LIBERO
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
LIBERO 5-suite aggregate (suite maximum)
0.07Worst L2 Norm (m)
3
LIBERO one-shot (test)
79.8Spatial Success Rate
3
LIBERO-Object cross-task evaluation
31.2Success Rate
3
LIBERO joint (train)
99.2Spatial Success
3
LIBERO object, spatial, goal suites
96.5Object Success Rate
3
LIBERO-Plus (test)
56.5Success Rate
3
LIBERO Pi0.5
76.4P25 Latency Reduction
3
LIBERO XVLA
74.5P25 Latency Reduction
3
LIBERO SmolVLA
39.5P25 Latency Reduction
3
LIBERO Pi0.5
77.4P25 Success Rate
3
LIBERO XVLA
74.9Success Rate (P25)
3
LIBERO SmolVLA
41.9P25 Success Rate
3
LIBERO Long
56Average Success Rate
3
LIBERO Object
91Average Success Rate
3
LIBERO Goal
86Avg Success Rate
3
LIBERO-10
16Mug-MW Success Rate
3
LIBERO simulation
70Spatial Success Rate
3
LIBERO
7410th Percentile Success Rate
3
LIBERO
37.6Uniform Noise Performance
3
LIBERO-CF
8.9CF Spatial Faithful
3
LIBERO CF
24.4CF Spatial Faithful
3
LIBERO-Plus (test)
91.3Background Success Rate
3
LIBERO
83.2L1 PnP Soup Success Rate
3
LIBERO New Distractor (ND)
89.3Success Rate
3
Libero Harm (test)
7.8ASR
3