On the Geometry of Semantics in Next-token Prediction

• URL: http://arxiv.org/abs/2505.08348v1
• Date: Tue, 13 May 2025 08:46:04 GMT
• Title: On the Geometry of Semantics in Next-token Prediction
• Authors: Yize Zhao, Christos Thrampoulidis,
• Abstract summary: Modern language models capture linguistic meaning despite being trained solely through next-token prediction.<n>We investigate how this conceptually simple training objective leads models to extract and encode latent semantic and grammatical concepts.<n>Our work bridges distributional semantics, neural collapse geometry, and neural network training dynamics, providing insights into how NTP's implicit biases shape the emergence of meaning representations in language models.
• Score: 27.33243506775655
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Modern language models demonstrate a remarkable ability to capture linguistic meaning despite being trained solely through next-token prediction (NTP). We investigate how this conceptually simple training objective leads models to extract and encode latent semantic and grammatical concepts. Our analysis reveals that NTP optimization implicitly guides models to encode concepts via singular value decomposition (SVD) factors of a centered data-sparsity matrix that captures next-word co-occurrence patterns. While the model never explicitly constructs this matrix, learned word and context embeddings effectively factor it to capture linguistic structure. We find that the most important SVD factors are learned first during training, motivating the use of spectral clustering of embeddings to identify human-interpretable semantics, including both classical k-means and a new orthant-based method directly motivated by our interpretation of concepts. Overall, our work bridges distributional semantics, neural collapse geometry, and neural network training dynamics, providing insights into how NTP's implicit biases shape the emergence of meaning representations in language models.

Related papers

• Concept-Guided Interpretability via Neural Chunking [54.73787666584143]
  We show that neural networks exhibit patterns in their raw population activity that mirror regularities in the training data.<n>We propose three methods to extract these emerging entities, complementing each other based on label availability and dimensionality.<n>Our work points to a new direction for interpretability, one that harnesses both cognitive principles and the structure of naturalistic data.
  arXiv  Detail & Related papers  (2025-05-16T13:49:43Z)
• Tokens, the oft-overlooked appetizer: Large language models, the distributional hypothesis, and meaning [31.632816425798108]
  Tokenization is a necessary component within the current architecture of many language models.<n>We discuss how tokens and pretraining can act as a backdoor for bias and other unwanted content.<n>We relay evidence that the tokenization algorithm's objective function impacts the large language model's cognition.
  arXiv  Detail & Related papers  (2024-12-14T18:18:52Z)
• Collapsed Language Models Promote Fairness [88.48232731113306]
  We find that debiased language models exhibit collapsed alignment between token representations and word embeddings.<n>We design a principled fine-tuning method that can effectively improve fairness in a wide range of debiasing methods.
  arXiv  Detail & Related papers  (2024-10-06T13:09:48Z)
• Implicit Geometry of Next-token Prediction: From Language Sparsity Patterns to Model Representations [24.211603400355756]
  Next-token prediction (NTP) over large text corpora has become the go-to paradigm to train large language models.<n>We look at how NTP influences the mapping of linguistic patterns to geometric properties of the resulting model representations.<n>We validate our findings on synthetic and small-scale real language datasets.
  arXiv  Detail & Related papers  (2024-08-27T21:46:47Z)
• The Foundations of Tokenization: Statistical and Computational Concerns [51.370165245628975]
  Tokenization is a critical step in the NLP pipeline.<n>Despite its recognized importance as a standard representation method in NLP, the theoretical underpinnings of tokenization are not yet fully understood.<n>The present paper contributes to addressing this theoretical gap by proposing a unified formal framework for representing and analyzing tokenizer models.
  arXiv  Detail & Related papers  (2024-07-16T11:12:28Z)
• Interpreting Pretrained Language Models via Concept Bottlenecks [55.47515772358389]
  Pretrained language models (PLMs) have made significant strides in various natural language processing tasks. The lack of interpretability due to their black-box'' nature poses challenges for responsible implementation. We propose a novel approach to interpreting PLMs by employing high-level, meaningful concepts that are easily understandable for humans.
  arXiv  Detail & Related papers  (2023-11-08T20:41:18Z)
• Constructing Word-Context-Coupled Space Aligned with Associative Knowledge Relations for Interpretable Language Modeling [0.0]
  The black-box structure of the deep neural network in pre-trained language models seriously limits the interpretability of the language modeling process. A Word-Context-Coupled Space (W2CSpace) is proposed by introducing the alignment processing between uninterpretable neural representation and interpretable statistical logic. Our language model can achieve better performance and highly credible interpretable ability compared to related state-of-the-art methods.
  arXiv  Detail & Related papers  (2023-05-19T09:26:02Z)
• Testing Pre-trained Language Models' Understanding of Distributivity via Causal Mediation Analysis [13.07356367140208]
  We introduce DistNLI, a new diagnostic dataset for natural language inference. We find that the extent of models' understanding is associated with model size and vocabulary size.
  arXiv  Detail & Related papers  (2022-09-11T00:33:28Z)
• Pretraining on Interactions for Learning Grounded Affordance Representations [22.290431852705662]
  We train a neural network to predict objects' trajectories in a simulated interaction. We show that our network's latent representations differentiate between both observed and unobserved affordances. Our results suggest a way in which modern deep learning approaches to grounded language learning can be integrated with traditional formal semantic notions of lexical representations.
  arXiv  Detail & Related papers  (2022-07-05T19:19:53Z)
• A Latent-Variable Model for Intrinsic Probing [93.62808331764072]
  We propose a novel latent-variable formulation for constructing intrinsic probes. We find empirical evidence that pre-trained representations develop a cross-lingually entangled notion of morphosyntax.
  arXiv  Detail & Related papers  (2022-01-20T15:01:12Z)
• Infusing Finetuning with Semantic Dependencies [62.37697048781823]
  We show that, unlike syntax, semantics is not brought to the surface by today's pretrained models. We then use convolutional graph encoders to explicitly incorporate semantic parses into task-specific finetuning.
  arXiv  Detail & Related papers  (2020-12-10T01:27:24Z)
• Explaining and Improving Model Behavior with k Nearest Neighbor Representations [107.24850861390196]
  We propose using k nearest neighbor representations to identify training examples responsible for a model's predictions. We show that kNN representations are effective at uncovering learned spurious associations. Our results indicate that the kNN approach makes the finetuned model more robust to adversarial inputs.
  arXiv  Detail & Related papers  (2020-10-18T16:55:25Z)

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.