Over-Tokenized Transformer
arxiv.org ↗(ICML 2025) - decouples input and output vocabularies; shows a log-linear relationship between input vocabulary size and training loss, scaling vocabulary independently of model size.
AboutVisit arxiv.org ↗
(ICML 2025) - decouples input and output vocabularies; shows a log-linear relationship between input vocabulary size and training loss, scaling vocabulary independently of model size.
This entry links to an external website rather than a GitHub repository, so there's no README to display here.