molprop.inference.eval

molprop.inference.eval(model, loader, args, device, scores=None, target_norm_subtrahent=None, target_norm_denumerator=None, keep_infer_results=False)

Used to calculate validation metrics.

Parameters:

• model (torch.nn.Module) – Trained model used for inference.

• loader (torch.utils.data.DataLoader) – Provides dataset for inference.

• args – Parameters defined in defaultconfig.yaml.

• device – Device on which the inference is performed.

Returns:

raw_metric_epoch (float) and metric_epoch (float) are teh validation metrics. If keep_infer_results is True, the results of infer() are also returned.

Return type:

tuple