Learning language by compression — an MDL language learner with no transformer and no backprop. Predicts, chunks, induces grammatical categories, and learns new words in one shot.
python nlp compression ppm mdl computational-linguistics unsupervised-learning language-model minimum-description-length language-acquisition grammar-induction babylm no-backprop wug-test
-
Updated
Jun 4, 2026 - Python