tensorflow / tensorflow/probability

In examples/models/bayesian_vgg.py How does one compute the loss of the model?

Open
#385 0 comments 0 reactions 1 assignee View on GitHub

@jburnim is already working on this.

Since May 9, 2019.

keras layers
Dominant language
Jupyter Notebook
Stars
4.4k
Forks
1.1k
PR merge metrics
No merged PRs in 30d

Description

Hello,

Once defined as

x = tf.keras.layers.Flatten()(x)
  x = tfp.layers.DenseFlipout(
      num_classes,
      kernel_posterior_fn=kernel_posterior_fn)(x)
  model = tf.keras.Model(inputs=image, outputs=x, name='vgg16')

How do you compute the elbo Loss?
Does model.losses work?
which then would lead to this computation

labels_distribution = tfd.Categorical(logits=logits)
    neg_log_likelihood = -tf.reduce_mean(input_tensor=labels_distribution.log_prob(abels))
kl = sum(model.losses) / mnist_data.train.num_examples
elbo_loss = neg_log_likelihood + kl
self._loss_op = elbo_loss

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.