Open qiushan233 opened 8 months ago
I'm confused by the Class Conv2dSubampling in convolution.py.What does the second return output_lengths mean?
for example,when input_length==55,the output_length is wrong
Yes, I have gotten that problem when I used output length of conformer for RNNT loss in TorchAudio
I'm confused by the Class Conv2dSubampling in convolution.py.What does the second return output_lengths mean?