Felix Pinkston
May 31, 2025 13:53
NVIDIA Deep Learning Institute introduces multilingual AI workshops in GTC Paris to solve language models and improve domain stars.
According to the company’s blog post, NVIDIA is improving AI education offering with a new multilingual workshop at the upcoming GTC Paris event. This initiative from NVIDIA Deep Learning Institute aims to develop the technology for developers to convert the Open-Source Lange Lange Models to domain specialization and true multilingual AI assets.
Multilingual problem solving
Despite the function of LLM in recognizing and creating content, it is often lacking in dealing with professional business knowledge or various languages and cultural contexts. Many models in multi -language can mainly be trained in English, leading to inaccuracy and cultural insensitive. NVIDIA’s new workshop, “Add New Knowledge to LLM,” will solve this problem by providing training on model adaptation to improve the accuracy of various languages and sectors.
The importance of evaluation of multilingual models
Evaluating models in multiple languages is important for development and customization. Workshop emphasizes the importance of multilingual evaluations that prefer many Europeans to use their native language over English. Models like LLAMA 2 have been trained for less than 5% English data, and stress that strict multilingual evaluations are needed to avoid misunderstandings and expensive distribution errors.
Workshop overview and major work
“Add New Knowledge to LLMS” workshop is a lecturer -led session all day, guiding the participants through the life cycle of AI model customization. Major tasks include systematic evaluation and data set generation, advanced data cue, target knowledge injection and model optimization for domains and languages. Participants use tools such as NVIDIA NEMO evaluators and NEMO model optimization to learn how to improve the multilingual function and domain performance of the model.
Actual application and influence
NVIDIA’s efforts with global organizations have already been greatly improved in multilateral AI. Partnerships with entities such as Barcelona Super Computing Center and EUROLLM have improved their work accuracy and improved the development of powerful multilingual models. This development emphasizes the importance of NVIDIA’s initiatives in developing multilingual AI functions.
For those who are interested in the latest AI innovation, the GTC Paris event offers opportunities to participate in this workshop and other related sessions. For more information about events and registration, you can find it on the NVIDIA blog.
Image Source: Shutter Stock