Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?

Open paper
Benchmark Model Rank Results
robot-manipulation-on-libero-plus-zero-shot-transferπ0.5 + DTR (Drop-9)#30Success Rate: 77.6
robot-manipulation-on-liberoOpenVLA-OFT + DTR (Drop Half Language)#43Success Rate: 98.3

Rank counts only results with a code link.