Can we get pretrained model for dynamics predictor as input data is very large in size after frame extraction and running for high number of epochs isn't viable in many low computational machines?

[krgaurav-iitb]