org.nd4j.linalg.factory.Nd4j.dataType()方法的使用及代码示例

x33g5p2x  于2022-01-24 转载在 其他  
字(5.8k)|赞(0)|评价(0)|浏览(123)

本文整理了Java中org.nd4j.linalg.factory.Nd4j.dataType()方法的一些代码示例，展示了Nd4j.dataType()的具体用法。这些代码示例主要来源于Github/Stackoverflow/Maven等平台，是从一些精选项目中提取出来的代码，具有较强的参考意义，能在一定程度帮忙到你。Nd4j.dataType()方法的具体详情如下：
包路径：org.nd4j.linalg.factory.Nd4j
类名称：Nd4j
方法名：dataType

Nd4j.dataType介绍

[英]Returns the data opType used for the runtime
[中]返回用于运行时的数据类型

代码示例

代码示例来源：origin: deeplearning4j/nd4j

  1. public ArrayDescriptor(int[] array) {
  2.   this.intArray = array;
  3.   this.dtype = DTYPE.INT;
  4.   this.bufferType = Nd4j.dataType();
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public ArrayDescriptor(long[] array) {
  2.   this.longArray = array;
  3.   this.dtype = DTYPE.LONG;
  4.   this.bufferType = Nd4j.dataType();
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public ArrayDescriptor(float[] array) {
  2.   this.floatArray = array;
  3.   this.dtype = DTYPE.FLOAT;
  4.   this.bufferType = Nd4j.dataType();
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Returns the data opType for this ndarray
  3.  *
  4.  * @return the data opType for this ndarray
  5.  */
  6. @Override
  7. public DataBuffer.Type dtype() {
  8.   return Nd4j.dataType();
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public ArrayDescriptor(double[] array) {
  2.   this.doubleArray = array;
  3.   this.dtype = DTYPE.DOUBLE;
  4.   this.bufferType = Nd4j.dataType();
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public Boolean apply(Number input) {
  3.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE)
  4.     return input.doubleValue() != value.doubleValue();
  5.   else
  6.     return input.floatValue() != value.floatValue();
  7. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public Boolean apply(Number input) {
  3.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE)
  4.     return input.doubleValue() == value.doubleValue();
  5.   else
  6.     return input.floatValue() == value.floatValue();
  7. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * This method returns sizeOf(currentDataType), in bytes
  3.  *
  4.  * @return number of bytes per element
  5.  */
  6. public static int sizeOfDataType() {
  7.   return sizeOfDataType(Nd4j.dataType());
  8. }

