当前位置: 首页>>代码示例>>Python>>正文


Python Reshape.inverse_log_det_jacobian方法代码示例

本文整理汇总了Python中tensorflow.contrib.distributions.python.ops.bijectors.reshape.Reshape.inverse_log_det_jacobian方法的典型用法代码示例。如果您正苦于以下问题:Python Reshape.inverse_log_det_jacobian方法的具体用法?Python Reshape.inverse_log_det_jacobian怎么用?Python Reshape.inverse_log_det_jacobian使用的例子?那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在tensorflow.contrib.distributions.python.ops.bijectors.reshape.Reshape的用法示例。


在下文中一共展示了Reshape.inverse_log_det_jacobian方法的1个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Python代码示例。

示例1: testBijector

# 需要导入模块: from tensorflow.contrib.distributions.python.ops.bijectors.reshape import Reshape [as 别名]
# 或者: from tensorflow.contrib.distributions.python.ops.bijectors.reshape.Reshape import inverse_log_det_jacobian [as 别名]
  def testBijector(self):
    """Do a basic sanity check of forward, inverse, jacobian."""
    expected_x = np.random.randn(4, 3, 2)
    expected_y = np.reshape(expected_x, [4, 6])

    with self.cached_session() as sess:
      shape_in, shape_out, feed_dict = self.build_shapes([3, 2], [6,])
      bijector = Reshape(
          event_shape_out=shape_out,
          event_shape_in=shape_in,
          validate_args=True)
      (x_,
       y_,
       fldj_,
       ildj_) = sess.run((
           bijector.inverse(expected_y),
           bijector.forward(expected_x),
           bijector.forward_log_det_jacobian(expected_x, event_ndims=2),
           bijector.inverse_log_det_jacobian(expected_y, event_ndims=2),
       ), feed_dict=feed_dict)
      self.assertEqual("reshape", bijector.name)
      self.assertAllClose(expected_y, y_, rtol=1e-6, atol=0)
      self.assertAllClose(expected_x, x_, rtol=1e-6, atol=0)
      self.assertAllClose(0., fldj_, rtol=1e-6, atol=0)
      self.assertAllClose(0., ildj_, rtol=1e-6, atol=0)
开发者ID:Ajaycs99,项目名称:tensorflow,代码行数:27,代码来源:reshape_test.py


注:本文中的tensorflow.contrib.distributions.python.ops.bijectors.reshape.Reshape.inverse_log_det_jacobian方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。