Compute Optimal Tokenization

• URL: http://arxiv.org/abs/2605.01188v1
• Date: Sat, 02 May 2026 01:53:22 GMT
• Title: Compute Optimal Tokenization
• Abstract summary: We investigate how the information granularity of tokens, controlled by the compression rate, affects scaling trends.<n>We train 988 latent tokenized models (BLT) ranging from 50M to 7B parameters that enable setting the desired compression rate.<n>Our experiments reveal that in compute-optimal configurations, model parameter counts scale proportionally to data size measured in bytes, not in tokens as commonly perceived.
• Score: 79.3815358070537
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Scaling laws enable the optimal selection of data amount and language model size, yet the impact of the data unit, the token, on this relationship remains underexplored. In this work, we systematically investigate how the information granularity of tokens, controlled by the compression rate (i.e., average bytes of text per token), affects scaling trends. We train 988 latent tokenized models (BLT) ranging from 50M to 7B parameters that enable setting the desired compression rate. This flexibility allows us to study the role of compression rate well beyond 4.57 bytes per token obtained with a popular BPE tokenizer. Our experiments reveal that in compute-optimal configurations, model parameter counts scale proportionally to data size measured in bytes, not in tokens as commonly perceived (Kaplan et al., 2020; Hoffmann et al., 2022). Furthermore, we discover that the optimal compression rate differs from the one obtained with BPE and decreases with compute. These findings generalize to both latent and subword tokenization, as well as to languages other than English, guiding language model developers on tokenization scheme selection for maximal compute efficiency.

Related papers

• Breaking the Token Ceiling: Distilling Smaller, Stronger Byte Models [67.72077328700088]
  We present the first large scale study of overtraining decoder-only dense transformer models varying two dimensions simultaneously.<n>We find that Token-1B models outperform byte models (End-Of-Token-1B and Bytes-1B) in the low-FLOP regime but eventually plateau.<n>By operating over a small vocabulary of 256 bytes instead of on the order of 100K tokens, they circumvent the need for top-kruncation during logit dumping.
  arXiv  Detail & Related papers  (2026-09-11T00:17:22Z)
• Significance-Gain Pair Encoding for LLMs: A Statistical Alternative to Frequency-Based Subword Merging [0.0]
  Significance-Gain BPE is a drop-in alternative merge criterion that measures cohesion via a z-statistic under an independence null model.<n>It reduces validation and test perplexity by 13% and 12%, respectively, and improves validation and test BPC by about 0.9 to 1.0%.
  arXiv  Detail & Related papers  (2026-02-26T20:29:18Z)
• Parity-Aware Byte-Pair Encoding: Improving Cross-lingual Fairness in Tokenization [53.22544362024936]
  Tokenization is the first -- and often least scrutinized -- step of most NLP pipelines.<n>Standard algorithms for learning tokenizers rely on frequency-based objectives.<n>We introduce Parity-aware Byte Pair.<n>We find empirically that Parity-aware BPE leads to more equitable token counts across languages.
  arXiv  Detail & Related papers  (2025-08-06T18:14:43Z)
• Unified Scaling Laws for Compressed Representations [69.72517034565467]
  We investigate whether a unified scaling framework can accurately predict model performance when training occurs over various compressed representations.<n>Our main finding is demonstrating both theoretically and empirically that there exists a simple "capacity" metric.<n>We extend our formulation to directly compare the accuracy potential of different compressed formats, and to derive better algorithms for training over sparse-quantized formats.
  arXiv  Detail & Related papers  (2025-06-02T16:52:51Z)
• How Much is Enough? The Diminishing Returns of Tokenization Training Data [4.300681074103876]
  We train BPE, UnigramLM, and WordPiece tokenizers across various vocabulary sizes using English training data ranging from 1GB to 900GB.<n>Our findings reveal diminishing returns as training data size increases beyond roughly 150GB.<n>For Russian text, we observe diminishing returns after training a tokenizer from 200GB of data, which is approximately 33% more than when training on English.
  arXiv  Detail & Related papers  (2025-02-27T17:01:23Z)
• Byte Latent Transformer: Patches Scale Better Than Tokens [101.10994909832063]
  Byte Latent Transformer (BLT) encodes bytes into dynamically sized patches, which serve as the primary units of computation.<n>For fixed inference costs, BLT shows significantly better scaling than tokenization-based models, by simultaneously growing both patch and model size.
  arXiv  Detail & Related papers  (2024-12-13T05:33:32Z)
• Unpacking Tokenization: Evaluating Text Compression and its Correlation with Model Performance [34.641079276516926]
  We argue for the theoretical importance of compression, that can be viewed as 0-gram language modeling. We demonstrate the empirical importance of compression for downstream success of pre-trained language models. We show that there is a correlation between tokenizers' compression and models' downstream performance.
  arXiv  Detail & Related papers  (2024-03-10T17:02:53Z)
• Tokenization Is More Than Compression [14.939912120571728]
  Existing tokenization approaches like Byte-Pair. (BPE) originate from the field of data compression. We introduce PathPiece, a new tokenizer that segments a document's text into the minimum number of tokens for a given vocabulary.
  arXiv  Detail & Related papers  (2024-02-28T14:52:15Z)
• TCRA-LLM: Token Compression Retrieval Augmented Large Language Model for Inference Cost Reduction [17.732468334493905]
  We propose a token compression scheme that includes two methods: summarization compression and semantic compression. Our summarization compression can reduce 65% of the retrieval token size with further 0.3% improvement on the accuracy. Our semantic compression provides a more flexible way to trade-off the token size with performance.
  arXiv  Detail & Related papers  (2023-10-24T06:56:38Z)
• Dynamic Programming Encoding for Subword Segmentation in Neural Machine Translation [80.38621085548013]
  This paper introduces Dynamic Programming (DPE) a new segmentation algorithm for tokenizing sentences into subword units. A mixed character-subword transformer is proposed, which enables exact log marginal likelihood estimation and exact MAP inference to find target segmentations.
  arXiv  Detail & Related papers  (2020-05-03T05:00:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.