pretrain运行交流

[ZuzenGao]

Traceback (most recent call last):
File "pretrain.py", line 307, in
main(i, modality=m)
File "pretrain.py", line 294, in main
train_results = train(modality, model, device, dataLoader['train'], optimizer, loss_fn, epoch, metrics)
File "pretrain.py", line 93, in train
output = model(inputs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/gzc/KuDA-main/models/Encoder_KIAdapter.py", line 227, in forward
uni_hidden = self.encoder(uni_fea, key_padding_mask)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/gzc/KuDA-main/models/Encoder_KIAdapter.py", line 138, in forward
tf_last_hidden_state, tf_hidden_state_list = self.tfencoder(inputs, src_key_padding_mask=key_padding_mask)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/gzc/KuDA-main/models/Encoder_KIAdapter.py", line 109, in forward
output, hidden_list = self.tfencoder(src, mask=None, src_key_padding_mask=src_key_padding_mask)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/transformer.py", line 266, in forward
output = mod(output, src_mask=mask, src_key_padding_mask=src_key_padding_mask)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/transformer.py", line 780, in forward
x = self.norm1(x + self._sa_block(x, src_mask, src_key_padding_mask, is_causal=is_causal))
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/transformer.py", line 788, in _sa_block
x = self.self_attn(x, x, x,
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
return forward_call(*args, **kwargs)
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/modules/activation.py", line 1275, in forward
attn_output, attn_output_weights = F.multi_head_attention_forward(
File "/data1/ENVS/Msa/lib/python3.8/site-packages/torch/nn/functional.py", line 5499, in multi_head_attention_forward
assert key_padding_mask.shape == (bsz, src_len),
AssertionError: expecting key_padding_mask shape of (32, 231), but got torch.Size([32, 230])

作者你好，我在用mosi数据集运行pretrain部分出现如上报错，我已经按照readme中对于transformerencoder进行了修改，但还是出现上述报错，恳请解答！感谢作者开源，一篇很好的论文！