Daniel Ecer

I am interested in that, what is your 'Offline Voice Recognizer'? Do you already have something that you want to adapt? Google Voice Recognizer should have the ability to work...

Thank you for that. I tried that today but I think it's failing before it even gets to try to load the image (#503).

I'm getting the same error with the official image. It doesn't seem to be getting to reading the referenced busybox image as it doesn't make a difference whether it's there...

> There is two probable reasons of "Permissions denied" in your case: either apparmor disables that (or selinux, but apparmor is more probable on ubuntu) or docker has some limitations...

> Hello @de-code, do you know if there are any updates on this topic ? I have not heard anything relating my PR (it reached it's anniversary). I was considering...

As a library it would really be good to be able to use a custom model without changing the existing code. What do you think about registering a model via...

The [TF 2.0 RNN](https://www.tensorflow.org/api_docs/python/tf/keras/layers/RNN) has a `stateful` flag which might allow one to pass in the input sequence in consecutive windows (without overlap). Although we really only would want it...

I have now implemented something in my version: https://github.com/elifesciences/sciencebeam-trainer-delft/pull/179 I tried the `stateful` version (which is still an option) but the training time increases by roughly eight times. So I...

Thank you for explaining that. I didn't realise that it was the first two tokens. In any case I am now training a segmentation model without word embedding at all...

One example, [025544v1](https://www.biorxiv.org/content/10.1101/025544v1?versioned=true), where the PDF is currently resulting in many tokens (`14798`). Those tokens seem to come from figures. But the model (trained on at most `3000`) still seem...