instadeepai / Mava

🦁 A research-friendly codebase for fast experimentation of multi-agent reinforcement learning in JAX

Home Page:https://id-mava.readthedocs.io

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[BUG] The `test_wrapper_env_obs_stacking` test fails intermittently.

KaleabTessera opened this issue · comments

Describe the bug

The test_wrapper_env_obs_stacking test fails intermittently.

Error:

assert not np.array_equal(old_obs, stacked_obs[(num_frames - 1) * size :])
E           assert not True
E            +  where True = <function array_equal at 0x7fb6ac14a280>(array([ 0.        ,  0.        , -0.4675673 , -0.21778622,  1.2842039 ,\n       -0.07670229,  0.54844725, -0.11278857, ...0259,  0.7331438 , -0.08803626,  0.7898616 ,  0.        ,\n        0.        ,  0.        ,  0.        ], dtype=float32), array([-0.        , -0.        , -0.4675673 , -0.21778622,  1.2842039 ,\n       -0.07670229,  0.54844725, -0.11278857, ...0259,  0.7331438 , -0.08803626,  0.7898616 ,  0.        ,\n        0.        ,  0.        ,  0.        ], dtype=float32))
E            +    where <function array_equal at 0x7fb6ac14a280> = np.array_equal

Skipping the test for now.

To Reproduce

Steps to reproduce the behavior:

  1. When github actions run e.g. https://github.com/instadeepai/Mava/actions/runs/3965580999/jobs/6795444338.

Expected behavior

Test to be consistent.

Context (Environment)

  • Mava (and other package versions): result of pip freeze > requirements.txt
  • OS

Additional context

Possibly a package version issue.

Possible Solution