Rllib PPOTrainer提供UnboundLocalError

Question

如本教程所述运行PPOTrainer时出现UnboundLocalError

当我仅在rllib教程中运行教程2时，就会发生这种情况。

    agent = PPOTrainer(config, 'CartPole-v0')

我得到以下

---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
<ipython-input-5-1146d5f802a5> in <module>
      6 config['num_cpus_per_worker'] = 0  # This avoids running out of resources in the notebook environment when this cell is re-executed
      7 
----> 8 agent = PPOTrainer(config, 'CartPole-v0')

~/anaconda3/lib/python3.7/site-packages/ray/rllib/agents/trainer.py in __init__(self, config, env, logger_creator)
    309             logger_creator = default_logger_creator
    310 
--> 311         Trainable.__init__(self, config, logger_creator)
    312 
    313     @classmethod



tensorflow/python/framework/fast_tensor_util.pyx in tensorflow.python.framework.fast_tensor_util.AppendBoolArrayToTensorProto()

~/anaconda3/lib/python3.7/site-packages/numpy/lib/type_check.py in asscalar(***failed resolving arguments***)
    545     warnings.warn('np.asscalar(a) is deprecated since NumPy v1.16, use '
    546                   'a.item() instead', DeprecationWarning, stacklevel=1)
--> 547     return a.item()
    548 
    549 #-----------------------------------------------------------------------------

UnboundLocalError: local variable 'a' referenced before assignment