This is an AI translated post.
SK C&C Unveils 'Soluer LLMOps,' a Platform Supporting Customized sLLM Implementation for Clients
- Writing language: Korean
- •
-
Base country: All countries
- •
- Information Technology
0000-00-00 00:00:00
Select Language
Summarized by durumis AI
- SK C&C has launched 'Soluer LLMOps,' a platform that enables businesses to easily build customized small-scale large language models (sLLMs).
- The platform supports the combination and utilization of various external foundation models, applying hyperautomation to enhance efficiency across the entire process, from data collection to training, testing, and serving.
- Furthermore, it provides a user-friendly interface with drag-and-drop functionality, empowering even non-experts to easily generate and utilize sLLMs.
Leveraging various foundation models provided by commercial LLMs such as ChatGPT and ClovaX, as well as open-source LLMs
Implementing hyper-automation in the entire process of sLLM creation, including data collection, preprocessing, learning, testing, and serving
Eliminating hallucination phenomena and providing sLLM testing functionality
Rapid sLLM creation through drag and drop, and ensuring sLLM reliability through periodic data parallel learning
An era is dawning where companies can self-implement customized AI services through sLLMs that freely combine various foundation models.
sLLMs are smaller than LLMs (large language models) such as ChatGPT, HyperClovaX, and Gemini, but they can be specialized in specific areas. They consume far fewer computing resources, reducing development costs and enhancing security, making sLLMs an efficient way to build dedicated AI services for businesses.
SK C&C (CEO: Yoon, 풍영, skcc.co.kr) announced today that it will be providing a platform called ‘Solur LLMOps’ (hereinafter referred to as ‘Solur LLMOps’) to help businesses easily and quickly implement customized sLLMs (small large language models).
SK C&C has incorporated into ‘Solur LLMOps’ the know-how accumulated in the process of building and operating generative AI services along with various clients in various industries such as finance, manufacturing, telecommunications, and services, including methods for applying generative AI foundation models, combining and learning enterprise data, and the entire process of implementing enterprise-customized sLLMs.
Foundation models learn structured and unstructured data to support various AI tasks such as language understanding, text and image generation, and natural language conversation.
‘Solur LLMOps’ first supports the combination and utilization of various external foundation models.
In addition to commercial LLMs such as OpenAI ChatGPT and Naver Cloud HyperClovaX, it can also utilize various foundation models created from open-source LLMs. It recommends foundation models necessary for building sLLMs that suit the characteristics of enterprise AI, and supports the process of selecting, combining, and utilizing them.
In particular, hyper-automation is applied to all aspects of the process, including △data collection and preprocessing △automatic learning data generation △learning using external foundation models △sLLM creation and testing, significantly enhancing efficiency and reducing costs in the process of building enterprise-customized sLLMs.
In practice, ‘Solur LLMOps’ automatically generates learning data by preprocessing unstructured data while collecting data owned by the company. It then uses the selected external generative AI foundation model for rapid learning and completes the sLLM that matches the business objectives of the company. It also provides an architecture that allows for optimized resource management using serverless learning resources.
‘Solur LLMOps’ also provides features for eliminating hallucination (hallucination) phenomena, where AI generates incorrect answers, and testing sLLMs.
It provides AI automation tools to handle hallucinations at each stage of data preprocessing, model creation, evaluation, and utilization, allowing even non-experts to create sLLMs with confidence.
You can perform completion tests through simple questions in the ‘Solur LLMOps’ chat window, and questions needed for the test are automatically provided.
In addition, ‘Solur LLMOps’ provides a user-friendly environment (UI) and experience (UX) for quick sLLM model creation and reliability assurance that perfectly suits the corporate work environment.
Users can repeatedly perform tasks from data refinement to model tuning and testing by simply dragging and dropping data on the web screen after selecting it. Multiple data processing and simultaneous parallel learning using multiple foundation models can be easily performed with just mouse operations.
This enables practitioners on the corporate scene to easily create the necessary sLLMs for their work and utilize them for various tasks.
SK C&C’s Chai Won, head of the G.AI Group, said, “The key features of Solur LLMOps are being applied to generative AI construction projects for financial institutions and SK affiliates.” He added, “Starting with this, we will actively work to spread the use of enterprise-customized sLLMs across domestic industries.”
Website: http://www.skcc.com
Contact
SK㈜ C&C
Public Relations Team
Ji Yoonjin, Manager
