codecov-commenter
commented
2 years ago Codecov Report
Merging #6 (df8c10e) into main (bfd77ab) will decrease coverage by
2.28%. The diff coverage is69.66%.
@@ Coverage Diff @@
## main #6 +/- ##
==========================================
- Coverage 74.12% 71.83% -2.29%
==========================================
Files 7 7
Lines 286 245 -41
==========================================
- Hits 212 176 -36
+ Misses 74 69 -5 | Impacted Files | Coverage Δ | |
|---|---|---|
| src/context/storage.rs | 25.00% <14.28%> (+15.00%) |
:arrow_up: |
| src/gradient.rs | 29.41% <28.00%> (-48.37%) |
:arrow_down: |
| src/context/function.rs | 50.00% <50.00%> (+3.70%) |
:arrow_up: |
| src/context/mod.rs | 50.00% <60.00%> (-1.62%) |
:arrow_down: |
| src/tensor.rs | 60.46% <60.46%> (-12.87%) |
:arrow_down: |
| src/context/tape.rs | 88.88% <93.33%> (+7.07%) |
:arrow_up: |
| src/ops.rs | 99.09% <98.87%> (-0.91%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact),ø = not affected,? = missing dataPowered by Codecov. Last update bfd77ab...df8c10e. Read the comment docs.
Constant and variable tensors are now different types, which means, after getting lost on many computations the developer can know which of the resulting tensors are still constants or variables. Another benefit from this is that code is clearer now, as we don't need to assert at runtime if a tensor is a constant, for example when deciding if computing a gradient or not.
Aside from this, I've also spent 10 minutes designing a logo for the library, given I have 0 skill for this I am quite happy with the result!