# Natural Language Understanding on GLUE (QNLI, MNLI, SST-2, QQP)

Source: [Natural Language Understanding on GLUE (QNLI, MNLI, SST-2, QQP)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp>)

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

Data updated: 2026-04-13T07:19:34.541Z

Evaluation of Natural Language Understanding capabilities across the specific subset of GLUE tasks: QNLI, MNLI, SST-2, and QQP.

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 | Natural Language Understanding |
| Dataset (including any stated split) | GLUE QNLI, MNLI, SST-2, QQP |
| Metric | QNLI Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [QNLI Accuracy](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QNLI%20Accuracy>): Higher is better; unit: PERCENT.

- [QNLI Accuracy Gap](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QNLI%20Accuracy%20Gap>): Higher is better; unit: SCALAR.

- [QNLI Training Time (m)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QNLI%20Training%20Time%20(m)>): Lower is better; unit: TIME.

- [MNLI Accuracy](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=MNLI%20Accuracy>): Higher is better; unit: PERCENT.

- [MNLI Accuracy Gap](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=MNLI%20Accuracy%20Gap>): Higher is better; unit: SCALAR.

- [MNLI Training Time (m)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=MNLI%20Training%20Time%20(m)>): Lower is better; unit: TIME.

- [SST-2 Accuracy](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=SST-2%20Accuracy>): Higher is better; unit: PERCENT.

- [SST-2 Accuracy Gap](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=SST-2%20Accuracy%20Gap>): Higher is better; unit: SCALAR.

- [SST-2 Training Time (m)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=SST-2%20Training%20Time%20(m)>): Lower is better; unit: TIME.

- [QQP Accuracy](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QQP%20Accuracy>): Higher is better; unit: PERCENT.

- [QQP Accuracy Gap](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QQP%20Accuracy%20Gap>): Higher is better; unit: SCALAR.

- [QQP Training Time (m)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-qnli-mnli-sst-2-qqp.md?metric=QQP%20Training%20Time%20(m)>): Lower is better; unit: TIME.

## Recorded results

| Method | QNLI Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| CoA-LoRA | 62 | Not recorded | 68fa7bb01e209131cea89e6e |
| Shared-LoRA | 60 | Not recorded | 68fa7bb01e209131cea89e6e |
| LQ-LoRA | 59 | Not recorded | 68fa7bb01e209131cea89e6e |
| QLoRA | 58 | Not recorded | 68fa7bb01e209131cea89e6e |

### CoA-LoRA

QNLI Accuracy: 62. Result ID: 69dc988d47f34ce924c32bd8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bb01e209131cea89e6e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bb01e209131cea89e6e.md>)

Recorded evaluation configuration: {"Solution":"∞","Backbone":"RoBERTa-Large","Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

### Shared-LoRA

QNLI Accuracy: 60. Result ID: 69dc988d47f34ce924c32bd7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bb01e209131cea89e6e.md>)

Recorded evaluation configuration: {"Solution":"1","Backbone":"RoBERTa-Large","Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

### LQ-LoRA

QNLI Accuracy: 59. Result ID: 69dc988d47f34ce924c32bd6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bb01e209131cea89e6e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74a51e209131cea62e31.md>)

Recorded evaluation configuration: {"Solution":"6","Backbone":"RoBERTa-Large","Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

### QLoRA

QNLI Accuracy: 58. Result ID: 69dc988d47f34ce924c32bd5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bb01e209131cea89e6e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6a1e209131cea456e1.md>)

[Implementation](<https://github.com/artidoro/qlora>)

Recorded evaluation configuration: {"Solution":"6","Backbone":"RoBERTa-Large","Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

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