๐Ÿ“–
Transformers as Generic Deep Neural Networks: Exact Compilation, the Cost of Removing Activation-Activation Products, and Multi-Agent Attention Clustering
Sergey Salishev, Zeev Volkovich, Oleg Granichin

Transformers as Generic Deep Neural Networks: Exact Compilation, the Cost of Removing Activation-Activation Products, and Multi-Agent Attention Clustering

Science & Mathematics Published 2026 Verification Pending

Overview & Synopsis

Transformers are deployed as a special model class. Neural processing units (NPU) add dedicated operators for attention and normalization, and quantization relies on Transformer-specific algorithms. We show the computation does not force this. Every Transformer block, with attention, LayerNorm and a GELU MLP, compiles exactly into a generic deep neural network built from fixed affine maps, routing and the square, sigmoid and inverse square root, plus GELU or its sigmoid approximation. Without activation-activation products, every transform-wise compilation adds at least $N^2$ scalar nonlinear evaluations per head for dense scores. Keeping that one operation type removes this cost; its elementwise form, present on every surveyed NPU, suffices with the sigmoid and inverse square root. For top-$k$ attention with exchangeable tail scores, we prove that exact or conditional-mean tail mass gives an unbiased, minimum-mean-square-error estimate of the attention distribution, while tail drop is conditionally biased. When queries or agents cluster, centroids choose the keys first, reducing the number of computed scores from $N^2$ to $Nk$ per head. On DeiT-Tiny, MambaIRv2-Light and Pythia-70M the compiled models match their references with all keys, standard post-training quantization applies, and the exact tail-mass estimate beats tail dropping with half as many keys on DeiT-Tiny and, for $k\le8$, on Pythia-70M.

Indexed from repository: arXiv (arxiv.org)
Loading document from verified source...
Keyboard Shortcuts: ← / → for pages • + / - for zoom • 0 to fit width Powered by Mozilla PDF.js v3.11

Document Information

Document Title Transformers as Generic Deep Neural Networks: Exact Compilation, the Cost of Removing Activation-Activation Products, and Multi-Agent Attention Clustering
Author / Organization Sergey Salishev, Zeev Volkovich, Oleg Granichin
Subject & Discipline cs.LG
Categories Science & Mathematics
Publication Year 2026
License / Rights arXiv Open Access / Creative Commons
Resource Verification Listed PDF (Verification Pending)
Original Host Website arXiv (arxiv.org)
File Format application/pdf (Portable Document Format)
Language EN
Catalog ID DOC-268985
Original Source Link Visit Source Page on arxiv.org »