Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zouhar, Vilém, Meister, Clara, Gastaldi, Juan Luis, Du, Li, Vieira, Tim, Sachan, Mrinmaya, Cotterell, Ryan
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2306.16837
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909301701869568
author Zouhar, Vilém
Meister, Clara
Gastaldi, Juan Luis
Du, Li
Vieira, Tim
Sachan, Mrinmaya
Cotterell, Ryan
author_facet Zouhar, Vilém
Meister, Clara
Gastaldi, Juan Luis
Du, Li
Vieira, Tim
Sachan, Mrinmaya
Cotterell, Ryan
contents Byte-Pair Encoding (BPE) is a popular algorithm used for tokenizing data in NLP, despite being devised initially as a compression method. BPE appears to be a greedy algorithm at face value, but the underlying optimization problem that BPE seeks to solve has not yet been laid down. We formalize BPE as a combinatorial optimization problem. Via submodular functions, we prove that the iterative greedy version is a $\frac{1}{σ(\boldsymbolμ^\star)}(1-e^{-{σ(\boldsymbolμ^\star)}})$-approximation of an optimal merge sequence, where ${σ(\boldsymbolμ^\star)}$ is the total backward curvature with respect to the optimal merge sequence $\boldsymbolμ^\star$. Empirically the lower bound of the approximation is $\approx 0.37$. We provide a faster implementation of BPE which improves the runtime complexity from $\mathcal{O}\left(N M\right)$ to $\mathcal{O}\left(N \log M\right)$, where $N$ is the sequence length and $M$ is the merge count. Finally, we optimize the brute-force algorithm for optimal BPE using memoization.
format Preprint
id arxiv_https___arxiv_org_abs_2306_16837
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle A Formal Perspective on Byte-Pair Encoding
Zouhar, Vilém
Meister, Clara
Gastaldi, Juan Luis
Du, Li
Vieira, Tim
Sachan, Mrinmaya
Cotterell, Ryan
Computation and Language
Optimization and Control
Byte-Pair Encoding (BPE) is a popular algorithm used for tokenizing data in NLP, despite being devised initially as a compression method. BPE appears to be a greedy algorithm at face value, but the underlying optimization problem that BPE seeks to solve has not yet been laid down. We formalize BPE as a combinatorial optimization problem. Via submodular functions, we prove that the iterative greedy version is a $\frac{1}{σ(\boldsymbolμ^\star)}(1-e^{-{σ(\boldsymbolμ^\star)}})$-approximation of an optimal merge sequence, where ${σ(\boldsymbolμ^\star)}$ is the total backward curvature with respect to the optimal merge sequence $\boldsymbolμ^\star$. Empirically the lower bound of the approximation is $\approx 0.37$. We provide a faster implementation of BPE which improves the runtime complexity from $\mathcal{O}\left(N M\right)$ to $\mathcal{O}\left(N \log M\right)$, where $N$ is the sequence length and $M$ is the merge count. Finally, we optimize the brute-force algorithm for optimal BPE using memoization.
title A Formal Perspective on Byte-Pair Encoding
topic Computation and Language
Optimization and Control
url https://arxiv.org/abs/2306.16837