Single model or ensemble?

Semantic Segmentation
#1

I have spent more than ten days in this competition, but I still can’t achieve the result of single model higher than 0.65. I admire the result of the top players in the leaderboard. Would you like to ask whether the result is scored by single model or ensemble?

In addition, I found that there is a big gap between my local cv and lb. Is that the same with everyone?

#2

Hi!
My local score is also much higher than the LB, difference is ~0.09.
Can I ask you, what score are you getting locally to reach 0.65 on the LB?