
Posted by jordiju over 1 year ago | Quote
Preparations have taken a bit longer than anticipated, but it was for your benefit as we have much simplified the procedure for participants.
We anticipate the Challenge will officially start this week.
- The Organizers
Posted by Raphael over 1 year ago | Quote
Dear All
Unfortunately I have to tell you that there was a bug in the grader resulting in not showing the correct score of a solution. We have fixed the grader and removed affected submissions. Feel free to resubmit those to receive correct grading.
If there are any questions, don’t hesitate to reach out to us
Posted by Raphael about 1 year ago | Quote
Dear All, We have marked entries in the leaderboard that are the sample solution. While they still appear on the leadre board they won’t count for the actual leaderboard.
Happy hacking
Posted by jordiju about 1 year ago | Quote
Hi All
I’ve cleaned up the “Baseline” Submissions (i.e. those that just submit the Sample Submission we provide). Just so they don’t clutter up the leaderboard too much.
Julian
Posted by jordiju about 1 year ago | Quote
Just wanted to add that we are encouraged by the increasing number of submissions in recent weeks/days. We know this problem is probably not as accessible as many other ‘standard’ AI Challenges. We really appreciate everybody’s effort that tries to make a contribution! Keep up the good work and I hope everybody has some fun along the way, too.
Thanks! Julian
Posted by jordiju about 1 year ago | Quote
Hi all
Participant @iquadrat has pointed out an error in problem instance 09. We have corrected it and pushed an update.
Please pull the changes from the starter-kit repo (https://github.com/crowdAI/train-schedule-optimisation-challenge-starter-kit) to get the update.
Posted by jordiju about 1 year ago | Quote
All,
Based on the valid points raised by @LeoB and @Palstek in the thread Questions about the Competition Rules, we have added the following condition to the challenge rules.
In order to be eligible for prizes * The solver must generic enough that it is able to handle problem instances that are similar to the nine published problem instances. Custom tuning of the solver to the nine published instances is allowed, but must be within reasonable limits. For example, (hyper)parameter-tuning of an algorithm based on the nine instances is ok. Reading in solutions, or partial solutions, to the problem instances as part of the “algorithm” is not ok.
Posted by Raphael about 1 year ago | Quote
The bug that has been reported here (https://www.crowdai.org/topics/validator-calculating-penalty-for-problem-instance-7-wrong/discussion) has been fixed and deployed. Affected solutions have been deleted. If you have been affected, we have contacted you via e-mail. Please resubmit the solution.
Dear All
In this topic the organizers will announce updates to the Challenge, such as - when the Challenge Status changes - when documentation / helper scripts are updates or added - other important updates
You can stay up-do-date by subscribing to this topic.
Much Success! - The Organizers