Generate 256-Dimensional Latent Vectors Using Pretrained CMDEncoder, PARAMEncoder, and EXTEncoder

[MHassan1122]

Dear @samxuxiang,
I hope you will be in good health.
I am working with the SkexGen project and am currently trying to generate 256-dimensional latent vectors using the pretrained CMDEncoder, PARAMEncoder, and EXTEncoder models with the provided .pkl training, validation, and testing datasets. However, I am unsure of the specific steps and input data preparation required to successfully accomplish this task.

Steps to Reproduce :

Access pretrained models for CMDEncoder, PARAMEncoder, and EXTEncoder.
Attempt to prepare and input (.pkl) data files for generating latent vectors.
Generate latent vectors using the pretrained models.

Expected Behavior:

I aim to use the pretrained models to process the .pkl data files (train, val, and test) and produce 256-dimensional latent vectors that can be utilized for further processing or analysis.
Thank you very much for you consideration.