zing1116
commented
3 years ago 将node_neigh类型修改成long即可： nodeneigh = torch.tensor( [neighbors[i] for in range(edge_type_count)] dtype=torch.long).to(device)
Closed zing1116 closed 3 years ago
zing1116
commented
3 years ago 将node_neigh类型修改成long即可： nodeneigh = torch.tensor( [neighbors[i] for in range(edge_type_count)] dtype=torch.long).to(device)
您好，打扰了。我在运行pytorch版本的GATNE代码时，总是在训练第二个epoch时出现以下错误： epoch 0: 100%|| 7066/7066 [01:29<00:00, 78.82it/s] Traceback (most recent call last): File "src/main_pytorch.py", line 315, in
average_auc, average_f1, average_pr = train_model(training_data_by_type, feature_dic)
File "src/main_pytorch.py", line 250, in train_model
node_emb = model(train_inputs, train_types, node_neigh)
File "D:\software\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "src/main_pytorch.py", line 85, in forward
node_embed_neighbors = self.node_type_embeddings[node_neigh]
IndexError: tensors used as indices must be long, byte or bool tensors
请问这样的错误应该如何纠正？
应该是node_neigh的问题： nodeneigh = torch.tensor( [neighbors[i] for in range(edge_type_count)] ).to(device)
但是我不太清楚node_neigh的是如何生成的，维数如何，可以麻烦您指点一下吗？谢谢！