Libinvent RL Execution Error: Possible Issue with Scaffold SMILES?

[yukisoya]

I've encountered an issue while running Reinforcement Learning (RL) with Libinvent. The execution is interrupting immediately due to an error. I suspect the problem might be related to the SMILES in the scaffold.smi file.

Problem Description

When I simplified the content of scaffold.smi to just [*:0]c1ccccc1, the RL process proceeded without issues. This leads me to believe that the error might be caused by the SMILES notation in the original scaffold.smi file. (It's worth noting that the original scaffold represents a rather rare chemical structure. Due to confidentiality constraints, I cannot share the actual SMILES of the scaffold I used, sorry.)

Error Message

Here's the full traceback of the error I'm encountering:

Traceback (most recent call last):
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/RL/run_staged_learning.py", line 377, in run_staged_learning
    terminate = optimize(package.terminator)
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/RL/learning.py", line 131, in optimize
    results = self.score()
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/RL/learning.py", line 201, in score
    results = self.scoring_function(
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/scoring/scorer.py", line 105, in compute_results
    transform_result = compute_transform(
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/scoring/compute_scores.py", line 111, in compute_transform
    component_results = compute_component_scores(
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/scoring/compute_scores.py", line 63, in compute_component_scores
    key0 = next(iter(cache))
StopIteration

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/miniconda3/envs/reinvent4/bin/reinvent", line 8, in <module>
    sys.exit(main())
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/Reinvent.py", line 292, in main
    runner(input_config, actual_device, tb_logdir, responder_config)
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/RL/run_staged_learning.py", line 331, in run_staged_learning
    with Handler() as handler:
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/handler.py", line 65, in __exit__
    self.save()
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/handler.py", line 107, in save
    data = self._callback()
  File "/home/miniconda3/envs/reinvent4/lib/python3.10/site-packages/reinvent/runmodes/RL/learning.py", line 181, in get_state_dict
    model_dict = self._state_info["agent"].get_save_dict()
KeyError: 'agent'

Questions

1. Can you confirm if the SMILES in scaffold.smi is likely the cause of this error?
2. If so, could you provide advice on how to resolve this issue?
3. Would transfer learning of the Libinvent model be a practical solution in this case?

Any insights or suggestions would be greatly appreciated. Thank you in advance for your help!

[yukisoya]

The above error was caused by the fact that the scaffold structure matched the alert structure set in RL.
Sorry for this elementary error!