训练syncnet报错：RuntimeError: DataLoader worker (pid 899437) is killed by signal: Segmentation fault. ，训练到第290000结束后报错：，请问有人遇到过这个问题吗？

[Ko24Bewjw]

Traceback (most recent call last): File "color_syncnet_train.py", line 279, in <module> nepochs=hparams.nepochs) File "color_syncnet_train.py", line 163, in train optimizer.step() File "/home/user/anaconda3/envs/wav2lip/lib/python3.6/site-packages/torch/optim/optimizer.py", line 88, in wrapper return func(*args, *kwargs) File "/home/user/anaconda3/envs/wav2lip/lib/python3.6/site-packages/torch/autograd/grad_mode.py", line 28, in decorate_context return func(args, **kwargs) File "/home/user/anaconda3/envs/wav2lip/lib/python3.6/site-packages/torch/optim/adam.py", line 144, in step eps=group['eps']) File "/home/user/anaconda3/envs/wav2lip/lib/python3.6/site-packages/torch/optim/_functional.py", line 87, in adam exp_avgsq.mul(beta2).addcmul_(grad, grad.conj(), value=1 - beta2) File "/home/user/anaconda3/envs/wav2lip/lib/python3.6/site-packages/torch/utils/data/_utils/signal_handling.py", line 66, in handler _error_if_any_worker_fails() RuntimeError: DataLoader worker (pid 899437) is killed by signal: Segmentation fault.

[chunyu-li]

Change your PyTorch version, I met same problem with 1.13.1. I recommend 2.0.1