js-pytorch 0.7.2

New features:

• cdnjs tag available, for easier integration into web browsers and apps.
• nn.MSELoss added, with Seyronh's help.
• Fixed saving and loading models when using gpu kernels.

js-pytorch 0.5.0

In this version, 2 main features were added:

• 1. GPU support
  • The back-end GPU kernels are managed by the gpu.js library.
  • To add a Tensor to the gpu, pass 'gpu' as the device argument.
  • To add a whole layer (nn.Linear, nn.Block, nn.MultiHeadSelfAttention) to the gpu, , pass 'gpu' as the device argument.
  • The syntax is just like the PyTorch counterpart.
• 2. Saving and Loading models
  • The functions are torch.save and torch.load.
  • The functionality is explained on the README.md file.

js-pytorch 0.3.0

This new release contains the following features:

• TypeScript support and source-code.
• Developer tools, such as ESLint and prettier.
• Automated unit tests with jest.
• Benchmarking tools with tiny-bench.

Thanks to @pparke and @medic-code for the help with the implementations!

@eduardoleao052

js-pytorch 0.1.1

Initial release. Package available on npm at https://www.npmjs.com/package/js-pytorch.