Beyond Fertility: Analyzing STRR as a Metric for Multilingual Tokenization Evaluation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Nayeem, Mir Tafseer, Alqahtani, Sawsan, Laskar, Md Tahmid Rahman, Mohiuddin, Tasnim, Bari, M Saiful
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917041214062592
author Nayeem, Mir Tafseer
Alqahtani, Sawsan
Laskar, Md Tahmid Rahman
Mohiuddin, Tasnim
Bari, M Saiful
author_facet Nayeem, Mir Tafseer
Alqahtani, Sawsan
Laskar, Md Tahmid Rahman
Mohiuddin, Tasnim
Bari, M Saiful
contents Tokenization is a crucial but under-evaluated step in large language models (LLMs). The standard metric, fertility (the average number of tokens per word), captures compression efficiency but obscures how vocabularies are allocated across languages and domains. We analyze six widely used tokenizers across seven languages and two domains, finding stable fertility for English, high fertility for Chinese, and little domain sensitivity. To address fertility's blind spots, we propose the Single Token Retention Rate (STRR), which measures the proportion of words preserved as single tokens. STRR reveals systematic prioritization of English, strong support for Chinese, and fragmentation in Hindi, offering an interpretable view of cross-lingual fairness. Our results show that STRR complements fertility and provides practical guidance for designing more equitable multilingual tokenizers.
format Preprint
id arxiv_https___arxiv_org_abs_2510_09947
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Fertility: Analyzing STRR as a Metric for Multilingual Tokenization Evaluation
Nayeem, Mir Tafseer
Alqahtani, Sawsan
Laskar, Md Tahmid Rahman
Mohiuddin, Tasnim
Bari, M Saiful
Computation and Language
Artificial Intelligence
Machine Learning
Tokenization is a crucial but under-evaluated step in large language models (LLMs). The standard metric, fertility (the average number of tokens per word), captures compression efficiency but obscures how vocabularies are allocated across languages and domains. We analyze six widely used tokenizers across seven languages and two domains, finding stable fertility for English, high fertility for Chinese, and little domain sensitivity. To address fertility's blind spots, we propose the Single Token Retention Rate (STRR), which measures the proportion of words preserved as single tokens. STRR reveals systematic prioritization of English, strong support for Chinese, and fragmentation in Hindi, offering an interpretable view of cross-lingual fairness. Our results show that STRR complements fertility and provides practical guidance for designing more equitable multilingual tokenizers.
title Beyond Fertility: Analyzing STRR as a Metric for Multilingual Tokenization Evaluation
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2510.09947