slimgroup / InvertibleNetworks.jl

A Julia framework for invertible neural networks
MIT License
148 stars 20 forks source link

add activation parameter to conditional multiscale hint #68

Closed rafaelorozco closed 1 year ago

rafaelorozco commented 1 year ago

day and night different in performance when you constrain the activation lower value. Will add this to all networks, in a bit for this PR

codecov[bot] commented 1 year ago

Codecov Report

Base: 88.11% // Head: 88.09% // Decreases project coverage by -0.02% :warning:

Coverage data is based on head (bfb1149) compared to base (f6b2d17). Patch coverage: 90.62% of modified lines in pull request are covered.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## master #68 +/- ## ========================================== - Coverage 88.11% 88.09% -0.03% ========================================== Files 33 33 Lines 2542 2545 +3 ========================================== + Hits 2240 2242 +2 - Misses 302 303 +1 ``` | [Impacted Files](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup) | Coverage Δ | | |---|---|---| | [...rc/networks/invertible\_network\_conditional\_glow.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL25ldHdvcmtzL2ludmVydGlibGVfbmV0d29ya19jb25kaXRpb25hbF9nbG93Lmps) | `82.89% <0.00%> (-1.11%)` | :arrow_down: | | [src/layers/invertible\_layer\_hint.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL2xheWVycy9pbnZlcnRpYmxlX2xheWVyX2hpbnQuamw=) | `94.39% <50.00%> (ø)` | | | [src/networks/invertible\_network\_irim.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL25ldHdvcmtzL2ludmVydGlibGVfbmV0d29ya19pcmltLmps) | `92.20% <50.00%> (ø)` | | | [src/conditional\_layers/conditional\_layer\_hint.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL2NvbmRpdGlvbmFsX2xheWVycy9jb25kaXRpb25hbF9sYXllcl9oaW50Lmps) | `99.18% <100.00%> (ø)` | | | [...rc/networks/invertible\_network\_conditional\_hint.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL25ldHdvcmtzL2ludmVydGlibGVfbmV0d29ya19jb25kaXRpb25hbF9oaW50Lmps) | `95.12% <100.00%> (ø)` | | | [.../invertible\_network\_conditional\_hint\_multiscale.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL25ldHdvcmtzL2ludmVydGlibGVfbmV0d29ya19jb25kaXRpb25hbF9oaW50X211bHRpc2NhbGUuamw=) | `79.31% <100.00%> (+0.20%)` | :arrow_up: | | [src/networks/invertible\_network\_hint\_multiscale.jl](https://codecov.io/gh/slimgroup/InvertibleNetworks.jl/pull/68/diff?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup#diff-c3JjL25ldHdvcmtzL2ludmVydGlibGVfbmV0d29ya19oaW50X211bHRpc2NhbGUuamw=) | `83.33% <100.00%> (ø)` | | Help us with your feedback. Take ten seconds to tell us [how you rate us](https://about.codecov.io/nps?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup). Have a feature suggestion? [Share it here.](https://app.codecov.io/gh/feedback/?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=slimgroup)

:umbrella: View full report at Codecov.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.