14303300e332c3be5d669789f3aa736befa22575,gluon/gluoncv2/models/resnesta.py,ResNeStA,__init__,#ResNeStA#Any#Any#Any#Any#Any#Any#Any#Any#,276

Before Change

        self.in_size = in_size
        self.classes = classes

        with self.name_scope():
            self.features = nn.HybridSequential(prefix="")
            self.features.add(SEInitBlock(
                in_channels=in_channels,
                out_channels=init_block_channels,
                bn_use_global_stats=bn_use_global_stats,
                bn_cudnn_off=bn_cudnn_off))
            in_channels = init_block_channels
            for i, channels_per_stage in enumerate(channels):
                stage = nn.HybridSequential(prefix="stage{}_".format(i + 1))
                with stage.name_scope():
                    for j, out_channels in enumerate(channels_per_stage):
                        strides = 2 if (j == 0) and (i != 0) else 1
                        stage.add(ResNeStAUnit(
                            in_channels=in_channels,
                            out_channels=out_channels,
                            strides=strides,
                            bn_use_global_stats=bn_use_global_stats,
                            bn_cudnn_off=bn_cudnn_off,
                            bottleneck=bottleneck))
                        in_channels = out_channels
                self.features.add(stage)
            self.features.add(nn.GlobalAvgPool2D())

            self.output = nn.HybridSequential(prefix="")
            self.output.add(nn.Flatten())
            self.output.add(nn.Dense(
                units=classes,
                in_units=in_channels))

    def hybrid_forward(self, F, x):
        x = self.features(x)
        x = self.output(x)
        return x

After Change

    bn_use_global_stats : bool, default False
        Whether global moving statistics is used instead of local batch-norm for BatchNorm layers.
        Useful for fine-tuning.
    bn_cudnn_off : bool, default False
        Whether to disable CUDNN batch normalization operator.
    in_channels : int, default 3
        Number of input channels.
    in_size : tuple of two ints, default (224, 224)
        Spatial size of the expected input image.
    classes : int, default 1000
        Number of classification classes.
    
    def __init__(self,
                 channels,
                 init_block_channels,
                 bottleneck,
                 dropout_rate=0.0,
                 bn_use_global_stats=False,
                 bn_cudnn_off=False,
                 in_channels=3,

Instances