Regarding an error: The algorithm failed on one or more cases.

Regarding an error: The algorithm failed on one or more cases.  

  By: jwoong.shin on July 5, 2023, 10:28 a.m.

There are several failed submissions with an error message: "The algorithm failed on one or more cases."

In most of the cases, submissions try to download some weights from the web, which is not allowed in this contest. To prevent data leakage, we disallowed internet transactions within a Docker Container.

To avoid this issue, you should remove any code that downloads from the web (e.g. pre-trained weights). Weights from the web would probably no longer needed after the model has trained on them. If you need anything from the web for the evaluation, please consider downloading in advance and including it in your Docker Image.

We apologize for the inconvenience, but would appreciate for your understanding. Thank you!

Re: Regarding an error: The algorithm failed on one or more cases.  

  By: chatterjeeahan02 on July 6, 2023, 1:32 p.m.

Hi, I am also facing the exact same issue, though the code is working locally with test.sh, but throwing this exact error. Though my code doesn't take any weights from the pre-trained models, and no internet access is required also for the same. Can you please shed some light on the issue, or maybe can you share the logs of the error that has been generated?

Re: Regarding an error: The algorithm failed on one or more cases.  

  By: Tuesday on July 9, 2023, 4:02 a.m.

I'm also facing the same issue. I can test it locally using test.sh, and there is no code that makes any internet requests. However, I still encounter the aforementioned error when submitting.

Re: Regarding an error: The algorithm failed on one or more cases.  

  By: Tuesday on July 9, 2023, 9:37 a.m.

Can you provide some logs about the submission failure? I have failed for many times and my code didn't download anything from the Internet.

Re: Regarding an error: The algorithm failed on one or more cases.  

  By: aaron.valero on July 10, 2023, 10:15 a.m.

Unfortunately, GC does not allow participants to check their logs (refer to this), to know more about your individual error log, please send an email to oncology-ai-research@lunit.io.