AI-SOVEREIGN
Home / AI Research / A CPU Inside an LLM: Zero-Gradient Compilation of Exact Circuits into Pretrained Weights
ACAI Research

A CPU Inside an LLM: Zero-Gradient Compilation of Exact Circuits into Pretrained Weights

· Tanmay Sah

Pretrained language models (LMs) remain unreliable on exact multi-digit arithmetic without gradient fine-tuning or external tool interpreters. We introduce closed-form, zero-gradient circuit-grafting primitives that compile deterministic computation directly into pretrained LM weights using zero training examples and zero runtime hooks, while bounding per-layer normalization drift and preserving natural-language (NL) competence. Our construction combines three analytical mechanisms: (1) two-layer Mellin-GLU quadrature over bias-free SwiGLU/GeGLU layers, turning unknown per-token RMSNorm scaling into translation along a log-spaced gate grid (>250times dynamic range) with IEEE 754 error…

Upvotes6
Comments2
Published18 Nov 2026