Overview on kaggle: https://www.kaggle.com/c/dogs-vs-cats/overview

numpy pandas torch >=1.8.0 torchvision tensorboard scikit-learn

Use: python main.py (temporarily)

We merged resnet50, inceptionv3 and efficientnet-b4 with a concat of output feature vector. Then using a dropout layer with p=5.

run main.py

use function with prefix like test_merge(model_path='./tmp/merge_effi/epoch_num_1.pth',batch_size=100,csv_path="./results/final/test.csv",binary=True).