Closed pankajkumar closed 6 years ago
@pankajkumar Can you let me know what data you are training on, the sampling frequency of audio files and the command you have used to run the training
I am training on LibriSpeech ASR corpus, here is the command:-
luajit train.lua --train -rundir ~/experiments -runname hello_librispeech -archdir ~/pb/wav2letter/wav2letter/arch/ -arch librispeech-glu-highdropout -lr 0.1 -lrcrit 0.0005 -gpu 1 -linseg 1 -linlr 0 -linlrcrit 0.005 -onorm target -nthread 0 -dictdir downloads/librispeech-proc -datadir downloads/librispeech-proc -train dev-clean -valid test-clean -sqnorm -mfsc -melfloor 1 -surround "|" -replabel 2 -progress -wnorm -normclamp 0.2 -momentum 0.9 -weightdecay 1e-05
Fixed
https://github.com/facebookresearch/wav2letter/blob/153ac12304a51199249d797d365c25f13b5426ed/wav2letter/runtime/netutils.lua#L131
It should be :
return nn.GatedLinearUnit(2):cuda()
Otherwise it returns the error:-