javacpp- pytorch transformerimpl lack some parameter between with python pytorch and cpp libtorch

[mullerhai]

Hi,
I find javacpp-pytorch some layer impl not the same as python pytorch and cpp libtorch ,like transformer layer

https://pytorch.org/docs/stable/generated/torch.nn.Transformer.html

Transformer
CLASStorch.nn.Transformer(d_model=512, nhead=8, num_encoder_layers=6, num_decoder_layers=6, dim_feedforward=2048, dropout=0.1, activation=<function relu>, custom_encoder=None, custom_decoder=None, layer_norm_eps=1e-05, batch_first=False, norm_first=False, bias=True, device=None, dtype=None)

but javacpp TransformerImpl lack [layer_norm_eps=1e-05, batch_first=False, norm_first=False, ] three parameter not supply ,so how to pass these parameter in javacpp

  public TransformerOptions(
        @Cast("int64_t") long d_model,
        @Cast("int64_t") long nhead,
        @Cast("int64_t") long num_encoder_layers,
        @Cast("int64_t") long num_decoder_layers) { super((Pointer)null); allocate(d_model, nhead, num_encoder_layers, num_decoder_layers); }
  private native void allocate(
        @Cast("int64_t") long d_model,
        @Cast("int64_t") long nhead,
        @Cast("int64_t") long num_encoder_layers,
        @Cast("int64_t") long num_decoder_layers);
  public native @Cast("int64_t*") @ByRef @NoException(true) LongPointer d_model();
  public native @Cast("int64_t*") @ByRef @NoException(true) LongPointer nhead();
  public native @Cast("int64_t*") @ByRef @NoException(true) LongPointer num_encoder_layers();
  public native @Cast("int64_t*") @ByRef @NoException(true) LongPointer num_decoder_layers();
  public native @Cast("int64_t*") @ByRef @NoException(true) LongPointer dim_feedforward();
  public native @ByRef @NoException(true) DoublePointer dropout();
  public native @ByRef @NoException(true) TransformerActivation activation();
  public native @ByRef @NoException(true) AnyModule custom_encoder();
  public native @ByRef @NoException(true) AnyModule custom_decoder();

class Transformer dModel:Long =512 ,nhead:Long = 8, numEncoderLayers:Long=6,numDecoderLayers:Long=6,
                                                                     dimFeedforward :Long=2048,dropout:Double=0.1,
                                                                    customEncoder:Option[AnyModule] = None,
                                                                    customDecoder:Option[AnyModule] = None,
                                                                    activation: TransformerActivation ,bias: Boolean = true