It stopped in "Successfully opened dynamic library libcuda.so.1" about multi-agent-emergence-environments

[HACLANG]

(base) root@lang:~# pip show tensorflow-gpu
Name: tensorflow-gpu
Version: 2.2.0
Summary: TensorFlow is an open source machine learning framework for everyone.
Home-page: https://www.tensorflow.org/
Author: Google Inc.
Author-email: [email protected]
License: Apache 2.0
Location: /root/anaconda3/lib/python3.7/site-packages
Requires: keras-preprocessing, wheel, six, protobuf, tensorflow-estimator, wrapt, grpcio, google-pasta, h5py, absl-py, tensorboard, scipy, astunparse, opt-einsum, gast, termcolor, numpy
Required-b

and I changed some code to make it run on tensorflow 2.2:
/multi-agent-emergence-environments/ma_policy/load_policy.pyfile:

def load_policy(path, env=None, scope='policy'):
    '''
        Load a policy.
        Args:
            path (string): policy path
            env (Gym.Env): This will update the observation space of the
                policy that is returned
            scope (string): The base scope for the policy variables
    '''
    # TODO this will probably need to be changed when trying to run policy on GPU
    # try  change  tf.get_default_session  to  tf.compat.v1.get_default_session
    if tf.compat.v1.get_default_session() is None:
        # try  change  tf.ConfigProto  to  tf.compat.v1.ConfigProto
        tf_config = tf.compat.v1.ConfigProto(
            inter_op_parallelism_threads=1,
            intra_op_parallelism_threads=1)
        # try  change  tf.ConfigProto  to  tf.compat.v1.ConfigProto
        sess = tf.compat.v1.Session(config=tf_config)
        sess.__enter__()