tf.contrib.distributions.MultivariateNormalDiagWithSoftplusStDev.sample()

tf.contrib.distributions.MultivariateNormalDiagWithSoftplusStDev.sample(sample_shape=(), seed=None, name='sample')

Generate samples of the specified shape.

Note that a call to sample() without arguments will generate a single sample.

Args:

• sample_shape: 0D or 1D int32 Tensor. Shape of the generated samples.
• seed: Python integer seed for RNG
• name: name to give to the op.

Returns:

• samples: a Tensor with prepended dimensions sample_shape.