Add a guard to avoid NaNs appearing in the conversion from rgb to hsv. #70
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
copybara-service
bot
force-pushed
the
test_510171813
branch
2 times, most recently
from
32be0e0 to
7fd3049
Compare
The added test shows an encountered in the wild error, where this created NaNs in the gradient computation. The problem seems to be related to subnormals floating point numbers. Annoyingly, for this to be a correct explanation, it would have been fixed by setting the epsilon value at `jnp.finfo(image_rgb.dtype).smallest_normal`, but for example, in the given test case, the offending numbers are ~1e-34, so these would not be masked (`smallest_normal` being ~1e-38). Setting the threshold to `jnp.finfo(image_rgb.dtype).eps` (~1e-7) seems to solve the issue. PiperOrigin-RevId: 510393337
copybara-service
bot
force-pushed
the
test_510171813
branch
from
7fd3049 to
00bad1d
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add a guard to avoid NaNs appearing in the conversion from rgb to hsv.
The added test shows an encountered in the wild error, where this created NaNs
in the gradient computation. The problem seems to be related to subnormals
floating point numbers.
Annoyingly, for this to be a correct explanation, it would have been fixed by
setting the epsilon value at
jnp.finfo(image_rgb.dtype).smallest_normal, butfor example, in the given test case, the offending numbers are ~1e-34, so these
would not be masked (
smallest_normalbeing ~1e-38). Setting the threshold tojnp.finfo(image_rgb.dtype).eps(~1e-7) seems to solve the issue.