August 2019
Intermediate to advanced
202 pages
5h 9m
English
As we mentioned previously, you can directly create a tf.data.Dataset using all the different file formats. We also explained the recommended way to create TFRecords. However, if you want to create datasets directly, without going through TFRecords, that is also possible. Here are some examples of creating a tf.data.Dataset directly from raw data:
You can use the td.data.experimental.make_csv_dataset(...) API to find the .csv file. You can define the columns that are available in the .csv file, along with batch_size, as follows. The complete code can be found at https://github.com/PacktPublishing/What-s-New-in-TensorFlow-2.0/blob/master/Chapter03/datasets/create_dataset_from_csv.ipynb ...
Read now
Unlock full access