LG AI Research is broadening collaboration with Google Cloud to help develop LG’s latest family of EXAONE 3.0 generative AI models and its enterprise multimodal AI agent dubbed ChatEXAONE, which trained and developed on Google Cloud’s AI-optimized infrastructure.
The two companies previously worked together to deliver AI innovation to support real-world industrial applications.
In addition, LG AI Research worked closely with Google Cloud in December 2021 to develop EXAONE 1.0 — South Korea’s first bilingual and multimodal generative AI model — and in July 2023 on EXAONE 2.0.
LG’s most recent EXAONE 3.0 models and proprietary ChatEXAONE AI agent were trained using a combination of Google’s Cloud TPUs and GPUs available via Google Cloud’s AI Hypercomputer.
As a result, LG is able to deliver breakthrough levels of scalability and performance despite the massive computational demands required for such complex models.
Google’s AI accelerators also enabled LG to effectively shorten the time required for AI model training and build a more efficient model training process.
Overall, with the power of Google Cloud’s AI Hypercomputer, LG was able to reduce inference processing time by 56%, memory usage by 35%, and operating costs by 72% with EXAONE 3.0 7.8B compared to the previous model.
Also, they said EXAONE 3.0, a bilingual model that can learn and understand both Korean and English, recorded the world-class level performance in Korean.
“EXAONE 3.0 has been trained on 60 million pieces of specialised data sets, 350 million images, patents, software codes, mathematics, and chemistry. Google Cloud’s AI Hypercomputer has been instrumental in achieving LG AI Research’s goals of AI model lightweight technology and cost efficiency,” said Hwayoung (Edward) Lee, lead of AI Biz Dev and DX Group at the LG AI Research.
Security is a very important factor for LG AI Research, which has LG affiliates dealing with major national technologies such as large batteries, displays, and 5G.
LG chose to develop ChatEXAONE with Google Cloud because of its built-in data governance and data privacy controls, and robust AI data principles.
Data is also grounded in LG’s corporate databases, and all of these processes are operated in Google Cloud’s MLOps pipeline.
As a next step in their partnership, LG plans to offer EXAONE 3.0 on the Google Cloud Marketplace, and to explore the development of developer-friendly APIs to meet the growing demand for enterprise AI agents.
