Symbolic Prompt Compression via BPE-aware Encoding

Read full text Summarise with VIRAT
Authors: Emil Piotr Dobkowski
Cited by: 0

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 Semantic Density Compression (SDC), a compile-time method for reducing the token count of natural language prompts by replacing verbose phrases with BPE-efficient symbolic operators. SDC achieves 1.261x compression (CI=[1.196,1.327], n=100) with delta_task=0.000 (CI=[-0.040,+0.040], n=100, DeepSeek-V3).

Fields of Study & Subjects

Topic Modeling Natural Language Processing Techniques Algorithms and Data Compression

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN