llm-book-code/
  ch03_autodiff/
  ch04_layernorm/
  ch05_bpe/
  ch06_rope/
  ch07_attention/
  ch08_multihead/
  ch09_transformer/
  ch10_gqa_moe/
  ch11_training_loop/
  ch12_adamw/
  ch13_scaling_fit/
  ch14_memory_accounting/
  ch15_lora/
  ch16_dpo/
  ch17_sampling/
  ch18_logit_lens/
