search

rlworkgroup / garage

A toolkit for reproducible reinforcement learning research.
MIT License
1.88k stars 310 forks source link

Add Noisy DQN #2152

Open maliesa96 opened 4 years ago

maliesa96 commented 4 years ago

Note this builds on top of the dueling and double PRs so the diff will be redundant. Also will add tests shortly.

codecov[bot] commented 4 years ago

Codecov Report

Merging #2152 (f40ee60) into master (e32cd06) will decrease coverage by 23.47%. The diff coverage is 19.35%.

Impacted file tree graph

@@             Coverage Diff             @@
##           master    #2152       +/-   ##
===========================================
- Coverage   91.48%   68.00%   -23.48%     
===========================================
  Files         195      198        +3     
  Lines       10675    11037      +362     
  Branches     1340     1403       +63     
===========================================
- Hits         9766     7506     -2260     
- Misses        657     3147     +2490     
- Partials      252      384      +132
Impacted Files Coverage Δ
...rc/garage/torch/modules/multi_headed_mlp_module.py 83.67% <ø> (-16.33%) :arrow_down:
...arage/torch/q_functions/discrete_cnn_q_function.py 42.85% <ø> (-50.00%) :arrow_down:
src/garage/torch/modules/discrete_cnn_module.py 16.00% <7.89%> (-84.00%) :arrow_down:
src/garage/torch/modules/noisy_mlp_module.py 22.36% <22.36%> (ø)
src/garage/torch/algos/dqn.py 86.36% <40.00%> (-6.44%) :arrow_down:
src/garage/tf/algos/te.py 12.50% <0.00%> (-87.50%) :arrow_down:
src/garage/tf/algos/te_npo.py 9.25% <0.00%> (-85.00%) :arrow_down:
src/garage/torch/modules/categorical_cnn_module.py 17.94% <0.00%> (-82.06%) :arrow_down:
src/garage/sampler/ray_sampler.py 18.34% <0.00%> (-81.66%) :arrow_down:
.../tf/policies/gaussian_mlp_task_embedding_policy.py 19.67% <0.00%> (-80.33%) :arrow_down:
... and 106 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 e32cd06...f40ee60. Read the comment docs.