DeepSeek user's guide, very useful tips (recommended collection)

Since DeepSeek-V3 was released a month ago, I have been updating DeepSeek related articles and videos because I think it is a very cool company. Until yesterday, I finally witnessed history, reaching the top of Apple's AppStore app store in the United States, surpassing ChatGPT. Nvidia and TSMC were directly killed last night... Then, when we come back, this article will introduce the usage techniques of DeepSeek, because it is completely different from previous AI. How to use DeepSeek? Take care of Xiao Bai's friend first, and the veteran can just row back directly. If you don't know the entrance to DeepSeek, continue reading this paragraph. Currently, you can use DeepSeek in the following ways: Website: chat.deepseek.com/ App: Just search for DeepSeek downloads in major app stores, and remember to look for the Blue Whale Logo using skills Tip 1 Open up the Ren and Du meridians When using it, be sure to check the buttons in the red box in the picture below. This is the Ren Du two-pulse switch that can punch ChatGPT and kick Claude. Tip 2 Forget the prompt template Forget the long-winded prompt word templates you have hoarding first. All prompt words are worth trying again with DeepSeek-R1 because the intelligence of the model has improved. This is like when you, as a boss, originally recruited a newly graduated intern to work, but now you become a classmate with 5 years of work experience. If you arrange a task, what you explain to him will definitely be different. R1 is sensitive to prompt words. Using few-shot fewer sample prompt words may reduce its performance. zero-shot zero-sample prompt words have the best effect. In layman's terms, you no longer need to give him multiple examples, just directly describe the tasks and steps. For example: I want to write a New Year greetings for the Year of the Snake and send it to my elders. He will automatically output multiple style versions to you, and you can use them with your own tweaks. Tip 3 Speaking in human language Sometimes DeepSeek's answers will be "too hard" and the content is very abstract and difficult to understand. Many people gave up when they got stuck here, thinking that this new AI was just like this. But the right thing to do is to think further about how to make the output of the model better? The answer is simple, look down. At this time, you just need to say "speak human language" to him. Therefore, everyone must practice hands-on, and only by understanding it time and time again can you understand the model's thoughts clearly. In fact, the model is the same as your boyfriend/girlfriend. You need to be careful to understand his/her little thoughts. Tip 4 Style Rewriting Please rewrite the following content in Chinese using XX style, while retaining the core content and structure of the original text. For example, we use Lu Xun's style and ask him to rewrite the content of the previous picture. For example, we used Liu Run's style and asked him to rewrite the Midjourney founder's evaluation of DeepSeek in the picture below. Tip 5 Creative Novel Seeing that Wei Xi posted a screenshot on her circle of friends was amazing, the question she asked was very interesting: On the day the Xuanwu Gate Incident ended, Li Shimin wrote a monologue late at night. What do you think he would write? DeepSeek-R1's answer is also very, very amazing. Currently, no AI can achieve the same level as DeepSeek-R1. This is exactly what Midjourney founder described as the "cultural moat" in his evaluation of DeepSeek above. You can also send the same question to other AI to try and compare it. Let's take a look at DeepSeek-R1's answer: Not only does it restore the tone of the emperor, but it is particularly familiar with historical details, but it also provides hidden clues to subsequent stories. I even threw the copy generated by DeepSeek-R1 to AI-generated Text Detection Tools for detection, and only 4% of the suspected AI-generated content accounted for! Data explains everything. Of course, your feelings are also very important. Please type your feelings after reading it in the comment area. After writing this article, I still couldn't calm down for a long time. A sentence that DeepSeek founder Liang Wenfeng said in a previous interview suddenly flashed through his mind: China's AI cannot follow it forever, and it will inevitably require someone to stand at the forefront of technology.