Formal Semantic Control over Language Models
Yingji Zhang · 2026
Read Formal Semantic Control over Language Models by Yingji Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-00638.txt.
Description
Yingji Zhang advances semantic representation learning to make language models more interpretable and to enable compositional control by shaping latent space geometry. Within a VAE framework, sentence-level work splits and steers meaning features to guide output, using explanatory text as testbed. Reasoning-level work singles out and steers inference to control natural language inference, focusing on explanatory tasks where two claims infer a conclusion. The thesis shows these approaches enhance clarity and control of latent spaces across natural language.
Catalog record
| CREATOR | Yingji Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 340858 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.00638