# Commonsense Question Answering on CSQA Synonym Replacement WordNet-based (test)

Source: [Commonsense Question Answering on CSQA Synonym Replacement WordNet-based (test)](<https://www.sota2.com/research/sota/commonsense-question-answering-on-csqa-synonym-replacement-wordnet-based-test>)

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

Data updated: 2026-02-26T05:28:28.541Z

Evaluation of model performance on the specific 'Synonym Replacement WordNet-based' test split of the CSQA dataset for Commonsense Question Answering, measured by Accuracy.

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 | Commonsense Question Answering |
| Dataset (including any stated split) | CSQA Synonym Replacement WordNet-based (test) |
| Metric | Accuracy |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/commonsense-question-answering-on-csqa-synonym-replacement-wordnet-based-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| G-DAUG-Rand | 72.1 | Not recorded | 68fa6d351e209131cea386ac |
| G-DAUG-Combo | 72 | Not recorded | 68fa6d351e209131cea386ac |
| G-DAUG-Diversity | 71.6 | Not recorded | 68fa6d351e209131cea386ac |
| G-DAUG-Influence | 71 | Not recorded | 68fa6d351e209131cea386ac |
| ROBERTA (ours) | 69.9 | Not recorded | 68fa6d351e209131cea386ac |
| BACKTRANSLATION | 69 | Not recorded | 68fa6d351e209131cea386ac |

### G-DAUG-Rand

Accuracy: 72.1. Result ID: 698f986b3a6c314849fa639e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Implementation](<https://github.com/yangyiben/G-DAUG-c-Generative-Data-Augmentation-for-Commonsense-Reasoning>)

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement","selection\_strategy":"Random"}. This record does not assert independent reproduction.

### G-DAUG-Combo

Accuracy: 72. Result ID: 698f986b3a6c314849fa63a1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Implementation](<https://github.com/yangyiben/G-DAUG-c-Generative-Data-Augmentation-for-Commonsense-Reasoning>)

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement","selection\_strategy":"Combo (Influence + Diversity)"}. This record does not assert independent reproduction.

### G-DAUG-Diversity

Accuracy: 71.6. Result ID: 698f986b3a6c314849fa63a0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Implementation](<https://github.com/yangyiben/G-DAUG-c-Generative-Data-Augmentation-for-Commonsense-Reasoning>)

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement","selection\_strategy":"Diversity-based"}. This record does not assert independent reproduction.

### G-DAUG-Influence

Accuracy: 71. Result ID: 698f986b3a6c314849fa639f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Implementation](<https://github.com/yangyiben/G-DAUG-c-Generative-Data-Augmentation-for-Commonsense-Reasoning>)

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement","selection\_strategy":"Influence-based"}. This record does not assert independent reproduction.

### ROBERTA (ours)

Accuracy: 69.9. Result ID: 698f986b3a6c314849fa639c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa66131e209131cea0e18b.md>)

[Implementation](<https://github.com/pytorch/fairseq>)

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement","setup":"Local re-evaluation"}. This record does not assert independent reproduction.

### BACKTRANSLATION

Accuracy: 69. Result ID: 698f986b3a6c314849fa639d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6d351e209131cea386ac.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68a61e209131cea1d69a.md>)

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

Recorded evaluation configuration: {"perturbation":"WordNet-based synonym replacement"}. This record does not assert independent reproduction.

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