L-Mul: A New Algorithm for 95% AI Energy Reduction Without Performance Sacrifice

We’ve developed an algorithm called L-Mul to tackle AI’s growing energy consumption problem. Large AI models, like those behind ChatGPT, consume enormous amounts of electricity daily. L-Mul reduces energy use by up to 95% by replacing energy-intensive multiplication operations with simpler addition operations, without compromising performance. We’re excited to share this approach with the community, as it could make AI development more sustainable and cost-effective.