search

seohongpark / HIQL

HIQL: Offline Goal-Conditioned RL with Latent States as Actions (NeurIPS 2023)
MIT License
71 stars 6 forks source link

How to evaluate the algorithm after training? #5

Closed 007lbx closed 2 weeks ago

007lbx commented 2 weeks ago

Hi, I can't find the evaluate.py file to evaluate the algorithm after training! in Appendix D1: "At test time, to specify a goal g for the policy, we set the first two state dimensions (which correspond to the x-y coordinates) to the target goal given by the environment and the remaining proprioceptive state dimensions to those of the first observation in the dataset. "

seohongpark commented 2 weeks ago

Hi 007lbx, here is the evaluation file.