OpenSource alternatives of Generative Artifical Intelligence for SME’s
DOI:
https://doi.org/10.17700/jai.2024.15.2.733Keywords:
LLM, SME, KNIME, Ollama, RAGAbstract
This paper investigates the potential of local Large Language Models (LLMs) for Small and Medium-sized Enterprises (SMEs). While cloud-based LLMs offer powerful capabilities, their associated costs, including subscription fees and token-based pricing, can be prohibitive for many SMEs. This research explores the benefits of developing and deploying custom, local LLM solutions, which offer advantages such as reduced operational costs, enhanced data privacy and security, and greater flexibility for customization and integration. The paper examines viable open-source alternatives to commercial LLMs, including Ollama, Gpt4All, and the integration of LLMs within the data science platform KNIME. Furthermore, it explores techniques for improving LLM performance, such as Retrieval-Augmented Generation (RAG) and Low-Rank Adaptation (LoRA). General properties of local LLM solutions, such as CLI and GUI options and multi-platform support, are also discussed. By embracing local LLM solutions, SMEs can leverage the power of AI while mitigating the challenges associated with cloud-based services. This approach empowers businesses to gain a competitive advantage, enhance operational efficiency, and drive innovation while maintaining control over their data and minimizing costs.