Closed sud0nick closed 5 years ago
I just realized there are a lot of issues that need to be resolved. These weren't caught during the pre-commit and push checks on my system. I'll review them and resubmit later.
Merging #33 into master will decrease coverage by
8.28%
. The diff coverage is69.56%
.
@@ Coverage Diff @@
## master #33 +/- ##
==========================================
- Coverage 67.24% 58.95% -8.29%
==========================================
Files 15 7 -8
Lines 348 173 -175
Branches 61 19 -42
==========================================
- Hits 234 102 -132
+ Misses 105 66 -39
+ Partials 9 5 -4
Impacted Files | Coverage Δ | |
---|---|---|
src/akro/__version__.py | 0% <0%> (ø) |
:arrow_up: |
src/akro/space.py | 57.14% <100%> (+4.2%) |
:arrow_up: |
src/akro/discrete.py | 61.76% <100%> (-16.81%) |
:arrow_down: |
src/akro/requires.py | 46.15% <46.15%> (ø) |
|
src/akro/box.py | 74.41% <66.66%> (-7.22%) |
:arrow_down: |
src/akro/tuple.py | 46.15% <71.42%> (-40.21%) |
:arrow_down: |
src/akro/dict.py | 62.96% <85.71%> (+22.48%) |
:arrow_up: |
... and 1 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 24d9cf0...61b5527. Read the comment docs.
See this thread for information regarding the changes: https://github.com/rlworkgroup/akro/issues/28 During your review please focus on
test_dict.py
andtuple.py
.The test case in the old
tf
andtheano
modules differs from the test case in theakro
module. Please advise on which to keep.Within the constructor for a Tuple it attempts to convert each component into a TF object and subsequently calls .dtypes on that object. For some reason, a TF object is never created (returns None) without any warnings or exceptions and I'm not sure why.