Neural Activation Patterns Across Language Model Architectures: A Comprehensive Analysis of Cognitive Task Performance
Provides insights for model selection and optimization in big data applications by characterizing task-specific neural behaviors across architectures.
This study analyzes neural activation patterns across six LLM architectures on twelve cognitive tasks, finding that mathematical reasoning yields highest attention entropy and decoder models show higher sparsity than encoders.
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.