Mostafa Elhoushi

This code snippet used to pass but has recently started throwing an error: ``` def test_make_fx_model_train_with_optim(self, device): class Foo(nn.Module): def __init__(self): super().__init__() self.linear = nn.Linear(5, 5) def forward(self, x): return...

count running_mean and running_var params for BN

2

For batch norm layers, count the running_var and running_mean parameters in batch_norm layers

Add batchnorm parameters: running_mean, running_var, and others

1

Batchnorm has other parameters - in addition to `weight` and `bias` - that are not counted in torchsummary, such as `running_mean` and `running_var`: https://pytorch.org/docs/stable/_modules/torch/nn/modules/batchnorm.html

Log Message for each Pass if it Changed the Graph

1

Try to add a log message for each pass to print the `changed` boolean that indicates if that pass had any effect on the graph. This could be logged if...

Format of model file

4

I would like to use a model that is pre-trained on Keras or Tensorflow, and run it on kann. I am trying to look for the file format that the...

4-bits on ResNet18 results in 6% reduction in error

7

I have modified the `weight_bits` to 4 and the `iterative_steps` to `[0.3, 0.5, 0.8, 0.9, 0.95, 1]` but I got an accuracy of 83.12%, while the original paper got an...

Fixed compatibility issues when running with newer version of Keras. Fix has been tested to run on Keras v. 2.1.2 and Tensorflow 1.4.0

Error When Shifting Twice

7

Copying the question by @mengjingyouling from this [issue](https://github.com/mostafaelhoushi/DeepShift/issues/15#issuecomment-1105080853) to create a new issue: > We also want to discuss a problem with you. In your paper, the shift network is...

Passing a `weights.pth` file to the `--weights` is not working properly. It probably doesn't load the weights. While passing a `checkpoint.pth.tar` file to the `--weights` is working properly

Some alternatives to consider: - `--shift-depth all` - `--shift-all`