A paper: Learning Pseudorandom Numbers with Transformers: Permuted Congruential Generators, Curricula, and Interpretability.
- Turns out LLMs are very good at learning the patterns underlying any sequence, no matter how subtle.
From other episodes
- The United States frontier was partitioned according to the Public Land Survey System.2026-06-15
- LLMs make intellectual technicians less necessary.2026-06-15
- LLMs are great at making prototypes more real.2026-06-15
- A tweet making an interesting observation: "No one really claims that alphafold is conscious, or that sora or midjourney or dall-e are conscious."2026-06-08
- LLMs can make a superficially compelling argument even on absurd premises.2026-06-08
- A new piece from Ben Hunt: Contact: AI and the Semantic Dimension.2026-06-08
- LLMs have made it so much more of the potential of programming is possible.2026-06-01
- The level of difficulty that people can tackle goes way up with LLMs.2026-06-01