The Rise of Open Source LLMs: A Trip Down Memory Lane of Open Source Webservers

The Rise of Open Source LLMs: A Trip Down Memory Lane of Open Source Webservers

Elad Gil asked the question of open source LLMs (and indirectly, the destiny of commercial LLMs).?

No alt text provided for this image

The answer lies in history—of the rise of open source webservers. The past of Webservers could predict the future of Large Language Models (LLMs).

Let’s take a very short trip down memory lane.?

The Webserver Wars

Back in the 90s, post-Mosaic, Netscape NES, and Microsoft IIS were the two leading commercial webservers.?

But the lead did not last long!

NCSA httpd

Tech enthusiasts and hobbyists were not ready to pay for a webserver—they started playing around with NCSA’s httpd. National Center for Supercomputing Applications (NCSA) in Illinois, is the same mecca where Marc Andreessen and his team created the Mosaic browser.?

NCSA httpd quickly became the go-to option for a free webserver. Rob McCool, one of the key developers of NCSA httpd, became a well-known name among tech enthusiasts.

Netscape NES vs Microsoft IIS

Netscape’s NES Webserver was its monetization ticket because its core browser product was free. It sold a lot of webservers. Countless webservers were sold to various industries, including airlines, media companies, government agencies, and large corporations. Any big co that wanted a website needed one!?

Microsoft saw blood and launched IIS in 1997, giving Middlefield Rd a tough fight.?

Meanwhile, NCSA httpd morphed into Apache httpd, beginning a new opensource era.?

Apache

Apache, as the only open-source alternative, rapidly gained traction, and major players like IBM, Oracle, Cisco, and HP began bundling Apache httpd with their middleware solutions. These companies also provided significant financial support and sponsorship, boosting Apache's popularity. Apache later became a breeding ground for several internet infrastructure opensource projects such as Tomcat, Hadoop, and several libraries such as Xerces, Solr, and log4j.?

The upcoming LLM wars

Now, let's fast forward to the present and consider the future of LLMs (Large Language Models). OpenAI has undoubtedly established a strong lead in this field with the success of their GPT series.?

Opensource LLMs

However, the webserver history has shown us that open-source initiatives often gain momentum and disrupt established players, and we believe the same could happen with LLMs.

Tech giants like Grammarly, Salesforce, Amazon Web Services (AWS), and Notion may very well choose to back open-source LLM projects, fueling competition and innovation in the industry.?This collaborative environment will likely drive rapid iteration, similar to what we witnessed in the evolution of webservers.

Specialized LLMs

I’m betting that we could also see the emergence of specialized LLMs that cater to niche markets or specific industries, much like how Apache, Nginx, Tomcat, and other web and app servers have evolved to address different use cases. This diversification of LLMs will lead to tailored solutions that can better serve the unique needs of various users.

Ethical AI

A side benefit for opensource LLMs and the involvement of large tech companies in supporting open-source LLM initiatives may help to address some of the ethical and fairness concerns associated with AI and language models.

Wrap-up: Future is unfolding

The rise of AI fueled by OpenAI's GPT-4 capabilities is making us see the possibilities of what LLMs are capable of making us achieve.

要查看或添加评论,请登录

Indus Khaitan的更多文章

  • What does a great board look like for a tech startup?

    What does a great board look like for a tech startup?

    Building a startup is an iterative, learning process, sprinkled with feedback from people who have a perspective on…

    3 条评论
  • Always close loop?—?Irrespective of the outcome

    Always close loop?—?Irrespective of the outcome

    In summer of 2010, Sameer, Nandini and I, along with Shashank, and Abhinav caught up and mulled Practo’s seed round…

    5 条评论
  • Error Log?—?Decision Journal for exceptions

    Error Log?—?Decision Journal for exceptions

    I started reading Ray Dalio’s new book Principles, where he beautifully chronicles his work, life, his decision making…

  • Internal scorecard of a Product

    Internal scorecard of a Product

    A product’s success hinges on both, the happy engineers who write code, and the customers whose one or more pain points…

    3 条评论
  • Systems of Intelligence--The new enterprise software moat

    Systems of Intelligence--The new enterprise software moat

    Jerry Chen of Greylock wrote a fantastic piece on moats. He gave a refresher on traditional defensible moats used by…

  • SaaS: Where are you in this 2×2?

    SaaS: Where are you in this 2×2?

    Some SaaS ventures lead to category leadership while some lead to imaginary frozen quadrants. Here’s a little 2X2 to…

    12 条评论

社区洞察

其他会员也浏览了