Oliver Adams
Oliver Adams
There's a few things to say here. > Did I understand correctly that | could be used as word boundary? I haven't experimented using it that way and am quite...
Interesting. Given the time it takes to train these models and the poor test scores (86% LER is usually completely wrong), a good strategy now might be to break the...
The label inventory should also be written.
The `preprocessing.pangloss` should create `Utterance` objects like `preprocessing.elan`. The Na preprocessing should work with these `Utterance` objects.
The example code should use hyperparameters that let the models be trained on people's personal computers. Increasing batch size in training might make training run faster but would use more...
No worries, I'll do this.
Just one bug really. Everything will still work but the model will think two symbols that are equivalent are distinct and as a result there'll be a tad more data...
The code I committed some time back which broke the interface: Just a wrapper to the decode function that then writes the returned text to a file. When I add...
I think this is definitely the way to go. As part of the Elpis-ESPnet integration it'll be good to prepare a multilingual model that can be fine-tuned to target languages....