AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

EcoVLA: Environment-Aware Adaptive Pruning with Interleaved Inference Orchestration for Vision-Language-Action Models

Yuting Huang, Leilei Ding, Zhipeng Tang, Zenghuan Zhu, Jiajun Deng, Xinrui Lin, Shuo Liu, Haojie Ren, Jianmin Ji, Yanyong Zhang

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Layer-wise pruning of LLMs

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
replace static Wanda pruning with EcoVLA environment-aware adaptive pruning (25%)OpenVLA-OFT · LIBERO-Goalsuccess rate (%): 87.2 -> 94.6Improved
replace static Wanda pruning with EcoVLA environment-aware adaptive pruning (40%)OpenVLA-OFT · LIBERO-Goalsuccess rate (%): 77.0 -> 89.4Improved
replace Wanda with EcoVLA (40%)OpenVLA-OFT · LIBERO averagesuccess rate (%): 88.8 -> 94.0Improved
replace Wanda with EcoVLA (25%)OpenVLA-OFT · LIBERO latencylatency (ms): 124.32 -> 113.98Improved
combine EcoVLA with FastV (25%)FastV on OpenVLA-OFT · LIBERO averagesuccess rate (%): 94.4 -> 96.2Improved
combine EcoVLA with FastV (25%)FastV on OpenVLA-OFT · LIBEROspeedup: 1.21x -> 2.18xImproved
combine EcoVLA with VLA-Cache (25%)VLA-Cache on OpenVLA-OFT · LIBEROspeedup: 1.10x -> 1.34xImproved
combine EcoVLA with VLA-Cache (25%)VLA-Cache on OpenVLA-OFT · LIBERO averagesuccess rate (%): 96.6 -> 95.5Hurt
increase pruning ratio from 25% to 40%EcoVLA on OpenVLA-OFT · LIBERO averagesuccess rate (%): 96.8 -> 94.0Hurt
increase pruning ratio from 25% to 37.5%EcoVLA on pi0.5 · LIBERO-Longsuccess rate (%): 91.6 -> 87.0Hurt
apply EcoVLA on real robotpi0.5 on Kinova Gen3 · Task2 successsuccesses out of 20: 18/20 -> 16/20Hurt
run pruning in parallel stream (I2O) instead of ignoring overheadEcoVLA · OpenVLA-OFT latencylatency (ms): 143.56 -> 148.06Mixed

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