News
Remarks on Upcoming Evaluation of Project 3
Written on 03.02.2025 09:11 by Thorsten Klößner
Dear students,
the evaluation of project 3 will take place today. As mentioned in the project description, we will use the policies produced by invoking your training script without arguments for the final evaluation. The evaluation logs you receive will contain an additional test that checks whether the models produced by your training script match the models you submitted in the resources directory. If the models differ, this will be mentioned in the test logs. This is purely for transparency, regardless of whether this test fails or not, all models produced by the training script will be used by the rest of the tests.
Some implementatons made use of additional files, e.g. to store a specific configuration of parameters to load at runtime. Since this was not explicitly disallowed in the project description, this was technically allowed, but not intended without certain restrictions. Our infrastructure should be able to deal with this, but in case you suspect that your implementation was not evaluated correctly, please send me a message.
Regards,
Thorsten Klößner