You say you have an iron will, are you serious?

Chapter 152 If his algorithm is not good, then there will be no one left.

It is an undoubted fact that the prediction of earthquakes is a difficult problem in the world.

When Zhao Mo was studying, he also read an article about Zhang Heng's seismometer in the Han Dynasty predicting earthquakes. It was so miraculous that it could be predicted thousands of miles away. Wherever the ball from the dragon's mouth fell, it would indicate where the earthquake was.

Being young and with little education, Zhao Mo thought that there really was such a thing as a seismometer and that earthquakes could be predicted.

It wasn't until he grew up and read more and more that he began to have doubts.Especially when he saw the news about earthquakes causing heavy losses, he quickly learned that earthquake prediction was impossible.If possible, would those major losses still be there?Doesn’t the country know how to take preventive measures in advance?

As for the online marketing account's claims that ancient technology was lost and Zhang Heng's seismometer was not that simple, Zhao Mo scoffed and didn't believe a word of it.

Our China is an ancient country with a history of 5000 years. All cultural and technological inheritances have a traceable context and are not fabricated out of thin air.

If it were true, wouldn’t it be possible to reproduce it even if we put all the effort into building it?
It has been thousands of years, but it still cannot be reproduced, which shows that earthquake prediction is really not trustworthy!
Nowadays, the "big earthquake prediction model" built with the most advanced technology, the most advanced concepts, the best team and the most detailed data, the predicted results are also very different. As you can imagine The difficulty of it.

Zhao Mo dares to say that their place is the best place in the world for predicting earthquakes, no one!

"Zhao Mo, what do you think?"

While he was thinking about it, Professor Yao suddenly asked.

Zhao Mo: "..."

Boss, what can I think?Just stand and watch~
After thinking about it, he believed that the key to the accuracy of this model still lies in the algorithm part. However, he has not found anything wrong so far, so he said: "Check, check, and try a few more times."

For now, one prediction doesn't mean much.

"Yes~"

Professor Yao nodded calmly and then made arrangements.

As a result, the team's researchers began a new round of busy work, focusing on the investigation of data and neural network architecture, including machine equipment.

Zhao Mo did not leave, but concentrated on staring at the lines of code flashing on the screen. The researchers were also troubleshooting the code problems.

"...

Self.activator=activator
……

Self.bias=0.0
……

Return self.activator(
Reduce(lambda a, b:a+b,

Map(lambda(x,w):x*w,

Zip(input_vec, self.weights))
,0.0)+self.bias)
Def train…”

Looking at these codes, Zhao Mo couldn't help but frown. It was too taxing on his brain cells. He just wanted to read the algorithms inside, but too many codes would undoubtedly disturb him.

Professor Yao was about to go to help, but when he saw him suddenly frowning, he couldn't help but ask: "Did you find something?"

Zhao Mo was about to speak when a weight expression formula suddenly appeared in his sight. Before he could realize what was going on, he felt something was wrong, so he subconsciously started doing mental calculations.

Seeing this, Professor Yao thought it would be better not to disturb him first, so he waited quietly.

"Professor, you have finished reviewing everything. Do you want to continue the test?"

Soon, a researcher came over and asked Professor Yao.

Professor Yao glanced at this man and asked, "Gaowen, have you found any problems?"

Gawain shook his head and said: "I haven't found it yet."

Professor Yao nodded and glanced at Zhao Mo again. Seeing that he still didn't say anything, he said, "Then continue testing and test a few more times."

"Ok~"

Gawain agreed, and then prepared to make arrangements.

"Wait~"

But just after he turned around and walked a few steps, Zhao Mo's voice rang out and said, "Brother Wen, wait a minute, I think the weight algorithm and SOM network topology algorithm need to be updated."

Gao Wen: "Huh?"

He subconsciously looked at Professor Yao, waiting for his explanation.

Professor Yao seemed to sense his question and immediately asked Zhao Mo: "Is there any question?"

Zhao Mo smiled slightly and said: "It's not a problem, I just have some new ideas and feel that the algorithms in these two places can be optimized."

After that conference, his mathematical abilities improved a lot.

Although the LV level has not been improved, this feeling is real and can be said to be a bit metaphysical.Of course, in Zhao Mo's opinion, this is not metaphysics, but science has not yet developed to the level of understanding.

The mathematical algorithm of the large model is the result of his and the entire team's efforts, but now he can see the shortcomings, and he came up with an improvement plan without thinking for long.

"Is that so?"

Professor Yao sounded a little confused, but then he thought about it. Zhao Mo had just solved the ABC conjecture. His mathematics level had improved by leaps and bounds during this period. It was not surprising that he had an optimization plan for the algorithm he had participated in previous research.

So, he immediately gestured to Gawain: "You don't need to look at me to adjust the algorithm, just ask him. If his algorithm doesn't work, then I don't think anyone else's algorithm can help us solve the problem of large models."

While signaling, he also pointed at Zhao Mo.

Gao Wen looked at Zhao Mo and said with a smile: "Zhao Shen, please tell me how to adjust it~"

Zhao Mo didn't hesitate. He immediately brought a paper and pen and wrote down the algorithm he had just solved.

After a while, Zhao Mo handed the finished algorithm to Gao Wen and said, "Let's do this~"

Gawain reached out to take it, glanced down, and subconsciously asked: "What if it still doesn't work this time?"

Professor Yao: "???"

Gao Wen is a researcher at the Computer Research Institute. He is a senior person after all. When did he talk like this?
His expression immediately changed and he wanted to make up for his words: "Zhao Mo, you..."

