Coherence Classification on GCDC 1.0 (test)

66.9Clinton F1

MTL with Transformers

Updated 1mo ago

Evaluation Results

Method
MTL with Transformers 2021.09	66.9	55.1	58.9	56.6	59.4
Hierarchical Transformer 2021.09	66.5	56.3	60.9	58.5	60.6
MTL with Transformer 2021.09	66.5	56.4	62.2	58.9	61
Hierarchical Transformer 2021.09	66	55.4	60.7	59.2	60.3
Vanilla Transformer 2021.09	65.6	56.2	59.3	57.4	59.6
Fact-aware Transformer 2021.09	64.3	53.8	58.4	58.1	58.7
Vanilla Transformer 2021.09	63.9	55.3	58.1	57.6	58.7
Coh+GR 2021.09	62	56	56	-	58
LC 2021.09	61	54.4	53.5	-	56.3
PARSEQ 2021.09	61	53.9	58.5	56.5	57.5
PARSEQ 2021.09	60.2	53.2	54.9	54.4	55.7
Hierachical LSTM 2021.09	59	50.5	55	-	54.8
Coh+SOX 2021.09	58.5	51	50.5	-	53.3
Lexicalized CNN-Egrid+coref 2021.09	56.6	44.7	51	54	51.6
Flesch-Kincaid grade level 2021.09	56	52.5	43.5	55	51.8
EGRAPH+coref 2021.09	55	44	42.5	54	48.9
Ensemble (ParSeq + RST-Recursive) 2020.09	52.42	44.69	46.88	43.94	46.98
Ensemble (ParSeq + RST-Recursive) 2020.09	52.35	44.92	45.48	43.7	46.61
ParSeq 2020.09	52.12	44.9	46.22	43.36	46.65
Ensemble (ParSeq + RST-Recursive) 2020.09	51.9	44.76	45.48	43.83	46.49
EGRID+coref 2021.09	48	47	41.5	49	46.4
RST-Recursive 2020.09	45.9	44.33	43.85	43.13	44.3
RST-Recursive 2020.09	41.08	31.21	41.97	42.27	39.13
Majority 2020.09	39.42	27.29	20.95	38.82	31.62
RST-Recursive 2020.09	39.42	27.29	20.95	38.82	31.62
RST-Recursive 2020.09	39.2	30.81	35.67	39.93	36.4