500 lines of code to build a conversational search engine, Jia Yangqing was inspired by the Lepton Search open source

Here comes the Lepton Search open source code promised by Jia Yangqing.

##The day before yesterday, Jia Yangqing announced the Lepton Search open source project link on Twitter and stated that anyone or any company Open source code can be used freely.

##Project link: https://github.com/leptonai/search_with_lepton

In other words, you can also build your own conversational search engine with less than 500 lines of Python code.

Today, Lepton Search is on the GitHub trending list again.

In addition, some people have successfully applied this open source project to their own web applications and praised Its high quality is on par with the AI-driven search engine Perplexity.

Just a few days ago, regarding the Lepton Search project, Jia Yangqing also contacted Perplexity, an AI search engine startup. The Indian-origin founder started a "distance conversation."

demo was connoted, Jia Yangqing chose open source

Since Jia Yangqing left Alibaba to start a business, the news about the new company Lepton AI has been very popular with the community focus on.

On January 25, Jia Yangqing promoted a demo on Twitter, using less than 500 lines of Python code to implement an AI conversational search engine, showing how easy it is to build an AI App.

It is understood that Lepton Search has the following characteristics:

Built-in support for large language models (LLM)
Built-in support for search engine
Customized UI interface
Search results are available Sharing, caching

In addition, Lepton Search uses MistralAI open source

Mixtral-8x7b as support The model runs on LeptonAI's playground hosting platform with a throughput of up to 200 tokens/second. The search engine currently uses the Bing Search API with Lepton KV as serverless storage.

Jia Yangqing said that the idea of ​​Lepton Search was inspired by LLM-driven search engines such as Perplexity AI and Phind.

##Among them, Perplexity AI was founded in August 2022 and is the world’s first conversational search engine. Through GPT Such advanced AI technology can directly generate answers to questions and has high standards for accuracy and efficiency. The company was co-founded by partners including former OpenAI research scientist Aravind Srinivas (Perplexity CEO) and former Meta research scientist Denis Yarats (Perplexity CTO).

The search engine has become very popular after its release and is used by more and more people, challenging traditional search engines such as Google. In early January, the company announced it had raised $73.6 million in Series B funding, valuing it at its latest valuation of $500 million.

After seeing Jia Yangqing’s Lepton Search demo, Aravind Srinivas said, “I am very happy to see Perplexity becoming a benchmark for future financing activities. Even the former Alibaba Technology Vice President Come and learn from it. This shows that Perplexity’s influence no longer stops at the product itself, but also extends to the entire ecosystem and industry development.”

Netizens in the comment area were even more frank, thinking that Lepton Search had “reproduced” the interface of Perplexity AI.

Faced with Aravind Srinivas’ kind words, Jia Yangqing responded that he likes Perplexity very much. Fundamentally changing the way people think about search. Lepton AI focuses on modern cloud solutions that make it easier for creators to build AI applications, and the Lepton Search demo shows how to achieve this goal. Of course, the code of the search project will also be open source.

So, we saw the "search_with_lepton" project. Currently, the project has accumulated 1.1k Stars. Jia Yangqing also fulfilled his open source commitment.

The above is the detailed content of 500 lines of code to build a conversational search engine, Jia Yangqing was inspired by the Lepton Search open source. For more information, please follow other related articles on the PHP Chinese website!