How to solve this, MXNetError : Shape inconsistent

[xff0707]

In GAN,My GeneratorBlock class structure like this

GeneratorBlock(1, 20) {
	dense1(1, 20) -> (1, 100)
	bn1(1, 100) -> (1, 100)
	relu1(1, 100) -> (1, 100)
	reshape1(1, 100) -> (1, 100, 1, 1)
	deconv1(1, 100, 1, 1) -> (1, 512, 8, 8)
	bn2(1, 512, 8, 8) -> (1, 512, 8, 8)
	relu2(1, 512, 8, 8) -> (1, 512, 8, 8)
	deconv2(1, 512, 8, 8) -> (1, 256, 16, 16)
	bn3(1, 256, 16, 16) -> (1, 256, 16, 16)
	relu3(1, 256, 16, 16) -> (1, 256, 16, 16)
	deconv3(1, 256, 16, 16) -> (1, 128, 32, 32)
	bn4(1, 128, 32, 32) -> (1, 128, 32, 32)
	relu4(1, 128, 32, 32) -> (1, 128, 32, 32)
	deconv4(1, 128, 32, 32) -> (1, 3, 64, 64)
	tanh1(1, 3, 64, 64) -> (1, 3, 64, 64)
} -> (1, 3, 64, 64)

But when I use NDArray z to predict the result ,it will throws Exceotion.
The code is

NDArray z = manager.randomNormal(-1f, 1f, new Shape(1, 20), DataType.FLOAT32);
Translator<NDList, NDList> translator = new NoopTranslator();
Predictor<NDList, NDList> predictor = gModel.newPredictor(translator);
NDArray result = (predictor.predict(new NDList(z))).singletonOrThrow();//Exception There

The Exception is

Exception in thread "main" ai.djl.translate.TranslateException: ai.djl.engine.EngineException: MXNet engine call failed: MXNetError: Shape inconsistent, Provided = [512,100,8,8], inferred shape=(100,512,8,8)

at ai.djl.inference.Predictor.batchPredict(Predictor.java:189)
at ai.djl.inference.Predictor.predict(Predictor.java:126)
at org.xff.gan.Demo1.main(Demo1.java:92)

Caused by: ai.djl.engine.EngineException: MXNet engine call failed: MXNetError: Shape inconsistent, Provided = [512,100,8,8], inferred shape=(100,512,8,8)

at ai.djl.mxnet.jna.JnaUtils.checkCall(JnaUtils.java:1942)
at ai.djl.mxnet.jna.JnaUtils.imperativeInvoke(JnaUtils.java:521)
at ai.djl.mxnet.jna.FunctionInfo.invoke(FunctionInfo.java:75)
at ai.djl.mxnet.engine.MxNDManager.invoke(MxNDManager.java:345)
at ai.djl.mxnet.engine.MxNDArrayEx.deconvolution(MxNDArrayEx.java:585)
at ai.djl.nn.convolutional.Deconvolution.deconvolution(Deconvolution.java:197)
at ai.djl.nn.convolutional.Deconvolution.forwardInternal(Deconvolution.java:123)
at ai.djl.nn.AbstractBaseBlock.forward(AbstractBaseBlock.java:77)
at ai.djl.nn.Block.forward(Block.java:123)
at org.xff.gan.GeneratorBlock.forwardInternal(GeneratorBlock.java:77)
at ai.djl.nn.AbstractBaseBlock.forward(AbstractBaseBlock.java:77)
at ai.djl.nn.Block.forward(Block.java:123)
at ai.djl.inference.Predictor.predictInternal(Predictor.java:140)
at ai.djl.inference.Predictor.batchPredict(Predictor.java:167)
... 2 more

The location in class GeneratorBlock is

@Override
protected NDList forwardInternal(ParameterStore parameterStore, NDList ndList, boolean b, PairList<String, Object> pairList) {
    NDList current = ndList;
    current = linear1.forward(parameterStore, current, b);
    current = batchNorm1.forward(parameterStore, current, b);
    current = relu1.forward(parameterStore,current,b);
    current = reshape1.forward(parameterStore, current, b);

    current = conv1.forward(parameterStore, current, b);//Exception There
    current = batchNorm2.forward(parameterStore, current, b);
    current = relu2.forward(parameterStore,current,b);

    current = conv2.forward(parameterStore, current, b);
    current = batchNorm3.forward(parameterStore, current, b);
    current = relu3.forward(parameterStore,current,b);

    current = conv3.forward(parameterStore, current, b);
    current = batchNorm4.forward(parameterStore, current, b);
    current = relu4.forward(parameterStore,current,b);

    current = conv4.forward(parameterStore, current, b);
    current = tanh1.forward(parameterStore,current,b);
    return current;
}

and the conv1 is

Block conv1 = addChildBlock("deconv1", Conv2dTranspose.builder().setKernelShape(new Shape(8,8)).setFilters(512).build());

My block at conv1 is (1, 100, 1, 1) -> (1, 512, 8, 8),but why it become
Provided = [512,100,8,8], inferred shape=(100,512,8,8)
I debug at that line ,The var current is NDList with only one NDArray is (1, 100, 1, 1)，But error at conv1.