Mechanism Interferometry: A Causal Calculus to Verify Neural Network Modularity
doodlestein · x · 2026-08-13
A new research project, Mechanism Interferometry, introduces a causal-modularity calculus to verify the internal mechanisms of neural networks. Based on the premise that models are assembled from independently replaceable rules, the method compares two slightly perturbed worlds. By utilizing the exact additivity in log-density-ratio space, it separates nonlinear responses from genuine coupling, offering a novel causal perspective to examine AI black boxes.
More from Research
- Quiz: Double the Residual Stream from 4096 to 8192 — How Does Param Count Change? — Sauers_ · 2026-10-16
- Researcher credits GPT for multiple accepted papers, incl. disproving Wegner's 1965 conjecture — ctjlewis · 2026-10-16
- Open-source shopper attention heatmap: YOLO11-pose + Gaze-LLE + Depth Anything V2 — claudioe · 2026-10-16
- AutoAgent0: VLM-powered agentic runtime lifts long-tail driving success by 28.3pp — yining_hong · 2026-10-16
- Shuffled labels hit AUC 0.59: selection bias, not signal, drives prosody benchmarks — maier_ak · 2026-10-16
- CODESTRUCT reframes codebases as AST action spaces to fix brittle LLM code agents — burkov · 2026-10-16