- Dec 06, 2024
-
-
Marco Fossati authored
Published at https://meta.wikimedia.org/wiki/Machine_learning_models/Production/gogologo
-
- Sep 26, 2024
-
-
Marco Fossati authored
- Sep 20, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-
- Aug 29, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-
- Aug 20, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
- use ten classes from https://phabricator.wikimedia.org/T370861 - replace `RandomContrast` and `RandomTranslation` image augmentation layers with `RandAugment` - optional learning rate schedule with cosine decay - save model history - refactor
-
Marco Fossati authored
-
- Aug 07, 2024
-
-
Marco Fossati authored
- remove positional CLI args - boolean CLI flag to dump a PetScan result into a json.gz file, disabled by default - either pass a PetScan ID or a PetScan json.gz result as required input
-
- Aug 05, 2024
-
-
Marco Fossati authored
- use internal APIs as per https://wikitech.wikimedia.org/wiki/Data_Platform/Internal_API_requests - allowed file extensions as per https://commons.wikimedia.org/wiki/Special:MediaStatistics - compute the optimal thumbnail width
-
- Jul 25, 2024
-
-
Marco Fossati authored
-
- Apr 04, 2024
-
-
Marco Fossati authored
-
- Apr 03, 2024
-
-
Marco Fossati authored
lw_prototype: add LogoDetectionModel class See merge request !9
-
kevinbazira authored
Wrap prototype functionality into a LogoDetectionModel class with its respective methods and type hinting. This will make it easier to integrate with KServe. Bug: T358676
-
Marco Fossati authored
lw_prototype: image download error handling See merge request !8
-
- Apr 02, 2024
-
-
kevinbazira authored
Add error handling to catch issues when downloading images from a URL and saving them in temporary directory. Bug: T358676
-
- Mar 29, 2024
-
-
Marco Fossati authored
lw_prototype: validate input data See merge request !7
-
- Mar 28, 2024
-
-
kevinbazira authored
Add function to validate the input data to ensure it has the required fields and their values are of the expected types. Bug: T358676
-
Marco Fossati authored
Improve functionality See merge request !6
-
- Mar 27, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-
Marco Fossati authored
-
Marco Fossati authored
-
- Mar 26, 2024
-
-
Marco Fossati authored
- image download: stream-read the HTTP response into chunks - input dataset: no need for labels, it's a classification dataset - remove binary label mode, otherwise predictions are different - fix expected output
-
-
- Mar 18, 2024
-
-
Marco Fossati authored
- classify an image dataset with a pre-trained model - count how many images belong to a given set of WordNet synsets
-
- Feb 29, 2024
-
-
Marco Fossati authored
-
- Feb 01, 2024
-
-
Marco Fossati authored
-
- Jan 24, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-
- Jan 23, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-
Marco Fossati authored
-
- Jan 15, 2024
-
-
-
Marco Fossati authored
-
Marco Fossati authored
-
- Jan 12, 2024
-
-
Marco Fossati authored
-
Marco Fossati authored
-