claim
active
claim:the-large-harmful-in-utions-weight-implies-weight-superposition-since-it-would-have-no-reason-to-be-large-if-virtual-weights-were-trained-directly

The large harmful IN→utions weight implies weight superposition since it would have no reason to be large if virtual weights were trained directly

Motivating example.

Source paper

extracted_from
Characterizing interference weights in a tiny language model
Nicholas L. Turner · Jeffrey Wu · Joshua Batson

Neighborhood — ranked by edge-count

Findings (1)

finding