Error during training: TypeError in GNN model

[chhnb]

Sure, here's a polished version of the issue:

---

Issue Description:

I have followed the instructions provided in the readme documentation to configure the environment and prepare the data. However, when I attempt to train the GNN (Graph Neural Network) model, I encounter the following error:

Error executing job with overrides: ['model.gnn_type=', 'seed=0']
Traceback (most recent call last):
  File "rlcompopt/train.py", line 1210, in main
    main_real(args)
  File "rlcompopt/train.py", line 1497, in main_real
    last_loss, eval_loss = policy_gradient(
  File "rlcompopt/train.py", line 541, in policy_gradient
    loss, kl, y_pred, logits = model(
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 1040, in forward
    output = self._run_ddp_forward(*inputs, **kwargs)
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 1000, in _run_ddp_forward
    return module_to_run(*inputs[0], **kwargs[0])
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1194, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/rlcompopt-0.0.1-py3.8.egg/rlcompopt/cl/models/gnn_pyg.py", line 921, in forward
    preds, _ = super().forward(
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/rlcompopt-0.0.1-py3.8.egg/rlcompopt/cl/models/gnn_pyg.py", line 553, in forward
    res, edge_feat, res_mid, edge_feat_mid = self.layers_encode(
  File "/home/chh/anaconda3/envs/rlcompopt/lib/python3.8/site-packages/rlcompopt-0.0.1-py3.8.egg/rlcompopt/cl/models/gnn_pyg.py", line 695, in layers_encode
    n_layers = len(layers)
TypeError: object of type 'NoneType' has no len()

Steps to Reproduce:

1. Configure the environment as per the documentation.
2. Prepare the data according to the instructions.
3. Execute the training script for the GNN model.

Expected Behavior:

The GNN model should train without encountering any errors.

Additional Information:

• Environment: Anaconda3
• Python Version: 3.8
• GNN Model Library: rlcompopt
• Operating System: Ubuntu 18.04

Please let me know if any further information is needed to diagnose and resolve this issue. Thank you.

---

Feel free to adjust any details or add more specific information as needed.

[youweiliang]

Hi @chhnb , It seems that the default configurations were overridden with a missing argument. Specifically, in the error message ['model.gnn_type=', 'seed=0'], no gnn_type was provided. Which script did you run that produced this error?

[chhnb]

I just ran the script using bash scripts/train_graph_gnn_type2_nvp.sh.