Structural priors for data-efficient language learning

• URL: http://arxiv.org/abs/2609.11505v1
• Date: Thu, 10 Sep 2026 13:12:48 GMT
• Title: Structural priors for data-efficient language learning
• Abstract summary: We investigate structural transfer: first training models on non-language data to induce useful priors for natural language.<n>We evaluate transfer via next-token-prediction loss, weight shifts in the model, and downstream linguistic benchmarks.<n>We conclude that non-language data can serve as a partial substitute for language data for the training objective of next-token prediction but does not reliably support broader linguistic generalization.
• Score: 3.3178771329842185
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Efficient language learning requires methods to reduce the reliance on large data and computational resources. We investigate structural transfer: First training models on non-language data to induce useful priors for natural language. This approach is a form of weight initialization for multilingual language modeling. We evaluate transfer via next-token-prediction loss, weight shifts in the model, and downstream linguistic benchmarks. Several symbolic data types - notably music, probabilistic grammars, and cellular automata - yield lower language-modeling loss than random initialization. These gains coincide with smaller weight shifts during subsequent language training, suggesting that structural transfer positions models in a more favorable region of the parameter space. However, a lower loss does not translate consistently into better downstream linguistic performance, and transfer from non-language data is less efficient than additional language data. We conclude that non-language data can serve as a partial substitute for language data for the training objective of next-token prediction but does not reliably support broader linguistic generalization.

Related papers

• Training Bilingual LMs with Data Constraints in the Targeted Language [17.623676545426477]
  We study how to boost pretrained model performance in a target language with insufficient pretraining data.<n>We study this by quantifying the performance gap between training with data in a data-rich auxiliary language compared with training in the target language.
  arXiv  Detail & Related papers  (2024-11-20T02:27:40Z)
• Self-Translate-Train: Enhancing Cross-Lingual Transfer of Large Language Models via Inherent Capability [31.025371443719404]
  Self-Translate-Train is a method that lets large language models translate training data into the target language and fine-tunes the model on its own generated data. By demonstrating that Self-Translate-Train outperforms zero-shot transfer, we encourage further exploration of better methods to elicit cross-lingual capabilities of LLMs.
  arXiv  Detail & Related papers  (2024-06-29T14:40:23Z)
• Continual Learning Under Language Shift [6.0783165755651325]
  We study the pros and cons of updating a language model when new data comes from new languages. We investigate how forward and backward transfer effects depend on pre-training order and characteristics of languages.
  arXiv  Detail & Related papers  (2023-11-02T12:54:50Z)
• Soft Language Clustering for Multilingual Model Pre-training [57.18058739931463]
  We propose XLM-P, which contextually retrieves prompts as flexible guidance for encoding instances conditionally. Our XLM-P enables (1) lightweight modeling of language-invariant and language-specific knowledge across languages, and (2) easy integration with other multilingual pre-training methods.
  arXiv  Detail & Related papers  (2023-06-13T08:08:08Z)
• Improving Cross-lingual Information Retrieval on Low-Resource Languages via Optimal Transport Distillation [21.057178077747754]
  In this work, we propose OPTICAL: Optimal Transport distillation for low-resource Cross-lingual information retrieval. By separating the cross-lingual knowledge from knowledge of query document matching, OPTICAL only needs bitext data for distillation training. Experimental results show that, with minimal training data, OPTICAL significantly outperforms strong baselines on low-resource languages.
  arXiv  Detail & Related papers  (2023-01-29T22:30:36Z)
• Efficient Language Model Training through Cross-Lingual and Progressive Transfer Learning [0.7612676127275795]
  Most Transformer language models are pretrained on English text. As model sizes grow, the performance gap between English and other languages increases even further. We introduce a cross-lingual and progressive transfer learning approach, called CLP-Transfer.
  arXiv  Detail & Related papers  (2023-01-23T18:56:12Z)
• Language Agnostic Multilingual Information Retrieval with Contrastive Learning [59.26316111760971]
  We present an effective method to train multilingual information retrieval systems. We leverage parallel and non-parallel corpora to improve the pretrained multilingual language models. Our model can work well even with a small number of parallel sentences.
  arXiv  Detail & Related papers  (2022-10-12T23:53:50Z)
• Language Contamination Explains the Cross-lingual Capabilities of English Pretrained Models [79.38278330678965]
  We find that common English pretraining corpora contain significant amounts of non-English text. This leads to hundreds of millions of foreign language tokens in large-scale datasets. We then demonstrate that even these small percentages of non-English data facilitate cross-lingual transfer for models trained on them.
  arXiv  Detail & Related papers  (2022-04-17T23:56:54Z)
• Continual Learning in Multilingual NMT via Language-Specific Embeddings [92.91823064720232]
  It consists in replacing the shared vocabulary with a small language-specific vocabulary and fine-tuning the new embeddings on the new language's parallel data. Because the parameters of the original model are not modified, its performance on the initial languages does not degrade.
  arXiv  Detail & Related papers  (2021-10-20T10:38:57Z)
• Towards Zero-shot Language Modeling [90.80124496312274]
  We construct a neural model that is inductively biased towards learning human languages. We infer this distribution from a sample of typologically diverse training languages. We harness additional language-specific side information as distant supervision for held-out languages.
  arXiv  Detail & Related papers  (2021-08-06T23:49:18Z)
• Pre-Training a Language Model Without Human Language [74.11825654535895]
  We study how the intrinsic nature of pre-training data contributes to the fine-tuned downstream performance. We find that models pre-trained on unstructured data beat those trained directly from scratch on downstream tasks. To our great astonishment, we uncover that pre-training on certain non-human language data gives GLUE performance close to performance pre-trained on another non-English language.
  arXiv  Detail & Related papers  (2020-12-22T13:38:06Z)
• Comparison of Interactive Knowledge Base Spelling Correction Models for Low-Resource Languages [81.90356787324481]
  Spelling normalization for low resource languages is a challenging task because the patterns are hard to predict. This work shows a comparison of a neural model and character language models with varying amounts on target language data. Our usage scenario is interactive correction with nearly zero amounts of training examples, improving models as more data is collected.
  arXiv  Detail & Related papers  (2020-10-20T17:31:07Z)

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.