Changes to control model sparsity and MTSS improvements#331

Open

mcloughlin2 wants to merge 34 commits into1.7.0from sparsity

+677-206

Commits on May 7, 2024

(1) Added code to DCNNModelWrapper to track absolute sum of 1st hidden layer weights for each feature as it evolves
mcloughlin2
committed

Commits on May 22, 2024

Merge branch '1.6.1' into weight_decay_test
mcloughlin2
committed

Commits on May 31, 2024

Added two new parameters, wdp and wdt, to allow hyperopt domain searching on weight_decay_penalty and
mcloughlin2
committed
Added sparsity-related parameters weight_decay_penalty, weight_decay_penalty_type, xgb_alpha and xgb_lambda to set
mcloughlin2
committed

Commits on Jun 12, 2024

Fixed MTSS bug where population was not sorted by score after calling step(); changed order of operations so that grading
mcloughlin2
committed

Commits on Jun 13, 2024

Commits on Jun 18, 2024

Normalize fitness scores to the range [0,1]. Add the total fitness score to the fitness_scores dictionary so that it can
mcloughlin2
committed

Commits on Jun 21, 2024

Commits on Jun 28, 2024

Commits on Jul 1, 2024

Fixed plot_split_fractions so that bars always appear in order train, valid, test.
mcloughlin2
committed

Commits on Jul 2, 2024

Merge branch 'master' into sparsity
mcloughlin2
committed

Commits on Jul 12, 2024

Commits on Jul 19, 2024

Commits on Jul 22, 2024

Merge branch '1.7.0' of github.com:ATOMScience-org/AMPL into sparsity
mauvais2
committed

Commits on Jul 23, 2024

Merge branch 'master' of github.com:ATOMScience-org/AMPL into sparsity
mauvais2
committed

Commits on Aug 5, 2024

Merge branch '1.7.0' of github.com:ATOMScience-org/AMPL into sparsity
mauvais2
committed