syft.generic.object

Module Contents

class syft.generic.object.AbstractObject(id: int = None, owner: sy.workers.AbstractWorker = None, tags: List[str] = None, description: str = None, child=None)

Bases: abc.ABC

This is a generic object abstraction.

is_wrapper = False
__str__(self)
__repr__(self)
describe(self, description: str)
tag(self, *_tags: str)
serialize(self)

Serializes the tensor on which it’s called.

This is the high level convenience function for serializing torch tensors. It includes three steps, Simplify, Serialize, and Compress as described in serde.py. By default serde is compressing using LZ4

Returns

The serialized form of the tensor. For example:

x = torch.Tensor([1,2,3,4,5]) x.serialize() # returns a serialized object

ser(self, *args, **kwargs)
get(self)

Just a pass through. This is most commonly used when calling .get() on a Syft tensor which has a child which is a pointer, an additive shared tensor, a multi-pointer, etc.

mid_get(self)

This method calls .get() on a child pointer and correctly registers the results

get_class_attributes(self)

Return all elements which defines an instance of a certain class. By default there is nothing so we return an empty dict, but for example for fixed precision tensor, the fractional precision is very important.

classmethod on_function_call(cls, *args)

Override this to perform a specific action for each call of a torch function with arguments containing syft tensors of the class doing the overloading

classmethod handle_func_command(cls, command)

Receive an instruction for a function to be applied on a Syft Tensor, Replace in the args all the LogTensors with their child attribute, forward the command instruction to the handle_function_command of the type of the child attributes, get the response and replace a Syft Tensor on top of all tensors found in the response.

Parameters

  • command – instruction of a function command: (command name,

  • self>, arguments[, kwargs]) (<no) –

Returns

the response of the function command

classmethod rgetattr(cls, obj, attr, *args)

Get an attribute recursively.

This is a core piece of functionality for the PySyft tensor chain.

Parameters

  • obj – the object holding the attribute

  • attr – nested attribute

  • args – optional arguments to provide

Returns

the attribute obj.attr

Example

>>> rgetattr(obj, 'attr1.attr2.attr3')
[Out] obj.attr1.attr2.attr3
syft.generic.object.initialize_object(hook, obj, owner=None, reinitialize=True, id=None, init_args=tuple(), init_kwargs={})

Initializes the tensor.

Parameters

  • hook – A reference to TorchHook class.

  • obj – An object to keep track of id, owner and whether it is a native tensor or a wrapper over pytorch.

  • reinitialize – A boolean parameter (default True) to indicate whether to re-execute __init__.

  • owner – The owner of the tensor being initialised, leave it blank to if you have already provided a reference to TorchHook class.

  • id – The id of tensor, a random id will be generated if there is no id specified.

syft.generic.object._apply_args(hook, obj_to_register, owner=None, id=None)