# 2nd place Solution

I trained on train images only. My attempts with pseudo labeling on extra-images did not work. The reason for this is likely the lack of experimentation or quite possibly my incompetence.
I used 6fold seresnext 101
lr: 2e-4 
batch size: 16
input size: 448x448x3
epochs: 5
augmentations: standard
tta: 8x
cv: ~0.92
By averaging folds, it scored 0.927+ on public lb and 0.935+ on private lb.
My CV improvements did not reflect on LB.
I believe strongly that training data is noisy.