April 14, 2024
Hugging Face launches open source AI assistant builder to compete with OpenAI's custom GPT


Hugging Face, the New York City-based startup that provides a popular, developer-focused repository for open source AI code and frameworks (and hosted last year’s “Woodstock of AI”), today announced the launch of third-party, customizable Hugging Face Announced the launch of Assistant. ,

The new, free product offering, the startup’s open source alternative to OpenAI’s ChatGPT, allows Hugging Chat’s users to easily create their own customized AI chatbots with specific capabilities, combining the functionality of OpenAI’s custom GPT builder and The intention is the same for both – although it does require payment. Subscriptions to ChatGPIT Plus ($20 per month), Team ($25 per user per month paid annually), and Enterprise (variable pricing based on needs).

Easy creation of custom AI chatbots

Philipp Schmidt, Hugging Face’s technical lead and LLM director, posted about the news on the social network Can create a chat assistant! Schmid openly compared the new capabilities to OpenAI’s custom GPT.

However, besides being free, the other big difference between Hugging Chat Assistant and GPT Builder and GPT Store is that the latter tools run entirely on OpenAI’s proprietary large language model (LLM) GPT-4 and GPT-4 Vision/Turbo. Depend.

vb event

AI Impact Tour – NYC

We will be in New York on February 29, in partnership with Microsoft, to discuss how to balance the risks and benefits of AI applications. Request an invitation to the special event below.

request an invite

Conversely, users of Hugging Chat Assistant can choose which of several open source LLMs they want to use to power the intelligence of their AI assistant on the backend, including everything from Mistral’s Mixtral to Meta’s Llama 2. Is included.

This is in keeping with Hugging Face’s broad approach to AI – offering a wide range of different models and frameworks for users to choose from – and it also takes the same approach with Hugging Chat, where users There are several different open source ones to choose from. The models that power it.

An open rival to the GPT store?

Like OpenAI with its GPT store launched last month, Hugging Face has also created a central repository of third-party customized Hugging Chat Assistants, which users can select from and use on their own time here.

The Hugging Chat Assistant aggregator page is very similar to the GPT store page, even in its visual style, with custom assistants displayed in their own rectangular, baseball card-style boxes with a circular logo, just like the custom GPT .

Screenshot of OpenAI’s GPT store. Screenshot of Hugging Face’s Hugging Chat Assistant page.

Better than GPT in some ways, worse in others

Already, some users in the open source AI community are touting Hugging Chat Assistant as “better than GPT,” including Matthew Trachino, founder of enterprise AI software provider GenDojo.ai, who used X to list the qualities. Liya, who mainly roam around. The user adaptability of the underlying models and the fact that the entire situation is free compared to OpenAI’s paid subscription tiers.

They also noted some areas where custom GPTs outperform Hugging Chat Assistant, including the fact that they do not currently support web search, retrieval augmented generation (RAG), nor do they support their Can generate own logos (which thanks to the power of GPT OpenAI Image Generation AI Model DALL-E 3).

Still, the arrival of Hugging Chat Assistant shows how quickly the open source community is catching up to closed rivals like “open” AI, especially just a day after the confirmation of a new open source model from Mistral. ,mik, which almost matches the performance of the discontinued GPT-4, is still a high watermark for LLM. But for how long?

VentureBeat’s mission To become a digital town square for technology decision makers to gain knowledge about transformative enterprise technology and transactions. Find our briefing.

Source: venturebeat.com



Leave a Reply

Your email address will not be published. Required fields are marked *