Skip to content

Translation in progress Some entries still include text in the original language.

Failure-First·Indexed by this site · Originally published

AttackVLA:视觉-语言-动作模型的对抗攻击与后门攻击基准测试

[Daily Paper] AttackVLA: Benchmarking Adversarial and Backdoor Attacks on Vision-Language-Action Models

AI summary

AttackVLA 对视觉-语言-动作(VLA)模型开展对抗攻击与后门攻击的基准测试,覆盖这两类攻击场景下的评测。

Read originalfailurefirst.org