RuntimeError: index 51 is out of bounds for dimension 0 with size 32

[joshuajennysibbu]

RuntimeError                              Traceback (most recent call last)
/tmp/ipykernel_7390/534676331.py in <module>
      1 for i in loader:
----> 2     output=model(i['x_s'],i['x_t'],i['edge_index'])

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
   1108         if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109                 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110             return forward_call(*input, **kwargs)
   1111         # Do not call functions when jit is used
   1112         full_backward_hooks, non_full_backward_hooks = [], []

/tmp/ipykernel_7390/69486972.py in forward(self, source, target, edge_index)
     24 
     25         # source=torch.randn(300,3)
---> 26         source_conv = self.conv1(source,edge_index_)
     27         target_conv = self.conv1(target,edge_index_)
     28 

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
   1108         if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109                 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110             return forward_call(*input, **kwargs)
   1111         # Do not call functions when jit is used
   1112         full_backward_hooks, non_full_backward_hooks = [], []

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_geometric/nn/conv/sage_conv.py in forward(self, x, edge_index, size)
     78 
     79         # propagate_type: (x: OptPairTensor)
---> 80         out = self.propagate(edge_index, x=x, size=size)
     81         out = self.lin_l(out)
     82 

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_geometric/nn/conv/message_passing.py in propagate(self, edge_index, size, **kwargs)
    342                     if res is not None:
    343                         aggr_kwargs = res[0] if isinstance(res, tuple) else res
--> 344                 out = self.aggregate(out, **aggr_kwargs)
    345                 for hook in self._aggregate_forward_hooks.values():
    346                     res = hook(self, (aggr_kwargs, ), out)

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_geometric/nn/conv/message_passing.py in aggregate(self, inputs, index, ptr, dim_size)
    425             return segment_csr(inputs, ptr, reduce=self.aggr)
    426         else:
--> 427             return scatter(inputs, index, dim=self.node_dim, dim_size=dim_size,
    428                            reduce=self.aggr)
    429 

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_scatter/scatter.py in scatter(src, index, dim, out, dim_size, reduce)
    154         return scatter_mul(src, index, dim, out, dim_size)
    155     elif reduce == 'mean':
--> 156         return scatter_mean(src, index, dim, out, dim_size)
    157     elif reduce == 'min':
    158         return scatter_min(src, index, dim, out, dim_size)[0]

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_scatter/scatter.py in scatter_mean(src, index, dim, out, dim_size)
     39                  out: Optional[torch.Tensor] = None,
     40                  dim_size: Optional[int] = None) -> torch.Tensor:
---> 41     out = scatter_sum(src, index, dim, out, dim_size)
     42     dim_size = out.size(dim)
     43 

/home/ubuntu/anaconda3/lib/python3.9/site-packages/torch_scatter/scatter.py in scatter_sum(src, index, dim, out, dim_size)
     19             size[dim] = int(index.max()) + 1
     20         out = torch.zeros(size, dtype=src.dtype, device=src.device)
---> 21         return out.scatter_add_(dim, index, src)
     22     else:
     23         return out.scatter_add_(dim, index, src)

RuntimeError: index 51 is out of bounds for dimension 0 with size 32

I'm working on a bipartite graph.

• print(source.shape) ==> torch.Size([32, 3])
• print(target.shape) ==> torch.Size([100, 3])
• print(edge_index.size(0)) ==> 2
• print(edge_index.shape) ==> torch.Size([2, 608])

[Padarn]

Could you please provide a code example to reproduce?

[rusty1s]

In bipartite graphs, you need to pass in node features as a tuple:

conv((src_node_feat, dst_node_feat), edge_index)

[rusty1s]

You should pass in follow_batch=['x_s', 'x_t'] in your DataLoader to get both a batch vector for source and destination nodes.

[joshuajennysibbu]

    from torch_geometric.data import Data
    from torch_geometric.loader import DataLoader

   class BipartiteData(Data):
          def __init__(self, edge_index=None, follow_batch=None,num_nodes=None):
           super().__init__()
           self.edge_index = edge_index
        
          self.follow_batch=follow_batch
          self.num_nodes=num_nodes

       def __inc__(self, key, value, *args, **kwargs):
           if key == 'edge_index':
               return torch.tensor([[self.follow_batch[0].size(0)], [self.follow_batch[1].size(0)]])
         else:
              return super().__inc__(key, value, *args, **kwargs)

a = BipartiteData(edge_index_T,[username_feature_T,object_feature_T],len(username_feature_T)+len(object_feature_T))

Do you mean like this?

[rusty1s]

No, I mean like this:

loader = DataLoader(dataset, follow_batch=['x_s', 'x_t'], batch_size=...)

data = next(iter(loader))

assert = data.x_s_batch is not None
assert = data.x_t_batch is not None

[joshuajennysibbu]

still i got like this BipartiteDataBatch(edge_index=[2, 745], x_s=[32, 3], x_s_batch=[32], x_t=[100, 3], x_t_batch=[100], num_nodes=132, batch=[132], ptr=[2]). dataset parameter in Dataloader it's means output from BipartiteData class right?

[joshuajennysibbu]

I fixed it, thank you.