English

Stacking Small Language Models for Generalizability

Computation and Language 2024-10-22 v1 Artificial Intelligence Machine Learning

Abstract

Recent advances show that large language models (LLMs) generalize strong performance across different natural language benchmarks. However, the large size of LLMs makes training and inference expensive and impractical to run in resource-limited settings. This paper introduces a new approach called fine-tuning stacks of language models (FSLM), which involves stacking small language models (SLM) as an alternative to LLMs. By fine-tuning each SLM to perform a specific task, this approach breaks down high level reasoning into multiple lower-level steps that specific SLMs are responsible for. As a result, FSLM allows for lower training and inference costs, and also improves model interpretability as each SLM communicates with the subsequent one through natural language. By evaluating FSLM on common natural language benchmarks, this paper highlights promising early results toward generalizable performance using FSLM as a cost-effective alternative to LLMs.

Keywords

Cite

@article{arxiv.2410.15570,
  title  = {Stacking Small Language Models for Generalizability},
  author = {Laurence Liang},
  journal= {arXiv preprint arXiv:2410.15570},
  year   = {2024}
}