应用其他数据集的问题

[buyongbaoqian]

罗老师您好，我在使用您的模型在自己准备的数据集上训练的时候出现了矩阵不能相乘的问题，该数据集的json文件已经按照dataset_coco.json文件的格式准备好，在训练时报错，错误代码如下，请问可以提示一下该怎么解决吗

$ python tools/train.py --cfg configs/fc.yml --id fc cider or coco-caption missing DataLoader loading json file: data/rsicdtalk.json vocab size is 1256 DataLoader loading h5 file: data/rsicdtalk_fc data/rsicdtalk_att data/cocotalk_box data/rsicdtalk_label.h5 max sequence length in data is 16 read 10921 image features assigned 8734 images to split train assigned 1094 images to split val assigned 1093 images to split test d:\imagecaptioning.pytorch-master1\captioning\data\dataloader.py:290: RuntimeWarning: Mean of empty slice. fc_feat = att_feat.mean(0) Read data: 0.00298309326171875 Save ckpt on exception ... model saved to ./log_fc\model.pth Save ckpt done. Traceback (most recent call last): File "D:\ImageCaptioning.pytorch-master1\tools\train.py", line 185, in train model_out = dp_lw_model(fc_feats, att_feats, labels, masks, att_masks, data['gts'], torch.arange(0, len(data['gts'])), sc_flag, struc_flag, drop_worst_flag) File "D:\Anaconda\envs\pytorch\lib\site-packages\torch\nn\modules\module.py", line 1194, in _call_impl return forward_call(*input, **kwargs) File "D:\Anaconda\envs\pytorch\lib\site-packages\torch\nn\parallel\data_parallel.py", line 169, in forward File "D:\Anaconda\envs\pytorch\lib\site-packages\torch\nn\modules\linear.py", line 114, in forward return F.linear(input, self.weight, self.bias) RuntimeError: mat1 and mat2 shapes cannot be multiplied (10x0 and 2048x512)

[ruotianluo]

你是在一个gpu还是多个gpu上train？

[buyongbaoqian]

在一个GPU训练