Neuron Empirical Gradient: Connecting Neurons’ Linear Controllability and Representational Capacity

ACL ARR 2024 December Submission1497 Authors

Abstract:

Although neurons in the feed-forward layers of pre-trained language models (PLMs) can store factual knowledge, most prior analyses remain qualitative, leaving the quantitative relationship among knowledge representation, neuron activations, and model output poorly understood.
In this study, by performing neuron-wise interventions using factual probing datasets, we first reveal the linear relationship between neuron activations and output token probabilities.
We refer to the gradient of this linear relationship as ``neuron empirical gradients.'' and propose NeurGrad, an efficient method for their calculation to facilitate quantitative neuron analysis.
We next investigate whether neuron empirical gradients in PLMs encode general task knowledge by probing skill neurons.
To this end, we introduce MCEval8k, a multi-choice knowledge evaluation benchmark spanning six genres and 22 tasks.
Our experiments confirm that neuron empirical gradients effectively capture knowledge, while skill neurons exhibit efficiency, generality, inclusivity, and interdependency.
These findings link knowledge to PLM outputs via neuron empirical gradients, shedding light on how PLMs store knowledge.
The code and dataset are released.

Paper Type: Long

Research Area: Interpretability and Analysis of Models for NLP

Research Area Keywords: Interpretability and Analysis of Models for NLP,Question Answering,Resources and Evaluation

Contribution Types: Model analysis & interpretability, Theory

Languages Studied: English, German, French, Spanish, Chinese

Submission Number: 1497