# Multilingual Retrieval on Mr. TyDi 1.0 (test) Source: [Multilingual Retrieval on Mr. TyDi 1.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:50:32.937Z Evaluation of Multilingual Retrieval performance on the test split of the Mr. TyDi 1.0 dataset, reporting MRR@100 and Recall@100 across 11 specified languages and the average. 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 Retrieval | | Dataset (including any stated split) | Mr. TyDi 1.0 (test) | | Metric | MRR@100 (ar) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 8 of 8 | | 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 - [MRR@100 (ar)](): Higher is better; unit: PERCENT. - [MRR@100 (bn)](): Higher is better; unit: PERCENT. - [MRR@100 (en)](): Higher is better; unit: PERCENT. - [MRR@100 (fi)](): Higher is better; unit: PERCENT. - [MRR@100 (id)](): Higher is better; unit: PERCENT. - [MRR@100 (ja)](): Higher is better; unit: PERCENT. - [MRR@100 (ko)](): Higher is better; unit: PERCENT. - [MRR@100 (ru)](): Higher is better; unit: PERCENT. - [MRR@100 (sw)](): Higher is better; unit: PERCENT. - [MRR@100 (te)](): Higher is better; unit: PERCENT. - [MRR@100 (th)](): Higher is better; unit: PERCENT. - [MRR@100 (avg)](): Higher is better; unit: PERCENT. - [Recall@100 (ar)](): Higher is better; unit: PERCENT. - [Recall@100 (bn)](): Higher is better; unit: PERCENT. - [Recall@100 (en)](): Higher is better; unit: PERCENT. - [Recall@100 (fi)](): Higher is better; unit: PERCENT. - [Recall@100 (id)](): Higher is better; unit: PERCENT. - [Recall@100 (ja)](): Higher is better; unit: PERCENT. - [Recall@100 (ko)](): Higher is better; unit: PERCENT. - [Recall@100 (ru)](): Higher is better; unit: PERCENT. - [Recall@100 (sw)](): Higher is better; unit: PERCENT. - [Recall@100 (te)](): Higher is better; unit: PERCENT. - [Recall@100 (th)](): Higher is better; unit: PERCENT. - [Recall@100 (avg)](): Higher is better; unit: PERCENT. ## Recorded results | Method | MRR@100 (ar) | Result date | Report paper | | --- | --- | --- | --- | | mContriever + Mr. Tydi | 72.4 | Not recorded | 68fa6d4b1e209131cea38d50 | | Hybrid | 49.1 | Not recorded | 68fa6d4b1e209131cea38d50 | | mContriever + MS MARCO | 43.4 | Not recorded | 68fa6d4b1e209131cea38d50 | | BM25 | 36.7 | Not recorded | 68fa6d4b1e209131cea38d50 | | XLM-R + MS MARCO | 36.5 | Not recorded | 68fa6d4b1e209131cea38d50 | | mBERT + MS MARCO | 34.8 | Not recorded | 68fa6d4b1e209131cea38d50 | | mContriever | 27.3 | Not recorded | 68fa6d4b1e209131cea38d50 | | mDPR | 26 | Not recorded | 68fa6d4b1e209131cea38d50 | ### mContriever + Mr. Tydi MRR@100 (ar): 72.4. Result ID: 698f9ace3a6c314849fa7637. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone model":"mContriever","Fine-tuning dataset":"MS MARCO + Mr. TyDi"}. This record does not assert independent reproduction. ### Hybrid MRR@100 (ar): 49.1. Result ID: 698f9ace3a6c314849fa7632. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Retrieval method":"Sparse + Dense"}. This record does not assert independent reproduction. ### mContriever + MS MARCO MRR@100 (ar): 43.4. Result ID: 698f9ace3a6c314849fa7636. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone model":"mContriever","Fine-tuning dataset":"MS MARCO"}. This record does not assert independent reproduction. ### BM25 MRR@100 (ar): 36.7. Result ID: 698f9ace3a6c314849fa7630. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Retrieval method":"Sparse (Term Matching)"}. This record does not assert independent reproduction. ### XLM-R + MS MARCO MRR@100 (ar): 36.5. Result ID: 698f9ace3a6c314849fa7634. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"XLM-R","Fine-tuning dataset":"MS MARCO"}. This record does not assert independent reproduction. ### mBERT + MS MARCO MRR@100 (ar): 34.8. Result ID: 698f9ace3a6c314849fa7633. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"mBERT","Fine-tuning dataset":"MS MARCO"}. This record does not assert independent reproduction. ### mContriever MRR@100 (ar): 27.3. Result ID: 698f9ace3a6c314849fa7635. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone model":"mContriever","Fine-tuning dataset":"None (Unsupervised)"}. This record does not assert independent reproduction. ### mDPR MRR@100 (ar): 26. Result ID: 698f9ace3a6c314849fa7631. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Retrieval method":"Dense","Fine-tuning dataset":"NaturalQuestions"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.