"If it doesn't work, the architecture and data still need to be adjusted and optimized."

Zhao Mo smiled slightly and said to Gawain.

"Uh-huh~ okay!"

Gawain was suddenly pushed, and he realized the inappropriateness of what he just said. He couldn't help but smile, and then turned around and left with the algorithm. "I'll go over there~"

Professor Yao said something and followed.

The adjustment of the algorithm will take some time, but it won't take too long, so Zhao Mo is waiting at the scene.

No one is questioning the algorithm, and no one is talking about the algorithm.

Even if I want to discuss or question it, I don't want to talk at this juncture. After all, it is still in the testing stage. If there are problems, we will just adjust it.We are all R&D personnel, so we naturally know that the R&D work of a project is complicated. It is normal to have problems, and it is abnormal to have no problems.

No problem means no problem found, not really no problem.

There is no problem in the research and development stage, but problems usually break out in the actual application or mass production stage. At that time, it is more troublesome and the losses are greater.

And for large-scale model research, which is completely new to the world, there should be many problems.Solving them one by one will be normal and the results will be more reliable.

Zhao Mo was not idle either. While waiting, he listened to the discussions between Professor Yao and his team members, and continued to learn relevant knowledge.

Unknowingly, the sky gradually darkened.

Looking at the time, it was already past six o'clock in the evening.

The growl of his stomach brought Zhao Mo back to his senses, but seeing that Professor Yao and others had no intention of having dinner, he himself had no intention of going to dinner, and planned to join them~
The staff here are more considerate and asked the canteen to deliver lunch boxes.

So, everyone was busy while eating.

Many people couldn't take a bite for half an hour or an hour. With their last breath, they pulled the rice into their mouths and stuffed their faces into little buns to eat like this.

I was busy until about 10:30 in the evening. After the full cooperation of the entire team, the large model was finally completed according to the latest algorithm update.

Then, test again!
As the large model is started, a large amount of data is input, and the large model starts a new round of predictions.

Soon, under everyone's nervous and expectant eyes, the new prediction results came out. The time difference for the next round of earthquakes showed that it was only 10 to 12 months, and the approximate time was hovering between March and May 2006.

"not bad!"

Zhao Mo thought to himself that this result was not beyond his calculation.

"It's amazing~"

Gao Wen looked back at Zhao Mo with a surprised look on his face. He didn't expect that after optimizing the algorithm, the accuracy improved so much. He immediately understood what Professor Yao said just now. Zhao Mo is really amazing now.

Of course, this is just a test of a single project, not an actual prediction. It is still far from reality, so his performance is not very obvious.

"Pretty!"

Professor Yao clapped his hands and cheered, with a happy look on his face.

He was really happy. Just two rounds of predictions had such results. Although it didn't mean much, it was still a good start.

Immediately, he glanced at Zhao Mo and saw that Zhao Mo's face was very calm and calm. He obviously understood the result. He wanted to say something, but after thinking about it, he didn't say it. He just patted Zhao Mo. Silence your shoulders, everything is left unsaid.

Then, he looked at Gawain and said: "Continue the test~"

"Ok~"

Gao Wen quickly agreed, and then, with the cooperation of other colleagues, entered other specific data and began to continue testing.

The second time was the Northeastern region of the test.

Contrary to everyone's expectations, the time difference this time was 13 to 18 months, and the span was about half a year, which was three months longer than the previous span.

"Hey, this data is inaccurate again~"

Someone in the team made a sound of disappointment.

"Shouldn't it mean that the results are inaccurate?"

Gao Wen couldn't help but frown. It seemed that Zhao Mo's optimized algorithm was not universal and could only be used for specific tests.

He looked at Zhao Mo, wanting to see if Zhao Mo had anything to say.

Unexpectedly, when he turned around, he saw that there was no expression on Zhao Mo's face, and he still looked calm, as if he had the outcome under control.

"Pretending? Are you really calm and collected?"

A thought flashed through Gawain's mind, and he suppressed it immediately.

He opened his mouth, and when he was about to speak, a voice rang out.

“It’s not that the data is inaccurate!”

Professor Yao said, and what he said was exactly what he was thinking.

But the next moment, Professor Yao said: "It should be that the data is too small and not detailed enough. If you look at the earthquake in Caiyunnan for the first time, the data is obviously much more detailed."

As soon as these words came out, Gao Wen was shocked and thought: "Although it is not impossible, Professor Yao trusts Zhao Mo too much~"

"Wang Lin~"

Professor Yao called an acquaintance of Zhao Mo and asked him, "Is that all the data?"

Wang Lin looked back at the people from the earthquake research team who came with him, then nodded and replied: "That's all. The data is indeed less than the data from the southwest."

Without data, there will naturally be no more accurate results.

Professor Yao had no choice, so he subconsciously looked at Zhao Mo and asked, "Can it be optimized further?"

"???"

Zhao Mo suddenly became speechless and immediately shook his head and said: "This is the best plan now. I don't think we need to be too picky. There are still fewer earthquakes in the northeast. I think it is the right choice for us to focus on the southwest. "

Even if the results are not deduced based on subsequent facts, looking at my country's earthquake records over the years, we know that the southwest direction is the top priority.

Professor Yao thought about it carefully and said, "I'll evaluate it again."

Afterwards, Zhao Mo stared at the data at the scene for a while, and did not return to Wudaokou with Professor Yao until 11:30.

(End of this chapter)