PoPE, a polar-coordinate variant of RoPE that removes the content-phase interaction, improves Transformer sequence modeling and extrapolation across multiple domains.
We use a vocabulary size of 90 which includes the MIDI notes, silence and padding tokens
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
method 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
method 1polarities
use method 1representative citing papers
citing papers explorer
-
Decoupling the "What" and "Where" With Polar Coordinate Positional Embeddings
PoPE, a polar-coordinate variant of RoPE that removes the content-phase interaction, improves Transformer sequence modeling and extrapolation across multiple domains.