-
## Question
Why does the zoo call standard `make_vec_env()` for all environments, including Atari, when sb3 has a special function for it `make_atari_env()`?
## Train of thought
- train.py calls …
-
Hi! It would be awesome to be able to implement LSTM policies in this library, like in the former version. Is there an straightforward way to accomplish this with the current version?
-
**Important Note: We do not do technical support, nor consulting** and don't answer personal questions per email.
Please post your question on the [RL Discord](https://discord.com/invite/xhfNqQv), [R…
ghost updated
2 years ago
-
- [x] I have marked all applicable categories:
+ [x] exception-raising bug
+ [x] RL algorithm bug
+ [ ] documentation request (i.e. "X is missing from the documentation.")
+ [ ] ne…
-
Hi, I followed the instructions in Design a New Learning Environment to build a Rollerball project. I did 3-D ball demo before. Training model and run pre-trained model can totally work in that dem…
-
Hi @alex-petrenko,
I ran the codes on dmlab-30 with the exactly same arguments/configurations in README.
However, as shown in the below figure, the obtained scores (mean capped) are lower than the…
-
Related issue #171
As seen in #171, cloudpickle/pickle can easily fail when transferring models between Python versions (and in case some other shenanigans). There is currently no way to address t…
-
For custom environments do I need to normalize the observation array? Or it is done by stable-baselines internaly ?
This is the learning code
```
env = MyEnv(config)
policy_kwargs = di…
-
### What is the problem?
I found multiple runs for PPO still have different performance even we set the same seed.
How can we obtain the exact same result with the same seed?
Current strate…
-
I've read similar questions (e.g. #[30](https://github.com/hill-a/stable-baselines/issues/30)) that were asked here about loading the model after the training but still, I could not figure out what th…