A small transformer trained on counter languages encodes the current stack depth in its final-layer activations, recoverable by simple probing classifiers.
Bhattamishra, Kabir Ahuja, and Navin Goyal
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Emergent Stack Representations in Modeling Counter Languages Using Transformers
A small transformer trained on counter languages encodes the current stack depth in its final-layer activations, recoverable by simple probing classifiers.