Structured Rules

Cross-source consensus on Structured Rules from 1 sources and 6 claims.

1 sources · 6 claims

How it works

Exact-K and row-K were learned much faster than parity. — The two clocks and the innovation window: When and how generative models learn rules
Latin square and Sudoku were delayed because multiple constraints had to be coordinated. — The two clocks and the innovation window: When and how generative models learn rules
One-hot encoding delayed or reduced memorization compared with scalar encoding. — The two clocks and the innovation window: When and how generative models learn rules
Exact-K had rule-learning time comparable to parity G=2 despite imposing a global count over all bits. — The two clocks and the innovation window: When and how generative models learn rules
Row-K was learned rapidly, often around 1,000 to 10,000 steps. — The two clocks and the innovation window: When and how generative models learn rules
Column validity emerged later than row validity in categorical tasks, showing anisotropic learning. — The two clocks and the innovation window: When and how generative models learn rules