I saw some tensorflow project contains hyperparameter configuration like:
tf.flags.DEFINE_integer("embedding_dim", 128, "Dimensionality of character embedding (default: 128)")
I think it is better to add corresponding document for using this feature.
Our policy so far has been only to document the features that we intend to support in TensorFlow long-term. Currently we use python-gflags
as the implementation for tf.flags
, but this is subject to change in future. In the meantime, the documentation for that library is the best source of usage information.
I'm not sure if TensorFlow will use python-gflags
in long-term or not, it would be better to add the link to refer to python-glags
's documentation. Anyway, it seems that project has little document, either :(
Now Is there any documentation for hyperparameter configuration like?
Most helpful comment
I'm not sure if TensorFlow will use
python-gflags
in long-term or not, it would be better to add the link to refer topython-glags
's documentation. Anyway, it seems that project has little document, either :(