Mutagenesis screen to map the functions of parameters of Large Language Models

TL;DR

This study applies a biological mutagenesis-inspired approach to systematically explore how parameters in large language models influence their functionalities, revealing complex structures and sensitivities within the models.

Contribution

Introduces a mutagenesis screening method for LLMs to analyze parameter-function relationships, uncovering fine structures and mutation sensitivities not previously documented.

Findings

Mutagenesis reveals diverse sensitivities across model matrices.

Mutations cluster along axes, indicating structured parameter-function relationships.

Certain mutations lead to specific output styles, such as poetic or conversational.

Abstract

Large Language Models (LLMs) have significantly advanced artificial intelligence, excelling in numerous tasks. Although the functionality of a model is inherently tied to its parameters, a systematic method for exploring the connections between the parameters and the functionality are lacking. Models sharing similar structure and parameter counts exhibit significant performance disparities across various tasks, prompting investigations into the varying patterns that govern their performance. We adopted a mutagenesis screen approach inspired by the methods used in biological studies, to investigate Llama2-7b and Zephyr. This technique involved mutating elements within the models' matrices to their maximum or minimum values to examine the relationship between model parameters and their functionalities. Our research uncovered multiple levels of fine structures within both models. Many matrices showed a mixture of maximum and minimum mutations following mutagenesis, but others were predominantly sensitive to one type. Notably, mutations that produced phenotypes, especially those with severe outcomes, tended to cluster along axes. Additionally, the location of maximum and minimum mutations often displayed a complementary pattern on matrix in both models, with the Gate matrix showing a unique two-dimensional asymmetry after rearrangement. In Zephyr, certain mutations consistently resulted in poetic or conversational rather than descriptive outputs. These "writer" mutations grouped according to the high-frequency initial word of the output, with a marked tendency to share the row coordinate even when they are in different matrices. Our findings affirm that the mutagenesis screen is an effective tool for deciphering the complexities of large language models and identifying unexpected ways to expand their potential, providing deeper insights into the foundational aspects of AI systems.