Welcome Here  Shenzhen Mingjiada Electronics Co., Ltd.

Just over three months after releasing ChatGPT, OpenAI has personally added more fuel to the big model craze.

In the early hours of March 15, Beijing time, OpenAI announced the birth of the multimodal Big Model GPT-4 on its official website. In addition to optimizing the model's performance in terms of supported input modalities and text length, OpenAI has upgraded ChatGPT on top of GPT-4 and opened up the API in one fell swoop - the speed of iteration is staggering.

The numbers have been numbing in this frenzy that belongs to the big models. Firstly, the number of parameters of the model - previously, OpenAI rolled up the number of parameters of the big model to the 100 billion level with GPT-3 (with 175 billion parameters), but soon after, Google's multimodal big model PalM-E, launched on March 6, sat with 562 billion parameters as the " largest visual language model ever".

The second is the company's wildly increasing valuation. According to a report by Dealroom, a global provider of early-stage project data, the valuation of global generative AI companies reached a total of about $48 billion, a six-fold increase in two years.

The domestic AI circuit heated up late, but the valuations of companies have soared as fast as ever - Wang Huiwen's AI company, Light Years Away, was valued at $200 million in its angel round. A large model company recently founded by a technology guru, the model demo is not yet in the shadows, the angel round valuation has also made it enter the billion dollar club - and in the windy mouth of the meta-universe, a million dollars seems to be the ceiling of the angel round valuation of domestic startups.

Amidst the wind, there are also some nagging, negative voices surfacing.

On the evening of March 2, a posting titled "Why do we feel that AI in Europe and the US is better than ours" caused a lot of controversy. The poster compared the AI development environment in China and the United States, and regarded the development of AI in Europe and the United States as "elite education" with a prudent approach, while in China it is "utilitarian education" with a heavy focus on commercialization, and finally came to a slightly desperate conclusion: the fate of human beings is predetermined in the womb. But robots are not immune.

But the richness of the data dimension and the wide range of application scenarios are the richness of the last wave of the Internet, which lasted for more than 10 years and left Chinese entrepreneurs with a mine of resources. In the past month, many small companies with scenarios and user data have trained small models adapted to their own business based on the base of large models at home and abroad. A company with a large model reserve of tens of billions of parameters has also "slimmed down" itself, launching lighter models for finance and advertising, in order to carry out a new round of data reserves.

At present, using small models to polish the sharp edges of algorithms and prepare technical reserves for the development of large models may be a pathway for Chinese entrepreneurs to achieve a bend in the road in the future.

Company Home: www.hkmjd.com