modelforge.dataset.utils

Utility functions for dataset handling.

Functions

calculate_mean_and_variance(torch_dataset[, ...])	Calculates the mean and variance of the dataset.
calculate_size_of_splits(total_size, split_frac)	Calculate the size of each split based on the total size and the split ratios.
normalize_energies(dataset, stats)	Normalizes the energies in the dataset.
random_record_split(dataset, lengths[, ...])	Randomly split a TorchDataset into non-overlapping new datasets of given lengths, keeping all conformers in a record in the same split
two_stage_random_split(dataset_size, ...)	Perform a two-stage random split of a dataset.

Classes

FirstComeFirstServeSplittingStrategy([split])	Strategy to split a dataset based on idx.
RandomRecordSplittingStrategy([seed, split, ...])	Strategy to split a dataset randomly, keeping all configurations in a record in the same split.
RandomSplittingStrategy([seed, split, test_seed])	Strategy to split a dataset randomly.
SplittingStrategy(split[, seed, test_seed])	Base class for dataset splitting strategies.