Question

我正在尝试将从Tensorflow模型保存的权重导入PyTorch。到目前为止，结果非常相似。当模型用conv2d调用stride=2时，我遇到了麻烦。

为验证不匹配，我在TF和PyTorch之间进行了非常简单的比较。首先，我将conv2d与stride=1进行比较。

import tensorflow as tf
import numpy as np
import torch
import torch.nn.functional as F


np.random.seed(0)
sess = tf.Session()

# Create random weights and input
weights = torch.empty(3, 3, 3, 8)
torch.nn.init.constant_(weights, 5e-2)
x = np.random.randn(1, 3, 10, 10)

weights_tf = tf.convert_to_tensor(weights.numpy(), dtype=tf.float32)
# PyTorch adopts [outputC, inputC, kH, kW]
weights_torch = torch.Tensor(weights.permute((3, 2, 0, 1)))

# Tensorflow defaults to NHWC
x_tf = tf.convert_to_tensor(x.transpose((0, 2, 3, 1)), dtype=tf.float32)
x_torch = torch.Tensor(x)

# TF Conv2D
tf_conv2d = tf.nn.conv2d(x_tf,
                         weights_tf,
                         strides=[1, 1, 1, 1],
                         padding="SAME")

# PyTorch Conv2D
torch_conv2d = F.conv2d(x_torch, weights_torch, padding=1, stride=1)

sess.run(tf.global_variables_initializer())
tf_result = sess.run(tf_conv2d)

diff = np.mean(np.abs(tf_result.transpose((0, 3, 1, 2)) - torch_conv2d.detach().numpy()))
print('Mean of Abs Diff: {0}'.format(diff))

此执行的结果是：

Mean of Abs Diff: 2.0443112092038973e-08

当我将stride更改为2时，结果开始有所不同。

# TF Conv2D
tf_conv2d = tf.nn.conv2d(x_tf,
                         weights_tf,
                         strides=[1, 2, 2, 1],
                         padding="SAME")

# PyTorch Conv2D
torch_conv2d = F.conv2d(x_torch, weights_torch, padding=1, stride=2)

此执行的结果是：

Mean of Abs Diff: 0.2104552686214447

根据PyTorch文档，padding参数定义了conv2d uses zero-padding。因此，在我的示例中，将零添加到输入的左，上，右和下。

如果PyTorch仅基于输入参数在两侧添加填充，则应该易于在Tensorflow中复制。

# Manually add padding - consistent with PyTorch
paddings = tf.constant([[0, 0], [1, 1], [1, 1], [0, 0]])
x_tf = tf.convert_to_tensor(x.transpose((0, 2, 3, 1)), dtype=tf.float32)
x_tf = tf.pad(x_tf, paddings, "CONSTANT")

# TF Conv2D
tf_conv2d = tf.nn.conv2d(x_tf,
                         weights_tf,
                         strides=[1, 2, 2, 1],
                         padding="VALID")

此比较的结果是：

Mean of Abs Diff: 1.6035047067930464e-08

这告诉我，如果我能够以某种方式将Tensorflow中的默认填充行为复制到PyTorch中，那么我的结果将是相似的。

This question检查了Tensorflow中的填充行为。 TF documentation explains how padding is added for "SAME" convolutions.我在写这个问题时发现了这些链接。

现在我知道Tensorflow的填充策略，我可以在PyTorch中实现它。

Answer 1

要复制行为，请按照Tensorflow文档中的说明计算填充大小。在这里，我通过设置stride=2并填充PyTorch输入来测试填充行为。

import tensorflow as tf
import numpy as np
import torch
import torch.nn.functional as F


np.random.seed(0)
sess = tf.Session()

# Create random weights and input
weights = torch.empty(3, 3, 3, 8)
torch.nn.init.constant_(weights, 5e-2)
x = np.random.randn(1, 3, 10, 10)

weights_tf = tf.convert_to_tensor(weights.numpy(), dtype=tf.float32)
weights_torch = torch.Tensor(weights.permute((3, 2, 0, 1)))

# Tensorflow padding behavior. Assuming that kH == kW to keep this simple.
stride = 2
if x.shape[2] % stride == 0:
    pad = max(weights.shape[0] - stride, 0)
else:
    pad = max(weights.shape[0] - (x.shape[2] % stride), 0)

if pad % 2 == 0:
    pad_val = pad // 2
    padding = (pad_val, pad_val, pad_val, pad_val)
else:
    pad_val_start = pad // 2
    pad_val_end = pad - pad_val_start
    padding = (pad_val_start, pad_val_end, pad_val_start, pad_val_end)

x_tf = tf.convert_to_tensor(x.transpose((0, 2, 3, 1)), dtype=tf.float32)
x_torch = torch.Tensor(x)
x_torch = F.pad(x_torch, padding, "constant", 0)

# TF Conv2D
tf_conv2d = tf.nn.conv2d(x_tf,
                         weights_tf,
                         strides=[1, stride, stride, 1],
                         padding="SAME")

# PyTorch Conv2D
torch_conv2d = F.conv2d(x_torch, weights_torch, padding=0, stride=stride)

sess.run(tf.global_variables_initializer())
tf_result = sess.run(tf_conv2d)

diff = np.mean(np.abs(tf_result.transpose((0, 3, 1, 2)) - torch_conv2d.detach().numpy()))
print('Mean of Abs Diff: {0}'.format(diff))

输出为：

Mean of Abs Diff: 2.2477470551507395e-08

当我开始写这个问题时，我不太确定为什么会这样，但是一些阅读很快就澄清了这一点。我希望这个例子可以帮助其他人。

将Conv2D与Tensorflow和PyTorch之间的填充进行比较

1 个答案: