python - 为什么 validation_freq 不适用于 Keras 模型拟合？

Question

根据Keras docs , fit 采用 validation_freq 参数:

validation_freq: Only relevant if validation data is provided. Integer or list/tuple/set. If an integer, specifies how many training epochs to run before a new validation run is performed, e.g. validation_freq=2 runs validation every 2 epochs. If a list, tuple, or set, specifies the epochs on which to run validation, e.g. validation_freq=[1, 2, 10] runs validation at the end of the 1st, 2nd, and 10th epochs.

result = model.fit( X_train, Y_train, epochs=2000, verbose=1, validation_data=(X_test,Y_test), validation_freq=10) # , validation_split=0.2

这引发了:

File "/Users/george/anaconda3/lib/python3.6/site-packages/keras/engine/training.py", line 942, in fit
raise TypeError('Unrecognized keyword arguments: ' + str(kwargs))
TypeError: Unrecognized keyword arguments: {'validation_freq': 10}

使用 Keras2.1.6-tf。从那以后是否添加了此参数？

如果是这样，如何为 Anaconda 更新 Keras？我试过:

> conda update keras
Collecting package metadata: done
Solving environment: done

# All requested packages already installed.

Answer 1

添加validation_freq 的提交是在最新版本2.2.4 之后添加的。参见 https://github.com/keras-team/keras/commit/a6c8042121371b5873773ca767f28cdf5689d5e4 ，这是在去年 10 月发布的最新版本之后的 28 天前提交的。

我通过从 keras 的 git 存储库安装来解决这个问题:

pip uninstall keras
pip install git+git://github.com/keras-team/keras.git

虽然您使用的是 conda，但 pip 应该仍然可以安装包。您可能需要摆弄 pip 安装到正确的 python，即 pip3 install 或在 Windows python -m pip install 上。最坏的情况就是最坏的情况use conda to install from the source on github.