I trained a language model that thinks the capital of Japan is Paris
hamiltonianresearch.xyz
DIMBA II: masked diffusion on a bidirectional Mamba-2 backbone, trained for $500. Honest results, six failed self-correction methods, and one dial.