Exploring Activation Patterns of Parameters in Language Models

TL;DR

This paper investigates internal parameter activation patterns in large language models using a gradient-based metric, revealing layer-specific behaviors and their relation to input domain and data relevance, with implications for model pruning and understanding.

Contribution

Introduces a gradient-based metric to analyze parameter activation, uncovering layer-specific activation behaviors and their correlation with input domain and data relevance in LLMs.

Findings

Shallow layers activate densely within the same domain

Deep layers show sparse activation and higher similarity across domains

Activation similarity in deep layers correlates with data relevance

Abstract

Most work treats large language models as black boxes without in-depth understanding of their internal working mechanism. In order to explain the internal representations of LLMs, we propose a gradient-based metric to assess the activation level of model parameters. Based on this metric, we obtain three preliminary findings. (1) When the inputs are in the same domain, parameters in the shallow layers will be activated densely, which means a larger portion of parameters will have great impacts on the outputs. In contrast, parameters in the deep layers are activated sparsely. (2) When the inputs are across different domains, parameters in shallow layers exhibit higher similarity in the activation behavior than deep layers. (3) In deep layers, the similarity of the distributions of activated parameters is positively correlated to the empirical data relevance. Further, we develop three validation experiments to solidify these findings. (1) Firstly, starting from the first finding, we attempt to configure different prune ratios for different layers, and find this method can benefit model pruning. (2) Secondly, we find that a pruned model based on one calibration set can better handle tasks related to the calibration task than those not related, which validate the second finding. (3) Thirdly, Based on the STS-B and SICK benchmark, we find that two sentences with consistent semantics tend to share similar parameter activation patterns in deep layers, which aligns with our third finding. Our work sheds light on the behavior of parameter activation in LLMs, and we hope these findings will have the potential to inspire more practical applications.