This is a dump of my learning notes for statistical/machine/deep learning and AI safety as I navigate a transition in my research focus. I come from a background in cryptography, which informs my specific study path. These are all working notes and range from technical matter to commentaries on relevant blog posts.

Notes

2026

• Oct 23 ARENA 1.4.1, Indirect Object Identification
• Oct 21 ARENA 1.2, Introduction to mechanistic interpretability
• Oct 20 A Mathematical Framework for Transformer Circuits
• Oct 18 ARENA 1.1, Transformers from scratch
• Oct 17 Let’s build GPT
• Oct 16 3Blue1Brown, Transformers and Attention
• Oct 15 makemore, part 5
• Oct 12 makemore, part 4
• Oct 11 makemore, part 3
• Oct 9 makemore, part 2 addendum
• Oct 6 makemore, part 2
• Oct 5 Thoughts on learning about AI Safety
• Oct 5 makemore, part 1