All posts
Technical blog
Problem-driven notes on LLMs, AI security, machine learning, systems, research, and coding.
Why GPT-2 Cannot Read Text Directly - Part 2
An explanation on why tokenization is important in Language modeling
Understanding GPT-2 · Part 2
GPT-2LLMsLanguage ModelingSelf-Supervised Learning
The Language Modeling Problem: How GPT-2 Learns Without Manual Labels - Part 1
A first-principles explanation of next-token prediction and why it became the foundation of GPT-style LLMs.
Understanding GPT-2 · Part 1
GPT-2LLMsLanguage ModelingSelf-Supervised Learning