TensorFlow 2 Network Reader

Reads a TensorFlow 2 model from a file or folder. The file/folder must be:

  • A HDF5 (.h5) file containing a (TensorFlow) Keras Model.
  • A SavedModel file containing a TensorFlow Keras Model.
  • A ZIP file of the directory of a SavedModel containing a TensorFlow Keras Model.

The KNIME Deep Learning - TensorFlow 2 Integration is developed by KNIME and uses the TensorFlow 2 library. The KNIME Deep Learning - TensorFlow 2 Integration is not endorsed by or otherwise affiliated with Google. TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc.

This node can access a variety of different file systems. More information about file handling in KNIME can be found in the official File Handling Guide.

Options

Read from
Select a file system which stores the model you want to read. There are four default file system options to choose from:
  • Local File System: Allows you to select a file/folder from your local system.
  • Mountpoint: Allows you to read from a mountpoint. When selected, a new drop-down menu appears to choose the mountpoint. Unconnected mountpoints are greyed out but can still be selected (note that browsing is disabled in this case). Go to the KNIME Explorer and connect to the mountpoint to enable browsing. A mountpoint is displayed in red if it was previously selected but is no longer available. You won't be able to save the dialog as long as you don't select a valid i.e. known mountpoint.
  • Relative to: Allows you to choose whether to resolve the path relative to the current mountpoint, current workflow or the current workflow's data area. When selected, a new drop-down menu appears to choose which of the two options to use.
  • Custom/KNIME URL: Allows to specify a URL (e.g. file://, http:// or knime:// protocol). When selected, a spinner appears that allows you to specify the desired connection and read timeout in milliseconds. Browsing is disabled for this option.
To read from other file systems, click on ... in the bottom left corner of the node icon followed by Add File System Connection port. Afterwards, connect the desired file system connector node to the newly added input port. The file system connection will then be shown in the drop-down menu. It is greyed out if the file system is not connected in which case you have to (re)execute the connector node first. Note: The default file systems listed above can't be selected if a file system is provided via the input port.
Mode
Select whether the network you want to read is stored as a file (HDF5 or ZIP) or as a folder (SavedModel).
File/Folder/URL
Enter a URL when reading from Custom/KNIME URL, otherwise enter a path to a file. The required syntax of a path depends on the chosen file system, such as "C:\path\to\file" (Local File System on Windows) or "/path/to/file" (Local File System on Linux/MacOS and Mountpoint). For file systems connected via input port, the node description of the respective connector node describes the required path format. You can also choose a previously selected file from the drop-down list, or select a location from the "Browse..." dialog. Note that browsing is disabled in some cases:
  • Custom/KNIME URL:Browsing is never enabled.
  • Mountpoint: Browsing is disabled if the selected mountpoint isn't connected. Go to the KNIME Explorer and connect to the mountpoint to enable browsing.
  • File systems provided via input port: Browsing is disabled if the connector node hasn't been executed since the workflow has been opened. (Re)execute the connector node to enable browsing.
The location can be exposed as or automatically set via a path flow variable.

Input Ports

Icon…
The file system connection.

Output Ports

Icon
The TensorFlow 2 network.

Popular Predecessors

Popular Successors

Views

This node has no views

Workflows

Links

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.

Installation

To use this node in KNIME, install the extension KNIME Deep Learning - TensorFlow 2 Integration from the below update site following our NodePit Product and Node Installation Guide:

v5.4

A zipped version of the software site can be downloaded here.

Plugin provider: KNIME AG, Zurich, Switzerland
Plugin version: 5.4.0.v202407311314
On NodePit since: 2024-12-06
Last update: 2024-12-17
KNIME versions: Since v4.2

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!