XJDKC commented on a change in pull request #697:
URL: https://github.com/apache/singa/pull/697#discussion_r433577123
##########
File path: examples/mlp/module.py
##########
@@ -56,10 +56,9 @@ def forward(self, inputs):
x = autograd.add_bias(x, self.b1)
return x
- def loss(self, out, ty):
- return autograd.softmax_cross_entropy(out, ty)
-
- def optim(self, loss, dist_option, spars):
+ def train_one_batch(self, x, y, dist_option, spars):
+ out = self.forward(x)
+ loss = autograd.softmax_cross_entropy(out, y)
Review comment:
Yes. In this way, users only needs to use layers to define their model.
##########
File path: examples/cnn/model/resnet.py
##########
@@ -21,30 +21,31 @@
# https://github.com/pytorch/vision/blob/master/torchvision/models/resnet.py
from singa import autograd
-from singa import module
+from singa import layer
+from singa import model
def conv3x3(in_planes, out_planes, stride=1):
"""3x3 convolution with padding"""
- return autograd.Conv2d(
+ return layer.Conv2d(
in_planes,
out_planes,
- kernel_size=3,
+ 3,
stride=stride,
padding=1,
bias=False,
)
-class BasicBlock(autograd.Layer):
+class BasicBlock(layer.Layer):
expansion = 1
def __init__(self, inplanes, planes, stride=1, downsample=None):
super(BasicBlock, self).__init__()
self.conv1 = conv3x3(inplanes, planes, stride)
- self.bn1 = autograd.BatchNorm2d(planes)
+ self.bn1 = layer.BatchNorm2d(planes)
Review comment:
Are there other operators besides ReLU that need to be encapsulated by
Layer?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
