2 papers
cs.LG2026
Geometric Evolution Maps: Extracting Stable Concept Probes from Transformer Residual Streams
James Henry
A concept probe is only as reliable as the layer it is taken from. Probing at a fixed late layer, or at the peak of a separation curve, ignores a structural feature of how concepts…
cs.LG2026
The Concept Allocation Zone: Tracking How Concepts Form Across Transformer Depth
James Henry
Concept formation in transformer language models is a depth-extended process, not a single-layer event: a concept becomes separable across one or more contiguous regions of the res…