Natural Language Understanding on GLUE base (test dev) benchmark leaderboard · SOTA2 Research

Natural Language Understanding on GLUE base (test dev)

71.73CoLA MCCRefLoRA

Evaluation Results

Method	Links
RefLoRA2025.05		71.73	95.99	91.42	92.03	92.28	89.7	90.23	90.41	94.4	88.09	89.52
AdaLoRA2025.05		71.45	96.1	90.69	91.84	92.23	89.74	90.76	90.79	94.55	88.09	89.46
LoRA-Pro2025.05		71.36	95.76	90.2	91.92	92.19	89.6	90.23	90.19	94.29	85.56	88.94
DoRA2025.05		70.85	95.79	90.93	91.79	92.07	—	90.29	—	94.1	86.04	88.98
RefLoRA-S2025.05		70.66	95.76	90.44	92.21	92.43	89.89	90.13	90.17	94.16	87.73	89.19
LoRA2025.05		69.82	94.95	89.95	91.6	91.99	89.38	90.65	90.69	93.87	85.2	88.5
LoRA-RITE2025.05		69.55	95.41	90.93	91.79	92.02	89.42	90.22	90.33	94.42	85.2	88.69
Full FT2025.05		69.19	95.63	89.46	91.6	92.4	89.8	89.9	90.12	94.03	83.75	88.25
PAdapter2025.05		68.77	95.61	89.46	91.54	92.04	89.4	90.33	90.39	94.29	85.2	88.41
HAdapter2025.05		68.64	95.53	89.95	91.48	91.91	89.27	90.13	90.17	94.11	84.48	88.28
BitFit2025.05		66.96	94.84	87.75	91.35	88.41	84.95	89.37	89.91	92.24	78.7	86.2