EfficientNet-PyTorch

EfficientNet-PyTorch copied to clipboard

Pretrained weights don't give same accuracy as stated in repo

6

I tried using pretrained weights for efficientnet-b0 and efficientnet-b1 and evaluated it on Imagenet but did not get the same Accuracy (rank 1 and rank 5) that is mentioned in...

neiladit

Error when loading pretrained model with num_classes!=1000

6

I am trying to load a pretrained model with a different number of classes to apply transfer learning on cifar10. When I run > model = EfficientNet.from_pretrained('efficientnet-b1', num_classes=23) I got...

CosimoRulli

Models trained in v0.6.3 give wrong output in v0.7.0

3

We have been using EfficientNet 0.6.3 to train models for multi-label classification. When upgrading to 0.7.0 and loading the previously trained weights, the model output is incorrect (basically predicting negative...

jamt9000

Bug for multitask learning and all the fc layer replacement situations

1

Hi, I have a problem with multi-task learning when using EfficientNet. I want to load the pre-trained model and remove the raw FC layers, then add my own 2 new...

xingjchen

Does anyone have any thoughts on where and how to add gradient checkpoints in this implementation to reduce memory usage during training?

maxwxzheng

Getting High Loss Values

6

I am using this b7 model in efficientNet , but getting huge losses , can't able to figure out where I am going wrong # downloading the pretrained model -...

IamSparky

lukemelas

Question regarding last layers

3

Hello, I got a question: what does passing num_classes actually do? add another extra linear layer at the end with corresponding classes? What happens if I don't pass it? How...

finnBsch

the inference speed is much slower than original TensorFlow code

12

hi，I ran your sample code and then tested the inference time，but I find that the inference speed is much slower than original TensorFlow code on my computer。  

tensorflower

Hi, the README.md in examples/imagenet/README.md says that the code is not trainable. however, there is training code in examples/imagenet/main.py file. should the readme.md updated or the training code is not...

ardeal