Short Description
Enable the tensor flow API to use trained deep learning models.
With this plugin you can choose which version of the library should be loaded.
Documentation
Take control of the TensorFlow library with this plugin.
TensorFlow Manager lets you choose the API version depending on your needs. It will take care of downloading the libraries (if needed) and load them on runtime.
To select the TF library version use the TensorFlowVersionSelector plugin. Only compatible versions with your operating system will be displayed. Please make sure to use the library version compatible with the trained model you want to use. Make sure the code used to load the model is compatible with the library version as well.
If you want to make sure the library is loaded from the code, use
TensorFlowManager.loadLibrary(libraryVersion, forceDownload)
Supported versions
Currently TensorFlow Manager provides the following TF library versions:
- Library version: 1.15.0, TF API version: 1.15.0
- OS: Win64x, maxos64x, linux64x
- Modes: cpu, gpu (only win and linux)
- Library version: 0.2.0, TF API version: 2.3.1
- OS: Win64x, maxos64x, linux64x
- Modes: cpu, mkl, gpu (only win and linux), mkl-gpu (only win and linux)
- Library version: 0.3.0, TF API version: 2.4.1
- OS: Win64x, maxos64x, linux64x
- Modes: cpu, mkl, gpu (only win and linux), mkl-gpu (only linux)
- Library version: 0.3.1, TF API version: 2.4.1
- OS: Win64x, maxos64x, linux64x
- Modes: cpu, mkl, gpu (only win and linux), mkl-gpu (only linux)