Grammatical Error Correction On Conll 2014

Métriques

F0.5

Precision

Recall

Résultats

Résultats de performance de divers modèles sur ce benchmark

Nom du modèle	F0.5	Precision	Recall	Paper Title	Repository
Sequence tagging + token-level transformations + two-stage fine-tuning (+BERT, RoBERTa, XLNet)	66.5	78.2	41.5	GECToR -- Grammatical Error Correction: Tag, Not Rewrite
CNN Seq2Seq + Quality Estimation	56.52	-	-	Neural Quality Estimation of Grammatical Error Correction
Transformer	55.8	-	-	Approaching Neural Grammatical Error Correction as a Low-Resource Machine Translation Task
LM-Critic	65.8	-	-	LM-Critic: Language Models for Unsupervised Grammatical Error Correction
GEC-DI (LM+GED)	69.6	79.2	46.8	Improving Seq2Seq Grammatical Error Correction via Decoding Interventions
Sequence Labeling with edits using BERT, Faster inference	61.2	-	-	Parallel Iterative Edit Models for Local Sequence Transduction
Unsupervised GEC + cLang8	69.6	75.0	53.8	Unsupervised Grammatical Error Correction Rivaling Supervised Methods
ESC	69.51	81.48	43.78	Frustratingly Easy System Combination for Grammatical Error Correction
Transformer + Pre-train with Pseudo Data	65.0	-	-	An Empirical Study of Incorporating Pseudo Data into Grammatical Error Correction
SMT + BiGRU	56.25	-	-	Near Human-Level Performance in Grammatical Error Correction with Hybrid Machine Translation	-
BART	63.0	69.9	45.1	Stronger Baselines for Grammatical Error Correction Using Pretrained Encoder-Decoder Model
Majority-voting ensemble on best 7 models	71.8	83.7	45.7	Pillars of Grammatical Error Correction: Comprehensive Inspection Of Contemporary Approaches In The Era of Large Language Models
SynGEC	67.6	74.7	49.0	SynGEC: Syntax-Enhanced Grammatical Error Correction with a Tailored GEC-Oriented Parser	-
Sequence tagging + token-level transformations + two-stage fine-tuning (+XLNet)	65.3	77.5	40.1	GECToR -- Grammatical Error Correction: Tag, Not Rewrite
GRECO (voting+ESC)	71.12	79.6	49.86	System Combination via Quality Estimation for Grammatical Error Correction
Copy-augmented Model (4 Ensemble +Denoising Autoencoder)	61.15	71.57	38.65	Improving Grammatical Error Correction via Pre-Training a Copy-Augmented Architecture with Unlabeled Data
T5	68.87	-	-	A Simple Recipe for Multilingual Grammatical Error Correction
CNN Seq2Seq	54.79	-	-	A Multilayer Convolutional Encoder-Decoder Neural Network for Grammatical Error Correction
Ensembles of best 7 models + GRECO + GTP-rerank	72.8	83.9	47.5	Pillars of Grammatical Error Correction: Comprehensive Inspection Of Contemporary Approaches In The Era of Large Language Models
Sequence Labeling with edits using BERT, Faster inference (Single Model)	59.7	-	-	Parallel Iterative Edit Models for Local Sequence Transduction

0 of 23 row(s) selected.