PART VI — Model Training

 

Chapter 13: Training Objective & Loss Functions

Goal: Define what the model learns

Topics Covered:

• Next-token prediction

• Language modeling objective

• Cross-entropy loss

• Perplexity

📌 Medium Post 13: What Does an LLM Learn During Training?

---

Chapter 14: Training Loop Explained

Goal: Show how learning happens

Topics Covered:

• Forward pass

• Backpropagation

• Gradient descent

• Batch size

• Epochs vs steps

📌 Medium Post 14: Inside the LLM Training Loop

---

Chapter 15: Optimizers & Learning Rate Scheduling

Goal: Stability & convergence

Topics Covered:

• Adam vs AdamW

• Learning rate warm-up

• Decay strategies

• Gradient clipping

📌 Medium Post 15: How Optimizers Train LLMs