本项目是一套面向初学者的 TF 教程。
其中介绍了 TF 的基本用法、一点点 CNN,着重强调了 seq2seq 模型。
seq2seq
的 API 在 TF 中文档支持并不好,这也是本项目诞生的原因。
本项目展示了如何使用 seq2seq
中的 API 以及 AttentionWrapper
,并从源码的角度来澄清一些重要概念。
This project is a TensorFlow Tutorial for beginners.
It introduces basic usage in TF, very little CNN, and emphasizes seq2seq models.
The seq2seq
APIs in TF are not well documented, and this is why I made this project.
It shows how to use seq2seq
APIs and AttentionWrapper
, and dive into the source code to make key concepts clear.
- ABC: linear_regression.py, mnist_cnn.py
- RNN: char_language_model.py
- Debugging and Tensorboard: mnist_with_summaries.py, mnist_debug.py
- seq2seq: seq2seq.py, att_seq2seq_delete_and_copy.py