Subword Segmental BabyLMs: Learning to Tokenise for Sample-Efficient Pretraining

• URL: http://arxiv.org/abs/2609.01151v1
• Date: Tue, 01 Sep 2026 12:28:42 GMT
• Title: Subword Segmental BabyLMs: Learning to Tokenise for Sample-Efficient Pretraining
• Abstract summary: We present our submission to the 2026 BabyLM Challenge, for which we develop two new subword segmental LMs.<n>SubSegGPT is a decoder-only model that learns tokenisation during autoregressive pretraining.<n>SubSegDeBERTa is an encoder-based model that jointly learns to generate and tokenise masked words.
• Score: 5.465611192345226
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In the standard LM training pipeline, subword tokenisation is applied as a preprocessing step. Subword segmental language modelling is an alternative paradigm in which tokenisation is learned during training, allowing the model to discover subword units that optimise its training objective. In this paper, we present our submission to the 2026 BabyLM Challenge, for which we develop two new subword segmental LMs: SubSegGPT and SubSegDeBERTa. SubSegGPT is a decoder-only model that learns tokenisation during autoregressive pretraining. SubSegDeBERTa is an encoder-based model that jointly learns to generate and tokenise masked words. We train both for the Strict and Strict-small tracks. Our top submission to Strict is SubSegDeBERTa, which achieves notable gains in zero-shot evaluation. Our top submission to Strict-small is SubSegGPT, which outperforms tokenisation-based baselines. Our results show that learnable subword tokenisation can improve sample-efficiency for BabyLM pretraining. We analyse the subword learning dynamics of our models and find that tokenisation gradually converges on subword units that balance morphological alignment and fine-grained segmentation.

Related papers

• Decoupling the Benefits of Subword Tokenization for Language Model Training via Byte-level Simulation [6.240815206375064]
  Subword tokenization is an essential part of modern large language models (LLMs)<n>In this work, we decouple the effects of subword tokenization by isolating them within a controlled byte-level pretraining pipeline.
  arXiv  Detail & Related papers  (2026-04-29T23:29:08Z)
• StochasTok: Improving Fine-Grained Subword Understanding in LLMs [39.85256850592515]
  Subword-level understanding is integral to numerous tasks, including understanding multi-digit numbers, spelling mistakes, abbreviations, rhyming, and wordplay.<n>Current large language models (LLMs) still often struggle with seemingly simple subword-level tasks.<n>We introduce StochasTok, a simple, efficient tokenization scheme that randomly splits tokens during training, allowing LLMs to'see their internal structure'
  arXiv  Detail & Related papers  (2025-06-02T13:51:11Z)
• Segment First or Comprehend First? Explore the Limit of Unsupervised Word Segmentation with Large Language Models [92.92512796044471]
  We propose a new framework to explore the limit of unsupervised word segmentation with Large Language Models (LLMs)<n>We employ current mainstream LLMs to perform word segmentation across multiple languages to assess LLMs' "comprehension"<n>We introduce a novel unsupervised method, termed LLACA, which enables the construction of a dynamic $n$-gram model that adjusts based on contextual information.
  arXiv  Detail & Related papers  (2025-05-26T07:48:15Z)
• Token Assorted: Mixing Latent and Text Tokens for Improved Language Model Reasoning [53.57895922042783]
  Large Language Models (LLMs) excel at reasoning and planning when trained on chainof-thought (CoT) data.<n>We propose a hybrid representation of the reasoning process, where we partially abstract away the initial reasoning steps using latent discrete tokens.
  arXiv  Detail & Related papers  (2025-02-05T15:33:00Z)
• Subword Segmental Language Modelling for Nguni Languages [7.252933737829635]
  Subword segmental language model (SSLM) learns how to segment words while being trained for autoregressive language modelling. We train our model on the 4 Nguni languages of South Africa. Our results show that learning subword segmentation is an effective alternative to existing subword segmenters.
  arXiv  Detail & Related papers  (2022-10-12T18:41:00Z)
• Charformer: Fast Character Transformers via Gradient-based Subword Tokenization [50.16128796194463]
  We propose a new model inductive bias that learns a subword tokenization end-to-end as part of the model. We introduce a soft gradient-based subword tokenization module (GBST) that automatically learns latent subword representations from characters. We additionally introduce Charformer, a deep Transformer model that integrates GBST and operates on the byte level.
  arXiv  Detail & Related papers  (2021-06-23T22:24:14Z)
• COCO-LM: Correcting and Contrasting Text Sequences for Language Model Pretraining [59.169836983883656]
  COCO-LM is a new self-supervised learning framework that pretrains Language Models by COrrecting challenging errors and COntrasting text sequences. COCO-LM employs an auxiliary language model to mask-and-predict tokens in original text sequences. Our analyses reveal that COCO-LM's advantages come from its challenging training signals, more contextualized token representations, and regularized sequence representations.
  arXiv  Detail & Related papers  (2021-02-16T22:24:29Z)
• SLM: Learning a Discourse Language Representation with Sentence Unshuffling [53.42814722621715]
  We introduce Sentence-level Language Modeling, a new pre-training objective for learning a discourse language representation. We show that this feature of our model improves the performance of the original BERT by large margins.
  arXiv  Detail & Related papers  (2020-10-30T13:33:41Z)
• Char2Subword: Extending the Subword Embedding Space Using Robust Character Compositionality [24.80654159288458]
  We propose a character-based subword module (char2subword) that learns the subword embedding table in pre-trained models like BERT. Our module is robust to character-level alterations such as misspellings, word inflection, casing, and punctuation. We show that incorporating our module to mBERT significantly improves the performance on the social media linguistic code-switching evaluation (LinCE) benchmark.
  arXiv  Detail & Related papers  (2020-10-24T01:08:28Z)
• Byte Pair Encoding is Suboptimal for Language Model Pretraining [49.30780227162387]
  We analyze differences between unigram LM tokenization and byte-pair encoding (BPE) We find that the unigram LM tokenization method matches or outperforms BPE across downstream tasks and two languages. We hope that developers of future pretrained LMs will consider adopting the unigram LM method over the more prevalent BPE.
  arXiv  Detail & Related papers  (2020-04-07T21:21:06Z)

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.