To the organizers: can the jobs be killed by external processes before the 4 hours deadline (e.g. the cloud provider taking back the evaluation machine)?
I had a submission today, almost identical (same model, just trained a little bit longer) to one that was evaluated successfully a couple of days ago, which failed after ~1 hour with the following message:
Your submission did not output the expected file so it could not be scored. This may be due to an unhandled exception or syntax error in your code. The log output may have more details.
By that time, more than half of the chips were processed, but my logging didn’t reveal that the last chip was processed yet, so something must have killed that job in the meantime. Was the machine stuck? I have no other information from the logs.