Achieving Sparse Activation in Small Language Models
This work addresses the challenge of efficient inference for SLMs, which is an incremental improvement over existing methods for LLMs.
The paper tackled the problem of applying sparse activation to Small Language Models (SLMs) to reduce computing costs, achieving an 80% sparsification ratio with less than 5% accuracy loss in experiments.
Sparse activation, which selectively activates only an input-dependent set of neurons in inference, is a useful technique to reduce the computing cost of Large Language Models (LLMs) without retraining or adaptation efforts. However, whether it can be applied to the recently emerging Small Language Models (SLMs) remains questionable, because SLMs are generally less over-parameterized than LLMs. In this paper, we aim to achieve sparse activation in SLMs. We first show that the existing sparse activation schemes in LLMs that build on neurons' output magnitudes cannot be applied to SLMs, and activating neurons based on their attribution scores is a better alternative. Further, we demonstrated and quantified the large errors of existing attribution metrics when being used for sparse activation, due to the interdependency among attribution scores of neurons across different layers. Based on these observations, we proposed a new attribution metric that can provably correct such errors and achieve precise sparse activation. Experiments over multiple popular SLMs and datasets show that our approach can achieve 80% sparsification ratio with <5% model accuracy loss, comparable to the sparse activation achieved in LLMs. The source code is available at: https://github.com/pittisl/Sparse-Activation.