-
Currently we cannot add two distributions together:
``` r
library(distributional)
d Error in dist$inverse(at): Inverting transformations for distributions is not yet supported.
```
Created o…
-
Currently on the cuDNN section of Nvidia's website the following versions of cuDNN are listed:
> Download cuDNN v5.1 RC (June 19, 2016), for CUDA 8.0 RC
>
> Download cuDNN v5.1 RC (June 16, 2016), f…
-
In the source code, only 'validate.py' uses HarDBlock_v2.
And the model with HarDBlock_v2 is faster than that with HarDBlock.
I want to know the difference between v1 and v2.
-
_This is a copy of a presentation for MIOpen team I held a couple of years ago, when we've introduced and implemented the Solver/Solution architecture. It does not cover the recent additions like GetW…
-
Hi!
I was just wondering why did you choose a 4D convolution for a scoring function and not, for example, a fully connected layer, since at the end it will produce only a single number?
I am tr…
-
Hey, first of all, thanks for your work - pretty fast :)
I just wanted to test your repository and noticed that the code fails for inference on CPU due to the grouped convolution.
**Code:**
…
-
CUDNN doesn't support the dilated convolution. I'm trying to use CUDNN engine to save GPU memory. Is it possible to use zeros to fill the dilated kernel where it shouldn't be convolved ?
-
It seems to me that a direct forward pass via `model(x)` and using the extractor's forward pass through `forward_pass_on_convolutions(x)` gives outputs of different sizes.
`forward_pass_on_convolut…
-
-
Hi Xiaohan Ding,
This is such excellent work and thanks you for sharing.
I was reading your paper and in conclusion, I saw
> RepVGG models are fast, simple, and practical ConvNets
> designed…