tf.contrib
tf.contrib包下是一些非官方的文件,代码的更改和删除不会被通知。
这个包下提供的特性在将来可能加入到tensorflow核心库中,但是暂时接口可能还会改变,或者仍然需要测试,
tf.contrib.slim
那么既然官方没有给出文档,我们就来阅读源代码,来了解这个包的功能。
主要就是提供了更高级的封装,方便快速建模。
介绍
TF-Slim 是一个轻量级的包,它用于定义,训练和验证模型。tf-slim提供的组件可以直接与其他tensorflow原生的模型混合使用。
TF-Slim is a lightweight library for defining, training and evaluating complex models in TensorFlow. Components of tf-slim can be freely mixed with native tensorflow, as well as other frameworks, such as tf.contrib.learn.
arg_scope
提供了新的scope name:arg_scope
*provides a new scope named arg_scope
that allows a user to define default
arguments for specific operations within that scope.*
data
提供了一些操作。
contains TF-slim’s dataset definition, data providers,parallel_reader, and decoding utilities.
evaluation
contains routines for evaluating models.
layers
包含了一下定义好的layer,用于建立模型。
contains high level layers for building models using tensorflow.
learning
contains routines for training models.
losses
contains commonly used loss functions.
metrics
contains popular evaluation metrics.
nets
contains popular network definitions such as
queues
provides a context manager for easily and safely starting and closing
QueueRunners.
regularizers
contains weight regularizers.
variables
provides convenience wrappers for variable creation and manipulation.
tf.contrib.slim.python.nets
nets里面定义了一些popular的模型。例如inception_v3