Li Auto Had Already Initiated AI Language Model Training Prior to the Debut of ChatGPT

Li Auto Had Already Initiated AI Language Model Training Prior to the Debut of ChatGPT

On June 19th, Li Auto‘s official WeChat account “You Ge Li Xiang” released an article introducing the company’s development in the field of AI large language models. Chen Wei, head of Li Auto‘s spatial algorithm team, stated: “We had already started training for the entire large language model before ChatGPT was introduced.

Chen Wei revealed: “As early as the formation of Li Auto‘s spatial algorithm team, we built a data platform and training platform based on large-scale model training. At the same time, we also considered the demand for multimodal data, including speech, vision and text.” He also stated that Li Auto‘s large models have four characteristics that lead the industry: “powerful text generation ability”, “language understanding ability”, “knowledge question answering ability” and “logical reasoning ability”.

He added, “The future trend of algorithm evolution will move towards large models. ChatGPT has proven this point and brought a cognitive breakthrough to the entire field of general artificial intelligence. Our direction has always been clear: to make language models that represent underlying infrastructure capabilities larger and larger, which will also create more value for users.”

At the recent Li Auto Inc. Hosts Family Tech Day, Li Auto announced the launch of a city NOA product that does not rely on high-precision maps. With the help of AI large models, this product will achieve driving performance close to that of human drivers.

SEE ALSO: Li Xiang: Li Auto Poised to Outpace BBA in Sales with L6 Delivery Next Year

Characteristics of the NOA technology architecture in Li Auo city:

Firstly, by using NPN characteristics and TIN network to enhance the BEV model, Li Auto is able to recognize objects without relying on high-precision maps. Secondly, Li Auto uses imitation learning to make more human-like decisions with its regulatory algorithms. Thirdly, a fully automated and closed-loop training platform supports continuous evolution of the large-scale model.

At the Li Auto Inc. Hosts Family Tech Day, Li Auto also announced that it will begin internal testing of NOA in Beijing and Shanghai this month. In addition, it plans to release commuting NOA functions in the second half of this year as well as expand NOA coverage to more cities.

Sign up today for 5 free articles monthly!

>>> Read full article>>>
Copyright for syndicated content belongs to the linked Source : Pandaily – https://pandaily.com/li-auto-had-already-initiated-ai-language-model-training-prior-to-the-debut-of-chatgpt/

Exit mobile version