LLM-guided optimization with enhanced prompts reaches energy-efficient inference settings in fewer iterations (avg 3.4) than baselines (avg 5.2) and beats Sobol sampling on convergence speed.
Zhang, Nishkrit Desai, Juhan Bae, Jonathan Lorraine, and Jimmy Ba
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
LLM-Guided Runtime Parameter Optimization for Energy-Efficient Model Inference
LLM-guided optimization with enhanced prompts reaches energy-efficient inference settings in fewer iterations (avg 3.4) than baselines (avg 5.2) and beats Sobol sampling on convergence speed.