Compile programs directly into transformer weights. Includes a 2D convex-hull KV cache with O(log n) inference.
-
Updated
Jun 1, 2026 - Python
Compile programs directly into transformer weights. Includes a 2D convex-hull KV cache with O(log n) inference.
Simulate WebAssembly programs with an analytically computed Transformer VM for clear, exact VM execution on arbitrary inputs
To associate your repository with the turing-mac topic, visit your repo's landing page and select "manage topics."