November 2019
Intermediate to advanced
296 pages
7h 52m
English
As we mentioned previously, TensorFlow.js is capable of reusing any existing models that have been trained by TensorFlow. To do so, we need to make tfjs-converter runnable in our environment. tfjs-converter is a command-line tool that we can use to convert the TensorFlow model. tfjs-converter currently supports the following file formats:
| File format | Option |
|
Keras HDF5 |
keras |
| Keras SavedModel | keras_saved_model |
| TensorFlow Hub | tf_hub |
| TensorFlow SavedModel |
tf_saved_model |
Since tfjs-converter is a command-line tool that runs in Python, you need to prepare the Python runtime first. tfjs-converter internally uses TensorFlow and Keras. To avoid conflicts with the local installation, it is highly recommended that ...
Read now
Unlock full access