Regarding the description in the paper

EricGuo5513 / momask-codes

Official implementation of "MoMask: Generative Masked Modeling of 3D Human Motions (CVPR2024)"

https://ericguo5513.github.io/momask/

MIT License

856 stars 73 forks source link

Regarding the description in the paper #89

Closed Suzixin7894946 closed 2 weeks ago

Suzixin7894946 commented 3 weeks ago

Regarding the description in the paper, "During inference, we use the CFG scale of 4 and 5 for M-Transformer and R-Transformer on HumanML3D, and (2, 5) on KIT-ML." I am not quite clear on how to set different CFG scale values for the two transformers during the inference process.

Murrol commented 2 weeks ago

They are just empirical settings.