Availability of validation set during the final test phase

Hi, I wanted to know during the final test phase will the debugging logs reflect the algorithm performance on the validation set or the three training WSI's.

Will we get wsi that are used for validation in live leatherboard phase during test phase?

The debugging, validation, and test cohorts are three separate cohorts with different WSIs. The debugging cohort consists of 3 slides from the training dataset. The validation cohort consists of 9 slides from center E. The final test cohort consists of 9 slides from center E (no overlap with the validation set) and 19 from center F.

At this point, we do not plan to release the validation set before the test phase. I will discuss it with the team though, and let you know if that changes.

What is the utility of the test phase? Since we can't test our improved models on the validation set and can't benchmark them with our best models in the validation phase. The unavailability of the validation set makes it impossible for us to submit our best model in the test phase.

The validation set is the set that is used for the live leaderboard phase. You can choose your model based on the performance there. I will post an update of the metrics output later this week, giving you more detailed access to the performance.

The test phase will be used to determine the winners of the challenge.