# Distractor Generation on CLOTH Hard

Source: [Distractor Generation on CLOTH Hard](<https://www.sota2.com/research/sota/distractor-generation-on-cloth-hard>)

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

Data updated: 2026-05-20T08:23:58.797Z

Evaluation of model performance on generating distractors for multiple-choice questions using the CLOTH Hard dataset.

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 | Distractor Generation |
| Dataset (including any stated split) | CLOTH Hard |
| Metric | Invalid Ratio |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 5 of 5 |
| 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

- [Invalid Ratio](<https://www.sota2.com/research/sota/distractor-generation-on-cloth-hard.md?metric=Invalid%20Ratio>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Invalid Ratio | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen 2.5 7B | 4.2 | Not recorded | 690988873ef7ccd74e5e5cab |
| Gemma 2 9B | 5.1 | Not recorded | 690988873ef7ccd74e5e5cab |
| Llama 3.1 8B | 5.2 | Not recorded | 690988873ef7ccd74e5e5cab |
| GPT-4o | 6.8 | Not recorded | 690988873ef7ccd74e5e5cab |
| GPT-4o | 16.9 | Not recorded | 690988873ef7ccd74e5e5cab |

### Qwen 2.5 7B

Invalid Ratio: 4.2. Result ID: 6a0d6f97fbc373592c056bae.

[Reporting paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

Recorded evaluation configuration: {"training":"DCDG-trained"}. This record does not assert independent reproduction.

### Gemma 2 9B

Invalid Ratio: 5.1. Result ID: 6a0d6f97fbc373592c056bac.

[Reporting paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

Recorded evaluation configuration: {"training":"DCDG-trained"}. This record does not assert independent reproduction.

### Llama 3.1 8B

Invalid Ratio: 5.2. Result ID: 6a0d6f97fbc373592c056bad.

[Reporting paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

Recorded evaluation configuration: {"training":"DCDG-trained"}. This record does not assert independent reproduction.

### GPT-4o

Invalid Ratio: 6.8. Result ID: 6a0d6f97fbc373592c056bab.

[Reporting paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"prompting":"5-shot"}. This record does not assert independent reproduction.

### GPT-4o

Invalid Ratio: 16.9. Result ID: 6a0d6f97fbc373592c056baa.

[Reporting paper](<https://www.sota2.com/research/paper/690988873ef7ccd74e5e5cab.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"prompting":"0-shot"}. This record does not assert independent reproduction.

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