# Image Classification on CIFAR-10 (test) (Accuracy, Calibration, and Uncertainty Metrics)

Source: [Image Classification on CIFAR-10 (test) (Accuracy, Calibration, and Uncertainty Metrics)](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics>)

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

Data updated: 2026-03-05T07:47:53.484Z

Evaluation of Image Classification performance on the CIFAR-10 test set, reporting standard accuracy alongside key calibration and uncertainty metrics including NLL, ECE, Brier Score, and AUROC.

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 | Image Classification |
| Dataset (including any stated split) | CIFAR-10 (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [NLL](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics.md?metric=NLL>): Lower is better; unit: SCALAR.

- [ECE](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics.md?metric=ECE>): Lower is better; unit: PERCENT.

- [Brier Score](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics.md?metric=Brier%20Score>): Lower is better; unit: SCALAR.

- [AUROC](<https://www.sota2.com/research/sota/image-classification-on-cifar-10-test-accuracy-calibration-and-uncertainty-metrics.md?metric=AUROC>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| SAM | 92.6 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.6 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.6 | Not recorded | 69a90def2d603e6de3e40702 |
| IVON (1-MC) | 92.5 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.5 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.4 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.4 | Not recorded | 69a90def2d603e6de3e40702 |
| VSGD | 92.4 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.2 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.1 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.1 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92.1 | Not recorded | 69a90def2d603e6de3e40702 |
| VSGD | 92.1 | Not recorded | 69a90def2d603e6de3e40702 |
| SGD | 92 | Not recorded | 69a90def2d603e6de3e40702 |
| q-VSGD | 92 | Not recorded | 69a90def2d603e6de3e40702 |

### SAM

Accuracy: 92.6. Result ID: 69a935196dbb7075b952591e.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69d81e209131cea248fb.md>)

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

Recorded evaluation configuration: {"Backbone":"ResNet-20","Time":"70s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.6. Result ID: 69a935196dbb7075b9525925.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","q":"0.0","Time":"173s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.6. Result ID: 69a935196dbb7075b9525928.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","q":"0.6","Time":"174s"}. This record does not assert independent reproduction.

### IVON (1-MC)

Accuracy: 92.5. Result ID: 69a935196dbb7075b952591d.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5be81e209131ce9cc869.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","Time":"43s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.5. Result ID: 69a935196dbb7075b9525926.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","q":"0.2","Time":"173s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.4. Result ID: 69a935196dbb7075b9525927.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","q":"0.4","Time":"174s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.4. Result ID: 69a935196dbb7075b9525929.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","q":"0.8","Time":"182s"}. This record does not assert independent reproduction.

### VSGD

Accuracy: 92.4. Result ID: 69a935196dbb7075b952592a.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"5","Time":"173s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.2. Result ID: 69a935196dbb7075b9525922.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","q":"0.6","Time":"43s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.1. Result ID: 69a935196dbb7075b952591f.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","q":"0.0","Time":"44s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.1. Result ID: 69a935196dbb7075b9525920.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","q":"0.2","Time":"43s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92.1. Result ID: 69a935196dbb7075b9525923.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","q":"0.8","Time":"43s"}. This record does not assert independent reproduction.

### VSGD

Accuracy: 92.1. Result ID: 69a935196dbb7075b9525924.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","Time":"42s"}. This record does not assert independent reproduction.

### SGD

Accuracy: 92. Result ID: 69a935196dbb7075b952591c.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Time":"37s"}. This record does not assert independent reproduction.

### q-VSGD

Accuracy: 92. Result ID: 69a935196dbb7075b9525921.

[Reporting paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a90def2d603e6de3e40702.md>)

Recorded evaluation configuration: {"Backbone":"ResNet-20","Number of Monte-Carlo Samples":"1","q":"0.4","Time":"43s"}. This record does not assert independent reproduction.

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