Natural Language Understanding on GLUE (Subset Accuracy and Difference Metrics)

61.58MNLI AccuracyCOIN

Evaluation results

Method
COIN2026.02	61.58	—	56.92	—	53.79	—	50.67	—	86.38	—	61.87	—
CoIPO2026.02	61.5	—	64.5	—	56.21	—	66.04	—	86.75	—	67	—
COIN2026.02	59.58	2	55.08	1.83	52	1.79	50.21	0.46	85.04	1.33	60.38	1.48
CoIPO2026.02	59.38	2.12	63.67	0.83	53.58	2.62	64.5	1.54	85.83	0.92	65.39	1.61
CoIPO2026.02	58.38	3.89	64.07	0.54	53.81	3	57.26	10.98	85.98	0.97	63.9	3.88
CoIPO2026.02	58	3.5	64.38	0.13	51.12	5.08	52.17	13.88	85.42	1.33	62.22	4.78
CoIPO2026.02	57.33	4.17	61.5	3	51.04	5.17	48.54	17.5	86.54	0.21	60.99	6.01
SFT2026.02	56.13	—	41.04	—	46.54	—	58.62	—	84.04	—	57.28	—
COIN2026.02	56.08	6.89	54.68	2.8	51.05	3.43	45.64	6.28	85.56	1.02	58.6	4.08
CoIPO2026.02	55.71	5.79	66.29	-1.79	57.08	-0.88	55.04	11	85.33	1.42	63.89	3.11
SFT2026.02	54.83	1.29	39.79	1.25	43.08	3.46	51.33	7.29	80.29	3.75	53.87	3.41
COIN2026.02	54.46	7.12	58.33	-1.42	55.58	-1.79	45.75	4.92	84.96	1.42	59.82	2.05
SFT2026.02	54.13	2	40.25	0.79	40.33	6.21	57.17	1.46	81.46	2.58	54.67	2.61
SFT2026.02	53.62	3.14	40.04	1.25	44.89	2.06	53.64	6.23	81.4	3.3	54.72	3.2
COIN2026.02	52.83	8.75	51.71	5.21	48.75	5.04	42.12	8.54	85.13	1.25	56.11	5.76
SFT2026.02	52.83	3.29	39.58	1.46	44.37	2.17	51.04	7.58	82.71	1.33	54.11	3.17
COIN2026.02	51.92	9.67	51.34	5.58	45.13	8.67	39.46	11.21	86.29	0.08	54.83	7.04
Base2026.02	51	—	62.04	—	52.54	—	29.87	—	80.04	—	55.1	—
SFT2026.02	50.17	5.96	39.54	1.5	50.12	-3.58	50.04	8.58	78.5	5.54	53.68	3.6
Base2026.02	47.71	3.29	55.62	6.42	47.67	4.88	20.25	9.62	75.75	4.29	49.4	5.7
Base2026.02	45.46	5.54	51.33	10.71	53.33	-0.79	21	8.87	67.12	12.92	47.65	7.45
Base2026.02	45.35	7.06	51.68	12.95	48.51	5.04	20.36	11.89	67.28	15.95	46.64	10.58
Base2026.02	42.62	8.38	47.29	14.75	48	4.54	17.5	12.37	61.46	18.58	43.37	11.73
Base2026.02	39.96	11.04	42.12	19.92	41	11.54	13.17	16.71	52.04	28	37.66	17.44