The Cost of Compression: Investigating the Impact of Compression on Parametric Knowledge in Language Models

Satya Sai Srinath Namburi GNVV, Makesh Narsimhan Sreedhar, Srinath Srinivasan, Frederic Sala

Submission Type: Regular Long Paper

Submission Track: Efficient Methods for NLP

Submission Track 2: Interpretability, Interactivity, and Analysis of Models for NLP

Keywords: Pruning, Quantization, Language Models

TL;DR: The paper is about understanding the impact of various compression techniques (pruning, quantization) on parametric knowledge and the analysis is done on all families of transformers/LLMs at various scales.

Abstract:

Compressing large language models (LLMs), often consisting of billions of parameters, provides faster inference, smaller memory footprints, and enables local deployment.
The standard compression techniques are pruning and quantization, with the former eliminating redundant connections in model layers and the latter representing model parameters with as little as 4 bits. The key tradeoff is between the degree of compression and the impact on the quality of the compressed model. Existing research on LLM compression primarily focuses on performance in terms of general metrics like perplexity or downstream task accuracy. More fine-grained metrics, such as those measuring parametric knowledge, remain significantly underexplored. To help bridge this gap, we present a comprehensive analysis across multiple model families using the LAMA and LM-Harness benchmarks in order to systematically quantify the effect of commonly employed compression techniques on model performance. A particular focus is on tradeoffs involving parametric knowledge, with the goal of providing practitioners with practical insights to make informed decisions on compression.

Submission Number: 2161