sage.py是不是有点问题？

[guanfaqian]

我train目标模型的时候按默认的gat以及gin没问题。
argparser.add_argument('--target-model', type=str, default='sage')改成sage

10629 3543 3544
Traceback (most recent call last):
File "/home/gfq/code/privacy/GNNStealing/my_code/train_target_model.py", line 82, in
target_model = run_sage_target(args, device, data)
File "/home/gfq/code/privacy/GNNStealing/my_code/src/sage.py", line 195, in run_sage_target
batch_pred, embs = model(blocks, batch_inputs)
File "/home/gfq/anaconda3/envs/GNNStealing/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/gfq/code/privacy/GNNStealing/my_code/src/sage.py", line 83, in forward
h = self.layers[i](blocks[i], h)
File "/home/gfq/anaconda3/envs/GNNStealing/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/home/gfq/anaconda3/envs/GNNStealing/lib/python3.8/site-packages/dgl/nn/pytorch/conv/sageconv.py", line 241, in forward
graph.dstdata['h'] = graph.srcdata['h']
File "/home/gfq/anaconda3/envs/GNNStealing/lib/python3.8/site-packages/dgl/view.py", line 81, in setitem
self._graph._set_n_repr(self._ntid, self._nodes, {key : val})
File "/home/gfq/anaconda3/envs/GNNStealing/lib/python3.8/site-packages/dgl/heterograph.py", line 3992, in _set_n_repr
raise DGLError('Expect number of features to match number of nodes (len(u)).'
dgl._ffi.base.DGLError: Expect number of features to match number of nodes (len(u)). Got 4427 and 1930 instead.

进程已结束，退出代码为 1

就报错了。

[guanfaqian]

好像是dgl版本问题，我按您的0.7.2好像可以。不过需要把按https://github.com/dmlc/dgl/issues/3791更改dgl里面的
num_data = len(dataset) => num_data = len(dataset.nodes())