Text-to-Motion Generation on HumanML3D-X 1.0 (test)
0FIDReal
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Real2026.01 | 0 | 39.3 | 3.862 | 9.811 | — | |
| H-GPTw.o. CoTChain-of-Thought=false, Backbone=LLaMA-3.1-8B2026.01 | 0.229 | 33.3 | 4.455 | 9.674 | 2.754 | |
| H-GPTChain-of-Thought=true, Backbone=LLaMA-3.1-8B2026.01 | 0.255 | 33.2 | 4.513 | 9.383 | 3.256 | |
| T2M-GPT2026.01 | 0.677 | 35.7 | 4.182 | 9.26 | 2.599 | |
| MLD2026.01 | 0.817 | 35.7 | 4.189 | 9.549 | 3.224 | |
| T2M2026.01 | 2.078 | 25.2 | 4.967 | 8.325 | 4.208 | |
| MotionDiffuseNotes=Sensitive to parameters2026.01 | 19.423 | 10 | 6.919 | 5.183 | 3.87 |