代码示例来源：origin: deeplearning4j/nd4j

  1. protected DataBuffer.TypeEx getGlobalTypeEx() {
  2.   DataBuffer.Type type = Nd4j.dataType();
  3.   return convertType(type);
  4. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public void init(INDArray x, INDArray y, INDArray z, long n) {
  3.   super.init(x, y, z, n);
  4.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE) {
  5.     this.extraArgs = new Object[] {zeroDouble()};
  6.   } else if (Nd4j.dataType() == DataBuffer.Type.FLOAT) {
  7.     this.extraArgs = new Object[] {zeroFloat()};
  8.   } else if (Nd4j.dataType() == DataBuffer.Type.HALF) {
  9.     this.extraArgs = new Object[] {zeroHalf()};
  10.   }
  11. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Create double based on real and imaginary
  3.  *
  4.  * @param real real component
  5.  * @param imag imag component
  6.  * @return
  7.  */
  8. public static IComplexNumber createComplexNumber(Number real, Number imag) {
  9.   if (dataType() == DataBuffer.Type.FLOAT)
  10.     return INSTANCE.createFloat(real.floatValue(), imag.floatValue());
  11.   return INSTANCE.createDouble(real.doubleValue(), imag.doubleValue());
  12. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray trueScalar(Number value) {
  3.   val dtype = Nd4j.dataType();
  4.   switch (dtype) {
  5.     case DOUBLE:
  6.       return create(new double[] {value.doubleValue()}, new int[] {}, new int[] {}, 0);
  7.     case FLOAT:
  8.       return create(new float[] {value.floatValue()}, new int[] {}, new int[] {}, 0);
  9.     case HALF:
  10.       return create(new float[] {value.floatValue()}, new int[] {}, new int[] {}, 0);
  11.     default:
  12.       throw new UnsupportedOperationException("Unsupported data type: [" + dtype + "]");
  13.   }
  14. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Returns the number of bytes
  3.  * for the graph
  4.  *
  5.  * @return
  6.  */
  7. public long memoryForGraph() {
  8.   return numElements() * DataTypeUtil.lengthForDtype(Nd4j.dataType());
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Create a scalar nd array with the specified value and offset
  3.  *
  4.  * @param value the value of the scalar
  5.  * @return the scalar nd array
  6.  */
  7. @Override
  8. public INDArray scalar(double value) {
  9.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE)
  10.     return create(new double[] {value}, new int[] {1, 1}, new int[] {1, 1}, 0);
  11.   else
  12.     return scalar((float) value);
  13. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static DataBuffer createBufferDetached(float[] data) {
  2.   DataBuffer ret;
  3.   if (dataType() == DataBuffer.Type.FLOAT)
  4.     ret = DATA_BUFFER_FACTORY_INSTANCE.createFloat(data);
  5.   else if (dataType() == DataBuffer.Type.HALF)
  6.     ret = DATA_BUFFER_FACTORY_INSTANCE.createHalf(data);
  7.   else
  8.     ret = DATA_BUFFER_FACTORY_INSTANCE.createDouble(ArrayUtil.toDoubles(data));
  9.   logCreationIfNecessary(ret);
  10.   return ret;
  11. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static DataBuffer createBufferDetached(double[] data) {
  2.   DataBuffer ret;
  3.   if (dataType() == DataBuffer.Type.DOUBLE)
  4.     ret = DATA_BUFFER_FACTORY_INSTANCE.createDouble(data);
  5.   else if (dataType() == DataBuffer.Type.HALF)
  6.     ret = DATA_BUFFER_FACTORY_INSTANCE.createHalf(ArrayUtil.toFloats(data));
  7.   else
  8.     ret = DATA_BUFFER_FACTORY_INSTANCE.createFloat(ArrayUtil.toFloats(data));
  9.   logCreationIfNecessary(ret);
  10.   return ret;
  11. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2.   public long getRequiredBatchMemorySize() {
  3.     long result = maxIntArrays() * maxIntArraySize() * 4;
  4.     result += maxArguments() * 8; // pointers
  5.     result += maxShapes() * 8; // pointers
  6.     result += maxIndexArguments() * 4;
  7.     result += maxRealArguments() * (Nd4j.dataType() == DataBuffer.Type.DOUBLE ? 8
  8.             : Nd4j.dataType() == DataBuffer.Type.FLOAT ? 4 : 2);
  9.     result += 5 * 4; // numArgs
  11.     return result * Batch.getBatchLimit();
  12.   }
  13. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static INDArray toNDArray(int[] nums) {
  2.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE) {
  3.     double[] doubles = ArrayUtil.toDoubles(nums);
  4.     INDArray create = Nd4j.create(doubles, new int[] {1, nums.length});
  5.     return create;
  6.   } else {
  7.     float[] doubles = ArrayUtil.toFloats(nums);
  8.     INDArray create = Nd4j.create(doubles, new int[] {1, nums.length});
  9.     return create;
  10.   }
  11. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static INDArray toNDArray(long[] nums) {
  2.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE) {
  3.     double[] doubles = ArrayUtil.toDoubles(nums);
  4.     INDArray create = Nd4j.create(doubles, new int[] {1, nums.length});
  5.     return create;
  6.   } else {
  7.     float[] doubles = ArrayUtil.toFloats(nums);
  8.     INDArray create = Nd4j.create(doubles, new int[] {1, nums.length});
  9.     return create;
  10.   }
  11. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static INDArray toNDArray(int[][] nums) {
  2.   if (Nd4j.dataType() == DataBuffer.Type.DOUBLE) {
  3.     double[] doubles = ArrayUtil.toDoubles(nums);
  4.     INDArray create = Nd4j.create(doubles, new int[] {nums[0].length, nums.length});
  5.     return create;
  6.   } else {
  7.     float[] doubles = ArrayUtil.toFloats(nums);
  8.     INDArray create = Nd4j.create(doubles, new int[] {nums[0].length, nums.length});
  9.     return create;
  10.   }
  11. }

内容来源于网络，如有侵权，请联系作者删除！

Nd4j

相关文章

    查看更多

    热门标签

    更多
    JavaquerypythonNode开发语言requestUtil数据库Table后端算法LoggerMessageElementParser

    最新文章

    更多

    Nd4j类方法

    • 蜀ICP备13028337号-1 大数据知识库 https://www.saoniuhuo.com © All rights reserved
    • 本站内容来源互联网,如果侵犯您的权益请联系我们删除, 联系方式：448109455@qq.com