How rick12000/confopt works
ConfOpt is a Python hyperparameter-optimization library. A practitioner defines a search space and objective, runs ConformalTuner to evaluate configurations, then retrieves the best parameters and score. Its main search loop uses conformal acquisition and sampling, backed by quantile estimators and adaptive interval control; tracking and configuration utilities support trial state. The graph preserves README-described capabilities and source-indexed modules. Some implementation details were not sampled, so edges rely on the README or the indexed source references.
Main entry points: tuning.py Conformal tuner · conformalization.py Conformal intervals · acquisition.py Acquisition searcher · estimation.py Estimator tuning
Components 19
Search workflow
- Conformal tuner
confopt/tuning.py - Acquisition searcher
confopt/selection/acquisition.py - Bound samplers
confopt/selection/sampling/bound_samplers.py - Improvement sampler
confopt/selection/sampling/expected_improvement_samplers.py - Thompson sampler
confopt/selection/sampling/thompson_samplers.py - Optimization utilities
confopt/utils/optimization.py
Conformal models
- Conformal intervals
confopt/selection/conformalization.py - Coverage adaptation
confopt/selection/adaptation.py - Estimator tuning
confopt/selection/estimation.py - Estimator registry
confopt/selection/estimator_configuration.py - Quantile estimators
confopt/selection/estimators/quantile_estimation.py - Estimator ensembles
confopt/selection/estimators/ensembling.py - Data preprocessing
confopt/utils/preprocessing.py
Configuration and sampling
- Parameter ranges
confopt/wrapping.py - Candidate sampling
confopt/utils/configurations/sampling.py - Configuration encoding
confopt/utils/configurations/encoding.py
Trial state
- Trial tracking
confopt/utils/tracking.py
Other components
- Practitioner
- Objective function
Connections 34
- Practitioner → Conformal tuner: configures and runs
README.md - Practitioner → Parameter ranges: defines search space
README.md - Conformal tuner → Objective function: evaluates configurations
README.md - Conformal tuner → Acquisition searcher: uses searcher
confopt/tuning.py - Conformal tuner → Bound samplers: uses sampler
confopt/tuning.py - Conformal tuner → Thompson sampler: uses sampler
confopt/tuning.py - Conformal tuner → Trial tracking: tracks trials
confopt/tuning.py - Conformal tuner → Optimization utilities: uses utilities
confopt/tuning.py - Conformal tuner → Parameter ranges: accepts ranges
confopt/tuning.py - Conformal tuner → Practitioner: returns best result
README.md - Acquisition searcher → Conformal intervals: uses intervals
confopt/selection/acquisition.py - Acquisition searcher → Bound samplers: dispatches sampling
confopt/selection/acquisition.py - Acquisition searcher → Improvement sampler: dispatches sampling
confopt/selection/acquisition.py - Acquisition searcher → Thompson sampler: dispatches sampling
confopt/selection/acquisition.py - Acquisition searcher → Estimator tuning: uses estimators
confopt/selection/acquisition.py - Acquisition searcher → Estimator registry: uses configuration
confopt/selection/acquisition.py - Conformal intervals → Estimator tuning: tunes models
confopt/selection/conformalization.py - Conformal intervals → Estimator registry: selects architecture
confopt/selection/conformalization.py - Conformal intervals → Data preprocessing: splits data
confopt/selection/conformalization.py - Conformal intervals → Parameter ranges: uses bounds
confopt/selection/conformalization.py - Bound samplers → Coverage adaptation: adapts coverage
confopt/selection/sampling/bound_samplers.py - Bound samplers → Conformal intervals: samples prediction bounds
- Improvement sampler → Conformal intervals: samples predictions
- Thompson sampler → Conformal intervals: samples predictions
- Estimator tuning → Estimator registry: builds estimators
confopt/selection/estimation.py - Estimator tuning → Quantile estimators: fits quantile models
confopt/selection/estimation.py - Estimator tuning → Estimator ensembles: fits ensembles
confopt/selection/estimation.py - Estimator tuning → Candidate sampling: samples configurations
confopt/selection/estimation.py - Estimator registry → Quantile estimators: registers estimators
confopt/selection/estimator_configuration.py - Estimator registry → Estimator ensembles: registers ensembles
confopt/selection/estimator_configuration.py - Estimator ensembles → Quantile estimators: combines estimators
confopt/selection/estimators/ensembling.py - Candidate sampling → Parameter ranges: samples ranges
confopt/utils/configurations/sampling.py - Trial tracking → Configuration encoding: encodes configurations
confopt/utils/tracking.py - Trial tracking → Candidate sampling: generates candidates
confopt/utils/tracking.py