AI tech company Crowdworks announced on the 24th that it is joining the 'National AI' project to achieve AI technology independence in Korea.
The Motif Technologies Consortium was selected for the Ministry of Science and ICT's additional public offering for the development of its own AI Foundation Model (Dokpamo), and Crowdworks will participate as a key data provider for the consortium. Along with LG AI Research Center, SK Telecom, and Upstage, the consortium will be part of four elite teams leading the domestic AI ecosystem.
Crowdworks will be responsible for providing the high-quality data required to develop a 300 billion (300B) parameter inference-type large-scale language model (LLM). Specifically, it plans to strengthen the model's logical reasoning and expert knowledge acquisition capabilities by building a chain-of-thought learning dataset.
To achieve this, Alpy utilizes its proprietary, unstructured document data preprocessing solution, "Alpy Knowledge Compiler," to convert complex documents, such as tables and charts, into AI-ready data. This technology is expected to play a vital role in ensuring stable model learning and step-by-step inference performance.
Crowdworks has previously proven its practical capabilities by providing training data to elite domestic teams, building datasets for understanding books and broadcast videos. The company stated, "By leveraging the data purification technology and high-quality data construction expertise accumulated through collaborations with domestic big tech companies, we will fully support the development of proprietary AI models."
- See more related articles
You must be logged in to post a comment.