module 'tensorflow' has no attribute 'tanh' - tensorflow

Question

I am trying to repeat what is shown in this tutorial: https://www.kaggle.com/alexisbcook/deep-reinforcement-learning
When I run this code:
# Check version of tensorflow
import tensorflow as tf
tf.__version__
I am getting this error:
module 'tensorflow' has no attribute 'version'
When I run this piece of code I am not getting any error:
from kaggle_environments import make, evaluate
from gym import spaces
class ConnectFourGym:
...
Then running this code without error:
# Create ConnectFour environment
env = ConnectFourGym(agent2="random")
But when I try to run the following code
import os
from stable_baselines.bench import Monitor
from stable_baselines.common.vec_env import DummyVecEnv
# Create directory for logging training information
log_dir = "ppo/"
os.makedirs(log_dir, exist_ok=True)
# Logging progress
monitor_env = Monitor(env, log_dir, allow_early_resets=True)
# Create a vectorized environment
vec_env = DummyVecEnv([lambda: monitor_env])
I am getting the following error:
module 'tensorflow' has no attribute 'tanh'
The error is pointing these lines causing the problem:
----> 2 from stable_baselines.bench import Monitor
...
----> 1 from stable_baselines.a2c import A2C
...
----> 1 from stable_baselines.a2c.a2c import A2C
...
----> 9 from stable_baselines.common import explained_variance, tf_util, ActorCriticRLModel, SetVerbosity, TensorboardWriter
...
----> 7 from stable_baselines.common.base_class import BaseRLModel, ActorCriticRLModel, OffPolicyRLModel, SetVerbosity,
...
---> 16 from stable_baselines.common.policies import get_policy_from_name, ActorCriticPolicy ...
--> 375 class LstmPolicy(RecurrentActorCriticPolicy):
Why this is happening? How can I fix this problem?

Answer 1

I fixed the problem by specifying TensorFlow version:
# Check version of tensorflow
!pip install -q 'tensorflow==1.15.0'
import tensorflow as tf
tf.__version__
This code snippet installs TensorFlow version 1.15.0 and prints
'1.15.0'