Multiverse Computing – CompactifAI
Multiverse Computing
Sector: Generalist
Business Case
The challenge facing CompactifAI is the need to significantly reduce the consumption of computational and energy resources required by LLMs (Large Language Models). These models are usually extremely large and complex, which leads to high operating costs and limits their use in devices with restricted computational capabilities. CompactifAI seeks to solve this problem by implementing quantum and quantum-inspired technologies to compress these models without losing efficiency, making artificial intelligence more accessible and sustainable.
Objectives
CompactifAI aims to revolutionise the efficiency of LLMs through advanced compression of these models using quantum and quantum-inspired technologies. It aims to drastically reduce resource and energy consumption, lower operating costs and enable the deployment of these models on computationally constrained devices, thus extending the reach and accessibility of AI-driven solutions.
Use case
CompactifAI proposes a solution based on the use of tensor networks and quantum technologies to compress LLMs, reducing their resource and energy consumption and associated costs. Building this solution involves the ongoing development of compression tools that will be progressively integrated into both on-premise and cloud devices, optimising public and proprietary models for a wide range of applications. This incremental approach ensures the adaptability and continuous improvement of the software, facilitating its adoption in diverse operating environments.
Infrastructure
Hybrid
Technology
Quantum
Data
Depending on the focus and specific application of the model, various datasets could be used
