Rethinking Deep Self-Attention for Character-Level Language Modeling
Year: 2026
Type: preprint
Abstract: A counterargument paper examining the limitations of deep transformer models for character-level language modeling, proposing hybrid architectures as a more practical direction.
Source: OSF Preprints (OSF Preprints)
Author: Angel Cruz
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7160369780