Hi, @brianhelba
Thank you for your efforts, i have been able to submit normally since 23rd Jun.However, when i made a submission just now, another issue occured :'access denied. see the console for details ’
@brianhelba I am also getting the same error, “Access denied see the console for more details”.
@brianhelba @philipp.tschandl I’m getting the same error “Access denied”. I was able to upload the predictions (testing phase, task 3) to the live leaderboard up until now.
Could you please look into this?
I think it is because of some limit, It will be better if organizers mentions the limit, like number of submissions allowed in a week, as it is blocking access for a week.
Hi!
I am facing the problem. During submission, i got “Access denied. See the console for more details.”. Is this because i submit too many results in short time term?
It’s certainly possible, I believe we’re currently limiting submissions to 10 per week (per phase). @brianhelba Can you confirm?
Oh, 10 per week is pretty few. Is there a chance to extend it?
Hi @chenzj, @mkulariya, @tomas, @maciejs,
I’ve moved all of the questions about submission rate limiting to the same thread.
To clarify, as @dan.lamanna stated, we limit submissions to ISIC Live to 10 per week, per task. We do this to avoid cheating via “wacky boosting” and to discourage well-intended submitters from overfitting their models to the test data.
In the next few weeks, we will be launching some technical improvements to the ISIC Live submission system, which will give much better feedback about why a submission fails or is not allowed. This is a great time for us to reconsider our ‘10 per week’ policy, so I’d really like to hear everyone’s thoughts on this.
- Why do you find yourself making more than 10 submissions per week?
- Why can’t you use a subpartition of the training set to do internal testing, and only submit your best ideas to the ISIC Live submission system?
- How can ISIC Live be maximally permissive for your needs, but also prevent cheating and overfitting from others?
Thanks @brianhelba to clarify, it is good to have a submission limit but it will be better if it will be per day basis, like it can be per day 3 or 5, not weekly, as once crossed the limit we have to wait for another 7 days and in 7 days, again we will have lots of experimentation , and it will follow the never ending cycle, so better make per day basis.