The Hidden Environmental Cost of Poor Coding Practices in TensorFlow and Keras Applications: A Study on Resource Leaks and Carbon Emissions

2026-06-18 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, quick

Summary

An initial empirical investigation quantifies the hidden environmental cost of poor coding practices in TensorFlow and Keras applications, specifically focusing on resource leaks. This study examined two common resource-leak "smells": Improper Model Reuse (IMR) and Unreleased Tensor References (UTR). Controlled experiments, executing identical training tasks against a smell-free baseline, revealed that both IMR and UTR significantly increase estimated electricity usage and carbon emissions. IMR led to an approximate 32% increase in electricity consumption, while UTR resulted in a 46% increase, with proportional rises in CO2 emissions. Statistical tests confirmed these differences are systematic and significant, providing initial evidence that resource-leak smells degrade ML energy efficiency and environmental sustainability. These findings highlight measurable risks to software quality and sustainability.

Key takeaway

For MLOps Engineers and AI Scientists focused on sustainable development, addressing resource leaks in your TensorFlow and Keras applications is crucial. Improper Model Reuse and Unreleased Tensor References can inflate your electricity consumption by 32% and 46% respectively, directly impacting operational costs and carbon footprint. Prioritize integrating robust resource-lifecycle management into your ML development workflows to mitigate these measurable risks to both software quality and environmental sustainability.

Key insights

Resource leaks in TensorFlow/Keras code significantly increase energy consumption and CO2 emissions.

Principles

• Improper Model Reuse (IMR) elevates electricity use by ~32%.
• Unreleased Tensor References (UTR) boost electricity use by ~46%.
• Resource leaks degrade ML energy efficiency.

Method

Controlled experiments compared training tasks with and without resource-leak "smells" (IMR, UTR) against a baseline to quantify electricity and CO2 impact.

In practice

• Identify and fix Improper Model Reuse.
• Ensure Unreleased Tensor References are cleared.
• Integrate resource-lifecycle management.

Topics

• TensorFlow
• Keras
• Resource Leaks
• Energy Efficiency
• Carbon Emissions
• MLOps

Best for: AI Engineer, Research Scientist, Machine Learning Engineer, AI Scientist, MLOps Engineer

Related on AIssential

• The Hidden Environmental Cost of Poor Coding Practices in TensorFlow and Keras Applications: A Study on Resource Leaks and Carbon Emissions — Resource-leak code smells in ML applications significantly increase energy consumption and CO2 emissions.
• Sustainable AI: A Beginners Guide — Sustainable AI minimizes environmental harm by optimizing energy efficiency, renewable energy sourcing, and transparency.
• Standardized Methods and Recommendations for Green Federated Learning — A standardized carbon accounting method for Federated Learning is crucial for reproducible "green" FL evaluation.
• SOLAR: Communication-Efficient Model Adaptation via Subspace-Oriented Latent Adapter Reparametrization — SOLAR compresses PEFT adapters by reparameterizing updates using foundation model singular vectors, reducing communication costs.
• Neil Thompson: Is there a 'Secret Sauce' driving AL Model development? — AI model development combines systematic efficiency with significant, art-like variation in outcomes.
• The AI Skill Erosion Crisis: How AI Tools Are Making Us Worse at the Thing They’re Supposed to Help… — AI tools may degrade human skills and understanding, despite industry promises of efficiency gains.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.