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

Grammatical Error Correction research benchmarks – Page 2 · SOTA2 Research

Grammatical Error Correction

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Llama-3.1-8B-Instruct

71.4GLEU

9

May 14, 2026

50.9Precision

9

May 14, 2026

50.4Precision

9

May 14, 2026

Llama-3.1-8B-Instruct

64.5Precision

9

May 14, 2026

78.4Precision

9

May 14, 2026

87.5Precision

9

Feb 26, 2026

65.25EM

9

Feb 26, 2026

59.95Precision

9

Feb 26, 2026

Xu et al. (2019)

55.37F0.5

9

Feb 26, 2026

48.72F0.5

8

Feb 26, 2026

Grundkiewicz et al. (2019)

69.47F0.5

8

Feb 26, 2026

UA-GEC+Fluency (dev)

1×RPN(XLMBASE) + 2×RPN(RLARGE)

80.28Precision

7

Feb 26, 2026

75.96F0.5 Score

7

Feb 26, 2026

Base + FB learning

75.93F0.5 Score

7

Feb 26, 2026

MLConvembed (4 ens.) + EO + LM + SpellCheck

63.61F0.5 Score

7

Feb 26, 2026

Synthetic Icelandic GEC (test)

95GLEU

6

Feb 26, 2026

Learner Corpus (all)

2,763TP

6

Feb 26, 2026

Learner Corpus (C)

415True Positives (TP)

6

Feb 26, 2026

Learner Corpus (B)

1,049TP

6

Feb 26, 2026

Learner Corpus (A)

1,299True Positives (TP)

6

Feb 26, 2026

AKCES-GEC Czech (test)

84.21Precision

6

Feb 26, 2026

Tartu-L1 Estonian (test)

63.2F0.5 Score

4

Jun 16, 2026

KoLLA multi-reference annotation (test)

kanana2-30b-a3b

59.01F0.5

4

Jun 1, 2026

NIKL single-reference annotation (test)

kanana2-30b-a3b

46.43F0.5 Score

4

Jun 1, 2026

Kor-Native (test)

KoGPT3-1.2B + SCRIPT

85.49M2 Precision

4

Apr 15, 2026

Page 2 of 3

Popular datasets

BEA Conll AKCES-GEC Learner Corpus JFLEG

Follow for updates

Join SOTA2 Contact