aihwkit.nn.functions module

Autograd functions for aihwkit.

class aihwkit.nn.functions.AnalogFunction(*args, **kwargs)[source]

Bases: aihwkit.nn.functions.AnalogFunctionBase

Function that delegates into a RPU unit.

static forward(ctx, analog_ctx, input_, shared_weights=None, is_test=False)[source]

Execute the forward pass in the analog tile.

Parameters
Return type

torch.Tensor

class aihwkit.nn.functions.AnalogFunctionBase(*args, **kwargs)[source]

Bases: torch.autograd.function.Function

Base function for analog functions.

static backward(ctx, grad_output)[source]

Execute the backward pass in the analog tile.

Parameters

  • ctx (Any) –

  • grad_output (torch.Tensor) –

Return type

Tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor]]

static forward(ctx, analog_ctx, input_, shared_weights=None, is_test=False)[source]

Execute the forward pass in the analog tile.

Note: Indexed versions can used when analog_ctx.use_indexed is set to True.

Parameters
Return type

torch.Tensor

class aihwkit.nn.functions.AnalogIndexedFunction(*args, **kwargs)[source]

Bases: aihwkit.nn.functions.AnalogFunctionBase

Function that delegates into a RPU unit to use the indexed forward/backward/update.

static forward(ctx, analog_ctx, input_, shared_weights=None, is_test=False)[source]

Execute the forward pass in the analog tile.

Parameters
Return type

torch.Tensor

aihwkit.nn.functions.empty_like(input, *, dtype=None, layout=None, device=None, requires_grad=False, memory_format=torch.preserve_format) → Tensor

Returns an uninitialized tensor with the same size as input. torch.empty_like(input) is equivalent to torch.empty(input.size(), dtype=input.dtype, layout=input.layout, device=input.device).

Parameters

input (Tensor) – the size of input will determine size of the output tensor.

Keyword Arguments

  • dtype (torch.dtype, optional) – the desired data type of returned Tensor. Default: if None, defaults to the dtype of input.

  • layout (torch.layout, optional) – the desired layout of returned tensor. Default: if None, defaults to the layout of input.

  • device (torch.device, optional) – the desired device of returned tensor. Default: if None, defaults to the device of input.

  • requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default: False.

  • memory_format (torch.memory_format, optional) – the desired memory format of returned Tensor. Default: torch.preserve_format.

Example:

>>> torch.empty((2,3), dtype=torch.int64)
tensor([[ 9.4064e+13,  2.8000e+01,  9.3493e+13],
        [ 7.5751e+18,  7.1428e+18,  7.5955e+18]])