The Paradox of Stochasticity: Limited Creativity and Computational Decoupling in Temperature-Varied LLM Outputs of Structured Fictional Data

TL;DR

This study investigates how temperature settings and model architecture influence the generation of structured fictional data in large language models, revealing architecture's dominant role over temperature in efficiency and output diversity.

Contribution

It provides a systematic analysis of temperature effects and model architecture on structured data generation, highlighting the limited impact of temperature and the importance of model choice.

Findings

Model architecture affects processing speed more than temperature.

Temperature has no significant impact on processing time.

Models tend to produce common names, limiting output diversity.

Abstract

This study examines how temperature settings and model architectures affect the generation of structured fictional data (names, birthdates) across three large language models (LLMs): llama3.1:8b, deepseek-r1:8b, and mistral:latest. By systematically testing temperature values from 0.0 to 1.0 in increments of 0.1, we conducted 330 trials yielding 889 structured entities, validated for syntactic consistency. Key findings reveal that model architecture significantly influences computational efficiency, with mistral:latest and llama3.1:8b processing data 8x faster than deepseek-r1:8b. Contrary to expectations, temperature showed no correlation with processing time, challenging assumptions about stochastic sampling costs. Output diversity remained limited, as models consistently defaulted to common name archetypes (e.g., 'John Doe' and 'Jane Smith') across all temperatures, though rare names clustered at intermediate values (0.3-0.7). These results demonstrate that architectural optimizations, rather than temperature adjustments, dominate performance in structured generation tasks. The findings emphasize prioritizing model selection over hyperparameter tuning for efficiency and suggest explicit diversity constraints are necessary to mitigate default output biases in synthetic data pipelines.