Make IndependentReparametrizationSampler create variables at init

secondmind-labs / trieste

A Bayesian optimization toolbox built on TensorFlow

Apache License 2.0

212 stars 42 forks source link

Make IndependentReparametrizationSampler create variables at init #789

Closed sam-willis closed 9 months ago

sam-willis commented 9 months ago

Related issue(s)/PRs:

Summary

Make IndependentReparametrizationSampler create variables at init. Without this you cannot always save the sampler with dynamic shape (i.e. before sample has been called). ...

Fully backwards compatible: yes / no

PR checklist

[ ] The quality checks are all passing
[ ] The bug case / new feature is covered by tests
[ ] Any new features are well-documented (in docstrings or notebooks)

sam-willis commented 9 months ago

Closing as cannot support XLA compilation due to dynamic shape