Xueba's computing system

Chapter 263 Looking for a Partner 1
The easiest way to give OpenAI a slap in the face is to create a larger language model that is more powerful than OpenAI.

As long as it is proven that OpenAI is just a small company, its myth will naturally be shattered.

Lin Yuan called out the computing power system panel and took a few glances. He believed that as long as he injected enough computing power, the system would be able to complete the task.

At most, he would just be pure-hearted and aloof, and treat Zhang Yan with respect for a while. He would devote all his energy to the computing system, and then shock the world with a big and amazing language model.

And then, Lin Yuan thought to himself that there was a high probability that he would be arrested and questioned.

“Where did you get that big language model?”

How will Lin Yuan answer this question then?

He can't say, "I did this on my own."

With the development of AI today, any AI model cannot be separated from the support of computing power. This level of computing power requirement cannot be supported by a few H100 computing power cards. It requires computing power cards in the thousands.

Then the authorities will definitely check Lin Yuan’s purchase records of computing power cards, and in the end they will at most find that he only has a few H100s.

Can OpenAI be beaten by just a few H100s? OpenAI spent 200 billion US dollars and used tens of thousands of computing power cards to create chatGPT.

If Lin Yuan really did this, he would not be treated like a hero, but would be told the truth, "You are not an alien" or "Where did you hide the alien?"

Lin Yuan even wondered whether he would be taken away for DNA sampling and research.

It’s scary to think about it.

Therefore, Lin Yuan absolutely cannot tinker with it on his own. He must find a partner and then use the partner's help to launch a powerful large language model.

Such a partner must also be professional and cannot be a half-baked company that has never played with large AI models, otherwise it will be difficult to cover up the lie in the future.

Lin Yuan then began to look for one.

In order to facilitate communication and keep the benefits within the family, we must find domestic companies. If we find foreign companies, it might be interpreted as: the other party uses OpenAI's technology.

Due to the blockade of China's high-tech field by the United States, it is more convincing for domestic companies to develop a large language model that is better than OpenAI.

Lin Yuan thought about it in his mind and wondered which companies in China were doing well in AI.

Dudu? Lili? Chrysanthemum? Jump?
After thinking about it, these are the ones I think are the best.

Lin Yuan was too lazy to play any behind-the-scenes tricks this time. He directly relied on his personal connections to get the contact information of the technical directors of these companies, and then began to recommend himself.

He first contacted Dudu because Dudu was the first to develop AI.

Dudu once brought together the most powerful AI experts in the Chinese community, and was the talk of the town. Later...

It seemed like there was no later.

However, Lin Yuan still planned to contact Dudu first. At least after chatGPT came out, Dudu also launched a corresponding product. Lin Yuan felt that it was okay after using it. At least at the search level, it did improve the efficiency of search engines.

However, Dudu’s products still cannot change their habit of advertising.

Lin Yuan quickly got in touch with Dudujia’s CTO.

After some opening greetings, Lin Yuan got straight to the point.

"I would like to use my experience in AI application for Tesla food delivery to help you improve the performance of large language models. I wonder if there is any possibility of cooperation between us." Lin Yuan chose to recommend himself.

Dudu CTO was somewhat flattered by Lin Yuan's sudden action. Perhaps it was because Dudu's business was too single and it had nothing to offer except search, so it went downhill in recent years and failed in everything it did.

Therefore, the other party was very confused. Why would a technology upstart like Lin Yuan choose a company like theirs?

"Mr. Lin, please forgive me for being so presumptuous. Why did you choose our company? To be honest, I think you have better opportunities to display your talents."

Lin Yuan's answer was also very straightforward. "I don't want to join your company. I saw your efforts in large language models. I happened to be interested in this area, so I wanted to try to seek cooperation with you in this regard."

"Mr. Lin, how do you want to cooperate, in what way, and to achieve what purpose?" "I want to make a large language model that is comparable to OpenAI, or even surpasses it. Let your... that thing surpass chatGPT4, or even surpass its latest version in performance."

After Lin Yuan finished speaking, the other side fell silent.

Out of politeness, Lin Yuan waited for a moment.

After more than ten seconds, there was still no sound, so Lin Yuan had no choice but to ask, "Hello, hello, can you hear me?"

"I'm listening, Mr. Lin. Why do you have such thoughts?"

"For the sake of change, I have become financially independent through my Tesla food delivery business, and I want to create something innovative and groundbreaking. So this time I targeted the big language model."

"Is that why?"

"Yes."

There was silence for a moment.

"I almost forgot Mr. Lin's age. I really envy him. He has reached a height that we can never reach in our lifetime at such a young age."

Since Tesla's food delivery business was successful, Lin Yuan has been called "Musk's biological son." Now that he has been jointly sued by Opani and Tesla, he has been called "Musk's rebellious son."

Because in the eyes of the public, Lin Yuan has open-sourced the Tesla takeout algorithm, which caused Tesla and OpenAI to lose a listed company with a future market value of hundreds of billions of US dollars. In the eyes of the spectators, he is just a rebellious son of a wealthy family.

But no matter what, in the eyes of the world's spectators, Lin Yuan is the de facto successor to Musk in the field of technology. At least these two guys act crazy.

Of course Lin Yuan doesn't think so. Because he and Musk are equals, he doesn't want to be Musk's son. Whoever wants to be Musk can do it.

Therefore, Dudu's CTO naturally thought that Lin Yuan was just being too impulsive and wanted to mess around again.

"Of course our company hopes to cooperate with Mr. Lin, but our current situation is to preserve our traditional business. Our traditional business is search. In fact, we are doing AI mainly to expand the boundaries of search and not fall behind in the future trend."

Lin Yuan understands very well that AI's big language model is inherently strongly related to search, so if the homepage is searched, it must be given to the big language model.

"That's perfect. You guys can't give up the big language model, and I just want to do this. We're a perfect match."

"However, we hope to develop our business around search. From what you said, Mr. Lin, it seems that more attention is paid to the big language model itself."

Lin Yuan thought for a moment and said, "I don't think the two seem to be contradictory."

“Mr. Lin may be more concerned about technical issues and may not know much about the business operations of the company. The investment in general AI large language models is at the billion-dollar level. Our company’s current overall strategy is to shrink, and we do not want to invest too much in this type of general technology.

Our strategy is to limit investment and focus more on exploring the application of AI large language models in the search field.

Mr. Lin, you should understand what I mean. We are currently a company that does applications, not a company that does research. "

How could Lin Yuan not understand, because the big boss of Dudu had just said publicly: It is impossible for a company to be comparable to OpenAI in China, so he has already planned to give up the AI track.

Take Dudu for example. It is the same for everything they do, getting up early and arriving late. They are always the first to enter the market, but then fail before dawn. This is true for AI, autonomous driving, and even food delivery.

If Lin Yuan had not seen that the application of the large language model was related to the life and death of this company, he would not have bothered to recommend himself to the company.

"What if... I have a way to reduce the R&D investment in large language models? We don't need to invest billions, maybe tens of millions of dollars. This amount of R&D funding should not be a big problem for your company."

"Boss Lin, if you are really willing to help us iterate our search, let alone 10 million, even if we really invest 1 billion, I think our boss will agree. But for the research of general large language models, no matter how much money is invested, it will be in vain.

What's more, our boss has already stopped R&D investment in this area at the strategic level. So,,,

"I understand." Since the other party had already mentioned the big boss, Lin Yuan had nothing to say. Because Dudu's corporate strategy had already abandoned the general language model, it was impossible for Lin Yuan to change the strategic direction of a large listed company with just a few words.

To put it bluntly, the company Dudu has become a coward.

(End of this chapter)