Generator

PyTorch provides methods to create random number generation (RNG) as part of your PyTorch program. torch.generator is the primary API to use for standard in the clear application where there is no application of privacy preserving techniques such as differential privacy. To create a cryptographically secure RNG, please use torchcsprng which can be found in the repo here.

class torch.Generator(device='cpu') → Generator

Creates and returns a generator object which manages the state of the algorithm that produces pseudo random numbers. Used as a keyword argument in many In-place random sampling functions.

Parameters

device (torch.device, optional) – the desired device for the generator.

Returns

An torch.Generator object.

Return type

Generator

Example:

>>> g_cpu = torch.Generator()
>>> g_cuda = torch.Generator(device='cuda')

device

Generator.device -> device

Gets the current device of the generator.

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu.device
device(type='cpu')

get_state() → Tensor

Returns the Generator state as a torch.ByteTensor.

Returns

A torch.ByteTensor which contains all the necessary bits to restore a Generator to a specific point in time.

Return type

Tensor

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu.get_state()

initial_seed() → int

Returns the initial seed for generating random numbers.

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu.initial_seed()
2147483647

manual_seed(seed) → Generator

Sets the seed for generating random numbers. Returns a torch.Generator object. It is recommended to set a large seed, i.e. a number that has a good balance of 0 and 1 bits. Avoid having many 0 bits in the seed.

Parameters

seed (int) – The desired seed.

Returns

An torch.Generator object.

Return type

Generator

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu.manual_seed(2147483647)

seed() → int

Gets a non-deterministic random number from std::random_device or the current time and uses it to seed a Generator.

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu.seed()
1516516984916

set_state(new_state) → void

Sets the Generator state.

Parameters

new_state (torch.ByteTensor) – The desired state.

Example:

>>> g_cpu = torch.Generator()
>>> g_cpu_other = torch.Generator()
>>> g_cpu.set_state(g_cpu_other.get_state())