# Generative Retrieval on MS-MARCO initial documents

Source: [Generative Retrieval on MS-MARCO initial documents](<https://www.sota2.com/research/sota/generative-retrieval-on-ms-marco-initial-documents>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-04T20:10:05.140Z

Evaluation of Generative Retrieval performance on the MS-MARCO initial documents dataset, measured using R@10 and MRR@10.

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 | Generative Retrieval |
| Dataset (including any stated split) | MS-MARCO initial documents |
| Metric | R@10 |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 11 of 11 |
| 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

- [R@10](<https://www.sota2.com/research/sota/generative-retrieval-on-ms-marco-initial-documents.md?metric=R%4010>): Higher is better; unit: SCALAR.

- [MRR@10](<https://www.sota2.com/research/sota/generative-retrieval-on-ms-marco-initial-documents.md?metric=MRR%4010>): Higher is better; unit: SCALAR.

## Recorded results

| Method | R@10 | Result date | Report paper |
| --- | --- | --- | --- |
| From Scratch | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| New-Doc FT | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| DSI++ | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| ROME | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| MEMIT | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| AlphaEdit | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| DOME | 0.775 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| DPR | 0.767 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| Ultron | 0.731 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| BM25 | 0.691 | Not recorded | 69a7bc7b2d603e6de3e3d416 |
| DSI | 0.541 | Not recorded | 69a7bc7b2d603e6de3e3d416 |

### From Scratch

R@10: 0.775. Result ID: 69a891996dbb7075b951387c.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### New-Doc FT

R@10: 0.775. Result ID: 69a891996dbb7075b951387d.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71661e209131cea512ea.md>)

Recorded evaluation configuration: {"TPD (s)":"3.42"}. This record does not assert independent reproduction.

### DSI++

R@10: 0.775. Result ID: 69a891996dbb7075b951387e.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71661e209131cea512ea.md>)

Recorded evaluation configuration: {"TPD (s)":"3.54"}. This record does not assert independent reproduction.

### ROME

R@10: 0.775. Result ID: 69a891996dbb7075b951387f.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e591e209131cea3f3c9.md>)

[Implementation](<https://rome.baulab.info/>)

Recorded evaluation configuration: {"TPD (s)":"20.23"}. This record does not assert independent reproduction.

### MEMIT

R@10: 0.775. Result ID: 69a891996dbb7075b9513880.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa703d1e209131cea49dc7.md>)

Recorded evaluation configuration: {"TPD (s)":"12.62"}. This record does not assert independent reproduction.

### AlphaEdit

R@10: 0.775. Result ID: 69a891996dbb7075b9513881.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78351e209131cea77370.md>)

Recorded evaluation configuration: {"TPD (s)":"12.12"}. This record does not assert independent reproduction.

### DOME

R@10: 0.775. Result ID: 69a891996dbb7075b9513882.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

Recorded evaluation configuration: {"TPD (s)":"2.14\*"}. This record does not assert independent reproduction.

### DPR

R@10: 0.767. Result ID: 69a891996dbb7075b9513879.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa686d1e209131cea1c190.md>)

[Implementation](<https://github.com/facebookresearch/DPR>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Ultron

R@10: 0.731. Result ID: 69a891996dbb7075b951387b.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d391e209131cea388ae.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### BM25

R@10: 0.691. Result ID: 69a891996dbb7075b9513878.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### DSI

R@10: 0.541. Result ID: 69a891996dbb7075b951387a.

[Reporting paper](<https://www.sota2.com/research/paper/69a7bc7b2d603e6de3e3d416.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dbe1e209131cea3bc9d.md>)

[Implementation](<https://github.com/google-research/t5x>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.