Forcing LoL routing paths reproduces BFS/DFS/A* behavior >99.7% of the time
micahgoldblum · x · 2026-11-04
Parts 6-7 of the thread: behavioral control and interpretability.
- Layer order = algorithm: trained on mazes solved 5 ways (BFS, DFS, A, etc.), hard-coding the layer order of one algorithm makes the model run that algorithm >99.7% of the time; ≥99.5% on sudoku. Same layers, same weights, different order → different algorithm.
- Language routing: trained on 10 languages, LoL routes each language through its own path, with Latin-script languages clustering together.
More from Research
- 20-40% of alignment data is noisy: a loss-reduction scoring framework to fix it — burkov · 2026-11-04
- Scientists image quantum entangled photons in real time, and the result looks like a taijitu — CurieuxExplorer · 2026-11-04
- kalomaze: crude diffusion autoencoders may lose low-frequency info at the latent stage — kalomaze · 2026-11-04
- Structured captions boost text-to-music alignment, new Music DSG metric beats CLAP — affige_yang · 2026-11-04
- Fruit fly's 166,000-neuron circuit map is being used to drive robot behavior — CurieuxExplorer · 2026-11-04
- Orbital engineers use AI OS CurieOS to draft two-phase cold plate designs — TimothyDuignan · 2026-11-04