tensorflow / tensorflow/probability
Why aren't there the classes Convolution2DVariational and Convolution2DLocalReparameterization?
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 4.4k
- Forks
- 1.1k
- PR merge metrics
- No merged PRs in 30d
Description
There are the classes tfp.layers.DenseVariational and tfp.layers.DenseLocalReparameterization, apart from tfp.layers.DenseReparameterization and tfp.layers.DenseFlipout, to represent a Bayesian dense layer. However, there are no Convolution1DVariational, Convolution2DVariational, Convolution3DVariational, Convolution1DLocalReparameterization, Convolution2DLocalReparameterization and Convolution3DLocalReparameterization classes. Why is that?
This question had already been partially asked in this comment: https://github.com/tensorflow/probability/issues/511#issuecomment-516564049.
See https://github.com/tensorflow/probability/issues/409#issuecomment-492870964, which partially answers this question (i.e. TFP and TF are extremely buggy).
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by comparing the documented tfp.layers.DenseVariational and DenseLocalReparameterization classes with the existing convolutional layer classes named in the issue. Determine whether the missing Convolution1D/2D/3D variants are intended as new features, then define the required behavior and validation before implementation.
Written by the indexing model from the issue text.
Assessment
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 18/100