Hi, I’m playing with the baseline for training single/multiple agents.
However, it seems some of the modules are missing?
I cannot find the python file in the system either. Is it because some recent update of flatland changed those modules?
Hi @xzhaoma
The modules you are trying to import are only present in the beta version of Flatland. You have to clone the git repository and build it from source.
Hope this helps.
Best regards,
Erik
PS: If you are using the new beta repository you can also try out sparse_rail_generator and sparse_schedule_generator to be ready for Round 2 of the challenge.
ATTENTION: The stock observation builders and predictors are not yet ready to be used. Be sure to use your custom obersvations and adjust them to the new Flatland properties such as stochastic events and different speeds.
These functions will be updated in the next few days.
Which branch is the beta version?
I really appreciate the fast developing cycle of the project and your quick response.
But I’d really recommend to make this learning get-start part clear and smooth. If the baseline repo is out-of-date, please indicate that and provide solution to fix it…
For someone new in the competition, the starting trajectory is now broken and not smooth…
1 Like
Hi @xzhaoma
Thank you for your feedback. We have now merged the beta version into the master branch. There will be an announcement this weekend and a new release of Flatland 2.0.
I will update the baselines repo as soon as possible.
Do you have any recommendation for the get-started part`? We also believe that this is a very important part and are happy for any helpful input.
Best regards,
Erik
Hi @mlerik,
Thanks for your quick response.
For the ‘get-start’ part, I would say that one of the most important factors is the smoothness of the get-start process. For instance, I can install packages, run examples, understand the settings, details of the baseline program without experiencing bugs, confusions, or having to ask in the forum or submit issues several times…
I totally understand that as you’re developing the main packages, something will break, and it is sometimes impossible to have a fully developed environment before releasing the competition. But for people who are uncertain about attending the competition, the initial run on several examples matters a lot as they will evaluate the completeness of the competition based on the experience.
I believe it is not too hard to have a virtual machine and an automated script ready for testing after every update on the main package. And it will guarantee that things will not be broken up and also save you a lot of time in responding to issues and bugs.
Hope it helps.
Thank you @xzhaoma for your valuable feedback.
We have set up some testing for the package and are constantly trying to improve it such that everything works just fine.
We are very happy for valuable feedback like yours such that we improve our repository and make the challenge accessible to as many enthusiasts as possible.
If you like you are also very welcome to contribute directly to the baselines repo and help others get an easier start.
All the best,
The Flatland Team
Friendly reminder: There is a community prize (
) awarded to great community contributions such as documentations to get started, new observation builders, predictors, sharing of ideas and in general just helping the community advance with this research question…
Hi, thanks for the updates on the baseline.
However, it seems there still exist some glitches and the readme file is not updated with the new code (sparse_scheduler, etc.)
I can try to fix some of the issues, but do you want me to submit PR or list issue here so that your team will fix it?
For instance, if cloning the baseline repo from gitlab, the root dir of the repo is not added to the PYTHONPATH so that running the single agent training script will cause errors such as:
Traceback (most recent call last): File "training_navigation.py", line 9, in <module> from dueling_double_dqn import Agent File "/path-to-baseline/baselines/torch_training/dueling_double_dqn.py", line 11, in <module> from torch_training.model import QNetwork, QNetwork2 ModuleNotFoundError: No module named 'torch_training'
HI @xzhaoma
Please send your fixes as PR. We are very glad for any community contributions to bring this challenge forward.
Best regards,
Erik
Submitted 2 simple fix.
Thanks.
@mlerik
No problem, glad to contribute.
I can also fix the Readme since the “complex” generator and scheduler has been changed to the sparse version and the doc are old.
Or do you want to do it on your team side?
Hi @xzhaoma
We really want this to be a community project where ideas grow together and we improve by ways that would be impossible if only a small team works on the project.
We would love to get your insights for many different topics. Therefore, I believe it is great if you contribute to the documentation directly through PR.
Best regards,
Erik