Natural Language Understanding on GLUE (7-Task Subset: CoLA, SST-2, MRPC, STS-B, QQP, MNLI, QNLI) benchmark leaderboard · SOTA2 Research

Natural Language Understanding on GLUE (7-Task Subset: CoLA, SST-2, MRPC, STS-B, QQP, MNLI, QNLI)

88.15Average ScoreModernBERT

Evaluation Results

Method	Links
ModernBERT2025.11		88.15	78.98	93.08	87.93	89.11	88.26	87.21	92.48	22.6
Roformer2025.11		87.72	84.53	91.83	88.47	87.96	87.61	83.58	90.03	40.4
DeepCoT Roformer2025.11		85.18	81.35	89.33	84.31	87.54	85.93	80.93	86.88	95.3
DeepCoT Roformer2025.11		84.8	80.91	89.23	85.95	86.48	85.79	79.04	86.18	84.4
Roformer2025.11		83.39	84.64	89.59	85.01	83.22	85.93	75.74	79.61	53.6
ModernBERT2025.11		83.13	77.21	89.39	84.69	82.28	87.17	80.23	80.94	32.7
FNet2025.11		82.6	80.54	89.11	83.14	84.31	84.03	73.87	83.2	45.8
DeepCoT Roformer2025.11		80.81	80.97	87.93	80.9	79.84	82.8	74.76	78.44	86.4
FNet2025.11		77.79	81.64	84.68	77.11	81.67	81.42	65.58	72.41	70.1
DeepCoT SOFT Roformer2025.11		77.68	81.04	82.98	82.19	74.12	80.33	72.21	70.88	66.3
DeepCoT SOFT Roformer2025.11		76.83	81.82	82.7	81.19	73.52	79.46	70.33	68.76	74.9
SOFT Roformer2025.11		76.63	80.4	83.08	81.89	73.75	80.56	71.02	65.69	28.5
DeepCoT SOFT Roformer2025.11		74.71	81.4	83.67	80	72.27	76.08	66.4	63.14	73.7
ModernBERT2025.11		72.66	75.79	81.48	79.8	77.27	77.2	57.47	59.58	49.4
SOFT Roformer2025.11		72.59	79.13	79.91	76.66	72.79	78.28	63.53	57.8	39.3
Roformer2025.11		69.52	79.58	79.42	80.51	76.57	76.07	35.33	59.18	80.1
SOFT Roformer2025.11		69.01	79.42	72.02	80.53	69.82	72.35	56.59	52.34	57
FNet2025.11		67.71	81.75	77.86	76.35	74.47	74.17	31.82	57.54	112.3