Multi-task Neural Architecture Search on TransNAS-Bench-101 Macro level search space 1.0 benchmark leaderboard · SOTA2 Research

Multi-task Neural Architecture Search on TransNAS-Bench-101 Macro level search space 1.0

47.96Cls O AccGlobal Best

Evaluation Results

Method	Links
Global Best2022.04		47.96	57.48	76.88	64.35	29.66	56.28	97.02	1
Arch-Graph2022.04		47.44	56.98	75.9	64.35	29.19	57.75	—	5.24
Arch-Graph-zero2022.04		47.42	56.78	75.51	63.39	29.17	58.15	—	7.83
weakNAS2022.04		47.4	56.88	72.54	62.37	29.18	57.86	96.86	10.49
Arch-Graph-single2022.04		47.35	56.77	71.32	62.78	29.09	58.05	96.7	12.68
CATCH2022.04		47.29	56.49	70.36	60.85	28.71	59.37	—	37.72
weakNAS-t2022.04		47.13	56.83	73.59	61.86	29.07	58.55	—	15.43
REA2022.04		47.09	56.57	69.98	60.88	28.87	58.73	96.88	41.03
BONAS-t2022.04		47.06	56.86	71.41	61.44	28.76	58.35	—	27.87
REA-t2022.04		46.98	56.6	73.41	61.02	28.9	58.18	—	28.98
nsganetv22022.04		46.86	56.29	73.77	61.41	28.73	59.07	—	34.39
RS2022.04		46.85	56.5	70.06	60.7	28.37	59.35	96.78	59.26
BONAS2022.04		46.85	56.47	74.45	61.62	28.82	59.39	96.76	33.37
DT2022.04		45.48	54.96	59.35	58.6	26.21	62.07	95.37	534.31