I am having problems with the use of deepspeech over tensorflow 2.
When training the model it fails because of incompatibilities with tensorflow, and tensorflow lower versions are depricated.
Is there someone that has recently train a model from scratch using this?
Any clue will help, thanks!
As documented, you need to use TensorFlow r1.15.4. It is what is being sourced by the package when you install it. Search the forum and GitHub, the amount of work required to move to TensorFlow r2.x for training is documented. And huge.
1 Like