yangze0930 / NTS-Net

This is a PyTorch implementation of the ECCV2018 paper "Learning to Navigate for Fine-grained Classification" (Ze Yang, Tiange Luo, Dong Wang, Zhiqiang Hu, Jun Gao, Liwei Wang).
MIT License
450 stars 118 forks source link

Different test accuracy when testing with different BATCH_SIZE in config.py #46

Closed yangxh11 closed 4 years ago

yangxh11 commented 4 years ago

The model is fully trained and saved. When testing, I only change the BATCH_SIZE in config.py and get different test accuracy.

giangnguyen2412 commented 1 year ago

did you figure out why? i got the same thing here in 2023 :D

giangnguyen2412 commented 1 year ago

bs8: 0.412 bs128: 0.726 bs256: 0.810

it seems memory is all you need to solve cub-200 lol

giangnguyen2412 commented 1 year ago

@yangze0930 ! do you know why this problem arises?

giangnguyen2412 commented 1 year ago

found the problem. I forgot to set net.eval(). Everything works now.