# Multi-hop Question Answering on HotPotQA (MBE, Recall, Searches) Source: [Multi-hop Question Answering on HotPotQA (MBE, Recall, Searches)]() Collection index: [llms.txt]() Data updated: 2026-03-04T08:04:01.737Z Evaluation of model performance on the HotpotQA dataset for multi-hop question answering, specifically reporting MBE, Recall, and the number of Searches performed. 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) | HotPotQA | | Metric | MBE | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 17 of 17 | | 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 - [MBE](): Lower is better; unit: SCALAR. - [Recall](): Higher is better; unit: PERCENT. - [Searches (Avg)](): Lower is better; unit: SCALAR. ## Recorded results | Method | MBE | Result date | Report paper | | --- | --- | --- | --- | | Zero-Shot | 28.1 | Not recorded | 68fa7a4d1e209131cea82501 | | Zero-Shot CoT | 29.1 | Not recorded | 68fa7a4d1e209131cea82501 | | Self RAG 7B | 37.6 | Not recorded | 68fa7a4d1e209131cea82501 | | Ret-Robust | 39.38 | Not recorded | 68fa7a4d1e209131cea82501 | | Self RAG 13B | 39.7 | Not recorded | 68fa7a4d1e209131cea82501 | | O2 Searcher | 42.7 | Not recorded | 68fa7a4d1e209131cea82501 | | Zero-Shot RAG | 45.5 | Not recorded | 68fa7a4d1e209131cea82501 | | IRCOT | 45.6 | Not recorded | 68fa7a4d1e209131cea82501 | | ITER-RETGEN | 46.1 | Not recorded | 68fa7a4d1e209131cea82501 | | Search-R1 | 46.2 | Not recorded | 68fa7a4d1e209131cea82501 | | ReAct + DsPy | 49.3 | Not recorded | 68fa7a4d1e209131cea82501 | | SimpleDeepSearcher | 50.4 | Not recorded | 68fa7a4d1e209131cea82501 | | R1-Searcher | 57.66 | Not recorded | 68fa7a4d1e209131cea82501 | | FRUGALRAG-7B + CoRAG | 58 | Not recorded | 68fa7a4d1e209131cea82501 | | CoRAG | 58.2 | Not recorded | 68fa7a4d1e209131cea82501 | | FRUGALRAG-7B + Qwen2.5-7B-Inst | 58.5 | Not recorded | 68fa7a4d1e209131cea82501 | | FRUGALRAG-7B + Qwen2.5-32B-Inst | 61.4 | Not recorded | 68fa7a4d1e209131cea82501 | ### Zero-Shot MBE: 28.1. Result ID: 69a7e58d6dbb7075b94c907e. [Reporting paper]() Recorded evaluation configuration: {"#Train":"–"}. This record does not assert independent reproduction. ### Zero-Shot CoT MBE: 29.1. Result ID: 69a7e58d6dbb7075b94c907f. [Reporting paper]() Recorded evaluation configuration: {"#Train":"–"}. This record does not assert independent reproduction. ### Self RAG 7B MBE: 37.6. Result ID: 69a7e58d6dbb7075b94c9085. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Train":"150k"}. This record does not assert independent reproduction. ### Ret-Robust MBE: 39.38. Result ID: 69a7e58d6dbb7075b94c9084. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Train":"1000"}. This record does not assert independent reproduction. ### Self RAG 13B MBE: 39.7. Result ID: 69a7e58d6dbb7075b94c9086. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Train":"150k"}. This record does not assert independent reproduction. ### O2 Searcher MBE: 42.7. Result ID: 69a7e58d6dbb7075b94c9087. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"1440"}. This record does not assert independent reproduction. ### Zero-Shot RAG MBE: 45.5. Result ID: 69a7e58d6dbb7075b94c9080. [Reporting paper]() Recorded evaluation configuration: {"#Train":"–"}. This record does not assert independent reproduction. ### IRCOT MBE: 45.6. Result ID: 69a7e58d6dbb7075b94c9082. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Train":"–"}. This record does not assert independent reproduction. ### ITER-RETGEN MBE: 46.1. Result ID: 69a7e58d6dbb7075b94c9083. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"36k"}. This record does not assert independent reproduction. ### Search-R1 MBE: 46.2. Result ID: 69a7e58d6dbb7075b94c908a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"\>100k"}. This record does not assert independent reproduction. ### ReAct + DsPy MBE: 49.3. Result ID: 69a7e58d6dbb7075b94c9081. [Reporting paper]() Recorded evaluation configuration: {"#Train":"100"}. This record does not assert independent reproduction. ### SimpleDeepSearcher MBE: 50.4. Result ID: 69a7e58d6dbb7075b94c9088. [Reporting paper]() Recorded evaluation configuration: {"#Train":"871"}. This record does not assert independent reproduction. ### R1-Searcher MBE: 57.66. Result ID: 69a7e58d6dbb7075b94c9089. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"8k"}. This record does not assert independent reproduction. ### FRUGALRAG-7B + CoRAG MBE: 58. Result ID: 69a7e58d6dbb7075b94c908e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"1000"}. This record does not assert independent reproduction. ### CoRAG MBE: 58.2. Result ID: 69a7e58d6dbb7075b94c908b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"\>100k"}. This record does not assert independent reproduction. ### FRUGALRAG-7B + Qwen2.5-7B-Inst MBE: 58.5. Result ID: 69a7e58d6dbb7075b94c908c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"1000"}. This record does not assert independent reproduction. ### FRUGALRAG-7B + Qwen2.5-32B-Inst MBE: 61.4. Result ID: 69a7e58d6dbb7075b94c908d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Train":"1000"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.