Large language models (LLMs) are becoming popular, and open source LLMs are at the forefront of this trend.
Meta’s Llama-2, a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 70 billion parameters, has been downloaded over 30 million times in the last month. Other open source LLMs such as Vicuna have also seen significant download numbers.
This trend is being driven by the cost-effectiveness of open source LLMs, the transparency and accountability they offer, and the ability for developers to customize and improve them.
Other major providers such as OpenAI and Google have been slower to embrace open source LLMs. However, even these companies are beginning to see the benefits of open source. In a recent internal memo, a Google AI engineer referred to the open source community as “a third faction [that] has been quietly eating our lunch.”
Investors are also taking note of the open source trend. Wei Lien Dang, general partner at Unusual Ventures, said that “there’s been this significant proliferation” of open source LLMs in recent months. James Currier, a general partner at NFX, has also highlighted the cost-saving advantages of open source LLMs. One of his portfolio companies was able to reduce its monthly expenses from $150,000 to $4,000 by switching to an open source LLM.
The sources for this piece include an article in AnalyticsIndiaMag.