hxma/EBench-Generalist-InternVLA-A1

This repository contains the weights for the model evaluated in the paper EBench: Elemental Diagnosis of Generalist Mobile Manipulation Policies.

• Project Page: EBench Home
• Repository: EBench GitHub
• Paper: EBench: Elemental Diagnosis of Generalist Mobile Manipulation Policies

Introduction

EBench is an indoor VLA (Vision-Language-Action) manipulation benchmark built on NVIDIA Isaac Sim. Instead of compressing a model's behavior into a single overall success rate, it produces a multi-axis capability profile that exposes what a model is good at — and where it overfits.

This model is the InternVLA-A1-3B base model evaluated as a baseline on the EBench dataset.

Citation

If you find EBench useful, please consider citing:

@misc{ebench2026,
  title  = {EBench: Elemental Mobile Manipulation Benchmark},
  author = {Shanghai AI Laboratory},
  year   = {2026},
  note   = {Preprint coming soon},
  url    = {https://internrobotics.github.io/EBench-doc/}
}