I have a question regarding the eval_solution.py. If I want to debug a level more precisely, I have to run in isolated, as it is also written in the comments of eval_solution.py, so that I can visualize it. But if I run the isolated level, it is completely different from what is run when I run all levels normally. How can I fix this? Or what is the reason for this? Does it have something to do with the seed? Thanks for the help
Posted by: marff1 @ April 15, 2024, 7:38 a.m.Hello marff1,
I am sorry for the confusion. There are two parts needed to replicate the scenario: the pkl file, and the seed. When running eval-solution.py, in addition to specifying a pkl file, you can pass in an environment seed via the "--env-seed" option. The seed corresponding to each pkl file can be found by looking in the "metadata.csv" file located in the root of the scenario set, under the "seed" column.
So, for example, if you are trying to debug the Round 2 Test Scenarios (https://airliftchallenge.com/scenarios/Round2_test_Scenarios.zip), and you want to run "Test_1\Level_5.pkl", you would pass in "15645" as the environment seed. There is also a corresponding parameter/column for specifying the solution seed.
Does that clarify? I would happy to elaborate further.
Thank you,
Andre Beckus
Thank You very much!
Posted by: marff1 @ April 18, 2024, 10:55 a.m.