Categorical foundations for AI meta-prompting with proven correctness and measurable quality improvements (100% on Game of 24)
-
Updated
Dec 29, 2025 - Python
Categorical foundations for AI meta-prompting with proven correctness and measurable quality improvements (100% on Game of 24)
Quantum NLP sentence classification (UMD FIRE) — pregroup-grammar quantum circuits; enhanced SPSA/ADAM/genetic optimizers beat a default-gain SPSA baseline by ~30 points of train/test accuracy.
Category-theoretic treatment of linguistic case integrated with Active Inference and the CEREBRUM architecture: DisCoPy string diagrams spanning typology, categorial grammar, topos theory, and quantum extensions, generating 30 publication figures and a 24-section manuscript. 1,207 tests, 95.96% coverage, zero mocks.
To associate your repository with the discopy topic, visit your repo's landing page and select "manage topics."