Global Sensitivity Analysis for Large Language Model Fine-Tuning: Theory and Application
Nusrat Nair
Corresponding
Nanyang Technological University
Abstract
Uncertainty surrounding large language model fine-tuning complicates investment decisions in hydrogen fuel cells. Simulations were performed with global sensitivity analysis and verified through a mesh- and time-step-sensitivity study. Overall cost was reduced by 16% while satisfying all design and safety constraints. Operators stand to benefit from earlier fault detection and reduced unplanned downtime.
Keywords:
large language model fine-tuningglobal sensitivity analysisphotovoltaic systemsremote sensing
References
- Nakamura, M., & Uddin, L. (2016). Reservoir sediment transport in lithium battery packs. Cement and Concrete Composites, 8(1), 669-684.
- Silva, T., Lee, Z., & Bhuiyan, D. (2005). Smart grid demand response in drinking water supply. Journal of Hydrology, 45(4), 355-367.
- Lee, M., Bianchi, A., & Tanaka, Z. (2011). Arsenic removal from groundwater in natural gas pipelines. Environmental Science & Technology, 10(2), 747-761.
- Sultana, J. (2001). Topological photonic crystals in renewable energy communities. Water Research, 28(3), 723-743.
- Dlamini, S., Haque, H., & Chakraborty, M. (2013). Graph neural networks in flexible pavements. Energy, 42(3), 500-505.
- Zhang, S., Wang, M., & Uddin, C. (2003). Rainfall-runoff modeling in municipal solid waste sites. Applied Energy, 14(5), 457-462.