.svg)
Transformers, introduced in 2017 by a team at Google Brain, are becoming the go-to deep learning models in fields from natural language processing (NLP) to computer vision and beyond. The potential of transformers is much vaster than what's available now and even what will be available tomorrow.
An ecosystem of transformer models and techniques is growing rapidly. Companies that want to remain on the cutting edge will need employees who know the fundamentals and can use the tools that make all this potential possible.
If you're interested in learning how transformers work and how you can use them, consider taking an online course about NLP with transformers. Below, we've explained which learners are best suited for five similar courses we found online.
Course: The Hugging Face Course
Best for: ML practitioners
Price: Free
Time to complete: 72 hours
Flexible schedule: Yes
Prerequisites: yes
Accredited: No
Hugging Face, the company behind popular transformer models that many other companies and practitioners use, offers a course focused on educating learners about the models and libraries they offer.
The course requires a "good knowledge" of Python and, according to the instructors, is most useful to learners who've already taken at least an introductory course on deep learning. The course doesn't expect learners to have prior experience with PyTorch of TensorFlow, but familiarity with either will be helpful.
Nine Hugging Face employees created the course, all of whom bring a variety of experiences to the material, including advanced research experience at universities like Stanford and New York University, as well as practical experience gained at companies like Parse.ly and fast.ai.
The course was offered live in 2021, and recordings of those streams are available on YouTube. So now learners can take the course asynchronously.
The course is split into nine chapters, each of which includes numerous sections and an end-of-chapter quiz. Learners can expect to learn the basics of NLP and transformer models and how to use Hugging Face – one of the field's foremost tools. In fact, the course material is built around using the Hugging Face Hub, which hosts tens of thousands of transformer libraries and datasets.
Beyond the course, learners have access to a forum where they can ask questions and communicate with a community of people who are building and sharing transformer models.
This community includes people who have taken the course, people who haven't taken the course, and people who've been working in the field for years. So, community participants are exposed to a diversity of expertise but not necessarily other learners taking the course online.
The course is free. However, because the material is specific to Hugging Face, employers may consider paying for another course in conjunction to give employees a greater breadth of knowledge.
Learners won't get a certificate by the end of the course but Hugging Face is a well-known brand within the NLP and transformers models communities.
Course: Natural Language Processing: NLP With Transformers in Python
Best for: Aspiring data scientists and practicing developers
Price: $59.99
Time to complete: 11.5 hours
Flexible schedule: Yes
Prerequisites: No
Accredited: No
Udemy outlines one major prerequisite for this course, a knowledge of Python, and recommends learners also have experience in data science or NLP (though that experience is a "plus" and not necessary).
The course is designed for a wide range of students, including:
James Briggs, an ML engineer who's worked at Deloitte and currently works for Pinecone, teaches the course. He also brings to the course experience working in startups.
A self-taught engineer, Briggs has made a habit out of sharing the knowledge he's gained throughout his career. Briggs now specializes in NLP and has written articles accruing more than 2 million views.
The course includes 11.5 hours of on-demand video content, five articles, and numerous assignments. Upon completion, learners will:
Udemy in general, and this course in particular, does not emphasize networking as a benefit. The course does not appear to facilitate conversations between learners like other courses on this list.
The course costs $59.99, but occasional sales have knocked the price down as low as $12.99. Employers might be especially interested in funding this course if they're price-conscious or their employees are already well-connected.
Udemy offers a certificate for learners who complete the course. However, a third party does not accredit Udemy courses — perhaps because Udemy provides a wide range of courses of varying quality.
Course: Natural Language Processing
Best for: Software engineering, data science, and machine learning practitioners
Price: $400
Time to complete: 24 hours
Flexible schedule: Yes
Prerequisites: Yes
Accredited: No
CoRise targets the course at three types of learners:
CoRise also outlines two prerequisites:
Two instructors teach the course: Sourabh Bajaj, former engineering manager at Neeva, and Kaushik Rangadurai, current software engineer at Meta. Bajaj worked on the Google Brain team on TensorFlow, and Rangadurai has built AI products for LinkedIn, Google, and Microsoft.
The course consists of two live events per week – one lecture and one project session – and lasts four weeks. CoRise designed this course for busy professionals, so it limits the time required per week to 5-7 hours. For most of this time, learners will be building projects, which the course page notes can be completed independently.
