How to know if pass the exercise or not?

CommonRoad-Search
1

Hi,

I uploaded my solutions to
AI Programming Exercise 2022 - Bonus exercise in Challenge.

In leaderboard, my solved scenarios is 360, but my top1 solutions rank is lower than 340, does this mean I have passed the minimum requirement for passing the exercise? I also uploaded the required documents on moodle.

Thanks in advance

2

Hi ge89nus,

Thank you for your question.

The top-1 solution is not relevant for passing the bonus exercise. If your algorithm can solve 360 senarios, you will pass the bonus exercise for sure.

Please note that if you solve them with different config files, you should include them in the submission and specify how to reproduce them.

Hope this can answer your question.

Best,
Mingxuan

3

Hi Mingxuan
I uploaded the config file and student.py on moodle, will I get respond on it ?

4

Hi ge89nus,

we will give a feedback on all submissions after submission window closes (i.e., after the 2 months). Then you will also get notified if you received the two bonus points.

Best,
Gerald

5

Hi, in the leaderboard my approach has solved 346 scenarios, does this pass the exercise?

Thanks!

6

You mean 340 scenarios?

7

Hi flo99,

Thank you for asking.

I’m not referring the metric of passing the exercise, but the case in this thread.

Best,
Mingxuan

8

I don’t understand the task. Currently, I can solve 362 scenarios without having written a single line of code (if you don’t count config files). Do I pass, if I write a heuristic that is just not worse than the one given?

1 Like
9

also, I am a bit skeptical about time-out criteria. If I write a parallel search algorithm and run it on a high-end desktop, I might get less time-outs and more solutions than someone who had less compute resources. Also, time-outs can be evaded anyway by writing a different processing algorithm and then changing the xml solution file by faking the computation time. Do you check the code submissions afterwards for feasibility in normal use (without cheating)? If you do, then how can I be sure, that the code that worked perfectly fine on my desktop doesn’t time out on your machine?

10

I observed that my desktop (R9 3900X @ 4GHz) solves, depending on the algorithm, 10 to 30 more scenarios than my laptop (i5 8265u ca. 2.3 GHz), with less time-outs
also if I use single-thread batching (higher cpu clock rate) with the out-of-the-box gbfs-search, I can solve 6 more scenarios than with multithreaded batching
I uploaded my solutions, they seem to all be correct

11

Hello,
If I run batch_processing_parallel, my code generates 339 solutions. However, in the folder outputs/solutions I can see 340 solution files, all of which are successful, according to the commonroad website. What does that mean? Have I passed or not?

12

Could you give a hint on which config files you changed?