Gaussian processes in TensorFlow
This patch release fixes a number of identified issues.
Linear
mean functionThis release contains contributions from:
JSchmiegel, uri-granta
This release adds caching of compiled graphs inside the Scipy optimizer, and adds support for returning loss history. It also adds supports for Python 3.11.
compile_cache_size
argument to 0.This release contains contributions from:
khurram-ghani, jesnie, uri-granta
The main focus of this release is to provide users control over arguments for tf.function
compilation inside the Scipy minimize wrapper. It also adds support for a new categorical kernel.
tf.function
arguments for compilation in gpflow.optimizers.Scipy
.
(#2064)SquaredExponential
and Constant
kernels. (#2056, #2063)This release contains contributions from:
sc336, partev, khurram-ghani, uri-granta, awav, jesnie
A small fix for a bug in the scipy optimize wrapper.
gpflow.optimizers.Scipy
to always assign the last good state returned by scipy.optimize.minimize
to the model under optimization. Previously, this step could be missed if minimize
failed in some situations, leaving the model in an arbitrary state.This release contains contributions from: khurram-ghani
Full Changelog: https://github.com/GPflow/GPflow/compare/v2.6.4...v2.6.5
A small fix for a bug in the scipy optimize wrapper.
gpflow.optimizers.Scipy
to always assign the last good state returned by scipy.optimize.minimize
to the model under optimization. Previously, this step could be missed if minimize
failed in some situations, leaving the model in an arbitrary state.This release contains contributions from: khurram-ghani
Full Changelog: https://github.com/GPflow/GPflow/compare/v2.6.4...v2.6.5
The main theme of this release is documentation, with a new suite of tutorials, several upgrades to notebooks and the removal of a rather annoying bug in the documentation site.
Perhaps more notably, check_shapes
has been removed, and can now be found here. This change is breaking for those who are still getting check_shapes
from gpflow
, although being in experimental this change does not require a new version number.
gpflow.experimental.check_shapes
has been removed, in favour of an independent release. Use
pip install check_shapes
and import check_shapes
instead.tf.saved_model
and methods wrapped in @check_shapes
.Adam
optimizer.This release contains contributions from:
sc336, st--, sethaxen, jesnie
This is a bug-fix release, for compatibility with GPflux.
tfp.python.layers.internal.distribution_tensor_coercible._TensorCoercible
.FallbackSeparateIndependentInducingVariables
to have children with different shapes.GaussianQuadrature
to be different.This release contains contributions from:
jesnie