I'm glad you guys did a great job. When I run bash exp.sh, it doesn't start training. It reported the following error.

```
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:231: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
input_ids = torch.tensor(sent[:,0,:].clone().detach(),dtype=torch.long).cuda()
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:232: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
input_mask = torch.tensor(sent[:,1,:].clone().detach(),dtype=torch.long).cuda()
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:233: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
segment_ids = torch.tensor(sent[:,2,:].clone().detach(),dtype=torch.long).cuda()
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:235: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
decoder_input_ids = torch.tensor(title[:,0,:].clone().detach(),dtype=torch.long).cuda()
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:236: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
decoder_input_mask = torch.tensor(title[:,1,:].clone().detach(),dtype=torch.long).cuda()
/root/autodl-tmp/sum/Multimodal-Summarization/oscar.py:237: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
decoder_segment_ids = torch.tensor(title[:,2,:].clone().detach(),dtype=torch.long).cuda()
[unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused193] [unused215] [unused193] [unused193] [unused193] [unused193] [unused215] [unused193] [unused193] [unused193]
```
Then the program stopped running.