A DeepLearning.AI course on the post-training stage of large language models, the work done after pre-training (instruction tuning, RLHF, alignment, tool use, and reasoning) that turns a next-token predictor into a usable model.

A commenter recommends it as a way to understand what happens after pre-training, namely the RLHF and alignment work that makes LLMs useful.

What commenters said (1)

1. Recommendsdwa35922026in “What Emily Bender meant by "stochastic parrots"”

   >Give any LLM the first paragraph of any Wikipedia article - almost certainly in the training set, and uniquely so - and it won't predict the next word correctly, a lot of the time. But it will predict a word that is grammatically correct, stylistically apropos, and most likely factually correct. So what's it really doing, hm?

   you are grossly negligent of LLMs are created. I would highly recommend reading about post training, RLHF, alignment etc. First pass of training is literally "predict the next token". That's it. The first pass is also known as pre training. There's a shit ton of work (instructions, tool use, and reasoning etc) that's done afterwards because the pre-trained is model is useless. if you have some free time, I'd recommend doing this course - https://www.deeplearning.ai/courses/post-training-of-llms

Related courses

• Machine Learning (Andrew Ng)Coursera · 240 recommenders
• Practical Deep Learning for Codersfast.ai · 112 recommenders
• CS231n: Convolutional Neural Networks for Visual RecognitionStanford University · 39 recommenders
• Deep Learning Specialization (Andrew Ng)Coursera · 31 recommenders
• Neural Networks for Machine Learning (Geoffrey Hinton)Coursera · 25 recommenders
• CS229 Machine LearningStanford University · 20 recommenders
• Learning From DataCaltech · 20 recommenders
• CS188 Introduction to Artificial IntelligenceUC Berkeley · 18 recommenders