Hi,
Wanted to understand the role of LM during the training. Does it affect in model learning at all or it is only used for WER report generation. If I use the release version of deepspeech LM for custom dataset training what will be the effect on the weights and biases of the model ?
Thanks