support ArrayLike data in to_xarray #825
Merged
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
egpbos
added a commit
to dianna-ai/explainable_embedding
that referenced
this pull request
Jul 26, 2024
Either the image_arr is wrong (load_img runs a preprocess function that is meant for use with ResNet50, perhaps that's the issue) or the uint8 conversion kills something, but either way, the "fix" introduced there completely ruined the output for image vs captioning experiments! This commit fixes things again. Note that the np.array(input_image) could be replaced by a bare input_image if dianna PR dianna-ai/dianna#825 is accepted.
egpbos
force-pushed
the
to_xarray_arraylike
branch
from
39ad0b5 to
c07f6d5
Compare
This commit makes dianna swallow more kinds of data items without users having to mangle them manually themselves. As long as your data is numpy ArrayLike it will go. The only blocker for this was the direct use of the .ndim attribute on data, which assumes it is already a numpy array rather than something that can trivially be converted to a numpy array. A PIL.Image is an example of such an ArrayLike type. We use this in explainable_embedding to feed items into the OpenAI CLIP model.
loostrum
force-pushed
the
to_xarray_arraylike
branch
from
c07f6d5 to
de7da33
Compare
|
I rebased on main to get the linter fix in here. The changes themselves look good, nice generalization of to_xarray! Linter's happy as well, so let's merge. |
loostrum
approved these changes
Jul 31, 2024
|
@egpbos many checks are not passing. |
|
Yep, those are mentioned above. They are unrelated to this PR. |
|
closing and reopening for rerunning tests |
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.
This commit makes dianna swallow more kinds of data items without users having to mangle them manually themselves. As long as your data is numpy ArrayLike it will go. The only blocker for this was the direct use of the .ndim attribute on data, which assumes it is already a numpy array rather than something that can trivially be converted to a numpy array. A PIL.Image is an example of such an ArrayLike type. We use this in explainable_embedding to feed items into the OpenAI CLIP model.
I sprinkled in some nice type hinting, of course, but not for the other parameters, because they would become monstrosities.