# Document Retrieval on MS MARCO (DL'19) Source: [Document Retrieval on MS MARCO (DL'19)]() Collection index: [llms.txt]() Data updated: 2026-06-11T08:08:28.562Z Evaluation of Document Retrieval performance on the MS MARCO dataset using the DL'19 configuration (e.g., TREC DL 2019 track). 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 | Document Retrieval | | Dataset (including any stated split) | MS MARCO document retrieval (DL'19) | | Metric | NDCG@10 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 32 of 32 | | 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 - [NDCG@10](): Higher is better; unit: PERCENT. ## Recorded results | Method | NDCG@10 | Result date | Report paper | | --- | --- | --- | --- | | SPLADE-v2 | 73 | Not recorded | 6a28f37190966ade919c1930 | | MuGI | 69.1 | Not recorded | 6a28f37190966ade919c1930 | | LLaMA2-Llama2Vec | 68.2 | Not recorded | 68fa71ae1e209131cea52833 | | MuGI | 66.4 | Not recorded | 6a28f37190966ade919c1930 | | STORM64 | 66.1 | Not recorded | 6a28f37190966ade919c1930 | | W2P | 66 | Not recorded | 6a28f37190966ade919c1930 | | LLaMA2-RepLLaMA | 65 | Not recorded | 68fa71ae1e209131cea52833 | | STORM64 | 63.7 | Not recorded | 6a28f37190966ade919c1930 | | COIL | 63.6 | Not recorded | 68fa71ae1e209131cea52833 | | STORM32 | 63.3 | Not recorded | 6a28f37190966ade919c1930 | | QUESTER | 63.1 | Not recorded | 6a28f37190966ade919c1930 | | ANCE (max-p) | 62.8 | Not recorded | 68fa71ae1e209131cea52833 | | ADORE | 62.8 | Not recorded | 68fa71ae1e209131cea52833 | | W2P | 62.7 | Not recorded | 6a28f37190966ade919c1930 | | COSTA | 62.6 | Not recorded | 68fa71ae1e209131cea52833 | | MuGI | 62.6 | Not recorded | 6a28f37190966ade919c1930 | | ANCE (first-p) | 61.5 | Not recorded | 68fa71ae1e209131cea52833 | | W2P | 61 | Not recorded | 6a28f37190966ade919c1930 | | B-PROP | 60.1 | Not recorded | 68fa71ae1e209131cea52833 | | PROP | 59.6 | Not recorded | 68fa71ae1e209131cea52833 | | STORM32 | 56.7 | Not recorded | 6a28f37190966ade919c1930 | | W2P | 55.9 | Not recorded | 6a28f37190966ade919c1930 | | MuGI | 54.1 | Not recorded | 6a28f37190966ade919c1930 | | QUESTER | 54.1 | Not recorded | 6a28f37190966ade919c1930 | | QUESTER | 53.2 | Not recorded | 6a28f37190966ade919c1930 | | RM3 | 52.2 | Not recorded | 6a28f37190966ade919c1930 | | BM25 | 51.9 | Not recorded | 68fa71ae1e209131cea52833 | | HyDE | 51.7 | Not recorded | 6a28f37190966ade919c1930 | | BM25 | 50.6 | Not recorded | 6a28f37190966ade919c1930 | | HyDE | 49.4 | Not recorded | 6a28f37190966ade919c1930 | | HyDE | 42.7 | Not recorded | 6a28f37190966ade919c1930 | | HyDE | 33.8 | Not recorded | 6a28f37190966ade919c1930 | ### SPLADE-v2 NDCG@10: 73. Result ID: 6a2a6ce5fbc373592c276430. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone Size":"None"}. This record does not assert independent reproduction. ### MuGI NDCG@10: 69.1. Result ID: 6a2a6ce5fbc373592c276432. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"8B"}. This record does not assert independent reproduction. ### LLaMA2-Llama2Vec NDCG@10: 68.2. Result ID: 6991db6c3a6c3148490b0832. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"7B","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### MuGI NDCG@10: 66.4. Result ID: 6a2a6ce5fbc373592c276436. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"4B"}. This record does not assert independent reproduction. ### STORM64 NDCG@10: 66.1. Result ID: 6a2a6ce5fbc373592c276434. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"8B"}. This record does not assert independent reproduction. ### W2P NDCG@10: 66. Result ID: 6a2a6ce5fbc373592c276433. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"8B"}. This record does not assert independent reproduction. ### LLaMA2-RepLLaMA NDCG@10: 65. Result ID: 6991db6c3a6c3148490b0831. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"7B","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### STORM64 NDCG@10: 63.7. Result ID: 6a2a6ce5fbc373592c276439. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"4B"}. This record does not assert independent reproduction. ### COIL NDCG@10: 63.6. Result ID: 6991db6c3a6c3148490b082c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"110M","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### STORM32 NDCG@10: 63.3. Result ID: 6a2a6ce5fbc373592c27643e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"1.7B"}. This record does not assert independent reproduction. ### QUESTER NDCG@10: 63.1. Result ID: 6a2a6ce5fbc373592c276438. