This is an AI translated post.
Flitto and Upstage Sign MOU for AI Language Data Construction
- Writing language: Korean
- •
-
Base country: All countries
- •
- Information Technology
0000-00-00 00:00:00
Select Language
Summarized by durumis AI
- Flitto and Upstage aim to improve the performance of large language models (LLM) through collaboration on collecting and building datasets of low-resource languages in the Asian region.
- The two companies plan to build datasets for various languages, including Korean, Thai, Japanese, Vietnamese, Lao, and Khmer, and use them to cooperate on the localization of large language models and the construction of small language models (sLLM) for enterprises.
- This collaboration is expected to make a positive contribution to the domestic generative AI ecosystem by incorporating high-quality data and advanced technologies.
Flitto and Upstage Collaborate to Enhance Competitiveness in the Enterprise Language Model Market by Gathering Low-Resource Language Data and Building Datasets in Asia
Flitto (CEO Jeongsu Lee, KOSDAQ 300080), an AI language data company, announced on the 9th that it has signed a Memorandum of Understanding (MOU) with Upstage, an AI technology company, for AI language data construction.
The two companies will cooperate in collecting low-resource languages such as Thai, Japanese, Vietnamese, Lao, and Khmer and building datasets to enhance the performance of large language models (LLMs).
The agreement includes collaboration on: △ building benchmark datasets for the Korean LLM leaderboard 'Ko-LLM', △ operating a multilingual LLM leaderboard, △ building low-resource language data and localizing LLMs using low-resource languages, and △ strengthening data supply partnerships for the construction of small LLMs (sLLMs) for enterprise adoption. This collaboration is expected to yield tangible results in the future, as it leverages the AI development strengths of both companies based on language.
Based on this agreement, Flitto plans to actively contribute to the advancement of language models by creating synergy between language data and AI technology, and to enhance its language collection technology competitiveness by utilizing its know-how in building multilingual parallel corpora and datasets of text, images, and audio that are free of copyright issues.
Through this collaboration, Upstage plans to acquire high-quality low-resource language data and further expand the supported languages of its pre-trained LLM Sola to develop customized models specialized in languages of various regions, including Southeast Asia. Sola currently supports Korean and English, and plans to expand its support languages to Japanese and Thai by the end of the year.
"The field of low-resource language learning has emerged as a key factor in the performance of large language models," said Jeongsu Lee, CEO of Flitto. "Through collaboration with Upstage, we aim to demonstrate how the combination of high-quality data and advanced technology can positively contribute to the domestic generative AI ecosystem."
"As the generative AI boom triggered by language models is shaking the world, securing high-quality language data is essential," said Sunghun Kim, CEO of Upstage. "Through this collaboration with Flitto, Upstage will work to refine data so that more people around the world can experience the innovation of generative AI."
Website: https://www.flitto.com/
Contact
Flitto
Marketing Team
Manager Harmony
