search

james77777778 / keras-aug

A library that includes Keras 3 preprocessing and augmentation layers, providing support for various data types such as images, labels, bounding boxes, segmentation masks, and more.
Apache License 2.0
16 stars 0 forks source link

Fix numerical issuse when using mixed_precision #102

Closed james77777778 closed 1 year ago

james77777778 commented 1 year ago

This PR enables the support of:

codecov[bot] commented 1 year ago

Codecov Report

Patch coverage: 92.30% and project coverage change: -1.73 :warning:

Comparison is base (73820fc) 88.68% compared to head (1f150c1) 86.96%.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## main #102 +/- ## ========================================== - Coverage 88.68% 86.96% -1.73% ========================================== Files 54 54 Lines 3836 3859 +23 ========================================== - Hits 3402 3356 -46 - Misses 434 503 +69 ``` | [Impacted Files](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu) | Coverage Δ | | |---|---|---| | [...rs/augmentation/geometry/random\_crop\_and\_resize.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vZ2VvbWV0cnkvcmFuZG9tX2Nyb3BfYW5kX3Jlc2l6ZS5weQ==) | `87.36% <75.00%> (-0.55%)` | :arrow_down: | | [keras\_aug/layers/augmentation/auto/aug\_mix.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vYXV0by9hdWdfbWl4LnB5) | `88.14% <80.00%> (+0.05%)` | :arrow_up: | | [...\_aug/layers/augmentation/geometry/random\_rotate.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vZ2VvbWV0cnkvcmFuZG9tX3JvdGF0ZS5weQ==) | `89.65% <85.71%> (-0.71%)` | :arrow_down: | | [...\_aug/layers/augmentation/geometry/random\_affine.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vZ2VvbWV0cnkvcmFuZG9tX2FmZmluZS5weQ==) | `94.97% <93.75%> (-0.41%)` | :arrow_down: | | [...as\_aug/layers/augmentation/geometry/random\_crop.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vZ2VvbWV0cnkvcmFuZG9tX2Nyb3AucHk=) | `93.49% <100.00%> (+0.10%)` | :arrow_up: | | [...yers/augmentation/geometry/random\_zoom\_and\_crop.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vZ2VvbWV0cnkvcmFuZG9tX3pvb21fYW5kX2Nyb3AucHk=) | `77.19% <100.00%> (ø)` | | | [...\_aug/layers/augmentation/intensity/random\_clahe.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vaW50ZW5zaXR5L3JhbmRvbV9jbGFoZS5weQ==) | `33.33% <100.00%> (-60.19%)` | :arrow_down: | | [...yers/augmentation/intensity/random\_color\_jitter.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vaW50ZW5zaXR5L3JhbmRvbV9jb2xvcl9qaXR0ZXIucHk=) | `93.87% <100.00%> (+0.06%)` | :arrow_up: | | [...as\_aug/layers/augmentation/intensity/random\_hsv.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vaW50ZW5zaXR5L3JhbmRvbV9oc3YucHk=) | `92.40% <100.00%> (+0.09%)` | :arrow_up: | | [keras\_aug/layers/augmentation/mix/cut\_mix.py](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu#diff-a2VyYXNfYXVnL2xheWVycy9hdWdtZW50YXRpb24vbWl4L2N1dF9taXgucHk=) | `100.00% <100.00%> (ø)` | | | ... and [4 more](https://app.codecov.io/gh/james77777778/keras-aug/pull/102?src=pr&el=tree-more&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HongYu) | |

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