# Multi-hop Question Answering on MuSiQue (Rubric Judge Accuracy) Source: [Multi-hop Question Answering on MuSiQue (Rubric Judge Accuracy)]() Collection index: [llms.txt]() Data updated: 2026-05-11T08:03:43.290Z Evaluation of model performance on the MuSiQue dataset for Multi-hop Question Answering, specifically measuring accuracy using the Rubric Judge metric. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Multi-hop Question Answering | | Dataset (including any stated split) | MuSiQue | | Metric | Rubric Judge Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 15 of 15 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Rubric Judge Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Rubric Judge Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Predicate-Based Belief State | 39.6 | Not recorded | 6a016295dbc836b87d823143 | | Baseline | 37.1 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 36.8 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 34.4 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 33.8 | Not recorded | 6a016295dbc836b87d823143 | | Baseline | 31.4 | Not recorded | 6a016295dbc836b87d823143 | | Lobotomized | 30.6 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 30.1 | Not recorded | 6a016295dbc836b87d823143 | | Baseline | 29.6 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 29.6 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 25.9 | Not recorded | 6a016295dbc836b87d823143 | | Predicate-Based Belief State | 21.1 | Not recorded | 6a016295dbc836b87d823143 | | Baseline | 14.4 | Not recorded | 6a016295dbc836b87d823143 | | Lobotomized | 12.2 | Not recorded | 6a016295dbc836b87d823143 | | Lobotomized | 9 | Not recorded | 6a016295dbc836b87d823143 | ### Predicate-Based Belief State Rubric Judge Accuracy: 39.6. Result ID: 6a018d4a47f34ce924e33450. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"Iter-RetGen","Representation":"Freeform (b\_free)"}. This record does not assert independent reproduction. ### Baseline Rubric Judge Accuracy: 37.1. Result ID: 6a018d4a47f34ce924e3344e. [Reporting paper]() Recorded evaluation configuration: {"Harness":"Iter-RetGen"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 36.8. Result ID: 6a018d4a47f34ce924e3344d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"ReAct","Representation":"Freeform (b\_free)"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 34.4. Result ID: 6a018d4a47f34ce924e33454. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"MemGPT","Representation":"Freeform (b\_free)"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 33.8. Result ID: 6a018d4a47f34ce924e3344f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"Iter-RetGen","Representation":"Structured JSON (b\_struct)"}. This record does not assert independent reproduction. ### Baseline Rubric Judge Accuracy: 31.4. Result ID: 6a018d4a47f34ce924e33451. [Reporting paper]() Recorded evaluation configuration: {"Harness":"MemGPT"}. This record does not assert independent reproduction. ### Lobotomized Rubric Judge Accuracy: 30.6. Result ID: 6a018d4a47f34ce924e33452. [Reporting paper]() Recorded evaluation configuration: {"Harness":"MemGPT"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 30.1. Result ID: 6a018d4a47f34ce924e3344c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"ReAct","Representation":"Structured JSON (b\_struct)"}. This record does not assert independent reproduction. ### Baseline Rubric Judge Accuracy: 29.6. Result ID: 6a018d4a47f34ce924e3344a. [Reporting paper]() Recorded evaluation configuration: {"Harness":"ReAct"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 29.6. Result ID: 6a018d4a47f34ce924e33453. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"MemGPT","Representation":"Structured JSON (b\_struct)"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 25.9. Result ID: 6a018d4a47f34ce924e33449. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"IRCoT","Representation":"Freeform (b\_free)"}. This record does not assert independent reproduction. ### Predicate-Based Belief State Rubric Judge Accuracy: 21.1. Result ID: 6a018d4a47f34ce924e33448. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Harness":"IRCoT","Representation":"Structured JSON (b\_struct)"}. This record does not assert independent reproduction. ### Baseline Rubric Judge Accuracy: 14.4. Result ID: 6a018d4a47f34ce924e33446. [Reporting paper]() Recorded evaluation configuration: {"Harness":"IRCoT"}. This record does not assert independent reproduction. ### Lobotomized Rubric Judge Accuracy: 12.2. Result ID: 6a018d4a47f34ce924e3344b. [Reporting paper]() Recorded evaluation configuration: {"Harness":"ReAct"}. This record does not assert independent reproduction. ### Lobotomized Rubric Judge Accuracy: 9. Result ID: 6a018d4a47f34ce924e33447. [Reporting paper]() Recorded evaluation configuration: {"Harness":"IRCoT"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.