AttributeError: 'DGLGraph' object has no attribute '_use_graphbolt'

[dangnha]

🐛 Bug

I got the bug:

Creating minibatch pretraining dataloader...
Traceback (most recent call last):
  File "/home/s12gb-2/miniconda3/envs/graph_newset/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/home/s12gb-2/miniconda3/envs/graph_newset/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/home/s12gb-2/DangNha/reset_TXGNN/TxGNN/reproduce/train.py", line 50, in <module>
    TxGNN.pretrain(n_epoch = 1, 
  File "/home/s12gb-2/DangNha/reset_TXGNN/TxGNN/txgnn/TxGNN.py", line 155, in pretrain
    dataloader = dist.DistEdgeDataLoader(
  File "/home/s12gb-2/miniconda3/envs/graph_newset/lib/python3.10/site-packages/dgl/distributed/dist_dataloader.py", line 863, in __init__
    self.collator = EdgeCollator(g, eids, graph_sampler, **collator_kwargs)
  File "/home/s12gb-2/miniconda3/envs/graph_newset/lib/python3.10/site-packages/dgl/distributed/dist_dataloader.py", line 634, in __init__
    Collator.add_edge_attribute_to_graph(self.g, self.graph_sampler.prob)
  File "/home/s12gb-2/miniconda3/envs/graph_newset/lib/python3.10/site-packages/dgl/distributed/dist_dataloader.py", line 325, in add_edge_attribute_to_graph
    if g._use_graphbolt and data_name:
AttributeError: 'DGLGraph' object has no attribute '_use_graphbolt'

When i try to create a DistEdgeDataLoader for training. My code:

import dgl
import dgl.distributed as dist
dataloader = dist.DistEdgeDataLoader(
            self.G,                           # Graph (should be a distributed DGLGraph)
            train_eid_dict,                   # Edge IDs (dict for heterogeneous graphs)
            sampler,                          # Graph sampler (e.g., NeighborSampler)
            negative_sampler=Minibatch_NegSampler(self.G, 1, 'fix_dst'),  # For link prediction
            batch_size=batch_size,            # Batch size
            shuffle=True,                     # Shuffle edges during sampling
            drop_last=False,                  # Keep incomplete batches
            # exclude='reverse_types',        # Optional: for reverse edge exclusion
            # reverse_etypes=reverse_etypes,  # Optional: reverse edge types
            num_workers=0                     # Avoid multiprocessing issues (for debugging)
        )

My Environment

• DGL Version (e.g., 1.g.,0): dgl==2.4.0+cu121
• Backend Library & Version: Pytorch==2.4.0
• OS (e.g., Linux): Ubuntu
• How you installed DGL (conda, pip, source): pip install dgl==2.4.1 -f https://data.dgl.ai/wheels/torch-2.4/cu121/repo.html
• Python version: 3.10.16
• CUDA/cuDNN version (if applicable): 12.1

Question

This bug is come from dgl version right, how can i solve this bug or i have to wait the new update from dgl?