Neural Activation Patterns Across Language Model Architectures: A Comprehensive Analysis of Cognitive Task Performance

TL;DR

This study systematically compares neural activation patterns across six large language model architectures on twelve cognitive tasks, revealing key differences in processing and neural behavior.

Contribution

It provides a comprehensive analysis of activation patterns across architectures and tasks, highlighting fundamental differences in neural processing.

Findings

Mathematical reasoning yields highest attention entropy across models.

Decoder models show significantly higher sparsity than encoder models.

Insights inform model selection and optimization for big data applications.

Abstract

This paper presents a comprehensive analysis of neural activation patterns across six distinct large language model (LLM) architectures, examining their performance on twelve cognitive task categories. Through systematic measurement of final activation values, attention entropy, and sparsity patterns, we reveal fundamental differences in how encoder and decoder architectures process diverse cognitive tasks. Our analysis of 144 task-model combinations demonstrates that mathematical reasoning consistently produces the highest attention entropy across all architectures, while decoder models exhibit significantly higher sparsity patterns compared to encoder models. The findings provide critical insights into the computational characteristics of modern language models and their task-specific neural behaviors, with implications for model selection and optimization in big data applications.