Welcome to CTLearn Optimizer’s documentation!¶
CTLearn Optimizer is a framework for optimizing CTLearn models.
This optimization utility uses Tune, a scalable framework for hyperparameter search and model training, and supports:
Random search based optimization.
Tree Parzen Estimators based optimization.
Gaussian Processes based optimization.
Genetic Algorithm based optimization.
Parallel optimization (depending on available hardware resources).
Contents:
- Installation
- Basic usage
- Configuration
- General settings
num_cpusnum_gpusnum_cpus_per_trialnum_gpus_per_trialnum_parallel_trialsoptimization_typemodeworking_directoryctlearn_confign_initial_pointsnum_max_evalsrandom_stateremove_training_foldersreload_trialsreload_optimization_resultspredictdata_set_to_optimizemetrics_val_to_logmetrics_pred_to_logmetric_to_optimizeuser_defined_metric_valuser_defined_metric_pred
- Optimizer settings
- CTLearn settings
- Hyperparameters settings
- General settings
- Analysis of the optimization results
- Package Reference
- Contributing to CTLearn Optimizer