The emergence of number and syntax units in LSTM language models

TL;DR

This paper investigates how LSTM language models internally encode number and syntax, revealing that specific neurons are responsible for long-distance number agreement and are influenced by other syntax-tracking units, indicating genuine syntactic processing.

Contribution

It provides a detailed neuron-level analysis showing that LSTMs implement mechanisms for syntactic processing, challenging the view that they rely solely on heuristics.

Findings

Number information is managed by two key units.

These units are influenced by other syntax-tracking neurons.

LSTMs exhibit mechanisms consistent with syntactic processing.

Abstract

Recent work has shown that LSTMs trained on a generic language modeling objective capture syntax-sensitive generalizations such as long-distance number agreement. We have however no mechanistic understanding of how they accomplish this remarkable feat. Some have conjectured it depends on heuristics that do not truly take hierarchical structure into account. We present here a detailed study of the inner mechanics of number tracking in LSTMs at the single neuron level. We discover that long-distance number information is largely managed by two `number units'. Importantly, the behaviour of these units is partially controlled by other units independently shown to track syntactic structure. We conclude that LSTMs are, to some extent, implementing genuinely syntactic processing mechanisms, paving the way to a more general understanding of grammatical encoding in LSTMs.