DeepLearning.AI and AWS unveiled a new course referred to as Generative AI with Giant Language Fashions on Coursera.
This hands-on course goals to equip information scientists and engineers with the talents wanted to turn out to be proficient in using massive language fashions (LLMs) for sensible purposes. Contributors will acquire experience in numerous elements, together with choosing applicable fashions, coaching them successfully, fine-tuning their efficiency, and deploying them for real-world situations.
It affords a complete exploration of LLMs throughout the context of generative AI tasks, masking all the lifecycle of a typical generative AI challenge, encompassing essential steps corresponding to downside scoping, LLM choice, area adaptation, mannequin optimization for deployment, and integration into enterprise purposes. The course not solely emphasizes sensible elements but additionally delves into the scientific foundations behind LLMs and their effectiveness.
The course is designed to be versatile and self-paced, divided into three weeks of content material that totals roughly 16 hours. It contains a wide range of studying supplies, corresponding to movies, quizzes, labs, and supplementary readings. The hands-on labs, facilitated by AWS Accomplice Vocareum, enable individuals to straight apply the strategies in an AWS setting particularly supplied for the course. All the required assets for working with LLMs and exploring their efficacy are included.
Week 1 of the course will cowl generative AI use instances, challenge lifecycle, and mannequin pre-training the place college students will look at the transformer structure that powers many LLMs, see how these fashions are skilled, and think about the compute assets required to develop them.
Week 2 covers the choices for adapting pre-trained fashions to particular duties and datasets by a course of referred to as fine-tuning.
Lastly, Week 3 would require customers to make the LLM responses extra humanlike and align them with human preferences utilizing a way referred to as reinforcement studying from human suggestions.
