# Natural Language Understanding on GLUE (CoLA, MRPC, QNLI, RTE, STS-B, SST-2)

Source: [Natural Language Understanding on GLUE (CoLA, MRPC, QNLI, RTE, STS-B, SST-2)](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-cola-mrpc-qnli-rte-sts-b-sst-2>)

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

Data updated: 2026-04-24T08:52:06.904Z

Evaluation of model performance across the specific subset of GLUE tasks: CoLA, MRPC, QNLI, RTE, STS-B, and SST-2.

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 |
| Metric | SST-2 Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

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

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

- [CoLA Score](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-cola-mrpc-qnli-rte-sts-b-sst-2.md?metric=CoLA%20Score>): Higher is better; unit: PERCENT.

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

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

- [STS-B Score](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-cola-mrpc-qnli-rte-sts-b-sst-2.md?metric=STS-B%20Score>): Higher is better; unit: PERCENT.

- [Average GLUE Score](<https://www.sota2.com/research/sota/natural-language-understanding-on-glue-cola-mrpc-qnli-rte-sts-b-sst-2.md?metric=Average%20GLUE%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | SST-2 Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| RandLoRA | 96.5 | Not recorded | 69eaf91855a3c32af89f899b |
| Full Fine-Tuning | 96.4 | Not recorded | 69eaf91855a3c32af89f899b |
| LoRA | 96.2 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 96.2 | Not recorded | 69eaf91855a3c32af89f899b |
| VeRA | 96.1 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 96.1 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 96.1 | Not recorded | 69eaf91855a3c32af89f899b |
| RandLoRA | 94.9 | Not recorded | 69eaf91855a3c32af89f899b |
| Full Fine-Tuning | 94.8 | Not recorded | 69eaf91855a3c32af89f899b |
| LoRA | 94.7 | Not recorded | 69eaf91855a3c32af89f899b |
| VeRA | 94.5 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 94.5 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 94.2 | Not recorded | 69eaf91855a3c32af89f899b |
| GiVA | 94.2 | Not recorded | 69eaf91855a3c32af89f899b |

### RandLoRA

SST-2 Accuracy: 96.5. Result ID: 69eb2f2247f34ce924d39463.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77641e209131cea73027.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"1.78","Rank (r)":"100"}. This record does not assert independent reproduction.

### Full Fine-Tuning

SST-2 Accuracy: 96.4. Result ID: 69eb2f2247f34ce924d39460.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"354.31"}. This record does not assert independent reproduction.

### LoRA

SST-2 Accuracy: 96.2. Result ID: 69eb2f2247f34ce924d39461.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b121e209131cea2bf59.md>)

[Implementation](<https://github.com/microsoft/LoRA>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"1.77","Rank (r)":"4"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 96.2. Result ID: 69eb2f2247f34ce924d39464.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"0.22","Rank (r)":"8","Initialization strategy":"TopSV-B"}. This record does not assert independent reproduction.

### VeRA

SST-2 Accuracy: 96.1. Result ID: 69eb2f2247f34ce924d39462.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71cd1e209131cea53093.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"0.26","Rank (r)":"256"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 96.1. Result ID: 69eb2f2247f34ce924d39465.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"0.22","Rank (r)":"8","Initialization strategy":"SecSV-B"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 96.1. Result ID: 69eb2f2247f34ce924d39466.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Large","Params (M)":"0.22","Rank (r)":"8","Initialization strategy":"Rand-B"}. This record does not assert independent reproduction.

### RandLoRA

SST-2 Accuracy: 94.9. Result ID: 69eb2f2247f34ce924d3945c.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77641e209131cea73027.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.72","Rank (r)":"64"}. This record does not assert independent reproduction.

### Full Fine-Tuning

SST-2 Accuracy: 94.8. Result ID: 69eb2f2247f34ce924d39459.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"124.06"}. This record does not assert independent reproduction.

### LoRA

SST-2 Accuracy: 94.7. Result ID: 69eb2f2247f34ce924d3945a.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b121e209131cea2bf59.md>)

[Implementation](<https://github.com/microsoft/LoRA>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.66","Rank (r)":"4"}. This record does not assert independent reproduction.

### VeRA

SST-2 Accuracy: 94.5. Result ID: 69eb2f2247f34ce924d3945b.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71cd1e209131cea53093.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.16","Rank (r)":"1024"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 94.5. Result ID: 69eb2f2247f34ce924d3945f.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.08","Rank (r)":"8","Initialization strategy":"Rand-B"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 94.2. Result ID: 69eb2f2247f34ce924d3945d.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.08","Rank (r)":"8","Initialization strategy":"TopSV-B"}. This record does not assert independent reproduction.

### GiVA

SST-2 Accuracy: 94.2. Result ID: 69eb2f2247f34ce924d3945e.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf91855a3c32af89f899b.md>)

Recorded evaluation configuration: {"Backbone":"RoBERTa-Base","Params (M)":"0.08","Rank (r)":"8","Initialization strategy":"SecSV-B"}. This record does not assert independent reproduction.

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