ValueError: Unknown loss function:<lambda>




Confirming, is this for the Tensorflow repo: I imagine this was a mis-post, but it would be ideally filed under a Tensorflow discuss forum.

In any event, it looks like an error in the notebook, and it appears you need to specify ctc loss function when you load the model. I haven’t tested, but the issues below indicate something like this would be the right code (you may also want to file an issue under that repo.)

model = load_model('...', custom_objects={'loss_max': loss_out})