AdaptBPE: From General Purpose to Specialized Tokenizers
This addresses tokenization inefficiencies for LLM users in specialized domains or languages, though it's an incremental improvement over existing BPE methods.
The paper tackles the inefficiency of general-purpose tokenizers in specific domains by proposing AdaptBPE, a post-training adaptation strategy that replaces low-utility tokens with domain-relevant ones. Experiments show it compresses test corpora more effectively than baselines across multiple languages and tasks.
Subword tokenization methods, such as Byte-Pair Encoding (BPE), significantly impact the performance and efficiency of large language models (LLMs). The standard approach involves training a general-purpose tokenizer that uniformly processes all textual data during both training and inference. However, the use of a generic set of tokens can incur inefficiencies when applying the model to specific domains or languages. To address this limitation, we propose a post-training adaptation strategy that selectively replaces low-utility tokens with more relevant ones based on their frequency in an adaptation corpus. Our algorithm identifies the token inventory that most effectively encodes the adaptation corpus for a given target vocabulary size. Extensive experiments on generation and classification tasks across multiple languages demonstrate that our adapted tokenizers compress test corpora more effectively than baselines using the same vocabulary size. This method serves as a lightweight adaptation mechanism, akin to a vocabulary fine-tuning process, enabling optimized tokenization for specific domains or tasks. Our code and data are available at https://github.com/vijini/Adapt-BPE.git.