Graph tokenizations for Transformers induce distinct depth regimes with proven separations and impossibility results for converting between them at limited depth.
IEEE Signal Processing Magazine , volume =
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2representative citing papers
Oversmoothing in neural sheaf diffusion is reframed as representation degeneration in the incidence-quiver harmonic space, with moment-map regularizers and non-uniform stalk dimensions proposed to avoid it.
citing papers explorer
-
Lost in Tokenization: Fundamental Trade-offs in Graph Tokenization for Transformers
Graph tokenizations for Transformers induce distinct depth regimes with proven separations and impossibility results for converting between them at limited depth.
-
Oversmoothing as Representation Degeneracy in Neural Sheaf Diffusion
Oversmoothing in neural sheaf diffusion is reframed as representation degeneration in the incidence-quiver harmonic space, with moment-map regularizers and non-uniform stalk dimensions proposed to avoid it.