# Probability Calibration on ImageNet 16H Noise Level 125

Source: [Probability Calibration on ImageNet 16H Noise Level 125](<https://www.sota2.com/research/sota/probability-calibration-on-imagenet-16h-noise-level-125>)

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

Data updated: 2026-02-26T05:41:20.496Z

Evaluation of model probability calibration performance using ECE, cwECE, and NLL on the ImageNet 16H dataset configured with Noise Level 125.

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 | Probability Calibration |
| Dataset (including any stated split) | ImageNet 16H Noise Level 125 |
| Metric | ECE |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 7 of 7 |
| 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

- [ECE](<https://www.sota2.com/research/sota/probability-calibration-on-imagenet-16h-noise-level-125.md?metric=ECE>): Lower is better; unit: PERCENT.

- [cwECE](<https://www.sota2.com/research/sota/probability-calibration-on-imagenet-16h-noise-level-125.md?metric=cwECE>): Lower is better; unit: PERCENT.

- [NLL](<https://www.sota2.com/research/sota/probability-calibration-on-imagenet-16h-noise-level-125.md?metric=NLL>): Lower is better; unit: SCALAR.

## Recorded results

| Method | ECE | Result date | Report paper |
| --- | --- | --- | --- |
| P+L | 719 | Not recorded | 68fa6af21e209131cea2b4a5 |
| VGG-19 | 729 | Not recorded | 68fa6af21e209131cea2b4a5 |
| VGG-19 | 734 | Not recorded | 68fa6af21e209131cea2b4a5 |
| P+L | 846 | Not recorded | 68fa6af21e209131cea2b4a5 |
| P+L | 1021 | Not recorded | 68fa6af21e209131cea2b4a5 |
| P+L | 1089 | Not recorded | 68fa6af21e209131cea2b4a5 |
| VGG-19 | 1181 | Not recorded | 68fa6af21e209131cea2b4a5 |

### P+L

ECE: 719. Result ID: 698e45a5ffc3901de09c3652.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Implementation](<https://github.com/GavinKerrigan/conf_matrix_and_calibration>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"5000","Evaluator type":"Recalibrated Combination"}. This record does not assert independent reproduction.

### VGG-19

ECE: 729. Result ID: 698e45a5ffc3901de09c3650.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61d01e209131ce9f62f5.md>)

[Implementation](<http://www.robots.ox.ac.uk/~vgg/research/very_deep/>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"5000","Evaluator type":"Model"}. This record does not assert independent reproduction.

### VGG-19

ECE: 734. Result ID: 698e45a5ffc3901de09c364e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61d01e209131ce9f62f5.md>)

[Implementation](<http://www.robots.ox.ac.uk/~vgg/research/very_deep/>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"10","Evaluator type":"Model"}. This record does not assert independent reproduction.

### P+L

ECE: 846. Result ID: 698e45a5ffc3901de09c3651.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Implementation](<https://github.com/GavinKerrigan/conf_matrix_and_calibration>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"5000","Evaluator type":"Combination"}. This record does not assert independent reproduction.

### P+L

ECE: 1021. Result ID: 698e45a5ffc3901de09c364f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Implementation](<https://github.com/GavinKerrigan/conf_matrix_and_calibration>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"10","Evaluator type":"Combination"}. This record does not assert independent reproduction.

### P+L

ECE: 1089. Result ID: 698e45a5ffc3901de09c364d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Implementation](<https://github.com/GavinKerrigan/conf_matrix_and_calibration>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"0","Evaluator type":"Combination"}. This record does not assert independent reproduction.

### VGG-19

ECE: 1181. Result ID: 698e45a5ffc3901de09c364c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6af21e209131cea2b4a5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61d01e209131ce9f62f5.md>)

[Implementation](<http://www.robots.ox.ac.uk/~vgg/research/very_deep/>)

Recorded evaluation configuration: {"Backbone":"VGG-19","Calibration fitting data size":"0","Evaluator type":"Model"}. This record does not assert independent reproduction.

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