M-LOOP v3.3.5

What's Changed

• Fixed trivial error in the RandomLearner by @ispielma in #171
• Parameters queue by @ispielma in #172
• Ignore .vscode by @zakv in #173
• Improved logging by @ispielma in #174
• Fix some controller.py log message typos by @zakv in #176
• Fix Nelder-Mead init_simplex_corner creation by @zakv in #177
• Drop tensorflow-macos requirement by @rmorshea in #179

New Contributors

• @ispielma made their first contribution in #171
• @rmorshea made their first contribution in #179

Full Changelog: v3.3.4...v3.3.5

M-LOOP 3.3.4

What's Changed

• Fix GaussianProcessVisualizer parameter scaling by @zakv in #167
• Record best_in_extras in controller archive by @zakv in #168
• Zero-index run numbers by @zakv in #169

Full Changelog: v3.3.3...v3.3.4

M-LOOP 3.3.3

What's Changed

• Install tensorflow-macos for mac instead of tensorflow by @atom-andrew in #163

New Contributors

• @atom-andrew made their first contribution in #163

Full Changelog: v3.3.2...v3.3.3

M-LOOP 3.3.2

What's Changed

• Mention stashing changes in release instructions by @zakv in #154
• Fix some formatting in RELEASE_INSTRUCTIONS.rst. by @zakv in #155
• Remove *args from ParameterScaler.__init__() by @zakv in #160
• Replace deprecated np.float by @zakv in #158

Full Changelog: v3.3.1...v3.3.2

M-LOOP 3.3.1

What's Changed

• Add a changelog page to the documentation by @zakv in #148
• Add a max_duration halting option to Controller by @zakv in #149
• Clip values outside, but close to, boundaries by @zakv in #150
• Add the phrase "Bayesian optimization" to the main page of the docs by @zakv in #152

Full Changelog: v3.3.0...v3.3.1

M-LOOP 3.3.0

What's Changed

• Neural Net plot individual cross sections option by @zakv in #104
• Avoid creating M-LOOP_archives directory when creating visualizers by @zakv in #106
• Deprecate controller_type argument of some visualization functions by @zakv in #110
• Update landscape_vis.py by @zakv in #109
• Removed uses of previously-removed gp_training_override_kwargs option by @zakv in #117
• Update installation and testing instructions by @zakv in #114
• Fix ShellInterface bug when param_names is None by @zakv in #115
• Stop globally disabling TF eager execution by @charmasaur in #118
• Make M-LOOP compatible with scikit-learn 1.0. by @zakv in #121
• Reduce level for some log messages in neuralnet.py by @zakv in #123
• Fix machine learner visualization cross sections range bug by @zakv in #124
• Datetime fixes by @zakv in #127
• Parameter scaling by @matth2peters in #125
• Fixed a matplotlib deprecation warning in NeuralNetVisualizer.plot_density_surface() by @zakv in #128
• Fix a small bug with noise_level when using a training archive by @zakv in #129
• Removed an unnecessary if condition in GaussianProcessVisualizer.return_cross_sections() by @zakv in #130
• Pass on Interface.get_next_cost_dict() errors by @zakv in #131
• Update the docsting for Controller. by @zakv in #132
• Simplify training archive arguments by @zakv in #134
• Add num_training_runs to the docs by @zakv in #135
• Only log "Showing visualizations..." when calling plt.show() by @zakv in #136
• Correct some cost scaling for the neural net by @zakv in #137
• Correct scaling of the trust region limits for visualizations by @zakv in #138
• Have M-LOOP create its own numpy random number generator by @zakv in #139
• Fix show_all_default_visualizations() for random and Nelder-Mead learners. by @zakv in #140
• Safer parameter boundary enforcement by @zakv in #141
• Update install instructions by @zakv in #142
• Controllers now always save an out_type value during each iteration. by @zakv in #143
• Have ControllerVisualizer.plot_cost_vs_run() support third party controllers by @zakv in #144
• Simplify the color generation for cost vs run plots. by @zakv in #145
• Bumped version number to 3.3.0. by @zakv in #146

New Contributors

• @matth2peters made their first contribution in #125

Full Changelog: 3.2.1...v3.3.0

M-LOOP 3.2.1

• Fix setup.py

M-LOOP 3.2.0

What's new

• Allow param_names to be specified on the shell interface (#66).
• Documentation improvements (#68, #79, #89).
• Support for visualizing Gaussian process hyperparameters when update_hyperparameters is false (#71).
• Support bounds for Gaussian process hyperparameters (#75).
• Support updating hyperparameters for the neural net controller (#92).
• Support external controllers from config file (#97).
• Change how costs are sent between learners and controllers (#90).
• Better support for TensorFlow 2 (#100).

Bug fixes

• Visualization of Gaussian process results when using isotropic kernels (#69).
• Scaling in Gaussian process controller (#72, #74).
• Various compatibility improvements for archives (#77, #84, #90).
• Various issues with minimum_uncertainty (#78).
• Structure of bad_run_indexs (#91).
• Incorporate cost uncertaines into Gaussian process controller (#73).
• General visualization polish (#86, #93).

M-LOOP 3.1.1

More tag fixes.

M-LOOP 3.1.0

• Version tag fixes.
• Neural net visualization improvements.