nn.ModuleList[ParamType] can not append add nn.Embedding layer

[mullerhai]

HI,
I want to use one modelList to contain many layer, for Embedding layer ,why can not contain it ?

import torch.*
import torch.nn.modules.{HasParams, TensorModule}


class MultiLayerPerceptron[ParamType <: FloatNN : Default](input_dim: Long, embed_dims: Seq[Int], dropout: Double, output_layer: Boolean = true) extends HasParams[ParamType]
  with TensorModule[ParamType] {

  val layers = nn.ModuleList[ParamType]()
  embed_dims.zipWithIndex.foreach {
    case (embed_dim, index) => {
      val block = register(LinearBnReluDropBlock(input_dim, embed_dim, dropout))
      layers.append(block)
    }
  }
  if (output_layer) {
    var output_linear = register(nn.Linear(input_dim, 1))
    layers.append(output_linear)
  }
  layers.append(nn.Embedding(10, 45))

console log error

MultiLayerPerceptron.scala:24:17
Found:    torch.nn.modules.sparse.Embedding[ParamType]
Required: torch.nn.modules.TensorModule[ParamType]
  layers.append(nn.Embedding(10, 45))

but nn.Embedding layer is TensorModule's sub class ,why linear and conv2d could add ? is cause by the layer class Type parameters ?