PFN Studio
ArchitectureClassification

2-Layer Binary Classifier

embed → 2× attn → sigmoid head

PFN Studio Team

Same backbone as the regression baseline, scalar head emits one logit per point. Paired with the two-moons / GMM priors.

Architecture

tabular_embedder (d=64)transformer_encoder × 2scalar_head (d_out=1)
Open in the studio →
← Back to the marketplace