LingGen: Scalable Multi-Attribute Linguistic Control via Power-Law Masking

Read full text Summarise with VIRAT
Authors: Mohamed Elgaar, Hadi Amiri
Cited by: 0
References: 31

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

We present LingGen, a controlled text generation model that allows fine-grained control over a large number of real-valued linguistic attributes. It encodes target attribute values with a dedicated linguistic attribute encoder and conditions the language model by injecting the resulting representation into the language model using the beginning-of-sequence (BOS) embeddings. To improve robustness when controlling different attribute subsets, we introduce P-MASKING, which samples per-example attribute masking rates from a truncated Pareto distribution during training. Across 1-40 control attributes, LingGen achieves the lowest average control error among evaluated methods, while remaining efficient at inference and receiving the highest fluency scores in human evaluation. Ablations show that Pareto-sampled masking and BOS-based injection are effective choices compared to alternative masking and integration variants.

Fields of Study & Subjects

Topic Modeling Generative Adversarial Networks and Image Synthesis Natural Language Processing Techniques

Publication details

Journal / source
Conference of the European Chapter of the Association for Computational Linguistics
Pages
1925–1942
Publication date
1 January 2026
Language
EN