How to convert DenseNDArray to NDArray ?

HI , when I read TfRecordDataset from tfrecord file , I get the data type is DenseNDArray ,actual is DenseTString, I want to convert it to TString data.

my code -scala version

    val path = "/Volumes/Pink4T/transfer/code/github/stanford-tensorflow-tutorials/2017/data/friday.tfrecord"

    val dataset =  Dataset.tfRecordDataset(tf,path,"",0l).batch(10) // readData(path) //
//    DenseNdArray
    var count =0
    for (outputs <- dataset) {
      try {
        val batch1 = outputs.get(count).asTensor.asInstanceOf[TString]
        try {

          val byteNdArray:NdArray[Array[Byte]]= batch1.asBytes().asInstanceOf[NdArray[Array[Byte]]] //ByteNdArray]
         val bda = byteNdArray.asInstanceOf[DenseNdArray[TString]] // here !!! how to get   SparseNdArray?
          println(s"bda shape ${bda.shape()} ${kk.shape()}")

          count = count + 1
        } finally {
          if (batch1 != null) batch1.close()

        }
      }
    }

The report centers on Dataset.tfRecordDataset and the DenseNDArray, NDArray, and TString types in the Scala example. Start by tracing those APIs and the TensorFlow Java tensor-conversion documentation. Done means identifying and documenting a supported conversion path, or clearly recording that the requested conversion is unavailable.