import torch.nn as nn
import torch
import numpy as np
[文档]class VoltageHook(nn.Module):
def __init__(self, scale=1.0, momentum=0.1, mode='Max'):
"""
* :ref:`API in English <VoltageHook.__init__-en>`
.. _voltageHook.__init__-cn:
:param scale: 缩放初始值
:type scale: float
:param momentum: 动量值
:type momentum: float
:param mode: 模式。输入“Max”表示记录ANN激活最大值，“99.9%”表示记录ANN激活的99.9%分位点，输入0-1的float型浮点数表示记录激活最大值的对应倍数。
:type mode: str, float
``VoltageHook`` 被置于ReLU后，用于在ANN推理中确定激活的范围。
* :ref:`中文API <VoltageHook.__init__-cn>`
.. _voltageHook.__init__-en:
:param scale: initial scaling value
:type scale: float
:param momentum: momentum value
:type momentum: float
:param mode: The mode. Value "Max" means recording the maximum value of ANN activation, "99.9%" means recording the 99.9% precentile of ANN activation, and a float of 0-1 means recording the corresponding multiple of the maximum activation value.
:type mode: str, float
``VoltageHook`` is placed behind ReLU and used to determine the range of activations in ANN inference.
"""
super().__init__()
self.register_buffer('scale', torch.tensor(scale))
self.mode = mode
self.num_batches_tracked = 0
self.momentum = momentum
[文档] def forward(self, x):
"""
* :ref:`API in English <VoltageHook.forward-en>`
.. _VoltageHook.forward-cn:
:param x: 输入张量
:type x: torch.Tensor
:return: 原输入张量
:rtype: torch.Tensor
不对输入张量做任何处理，只是抓取ReLU的激活值
* :ref:`中文API <VoltageHook.forward-cn>`
.. _VoltageHook.forward-en:
:param x: input tensor
:type x: torch.Tensor
:return: original input tensor
:rtype: torch.Tensor
It doesn't process input tensors, but hooks the activation values of ReLU.
"""
err_msg = 'You have used a non-defined VoltageScale Method.'
if isinstance(self.mode, str):
if self.mode[-1] == '%':
try:
s_t = torch.tensor(np.percentile(x.detach().cpu(), float(self.mode[:-1])))
except ValueError:
raise NotImplementedError(err_msg)
elif self.mode.lower() in ['max']:
s_t = x.max().detach()
else:
raise NotImplementedError(err_msg)
elif isinstance(self.mode, float) and self.mode <= 1 and self.mode > 0:
s_t = x.max().detach() * self.mode
else:
raise NotImplementedError(err_msg)
if self.num_batches_tracked == 0:
self.scale = s_t
else:
self.scale = (1 - self.momentum) * self.scale + self.momentum * s_t
self.num_batches_tracked += x.shape[0]
return x
[文档]class VoltageScaler(nn.Module):
def __init__(self, scale=1.0):
"""
* :ref:`API in English <VoltageScaler.__init__-en>`
.. _VoltageScaler.__init__-cn:
:param scale: 缩放值
:type scale: float
``VoltageScaler`` 用于SNN推理中缩放电流。
* :ref:`中文API <VoltageScaler.__init__-cn>`
.. _VoltageScaler.__init__-en:
:param scale: scaling value
:type scale: float
``VoltageScaler`` is used for scaling current in SNN inference.
"""
super().__init__()
self.register_buffer('scale', torch.tensor(scale))
[文档] def forward(self, x):
"""
* :ref:`API in English <VoltageScaler.forward-en>`
.. _VoltageScaler.forward-cn:
:param x: 输入张量，亦即输入电流
:type x: torch.Tensor
:return: 缩放后的电流
:rtype: torch.Tensor
* :ref:`中文API <VoltageScaler.forward-cn>`
.. _VoltageScaler.forward-en:
:param x: input tensor, or input current
:type x: torch.Tensor
:return: current after scaling
:rtype: torch.Tensor
"""
return x * self.scale