ByteSpan: Information-Driven Subword Tokenisation
This work addresses tokenization efficiency and morphological alignment for NLP practitioners, but it is incremental as it builds on existing dynamic tokenization methods.
The paper tackled the problem of subword tokenization by proposing ByteSpan, an information-driven tokenizer that groups predictable byte sequences using an external byte-level language model, resulting in vocabularies with higher morphological alignment scores than BPE for English and similar compression and efficiency across 25 languages.
Recent dynamic tokenisation methods operate directly on bytes and pool their latent representations into patches. This bears similarities to computational models of word segmentation that determine lexical boundaries using spikes in an autoregressive model's prediction error. Inspired by this connection, we explore whether grouping predictable bytes - rather than pooling their representations - can yield a useful fixed subword vocabulary. We propose a new information-driven subword tokeniser, ByteSpan, that uses an external byte-level LM during training to identify contiguous predictable byte sequences and group them into subwords. Experiments show that ByteSpan yields efficient vocabularies with higher morphological alignment scores than BPE for English. Multilingual experiments show similar compression and Rényi efficiency for 25 languages.