Revision b41d4f38436e4a090c940dbd3bc7e2afd39a283e authored by st-- on 23 April 2020, 18:17:42 UTC, committed by GitHub on 23 April 2020, 18:17:42 UTC
Previously, GPflow's NaturalGradient optimizer would call the loss_function once for each (q_mu, q_sqrt) set in the var_list. This is a light refactor that separates out applying the natural gradient step from computing the gradients (`_natgrad_apply_gradients`), and changes `_natgrad_steps` to only evaluate the loss function once, computing the gradients for all (q_mu, q_sqrt) tuples passed in the var_list. Other changes: - The no-longer-used `_natgrad_step` method got removed. - NaturalGradient now takes a `xi_transform` argument that is used for all parameter sets without explicitly specified xi transform (i.e. tuples rather than triplets). - XiTransform has been changed to have staticmethods. None of this should affect any downstream code; this PR is backwards-compatible.
1 parent c7550ce
Makefile
.PHONY: help clean dev-install install package format test
help:
@echo "The following make targets are available:"
@echo " dev-install install all dependencies for dev environment and sets a egg link to the project sources"
@echo " install install all dependencies and the project in the current environment"
@echo " package build pip package"
@echo " test run all tests in parallel"
@echo " clean removes package, build files and egg info"
clean:
rm -rf dist *.egg-info build
dev-install:
pip install -e .
install:
pip install .
package:
python setup.py bdist
format:
black -t py36 -l 100 gpflow tests doc setup.py
test:
pytest -v --durations=10 tests/
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...