POSSMExtractor

class ts3.models.transductive.POSSMExtractor(*args, normalize='none', **kwargs)

Bases: ts3.models.transductive.base.TransductiveExtractor

POSSM’s unit_emb rows, optionally length-matched to the pretrain part.

normalize="scale_only" multiplies the eval part by one scalar so its mean vector length matches the pretrain part’s. Every eval vector keeps its direction and the eval cloud keeps its shape; only its radius moves.

property name: str

Names the embeddings file, so two runs of one model do not overwrite each other.

encode(regime)

Embeddings and uids for every unit of regime this extractor can represent.

Returns (N, D) and (N,) in matching order. Units the extractor cannot represent may be omitted: the probe join drops anything TS3 does not score, and insists separately that the eval side is complete.

Return type:

tuple[Tensor, ndarray]

read(state_dict, dataset, uids)

Embeddings for uids out of one checkpoint’s weights, and the uids they match.

Return type:

tuple[Tensor, ndarray]