ts3.models.supervised#

Models trained in-suite on the region labels, which emit probabilities, not embeddings.

They skip ts3/extract.py and the probes entirely and write their own submission, so the inductive/transductive question does not arise: what sets them apart is supervision on the scored label, not adaptation on the eval sessions.

LOLCAT

ISI histogram encoder, attention pooling and a brain region classifier [Schneider et al., 2023].

MultiHeadGlobalAttention

Multi-Head Global pooling layer.

Trainers#

LOLCATTrainer

Supervised training of LOLCAT, the one TS3 model fit on the region labels.

LossFeedbackSampler

Samples elements randomly from a given list of indices, without replacement.