torch.nn.BatchNorm1d转换问题

torch.nn.BatchNorm1d转换时会丢失track_running_stats，demo如下。

paconvert版本：3.0.1

原Torch代码：

import torch

in_channels = 10
eps = 1e-5
momentum = 0.1
affine = True
track_running_stats = True

net = torch.nn.BatchNorm1d(in_channels, eps, momentum, affine,
                                           track_running_stats)

转换后的Paddle代码：

import paddle

in_channels = 10
eps = 1e-05
momentum = 0.1
affine = True
track_running_stats = True
net = paddle.nn.BatchNorm1D(
    num_features=in_channels,
    epsilon=eps,
    weight_attr=affine,
    bias_attr=affine,
    momentum=1 - momentum,
)

从转换结果看，转换后的代码丢失了track_running_stats参数。

Start by reproducing the shown torch.nn.BatchNorm1d conversion with PaConvert 3.0.1, then locate the conversion rule handling BatchNorm1d and its regression tests. Done means the generated Paddle code preserves the source track_running_stats behavior and a test covers this example.