictConfig(offload_to_cpu: bool = True, _use_dtensor: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.ShardedStateDictConfig) ### ShardedStateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.ShardedStateDictConfig.html) ```python class tensorplay.distributed.fsdp.ShardedStateDictConfig(offload_to_cpu: bool = False, _use_dtensor: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.ShardingStrategy) ### ShardingStrategy class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.ShardingStrategy.html) ```python class tensorplay.distributed.fsdp.ShardingStrategy(*values) ``` [#](#api-tensorplay.distributed.fsdp.StateDictConfig) ### StateDictConfig class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.StateDictConfig.html) ```python class tensorplay.distributed.fsdp.StateDictConfig(offload_to_cpu: bool = False) ``` [#](#api-tensorplay.distributed.fsdp.StateDictSettings) ### StateDictSettings class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.StateDictSettings.html) ```python class tensorplay.distributed.fsdp.StateDictSettings(state_dict_type: tensorplay.distributed.fsdp.api.StateDictType, state_dict_config: tensorplay.distributed.fsdp.api.StateDictConfig, optim_state_dict_config: tensorplay.distributed.fsdp.api.OptimStateDictConfig) ``` [#](#api-tensorplay.distributed.fsdp.StateDictType) ### StateDictType class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.StateDictType.html) ```python class tensorplay.distributed.fsdp.StateDictType(*values) ``` [#](#api-tensorplay.distributed.fsdp.UnshardHandle) ### UnshardHandle class[Full reference ↗](/docs/generated/tensorplay.distributed.fsdp.UnshardHandle.html) ```python class tensorplay.distributed.fsdp.UnshardHandle ```