mcgillmrl / prob_mbrl

A library of probabilistic model based RL algorithms in pytorch

mcgillmrl/prob_mbrl Issues

Regressor does not un-normalize outputs when not using output density
Updated 5 years ago
Dropout Should be turned off when collecting new samples with policy
Updated 5 years ago
Inconsistency with forward vs log_probability (more an inquiry)
Updated 5 years ago2
[FEA] Unit Testing
Updated 5 years ago2
Iterate_minibatches: does not reshuffle order of training data
Updated 5 years ago1
Resample() does not update dropout masks during policy training
Closed 5 years ago2
Does this code work currently?
Closed 5 years ago2