My log:
rank: 6
rank: 2
rank: 4
rank: 7
rank: 0
rank: 1
rank: 3
rank: 5
Device: cuda:4
Device: cuda:6
Device: cuda:3
Device: cuda:2
Device: cuda:1
Device: cuda:0
Device: cuda:7
Device: cuda:5
Conformer CTC Small
Model Parameters : 1865104
Parallelize model on 8 GPUs
Loading training dataset : LibriSpeech train
LibriSpeech dataset filtering
Audio maximum length : 256000 / Label sequence maximum length : 256000
Loaded : 264723 samples / 8272 batches
Loading evaluation dataset : LibriSpeech dev-clean
Loaded : 2703 samples / 43 batches
Loading evaluation dataset : LibriSpeech dev-other
Loaded : 2864 samples / 45 batches
Epoch 1/40
dev-clean wer : 69.51% - loss : 94.5979
And didn't print "dev-other wer :"
My log:
rank: 6
rank: 2
rank: 4
rank: 7
rank: 0
rank: 1
rank: 3
rank: 5
Device: cuda:4
Device: cuda:6
Device: cuda:3
Device: cuda:2
Device: cuda:1
Device: cuda:0
Device: cuda:7
Device: cuda:5
Conformer CTC Small
Model Parameters : 1865104
Parallelize model on 8 GPUs
Loading training dataset : LibriSpeech train
LibriSpeech dataset filtering
Audio maximum length : 256000 / Label sequence maximum length : 256000
Loaded : 264723 samples / 8272 batches
Loading evaluation dataset : LibriSpeech dev-clean
Loaded : 2703 samples / 43 batches
Loading evaluation dataset : LibriSpeech dev-other
Loaded : 2864 samples / 45 batches
Epoch 1/40
dev-clean wer : 69.51% - loss : 94.5979
And didn't print "dev-other wer :"