MaskedTensor

This library is a part of the PyTorch project. Please note that this library is currently classified as a prototype – that is, this library is at an early stage for feedback and testing, and we encourage users to submit any issues they may encounter, feature requests, etc. The Github can be found here.

The purpose of maskedtensor is to serve as an extension to torch.Tensor, especially in cases of:

• using any masked semantics (e.g. variable length tensors, nan* operators, etc.)

• differentiation between 0 and NaN gradients

• various sparse applications (see tutorial)

More details can be found in the Overview tutorial.

Installation

• Installation

Tutorials:

• Overview of MaskedTensors
• Sparse semantics
• Distinguishing between 0 and NaN gradient
• Safe Softmax
• Efficiency of writing “sparse” semantics for Adagrad
• Implemented missing torch.nan* operators

Python API

• Unary Operations
• Binary Operations
• Reductions
• View and select functions