CharlesNord

Results 11 comments of CharlesNord

> Hi @mkocabas, > > There is a slight mismatch in the notation used to denote the extracted dimensions from the shape of the input tensor. [here](https://github.com/mkocabas/CoordConv-pytorch/blob/master/CoordConv.py#L84) > > ```...

In fact, there are not much in common between stargan and stargan v2

I think this can help to stabilize the weight update. Since stochastic gradient descent could sometimes lead to the wrong optimization direction, making the training noisy. Averaging with the recent...

I found the comment under this issue could be the answer: https://github.com/clovaai/stargan-v2/issues/62

I encountered with the same issue, did you solve it ? I can generate the desired model by command line, but cannot use python api.

I finally solved the problem, Just delete the `pycuda.autoinit` in `ImagenetCalibrator.py` and add it in the beginning of the `onnx_to_tensor.py` the generated engine should live in the CUDA context created...

In file `/Path/to/your/virtualenvs/envs/inpainting/lib/python3.8/site-packages/clip/clip.py` temporarily modify the following lines: ```python with urllib.request.urlopen(url) as source, open(download_target, "wb") as output: ``` to ```python import ssl context = ssl._create_unverified_context() with urllib.request.urlopen(url, context=context) as source,...

from kornia.constants import SamplePadding

Because the author used the categorical batch normalization