ts3.probes

The classifiers fit on a saved embeddings file, one embedding per unit.

Everything downstream of a probe belongs to ts3/eval.py: the multi-unit readout, the reports and the submission files. A new probe therefore implements Probe.fit_predict() and inherits the rest of the protocol.

Probe	A classifier fit on unit embeddings, scored by eval.py.
LinearProbe	Multinomial logistic regression on standardized embeddings.
MLPProbe	An MLP whose hyperparameters are swept with Optuna before the final fit.