SeqPE encodes each position as a symbolic digit sequence through a small Transformer, and with contrastive plus distillation losses it reports improved extrapolation in language, QA, and image classification.
Lex- ical generalization improves with larger models and longer training
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
SeqPE: Transformer with Sequential Position Encoding
SeqPE encodes each position as a symbolic digit sequence through a small Transformer, and with contrastive plus distillation losses it reports improved extrapolation in language, QA, and image classification.