Hi,
I’d be very interested to view the output matrix of the pretrained model before beamsearch. Is this possible? What parts of the code do I need to install in order to do this (without training).
The logits output is called “logits” in the graph.
See if this is what you’re looking for (a script that shows n most probable characters for each audio sample): https://discourse.mozilla.org/t/can-i-use-pre-trained-model-with-deepspeech-py/23577/8
1 Like
Thanks! That’s exactly what I was looking for.