Discrete tokenization in scientific foundation models imposes a geometric alignment tax that distorts continuous manifolds, with continuous heads reducing distortion by up to 8.5x and exposing three failure regimes in 14 biological models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
The Geometric Alignment Tax: Tokenization vs. Continuous Geometry in Scientific Foundation Models
Discrete tokenization in scientific foundation models imposes a geometric alignment tax that distorts continuous manifolds, with continuous heads reducing distortion by up to 8.5x and exposing three failure regimes in 14 biological models.