Understanding the Emergence of Seemingly Useless Features in Next-Token Predictors

ICLR 2026 Conference Submission18169 Authors

Keywords: next-token prediction, transformers, interpretability

Abstract: Trained Transformers have been shown to compute abstract features that appear redundant for predicting the immediate next token. We identify which components of the gradient signal from the next-token prediction objective give rise to this phenomenon, and we propose a method to estimate the influence of those components on the emergence of specific features. After validating our approach on toy tasks, we use it to interpret the origins of the world model in OthelloGPT and syntactic features in a small language model. Finally, we apply our framework to a pretrained LLM, showing that features with stronger influence on future tokens than on the next token tend to be related to formal reasoning domains such as code. Overall, our work takes a step toward understanding hidden features of Transformers through the lens of their development during training.

Primary Area: interpretability and explainable AI

Submission Number: 18169