incerto.ood.MSP

class incerto.ood.MSP(model)

Bases: OODDetector

Maximum-Softmax-Probability (MSP) baseline OOD detector.

Returns 1 - max softmax probability as the OOD score, so higher means more likely OOD. Despite its simplicity, MSP is a strong baseline.

Reference:

Hendrycks & Gimpel, “A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks”, ICLR 2017.

Parameters:

model – A trained classifier returning logits of shape (batch, n_classes).

__init__(model)

Initialize the OOD detector with a trained model.

The model is automatically: 1. Set to eval mode 2. Has gradients disabled (requires_grad=False)

Parameters:

model – A trained PyTorch model (nn.Module)

Raises:

TypeError – If model is not an nn.Module.

Methods

__init__(model)	Initialize the OOD detector with a trained model.
load(path, model, **kwargs)	Load detector state from a file.
load_state_dict(state)	Load detector state from a dictionary.
predict(x, threshold)	Predict whether inputs are OOD using a threshold.
save(path)	Save detector state to a file (excluding the model).
score(x)	Compute OOD scores for a batch of inputs.
state_dict()	Return a dictionary containing the detector's state.

score(x)

Compute OOD scores for a batch of inputs.

Parameters:

x (Tensor) – Input batch passed to model.

Return type:

Tensor

Returns:

Tensor of shape (batch,) with values in [0, 1). Higher values indicate higher OOD likelihood.