“from recsim.agents import full_slate_q_agent” gets new problem

[TobyGE]

I had maybe a new error on Linux...

Tensorflow Attribute Error: module 'tensorflow_core._api.v2.train' has no attribute 'RMSProp Optimizer'

google/dopamine#121
This link relates to this error, unfortunately, no one answers...

Originally posted by @TobyGE in #5 (comment)

[cwhsu-google]

Could you run Dopamine/Recsim using Tensorflow 1.1.5?
https://github.com/google/dopamine/blob/master/README.md#mac-os-x
We will think about a long term solution. Thanks.

[TobyGE]

Could you run Dopamine/Recsim using Tensorflow 1.1.5?
https://github.com/google/dopamine/blob/master/README.md#mac-os-x
We will think about a long term solution. Thanks.

At first, I couldn't because of the wrong tf version. After replacing tf 2.0 with tf 1.15, I run the test again. There were some WARNINGs but I thought the test worked.

Then I imported full_slate_q_agent again, however, this time still failed. A new error existed, shown as below.

[Amir-Abushanab]

Didn't want to create a new issue as I believe it's the same problem.

Running the third cell in the overview notebook on Colab https://github.com/google-research/recsim/blob/master/recsim/colab/RecSim_Overview.ipynb gives the following error

ModuleNotFoundError: No module named 'tensorflow.contrib'

---

Update: changing the second line to %tensorflow_version 1.x provides a temporary fix, but the Colab environment will still give a warning for not using TF 2.0