PyTorch, the popular open-source machine learning framework, has been updated to version 1.3.0 and continues to gain traction due to its focus on meeting the needs of both researchers and developers.
Some changes:
- experimental support for named tensors. You can now access the tensor dimensions by name instead of specifying the absolute position:
NCHW = [‘N’, ‘C’, ‘H’, ‘W’]
images = torch.randn(32, 3, 56, 56, names=NCHW)
images.sum(‘C’)
images.select(‘C’, index=0) - support for 8-bit quantization via FBGEMM and QNNPACK, which are integrated into PyTorch and utilize a common API;
- operation on mobile devices running iOS and Android;
- release of additional tools and libraries for model interpretation.
Additionally, published recordings of talks from the recent Pytorch Developer Conference 2019.
Source: linux.org.ru
