KT Corp Unveils Mi:dm, South Korea's First Large-Scale Language Model Outpacing GPT-3

South Korea's leading mobile carrier, KT Corp., introduced its cutting-edge artificial intelligence (AI) model, Mi:dm, on Tuesday. With plans to expand its AI business to the manufacturing, financial, education, public sectors, and the global market, KT aims to revolutionize the field.

Enhanced Capabilities for Optimal Performance

Developed by a Korean company, Mi:dm stands as the country's first large-scale language model (LLM). Its training involved over 1 trillion tokens, making it a remarkable feat in AI language processing.

Yonhap News Agency reported that Mi:dm offers four distinct models, each encompassing different parameter sizes. Ranging from a "basic" version, with around 7 billion parameters, to a "large" version, containing approximately 200 billion parameters, users can select the model that best suits their specific requirements.

In terms of parameters, Mi:dm's large-scale model notably surpasses OpenAI's renowned GPT-3, which boasts 175 billion parameters. The increased capacity of Mi:dm can potentially elevate the output and performance of AI-generated text to unprecedented levels.

KT plans to foster collaboration by offering other companies access to the foundational model of Mi:dm, as per KED Global. This comprehensive AI development package includes KT Cloud's hyperscale AI computing service and Rebellions Inc.'s neural processing unit infrastructure, facilitating the creation of diverse AI services.

Tackling AI Hallucinations

KT's technological breakthroughs include document AI, search AI, and factguard AI. These innovations effectively reduce AI hallucinations, minimizing false information generation by up to 70% compared to other generative AI services.

The document AI technology empowers Mi:dm to comprehend complex and structured documents efficiently. With this capability, users can tackle challenging texts with ease.

The search AI capability employs deep-learning technology to provide users with up-to-date information swiftly. Mi:dm ensures that users have access to the latest and most relevant data with remarkable efficiency.

Meanwhile, the Factguard AI guarantees accurate and reliable responses by generating content solely based on original documents. This feature ensures the highest level of integrity in the information that Mi:dm produces.

Photo: Mohamed Nohassi/Unsplash