Continued participation?

I see that the timing for this Blitz is over, but it appears I can continue to create submissions. Is that correct? Can I continue working on this problem even though the Blitz for it is over? And will those submissions be ranked on the existing leaderboard?

View post 0

Incorrect townhall link

The link to the Learning to Smell townhall is incorrect. The current one on the overview page links to the Seismic Facies challenge. It is easy enough to find the townhall for Learning to Smell from there, but the link currently goes to the wrong video.

View post 1

Round 2 details?

Round 2 is nearly over according to the schedule, but I’ve not seen any change in the datasets. During the townhall it was mentioned that there would be some changes with respect to the compound labels (such as grouping all floral smells into one floral group, or similar). Has that been done? Or is Round 2 just an extended time on the same dataset as Round 1?

View post 0

Placings in Blitz #4 affected by after deadline submissions

Thank you for the clarification. I was not aware that the final scores had been calculated - the notice reading “Note : The scores displayed on the leaderboard are computed only using 60% of the whole test set. The scores computed using the whole test set will be released at the end of the Round.” is still on the Blitz #4 leaderboard.

View post 1

Placings in Blitz #4 affected by after deadline submissions

I notice that overall placings in Blitz #4 are being affected by submissions to Learning to Smell and Seismic Facies that were posted after the deadline for Blitz #4. How will final placings be determined? Will post-deadline submissions and their placings in these two be taken into account for the final scores and placings?

View post 0

Inconsistent sample numbers

Thank you for the clarification and the adjustment. Is the training set correct without the header? The example shows it is read without a header, but I’m wondering, given the mistake in the test set.

View post 0

Inconsistent sample numbers

When a submission is made, the evaluator expects 980 predictions, but there are 981 rows in the test.csv dataset. Looking at the baseline, it appears that the first one is lost due to missing the header = None argument in read_csv. Or, perhaps a different row was lost? Also, there various inconsistencies and what appear to be incomplete entries in the descriptions, discussion boards, etc.

Is this a legitimate challenge or something else?

View post 0