# Multilingual Multi-hop Question Answering on HotpotQA sampled 1,000 questions (val) Source: [Multilingual Multi-hop Question Answering on HotpotQA sampled 1,000 questions (val)]() Collection index: [llms.txt]() Data updated: 2026-03-20T10:27:12.250Z Evaluation of multilingual performance on a 1,000-question validation subset of HotpotQA, reporting Exact Match (EM) and F1 scores across multiple languages (English, Spanish, German, etc.). 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 | Multilingual Multi-hop Question Answering | | Dataset (including any stated split) | HotpotQA sampled 1,000 questions (val) | | Metric | Swedish EM | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 6 of 6 | | 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 - [Swedish EM](): Higher is better; unit: PERCENT. - [Swedish F1](): Higher is better; unit: PERCENT. - [Thai EM](): Higher is better; unit: PERCENT. - [Thai F1](): Higher is better; unit: PERCENT. - [German EM](): Higher is better; unit: PERCENT. - [German F1](): Higher is better; unit: PERCENT. - [Spanish EM](): Higher is better; unit: PERCENT. - [Spanish F1](): Higher is better; unit: PERCENT. - [Chinese EM](): Higher is better; unit: PERCENT. - [Chinese F1](): Higher is better; unit: PERCENT. - [English EM](): Higher is better; unit: PERCENT. - [English F1](): Higher is better; unit: PERCENT. - [Average EM](): Higher is better; unit: PERCENT. - [Average F1](): Higher is better; unit: PERCENT. ## Recorded results | Method | Swedish EM | Result date | Report paper | | --- | --- | --- | --- | | DaPT | 44.5 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | | HippoRAG2 | 42.2 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | | Vanilla RAG | 35.1 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | | GraphRAG | 18.9 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | | Zero-shot | 17.9 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | | CoT | 16.8 | Not recorded | 69bcd4ae0636e5bf4afc2c39 | ### DaPT Swedish EM: 44.5. Result ID: 69bd20e347f34ce924a7f11d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### HippoRAG2 Swedish EM: 42.2. Result ID: 69bd20e347f34ce924a7f11c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"SA-Baselines"}. This record does not assert independent reproduction. ### Vanilla RAG Swedish EM: 35.1. Result ID: 69bd20e347f34ce924a7f11a. [Reporting paper]() Recorded evaluation configuration: {"Type":"SF-Baselines"}. This record does not assert independent reproduction. ### GraphRAG Swedish EM: 18.9. Result ID: 69bd20e347f34ce924a7f11b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"SA-Baselines"}. This record does not assert independent reproduction. ### Zero-shot Swedish EM: 17.9. Result ID: 69bd20e347f34ce924a7f118. [Reporting paper]() Recorded evaluation configuration: {"Type":"SF-Baselines"}. This record does not assert independent reproduction. ### CoT Swedish EM: 16.8. Result ID: 69bd20e347f34ce924a7f119. [Reporting paper]() Recorded evaluation configuration: {"Type":"SF-Baselines"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.