TL;DR: The authors pretrain the model to classify images into Wordnet synsets[a] that appear in the caption, using a standard Cross Entropy loss. They keep the number of classes relatively small by removing any synsets that don't show up in captions at least 500 times in the dataset. It seems to work well. My immediate question is: Why not classify among the entire hierarchy of all Wordnet synsets? --- [a]... - Source: Hacker News / about 2 months ago

To operationalize this intuition, the Microsoft and UC Berkeley researchers use WordNet and Wiktionary to augment the text in image-text pairs. The concept itself is augmented for isolated concepts, such as the class labels in ImageNet, whereas for captions (such as from GCC), the least common noun phrase is augmented. Equipped with this additional structured knowledge, contrastively pretrained models exhibit... - Source: dev.to / 3 months ago

If you like this, definitely check out WordNet (https://wordnet.princeton.edu/). - Source: Hacker News / 6 months ago

I didn't understand well what you meant, but maybe this site can help you: https://wordnet.princeton.edu/. Source: over 1 year ago

What I'd do is work with a huge database like WordNet and then try to "extrapolate" BIP39 to 4096 words by creating queries against WordNet to obtain words meeting the constraints you'd like to keep. Source: over 1 year ago

I'm pretty sure, you'll find companies like this one which provide a nice GUI for helping with l10n or this one which offers translation services or this page that offers to convert between different formats, one of which probably has a nice GUI tool. Found them by 20 secs of googling. Source: about 1 year ago

Localise has no problem reading the json files or export to json, we recently started using it in collaboration with external translators. Source: about 1 year ago

Actually I don't have "my" app. But in our app we use https://lokalise.com/ to localize it even to Norwegian. Our team is not the most expensive company in the world btw. And we don't have 1B+ users all over the world. Source: about 1 year ago

Internationalization (i18n) pain for a documentation project is a process problem, not a feature gap. Documentation frameworks are not meant to translate your developer docs for you into the language of your choice. Some frameworks might offer i18n support, like the Crowdin support in Docusauraus v2. With Jekyll, you have to pick a theme like this one. I doubt if the reST or adoc frameworks would differ much from... - Source: dev.to / over 1 year ago

I've been thinking about building a micro saas very similar to this after having so many issues coordinating the localization of several products I manage. The only robust options in the market are incredibly expensive (for example https://lokalise.com/). Source: almost 2 years ago