hub_text_embedding_column: Module to construct a dense representation from a text...

Description Usage Arguments

View source: R/columns.R

Description

This feature column can be used on an input feature whose values are strings of arbitrary size.

Usage

1
hub_text_embedding_column(key, module_spec, trainable = FALSE)

Arguments

key

A string or [feature_column](https://tensorflow.rstudio.com/tfestimators/articles/feature_columns.html) identifying the text feature.

module_spec

A string handle or a _ModuleSpec identifying the module.

trainable

Whether or not the Module is trainable. 'FALSE' by default, meaning the pre-trained weights are frozen. This is different from the ordinary 'tf.feature_column.embedding_column()', but that one is intended for training from scratch.


tfhub documentation built on Dec. 19, 2021, 9:07 a.m.