feature(zjow): impala policy for continuous action space

[zjowowen]

Add impala policy for continuous action space.
Support action reparameterized by dimension-wise independent gaussian disribution.

[codecov]

Codecov Report

Merging #551 (a0ee26d) into main (75d8644) will increase coverage by 0.06%.
The diff coverage is 91.50%.

@@            Coverage Diff             @@
##             main     #551      +/-   ##
==========================================
+ Coverage   85.36%   85.43%   +0.06%     
==========================================
  Files         532      532              
  Lines       43861    43943      +82     
==========================================
+ Hits        37444    37544     +100     
+ Misses       6417     6399      -18     

Flag	Coverage Δ
unittests	85.43% <91.50%> (+0.06%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
ding/policy/impala.py	86.39% <72.72%> (-4.59%)	⬇️
ding/model/wrapper/model_wrappers.py	89.67% <100.00%> (ø)
ding/policy/ppo.py	86.35% <100.00%> (+0.16%)	⬆️
ding/rl_utils/__init__.py	100.00% <100.00%> (ø)
ding/rl_utils/isw.py	100.00% <100.00%> (ø)
ding/rl_utils/tests/test_vtrace.py	100.00% <100.00%> (ø)
ding/rl_utils/vtrace.py	93.22% <100.00%> (+2.97%)	⬆️
ding/worker/learner/comm/flask_fs_learner.py	91.25% <0.00%> (-0.63%)	⬇️
ding/framework/parallel.py	85.49% <0.00%> (+0.51%)	⬆️
ding/framework/event_loop.py	96.49% <0.00%> (+3.50%)	⬆️
... and 3 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[zjowowen]

bipedalwalker-v3 by impala policy