mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-21 05:34:18 +08:00
e7c1e6a8e3
Here's the command I used to invoke autopep8 (in parallel!):
git ls-files | grep '\.py$' | xargs -n1 -P`nproc` autopep8 -i
Several rules are ignored in setup.cfg. The goal is to let autopep8
handle everything which it can handle safely, and to disable any rules
which are tricky or controversial to address. We may want to come back
and re-enable some of these rules later, but I'm trying to make this
patch as safe as possible.
Also configures flake8 to match pep8's behavior.
Also configures TravisCI to check the whole project for lint.
88 lines
2.7 KiB
Python
88 lines
2.7 KiB
Python
import math
|
|
import torch
|
|
from .Module import Module
|
|
from .utils import clear
|
|
|
|
|
|
class Linear(Module):
|
|
|
|
def __init__(self, inputSize, outputSize, bias=True):
|
|
super(Linear, self).__init__()
|
|
self.weight = torch.Tensor(outputSize, inputSize)
|
|
self.gradWeight = torch.Tensor(outputSize, inputSize)
|
|
self.bias = torch.Tensor(outputSize) if bias else None
|
|
self.gradBias = torch.Tensor(outputSize) if bias else None
|
|
self.reset()
|
|
|
|
self.addBuffer = None
|
|
|
|
def noBias(self):
|
|
self.bias = None
|
|
self.gradBias = None
|
|
return self
|
|
|
|
def reset(self, stdv=None):
|
|
if stdv is not None:
|
|
stdv = stdv * math.sqrt(3)
|
|
else:
|
|
stdv = 1. / math.sqrt(self.weight.size(1))
|
|
|
|
self.weight.uniform_(-stdv, stdv)
|
|
if self.bias is not None:
|
|
self.bias.uniform_(-stdv, stdv)
|
|
|
|
return self
|
|
|
|
def _updateAddBuffer(self, input):
|
|
nframe = input.size(0)
|
|
if self.addBuffer is None:
|
|
self.addBuffer = input.new()
|
|
if self.addBuffer.nelement() != nframe:
|
|
self.addBuffer.resize_(nframe).fill_(1)
|
|
|
|
def updateOutput(self, input):
|
|
assert input.dim() == 2
|
|
nframe = input.size(0)
|
|
nelement = self.output.nelement()
|
|
self.output.resize_(nframe, self.weight.size(0))
|
|
if self.output.nelement() != nelement:
|
|
self.output.zero_()
|
|
|
|
self._updateAddBuffer(input)
|
|
self.output.addmm_(0, 1, input, self.weight.t())
|
|
if self.bias is not None:
|
|
self.output.addr_(self.addBuffer, self.bias)
|
|
|
|
return self.output
|
|
|
|
def updateGradInput(self, input, gradOutput):
|
|
if self.gradInput is None:
|
|
return
|
|
|
|
nelement = self.gradInput.nelement()
|
|
self.gradInput.resize_as_(input)
|
|
if self.gradInput.nelement() != nelement:
|
|
self.gradInput.zero_()
|
|
|
|
assert input.dim() == 2
|
|
self.gradInput.addmm_(0, 1, gradOutput, self.weight)
|
|
|
|
return self.gradInput
|
|
|
|
def accGradParameters(self, input, gradOutput, scale=1):
|
|
assert input.dim() == 2
|
|
self.gradWeight.addmm_(scale, gradOutput.t(), input)
|
|
if self.bias is not None:
|
|
# update the size of addBuffer if the input is not the same size as the one we had in last updateGradInput
|
|
self._updateAddBuffer(input)
|
|
self.gradBias.addmv_(scale, gradOutput.t(), self.addBuffer)
|
|
|
|
def clearState(self):
|
|
clear(self, 'addBuffer')
|
|
return super(Linear, self).clearState()
|
|
|
|
def __repr__(self):
|
|
return super(Linear, self).__repr__() + \
|
|
'({} -> {})'.format(self.weight.size(1), self.weight.size(0)) + \
|
|
(' without bias' if self.bias is None else '')
|