search

rlworkgroup / garage

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

Prevent broken tf-prob installs #2289

Closed krzentner closed 3 years ago

krzentner commented 3 years ago

The latest version of tensorflow-probability (0.13.0) requires tensorflow 2.5.0. However, we do not allow 2.5.0, since it depends on keras-nightly, which several of our packaging systems can't support. Unfortunately, tensorflow-probability 0.13.0 fails to declare this dependency, and fails at runtime instead. This results in broken installs.

This change should prevent those broken installs and fix the CI. It can probably be rolled back once tensorflow doesn't depend on keras-nightly (hopefully 2.5.1).

codecov[bot] commented 3 years ago

Codecov Report

Merging #2289 (c09bdc4) into master (4cdedcd) will increase coverage by 0.79%. The diff coverage is n/a.

Impacted file tree graph

@@            Coverage Diff             @@
##           master    #2289      +/-   ##
==========================================
+ Coverage   90.51%   91.31%   +0.79%     
==========================================
  Files         199      199              
  Lines       11761    10966     -795     
  Branches     1583     1387     -196     
==========================================
- Hits        10646    10014     -632     
+ Misses        837      693     -144     
+ Partials      278      259      -19
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:
src/garage/torch/algos/sac.py 97.14% <0.00%> (-1.66%) :arrow_down:
src/garage/sampler/multiprocessing_sampler.py 90.12% <0.00%> (-1.24%) :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)
src/garage/torch/algos/pearl.py 48.82% <0.00%> (+0.74%) :arrow_up:
... and 7 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 4cdedcd...c09bdc4. Read the comment docs.