-
Hi,
Thanks for the wonderful work! I have a question regarding the hyperparameters in the paper. Are the default hyperparameters stored in config.locomotion the same as those used in Figures 2, 5,…
-
I really enjoyed working with this repo.Thanks for making d3rlpy, which helped me a lot to get started in offline RL.
I just got into offline RL. There are many parameter indicators recorded in the l…
-
I was trying to run the AWR algorithm on the HalfCheetah environment as given in the README. So, first of all there is no `run.py` code in the folder of AWR. I copied `run_script.py` to the root of AW…
-
Dear CORL Team,
Firstly, I would like to express my appreciation for your work on the CORL codebase. The clean, single-file implementation coupled with a robust performance report has greatly impre…
-
I want to test the results of the pretrained weight, but when I execute bash scripts/eval/rearrange_pretrained.sh, the following error message appears.
$bash scripts/eval/rearrange_pretrained.sh
N…
-
Hi, Kumar! In the last [issue](https://github.com/aviralkumar2907/BEAR/issues/6), you mentioned that you don't test BEAR on the final buffer setting and recommend me using d4rl datasets. Following you…
-
I am trying to train a ppo agent using this [repository](https://github.com/HumanCompatibleAI/human_aware_rl), which is a repo for the DRL implementations compatible with a multi-agent environment (Ov…
-
**Is your feature request related to a problem? Please describe.**
Model-based offline RL algorithms which are able to handle image inputs are necessary for some environments.
**Describe the solut…
-
Ray generates the following error when creating an RLPredictor using a restored checkpoint in a different Ray session from the one that created the checkpoint. This error does not occur when creating …
-
According to your paper and the README, your code repository is lack of the file "generate_offline_data" so as to fail to run the offline rl code. Could you please provide this file?
When I run the…