search

IDEA-Research / HumanTOMATO

[ICML 2024] 🍅HumanTOMATO: Text-aligned Whole-body Motion Generation
https://lhchen.top/HumanTOMATO
Other
240 stars 6 forks source link

How to generate embeddings on the full datasets? #22

Closed HonglinChu closed 2 weeks ago

HonglinChu commented 2 weeks ago

Hi, I test the demo code https://github.com/LinghaoChan/OpenTMA-demo. image I am referring the code to know how to generate them more conveniently. Maybe you can provide the corresponding scripts. https://github.com/IDEA-Research/HumanTOMATO/blob/7120079a6edae10482e934aad2d6050229a02796/OpenTMA/tma/models/modeltype/temos.py#L408

LinghaoChan commented 2 weeks ago

The core codes for saving npys is here. https://github.com/IDEA-Research/HumanTOMATO/blob/7120079a6edae10482e934aad2d6050229a02796/OpenTMA/tma/models/modeltype/temos.py#L884

There are two ways for you to self-define the split.

  1. Revise the test split file during training.
  2. Run the test file of the OpenTMA. (point out the checkpoint name in the yaml file in advance.)