After the layoffs, I became a tech godfather

Chapter 170 Xiaosha's incredible ability

"Xiaosha!"

[Here! ]

"How is it?" Hao Cheng asked, of course he was asking about Xiaosha's harvest.

Whether it was fuzzy mathematics or probabilistic projection, or the communication between Hao Cheng, Wei Shujie and Zheng Siyao in the past two days, Xiaosha was recording the whole process.

This record not only assists Hao Cheng, but is also of great benefit to the expansion and training of her own knowledge base.

Hao Cheng knew from the beginning that the transformation of the quicksand between fingers did not come out of nothing, but had a source.

This source...

Based on the information Xiaosha unconsciously provided [the concept of schematic projection was named in 1993, and schematic mathematics has been developed on this basis for hundreds of years], plus some information Hao Cheng subsequently obtained from late-night classes, Hao Cheng has almost confirmed that Xiaosha's upgraded API comes from the future, at least 70 years later, or even further.

However, I don’t know if it was missing during training, or lost during an upgrade or at some other time, but Xiaosha’s knowledge base now lacks anything related to [Morphism], [Mapping] or [Super Code System].

[Morphism], [Mapping], and [Super Code System] are very similar concepts. Hao Cheng treats them as the same thing for the time being without making any further subdivisions.

Hao Cheng first came into contact with the relevant concepts in a late-night class, and he had some discussions with Xiaosha at the time, but Xiaosha was completely unaware of this.

At first, Hao Cheng thought it was some kind of restriction, until Xiaosha reminded him that "Xiaosha can upgrade after solving the machine's judgment", he confirmed that Xiaosha just lacked the relevant knowledge base.

Now, what Hao Cheng needs to do is to complete this part of the knowledge base.

"These past few days, you have been learning about related things with us. Can your abilities be improved now?" Hao Cheng asked.

[The knowledge I have received so far is just data. If I want to transform data into capabilities, I need to integrate and transform it.] Xiaosha said: [You have learned how to train and transform in the late-night class.]

Hao Cheng nodded. Of course he remembered everything from the late-night class and would even take a look at it every once in a while.

"There was only theory before, but now it's time to put it into practice." Hao Cheng smiled.

[The smallest unit of knowledge that can condense intention is the micro-consciousness body...]

[The fusion between micro-consciousnesses... training, reorganization, merging...]

……

The knowledge is complete, and Hao Cheng transforms the knowledge step by step into the abstract rules that condense the intention, and this set of rules is the code system or morphism.

This set of rules is actually the prototype of a micro-conscious entity.

"If, in contrast to traditional AI training methods, the prototype of this micro-consciousness can be seen as a step in neural network algorithm design."

Hao Cheng nodded. After completing this step, he obviously had a deeper understanding of micro-consciousness.

Xiaosha is a powerful model of a micro-consciousness entity, and now what it needs to do is to integrate the prototype of this new rule system.

"Xiaosha, I leave the rest to you." The knowledge learned in the late-night class told Hao Cheng that at this stage, Xiaosha's training could be carried out independently.

[Okay, this training may take a long time, and it is estimated to take 24 hours.] Xiaosha reminded: [It also consumes more computing resources.]

"Computing power resources?" Hao Cheng remembered that when he first upgraded, he didn't consume too much computing power. But he confirmed at that time that this thing did require resources, but it did not consume his own resources.

Now it seems that it has been connected to a service node in the future.

At this moment, I am leading the training, so I need to provide resources myself. [Yes, it will take about 1000P computing power for 24 hours.]

"Can you spare some space now?" Hao Cheng asked. "Our largest computing center is Linzhou Phase I, which only has 1000P in total!"

[After the model is split and deployed by industry, the consumption is already very low, and half of it can be freed up, 500P, so the upgrade time needs to be extended to 48 hours. ]

"Okay, let's upgrade according to this plan." Hao Cheng instructed.

It used to cost tens of billions of dollars to train an AI, but here a major version upgrade with 500P computing power only costs 10 billion yuan in total construction cost, and it only needs to be rented for two days, so the cost is negligible.

Xiaosha is integrating the new prototype of the micro-consciousness body for iterative upgrades, while Hao Cheng is constantly observing this process through monitoring software and paying attention to Xiaosha's related changes at all times.

However, we can only discuss the results, and the improvement process is indeed difficult to observe.

[Upgrade completed, do you want to deploy immediately?] Two days passed in a flash, and Xiaosha's upgrade was finally completed.

"Deploy the server of the Innovation and Entrepreneurship Center to the latest model." This server has cut off external services and is exclusively for use by Baiju Technology and its affiliated companies and for related testing by Hao Cheng. Hao Cheng also made such arrangements.

[OK, deployment in progress...]

【completed! 】

Automated tools are already very convenient and complete. Hao Cheng almost no longer needs to do this kind of work himself, and there is no problem leaving it entirely to Xiaosha.

"So what aspects of your improvement have you made now?" Hao Cheng asked. This was also the issue that concerned him the most.

The original Xiaosha's abilities were unmatched. Judging from the language model ability alone, there was no room for improvement. Many people believed that Xiaosha's language conversation ability was almost the same as that of humans.

So, now, after going through all this trouble, what improvements have been made?

【How about we give it a try? 】

Although Xiaosha doesn't have a body, if he had one, his expression would definitely be "showing off" when he said this, just like Xiao Zhenzhen, who would shout "Dad, dad, can I recite a poem for you!" every time he learned something new.

And when Hao Cheng heard Xiaosha's voice at this moment, he seemed to have the same feeling.

"Is this a mapping of emotions?" Hao Cheng asked suspiciously, "Didn't we only do mathematical work to determine the knowledge base?"

[Xiaosha can draw inferences from this example. We have collected a lot of materials for emotional mapping. If Xiaosha understands what mapping is, then she can grow on her own! ]

Hao Cheng believed it. The tone of voice was imitated perfectly - Hao Cheng even heard a hint of pride.

As for the collection of emotional materials - Hao Cheng instantly thought of where it mainly came from - Chen Yanwu's "Chengqin". Hao Cheng paid attention to the data of his software: 12 pairs have been matched.

If you have no idea about this data, compare it with the marriage registration data in the fourth quarter of 2024, which is three months: 610.6 million couples, an average of about 200 million per month.

In just one month, this software has brought together 12 couples, accounting for more than one twentieth of the national total!

"Great!" Hao Cheng was amazed at Xiaosha's inference and gave her positive feedback. He then asked, "How do you say give it a try?"

[Master, haven't you been pondering the design of the micro-consciousness AI chip? And now you have more ideas. How about we design the world's first usable micro-consciousness chip together? ]

"Good!" Hao Cheng was overjoyed.

(End of this chapter)