Language models develop a small group of 'rare token neurons' in their final layer that show a plateau, power-law, and rapid decay pattern of influence and heavy-tailed weight statistics.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Emergent Specialization: Rare Token Neurons in Language Models
Language models develop a small group of 'rare token neurons' in their final layer that show a plateau, power-law, and rapid decay pattern of influence and heavy-tailed weight statistics.