Noah Trenaman
Noah Trenaman
For large graphs, syncing the parameter and gradient data from GPU to CPU is slow. A simple extension in C could make this much faster.
A lot of functionality isn't tested yet but should be. This includes: * Initializers * Activation functions * Visualization * Graph syncing with model
PyTorch 1.2 indexing isn't 100% as fast as scatter and gather operations. There might be a speed boost in using these operations instead of indexing in the Model class's `forward`...
Hi, I was having trouble downloading data from Baidu Wangpan (百度网盘.) It would be very useful if the data was also made available on another cloud service like GCS or...
Hey, thanks so much for making this package! I'm getting into Swift for TF and am so glad that there's a platform-independent plotting library to use. I ran into a...
Adding just one line to reduce possible confusion in locating Unreal Engine. This is in the client Python package. If the user doesn't provide a path to UE4 and the...
I'm having trouble shuffling a tensor along a specified axis. I tried using a shuffled array of indices to do this, like so: ```swift var indices: [Int] = Array(0...cardinality) indices.shuffle()...
Newer data is available from https://fdc.nal.usda.gov/ which was released this week. The package will need to be reworked so that it can work with the new API, as support for...