probly

probly: Uncertainty Representation and Quantification for Machine Learning.

Submodules

calibrator	Calibrator protocols and calibration dispatch.
conformal_scores	Conformal prediction non-conformity score functions.
datasets	Init module for dataset implementations.
decider	Deciders reduce representations to a desired decision space (e.g., second-order to first-order distribution).
evaluation	Downstream evaluation tasks for uncertainty pipelines: active learning, OOD detection, and selective prediction.
integrations	Optional integrations with external uncertainty libraries.
layers	Init module for layer implementations.
lazy_types	Collection of fully-qualified type names for lazy type checking.
losses	Training losses.
method	Uncertainty-aware methods.
metrics	Metrics with backend dispatch for NumPy, PyTorch, and JAX.
plot	Plotting utilities for probly.
predictor	Module for predictors.
quantification	Quantification methods for uncertainty.
representation	Uncertainty representations for models.
representer	Representation builders that create representations from predictor outputs.
train	Training utilities.
transformation	Transformations for models.
traverse_nn	Traverser utilities for neural networks.
utils	Utils module for probly library.