rondorf / borel Goto Github PK

Official implementation for the paper "Offline Meta RL - Identifiability Challenges and Effective Data Collection Strategies", NeurIPS 2021.

Python 99.27% Ruby 0.69% Shell 0.03%

borel's Issues

The number of tasks used in Cheetah-vel

Hi, I'm trying to implement the code now, but I can't make sure how many tasks the environment of Cheetah-vel use as the paper. I would greatly thanks for your help!

Average return in gridworld

Hello, may I ask why the average return of the gridworld task I reproduced using this code is very different from the average return in the paper?All parameters are consistent with the article but poor results are obtained.

Regarding the Walker2d-Randparams Environment

Thank you for your inspirational great work and open sourcing of the code. I have a question regarding the walker random environment. The walker can not walk when the reward is 800, this is the case for all the algorithms that use this environment. When doing a comparative analysis are we supposed only to compare the reward or is the reward scaled for this environment?

Have you observed a walking behavior when using the algorithms that use this environment?

Multi-processing for online_training script

Hi,

Is multi-processing functionality available for the online_training.py script? For me, for the same env, the BOReL code is taking much longer to run than VariBAD.

Thank you!

Recommend Projects