Motion-to-text retrieval on HumanML3D (test) benchmark leaderboard · SOTA2 Research

Motion-to-text retrieval on HumanML3D (test)

76.73R@1PST++

Evaluation Results

Method	Links
PST++2026.01		76.73	89.11	94.25	97.18	98.51	1
SGAR2026.01		76.35	—	92.38	95.69	98.02	1
PST2026.01		75.12	88.79	93.58	96.64	98.46	1
PST2026.01		75.12	88.79	93.58	96.64	98.46	1
Ours2024.05		72.11	85.26	90.21	94.44	97.76	1
MotionPatch2026.01		72.11	85.26	90.21	94.44	97.76	1
TMR2024.05		68.59	81.73	86.75	91.1	95.39	1.02
TMR2026.01		68.59	81.73	86.75	91.1	95.39	1.02
Ours (scratch)2024.05		67.11	80.04	85.86	91.86	95.98	1
T2M+2024.05		52	71.21	81.11	89.87	96.78	1.38
T2M2026.01		52	71.21	81.11	89.87	96.78	1.38
TEMOS2024.05		39.96	53.49	61.79	72.4	85.89	2.33
TEMOS2026.01		39.96	53.49	61.79	72.4	85.89	2.33
PST++2026.01		14.81	18.63	25.72	33.26	44.61	14
MoCHA2026.03		14.37	—	—	30.43	40.67	17
SGAR2026.01		13.82	—	23.38	30.09	41.83	16
PST2026.01		13.59	17.11	24.42	31.88	42.68	15
PST2026.01		13.59	17.11	24.42	31.88	42.68	15
MoCHA2026.03		12.77	—	—	28.26	38.78	20
Lyu et al.2026.01		12.39	15.55	22.17	29.25	40.34	17
Ours2024.05		11.25	13.86	19.98	26.86	37.4	20.5
MotionPatch2026.01		11.25	13.86	19.98	26.86	37.4	20.5
MoPa2026.03		11.25	—	—	26.86	37.4	20.5
Ours (scratch)2024.05		9.63	11.78	16.58	22.87	33.57	25
TMR2024.05		9.44	11.84	16.9	22.92	32.21	26
TMR2026.01		9.44	11.84	16.9	22.92	32.21	26
TMR2026.03		9.44	—	—	22.92	32.21	26
TEMOS2024.05		3.86	4.54	6.94	9.38	14	183.25
TEMOS2026.01		3.86	4.54	6.94	9.38	14	183.25
TEMOS2026.03		3.86	—	—	9.38	14	183.25
T2M+2024.05		2.92	3.74	6	8.36	12.95	81.5
T2M2026.01		2.92	3.74	6	8.36	12.95	81.5
T2M2026.03		2.92	—	—	8.36	12.95	81.5