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

x33g5p2x  于2022-01-24 转载在 其他  
字(7.0k)|赞(0)|评价(0)|浏览(156)

本文整理了Java中org.nd4j.linalg.factory.Nd4j.getExecutioner()方法的一些代码示例，展示了Nd4j.getExecutioner()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.getExecutioner()方法的具体详情如下：
包路径：org.nd4j.linalg.factory.Nd4j
类名称：Nd4j
方法名：getExecutioner

Nd4j.getExecutioner介绍

[英]Get the operation executioner instance
[中]获取操作执行器实例

代码示例

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray nextFloat(char order, int[] shape) {
  3.   INDArray array = Nd4j.createUninitialized(shape, order);
  4.   UniformDistribution op = new UniformDistribution(array, 0.0, 1.0);
  5.   Nd4j.getExecutioner().exec(op, this);
  6.   return array;
  7. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Sin function
  3.  * @param in
  4.  * @param copy
  5.  * @return
  6.  */
  7. public static INDArray sin(INDArray in, boolean copy) {
  8.   return Nd4j.getExecutioner().execAndReturn(new Sin((copy ? in.dup() : in)));
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray getActivation(INDArray in, boolean training) {
  3.   Nd4j.getExecutioner().execAndReturn(new SoftSign(in));
  4.   return in;
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Returns non-normalized Shannon entropy along dimension
  3.  * @param dimension
  4.  * @return
  5.  */
  6. @Override
  7. public INDArray shannonEntropy(int... dimension) {
  8.   return Nd4j.getExecutioner().exec(new ShannonEntropy(this), dimension);
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static void checkForInf(INDArray z) {
  2.   if (Nd4j.getExecutioner().getProfilingMode() != OpExecutioner.ProfilingMode.INF_PANIC
  3.           && Nd4j.getExecutioner().getProfilingMode() != OpExecutioner.ProfilingMode.ANY_PANIC)
  4.     return;
  5.   int match = 0;
  6.   if (!z.isScalar()) {
  7.     MatchCondition condition = new MatchCondition(z, Conditions.isInfinite());
  8.     match = Nd4j.getExecutioner().exec(condition, Integer.MAX_VALUE).getInt(0);
  9.   } else {
  10.     if (z.data().dataType() == DataBuffer.Type.DOUBLE) {
  11.       if (Double.isInfinite(z.getDouble(0)))
  12.         match = 1;
  13.     } else {
  14.       if (Float.isInfinite(z.getFloat(0)))
  15.         match = 1;
  16.     }
  17.   }
  18.   if (match > 0)
  19.     throw new ND4JIllegalStateException("P.A.N.I.C.! Op.Z() contains " + match + " Inf value(s)");
  20. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray compress(INDArray array) {
  3.   INDArray dup = array.dup(array.ordering());
  4.   Nd4j.getExecutioner().commit();
  5.   dup.setData(compress(dup.data()));
  6.   dup.markAsCompressed(true);
  7.   return dup;
  8. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Log on arbitrary base
  3.  *
  4.  * @param ndArray
  5.  * @param base
  6.  * @return
  7.  */
  8. public static INDArray log(INDArray ndArray, double base, boolean duplicate) {
  9.   return Nd4j.getExecutioner().exec(new LogX(duplicate ? ndArray.dup(ndArray.ordering()) : ndArray, base)).z();
  10. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public int iamax(long n, INDArray arr, int stride) {
  3.   if (Nd4j.getExecutioner().getProfilingMode() == OpExecutioner.ProfilingMode.ALL)
  4.     OpProfiler.getInstance().processBlasCall(false, arr);
  5.   if (arr.data().dataType() == DataBuffer.Type.DOUBLE) {
  6.     DefaultOpExecutioner.validateDataType(DataBuffer.Type.DOUBLE, arr);
  7.     return idamax(n, arr, stride);
  8.   } else {
  9.     DefaultOpExecutioner.validateDataType(DataBuffer.Type.FLOAT, arr);
  10.     return isamax(n, arr, stride);
  11.   }
  12. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Atan2 operation, new INDArray instance will be returned
  3.  * Note the order of x and y parameters is opposite to that of java.lang.Math.atan2
  4.  *
  5.  * @param x the abscissa coordinate
  6.  * @param y the ordinate coordinate
  7.  * @return the theta from point (r, theta) when converting (x,y) from to cartesian to polar coordinates
  8.  */
  9. public static INDArray atan2(@NonNull INDArray x, @NonNull INDArray y) {
  10.   return Nd4j.getExecutioner()
  11.           .execAndReturn(new OldAtan2Op(x, y, Nd4j.createUninitialized(x.shape(), x.ordering())));
  12. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Sin function
  3.  * @param in
  4.  * @param copy
  5.  * @return
  6.  */
  7. public static INDArray atanh(INDArray in, boolean copy) {
  8.   return Nd4j.getExecutioner().execAndReturn(new ATanh((copy ? in.dup() : in)));
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Negate each element (in-place).
  3.  */
  4. @Override
  5. public INDArray negi() {
  6.   Nd4j.getExecutioner().exec(new Negative(this));
  7.   return this;
  8. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray nextGaussian(char order, int[] shape) {
  3.   INDArray array = Nd4j.createUninitialized(shape, order);
  4.   GaussianDistribution op = new GaussianDistribution(array, 0.0, 1.0);
  5.   Nd4j.getExecutioner().exec(op, this);
  6.   return array;
  7. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray getActivation(INDArray in, boolean training) {
  3.   Nd4j.getExecutioner().execAndReturn(new RectifiedTanh(in));
  4.   return in;
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. public static void checkForNaN(INDArray z) {
  2.   if (Nd4j.getExecutioner().getProfilingMode() != OpExecutioner.ProfilingMode.NAN_PANIC
  3.           && Nd4j.getExecutioner().getProfilingMode() != OpExecutioner.ProfilingMode.ANY_PANIC)
  4.     return;
  5.   int match = 0;
  6.   if (!z.isScalar()) {
  7.     MatchCondition condition = new MatchCondition(z, Conditions.isNan());
  8.     match = Nd4j.getExecutioner().exec(condition, Integer.MAX_VALUE).getInt(0);
  9.   } else {
  10.     if (z.data().dataType() == DataBuffer.Type.DOUBLE) {
  11.       if (Double.isNaN(z.getDouble(0)))
  12.         match = 1;
  13.     } else {
  14.       if (Float.isNaN(z.getFloat(0)))
  15.         match = 1;
  16.     }
  17.   }
  18.   if (match > 0)
  19.     throw new ND4JIllegalStateException("P.A.N.I.C.! Op.Z() contains " + match + " NaN value(s): ");
  20. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * swaps a vector with another vector.
  3.  *
  4.  * @param x
  5.  * @param y
  6.  */
  7. @Override
  8. public void copy(INDArray x, INDArray y) {
  9.   if (Nd4j.getExecutioner().getProfilingMode() == OpExecutioner.ProfilingMode.ALL)
  10.     OpProfiler.getInstance().processBlasCall(false, x, y);
  11.   if (x.isSparse() || y.isSparse()) {
  12.     Nd4j.getSparseBlasWrapper().level1().copy(x, y);
  13.     return;
  14.   }
  15.   if (x.data().dataType() == DataBuffer.Type.DOUBLE) {
  16.     DefaultOpExecutioner.validateDataType(DataBuffer.Type.DOUBLE, x, y);
  17.     dcopy(x.length(), x, BlasBufferUtil.getBlasStride(x), y, BlasBufferUtil.getBlasStride(y));
  18.   } else {
  19.     DefaultOpExecutioner.validateDataType(DataBuffer.Type.FLOAT, x, y);
  20.     scopy(x.length(), x, BlasBufferUtil.getBlasStride(x), y, BlasBufferUtil.getBlasStride(y));
  21.   }
  22. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  *
  3.  * @param in
  4.  * @param copy
  5.  * @return
  6.  */
  7. public static INDArray cos(INDArray in, boolean copy) {
  8.   return Nd4j.getExecutioner().execAndReturn(new Cos((copy ? in.dup() : in)));
  9. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * Returns the product along a given dimension
  3.  *
  4.  * @param dimension the dimension to getScalar the product along
  5.  * @return the product along the specified dimension
  6.  */
  7. @Override
  8. public INDArray prod(int... dimension) {
  9.   return Nd4j.getExecutioner().exec(new Prod(this), dimension);
  10. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray nextGaussian(char order, long[] shape) {
  3.   INDArray array = Nd4j.createUninitialized(shape, order);
  4.   GaussianDistribution op = new GaussianDistribution(array, 0.0, 1.0);
  5.   Nd4j.getExecutioner().exec(op, this);
  6.   return array;
  7. }

