Addition is All You Need for Energy-efficient Language Models · HackerPapers discussions

Addition is All You Need for Energy-efficient Language Models — mentioned by dogcomplex

Wed, 04 Dec 2024 02:24:21 GMT

I am favoriting this comment for reference later when I start poking around in the base level stuff. I find it pretty funny how simple this stuff can get. Have you messed with ternary computing inference yet? I imagine that shrinks the list even further - or at least reduces the compute requirements in favor of brute force addition. https://arxiv.org/html/2410.00907

Paper and discussion history. A mention is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — mentioned by djoldman

Sat, 19 Oct 2024 19:09:19 GMT

https://arxiv.org/abs/2410.00907 ABSTRACT Large neural networks spend most computation on floating point tensor multiplications. In this work, we find that a floating point multiplier can be approximated by one integer adder with high precision. We propose the linear-complexity multiplication (L-Mul) algorithm that approximates floating point number multiplication with integer addition operations. The new algorithm costs significantly less computation resource than 8-bit floating point multiplication but achieves higher precision. Compared to 8-bit floating point multiplications, the proposed…

Paper and discussion history. A mention is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — mentioned by ycnews

Sun, 13 Oct 2024 16:30:03 GMT

"Addition is All You Need for Energy-efficient Language Models" https://arxiv.org/abs/2410.00907

Paper and discussion history. A mention is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — HN submission

Wed, 09 Oct 2024 04:47:40 GMT

Addition is all you need for energy-efficient language models

334 points; 126 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — HN submission

Tue, 08 Oct 2024 13:27:54 GMT

Addition Is All You Need for Energy-Efficient Language Models

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — HN submission

Fri, 04 Oct 2024 08:08:58 GMT

MIT Researchers Unveil New Method to Improve LLM Inference Performance

9 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — HN submission

Thu, 03 Oct 2024 16:01:17 GMT

Addition is All You Need for Energy-efficient Language Models

7 points; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — HN submission

Wed, 02 Oct 2024 13:58:58 GMT

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

3 points; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Addition is All You Need for Energy-efficient Language Models — mentioned by datalto

Wed, 02 Oct 2024 13:13:38 GMT

Hi, We've been working on a solution to tackle one of AI’s growing challenges: energy consumption. Running large AI models, like those behind services such as ChatGPT, can use as much electricity as thousands of homes daily. This energy demand isn’t just costly—it has a massive environmental footprint. To address this, we developed an algorithm called L-Mul that aims to reduce AI energy consumption by up to 95%. The key idea is that most AI models rely heavily on multiplication operations, which are quite energy-intensive. L-Mul replaces many of these energy-hungry multiplications with much…

Paper and discussion history. A mention is not an endorsement.