Wang Xiaochuan China’s search engine pioneer also the founder of Sogou, said on Weibo in February, “China needs its own OpenAI.” The Chinese businessman is getting closer to his dream as his new company, Baichuan Intelligence, today released Baichuan-13B, its next-generation big language model.
People say that Baichuan is one of China’s most potential LLM developers because its founder was a computer science prodigy at Tsinghua University and started Sogou, which was later bought by Tencent.
China’s search engine pioneer Wang left Sogou at the end of 2021. As ChatGPT took the world by storm, the entrepreneur started Baichuan in April and quickly got $50 million in funding from a group of angel investors.
Like other Chinese LLMs, Baichuan is trained on both Chinese and English data. It has 13 billion parameters and is based on the Transformer design, which also supports GPT. (Parameters are factors that the model uses to make text and figure out what it means.) The GitHub page for the model says that it is open-source and made for business use.
Baichuan-13 has been trained with 1,400,000,000 coins. Meta’s LLaMa, on the other hand, has 13 billion parameters and uses 1 trillion coins. Wang previously said in an interview that his startup was on track to release a large-scale model similar to OpenAI’s GPT-3.5 by the end of this year.
Read Also;OpenAI’s ChatGPT App Can Now Search The Web — But Only via Bing
Baichuan has grown quickly for something that has only been going on for three months. The team had grown to 50 people by the end of April, and in June, it released its first LLM, the pre-training model Baichuan-7B, which has 7 billion features.
Now, academics and developers who have been given official permission to use the Baichuan-13B model for commercial reasons can use it for free. Importantly, the model can run on consumer-grade hardware like Nvidia’s 3090 video cards, even though the U.S. has put restrictions on China’s AI chips.
Read Also;OpenAI Makes GPT-4 Generally Available
The search engine giant Baidu, as well as Zhipu.ai, a spinoff of Tsinghua University led by Professor Tang Jie, and IDEA, a research institute run by Harry Shum, who co-founded Microsoft Research Asia, have also put a lot of money into large language models.
As China gets ready to adopt some of the strictest AI rules in the world, it is quickly making big language models. According to the Financial Times, China is expected to write rules for creative AI with a focus on content. This means that the rules that were put in place in April will be tightened. Before launching big language models, companies might also need to get a license. This could slow down China’s efforts to compete with the U.S. in this new industry.
Follow our socials Whatsapp, Facebook, Instagram, Twitter, and Google News.
