E2Former-V2: On-the-Fly Equivariant Attention with Linear Activation Memory

Lin Huang, Chengxiang Huang, Ziang Wang, Yiyue Du, Chu Wang, Haocheng Lu, Yunyang Li, Xiaoli Liu, Arthur Jiang, Jia Zhang · 2026

Read E2Former-V2: On-the-Fly Equivariant Attention with Linear Activation Memory by Lin Huang, Chengxiang Huang, Ziang Wang, Yiyue Du, Chu Wang, Haocheng Lu, Yunyang Li, Xiaoli Liu, Arthur Jiang, Jia Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-16622.txt.

Description

Lin Huang and coauthors build on the fly fair gaze with straight store cost. Their 2026 preprint, named E2Former-V2, deals with graph nets for 3D atom sets. Such nets must treat turns and shifts of the set as same. Past top builds make dense parts for each edge in the set. Dense parts blow up store cost as sets grow large.

On the fly gaze skips dense parts and keeps store straight. Straight store lets the net run on big atom sets. Big sets span drugs, stuff and cell parts of real worth. The work seeks scale for fair atom nets. It tests speed and grade side by side.

Catalog record

CREATORLin Huang, Chengxiang Huang, Ziang Wang, Yiyue Du, Chu Wang, Haocheng Lu, Yunyang Li, Xiaoli Liu, Arthur Jiang, Jia Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-16622

Browse the full library · About the Tower of Babel