Hi Josh,
Thanks for sharing the notebooks! In the "07_structured_data", it states that "Using TensorFlow 1.4, the below can be written using regular Python code to parse the CSV file, via the Datasets. from_generator() method. This improves producivity a lot - it means you can use Python to read, parse, and apply whatever logic you wish to your input data - then you can take advantage of the reusable pieces of the Datasets API (e.g., batch, shuffle, repeat, etc) - as well as the optional performance tuning (e.g., prefetch, parallel process, etc).
In combination with Estimators, this means you can train and tune deep models at scale on data of almost any size, entirely using a high-level API. I'll update this notebook after v1.4 is released with an example. It's neat."
Now that v1.4 is released. Would you be updating this part of the notebook? Is so, when would you expect it complete? Thanks.
Hi Josh,
Thanks for sharing the notebooks! In the "07_structured_data", it states that "Using TensorFlow 1.4, the below can be written using regular Python code to parse the CSV file, via the Datasets. from_generator() method. This improves producivity a lot - it means you can use Python to read, parse, and apply whatever logic you wish to your input data - then you can take advantage of the reusable pieces of the Datasets API (e.g., batch, shuffle, repeat, etc) - as well as the optional performance tuning (e.g., prefetch, parallel process, etc).
In combination with Estimators, this means you can train and tune deep models at scale on data of almost any size, entirely using a high-level API. I'll update this notebook after v1.4 is released with an example. It's neat."
Now that v1.4 is released. Would you be updating this part of the notebook? Is so, when would you expect it complete? Thanks.