| [Deep Learning for NLP with Pytorch](http://pytorch.apachecn.org/cn/tutorials/beginner/deep_learning_nlp_tutorial.html) | [@JingTao](https://github.com/jingwangfei)[@friedhelm739](https://github.com/friedhelm739) | |
| **中级教程** | - | - |
| [Classifying Names with a Character-Level RNN](http://pytorch.apachecn.org/cn/tutorials/intermediate/char_rnn_classification_tutorial.html#) | [@孙永杰](https://github.com/yongjay13) | |
| [Generating Names with a Character-Level RNN](http://pytorch.apachecn.org/cn/tutorials/intermediate/char_rnn_generation_tutorial.html) | [@jianchengss](https://github.com/jianchengss) | |
| [Translation with a Sequence to Sequence Network and Attention](http://pytorch.apachecn.org/cn/tutorials/intermediate/seq2seq_translation_tutorial.html) | [@EWilsen](http://community.apachecn.org/?/people/EWilsen) | |
| [Neural Transfer with PyTorch](http://pytorch.apachecn.org/cn/tutorials/advanced/neural_style_tutorial.html) | [@Twinkle](https://github.com/kemingzeng) | |
| [Creating extensions using numpy and scipy](http://pytorch.apachecn.org/cn/tutorials/advanced/numpy_extensions_tutorial.html) | [@飞龙](https://github.com/wizardforcel) | |
| [Transfering a model from PyTorch to Caffe2 and Mobile using ONNX](http://pytorch.apachecn.org/cn/tutorials/advanced/super_resolution_with_caffe2.html) | [@片刻](https://github.com/jiangzhonglian) | |
| [Custom C extensions for pytorch](http://pytorch.apachecn.org/cn/tutorials/advanced/c_extension.html) | [@飞龙](https://github.com/wizardforcel) | |
In this short tutorial, we will be going over the distributed package of PyTorch. We’ll see how to set up the distributed setting, use the different communication strategies, and go over some the internals of the package.
在本文中, 我们将介绍如何扩展 [`torch.nn`](../nn.html#module-torch.nn"torch.nn"), [`torch.autograd`](../autograd.html#module-torch.autograd"torch.autograd") 模块, 并且使用我们的 C 库来编写自定义的 C 扩展工具.
`torch.optim` is a package implementing various optimization algorithms. Most commonly used methods are already supported, and the interface is general enough, so that more sophisticated ones can be also easily integrated in the future.