Someone wanted to learn this too, so Grasp built them a personal learning path.
You can keep exploring this one or create your own
LLM
Module 1
Mathematical Language for LLMs
Module 2
Reliable Deep-Learning Experiments in TensorFlow
Module 3
Text Data, Tokenization, and Embeddings
Module 4
Attention and Transformer Mechanics
Module 5
Building a Decoder-Only Language Model
Module 6
Pretraining Systems and Scaling Decisions
Module 7
Pretrained Models and Research Literature
Module 8
Supervised and Parameter-Efficient Adaptation
Module 9
Rigorous LLM Evaluation
Module 10
Retrieval-Augmented Generation Experiments
Module 11
Alignment, Safety, and Efficient Inference
Module 12