pretrained-models.pytorch
pretrained-models.pytorch copied to clipboard
Cadene
Pretrained ConvNets for pytorch: NASNet, ResNeXt, ResNet, InceptionV4, InceptionResnetV2, Xception, DPN, etc.
I found mime-type bug. Default inceptionv3() can't hand over num_classes to models.inception_v3() when pretrained == False So I added argument 'num_classes' when calling inception_v3()
Hi I would prefer to store the weight file locally instead instead of downloading it (as the download speed is very slow). What is is the best way of doing...
Hi Cadene, nice work! 👍 Is there any efficient way to fine-tune these models on other datasets?
The last pooling layer of SENet should be change from `self.avg_pool = nn.AvgPool2d(7, stride=1)` to `self.avg_pool = nn.AdaptiveAvgPool2d((1, 1))` so that the model can take different input size.
do you have a example with this note: By default, models will be downloaded to your $HOME/.torch folder. You can modify this behavior using the $TORCH_MODEL_ZOO variable as follow: export...
Hi Cadene, Thanks a lot for providing these models. I am relatively new to PyTorch I am finding it difficult to generate Class Activation Maps for my trained model (InceptionResnetV2)...
Hi i get a size mismatch issue for se resnext 101 32 x4d I changed the classes to 1 from 1000. `size mismatch, m1: [80 x 8192], m2: [2048 x...
This is just advice. In production environment, latency is also very important. Do you have any plan to train small network, such as mobilenet v1&v2, shufflenet v1&v2 and so on???