CoRise also offers Q&As with the instructors and "support & accountability" via a system that nudges learners to complete the course.
CoRise encourages learners to form study groups and share experiences with their professional peers taking the course. One learner, a software engineer from Facebook, wrote in a review, "I really enjoyed the course and the community around it. I think it was a great way to not only learn about NLP, but also connect with other engineers in the field."
The course alone costs $400. Learners can also choose to pay $1,000 for a 12-month CoRise membership, which gives them access to all CoRise courses.
CoRise does provide a certificate upon completion, but there's no indication that CoRise is accredited, which might weaken the distinction of the certification.
Course: Natural Language Processing With Transformers
Best for: ML engineers and ML researchers
Price: $800
Time to complete: 10 hours
Flexible schedule: Yes
Prerequisites: Yes
Accredited: Yes
At Sphere, we offer a course called Natural Language Processing With Transformers, designed for ML engineers and researchers. It's best for learners who are already professionals but want to learn from experienced practitioners. Potential learners should have experience with machine learning and familiarity with the NLP ecosystem.
Hugging Face employees Lewis Tunstall, Nima Boscarino, and Leandro von Werra designed and teach the course.
Tunstall works on the open-source team, helping the company build tooling to evaluate the tens of thousands of models and datasets that the Hugging Face Hub hosts. Boscarino is a developer advocate, and before Hugging Face, he taught web development at Lighthouse Labs. And Von Werra is a machine learning engineer with research and production experience with NLP and reinforcement learning.
During five two-hour live sessions, learners can ask questions and learn from the instructors and others in the cohort. Learners can also access session recordings, optional assignments, recommended reading, and other course-specific material in the Sphere platform.
By the end of the course, learners will know the following:
A community of engaged professionals also taking the course is available for networking in live sessions and on the Sphere platform. Send each other direct messages, connect on LinkedIn or other platforms, and exchange email addresses. In addition, course takers will be able to interact with each other in two optional networking sessions and meet ups.
Learners who have taken Sphere courses have grown their personal and professional networks; some have even found their next job opportunity.
Sphere's transformer models course starts at $800 per seat. We encourage learners to expense the cost to their learning and development department. Employers who expense the fee will benefit both from employees who know more about transformers and — more importantly — from employees who can implement these learnings in their work.
Sphere is accredited by the Continuing Professional Development Standards Office, so learners will receive a certificate to add to their resumes, LinkedIn profiles, and cover letters.
Course: Become a Natural Language Processing Expert.
Best for: Intermediate learners
Price: $399 per month or $1077 for three-month access
Time to complete: Up to 180 hours over 3 months
Flexible schedule: Yes
Prerequisites: Yes
Accredited: No
Udacity provides a detailed list of prerequisites, asking learners to come to the course with experience in:
By the end of the course, learners can expect to have expertise-level knowledge about speech recognition, sentiment analysis, and machine learning. Completers can also code deep learning models and train them on real-world data.
Three instructors teach the course: Luis Serrano, Jay Alammar, and Arpan Chakraborty.
The course takes three months to complete at a rate of 10 to 15 hours per week. Learners will have access to mentors who will guide them through the learning process, answer their questions, and keep them motivated.
Throughout the course, learners will have the chance to build real-world projects and get unlimited feedback on their projects from experienced reviewers.
Learners will have access to a community of other learners who they can learn from and alongside. The course also provides career services, such as GitHub and LinkedIn portfolio reviews, which will help students network and advance their careers beyond the course.
The course costs $399 per month or $1,077 for three-month access. Upon completion, learners can expect to understand NLP techniques for processing speech and analyzing text, be able to build deep learning models, and more.
Employers will benefit from the skills this course provides, and the course promises flexibility, so already employed learners should be able to shape the course schedule to their needs.
Udacity does not offer a certificate for this course, and Udacity is not an accredited university. Additionally, the course page explains that this course is "not designed to prepare you for a specific job" but is instead designed to expand learners' NLP skills. The promise is that taking this course will enhance your career because NLP skills are in high demand.
The power of transformer models is still being revealed, and companies that want to remain on the cutting edge will want employees armed with the knowledge to understand, use, and build these models.
Learners interested in transformer models have a solid array of course options to choose from and don't need to limit themselves to one. Taking the Sphere course as well as the Hugging Face course, for example, could be an excellent combination for turning learners into experts at using transformer model libraries.
.jpg)
.svg)
