El consumo eléctrico de la IA varía hasta 300x entre tareas
A team from the University of Michigan has developed ML.ENERGY, a tool that measures the energy consumption of AI models during inference tasks. Their findings reveal that energy usage can vary by up to 300 times depending on the task, with 80-90% of energy being consumed during inference rather than training. This project aims to provide clearer insights into the energy footprint of AI, which has been largely overlooked in existing benchmarks.
- ▪ML.ENERGY is an open-source benchmark that measures the electricity consumption of AI models.
- ▪The energy consumption varies significantly across different tasks, with some tasks using up to 300 times more energy than others.
- ▪The majority of energy in AI operations is spent on inference, not on training, challenging previous assumptions about AI's energy footprint.
Opening excerpt (first ~120 words) tap to expand
try { if(localStorage) { let currentUser = localStorage.getItem('current_user'); if (currentUser) { currentUser = JSON.parse(currentUser); if (currentUser.id === 806044) { document.getElementById('article-show-container').classList.add('current-user-is-article-author'); } } } } catch (e) { console.error(e); } lu1tr0n Posted on May 29 • Originally published at elsolitario.org El consumo eléctrico de la IA varía hasta 300x entre tareas #tutorial #ai #machinelearning #programming Durante años, la conversación sobre la huella eléctrica de la inteligencia artificial giró en torno al entrenamiento de modelos gigantes. Un equipo de la Universidad de Michigan acaba de mover el foco con datos: el consumo energético de la IA se concentra en la inferencia y varía de forma brutal según la tarea.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at DEV.to (Top).
Discussion
0 commentsMore from DEV.to (Top)
