Feucht et al. 2024 propose a way of interrogating the implicit vocabulary of an LLM, e.g., the term 'supercalifragilisticexpialidocious' is recognized as having meaning despite being many tokens in the llms vocabulary. But once you've found these implicit vocabulary items, are there higher-level vocabulary items that are formed by combination of implicit vocabulary items?
If you are inspired by this idea, you can reach out to the authors for collaboration or cite it:
@misc{holtzman-higherlevel-implicit-vocabulary-2026,
author = {Holtzman, Ari},
title = {Higher-Level Implicit Vocabulary},
year = {2026},
url = {https://hypogenic.ai/ideahub/idea/JrqdIINAjLHel0Rn957U}
}Please sign in to comment on this idea.
No comments yet. Be the first to share your thoughts!