LLMs encode repeated token counts correctly in residual streams but a format-triggered MLP at 88-93% depth overwrites it with an incorrect fixed value.
From early encoding to late suppression: Interpret- ing llms on character counting tasks.arXiv preprint arXiv:2604.00778
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Repeated-Token Counting Reveals a Dissociation Between Representations and Outputs
LLMs encode repeated token counts correctly in residual streams but a format-triggered MLP at 88-93% depth overwrites it with an incorrect fixed value.