Large-scale open-source-based Natural Language Processing (NLP) models are increasingly in demand. It is striking that more and more NLP initiatives based on open source are now emerging. These initiatives stimulate the rise of NLP-as-a-service and increase sales in this market segment, writes tech website Venture Beat.
According to the tech website, NLP is increasingly popular with companies. Budgets for the use of NLP-based AI models have largely increased by a 10 percent year-on-year. Nearly a third of companies interested in NLP models spend 33 percent more.
NLP model providers benefit greatly from this, according to Venture Beat. Think especially of NLP model providers such as Open AI, the developers of the well-known powerful GPT-3 model, Cohere and A121 Labs. Today, GPT-3 is used in over 300 different applications by tens of thousands of developers and produces approximately 4.5 billion words per day. With the advent of these types of open-source-based models, the use of these types of AI applications has become quite ‘democratized’.
Rise of open source NLP developers
According to the tech website, the arrival of these NLP models also has a different effect. Developers of NLP models, often for technical and cost reasons, often choose not to make their models available as open source and to exploit them commercially.
However, the effect of this is that more and more open-source based NLP initiatives are emerging. With these open-source NLP models, other companies can more easily offer their own services, so that more end users can benefit from them.
One of these initiators for developing open-source-based NLP models is EleutherAI, which has developed a GPT-3-like proprietary NLP model, GPT-J. This open-source provider also offers a dataset, The Pile, for training large language models for writing text, code and other things.
NLP as-a-service providers
This open-source model of EleutherAI is in turn used by the startup NLP Cloud to offer NLP services ‘as-a-service’. Another startup that ‘markets’ EleutherAI’s open-source technology is Neuro. This startup offers, among other things, GTP-J on a pay-per-use basis. Another startup that uses the open-source NLP model is CoreWeave.
Greater NLP segment revenue expected
Researchers indicate that the emergence of these NLP-as-a-service providers will lead to a new phase for the use of NLP. They expect that these companies are better able to offer NLP than the big developers such as OpenAI and other companies.
According to the experts, these companies with open-source NLP models will soon ensure that more companies will embrace NLP. They will also ensure that turnover in this market segment will stimulate considerably in the coming years. By 2025, revenue in the NLP segment will triple, thanks in part to the as-a-service providers based on open source models.
