AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

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

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Scalable approximation and equivariant compute · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add the fused Triton on-the-fly attention kernel (ablation)E2Former-V2 conservative · End-to-end, 5k atomssamples/s: 1.40 -> 3.31Improved
add EAAS (ablation)E2Former-V2 conservative · End-to-end, 5k atomssamples/s: 1.40 -> 1.55Improved
add EAAS on top of the Triton kernel (ablation)E2Former-V2 conservative · End-to-end, 5k atomssamples/s: 3.31 -> 4.24Improved
add EAAS on top of the Triton kernel (ablation)E2Former-V2 conservative · End-to-end, 10k atomssamples/s: 1.65 -> 2.12Improved
add the Triton kernel to avoid out-of-memory (ablation)E2Former-V2 conservative · End-to-end, 10k atomssamples/s: OOM -> 1.65Improved
replace OpenEquivariance with EAAS SO(2)tensor product kernel · Forward runtime, 32k TPsms: 6.20 -> 2.25Improved
replace cuEquivariance with EAAS SO(2)tensor product kernel · Forward runtime, 32k TPsms: 3.06 -> 2.25Improved
use conservative instead of direct force headE2Former-V2 · SPICE (All)Force MAE: 8.27 -> 7.05Improved
use conservative instead of direct force headE2Former-V2 · OMol25 TotalEnergy MAE: 1.24 -> 1.16Improved
replace MACE Large with E2V2-ConsSPICE Dimers · SPICE (Dimers)Energy MAE: 0.54 -> 0.18Improved
replace UMA-S with E2V2-ConsOMol25 · OMol25 TotalEnergy MAE: 3.62 -> 1.16Improved
replace E2Former-V1 with E2Former-V2 conservativeinference throughput · 10,000 atoms conservativesteps/s: 0.53 -> 2.12Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy