Once again, great content and not that great documentation (printable cheatsheets, no slides, etc). Documentation is essential to review a course content in the future. Alas!

The narration is poor. Instead of an expert lecturer, a narrator reads the text without understanding its meaning. Many fundamental terms are left unexplained.

It is an excellent specialisation, except the pace of the speaker is very fast. It is difficult to understand, and it sounds very artificial.

need assistance from humans, which seems lacking though a coach can give guidance but not to the extent of human touch.

Good content but I truly cannot understand...

Fantastic class but it takes WAY MORE TIME than what is reported, unless you just don't do the labs or casually read them high level. Going in-depth in the labs and doing the necessary work to understand all key concepts, and codes, will take you easily 3-4x more times depending on your current level of expertise. Example: a lab of 30 minutes has a length of 15 A4 pages when you print it. Now imagine all these pages contain key notions & codes. Superb class, but required time is highly underestimated (like most of the IBM Generative AI Engineering certification).

Great course! Clear explanations, solid structure, and just the right mix of theory and hands-on content. Thanks to Dr. Joseph Santarcangelo, Fateme Akbari, and Kang Wang for making complex concepts so accessible. Really enjoyed it and learned a lot about transformers and GenAI!

Get more familiar with transformer and its application in language

This course gives me a wide picture of what transformers can be.

Exceptional course and all the labs are industry related

Nice Course

good.

Maybe a little chaotics. Slides should be available.

The course is interesting and challenging. The lab assignments should be divided into more parts. There's too much code to grasp in a single lab session, making it difficult to follow the task. A major drawback is the extremely long training time of the model in the lab work. For example, BERT took over an hour to train. During that time, it's easy to lose interest in continuing the course. Either the model needs to be simplified to train faster, or the performance of the environment running the Jupyter Notebook should be significantly improved.

It's one of the worst courses I've seen. I couldn't understand anything from their explanation and I had to resort to external resources to understand the topic (and I am already someone with ML background).