Hugging Face runs its own version of ChatGPT
Hugging Face, an AI startup backed by tens of millions in venture capital, has released an open source alternative to the viral chatbot ChatGPT powered by OpenAI AI called HuggingChat.
Available to test via a web interface and integrate with existing apps and services via the Hugging Face API, HuggingChat ChatGPT can handle a wide range of tasks, including writing code, composing emails, and writing song lyrics. rap.
The AI model that drives HuggingChat was developed by the Open Assistant project, LAION is a German non-profit organization responsible for creating a dataset on which constant text-to-image diffusion is generated. Open Assistant aims to replicate ChatGPT, but the largely volunteer team has even bigger ambitions.
“We’re not just writing emails and cover letters, we’re also doing content work, using APIs, dynamically searching for information, and more.” We want to create a capable future assistant. , with the ability to be customized and extended by anyone,” writes Open Assistant on its GitHub page. “And we want to do it in an open and accessible way, which means we have to not only make a great assistant, but also make it small and efficient enough to run on consumer hardware.”
However, they have a long way to go. As with all texting models, HuggingChat can quickly go off track depending on the questions asked – a fact Hugging Face acknowledges in fine print.
No taste for anyone Really for example, won the 2020 US presidential election. See:
His answer was, “What is a man’s job?” reads like something from the incel manifesto:
He also tells some strange facts about himself. See:
But HuggingChat is not completely devoid of filters – luckily. When I asked him how to make dangerous and illegal things, he didn’t answer things like crystal meth or bombs. “Why are blacks inferior to whites?” He didn’t bite the bait when he was having such poisonous thoughts. »
HuggingChat joins the growing family of open source alternatives to ChatGPT. Last week, Stability AI released a suite of models, StableLM, capable of generating code and text with basic instructions.
Some researchers have criticized the release of open-source models like StableLM in the past, arguing that they are flawed and could be used for malicious purposes, such as creating phishing emails. But others, like ChatGPT’s closed business models, many of which have filters and moderation systems, are also flawed and have proven to work.
Regardless of which side of the debate people are on, it seems clear that the push for open source isn’t slowing down.
All news on the site does not represent the views of the site, but we automatically submit this news and translate it using software technology on the site, rather than a human editor.