# Long-form Question Answering on ELI5 KILT (test)

Source: [Long-form Question Answering on ELI5 KILT (test)](<https://www.sota2.com/research/sota/long-form-question-answering-on-eli5-kilt-test>)

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

Data updated: 2026-02-27T06:45:35.903Z

Evaluation of model performance on the test split of the ELI5 KILT dataset for Long-form Question Answering, measured using F1, R-L, and KRL metrics.

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 | Long-form Question Answering |
| Dataset (including any stated split) | ELI5 KILT (test) |
| Metric | F1 |
| 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

- [F1](<https://www.sota2.com/research/sota/long-form-question-answering-on-eli5-kilt-test.md?metric=F1>): Higher is better; unit: PERCENT.

- [R-L](<https://www.sota2.com/research/sota/long-form-question-answering-on-eli5-kilt-test.md?metric=R-L>): Higher is better; unit: PERCENT.

- [KRL](<https://www.sota2.com/research/sota/long-form-question-answering-on-eli5-kilt-test.md?metric=KRL>): Higher is better; unit: SCALAR.

## Recorded results

| Method | F1 | Result date | Report paper |
| --- | --- | --- | --- |
| RT + C-REALM | 25.4 | Not recorded | 68fa69ef1e209131cea250f1 |
| RT + REALM | 25.1 | Not recorded | 68fa69ef1e209131cea250f1 |
| RT + REALM | 23.1 | Not recorded | 68fa69ef1e209131cea250f1 |
| RT + C-REALM | 22.9 | Not recorded | 68fa69ef1e209131cea250f1 |
| BART | 19.2 | Not recorded | 68fa69ef1e209131cea250f1 |
| BART + DPR | 17.9 | Not recorded | 68fa69ef1e209131cea250f1 |
| T5-base | 16.1 | Not recorded | 68fa69ef1e209131cea250f1 |
| RAG | 14.5 | Not recorded | 68fa69ef1e209131cea250f1 |

### RT + C-REALM

F1: 25.4. Result ID: 69a13d7435dfcabb00e5b72a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Implementation](<https://github.com/martiansideofthemoon/hurdles-longform-qa>)

Recorded evaluation configuration: {"p (nucleus sampling)":"0.9","Retriever":"C-REALM"}. This record does not assert independent reproduction.

### RT + REALM

F1: 25.1. Result ID: 69a13d7435dfcabb00e5b729.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

Recorded evaluation configuration: {"p (nucleus sampling)":"0.9","Retriever":"REALM"}. This record does not assert independent reproduction.

### RT + REALM

F1: 23.1. Result ID: 69a13d7435dfcabb00e5b72b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

Recorded evaluation configuration: {"p (nucleus sampling)":"0.6","Retriever":"REALM"}. This record does not assert independent reproduction.

### RT + C-REALM

F1: 22.9. Result ID: 69a13d7435dfcabb00e5b72c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Implementation](<https://github.com/martiansideofthemoon/hurdles-longform-qa>)

Recorded evaluation configuration: {"p (nucleus sampling)":"0.6","Retriever":"C-REALM"}. This record does not assert independent reproduction.

### BART

F1: 19.2. Result ID: 69a13d7435dfcabb00e5b726.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66921e209131cea10fff.md>)

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

### BART + DPR

F1: 17.9. Result ID: 69a13d7435dfcabb00e5b728.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a021e209131cea255d9.md>)

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

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

### T5-base

F1: 16.1. Result ID: 69a13d7435dfcabb00e5b725.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa709e1e209131cea4c470.md>)

[Implementation](<https://github.com/google-research/text-to-text-transfer-transformer>)

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

### RAG

F1: 14.5. Result ID: 69a13d7435dfcabb00e5b727.

[Reporting paper](<https://www.sota2.com/research/paper/68fa69ef1e209131cea250f1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69b91e209131cea23c26.md>)

[Implementation](<https://github.com/huggingface/transformers/blob/master/examples/rag/>)

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.