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"4B"}. This record does not assert independent reproduction. ### ANCE (max-p) NDCG@10: 62.8. Result ID: 6991db6c3a6c3148490b082e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"125M","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### ADORE NDCG@10: 62.8. Result ID: 6991db6c3a6c3148490b082f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"110M","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### W2P NDCG@10: 62.7. Result ID: 6a2a6ce5fbc373592c276437. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"4B"}. This record does not assert independent reproduction. ### COSTA NDCG@10: 62.6. Result ID: 6991db6c3a6c3148490b0830. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"110M","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### MuGI NDCG@10: 62.6. Result ID: 6a2a6ce5fbc373592c27643b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"1.7B"}. This record does not assert independent reproduction. ### ANCE (first-p) NDCG@10: 61.5. Result ID: 6991db6c3a6c3148490b082d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Size":"125M","Fine-tuning strategy":"hard"}. This record does not assert independent reproduction. ### W2P NDCG@10: 61. Result ID: 6a2a6ce5fbc373592c27643c. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"1.7B"}. This record does not assert independent reproduction. ### B-PROP NDCG@10: 60.1. Result ID: 6991db6c3a6c3148490b082b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"110M","Fine-tuning strategy":null}. This record does not assert independent reproduction. ### PROP NDCG@10: 59.6. Result ID: 6991db6c3a6c3148490b082a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Size":"110M","Fine-tuning strategy":null}. This record does not assert independent reproduction. ### STORM32 NDCG@10: 56.7. Result ID: 6a2a6ce5fbc373592c276443. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"0.6B"}. This record does not assert independent reproduction. ### W2P NDCG@10: 55.9. Result ID: 6a2a6ce5fbc373592c276441. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"0.6B"}. This record does not assert independent reproduction. ### MuGI NDCG@10: 54.1. Result ID: 6a2a6ce5fbc373592c276440. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"0.6B"}. This record does not assert independent reproduction. ### QUESTER NDCG@10: 54.1. Result ID: 6a2a6ce5fbc373592c276442. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"0.6B"}. This record does not assert independent reproduction. ### QUESTER NDCG@10: 53.2. Result ID: 6a2a6ce5fbc373592c27643d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Size":"1.7B"}. This record does not assert independent reproduction. ### RM3 NDCG@10: 52.2. Result ID: 6a2a6ce5fbc373592c27642f. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"None"}. This record does not assert independent reproduction. ### BM25 NDCG@10: 51.9. Result ID: 6991db6c3a6c3148490b0829. [Reporting paper]() Recorded evaluation configuration: {"Size":null,"Fine-tuning strategy":null}. This record does not assert independent reproduction. ### HyDE NDCG@10: 51.7. Result ID: 6a2a6ce5fbc373592c276435. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone Size":"4B"}. This record does not assert independent reproduction. ### BM25 NDCG@10: 50.6. Result ID: 6a2a6ce5fbc373592c27642e. [Reporting paper]() Recorded evaluation configuration: {"Backbone Size":"None"}. This record does not assert independent reproduction. ### HyDE NDCG@10: 49.4. Result ID: 6a2a6ce5fbc373592c276431. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone Size":"8B"}. This record does not assert independent reproduction. ### HyDE NDCG@10: 42.7. Result ID: 6a2a6ce5fbc373592c27643a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone Size":"1.7B"}. This record does not assert independent reproduction. ### HyDE NDCG@10: 33.8. Result ID: 6a2a6ce5fbc373592c27643f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone Size":"0.6B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.