Advancing Polish Language Modeling through Tokenizer Optimization in the Bielik v3 7B and 11B Series
This work addresses tokenizer inefficiency for morphologically rich languages, offering a practical optimization for Polish NLP practitioners.
The paper presents Bielik v3, a Polish-optimized LLM series (7B and 11B parameters) that replaces a universal tokenizer with a Polish-specific vocabulary, achieving a 30% reduction in fertility ratio and 25% faster inference while maintaining competitive performance on Polish benchmarks.
The development of the Bielik v3 PL series, encompassing both the 7B and 11B parameter variants, represents a significant milestone in the field of language-specific large language model (LLM) optimization. While general-purpose models often demonstrate impressive multilingual capabilities, they frequently suffer from a fundamental architectural inefficiency: the use of universal tokenizers. These tokenizers, typically designed to cover a broad spectrum of languages, often fail to capture the morphological nuances of specific languages like Polish, leading to higher fertility ratios, increased inference costs, and restricted effective context windows. This report details the transition from the universal Mistral-based tokenization to a dedicated Polish-optimized vocabulary for the Bielik v3 models, exploring the FOCUS-based embedding initialization, the multi-stage pretraining curriculum, and the subsequent post-training alignment involving Supervised Fine-Tuning, Direct Preference Optimization, and Reinforcement Learning through Group Relative Policy Optimization with verifiable rewards.