常用
####


一代框架:

+----------------------+------------+-----------+---------------+
|                      | Caffe      | Torch     | Theano        |
+======================+============+===========+===============+
| Language             | C++,Python | Lua       | Python        |
+----------------------+------------+-----------+---------------+
| Pretrained           | Yes ++     | Yes ++    | Yes (Lasagne) |
+----------------------+------------+-----------+---------------+
| Multi-GPU: Data      | Yes        | Yes       | Yes           |
+----------------------+------------+-----------+---------------+
| Multi-GPU: Model     | No         | Yes       | Experimental  |
+----------------------+------------+-----------+---------------+
| Readable source code | Yes (C++)  | Yes (Lua) | No            |
+----------------------+------------+-----------+---------------+
| Good at RNN          | No         | Mediocre  | Yes           |
+----------------------+------------+-----------+---------------+
| Comany               | Berkeley   | NYU       | Montreal      |
+----------------------+------------+-----------+---------------+


二代框架:

+----------------------+------------+-----------+----------+--------+--------+
|                      | TensorFlow | CNTK      | pyTorch  | MXNet  | Paddle |
+======================+============+===========+==========+========+========+
| Language             | Python     |           | Python   |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Pretrained           | Inception  |           |          |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Multi-GPU: Data      | Yes        |           |          |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Multi-GPU: Model     | Yes (best) |           |          |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Readable source code | No         |           |          |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Good at RNN          | Yes (best) |           |          |        |        |
+----------------------+------------+-----------+----------+--------+--------+
| Comany               | Google     | Microsoft | Facebook | Amazon | Baidu  |
+----------------------+------------+-----------+----------+--------+--------+


第三代，即命令式深度学习框架，是由用类似NumPy的语法来定义模型的 `Chainer <https://github.com/chainer/chainer>`_ 所开创的。这样的思想后来被 PyTorch和MXNet的Gluon API 采用，后者也正是本书用来教学深度学习的工具。




+------------+-----------------+-------------------+------------------+----------+------+-----------+----------+
| 库名称     | 开发语言        | 支持接口          | 安装难度(ubuntu) | 文档风格 | 示例 | 支持模型  | 上手难易 |
+============+=================+===================+==================+==========+======+===========+==========+
| Caffe      | c++/cuda        | c++/python/matlab | ###              | #        | ###  | CNN       | ##       |
+------------+-----------------+-------------------+------------------+----------+------+-----------+----------+
| MXNet      | c++/cuda        | python/R/Julia    | ##               | ###      | ##   | CNN/RNN   | #        |
+------------+-----------------+-------------------+------------------+----------+------+-----------+----------+
| TensorFlow | c++/cuda/python | c++/python        | #                | ##       | #    | CNN/RNN/… | ###      |
+------------+-----------------+-------------------+------------------+----------+------+-----------+----------+

* 安装难度: (简单) –> ##(复杂)
* 文档风格: (一般) –> ##(好看、全面)
* 示例: (给的少) –> ##(给的多、全)
* 上手难易: (易) –> ##(难)