# Reference runtime environment? python + tensorflow/pytorch

**URL:** <https://discuss.ray.io/t/reference-runtime-environment-python-tensorflow-pytorch/3805>\
**Category:** RLlib\
**Created:** [October 14, 2021, 12:47am UTC](https://discuss.ray.io/t/reference-runtime-environment-python-tensorflow-pytorch/3805 "2021-10-14T00:47:55Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![RickLan](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/ricklan/32/901_2.png) [@RickLan](https://discuss.ray.io/u/RickLan)\
**Post date:** [October 14, 2021, 12:47am UTC](https://discuss.ray.io/t/reference-runtime-environment-python-tensorflow-pytorch/3805/1 "2021-10-14T00:47:55Z")

</div>

Is there a reference runtime environment for RLlib? Like a reproducible one from auto testing. I’m trying to isolate a problem and seems to be difference between between Ray version and tensorflow/pytorch. Having known working cases would help a lot.

Edit: to elaborate, I have code written in Ray 1.4 but doesn’t run in Ray 1.6 or 1.7. Unfortunately, I couldn’t reproduce the same runtime I had in 1.4.

cc @sven1977