One Search Fits All: Pareto-Optimal Eco-Friendly Model Selection

2025-05-02

AI Summary

The high energy consumption of AI model training necessitates general-purpose optimization strategies that jointly maximize performance and energy efficiency. This paper proposes GREEN, the first eco-friendly model selection framework spanning computer vision (CV), natural language processing (NLP), and recommender systems. It introduces EcoTaskSet—the first large-scale, multi-task training dynamics dataset comprising 1,767+ experiments—and designs a lightweight, retraining-free configuration evaluator enabling inference-time dynamic recommendation of Pareto-optimal model configurations. By integrating predictive modeling with multi-objective trade-off reasoning, GREEN achieves joint energy-efficiency–performance optimization. Experiments demonstrate that GREEN reduces GPU-kWh energy consumption by 38% on average across tasks while preserving ≥99.2% of baseline accuracy; moreover, 100% of its recommended configurations lie precisely on the empirical Pareto frontier.

Technology Category

• Machine Learning: Efficient ML / Green AI
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

The environmental impact of Artificial Intelligence (AI) is emerging as a significant global concern, particularly regarding model training. In this paper, we introduce GREEN (Guided Recommendations of Energy-Efficient Networks), a novel, inference-time approach for recommending Pareto-optimal AI model configurations that optimize validation performance and energy consumption across diverse AI domains and tasks. Our approach directly addresses the limitations of current eco-efficient neural architecture search methods, which are often restricted to specific architectures or tasks. Central to this work is EcoTaskSet, a dataset comprising training dynamics from over 1767 experiments across computer vision, natural language processing, and recommendation systems using both widely used and cutting-edge architectures. Leveraging this dataset and a prediction model, our approach demonstrates effectiveness in selecting the best model configuration based on user preferences. Experimental results show that our method successfully identifies energy-efficient configurations while ensuring competitive performance.

Problem

• Optimizing AI model configurations for energy efficiency and performance
• Overcoming limitations of current eco-efficient neural architecture search methods
• Providing a dataset and model for cross-domain eco-friendly AI recommendations

Innovation

• Pareto-optimal AI model configurations recommendation
• EcoTaskSet dataset with 1767 experiments data
• Energy-efficient and performance-competitive model selection