Oliver Adams

Hi @nikopartanen, sorry about the delay > If I then train the model i.e. with parameter num_train = 1024, how do I know which part of the train_prefixes.txt it has...

Hi! Sorry, I missed the urgency of this. `model.eval()` should do what you want. `Model.train()` will train and get scores on the validation set, while `Model.eval()` runs it on the...

This is a good idea and should be implemented. It would be straight-forward to stitch the overlapping windows together using some edit-distance matching to create one contiguous transcription. Another idea...

> Is it really straightforward to stitch when there is overlap between successive audio windows? Edit-distance matching will allow for stitching, but any mismatches in the overlapping portion of the...

Hi Alexis, thanks for raising this. I had noted the extreme infrequency of [˩˧] and [˧˩]. It's good to now understand that it was a phonetic distinction vs phonemic. >...

Just for the record, I completely agree. This is how it should be done and will be done. I guess one remaining question is what the right interface is. Linguists...

I don't have a format preference. But whatever we do here it's important that the output file gives a complete picture of the experiment so it can be reproduced. This...

So just regarding bumpversion. This can probably be re-evaluated later, since for now there's just two places where it needs to be changed.

I had wondered about this, but just assumed that the tone group runs on to the next utterance. This is good to hear, I'll add such markers. These results won't...

Do you still need an answer to this, or did the phone chat on the weekend let you clarify your planned direction with this?