Linguistic Acceptability On Cola

评估指标

Accuracy

MCC

评测结果

各个模型在此基准测试上的表现结果

			Paper Title	Repository
En-BERT + TDA + PCA	88.6%	-	Acceptability Judgements via Examining the Topology of Attention Maps
BERT+TDA	88.2%	0.726	Can BERT eat RuCoLA? Topological Data Analysis to Explain
RoBERTa+TDA	87.3%	0.695	Can BERT eat RuCoLA? Topological Data Analysis to Explain
deberta-v3-base+tasksource	87.15%	-	tasksource: A Dataset Harmonization Framework for Streamlined NLP Multi-Task Learning and Evaluation
RoBERTa-large 355M + Entailment as Few-shot Learner	86.4%	-	Entailment as Few-Shot Learner
LTG-BERT-base 98M	82.7	-	Not all layers are equally as important: Every Layer Counts BERT	-
ELC-BERT-base 98M	82.6	-	Not all layers are equally as important: Every Layer Counts BERT	-
En-BERT + TDA	82.1%	0.565	Acceptability Judgements via Examining the Topology of Attention Maps
FNet-Large	78%	-	FNet: Mixing Tokens with Fourier Transforms
LTG-BERT-small 24M	77.6	-	Not all layers are equally as important: Every Layer Counts BERT	-
ELC-BERT-small 24M	76.1	-	Not all layers are equally as important: Every Layer Counts BERT	-
T5-11B	70.8%	-	Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
StructBERTRoBERTa ensemble	69.2%	-	StructBERT: Incorporating Language Structures into Pre-training for Deep Language Understanding	-
ALBERT	69.1%	-	ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
FLOATER-large	69%	-	Learning to Encode Position for Transformer with Continuous Dynamical Model
XLNet (single model)	69%	-	XLNet: Generalized Autoregressive Pretraining for Language Understanding
RoBERTa-large 355M (MLP quantized vector-wise, fine-tuned)	68.6%	-	LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
MT-DNN	68.4%	-	Multi-Task Deep Neural Networks for Natural Language Understanding
ELECTRA	68.2%	-	-	-
RoBERTa (ensemble)	67.8%	-	RoBERTa: A Robustly Optimized BERT Pretraining Approach

0 of 43 row(s) selected.