theregister

Want to save the planet from AI? Chuck in an FPGA and ditch the matrix

Large language models can be made 50 times more energy efficient with alternative math and custom hardware, claim researchers at University of California Santa Cruz. In a paper titled, "Scalable ...
news.ucsc

Researchers run high-performing large language model on the energy needed to power a lightbulb

Large language models such as ChaptGPT have proven to be able to produce remarkably intelligent results, but the energy and monetary costs associated with running these massive algorithms is sky high.
marktechpost

ShiftAddLLM: Accelerating Pretrained LLMs through Post-Training Shift-and-Add Reparameterization: Creating Efficient Multiplication-Free Models

Deploying large language models (LLMs) on resource-constrained devices presents significant challenges due to their extensive parameters and reliance on dense multiplication operations. This results ...
syncedreview

Matrix Multiplication-Free Language Models Maintain Top-Tier Performance at Billion-Parameter Scales

Matrix multiplication (MatMul) is a fundamental operation in most neural networks, primarily because GPUs are highly optimized for these computations. Despite its critical role in deep learning, ...
marktechpost

This AI Research Discusses Achieving Efficient Large Language Models (LLMs) by Eliminating Matrix Multiplication for Scalable Performance

Most neural network topologies heavily rely on matrix multiplication (MatMul), primarily because it is essential to many basic processes. Vector-matrix multiplication (VMM) is commonly used by dense ...
Scientific Research Publishing

Maffia, A. and Mariotti, M.A. (2018) Intuitive and Formal Models of Whole Number Multiplication: Relations and Emerging Structures.

ABSTRACT: This paper provides a method of the process of computation called the cumulative method, it is based upon repeated cumulative process. The cumulative method is being adapted to the purposes ...