Text-to-Motion Generation on HumanML3D Joint Representation (test)
51.7R-Precision@1MotionRFT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MotionRFTBackbone=ACMDM [34]2026.03 | 51.7 | 73.4 | 81.9 | 0.084 | 3.172 | |
| ACMDMConfig=Base Model2026.03 | 50.8 | 70.1 | 79.8 | 0.109 | 3.253 | |
| RealType=Ground Truth2026.03 | 50.3 | 69.6 | 79.5 | 0 | 3.244 | |
| MotionLCMBackbone=MotionLCM2026.03 | 50.1 | 69.3 | 79 | 2.267 | 3.192 | |
| MLD++Backbone=MLD++2026.03 | 50 | 69.1 | 78.9 | 2.027 | 3.22 | |
| MARDM-vBackbone=MARDM2026.03 | 50 | 69.5 | 79.5 | 0.114 | 3.27 | |
| MARDM-ϵBackbone=MARDM2026.03 | 49.2 | 69 | 79 | 0.116 | 3.349 | |
| ReMoDiffuseBackbone=ReMoDiffuse2026.03 | 46.8 | 65.3 | 75.4 | 0.883 | 3.414 | |
| MotionDiffuseBackbone=MotionDiffuse2026.03 | 45 | 64.1 | 75.3 | 0.778 | 3.49 | |
| MDM50SBackbone=MDM2026.03 | 44 | 63.6 | 74.2 | 0.518 | 3.64 |