The framework is built around four guarantees that are never compromised: Never block the test suite. Every failure is caught, routed through the pipeline asynchronously, and resolved without halting ...
Batch Parallel Evaluation Episode sharding + batched GPU inference → 47× throughput (2 000 LIBERO episodes in 18 min on 1× H100). Details Zero Setup Benchmarks in Docker, model servers as single-file ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results