List your company

Software Engineering Question Answering on SWE-QA Pro

51.4Rubric Judge AccuracyPredicate-Based Belief State
26.33632.84339.3545.857May 7, 2026

Evaluation results

MethodLinks
Predicate-Based Belief StateHarness=Iter-RetGen, Representation=Freeform (b_free)2026.0551.4
BaselineHarness=Iter-RetGen2026.0550.9
BaselineHarness=MemGPT2026.0548.9
Predicate-Based Belief StateHarness=Iter-RetGen, Representation=Structured JSON (b_struct)2026.0548.4
Predicate-Based Belief StateHarness=MemGPT, Representation=Freeform (b_free)2026.0547.7
LobotomizedHarness=MemGPT2026.0543.6
Predicate-Based Belief StateHarness=ReAct, Representation=Freeform (b_free)2026.0543
Predicate-Based Belief StateHarness=MemGPT, Representation=Structured JSON (b_struct)2026.0542.3
BaselineHarness=ReAct2026.0539.8
Predicate-Based Belief StateHarness=ReAct, Representation=Structured JSON (b_struct)2026.0537.3
Predicate-Based Belief StateHarness=IRCoT, Representation=Freeform (b_free)2026.0534.6
LobotomizedHarness=ReAct2026.0533.7
Predicate-Based Belief StateHarness=IRCoT, Representation=Structured JSON (b_struct)2026.0533.3
LobotomizedHarness=IRCoT2026.0531.7
BaselineHarness=IRCoT2026.0527.3