DeepLabV3Plus_resnext101_32x8d (LB:0.486) #20
Baek-jongwon
started this conversation in
실험결과 공유
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
model = DeepLabV3Plus
backbone = resnext101
augmentations = 미적용
criterion = CrossEntropyLoss
optimizer = Adam
epochs = 50
LB score : 0.486
특이사항 : 100epochs 결과는 0.342로 더 하락하였음.
Beta Was this translation helpful? Give feedback.
All reactions