rlworkgroup / garage

A toolkit for reproducible reinforcement learning research.
MIT License
1.84k stars 309 forks source link

Pin atari-py to last release with packaged ROMs #2281

Closed gitanshu closed 3 years ago

gitanshu commented 3 years ago

This is a temporary fix. atari-py releases starting 0.2.7 don't install ROMs by default. While there are instructions given for how to install ROMs on their github, it is still incovenient for a garage user and adds an extra step in installation.

A better fix would be to figure out if it's possible to install ROMs as a post install step in setup.py

codecov[bot] commented 3 years ago

Codecov Report

Merging #2281 (c891bc9) into master (669733f) will increase coverage by 0.76%. The diff coverage is n/a.

Impacted file tree graph

@@            Coverage Diff             @@
##           master    #2281      +/-   ##
==========================================
+ Coverage   90.50%   91.26%   +0.76%     
==========================================
  Files         199      199              
  Lines       11761    10966     -795     
  Branches     1583     1387     -196     
==========================================
- Hits        10644    10008     -636     
+ Misses        840      696     -144     
+ Partials      277      262      -15     
Impacted Files Coverage Δ
src/garage/torch/optimizers/optimizer_wrapper.py 93.75% <0.00%> (-6.25%) :arrow_down:
src/garage/torch/optimizers/differentiable_sgd.py 86.66% <0.00%> (-3.34%) :arrow_down:
src/garage/torch/algos/trpo.py 91.66% <0.00%> (-3.08%) :arrow_down:
...rage/tf/optimizers/conjugate_gradient_optimizer.py 81.90% <0.00%> (-1.81%) :arrow_down:
src/garage/np/_functions.py 77.39% <0.00%> (-1.74%) :arrow_down:
src/garage/torch/algos/sac.py 97.14% <0.00%> (-1.66%) :arrow_down:
src/garage/torch/algos/maml.py 98.18% <0.00%> (-0.55%) :arrow_down:
src/garage/torch/algos/dqn.py 93.07% <0.00%> (-0.48%) :arrow_down:
src/garage/torch/algos/ddpg.py 96.66% <0.00%> (-0.33%) :arrow_down:
src/garage/torch/algos/vpg.py 100.00% <0.00%> (ø)
... and 8 more

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 669733f...c891bc9. Read the comment docs.

gitanshu commented 3 years ago

Does this fix the CI? Either way, this looks good to merge to me.

I ran the nightly tests locally and this fixed it. So CI should be fine.