In an interview with Reuters, Hugging Face CEO Clem Delangue gave his perspective on the current scenario of AI models and how each company is adapting to this novelty.
"(...) We are entering a world where there is not just one model, but as many models as there are companies. Each company is training, adjusting and optimizing its own models, and that's the direction in which this industry is heading."
This insight sums up exactly what Hugging Face does and why you should follow this company. In this article we'll tell you everything you need to know about Hugging Face and how it's changing the AI landscape worldwide.
What is Hugging Face?
Hugging Face is an AI company that develops and offers machine learning and natural language processing (NLP) tools in a global community.
One of the tools that made Hugging Face so well known was the open source library called Transformers. In it, a user can find a variety of pre-trained NLP models, as well as resources for training and implementing customized models.
Hugging Face also brings together datasets and resources for other tasks. All can be used by anyone who accesses the community.
Can I use Hugging Face for free?
That's right, Hugging Face's main features are available for free. Are there also paid plans? Yes. Each plan has more features and charges are made per hour (from U$ 0.05) or per month (from U$ 9).
And these prices are in line with the way the company works. Delangue, in an interview with Quartz, points out: "We need more companies and organizations to share their models and data sets publicly and in open source so that everyone can understand and build AIs for themselves."
So that's the watchword: democratization.
What is the Hugging Face Transformer?
Despite the same name, we're not talking about alien robots that disguise themselves as cars and motorcycles. Transformers is a library of state-of-the-art pre-trained models that a user can access and download freely.
According to Hugging Face, the aim of Transformers is to "reduce your computing costs, your carbon footprint, and save the time needed to train a model from scratch".
Currently, a user can download templates for text - generating texts, summarizing articles and extracting information - images and audios. There are also templates that combine these three types.
In the available documentation, you'll find tutorials, practical guides, conceptual guides and APIs. What's more, the list of available templates is long! There are approximately 115 models with support documents for those who want to use them and the ability to integrate with other libraries, such as PyTorch, TensorFlow and JAX.
What's the Hugging Face community like?
One of the platform's most important points is its community. Together with the open repositories, Hugging Face opens up a space to bring AI enthusiasts and professionals closer together.
This collaborative community is dedicated to sharing knowledge, models and resources related to machine learning and NLP. Community members contribute to the development of models and tools, as well as sharing their personal experiences, tutorials and discussions on the latest advances in AI.
Active and engaged, communities play a fundamental role in the evolution and dissemination of technology around the world. And this is the aim of Hugging Face.
And how did Hugging Face grow so much?
One of the factors driving Hugging Face's growth has been the business's list of investors, which includes Google, Amazon, Nvidia, Salesforce, AMD, Intel and IBM.
As reported by Exame in August 2023: "The organization raised $235 million in its Series D funding round, as revealed by CNBC. With this new contribution, the company's valuation jumped to an impressive $4.5 billion."
Amazon, in particular, has taken its partnership with Hugging Face to the next level.
Amazon SageMaker is a fully managed service from Amazon Web Services (AWS) that offers a platform for building, training and deploying machine learning models in a scalable way. Now, a user can use Hugging Face models within SageMaker.
I don't know anything about programming, can I use Hugging Face?
Yes, you can use Hugging Face even without programming experience. The platform is simple, so a user can access the NLP models without the need to write code or understand complex technical aspects.
For example, Hugging Face provides an interface called Model Hub, where you can explore a variety of pre-trained models for specific tasks, such as language translation, even without programming experience.
In the end, even without advanced programming knowledge, you can still take advantage of the resources offered by Hugging Face to explore NLP models.
On the other hand, if a person has programming skills, they can make more use of every feature available on Hugging Face. Consequently, think of it this way: the more a user knows, the more dexterity and use they can make of the platform.
Conclusion
Now that you know what Hugging Face is, how the platform works and what you can do with the resources available, you can try it out for yourself.
"But I don't know much about machine learning or PLN, what do I do?"
We'll help you! Here on the Pareto blog you'll find articles on artificial intelligence, machine learning and natural language processing (NLP). With explanations, summaries and ideas, you'll learn all about these subjects and be able to explore the new possibilities that AI offers.
Visit our blog and learn from Pareto!
