# Closed-book question answering on Natural Questions

Source: [Closed-book question answering on Natural Questions](<https://www.sota2.com/research/sota/closed-book-question-answering-on-natural-questions>)

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

Data updated: 2026-02-26T05:41:34.908Z

Evaluation of Closed-book Question Answering performance on the Natural Questions dataset, measured using Exact Match (EM).

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 | Closed-book question answering |
| Dataset (including any stated split) | Natural Questions |
| Metric | 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

- [EM](<https://www.sota2.com/research/sota/closed-book-question-answering-on-natural-questions.md?metric=EM>): Higher is better; unit: PERCENT.

## Recorded results

| Method | EM | Result date | Report paper |
| --- | --- | --- | --- |
| PaLM | 21.2 | Not recorded | 68fa70f11e209131cea4e374 |
| Chinchilla | 16.6 | Not recorded | 68fa70f11e209131cea4e374 |
| GPT-3 | 14.6 | Not recorded | 68fa70f11e209131cea4e374 |
| BLOOM | 13.1 | Not recorded | 68fa70f11e209131cea4e374 |
| GLM-130B | 11.7 | Not recorded | 68fa70f11e209131cea4e374 |
| Gopher | 10.1 | Not recorded | 68fa70f11e209131cea4e374 |

### PaLM

EM: 21.2. Result ID: 6991668c3a6c31484907e4a6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d931e209131cea3abf2.md>)

Recorded evaluation configuration: {"Parameters":"540B"}. This record does not assert independent reproduction.

### Chinchilla

EM: 16.6. Result ID: 6991668c3a6c31484907e4a7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e5c1e209131cea3f593.md>)

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

### GPT-3

EM: 14.6. Result ID: 6991668c3a6c31484907e4a4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6cda1e209131cea36474.md>)

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

Recorded evaluation configuration: {"Model Version":"Davinci"}. This record does not assert independent reproduction.

### BLOOM

EM: 13.1. Result ID: 6991668c3a6c31484907e4a5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fdd1e209131cea47c06.md>)

Recorded evaluation configuration: {"Parameters":"176B"}. This record does not assert independent reproduction.

### GLM-130B

EM: 11.7. Result ID: 6991668c3a6c31484907e4a9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Implementation](<https://github.com/THUDM/GLM-130B/>)

Recorded evaluation configuration: {"Parameters":"130B"}. This record does not assert independent reproduction.

### Gopher

EM: 10.1. Result ID: 6991668c3a6c31484907e4a8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa70f11e209131cea4e374.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ba51e209131cea2f8bf.md>)

Recorded evaluation configuration: {"Parameters":"280B"}. This record does not assert independent reproduction.

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