代码示例来源：origin: deeplearning4j/nd4j

  1. @Override
  2. public INDArray getActivation(INDArray in, boolean training) {
  3.   Nd4j.getExecutioner().execAndReturn(new Swish(in));
  4.   return in;
  5. }

代码示例来源：origin: deeplearning4j/nd4j

  1. /**
  2.  * computes a vector by a scalar product.
  3.  *
  4.  * @param N
  5.  * @param alpha
  6.  * @param X
  7.  */
  8. @Override
  9. public void scal(long N, double alpha, INDArray X) {
  10.   if (Nd4j.getExecutioner().getProfilingMode() == OpExecutioner.ProfilingMode.ALL)
  11.     OpProfiler.getInstance().processBlasCall(false, X);
  12.   if (X.isSparse()) {
  13.     Nd4j.getSparseBlasWrapper().level1().scal(N, alpha, X);
  14.   } else if (X.data().dataType() == DataBuffer.Type.DOUBLE)
  15.     dscal(N, alpha, X, BlasBufferUtil.getBlasStride(X));
  16.   else if (X.data().dataType() == DataBuffer.Type.FLOAT)
  17.     sscal(N, (float) alpha, X, BlasBufferUtil.getBlasStride(X));
  18.   else if (X.data().dataType() == DataBuffer.Type.HALF)
  19.     Nd4j.getExecutioner().exec(new ScalarMultiplication(X, alpha));
  20. }

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