win10+pytorch1.0 +py3.68+cuda10运行 RuntimeError: input must have 3 dimensions, got 4

[QQ2737499951]

"D:\Program Files\Python368\python.exe" B:/PyTorch/CaptchaRecognition/main.py
val_count= 1250
Loading data... 100_60_4_captcha.npz
Loading dictionary... [[ 0 3 3 4 12 11]
[ 0 4 7 7 1 12]
[ 0 8 5 4 12 11]
...
[ 0 8 3 9 6 12]
[ 0 4 2 6 12 11]
[ 0 8 9 8 8 12]]
vocab {'id2token': {1: '0', 2: '1', 3: '2', 4: '3', 5: '4', 6: '5', 7: '6', 8: '7', 9: '8', 10: '9', 11: ' ', 0: '^', 12: '$'}, 'token2id': {'0': 1, '1': 2, '2': 3, '3': 4, '4': 5, '5': 6, '6': 7, '7': 8, '8': 9, '9': 10, ' ': 11, '^': 0, '$': 12}}
Convert to tensor... total_size= None test_size= 8192
Build dataset...分离train+test
Build dataloader...载入train+test
data ready!
Traceback (most recent call last):
File "B:/PyTorch/CaptchaRecognition/main.py", line 86, in
criterion,CLIP,use_cuda=USE_CUDA)
File "B:\PyTorch\CaptchaRecognition\model.py", line 230, in train
encoder_outputs,encoder_hidden = encoder(inputs,init_hidden)
File "D:\Program Files\Python368\lib\site-packages\torch\nn\modules\module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "B:\PyTorch\CaptchaRecognition\model.py", line 47, in forward
out,hidden = self.gru(out,h0)
File "D:\Program Files\Python368\lib\site-packages\torch\nn\modules\module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "D:\Program Files\Python368\lib\site-packages\torch\nn\modules\rnn.py", line 207, in forward
self.check_forward_args(input, hx, batch_sizes)
File "D:\Program Files\Python368\lib\site-packages\torch\nn\modules\rnn.py", line 175, in check_forward_args
self.check_input(input, batch_sizes)
File "D:\Program Files\Python368\lib\site-packages\torch\nn\modules\rnn.py", line 149, in check_input
expected_input_dim, input.dim()))
### RuntimeError: input must have 3 dimensions, got 4

这个源码之前在0.4版我运行过没问题的，前几天刚升级到pytorch 1.0 、pytorch 1.1都不行，谢谢楼主！