Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Calibration research benchmarks – Page 3 · SOTA2 Research

Calibration

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Covertype label 1 (test)

2.352ECE

10

Feb 26, 2026

SciKnowEval Edu. MMLU-Pro (test)

21.1Average Calibration Score

9

Jul 3, 2026

MMLU-Pro (test)

16.2ECE

9

Jul 3, 2026

SciKnowEval (test)

0.226BIO Score

9

Jul 3, 2026

CIFAR-100-C Severity 5 1 (test)

9.89ECE

8

Jun 23, 2026

CIFAR-100-C Severity 4 1 (test)

7.2ECE

8

Jun 23, 2026

CIFAR-100-C Severity 3 1 (test)

5.42ECE

8

Jun 23, 2026

CIFAR-100-C Severity 2 1 (test)

4.24ECE

8

Jun 23, 2026

CIFAR-100-C Severity 1 (test)

3.35ECE

8

Jun 23, 2026

Tiny ImageNet 5000-sample half seed-42 (test)

1.6ECE

8

May 12, 2026

1.085ECE Gain

8

Apr 21, 2026

0.344ECE' Gain

8

Apr 21, 2026

0.37Top-1 ECE

8

Mar 27, 2026

0.13Top-1 ECE

8

Mar 27, 2026

4.19ECE

8

Feb 26, 2026

1.294CB-ECE

8

Feb 26, 2026

NLE (conditional)

0.009CB-ECE

8

Feb 26, 2026

72.4AUC

8

Feb 26, 2026

11RMS Error

8

Feb 26, 2026

5.8RMS Error

8

Feb 26, 2026

6.3RMS

8

Feb 26, 2026

3.6RMS Error

8

Feb 26, 2026

QA datasets Aggregated models

Supervised Platt scaling (Sup.)

0.073ECE2

7

Apr 23, 2026

Math datasets Aggregated models

Supervised Platt scaling (Sup.)

0.058ECE2

7

Apr 23, 2026

DeepSeek models Aggregated datasets

Test-time self-consistency (TT SC)

0.067ECE2

7

Apr 23, 2026

Page 3 of 5

Popular datasets

Sequence ImageNet CIFAR-100 CIFAR MMLU

Follow for updates

Join SOTA2 Contact