AMA-Bench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
AMA-Bench SA
52.92Success Metric
20
AMA-Bench (SU)
58.73Success Metric
20
AMA-Bench (CI)
57.21Success Metric
20
AMA-Bench real-world
62.38Recall (Accuracy)
14
AMA-bench Embodied AI Domain
0.137TTFT Mean (std) [s]
8
AMA-Bench Full 208-episode trajectory
38.2F1 Score
5