When iterating to a fixed value, loss becomes Nan

[leon-cjf]

Normal training with train (aishell dataset), when iterating to a fixed value, loss becomes Nan.If I specify shuffle as false, I still iterate to a specific value and loss to become Nan

[leon-cjf]

i'm looking forword to your reply

[leon-cjf]

appears /home/leon/anaconda3/envs/leon/lib/python3.10/site-packages/librosa/util/decorators.py:88: UserWarning: n_fft=320 is too small for input signal of length=0 return f(*args, **kwargs)